No title

Aims and Scope The Journal of Hazardous Materials publishes full length research papers, reviews, case studies and short communications which improve our understanding of the hazards and risks certain materials pose to people and the environment or deal with ways of controlling these hazards and associated risks. The Editors particularly welcome proposals for review articles. To limit the scope the following areas are excluded: standard municipal and waste and wastewater treatment processes; manufacturing of explosives. The Journal publishes high-impact contributions on: • Characterization of the harmful effects of chemicals and materials • Measurement and monitoring of hazardous materials • Transport and fate of HM in environment • Risk assessment and management • Physico-chemical and separation processes • Biological processes • Advanced oxidation processes • Safer and cleaner technologies The Journal publishes papers with significant novelty and scientific impact. The Editors reserve the right to decline, without external review, papers that do not meet these criteria, including papers that: • Are very similar to previous publications, with changed target substrates, employed materials, analyzed sites or experimental methods, • Deal with parameter optimization of known processes without new concepts and/or interpretations, • Report the environmental analysis and monitoring of specific geographic areas without presenting new insights and/or hypothesis testing, or • Do not focus on the environmental relevance and significance of the studied systems or materials. Editors Professor Diana S. Aga, Department of Chemistry, State University of New York (SUNY) at Buffalo, 359 Natural Sciences Complex, Buffalo, NY 142603000, USA, Tel: (716) 645-4220, E-mail: [email protected] Professor Wonyong Choi, School of Environmental Science & Engineering Pohang University of Science and Technology (POSTECH) Pohang, Korea 790-784 Phone: +82-54-279-2283 Fax: 82-54-279-8299 E-mail:[email protected] Professor Andrew Daugulis, Department of Chemical Engineering, Queen’s University, Dupuis Hau, Kingston, K7L 3N6, Canada, Tel.: (613) 533-2784; Fax: (613) 533-6637; E-mail: [email protected] Professor Gianluca Li Puma, Department of Chemical Engineering, Loughborough University, Loughborough LE11 3TU, United Kingdom. Tel: +44 (0) 1509 222 510; Fax: +44 (0) 1509 223 923; E-mail: [email protected] Professor Gerasimos Lyberatos, School of Chemical Engineering, National Technical University of Athens, Zografou Campus, Athens 15780, Greece. Tel.: +302107723256; Fax: +302107723155; E-mail: [email protected] Professor Tay Joo Hwa, Division of Environmental & Water Resources Engineering, School of Civil & Environmental Engineering, Nanyang Technology University, Nanyang Avenue, Singapore 639798, Tel: +65 6790 6228, Fax: +65 6791 0676, E-mail: [email protected] Associate Editors Jaesang Lee, Water Environment Center, National Agenda Research Division, Korea Institute of Science and Technology (KIST), 39-1 Hawolgok-Dong, Seongbuk-Gu, 136-791 Seoul, South Korea, Tel: +82-2-958-6947, E-mail: [email protected] Constantina Marazioti, Laboratory of Biochemical & Environmental Technology, Dept. of Chemical Engineering, University of Patras, Patras, Greece, Tel: +30 6976 128157, E-mail: [email protected] Editorial Board R. Andreozzi (Naples, Italy) J. Baeyens (Tremelo, Belgium) A. Ballester (Madrid, Spain) E. Brillas (Barcelona, Spain) V. Cozzani (Bologna, Italy) G. de la Rosa (Guanajuato, Mexico) D. Dong (Changchun, China) M. Erdem (Elazig˘, Turkey) S. Esplugas (Barcelona, Spain) M. Fingas (Alberta, Canada) V. K. Gupta (Roorkee, India) O. Hamdaoui (Annaba, Algeria)

W.-J. Lee (Tainan, Taiwan, ROC) P. Lianos (Patras, Greece) E. Lima (Porto Alegre, Brazil) T. Mangialardi (Rome, Italy) A. Mittal (Bhopal, India) D. Mohan (New Delhi, India) C. Mulligan (Montreal, Canada) A. Nzihou (Albi, France) J.M. Park (Pohang-si, Republic of Korea) H.J. Pasman (Delft, The Netherlands) S. Pavlostathis (Atlanta, GA, USA) H. Potgieter (WITS, South Africa)

P. Randall (Cincinnatti, OH, USA) K. Reddy (Chicago, IL, USA) J. Sneddon (Lake Charles, Louisiana, USA) M. Soylak (Kayseri, Turkey) M. Talawar (The Hague, Netherlands) T. Townsend (Gainesville, FL, USA) S. Tunali Akar (Eskis¸ehir, Turkey) S. Wang (Perth, WA, Australia) H.-Q. Yu (Hefei, China) J. Yu (Wuhan, China) S. Zeman (Pardubice, Czech Republic) A.I. Zouboulis (Thessaloniki, Greece)

Publication schedule and subscription information

Journal of Hazardous Materials (ISSN 0304-3894). For 2011 volumes 182–192 are scheduled for publication. Subscription prices are available upon request from the Publisher or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/jhazmat). Further information is available on this journal and other Elsevier Products through Elsevier’s website (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside Europe). Priority rates are available upon request. Claims for missing issues should be made within six months of the date of dispatch. Orders, claims, and journal enquiries: please contact the Elsevier Customer Service Department nearest you: St. Louis: Elsevier Customer Service Department, 3251 Riverport Lane, Maryland Heights, MO 63043, USA; phone: (877) 8397126 [toll free within the USA]; (+1) (314) 4478878 [outside the USA]; fax: (+1) (314) 4478077; e-mail: [email protected] Oxford: Elsevier Customer Service Department, The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865) 843970; e-mail: [email protected] Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3) 5561 5037; fax: (+81) (3) 5561 5047; e-mail: [email protected] Singapore: Elsevier Customer Service Department, 3 Killiney Road, #08-01 Winsland House I, Singapore 239519; phone: (+65) 63490222; fax: (+65) 67331510; e-mail: [email protected] doi:10.1016/S0304-3894(11)01319-7

Journal of Hazardous Materials 196 (2011) 1–15

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Review

Effects of the presence of sulfonamides in the environment and their influence on human health b ´ Wojciech Baran a , Ewa Adamek a,∗ , Justyna Ziemianska , Andrzej Sobczak a,b a b

Silesian Medical University, Department of General and Analytical Chemistry, Jagiello´ nska 4, 41-200 Sosnowiec, Poland Institute of Occupational Medicine and Environmental Health, Ko´scielna 13, 41-200 Sosnowiec, Poland

a r t i c l e

i n f o

Article history: Received 10 March 2011 Received in revised form 22 July 2011 Accepted 31 August 2011 Available online 6 September 2011 Keywords: Sulfonamides Biotransformation Ecotoxicity Environmental risk Drug resistance

a b s t r a c t World production and consumption of pharmaceuticals has been steadily increasing. Anti-infectives have been particularly important in modern therapy of microbial infection. Sulfonamides have been widely used for a long time as anti-infectives and are still widely prescribed today. This review presents the most common types of sulfonamides used in healthcare and veterinary medicine and discusses the problems connected with their presence in the biosphere. Based on the analysis of over 160 papers, it was found that small amounts of sulfonamides present in the environment were mainly derived from agricultural activities. These drugs have caused changes in the population of microbes that could be potentially hazardous to human health. This human health hazard could have a global range, and administrative activities have been ineffective in risk reduction. © 2011 Elsevier B.V. All rights reserved.

Contents 1. 2. 3. 4. 5. 6. 7. 8.

9. 10. 11. 12.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Physicochemical properties of SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanism of antibacterial activity of SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Use of SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Estimated usage of SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Occurrence of SNs in the environment and food . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ecotoxicity of SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Degradation of SNs in organisms and in the environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1. Metabolism of SNs in mammals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2. Biodegradability of SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3. Physicochemical methods of degradation of SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Removal of SNs from wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The environmental risk assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Generation of drug resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction World production and consumption of pharmaceuticals has been steadily increasing at a rate higher than the rate of global

∗ Corresponding author. Tel.: +48 032 364 15 62. E-mail address: [email protected] (E. Adamek). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.082

1 2 2 3 4 5 6 7 7 7 8 8 8 9 12 12 12

population growth. After use, large amounts of drugs have been discharged into the environment in the form of human and animal excretions and unused waste [1]. The persistence of pharmaceuticals in the environment, the rate of their spreading and their ability to accumulate in the biosphere has differed. However, their high biological activity indicates that these drugs, even in trace amounts, could cause significant changes in the biosphere. An example of such changes in the last decade of the 20th century is

2

W. Baran et al. / Journal of Hazardous Materials 196 (2011) 1–15

O 4

H2N

S

N1 R

O

H

Fig. 2. Chemical structure of SNs with bacteriostatic properties.

• excessive amounts of SNs are introduced to the biosphere (a common practice is illegal and/or without control administration of SNs to healthy farm animals), • a local concentration of SNs in the environment and risk associated with this issue are very high, • SNs can remain in the environment for a long time, • SNs present in the environment are active in the generation of drug resistance in bacterial cells (including cross resistance to drugs). 2. Physicochemical properties of SNs Fig. 1. The possible fates of SNs residues and resistance genes (SNsR) in the environment.

the phenomenon of feminization of fish by sex hormones caused by anthropogenic pollution of European rivers [2]. For these reasons, pharmaceuticals have been classified as particularly dangerous pollutants for the environment. As a result, research and multinational projects (e.g., REMPHARMAWATER [3], POSEJDON [4], KNAPPE [5], ERAPHARM [6], and ECO-SENS [7]) have been carried out to find answer to the following questions: • Which pharmaceuticals have the greatest environmental risk? • How can we effectively control the amounts and effects of drugs on the environment? • How can we successfully reduce their release into the environment? Antibiotics are a group of pharmaceuticals with effects on the environment that could be particularly harmful to human health. Unfortunately, their frequency in environmental samples is very high [1,5,8–18]. Historically, sulfonamides (SNs) have been used as synthetic antibiotics the longest. Recently the large quantities of SNs are used in animal husbandry in particular as veterinary medicines. Based on these drugs, we can obtain a reliable estimation of the effects and consequences of prolonged use of anti-infectives on people’s health and on the environment. A report for State Office for Nature, Environment and Consumer Protection of North Rhine-Westphalia (Germany) published in 2007 has been presented the literature review on effects of the introduction of SNs to the environment [1]. In the majority of published articles, the authors assessed the risk caused by SNs almost exclusively on the basis of their use, toxicity, and removal efficiency from the environment. The data presented in this context led to the conclusion that the presence of drugs in the environment was a negligible problem regarding quality of life. However, in the majority of articles, the effect of antiinfectives in the generation of drug resistance in microbes was not considered. The effect of antibiotics occurring in the environment to the generation and prevalence of drug-resistant microorganisms is essential from the human health point of view (Fig. 1.). This influence has been much more widespread in the last decades due to the globalization process.The aim of the work is to show that:

Since the early 1940s, over 150 SNs, sulfanilamide derivatives, have been applied in human and veterinary medicine as antibacterial drugs [19]. The formula of structure presented in Fig. 2 corresponds to the synthetic antimicrobial agents that contain the sulfonamide group. Such a molecule should have a free amino group (–N4 H2 ) at one end. SNs are a group of synthetic bacteriostatic drugs classified by the Anatomical Therapeutic Chemical (ATC) classification index as a group of antibacterial drugs for systemic use (the subgroup J01E) [20]. Many SN derivatives have also been used as antiprotozoal agents [21] and herbicides [19], and complexes of SNs with Ag+ and Zn2+ have been used as antifungals [22]. Moreover, SNs have been the most commonly used components of more composite drugs with trimethoprim (TMP). The characteristics of commonly used SNs are presented in Table 1. SNs are polar molecules with amphoteric properties. Their amino nitrogen (N4 ) is protonated at pH 2–3, while the amide nitrogen (N1 ) is deprotonated at pH 4.5–11 [10,23]. The SNs presented in this text are small molecules (molar mass 177–300 g mol−1 ), are water soluble (with the exception of SGM and sulfasalazine) and have low Henry’s constant (1.3 × 10−12 –1.8 × 10−8 ) values [9,10,24]. They are slightly sorbed by soil (the soil partition coefficient values are 0.6–7.4 l kg−1 ) [9]. Thus, these SNs are easily and quickly spread in the environment, but their properties should limit their accumulation in defined biotopes. SNs do not easily adsorb onto activated carbon [1,4]. They are classified as photo- and thermally stable substances at the degradation half-life (DT50 ) >1 year [24]. They can undergo alkaline hydrolysis and coupling reactions with phenols and amines and easily react with the hydroxyl radical HO• [10,11,25]. 3. Mechanism of antibacterial activity of SNs As shown in Fig. 3, antibacterial SNs act as competitive inhibitors of the enzyme dihydropteroate synthease (DHPS) which catalyses the conversion of para-aminobenzoate (PABA) to dihydropteroate (AHHMD), a precursor of folate synthesis. Tetrahydrofolic acid (THF) participates in the synthesis of nucleic acids that are essentials as building blocks of DNA and RNA.A mechanism of action of herbicidal SNs is similar.As a result, it is possible to inhibit the synthesis of nucleic acids and thus proteins [18,27]. SNs also inhibit the permeability of the bacterial cell wall for glutamic acid, which is also an essential component in folic acid synthesis. However, SNs do not inhibit the growth of microorganisms that:

W. Baran et al. / Journal of Hazardous Materials 196 (2011) 1–15

3

Table 1 Common names, CAS number and structure of selected SNs. Common name of SNs

CAS number

ATC classification index

Abbreviation

–R

Sulfanilamide Sulfacetamide Sulfacarbamide Asulam (herbicide) Carbutamide Sulfathiourea Sulfaguanidine

63-74-1 144-80-9 547-44-4 3337-71-1 339-43-5 515-49-1 57-67-0

J01EB06, D06BA05, QJ01EQ06 S01AB04 J01EC20 (with SDZ and SDM) – A10BB06 J01EB08 A07AB03

SAD SCT SC

–H –COCH3 –CONH2 –COOCH3 –CONH(CH2 )3 CH3 –CSNH2 C(NH2 )2

Sulfathiazole

72-14-0

D06BA02, J01EB07, QJ01EQ07

STZ

STU SGM

S N

H3C Sulfafurazole, Sulfisoxazole

127-69-5

J01EB05, S01AB02, QJ01EQ05

CH3

SSZ

N

O N Sulfamethoxazole

723-46-6

J01EC01, QJ01EQ11

O

SMX

CH3

Sulfamoxole

Sulfapyridine

729-99-7

144-83-2

J01EC03

SMM

J01EB04, QJ01EQ04

SPY

O

CH3

N

CH3

N N

Sulfadiazine

68-35-9

J01EC02, QJ01EQ10

SDZ

N

N Sulfamethoxine, Sulfamethoxydiazine

651-06-9

J01ED04

OCH 3

SMO

N

N Sulfamerazine

127-79-7

J01ED07

SMR

N CH3 CH 3

N Sulfamethazine, Sulfadimidine

57-68-1

J01EB03, QJ01EQ03, QP51AG01

SDM

N CH 3 OCH3

Sulfadimethoxine

122-11-2

J01ED02, QJ01EQ09, QP51AG02

SDT

N N OCH3 N N

Sulfamethoxypyridazine

80-35-3

Sulfachloropyridazine

80-32-0

J01ED05, QJ01EQ15

OCH 3

SMP

N N SCP

Cl

N N Sulfadoxine

2447-57-6

QJ01EQ13

SDX

H3CO

• need the presence of folic acid in the environment, • possess a high concentration of PABA, or • have modified metabolic pathways (drug resistance). 4. Use of SNs SNs are active against a broad spectrum of Gram-positive and many Gram-negative bacteria including species of the genus

OCH3

Streptococcus, Staphylococcus, Escherichia, Neisseria, Shigella, Salmonella, Nocardia, Chlamydia and Clostridium. Moreover, SNs have used against protozoa (e.g., Toxoplasma gondii), parasites (e.g., Plasmodium malariae), and fungi (e.g., Pneumocystis carinii). SMX, SCT or sulfasalazine belong to SNs commonly used in human medicine while SDM, SDT, SMR, SDZ, STZ are used most frequently in veterinary medicine (different SNs have been used in different countries). Moreover, SNs have been added to animal feed

4

W. Baran et al. / Journal of Hazardous Materials 196 (2011) 1–15

N

H2N N 2-Amino-4-hydroxy-6-hydroxymethyl7,8 dihydropteridine diphosphate

H N

H H

N

CH2

H2N

O

H N

N N

P ~P

N

ATP AMP

OH

H H CH2

OH

OH (2-Amino-4-hydroxy6-hydroxymethyl-7,8 dihydropteridine

(AHHMD)

(AHHMP) dihydropteroate synthetase (DHPS) EC 2.5.1.15 - P-P and H2O

SO2NHR

H 2N

Sulfonamide

COOH

H2N

4-Aminobenzoic acid (PABA)

X

H N

N

H2N N

H H

Dihydropteroic acid COOH

CH2 HN

N OH

CH2

CH2

COOH

H2N CH

Glutamic acid

ATP

COOH H N

N

H2N N

N

H H

O

CH2

CH2 HN

C HN

CH

CH2

COOH [O] [H]

OH

COOH

Dihydrofolic acid (DHF) dihydrofolate reductase (DHFR) EC 1.5.1.3

Folic acid N-[p-[[(2-Amino-4-hydroxy-6-pteridinyl) methyl]amino]benzoyl]-L-glutamic acid

[O]

[H]

N

H2N N

OH

H N N H

H H H CH2 HN

C

CH2

CH2

O

HN CH

COOH

Tetrahydrofolic acid (THF)

COOH

Fig. 3. The schema of SNs pathways, based on Wilson & Gisvold’s Textbook of Organic Medicinal and Pharmaceutical Chemistry [26].

premix used in young animals feeding. For example in Denmark in 2009 the consumption of SNs with TMP per kg of meat produced was as follows [28]:

However, the use of Asulam could lead to the contamination of honey with SNs residues [29]. In 2008, it has been withdrawn from use in the EU countries.

• • • •

5. Estimated usage of SNs

pigs 4.82 mg, cattle 17.2 mg, broilers 0.033 mg, farmed fish (aquaculture) 58.5 mg.

Moreover, SNs can be used in commercial beekeeping (they protect honey bees against bacterial diseases e.g., American foulbrood). In agriculture, sulfonamide Asulam has been widely used as a herbicide. It is effective against dicotyledonous weeds e.g., barnyard grass (Echinochloa crus-galli), velvet grass (Holcus lanatus), wild oat (Avena fatua) and broadleaf dock (Rumex obtusifolius).

An accurate assessment of the global consumption of all drugs would be difficult, if not impossible. The authors of the KNAPPE project have estimated that the global consumption of pharmaceuticals used in human and veterinary medicine has reached 100,000 tonnes per year [5]. Based on information from the Union of Concerned Scientists, Sarmach et al. indicated that, at the beginning of the 21st century, Americans consumed 16,000 tonnes of antibiotics per year [9]. SNs used in veterinary medicine accounted for approximately 2.3% of the total amount of antibiotics. In European

W. Baran et al. / Journal of Hazardous Materials 196 (2011) 1–15

18 16

active compounds (t)

14 12 10 8 6 4 2 0 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

year Fig. 4. The dynamics of consumption of SNs and TMP in Denmark in the years 1990–2009 [28,31].

countries, this value ranged from 11 to 23% [9]. According to other authors, the worldwide consumption of antibiotics (anti-infective drugs) ranged from 100,000 to 200,000 tonnes per year, including 50–75% that were used in veterinary medicine and animal husbandry [1,24]. It has been possible that each year more than 20,000 tonnes of SNs, with bacteriostatic properties, have been introduced into the biosphere (not counting drugs introduced as herbicides). Since the end of 20th century, Scandinavia and other countries in Europe and North America have imposed restrictions on the use of antibiotics (including SNs) in animal husbandry. The use of antibiotics as growth promoters in animal husbandry in the European Union has been banned since January 1, 2006 [30]. However, reports on the consumption of pharmaceuticals in different countries have not shown a reduction in the use of these drugs. Fig. 4 presents the dynamics of SN with TMP consumption in Denmark in the years 1990–2009 [28,31]. A decrease in the use of SNs in animal husbandry occurred in the mid-1990s and is associated with the introduction of administrative restrictions related to the application of these drugs in animal feed. Although the ban has been still in place, the use of SNs in agriculture is similar as in 1994. In our opinion, the plots in Fig. 4 illustrate global trends in consumption of SNs in livestock farming and medicine. 6. Occurrence of SNs in the environment and food The first publication containing quantitative data about the presence of SNs in river water was published in 1982 [9]. However, systematic studies on the quantitative determinations of SNs in environmental matrices became possible after the development of highly sensitive analytical methods. According to data from the U.S. Environmental Protection Agency the limit of detection during routine analytical procedures using SPE/HPLC-MS/MS techniques for the selected SNs was below 10−9 g l−1 (e.g., for SDT, the limit of detection was 1 × 10−10 g l−1 ). A detailed statement of the analytical techniques and limits of detection of drugs (including SNs) in environmental samples has been discussed by García-Galán et al. [18] and Seifrtová et al. [32]. At the described level of detection, SNs were detected in 27% of rivers and streams in the USA [11], in almost all surface waters in France and Tajwan [33,34], and in 100% of wastewater samples [13,35,36]. According to Vulliet and CrenOlivé, in the Rhônee Alpes region of Frances the frequencies of SMX in surface and groundwater were 37 and 66%, respectively [37]. In commercially available, Italian natural mineral water the frequency

5

of SNs was 50% (in 4 of the 8 investigated samples) [38]. GarcíaGalán et al. [36] described in detail the frequency of occurrence of 19 selected SNs in wastewater. Moreover, metabolites of SNs, mainly N4 -acetyl sulfonamides (N4 -AcSNs), were also identified in environmental samples [11,39]. SNs concentrations in the environment underwent significant fluctuations, which were mainly dependent on the type of matrix and the type of SN [36]. Additionally, the results obtained may have depended on the sampling site, the day of the week [40] and even the time of day [41]. However, it was important to note that the data concerning the determination of SNs in environmental samples could contain significant errors. The cause of this may be imperfection of the analytical procedure used and the incorrect (incomplete) extraction of samples. For example, the recovery of SNs from soil samples ranged from 5 to nearly 294% but the authors have found that the “presented method is characterized by good selectivity” [42]. The recovery efficiency depends on various parameters including extraction/purification strategies [43] and the type of matrix [44]. A summary of the occurrence of SNs, depending on the matrix, is shown in Fig. 5, and the maximal values are given in Table 2. The presented data are based on the maximum values described in the literature. SNs concentrations in samples increased as follows: seawater < ground water < surface water < treated sewage < untreated (raw) municipal sewage < hospital sewage < activated sludge < soil < runoff from farmland < leachates from landfill < manure. Due to the low concentrations and low abundance of SNs, the presence of trace amounts of these drugs in drinking water was not considered a significant problem. The maximum concentrations were found in freshly removed bedding [58] and manure from pigs fed diets that contained SNs, mainly SDM [59]. This SN occurred in almost 50% of samples (the average concentration of the drug was 7 mg kg−1 ). Additionally, other SNs were identified in tested samples (e.g., for SDZ, the maximum concentration was 35.2 mg kg−1 ). Fortunately, even short-term storage of manure could result in a significant reduction in the concentration of SNs [58]. The highest allowed concentrations of SNs in food were established in administrative regulations. The European Union adopted a maximum SN concentration of 100 ␮g kg−1 in animal foodstuffs [61]. In Poland, the maximum permitted concentration of Asulam in fruits and vegetables is 0.5 mg kg−1 [62]. The occurrence of SNs in tissues of farmed fish has been incidental, e.g., in Slovenia, SNs residues were found in 14 of the 2363 samples [63]). SNs residues in edible marine food were detected rarely, however the concentration of SNs in tissue of common eel (Anguilla anguilla) was above 5 mg kg−1 [64]. In EU countries, the occurrence of SNs residues in edible tissues of farm animals has been insignificant. According to “Report for 2006 on the results of residue monitoring in food of animal origin in the Member States” SNs, at concentrations above their maximum allowed limits, were detected in 0.006, 0.05, 0.08, 0.97 and 3.86% of samples of poultry, bovines, pigs, eggs and rabbits, respectively [65]. Although the use of SNs in beekeeping is banned in the EU, the frequency of these drugs in honey samples is high. In Poland, it has been estimated that almost 10% of honey samples contain excessive amounts of SNs i.e., above the allowed maximum concentration. The results reported by the Chinese researchers are much less optimistic. High concentrations of SNs were determined in pig offal (almost 74 mg kg−1 of SDT and 73 mg kg−1 of STZ) and poultry offal (46 mg kg−1 of SDZ) [66]. Even more worrying is the fact that 75% of meat samples contained SNs at the total concentration >100 ␮g kg−1 [67]. SNs could be absorbed and accumulated by plants fertilized with manure (the highest concentrations of SNs are determined in roots and leaves [9,68–71]. For example, the maximum concentration

6

W. Baran et al. / Journal of Hazardous Materials 196 (2011) 1–15

1000 000 100 000

-1

SNs concentration (ug l )

10 000 1 000 100 10 1 0,1 0,01

Drinking Bottled Groundwater waters mineral water

SeawaterLeachate

Wastewater /influent

Hospital wastewater

Effluent

SDZ

Ac-SDM

SNs

Biosolid Soil /sludge

SDM

SPY

STZ

SMX

SDZ

SPY

SDZ

SMX

SMX

STZ

SDZ

SDT

SDM

SMX

SCP

SDM

SDM

SDZ

Surface

SMX

STZ

SDM

SMX

SDT

SDM

SMX

SDT

STZ

SMX

SMX

0 ,0 001

Ac-SDM

0,001

Manure

Fig. 5. Occurrence of the selected SNs depending on the matrix.

of SDM determined in corn, tomatoes and lettuce was 0.1 mg kg−1 [69]. Migliore et al. [71] reported that cosmopolitan weeds (Amaranthus retroflexus and Plantago major) showed a high tendency to bioaccumulate. In the tissues of these plants cultured in the medium containing SDT the accumulation rates were 2314 and 6065 mg kg−1 , respectively [71]. In our opinion, although the data presented in this section are based on the maximum values, it is possible that they can be underestimated. A commonly routine practice is the excessive and prophylactic use of antibiotics in animal husbandry and the use of manure as a fertilizer. Therefore, the local real concentrations of

SNs in the biosphere are much higher. Additionally, this effect is difficult to control due to the high mobility of SNs in the environment. For these reasons, the excessive use of manure containing SNs as fertilizer should be banned. 7. Ecotoxicity of SNs The toxicity of SNs to higher organisms (vertebrates) is not high. According to the EU directive 93/67/EEC, SNs under investigation can be classified as non-toxic or harmful [72]. The results described in the literature indicate that SNs do not exhibit mutagenic or

Table 2 Concentrations of SNs in the environment. nb

Maximal values

2.1 (0–8.5 [8]) ␮g l (SMX); 0.011 ␮g l (SMX) [37] 0.164 ng l−1 (SDT) [38] 0.047 (0.013–0.080) ␮g l−1 (SMX) [38] 0.80 (0.0099–1.11 [14]) ␮g l−1 (SMX) 0.053 (0.0002–0.09148 [45]) ␮g l−1 (SDT) 0.87 (0.015–18 [8]) ␮g l−1 (SMX) 2.26 (0.0108–19.2 [46]) ␮g l−1 (SDM) 0.0475 ␮g l−1 (SMX) [47] 379.78 (0.66–703.2 [48]) ␮g l−1 (SCP) 46.58 (0.05–1340 [33]) ␮g l−1 (SMX)

4 2 11 3 39 12 1 7 31

61.11 (0.0269–500 [49]) ␮g l−1 (SDM)

17

17.78 (0.3–79.9 [51]) ␮g l−1 (SMX) 1.28 (0.353–2.2 [51]) ␮g l−1 (SDZ) 0.517 (0.00366–6.0 [53]) ␮g l−1 (SMX) 1.26 (0.005–4.27 [50]) ␮g l−1 (STZ)

6 2 30 4

Soil

22.56 (0.01–113 [54]) ␮g kg−1 (SMX) 99.1 (1.2–197 [54]) ␮g kg−1 (SPY) 211.6 (0.16–860 [56]) ␮g kg−1 (SNs)

6 2 10

Manure

27.30 (0.23–167 [1]) mg kg−1 (SDM)

7

59.07 (35.2–91 [1]) mg kg−1 (SDZ)

3

8.5 ␮g l−1 (PECc for SMX) [8] 0.080 ␮g l−1 (SMX) [38] 3.461 ␮g l−1 (SCT) [45] 25 ␮g l−1 (all SNs) [43] 0.0475 ␮g l−1 (SMX) [47] 703.2 ␮g l−1 (SCP) [48] 1340 ␮g l−1 (SMX; from pharmaceutical production) [33] 1158.68 ␮g l−1 (STZ; agricultural wastewater) [50] 12.8 ␮g l−1 (SMX) [52] PEC 92.8 ␮g l−1 (all SNs) [51] 6.0 ␮g l−1 (SMX) [53] 4.27 ␮g l−1 (STZ; effluent of agricultural WWTP) [50] 197 ␮g kg−1 dwd (SPY) [54] 31 ␮g kg−1 (SDM) [55] 400 ␮g kg−1 (STZ; agricultural soil) [57] PEC 860 ␮g kg−1 (SCP; soil pore water estimation) [56] 395.730 mg kg−1 (SDT; in bedding – day 0) [58] 167 mg kg−1 (SDM) [59] 1600 ␮g l−1 (all SNs) [60]

Matrix Drinking waters Bottled mineral water Ground water Surface water Sea water Drainflow/leachate Influent/wastewater

Hospitals wastewater Effluent (after WWTP)

Sludge (after WWTP)

Meana /the most described SNs −1

Waste landfill a b c d

Calculated based on maximal values given in tables. Number of papers. Predicted environmental concentration. Dry weight.

−1

W. Baran et al. / Journal of Hazardous Materials 196 (2011) 1–15

carcinogenic (teratogenic) activity [73]. On the other hand, in the report “Environmentally Classified Pharmaceuticals 2009”, SNs were considered as highly toxic drugs [74]. The discrepancies between these reports probably result from different criteria used to define a risk. Directive 93/67/EEC is based on the environmental risk posses by pharmaceutical substances while “Environmentally Classified Pharmaceuticals” report assesses both environmental risk (based on the acute toxic risk to the aquatic environment) and additionally persistence and bioaccumulation of SNs in the environment (based on the information published by the Swedish Association of the Pharmaceutical Industry [74]). Fig. 6 illustrates the toxicity of SMX to selected test organisms. Important data on the SNs ecotoxicity were summarized in articles by García-Galán et al. [18] and Isidori et al. [73]. SNs are practically non-toxic to most microorganisms tested [4,18,73,75], including selected strains of bacteria, such as Vibrio fischeri, Enterococcus faecalis, Escherichia coli, Pseudomonas aeruginosa, and Staphylococcus aureus. For example, the L(E)C50 values determined using the Microtox® test (V. fischeri) ranged from 16.9 to 118.7 mg l−1 (for SMX) to >1000 mg l−1 (for STZ) [73,76,77]. Strong bacteriostatic properties caused by the SNs could significantly change the functioning of microorganisms living in the environment, for example a significant reduction of their microbial activity [78]. Additionally the number of less sensitive (resistant) strains has increased and the number of strains sensitive to SNs has decreased. Thiele-Bruhn and Beck showed that the disposing of urine that contained even a low concentration of SPY (0.02 mg kg−1 ) into the soil resulted in a significant reduction of microbial activity [78]. It was found that, in the case of SPY, the EC10 values for soil organisms ranged from 0.00014 to 0.16 mg kg−1 (the microbial Fe(III) reduction test) and from 0.0071 to 0.056 mg kg−1 (the substrate-induced respiration test) [79]. However, the most sensitive assays for the presence of SNs are bioindicators containing chlorophyll [9,18,73]. A highly toxic effect of SMX on Synechococcus leopoliensis (EC50 = 0.0268 mg l−1 ) was described by Ferrari et al. [77]. In the case of SMX, the no observed effect concentrations (NOECs) for algae (Pseudokirchneriella subcapitata and S. leopoliensis) and gibbous duckweed (Lemna gibba) were 0.090 [77], 0.0059 [77] and 0.01 mg l−1 [4], respectively. This indicates that even low concentrations of SNs may significantly affect the growth and development of plants. SNs can accumulate in various organisms in the food chain, and this accumulation could lead to a local increase in toxic effects induced by these drugs [9,10,70,71]. In addition, the toxic effects of SNs and other pollutants could exhibit a synergism [11,80,81]. At environmental exposure levels (samples contained 13 micropolutans, including SMX at the concentration of 46 ng l−1 ) the drug mix inhibited the growth of human embryonic cells HEK293, with the highest effect observed as a 30% decrease in cell proliferation compared to controls [81]. Since there has not been sufficient extensive experiments in patients with a single overdose of SNs the maximum tolerated dose in humans are unknown [82]. In laboratory experiments, acute oral overdoses of SNs in animals (LD50 ) were as follows: • in rats 10,000 mg kg−1 (SSZ), • in rabbits 2000 mg kg−1 (SSZ), • in mice 5700 mg kg−1 (SSZ), 16,500 mg kg−1 3700–4200 mg kg−1 (SAD), 4500 mg kg−1 (STZ) [83].

(SCT),

Exemplary adverse effects associated with overdosage of SNs in humans include nausea and cutaneous hypersensitivity reactions. Other adverse effects e.g., stomatitis, hemolysis, methemoglobinemia, hepatotoxicity and renal toxicity occur rarely. SNs can cause interaction with other drugs, for example with methotrexate,

7

sulfonylureas, wafarin, mercaptopurine, cyclosporine or didanosine [84]. In our opinion, direct toxic effects caused by SNs occurring in the environment do not appear to be a significant threat to public health. Potential possible cases of direct toxic effects of SNs on human may be sporadic. On the other hand, the occurrence of SNs residues in food, particularly in the case of illegal or improper use of these drugs, can be a more serious problem. According to Dolliver et al., SNs residues in food products do not pose a threat and/or adverse effect to human health but “development and spread of antibiotic resistance, which is a major problem globally” [69]. 8. Degradation of SNs in organisms and in the environment Possible products of the biotransformation and degradation of SNs are shown in Fig. 6. A detailed discussion of these processes is presented in the next sections. 8.1. Metabolism of SNs in mammals A large part of the SNs dose is excreted from organisms as unchanged compounds. For example, 75% of SMR could be excreted from the body in its parent form [1]. However, in general, over 80% of an SN dose undergoes biotransformation in mammals. The degree of transformation of each SN depends both on its type and the features of the organism. Biotransformation of SNs is mainly based on oxidation, acetylation or hydroxylation at the N4 nitrogen atom or glucuronidation of the N1 - or N4 -nitrogen atoms [1,10,11]. It is assumed that, after oral administration, 50–70% of the dose is excreted in urine as N4 -AcSNs, and 15–20% as N1 glucuronides [1,10]. The metabolites of SNs do not possess high biological activity as unchanged SNs. However, this activity could be easily restored during in vitro conditions [11,85]. The concentrations of metabolites other than those listed above are small and are likely not significant in the environment. Reviews of possible paths of SNs biotransformation were described in the papers of Sukul and Spiteller [10] and García-Galán et al. [11]. 8.2. Biodegradability of SNs The opinions of researchers on the biodegradability of SNs have been divided [1,4,10,17,24,86]. The cause of this may be the differences in microbial activity of the matrix, the inoculum used, and the applied methods used to assess SN degradation (Table 3). The stability of various SNs is also different; for example, SDM is more (10x) resistant to biodegradation than STZ. The results of standardized tests, such as the ISO 11734:1995 and OECD 301D, and the assessment of soil microbial activity suggest that most of the SNs do not undergo natural biodegradation. One of the most often described SNs in the literature is SMX, which has been regarded as a non-biodegradable compound (in pure water, seawater, natural water and wastewater or active sludge) in 9 of 24 articles [1,4,24,86,89,91,97–99]. According to Weifen et al. [114], in the presence of shrimp (Penaeus chinensis), the DT50 value for SMX is 5.68 h. Ingerslev and Halling-Sørensen [92] found that, in the presence of microorganisms in activated sludge, the DT50 of SNs is only ∼7 h. De Liguoro et al. [58] stated that, in the case of SDT, the DT50 for microbial degradation in fresh bedding is ∼1 day. Similarly, equally rapid degradation of SDT has been described by Wang et al. [87]. These authors have observed an increase in the DT50 value with increasing initial concentrations of SDT in fresh and sterile manure. In these cases, most of the SNs were incorporated into microorganisms and/or underwent only reversible transformations, such as acetylation [11,85]. The rapid disappearance of SNs in soil and manure could be an effect of binding between SNs and organic or mineral particles [85,88,93] or could be caused by

8

W. Baran et al. / Journal of Hazardous Materials 196 (2011) 1–15

100000

L(E)C50 (mg l-1) or (mg kg -1)

10000 1000 100 10 1 0,1 0,01

Bacterium

Diatom

Algae

Aqua Gnitic daria plant

Crustacean

Rotifer

In verte brate

Fish

Children (PNEC)

D. rerio (96h)

O. latipes (96h)

O. mykiss (EA)

T. platyurus (24h) B. calciflorus (24h) B. calciflorus (48h) M. macrocopa (48h)

C. dubia (7d)

C. dubia (48h)

H. attenuata (96h) D. magna (24h) D. magna (48h) D. magna (48h)

L. gibba (7d)

Actived sludge (20d) AMES test (72h) C. meneghiniana S. capricornutum P. subcapitata (96h) C. vulgaris (48h) S. subspicatus (72 h) S. leopolensis (96h)

C. freudii (24h)

P. agglomerans P. aeruginosa (48h) V.fischeri (5min) V. fischeri (15min) V. fischeri (30min)

0,001

Mam mals

Fig. 6. Comparison of SMX toxicity to selected test organisms.

photochemical processes (on the soil surface, in the presence of Fe compounds and nitrates) [107,115]. Most researchers recognized SNs as poor or non-biodegradable compounds in the environment (in pure water, surface water and in soil with a DT50 > 30 days) [24,74]. The fact that SNs occurred so often in test samples could also be considered as evidence of their persistence in the environment. In our unpublished study on the biodegradation of SAD, STZ, SMX and SDZ applied to natural matrices, we found that, in individual cases (STZ under aerobic conditions, in wastewater and water from the swamp), the DT50 was less than 2.5 days. In the remaining cases, the DT50 values for SDZ, SAD and SMX were >5, >8 and 31 days, respectively. In our opinion the read stability data related to SNs residues in the environment (especially for SMX) are generally much higher than the data reported by other researchers in the cited articles. The above-described high frequency of SNs in environmental samples can confirm this assumption. 8.3. Physicochemical methods of degradation of SNs The efficiency of SNs degradation using the most commonly used chemical and physicochemical methods is presented in Table 3. The high degradation efficiency of SNs in wastewater was obtained using various advanced oxidation processes (AOP) [1,4,101,105,108,116,117], such as the use of O3 , Cl2 , and ClO2 [1,101,116–118], the Fenton reaction [105,116] or photocatalytic processes [85,105,116,117]. Unfortunately, the application of these methods is costly and could be harmful to the environment due to the formation of highly toxic intermediates [118]. Moreover, a decrease in the efficiency of AOP with an increase in overall wastewater pollution was observed [108]. This fact made it difficult to apply these methods directly to remove SNs from manure. In Table 3, examples of other methods used to remove contaminants from the aquatic environment without their degradation or transformation (non-destructive methods) are presented. SNs can be removed from wastewater with nearly 100% efficiency by reverse osmosis [1,4,24,109,110]. However, with this method, there could be a problem with wastewater containing concentrated solutions

of toxins (including SNs) [110]. In the case of substances resistant to biodegradation, there could be a local, risky increase in the concentration of these toxins in a small area [119]. The physico-chemicals methods (particularly AOP) can be effective and very useful in the degradation of SNs. In our opinion, it is not excluded that their environmental degradation is the result of photochemical reactions initiated by sunlight in the presence of natural photosensitizers, and not only of biodegradation processes. 9. Removal of SNs from wastewater Opinions on the efficiency of SN removal in conventional biological-mechanical treatment plants are divergent. Similar differences occur during the assessment of the biodegradability of drugs (Table 3). Onesios et al. [120] analyzed 49 cases of the removal of SDZ, SDM, SMX and SPY from wastewater in wastewater treatment plants (WWTPs). Based on the analysis of recent publications, −280 to 100% of SNs were removed using activated sludge (AS) (Table 4). The mean degree of SN removal in these cases was ∼24%. According to the data published in 2010, SMX was removed from the selected WWTPs in Spain in the range of 30–92% [35]. However, there are also cases in which the concentration of SNs in effluent was higher than that in the influent [4,86,134,138]. This effect was described in a pilot WWTP in Austria [90] and in Switzerland [138]. This effect is likely caused by hydrolysis of the N4 -AcSNs present in wastewater to the parent SNs [11]. A conclusion of this problem may be found in the data from the study by Turkdogan and Yetilmezsoy [109]. These authors have estimated that 80% of used antibiotics enter the environment despite the use of various processes in WWTPs (based on the data from Turkey, without regard to SNs). Importantly, a large part of SNs may be adsorbed in WWTPs by biomass [132] and could return again to the environment. 10. The environmental risk assessment The majority of researchers have used the method recommended by the European Medicines Evaluation Agency (EMEA) for environmental risk assessment. This method uses the results of

W. Baran et al. / Journal of Hazardous Materials 196 (2011) 1–15

9

Table 3 Biotransformation, degradation and other methods of SNs removal. Matrix

Methods

Biotransformation Human Animal Manure

Efficiency

DT50

99.0%) was from Aldrich and used as received. Hexane, sodium chloride, and hydrochloric acid were all of analytical purity. Deionized water (>18.0 m) was used for solution preparation and dilution. Commercial HA powder was obtained from Lemandou Chemicals Co., Ltd., China, which was derived from lignite (Sinkiang, China). The HA was further purified according to the procedures recommended by the International Humic Substances Society (IHSS) before use [14]. The element composition of the purified HA was analyzed on a Vario Micro element analyzer (Elementar, Germany). The HA comprises 48.3% C, 2.6% H, 29.3% O, 1.1% N, and 0.3% S. Kaolin (chemical purity) obtained from Shanghai Qingpu Chemical CO. Ltd. was selected as the model clay. The organic content of kaolin was 0.12%, and the cation exchange capacity (CEC) measured by BaCl2 –H2 SO4 method (ISO 11260-1997) was 16.9 cmol kg−1 . The BET specific surface area measured by a surface area apparatus (Micromeritics Tristar 3000) was 19.4 m2 g−1 . X-ray fluorescence analysis (Genesis, EDAX Inc.) shows that its main mineral elements were Si (59.9% in wt) and Al (36.7%). Other physicochemical properties associated with the kaolin are listed as Table S1 in Supplementary Materials.

2.2. Influence of HA–TX100 interaction on HA and TX100 sorption to kaolin All the sorption experiments were carried out in triplicate using batch equilibrium technique in glass vials sealed with Teflon screw caps. Data processing and fitting (including isotherm model fitting) were performed using Origin v. 8.0. To begin with, the sorption of HA or TX100 in a kaolin/water system was studied respectively. A total of 0.5 g kaolin was mixed with 10 mL HA (0–2500 mg L−1 ) or TX100 solutions (0–10 mmol L−1 ). The vials were agitated in a reciprocating shaker end over end for 48 h (25 ± 1 ◦ C), and centrifuged at 4000 rpm for 10 min. The

supernatant was filtered through a 0.45 ␮m cellulose acetate membrane and subjected for HA or TX100 analysis. The influence of HA–TX100 interaction on TX100 and HA sorption to kaolin was further investigated in two different manners (0.5 g kaolin in 10 mL solutions): with constant HA total concentration of 400 mg L−1 and varying TX100 total concentrations of 0–10 mmol L−1 , or with constant TX100 total concentration of 2 mmol L−1 and varying HA total concentrations of 0–2500 mg L−1 . The pH of the slurries was adjusted to 7.0 ± 0.2 with appropriate concentrations of HCl or NaOH (mostly 1 mol L−1 while 10 mol L−1 HCl or NaOH was used for samples with high HA concentrations). Furthermore, 0.01 mol L−1 NaCl was contained as the background electrolyte. Equilibrium aqueous concentrations of HA and TX100 were both analyzed for each sample. 2.3. Influence of HA or TX100 on HCB sorption to kaolin In this section firstly the influence of HA addition on HCB sorption to kaolin was investigated. A total of 0.5 g kaolin was mixed with 10 mL HA solutions (0–2500 mg L−1 , final pH 7). Then 40 ␮L of HCB acetone solution was added into the vials by a microsyringe before shaking (acetone fraction was below 0.5%). The total concentration of HCB was 0.5 mg L−1 (or 10 mg kg−1 kaolin). After equilibrium and centrifugation, the supernatant was filtered through a 0.45 ␮m cellulose acetate membrane. HCB in the supernatant was then extracted by hexane via liquid-liquid extraction in 11-mL glass vials (with an extraction ratio of 1:2, agitated in a shaker for 2 h) and analyzed by gas chromatography (GC). Meanwhile HA in the filtrate was measured. Secondly, the influence of TX100 on HCB sorption was investigated which followed the same procedure as described above. TX100 concentration used was 0–10 mmol L−1 . Both TX100 and HCB equilibrium concentrations were determined for each sample. 2.4. HCB partitioning in a solid HA/TX100/water system Predetermined volumes of HA and TX100 stock solutions (HA 5 g L−1 and TX100 20 mmol L−1 ) were pipetted into 50 mL glass flasks, and diluted with deionized water to 20 mL. The amounts of HA added were as 0–2.5 g L−1 , and the total TX100 concentration was 2 mmol L−1 . The pH of the mixture was adjusted to 2–3 with HCl to precipitate HA thoroughly. Then 80 ␮L HCB acetone solution was added before shaking. The total concentration of HCB was 0.5 mg L−1 . After equilibrium and centrifugation, the supernatant was filtered through a 0.45 ␮m cellulose acetate membrane and subjected for TX100 and HCB analysis. 2.5. HCB partitioning in a HACK/TX100/water system The sorption of HCB to HACK (detailed preparation procedures are provided in Supplementary Materials) in the presence of TX100 was further investigated. A total of 0.5 g HACK with varied HA coating amounts (0.25–5%) was mixed with 10 mL TX100 solutions (0–10 mmol L−1 ). The pH of the slurry was then adjusted to 7.0 ± 0.2 or 3.0 ± 0.2 (pH 3.0 could avoid HA dissolution from HACK, thereby better reveal the role of solid HA in the partitioning of TX100 and HCB). Then 40 ␮L HCB acetone solution was added into the vials by a micro-syringe before shaking (the final acetone fraction was below 0.5%). Both TX100 and HCB equilibrium concentrations were determined for each sample. 2.6. Chemical analysis Aqueous TX100 was analyzed by a high performance liquid chromatography (Hitachi L7100, Japan) equipped with an

J. Wan et al. / Journal of Hazardous Materials 196 (2011) 79–85

Fig. 1. (a) TX100 sorption isotherm at 0 and 400 mg L−1 HA; (b) influence of HA amount on TX100 (2 mmol L−1 ) sorption to kaolin.

L-7420 ultraviolet–visible (UV–vis) detector and an Agilent ZORBAX Eclipse XDB-C18 column (Agilent, USA). The wavelength was set at 223 nm. The mobile phase was 90% methanol plus 10% water, with a flow rate of 1.0 mL min−1 . HCB in the hexane was determined on a Hewlett–Packard 6890 GC equipped with an electron capture detector and a ZB-5 capillary column (Phenomenex, USA). Detailed information for GC procedure was included in our previous study [15]. The aqueous concentration of HA was measured by a Cary 50 UV–vis spectrophotometer (Varian, USA) at 254 nm (although TOC analysis is frequently used to quantify the humic substances (including HA), herein the co-presence of TX100 especially above 2 mmol L−1 may seriously interfere the TOC analysis of HA. However, a much higher absorptivity of HA under UV-254 than TX100 suggest that the spectrophotometer measurement might be more appropriate). The linear range of the HA working curve was 0–24 mg L−1 (r = 0.9999). The absorbance of HA in the absence of TX100 can be obtained directly. Based on the preliminary observation that the absorbance of TX100 and HA mix at 254 nm was additive, HA content could be deduced by subtracting the absorbance contributed by TX100 from the absorbance of the TX100-HA mixture (the absorbance of TX100 can be calculated from its working curve at 254 nm and the corresponded concentration obtained by HPLC). 3. Results and discussion 3.1. Effects of SOM on TX100 sorption in a kaolin–water system Fig. 1a displays the sorption isotherms of TX100 in the absence and presence of HA (400 mg L−1 ) in a kaolin/water system. The two isotherms can be well fitted by the Langmuir model in the range of 0–10 mmol L−1 , similar to those reported in the literature [7,16]. Comparison of the maximal sorption amounts of TX100 between HA and HA-free systems (24.0 vs 19.1 mmol kg−1 ) indicates that

81

the presence of HA at 400 mg L−1 increased the sorption of TX100. Enhanced sorption of surfactant to soil as an effect of SOM has been reported by a number of researchers [17–19]. Zhu et al. [20] revealed that the sorption capacity of HA for TX100 was almost one order larger than that of kaolin. Zhang et al. [21] also reported that much higher sorption capacity and partitioning coefficient (Kd ) for TX100 was exhibited by HA relative to original soil. Furthermore, as shown in Fig. 1b, for the observed HA concentration range, the sorption of TX100 firstly increased and then decreased, with a maximum at HA concentration of about 400 mg L−1 . The initial increase in TX100 sorption was strongly correlated with the partitioning of surfactant to the clay-bound HA. However, as the added HA increased above 400 mg L−1 , the amount and more importantly, the fraction of the dissolved HA increased considerably (as indicated by decreasing Kd values in Table S2). The dissolved HA may compete with the clay-bound HA for the partitioning of TX100, since HA molecule contains both hydrophilic and hydrophobic structure, very similar to surfactants. Another explanation may be that the dissolved HA could reduce the TX100 adsorption by forming HA–TX100 complexes, similar to the complexing of HOCs to DOM [10,22,23]. Lee et al. [22] found that the sorption isotherm of TX100 to a Florida peat was of a skewedGaussian shape, with a sharp decrease in TX100 sorption coefficient at equilibrium concentration above 1000 mg L−1 . It was suggested that at higher TX100 concentrations, more SOM was dissolved, and the DOM was expected to promote the “dissolution” of TX100 and reduce the probability of TX100 sorption [22]. Inspection of Fig. 1b further reveals that even at relatively high concentrations of dissolved HA (100–400 mg L−1 ), the sorption of TX100 was higher than sorption at HA = 0, suggesting a considerable binding of TX100 to the kaolin-bound HA. It can be also estimated from Fig. 1b that the critical HA amount when the dissolved HA outcompeted the clay-bound fraction for TX100 partitioning was approximately 1500 mg L−1 . As a result, it is expected that the effect of SOM on the sorption of nonionic surfactants is dependent on the content of SOM, and only at content high enough as to induce a significant amount of DOM, SOM can reduce the sorption of surfactants, thereby functioning positively to SER.

3.2. Effects of TX100 on HA sorption in a kaolin/water system The partitioning of HA within a kaolin/water system is presented as Fig. 2a. The sorption isotherm of HA in the TX100-free system can be described as a two phase linear relationship: with a steep increase for dissolved HA concentration of 0–7 mg L−1 and a followed slowing increase in the range of 7–360 mg L−1 . Moreover, a linear regression can be applied for HA sorption in the presence of TX100 when the aqueous concentration of HA ranged from 29 to 610 mg L−1 . Inspection of Fig. 2a reveals that the copresence of TX100 at 2 mmol L−1 decreased HA sorption to kaolin. Furthermore, the slope of HA isotherm in the presence of TX100 is lower than the HA isotherm in the absence of TX100 (5.7 vs 7.7), suggesting an increased reduction in HA sorption as the dissolved HA increases. Additionally, Fig. 2b shows a consistent decrease in HA sorption with an increase of TX100 concentration (0–8 mmol L−1 ). In particular, the presence of 8 mmol L−1 TX100 could reduce the sorption of HA by 38% compared with TX100-free system. Similar observations were also recorded by Cheng and Wong [13], therein the presence of Tween 80 at 150 mg L−1 dramatically reduced the sorption of DOM to soil. Preferential binding of Tween 80 molecules to soil with respect to DOM was hypothesized to be relevant to the above results [13]. Despite the lower calculated partitioning coefficients for TX100 (2–34 L kg−1 at 1–10 mmol L−1 , Table S1) in comparison with HA (23.5–80 L kg−1 at 350–1000 mg L−1 , Table S1), it may still be reasonable to suggest that the competitive

82

J. Wan et al. / Journal of Hazardous Materials 196 (2011) 79–85

Fig. 2. (a) HA sorption isotherm at 0 and 2 mmol L−1 TX100; (b) influence of TX100 concentration on HA (400 mg L−1 ) sorption to kaolin.

adsorption between the two molecules to the clay surface contributed primarily to the declined sorption of HA as the aqueous TX100 increased. 3.3. HCB partitioning within a HACK/TX100/water system Although the influence of TX100-HA interaction on TX100 and HA sorption to kaolin has been verified, the resultant effects on HCB partitioning in a clay/water system remain unclear. In fact, in a complicated surfactants/SOM/water/clay system, a series of sorption and complexing processes may occur and affect HCB partitioning behavior. To be more specific, these interactions include (1) the formation of DOM-HCB complex that may impede HCB sorption [10,24]; (2) the role of clay-bound SOM as a hydrophobic sorption domain for HCB [23,25,26]; (3) partitioning of HCB to TX100 micelles that can effectively reduce the potential of HCB sorption [3,27]; (4) the influence of TX100-HA interaction on TX100 and HA sorption to clay surface that may further affect the partitioning of HCB; (5) the possible sorption of HCB by the clay-bound TX100 and by the mineral [27–29]. Given the complexity of HOCs partitioning in the co-presence of TX100 and SOM, HCB sorption in SOM and TX100 individual system was investigated first. It can be found from Fig. 3a that HCB sorption correlated negatively with the amount of HA added at HA > 250 mg L−1 . The fraction of sorbed HCB decreased steadily from 1.0 to 0.5 as HA increased to 2500 mg L−1 . With the added HA increasing, the dissolved HA became increasingly abundant, therefore impeding the sorption of HCB. Nevertheless, at a low HA range of 0–150 mg L−1 , HCB sorption increased slightly. It is generally accepted that HA could bind HOCs through hydrophobic interactions to form HA–HOCs complexes [10,24,30], by which means the soil-bound HA sorb HOCs from solution while the dissolved HA desorb HOCs from soils. On the other hand, the formation of micelle-like structure of HA due to molecule aggregation is also proposed as the mechanism for the solubilization/enhanced desorption of HA for HOCs [9,27]. Furthermore, Fig. 3b indicates

Fig. 3. Influence of (a) HA and (b) TX100 on HCB sorption to kaolin individually. HCB 0.5 mg L−1 , pH 7.0.

that the sorption of HCB experienced a steady decline with the increase of TX100 added (0–2 mmol L−1 , Fig. 3b). Particularly, when 2 mmol L−1 of TX100 was added, the fraction of HCB sorbed was about 0.4. It is suggested that only at total surfactant concentration above critical desorption concentration (CDC), i.e., the aqueous surfactant concentration above CMC, reliable enhancement in HCB desorption by surfactants can be obtained [3,31]. The CDC of TX100 herein can be estimated from following equation: CDC = CMC + CsCMC wherein CsCMC is the concentration of TX100 in soil when the corresponding aqueous concentration is at the CMC, which can be calculated from CsCMC = 19.1CMC/(0.085 + CMC) (Fig. 1a). By substituting the measured CMC value of TX100 of 0.1 mmol L−1 into above equations, the estimated CDC is 0.62 mmol L−1 . As a consequence, no apparent desorption of HCB was recorded at total TX100 concentration of 0.5 mmol L−1 , while remarkable HCB desorption was obtained at TX100 dosage above 1 mmol L−1 . The partitioning of HCB in a HACK/TX100/water system, and the effects of clay-bound HA and dissolved HA on the sorption of HCB to HACK are illustrated in Fig. 4a–c. From Fig. 4a it is evident that the sorption of TX100 and HCB to solid HA both correlated positively with the HA content. With the HA concentration increasing from 0 to 2.5 g L−1 , the aqueous TX100 (initially 1 mmol L−1 ) decreased substantially to nearly undetectable, and correspondingly, the sorption of HCB (initial aqueous concentration was 0.5 mg L−1 ) increased steadily to 1.0. The reduced aqueous TX100 concentration was directly associated with the high affinity of HA for TX100, as the Kd values for TX100 herein were estimated as (0.6–8) × 103 L kg−1 , which are 1.8–2.9 orders of magnitude larger than that of TX100 to kaolin (10.9 L kg−1 at TX100 of 2 mmol L−1 ). The reduced aqueous HCB concentration, however, was mainly relevant to the strong affinity of both HA and HA-bound TX100 [27]. The binding constant (Kb ) for the HA–HCB complex was estimated as 5.5 × 104 L kg−1 by using the solubility enhancing method [24,32]

J. Wan et al. / Journal of Hazardous Materials 196 (2011) 79–85

83

Fig. 5. Estimation of factors contributed to HCB sorption, wherein “HA only” and “TX100 only represent HCB sorption as the inherent effect of HA (Fig. 3a) and 2 mmol L−1 TX100 only (Fig. 3b), respectively. “Theoretical effect of TX100 in TX100/HACK system” represents the theoretical HCB sorption by TX100 under the interference of HA coating, calculated from the aqueous TX100 content (Fig. 4c) and the inherent effect of TX100 on HCB sorption (Fig. 3b).

of HA coating amount of 0.5–5% (Fig. S2, the slope of the “SOM content” linear was less than 0.5). As a consequence, the formation of aqueous HA–HCB complex became increasingly pronounced. Furthermore, the reduced sorption of TX100 by the presence of HA may be further responsible for the decreasing sorption of HCB. As indicated in Fig. 4c, the dissolved fraction of TX100 increased from 1.3 to above 1.6 mmol L−1 as the amount of HA coating increased, slightly deviated from the results of Fig. 1b. A plausible explanation for this deviation was that for HACK the clay surface was preliminarily occupied by HA, and the replacement of TX100 for the bound-HA was somewhat harder than the competitive sorption with the dissolved HA (Section 3.1). Finally, the possibly synergistic effect of HA–TX100 complexes on HCB solubilization and reduced adsorption in the HACK/surfactant/water systems cannot be ruled out [13]. 3.4. Contributions of SOM and TX100 to HCB sorption

Fig. 4. Partitioning of TX100 and HCB within (a) solid HA/water, (b) HACK/water (pH 3.0) and (c) HACK/water (pH 7.0) system. Initial concentrations for HCB and TX100 were 0.5 mg L−1 and 2 mmol L−1 , respectively.

(detailed in the notification of Fig. S1 in Supplementary Materials), suggesting a rather strong interaction between HA and HCB. Furthermore, the reduced aqueous TX100 concentration as a function of solid HA could also contribute to the increased sorption of HCB considering the poor inherent solubility of HCB in water. Similar trends for TX100 and HCB sorption were also observed in the HACK/water system at pH 3 (Fig. 4b). Note that much lower aqueous TX100 (about 0.9 mmol L−1 ) and higher HCB sorption (over 0.6) were recorded at a very low HA content of 0.25% (Fig. 4b), which can be mainly attributed to the sorption by kaolin. The important role of mineral components in the partitioning of either surfactants or HOCs especially at low organic carbon contents has been addressed by other researchers [33,34]. Fig. 4c depicts the partitioning of TX100 and HCB at pH 7.0 within the HACK/water system. Contrary to the trends at pH 3.0 (Fig. 4b), the fraction of sorbed HCB decreased from 0.47 to 0.26 with the HA coating ranging from 0.25% to 5%. The declining sorption of HCB was consistent with the aforementioned observations in the HA/kaolin system (Fig. 3a). It is noteworthy to address that although an increase in HA coating may result in a direct increment in SOM content of kaolin, the fraction of dissolved HA was apparently higher than the fraction bound to kaolin all along the range

Based on above results of HCB partitioning behaviors in individual systems, in this section we will further summarize and compare the individual and combined roles of HA and TX100 in HCB sorption. For this purpose, Fig. 5 and Table 1 were constructed. The dashed curve (designated as “HA only”, i.e., Fig. 3a) and dotted curve (designated as “TX100 only”, i.e., Fig. 3b) represent HCB sorption on clay when HA (125–2500 mg L−1 ) and TX100 (2 mmol L−1 ) were added individually. The solid curve (designated as “HA & TX100 both”, i.e., Fig. 4c) displays HCB sorption due to the copresence of TX100 and HA. The curve of dash-dot (designated as “Theoretical effect of TX100 in TX100/HACK system”) is the theoretical HCB sorption caused by 2 mmol L−1 TX100 in the TX100/HACK system, in which the effect of additional aqueous TX100 (resulted from HA coating, Section 3.3 and Fig. 4c) on HCB sorption was involved (see detailed deduction procedure from Notes of Table 1). It should be mentioned that a neutral pH of 7.0 was set for all above systems, thus the influence of pH could be neglected. Inspection of Fig. 5 indicates that TX100 played a dominant role in the reduction of HCB sorption. Other than increasing HCB solubility by the inherent aqueous TX100 (in HA-free system), the reduced sorption of TX100 due to the co-presence of HA may lower the sorption of HCB additionally. This additional fraction of sorbed HCB was expected to be about 0.04, as revealed by comparing the result of “TX100-observed” and “TX100-Theoretical in TX100/HACK system” (Table 1). However, the effect of HA coating on HCB partitioning was dual depending on the coating amount. Although a dramatic decline in HCB sorption was obtained as a function of increasing HA coating in the whole range, the presence of HA was found to promote the HCB sorption

84

J. Wan et al. / Journal of Hazardous Materials 196 (2011) 79–85

Table 1 Contributions of TX100 and HA to HCB desorption from HACK. HA coating (%)

0.5 1.0 2.0 3.0 4.0 5.0

Fraction of HCB desorbed in different systems, Ce/C0 HA-observeda

TX100observeda

TX100-theoretical in HACK/TX100 systemb

HA & TX100 both-theoreticalc

Sum-HA & TX100 individuallyd

0.03 0.24 0.26 0.40 0.44 0.50

0.59 0.59 0.59 0.59 0.59 0.59

0.61 0.63 0.63 0.63 0.63 0.63

0.53 0.54 0.58 0.62 0.65 0.74

0.61 0.83 0.84 0.99 1.03 1.10

a

The observed fraction of HCB desorption by HA individually (results of Fig. 3a). Similar for “TX100-observed” (results of Fig. 3b). The theoretical fraction of HCB desorption as an effect of 2 mmol L−1 TX100 in an TX100/HACK system, calculated by substituting the value of the aqueous TX100 (Fig. 3c) to the following equation that correlates HCB sorption to aqueous TX100 concentration: b

FHCB =

0.109 + 1.08 1 + 1.95CeTX100 0.946

wherein FHCB is the fraction of HCB sorbed to HACK, CeTX100 is the aqueous TX100 concentration. c The observed fraction of HCB desorption as the effects of both HA coating and TX100 addition. d The theoretical fraction of HCB desorption by TX100 and HA, i.e., the sum of fraction of HCB desorption by HA and TX100 individually.

at lower coating amounts if taking the result of “TX100 only” (Fig. 5) as the reference. As discussed above, the competition between the aqueous and clay-bound HA determines the effect of HA coating on HCB sorption [10,23], regardless of the interference by TX100. Note that the intersection point of the “HA & TX100 both” and the “TX100 only” curve (P1 in Fig. 5) represents an equilibrium of the facilitating and impeding role of HA coating in HCB sorption. It is speculated that only at HA coating higher than 2.4%, the dissolved HA outcompeted the clay-bound HA for HCB partitioning. Excess HCB desorption from clay-bound HA could be reached due to this dissolved HA, compared to HCB desorptoion by TX100 alone. If we actually take the additional effect of sorption reduction of TX100 into account, the intersection would shift toward the higher HA coating, as depicted in Fig. 5 (P2), which means the positive effects on the reduction of HCB sorption by HA is expected at the coating amount above 3.3%. Inspection of Fig. 5 further reveals that P1 is actually the critical point at which the combined effects of HA and TX100 on HCB sorption equals to that of TX100 alone. That implies that at HA coating higher than P1, the co-presence of TX100 and HA resulted in less HCB sorption than either HA or TX100 alone. Nevertheless, as indicated by Table 1, the combined effect on reduction of HCB sorption was less than the sum of TX100 and HA alone. The results seemingly conflict with previous observations that a synergistic effect on HOCs desorption from soil by co-addition of Tween 80 and DOM was recorded [13]. However, note that the soil texture (one loam sandy soil) and the procedures for agents adding (simultaneous addition of Tween 80 and DOM into the system) therein would probably generate less sorption of DOM as compared with the HACK in our study. Furthermore, since the fractions of HCB “desorption” were 60% and over 40% for TX100 (2 mmol L−1 ) and HA (at HA-coating ≥2%) individually (Table 1), no synergistic or even additive effect can be expected by the combination of TX100 and HA; while as reported by Cheng and Wong [13], even when a synergistic effect was reached, the maximal desorption ratio was as low as 16.2% and 10.9% for phenanthrene and pyrene, respectively. We presumed that similar synergistic or additive effect by the copresence of TX100 and HA can be reached if a proper TX100 and HA concentration was applied (at least, the sum of HCB desorption efficiency by TX100 and HA alone was apparently below 100%).

4. Conclusions A deeper insight into the combined effect of surfactants, SOM, and DOM on HOC adsorption in a soil/water system is required to

better predict the efficacy of SER. Herein we investigated the partitioning of HCB in a HACK/water system in the presence of TX100. Main conclusions can be summed up as follows: (1) Appreciable influence of TX100-HA interaction on TX100 and HA sorption to clay was observed. The addition of HA at lower and higher amount than 1500 mg L−1 was found enhancing and reducing the sorption of TX100 to kaolin, respectively. Furthermore, the presence of TX100 suppressed the sorption of HA to kaolin over the entire observed concentration range of 0.5–2 mmol L−1 . The presented results suggest that for soils of high organic contents, the surfactant-SOM interaction may be beneficial to SER. In addition, a properly high dosage of surfactant may be more preferable to dissolve the SOM and exert such extra desorption. (2) TX100 contributed primarily to HCB desorption in a HACK/TX100/water system. DOM also showed encouraging enhancement in HCB desorption. The coated HA, however, imposed a negative and positive influence on TX100-enhanced desorption of HCB at coating amounts below and above 2.4%, respectively. The combined effect of HA and TX100 on HCB desorption was less than the sum of TX100 and HA alone. (3) It should be noted that HACK may or may not fully represent the whole soil in nature. However, the results we obtained are still of practical significance for SER since HA is often intimately associated with clay minerals in soil, and both HA and clay minerals are primary soil compounds that determine the sorption of surfactants and HOCs, i.e., the performance of SER. (4) Our study suggests that for soils of high organic contents, a combined influence of SOM and surfactants on HOCs removal can be expected, which implies a higher performance of SER, or a smaller dosage of surfactant may be required. However, the critical content of SOM that differentiates the positive role of SOM from the negative one in the SER has been revealed herein, the value may vary from site to site, depending on a series of factors such as characteristics of soil, SOM, surfactants, and HOCs. Furthermore, pH as an important environmental variable may influence the amount of DOM, therefore influencing the SER process. However, for the consideration of system simplification, in present study we only chose pH 7 as a typical circumstance. As a result, further studies are still required to focus on the combined effect of surfactants and SOM on SER for real soils with different organic contents, a broader range of pH conditions and a variety of HOCs.

J. Wan et al. / Journal of Hazardous Materials 196 (2011) 79–85

Acknowledgements This work was supported by the National Natural Science Foundation of China (Grants 20777024), the National High-Tech Research and Development (863) Programme (2009AA063103), and Shanghai Tongji Gao Tingyao Environtal Protection Sci. & Tech. Development Foundation. The Analytical and Testing Center of Huazhong University of Science and Technology is thanked for its help in kaolin and HA characterization. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.072. References [1] M. Svab, M. Kubala, M. Muellerova, R. Raschman, Soil flushing by surfactant solution: pilot-scale demonstration of complete technology, J. Hazard. Mater. 163 (2009) 410–417. [2] S.H. Yuan, Z. Shu, J.Z. Wan, X.H. Lu, Enhanced desorption of hexachlorobenzene from kaolin by single and mixed surfactants, J. Colloid Interface Sci. 314 (2007) 167–175. [3] K. Yang, L.Z. Zhu, B.S. Xing, Enhanced soil washing of phenanthrene by mixed solutions of TX100 and SDBS, Environ. Sci. Technol. 40 (2006) 4274–4280. [4] C.N. Mulligan, R.N. Yong, B.F. Gibbs, Surfactant-enhanced remediation of contaminated soil: a review, Eng. Geol. 60 (2001) 371–380. [5] W. Huang, W.J. Weber, A distributed reactivity model for sorption by soils and sediments. 10. Relationships between desorption, hysteresis, and the chemical characteristics of organic domains, Environ. Sci. Technol. 31 (1997) 2562–2569. [6] R. Chefetz, A.P. Deshmukh, P.G. Hatcher, Pyrene sorption by natural organic matter, Environ. Sci. Technol. 34 (2000) 2925–2930. [7] M.J. Salloum, M.J. Dudas, W.B. Mcgill, S.M. Murphy, Surfactant sorption to soil and geologic samples with varying mineralogical and chemical properties, Environ. Toxicol. Chem. 19 (2000) 2436–2442. [8] F.J. Ochoa-Loza, W.H. Noordman, D.B. Jannsen, M.L. Brusseau, R.M. Maier, Effect of clays, metal oxides, and organic matter on rhamnolipid biosurfactant sorption by soil, Chemosphere 66 (2007) 1634–1642. [9] P. Conte, A. Agretto, R. Spaccini, A. Piccolo, Soil remediation: humic acids as natural surfactants in the washings of highly contaminated soils, Environ. Pollut. 135 (2005) 515–522. [10] M. Rebhun, F. De Smedt, J. Rwetabula, Dissolved humic substances for remediation of sites contaminated by organic pollutants. Binding-desorption model predictions, Water Res. 30 (1996) 2027–2038. [11] H.H. Cho, J. Choi, M.N. Goltz, J.W. Park, Combined effect of natural organic matter and surfactants on the apparent solubility of polycyclic aromatic hydrocarbons, J. Environ. Qual. 21 (2002) 275–280. [12] H. Lippold, U. Gottschalch, H. Kupsch, Joint influence of surfactants and humic matter on PAH solubility. Are mixed micelles formed? Chemosphere 70 (2008) 1979–1986. [13] K.Y. Cheng, J.W.C. Wong, Combined effect of nonionic surfactant Tween 80 and DOM on the behaviors of PAHs in soil–water system, Chemosphere 62 (2006) 1907–1916.

85

[14] B. Wen, J.J. Zhang, S.Z. Zhang, X.Q. Shan, S.U. Khan, B.S. Xing, Phenanthrene sorption to soil humic acid and humin fractions, Environ. Sci. Technol. 41 (2007) 3165–3171. [15] J.Z. Wan, S.H. Yuan, K.T. Mak, J. Chen, T.P. Li, L. Lin, X.H. Lu, Enhanced washing of HCB contaminated soils by methyl-beta-cyclodextrin combined with ethanol, Chemosphere 75 (2009) 759–764. [16] P. Wang, A.A. Keller, Partitioning of hydrophobic organic compounds within soil–water-surfactant systems, Water Res. 42 (2008) 2093–2101. [17] S. Paria, Surfactant-enhanced remediation of organic contaminated soil and water, Adv. Colloid Interface Sci. 138 (2008) 24–58. [18] I.F. Paterson, B.Z. Chowdhry, P.J. Carey, S.A. Leharne, Examination of the adsorption of ethylene oxide-propylene oxide triblock copolymers to soil, J. Contam. Hydrol. 40 (1999) 37–51. [19] C.K.J. Yeh, L.C. Lin, Sorption and desorption kinetics of surfactants TX-100 and DPC on different fractions of soils, J. Environ. Sci. Health A Toxic/Hazard. Subst. Environ. Eng. 38 (2003) 1145–1157. [20] L.Z. Zhu, K. Yang, B.F. Lou, B.H. Yuan, A multi-component statistic analysis for the influence of sediment/soil composition on the sorption of a nonionic surfactant (Triton X-100) onto natural sediments/soils, Water Res. 37 (2003) 4792–4800. [21] G.Z. Zhang, H. Hu, W.L. Sun, J.R.J. Ni, Sorption of Triton X-100 on soil organic matter fractions: kinetics and isotherms, Environ. Sci. 21 (2009) 795–800. [22] J.F. Lee, P.M. Liao, C.C. Kuo, H.T. Yang, C.T. Chiou, Influence of a nonionic surfactant (Triton X-100) on contaminant distribution between water and several soil solids, J. Colloid Interface Sci. 229 (2000) 445–452. [23] Y. Laor, W.J. Farmer, Y. Aochi, P.F. Strom, Phenanthrene binding and sorption to dissolved and to mineral-associated humic acid, Water Res. 32 (1998) 1923–1931. [24] C. Rav-Acha, M. Rebhun, Binding of organic solutes to dissolved humic substances and its effects on adsorption and transport in the aquatic environment, Water Res. 26 (1992) 1645–1654. [25] S. Kleineidam, H. Rügner, B. Ligouis, P. Grathwohl, Organic matter facies and equilibrium sorption of phenanthrene, Environ. Sci. Technol. 32 (1999) 1637–1644. [26] Y. Ran, K. Sun, Y. Yang, B.S. Xing, E. Zeng, Strong sorption of phenanthrene by condensed organic matter in soils and sediments, Environ. Sci. Technol. 41 (2007) 3952–3958. [27] S.O. Ko, M.A. Schlautman, E.R. Carraway, Partitioning of hydrophobic organic compounds to sorbed surfactants. 1. Experimental studies, Environ. Sci. Technol. 32 (1998) 2769–2775. [28] J. Hur, M.A. Schlautman, Effects of mineral surfaces on pyrene partitioning to well-characterized humic substances, J. Environ. Qual. 33 (2004) 1733–1742. [29] Y.J. Zhang, D.Q. Zhu, H.X. Yu, Sorption of aromatic compounds to clay mineral and model humic substance–clay complex: effects of solute structure and exchangeable cation, J. Environ. Qual. 37 (2008) 817–823. [30] F. De Paolis, J. Kukkonen, Binding of organic pollutants to humic and fulvic acids: influence of pH and the structure of humic material, Chemosphere 34 (1997) 1693–1704. [31] W. Zhou, L. Zhu, Efficiency of surfactant-enhanced desorption for contaminated soils depending on the component characteristics of soil-surfactant-PAHs system, Environ. Pollut. 147 (2007) 66–73. [32] A.D. Site, Factors affecting sorption of organic compounds in natural sorbent/water systems and sorption coefficients for selected pollutants, J. Phys. Chem. Ref. Data 30 (2001) 187–439. [33] J.C. Joo, C.D. Shackelford, K.F. Reardon, Sorption of nonpolar neutral organic compounds to humic acid-coated sands: contributions of organic and mineral components, Chemosphere 70 (2008) 1290–1297. [34] X. Shi, L. Ji, D. Zhu, Investigating roles of organic and inorganic soil components in sorption of polar and nonpolar aromatic compounds, Environ. Pollut. 158 (2010) 319–324.

Journal of Hazardous Materials 196 (2011) 86–92

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Geopolymers prepared from DC plasma treated air pollution control (APC) residues glass: Properties and characterisation of the binder phase Ioanna Kourti a , Amutha Rani Devaraj a,b , Ana Guerrero Bustos c , David Deegan d , Aldo R. Boccaccini b,e , Christopher R. Cheeseman a,∗ a

Department of Civil and Environmental Engineering, Imperial College London, London SW7 2AZ, UK Department of Materials, Imperial College London, London SW7 2BP, UK c Institute of Construction Science Eduardo Torroja (CSIC), C/Serrrano Galvache, 4, 28033 Madrid, Spain d Tetronics Ltd., South Marston Business Park, Swindon, Wiltshire SN3 4DE, UK e Institute of Biomaterials, Department of Materials Science and Engineering University of Erlangen-Nuremberg, Cauerstr. 6, 91058 Erlangen, Germany b

a r t i c l e

i n f o

Article history: Received 14 February 2011 Received in revised form 30 August 2011 Accepted 31 August 2011 Available online 6 September 2011 Keywords: Geopolymers Plasma Incineration Energy Waste APC residues

a b s t r a c t Air pollution control (APC) residues have been blended with glass-forming additives and treated using DC plasma technology to produce a high calcium aluminosilicate glass (APC glass). This has been used to form geopolymer–glass composites that exhibit high strength and density, low porosity, low water absorption, low leaching and high acid resistance. The composites have a microstructure consisting of un-reacted residual APC glass particles imbedded in a complex geopolymer and C–S–H gel binder phase, and behave as particle reinforced composites. The work demonstrates that materials prepared from DC plasma treated APC residues have potential to be used to form high quality pre-cast products. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The air pollutions control systems at energy from waste (EfW) plants burning municipal solid waste (MSW) produce granular air pollution control (APC) residues. These are a hazardous waste with an absolute entry in the European Waste Catalogue (19 01 07*), and they contain fly ash, excess lime, carbon, relatively high concentrations of volatile heavy metals and soluble salts, particularly leachable chlorides. They also contain trace levels of organics including dioxins and furans. DC plasma technology provides a sustainable treatment for APC residues that meets the aims of the EU waste policy as it is a recycling/recovery option higher in the waste management hierarchy than alternative options [1,2]. In the DC plasma treatment process APC residues are combined with glass-forming additives and melted to produce inert APC glass [2]. There is increasing interest in developing sustainable construction products which contain recycled materials. Reuse of APC glass would have significant economic and environmental benefits, and

∗ Corresponding author. Tel.: +44 207 594 5971; fax: +44 207 823 9401. E-mail address: [email protected] (C.R. Cheeseman). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.081

would help to make DC plasma treatment of APC residues commercially attractive. Previous work has investigated the use of APC glass in glass-ceramics [3,4] and sintered ceramic tiles [5]. Geopolymers are synthetic alumino-silicates consisting of silica (SiO4 ) and alumina (AlO4 ) tetrahedra, linked by shared oxygen atoms [6]. Their formation is based on the chemistry of alkaliactivated inorganic binders and involves the chemical reaction of geopolymeric precursors, such as alumino–silicate oxides with alkali poly-silicates, to form polymeric Si–O–Al bonds [7]. The negative charge of Al3+ in four-fold coordination is balanced by the presence of positive ions such as Na+ , K+ and Ca2+ in framework cavities [8]. The empirical formula of geopolymers is therefore: Mn (–(SiO2 )z –AlO2 )n ,wH2 O

(1)

where M is a cation such as Na+ , K+ or Ca2+ , z is 1, 2 or 3, and n is the degree of poly-condensation. Geoploymers are associated with low CO2 emissions compared to Portland cement [9]. Using APC residue glass to form a geopolymer would provide a low-carbon emission reuse option that does not involve thermal processing. The microstructure and properties of geopolymers are determined by the raw materials used, and they can have high early

I. Kourti et al. / Journal of Hazardous Materials 196 (2011) 86–92 Table 1 Chemical composition of APC glass. Oxide

Composition (wt%)

Na2 O MgO Al2 O3 SiO2 P2 O5 K2 O CaO TiO2 Mn3 O4 Cr2 O3 Fe2 O3 Cl

2.88 2.31 14.78 41.10 0.77 0.03 32.59 1.19 0.23 0.06 4.07 2.5

compressive strengths, low shrinkage, rapid or slow setting, good acid resistance and fire resistance, and low thermal conductivity [10–12]. The use of APC glass in geopolymers results in geopolymer–glass composites in which residual APC glass particles act as rigid inclusions in a geopolymer matrix [13]. The presence of calcium in geopolymer systems can result in the formation of calcium silicate hydrate (C–S–H) gel and Al substituted C–S–H gel, and these are reported to decrease porosity and increase the strength of geopolymers [14–25]. This coexistence of geopolymer gel and hydration products has also been observed in alkali activated fly ash-Portland cement blends [26]. Recent research on the effect of alkalis and Al in C–S–H gel has confirmed that the structure is modified by alkali metals and Al [27,28]. The formation of a geopolymer phase or C–S–H gel is determined by the chemical and mineralogical composition, the physical properties of the aluminosilicate and Ca sources, the alkalinity of the activator and the percentage Ca in the system [17–19,22,24]. This work follows from previous research [13] in which novel geopolymers prepared using APC glass were described. The effect of processing parameters on geopolymerisation of APC glass was examined. In the present paper the properties of optimised APC glass geopolymers and the formation of a complex binder phase are investigated in detail. This provides new insight and information on the potential applications of this material and the link between the final geopolymer composite properties and microstructure. 2. Materials and methods 2.1. Materials Glass produced by DC plasma treatment of APC residues was supplied by Tetronics Ltd. (Swindon, UK) in the form of a coarse granular material with Pb2+ > Ni2+ > Cd2+ and with a saturation loading capacity of 0.86 mmol of Cu/g. I.M. El-Nahhal and co-workers have prepared a series of amino group functionalized polysiloxane-immobilized ligand systems via the sol–gel process and found their application in separation of heavy metal ions from aqueous solution [14–18]. In recent years, there has been intense interest in the preparation and application of functional polysilsesquioxane particles [19–21]. We have reported the strong adsorbability of Ag(I) ions onto poly(3-mercaptopropylsilsesquioxane) (PMPSQ) microspheres recently [22]. Beari et al. [23] have studied the hydrolytic condensation of 3-aminopropyltriethoxysilane (APTES) in aqueous solutions and found the hydrolysis and condensation products of APTES was not precipitated from the solution even after several weeks due to their excellent water solubility. In this paper, we report a one-step synthetic process for synthesizing amino-functionalized polysilsesquioxane having high content of amino groups to develop an efficient adsorbent of

X. Lu et al. / Journal of Hazardous Materials 196 (2011) 234–241

Nomenclature APTES b C0 Ce h k1 k2

3-aminopropyltriethoxysilane Langmuir constant (L/mmol) initial metal ions concentration (mmol/L) equilibrium metal ions concentration (mmol/L) initial adsorption rate (mmol g−1 min−1 ) pseudo-first-order rate constant (min−1 ) pseudo-second-order rate constant (g mmol−1 min−1 ) m mass of the adsorbent (g) MTMS methyltrimethoxysilane PAMSQ poly(aminopropyl/methyl)silsesquioxane PMSQ poly(methylsilsesquioxane) adsorption capacity (mmol/g) qe qm theoretical saturation adsorption capacity (mmol/g) qt adsorption capacity at t (mmol/g) regression coefficient R2 t time (min) the solution volume (L) V

heavy metals. Poly(aminopropyl/methyl)silsesquioxane (PAMSQ) particles were obtained by hydrolytic co-condensation of 3aminopropyltriethoxysilane (APTES) with methyltrimethoxysilane (MTMS) in aqueous medium. The PAMSQ particles have the ability to effectively remove the Cu(II) and Pb(II) ions from the aqueous solution. The effect of adsorption time, initial metal ions concentration, and solution pH was studied by a static adsorption method to optimize the Cu(II) and Pb(II) adsorbability of PAMSQ particles. 2. Experimental 2.1. Materials 3-Aminopropyltriethoxysilane (APTES, ≥98.0%) was purchased from Diamond Advanced Material of Chemical Inc. Methyltrimethoxysilane (MTMS, ≥98.0%) was purchased from Jiangsu Danyang Organosilicon Material Industrial Corporation. Ammonium hydroxide solution (NH4 OH, 25%), copper sulfate pentahydrate and lead nitrate of analytical reagent grade were commercially obtained and used as received.

235

constant temperature of 20 ◦ C at 100 rpm. After a desired period of adsorption, the particles were filtered from the solution with millipore filter membrane (0.22 ␮m). The final concentrations of the metal ions in the solution were analyzed by inductively coupled plasma, atomic emission spectrometry (ICP-AES, IRIS 1000, Thermo Elemental). The equilibrium adsorption capacity was calculated from Eq. (1). qe =

(C0 − Ce )V m

(1)

where qe (mmol/g) is the adsorption capacity and C0 (mmol/L) and Ce (mmol/L) are respectively the initial and equilibrium metal concentrations. V (L) is the solution volume, and m (g) is the weight of the adsorbent. 2.4. Measurements Elemental analyses were performed with a Vario EL III elemental analyzer (Elementar Analysen systeme GmbH, Germany). FT-IR spectra were recorded on a iS10 FT-IR spectrophotometer (Nicolet, USA). The samples were mixed with potassium bromide and pressed to a disk to measure the absorption spectrum. High-resolution solid-state 29 Si NMR spectra were measured at room temperature on a Bruker Avance 400 MHz spectrometer (silicon frequency 99.36 MHz) equipped with a Bruker solid-state accessory. Spectra were obtained using a broadband probehead with a 4 mm double air bearing magic-angle spinning assembly. Chemical shifts of silicon atoms in silsesquioxane compounds are referred to using the traditional terminology Tn , where the superscript corresponds to the number of oxygen bridges to other silicon atoms. Thus, an uncondensed monomer was designated T0 and a fully condensed polymer with no residual silanols was assigned as T3 silicon atoms [24]. A field emission SEM (JSM-7401F FE-SEM, JEOL Ltd., Japan) was utilized to study the morphology of the PAMSQ particles. The particles were sputter-coated with gold for SEM observations. Specific surface area data were performed on a Micromeritics ASAP 2020 surface area and porosity analyzer with BET method. TGA was performed on a SDT Q600 (TA Instruments, USA) thermal analyzer at a heating rate of 10 ◦ C min−1 in air. 3. Results and discussion 3.1. Synthesis and properties of PAMSQ particles

2.2. Synthesis of PAMSQ particles by hydrolytic co-condensation process PAMSQ particles were prepared using base catalyzed sol–gel process in aqueous medium. The mixture of APTES and MTMS at different molar ratios was added to 100 mL of water, maintaining 10% weight percentage. Ammonium hydroxide solution (0.16 mL) was added into the above solution. The reaction was continued overnight at room temperature. Then the resulting precipitate was filtrated with millipore filter membrane (0.22 ␮m) and rinsed thoroughly with distilled water and ethanol several times to remove the residual NH4 OH as well as unreacted monomers or oligomers. Finally, the products were dried in vacuum and the PAMSQ particles were attained. 2.3. Adsorption of Cu(II) and Pb(II) onto the PAMSQ particles Batch adsorption experiments were conducted using PAMSQ particles as a adsorbent to adsorb Cu(II) or Pb(II) ions from aqueous single metal ion solutions. The sample pH was adjusted to the desired value with HNO3 or ammonia solution. The batch adsorption experiments were conducted in a shaker bath kept at

Synthesis of PAMSQ particles with controllable amount of aminopropyl functional groups using APTES and MTMS as precursors by hydrolytic co-condensation process were conducted (Scheme 1). Generally, hydrolytic condensation of the organotrimethoxysilanes in water or ethanol–water was quite rapid under basic conditions [23–25]. Initial hydrolysis of the APTES and MTMS resulted in silanol oligomers. Silanol (Si–OH) was very reactive and then condensed to form polysilsesquioxanes in the presence of base catalyst. The amino groups of APTES could increase the pH of solution and accelerates the hydrolytic co-condensation process. Simultaneous hydrolysis of APTES and MTMS led to cocondensation, but the product state was different with the variation of APTES/MTMS molar ratios. If the APTES molar ratio in the precursors was less than 40%, white precipitate appeared. However, the co-condensation products were not precipitated from the solution when the APTES molar ratio in the precursors was above 50% due to the excellent solubility of co-condensed PAMSQ in water. SEM images show morphology of polysilsesquioxanes particles (see Fig. 1). The PMSQ particles prepared from MTMS alone are spherical with a medium size of 2.0 ␮m. The particle aggregation was quite evident for the copolymerized PAMSQ particles

236

X. Lu et al. / Journal of Hazardous Materials 196 (2011) 234–241

Scheme 1. Schematic illustration of the synthetic strategy for PAMSQ particles. Note that “OR” can present either silanol (OH) or other silane units

and the particle size decreased with increasing APTES amount in the APTES/MTMS mixtures, as seen in Fig. 1b–d. The results could be due to the excellent solubility of APTES in water, which makes the copolymerized PAMSQ particles to be more hydrophilic. Additionally, the amino groups of APTES raised pH of reaction solution and catalyze the hydrolytic co-condensation reactions of APTES and MTMS. Increase in hydrolysis rates with increasing pH led to a higher nucleation rate, which also resulted in a larger number of particles but a smaller final particle size [24]. Table 1 shows the properties of polysilsesquioxanes particles. Amino group contents of PAMSQ particles determined by element analysis were lower than the theoretical values. The results could be due to the excellent solubility of APTES and its hydrolysis and condensation products in water [23]. These results are consistent with those reported by Liu et al. [20] that copolymerized aminopropyl/phenylsilsesquioxane microparticles synthesized from the hydrolytic co-condensation of APTES and phenyltriethoxysilane (PTES). The specific surface areas of the polysilsesquioxanes particles were evaluated using the BET method as shown in Table 1.

The relatively small BET surface area values were due to complete condensation of polysilsesquioxanes. The influence of amino groups content on Cu(II) and Pb(II) adsorption onto the PAMSQ particles was investigated. As shown in Table 1, the aminopropyl functionalized PAMSQ samples revealed a high affinity towards Cu(II) and Pb(II), and the adsorption capacity of metal ions onto the PAMSQ particles increase with increasing amino groups content. However, the unmodified PMSQ without amino groups adsorbed only small amount of Cu(II) and Pb(II) ions. The results of Table 1 suggest that the mechanism of adsorption involves primarily metal ion complexation by the amino groups. High amino group contents make PAMSQ3 fit as a representative copolymerized product to be used for further studies in adsorption experiment. Solid-state 29 Si NMR spectra are powerful methods for characterizing the chemical structure of polysilsesquioxane frameworks. According to Arkhireeva et al. [26], in the case of the polysilsesquioxane derived from MTMS, resonances at −65.9, −57.1, −48.5 ppm can be assigned to T3 , T2 and T1 species, respectively. And Caravajal et al. [27] reported the 29 Si NMR spectra

Fig. 1. SEM images of polysilsesquioxanes particles: (a) PMSQ, (b) PAMSQ1, (c) PAMSQ2 and (d) PAMSQ3.

X. Lu et al. / Journal of Hazardous Materials 196 (2011) 234–241

237

Table 1 Properties of polysilsesquioxanes particles. Samples

APTES in comonomers (mol%)

a

Amino content in PAMSQ (mol%)

BET surface area (m2 /g)

b Cu(II) adsorption capacity (mmol/g)

b Pb(II) adsorption capacity (mmol/g)

PMSQ PAMSQ1 PAMSQ2 PAMSQ3

0 20 30 40

0 15 22 27

4.3 4.8 6.7 5.0

0.13 0.80 1.62 2.25

0.17 0.58 0.90 1.14

a b

The amino content in PAMSQ is determined by elemental analysis of nitrogen content. Initial metal ions concentration: 10 mM, adsorption time: 5 h, adsorbent dose: 2 g/L.

1482 cm−1 (ıN–H ), 3376 cm−1 (N–H ), 2934 and 2970 cm−1 (C–H ). A broad band around 3400 cm−1 could be attributed both to the adsorbed water and to the Si–OH group [29]. The thermal stability of the PMSQ and PAMSQ3 in air was investigated using TGA (Fig. 4). Weight loss in the 100–250 ◦ C range for polysilsesquioxane is probably due to the residual reaction of alkoxysilyl groups [26]. The thermal reduction of polysilsesquioxane in the 250–700 ◦ C range appeared to be mainly due to the decomposition of organic moieties groups [24]. The thermal decomposition temperatures of the PAMSQ3 particles in air, at 5% weight loss and at 10% weight loss are 257 ◦ C and 370 ◦ C, respectively. The TGA result shows that both the PMSQ and PAMSQ3 particles have good thermal stability.

Fig. 2. Solid-state

29

Si NMR spectrum of PAMSQ3 particles.

3.2. Adsorption kinetics of Cu(II) and Pb(II) onto the PAMSQ particles

exhibited major peaks in the regions of −66, −58 and −49 ppm, due to the silicons of the attached CH2 CH2 CH2 NH2 moiety of APTES-modified silica. Fig. 2 shows solid-state 29 Si NMR spectrum of PAMSQ3 particles. There is one large peak at −62.1 ppm and a weak shoulder peak at about −53 ppm assigned to fully condensed T3 and linear T2 species, respectively. The formation of T1 and T0 species is insignificant, suggesting that the co-condensation is quite completed [28]. Fig. 3 shows IR spectra of PMSQ and PAMSQ3particles. PMSQ and PAMSQ3 exhibit well-defined methyl group and Si–O–Si absorption bands at: 2970, ca. 2921–2934 cm−1 (C–H ), 1410 cm−1 (ıC–H in Si–R), 1272 cm−1 (ıC–H in Si–R), ca. 1119–1127, ca. 1032–1036 cm−1 (Si–O–Si ), and 778 cm−1 (Si–C ) [29,30]. Si–O–Si stretching peaks at 1119–1127 cm−1 indicate the presence of cagestructure while adsorption at 1032–1036 cm−1 show that the ordered structure is probable ladderlike or layered [20,30]. The spectrum of PAMSQ3 showed bands due to aminopropyl groups at

The kinetics of adsorption is one of the important characteristics that define the efficiency of adsorption. The effect of contact time on the adsorption of Cu(II) and Pb(II) onto the PAMSQ3 particles is shown in Fig. 5. The kinetic curve showed that the adsorption was rapid for the first 10 min, when the adsorption capacity reaches up to 1.49 mmol/g and 0.58 mmol/g for Cu(II) and Pb(II), respectively, and then slowed gradually. The initial rapid step of metal ions adsorption may be attributed to the physical and reactive adsorption between metal ions and the amino groups on the surface of the PAMSQ particles. However, the subsequent slow step is attributable to the adsorption inside the particles, representing the diffusion of Cu(II) and Pb(II) ions into the inner of the particles over a long period. Experimental results suggest that the amount of metal ions adsorbed increased with increasing adsorption time and reached equilibrium at 300 min for Cu(II) and Pb(II). Hence, in the present study, we used 300 min contact time for further experiments. Pseudo-first-order and pseudo-second-order models [31] were used to test the experimental data and thus elucidate the

Fig. 3. IR spectra of PMSQ and PAMSQ3 particles.

Fig. 4. TGA curves of PMSQ and PAMSQ3 in air.

238

X. Lu et al. / Journal of Hazardous Materials 196 (2011) 234–241

Fig. 5. Effect of adsorption time on Cu(II) and Pb(II) adsorption onto PAMSQ3 particles (initial metal ions concentration: 10 mM; adsorbent dose: 2 g/L).

adsorption kinetic process. The Lagergren pseudo-first-order kinetic model, represented as: k1 log(qe − qt ) = log qe − t 2.303

(2)

where qe and qt are the amounts of metal ions adsorbed (mmol/g) at equilibrium and time t, respectively, and k1 (min−1 ) is the pseudo-first-order rate constant. The qe and rate constant k1 were calculated by plotting the log (qe − qt ) vs. t. As seen from Fig. 6, the pseudo-first-order model does not fit the data well. The experimental and calculated qe values, pseudo-first-order rate constants and regression coefficient (R2 ) values are presented in Table 2. The calculated qe values in the pseudo-first-order model were not in agreement with the experimental qe values, suggesting that the adsorption of Cu(II) and Pb(II) does not follow pseudo-firstorder kinetics. In order to find a more reliable description of the adsorption kinetics of Cu(II) and Pb(II) ions onto the particles, a pseudo-second-order kinetic model was applied to the experimental data. The pseudo-second-order equation can be written as: t 1 1 = + t qt qe k2 q2e

(3)

where qe and qt are defined as in the pseudo-first-order kinetic model; k2 is the pseudo-second-order rate constant. The slope and intercept of the linear plot t/qt vs. t in Fig. 7 yielded the values

Fig. 6. Pseudo-first-order kinetic plots for the adsorption of Cu(II) and Pb(II) onto PAMSQ3 particles.

Fig. 7. Pseudo-second-order kinetic plots for the adsorption of Cu(II) and Pb(II) onto PAMSQ3 particles.

of qe and k2 . Additionally, the initial adsorption rate (h) can be determined from k2 and qe values using h = k2 q2e . The regression coefficients (R2 ) and several parameters obtained from the pseudosecond-order kinetic model are also shown in Table 2. As seen from Table 2, the calculated qe values are in good agreement with experimental qe values. Moreover, the obtained R2 values for Cu(II) and Pb(II) adsorption both are above 0.99. Hence, the adsorption kinetics could well be approximated more favorably by pseudosecond-order kinetic model for Cu(II) and Pb(II) onto the PAMSQ particles. The pseudo-second-order model was developed based on the assumption that the determining rate step may be chemisorption promoted by covalent forces through the electron exchange, or valency forces through electrons sharing between adsorbent and adsorbate [31], indicating that the adsorption of Cu(II) and Pb(II) on PAMSQ3 particles is mainly the chemically reactive adsorption. 3.3. Effect of initial metal ions concentration and adsorption isotherm The effect of the initial metal ions concentration on adsorption of Cu(II) and Pb(II) onto the PAMSQ3 particles is shown in Fig. 8. At a lower initial metal ions concentration, abundant aminopropyl groups on the surface of the PAMSQ particles can react with metal ions, resulting in a significantly increased adsorption of Cu(II) and Pb(II). Then the adsorption process gradually becomes slow with increasing initial metal ions concentration.

Fig. 8. Effect of initial metal ions concentration on adsorption of Cu(II) and Pb(II) on PAMSQ3 particles (initial metal ions concentration: 1–20 mM; adsorption time: 5 h; adsorbent dose: 2 g/L).

X. Lu et al. / Journal of Hazardous Materials 196 (2011) 234–241

239

Table 2 Kinetic model equations for Cu(II) and Pb(II) adsorption onto the PAMSQ3 particles. Metal ions

Cu(II) Pb(II)

qe (exp.) (mmol/g)

2.25 1.14

Pseudo-first-order

Pseudo-second-order

k1 (min−1 )

qe (cal.) (mmol/g)

R2

k2 (g mmol−1 min−1 )

h (mmol g−1 min−1 )

qe (cal.) (mmol/g)

R2

0.01073 0.01262

1.37 0.85

0.8695 0.8830

0.021 0.028

0.11 0.04

2.30 1.22

0.9915 0.9908

facile method under moderate condition would have a promising application as a cost-effective adsorbent. 3.4. Effect of solution pH on adsorption

Fig. 9. Langmuir plots for Cu(II) and Pb(II) adsorption on the PAMSQ3 particles (initial metal ions concentration: 1–20 mM; adsorption time: 5 h; adsorbent dose: 2 g/L).

Fig. 10 shows the effect of the solution pH on the adsorptions of Cu(II) and Pb(II) by PAMSQ3 particles, respectively. The pH in a range of 2.0–5.0 was chosen to avoid the precipitations of Cu(OH)2 and Pb(OH)2 . The adsorption capacities increased with an increase in solution pH in the pH range of 2.0–5.0 and no adsorption was observed at pH 2.0. This could be attributable to a competitive adsorption between metal ions and H+ ions on PAMSQ3 particles. At low pH value, the adsorption of metal ions is decreased because high concentrations of competitive H+ ions occupy the adsorption sites, whereas the protonated amino groups are deprotonated with increasing pH value, enhancing metal ions adsorbability [15,44]. Therefore, the solution pH around 5.0 could be optimal for the application of the PAMSQ3 particles as efficient Cu(II) and Pb(II) adsorbent. 3.5. Adsorption mechanism of metal ions onto PAMSQ particles

Fig. 9 shows the adsorption isotherms of Cu(II) and Pb(II) by the PAMSQ3 particles at 20 ◦ C. The adsorption data were plotted according to Langmuir equation: Ce 1 Ce = + qe qm qm b

(4)

where qm and b are the characteristic Langmuir parameters. qm is the theoretical saturation adsorption capacity of the monolayer (mmol/g) and b is a constant related to the intensity of adsorption. Plotting Ce /qe against Ce gives straight lines as shown in Fig. 9. Table 3 displays the coefficients of the Langmuir model along with regression coefficients (R2 ). As seen from Table 3, the R2 values for the Langmuir isotherm models were both above 0.99, suggesting that the Langmuir model closely fits the experimental results. The calculated qm values are in good agreement with those experimentally found. PAMSQ3 particles possess a strong capability to adsorb Cu(II) and Pb(II) ions from aqueous solutions, indicating a great potential as a high efficiency absorbent. The variations of metal ions uptake on various adsorbents are associated with adsorbent properties such as structure, functional groups, and specific surface area. Table 4 shows the comparison of the maximum adsorption capacity of PAMSQ for Cu(II) and Pb(II) onto various adsorbents reported in the literature. The results demonstrate that the adsorption capacities of PAMSQ3 particles for Cu(II) and Pb(II) were high when compared to several other adsorbents. Therefore, it could be believed that the PAMSQ3 particles synthesized from common silane coupling agent through a

Sorption is broadly defined as the transferring of ions from the solution phase to the solid phase via various mechanisms such as physical and chemical adsorption, surface precipitation, or solidstate diffusion or fixation [45]. According to hard and soft acids and bases theory of Pearson [11], aminopropyl group functionalized PAMSQ has bonding ability with heavy metal ions such as Cu(II) and Pb(II). The FTIR spectra of PAMSQ3 particles before and after adsorption of Cu(II) ion are shown in Fig. 11. Appearance of a sharp peak at 619 cm−1 after adsorption of Cu(II) on PAMSQ3 is assigned to the stretching vibration of N–Cu bond formed during complexation process [37]. The FTIR results confirm that nitrogen of PAMSQ3 particles are actively participated during the adsorption process through complexation with Cu(II) ion. So the adsorption mechanism of metal ions onto PAMSQ3 particles involves primarily metal ions complexation by the amino groups.

Table 3 Coefficients of Langmuir isotherms. Metal ions

qm (exp.) (mmol/g)

qm (cal.) (mmol/g)

b (L/mmol)

R2

Cu(II) Pb(II)

2.29 1.31

2.26 1.47

11.82 0.48

0.9986 0.9925

Fig. 10. Effect of the pH on adsorption of Cu(II) and Pb(II) on PAMSQ3 particles (initial metal ions concentration: 2.5 mM; adsorption time: 5 h; adsorbent dose: 2 g/L).

240

X. Lu et al. / Journal of Hazardous Materials 196 (2011) 234–241

Table 4 Comparison of maximum adsorption capacity of PAMSQ for Cu(II) and Pb(II) onto various adsorbents reported in the literature. Adsorbents

Adsorption capacity (mg/g)

SBA-15 mesoporous silica with melamine-based dendrimer amines Silica gel chemically modified by triethylenetetraminomethylenephosphonic acid Silica gel modified with 5-amino-1,3,4-thiadiazole-2-thiol Ethylenediaminetriacetic acid functionalized silica–gel 4-Amine-2-mercaptopyrimidine modified silica gel 2-Aminophenylaminopropylpolysiloxane Epichlorohydrin cross-linked xanthate chitosan Aminated polyacrylonitrile nanofibers Porous chitosan monoliths PS-EDTA resin 2-((2-Aminoethylamino)methyl)phenol-functionalized activated carbon Iron oxide coated sewage sludge Ulva lactuca algae Potassium hydroxide treated pine cone powder PAMSQ particles

Cu(II)

Pb(II)

Ref.

126 19.8 1.21 99.4

130 16.8 1.54

[32] [33] [34] [35] [36] [16] [37] [38] [39] [40] [41] [42] [9] [43] This work

80.19 130.91 43.47 116.52 141.8 42.1 12.1 17.3 112 19.22 146

32.1 16.2 42.4 230 26.27 272

Fundamental Research Funds for the Central Universities (Project no. WA1013012), and East China University of Science and Technology (ECUST) through fostering the Undergraduates Innovating Experimentation Project (No. X0807).

References

Fig. 11. FTIR spectra of PAMSQ3 particles (a) before adsorption and (b) after adsorption of Cu(II). Inset: scheme of copper ions binding.

4. Conclusions Amino-functionalized polysilsesquioxane particles have been synthesized by hydrolytic co-condensation using APTES and MTMS as precursors in the presence of base catalyst in aqueous medium. The process is a one-step co-condensation synthetic route where the functionalities of the particles can be easily controlled by changing the organosilanes feed ratio. The results of solid-state NMR spectroscopy, FT-IR analysis, and elemental analysis confirmed the co-condensation between organosilanes. The PAMSQ particles had shown as an efficient adsorbent for the removal of Cu(II) and Pb(II). The adsorption behavior of Cu(II) and Pb(II) onto PAMSQ particles is influenced by the adsorption time, initial concentration of metal ions, and solution pH. The kinetic studies indicated that the adsorption process well fits the pseudo-second-order kinetics with a rapid initial adsorption rate. The experimental data was well fit by the Langmuir isotherm model. The PAMSQ particles demonstrate the highest Cu(II) and Pb(II) adsorption capacity of 2.29 mmol/g and 1.31 mmol/g at an initial metal ions concentration of 20 mM, respectively. Therefore, there are good prospects for the PAMSQ particles in practical applications for the removal of Cu(II) and Pb(II) ions from their aqueous solutions. Acknowledgments This work was financially supported by the National Natural Science Foundation of China (Project no. 21006025), the

[1] X.G. Li, X.L. Ma, J. Sun, M.R. Huang, Powerful reactive sorption of silver(I) and mercury(II) onto poly(o–phenylenediamine) microparticles, Langmuir 25 (2009) 1675–1684. [2] I.B. Kim, U.H.F. Bunz, Modulating the sensory response of a conjugated polymer by proteins: an agglutination assay for mercury ions in water, J. Am. Chem. Soc. 128 (2006) 2818–2819. [3] N. Lihareva, L. Dimova, O. Petrov, Y. Tzvetanova, Ag+ sorption on natural and Naexchanged clinoptilolite from Eastern Rhodopes, Bulgaria, Micropor. Mesopor. Mater. 130 (2010) 32–37. [4] M.J. Manos, C.D. Malliakas, M.G. Kanatzidis, Heavy-metal-ion capture, ionexchange, and exceptional acid stability of the open-framework chalcogenide (NH4 )4 In12 Se20 , Chem. Eur. J. 13 (2007) 51–58. [5] B. Pollet, J.P. Lorimer, S.S. Phull, J.Y. Hihn, Sonoelectrochemical recovery of silver from photographic processing solutions, Ultrason. Sonochem. 7 (2000) 69–76. [6] A.R. Ladhe, P. Frailie, D. Hua, M. Darsillo, D. Bhattacharyya, Thiol-functionalized silica-mixed matrix membranes for silver capture from aqueous solutions: experimental results and modeling, J. Membr. Sci. 326 (2009) 460–471. [7] P. Vassileva, P. Tzvetkova, L. Lakov, O. Peshev, Thiouracil modified activated carbon as a sorbent for some precious and heavy metal ions, J. Porous Mater. 15 (2008) 593–599. [8] C.Y. Chen, M.S. Lin, K.R. Hsu, Recovery of Cu(II) and Cd(II) by a chelating resin containing aspartate groups, J. Hazard. Mater. 152 (2008) 986–993. [9] S. Zakhama, H. Dhaouadi, F. M’Henni, Nonlinear modelisation of heavy metal removal from aqueous solution using Ulva lactuca algae, Bioresour. Technol. 102 (2011) 786–796. [10] H. Yang, R. Xu, X.M. Xue, F.T. Li, G.T. Li, Hybrid surfactant-templated mesoporous silica formed in ethanol and its application for heavy metal removal, J. Hazard. Mater. 152 (2008) 690–698. [11] R.G. Pearson, Hard and soft acids and bases, J. Am. Chem. Soc. 85 (1963) 3533. [12] K.F. Lam, K.L. Yeung, G. McKay, A rational approach in the design of selective mesoporous adsorbents, Langmuir 22 (2006) 9632–9641. [13] W. Yantasee, Y.H. Lin, G.E. Fryxell, K.L. Alford, B.J. Busche, C.D. Johnson, Selective removal of copper(II) from aqueous solutions using fine-grained activated carbon functionalized with amine, Ind. Eng. Chem. Res. 43 (2004) 2759–2764. [14] N. El-Ashgar, I. El-Nahhal, Preconcentration and separation of copper(II) by 3aminopropylpolysiloxane immobilized ligand system, J. Sol–Gel Sci. Technol. 34 (2005) 165–172. [15] N.M. El-Ashgar, I.M. El-Nahhal, Separation of Cu(II), Ni(II), and Co(II) ions using iminobis(N-2-aminoethylacetamide) polysiloxane-immobilized ligand system, J. Disper. Sci. Technol. 27 (2006) 915–919. [16] N.M. El-Ashgar, I.M. El-Nahhal, M.M. Chehimi, F. Babonneau, J. Livage, Extraction of Co, Ni, Cu, Zn and Cd ions using 2aminophenylaminopropylpolysiloxane, Environ. Chem. Lett. 8 (2010) 311–316. [17] I.M. El-Nahhal, N.M. El-Ashgar, A review on polysiloxane-immobilized ligand systems: synthesis, characterization and applications, J. Organomet. Chem. 692 (2007) 2861–2886. [18] I.M. El-Nahhal, N.M. El-Ashgar, M.M. Chehimi, P. Bargiela, J. Maquet, F. Babonneau, J. Livage, Metal uptake by porous iminobis(N-2-aminoethylacetamide)modified polysiloxane ligand system, Micropor. Mesopor. Mater. 65 (2003) 299–310.

X. Lu et al. / Journal of Hazardous Materials 196 (2011) 234–241 [19] H. Mori, M.G. Lanzendorfer, A.H.E. Muller, J.E. Klee, Silsesquioxane-based nanoparticles formed via hydrolytic condensation of organotriethoxysilane containing hydroxy groups, Macromolecules 37 (2004) 5228–5238. [20] S.M. Liu, X.M. Lang, H. Ye, S.J. Zhang, J.Q. Zhao, Preparation and characterization of copolymerized aminopropyl/phenylsilsesquioxane microparticles, Eur. Polym. J. 41 (2005) 996–1001. [21] H.W. Du, P.D. Hamilton, M.A. Reilly, A. d’Avignon, P. Biswas, N. Ravi, A facile synthesis of highly water-soluble, core–shell organo-silica nanoparticles with controllable size via sol–gel process, J. Colloid Interface Sci. 340 (2009) 202–208. [22] X. Lu, Q.F. Yin, Z. Xin, Z.Q. Zhang, Powerful adsorption of silver(I) onto thiolfunctionalized polysilsesquioxane microspheres, Chem. Eng. Sci. 65 (2010) 6471–6477. [23] F. Beari, M. Brand, P. Jenkner, R. Lehnert, H.J. Metternich, J. Monkiewicz, H.W. Siesler, Organofunctional alkoxysilanes in dilute aqueous solution: new accounts on the dynamic structural mutability, J. Organomet. Chem. 625 (2001) 208–216. [24] S. Sankaraiah, J.M. Lee, J.H. Kim, S.W. Choi, Preparation and characterization of surface-functionalized polysilsesquioxane hard spheres in aqueous medium, Macromolecules 41 (2008) 6195–6204. [25] I. Noda, T. Kamoto, M. Yamada, Size-controlling synthesis of narrowly distributed particles of methylsilsesquioxane derivatives, Chem. Mater. 12 (2000) 1708–1714. [26] A. Arkhireeva, J.N. Hay, W. Oware, A versatile route to silsesquioxane nanoparticles from organically modified silane precursors, J. Non-Cryst. Solids 351 (2005) 1688–1695. [27] L.D.G.S. Caravajal, G.R. Quinting, G.E. Maciel, Structural characterization of (3-aminopropyl)triethoxysilane-modified silicas by silicon-29 and carbon-13 nuclear magnetic resonance, Anal. Chem. 60 (1988) 1776–1786. [28] A. Arkhireeva, J.N. Hay, Synthesis of sub-200 nm silsesquioxane particles using a modified Stober sol–gel route, J. Mater. Chem. 13 (2003) 3122–3127. [29] T. Matsumoto, Y. Takayama, N. Wada, H. Onoda, K. Kojima, H. Yamada, H. Wakabayashi, Acid-free synthesis of poly-organo-siloxane spherical particles using a W/O emulsion, J. Mater. Chem. 13 (2003) 1764–1770. [30] B.P. Nair, C. Pavithran, Bifunctionalized hybrid silica spheres by hydrolytic cocondensation of 3-aminopropyltriethoxysilane and vinyltriethoxysilane, Langmuir 26 (2010) 730–735. [31] Y.S. Ho, G. McKay, The kinetics of sorption of basic dyes from aqueous solution by sphagnum moss peat, Can. J. Chem. Eng. 76 (1998) 822–827. [32] A. Shahbazi, H. Younesi, A. Badiei, Functionalized SBA-15 mesoporous silica by melamine-based dendrimer amines for adsorptive characteristics of Pb(II), Cu(II) and Cd(II) heavy metal ions in batch and fixed bed column, Chem. Eng. J. 168 (2011) 505–518.

241

[33] Y.A. Tian, P. Yin, R.J. Qu, C.H. Wang, H.G. Zheng, Z.X. Yu, Removal of transition metal ions from aqueous solutions by adsorption using a novel hybrid material silica gel chemically modified by triethylenetetraminomethylenephosphonic acid, Chem. Eng. J. 162 (2010) 573–579. [34] P. Tzvetkova, P. Vassileva, R. Nickolov, Modified silica gel with 5-amino-1,3,4thiadiazole-2-thiol for heavy metal ions removal, J. Porous Mater. 17 (2010) 459–463. [35] N.M. El-Ashgar, I.M. El-Nahhal, M.M. Chehimi, F. Babonneau, J. Livage, Preparation of ethylenediaminetriacetic acid silica-gel immobilised ligand system and its application for trace metal analysis in aqueous samples, Int. J. Environ. Anal. Chem. 89 (2009) 1057–1069. [36] A.S. Pereira, G. Ferreira, L. Caetano, R.S.D. Castro, A. dos Santos, P.M. Padilha, G.R. Castro, 4-Amine-2-mercaptopyrimidine modified silica gel applied in Cd(II) and Pb(II) extraction from an aqueous medium, Pol. J. Chem. Technol. 12 (2010) 7–11. [37] B. Kannamba, K.L. Reddy, B.V. AppaRao, Removal of Cu(II) from aqueous solutions using chemically modified chitosan, J. Hazard. Mater. 175 (2010) 939–948. [38] P.K. Neghlani, M. Rafizadeh, F.A. Taromi, Preparation of aminatedpolyacrylonitrile nanofiber membranes for the adsorption of metal ions: comparison with microfibers, J. Hazard. Mater. 186 (2011) 182–189. [39] F.Q. Xu, N.N. Zhang, Y. Long, Y.M. Si, Y. Liu, X. Mi, X.D. Wang, F.B. Xing, X.D. You, J.P. Gao, Porous CS monoliths and their adsorption ability for heavy metal ions, J. Hazard. Mater. 188 (2011) 148–155. [40] L.Y. Wang, L.Q. Yang, Y.F. Li, Y. Zhang, X.J. Ma, Z.F. Ye, Study on adsorption mechanism of Pb(II) and Cu(II) in aqueous solution using PS-EDTA resin, Chem. Eng. J. 163 (2010) 364–372. [41] Q. He, Z. Hu, Y. Jiang, X.J. Chang, Z.F. Tu, L.N. Zhang, Preconcentration of Cu(II), Fe(III) and Pb(II) with 2-((2-aminoethylamino)methyl)phenol-functionalized activated carbon followed by ICP-OES determination, J. Hazard. Mater. 175 (2010) 710–714. [42] T. Phuengprasop, J. Sittiwong, F. Unob, Removal of heavy metal ions by iron oxide coated sewage sludge, J. Hazard. Mater. 186 (2011) 502–507. [43] A.E. Ofomaja, E.B. Naidoo, S.J. Modise, Biosorption of copper(II) and lead(II) onto potassium hydroxide treated pine cone powder, J. Environ. Manage. 91 (2010) 1674–1685. [44] I.M. El Nahhal, M.M. Chehimi, C. Cordier, G. Dodin, XPS, NMR and FTIR structural characterization of polysiloxane-immobilized amine ligand systems, J. NonCryst. Solids 275 (2000) 142–146. [45] X.F. Liang, Y.M. Xu, G.H. Sun, L. Wang, Y. Sun, X. Qin, Preparation, characterization of thiol-functionalized silica and application for sorption of Pb2+ and Cd2+ , Colloids Surf. A 349 (2009) 61–68.

Journal of Hazardous Materials 196 (2011) 242–247

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Decontamination of waters polluted with simazine by sorption on mesoporous metal oxides Veria Addorisio a , Domenico Pirozzi b , Serena Esposito c , Filomena Sannino a,∗ a Dipartimento di Scienze del Suolo, della Pianta, dell’Ambiente e delle Produzioni Animali, Facoltà di Scienze Biotecnologiche, Università di Napoli “Federico II”, Via Università 100, 80055 Portici, Napoli, Italy b Dipartimento di Ingegneria Chimica, Facoltà di Ingegneria, Università degli Studi di Napoli “Federico II”, P.le Tecchio, 80, 80125 Napoli, Italy c Laboratorio Materiali del Dipartimento di Meccanica, Strutture, Ambiente e Territorio, Facoltà di Ingegneria dell’Università di Cassino, Via G. di Biasio 43, I-03043, Cassino (Fr), Italy

a r t i c l e

i n f o

Article history: Received 19 April 2011 Received in revised form 6 September 2011 Accepted 6 September 2011 Available online 12 September 2011 Keywords: Decontamination Sorption Simazine Mesoporous metal oxides Regeneration

a b s t r a c t Two mesoporous metal oxides, Al2 O3 and Fe2 O3 , were evaluated as regards their ability to remove simazine, a highly persistent herbicide of s-triazines, using a batch equilibrium method. The effect of several experimental parameters such as pH, contact time, initial concentration and sorbent dosage on the sorption of the herbicide was investigated. The maximum sorption of simazine on Al2 O3 and Fe2 O3 was observed at pH 6.5 and 3.5, respectively. The different sorption capacities of the two oxides were explained considering a set of factors affecting the sorption process such as the surface area and the porosity. The kinetics of sorption on both oxides was described using a pseudo second-order model. The sorption of simazine on Fe2 O3 was faster in comparison to that observed on Al2 O3 . It was shown that aluminum oxide can be regenerated by incineration, and consequently can be considered for industrial treatment systems designed to mitigate the pesticide pollution in the aquatic environments. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Water pollution by pesticides has been recognized in agricultural areas of the world for many years, and considerable evidences suggest that many water resources are contaminated by organic pesticides. Common agricultural practices, accidental spillage or uncontrolled release of contaminated waters due to washing of pesticide containers or industrial effluents in the environment have resulted in the contamination of air, soils, surface and ground waters, as well as of living organisms. Thus, in order to protect the environment and the human health, it is important to develop new remediation technologies. Currently, sorption is believed to be a simple and effective technique for water and wastewater treatment and its success largely depends on the development of efficient sorbents. Activated carbon [1], clay minerals [2], biomaterials [3], zeolites [4], and some industrial solid wastes [5] have been widely used with varying efficiency. In a wastewater treatment process that utilizes sorption,

∗ Corresponding author at: Dipartimento di Scienze del Suolo, della Pianta, dell’Ambiente e delle Produzioni Animali, Università degli Studi di Napoli Federico II, Via Università 100, 80055 Portici (NA), Italy. Tel.: +39 081 2539187/2539183; fax: +39 081 2539186. E-mail address: [email protected] (F. Sannino). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.022

the regeneration of the sorbent is crucially important. However, the high costs associated with the regeneration of the sorbents or the necessity of an extraction achieved by acid or alkali solutions represent a serious problem. New sorbents are required to remove organic pollutants in water decontamination processes. An ideal sorbent should have high surface area (i.e. high density of sorption sites), uniformly accessible pores and physical and/or chemical stability [6]. It is believed that the sorption capacity of a sorbent is largely determined by its surface area, which usually increases with decreasing the particle size, although the pore size distribution is also decisive for an optimal sorption process. Therefore, thanks to the introduction of nano-structured oxide materials, the pollutant removal efficiency can be increased dramatically. Mesoporous materials, a class of nanoporous materials, have attracted a lot of attention in both the scientific and the industrial communities since the introduction of well-ordered mesoporous silicas which possess large surface areas and uniform and tunable pore sizes (2–50 nm) [7,8]. The great interest of these materials as adsorbents for environmental remediation is due not only to their high surface area but also to their fast contaminant sorption kinetics. Recent works [9–11] have shown that mesoporous materials can have large adsorption capacity, good selectivity and improved recoverability for the removal of toxic compounds from aqueous solutions.

V. Addorisio et al. / Journal of Hazardous Materials 196 (2011) 242–247

The encouraging results obtained from these studies prompted us to investigate the sorption of simazine (2-chloro-4,6bis(ethylamino)-s-triazine), a basic herbicide belonging to striazine family, on the mesoporous metal oxides. The s-triazines are selective persistent herbicides, widely investigated due to their still large application in forestry and pre- and post-emergence in agricultural soils [12]. Even though these herbicides are now forbidden in some countries, the recalcitrance of s-triazines against chemical and biological degradation has led to their accumulation in the environment [12]. In Italy, the annex number 5 included in the Legislative Declaration 152/2006 on the environment, states the safe limit of atrazine (s-triazine herbicide) in soil that varies from 0.01 to 1.0 mg kg−1 , whereas the limit in waters is 0.3 ␮g L−1 . Simazine is a synthetic s-triazine herbicide used for pre-emergence control of broad-leaf weeds and annual grasses in agricultural and non-crop fields [13,14]. It is the second most commonly detected pesticide in surface and groundwater in the United States, Australia and Europe [15]. Due to the carcinogenic potential of s-triazines, simazine presence in water is of increasing concern [16]. A significant research on the removal of s-triazines by sorption on soils and different organic and inorganic sorbents has been performed [17,18]. Nevertheless, as far as we know, no papers have been published on the sorption capacity of mesoporous oxides towards triazine. Therefore, the objective of this work was to evaluate two commercial metal oxides with mesoporous structure (Al2 O3 and Fe2 O3 ) as regards their ability to remove simazine from aqueous solutions. In view of future applications, the regeneration of these materials is also discussed.

243

2.3. Sorption experiments A stock solution of herbicide was prepared by dissolving 2 mg of simazine in 500 mL of KCl 0.03 M (final concentration 20 ␮mol L−1 ). This solution was then kept refrigerated. All sorption experiments were carried out by adding 10 mg of sorbent to 20 mL of simazine solution in glass vials with Teflon caps at a temperature of 20 ◦ C. The samples, after incubation for 24 h in a rotatory shaker (35 rpm), were centrifuged at 7000 rpm for 20 min. The supernatant was analyzed to evaluate the herbicide concentration using high pressure liquid chromatography (HPLC) technique as described below. The amount of simazine sorbed on the oxides was calculated as the difference between the initial quantity of herbicide added and that present at the equilibrium. Blanks of simazine in KCl 0.03 M were analyzed in order to check the pesticide stability and the sorption to vials. Several experiments were carried out to study the effect of different factors affecting the sorption of simazine on Al2 O3 and Fe2 O3 , as summarized below:

2-Chloro-4,6-bis(ethylamino)-1,3,5-triazine (simazine) (Fig. S1 of Supporting Information) was purchased from Sigma–Aldrich Chemical Company (Poole, Dorset, UK; 99.0% purity). All solvents were of HPLC grade (Carlo Erba, Milan, Italy) and were used without further purification. The water used in the preparation of all solutions was obtained from a Millipore Waters Milli-Q water purification system. All other chemicals were obtained from Sigma–Aldrich unless otherwise specified. ␥-Aluminum (Al2 O3 ) and iron(III) (Fe2 O3 ) oxides were purchased from IoliTec Nanomaterials (Denzlingen, Germany; 99.9 and 99.5% purity for Al2 O3 and Fe2 O3 , respectively).

(a) Effect of pH: Experiments were carried out by adding pesticide solutions at fixed concentration (10 ␮mol L−1 ) and different pH values from 3.0 to 7.0. The pH was controlled by the addition of a 0.10 or 0.01 mmol L−1 solution of HCl or KOH. The samples were shaken for 24 h and subsequently, after centrifugation, analyzed as described below. (b) Effect of sorbent amount: The experiments were carried out by adding pesticide solutions at two concentrations (5 and 10 ␮mol L−1 ), at different solid/liquid ratios. Ratios of 0.1, 0.5, 1.0 and 2.0 were obtained by adding 2.0, 10, 20 and 40 mg, respectively, of Al2 O3 or Fe2 O3 to a final volume of 20 mL, at 20 ◦ C. The samples were incubated at pH values of 6.5 (tests with Al2 O3 ) and 3.5 (tests with Fe2 O3 ), for 24 h. (c) Effect of incubation time: Kinetic studies were performed using 10 ␮mol L−1 solutions of simazine at pH 6.5 (tests with Al2 O3 ) and pH 3.5 (tests with Fe2 O3 ). The solutions were stirred for 2.0, 5.0, 20, 40, 60, 90, 120, 320, 960 and 1800 min. (d) Sorption isotherm: Different volumes of a stock solution of herbicide (20 ␮mol L−1 ) were added to each oxide to give an initial simazine concentration ranging from 0.50 to 10.69 ␮mol L−1 . The pH of each solution was kept constant at pH 6.5 (tests with Al2 O3 ) and 3.5 (tests with Fe2 O3 ), by the addition of 0.10 or 0.01 mmol L−1 solutions of HCl or KOH. The samples were incubated for 20 min (tests with Al2 O3 ) and 180 min (tests with Fe2 O3 ); then, after centrifugation, the supernatants were analyzed as described below.

2.2. Chemical and physical analysis of Al2 O3 and Fe2 O3

2.4. Analytical determination

The determination of point of zero charge (pzc) of Al2 O3 and Fe2 O3 was performed according to the methods described by Addorisio et al. [11]. The specific surface area (SSA) of Al2 O3 and Fe2 O3 was calculated by the Brunauer–Emmett–Teller (BET) method [19]. N2 adsorption–desorption isotherms at 77 K were obtained by a Micromeritics Gemini II 2370 apparatus. Before each measurement the sample was degassed at 250 ◦ C for 2 h under N2 flow. Pore volumes were determined from the amounts of adsorbed N2 at P/P◦ = 0.98 (desorption curve), assuming the presence of liquid N2 (density = 0.807 g cm−3 ) in the pores under these conditions. The average values of the pore diameter dp were calculated from the relation: dp = 4V/ABET , where V is total pore volume. The Barrett–Joyner–Halenda (BJH) approach [19] was used to calculate the pore size distribution of the sample using the desorption data.

Simazine was analyzed with an Agilent 1200 Series HPLC apparatus (Wilmington, U.S.A.), equipped with a DAD array and a ChemStation Agilent Software. The procedure of analysis is described in detail in Supporting Information.

2. Materials and methods 2.1. Materials

2.5. Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) analysis The procedure of sample preparation for DRIFTS determinations is reported in detail in Supporting Information. 2.6. Scanning Electron Microscopy (SEM) analysis The SEM analysis of Al2 O3 samples at pH 4.0 and 6.5 was carried out by a FEI Quanta 200 ESEM.

244

V. Addorisio et al. / Journal of Hazardous Materials 196 (2011) 242–247

-1

Simazine sorbed (µmol kg )

Al2O3

Fe2O3

6000 5000 4000 3000 2000 1000 0

3.0

4.0

5.0

pH

6.0

7.0

8.0

Fig. 1. Effect of pH on the sorption of simazine by Al2 O3 and Fe2 O3 at solid/liquid ratio of 0.5.

2.7. Analysis of the data All the experiments were performed in triplicate and the relative standard deviation was in all cases lower than 3%. 3. Results and discussion 3.1. Effect of pH In order to evaluate the optimum pH to be used in the subsequent experiments, sorption tests were carried out to study the effect of pH at a 0.5 solid/liquid ratio, preliminarily found as the optimal value for sorption. The results reported in Fig. 1 show that the greatest sorbed amount of simazine was observed at pH 6.5 when using Al2 O3 , and at pH 3.5 in the tests with Fe2 O3 . In aqueous solution, triazines such as simazine exist in either neutral or protonated form, depending on the pKa of the compound (the pKa of simazine is 1.70) and on the pH of the system. The ring nitrogen atom, located in the 3-position between the electron-rich alkyl-side chains, is the most basic and hence the most likely site of protonation. At low pH values (e.g. 3.0–3.5) the surfaces of Feoxide and of soluble species are strongly protonated [20], so that the most basic triazinic nitrogen (N-3) could easily form coordination bonds with Fe2 O3 deriving from the overlap of nitrogen lone pair electrons and partially filled metal d-orbitals on iron, being this latter a transition metal (d-block element). The adsorption of simazine onto Al2 O3 as a function of pH is quite different. A possible explanation of our findings is that the key role in the sorption of the herbicide is played by the aggregation state of the oxide, which is greatly influenced by the pH of the medium. In particular, as reported in Fig. 2, large aggregates and small particles were observed at pH 4.0 and pH 6.5, respectively. Consequently, the textural properties of Al2 O3 are expected to be modified by the pH. To support the latter hypothesis, a physical analysis on Al2 O3 sample at pH 4.0 and pH 6.5 was performed through the analysis of the relative N2 adsorption–desorption isotherms. As shown in Table 1, the surface area of Al2 O3 sample at pH 4.0 (157 m2 g−1 ) results comparable to that at pH 6.5 (150 m2 g−1 ). Table 1 Physical properties of Al2 O3 at pH 4.0 and pH 6.5. Sample

ABET (m2 g−1 )

Pore volume (cm3 g−1 )

Average dp (nm)

Al2 O3 (pH 4.0) Al2 O3 (pH 6.5)

157 150

0.352 0.643

8.9 17.3

Fig. 2. SEM image of Al2 O3 at 500× magnification at pH 4.0 (a) and at pH 6.5 (b), respectively.

However, more interesting indications can be obtained from a comparison between the pore size distributions (Table 1), obtained by the elaboration of the desorption data by the BJH method. Clearly, the contact with solutions at different pH strongly affects the intrinsic organization of Al2 O3 particles, generating a porosity made of smaller cavities in the case of Al2 O3 sample at pH 4.0. Moreover, the pore volume of the samples at pH 6.5 is much greater than that observed at pH 4.0 (Table 1). These observations are confirmed by the SEM analysis (Fig. 2). Finally, at pH 6.5, the herbicide could give an acid–base reaction with Al2 O3 , that at this pH is present as Al[(H2 O)6 ]3+ . Alternatively, being simazine more nucleophilic than water molecules, it can replace some water molecules in the hexacoordinate complex. Chappell et al. [21] showed the interaction of atrazine with smectite surfaces through hydrogen bonding, and at the same time demonstrated that alkyl tails of the herbicide may interact with hydrophobic nanosites on the smectite basal surfaces. Other studies demonstrated that noncovalent binding forces, cation-␲, may occur between s-triazine and metallic cation [22]. In the literature, the information about the binding mechanism of triazine herbicides on the oxides is very scarce. Consequently, the explanation of the observed behaviour reported above deserves a close attention.

V. Addorisio et al. / Journal of Hazardous Materials 196 (2011) 242–247

245

-1

Simazine sorbed (µmol kg )

Al O 2

1.2 104 1 10

Fe O

3

2

3

4

8000 6000 4000 2000 0

0

1

2

3

4

5

6

7

8

Equilibrium concentration of Simazine ( µmol L ) -1

Fig. 3. Effect of solid/liquid ratio on the sorption of two different concentrations of simazine by Al2 O3 and Fe2 O3 at pH 3.5 and 6.5, respectively.

Alternatively, the sorption kinetic curves were analyzed adopting the pseudo second-order kinetic model:

3.2. Effect of solid/liquid ratio Sorption studies of simazine were carried out using Al2 O3 (at pH 6.5) and Fe2 O3 (at pH 3.5), varying the amount of sorbent and adding two different concentrations of herbicide. The results reported in Fig. 3 show, for both the oxides and regardless of the herbicide concentration, a higher sorption at solid/liquid ratio 0.5. In particular, the amount of sorbed herbicide on Al2 O3 was already significant at the lowest solid/liquid ratio (0.1) and was greatly increased by increasing the amount of oxide. However, in the presence of 20 and 40 mg of oxide, no sorption of simazine was observed. Evidently, the greater the amount of oxide, the greater the resistance to the diffusion in the mesoporous structure, which results in a lower sorption of the herbicide.

3.3. Effect of incubation time The kinetic data were analyzed adopting a pseudo first-order kinetic equation [23]: dq = k1 · (qe − q) dt

(E1)

where qe and q are the amounts of herbicide sorbed (␮mol kg−1 ) at equilibrium and at time t, respectively, k1 is the rate constant of sorption (min−1 ) and t is the time (min). Integrating with the boundary condition q|t=0 = 0 Eq. (E1), the following expression is obtained: log(qe − q) = log qe −

Ka t 2.303

0.20

(E2)

Al O 2

3

Fe O 2

3

-1

t/q (min kg µmol )

Fig. 5. Sorption isotherm of simazine by Al2 O3 and Fe2 O3 .

0.15

dq = k2 · (qe − q)2 dt

(E3)

where k2 is the rate constant of sorption (kg ␮mol−1 min−1 ). Upon integration with the boundary condition q|t=0 = 0, Eq. (E3) yields the following expression: t 1 t − = q qe k2 · q2e

(E4)

The best model to describe the sorption kinetics data was the pseudo second-order model (i.e. Eq. (E3)), as shown by the linear behaviour of the (t/q) versus time plot (Fig. 4). The corresponding model parameters (qe and k2 ) were estimated with reference to the simazine sorption on Al2 O3 (qe = 6098 ␮mol kg−1 , k2 = 2.85 × 10−5 kg ␮mol−1 min−1 , r2 = 0.99), and on Fe2 O3 (qe = 1695 ␮mol kg−1 , k2 = 9.03 × 10−4 kg ␮mol−1 min−1 , r2 = 0.99). The sorption on Fe2 O3 , reaching the equilibrium after 5 min, was faster in comparison to that pertaining Al2 O3 , showing an equilibrium time of 120 min. Therefore, all the equilibrium determinations were carried out adopting an incubation period of 20 min for Fe2 O3 and 180 min for Al2 O3 . 3.4. Sorption isotherm The sorption isotherms of simazine on Al2 O3 and Fe2 O3 are displayed in Fig. 5. The obtained data were analyzed according to the Freundlich equation: x = Kc 1/N

(E5)

where x is the amount of pesticide sorbed (␮mol kg−1 ), c is the equilibrium concentration of pesticide (␮mol L−1 ), K [(␮mol kg−1 )/(␮mol L−1 )1/N ] and N (dimensionless) are constants that give estimates of the sorptive capacity and intensity, respectively, according to Giles et al. [24]. The sorption isotherms of simazine on Al2 O3 and Fe2 O3 , shown in Fig. 5, were well-fitted by the linearized form of Freundlich equation (r2 > 0.99) (Table 2). According to the classification of Giles et al.

0.10 Table 2 Freundlich parameters for the sorption of simazine on Al2 O3 and Fe2 O3 .

0.05

Freundlich parameters

0

0

50

100

150

Time (min) Fig. 4. Effect of time on the sorption of simazine by Al2 O3 and Fe2 O3 .

200

Al2 O3 Fe2 O3 a

K (␮mol kg−1 )/(␮mol L−1 )1/N

N (dimensionless)

r2 a

168.11 156

0.44 0.56

0.99 0.99

Correlation coefficient.

246

V. Addorisio et al. / Journal of Hazardous Materials 196 (2011) 242–247

Table 3 Comparison of surface area, pore volume and average pore diameter of Al2 O3 and Fe2 O3 samples. Sample Al2 O3 Al2 O3 500 Fe2 O3 Fe2 O3 500

ABET (m2 g−1 ) 195 200 106 33

Pore volume (cm3 g−1 )

Average dp (nm)

0.723 0.770 0.239 0.0650

14.8 14.7 9.2 7.9

[24], the experimental sorption isotherms were of S-type for Al2 O3 and C-type for Fe2 O3 . In particular, at low equilibrium concentrations the sorbed amount of simazine on Fe2 O3 was similar to that detected on Al2 O3 , whereas at concentrations greater than 3 ␮mol L−1 there was a marked difference in the behaviour of the two mesoporous oxides. As a matter of facts, at a 6.0 ␮mol L−1 equilibrium concentration of simazine, the amount of herbicide sorbed on Fe2 O3 was 4000 ␮mol kg−1 , whereas that sorbed on Al2 O3 was ∼8000 ␮mol kg−1 . The S-type isotherm of simazine on Al2 O3 indicates that the presence of molecules of herbicide already sorbed on the surface favours the sorption process by a cooperative effect. This effect can be explained assuming that the molecules already sorbed modify the affinity of the sorption sites towards the molecules present in solution. On the contrary, the C-type isotherm of simazine on Fe2 O3 was characterized by a straight line trend, indicative of a constant partition of the herbicide between solution and sorbent until reaching saturation. The Freundlich constants (K and N) (Table 2) showed that Al2 O3 sorbed the herbicide with a higher sorptive capacity and a lower affinity in comparison to Fe2 O3 . The presence of secondary small pores at the boundary of micropores region in Al2 O3 may affect positively the sorption of small organic molecules such as simazine (0.784 nm), as it is possible that the sorption energy increases in those pores whose dimensions approach the herbicide dimensions (0.7–0.9 nm). In fact, as reported in Fig. S2 of Supporting Information, the pore size distribution of Al2 O3 appears to be bimodal, characterized by two maxima at about 3 nm and at about 15 nm. On the contrary, Fe2 O3 shows a unimodal distribution and most of the N2 volume is adsorbed in the pore size range 6–10 nm [11]. A combination of factors such as the surface area and the porosity concurs significantly to influence the highest sorption capacity of Al2 O3 than Fe2 O3 (see Table 3). Finally, DRIFT analyses were carried out on each metal oxide after the sorption of the herbicide. The DRIFT spectra of Al2 O3 and Fe2 O3 -simazine complexes were recorded and compared with those of simazine and untreated oxides (Fig. S3 of Supporting Information). In particular, in Fig. S3a, the characteristic sorption bands of simazine corresponding to NH stretching (3260 cm−1 ) and C N stretching (1637, 1565 and 1406 cm−1 ) were observed [27]. Fig. S3b and c shows that, after the sorption of the herbicide, the sorption band at 3440 cm−1 , corresponding to –OH stretching of each oxide, was reduced more or less strongly due to a possible coordination interactions simazine-Fe2 O3 and acid–base reactions or replacement of the herbicide with water molecules in the acid hexacoordinate complex [Al(H2 O)6 ]3+ , respectively. In a sorption–desorption study of atrazine and simazine by model soil colloidal components, Celis et al. [25] demonstrated that ferrihydrite does not adsorb triazine herbicides. Enhanced sorption of these herbicides on montmorillonite was measured after increasing the surface acidity of the clay. On the contrary, a carbonrich product (biochar) generated from biomass through pyrolysis sorbed an amount of simazine of ∼2480 ␮mol kg−1 [26].

3.5. Regeneration of Al2 O3 and Fe2 O3 In a wastewater treatment involving a sorption process, the regeneration of the sorbent is crucially important. Nowadays, in many applications, the reuse of the sorbent through regeneration of its sorption properties is an economic necessity. Desorption agents (e.g. sodium hydroxide solution) are commonly used to recover sorbents such as Fe- and Al-based supports [28]. However, the utilization of a desorption agent has some disadvantages because it increases the operating cost, and the waste solution containing NaOH discarded from the regeneration of the sorbent causes environmental pollution. The incineration method could be considered as an alternative way for the regeneration of the sorbents, as it avoids the use of hazardous desorption agents. To assess the feasibility of this option, Al2 O3 and Fe2 O3 were annealed at 500 ◦ C for 1 h; subsequently, to ascertain whether the textural properties were retained, a physical characterization on heat treated oxides was performed through the analysis of the relative N2 adsorption–desorption isotherms. The porosities of Al2 O3 and Fe2 O3 have been previously analyzed by the authors [11]; herein we report a comparison with the physical properties of the heat treated samples. The notation used for the samples is referred to the chemical formula followed by a number indicating the temperature of the heat treatment, i.e. Al2 O3 500 and Fe2 O3 500; the samples before the heat treatment are simply denoted with the chemical formula, as in the text. The perfect correspondence between the isotherms obtained with Al2 O3 500 and Al2 O3 [11], indicated that the mesoporous structure was not damaged by the annealing. The adsorption isotherms were elaborated using the BET method, to obtain the corresponding surface areas reported in Table 3 together with the total pore volume and the estimated average pore diameter. As clearly shown by the data in Table 3, all the textural properties of Al2 O3 are well preserved after the heat treatment. These results drive us to consider the incineration method as an effective option for the regeneration of aluminum oxide. On the contrary, the thermal stability of Fe2 O3 was not comparable to that of Al2 O3 as regards the pore structure. As a matter of facts, the heat treatment strongly altered the textural properties of the sample and a drastic collapse of the surface area was observed (Table 3). To observe to which extent the pore size distribution of iron oxide was modified by the annealing procedure, the desorption data were elaborated by the BJH method. The comparison between the pore size distributions of Fe2 O3 and Fe2 O3 500 samples (Fig. S4 of Supporting Information) indicates beyond doubt that the mesoporous structure was completely destroyed by the heat treatment at 500 ◦ C, making the iron oxide not recoverable by incineration.

4. Conclusions In this study, two metal oxides with mesoporous structure, Al2 O3 and Fe2 O3 , showed different capacities to adsorb simazine, a highly persistent herbicide. In particular, the optimum pH for sorption was found to be 6.5 for Al2 O3 and 3.5 for Fe2 O3 . The different sorption capacities of the two oxides were explained by considering a set of factors significantly concurring to influence the sorption process, such as the surface area and the porosity. The kinetics of sorption was described by a pseudo secondorder model, demonstrating that Fe2 O3 adsorbs simazine faster than Al2 O3 . Finally, we demonstrated that Al2 O3 can be regenerated by incineration and could be considered for industrial treatment systems, to remove effectively simazine from the aquatic environments and eventually to mitigate the pesticide pollution.

V. Addorisio et al. / Journal of Hazardous Materials 196 (2011) 242–247

Supplementary data The chemical formula of simazine and its analytical determination, the Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) analysis, the pore size distribution of Al2 O3 and Fe2 O3 , and the pore size distribution of Fe2 O3 at 500 ◦ C are reported. Acknowledgement This manuscript is contribution DiSSPAPA 248. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.09.022. References [1] S. Baup, C. Jaffre, D. Wolbert, A. Laplanche, Adsorption of pesticides onto granular activated carbon: determination of surface diffusivities using simple batch experiments, Adsorption 6 (2000) 219–228. [2] F. Bruna, I. Pavlovic, C. Barriga, J. Cornejo, M.A. Ulibarri, Adsorption of pesticides carbetamide and amitron on organohydrotalcite, Appl. Clay Sci. 33 (2006) 116–124. [3] G. Crini, Recent developments in polysaccharide-based materials used as adsorbents in wastewater treatment, Prog. Polym. Sci. 30 (2005) 38–70. [4] S. Wang, Y. Peng, Natural zeolites as effective adsorbents in water and wastewater treatment, Chem. Eng. J. 156 (2010) 11–24. [5] N. Ratola, C. Botelho, A. Alves, The use of pine bark as a natural adsorbent for persistent organic pollutants—study of lindane and heptachlor adsorption, J. Chem. Technol. Biotechnol. 78 (2003) 347–351. [6] H. Yoshitake, T. Yokoi, T. Tatsumi, Adsorption of chromate and arsenate by amino functionalized MCM-41 and SBA-1, Chem. Mater. 14 (2002) 4603–4610. [7] C. Lee (Ed.), Adsorption Science and Technology, World Scientific, Singapore, 2003, pp. 605–609. [8] Y. Kim, C. Kim, I. Choi, S. Rengaraj, J. Yi, Arsenic removal using mesoporous alumina prepared via a templating method, Environ. Sci. Technol. 38 (2004) 924–931. [9] M. Anbia, N. Mohammadi, K. Mohammadi, Fast and efficient mesoporous adsorbents for the separation of toxic compounds from aqueous media, J. Hazard. Mater. 176 (2010) 965–972. [10] P. Wang, I.M.C. Lo, Synthesis of mesoporous magnetic ␥-Fe2 O3 and its application to Cr(VI) removal from contaminated water, Water Res. 43 (2009) 3727–3734.

247

[11] V. Addorisio, S. Esposito, F. Sannino, Sorption capacity of mesoporous metal oxides for the removal of MCPA from polluted waters, J. Agric. Food Chem. 58 (2010) 5011–5016. [12] G. Celano, D. Smejkalova, R. Spaccini, A. Piccolo, Interactions of three striazines with humic acids of different structure, J. Agric. Food Chem. 56 (2008) 7360–7366. [13] A. Garcia-Valcarcel, J. Tadeo, Influence of soil moisture on sorption and degradation of hexazinone and simazine in soil, J. Agric. Food Chem. 47 (1999) 3895–3900. [14] A.S. Gunasekara, J. Troiano, K.S. Goh, R.S. Tjeerdema, Chemistry and fate of simazine, Rev. Environ. Contam. Toxicol. 189 (2007) 1–23. [15] J. Troiano, D. Weaver, J. Marade, F. Spurlock, M. Pepple, C. Nordmark, D. Bartkowiak, Summary of well water sampling in California to detect pesticide residues resulting from nonpoint-source applications, J. Environ. Qual. 30 (2001) 448–459. [16] T. Hayes, P. Case, S. Chui, D. Chung, C. Haeffele, K. Haston, Pesticide mixtures, endocrine disruption, and amphibian declines: are we underestimating the impact, Environ. Health Perspect. 1 (2006) 40–50. [17] C. Flores, V. Morgante, M. Gonzalez, R. Navia, M. Seeger, Adsorption studies of the herbicide simazine in agricultural soils of the Aconcagua valley, central Chile, Chemosphere 74 (2009) 1544–1549. [18] M. Lucio, P. Schmitt-Kopplin, Modelling the binding of triazine herbicides to humic substances using capillary electrophoresis, Environ. Chem. Lett. 4 (2006) 15–21. [19] F. Rouquerol, J. Rouquerol, K. Sing, Adsorption by Powders and Porous Solids: Principles Methodology and Application, Academic Press, London, 1999. [20] D.L. Sparks, Environmental Soil Chemistry, Academic Press, London, 1995. [21] M. Chappell, D.A. Laird, M.L. Thompson, H. Li, B.J. Teppen, V. Aggarwal, C.T. Johnston, S.A. Boyd, Influence of smectite hydration and swelling on atrazine sorption behaviour, Environ. Sci. Technol. 39 (2005) 3150–3156. [22] C. Garau, D. Quinonero, A. Frontiera, P. Ballester, A. Costa, P.M. Deya, Dual binding mode of s-triazine to anions and cations, Org. Lett. 5 (2003) 2227–2229. [23] M. Ozacar, I.A. Sengil, A two-stage batch adsorber design for methylene blue removal to minimize contact time, J. Environ. Manage. 80 (2006) 372–379. [24] C.H. Giles, D. Smith, A. Huitson, A general treatment and classification of the solute adsorption isotherms. I. Theoret., J. Colloid Interface Sci. 47 (1974) 755–765. [25] R. Celis, J. Cornejo, M.C. Hermosin, W.C. Koskinen, Sorption–desorption of atrazine and simazine by model soil colloidal components, Soil Sci. Soc. Am. J. 61 (1997) 436–443. [26] W. Zheng, M. Guo, T. Chow, D.N. Bennett, N. Rajagoplan, Sorption properties of greenwaste biochar for two triazine pesticides, J. Hazard. Mater. 181 (2010) 121–126. [27] L.J. Bellamy, The Infrared Spectra of Complex Molecules, vol. I, 3rd ed., Chapman and Hall, New York, 1975. [28] Y. Kim, B. Lee, J. Yi, Preparation of functionalized mesostructured silica containing magnetite (MSM) for the removal of copper ions in aqueous solutions and its magnetic separation, Sep. Sci. Technol. 38 (2003) 2533–2548.

Journal of Hazardous Materials 196 (2011) 287–294

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Valorisation of electric arc furnace steel slag as raw material for low energy belite cements R.I. Iacobescu a , D. Koumpouri b , Y. Pontikes c , R. Saban a , G.N. Angelopoulos b,∗ a b c

Department of Materials Science and Engineering, Politehnica University of Bucharest, Splaiul Independentei 313, 060032 Bucharest, Romania Laboratory of Materials and Metallurgy, Department of Chemical Engineering, University of Patras, 26500 Rio, Greece Department of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven, Kasteelpark Arenberg 44 bus 2450, B-3001 Heverlee (Leuven), Belgium

a r t i c l e

i n f o

Article history: Received 5 July 2011 Received in revised form 7 September 2011 Accepted 7 September 2011 Available online 12 September 2011 Keywords: Electric arc furnace slag Belite cement

a b s t r a c t In this paper, the valorisation of electric arc furnace steel slag (EAFS) in the production of low energy belite cements is studied. Three types of clinkers were prepared with 0 wt.% (BC), 5 wt.% (BC5) and 10 wt.% (BC10) EAFS, respectively. The design of the raw mixes was based on the compositional indices lime saturation factor (LSF), alumina ratio (AR) and silica ratio (SR). The clinkering temperature was studied for the range 1280–1400 ◦ C; firing was performed at 1380 ◦ C based on the results regarding free lime and the evolution of microstructure. In order to activate the belite, clinkers were cooled fast by blown air and concurrent crushing. The results demonstrate that the microstructure of the produced clinkers is dominated by belite and alite crystals, with tricalcium aluminate and tetracalcium-alumino-ferrite present as micro-crystalline interstitial phases. The prepared cements presented low early strength development as expected for belite-rich compositions; however the 28-day results were 47.5 MPa, 46.6 MPa and 42.8 MPa for BC, BC5 and BC10, respectively. These values are comparable with OPC CEMI 32.5 N (32.5–52.5 MPa) according to EN 197-1. A fast setting behaviour was also observed, particularly in the case of BC10, whereas soundness did not exceed 1 mm. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Recent years have seen the cement industry growing dynamically with most of the activity taking place in emerging economies. Despite financial turbulence, population growth and the resulting need for housing along with state investments in infrastructure are strong drivers to offset the downturn in cement markets. Globally, cement production increased from 2.568 Mt in 2006 to 3.294 Mt in 2010 [1]. Unavoidably, as with any industrial activity, cement production has its own environmental footprint. Estimations suggest that cement production is responsible for 5–7% of the worldwide CO2 emission [2,3]. If all the greenhouse gases emitted by anthropogenic activities are considered, the cement manufacturing industry contributes about 3% of the total anthropogenic greenhouse gases emissions [2]. This is predominantly the result of the fuels used to generate the required energy, estimated at 0.37 kg/kg clinker, and of the de-carbonation of limestone (CaCO3 ) which takes place during cement production, estimated at 0.53 kg/kg clinker CO2 [2]. Consequently, reducing the limestone in the raw meal and thus changing its chemistry, could lead to lower CO2 emissions. This potential has resulted in increased scientific interest in innovative

∗ Corresponding author. Tel.: +30 2610969530; fax: +30 2610990917. E-mail address: [email protected] (G.N. Angelopoulos). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.024

types of cement [4], and more specifically, in belite-rich cements over the last 20 years [5,6]. This type of cement, unlike conventional OPC, contains a higher percentage of belite (C2 S) and a lower percentage of alite (C3 S). In order to reach the desirable percentages of C2 S and C3 S, the lime saturation factor (LSF) must be between 78% and 83% [7]. The environmental benefits of the belite type cements over OPC can be summarized as follows: energy saving could rise up to 16% [8], burning temperatures could be reduced by 6–10% and levels of emitted CO2 and NOx could fall [9,10]. However, the early strength of such cements is lower and milling energy might be increased due to the hardness of C2 S. By combining the production of belite cements with alternative raw materials as a substitute for limestone, such as metallurgical slag’s, additional benefits may be obtained [11]. During the production of iron and steel, several types of slag are generated. These include blast furnace (BF), basic oxygen furnace (BOF), electric arc furnace (EAF) and stainless steel (SS-EAF and SS-AOD) slag. Nearly 50 Mt/y of steel slag is produced globally and 12 Mt/y is produced in Europe [12]. About 65% of this is used in qualified fields of applications, mainly construction, while the remainder is stored or used for other small purposes [13]. Around 37% of the steel slag produced in Europe in 2010 was used for cement production [14]. Nowadays more than 40% of global steel production takes place in EAFs [15] and is associated with 20 Mt/y slag generation. Greece has a cement production capacity of approximately 18 Mt/y [16],

288

R.I. Iacobescu et al. / Journal of Hazardous Materials 196 (2011) 287–294

and a steel production capacity of 3.5 Mt/y. The annual EAF slag (EAFS) production varies from 300.000 t/y to 400.000 t/y. Of the total amount of EAFS processed annually, around 55% is used in the production of coarse aggregates for road construction. The main environmental problems associated with the disposal of the wasted EAFS are the “dusting” of the slag and the release of leachates. Thus, quite apart from the needs of the cement industries, steel producers have their own motivations for finding a use for their slag. In principle, there are two methods of incorporating slag in cement production: either in the raw meal or in a later stage, as a (latent) hydraulic or pozzolanic material [17]. Prior studies indicate that an EAFS addition of up to 10 wt.% in the raw meal is effective without any detrimental effect on the technical properties of the resultant cement [17]. Other authors reached similar results, in terms of sintering, microstructure, as well as, hydrating properties of the final clinker with 10.5 wt.% EAFS addition. In addition to the above, other authors suggest that the addition of EAFS in clinker production will reduce the sintering temperature of raw meal and the theoretic heat consumption [18]. Moreover, cement containing steel slag as BOF and/or EAF can also have improved corrosion resistance than conventional Portland cement [19]. Finally, combinations of EAFS, BOF and AOD slags were tested for the production of sulfo-aluminate belite cements with encouraging results [20]. Despite the work in the field, no studies were focused on the use of EAFS for belite-rich cements to the best of our knowledge. In terms of disadvantages, the heavy metals content (Cr, V, etc.) in steel slag is an issue of concern. In the European Union (EU), a directive regarding Cr(VI) came into effect in 2005 and prohibits the use or supply of cements containing more than 2 ppm water-soluble chromium by mass of cement [21]. Typically, Cr(VI) compounds are more water-soluble (although insoluble do exist as well), thus more likely to participate in leaching. A number of adverse health effects have been associated with Cr(VI) exposure, ranging in severity. According to NIOSH [22] all Cr(VI) compounds are considered potential occupational carcinogens. Reducing agents, such as ferrous sulphate, either the monohydrate (FeSO4 ·H2 O) or heptahydrate (FeSO4 ·7H2 O) form, or stannous sulphate (SnSO4 ) are added to control the oxidation state of chromium [23]. The present work explores how EAFS can be exploited as a raw material in the production of low energy belite cements. The clinkers produced were characterised by SEM/EDS and Rietveld QXRD. Water demand, initial setting time, soundness and compressive strength were measured on both cement and cement paste. The hydration behaviour of these cements as well as their leaching potential is addressed in a separate work.

2. Materials and methods The raw materials used in the preparation of the raw meals were limestone, flysch and EAFS. The chemical analysis was performed by X-ray fluorescence spectrometry (XRF, Philips PW 2400). The crystalline phases of the raw materials were identified by Xray diffraction analysis (D5000 Siemens). Qualitative analysis was performed by the DIFFRACplus EVA® software (Bruker-AXS) based on the ICDD Powder Diffraction File. The mineral phases were quantified using a Rietveld-based quantification routine with the TOPAS® software (Bruker-AXS). This routine is based on the calculation of a single mineral-phase pattern and the refinement of the pattern using a non-linear least squares routine [24]. A number of corrections, including adjustments to the instrument’s geometry, background, sample displacement, detector type and mass absorption coefficients of the refined phases, were applied in order to achieve the optimum pattern fitting. Diffraction patterns were

measured in 2 range of 5–70◦ using CuK␣ radiation of 40 kV and 30 mA, with a 0.01◦ step size and step time of 1 deg/min. The design of the raw meals was based on the predictions of Bogue equations. In order to produce high belite cement, the lime saturation factor (LSF) was adjusted between 78% and 83% [7] whereas the alumina (AR) and silica ratios (SR) varied from 1.00% to 1.87% and 1.96% to 3.29%, respectively, similar to those adopted in the production of OPC. The quality indices LSF, AR and SR were calculated according to Eqs. (1)–(3) [5,23]:

LSF =

%CaO 2.8 ∗ %SiO2 + 1.2 ∗ %Al2 O3 + 0.65 ∗ %Fe2 O3

(1)

SR =

%SiO2 %Al2 O3 +%Fe2 O3

(2)

AR =

%Al2 O3 %Fe2 O3

(3)

Based on the above and on the chemical analysis of the raw materials, an MS Excel© worksheet was used in order to derive the syntheses of the raw meals. Three types of clinker were prepared: one as a reference (named BC), a second with the addition of 5 wt.% EALS (BC5) and third with the addition of 10 wt.% EAFS (BC10). The obtained meal contents in limestone/flysch/EAFS were in wt.%: 84.0/16.0/0.0, 80.5/14.5/5.0 and 77.0/13.0/10.0 for BC, BC5 and BC10, respectively. The results of the quality indices are presented in Table 3. BC10 presents a maximum in terms of EAFS addition (10 wt.%), while keeping LSF within the desired limits. The mineralogical phases of the clinkers were calculated also by the Rietveld method, besides the estimations derived from the Bogue equations (Table 4). For the preparation of the clinkers, raw materials were individually milled in a Siebtechnik® planetary mill at a particle size below 90 ␮m. After mixing and homogenizing, pellets of approximately 15–20 mm diameter were formed by hand with a minimum water addition. The pellets were dried for 24 h at 110 ◦ C, followed by calcination at 1000 ◦ C for 4 h. Firing of the clinker was performed in a Nabertherm® type Super Kanthal resistance furnace at 1380 ◦ C. Optimum clinkering temperature was determined by burnability tests at 1280 ◦ C, 1300 ◦ C, 1320 ◦ C, 1350 ◦ C, 1380 ◦ C and 1400 ◦ C, with 40 min of soaking time to determine the free lime content according to ASTM C114-03, as well as by SEM observations which evaluated the quality of the clinker. For the stabilization of ␣ - and ␤- C2 S polymorphic forms, fast cooling was applied by simultaneously applying blown air and crushing by means of a hammer. Clinkers were characterised by QXRD and SEM/EDS microanalysis (Jeol JSM 6300 and LINK PentaFET 6699, Oxford Instruments). Carbon coated samples, fractured, polished as well as etched with 1% Nital, were used. All EDS analyses were undertaken well away from phase boundaries. In the case of belite and alite crystals, spot analyses were performed. In the case of the interstitial phase, due to the micro-crystalline texture of the individual phases finely distributed within the amorphous one, analysis of an approximately 4 ␮m × 4 ␮m area away from alite and belite boundaries was performed. The default standards of LINK ISIS have been used. For the preparation of the cement, clinker was milled by means of the aforementioned planetary mill to finesse in the range of 4000–4100 cm2 /g. After milling, 5 wt.% gypsum with grain size lower than 90 ␮m was added. Specific surface (Blaine method) was measured according to EN 196-6 [25], setting time and soundness according to EN 196-3 [26] and compressive strength according to EN 196-1 [27].

R.I. Iacobescu et al. / Journal of Hazardous Materials 196 (2011) 287–294 Table 1 Chemical composition of the raw materials (wt.%). Oxides

EAFS

Limestone

Flysch

CaO FeOtotal SiO2 Al2 O3 MnO MgO Cr2 O3 P2 O5 TiO2 SO3 Cl BaO Na2 O K2 O V2 O5 LOI

32.50 26.30 18.10 13.30 3.94 2.53 1.38 0.48 0.47 0.44 0.14 0.14 0.13 n.d. 0.06 0.00

48.90 1.00 9.00 1.36 n.d. 0.65 n.d. n.d. n.d. n.d. n.d. n.d. 0.10 0.15 n.d. 38.00

5.55 5.90 58.25 13.75 n.d. 2.86 n.d. n.d. n.d. 0.05 n.d. n.d. 1.10 2.50 n.d. 9.80

Total

99.91

99.16

99.76

LOI, loss on ignitions; n.d., not determined.

3. Results and discussion 3.1. Characterisation of raw materials XRF chemical analyses of the raw materials, limestone, flysch and EAFS are given in Table 1. It is observed that EAFS contains elements such as Cr, P, Ti, S and Ba that are considered as dopants for belite activation. The introduction of such ions into the crystal lattice of C2 S, can stabilise ␣ - and ␤- polymorphs; ␣ -C2 S being more active than ␤-C2 S [7,28]. XRD analyses of the raw materials are shown in Fig. 1 whereas the results for the semi-quantitative mineralogical analysis are presented in Table 2. The main mineralogical phases identified are calcite and quartz for limestone, quartz, illite, dolomite, albite and clinochlore for flysh. EAFS contains significant amounts of larnite (␤-belite), gehlenite, wüstite, magnetite, and brownmillerite. The Rietveld analysis results are 41.0 wt.%, 14.7 wt.%, 12 wt.%, 10.0 wt.% and 9.4 wt.% for the above phases, respectively.

289

Table 2 Mineralogical composition of the raw materials, wt.%, according to Rietveld analysis, normalised. Limestone Calcite Quartz Illite Microcline Muscovite Kaolinite Hematite Clinochlore Cristobalite Total

Flysch 90.5 5.7 1.1 1.1 0.6 0.5 0.2 0.2 0.1

Illite Quartz Kaolinite Dolomite Albite Calcite Microcline Muscovite Clinochlore Hematite

100

EAFS 34.1 28.2 8.5 6.8 6.7 5.9 5.7 2.2 1.3 0.6

Larnite Gehlenite Wüstite Magnetite Brownmillerite Mayenite Merwinite Spinel

100

41.0 14.7 12.0 10.0 9.4 7.2 3.7 2.0

100

3.2. Clinker quality as a function of firing temperature The composition of the produced clinkers, as well as the quality indices are presented in Table 3. In the firing tests, the maximum free lime content was 1.6 wt.% for both BC and BC10 at 1280 ◦ C. For the temperature range 1300–1400 ◦ C, free lime varied from 0.4 wt.% to 0.2 wt.% for all mixtures tested. For firing higher than 1300 ◦ C, therefore, free lime values are well below the commonly defined threshold of 1 wt.% for OPC clinker. In Figs. 2 and 3, backscattered electron images revealing the development of clinker microstructure at different firing temperatures for BC and BC10 are presented. In both cases for temperatures up to 1320 ◦ C, clinker microstructure is poorly developed. An extended interstitial phase is also observed. At temperatures exceeding 1350 ◦ C, the dissolution and transport phenomena through the melt are enhanced: the precipitation of stable, rounded belite is apparent in conjunction with the formation, at a lesser extent, of angular alite. The resulting microstructures are characterised by uniform distribution and development of the phases. Firing at 1400 ◦ C has no detectable difference in terms of phase morphology and growth compared to 1380 ◦ C. Comparing BC and BC10 microstructures, slag addition disfavours alite formation and promotes the formation of the interstitial phase as well as that of Table 3 Composition of raw metals, resulted chemical composition of the produced clinkers and quality indexes results.

Raw material EAFS Limestone Flysch Oxides SiO2 Al2 O3 Fe2 O3 CaO MgO K2 O Na2 O SO3 MnO Cr2 O3 P2 O5 TiO2 Cl BaO V2 O5 Fig. 1. XRD patterns of the raw materials. The main minerals identified are: 1, calcite (CaCO3 ); 2, quartz (SiO2 ); 3, illite ((K,H3 O)(Al,Mg,Fe)2 (Si,Al)4 O10 ((OH)2 ,(H2 O))); dolomite (CaMg(CO3 )2 ); 5, albite (NaAlSi3 O8 ); 6, clinochlore 4, (Mg2.5 Fe1.65 Al1.5 Si2.2 Al1.8 O10 (OH)8 ); 7, larnite (␤-Ca2 SiO4 ); 8, gehlenite (Ca2 Al(AlSi)O7 ); 9, wüstite (FeO); 10, magnetite (Fe3 O4 ); 11, brownmillerite (Ca2 (AlFe3 )2 O5 ); 12, mayenite (Ca12 Al14 O33 ).

Total Quality indexes LSF AR SR

0 wt.% BC

5 wt.% BC5

10 wt.% BC10

0.0 84.0 16.0

5.0 80.5 14.5

10.0 77.0 13.0

25.38 5.03 2.68 63.09 1.51 0.79 0.39 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00

24.41 5.52 4.38 61.47 1.57 0.71 0.36 0.04 0.29 0.10 0.04 0.03 0.01 0.01 0.00

23.48 6.00 6.00 59.92 1.62 0.63 0.34 0.07 0.57 0.20 0.07 0.07 0.02 0.02 0.01

98.88

98.95

99.01

80.13 1.87 3.29

79.00 1.26 2.47

78.10 1.00 1.96

290

R.I. Iacobescu et al. / Journal of Hazardous Materials 196 (2011) 287–294

Fig. 2. Backscattered images of preliminary firings. BC clinker formed at different firing temperatures: (a) 1280 ◦ C, (b) 1300 ◦ C, (c) 1320 ◦ C, (d) 1350 ◦ C, (e) 1380 ◦ C and (f) 1400 ◦ C.

belite. The size of crystals varies from 5 ␮m to 50 ␮m. These values are typical for a good quality clinker. According to the above results, it was decided that the firing of the clinker should be performed at 1380 ◦ C. 3.3. Clinker characterisation The XRD patterns of the prepared clinkers are depicted in Fig. 4. Table 4 presents the mineralogical compositions calculated by Rietveld and the estimations obtained by Bogue equations. As expected, discrepancies exist between the results obtained by Bogue and Rietveld, the most significant being Bogue’s underestimation of C3 S and overestimation of C2 S. These are attributed

to the fact that Bogue’s method is based on ideal stoichiometries for the clinker phases, without taking into consideration solid solutions, and also implies a certain reaction and solidification path. On the other hand, Rietveld analysis can reflect the changing conditions induced by the different raw materials as well as the non-equilibrium conditions during firing and cooling. Nonetheless, the qualitative trend is soundly predicted by Bogue. Based on the X-ray diffraction results, the main mineralogical phases identified were: alite (C3 S), belite (C2 S), tricalcium aluminate (C3 A), and tetracalcium-alumino-ferrite (C4 AF). The specific polymorphism is important as affects the hydration and, as a consequence, the development of microstructure and mechanical properties. The main polymorphic form of stabilised alite is tri-

Fig. 3. Backscattered images of preliminary firings. BC10 clinker formed at different firing temperatures: (a) 1280 ◦ C, (b) 1300 ◦ C, (c) 1320 ◦ C, (d) 1350 ◦ C, (e) 1380 ◦ C and (f) 1400 ◦ C.

R.I. Iacobescu et al. / Journal of Hazardous Materials 196 (2011) 287–294

291

Fig. 4. X-ray patterns of the prepared clinkers. The main minerals identified: 1, C3 S; 2, C2 S; 3, C3 A; 4, C4 AF.

clinic (T), which decreases as the slag content increases towards the monoclinic (M) one. This is expected to affect early strength development towards lower values [29]. The rhombohedral (R) alite is reacting slightly more rapidly compared with T and M polymorphic forms [29,30] but without a significant effect in the present case due to the low content. Regarding belite, mainly the ␣ polymorph was identified. The ␥- and ␤-C2 S polymorphs were also detected and increased as the slag content rose. Beta, ␤- and ␣ -C2 S play an important role during late days of hydration. On the contrary, ␥C2 S does not present notable hydraulic properties. C3 A was formed as cubic and orthorhombic in BC, whereas only the orthorhombic form was found in BC5 and BC10. The predominance of orthorhombic C3 A in the last two mentioned clinkers was most probably due to the higher sulphate content in the EAFS which inhibits the formation of cubic C3 A [31]. The identified orthorhombic C3 A polymorph will react faster in the presence of gypsum than its cubic counterpart [32]. In general, the increase in EAFS content results in a decrease in C3 S and C3 A and an increase in C4 AF; this is attributed to the high iron content of the slag.

Table 4 Estimated (Bogue) and calculated (Rietveld) mineralogical composition of the prepared clinkers. Phases

BC Rietveld

BC5 Bogue

Rietveld

BC10 Bogue

Bogue

1.0 0.3 33.3 0.7

Total C2 S-␣’ C2 S-␤ C2 S-␥

35.30 45.5 1.9 0.0

27.43

28.90 41.6 4.0 1.5

22.39

21.80 42.2 5.8 0.8

17.59

Total C3 A – Cubic C3 A – Orth.

47.40 2.7 6.0

54.96

47.10 0.0 4.9

56.16

48.80 0.0 4.3

57.32

8.70 7.50 0.4 0.7 100

9.13 8.49 0.0 0.0 100

4.90 18.50 0.0 0.6 100

7.54 13.90 0.0 0.0 100

4.30 24.10 0.0 1.0 100

6.01 19.08 0.0 0.0 100

Total C4 AF Lime MgO Total

3.4 1.3 20.8 3.4

Rietveld

C3 S (M1) C3 S (M3) C3 S-T C3 S-R

4.0 1.5 15.4 0.9

Fig. 5. Backscattered images of clinkers in polished section: (a) BC, (b) BC5 and (c) BC10.

Backscattered electron images (BEI) are presented in Fig. 5a, b and c for BC, BC5 and BC10, respectively. In all cases the microstructures consist predominantly of well-developed crystals of type I belite according to Insley’s classification [33]. Their diameter varies from 5 ␮m to 50 ␮m. Most of the belite crystals display complex twin lamellae. The formed striations arise as a consequence of phase transformation during cooling. In type I belite, it has been reported that this is a skeleton structure rather than polysynthetic twinning, consisting of beta and alpha forms of belite [33]. Crystals with parallel striations are less noticeable. Angular, euhedral and subhedral alite crystals are also present. In the case of BC10, alite crystals

292

R.I. Iacobescu et al. / Journal of Hazardous Materials 196 (2011) 287–294

they undergo inversions on cooling which cause twinning resulting in strain accumulations [34] and (b) they were caused by the etching [33]. Representative EDS micro-analyses of C3 S and C2 S crystals and interstitial phase for each clinker prepared are presented in Table 5. Ferrite and aluminate phases are not presented individually due to the difficulties imposed by their micro-crystalline texture, but analysis of the interstitial phase is denoted as C3 A + C4 AF in the table. As a general remark, it was observed that alite and belite become enriched with iron as the slag content rises, whereas Ba, Cr, Ti and P are more likely to be found in the belite crystals. According to the present QXRD and SEM/EDS results and under the adopted experimental conditions, no clear conclusions can be drawn concerning the influence of slag addition on the belite crystals, although it is a host of elements such as P, S, Cr which are known as belite stabilisers. However, this will be the subject of a forthcoming communication dealing with the composition and substitutions in the individual phases. 3.4. Specific surface, setting time and soundness measurements In Table 6 the results of water demand, initial setting time and soundness of the cement pastes are presented. To prepare the cement paste with the standard consistency [26] the water demand was 27.6 wt.% for all cases. Initial setting times obtained for BC, BC5 and BC10 were 240 min, 170 min and 20 min, respectively according to EN 197-1 standard in the first 2 cases. It is observed that the use of slag decreases the setting time, with the BC10 behaving as fast-setting cement. This is attributed to the increase of the molten phase, forming upon cooling higher amounts of C4 AF, and the interaction with C3 A and gypsum during hydration. In more detail, as early hydration of cement is principally controlled by the amount and activity of C3 A, setting is balanced by the amount and type of sulphate interground with the cement. Tetracalcium-alumino-ferrite (C4 AF) reacts much like C3 A, i.e., forming ettringite in the presence of gypsum. The higher amounts of C4 AF in the clinkers with slag (more than two-fold and three-fold increase for BC5 and BC10, respectively compared to BC) consume also higher amounts of gypsum. Hence, it is likely that setting occurs due to the uncontrolled reaction of the C3 A, after depletion of the sulphate by reaction with the C4 AF. Addition of higher amounts of gypsum or of retarders, possibly of organic nature, could control the setting behaviour. Expansion was 1 mm for all prepared cements. In order to obtain Blaine of 4000 cm2 /g, 4080 cm2 /g and 4057 cm2 /g, the required milling times were 60 sec, 80 sec and 103 sec for BC, BC5 and BC10, respectively. Notably, increasing the slag addition results in an increased milling time for comparable finesse. 3.5. Compressive strength of the BC, BC5 and BC10 cements Fig. 6. Backscattered images of clinkers in fractured surface: (a) BC, (b) BC5 and (c) BC10.

contain inclusions of belite in some cases. The crystallised-duringcooling interstitial phase presents a micro-crystalline texture and mainly consists of a mixture of ferrite and C3 A. It partially separates the primary alite and belite crystals. The addition of slag favours the formation of belite and ferrite phases as it disfavours the formation of alite. Ferrite crystals pipework exposed on a broken surface are presented in Fig. 6b and c for BC5 and BC10, respectively. Voids in their structure are presumably occupied by aluminate. The lamellar belite structure is slightly visible. In Fig 6b micro-cracks that developed on the belite crystals are clearly visible. Plausible hypotheses for the formation of these micro-cracks are: (a) as dicalcium silicate crystals are considered to be complex, containing point defects,

In Fig. 7 the compressive strength results are presented. As expected for belite cements, the early days strength development is significantly lower than that of the OPC. For BC, the results for 2 days were 6.5 MPa whereas for BC5 and BC10 they were even lower, at 2.5 MPa and 1.6 MPa, respectively. However the 28-day results for BC, BC5 and BC10 were 47.5 MPa, 46.6 MPa and 42.8 MPa, respectively, which are comparable to OPC CEMI 32.5N (32.5–52.5 MPa) according to EN 197-1 [35]. These results concur with previously reported results [36–39]. The low early strength development observed in the cements with slag addition is attributed, as in cases of fast setting behaviour, to the extended formation of ferrite. The higher results for BC at early days are attributed to the higher C3 S and C3 A content, compared to BC5 and BC10. Compared to other belite cements incorporating wastes, belite cement with EAFS has

R.I. Iacobescu et al. / Journal of Hazardous Materials 196 (2011) 287–294

293

Table 5 Typical EDS micro-analyses for clinkers: (a) BC, (b) BC5 and (c) BC10, in wt.%. Phase

BC

BC5

BC10

C3 S

C2 S

C4 AF + C3 A

C3 S

C2 S

C4 AF + C3 A

C3 S

C2 S

C4 AF + C3 A

NaO MgO Al2 O3 SiO2 P2 O5 SO3 K2 O CaO TiO2 MnO FeOtotal V2 O5 BaO Cr2 O3

0.10 0.40 1.23 24.05 0.00 0.00 1.43 71.25 0.00 0.00 1.54 0.00 0.00 0.00

0.20 0.36 1.73 30.65 0.00 0.20 1.44 64.10 0.00 0.00 1.32 0.00 0.00 0.00

0.67 1.45 20.74 5.35 0.00 0.08 0.50 59.14 0.00 0.00 12.07 0.00 0.00 0.00

0.11 0.30 1.66 23.66 0.00 0.00 1.36 69.97 0.04 0.36 2.19 0.00 0.00 0.35

0.01 0.36 0.84 30.88 0.23 0.03 1.35 61.38 0.10 0.39 3.22 0.00 0.09 0.22

0.28 1.53 20.04 4.16 0.00 0.14 0.50 56.41 0.06 1.35 15.43 0.00 0.05 0.05

0.09 0.81 2.25 22.49 0.01 0.10 1.50 68.01 0.08 0.94 3.34 0.00 0.03 0.35

0.12 0.97 2.82 30.25 0.43 0.00 0.81 59.03 0.04 0.38 4.52 0.00 0.13 0.50

0.19 1.35 18.99 4.78 0.00 0.13 0.54 53.59 0.09 1.38 18.86 0.02 0.06 0.02

Total

100

100

100

100

100

100

100

100

100

• Early days compressive strength results are low, as is expected for belite cements, however the 28-day results for BC, BC5 and BC10 were 47.5 MPa, 46.6 MPa and 42.8 MPa, respectively which are comparable to EN 197-1, OPC CEMI 32.5N ones. • The addition of slag did not affect the water demand of the cements and soundness did not exceed 1 mm, although setting time was decreased for BC10, which behaved like “flash set” cement. • The fast setting that occurred for the cements with slag addition is attributed to the extended formation of the molten phase which forms ferrite upon cooling and the interaction with C3 A and gypsum during hydration. Acknowledgements

Fig. 7. Compressive strength results of the BC, BC5 and BC10 cements.

Table 6 Physical properties of the cement and cement pastes. Cement types Specific surface (cm2 /g) Initial setting time (min) Water demand (wt.%) Soundness (mm)

BC 4000 240 27.6 1

BC5

BC10

4080 170 27.6 1

4057 20 27.6 1

R.I. Iacobescu and R. Saban acknowledge the support of the Sectoral Operational Programme for Human Resources Development 2007–2013 of the Romanian Ministry of Labour, Family and Social Protection through the Financial Agreement POSDRU/6/1.5/S/16. D. Koumpouri and G.N. Angelopoulos acknowledge the support of University of Patras through the “Karatheodoris” 2011 research program. Y. Pontikes is thankful to the Research Foundation – Flanders for the post-doctoral fellowship. TITAN Cement Company S.A. and SOVEL S.A. metallurgy industries are gratefully acknowledged for providing raw materials as well as their technical assistance. References

lower early and late compressive strength than belite with Bayer’s process red mud [38] (5.0 MPa at 1 day and 53.7 MPa at 28 days) and higher early and late compressive strength than belite cement with boron waste [39] (1.5 MPa at 1 day and 32.3 MPa at 28 days). 4. Conclusion The production of belite cements with EAFS is feasible and can offer significant environmental advantages. Specifically the characteristics of the cements are as follows: • Clinkers predominantly contain well-formed belite crystals. Alite crystals are also present. • The interstitial phase is a mixture of C4 AF and C3 A, and partially separates the primary alite and belite crystals. • Slag addition favours the formation of the belite and ferrite phase and disfavours the formation of alite, in accordance with Bogue’s predictions under the requirement of comparable quality indices.

[1] International Cement Review, The Global Cement Report, ninth ed., World Overview, 2011. [2] J.S. Damtoft, J. Lukasik, D. Herfort, D. Sorrentino, E.M. Gartner, Sustainable development and climate change initiatives, Cem. Concr. Res. 38 (2008) 115–127. [3] V.M. Malhotra, Global warming, and role of supplementary cementing materials and superplasticisers in reducing greenhouse gas emissions from the manufacturing of portland cement, Int. J. Struct. Eng. 1 (2010) 116–130. [4] M. Schneider, M. Romer, M. Tschudin, H. Bolio, Sustainable cement production—present and future, Cem. Concr. Res. 41 (2011) 642–650. [5] J.I. Bhatty, F.M. Miller, S.H. Kosmatka, Innovations in Portland Cement Manufacturing, Portland Cement Association, 2004. [6] C.D. Popescu, M. Muntean, J.H. Sharp, Industrial trial production of low energy belite cement, Cem. Concr. Compos. 25 (2003) 689–693. [7] C.D. Lawrence, The production of low-energy cements, in: Lea’s Chemistry of Cement and Concrete, 4th ed., Butterworth-Heinemann, Oxford, 2003. [8] H. Uchikawa, Management strategy in cement technology for the next century. Part 3, World Cem. November (1994) 47. [9] E. Gartner, Industrially interesting approaches to “low-CO2 ” cements, Cem. Concr. Res. 34 (2004) 1489–1498. [10] K. Quillin, Performance of belite-sulfoaluminate cements, Cem. Concr. Res. 31 (2001) 1341–1349. [11] US Environmental Protection Agency, Materials characterization paper, in support of the final rulemaking: identification of nonhazardous secondary

294

[12] [13] [14] [15] [16] [17] [18]

[19]

[20] [21] [22]

[23] [24]

[25] [26]

R.I. Iacobescu et al. / Journal of Hazardous Materials 196 (2011) 287–294 materials that are solid waste steel furnace Slag (used as an ingredient in clinker manufacture and bituminous concrete), in: Steel Furnace Slag, 2011, pp. 1–5. I. Akln Altun, Y. Ismail, Study on steel furnace slags with high MgO as additive in Portland cement, Cem. Concr. Res. 32 (2002) 1247–1249. H. Motz, J. Geiseler, Products of steel slags an opportunity to save natural resources, Waste Manag. 21 (2001) 285–293. European Slag Association, www.euroslag.org, 2011. International Iron and Steel Institute, IISI Steel statistical Yearbook, Brussels, Belgium, 2004. http://www.worldsteel.org/. Hellenic Cement Industry Association, http://www.hcia.gr. S.R. Rao, Waste management series 7, in: Resource Recovery and Recycling from Metallurgical Wastes, 2006. J. Paceagiu, E. Radulescu, A.M. Dragomir, R. Hotnog, Implications of the use of steel slags to clinker manufacture: laboratory test results, Romanian J. Mater. 40 (2010) 306–314. C. Shi, Corrosion resistant cement made with steel mill by-products, Proceedings of International Symposium of the Utilisation of Metallurgical Slags, Chinese Society for Metals, Beijing, 1999, pp. 171–178. D. Adolfsson, N. Menad, E. Viggh, B. Björkman, Steelmaking slags as raw material for sulphoaluminate belite cement, Adv. Cem. Res. 19 (2007) 147–156. European Parliament, Directive 2003/53/EC of the European Parliament and of the Council, 2003. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health, NIOSH, Criteria document update, Occupational Exposure to Hexavalent Chromium, External Review Draft, 2008. B.W. Nicholas, Understanding Cement, WHD Microanalysis Consultant Ltd., United Kingdom, 2010. C.R. Ward, J.C. Taylor, C.E. Matulis, L.S. Dale, Quantification of mineral matter in the Argonne Premium Coals using interactive Rietveld-based X-ray diffraction, Int. J. Coal Geol. 46 (2001) 67–82. European Committee for Standardization, EN 196-6, Methods of testing cement. Determination of Fineness, 1989. European Committee for Standardization, EN 196-3, Methods of testing cement. Part 3. Determination of Setting Time and Soundness, 1994.

[27] European Committee for Standardization, EN 196-1, Methods of testing cement. Part 1. Determination of Strength, 1994. [28] S.N. Ghosh, P.B. Rao, A.K. Paul, K. Raina, Review. The chemistry of dicalcium silicate mineral, J. Mater. Sci. 14 (1979) 1554–1566. ´ The influence of the alite polymorphism on the strength [29] T. Stanek, P. Sulovsky, of the Portland cement, Cem. Concr. Res. 32 (2002) 1169–1175. [30] R.T.H. Aldous, The hydraulic behaviour of rhombohedral alite, Cem. Concr. Res. 13 (1983) 89–96. [31] L. Gobbo, L. Sant’ Agostino, L. Garcez, C3 A polymorphs related to industrial clinker alkalies content, Cem. Concr. Res. 34 (2004) 657–664. [32] A. Kirchheim, V. Fernàndez-Altable, P. Monteiro, D. Dal Molin, I. Casanova, Analysis of cubic and orthorhombic C3 A hydration in presence of gypsum and lime, J. Mater. Sci. 44 (2009) 2038–2045. [33] D.H. Campbell, Microscopical Examination and Interpretation of Portland Cement and Clinker, Portland Cement Association, Skokie, IL 60077-1083, USA, 1999. [34] S.N. Ghosh, Advances in Cement Technology: Critical Reviews and Case Studies on Manufacturing, Quality Control, Optimisation and Use, New Delhi, India, 1983. [35] European Committee for Standardization, EN 197-1, Cement. Part 1. Composition, Specifications and Conformity Criteria for Common Cements, 2000. [36] J. Stark, A. Müller, R. Schrader, K.R. Rumpler, Existence Conditions of Hydraulically Active Belite, Zement-Kalk-Gips, Bauverlag GmbH, Wiesebaden, Germany, 1981. [37] I. Vangelatos, Valorisation of Red Rud in the Cement Industry, Department of Chemical Engineering, University of Patras, Patras, 2008. [38] I. Vangelatos, Y. Pontikes, G.N. Angelopoulos, Ferroalumina as a raw material for the production of “Green” belite type cements, in: SERES’ 09. I. International Ceramic, Glass, Porcelain Enamel, Glaze and Pigment Congress, Eskisehir, Turkey, 2009. [39] T. Kavas, I. Vangelatos, S. Koyas, Y. Tabak, G.N. Angelopoulos, Wastes from alumina and boron production as raw materials for belite cement, in: J. Heinrich, C. Aneziris (Eds.), 10th Conference and Exhibition of the European Ceramic Society, Berlin, Germany, 2007, pp. 1799–1803.

Journal of Hazardous Materials 196 (2011) 248–254

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

The influence of preparation method, nitrogen source, and post-treatment on the photocatalytic activity and stability of N-doped TiO2 nanopowder Shaozheng Hu ∗ , Fayun Li, Zhiping Fan Institute of Eco-environmental Sciences, Liaoning Shihua University, Fushun 113001, PR China

a r t i c l e

i n f o

Article history: Received 19 May 2011 Received in revised form 6 September 2011 Accepted 6 September 2011 Available online 10 September 2011 Keywords: Lattice-nitrogen TiO2 Photocatalysis Stability Post-treatment

a b s t r a c t NH3 plasma, N2 plasma, and annealing in flowing NH3 were used to prepare N doped TiO2 , respectively. XRD, UV–vis spectroscopy, N2 adsorption, FT-IR, Zeta-potential measurement, and XP spectra were used to characterize the prepared TiO2 samples. The nitridation procedure did not change the phase composition and particle sizes of TiO2 samples, but extended its absorption edges to the visible light region. The photocatalytic activities were tested in the degradation of an aqueous solution of a reactive dyestuff, methylene blue, under visible light. The photocatalytic activity and stability of TiO2 prepared by NH3 plasma were much higher than that of samples prepared by other nitridation procedures. The visible light activity of the prepared N doped TiO2 was improved by increasing the lattice-nitrogen content and decreasing adsorbed NH3 on catalyst surface. The lattice-nitrogen stability of N-doped TiO2 samples improved after HCl solution washing. The possible mechanism for the photocatalysis was proposed. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Nanocrystalline TiO2 has great potential for many applications such as photocatalysis, solar energy conversion, and gas sensor [1,2]. However, with a wide band gap energy of 3.0–3.2 eV, TiO2 cannot be activated to generate photoexcited electrons and holes to promote redox reaction unless it is irradiated by ultraviolet. This hinders the application of TiO2 as a photocatalyst with response to solar light or even indoor light. Therefore, it is highly desirable to shift the absorption edge of TiO2 to the visible light region. In 2001, Asahi et al. [3] prepared nitrogen doped TiO2 films by sputtering TiO2 in a N2 /Ar gas mixture, and concluded that the doped N atoms narrowed the band gap of TiO2 by mixing N 2p and O 2p states, therefore demonstrating the activity for the decomposition of acetone and methylene blue. Since then, N-doping has become a hot topic and been widely investigated. Heating TiO2 powders in N2 and/or NH3 at elevated temperatures is the conventional method to prepare nitrogen-doped TiO2 [3]. Besides the energy waste, the treatment at such high temperature usually results in the low surface area due to grain growth, which would decrease the number of photoactive sites. Therefore, new strategies for preparing nitrogen-doped TiO2 , such as sputtering [4], sol–gel [5], ion implantation [6], pulsed laser deposition [7], hydrothermal synthesis [8], and plasma treatment [9] have been proposed more recently.

∗ Corresponding author. Tel.: +86 24 23847473. E-mail address: [email protected] (S. Hu). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.021

Non-thermal plasma is composed of atoms, ions and electrons, which are much more reactive than their molecule precursors. Plasma is able to initiate a lot of reactions, which take place efficiently only at elevated temperatures and high pressures, under mild conditions. So far, some literatures on preparation of N doped TiO2 by plasma treatment have been reported [9–13]. Yamada et al. [9–11] investigated the photocatalytic activity of TiO2 thin films prepared by plasma treatment using N2 as nitrogen source. They suggested that the substitutional N-doping contributed to the band gap narrowing, therefore absorbing visible light and demonstrating the photocatalytic activity. Abe et al. [12] prepared N doped TiO2 by NH3 (10%)/Ar plasma. The influence of the NH3 /Ar gas pressures (50, 300 and 1000 Pa) on the physical and photocatalytic property of the powder was investigated. Miao et al. [13] reported the structural and compositional properties of TiO2 thin films prepared by N2 –H2 plasma treatment. HRTEM results indicated that the primitive lattice cells of anatase TiO2 films are distorted after plasma treatment in comparison with that of bulk TiO2 , which confirmed the N doping by N2 –H2 plasma. It is shown from the above literatures that N2 and NH3 were usually used as nitrogen source to prepare N doped TiO2 under plasma treatment. However, few literature on the comparison of N2 and NH3 plasma treated TiO2 were reported. In this work, NH3 plasma, N2 plasma, and annealing in flowing NH3 were used to prepare N doped TiO2 , respectively. The structural and optical properties of prepared N doped TiO2 were compared. The photocatalytic performance was evaluated in the degradation of methylene blue under visible light. The possible mechanism for the photocatalysis was proposed.

S. Hu et al. / Journal of Hazardous Materials 196 (2011) 248–254

249

2. Experimental 2.1. Preparation and characterization

2.2. Photocatalytic reaction Methylene blue (MB) was selected as model compound to evaluate the photocatalytic performance of the prepared TiO2 particles in an aqueous solution under visible light irradiation. 0.1 g TiO2 powders were dispersed in 100 ml aqueous solution of MB (initial concentration C0 = 50 ppm, pH 6.8) in an ultrasound generator for 10 min. The suspension was transferred into a self-designed glass reactor, and stirred for 30 min in darkness to achieve the adsorption equilibrium. The concentration of MB at this point was considered as the absorption equilibrium concentration C0 . The adsorption capacity of a catalyst to MB was defined by the adsorption amount of MB on the photocatalyst (C0 − C0 ). In the photoreaction under visible light irradiation, the suspension was exposed to a 110 W high-pressure sodium lamp with main emission in the range of 400–800 nm, and air was bubbled at 130 ml min−1 through the solution. The UV light portion of sodium lamp was filtered by 0.5 M NaNO2 solution [14]. The light intensity is 130 mW cm−2 . All runs were conducted at ambient pressure and 30 ◦ C. At given time intervals, 4 ml suspension was taken and immediately centrifuged to separate the liquid samples from the solid catalyst. The concentrations of MB before and after reaction were measured by means of a UV–vis spectrophotometer at a wavelength of 665 nm. It is the linear relationship between absorbance and concentration of liquid sample in the experimental concentration range. Therefore, the percentage of degradation D% was determined as follows: D% =

A0 − A × 100% A0

TO-PN2

TO-CNH3

P25

20

30

40

50

60

2θ / deg. Fig. 1. XRD patterns of P25 and prepared N-doped TiO2 samples.

where A0 and A are the absorbances of the liquid sample before and after degradation, respectively. 3. Results and discussion It is reported that the phase composition and particle size of TiO2 have significant influence on its photocatalytic activity [2]. The XRD patterns of P25 and prepared N-doped samples (Fig. 1) indicate that all TiO2 samples were mixtures of anatase and rutile phases. The phase contents and the particle sizes of the samples were calculated by their XRD patterns according to the method of Spurr [15] and Debye–Scherrer equation [16], respectively. The results (Table 1) indicate that there were no remarkable changes in phase composition and particle sizes. Up to date, the mechanism of the enhancement by N-doping is still controversial. Asahi et al. [3] concluded that the doped N atoms narrowed the band gap of TiO2 by mixing N 2p and O 2p states, therefore demonstrating the activity. Irie et al. [17] argued that the isolated narrow band located above the valence band is responsible for the visible light response. Lee et al. [18] suggested that substitutional N-doping would narrow the band gap by the coupling of the O 2p and N 2p orbitals, while interstitial N-doping would create an isolated defect state between the conduction band and valence band. Fig. 2 shows the UV–vis spectra of P25 and prepared N-doped TiO2 samples. Compared with the spectra of P25, obvious red-shifts of the absorption bands were observed for prepared N-doped TiO2 . 1.0

TO-PNH3

0.8

TO-CNH3 TO-PN2

0.6

Abs.

The doping of TiO2 was conducted in a dielectric barrier discharge (DBD) reactor, consisting of a quartz tube and two electrodes. The high voltage electrode was a stainless-steel rod (2.5 mm), which was installed in the axis of the quartz tube and connected to a high voltage supply. The grounding electrode was an aluminum foil, which was wrapped around the quartz tube. For each run, 0.4 g commercial TiO2 powder (P25) was charged into the quartz tube. At a constant NH3 flow (40 ml min−1 ), a high voltage of 9–11 kV was supplied by a plasma generator at an overall power input of 50 V × 0.4 A. The discharge frequency was fixed at 10 kHz, and the discharge was kept for 15 min. After discharge, the reactor was cooled down to room temperature. The obtained TiO2 sample was denoted as TO–PNH3 . When N2 was used to replace NH3 following the same procedure in the preparation of TO–PNH3 , the product is denoted as TO–PN2 . For comparison, P25 was calcined under NH3 flow (40 ml min−1 ) for 15 min at 500 ◦ C. The obtained sample was denoted as TO–CNH3 . XRD patterns of the prepared TiO2 samples were recorded on a ˚ Rigaku D/max-2400 instrument using Cu K␣ radiation ( = 1.54 A). UV–vis spectroscopy measurement was carried out on a Jasco V-550 spectrophotometer, using BaSO4 as the reference sample. FT-IR spectra were obtained on a Nicolet 20DXB FT-IR spectrometer in the range of 400–2300 cm−1 . The zeta-potential of the catalyst was measured at room temperature on Zetasizer Nano S90 (Malvern Instruments). The pH was adjusted by dropwise addition of dilute HCl or NaOH solution. Photoluminescence (PL) spectra were measured at room temperature with a fluorospectrophotometer (FP-6300) using a Xe lamp as excitation source. XPS measurements were conducted on a Thermo Escalab 250 XPS system with Al K␣ radiation as the exciting source. The binding energies were calibrated by referencing the C1s peak (284.6 eV) to reduce the sample charge effect.

Intensity / a.u.

TO-PNH3

P25 0.4

0.2

0.0 300

400

500

600

700

Wavelength / nm (1) Fig. 2. UV–vis spectra of P25 and prepared N-doped TiO2 samples.

800

250

S. Hu et al. / Journal of Hazardous Materials 196 (2011) 248–254

Table 1 Summary of physical properties of P25 and prepared N-doped TiO2 samples. Sample

Size (nm)

XA (%)a

SBET (m2 g−1 )

Pore volume (cm3 g−1 )

Central pore size (nm)

Eg (eV)

Nfresh (at.%)b

Nused (at.%)c

P25 TO–PN2 TO–CNH3 TO–PNH3

28.2 28.5 29.3 28.1

74.6 74.7 74.4 75.1

43 40 36 41

0.07 0.06 0.05 0.06

3.6 3.2 3.1 3.4

3.10 2.92 2.75 2.67

0 1.32 1.64 1.95

0 0.76 1.17 1.91

c

XA represents the phase composition of anatase. Nfresh represents the lattice-nitrogen content before photocatalytic reaction Nused represents the lattice-nitrogen content after photocatalytic reaction.

The band gap energies of TiO2 samples which calculated according to the method of Oregan and Gratzel [19] (Table 1) indicate that the prepared N-doped TiO2 samples exhibited much narrowed band gap energies. According to the previous result [3,18], this indicated that substitutional N-doping existed in the prepared N-doped TiO2 samples. It is shown that the band gap energy decreased in the order: TO–PN2 > TO–CNH3 > TO–PNH3 , which is probably due to the different doping N content in prepared N-doped TiO2 samples. Xu et al. [20] prepared N doped TiO2 by pulsed laser deposition and suggested that more absorption edge red-shift indicated higher nitrogen concentration. Besides, the distinct differences in visible light absorption are observed between calcination and plasma treated samples. In the spectrum of TO–CNH3 , the obvious absorption in 400–550 nm is observed, which is a typical absorption region for N doped TiO2 materials. This typical absorption is due to the electronic transition from the isolated N 2p level, which is formed by incorporation of nitrogen atoms into the TiO2 lattice, to the conduction band [21]. However, the spectra of plasma treated samples are obvious different. The broad absorptions in the whole visible light region are observed in the spectra of TO–PN2 and TO–PNH3 . Huang et al. [22] prepared the visible light responsive TiO2 by nitrogen-plasma surface treatment, and found the similar broad absorption in visible light region. Abe et al. [12] prepared N doped TiO2 by NH3 /Ar plasma, and suggested that such broad absorption is attributed to the presence of Ti3+ , which might be formed by plasma treatment. It is noted that TO–PNH3 showed much stronger broad absorption in visible light region than TO–PN2 . This is probably due to that NH3 plasma consists of not only various active nitrogen species but excited hydrogen, leading to Ti4+ reduced easily, thus more Ti3+ were formed. Therefore, according to the conclusion of Lee et al. [18], it is proposed that substitutional and interstitial Ndoping existed simultaneously in TO–CNH3 which caused the band gap narrowing and remarkable absorption in 400–550 nm, whereas only substitutional N-doping existed in TO–PN2 and TO–PNH3 . The

N 1s

TO-PN2

Intensity / a.u.

a b

TO-PNH3

TO-CNH3

396

398

400

Fig. 3. XP spectra of prepared N-doped TiO2 samples in the region of N1s.

broad absorptions of TO–PN2 and TO–PNH3 in the whole visible light region were due to the presence of Ti3+ caused by the N-doping. XPS is an effective surface test technique to characterize elemental composition and chemical states. According to the previous literatures [10,11], the peaks around 396 and 400 eV are attributed to the formation of lattice-nitrogen and other surface N species such as N–N and N–O bond. The XP spectra in the region of N1s (Fig. 3) indicated that most N species in prepared TiO2 using NH3 as nitrogen source existed in lattice-nitrogen, whereas other surface N species such as N–N and N–O bond were dominant in TO–PN2 which using N2 as nitrogen source. The lattice-nitrogen content calculated by XPS data were shown in Table 1. The Nfresh content decreased in the order: TO–PNH3 > TO–CNH3 > TO–PN2 , which indicated that NH3 plasma treatment is more effective than another two methods

(B)

(A)

O 1s

Intensity / a.u.

Intensity / a.u.

Ti 2p

TO-PNH3 TO-CNH3

TO-PNH3 TO-PN2 TO-CNH3

TO-PN2

P25

P25

456

458

460

462

Binding Energy / eV

464

466

402

Binding Energy / eV

529

530

531

532

Binding Energy / eV

Fig. 4. XP spectra of P25 and prepared N-doped TiO2 samples in the region of Ti 2p (A) and O1s (B).

533

534

S. Hu et al. / Journal of Hazardous Materials 196 (2011) 248–254

251

P25 TO-CNH3

20

TO-P HN3

TO-PN2

1070 cm

TO-C HN3

-1

1224 cm

-1

P25 1402 cm -1 1620 cm-1

655 cm-1

Zeta potential / mV

Absobance / a.u.

10

TO-PNH3 0 2

4

6

8

10

12

-10

-20

-30

pH 400

600

800

1000

1200

1400

1600

Wavenumber / cm

1800

2000

2200

-1

Fig. 6. Plots of the zeta-potential as a function of pH for P25 and prepared N-doped TiO2 suspensions in the presence of NaCl (10−3 M).

Fig. 5. FT-IR spectra of P25, TO–CNH3 and TO–PNH3 .

to form lattice-nitrogen. This is probably due to that NH3 plasma consists of various nitriding species, leading to the formation of lattice-nitrogen easier [23]. Besides, trace N species, which located at 395.3 eV were present in TO–PNH3 and TO–CNH3 . Li et al. [24] prepared N doped TiO2 in NH3 /ethanol fluid under supercritical condition, and suggested that the N species with binding energy at 395.3 eV was attributed to the surface adsorbed NH3 molecules. In this investigation, the N species located at 395.3 eV only existed in TO–PNH3 and TO–CNH3 , which prepared using NH3 as nitrogen source. This confirmed the result of Li et al. Furthermore, the peak intensity of TO–CNH3 at 395.3 eV was obvious higher than that of TO–PNH3 , indicating more NH3 molecules adsorbed on TO–CNH3 surface. Fig. 4 shows the XP spectra of P25 and prepared N-doped TiO2 samples in the region of Ti 2p and O1s. Compared with the spectra of P25, obvious shifts to lower binding energies were observed for N-doped TiO2 samples in the Ti 2p region (458.4 eV) as well as the O1s region (529.7 eV). This is probably attributed to the change of chemical environment after N doping [24]. It is known that the binding energy of the element is influenced by its electron density. A decrease in binding energy implies an increase of the electron density. The electrons of N atoms may be partially transferred from N to Ti and O, due to the higher electronegativity of oxygen, leading to increased electron densities on both Ti and O. The peaks around 530 and 532 eV in the O1s region are attributed to crystal lattice oxygen (Ti–O) and surface hydroxyl group (O–H) of TiO2 . The ratio of these two peak areas (SO–H /STi–O ) represents the abundance of surface hydroxyl groups. The calculated results indicated that SO–H /STi–O ratio for TO–CNH3 was 0.06, much lower than that of P25 (0.14). Whereas, SO–H /STi–O ratios for TO–PN2 and TO–PNH3 were 0.13 and 0.11, which were slight lower than P25. This indicated the content of surface hydroxyl groups decreased more drastically after the calcination procedure under NH3 flow compared with plasma treatment. Those surface hydroxyl groups are known to play an important role in photocatalysis. They react with photogenerated holes, producing active hydroxyl radicals, which are responsible for the photo-degradation [25]. The FT-IR spectra of P25 and TO–PNH3 were shown in Fig. 5. The absorption peak at 1620 cm−1 is attributed to bending vibration of hydroxyl group. The band at around 655 cm−1 belongs to O–Ti–O structure of TiO2 . There are three bands at 1402, 1224, and 1070 cm−1 which were observed in the spectra of TO–CNH3 and TO–PNH3 , but not in that of P25. The band at 1402 cm−1 is attributed to the surface adsorbed NH3 species on Brönsted acid sites (–OH)

[26]. It is known that NH3 can adsorb on Brönsted acid sites (–OH) located at 1400 cm−1 and Lewis acid sites (Ti4+ ) located at 1225 and 1190 cm−1 [26,27]. However, in Fig. 5, no NH3 adsorbed on the Lewis acid sites was observed. There are many previous literatures, which report the FT-IR results of NH3 adsorbed on TiO2 materials. Some of them reported that NH3 adsorbed on both Brönsted acid and Lewis acid sites [26,27]. Other results showed that only adsorption on Brönsted acid sites was obtained which is consistent with the result of Fig. 5 [24,28]. Therefore, it is proposed that the preparation methods and conditions probably affect the adsorption state, leading to the adsorption site different from different literatures. It is known that the TiO2 surface is hydrophilic. In this investigation, TiO2 materials were treated under calcination and plasma condition only for 15 min, leading to most of the H2 O adsorbed on Ti4+ still existed on TiO2 surface. It is reported that when Ti4+ sites are saturated by hydroxyl groups, NH3 will adsorb mainly on Brönsted acid sites by the formation of an N· · ·HO bond [29]. Therefore, only adsorption on Brönsted acid sites was obtained. In Fig. 5, the peaks at 1224 and 1070 cm−1 could be attributed to the nitrogen atoms embedded in the TiO2 network, which is consistent with XPS result [28]. These results confirmed the formation of doping N species in the TiO2 lattice. Fig. 6 shows the plots of the zeta-potential as a function of pH for P25 and prepared N-doped TiO2 suspensions in the presence of NaCl (10−3 M). It is known that the point of zero charge (PZC) of TiO2 is around 3–6, indicating the surface of TiO2 particles is positively charged. Compared with P25, the distinct shifts to lower value of the PZC were observed for all the N doped TiO2 , indicating the positive charge on TiO2 surface decreased. The PZC value decreased in the order: P25 > TO–PN2 > TO–PNH3 > TO–CNH3 . It is possible that the lone electron pair of doping N counteract a few of positive charge. Besides, NH3 readily adsorbed on the catalyst surface during the nitridation process, due to the numerous acidic hydroxyl groups on the TiO2 surface. The presence of these surface-adsorbed NH3 decreased the number of acidic hydroxyl groups, resulting in lower PZC value of TO–PNH3 and TO–CNH3 . Furthermore, plasma treatment caused the NH3 decomposition more drastically, leading to less NH3 adsorbed on TO–PNH3 surface compared with TO–CNH3 . Therefore, the PZC value of TO–PNH3 is higher than TO–CNH3 . The adsorption of MB on TiO2 -based catalysts was measured by the equilibrium adsorption capacity. The adsorption capacities of all the N doped TiO2 samples were lower than that of P25 (Fig. 7). The BET specific surface area (SBET ), pore volume, and central pore size are listed in Table 1. Compared with P25, the SBET , pore volume, and central pore size of prepared samples decreased.

252

S. Hu et al. / Journal of Hazardous Materials 196 (2011) 248–254

6

100

TO-PNH3

5

TO-PNH3(HCl)

TO-PN2 4

D/%

TO-PNH3 3

TO-CNH3

60

40

2 20

1

0

0

0

Fig. 7. Adsorption capacity of MB on P25 and N-doped TiO2 samples.

This probably caused the decreased equilibrium adsorption capacity shown in Fig. 7. Besides, it is possibly that such decreased adsorption of MB is attributed to the coverage of TiO2 surface by excess surface N species. It is noted that the equilibrium adsorption capacity decreased in the order: P25 > TO–PN2 > TO–PNH3 > TO–CNH3 , which is completely consistent with the order of surface hydroxyl groups content. This indicated that the content of surface hydroxyl groups influenced significantly on the equilibrium adsorption capacity. It is shown that equilibrium adsorption capacities of TO–PNH3 and TO–CNH3 were lower than TO–PN2 , which using N2 as nitrogen source. Besides the lower surface hydroxyl groups content than that of TO–PN2 , large numbers of NH3 molecules adsorbed on hydroxyl groups of TO–PNH3 and TO–CNH3 , caused the reduced surface sites for adsorbing MB, leading to lower equilibrium adsorption capacity than TO–PN2 . The photocatalytic performances under visible light shown in Fig. 8 indicate that prepared N-doped TiO2 samples exhibited much higher activities than that of P25. Since no obvious change were observed in phase compositions and particle sizes between P25 and prepared N-doped TiO2 samples, the enhanced photocatalytic activity must result from the doping of nitrogen in TiO2 , which gave rise to the narrowed band gap and thus to the enhanced absorption in the visible region. Moreover, it is shown that the photocatalytic activity increased in the order: TO–PN2 > TO–CNH3 > TO–PNH3 , which is in agreement with the order of lattice-nitrogen content (Nfresh ). This proved that the lattice-nitrogen significantly 100

P25 TO-PN2

80

TO-CNH3 TO-PNH3

D/%

TO-PNH3(H2O)

80

-6

Adsorption amount / 10 g/g

P25

60

40

20

0 0

1

2

3

4

t/h Fig. 8. Photocatalytic performances of P25 and prepared N-doped TiO2 samples in the degradation MB under visible light irradiation.

1

2

3

4

t/h Fig. 9. Photocatalytic performances of TO–PNH3 , TO–PNH3 (H2 O), and TO–PNH3 (HCl) in the degradation MB under visible light irradiation.

influenced the visible light activity, which is consistent with the earlier results of Yamada [10]. On the other hand, the stronger absorption in visible light region of TO–PNH3 caused the visible light utilization more effectively, thus leading to the much higher activity than that of TO–PN2 and TO–CNH3 . Nused is calculated and shown in Table 1. Obviously, Nused of TO–CNH3 and TO–PN2 are much lower than that of Nfresh , whereas lattice-nitrogen of TO–PNH3 is relatively stable. It is reported that the lattice-nitrogen was oxidated by photogenerated holes during the degradation reaction, leading to the decrease of lattice-nitrogen content [30]. Therefore, it is deduced that the oxidation of lattice-nitrogen of TO–PNH3 is more difficult than that of another two samples. This difference in lattice-nitrogen stability is probably due to the different preparation method among three samples. Besides, Chen et al. [30] prepared N doped TiO2 by heating TiO2 powders in NH3 flow and found that the presence of surface-adsorbed NH3 decreased the number of surface sites accessible for reactants, resulting in low photocatalytic activity. In this investigation, compared with TO–PNH3 , more NH3 adsorbed on TO–CNH3 surface, thus leading to the lower adsorption capacity and photocatalytic activity of TO–CNH3 . To confirm the detrimental effect of NH3 , Chen et al. [30] washed the prepared N doped TiO2 with pure water for several times to remove adsorbed NH3 . The photocatalytic activity of obtained sample was improved after washing, but still much lower than that of postcalcination sample (NT400). This is probably due to that NH3 was not removed completely by washing with pure water. In this investigation, TO–PNH3 was washed with HCl (0.1 M) to remove the adsorbed NH3 , and then cleaned with deionized water. The obtained sample was denoted as TO–PNH3 (HCl). For comparison, TO–PNH3 (H2 O) was obtained by washing TO–PNH3 with deionized water directly. The FT-IR results (not shown) indicated that adsorbed NH3 were removed completely after HCl washing, whereas residual NH3 still existed on TO–PNH3 (H2 O) surface. The photocatalytic performances (Fig. 9) show that the activity increased in the order: TO–PNH3 < TO–PNH3 (H2 O) < TO–PNH3 (HCl), which confirmed NH3 detrimental effect on photocatalytic activity. When TO–PN2 and TO–CNH3 were used to replace TO–PNH3 following the same procedure as in the preparation of TO–PNH3 (HCl), the product were denoted as TO–PN2 (HCl) and TO–CNH3 (HCl), respectively. The photocatalytic performances of TO–PN2 (HCl), TO–CNH3 (HCl), and TO–PNH3 (HCl) were investigated in three cycles to check the photocatalytic stability (Fig. 10). It is shown that the activity of TO–PNH3 (HCl) decreased slightly in 1st reuse and kept stable in the next two cycles. However, for TO–PN2 (HCl) and

S. Hu et al. / Journal of Hazardous Materials 196 (2011) 248–254

253

120 110

TO-PNH3(HCl)

TO-CNH3(HCl)

+

N 1s

TO-PN2(HCl)

after Ar etching

100 90

Intensity / a.u.

D/%

80 70 60 50 40

TO-PNH3(HCl)

30

TO-CNH3(HCl)

20 10 0

TO-PN2(HCl) fresh

1st reuse

2nd reuse

3rd reuse

Fig. 10. Photocatalytic stability of prepared N-doped TiO2 samples in the degradation of MB.

TO–CNH3 (HCl), the activities decreased gradually, from 41.8% and 55.1% for fresh catalyst to 30.9% and 49.6% for 3rd reused catalyst. This hinted that the photocatalytic stability of TO–PNH3 (HCl) was much better than TO–PN2 (HCl) and TO–CNH3 (HCl). It is proposed that this difference in photocatalytic stability is attributed to the different lattice-nitrogen stability among the three samples. Therefore, the lattice-nitrogen contents of fresh and reused TO–PN2 (HCl), TO–CNH3 (HCl), and TO–PNH3 (HCl) were calculated according to the relevant XPS data (Table 2). The lattice-nitrogen contents of TO–PN2 (HCl) and TO–CNH3 (HCl) decreased gradually from 1.28 at.% and 1.62 at.% to 0.46 at.% and 1.04 at.% after three cycles which confirmed that the lattice-nitrogen significantly influenced the visible light activity. However, lattice-nitrogen content of TO–PNH3 (HCl) decreased slightly from 1.94 at.% to 1.78 at.% for the 1st reuse, and then kept stable in the next two cycles. This indicated that the lattice-nitrogen atoms in TO–PNH3 (HCl) remained relatively stable. As mentioned above, Nused of TO–CNH3 and TO–PN2 are much lower than that of Nfresh , whereas lattice-nitrogen of TO–PNH3 is stable (Table 1). This difference in lattice-nitrogen stability is probably due to the different preparation method among three samples. In order to elucidate why the photocatalytic stability is different among the samples, the XP spectra of fresh TO–PN2 (HCl), TO–CNH3 (HCl), and TO–PNH3 (HCl) in N1s region after Ar+ ion etching were measured and shown in Fig. 11. Apparently, the surface adsorbed NH3 species located at 395.3 eV and N–N (N–O) species located at 400 eV were removed after Ar+ ion etching to get rid of the surface layer. Only one peak around 396 eV, which attributed to lattice-nitrogen was observed for all the three samples. The calculation according to the relevant XPS data revealed that the lattice-nitrogen content of TO–PN2 (HCl), TO–CNH3 (HCl), and TO–PNH3 (HCl) were 0.32, 0.74, and 1.58 at.%, respectively. Compared with the data of fresh catalyst in Table 2, more than 50% and 75% lattice-nitrogen in TO–CNH3 (HCl) and TO–PN2 (HCl) was eliminated after Ar+ ion etching. This indicated that a great number of N atoms doped only into the surface layer of TO–CNH3 (HCl) and TO–PN2 (HCl), which were oxidated easily by photogenerated holes Table 2 Lattice-nitrogen content of fresh and reused TO–PNH3 (HCl), TO–CNH3 (HCl), and TO–PN2 (HCl) determined by XPS data. Sample

Fresh catalyst (at.%)

1st recycle (at.%)

2nd recycle (at.%)

3rd recycle (at.%)

TO–PNH3 (HCl) TO–CNH3 (HCl) TO–PN2 (HCl)

1.94 1.62 1.28

1.78 1.31 1.02

1.78 1.22 0.75

1.75 1.04 0.46

390

395

400

405

Binding Energy / a.u. Fig. 11. XP spectra of TO–PN2 (HCl), TO–CNH3 (HCl), and TO–PNH3 (HCl) in the region of N1s after Ar+ ion etching.

Table 3 Comparison of lattice-nitrogen stability of N-doped TiO2 samples before and after HCl solution washing. Sample

Fresh catalyst (at.%)

1st reuse (at.%)

Retention ratea

TO–PNH3 (HCl) TO–CNH3 (HCl) TO–PN2 (HCl) TO–PNH3 TO–CNH3 TO–PN2

1.94 1.62 1.28 1.95 1.64 1.32

1.78 1.31 1.02 1.91 1.17 0.76

0.92 0.81 0.80 0.98 0.71 0.58

a Retention rate is equal to the ratio of lattice-nitrogen content in 1st reused catalyst to that of fresh catalyst.

during the degradation reaction, leading to the decrease of latticenitrogen content. Therefore, TO–PN2 and TO–CNH3 exhibited the poor photocatalytic stability. On the contrary, compared with the data of fresh TO–PNH3 (HCl) in Table 2, less than 20% lattice-nitrogen of TO–PNH3 (HCl) was removed after Ar+ ion etching. This is probably due to that the excited hydrogen species produced by NH3 plasma made the N atoms doped into crystal lattice of deeper layer, thus caused it oxidated difficulty by photogenerated holes. Therefore, the photocatalytic stability of TO–PNH3 (HCl) was much higher than that of TO–PN2 and TO–CNH3 . The retention rate of lattice-nitrogen, which represents the lattice-nitrogen stability is calculated and shown in Table 3. Compared with the sample before HCl washing, more than 10% and 20% enhancement of retention rate are observed in TO–CNH3 (HCl) and TO–PN2 (HCl), whereas only slight decrease of retention rate is shown in TO–PNH3 (HCl). This indicated the lattice-nitrogen stability of N-doped TiO2 samples improved after HCl solution washing. This is probably due to that the surface N species absorbed on Brönsted acid sites (–OH) were removed by HCl solution, leading to more surface hydroxy groups are available to trap the photogenerated holes, thus restrain the oxidation of lattice-nitrogen by photo-generated holes. 4. Conclusion NH3 plasma, N2 plasma, and annealing in flowing NH3 were used to prepare N doped TiO2 respectively to investigate the influence of preparation method, nitrogen source, and post-treatment on the photocatalytic activity and stability. The photocatalytic activity increased in the order: TO–PN2 < TO–CNH3 < TO–PNH3 , indicating NH3 plasma is most effective among the three

254

S. Hu et al. / Journal of Hazardous Materials 196 (2011) 248–254

methods. The lattice-nitrogen significantly influenced the visible light activity. NH3 which adsorbed on catalyst surface led to the low adsorption capacity for reactant MB, thus decreased photocatalytic activity. After removal NH3 by HCl washing, the obtained catalysts exhibited much higher activities under visible light, which confirmed the detrimental effect of NH3 . The photocatalytic stability of N doped TiO2 prepared by NH3 plasma was much higher than that of samples prepared by other nitridation procedures. This is proposed that the excited hydrogen species produced by NH3 plasma made the N atoms doped into crystal lattice of deeper layer, which caused it oxidated more difficult by photogenerated holes than other catalysts, thus leading to the stable lattice-nitrogen. Besides, the lattice-nitrogen stability of N-doped TiO2 samples improved after HCl solution washing. This is probably due to that the surface N species absorbed on Brönsted acid sites (–OH) were removed by HCl solution, leading to more surface hydroxy group are available to trap the photo-generated holes, thus restrain the oxidation of lattice-nitrogen. This stable lattice-nitrogen during the degradation reaction caused the high photocatalytic stability. Acknowledgments This work was supported by National Natural Science Foundation of China (no. 41071317, 30972418), National Key Technology R & D Programme of China (no. 2007BAC16B07), the Natural Science Foundation of Liaoning Province (no. 20092080). The authors would like to thank Prof. Anjie Wang, Dalian University of Technology, for the contribution to the manuscript. References [1] A. Fujishima, T.N. Rao, D.A. Tryk, Titanium dioxide photocatalysis, J. Photochem. Photobiol. C 1 (2000) 1–21. [2] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann, Environmental applications of semiconductor photocatalysis, Chem. Rev. 95 (1995) 69–96. [3] R. Asahi, T. Morikawa, T. Ohwaki, A. Aoki, Y. Taga, Visible-light photocatalysis in nitrogen-doped titanium oxides, Science 293 (2001) 269–271. [4] T. Lindgren, J.M. Mwabora, E. Avendano, J. Jonsson, A. Hoel, C.G. Granqvist, S.E. Lindquist, Photoelectrochemical and optical properties of nitrogen doped titanium dioxide films prepared by reactive DC magnetron sputtering, J. Phys. Chem. B 107 (2003) 5709–5716. [5] M. Qiao, S.S. Wu, Q. Chen, J. Shen, Novel triethanolamine assisted sol–gel synthesis of N-doped TiO2 hollow spheres, Mater. Lett. 12 (2010) 1398–1400. [6] H. Shen, L. Mi, P. Xu, W.D. Shen, P.N. Wang, Visible-light photocatalysis of nitrogen-doped TiO2 nanoparticulate films prepared by low-energy ion implantation, Appl. Surf. Sci. 17 (2007) 7024–7028. [7] L. Zhao, Q. Jiang, J.S. Lian, Visible-light photocatalytic activity of nitrogen-doped TiO2 thin film prepared by pulsed laser deposition, Appl. Surf. Sci. 15 (2008) 4620–4625. [8] S.Z. Hu, A.J. Wang, X. Li, H. Löwe, Hydrothermal synthesis of well-dispersed ultrafine N-doped TiO2 nanoparticles with enhanced photocatalytic activity under visible light, J. Phys. Chem. Solid 71 (2010) 156–162.

[9] K. Yamada, H. Nakamura, S. Matsushima, H. Yamane, T. Haishi, K. Ohira, K. Kumada, Preparation of N-doped TiO2 particles by plasma surface modification, C. R. Chimie 9 (2006) 788–793. [10] K. Yamada, H. Yamane, S. Matsushima, H. Nakamura, K. Ohira, M. Kouya, K. Kumada, Effect of thermal treatment on photocatalytic activity of N-doped TiO2 particles under visible light, Thin Solid Films 516 (2008) 7482–7487. [11] K. Yamada, H. Yamane, S. Matsushima, H. Nakamura, T. Sonoda, S. Miura, K. Kumada, Photocatalytic activity of TiO2 thin films doped with nitrogen using a cathodic magnetron plasmatreatment, Thin Solid Films 516 (2008) 7560–7564. [12] H. Abe, T. Kimitani, M. Naito, Influence of NH3 /Ar plasma irradiation on physical and photocatalytic properties of TiO2 nanopowder, J. Photochem. Photobiol. A 183 (2006) 171–175. [13] L. Miao, S. Tanemura, H. Watanabe, S. Toh, K. Kaneko, Structural and compositional characterization of N2 –H2 plasma surface-treated TiO2 thin films, Appl. Surf. Sci. 244 (2005) 412–417. [14] F.B. Li, X.Z. Li, M.F. Hou, K.W. Cheah, W.C.H. Choy, Enhanced photocatalytic activity of Ce3+ –TiO2 for 2-mercaptobenzothiazole degradation in aqueous suspension for odour control, Appl. Catal. A 285 (2005) 181–189. [15] R.A. Spurr, H. Myers, Quantitative analysis of anatase–rutile mixtures with an X-ray diffractometer, Anal. Chem. 29 (1957) 760–762. [16] J. Lin, Y. Lin, P. Liu, M.J. Meziani, L.F. Allard, Y.P. Sun, Hot-fluid annealing for crystalline titanium dioxide nanoparticles in stable suspension, J. Am. Chem. Soc. 124 (2002) 11514–11518. [17] H. Irie, Y. Watanaba, K. Hashimoto, Nitrogen-concentration dependence on photocatalytic activity of TiO2−x Nx powders, J. Phys. Chem. B 107 (2003) 5483–5486. [18] S. Lee, I. Cho, D.K. Lee, D.W. Kim, T.H. Noh, C.H. Kwak, S. Park, K.S. Hong, J. Lee, H.S. Jung, Influence of nitrogen chemical states on photocatalytic activities of nitrogen-doped TiO2 nanoparticles under visible light, J. Photochem. Photobiol. A 213 (2010) 129–135. [19] B. Oregan, M. Gratzel, A low-cost, high-efficiency solar-cell based on dyesensitized colloidal TiO2 films, Nature 353 (1991) 737–740. [20] P. Xu, L. Mi, P.N. Wang, Improved optical response for N-doped anatase TiO2 films prepared by pulsed laser deposition in N2 /NH3 /O2 mixture, J. Cryst. Growth 289 (2006) 433–439. [21] H. Ozaki, N. Fujimoto, S. Iwamoto, M. Inoue, Photocatalytic activities of NH3 treated titanias modified with other elements, Appl. Catal. B 70 (2007) 431–436. [22] C.M. Huang, L.C. Chen, K.W. Cheng, G.T. Pan, Effect of nitrogen-plasma surface treatment to the enhancement of TiO2 photocatalytic activity under visible light irradiation, J. Mol. Catal. A 261 (2007) 218–224. [23] S. Luan, G.W. Neudeck, Effect of NH3 plasma treatment of gate nitride on the performance of amorphous silicon thin film transistors, J. Appl. Phys. 68 (1990) 3445–3450. [24] H.X. Li, J.X. Li, Y.N. Huo, Highly active TiO2 N photocatalysts prepared by treating TiO2 precursors in NH3 /ethanol fluid under supercritical conditions, J. Phys. Chem. B 110 (2006) 1559–1565. [25] S.Z. Hu, A.J. Wang, X. Li, Y. Wang, H. Löwe, Hydrothermal synthesis of ionic liquid [Bmim]OH modified TiO2 nanoparticles with enhanced photocatalytic activity under visible light, Chem. Asian J. 5 (2010) 1171–1177. [26] J.M.G. Amores, V.S. Escribano, G. Ramis, G. Busca, An FT-IR study of ammonia adsorption and oxidation over anatase-supported metal oxides, Appl. Catal. B 13 (1997) 45–58. ˜ B.S. Uphade, P.G. Smirniotis, TiO2 -supported metal oxide catalysts [27] D.A. Pena, for low-temperature selective catalytic reduction of NO with NH3 I. Evaluation and characterization of first row transition metals, J. Catal. 221 (2004) 421–431. [28] J.H. Xu, W.L. Dai, J.X. Li, Y. Cao, H.X. Li, H.Y. He, K.N. Fan, Simple fabrication of thermally stable apertured N-doped TiO2 microtubes as a highly efficient photocatalyst under visible light irradiation, Catal. Commun. 9 (2008) 146–152. [29] A. Markovits, J. Ahdjoudj, C. Minot, A theoretical analysis of NH3 adsorption on TiO2 , Surf. Sci. 365 (1996) 649–661. [30] X.F. Chen, X.C. Wang, Y.D. Hou, J.H. Huang, L. Wu, X.Z. Fu, The effect of postnitridation annealing on the surface property and photocatalytic performance of N-doped TiO2 under visible light irradiation, J. Catal. 255 (2008) 59–67.

Journal of Hazardous Materials 196 (2011) 255–262

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Enhanced photocatalytic activity of Bi2 WO6 with oxygen vacancies by zirconium doping Zhijie Zhang, Wenzhong Wang ∗ , Erping Gao, Meng Shang, Jiehui Xu State Key Laboratory of High Performance Ceramics and Superfine Microstructures, Shanghai Institute of Ceramics, Chinese Academy of Sciences, 1295 Dingxi Road, Shanghai 200050, PR China

a r t i c l e

i n f o

Article history: Received 3 July 2011 Received in revised form 5 September 2011 Accepted 6 September 2011 Available online 10 September 2011 Keywords: Zr4+ -doped Bi2 WO6 Oxygen vacancy Photocatalysis RhB Phenol

a b s t r a c t To overcome the drawback of low photocatalytic efficiency brought by electron–hole recombination, Bi2 WO6 photocatalysts with oxygen vacancies were synthesized by zirconium doping. The oxygen vacancies as the positive charge centers can trap the electron easily, thus inhibiting the recombination of charge carriers and prolonging the lifetime of electron. Moreover, the formation of oxygen vacancies favors the adsorption of O2 on the semiconductor surface, thus facilitating the reduction of O2 by the trapped electrons to generate superoxide radicals, which play a key role in the oxidation of organics. Visible-light-induced photodegradation of rhodamine B (RhB) and phenol were carried out to evaluate the photoactivity of the products. The results showed that oxygen-deficient Bi2 WO6 exhibited much enhanced photoactivity than the Bi2 WO6 photocatalyst free of oxygen deficiency. This work provided a new concept for rational design and development of high-performance photocatalysts. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Semiconductor-based photocatalysis has been attracting a great deal of attention due to its potential applications in renewable energy and environment fields such as dye-sensitized solar cells, hydrogen generation from water splitting and photocatalytic water/air purification [1–5]. Because these applications are based on the photogeneration of charge carriers such as electrons and holes, success in the applications relies on the transfer efficiency of electron or hole, which is closely related to the recombination rates of the photogenerated charge carriers. Unfortunately, due to the much faster recombination rate (nanoseconds) than the interfacial transfer rate (microseconds to milliseconds), many charge carriers recombine and dissipate the input energy as heat, which seriously limits the overall quantum efficiency for photocatalysis [6]. Therefore, to improve the photocatalytic activity of the semiconductors, it is important to control the recombination dynamics of the photogenerated charge carriers. If a suitable scavenger or surface defect state is available to trap the electron or hole, recombination is inhibited and ensuing redox reaction may occur. It was reported that oxygen vacancies may act as electron capture centers, and thus play an important role in retarding the recombination of charge carriers, which can lead to an enhanced photocatalytic activity of the photocatalysts [7–9]. Moreover, the existing oxygen vacancies

can act importantly as specific reaction sites for reactant molecules in heterogeneous reactions [10]. Therefore, introducing oxygen vacancies into the photocatalysts can be a feasible approach for developing highly active photocatalysts. As one of the simplest Aurivillius oxides with layered structure, Bi2 WO6 has recently attracted considerable attention for its good photocatalytic performance in water splitting and organic contaminant decomposing under visible light irradiation [11–15]. Up to now, much work has been done to facilitate the electron–hole separation and enhance the photocatalytic activity of Bi2 WO6 , including surface modification [16,17], anion doping [18], and coupling with other semiconductors [19,20]. However, to the best of our knowledge, the effect of oxygen vacancies on the photocatalytic activity of Bi2 WO6 has seldom been reported. Here for the first time we introduce oxygen vacancies into Bi2 WO6 through zirconium doping and the relationship between oxygen vacancies and the photocatalytic activity of Bi2 WO6 has been investigated. Bi2 WO6 does not contain oxygen vacancies and it was reported that substitution of W by appropriate cations with lower valence states could lead to an extrinsic oxygen deficiency by charge compensation [21,22]. Defect calculations show that the low solution energy (0.05 eV) is favorable for the substitution of ZrIV at WVI site with the creation of extrinsic oxygen vacancies [22], which may be described by defect reactions written as: WO

3 ZrO2 −→Zr

∗ Corresponding author. Tel.: +86 21 5241 5295; fax: +86 21 5241 3122. E-mail address: [email protected] (W. Wang). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.017



W

+ VO •• + 2OO

In this study, we succeeded in preparing oxygen deficient Bi2 WO6 phases by substitution for WVI with ZrIV . The photoactivity

256

Z. Zhang et al. / Journal of Hazardous Materials 196 (2011) 255–262

Fig. 1. (A) The XRD patterns of the as-synthesized products; (B) diffraction peak positions of the (1 3 1) plane in the range of 2 = 27.5–29◦ .

evaluation, via the photocatalytic degradation of RhB and phenol under visible light, demonstrated that the photocatalytic activity is dependent on the concentration of oxygen vacancy, and the Zr4+ -doped Bi2 WO6 exhibit much better photocatalytic performance than undoped Bi2 WO6 sample. Moreover, the role of oxygen vacancy in promoting the separation of charge carriers and enhancing the photocatalytic activities is elucidated in detail.

2. Experimental 2.1. Preparation of Zr4+ -doped Bi2 WO6 photocatalysts The Zr4+ -doped Bi2 WO6 photocatalysts were prepared by a hydrothermal method. In a typical process, 2 mmol of Bi(NO3 )3 ·5H2 O and 1 mmol of Na2 WO4 ·2H2 O were dissolved in 2 mL of 2 M nitric acid and 30 mL of deionized water, respectively. After that, these two solutions were mixed together and stirred for 30 min. Then aqueous solution containing desired amounts of ZrOCl2 ·8H2 O was added for Zr4+ -doped Bi2 WO6 . The molar ratios of Zr to Bi2 WO6 were set as 0, 2.0%, 3.0% and 4.0%, respectively, and the corresponding products were named as Zr-0, Zr-0.02, Zr0.03 and Zr-0.04. The pH value of the final suspension was adjusted to about 7 and the mixture was stirred for several hours at room temperature. Afterward, the suspensions were added into a 50 mL Teflon-lined autoclave up to 80% of the total volume. The autoclave was sealed in a stainless steel tank and heated at 160 ◦ C for 24 h. Subsequently, the autoclave was cooled to room temperature naturally. The products were collected by filtration, washed with distilled water for several times, and then dried at 60 ◦ C in air for 12 h.

2.2. Characterization The phase and composition of the as-prepared samples were measured by X-ray diffraction (XRD) studies using an X-ray diffractometer with Cu K␣ radiation under 40 kV and 100 mA and with the 2 ranging from 20◦ to 60◦ (Rigaku, Japan). The morphologies and microstructures of the as-prepared samples were investigated by transmission electron microscopy (TEM, JEOL JEM-2100F). UV–vis diffuse reflectance spectra (DRS) of the samples were recorded with an UV–vis spectrophotometer (Hitachi U-3010) using BaSO4 as reference. Chemical compositions of the derived products were analyzed using X-ray photoelectron spectroscopy (XPS) analysis (Thermo Scientific Escalab 250). All binding energies were referenced to the C 1s peak (284.8 eV) arising from adventitious carbon. The photoluminescence (PL) spectra of the samples were recorded

with a Perkin Elmer LS55. Total organic carbon (TOC) analysis was carried out with an elementar liqui TOC II analyzer. 2.3. Photocurrent measurement Photocurrent measurements were carried out by using a CHI 660C electrochemical workstation. 25 mg of photocatalyst was suspended in de-ionized water (50 mL) containing acetate (0.1 M) and Fe3+ (0.1 mM) as an electron donor and acceptor, respectively. A Pt plate (both sides exposed to solution), a saturated calomel electrode (SCE), and a Pt gauze were immersed in the reactor as working (collector), reference, and counter electrodes, respectively. Photocurrents were measured by applying a potential (+1 V vs SCE) to the Pt electrode using a potentiostat (EG&G). 2.4. Measurement of photocatalytic activities Photocatalytic activities of the Zr4+ -doped Bi2 WO6 photocatalysts were measured by monitoring photo-degradation of rhodamine B (RhB) and phenol in aqueous solution. 100 mg of the photocatalysts were dispersed in a 100 mL solution of RhB (10−5 mol/L) or phenol (20 mg/L). Before illumination, the suspensions were magnetically stirred in the dark for 1 h to ensure adsorption/desorption equilibrium of RhB or phenol with the photocatalyst powders, and then exposed to visible light from a 500 W Xe lamp with a 420 nm cutoff filter. After a certain period of irradiation, 3 mL suspension was sampled and centrifuged to remove the photocatalysts. After that, the supernatant was taken out to measure the absorption spectral change of RhB or phenol through a UV–vis spectrophotometer (Hitachi U-3010) to monitor the photodegradation rate. The concentration change of rhodamine B and phenol were determined by monitoring the optical intensity of absorption spectra at 553 nm and 270 nm, respectively. 3. Results and discussion 3.1. Crystal structure and morphology of the products The XRD diffraction patterns of the pure Bi2 WO6 and Zr4+ doped Bi2 WO6 samples are shown in Fig. 1(A). All of the diffraction peaks match the standard data for a Russellite Bi2 WO6 structure (JCPDS 39-0256), and no characteristic peaks of any impurities are detected in the patterns, which demonstrates that doping with zirconium does not result in the development of new phases. However, a careful comparison of the (1 3 1) diffraction peaks in the range of 2 = 27.5–29◦ (Fig. 1(B)) shows that the peak position of Bi2 WO6 shifts slightly toward a lower 2 value with the increase

Z. Zhang et al. / Journal of Hazardous Materials 196 (2011) 255–262

257

Fig. 2. TEM micrograph of (A) Zr-0 and (B) Zr-0.03.

of zirconium contents. The same results are also presented in other diffraction peaks. According to Bragg’s law, d(h k l) = /(2 sin ), where d(h k l) is the distance between crystal planes of (h k l),  is the X-ray wavelength, and  is the diffraction angle of the crystal plane (h k l) [23], the decrease in 2 value should result from the increase in lattice parameters (d(1 3 1) value). Because the ionic radius of Zr4+ (0.080 nm) is smaller than that of Bi3+ (0.108 nm) but larger than that of W6+ (0.062 nm), the observed shift of diffraction peak toward lower angles should be due to the larger lattice parameter expected for substitution of W6+ by Zr4+ . In order to obtain detailed information about the microstructure and morphology of the as-synthesized samples, TEM observations

are carried out. Fig. 2(A) and (B) shows the representative TEM images of pure Bi2 WO6 sample and Bi2 WO6 sample doped with a zirconium content of 3.0 mol%, respectively. Both samples exhibit sheet-like morphology, which indicates that zirconium doping has no obvious influence on the morphology of Bi2 WO6 . 3.2. X-ray photoelectron spectroscopic (XPS) analysis The surface composition and elementary oxidation states of the as-prepared sample with a zirconium content of 3.0 mol% is investigated using XPS analysis, and the corresponding experiment results are shown in Fig. 3. The overall XPS spectra shown in Fig. 3(A)

Fig. 3. XPS spectra of Zr-0.03. (A) The overall XPS spectra of the sample; (B) Bi 4f spectrum; (C) W 4f spectrum and (D) Zr 3d spectrum.

258

Z. Zhang et al. / Journal of Hazardous Materials 196 (2011) 255–262

Fig. 4. High-resolution XPS spectra of O 1s for (A) Zr-0 and (B) Zr-0.03.

indicates that all of the peaks on the curve are ascribed to Bi, W, O, Zr and C elements and no peaks of other elements are observed. The presence of C comes mainly from carbon tape used for XPS measurement. Parts B–D of Fig. 3 display the high-resolution spectrum for Bi, W and Zr species, respectively. According to Fig. 3(B), the binding energies of Bi 4f7/2 and Bi 4f5/2 are 159.2 eV and 164.4 eV, respectively, which correspond to the characteristic peak of Bi3+ . The W 4f orbital is clearly resolved into W 4f5/2 and W 4f7/2 contributions, centered upon 37.3 eV and 35.2 eV, respectively (Fig. 3(C)), which are very close to previously reported values [19], suggesting that the tungsten in the Zr4+ -doped Bi2 WO6 sample exists as W6+ . The Zr 3d spectra shown in Fig. 3(D) consists of Zr 3d3/2 and Zr 3d5/2 main peaks with a peak separation of 2.4 eV, which is in agreement with the literature data of Zr4+ [24]. Moreover, we investigated the presence of oxygen vacancies by the XPS spectra. The high-resolution O 1s XPS spectra of undoped Bi2 WO6 and Zr4+ -doped Bi2 WO6 samples were presented in Fig. 4(A) and (B), respectively. Both profiles are asymmetric and can be fitted to two Gaussian features, which are normally assigned as the low binding energy component (LBEC) and the high binding energy component (HBEC), indicating two different kinds of O species in the sample. The LBEC and HBEC can be attributed to the lattice oxygen and chemisorbed oxygen caused by the surface chemisorbed species such as hydroxyl and H2 O, respectively [25]. It has been previously reported that the HBEC component develops with the increase of oxygen vacancies [26], which can lead to the asymmetry of the main peak. The high-resolution O 1s XPS spectra indicate that the peak area of HBEC is obviously larger in the zirconium doped sample as compared to the undoped one. Moreover, the calculated ratios of the adsorbed oxygen to the lattice oxygen are 1.14 and 0.43 for the zirconium doped sample and undoped sample, respectively, which strongly suggests the presence of oxygen deficiencies in the zirconium doped sample.

associated with oxygen vacancies just below the conduction band minimum [27]. The oxygen vacancies are positive charges centers, which bound electrons easily. Excitation of the electrons from such local states to the conduction band can lead to better visible light absorbance. Therefore, with more zirconium dopant concentration, more oxygen vacancies are created, and optical absorption properties of the samples become stronger. 3.4. Photoluminescence spectra and photoelectrochemical measurements Since photoluminescence (PL) emission mainly results from the recombination of free carriers, PL spectra is useful in determining the migration, transfer, and recombination processes of the photogenerated electron–hole pairs in a semiconductor. A weaker PL intensity implies a low recombination rate of the electron–hole under light irradiation [28]. Fig. 6(A) shows the PL spectra of undoped Bi2 WO6 and Zr4+ -doped Bi2 WO6 (Zr-0.03) when the excitation wavelength was 300 nm. There was a significant decrease in the intensity of PL spectra of Zr4+ -doped Bi2 WO6 , which confirmed that zirconium doping could effectively inhibit the recombination of photogenerated charge carriers. Photocurrent reflects indirectly the semiconductor’s ability to generate and transfer the photogenerated charge carriers, which correlates with the photocatalytic activity [29]. To investigate the photo-induced charges separation efficiency of undoped and zirconium doped samples, the photocurrent measurement was carried out under visible light irradiation. As shown in Fig. 6(B), the Zr0.03 sample generates higher photocurrent than undoped Bi2 WO6 ,

3.3. Optical properties of the products The UV–vis diffuse reflectance spectra (DRS) of Zr4+ -doped Bi2 WO6 samples in comparison with pure Bi2 WO6 are shown in Fig. 5. Pure Bi2 WO6 sample presented the photoabsorption ability from the UV light region to the visible light with the wavelength shorter than 450 nm. It was noteworthy that the absorption onset of Zr4+ -doped Bi2 WO6 samples was red-shifted apparently. With the increasing zirconium doping amount, the visible light absorption intensity of the samples became stronger, which may be attributed to excitations of trapped electrons in localized states

Fig. 5. UV–vis diffuse reflectance spectra of the as-prepared samples.

Z. Zhang et al. / Journal of Hazardous Materials 196 (2011) 255–262

259

Fig. 6. (A) The room temperature photoluminescence (PL) spectrum of Zr-0.03 and Zr-0 (Ex = 300 nm); (B) photocurrent generated with visible light irradiation time over Zr-0.03 and Zr-0 suspended with acetate and Fe3+ .

which indicates that zirconium doping can effectively promote the charge carrier transfer and reduce the electron–hole recombination. 3.5. Photocatalytic performance of Zr4+ -doped Bi2 WO6 samples RhB, a hazardous compound as well as a common model pollutant, was chosen as a representative pollutant to evaluate the photocatalytic performance of the photocatalysts. The RhB concentration variation versus the reaction time in the presence of Zr4+ -doped Bi2 WO6 samples compared with pure Bi2 WO6 is plotted in Fig. 7(A). The results demonstrate that the photoactivity of the samples was strongly dependent on the zirconium doping concentration. With zirconium concentration increasing from 0 mol% to 3.0 mol%, the photoactivity of the samples for the RhB photodegradation was enhanced. When the zirconium concentration increased to 4.0 mol%, the photoactivity decreased as compared to that of 3.0 mol%, but still higher than that of undoped Bi2 WO6 . The maximum photoactivity was observed for Zr-0.03, which can degrade RhB completely in 20 min, while only 65.4% of RhB was degraded in the presence of pure Bi2 WO6 within the same time period. Moreover, the comparison of the apparent rate constant k in Fig. 7(B) demonstrated that Zr-0.03 had the highest k value in the photodegradation of RhB, while that of Zr-0.04 decreased compared with that of Zr-0.03. The reason can be interpreted as follows: appropriate amount of oxygen vacancies can trap the electrons, resulting in the holes free to diffuse to the semiconductor surface where oxidation of organic species can occur.

Therefore, appropriate content of oxygen vacancies will improve the photocatalytic process by separating the electron–hole pairs effectively. If it exceeded the optimum value, however, the oxygen vacancies would act as the recombination centers for the photoinduced electrons and holes, which is unfavourable to the photocatalytic performance [27,30]. When the Zr doping concentration was 3.0 mol%, appropriate content of oxygen vacancies were generated and this photocatalyst exhibited the highest photocatalytic activity. When the doping concentration of Zr was increased further, the excess oxygen vacancies generated led to a poor photocatalytic performance. Therefore, appropriate zirconium doping amount can significantly enhance the photocatalytic activity of Bi2 WO6 . Photocatalytic activities of the above-mentioned photocatalysts can be further tested by the degradation of some other organic compound, such as phenol that has no light absorption property in the visible light region and no photosensitization, as shown in Fig. 8. Obviously, upon visible-light irradiation, phenol is degraded more efficiently by Zr-0.03 than by pure Bi2 WO6 (Fig. 8(A)). About 62.5% and 14.2% degradation efficiency was reached within 120 min by Zr-0.03 and pure Bi2 WO6 , respectively. In addition, due to pseudofirst-order kinetics of phenol photodegradation on Bi2 WO6 , the apparent rate constant k is calculated to be 0.0012 min−1 and 0.0083 min−1 for pure Bi2 WO6 and Zr-0.03, respectively (Fig. 8(B)). In other words, the photocatalytic activity of Zr-0.03 is about 7 times that of pure Bi2 WO6 . In order to further investigate the photodegradation of phenol, total organic carbon (TOC), which has been widely used to evaluate

Fig. 7. (A) Photocatalytic degradation of RhB under visible light ( > 420 nm) as a function of irradiation time by the as-prepared samples; (B) the comparison of rate constant k.

260

Z. Zhang et al. / Journal of Hazardous Materials 196 (2011) 255–262

Fig. 8. (A) Photocatalytic degradation of phenol under visible-light irradiation by Zr-0.03 and Zr-0, respectively; (B) the comparison of rate constant k; (C) TOC removal efficiency during the course of photocatalytic degradation of phenol in the presence of Zr-0.03 and Zr-0, respectively; (D) cycling runs in the photocatalytic degradation of phenol under visible-light irradiation.

the degree of mineralization of organic species, was measured in the photodegradation process by the as-prepared samples under visible light, as shown in Fig. 8(C). The results confirm that phenol is steadily mineralized by the as-prepared samples. Moreover, the TOC removal efficiency in the presence of Zr-0.03 reaches a value of 29.2% after 120 min of irradiation, while that of Bi2 WO6 is only 5.3%. Based on the above results, it can be deduced that Zr-0.03 is a much superior photocatalyst to pure Bi2 WO6 . To check the stability of the Zr4+ -doped Bi2 WO6 photocatalyst, the circulating runs in the photocatalytic degradation of phenol were performed under visible light. As shown in Fig. 8(D), after five recycles for the photodegradation of phenol, the catalyst did not exhibit any significant loss of activity, confirming the Zr4+ -doped Bi2 WO6 is not photocorroded during the photocatalytic oxidation of the pollutant molecules, which is especially important for its application.

groups or H2 O to form surface-bound hydroxyl radicals (• OH) and the conduction band electrons can interact with adsorbed O2 to form superoxide radicals (O2 •− ), which are both strong oxidative species and play crucial roles in the oxidative degradation of organics [32,33]. However, for the Bi2 WO6 system, the holes could not react with OH− /H2 O to form • OH due to the more negative redox potential of BiV /BiIII (+1.59 V) than that of • OH/OH− (+1.99 V) [34]. In order to ascertain the active species in the degradation process, holes and hydroxyl radicals scavengers were added into the degradation system. Fig. 9 showed that the addition of isopropanol (IPA) as hydroxyl radicals scavenger [34] caused a minor change in the photocatalytic degradation of phenol, indicating that • OH is not the major oxidation species in this process. However, when holes scavenger, EDTA [35] was introduced, the degradation rates of phenol

3.6. Mechanism of enhanced photoactivities Photocatalysis generally involves four processes [31]: (i) lightinduced generation of conduction band electrons and valence band holes; (ii) transfer of the photogenerated charge carriers to the photocatalyst surface; (iii) subsequent reduction/oxidization of the adsorbed reactants directly by electrons/holes or indirectly by reactive oxygen species; and (iv) recombination of the photogenerated electron–hole pairs. The photocatalysis efficiency is determined by the competition between the charge separation process and the charge recombination process. Desired photocatalysts are expected to promote the charge transfer processes while suppressing recombination process. Typically, for photocatalysts in an aqueous solution, the photoinduced valence band holes can react with chemisorbed hydroxyl

Fig. 9. Photocatalytic degradation of phenol with the addition of hole and hydroxyl radical scavengers and in N2 -saturated solutions under visible light irradiation ( > 420 nm).

Z. Zhang et al. / Journal of Hazardous Materials 196 (2011) 255–262

261

more superoxide anions on the photocatalyst surface. Generation of superoxide anions was a process of trapping the photoinduced electrons, which facilitated the charge separation and resulted in a lower electron–hole recombination rate. Our work suggests that the idea of oxygen vacancies introduction can be a plausible strategy to develop efficient visible-light-driven photocatalysts for environmental remediation. Acknowledgements

Fig. 10. Proposed photocatalytic mechanism of oxygen deficient Bi2 WO6 . OP: organic pollutant; DP: degradation product.

were depressed to a large extent. Therefore, holes play an important part in Bi2 WO6 photocatalysis. The superoxide radical is another important intermediate for oxidative degradation of organics [36,37]. In order to examine the role of superoxide radical in the photocatalysis, photocatalytic degradation of phenol was carried out under N2 -saturated conditions. The result shown in Fig. 9 indicated that under the anoxic condition, the photodegraded rate of phenol was largely suppressed, suggesting that superoxide radical is an important oxidation species in the photocatalytic process. Oxygen molecules as the electron scavengers play a crucial role in photocatalysis by reacting with electrons to generate superoxide radicals. However, when the rate of O2 reduction by electrons is not sufficiently fast to match the rate of reaction of holes, an excess of electrons will accumulate on the photocatalyst particles, and the electron–hole recombination rate will increase consequently. In this case, electrons transfer to O2 may be the rate limiting step in photocatalysis [38,39]. This, however, can be overcome by the introduction of oxygen vacancies into the photocatalyst. In order to facilitate the reaction between oxygen and electrons, strong oxygen adsorption on the photocatalyst surface and longer lifetime of electrons are indispensable. It was reported that adsorption of O2 molecules is mainly mediated by oxygen vacancies and oxygen physisorbs on defect-free oxide surfaces but interacts strongly with oxygen vacancies [40,41]. So the adsorption of oxygen can be promoted by the formation of oxygen vacancies. On the other hand, if an electron is freely mobile in the semiconductor particle, it is hardly possible for it to escape destiny of recombination. However, this can be prevented if the electrons are transiently but efficiently trapped in the particles. The positively charged VO •• defects can work as electron acceptors and can trap the photogenerated electrons temporarily to reduce the surface recombination of electrons and holes [8,9]. Therefore, the creation of oxygen vacancies can not only favor oxygen adsorption but also retard the recombination of charge carriers, which facilitates the O2 reduction rate to generate more superoxide anions (O2 •− ) on the photocatalyst surface (Fig. 10), and thus lead to an enhanced photocatalytic activity. 4. Conclusions Oxygen-deficient Bi2 WO6 photocatalysts were synthesized by zirconium doping, and the relationship between oxygen vacancies and photocatalytic activities of Bi2 WO6 was investigated. The visible-light-induced photo-degradation of RhB and phenol demonstrated that zirconium doping could significantly enhance the photocatalytic performance of Bi2 WO6 . The higher photocatalytic activity was attributed to the formation of oxygen vacancies, which promote the O2 adsorption and O2 reduction rate, leading to

We acknowledge the financial support from the National Natural Science Foundation of China (50972155, 50902144, 50732004), National Basic Research Program of China (2010CB933503) and Science Foundation for Youth Scholar of State Key Laboratory of High Performance Ceramics and Superfine Microstructures (SKL 200904). References [1] A.L. Linsebigler, G.Q. Lu, J.T. Yates, Photocatalysis on TiO2 surfaces: principles mechanisms, and selected results, Chem. Rev. 95 (1995) 735–758. [2] A. Hagfeldt, M. Gratzel, Light-induced redox reactions in nanocrystalline systems, Chem. Rev. 95 (1995) 49–68. [3] X.B. Chen, S.S. Mao, Titanium dioxide nanomaterials: synthesis, properties, modifications, and applications, Chem. Rev. 107 (2007) 2891–2959. [4] W. Morales, M. Cason, O. Aina, N.R. Tacconi, K. Rajeshwar, Combustion synthesis and characterization of nanocrystalline WO3 , J. Am. Chem. Soc. 130 (2008) 6318–6319. [5] W.J. Youngblood, S.H.A. Lee, K. Maeda, T.E. Mallouk, Visible light water splitting using dye-sensitized oxide semiconductors, Acc. Chem. Res. 42 (2009) 1966–1973. [6] M.R. Hoffmann, S.T. Martin, W.Y. Choi, D.W. Bahnemann, Environmental applications of semiconductor photocatalysis, Chem. Rev. 95 (1995) 69–96. [7] T.J. Kuo, C.N. Lin, C.L. Kuo, M.H. Huang, Growth of ultralong ZnO nanowires on silicon substrates by vapor transport and their use as recyclable photocatalysts, Chem. Mater. 19 (2007) 5143–5147. [8] J.C. Wang, P. Liu, X.Z. Fu, Z.H. Li, W. Han, X.X. Wang, Relationship between oxygen defects and the photocatalytic property of ZnO nanocrystals in nafion membranes, Langmuir 25 (2009) 1218–1223. [9] Y.H. Zheng, C.Q. Chen, Y.Y. Zhan, X.Y. Lin, Q. Zheng, K.M. Wei, J.F. Zhu, Y.J. Zhu, Luminescence and photocatalytic activity of ZnO nanocrystals: correlation between structure and property, Inorg. Chem. 46 (2007) 6675–6682. [10] X.Q. Gong, A. Selloni, M. Batzill, U. Diebold, Steps on anatase TiO2 (1 0 1), Nat. Mater. 5 (2006) 665–670. [11] A. Kudo, S. Hijii, H2 or O2 evolution from aqueous solutions on layered oxide photocatalysts consisting of Bi3+ with 6s2 configuration and d0 transition metal ions, Chem. Lett. 28 (1999) 1103–1104. [12] C. Zhang, Y.F. Zhu, Synthesis of square Bi2 WO6 nanoplates as high-activity visible-light-driven photocatalysts, Chem. Mater. 17 (2005) 3537–3545. [13] H.B. Fu, L.W. Zhang, W.Q. Yao, Y.F. Zhu, Nanosized Bi2 WO6 catalysts prepared by hydrothermal synthesis and their photocatalytic properties, Appl. Catal. B: Environ. 66 (2006) 100–110. [14] F. Amano, A. Yamakata, K. Nogami, M. Osawa, B. Ohtani, Visible light responsive pristine metal oxide photocatalyst: enhancement of activity by crystallization under hydrothermal treatment, J. Am. Chem. Soc. 130 (2008) 17650–17651. [15] G.S. Li, D.Q. Zhang, J.C. Yu, M.K.H. Leung, An efficient bismuth tungstate visiblelight-driven photocatalyst for breaking down nitric oxide, Environ. Sci. Technol. 44 (2010) 4276–4281. [16] J. Ren, W.Z. Wang, S.M. Sun, L. Zhang, J. Chang, Enhanced photocatalytic activity of Bi2 WO6 loaded with Ag nanoparticles under visible light irradiation, Appl. Catal. B: Environ. 92 (2009) 50–55. [17] Y.Y. Li, J.P. Liu, X.T. Huang, J.G. Yu, Carbon-modified Bi2 WO6 nanostructures with improved photocatalytic activity under visible light, Dalton Trans. 39 (2010) 3420–3425. [18] M. Shang, W.Z. Wang, L. Zhang, H.L. Xu, Bi2 WO6 with significantly enhanced photocatalytic activities by nitrogen doping, Mater. Chem. Phys. 120 (2010) 155–159. [19] Q. Xiao, J. Zhang, C. Xiao, X.K. Tan, Photocatalytic degradation of methylene blue over Co3 O4 /Bi2 WO6 composite under visible light irradiation, Catal. Commun. 9 (2008) 1247–1253. [20] L.S. Zhang, K.H. Wong, Z.G. Chen, J.C. Yu, J.C. Zhao, C. Hu, C.Y. Chan, P.K. Wong, AgBr–Ag–Bi2 WO6 nanojunction system: a novel and efficient photocatalyst with double visible-light active components, Appl. Catal. A 363 (2009) 221–229. [21] C.K. Lee, L.T. Sim, A.M. Coats, A.R. West, On possible Cu doping of Bi2 WO6 , J. Mater. Chem. 11 (2001) 1096–1099. [22] M.S. Islam, S. Lazure, R.N. Vannier, G. Nowogrocki, G. Mairesse, Structural and computational studies of Bi2 WO6 based oxygen ion conductors, J. Mater. Chem. 8 (1998) 655–660. [23] G.L. Huang, Y.F. Zhu, Enhanced photocatalytic activity of ZnWO4 catalyst via fluorine doping, J. Phys. Chem. C 111 (2007) 11952–11958.

262

Z. Zhang et al. / Journal of Hazardous Materials 196 (2011) 255–262

[24] I. Atribak, A.B. Lopez, A.G. Garcia, B. Azambre, Contributions of surface and bulk heterogeneities to the NO oxidation activities of ceria–zirconia catalysts with composition Ce0.76 Zr0.24 O2 prepared by different methods, Phys. Chem. Chem. Phys. 12 (2010) 13770–13779. [25] Y.H. Zheng, L.R. Zheng, Y.Y. Zhan, X.Y. Lin, Q. Zheng, K.M. Wei, Ag/ZnO heterostructure nanocrystals: synthesis characterization and photocatalysis, Inorg. Chem. 46 (2007) 6980–6986. [26] M. Naeem, S.K. Hasanain, M. Kobayashi, Y. Ishida, A. Fujimori, S. Buzby, S.I. Shah, Effect of reducing atmosphere on the magnetism of Zn1−x Cox O (0 ≤ x ≤ 0.10) nanoparticles, Nanotechnology 17 (2006) 2675–2680. [27] J. Wang, D.N. Tafen, J.P. Lewis, Z.L. Hong, A. Manivannan, M.J. Zhi, M. Li, Ni.Q. Wu, Origin of photocatalytic activity of nitrogen-doped TiO2 nanobelts, J. Am. Chem. Soc. 131 (2009) 12290–12297. [28] K. Fujihara, S. Izumi, T. Ohno, M. Matsumura, Time-resolved photoluminescence of particulate TiO2 photocatalysts suspended in aqueous solutions, J. Photochem. Photobiol. A: Chem. 132 (2000) 99–104. [29] H.G. Kim, P.H. Borse, W.Y. Choi, J.S. Lee, Photocatalytic nanodiodes for visiblelight photocatalysis, Angew. Chem. Int. Ed. 44 (2005) 4585–4589. [30] H.H. Wang, C.S. Xie, The effects of oxygen partial pressure on the microstructures and photocatalytic property of ZnO nanoparticles, Physica E 40 (2008) 2724–2729. [31] N.Q. Wu, J. Wang, D.N. Tafen, H. Wang, J.G. Zheng, J.P. Lewis, X.G. Liu, S.S. Leonard, A. Manivannan, Shape-enhanced photocatalytic activity of singlecrystalline anatase TiO2 (1 0 1) nanobelts, J. Am. Chem. Soc. 132 (2010) 6679–6685. [32] K. Ishibashi, A. Fujishima, T. Watanabe, K. Hashimoto, Generation and deactivation processes of superoxide formed on TiO2 film illuminated by very weak UV light in air or water, J. Phys. Chem. B 104 (2000) 4934–4938.

[33] A.M. Peiro, C. Colombo, G. Doyle, J. Nelson, A. Mills, J.R. Durrant, Photochemical reduction of oxygen adsorbed to nanocrystalline TiO2 films: a transient absorption and oxygen scavenging study of different TiO2 preparations, J. Phys. Chem. B 110 (2006) 23255–23263. [34] H.B. Fu, C.S. Pan, W.Q. Yao, Y.F. Zhu, Visible-light-induced degradation of rhodamine B by nanosized Bi2 WO6 , J. Phys. Chem. B 109 (2005) 22432–22439. [35] F. Zhou, R. Shi, Y.F. Zhu, Significant enhancement of the visible photocatalytic degradation performances of ␥-Bi2 MoO6 nanoplate by graphene hybridization, J. Mol. Catal. A: Chem. 340 (2011) 77–82. [36] S. Kim, H. Park, W. Choi, Comparative study of homogeneous and heterogeneous photocatalytic redox reactions: PW12 O40 3− vs. TiO2 , J. Phys. Chem. B 108 (2004) 6402–6411. [37] L. Cermenati, P. Pichat, C. Guillard, A. Albini, Probing the TiO2 photocatalytic mechanism in water purification by use of quinoline, photo-fenton generated OH radicals and superoxide dismutase, J. Phys. Chem. B 101 (1997) 2650–2658. [38] C.M. Wang, A. Heller, H. Gerischer, Palladium catalysis of O2 reduction by electrons accumulated on TiO2 particles during photoassisted oxidation of organic compounds, J. Am. Chem. Soc. 114 (1992) 5230–5234. [39] H. Gerischer, A. Heller, The role of oxygen in photooxidation of organic molecules on semiconductor particles, J. Phys. Chem. 95 (1991) 5261–5267. [40] D. Pillay, Y. Wang, G.S. Hwang, Prediction of tetraoxygen formation on rutile TiO2 (1 1 0), J. Am. Chem. Soc. 128 (2006) 14000–14001. [41] G.A. Kimmel, N.G. Petrik, Tetraoxygen on reduced TiO2 (1 1 0): oxygen adsorption and reactions with bridging oxygen vacancies, Phys. Rev. Lett. 100 (2008) 196102.

Journal of Hazardous Materials 196 (2011) 295–301

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Monitoring of PCBs at facilities related with PCB-containing products and wastes in South Korea Guang-Zhu Jin a,1 , Ming-Liang Fang a , Jung-Ho Kang a , Hyokeun Park a , Sang-Hyup Lee b , Yoon-Seok Chang a,b,∗ a b

School of Environmental Science and Engineering, Pohang University of Science and Technology (POSTECH), San 31, Hyoja-dong, Nam-gu, Pohang 790-784, Republic of Korea Water Research Center, Korea Institute of Science and Technology (KIST), Hwarangno, 14-Gil 5, Seongbuk-gu, Seoul 136-791, Republic of Korea

a r t i c l e

i n f o

Article history: Received 1 March 2011 Received in revised form 8 September 2011 Accepted 8 September 2011 Available online 28 September 2011 Keywords: PCBs Inventory Emission factor South Korea

a b s t r a c t Polychlorinated biphenyl (PCB) contents were analyzed in samples collected from facilities related to PCB-containing products or wastes in South Korea. Average concentrations of the atmospheric 209 PCBs were 7420 (37.0–104,048) pg m−3 and 16.8 (ND–34.2) fg WHO-TEQ m−3 in indoor air samples; and 1670 (106–13,382) pg m−3 and 5.64 (ND–36.0) fg WHO-TEQ m−3 in outdoor air samples. The highest levels were observed in indoor air samples from disposal facilities (7336–104,048 pg m−3 ), followed by production (330–25,057 pg m−3 ), recycling, and storage facilities, indicating that PCB emissions from PCB-containing products and wastes remains very high and the facilities related with those may be an important source to atmospheric PCBs. Principal component analysis of PCB profiles showed that the homologue patterns of PCBs in outdoor and indoor air samples collected from the facilities were similar to those of boundary air samples and PCB commercial products, e.g. Aroclor 1016, 1221, 1232 and 1242. Evaluation of the PCB mass balance in a facility, dismantling and solvent-washing PCB-contaminated transformers, showed that of the total PCBs treated in this facility, approximately 0.0022% was emitted to the atmosphere, and most was transferred to waste oil for disposal by incineration or chemical methods. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Polychlorinated biphenyls (PCBs) are classified and regulated as one of the 12 persistent organic pollutants (POPs) under the Stockholm Convention on POPs [1]. The sources of PCBs can be divided into two major categories: intentional chemicals produced in the chemical industries, and unintentionally de novo synthesized by-products during thermal processes [2,3]. The production and consumption of global PCBs for industrial purposes are relatively well established. PCBs were mostly produced commercially from 1929 to the early 1970s. During this period, total global production of PCBs was estimated approximately 1.3 million tons [4]. The commercial PCBs are known by a variety of trade names, such as Aroclor (USA, UK), Kanechlor (Japan), Sovol (Russia), Chlophen (Germany, Poland), and Phenoclor (France) [5,6].

∗ Corresponding author at: School of Environmental Science and Engineering, Pohang University of Science and Technology (POSTECH), San 31, Hyoja-dong, Namgu, Pohang 790-784, Republic of Korea. Tel.: +82 54 279 2281; fax: +82 54 279 8299. E-mail address: [email protected] (Y.-S. Chang). 1 Present address: Key Laboratory of Nature Resource of the Changbai Mountain and Functional Molecular (Yanbian University), Ministry of Education, 133002, Park Road 977, Yanji, Jilin Province, China. 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.030

In South Korea, industrial PCB mixtures have never been produced and their use in electronic equipments has been banned since 1979, and their import and use was completely banned in 1996. Kim et al. [7] reported that the ambient air in Korea was more influenced by combustion processes than that in Japan and also the contribution of PCB commercial products was relatively small. PCB levels in iron and steel complexes in South Korea have been reported to be higher than those in residential areas, indicating that iron and steel complexes are probably an important source of PCBs [8]. However, the emission of PCBs caused by de novo synthesis is not believed to contribute significantly to the global historical PCB mass balance [9]. The relative importance of atmospheric emissions from various source categories is not well known with considerable uncertainty [10]. Jamshidi et al. [11] reported that the principal contemporary source of PCBs in UK conurbation was ventilation of indoor air and not volatilization from soil. According to the Korean Law, wastes that contain PCBs (>0.0001 mg kg−1 in solids or >0.01 mg kg−1 in liquids) are considered as “PCB-containing wastes” which must be treated by specialized methods [12]. Recycling of PCB-containing wastes only limits for wastes which contain less than 2 mg kg−1 PCBs. In 2007, the amount of PCB-containing wastes generated in South Korea that were contaminated with >2 mg kg−1 PCBs was 2543 tons [13]. Therefore, the emission of PCBs from PCB-containing products and

296

G.-Z. Jin et al. / Journal of Hazardous Materials 196 (2011) 295–301

Fig. 1. PCB levels in air samples collected using HVAS from facilities related to PCB-containing products or wastes. Cross bars are mean values and vertical represent maximum and minimum concentrations.

wastes remains very high, even 30 years after PCB production ceased. Emission inventories are essential for identifying, evaluating, and prioritizing sensible control strategies on a regional or a global scale [1,14]. Also, to understand and predict the long range transport features and environmental fates of these substances, quantitative information on their atmospheric releases is deemed essential [15,16]. Hosomi et al. have estimated volatilization of PCBs from PCB-containing ballast in a fluorescent lamp [17]. However, few studies have investigated PCB contamination in facilities related to PCB-containing products and wastes rather than unintentional sources such as incinerators; no such evaluation has been performed in South Korea. In this study, we investigated atmospheric levels and distribution of PCBs in facilities related to PCB-containing products and wastes. These facilities include production, in-use, recycling, storage and disposal facilities across South Korea. We also evaluated PCB emission factors and the mass balance in a PCB disposal facility. The emission of PCBs caused by de novo synthesis was not considered in this study. This is the first study to investigate PCB emissions from facilities related to PCB-containing products and wastes in South Korea, providing valuable data in planning for comprehensive management and final elimination of PCB-containing products and wastes. 2. Experiment and method 2.1. Sampling Air samples were collected from 44 sites (9 production, 8 inuse, 14 storage, 10 recycling, 1 disposal, and 2 boundary) related to PCB-containing products or wastes across South Korea from October 2007 to July 2008 (Fig. S1). The specific information of sampling was shown in the Table S1. Samples were collected using high volume air sampling (HVAS, DHA-1000S, SIBATA). A glass fiber filter (GFF) and two consecutive polyurethane foam plugs (PUF) were used to collect airborne particles and vaporphase PCBs, respectively. Before sampling, the GFFs were baked at 450 ◦ C for 12 h, and the PUF disks were Soxhlet extracted for 16 h with acetone, then for 16 h with dichloromethane, then dried in a desiccator under vacuum for 24 h. A total of 20 outdoor air samples and 39 indoor air samples were collected for 24 h and at a flow rate of 700 L/min. It is important to note that

room sizes of a few in-use facilities were smaller than collected air volumes (1000 m3 ). Therefore, PCB concentrations could be underestimated by dilution effects in those small facilities. Outdoor air samples were collected within 5 m from the facilities (or rooms). Boundary PCB concentrations were measured at sites situated at the boundaries (500–800 m) of facilities. An additional 89 bottom samples were collected at the 37 facilities by wiping floor dust with hexane rinsed glass wool. At each site, 1–3 bottom samples were collected according to its facility size. For mass balance case study, several final product samples, such as copper, silicon steel plate, waste paper, waste oil etc., were collected from a dismantling and cleaning facility of PCB-containing wastes. 2.2. Analytical methods In the laboratory, samples were treated, extracted and analyzed according to the methods established at the US EPA’s method 1668A [18]. Briefly, the samples were spiked with the internal standard containing 27 13 C-labled PCB congeners (1, 3, 4, 15, 19, 37, 54, 77, 81, 104, 105, 114, 118, 123, 126, 155, 156, 157, 167, 169, 188, 189, 202, 205, 206, 208, and 209) (Wellington, 1668-LCS), then Soxhlet-extracted for 24 h using toluene. The extracts were then washed with concentrated H2 SO4 followed by hexane-saturated H2 O. Sample cleanup was performed using multi-layer silica and florisil columns. The eluent was reduced to 0.5 mL by rotary evaporation and a gentle stream of N2 gas. Finally, the extracts were transferred to GC vials, and 13 C-labled PCBs (9, 28, 52, 101, 111, 138, 178, and 194) were added as recovery standards. PCB contents were analyzed using an Agilent Hewlett-Packard 6890 gas chromatograph/Jeol JMS-700T high resolution mass spectrometer (GC/HRMS) with a DB-5MS column (J&W Scientific, 60 m length, 0.25 mm ID, 0.25 ␮m film thickness). The instrument was operated using He as the carrier gas with a constant flow of 1 mL min−1 . The temperature program of the GC oven was as follows: the temperature was held at the initial value of 110 ◦ C for 2 min, then raised at 40–200 ◦ C min−1 and held for 3 min, then raised at 2–230 ◦ C min−1 , then raised at 7–300 ◦ C min−1 and held for 7 min. 1 ␮L sample was injected at a temperature of between 280 and 300 ◦ C for the analysis of PCB contents. The GC/HRMS was operated under positive EI conditions (38 eV) with a resolution of 10,000. Data were obtained in the selected ion monitoring (SIM) mode.

G.-Z. Jin et al. / Journal of Hazardous Materials 196 (2011) 295–301

297

Fig. 2. PCB levels in bottom samples. Cross bars are mean values and vertical represent maximum and minimum concentrations.

Peak assignment was conducted to quantify 209 PCB congeners, but typically only 120 PCB congeners were detected. Bottom samples were analyzed using a GC with an electron capture detector (HP 6890, Agilent) following the Korean waste official method [19].

3. Results and discussion 3.1. PCB levels Detected levels of PCBs were generally lower in outdoor air samples than in indoor air samples, although the range was very large (Table 1, Fig. 1). The mean PCB concentration (209 PCBs) in outdoor air samples was 1670 pg m−3 (5.64 fg WHO-TEQ m−3 ) and ranged from 106 pg m−3 at a PCB-containing waste storage site to 13,400 pg m−3 at a PCB disposal (dismantling and cleaning) facility. These PCB concentrations in outdoor air samples were consistent with the PCB levels in the ambient air of South Korea in a previous study [7] and comparable with those in global urban sites (mean: 1700 pg m−3 ) [20]. The mean PCB concentrations of indoor air samples were 7420 pg m−3 (16.8 fg WHO-TEQ m−3 ) and ranged from 37 pg m−3 at an indoor transformer site (containing 46.6 tons of transformer oil contaminated with 0.15 mg kg−1 PCBs) to 104,048 pg m−3 at a PCB disposal facility. PCB levels in both indoor and outdoor air samples were highest at the disposal facility followed by the production facility, the recycling facility, the storage site and the in-use site. The high concentration at the PCB disposal facility might be due to the volatilization of

2.3. Quality assurance/quality control Several steps were taken to obtain data that would allow an assessment of the accuracy and reliability of the data. Analytical blanks were included at a rate of one per 10 samples. All data have been blank corrected. The average recoveries of 27 13 C-labled PCB congeners ranged from 25 to 93% (Table S2), which satisfied the criteria (25–150%) recommended by US EPA method 1668A. Recovery statistics are given in Table S6. The method detection limit was calculated as 3 times the standard deviation of seven blank replicate samples (Table S3). The criteria for the quantification of analytes were as follows: retention time within 2 s of that of the standard, isotope ratio within 20% of that of the standard, and signal-to-noise ratio ≥3.

Table 1 PCB levels (209 PCBs) in indoor and outdoor air samples from different sites using HVAS. Site type

Sample type

Mean PCB concentration 3

Boundary Production In use (indoor) In use (indoor) In use (outdoor) Storage (indoor) Storage (indoor) Storage (outdoor) Recycling Recycling Disposal Disposal Background Industrial Residential In use condenser (containing PCBs) Industrial area Urban area Background

Outdoor (n = 2) Indoor (n = 9) Indoor (n = 5) Outdoor (n = 3) Outdoor (n = 3) Indoor (n = 11) Outdoor (n = 9) Outdoor (n = 1) Indoor (n = 10) Outdoor (n = 1) Indoor (n = 4) Outdoor (n = 2) Outdoor

Indoor Outdoor (n = 3) Outdoor (n = 3) Outdoor

References −3

(pg/m )

(fg WHO-TEQ m

153 (146–161) 8722 (330–25,057) 788 (37–2273) 815 (199–1159) 1017(671–1561) 1469 (353–5404) 790 (106–2527) 2539 (2539–2539) 5731 (160–17,710) 1667 (1667–1667) 33,692 (7336–104,048) 8257 (3131–13,382) 180–280 2080–5820 240–28,000 26,000–110,000

1.44 (0.029–2.85) 11.50 (0.073–31.0) 4.79 (0.157–12.3) 8.20 (ND–24.6) 8.76 (ND–14.9) 0.72 (ND–3.74) 6.81 (ND–36.0) 0.033 6.05 (0.018–41.3) ND 114 (ND–342) 2.31 (1.70–2.91)

21–27 19–46 0.6 (1–1.9)

) This study This study This study This study This study This study This study This study This study This study This study This study Kim et al. [7]

Hosomi [17] Martínez et al. [29] Menichini et al. [30]

298

G.-Z. Jin et al. / Journal of Hazardous Materials 196 (2011) 295–301

more-volatile PCB congeners during transformer dismantling and solvent washing. PCB levels were also investigated in 89 bottom samples collected from 36 facilities (Fig. 2). PCBs were detected in about 80% of bottom samples. PCB concentrations ranged from ND to 342 ng cm−2 and the highest level was found in samples from a transformer oil production site (ND–342 ng cm−2 ), followed by a disposal facility (4–152 ng cm−2 ). PCB concentrations of 26,000–110,000 pg m−3 have been observed in the indoor air of an office where fluorescent lamps with PCB-containing ballast had been used [17]; the PCB volatilization rates from this ballast were temperature-dependent and the PCB composition of the emission gas was similar to that observed in the ballast samples collected. Our samples were not collected to quantify the effect of temperature on PCB levels in indoor air samples; however, the highest PCB concentrations were observed in samples collected in a disposal facility during summer (104,000 pg m−3 in July vs. 7340 pg m−3 in October), followed by a production facility (25,100 pg m−3 ). In a major UK conurbation, the principal contemporary source of PCBs has been reported to be not the volatilization from soil but the ventilation of indoor air; existing structures, especially older buildings in which PCBs had been used in the past, were the major source of PCBs in outdoor air [11]. Generally, urban areas are more polluted by PCBs than rural areas [21]. Based on our data and previous studies, it seems that PCBs volatilized from the PCBcontaining products or wastes are important sources of PCBs in the ambient air in South Korea. Future reductions in PCB concentrations in the outdoor air and ultimately in human exposure may be best achieved by actions to these remaining sources of PCBs from PCB-containing products and wastes. In our air samples, the average contribution of gas phase PCBs to total PCBs was about 96%, which was consistent with the previous study of 24 PCB congeners in South Korea [22]. Gas-particle partitioning of PCBs in air samples from each type of facilities related to PCB-containing products or wastes showed similar patterns (Fig. S5). Gas phase contribution to total PCBs decreased from 89 (mono-CBs) to 24% (deca-CB) in indoor air samples, and from 92 (mono-CBs) to 22% (deca-CB) in outdoor air samples (Fig. S6). These results suggest that PCBs in the air exist predominantly in the gas phase and that the contribution of PCB congeners to the gas phase decreases as congeners become more-highly chlorinated (i.e., less volatile). 3.2. Homologue patterns The homologue patterns of PCBs in air samples were similar at all sampling sites (Fig. S2). In all air samples, the dominant PCB homologues found were low chlorinated PCBs such as mono-, diand tri-PCBs which accounted for about 14%, 35%, and 33% of total PCBs in indoor air samples and about 15%, 41%, and 30% of total PCBs in outdoor air samples, on average, respectively (Fig. S3). Many sources of PCBs can influence atmospheric PCB levels, including incinerators, industrial thermal processes, and PCBcontaining products and wastes [23]; homologue patterns can provide clues to where and how these substances originated [24]. PCBs in the outdoor air in this study were apparently influenced by the indoor air PCBs due to the higher levels of PCBs in the indoor environment. For further source identification, principal component analysis (PCA) of the data was conducted using SPSS 12.0 software (SPSS, Inc.) and homologue patterns of air samples from this study were compared to those of commercial mixtures of Aroclor 1016, 1221, 1232, 1242, 1248, 1254, 1260, 1262 and 1268 from other studies [6,25]. Total concentrations of each homologue PCBs (i.e. 1 Cl, 2 Cls, . . ., 10 Cls) were used for PCA analysis. PCB data were normalized by dividing by the total PCB concentrations for each sample, producing data ranging from 0 to 1. Finally, these normalized PCB compositions were used as input data for PCA. As a result,

PC1 and PC2 accounted for 60% of the total variance (Fig. 3). In the loading plot, the variables are well grouped by the number of chlorine. The homologue patterns of PCBs in air samples from various sites in this study were similar to commercial mixtures such as Aroclor 1016, 1221, 1232, and 1242, suggesting that the homologue patterns of many air samples were simultaneously influenced by these commercial mixtures. It is consistent with another previous study that the homologue patterns of PCBs found in sediments in South Korea indicated that their sources were commercial mixtures such as Aroclor 1016, 1242, 1254, and 1260 or corresponding Kanechlor products [26]. PCA was also used to compare homologue patterns of air samples in this study to those of ambient soil [27], incineration flue gas, and cement plant flue gas samples from other researches [6,28]. As a result, PC1 and PC2 accounted for 90% of the total variance (Fig. 4). The homologue patterns of PCBs in our air samples from various sites in South Korea were different from those of ambient soil, incineration flue gas, and cement plant flue gas samples, suggesting that there are other significant sources. However, all our air samples including boundary air samples had similar homologue patterns with the general ambient air samples (Korea, n = 15; Japan, n = 11) [7], indoor air samples from a disposal facility (Japan, n = 5), and indoor air samples of a room where PCB-containing sealant was used (Japan, n = 3) [6] (Fig. S4). Kim et al. [7] reported that the PCB levels in the ambient air of South Korea were more influenced by combustion processes than that in Japan, and also that the contribution of commercial PCB products was relatively small. However, our results strongly suggest that the ambient air in South Korea is contaminated by mixtures of commercial Aroclor products with various chlorine contents, particularly lowly chlorinated mixtures. 3.3. A case study of PCB mass balance To evaluate the PCB mass balance, a PCB disposal facility was selected, where PCBs in the indoor air showed the highest level. A series of air, bottom, and final product samples were collected from each process of dismantling and cleaning. This facility mainly treats waste transformers which contain PCB-contaminated transformer oil (>2 mg kg−1 ). The main processes are removal of transformer oil, dismantling of outer transformer cases, first extraction with toluene, dismantling of transformer inner assemblies, and second extraction with toluene. The final products are recycled (metals) or passed on for further disposal (waste oil). The mean PCB concentrations were measured as 14,560 pg m−3 in indoor air samples and 8130 pg m−3 in outdoor air samples (Fig. 5). Since all PCB sampling was conducted only in autumn season, no seasonal variation of PCBs in outdoor air was reflected. This is one of the limitations in the present study. Limited field monitoring data need further study in the future. If the concentrations of PCBs in the indoor air are relatively constant over the operation period, air PCB equilibrium between indoor and outdoor air can be described by the following Eq. (1): V

dC1 = Ea Q − VR (C1 − C0) = 0 dt

(1)

and the PCB air emission factor Ea in this facility can be calculated using Eq. (2), Ea =

(C1 − C0)VR Q

(2)

where V is the volume of the room (m3 ), C1 is the concentration of PCBs in the indoor air (ng m−3 ), C0 is the concentration of PCBs in the outdoor air (ng m−3 ), t is time, Q is the quantity of PCBs treated in the facility, and R is the natural air exchange rate at which outdoor air replaces the whole indoor air. Generally, natural air exchange rate of the concrete building is 7–24 day−1 [17],

G.-Z. Jin et al. / Journal of Hazardous Materials 196 (2011) 295–301

Fig. 3. Comparison of homologue patterns of air samples from this study with commercial mixtures of PCBs.

Fig. 4. Comparison of homologue patterns of air samples from this study with samples from other sources.

299

300

G.-Z. Jin et al. / Journal of Hazardous Materials 196 (2011) 295–301

PCB-containing products and wastes in South Korea. The total PCB concentrations ranged from 37.0 to 104,048 pg m−3 in indoor air samples and from 106 to 13,382 pg m−3 in outdoor air samples. The homologue patterns of PCBs in outdoor and indoor air samples collected from various facilities were similar to those of boundary air samples and the PCB commercial mixtures of Aroclor 1016, 1221, 1232 and 1242. These results suggest that PCB emissions during the production, recycling, in-use and disposal of PCB-containing products and wastes can be an important source of atmospheric PCBs. Therefore, it provides valuable data in planning for comprehensive management and final elimination of PCB-containing products and wastes. Fig. 5. PCB mass balance in a PCB disposal facility.

Acknowledgements which was used in this study. Using the data observed in this facility (Table S4), Ea was estimated to be 9.8 × 10−4 to 3.4 × 10−3 g-PCB gPCB−1 yr−1 . This value is comparable or slightly higher than that reported in a previous study [4]; Ea ranged from 1.58 × 10−5 to 2.56 × 10−2 g-PCB g-PCB−1 yr−1 for open in-use and storage sites, and from 3.38 × 10−9 to 5.22 × 10−4 g-PCB g-PCB−1 yr−1 for closed in-use and storage sites [9]. In the previous study, Breivik et al. [4] has reported emission factors of only 22 PCB congeners, which have high uncertainties. The specific amounts of 22 congeners in the Aroclor mixtures in South Korea are not available. However, direct comparison of PCB air emission factors reported in this study with those above might be reasonable, because the 22 congeners were dominant congeners of PCBs and both air emission factors had the same unit. The uncertainty of air emission factor calculation in this study mainly comes from natural air exchange rate, variation of PCB concentrations in indoor and outdoor air samples, temperature etc. PCB bottom emission factor Eb in this facility can be calculated using Eq. (3): Eb =

Cb A Q

(3)

where Cb is the average concentration of PCBs in bottom samples, and A is the area of the facility. Using the data measured in this facility, Eb was estimated for bottom samples to be 3.0 × 10−4 gPCB g-PCB−1 yr−1 . The PCB mass balance in this disposal facility was calculated (Fig. 5). There are major uncertainties, like Breivik et al. [9], which mainly come from natural air exchange rate, variation of PCB concentrations in air samples and temperature etc. Of the total PCBs disposed in this facility, approximately 0.0022% was emitted to the atmosphere and 0.03% was deposited to the indoor bottom as dust particles or transformer oil leakage. Meanwhile, most PCBs (98.7%) were transferred as waste transformer oil for later disposal by incineration or chemical treatment. If this facility were to operate at its maximum capacity of 100 tons/week, the estimated maximum 209 PCB emission to the air would be 13 g/yr. This is much smaller than the previous estimation, where estimated PCB emissions to air in South Korea was 199 kg (for 22 PCB congeners, mid scenario, maximum is hundreds-fold of minimum scenario) in the reference year 2008 [4]. Although the production, import and use of PCBs have been banned in South Korea since 1999, however, the amounts of in-use PCB-containing products are still huge. In South Korea, the amount of PCB-containing waste, which is contaminated with >2 mg kg−1 PCBs, was 2543 tons in 2007 [13]. Therefore, atmospheric emission of PCBs from PCB-containing products and wastes still can be a significant source for some period by the time of their complete elimination. 4. Conclusion In this study, we investigated PCBs from various types of facilities and calculated the PCB mass balance in a facility related to

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (No. 2011-1128723), and partially supported by the Korea Institute of Science and Technology (KIST) as the Institutional Program (2E22173). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.09.030. References [1] UNEP, The Stockholm convention on persistent organic pollutants (POPs), Chemicals, United Nations Environment Programme, 2001. [2] S. Sakai, M. Hiraoka, N. Takeda, K. Shiozaki, Formation and emission of nonortho CBs and mono-ortho CBs in municipal waste incineration, Chemosphere 29 (1994) 1979–1986. [3] B. Wyrzykowska, N. Hanari, A. Orlikowska, N. Yamashita, J. Falandysz, Dioxinlike compound compositional profiles of furnace bottom ashes from household combustion in Poland and their possible associations with contamination status of agricultural soil and pine needles, Chemosphere 76 (2009) 255–263. [4] K. Breivik, A. Sweetman, J.M. Pacyna, K.C. Jones, Towards a global historical emission inventory for selected PCB congeners – a mass balance approach. 3. An update, Sci. Total Environ. 377 (2007) 296–307. [5] S.K. Shin, T.S. Kim, Levels of polychlorinated biphenyls (PCBs) in transformer oils from Korea, J. Hazard. Mater. 137 (2006) 1514–1522. [6] Y. Ishikawa, Y. Noma, Y. Mori, S.-i. Sakai, Congener profiles of PCB and a proposed new set of indicator congeners, Chemosphere 67 (2007) 1838–1851. [7] K.S. Kim, B.-J. Song, J.-G. Kim, K.-K. Kim, A study on pollution levels and source of polychlorinated biphenyl (PCB) in the ambient air of Korea and Japan, J. KSEE 27 (2005) 170–176. [8] S.-D. Choi, S.-Y. Baek, Y.-S. Chang, Passive air sampling of persistent organic pollutants in Korea, Toxicol. Environ. Health Sci. 1 (2009) 75–82. [9] K. Breivik, A. Sweetman, J.M. Pacyna, K.C. Jones, Towards a global historical emission inventory for selected PCB congeners – a mass balance approach: 2. Emissions, Sci. Total Environ. 290 (2002) 199–224. [10] K. Breivik, R. Alcock, Y.F. Li, R.E. Bailey, H. Fiedler, J.M. Pacyna, Primary sources of selected POPs: regional and global scale emission inventories, Environ. Pollut. 128 (2004) 3–16. [11] A. Jamshidi, S. Hunter, S. Hazrati, S. Harrad, Concentrations and chiral signatures of polychlorinated biphenyls in outdoor and indoor air and soil in a major UK conurbation, Environ. Sci. Technol. 41 (2007) 2153–2158. [12] Korea, Persistent Organic Pollutant Special Management Law, Ministry of Environment, 2008. [13] Korea, Generation and Disposal Status of Designated Wastes, Ministry of Environment, 2008. [14] UN/ECE, The 1998 Aarhus protocol on POPs, United Nations/Economic Council for Europe, 1998 www.unece.org/env/lrtap/pops h1.htm. [15] K. Breivik, B. Bjerkeng, F. Wania, A. Helland, J. Magnusson, Modeling the fate of polychlorinated biphenyls in the Inner Oslofjord, Norway, Environ. Toxicol. Chem. 23 (2004) 2386–2395. [16] H. Hung, C.L. Sum, F. Wania, P. Blanchard, K. Brice, Measuring and simulating atmospheric concentration trends of polychlorinated biphenyls in the Northern Hemisphere, Atmos. Environ. 39 (2005) 6502–6512. [17] M. Hosomi, Volatilization of PCBs from PCB-containing Ballast in Fluorescent Lamp and Indoor PCB pollution: odor of PCBs, J. Japan Assoc. Odor Environ. 36 (2005) 323–330. [18] USEPA, Method 1668, Revision A: Chlorinated Biphenyl Congeners in Water, Soil, Sediment, and Tissue by HRGC/HRMS, 1999. [19] Korea, Official Analysis Method of PCBs in Wastes Samples, Ministry of Environment, 2005.

G.-Z. Jin et al. / Journal of Hazardous Materials 196 (2011) 295–301 [20] Y.-F. Li, T. Harner, L. Liu, Z. Zhang, N.-Q. Ren, H. Jia, J. Ma, E. Sverko, Polychlorinated biphenyls in global air and surface soil: distributions air–soil exchange, and fractionation effect, Environ. Sci. Technol. 44 (2010) 2784–2790. [21] H.-G. Yeo, M. Choi, M.-Y. Chun, T.-W. Kim, K.-C. Cho, Y. Sunwoo, Concentration characteristics of atmospheric PCBs for urban and rural area, Korea, Sci. Total Environ. 324 (2004) 261–270. [22] H.-G. Yeo, M. Choia, M.-Y. Chunb, Y. Sunwoo, Gas/particle concentrations and partitioning of PCBs in the atmosphere of Korea, Atmos. Environ. 37 (2003) 3561–3570. [23] S. Du, L.A. Rodenburg, Source identification of atmospheric PCBs in Philadelphia/Camden using positive matrix factorization followed by the potential source contribution function, Atmos. Environ. 41 (2007) 8596–8608. [24] W.S. Lee, G.P. Chang-Chien, L.C. Wang, W.J. Lee, P.J. Tsai, K.Y. Wu, C. Lin, W.-S. Lee, G.-P. Chang-Chien, L.-C. Wang, W.-J. Lee, P.-J. Tsai, K.-Y. Wu, C. Lin, Source identification of PCDD/Fs for various atmospheric environments in a highly industrialized city, Environ. Sci. Technol. 38 (2004) 4937–4944. [25] T. Takasuga, K. Senthilkumar, T. Matsumura, K. Shiozaki, S.-i. Sakai, Isotope dilution analysis of polychlorinated biphenyls (PCBs) in transformer oil and global

[26]

[27] [28]

[29]

[30]

301

commercial PCB formulations by high resolution gas chromatography–high resolution mass spectrometry, Chemosphere 62 (2006) 469–484. S.-K. Kim, J.S. Khim, K.-T. Lee, J.P. Giesy, K. Kannan, D.-S. Lee, C.-H. Koh, Emission, Contamination and Exposure, Fate and Transport, and National Management Strategy of Persistent Organic Pollutants in South Korea, vol. 7, D.i.E Science, 2007, pp. 3–157 (Chapter 2). H.-G. Yeo, M.-k. Choi, M.-Y. Chun, S.-w. Young, Air–soil partitioning of PCBs in rural area, J. Korean Soc. Atmos. Environ. 19 (2003) 1–9. K.S. Kim, Y. Hirai, M. Kato, K. Urano, S. Masunaga, Detailed PCB congener patterns in incinerator flue gas and commercial PCB formulations (Kanechlor), Chemosphere 55 (2004) 539–553. K. Martínez, J.R. Austrui, E. Jover, M. Ábalos, J. Rivera, E. Abad, Assessment of the emission of PCDD/Fs and dioxin-like PCBs from an industrial area over a nearby town using a selective wind direction sampling device, Environ. Pollut. 158 (2010) 764–769. E. Menichini, N. Iacovella, F. Monfredini, L. Turrio-Baldassarri, Atmospheric pollution by PAHs, PCDD/Fs and PCBs simultaneously collected at a regional background site in central Italy and at an urban site in Rome, Chemosphere 69 (2007) 422–434.

Journal of Hazardous Materials 196 (2011) 263–269

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Rhizodegradation gradients of phenanthrene and pyrene in sediment of mangrove (Kandelia candel (L.) Druce) Haoliang Lu a,b , Yong Zhang a,∗ , Beibei Liu a , Jingchun Liu b , Juan Ye b , Chongling Yan b a b

State Key Laboratory of Marine Environmental Science (Xiamen University), Environmental Science Research Center, Xiamen University, Xiamen 361005, Fujian Province, PR China Key Laboratory of Ministry of Education for Coastal and Wetland Ecosystems, and School of Life Sciences, Xiamen University, Xiamen 361005, Fujian Province, PR China

a r t i c l e

i n f o

Article history: Received 21 January 2011 Received in revised form 7 September 2011 Accepted 7 September 2011 Available online 14 September 2011 Keywords: Rhizodegradation Phenanthrene Pyrene Mangrove Kandelia candel (L.) Druce

a b s t r a c t A greenhouse experiment was conducted to evaluate degradation gradient of spiked phenanthrene (Ph, 10 mg kg−1 ) and pyrene (Py, 10 mg kg−1 ) in rhizosphere of mangrove Kandelia candel (L.) Druce. Rhizosphere model system was set up using a self-design laminar rhizoboxes which divided into eight separate compartments at various distances from the root surface. After 60 days of plant growth, presence of the plant significantly enhanced the dissipation of Ph (47.7%) and Py (37.6%) from contaminated sediment. Higher degradation rates of the PAHs were observed at 3 mm from the root zone (56.8% Ph and 47.7% Py). The degradation gradient followed the order: near rhizosphere > root compartment > far-rhizosphere soil zones for both contaminants where mangrove was grown. Contribution of direct plant uptake and accumulation of Ph and Py were very low compared to the plant enhanced dissipation. By contrast, plant-promoted biodegradation was the predominant contribution to the remediation enhancement. The correlation analysis indicates a negative relation between biological activities (microbial biomass carbon, dehydrogenase, urease, and phosphatase activity) and residual concentrations of Ph and Py in planted soils. Our results suggested that mangrove rhizosphere was effective in promoting the depletion of aromatic hydrocarbons in contaminated sediments. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous pollutants persisting in the environment. Anthropogenic inputs of PAHs from oil spills, ship traffic, urban runoff and emission from combustion and industrial processes have caused significant accumulation of PAHs in coastal mangrove wetlands especially those near urban centers and industrial cities [1,2]. Phytoremediation of PAHs is a promising alternative approach to sediment remediation due to its cost effectiveness, convenience and environmental acceptability [1,3]. There are several branches of phytoremediation identified by the USEPA (2000), including phytoextraction, rhizofiltration, phytovolatization, rhizodegradation and phytodegradation, and phytostabilization. Rhizodegradation refers to the microbial breakdown of organic contaminants in the root zone (rhizosphere) soil and sediment. This process uses the natural ability of plants to manipulate the biological, chemical and physical characteristics of the rhizosphere for reducing organic contaminant concentrations in soil and sediment [4,5]. In sediment, rhizoremediation was suggested to be the primary mechanism responsible for PAH

∗ Corresponding author. Tel.: +86 592 2188685; fax: +86 592 2184977. E-mail address: [email protected] (Y. Zhang). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.031

degradation in plant-assisted remediation efforts [6,7]. In this case, roots contribute to the dissipation of hydrocarbon contaminants through an increase in the number of microbes, improvement of physical and chemical soil conditions, increased root exudates and humification, and adsorption of pollutants in the rhizosphere was investigated. Mangrove ecosystems, important inter-tidal estuarine wetlands along coastlines of tropical and subtropical regions, are closely tied to industrial activities and are subject to contamination [8,9]. Mangrove may contribute to the dissipation of organic contaminants through an increase in the number of microbes, improvement of physical and chemical soil conditions, increased humification and adsorption of pollutants in the rhizosphere, but the impact of each process has not been clearly elucidated. A number of bacterial strains able to degrade PAHs have been isolated from surface mangrove sediment and the degradation of PAHs by these consortia and isolates in culture medium and in sediment slurry have been studied [10,11]. Nevertheless, the question of how far a mangrove rhizosphere effect on degradation of PAHs may extend has, however, never been approached, but the preferential use of plants with fibrous root systems for rhizodegradation indicates that it was rather narrow. The production of protons, exudates and metabolites is released by plant roots in rhizosphere soil which led to significant

264

H. Lu et al. / Journal of Hazardous Materials 196 (2011) 263–269

differences between rhizosphere and non-rhizosphere in soil properties have been reviewed in previous reports [12,13]. The rhizosphere, a layer of soil surrounding plant roots, was difficult to physically sample and manipulate with precision. Rhizosphere soil was commonly separated from plant root by gentle shaking. In an attempt to overcome some of these problems, a rhizobox was designed where soil in close proximity to roots which allowed for the harvesting of thinner consecutive sections (1–5 mm, and >5 mm) of rhizosphere soil in the lab [14]. Recently, phytodegradation of PAH contaminated sediments using mangrove plants has been the subject of several studies [2,8]. However, limited by sampling techniques of rhizosphere soil in the proximity to roots, the distance-dependent microscale depletion of PAHs in root–soil interface along the rhizosphere gradient has thus far been seldom studied. Therefore, we hypothesized that the differences with distance in rhizosphere effects would coincide with the degradation gradients of PAHs. The objective of this study was therefore to investigate the rhizosphere effect on the removal process of PAHs in a specially designed rhizobox which permitted the separation of rhizosphere soil (root compartment) and soil affected by root exudation (root-free compartment). The study also attempts to reveal the effect of the increasing distance on PAH removal. 3-Ring PAHs Ph, and 4-ring PAHs Py were used as target PAHs. Kandelia candel (L.) Druce (K. candel), a common red mangrove species in China, was chosen as the model plant. 2. Materials and methods 2.1. Chemicals Ph and Py with a purity of 99.9% were obtained from Sigma–Aldrich Co. Ltd., UK. All the other chemicals used in the study were of analytical purity. 2.2. Preparation of PAH-spiked sediment Bulk samples of surface sediments were collected from Jiulong Estuary mangrove wetland, PR China, and sieved through a 0.5 cm sieve to remove coarse debris, homogenized, and then stored at 4 ◦ C until use. The physical and chemical properties of the sediments were measured in the laboratory as follows: pH 6.63, moisture content 49.5%, total organic content 2.1% and total nitrogen amount 0.90 g kg−1 dry sediment, total phosphate amount 0.62 g kg−1 dry sediment and cation exchange capacity 15.8 cmol kg−1 . PAHs were detected in the sediment samples with concentration of 19.3 ␮g kg−1 Ph and 24.56 ␮g kg−1 Py, respectively. Sediment was taken and spiked with PAHs as follows: a portion of the sediment was accurately weighed in the vessel. Then, a volume of the PAHs dissolved in acetone is added and allowed to equilibrate with the matrix, stored in the dark and allowed to dry. Mass balance is used to determine the evaporation of acetone. The acetone was evaporated 12 h and the portion of spiked sediment was first mixed with near 25% of total sediment, and then to mix with the remaining 75% of wet sediments followed by mechanical mixing. After aging for 7 days, the sediment was used for rhizobox experiment. The detected concentration of Ph and Py was 10 ± 0.5 and 10 ± 0.4 mg kg−1 , respectively in 7 days aged sediment. No nutrient amendments were added to the soil during the experiment. 2.3. Experimental design A laboratory rhizobox modified from our previous study [15] (Fig. 1) was used to plant K. candel. The dimension of the rhizobox (Fig. 1) was 150 mm × 300 mm × 200 mm

Fig. 1. Sketch diagram of rhizobox (modified from Lu et al. [15]). S0: sediment for seedling growth; S1: rhizosphere; S2: near rhizosphere; S3: near bulk soil; and S4: bulk soil.

(length × width × height). The rhizobox was divided into five sections from central to left or right boundary of rhizobox which were surrounded by nylon cloth (400 mesh), viz. a central zone for plant growth (20 mm in width), rhizosphere zones (1 mm in width), near rhizosphere zones (2 mm in width), near bulk soil zones (10 mm in width) and bulk soil zones (52 mm in width). In the rhizobox soil for seedlings growth, rhizosphere, near rhizosphere, near bulk soil and bulk soil zones were designated as S0, S1, S2, S3 and S4, respectively. The design successfully prevents root hairs from entering the adjacent soil zones as well as keeping the soil zones separated, while permitting the transfer of soil microfauna and root exudates between the compartments. About 12 kg of the treated sediment was added to each rhizobox, each treatment had three replicates. Five K. candel seedlings were planted in the central zone of the box. The plants were grown under greenhouse conditions with natural illumination and the relative humidity of 85%, the temperature ranging from 26 to 32 ◦ C for 60 days. Sediment moisture content was adjusted to 100% of the water holding capacity by watering with fresh water to minimize drainage and simulates anoxic, waterlogged conditions. Rhizobox without plant was used as control. Rhizoboxes were arranged in a randomized design within the glasshouse and their position was rotated regularly to ensure uniform conditions. Before harvesting, rhizoboxes were withheld from watering for 2 days. Harvesting involved the sequential dismantling of each rhizobox, separating the layers of each soil zone of the rhizobox and removing the plants from the root compartment. Roots and shoots were manually separated from soils washed with deionized water, and then blotted dry with filter paper. The soil samples from different soil zones of each rhizobox were homogenized separately before analysis. 2.4. Analyses 2.4.1. Measurements of enzymatic activities Soil microbial biomass carbon (Cmic ) was determined by the chloroform-fumigation–extraction method [16,17]. Sediment dehydrogenase activity was measured by the reduction of 2,3,5triphenyl tetrazolium chloride (TTC) to 1,3,5-triphenyl formazan (TPF). Briefly, 5.0 g of freeze-dried sediment sample was incubated for 24 h at 37 ◦ C in 5.0 mL of TTC solution (5.0 g L−1 in 0.2 mol/L Tris–HCl buffer, pH 7.4). Two drops of concentrated H2 SO4 were immediately added after incubation to stop the reaction. The sample was then blended with 5.0 mL of toluene to extract TPF and shaken for 30 min at 250 rpm (25 ◦ C), followed by centrifugation

H. Lu et al. / Journal of Hazardous Materials 196 (2011) 263–269

at 5000 rpm for 5 min, and absorbance in the extract was measured at 492 nm. Finally, soil dehydrogenase activity was calculated as 1.00 ␮g TPF g−1 dry sediment [18]. Colorimetric method was utilized to determine the urease and phosphatase activities [19]. Enzyme activities expressed as mg NH4 –N released kg−1 dry sediment at 37 ◦ C and mg P released kg−1 dry sediment at 37 ◦ C each for urease and phosphatase, respectively. 2.4.2. PAH analysis Sediment samples were freeze-dried, meshed, and extracted with an accelerated microwave extraction system modified from Zhang et al. [20]. Briefly, 10.00 g of freeze-dried sediment was extracted with 50 mL mix solvent (n-hexane/acetone 1:1, v/v) using microwave extraction system (CEM Co., Matthews, NC, USA). The surrogates Ph-d10 and chrysene-d12 (Chy-d12 ) (Sigma–Aldrich, UK) were added to the samples prior to extraction. Activated copper (stirring copper with 5% of iodide/acetone solution for about 10 min) was added into the extract for desulphurization, and then pre-concentrated to 2 mL by a rotary evaporator (Buchi Vac V-800, Switzerland). Concentrated extracts were fractionated with alumina/silica gel (100–200 mesh) column chromatography (40 cm × 1.5 cm i.d.) packed from the bottom with glass wool, 10.00 g neutral aluminum oxide (100–200 mesh, dried at 440 ◦ C for 4 h), 18.00 g silica gel (100–200 mesh, dried at 170 ◦ C for 4 h) and 2.00 g anhydrous sodium sulphate. Target analytes were eluted from the column with 150 mL of mix solvent n-hexane/methylene chloride (1:1, v/v). This fraction was then concentrated to 2 mL by rotary vacuum evaporation in a water bath at 60 ◦ C and solventexchanged to n-hexane. The PAH fraction was finally concentrated to 1 mL under a gentle stream of nitrogen before GC/MS analysis. Plant samples were ground and homogenized, and extracted using the same method as to sediment. The concentrations of the PAHs in the extracts were determined by a Hewlett-Packard 6890 gas chromatography equipped with a mass spectroscopy detector (HP5975B). The HP-5MS column (Agilent Co., USA) was 30 m in length, with an internal diameter of 0.25 mm and a film thickness of 0.25 ␮m. The temperature was raised from 60 ◦ C to 150 ◦ C at a rate of 15 ◦ C min−1 , increased to 220 ◦ C at 5 ◦ C min−1 , and increased to 300 ◦ C at 10 ◦ C min−1 , then held at 300 ◦ C for 5 min. Helium was used as the carrier gas. The injector and detector temperatures were 280 ◦ C and 300 ◦ C, respectively. The electron-impact energy was 70 eV and the mass to charge ratio scan (m/z) was from 50 to 400 amu. The selected ion mode (SIM) was chosen. Detection limits derived from replicate and procedural blanks were 2.2 and 1.6 ␮g kg−1 dry weight for Ph and Py, respectively. All data were subject to strict quality control procedures. Matrix spikes, laboratory sample duplicates, and laboratory blanks were processed with each batch of samples (10 samples per batch) as part of the laboratory internal quality control. The mean recoveries of deuterated surrogate were 87.2 ± 2.1% for Ph-d10 and 90.3 ± 1.8% for Chy-d12 , respectively (n = 3). Spiked samples in each batch were analyzed with mean recoveries of 86.7 ± 2.6% for Ph and 89.6 ± 1.4% for Py, respectively (n = 3). Each extract was analyzed in duplicate form and relative standard deviations were less than 20%. Any analyses not meeting quality assurance requirements were re-analyzed. 2.5. Statistical analyses All of these experiments were performed in triplicates and the results presented were average values of the three replicates. Data were analyzed statistically using analysis of variance (ANOVA) and the Duncan’s multiple range tests was employed to determine the significance of the differences between the parameters. The

265

Table 1 Removal percentages of phenanthrene and pyrene in various sampling zones in planted and unplanted treatments after 60 days of K. candel growth. Zones

Treatment Phenanthrene (%) Unplanted

S0 S1 S2 S3 S4

27.1 26.5 25.6 26.1 25.5

± ± ± ± ±

3.5Cb 3.4Cb 3.1Cb 2.9Cb 3.2Cb

Pyrene (%) Planted 47.5 53.6 56.8 43.2 37.3

± ± ± ± ±

Unplanted 4.1Aa 6.4Aa 6.2Aa 3.4Aa 3.2Ba

23.5 24.4 21.9 23.2 20.9

± ± ± ± ±

2.9Cb 2.6Cb 2.3Cb 2.8Cb 2.6Ca

Planted 32.4 46.2 47.7 38.1 23.5

± ± ± ± ±

3.7Ba 5.1Aa 4.9Aa 4.2Ba 2.6BCa

Note: Values in each column followed with different capital letters (A, B, C and D) indicated significant (p < 0.05) differences among different distances (0, 1, 2, 4 and 6 mm) from roots, and in each row followed with different lowercase letters (a and b) indicated significant difference between planted and unplanted soils by statistically using Duncan’s multiple range tests. Values represent means ± standard deviation. S0–S4 represented the distance of 0, 1, 2, 4 and 6 mm far from the root surface.

statistical package used was SPSS statistical software package (Version 11.0) and the confidence limit was 95%. 3. Results and discussion 3.1. Dissipation gradients of Ph and Py in sediment At the beginning of the experiment, 10 ± 0.5 and 10 ± 0.4 mg kg−1 of the added Ph and Py in the sediment slurry were adsorbed onto the sediments, respectively. This indicates that evaporation of acetone did not cause any significant loss of the spiked PAHs in sediment slurries. At the end of 60 days experiment, the results showed initial Ph (10.0 mg kg−1 ) and Py (10.0 mg kg−1 ) concentrations significantly decreased in the planted sediment as well as in unplanted control, but a more marked rate of disappearance was evident when plants were presented. The removal percentages for Ph and Py were 37.3–56.8% and 23.5–47.7%, respectively, in different gradient zones of planted sediment, which were significantly higher compared to unplanted treatments (25.5–27.1% for Ph and 20.9–24.4% for Py) (Table 1). The PAH concentration in the sediment after growing mangrove was affected by proximity to the roots. At both spiked Ph and Py sediment with planted treatments, the general trend in the degradation of the PAHs was typically rhizosphere > compartment > far-rhizosphere apart from a slight difference in root zones (Table 1). The mass balance results suggest that although the spiked PAHs were adsorbed tightly onto the sediments at the beginning of the experiment, PAH-degraders had the ability to utilize and degrade the sorbed PAHs efficiently (Table 2). The loss of the PAHs from mangrove sediment could be due to biotransformation, Table 2 Mass balance of phenanthrene and pyrene in rhizobox sediment after 60 days of K. candel growth. PAH (mg)

Non-vegetation

Plant-treatment

Ph

Input Leachate Plant uptake Remain in sediment Losses

100.0 ± 5.0 ND NA 74.2 ± 4.5 25.8 ± 1.7

100.0 ± 5.0 ND NA 59.8 ± 3.7 40.2 ± 2.3

Py

Input Leachate Plant uptake Remain in sediment Losses

100.0 ± 4.0 ND NA 78.4 ± 5.2 21.6 ± 1.2

100.0 ± 4.0 ND NA 72.3 ± 4.4 27.7 ± 1.8

Note: ND, not detected; NA, not applicable because no plants were grown in the rhizobox or PAHs uptake was negligible. Values represent means ± standard deviation.

H. Lu et al. / Journal of Hazardous Materials 196 (2011) 263–269

Phenanthrene Planted

16

Cmic(mg kg-1)

biodegradation, plant uptake, or abiotic dissipation, including leaching and volatilization [21,22]. In this study, abiotic losses by leaching were insignificant because non leachate was produced during the experiment. The losses of Ph and Py via volatilization from sediment are also unlikely to occur due to whole water covered condition and low vapor pressure of the PAHs (10−1.00 and 10−2.05 L atm mmol−1 for Ph and Py, respectively). Our data showed that mangrove plant only accumulate a little PAHs (detailed in Section 3.4), thus, the loss of the PAHs from soil by plant uptake/accumulation can be assumed to be negligible. In rhizosphere, Reilley’s results suggested that abiotic dissipation (chemical degradation and irreversible sorption) was not a possible pathway of loss of anthracene and Py [23]. Therefore, our results indicated that the enhanced dissipation of the PAHs might be caused by increased rhizosphere microbial density and activity compared to unplanted soil, since the root exudates and plant litter could enhance the bioavailability of the contaminant, provide more substrate for co-metabolic degradation, and modify the soil environment to be more suitable for microbial transformation. The roots are known to release some organic compounds, such as amino acids, organic acids, sugars, enzyme and complex carbohydrates, providing carbon source and energy for the growth of rhizosphere microorganisms [13,24]. The increased dissipation of PAHs in the rhizosphere may also be due to the decreased extractability of the PAHs with the formation of bound residues. The rhizosphere could stabilize pollutants by polymerization reactions such as humification [12,25]. The PAH degradation gradients observed in the rhizobox showed that the dissipation of Ph and Py was higher in the sediment only receiving root exudates than the soil with root exudates and plant roots. This is not consistent with the gradients in root exudates and plant enzymes or the depletion zones of the most diffusion limited mineral nutrients in lots of reports [26,27]. This is an important but interesting conclusion. It might be the result of the competition between plant roots and soil microbes for soil nutrients influencing the activities of soil microbes, especially in low organic matter soil such as sediment used in this study. In addition, the Ph of suberization induced by root senescence might render the PAHs more hydrophobic and potentially interfere with their availability through adsorption [28]. While PAHs accumulated in plants only accounted for a small amount of removed PAHs, whether this plant K. candel itself was able to produce enzymes for PAHs degradation is unknown. However, the synergism mechanism still need to be confirmed by further studies, in which more plants should be involved and different plant species should be studied.

Unplanted

12 8 4 0 S0

S1

S2

S3

S4

Pyrene

16

Cmic(mg kg-1)

266

Planted

Unplanted

12 8 4 0 S0

S1

S2

S3

S4

Fig. 2. The amount of microbial biomass carbon (Cmic ) in various distances proximity to K. candel roots grown in the sediment treated with phenanthrene and pyrene. Bars are the standard error of means of three replicates.

sediment with plants compared to unplanted treatments. In the PAHs-treated sediment, enzyme activities were largest in the rhizosphere or root compartment, and then decreased with increased distance from the root surface (Figs. 2–5). However, enzyme activities did not decrease with distance in unplanted soils. This matches well the PAH degradation data (Table 1). The most probable number of PAH degraders was influenced by planting regime. Our microbial biomass data support the hypothesis that micro-organisms were responsible for the observed PAHs degradation. Planted treat-

3.2. Microbial biomass and enzyme activities of the sediment The content of soil microbial biomass carbon and activities of soil dehydrogenase, urease and phosphatase were measured to evaluate the gradient effect of the rhizosphere on the PAH degradation. The different gradient sediments from roots displayed different responses to the presence of the PAHs in the rhizobox. Overall, in the unplanted sediment, microbial biomass measured as total Cmic was the same in various compartments but was lower than the planted sediment (Fig. 2). Likewise, Cmic was 16–234% greater with, than without, plants. The largest Cmic concentration (12.65 mg kg−1 for Ph and 10.68 mg kg−1 for Py, respectively) at either gradient zone was found in the rhizosphere (S1). The activities of soil dehydrogenase, urease and phosphatase in the planted soils were higher than those of unplanted treatments over the process of 60 days biodegradation (Figs. 3–5). In our study, it was shown that the relative lower concentrations of Ph and Py (10.0 mg kg−1 ) had a stimulatory effect on enzyme activity in sediment. However, other researchers found that higher concentration of PAHs could inhibit the enzyme in the soil [10]. Our data indicated that rhizosphere effects caused the increased response characteristics in the

Fig. 3. Urease activities in various distances proximity to K. candel roots grown in the sediment treated with phenanthrene and pyrene. Bars are the standard error of means of three replicates.

H. Lu et al. / Journal of Hazardous Materials 196 (2011) 263–269

267

Table 3 The linear regression between residual PAH concentrations (Y) in sediment and different biological parameters in rhizosphere (S1) (Y = ax + b). Spiked-PAHs

Indexes (x)

a

b

r2

Phenanthrene

Cmic Urease Dehydrogenase Phosphatase

−0.36 −0.15 −5.76 −0.03

8.79 13.43 10.01 8.50

0.792* 0.693* 0.749* 0.508*

Pyrene

Cmic Urease Dehydrogenase Phosphatase

−0.41 −0.17 −7.22 −0.04

9.37 12.69 10.05 9.77

0.645* 0.745* 0.761* 0.459*

* Significant (p < 0.05) difference between residual PAHs concentrations and biological parameters.

of the soil biological status [29]. In our experiment, although the responses of activities of urease and phosphatase were different and inconsistent to some extent with the degradation of PAHs, it could still be concluded that increased urease and phosphatase activity occurred in the planted soil, especially in the rhizosphere, compared to the unplanted soil. The reason for different enzyme activities in different soil zones might relate to the gradient impact of root exudates. Fig. 4. Dehydrogenase activities in various distances proximity to K. candel roots grown in the sediment treated with phenanthrene and pyrene. Bars are the standard error of means of three replicates.

3.3. Correlation between microbial activities and PAH dissipation in rhizosphere

ments, especially rhizosphere, contained a significantly increased and large microbial biomass that could mediate the enhanced degradation of PAHs. The differences observed between soil with and without plants, as well as among various sampling zones in proximity to roots of the planted soils, were expected on the basis of microbial growth and community structure modified by both PAHs and root exudates. Overall microbial activity, as determined by dehydrogenase, urease and phosphatase activities is indicator

The successful application of rhizoremediation is largely dependent on the capacity of contaminant degraders or plant growth promoting microbes to efficiently colonize growing roots. Table 3 lists the relationships between microbial activities and the PAH dissipation in rhizosphere after 60 days of cultivation. A significant negative correlation was found between residual contaminant concentrations and soil enzymes in the rhizosphere. Statistical correlations (r2 ) of both spiked PAHs, especially for Cmic and dehydrogenase two indexes, had better values (rC2 = 0.792 and mic

2 0.645, p < 0.001; rdehydrogenase = 0.749 and 0.761, p < 0.001). For phosphatase, it showed a relatively poorer correlation. Some plant species appear to increase the numbers of degradative microbes in a large volume of soil that extends beyond the rhizosphere. The release of compounds or enzymes from roots is presumed to be associated with rhizosphere biodegradation and plant types vary in the nature and quantity of compounds released, it follows that the plant species used could be a significant factor influencing the efficacy of phytoremediation. Parrish et al. [30] reported that after 12 months of plant growth, the PAH degrading microbial populations in vegetated treatments were more than 100 times greater than those in unvegetated controls. This microbial consortia can provide various benefits to plants, including the synthesis of compounds that protect the plants by decreasing plant stress hormone levels; chelators for delivering key plant nutrients; protection against plant pathogens; and degradation of contaminants before they can negatively impact the plants [31]. Therefore, differences between rhizosphere soils and nonrhizosphere soils could be explained by the rhizosphere effect.

3.4. Accumulative potential of Ph and Py in plant tissues

Fig. 5. Phosphatase activities in various distances proximity to K. candel roots grown in the sediment treated with phenanthrene and pyrene. Bars are the standard error of means of three replicates.

When using spiked sediments for remediation experiments, the focus has often been on the ability of a given plant to accumulate a specific compound and can be removed along with the biomass for sequestration or incineration. In order to acquire a comprehensive understanding about the mechanisms of the PAH degradation, the

268

H. Lu et al. / Journal of Hazardous Materials 196 (2011) 263–269

Table 4 Phenanthrene and pyrene concentrations (mg kg−1 ) and concentration factors (CFs) in plant after 60 days of plant growth. Treatments

Root

RCFs

Stem

SCFs

Leaf

LCFs

Phenanthrene Pyrene

0.83 ± 0.11 1.56 ± 0.25

0.18 0.29

0.55 ± 0.07 1.83 ± 0.24

0.12 0.34

0.32 ± 0.04 0.59 ± 0.07

0.07 0.11

uptake of Ph and Py by K. candel was measured (Table 4). The concentrations of Ph and Py in root were higher than those in the stem and leaf, and the concentrations of Ph in leaf were lowest among plant tissues (Table 4). Plant concentration factors (CFs) were calculated as the ratio of the PAH concentrations in plant tissues (root, shoot and leaf) and in sediments on a dry weight basis. The results also indicated that root concentration factors (RCFs) of Ph (0.18) were much lower than those of Py (0.29) treatment. It might be explained by the higher Kow (octanol–water partition coefficient) value of Py than Ph [32]. It was demonstrated that hydrophobic compounds with log Kow > 4 are not readily taken up by plants through transpiration due to their hydrophobicity; log Kow for Ph and Py was 4.17 and 5.13, respectively [33]. Our data indicated that K. candel was not a hyper accumulation plant for PAHs (CFs from 0.07 to 0.34). There were no significant correlations between concentration of Ph and Py in roots and the dissipation of Ph and Py from rhizosphere sediments (S1) as well as other gradient sediments (S2–S4) were found. This indicated that accumulation of Ph and Py by roots was not the major factor contributing to the removal of PAHs from soil. This result is similar to previous report [34] which indicated contribution of K. candel accumulation and plant uptake to the removal of Py from contaminated sediments was insignificant. 4. Conclusion We investigated the rhizodegradation gradient of mangrove plant K. candel for the PAH contaminated sediment. The presence of mangrove plant significantly increased the dissipation of Ph and Py in contaminated sediment. Effect of root proximity was important in the removal process of Ph and Py, which was depended on the distance from the root surface. Enhanced dissipation rate in different gradients of planted versus unplanted sediment was 11.8–29.9% for Ph and 2.9–25.8% for Py. Accumulation of the PAHs in plant parts showed negligible contributions to the total remediation. Plant root-promoted dissipation was the predominant contribution to the remediation enhancement for sediment Ph and Py in the presence of K. candel. Our results suggested that the enhancement of Ph and Py disappearance is caused by an increase in the rhizosphere biological activity compared to root free sediment. Moreover, there is a scope for future work particularly regarding underlying mechanisms responsible for observed rhizoremediation outcomes. These future directions include elucidation of the complex processes at the interface of soil, microorganisms and roots. More effort also should be made to investigate of root exudates being deposited into the rhizosphere and involved microbe activities during the remediation process, and the achievable outcome using the mangrove under field trials. Acknowledgements The present study was supported by the China Postdoctoral Science Foundation Found Project and National Natural Science Foundation of China (10805036, 20777062, and 30710103908). The authors wish to thank Bosen Weng and Yong Huang for assistance in the sampling work, and Dr. Youwei Hong for the PAH analysis.

References [1] L. Ke, T.W.Y. Wong, Y.S. Wong, N.F.Y. Tam, Fate of polycyclic aromatic hydrocarbon (PAH) contamination in a mangrove swamp in Hong Kong following an oil spill, Mar. Pollut. Bull. 45 (2002) 339–347. [2] N.F.Y. Tam, Y.S. Wong, Effectiveness of bacterial inoculum and mangrove plants on remediation of sediment contaminated with polycyclic aromatic hydrocarbons, Mar. Pollut. Bull. 57 (2008) 716–726. [3] S.A. Cheema, M. Imran Khan, C. Shen, X. Tang, M. Farooq, L. Chen, C. Zhang, Y. Chen, Degradation of phenanthrene and pyrene in spiked soils by single and combined plants cultivation, J. Hazard. Mater. 177 (2010) 384–389. [4] U. EPA, Introduction to Phytoremediation, in, EPA/600/R-99/107, 2000. [5] K.S.H. Yu, A.H.Y. Wong, K.W.Y. Yau, Y.S. Wong, N.F.Y. Tam, Natural attenuation, biostimulation and bioaugmentation on biodegradation of polycyclic aromatic hydrocarbons (PAHs) in mangrove sediments, Mar. Pollut. Bull. 51 (2005) 1071–1077. [6] M. Mackova, D.N. Dowling, T. Macek, Phytoremediation Rhizoremediation, Springer Verlag, 2006. [7] A. Yateem, T. Al-Sharrah, A. Bin-Haji, Investigation of Microbes in the rhizosphere of selected trees for the rhizoremediation of hydrocarbon-contaminated soils, Int. J. Phytoremediation 10 (2008) 311–324. [8] L. Ke, K.S.H. Yu, Y.S. Wong, N.F.Y. Tam, Spatial and vertical distribution of polycyclic aromatic hydrocarbons in mangrove sediments, Sci. Total Environ. 340 (2005) 177–187. [9] N.F.Y. Tam, L. Ke, X.H. Wang, Y.S. Wong, Contamination of polycyclic aromatic hydrocarbons in surface sediments of mangrove swamps, Environ. Pollut. 114 (2001) 255–263. [10] C.H. Li, H.W. Zhou, Y.S. Wong, N.F.Y. Tam, Vertical distribution and anaerobic biodegradation of polycyclic aromatic hydrocarbons in mangrove sediments in Hong Kong, South China, Sci. Total Environ. 407 (2009) 5772–5779. [11] Y. Tian, H.J. Liu, T.L. Zheng, K.K. Kwon, S.J. Kim, C.L. Yan, PAHs contamination and bacterial communities in mangrove surface sediments of the Jiulong River Estuary, China, Mar. Pollut. Bull. 57 (2008) 707–715. [12] H.B. Shao, L.Y. Chu, C.A. Jaleel, P. Manivannan, R. Panneerselvam, M.A. Shao, Understanding water deficit stress-induced changes in the basic metabolism of higher plants-biotechnologically and sustainably improving agriculture and the ecoenvironment in arid regions of the globe, Crit. Rev. Biotechnol. 29 (2009) 131–151. [13] W. Shi, H. Shao, H. Li, M. Shao, S. Du, Progress in the remediation of hazardous heavy metal-polluted soils by natural zeolite, J. Hazard. Mater. 170 (2009) 1–6. [14] J.L. Chen, K.C. Au, Y.S. Wong, N.F.Y. Tam, Using orthogonal design to determine optimal conditions for biodegradation of phenanthrene in mangrove sediment slurry, J. Hazard. Mater. 176 (2010) 666–671. [15] H. Lu, C. Yan, J. Liu, Low-molecular-weight organic acids exuded by Mangrove (Kandelia candel (L.) Druce) roots and their effect on cadmium species change in the rhizosphere, Environ. Exp. Bot. 61 (2007) 159–166. [16] J. Wu, R.G. Joergensen, B. Pommerening, R. Chaussod, P.C. Brookes, Measurement of soil microbial biomass C by fumigation–extraction: an automated procedure, Soil Biol. Biochem. 22 (1990) 1167–1169. [17] R.K. Lu, Soil Agrochemistry and Analytical Methods, Chinese Agricultural Science and Technology Press, Beijing, 2000. [18] S.A. Cheema, M.I. Khan, X. Tang, C. Zhang, C. Shen, Z. Malik, S. Ali, J. Yang, K. Shen, X. Chen, Enhancement of phenanthrene and pyrene degradation in rhizosphere of tall fescue (Festuca arundinacea), J. Hazard. Mater. 166 (2009) 1226–1231. [19] S.Y. Guan, Soil Enzymes and their Methodology, Agricultural Press, Beijing, 1986. [20] J. Zhang, L. Cai, D. Yuan, M. Chen, Distribution and sources of polynuclear aromatic hydrocarbons in Mangrove surficial sediments of Deep Bay, China, Mar. Pollut. Bull. 49 (2004) 479–486. [21] K.E. Gerhardt, X.D. Huang, B.R. Glick, B.M. Greenberg, Phytoremediation and rhizoremediation of organic soil contaminants: potential and challenges, Plant Sci. 176 (2009) 20–30. [22] N.F.Y. Tam, C.L. Guo, W.Y. Yau, Y.S. Wong, Preliminary study on biodegradation of phenanthrene by bacteria isolated from mangrove sediments in Hong Kong, Mar. Pollut. Bull. 45 (2002) 316–324. [23] K.A. Reilley, M.K. Banks, A.P. Schwab, Dissipation of polycyclic aromatic hydrocarbons in the rhizosphere, J. Environ. Qual. 25 (1996) 212–219. [24] W. Ling, Y. Zeng, Y. Gao, H. Dang, X. Zhu, Availability of polycyclic aromatic hydrocarbons in aging soils, J. Soil Sediment 10 (2010) 799–807. [25] S.H. Lee, W.S. Lee, C.H. Lee, J.G. Kim, Degradation of phenanthrene and pyrene in rhizosphere of grasses and legumes, J. Hazard. Mater. 153 (2008) 892–898. [26] T.A. Anderson, E.A. Guthrie, B.T. Walton, Bioremediation in the rhizosphere, Environ. Sci. Technol. 27 (1993) 2630–2636. [27] E.J. Joner, I. Jakobsen, Growth and extracellular phosphatase activity of arbuscular mycorrhizal hyphae as influenced by soil organic matter, Soil Biol. Biochem. 27 (1995) 1153–1159.

H. Lu et al. / Journal of Hazardous Materials 196 (2011) 263–269 [28] T.R. Sun, L. Cang, Q.Y. Wang, D.M. Zhou, J.M. Cheng, H. Xu, Roles of abiotic losses, microbes, plant roots, and root exudates on phytoremediation of PAHs in a barren soil, J. Hazard. Mater. 176 (2010) 919–925. [29] X. Zhuang, J. Chen, H. Shim, Z. Bai, New advances in plant growth-promoting rhizobacteria for bioremediation, Environ. Int. 33 (2007) 406–413. [30] Z.D. Parrish, M.K. Banks, A.P. Schwab, Assessment of contaminant lability during phytoremediation of polycyclic aromatic hydrocarbon impacted soil, Environ. Pollut. 137 (2005) 187–197. [31] G.L. Mena-Benitez, F. Gandia-Herrero, S. Graham, T.R. Larson, S.J. McQueenMason, C.E. French, E.L. Rylott, N.C. Bruce, Engineering a catabolic pathway

269

in plants for the degradation of 1,2-dichloroethane, Plant Physiol. 147 (2008) 1192–1198. [32] A. Dietz, J. Schnoor, Advances in phytoremediation, Environ. Health Perspect. 109 (2001) 163–168. [33] R.P. Schwarzenbach, P.M. Gschwend, D.M. Imboden, Environmental Organic Chemistry, John Wiley & Sons, New York, 1993. [34] L. Ke, W.Q. Wang, T.W.Y. Wong, Y.S. Wong, N.F.Y. Tam, Removal of pyrene from contaminated sediments by mangrove microcosms, Chemosphere 51 (2003) 25–34.

Journal of Hazardous Materials 196 (2011) 270–277

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Amphiphilic hollow carbonaceous microspheres for the sorption of phenol from water Zhengrong Guan, Li Liu, Lilu He, Sen Yang ∗ College of Resources and Environmental Sciences, Biomass Engineering Center, China Agricultural University, Beijing 100193, People’s Republic of China

a r t i c l e

i n f o

Article history: Received 29 June 2011 Received in revised form 19 August 2011 Accepted 7 September 2011 Available online 14 September 2011 Keywords: Carbonaceous spheres Removal Partition Phenol

a b s t r a c t Amphiphilic porous hollow carbonaceous spheres (PHCSs) were synthesized via mild hydrothermal treatment of yeast cells and further pyrolyzing post treatment. The morphology, chemical composition, porosity, and structure of the carbonaceous materials were investigated. It is evident that the carbonaceous materials were composed of the carbonized organic matter (COM) and the noncarbonized organic matter (NOM), and the relative COM and NOM fractions could be adjusted through changing the temperature of hydrothermal and/or pyrolyzing treatment. The phenol sorption properties of the carbonaceous materials had been investigated and the sorption isotherms fit well to the modified Freundlich equation. It was found that the sorption isotherm of phenol onto PHCSs was practically linear even at extreme high concentrations, which was fewer reported for activated carbon or other inorganic materials. This type of sorption isothermals was assigned to a partition mechanism, and the largest value of the partition coefficient (Kf ) and carbon-normalized Kf (Koc ) is 56.7 and 91.5 mL g−1 , respectively. Moreover, PHCSs exhibit fast sorption kinetic and facile regeneration property. The results indicate PHCSs are potential effective sorbents for removal of undesirable organic chemicals in wastewater, especially at high concentrations. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Phenol-like compounds have attracted global concerns because of their toxicity and the frequency and quantity of their presence in wastewaters of various industries, such as refineries (6–500 mg L−1 ), coking operations (28–3900 mg L−1 ), coal processing (9–6800 mg L−1 ), and manufacture of petrochemicals (2.8–1220 mg L−1 ) [1–4]. Phenol-containing wastewater usually involves multiple, different contaminants and the concentration is different from low-concentration of several mg L−1 to highconcentration of several thousands mg L−1 . Those meanings the best abatement technologies for phenol from wastewaters to be applied strongly depends on single cases, in particular on the concentration of phenol in the stream, the co-presence of other contaminants, the nature of the plant where this problem is found [2]. Now, a number of strategies such as oxidation with ozone/hydrogen peroxide [5], biological methods [6,7], membrane filtration [8], ion exchange [9], electrochemical oxidation [10], photocatalytic degradation [11], and adsorption [12] have been used for the removal of phenol. Review on available technologies for phenol removal from fluid streams has been recently published providing comparison of the experimental conditions and the performances of different techniques [2]. Adsorption is generally considered an

∗ Corresponding author. Tel.: +86 10 62733470; fax: +86 10 62733470. E-mail address: [email protected] (S. Yang). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.025

operationally simple, effective and widely used process for the removal and recovery of the phenol. Among vast number of different adsorbents, activated carbon (AC) is the most commonly used adsorbent in industrial scale and experimental research [1]. However, AC is not an ideal adsorbent for practical applications at high phenol concentrations, since the adsorption amount of AC will soon saturate. The demanding regeneration and poor mechanical rigidity of AC are also problems for its wider application [13]. Consequently, a large variety of non-conventional adsorbents have been studied for removal of phenol and phenolic pollutants including organoclay [14], polymer [13], carbon nanotube [15], sewage sludge-based adsorbents [16] and activated carbon cloth [17]. However, development of novel adsorbent materials for removal and recovery of the phenol and phenolic pollutants with high concentrations still remains an important challenge. Yeast, a by-product of the brewing industry, is considered as an industrial organic waste that causes a great deal of concern [18]. In a previous study, we reported a facile method for fabricating porous hollow carbonaceous spheres (PHCSs) with controlled shell porosity from Saccharomyces cerevisiae (S. cerevisiae) cells [19,20]. Through mild hydrothermal treatment of these tiny unicellular organisms, hollow microspheres with controllable mesoand macroporous shells were synthesized. Most interestingly, the surfaces of these hollow spheres were found to be covered with both hydrophobic and hydrophilic functional groups, endowing the as-obtained microspheres with amphiphilic property. Actually, we found that PHCSs could be well dispersed not only in water but also

Z. Guan et al. / Journal of Hazardous Materials 196 (2011) 270–277

in nonpolar solvents such as toluene and chloroform. This inspired us to speculate that PHCSs may be promising sorbent for organic pollutant from aqueous solution. Herein, four kinds of PHCSs were synthesized via mild hydrothermal treatment of yeast cells and further pyrolyzing post treatment. And the morphology, chemical composition, porosity, and structure of the carbonaceous materials were investigated. The sorption behavior (i.e., sorption isotherms, kinetics models, and the influence of pH and temperature) of PHCSs toward phenol was examined. The relationship existing between the specific surface area and the chemical composition of PHCSs and their phenol sorption capability was elucidated. Lastly, principal sorption mechanisms were clarified. 2. Experimental 2.1. Synthesis of PHCS PHCSs were synthesized via mild hydrothermal treatment of yeast cells using modified methods described in our previous studies [19]. Typically, S. cerevisiae cells (3–4 g, purchased from Angel Yeast Co., Ltd., China) pre-washed with acetone were dispersed in 2–3% (v/v) glutaraldehyde and diluted (less than 0.01 mol L−1 ) nitric acid aqueous solution (40 mL), which was then placed in a 50 mL Teflon-sealed autoclave and maintained at 180 or 230 ◦ C for 8 h. The puce solid products were centrifugal separated, then washed by three cycles of centrifugation/washing/redispersion in deionized water and alcohol, and oven-dried at 80 ◦ C for 4 h. The PHCSs samples hydrothermally treated at 180 and 230 ◦ C were denoted as P180 and P230, respectively. The samples denoted as P350 and P700 were prepared via further pyrolyzing P180 at temperature of 350 ◦ C and 700 ◦ C for 1 h, respectively, by a tubular reactor in a flow of nitrogen (15 mL min−1 ). 2.2. Characterization of PHCS The morphology of the materials was inspected with a field emission scanning electron microscope (SEM, JEOL, JSM-6700F, Japan). Surface area and pore volume of the materials were measured by N2 adsorption/desorption isotherms at 77 K with a Physisorption Analyzer (Micromeritics, ASAP 2020, U.S.A.). Fourier-transform infrared (FTIR) analysis was performed by a Micro FTIR spectrometer (Nicolet, Magna 750 Nic-Plan FTIR Microscope, U.S.A.) in the spectral region of 4000–650 cm−1 . Solid-state cross-polarization magic angle-spinning and totalsideband-suppression 13 Carbon nuclear magnetic resonance (13 C NMR) spectrum (CPMAS-TOSS) were obtained by a Bruker Avance 400 MHz spectrometer (Karlsruhe, Germany). The C, H, N, O contents of the samples were determined using elemental analyzer (Elementar, Vario EL, Germany). The atomic ratio of (O + N)/C, O/C, H/C were calculated with the element content. 2.3. Sorption experiments Batch experiments were carried out using a series of 15 mL screw cap centrifuge tube covered with Teflon sheets to prevent the introduction of any foreign particle contamination. In typical batch experiment, 20 mg of the sorbent was added to 10 mL phenol solution at various concentrations (0–10,000 mg L−1 ) taken in sealed tubes, which were placed in the thermostat shaking assembly. The solutions were shaken at 150 rpm and constant temperatures for 24 h to achieve equilibration. After equilibrium, the mixtures were filtrated through 0.45 ␮m nitrocellulose syringe filters. The phenol concentrations in the filtrates were determined by a UV-visible spectrometer (Persee, TU-1810, China) at the maximum adsorption wavelength of phenol (270 nm) and pH 6 (pH adjusted with 0.5 M

271

HCl or NaOH). Isotherms were performed by taking different concentrations of phenol at designed temperatures and pH values. All the experiments were carried in triplicate and the averaged data were reported. Standard deviations were found to be within 2.0%. Furthermore, the error bars for the figures were smaller than the symbols used to plot the graphs and hence are not shown in the figures. 2.4. Sorption models and statistical analysis Freundlich model was applied for sorption data fitting in this work due mainly to its advantages in isotherm nonlinearity investigation. To facilitate direct comparisons of sorption affinities among the samples tested, and investigate the effect of temperature on sorption, the modified Freundlich equation was applied for sorption data fitting in this work: log qe = log KF + n log Cr Cr =

Ce Sw

(1) (2)

where qe is the solid-phase concentration (mg g−1 ), Ce is the liquidphase equilibrium concentration (mg L−1 ); Sw is the solubility of phenol for a given temperature (mg L−1 ) and Cr is dimensionless since the value of Sw is constant for a given temperature and is expressed in the same unit as Ce ; K  F and n are the modified Freundlich adsorption parameters, K  F (mg g−1 ) is the sorption capacity coefficient, which represents the mass of phenol sorbed per unit mass of sorbent when the Ce concentration approaches saturation, and n (dimensionless) is an indicator of isotherm nonlinearity related to the heterogeneity of sorption sites[21]. The partition–adsorption model for describing sorption from aqueous solutions on heterogeneous solids was also analyzed [14,22]: QT = QA + QP

(3)

where QT is the total amount of phenol sorbed onto the sorbent; QA and QP are the amounts contributed by adsorption and partition, respectively. According to the partition–adsorption model, the partition effect is favored progressively by increasing the solute concentration, whereas the adsorption contribution reaches saturation more rapidly with the solute concentration. The isotherm at high concentrations should approach linearity. Therefore, at the high solute concentration range the adsorption becomes saturation and the linear partition remains. Thus, Eq. (3) can be transformed to: QT = QAmax + QP = QAmax + Kf Ce Koc =

Kf foc

(4) (5)

where QAmax is the saturated adsorption capacity estimated from the high concentration data; Kf Ce is the partition contribution at high concentration with Kf being the partition coefficient; Ce is the solute equilibrium concentration (mg L−1 ); Koc is carbon-normalized Kf , and foc is the percentage of carbon contents of sorbent. Linear regression between QT and Ce were conducted at high solute concentration range, and the QAmax corresponds to the y-axis intercept of the line, and Kf to the slope. 2.5. Sorption kinetics For kinetic studies the batch technique was used. Typically, 20 mg of PHCSs were added 10 mL phenol solution (2000 mg L−1 ) into series of screw cap centrifuge tubes, and then shaken in constant temperature rotary shaker (150 rpm) at 25 ± 0.5 ◦ C. The

272

Z. Guan et al. / Journal of Hazardous Materials 196 (2011) 270–277

Fig. 1. SEM images of (a) P180, (b) P230, (c) P350 and (d) P700.

concentrations of phenol in solution were sampled and analyzed with different time intervals, and the averaged data were reported. Linear pseudo-first and pseudo-second order models were given in the following equations [23,24]: ln(qe − qt ) = ln qe − k1 t

(6)

t 1 t = + qt qe k2 q2e

(7)

t1/2 =

1 k2 qe

(8)

h = k2 q2e

(9) mg g−1 )

where qe and qt (both in are the amount of phenol sorbed on per unit mass of sorbent at equilibrium and at time t (h), respectively; k1 (h−1 ) and k2 (g mg−1 h−1 ) are the pseudo-first-order and pseudo-second-order rate constants, respectively. Moreover, for the pseudo-second-order kinetic model, the half sorption time (t1/2 ) and the initial sorption rate (h) are given (Eqs. (8) and (9)). The value of t1/2 is the time required to uptake half of the maximal sorbed amount of sorbate at equilibrium. 3. Results and discussion 3.1. Characterization of PHCSs SEM images and optical microscopy photos of the carbonaceous products are shown in Figs. 1 and S1, respectively. It was found that the hydrothermal products (P180 and P230) were porous hollow microspheres in the size range 2.0–4.0 ␮m, which was consistent

with our previous results [19]. The thermal stability of the microspheres was satisfactory. After post pyrolyzing treatment of P180 at 350 ◦ C, the microsphere structure was still well preserved (Fig. 1c for P350). Further increasing the pyrolyzing treatment temperature to 700 ◦ C, however, the morphology of materials changed from microspheres into larger carbon blocks (Fig. 1d for P700). Surface area and pore volume of the materials were measured by N2 adsorption/desorption isotherms at 77 K and all the BET surface areas of the materials are below 10 m2 g−1 (Table 1). N2 adsorption/desorption isotherm of the materials at 77 K with corresponding pore size distributions are shown in Fig. S2. Fig. 2 shows the solid-state 13 C NMR spectra of the carbonaceous products. Peaks at ı = 26 and 31 ppm can be attributed to methyl and methylene, respectively, and those in the ı = 120–150 ppm region can be assigned to long-range conjugated C C bonds and oxygen-substituted C C bonds, revealing the existence of aromatic furan ring compounds [25,26]. Moreover, oxygenated functional groups, including carbonyl, carboxy, hydroxy, ether, and ester groups were also detected. The content of aromatic species was enhanced remarkably by pyrolyzing treatment of the hydrothermal products, indicating high carbonization degree for P350 and especially P700. Moreover, the average chemical shift decreases with the sequence of the carbonization, resembling to some extent the resonance pattern of graphite. Corresponding Fourier transform infrared (FTIR) spectra of the carbonaceous materials are shown in Fig. 3. Both aliphatic and aromatic species were revealed for P180, P230 and P350. The bands at 2925, 2861, 1442, and 1372 cm−1 are assigned mainly to CH2 units [27]. Those at 1693 and 1160 cm−1 are assigned to C O and C–O stretching vibrations of ester bonds, respectively. And the band at 1602 cm−1 is assigned to C C and C O

Z. Guan et al. / Journal of Hazardous Materials 196 (2011) 270–277

273

Table 1 Elemental compositions, atomic ratios, BET surface area (SA), and total pore volume (TPV) of the carbonaceous materials. Sample

Treatment temperature (◦ C)

C (wt.%)

H (wt.%)

O (wt.%)

N (wt.%)

(O + N)/C ratio

O/C ratio

H/C ratio

SA (m2 g−1 )

TPV (mL g−1 )

P180 P230 P350 P700

180 230 350 700

62.00 72.96 73.22 77.17

6.25 6.99 4.44 1.94

21.16 12.15 10.61 5.71

6.08 4.88 6.27 5.06

0.34 0.18 0.18 0.11

0.26 0.13 0.11 0.06

1.21 1.14 0.72 0.30

9.504 2.635 1.960 2.830

0.0498 0.0095 0.0052 0.0080

H/C: atomic ratio of hydrogen to carbon. O/C: atomic ratio of oxygen to carbon. (O + N)/C: atomic ratio of sum of nitrogen and oxygen to carbon.

Fig. 2. Solid-state 13 C NMR spectrum of (a) P180, (b) P230, (c) P350 and (d) P700.

stretching in the aromatic ring [22]. The peak of 786, 706 cm−1 can be assigned to the aromatic CH out-of-plane deformation. In the case of P700, almost all of the band intensities mentioned above are dramatically decreased or disappeared, indicating destruction of

the surface function groups and chemical structure after pyrolyzing treatment at high temperatures. The elemental compositions shown in Table 1 agree well with the observations of FTIR. O content decreased from 21.16% for

Fig. 3. FTIR spectrum of (a) P180, (b) P230, (c) P350 and (d) P700.

274

Z. Guan et al. / Journal of Hazardous Materials 196 (2011) 270–277

Fig. 4. Sorption isotherms of phenol on carbonaceous products in aqueous solution at 25 ◦ C and pH 6, inset: sorption isotherms on P230 at high phenol concentrations.

P180 to 12.15% for P230 and further declined slightly to 10.61% for P350. When the pyrolyzing treatment temperature was increased to 700 ◦ C, the O content declined drastically to 5.71% (P700). The H content of P350 and P700 declined dramatically from about 6.25% (P180) to 4.44% and 1.94%, respectively. These results mean that the degree of carbonation of samples increased with higher hydrothermal and pyrolyzing temperature. The H/C ratio and O/C ratio decreases with deep carbonization, indicating the surfaces of the products become less hydrophilic [28]. The decrease of the polarity index [(O + N)/C] with the degree of carbonization reveals a reduction of the surface polar functional groups [22]. All the above results reveal that both hydrothermal and pyrolyzing treatment changed the chemical composition of PHCSs to a large degree. Due to the partial loss of the hydrophilic groups and aromatization of the molecule networks, wettability characteristics of PHCSs were supposed to be very different from that of the original hydrophilic yeast cells. We found all samples except P700 could be well dispersed not only in water but also in nonpolar solvents (Fig. S3), suggesting that PHCSs possess amphiphilic surfaces, as we reported previously [19,20]. Moreover, there are two types carbon species in the PHCSs, namely alkyl carbon and aromatic carbon. The alkyl carbon is mainly derived form the noncarbonized organic matter (NOM) of the yeast cells, and the aromatic carbon is the products of carbonization, i.e., the carbonized organic matter (COM). It is evident that the relative COM and NOM fractions could be adjusted through changing the temperature of hydrothermal and/or pyrolyzing treatment. While there are large amounts of NOM in the samples of P180, P230 and P350, COM is dominant in P700. The amphiphilic property and the specific chemical composition of the PHCSs make them may be excellent potential sorbents for organic compounds removal from wastewater. Subsequently, the sorption isotherms of phenol were carried out. 3.2. Sorption isotherms Sorption of phenol onto the carbonaceous materials at 25 ◦ C was investigated and the results are presented in Fig. 4. The sorption of P700 was saturated at very low phenol concentration (86 mg L−1 ) and the sorption amount was only 5.5 mg g−1 , which may be due to the very low surface area (2.8 m2 g−1 ). Because P700 was produced by pyrolyzing P180 at 700 ◦ C for 1 h, the high carbonization degree and few surface function groups (Figs. 2 and 3) suggested that the sorption mechanism of P700 is physical adsorption.

The result means P700 cannot be used as sorbents for organic compounds, and so we did not discuss about P700 any more. The surface area of P180, P230 and P350 was all less than 10 m2 g−1 , however, they exhibited quite different sorption behaviors: surface area-independent sorption amount and the almost linear isotherm without a saturated sorption. This sorption characteristic was fewer reported for AC [29,30] or other inorganic adsorbent materials [31], which suggests that the sorption mechanism of PHCSs is not simple physical or chemical adsorption. Modified Freundlich model was applied to investigate the isotherm nonlinearity of PHCSs. The sorption isotherms fit well to the modified Freundlich equation and the calculated parameters are listed in Table 2. The isotherm of phenol onto P180 is practically linear, with Freundlich n = 1.016 ± 0.025, and the isotherms for P230 and P350 display different nonlinearity of a concavedownward curvature at low solute concentrations but exhibit a practically linear shape at moderate to high concentrations. The nonlinear effects are relatively more visible for the P350 than for the P230. Similar results were observed for sorption of organic solutes from water over a wide range of Ce /Sw by black carbons (BCs) or biochars, derived mainly from the incomplete combustion of biomass and fossil fuel [22,32,33]. The unique isotherm shape, i.e., nonlinear at low Ce /Sw but virtually linear at other Ce /Sw , suggests that more than a single mechanism be operative over the entire concentration range [34]. To explain the nonlinearity of sorption isotherms, dual-mode sorption models (DMSM) or dual-reactive domain models (DRDM) were suggested. According to these models [14,22,32,33,35], the sorbent was considered to be a heterogeneous substance, and a concept of NOM (or “soft carbon”, expanded, rubbery state) vs. COM (or “hard carbon”, condensed, glassy state) was been invoked to operationally delineate chemical heterogeneity of sorbent and to elucidate the mechanisms for sorption by soils, sediments, biochars and charcoal. The COM is expected to behave as an physical adsorbent, producing isotherm nonlinearity, and the NOM can uptake pollutants via a partition (sorption) mechanism [22,32,33]. One unique feature of the partition process is that the ratio of solid phase to aqueous-phase concentrations remains unchanged with the variation of the solute concentration. In this sense, the sorptive uptakes are determined by the relative carbonized and noncarbonized fractions and their surface and bulk properties. Thus, the linearity of P180 can be assigned to sorption of phenol to NOM via partition mechanism, since NOM is the mainly carbon species in this sample. The nonlinearity of P230 and P350 at low Ce /Sw is attributed to a combined physical adsorption on a small amount of COM and a partition effect of NOM. At moderate to high Ce /Sw , the physical adsorption becomes largely saturated and the partition in NOM predominates to give an essentially linear isotherm. The larger curvature at low Ce /Sw of P350 is presumably due to the higher content of aromatic moieties with increased carbonization degree. The adsorption behavior of P700 with highest carbonization degree, validate this speculation. High isotherm nonlinearity was also observed for high-temperature chars [33]. From the partition–adsorption model, QAmax , Kf and Koc were calculated (Table 2), The Kf and Koc increase in the order of P350 < P230 < P180, which is in inverse with their carbonization degree. The physicochemical nature of the organic carbon has been suggested as a major factor controlling sorption of organic compounds to natural or modified organic sorbent. According to FTIR and NMR data (Figs. 2 and 3), the major partition phase in P180, P230 and P350 is a polymeric aliphatic fraction preserved during the hydrothermal process, and the content of aliphatic fraction increased in the order of P350< P230 < P180, which is in the line with the increased partition effect. The similar conclusions were also presented by other researchers [22,36]. Chefetz et al. [36] tested the sorption of pyrene to a series of sorbents comprised of different levels of aromaticity and aliphaticity. In that study, a

Z. Guan et al. / Journal of Hazardous Materials 196 (2011) 270–277

275

Table 2 Modified Freundlich model parameters, partition coefficients and saturated adsorption of phenol to PHCs. Sample

Modified Freundlich model parameters log KF a

P180 P230d P230 P230e P350

3.733 3.102 3.434 3.230 2.543

± ± ± ± ±

nb 0.049 0.024d 0.058 0.018e 0.041

1.016 0.661 0.822 0.700 0.378

± ± ± ± ±

0.025 0.013d 0.025 0.009e 0.019

Partition–adsorption model R2

Kf (mL g−1 )c

QAmax (mg g−1 )c

0.996 0.997d 0.987 0.991e 0.986

56.7 25.8d 38.2 39.4e 17.3

0 67.3d 45.8 34.3e 45.15

Koc (mL g−1 )

max QA,SA (mg g−1 m−2 )

91.5 35.4d 52.5 54.0e 23.6

0 25.5 17.4 13.0 23.0

max Koc is carbon-normalized Kf . QA,SA is the SA-normalized QAmax . The solubility of phenol in water at 15, 25 and 35 ◦ C is 8200 mg/100 mg, 8660 mg/100 mg and 9910 mg/100 mg, respectively. a 95% confidence interval of log KF . b 95% confidence interval on n. c The slope and y-axis intercept of the linear equation were used to calculate partition coefficient (Kf ) and the maximum adsorption capacity (QAmax , mg g−1 , Chen et al. [22]), respectively. d Tested at 15 ◦ C. e Tested at 35 ◦ C.

positive trend was observed between the Koc level and the aliphaticity of the set of samples. Chen et al. [22] performed sorption experiments with biochars, produced by pyrolysis of pine needles at different temperatures. This study clearly demonstrated that a greater sorption affinity of naphthalene, nitrobenzene, and m-dinitrobenzene with aliphatic-rich biochars than with aromaticrich biochars. max of P230 and P350 was listed in Table 2, which exceeds The QA,SA greatly the amount accountable by the small surface area of the sorbent. Some researchers also reported the higher adsorption of polar solutes compared with the little sorbent surface area. To explain the higher sorption of polar pesticides at low (relative) concentrations, Spurlock and Biggar [37] suggested the specific-interaction model. The model postulates that the specific interaction of polar solutes with highly active sites of organic carbon phase. That implies that more highly active site, more specific interaction, and max . P350 showed less oxygen content and highly active higher QA,SA max site because of pyrolyzed dehydration at 350 ◦ C, however, the QA,SA of P350 is 23.0 mg g−1 m−2 , higher than the 17.4 mg g−1 m−2 of P230. We speculated the possible phenol–sorbent interaction may be hydrophobic effect since P350 showed higher hydrophobicity compared with P230.

3.3. Sorption kinetics Fig. 5 shows the sorption kinetics of phenol onto P180 and P230. Apparently, the sorption rates were very fast. For example, given the tested conditions, approximately 86 and 85% of sorption was accomplished within 0.5 h for P180 and P230, respectively. For quantitative comparison of apparent sorption kinetics between P180 and P230, the data were fitted to the pseudo-first order and pseudo-second order models, respectively. The sorption kinetic constants were listed in Table 3. The regression coefficient (R2 ) for the pseudo-first order model varied from 0.9830 to 0.7022 and the Qe values calculated from the model deviated tremendously from the experimental values, together indicating the invalidity of the model. However, the pseudo-second-order model provided the best fitting for the all experimental data. The plots show regression coefficients higher than 0.9997 for P180 and P230. The value of the constant k2 of P180 is higher than that of P230, which inversely correlates with the Kf . Thus, it could be speculated that kinetics of phenol following the pseudo-second-order model are controlled by an adsorption process and adsorption was the main sorption rate-limiting step. 3.4. Effect of temperature on sorption To investigate the effect of temperature on the sorption of phenols, sorption experiments were conducted at 15, 25 and 35 ◦ C for P230. The results are shown in Fig. 6. The parameters of

Fig. 5. Sorption kinetics plotted as sorbed amount of phenol vs. time on P180 and P230 with a 2000 mg L−1 initial phenol concentration and 25 ◦ C.

Fig. 6. Sorption isotherms of phenol on P230 at 15, 25 and 35 ◦ C at pH 6.

276

Z. Guan et al. / Journal of Hazardous Materials 196 (2011) 270–277

Table 3 Adsorption rate constants for two kinetic models at 25 ◦ C and pH of 6. Sample

P180 P230 a b

qe (exp)a (mg g−1 )

98.63 107.85

First-order model

Second-order model

K1 (h−1 )

qe (cal)b (mg g−1 )

R2

K2 (g mg−1 h−1 )

qe (cal)b (mg g−1 )

t1/2 (h)

h (mg g−1 h−1 )

R2

3.17 0.12

82.92 19.86

0.9830 0.7022

0.1028 0.0660

110.31 106.80

0.099 0.14

1000 769

0.9997 0.9998

Experimental data. Calculated data from models.

modified Freundlich adsorption model and partition–adsorption model are incorporated in Table 2. A comparison of modified Freundlich parameter n shows that the biggest value was got at 25 ◦ C and the smallest value at 15 ◦ C. This means that the effect of temperature on the adsorption and partition is different. Comparing with QAmax and Kf of P230 at 15, 25 and 35 ◦ C, we can find that partition increase with an increase in temperature, meaning that temperature may play an important role for the partition coefficient Kf . To date, fewer Kf values have been reported for hydrophobic organic contaminants (HOCs) at temperatures other than 25 ◦ C, and the reported conclusion is different [38]. Chen and Pawliszyn [39] evaluated the effect of temperature on Kf for BTEX compounds and found that Kf should not be greatly dependent on temperature. Muijs and Jonker [40] determined Kf values for several PAHs and found that temperature can play a significant role. We speculated that the increasing Kf values of P230 for phenol with increasing temperature maybe due to the polarity of phenol and the structure of P230. However, the adsorption decreases with an increase in temperature indicate that the process is apparently exothermic.

3.5. Effect of pH value on sorption The effect of pH on the sorption ability of P230 was investigated and the results are shown in Fig. 7. According to the model proposed by Deryło-Marczewska and Marczewski [41], the adsorption of non-ionized compound does not depend on the pH and the surface charge. The sorption isotherms at the pH of 3 and 6 coincided. This may be due to the dissociation degree of phenol is very low (pKa = 9.89) in acidic conditions. However, the sorption at pH of 11 obviously decreased since phenol is highly dissociated at pH 11 and dissolved into water.

3.6. Sorbent regeneration Important goals in the development of sorbent materials include simple regeneration and sorbate isolation [42]. Regeneration allows for the repeated use of the sorbent material and decreasing costs. It was found that desorption of phenol from the loaded P230 using 0.01 M NaOH solution was rapid. For example, after placing P230 that had sorbed 141 mg g−1 phenol in a 0.01 M NaOH solution for 10 min, the regenerated P230 showed greater than 98% of the phenol was removed from the sorbent. The facile regeneration is due to the high solubility of the sodium salt of phenol in water. 4. Conclusions Amphiphilic PHCSs were synthesized via mild hydrothermal treatment of yeast cells and further pyrolyzing post treatment. The sorption properties of PHCSs for phenol in aqueous solutions were investigated, and then reached the following conclusions: (1) PHCSs were composed of COM and NOM, and the relative COM and NOM fractions could be adjusted through changing the temperature of hydrothermal and/or pyrolyzing treatment. (2) The sorption isotherm of phenol onto PHCSs was practically linear even at extreme high concentrations. This type of sorption isothermals was assigned to a partition mechanism, and the largest value of the partition coefficient (Kf ) and carbonnormalized Kf (Koc ) is 56.7 and 91.5 mL g−1 , respectively. (3) PHCSs exhibit fast sorption kinetic and facile regeneration property. The results indicate they are potential effective sorbents for removal and recovery of undesirable organic chemicals in water treatment, especially at high concentrations. Acknowledgments This work was supported by the National Natural Science Foundation of China (20703065, 20877097, and 20806089), the Ministry of Science and Technology of China (2008AA06Z324) and Chinese Universities Scientific Fund (2011JS160). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.09.025. References

Fig. 7. Sorption isotherms of phenol on P230 at pH 3, 6 and 11 at 25 ◦ C.

[1] M. Ahmaruzzaman, Adsorption of phenolic compounds on low-cost adsorbents: a review, Adv. Colloid Interface Sci. 143 (2008) 48–67. [2] G. Busca, S. Berardinelli, C. Resini, L. Arrighi, Technologies for the removal of phenol from fluid streams: a short review of recent developments, J. Hazard. Mater. 160 (2008) 265–288. [3] A. Bahdod, S. El Asri, A. Saoiabi, T. Coradin, A. Laghzizil, Adsorption of phenol from an aqueous solution by selected apatite adsorbents: kinetic process and impact of the surface properties, Water Res. 43 (2009) 313–318.

Z. Guan et al. / Journal of Hazardous Materials 196 (2011) 270–277 [4] C.R. Ispas, M.T. Ravalli, A. Steere, S. Andreescu, Multifunctional biomagnetic capsules for easy removal of phenol and bisphenol A, Water Res. 44 (2010) 1961–1969. [5] O. Chedeville, M. Debaccq, M.F.A. Almanza, C. Porte, Use of an ejector for phenol containing water treatment by ozonation, Sep. Purif. Technol. 57 (2007) 201–208. [6] G. Annadurai, R.S. Juang, D.J. Lee, Microbiological degradation of phenol using mixed liquors of Pseudomonas putida and activated sludge, Waste Manage. 22 (2002) 703–710. [7] N. Massalha, A. Shaviv, I. Sabbah, Modeling the effect of immobilization of microorganisms on the rate of biodegradation of phenol under inhibitory conditions, Water Res. 44 (2010) 5252–5259. [8] W. Cichy, J. Szymanowski, Recovery of phenol from aqueous streams in hollow fiber modules, Environ. Sci. Technol. 36 (2002) 2088–2093. [9] Z.M. Ahmed, S. Lyne, R. Shahrabani, Removal and recovery of phenol from phenolic wastewater via ion exchange and polymeric resins, Environ. Eng. Sci. 17 (2000) 245–255. [10] L.Z. Wang, Y.M. Zhao, J.F. Fu, The influence of TiO2 and aeration on the kinetics of electrochemical oxidation of phenol in packed bed reactor, J. Hazard. Mater. 160 (2008) 608–613. [11] R. Vinu, G. Madras, Kinetics of simultaneous photocatalytic degradation of phenolic compounds and reduction of metal ions with nano-TiO2 , Environ. Sci. Technol. 42 (2008) 913–919. [12] F.A. Banat, B. Al-Bashir, S. Al-Asheh, O. Hayajneh, Adsorption of phenol by bentonite, Environ. Pollut. 107 (2000) 391–398. [13] B.C. Pan, W. Du, W.M. Zhang, X. Zhang, Q.R. Zhang, B.J. Pan, L. Lv, Q.X. Zhang, J.L. Chen, Improved adsorption of 4-nitrophenol onto a novel hyper-cross-linked polymer, Environ. Sci. Technol. 41 (2007) 5057–5062. [14] L.Z. Zhu, B.L. Chen, X.Y. Shen, Sorption of phenol, p-nitrophenol, and aniline to dual-cation organobentonites from water, Environ. Sci. Technol. 34 (2000) 468–475. [15] D.H. Lin, B.S. Xing, Adsorption of phenolic compounds by carbon nanotubes: role of aromaticity and substitution of hydroxyl groups, Environ. Sci. Technol. 42 (2008) 7254–7259. [16] K.M. Smith, G.D. Fowler, S. Pullket, N.J.D. Graham, Sewage sludge-based adsorbents: a review of their production, properties and use in water treatment applications, Water Res. 43 (2009) 2569–2594. [17] E. Ayranci, O. Duman, Adsorption behaviors of some phenolic compounds onto high specific area activated carbon cloth, J. Hazard. Mater. 124 (2005) 125–132. [18] W. Lamoolphak, M. Goto, M. Sasaki, M. Suphantharika, C. Wangnapoh, C. Prommuag, A. Shotipruk, Hydrothermal decomposition of yeast cells for production of proteins and amino acids, J. Hazard. Mater. 137 (2006) 1643–1648. [19] D.Z. Ni, L. Wang, Y.H. Sun, Z.R. Guan, S. Yang, K.B. Zhou, Amphiphilic hollow carbonaceous microspheres with permeable shells, Angew. Chem.: Int. Ed. 49 (2010) 4223–4227. [20] L. Wang, D.Z. Ni, D. Yang, K.B. Zhou, S. Yang, Single-hole carbonaceous microcapsules, Chem. Lett. 39 (2010) 451–453. [21] A.M. Carmo, L.S. Hundal, M.L. Thompson, Sorption of hydrophobic organic compounds by soil materials: application of unit equivalent Freundlich coefficients, Environ. Sci. Technol. 34 (2000) 4363–4369. [22] B.L. Chen, D.D. Zhou, L.Z. Zhu, Transitional adsorption and partition of nonpolar and polar aromatic contaminants by biochars of pine needles with different pyrolytic temperatures, Environ. Sci. Technol. 42 (2008) 5137–5143. [23] G.F. Liu, J. Ma, X.C. Li, Q.D. Qin, Adsorption of bisphenol A from aqueous solution onto activated carbons with different modification treatments, J. Hazard. Mater. 164 (2009) 1275–1280.

277

[24] Y.S. Ho, J.C.Y. Ng, G. McKay, Kinetics of pollutant sorption by biosorbents: review, Sep. Purif. Methods 29 (2000) 189–232. [25] N. Baccile, G. Laurent, F. Babonneau, F. Fayon, M.M. Titirici, M. Antonietti, Structural Characterization of hydrothermal carbon spheres by advanced solid-state MAS C-13 NMR investigations, J. Phys. Chem. C 113 (2009) 9644–9654. [26] M.M. Titirici, M. Antonietti, N. Baccile, Hydrothermal carbon from biomass: a comparison of the local structure from poly- to monosaccharides and pentoses/hexoses, Green Chem. 10 (2008) 1204–1212. [27] B.L. Chen, E.J. Johnson, B. Chefetz, L.Z. Zhu, B.S. Xing, Sorption of polar and nonpolar aromatic organic contaminants by plant cuticular materials: role of polarity and accessibility, Environ. Sci. Technol. 39 (2005) 6138–6146. [28] G. Cornelissen, O. Gustafsson, T.D. Bucheli, M.T.O. Jonker, A.A. Koelmans, P.C.M. Van Noort, Extensive sorption of organic compounds to black carbon, coal, and kerogen in sediments and soils: mechanisms and consequences for distribution, bioaccumulation, and biodegradation, Environ. Sci. Technol. 39 (2005) 6881–6895. [29] S. Kumar, M. Zafar, J.K. Prajapati, S. Kumar, S. Kannepalli, Modeling studies on simultaneous adsorption of phenol and resorcinol onto granular activated carbon from simulated aqueous solution, J. Hazard. Mater. 185 (2011) 287–294. [30] K. Laszlo, P. Podkoscielny, A. Dabrowski, Heterogeneity of polymer-based active carbons in adsorption of aqueous solutions of phenol and 2,3,4-trichlorophenol, Langmuir 19 (2003) 5287–5294. [31] L. Damjanovic, V. Rakic, V. Rac, D. Stosic, A. Auroux, The investigation of phenol removal from aqueous solutions by zeolites as solid adsorbents, J. Hazard. Mater. 184 (2010) 477–484. [32] D.Q. Zhu, S. Kwon, J.J. Pignatello, Adsorption of single-ring organic compounds to wood charcoals prepared under different thermochemical conditions, Environ. Sci. Technol. 39 (2005) 3990–3998. [33] Y. Chun, G.Y. Sheng, C.T. Chiou, B.S. Xing, Compositions and sorptive properties of crop residue-derived chars, Environ. Sci. Technol. 38 (2004) 4649–4655. [34] C.T. Chiou, D.E. Kile, Deviations from sorption linearity on soils of polar and nonpolar organic compounds at low relative concentrations, Environ. Sci. Technol. 32 (1998) 338–343. [35] Y.N. Yang, G.Y. Sheng, Enhanced pesticide sorption by soils containing particulate matter from crop residue burns, Environ. Sci. Technol. 37 (2003) 3635–3639. [36] B. Chefetz, A.P. Deshmukh, P.G. Hatcher, E.A. Guthrie, Pyrene sorption by natural organic matter, Environ. Sci. Technol. 34 (2000) 2925–2930. [37] F.C. Spurlock, J.W. Biggar, Thermodynamics of organic-chemical partition in soils. 2. Nonlinear partition of substituted phenylureas from aqueous-solution, Environ. Sci. Technol. 28 (1994) 996–1002. [38] E.L. Difilippo, R.P. Eganhouse, Assessment of PDMS–water partition coefficients: implications for passive environmental sampling of hydrophobic organic compounds, Environ. Sci. Technol. 44 (2010) 6917–6925. [39] Y. Chen, J. Pawliszyn, Kinetics and the on-site application of standards in a solid-phase microextraction fiber, Anal. Chem. 76 (2004) 5807–5815. [40] B. Muijs, M.T.O. Jonker, Temperature-dependent bioaccumulation of polycyclic aromatic hydrocarbons, Environ. Sci. Technol. 43 (2009) 4517–4523. [41] A. DeryloMarczewska, A.W. Marczewski, A general model for adsorption of organic solutes from dilute aqueous solutions on heterogeneous solids: application for prediction of multisolute adsorption, Langmuir 13 (1997) 1245–1250. [42] M.C. Burleigh, M.A. Markowitz, M.S. Spector, B.P. Gaber, Porous polysilsesquioxanes for the adsorption of phenols, Environ. Sci. Technol. 36 (2002) 2515–2518.

Journal of Hazardous Materials 196 (2011) 302–310

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Partitioning behavior and stabilization of hydrophobically coated HfO2 , ZrO2 and Hfx Zr1−x O2 nanoparticles with natural organic matter reveal differences dependent on crystal structure Divina A. Navarro 1 , Sean W. Depner, David F. Watson, Diana S. Aga ∗ , Sarbajit Banerjee ∗∗ Department of Chemistry, University at Buffalo, State University of New York, Buffalo, NY 14260-3000, USA

a r t i c l e

i n f o

Article history: Received 11 April 2011 Received in revised form 7 September 2011 Accepted 8 September 2011 Available online 14 September 2011 Keywords: Twin-metal oxides Natural organic matter Phase transfer Nanoparticle structure Environmental mobility Colloidal interactions

a b s t r a c t The interactions of engineered nanomaterials with natural organic matter (NOM) exert a profound influence on the mobilities of the former in the environment. However, the influence of specific nanomaterial structural characteristics on the partitioning and colloidal stabilization of engineered nanomaterials in various ecological compartments remains underexplored. Herein, we present a systematic study of the interactions of humic acid (HA, as a model for NOM) with monodisperse, well-characterized, ligandpassivated HfO2 , ZrO2 , and solid-solution Hfx Zr1−x O2 nanoparticles (NPs). We note that mixing with HA induces the almost complete phase transfer of hydrophobically coated monoclinic metal oxide (MO) NPs from hexane to water. Furthermore, HA is seen to impart appreciable colloidal stabilization to the NPs in the aqueous phase. In contrast, phase transfer and aqueous-phase colloidal stabilization has not been observed for tetragonal MO-NPs. A mechanistic model for the phase transfer and aqueous dispersal of MO-NPs is proposed on the basis of evidence from transmission electron microscopy, ␨-potential measurements, dynamic light scattering, Raman and infrared spectroscopies, elemental analysis, and systematic experiments on a closely related set of MO-NPs with varying composition and crystal structure. The data indicate the synergistic role of over-coating (micellar), ligand substitution (coordinative), and electrostatic processes wherein HA acts both as an amphiphilic molecule and a charged chelating ligand. The strong observed preference for the phase transfer of monoclinic instead of tetragonal NPs indicates the importance of the preferential binding of HA to specific crystallographic facets and suggests the possibility of being able to design NPs to minimize their mobilities in the aquatic environment. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The imminent large-scale commercialization of engineered nanomaterials (ENMs) has raised concerns regarding their potential environmental impact [1–4]. Some preliminary data are starting to become available regarding the toxicity of ENMs at the sub-cellular, cellular, and organism levels [5–8]. Among these materials, transition metal oxide (MO)-based nanoparticles (NPs) are finding various applications as nanoceramic fillers within composite materials, magnetic recording media, catalyst supports, and sensing elements [9]. Most notably, HfO2 and ZrO2 NPs have found widespread applications in optical and protective coating

∗ Corresponding author. Tel.: +1 716 645 4220; fax: +1 716 645 6963. ∗∗ Corresponding author. Tel.: +1 716 645 4140; fax: +1 716 645 6963. E-mail addresses: [email protected] (D.S. Aga), [email protected] (S. Banerjee). 1 Current address: CSIRO Land and Water, Advanced Materials Transformational Capability Platform, Nanosafety, Biogeochemistry Program, Waite Campus, Waite Rd, SA 5155, Australia. 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.028

technologies due to their thermal stability and high dielectric constants [10]. In particular, these materials are promising alternatives to SiO2 as gate dielectric layers for flexible electronics [11–13]. The underlying premise of flexible electronics has been affordability and ubiquitous availability on standard media such as paper and cloth. Consequently, with increasing commercial production, consumer use, and end-user disposal, the release of these NPs to different environmental compartments (especially water and soil) is inevitable. In particular, waste generated during manufacturing processes will have a high concentration of NPs. Although, the likelihood of the release of MO NPs affixed within device structures is not high, environmental discharge may occur over a protracted period of time, especially as a result of material abuse and towards the end of product life. Given the low proposed cost of flexible electronic devices and lack of specifications regarding disposal outlined by manufacturers, several of these materials may eventually enter landfills and run-off streams through household waste disposal. A systematic understanding of partitioning behavior, potential mobilities, and persistence of MO NPs is thus necessary for

D.A. Navarro et al. / Journal of Hazardous Materials 196 (2011) 302–310

evaluating potential ecological hazards and framing informed policy [1]. Up till now, the environmental fate and transport of ENMs have been characterized under different environmental conditions (i.e., pH, ionic strength, organic colloids, etc.). The influence of natural organic matter (NOM) on NP behavior in the environment has been emphasized in many studies because of the ubiquity of the former in aquatic and soil environments. Indeed, the nature and amount of NOM in water have been demonstrated (both theoretically and experimentally) to affect the stability and bioavailability of a variety of NPs [14–20]. In many cases, the stability of NPs in aqueous suspension is often attributed to the adsorption of NOM. However, these reports only tend to deem adsorption as a primary mechanism for interaction based on indirect measures such as transmission electron microscopy (TEM), ␨-potential, and light scattering, which are somewhat limited in characterizing surfacerelated interactions. An evaluation of the chemical structure of NOM and colloidal MO-NPs suggests several distinctive processes that can facilitate the stabilization of NPs in the aqueous phase. NOM has a highly complex molecular structure, which includes a skeleton of alkyl and aromatic units with pendant functional groups including carboxylic acid, phenolic hydroxyl, and quinone moieties [21–23]. MO-NPs, on the other hand, comprise an inorganic core passivated by a layer of organic ligands [11,24]. The crystalline core can adopt different crystal structures depending on the specifics of composition and stoichiometry (HfO2 and ZrO2 NPs adopt monoclinic and tetragonal crystal structures, respectively) [25,11]. The intrinsic morphological, energetic, and surface chemical properties of NOM enable them to interact with and stabilize different species via amphiphilic and metal-chelating processes. In addition to the organic ligands that surround the crystalline core, MO-NPs also have highly reactive surfaces (edge and corner sites) that greatly influence their behavior and reactivity [26–28]. While these structural and surface characteristics are well known to be important in surface science, these details have typically been overlooked in many fate and transport studies. Very few studies thus far have focused on the transformations of ligand-passivated NPs prepared by hot colloidal chemistry methods upon interactions with NOM. Such ligand-capped NPs are indeed likely to be the mainstay of most nanoscience-enabled technologies [29–32]. Herein, we describe systematic studies on the interaction of ligand-passivated HfO2 , ZrO2 , and Hfx Zr1−x O2 NPs with Suwannee River humic acid (HA) as a model for NOM. The following topics have been addressed in this work: (1) the partitioning of hydrophobically coated MO-NPs, with or without HA, in the aqueous phase; (2) examination of interactions that enable HA to colloidally stabilize MO-NPs (i.e., electrostatic, coordinative, and dispersive interactions); and (3) the importance of both NP surface (crystal structure: monoclinic or tetragonal) and passivating ligands (surface coating: tri-n-octylphosphine oxide (TOPO)) on the stabilization of NPs by HA in water; TOPO is a ubiquitous ligand in nanoscience and is commonly used in hot colloidal synthesis. This study provides a mechanistic report of the aqueous-phase stabilization of different types of hydrophobically coated MO-NPs with and without HA. In particular, we have attempted to directly characterize processes that govern the adsorption and agglomeration of MO-NPs with HA.

303

were synthesized by the non-hydrolytic sol–gel condensation of metal alkoxides with metal halides using TOPO (Strem Chemicals, MA, USA) as the coordinating ligand [11]. This synthetic approach provides monodispersity and excellent control over crystal structure, size, and stoichiometry. The TOPO ligand coordinates to surficial atoms, completing the coordination shell for undercoordinated metal sites, and thereby serving as a passivating coating. The use of monodisperse systems provides standardization required for careful mechanistic studies and precludes obscuration from polydispersities in particle size, surface capping, and crystal structure, which would substantially complicate studies of MO-NPs with NOM. All MO-NP powders were readily dispersible in hexane. Approximately 500–1000 mg/L NP suspensions were prepared and used for the phase transfer experiments; these concentrations were chosen for ease in detection of NPs. No consensus has yet emerged on what constitutes an environmentally realistic concentration of MO NPs. The use of the said concentrations allows us to deploy standard microscopy and spectroscopy tools for elucidation of the nature of NP–NOM interactions and indeed such interactions are likely to persist even at low concentrations. For aqueous solubility tests, NP powders were dispersed in water. Suwannee River HA (SRHA-II) standard was purchased from International Humic Substances Society (St. Paul, MN, USA). The use of well-characterized SRHA-II also enables standardization, articulated at various international workshops as an urgent goal for establishing generalizable means of evaluation of ENM fate and transport. Given our primary goal of elucidating the mechanistic basis for ENM phase transfer, the use of well-characterized NOM acquires paramount importance. For Hf/Zr analysis, metal standards (fluoride-soluble metals) and Aristar Ultra grade concentrated HF and HNO3 from BDH Chemicals (West Chester, PA, USA) were used in standard preparation and acid digestions, respectively. Deionized (DI) water from a Barnstead NANOpure (USA) water system was used to prepare all aqueous solutions (resistivity = 18.2 M/cm). 2.2. Phase transfer experiments A 5-mL aliquot of the MO-NP suspension (in hexane) was mixed with 5 mL of 20 mg/L HA in DI water (pH ∼ 4.4) in a clear vial. This experimental construct is referred to as a “phase transfer set-up”. A 20 mg/L HA solution contains 12.5 mg/L dissolved organic carbon that is within levels typical of natural waters (0.1–200 mg/L) [33]. The low natural pH used here is also representative of the low pH of the Suwanee River where the HA was sampled. Similar conditions were used in our previous work on CdSe QDs [30,31]. Phase transfer set-ups were also prepared in DI water (no HA) to serve as controls. Our intention was to study the interactions of MO-NPs with analogs of actual environmental samples, containing controlled concentrations of HA. In between measurements, each set-up was protected from light using Al foil, and was stirred continuously at room temperature using a rocking platform shaker to stimulate natural mixing and fluid diffusion processes. Mixing was performed for 15 days (∼2 weeks). For metal analyses, individual phase transfer set-ups (0, 1, 3, 5, 10, and 15 days) were prepared; this avoids sampling errors due to interfacial aggregation (i.e., removal of flocculated aggregates as the volume of solution is reduced).

2. Experimental

2.3. Analysis and instrumentation

2.1. Materials

Qualitative and quantitative analyses were performed using ␨-potential measurements, dynamic light scattering (DLS), TEM, Raman and Fourier transform infrared (FTIR) spectroscopies and inductively coupled plasma mass spectrometry (ICP-MS).

The monoclinic (m-) HfO2 and Hf0.37 Zr0.63 O2 , and tetragonal (t) ZrO2 and Hf0.37 Zr0.63 O2 NP powders used in these experiments

304

D.A. Navarro et al. / Journal of Hazardous Materials 196 (2011) 302–310

Fig. 1. HRTEM images of (A1–2) the m-HfO2 and t-ZrO2 NPs and (B3–4) the HA-transferred MO-NPs in aqueous solution. Insets (i–ii) highlight the predominant crystal planes of the NPs. Lattice spacings were assigned based on monoclinic (JCPDS# 780050) and tetragonal (JCPDS# 881997) structures.

␨-Potential and DLS data were acquired using a Zetasizer Nano ZS90 instrument (Malvern Instruments, Malvern Hills, UK). TEM measurements were performed using a JEOL JEM-2010 (Tokyo, Japan) operating at an accelerating voltage of 200 kV. Raman spectra were acquired at room temperature using a Horiba JobinYvon (Villeneuve d’Ascq, France) Labram HR Raman spectrometer using 784.51 nm laser excitation from a diode laser. FTIR spectra were collected using a Nicolet-Magna (USA) 550 spectrometer purged with dry air with a spectral resolution of 4 cm−1 . Total Hf and Zr concentrations present in the organic and aqueous phase were quantitatively determined by ICP-MS. ICP-MS measurements were conducted using a Thermo Scientific (Germany) X-Series 2 instrument. Concentrations of Hf and/or Zr in the samples were determined using an external calibration curve. Details on sampling and the acid digestion protocol are described in the Supporting information (SI).

3. Results and discussion 3.1. Characteristics of the MO-NPs Figs. 1A and S1A depicts TEM and lattice-resolved high-resolution TEM (HRTEM) images of m-HfO2 , t-ZrO2 , mHf0.37 Zr0.67 O2 , and t-Hf0.37 Zr0.67 O2 NPs. All NP surfaces are passivated with TOPO and phosphonate ligands. The m-HfO2 particles adopt an elongated rice-grain-type morphology with aspect ratios ranging from 3 to 4. In contrast, the t-ZrO2 NPs adopt a quasi-spherical morphology with an average diameter of 3.3 nm. Solid-solution m-Hf0.37 Zr0.67 O2 NPs are slightly elongated, whereas t-Hf0.37 Zr0.67 O2 NPs are quasi-spherical [11]. Representative XRD patterns of the MO-NPs used in the phase transfer experiments are shown in Fig. S2. Figs. 1A and S1A demonstrate the monodispersity of the NPs used. The HRTEM images indicate

D.A. Navarro et al. / Journal of Hazardous Materials 196 (2011) 302–310

the exposed crystal facets for each set of particles. The monoclinic NPs consistently show a preference for {1 0 0} and {1 1 1} crystal facets, whereas the tetragonal NPs predominantly exhibit {1 0 1} surface-terminating planes. These assignments are consistent with surface energy calculations for m-HfO2 and t-ZrO2 , which indicate preferential energetic stabilization for these planes in the monoclinic and tetragonal crystal structures [34,35]. Given the hydrophobic nature of TOPO, it would be reasonable to expect to first approximation that such NPs will have insignificant dispersibilities or mobilities in the aquatic environment. Indeed, these NPs exhibit very low aqueous solubilities: 2 weeks. In the absence of HA, phase transfer occurs at a relatively slower rate and the transferred MONPs tend to flocculate and eventually settle at the bottom of the vial. Indeed, the TEM images, ␨-potential, DLS, and FTIR spectral data all suggest direct interaction between the MO-NPs and HA. Three distinctive interactions of HA with the MO-NPs leading to aqueous phase dispersal and stabilization can be envisaged [30,31,45]. The first process involves overcoating of the MO-NPs with HA through non-specific adsorption of humics onto the TOPO ligand shell. The amphiphilic characteristics of HA enable the formation of pseudomicellar agglomerates that overcoat the NPs, wherein hydrophobic aromatic and heteroaliphatic regions form a hydrophobic interior cavity, whereas pendant hydrophilic carboxylic acid, phenolic, and amine moieties are directed outwards imparting colloidal stability in water via electrostatic stabilization [38,46]. Provided that the NP surface ligands are engaged in this interaction, this mechanism is not expected to show any discrimination between monoclinic and tetragonal crystal structures. The second process involves dative interactions between HA and the MO surfaces, where the carboxylic acid (estimated to be 10% in SRHA-II, [47]), phenolic, and amine moieties that are abundant in the HA structure serve as versatile polydentate chelating ligands [38,48–51]. Interaction involving the NP surface is conceivable given the highly reactive surfaces of nanoscale materials, wherein most of the constituent atoms reside at or near the surface [27,28,52]. This mechanism is predicated on the availability and accessibility of metal sites on the MO-NP surface that can participate in coordinative interactions. Different crystal structures have distinctive planes, corners, and edges exposed at the surface (Fig. 1). Hence, in contrast to the overcoating mechanism, this interaction can possibly provide some discrimination between monoclinic and tetragonal crystal structures. The third process involves electrostatic interactions between the coordinatively unsaturated surface sites on the MO-NPs (positively and negatively charged) and the carboxylic and phenolic moieties (negatively charged) in the HA structure. these processes (micellar/overcoating, coordinaAll tive/substitutional, and electrostatic interactions) appear to work synergistically to facilitate the phase transfer and subsequent aqueous-phase dispersal of monoclinic MO-NPs passivated with TOPO but not the tetragonal NPs. The differences in the phase transfer behavior observed between the two different polymorphs suggest that specific rather than non-specific interactions mediate

phase transfer; in this case, the surface-structure-dependent coordinative/subsitutional interactions are the likely genesis of the distinctive reactivity. Upon mixing with H2 O/HA, some of the TOPO ligands are likely displaced; FTIR spectra of phasetransferred MO-HA agglomerates suggest that the amount of TOPO is diminished compared to the TOPO-coated NPs. Removal of TOPO provides access to oxophilic, coordinatively unsaturated cationic sites on the MO-NP surfaces that can be accessed by the carboxylic acid moieties of HA (electrostatic interactions likely induce the initial approach of the HA moieties and NPs). Apart from ligand displacement, incomplete coverage/passivation of the initial NPs by TOPO would also make Hf/Zr surface sites available for binding to HA. Impurities in technical grade TOPO (90%), particularly alkylphosphonic and alkylphosphinic acids that have been shown to play an active role in passivating surfaces of CdSe quantum dots, rods, and wires [53–56], could also influence the interactions between HA and the NP surface [31]. The fractional surface coverage of the coating groups and the precise nature of the ligand passivation shell are beyond the scope of this study (ligand passivation shells remain to be adequately characterized even for CdSe quantum dots that are possibly the most mature of this class of materials). Nonetheless, as suggested by FTIR, the formation of metal–humate linkages tethers the MO-NPs to the humic colloids, which likely draws the NPs to the hexane/water interface such that both the hydrophobic NPs and the hydrophilic HA colloids can be adequately solvated. Subsequently, as described in the literature [46,49,57], the flexible humic colloid can undergo molecular rearrangement and cross-linking with proximal HA moieties at the hexane/water interface to optimize hydrophobic interactions with the pendant aliphatic chains on the TOPO ligand [41–43,58]. In addition, our results indicate that H2 O by itself enables some phase transfer of NPs. The oxophilicity of early transition metal oxide surfaces may also allow for facile ligand substitution by H2 O, which can eventually result in appreciable hydroxylation of the MO surfaces. The affinities of different crystallographic facets for HA may reasonably be assumed to parallel the likelihood of ligand substitution by H2 O, which may explain the significant phase transfer observed for the monoclinic NPs even in the absence of HA. Phase transfer only shows differences between H2 O and HA on shorter timescales, when the MO-H2 O interaction is likely limited by ligand substitution. The interfacial turbidity and shorter-lived phase transfer noted in control samples in the absence of HA likely arise from the displacement of some TOPO ligands by H2 O. Consistent with the proposed mechanism, since surface-coordinated H2 O molecules lack the amphiphilic characteristics of HA, they are not able to adequately stabilize the TOPO-coated MO-NPs in the aqueous phase. In other words, ligand substitution of TOPO by H2 O can induce sedimentation at the hexane/water interface but does not permit colloidal stabilization in the aqueous phase in the absence of HA. Preferential binding of HA or H2 O to monoclinic instead of tetragonal surfaces may form the basis for the observed selectivity of phase transfer such as between m-HfO2 and t-ZrO2 NPs, and between m-Hf0.37 Zr0.63 O2 and t-Hf0.37 Zr0.63 O2 . As shown in Fig. 1A, the m-HfO2 (and m-Hf0.37 Zr0.63 O2 ) NPs preferentially expose {1 0 0} and {1 1 1} surfaces, whereas for t-ZrO2 (and tHf0.37 Zr0.63 O2 ) NPs {1 0 1} surfaces are energetically preferred. While the difference between m-HfO2 and t-ZrO2 NPs with regard to phase transfer could be related to the extent/strength of the Hf-HA bonds vs. the Zr-HA bonds, results from m-Hf0.37 Zr0.63 O2 and t-Hf0.37 Zr0.63 O2 NPs (same chemical composition) suggest that phase transfer is more responsive to changes in crystal structure. Consequently, despite both having potentially reactive surfaces, we speculate that the differences in phase transfer of the NPs may originate from (a) the relative binding affinities of the different surfaces in monoclinic and tetragonal NPs for TOPO and HA; (b) the degree of

D.A. Navarro et al. / Journal of Hazardous Materials 196 (2011) 302–310

coordinative unsaturation and steric hindrance for transition metal sites in the different surface planes; and (c) the density of exposed transition metal cation sites. A combination of these factors could make metal sites less available and the displacement of TOPO ligands by HA functionalities more difficult for tetragonal NPs. In the absence of metal–humate linkages, phase transfer may not be as readily initiated resulting in the poor phase transfer efficiencies observed for t-ZrO2 and t-Hf0.37 Zr0.63 O2 NPs. An analogous preference for binding different surfaces is also expected for coordinative interactions with H2 O molecules. Calculation of binding affinities, extent of surface unsaturation, and density of cation and anions on the NP surface is beyond the scope of this study; these measurements have indeed not been experimentally validated for ligand-passivated colloidal NP systems with any degree of accuracy. Nonetheless, our extensive characterization data is adequate to draw some conclusions with regard to the mechanisms that dictate phase transfer and aqueous phase stabilization of these NPs. 4. Conclusions The interactions between HA and MO-NPs indicate the role of HA as both a coordinating ligand and an amphiphilic surfactant. Our results further provide experimental validation of theoretical predictions [14] and experimental observations [15–17] of modifications to the colloidal stability of MO-NPs upon the acquisition of NOM coatings. In this study, HA and H2 O both exhibit a distinct preference for monoclinic rather than tetragonal MONPs, possibly because of stronger binding affinities to monoclinic surfaces (coordinative/substitutional interactions) and the ease of formation of cylindrical pseudo-micellar structures (overcoating/micellar interaction). The extent of phase transfer and degree of colloidal stabilization observed for well-defined MO-NPs with hydrophobic coatings in the presence of HA also underline the importance of developing a detailed understanding of the potential environmental transformations of ENPs. The distinctive selectivity in the HA-induced phase transfer and dispersion of different polymorphs suggests that interactions of different inorganic ENMs with NOM, which dictate the extent of NP transport and stabilization, are not necessarily generalizable and that it may be possible to design NMs to minimize their residence time in the aquatic environment. In this regard, further research is required to determine the actual binding affinities of HA onto monoclinic and tetragonal surfaces and to investigate the influence of different surface coatings. Acknowledgements This work was primarily supported by the US Environmental Protection Agency (Grant# R833861). SB acknowledges partial support of this work from National Science Foundation under DMR 0847169. We acknowledge the NSF MRI Program CHE 0959565 for acquisition of the ICP-MS instrument. Both CSIRO and the US EPA have not subjected this manuscript to internal peer and policy review. Therefore, no official endorsement should be inferred. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.09.028. References [1] P.J.J. Alvarez, V.L. Colvin, J. Lead, V. Stone, Research priorities to advance ecoresponsible nanotechnology, ACS Nano 3 (2009) 1616–1619. [2] A. Maynard, R.J. Aitken, T. Butz, V. Colvin, K. Donaldson, G. Oberdorster, M.A. Philbert, J. Ryan, A. Seaton, V. Stone, S.S. Tinkle, L. Tran, N.J. Walker, D.B. Warheit, Safe handling of nanotechnology, Nature 444 (2006) 267–269.

309

[3] M.R. Wiesner, G.V. Lowry, P. Alvarez, D. Dionysiou, P. Biswas, Assessing the risks of manufactured nanomaterials, Environ. Sci. Technol. 40 (2006), 4336–4335. [4] J.R. Peralta-Videa, L.J. Zhao, M.L. Lopez-Moreno, G. de la Rosa, J. Hong, J.L. Gardea-Torresdey, Nanomaterials and the environment: a review for the biennium 2008–2010, J. Hazard. Mater. 186 (2011) 1–15. [5] E. Casals, S. Vazquez-Campos, N.G. Bastus, V. Puntes, Distribution and potential toxicity of engineered inorganic nanoparticles and carbon structures in biological systems, TrAC, Trends Anal. Chem. 27 (2008) 672–684. [6] S. Harper, C. Usenko, J.E. Hutchison, B.L.S. Maddux, R.L. Tanguay, In vivo biodistribution and toxicity depends on nanomaterial composition, size, surface functionalisation and route of exposure, J. Exp. Nanosci. 3 (2008) 195–206. [7] L.S. Karlsson, J. Gustafsson, P. Cronholm, L. Moller, Size-dependent toxicity of metal oxide particles: a comparison between nano- and micrometer size, Toxicol. Lett. 188 (2008) 112–118. [8] M. Farre, K. Gajda-Schrantz, L. Kantiani, D. Barcelo, Ecotoxicity and analysis of nanomaterials in the aquatic environment, Anal. Bioanal. Chem. 393 (2009) 81–95. [9] A.M.E. Maynard, The Nanotechnology Consumer Products Inventory, Woodrow Wilson International Center for Scholars, The PEW Charitable Trusts, 2006. [10] A.I. Kingon, J.-P. Maria, S.K. Streiffer, Alternative dielectrics to silicon dioxide for memory and logic devices, Nature (London U. K.) 406 (2000) 1032–1038. [11] S.W. Depner, K.R. Kort, S. Banerjee, Precursor control of crystal structure and stoichiometry in twin metal oxide nanocrystals, CrystEngComm 11 (2009) 841–846. [12] A. Facchetti, M.-H. Yoon, T.J. Marks, Gate dielectrics for organic field-effect transistors: new opportunities for organic electronics, Adv. Mater. (Weinheim, Germany) 17 (2005) 1705–1725. [13] J. Tang, J. Fabbri, R.D. Robinson, Y. Zhu, I.P. Herman, M.L. Steigerwald, L.E. Brus, Solid-solution nanoparticles: use of a nonhydrolytic sol–gel synthesis to prepare HfO2 and Hfx Zr1−x O2 nanocrystals, Chem. Mater. 16 (2004) 1336–1342. [14] N.T. Loux, N. Savage, An assessment of the fate of metal oxide nanomaterials in porous media, Water Air Soil Pollut. 194 (2008) 227–241. [15] S.-W. Bian, I.A. Mudunkotuwa, T. Rupasinghe, V.H. Grassian, Aggregation and dissolution of 4 nm ZnO nanoparticles in aqueous environments: influence of pH, ionic strength, size, and adsorption of humic acid, Langmuir 27 (2011) 6059–6068. [16] J.D. Hu, Y. Zevi, X.M. Kou, J. Xiao, X.J. Wang, Y. Jin, Effect of dissolved organic matter on the stability of magnetite nanoparticles under different pH and ionic strength conditions, Sci. Total Environ. 408 (2010) 3477–3489. [17] Y. Zhang, Y.S. Chen, P. Westerhoff, J. Crittenden, Impact of natural organic matter and divalent cations on the stability of aqueous nanoparticles, Water Res. 43 (2009) 4249–4257. [18] H. Hyung, J.D. Fortner, J.B. Hughes, J.-H. Kim, Natural organic matter stabilizes carbon nanotubes in the aqueous phase, Environ. Sci. Technol. 41 (2007) 179–184. [19] E. Navarro, A. Baun, R. Behra, N.B. Hartmann, J. Filser, A.-J. Miao, A. Quigg, P.H. Santschi, L. Sigg, Environmental behavior and ecotoxicity of engineered nanoparticles to algae, plants, and fungi, Ecotoxicology 17 (2008) 372–386. [20] A.A. Keller, H.T. Wang, D.X. Zhou, H.S. Lenihan, G. Cherr, B.J. Cardinale, R. Miller, Z.X. Ji, Stability, Aggregation of metal oxide nanoparticles in natural aqueous matrices, Environ. Sci. Technol. 44 (2010) 1962–1967. [21] M.F. Benedetti, C.J. Milne, D.G. Kinniburgh, W.H. Van Riemsdijk, L.K. Koopal, Metal ion binding to humic substances: application of the non-ideal competitive adsorption model, Environ. Sci. Technol. 29 (1995) 446–457. [22] E.M. Perdue, J.H. Reuter, R.S. Parrish, A statistical model of proton binding by humus, Geochim. Cosmochim. Acta 48 (1984) 1257–1263. [23] D.G. Lumsdon, A.R. Fraser, Infrared spectroscopic evidence supporting heterogeneous site binding models for humic substances, Environ. Sci. Technol. 39 (2005) 6624–6631. [24] S.W. Depner, K.R. Kort, C. Jaye, D.A. Fischer, S. Banerjee, Nonhydrolytic synthesis and electronic structure of ligand-capped CeO2-␦ and CeOCl nanocrystals, J. Phys. Chem. C 113 (2009) 14126–14134. [25] R.D. Robinson, J. Tang, M.S. Steigerwald, L.E. Brus, I.P. Herman, Raman scattering in Hfx Zr1−x O2 nanoparticles, Phys. Rev. B Condens. Matter Mater. Phys. 71 (2005) 115408. [26] D. Schodek, P. Ferreira, M. Ashby, Nanomaterials Nanotechnologies and Design, Elsevier, Burlington, MA, 2009. [27] I.A. Mudunkotuwa, V.H. Grassian, The devil is in the details (or the surface): impact of surface structure and surface energetics on understanding the behavior of nanomaterials in the environment, J. Environ. Monit. 13 (2011) 1135–1144. [28] V.H. Grassian, When size really matters: size-dependent properties and surface chemistry of metal and metal oxide nanoparticles in gas and liquid phase environments, J. Phys. Chem. C 112 (2008) 18303–18313. [29] A.M. Derfus, W.C.W. Chan, S.N. Bhatia, Probing the cytotoxicity of semiconductor quantum dots, Nano Lett. 4 (2004) 11–18. [30] D.A. Navarro, D.F. Watson, D.S. Aga, S. Banerjee, Natural organic mattermediated phase transfer of quantum dots in the aquatic environment, Environ. Sci. Technol. 43 (2009) 677–682. [31] D.A. Navarro, S. Banerjee, D.S. Aga, D.F. Watson, Partitioning of hydrophobic CdSe quantum dots into aqueous dispersions of humic substances: influence of capping-group functionality on the phase-transfer mechanism, J. Colloid Interface Sci. 348 (2010) 119–128. [32] W.W. Yu, E. Chang, J.C. Falkner, J. Zhang, A.M. Al-Somali, C.M. Sayes, J. Johns, R. Drezek, V.L. Colvin, Forming biocompatible and nonaggregated nanocrystals in water using amphiphilic polymers, J. Am. Chem. Soc. 129 (2007) 2871–2879.

310

D.A. Navarro et al. / Journal of Hazardous Materials 196 (2011) 302–310

[33] J.A. Leenheer, J.-P. Croué, Characterizing aquatic dissolved organic matter, Environ. Sci. Technol. 37 (2003) 18A–26A. [34] I.M. Iskandarova, A.A. Knizhnik, E.A. Rykova, A.A. Bagatur‘yants, B.V. Potapkin, A.A. Korkin, First-principle investigation of the hydroxylation of zirconia and hafnia surfaces, Microelectron. Eng. 69 (2003) 587–593. [35] A.B. Mukhopadhyay, J.F. Sanz, C.B. Musgrave, First-principles calculations of structural and electronic properties of monoclinic hafnia surfaces, Phys. Rev. B Condens. Matter Mater. Phys. 73 (2006) 115330. [36] F.R. Rizzi, S. Stoll, N. Senesi, J. Buffle, A transmission electron microscopy study of the fractal properties and aggregation processes of humic acids, Soil Sci. 169 (2004) 765–775. [37] S. Jia, S. Banerjee, I.P. Herman, Mechanism of the electrophoretic deposition of CdSe nanocrystal films: influence of the nanocrystal surface and charge, J. Phys. Chem. C 112 (2008) 162–171. [38] B. Sutton, G. Sposito, Molecular structure in soil humic substances: the new view, Environ. Sci. Technol. 39 (2005) 9009–9015. [39] K.M. Elkins, D.J. Nelson, Spectroscopic approaches to the study of the interaction of aluminum with humic substances, Coord. Chem. Rev. 228 (2002) 205–225. [40] SciFinder Scholar Substance Identifier as calculated using Advanced Chemistry Development (ACD/Labs) software V11.02 © 1994–2011, in. [41] C.T. Chiou, D.E. Kile, T.I. Brinton, R.L. Malcolm, J.A. Leenheer, P. MacCarthy, A comparison of water solubility enhancements of organic solutes by aquatic humic materials and commercial humic acids, Environ. Sci. Technol. 21 (1987) 1231–1234. [42] H.-H. Cho, J.-W. Park, C.C.K. Liu, Effect of molecular structures on the solubility enhancement of hydrophobic organic compounds by environmental amphiphiles, Environ. Toxicol. Chem. 21 (2002) 999–1003. [43] D.E. Kile, C.T. Chiou, Water solubility enhancements of DDT and trichlorobenzene by some surfactants below and above the critical micelle concentration, Environ. Sci. Technol. 23 (1989) 832–838. [44] N. Rawat, S. Kumar, B. Tomar, The time differential perturbed angular correlation study of binding of hafnium to humic acid and its model compound, Polyhedron 28 (2009) 1399–1402. [45] S. Diegoli, A.L. Manciulea, S. Begum, I.P. Jones, J.R. Lead, J.A. Preece, Interaction between manufactured gold nanoparticles and naturally occurring organic macromolecules, Sci. Total Environ. 402 (2008) 51–61.

[46] R.L. Wershaw, A new model for humic materials and their interactions with hydrophobic organic chemicals in soil–water or sediment–water systems, J. Contam. Hydrol. 1 (1986) 29–45. [47] 13C NMR Estimates of Carbon Distribution in IHSS Samples, in, International Humic Substances Society. [48] M. Baalousha, M. Montelica-Heino, P. Le Coustumer, Conformation and size of humic substances: effects of major cation concentration and type pH, salinity, and residence time, Colloids Surf. A 272 (2006) 48–55. [49] C.E. Clapp, M.H.B. Hayes, Sizes and shapes of humic substances, Soil Sci. 164 (1999) 777–789. [50] M. Schnitzer, S.I.M. Skinner, Organometallic interactions in soils. III. Properties of iron- and aluminum-organic matter complexes prepared in the laboratory and extracted from a soil, Soil Sci. 98 (1964) 197–203. [51] M. Schnitzer, S.I.M. Skinner, Organo metallic interactions in soils. IV. Carboxyl and hydroxyl groups in organic matter and metal retention, Soil Sci. 99 (1965) 278–284. [52] C. Burda, X. Chen, R. Narayanan, M.A. El-Sayed, Chemistry and properties of nanocrystals of different shapes, Chem. Rev. (Washington DC, United States) 105 (2005) 1025–1102. [53] J.T. Kopping, T.E. Patten, Identification of acidic phosphorus-containing ligands involved in the surface chemistry of CdSe nanoparticles prepared in tri-noctylphosphine oxide solvents, J. Am. Chem. Soc. 130 (2008) 5689–5698. [54] F. Wang, R. Tang, W.E. Buhro, The trouble with TOPO; identification of adventitious impurities beneficial to the growth of cadmium selenide quamtum dots, rods, and wires, Nano Lett. 8 (2008) 3521–3524. [55] F. Wang, R. Tang, J.L.-F. Kao, S.D. Dingman, W.E. Buhro, Spectroscopic identification of tri-n-octylphosphine oxide (TOPO) impurities and elucidation of their roles in cadmium selenide quantum-wire growth, J. Am. Chem. Soc. 131 (2009) 4983–4994. [56] J.S. Owen, J. Park, P.-E. Trudeau, A.P. Alivisatos, Reaction chemistry and ligand exchange at cadmium–selenide nanocrystal surfaces, J. Am. Chem. Soc. 130 (2008) 12279–12281. [57] L.T. Sein, J.M. Varnum, S.A. Jansen, Conformational modeling of a new building block of humic acid: approaches to the lowest energy conformer, Environ. Sci. Technol. 33 (1999) 546–552. [58] Y. Lu, J.J. Pignatello, Sorption of apolar aromatic compounds to soil humic acid particles affected by aluminum(III) ion crosslinking, J. Environ. Qual. 33 (2004) 1314–1321.

Journal of Hazardous Materials 196 (2011) 311–317

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Reduction of As(V) to As(III) by commercial ZVI or As(0) with acid-treated ZVI Fenglong Sun a , Kwadwo A. Osseo-Asare b , Yongsheng Chen c , Brian A. Dempsey a,∗ a

Department of Civil and Environmental Engineering, Penn State University, UP, United States Department of Materials Science and Engineering, Penn State University, UP, United States c Department of Energy and Mineral Engineering, Penn State University, UP, United States b

a r t i c l e

i n f o

Article history: Received 24 April 2011 Received in revised form 7 September 2011 Accepted 8 September 2011 Available online 14 September 2011 Keywords: Arsenic Zero-valent iron Reduction X-ray absorption spectroscopy (XAS)

a b s t r a c t Zero-valent iron (ZVI) consists of an elemental iron core surrounded by a shell of corrosion products, especially magnetite. ZVI is used for in situ removal or immobilization of a variety of contaminants but the mechanisms for removal of arsenic remain controversial and the mobility of arsenic after reaction with ZVI is uncertain. These issues were addressed by separately studying reactions of As(V) with magnetite, commercial ZVI, and acid-treated ZVI. Strictly anoxic conditions were used. Adsorption of As(V) on magnetite was fast with pH dependence similar to previous reports using oxic conditions. As(V) was not reduced by magnetite and Fe(II) although the reaction is thermodynamically spontaneous. As(V) reactions with ZVI were also fast and no lag phase was observed which was contrary to previous reports. Commercial ZVI reduced As(V) to As(III) only when As(V) was adsorbed, i.e., for pH < 7. As(III) was not released to solution. Acid-treated ZVI reduced As(V) to As(0), shown using wet chemical analyses and XANES/EXAFS. Comparisons were drawn between reactivity of acid-treated ZVI and nano-ZVI; if true then acid-treated ZVI could provide similar reactive benefits at lower cost. © 2011 Elsevier B.V. All rights reserved.

1. Introduction: High arsenic concentrations in groundwater have been reported throughout the world, notably in Bangladesh and Taiwan [1]. Arsenic in aquatic environments usually exists in inorganic forms as arsenate (As(V)) and arsenite (As(III)). As(III) is usually more mobile and toxic than As(V) although As(III) can also become immobilized in the presence of sulfide. Elevated arsenic concentrations in groundwater can occur due to reductive dissolution of ferric oxide sorbents and consequent reduction and mobilization of As(III), desorption of As(V) under alkaline pH conditions especially in the presence of phosphate or other competing adsorbates, or oxidation of sulfidic materials [1,2]. Zero-valent iron (ZVI) has been used to remove organic and inorganic contaminants including chlorinated solvents, nitrate, uranyl ion, chromate, lead, and arsenic [3,4]. ZVI can be incorporated into permeable reactive barriers or nano-ZVI (nZVI) can be injected into contaminated soils [5]. ZVI is also found in some point-ofuse potable water treatment systems [6]. ZVI is usually reported to have a core–shell structure. The shell contains oxidized iron that is mostly magnetite and often with maghemite (␥-Fe2 O3 ) or lepidocrocite (␥-FeOOH) [3,7–10].

∗ Corresponding author. Tel.: +1 814 865 1226; fax: +1 814 863 7304. E-mail address: [email protected] (B.A. Dempsey). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.029

ZVI immobilizes arsenic by adsorption of As(V) or As(III) onto iron corrosion products in the shell surrounding the elemental iron core and this is sometimes accompanied by reduction of As(V) to As(III) [11–14]. Detailed mechanisms have been reported for removal of many contaminants by ZVI. Fe(0), dissolved Fe(II), solidbound Fe(II), and H2 have been proposed as elementary reductants [3,9,15]. However there is still controversy about the mechanisms for removal of arsenic especially regarding redox reactions. The rate and extent of As(V) reduction by ZVI may depend on the experimental conditions. In different studies, 25% of initial As(V) was reduced to As(III) by nano-ZVI at neutral pH after 90 days [13,16], As(V) was partially reduced to As(III) by commercial ZVI at slightly basic pH after 60 days [17], and there was no As(V) reduction using iron wires [18]. It was also reported that As(III) was reduced to As(0) with acid-treated iron filings [19]. Magnetite is often observed to be a dominant component in the corroded ZVI shell. In an effort to identify mechanisms by which ZVI immobilized contaminants, Lago and co-workers [20,21] used mechanical grinding to produce a magnetite/ZVI reactant that reduced methylene blue, H2 O2 , and Cr(VI). Other iron oxides (␣Fe2 O3 , FeOOH, or ␥-Fe2 O3 ) mixed with ZVI were much less reactive. It was suggested that the semi-conductor behavior of magnetite was important for effective reduction of contaminants. The reactivity of magnetite may depend on whether the ZVI has been in contact with air. In this context White and Peterson [22] showed that magnetite reduced Cr(VI) at a much faster rate under anoxic conditions than under oxic conditions. It has also been shown that

312

F. Sun et al. / Journal of Hazardous Materials 196 (2011) 311–317

“stoichiometric magnetite” that is produced and maintained under strictly anoxic conditions has a Fe(II)/Fe(III) ratio close to 0.5 and is a stronger reducing agent than “non-stoichiometric magnetite” that has been produced or stored in the presence of O2 resulting in a Fe(II)/Fe(III) ratio k( FeAsO4 2− ), which is consistent with decreased As(V) adsorption with increasing pH. As(V) adsorption on magnetite formed similar inner-sphere complexes as on HFO and goethite [42] and similar adsorption fonts have been reported [24–26,42]. As(V) adsorption on magnetite under anoxic conditions in this study showed similar pH dependence as previously reported for oxic conditions. All of the dissolved and adsorbed arsenic in the magnetite experiments was recovered as As(V), even when the experiment was conducted in the presence of excess dissolved Fe(II). Information about the arsenic oxidation state is shown in Fig. S1. Although we used stoichiometric magnetite (a stronger reducing agent than non-stoichiometric magnetite [22,23]), a rigorous anoxic

Fig. 5. As(V) adsorption and reduction to As(III) using commercial ZVI (2.0 g/0.02 L) in an anoxic environment: (a) pH = 6.5 ± 0.5; (b) pH = 8.5 ± 0.5; (c) pH = 10.

environment, and a stoichiometric excess of dissolved Fe(II), magnetite did not reduce As(V) at pH 5–10. 3.3. Reactions of As(V) with commercial ZVI The results of experiments with As(V) and commercial ZVI at pH 6.5 ± 0.5, 8.5 ± 0.5 and 10 are shown in Fig. 5 in which concentrations of dissolved and adsorbed As(V) and As(III) are plotted against time. There was partial reduction of As(V) to As(III) at the two lower pH ranges and some loss of As(V) + As(III) at pH 8.5 ± 0.5.

F. Sun et al. / Journal of Hazardous Materials 196 (2011) 311–317

Fig. 6. As(V) adsorption and reduction to As(0) using acid-treated ZVI (2 g in 0.02 L) in an anoxic environment at pH 7. No As(III) was produced.

The t1/2 for adsorption and for reduction to As(III) were Mn2+ (Table 4). This implied that the negative charge of birnessite layer was mainly balanced by H+ and K+ . Mn2+ was not detected during Pb2+ adsorption by HB, indicating the presence of small amounts of Mn2+/3+ in the interlayer, which was consistent with the XPS results. The release amount of H+ and K+ was positively related to the Pb2+ adsorbing capacity, however, that of Co2+ and Mn2+ increased with the increase of cobalt content. With an increase in cobalt content, more Mn2+ ions were released into the solution. These low valence Mn ions were formed from the reduction of Mn3+/4+ by Co2+ . About 9.6%, 7.4%, 7.9% and 13.1% of the total cobalt in HC2, HC5, HC10 and HC20 was released, respectively. It indicated that there might be some Co2+ , existing in the interlayer or on the surface of birnessite besides Co3+ when the initial Co2+ concentration was high [21]. The Co2+ can be replaced by Pb2+ during the adsorption. Furthermore, the Co3+ in the interlayer might also be driven by Pb2+ into the solution coupled with redox reactions, but the underlying mechanisms are not clear yet. The aqueous speciation of Pb2+ is shown in Fig. 7 as a function of pH in the range of 3–11 (calculated using ECOSAT4.9 [54]). When pH was below 5.4, Pb2+ and [Pb(NO3 )]+ were the main species. As it increased to 5.4, the formation of Pb(OH)2 (s) began to limit the concentrations of aqueous species. At ∼pH 5.5, Pb(OH)2 (s) was the major form present in the system. When pH was above ∼6.5, Pb(II) occurred predominantly as Pb(OH)2 (s). At pH 5.0, the species of Pb(II) in the adsorption systems existed as Pb2+ (61.21%), [Pb(NO3 )]+ (34.84%), and Pb(NO3 )2 (3.95%). According to the charge conservation law in the process of ion exchange, two moles of H+ and/or K+ or a mole of divalent cation (Me2+ ) would be released after the adsorption of one mole of Pb2+ whereas adsorption of [Pb(NO3 )]+ would drive only one mole of H+ or K+ or 1/2 mol of Me2+ away from the surface of birnessite. This was confirmed by the algebraic relationship of the amounts of

n(Pb(II)) × (0.6121 × 2 + 0.3484) ≈ n(H+ ) + n(K+ ) + (n(Mn2+) + n(Co2+ )) × 2 where n denotes the amount of adsorbed/released ions, and was listed in Table 4. 3.4. Effects of Co2+ -exchange on the oxidation of As(III) Arsenic contamination is an issue of great concern. Depending on its source, arsenic concentrations in natural waters may range up to several hundred milligrams per liter. Due to its acute toxicity to humans, a maximum contaminant level (MCL) should be less than 10 ␮g L−1 for arsenic in drinking water [55]. Manganese oxides are reactive oxidants for the transformation of As(III) to As(V) under natural conditions. As-prepared birnessites were used to study the oxidative transformation of sodium arsenite at the interface of minerals and water. Arsenite oxidation first occurred quickly and the reaction rate then decreased to keep a balance after 1–2 h. HC2 and HC5 had the same shape as that of HB. However, in the case of HC10 and HC20, the oxidation amount of As(III) gradually increased as the reaction progressed (Fig. 8). An apparent oxidation capacity of As(III) to As(V) was calculated to evaluate the oxidation ability of birnessite, due to the fact that adsorption and fixation of As(III) and As(V) occurred simultaneously during the oxidation process [8]. The calculated apparent oxidation capacity of As(III) by HB is 77.3% at equilibrium. After reaction for 7 h, the conversion of As(III) to As(V) by HC2, HC5, HC10 and HC20, were 91.0%, 93.2%, 88.4% and 60.2%, respectively. High oxidation ability towards As(III) was ascribed to the participation of Co(III) in the reaction, since the standard reduction potential for Co3+ /Co2+ (E◦ = 1.92 V) is higher than the MnO2 /Mn2+ (E◦ = 1.224 V) and Mn3+ /Mn2+ (E◦ = 1.5415 V) half reactions [23]. This was already confirmed in our previous work using XPS analysis [24]. The relationship of As(III) concentration with time was analyzed by fitting a first-order rate equation to the 0–0.33 h portions of all the five systems (Fig. 9). The apparent reaction rate constants (kobs ) of HB, HC2, HC5, HC10, and HC20 were calculated to be 0.0226, 0.0175, 0.0161, 0.0123, and 0.0035 min−1 , respectively. The higher initial reaction rate of As(III) oxidation for HB than for Co-containing ones can be ascribed to several reasons. Firstly, in Cocontaining birnessites, oxygen atoms bound to Co3+ will be more strongly held than those bound to Mn3+/4+ due to the high crystal field stabilization energy (CFSE) of the low-spin Co3+ ion [15]. This would increase the activation energy at these sites, resulting in a slower reaction rate [56]. Secondly, the As(III) oxidation by birnessites is a complex process. Investigation on the arsenite oxidation by a poorly crystalline manganese-oxide exhibited that As(III) oxidation and As(V) sorption is greatly affected by Mn AOS in

H. Yin et al. / Journal of Hazardous Materials 196 (2011) 318–326

Fig. 9. Linear regression analysis of normalized As(III) uptake by as-obtained birnessite.

the ␦-MnO2 structure [57]. The Mn(III) reactive sites on Mn-oxide surfaces were expected to be less reactive than Mn(IV) reactive sites in terms of As(III) oxidation [39,58,59]. Moreover, analysis of XPS and the release of ions during Pb2+ adsorption revealed that certain amounts of Mn2+/3+ and Co2+/3+ were adsorbed on the surface of birnessites. These low valence ions also blocked reactive sites on the mineral surface. Release of Mn2+ , Co2+ and K+ into the solution was also monitored. The concentrations of K+ released by HB, HC2, HC5, HC10 and HC20 during the arsenite oxidation process were 1998, 1523, 1247, 1068 and 669 mmol kg−1 , respectively. However, Mn2+ and Co2+ were not detected when HB, HC2, and HC5 were used as oxidants. For HC10 and HC20, 13 and 0, 36 and 8 mmol kg−1 of Mn2+ and Co2+ were determined, respectively. There were two reasons for this low Mn2+ /Co2+ release: (i) MnO2 particle surface contained negatively charged surface functional groups (≡Mn–O− ), thus soluble Mn2+/3+ and Co2+/3+ formed by reductive dissolution of Co-containing birnessites were adsorbed on the surface at pH 7.00 in the present study, and (ii) the precipitate of Co3 (AsO4 )2 (pKsp = 28.17) [60] and the possible Krautite was formed [61]. The oxidation of As(III) by manganese oxide was an important reaction in both the natural cycling of As and in developing remediation technology for lowering the As(III) concentration in drinking water. In the presence of Co-containing birnessite, As(III) in wastewater or underground water would be oxidized to As(V). As(III) has higher mobility and weaker adsorption, and thereby is more poisonous than As(V) [8]. Because As(V) exists as deprotonated oxyanions in broad pH ranges [23] and has high affinity of mineral surfaces, oxidation of As(III) to As(V) not only reduces its toxicity, but also facilitates the removal of As species. Metal compounds (Fe/Al oxides, hydroxides, etc.) are the most widely used adsorbents for As, for their higher removal efficiency at lower cost versus many other adsorbents [4]. The maximum amount of As(V) produced during the oxidation of As(III) by birnessite was greatly enhanced in the presence of goethite. The combined effects of the oxidation (by birnessite)-adsorption (by goethite) led to rapid oxidation and immobilization of As and alleviation of the As toxicity in the environments [62]. Hence, the powerful oxidation of arsenite by manganese oxides, followed by adsorption of arsenate by Fe/Al compounds as adsorbents, is an applicable approach for the treatment of As(III) contaminated water systems, and is worth further investigation. 4. Conclusions Co2+ ion exchange with birnessite was conducted at different Co2+ concentrations. Co2+ ions were totally retained by birnessite at low concentrations. However, when initial Co/Mn molar ratio was increased to 0.2, only 80% of Co2+ could be located in the

325

structure. Induction of Co2+ had no effect on the crystal structure and morphology of birnessite. The crystallinity of Cocontaining birnessites gradually decreased and specific surface areas increased. The valence of Co was exclusively +3 and Mn AOS of Co-containing birnessites gradually decreased. The content of hydroxyl groups in the structure of Co-containing birnessites gradually decreased, which accounted for the reduced Pb2+ adsorption capacities. As(III) oxidation by Co-containing birnessite was enhanced. Conversely, with an increase in cobalt concentration, the initial reaction rate constant was greatly reduced. During the process of Co2+ oxidation, Mn(IV) was more likely the electron sink, and subsequently reduced to Mn(III). The present work provides a new insight into the environmental chemical behavior and interaction mechanism of cobalt and manganese oxides. Further, these modified materials have higher adsorption capacity for Pb2+ than many other adsorbents. Simultaneously, their enhanced oxidation ability for As(III) to As(V) can greatly reduce the toxicity of As(III) in the environment. These as-obtained birnessites have great potential applications in the remediation of heavy metal-contaminated soil and water. Acknowledgements The authors gratefully thank the National Natural Science Foundation of China (Grant numbers: 40830527, 41171375) and the Fundamental Research Funds for the Central Universities (Program number: 2011PY015) for financial support. The authors also acknowledge research assistant Homer Genuino at University of Connecticut for improving English writing in the paper. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.09.027. References [1] Y.N. Vodyanitskii, Mineralogy and geochemistry of manganese: a review of publications, Eurasian Soil Sci. 42 (2009) 1170–1178. [2] F. Liu, C. Colombo, P. Adamo, J.Z. He, A. Violante, Trace elements in manganese–iron nodules from a Chinese Alfisol, Soil Sci. Soc. Am. J. 66 (2002) 661–670. [3] R.P. Han, W.H. Zou, Z.P. Zhang, J. Shi, J.J. Yang, Removal of copper(II) and lead(II) from aqueous solution by manganese oxide coated sand. I. Characterization and kinetic study, J. Hazard. Mater. B137 (2006) 384–395. [4] D. Mohan, C.U. Pittman Jr., Arsenic removal from water/wastewater using adsorbents—A critical review, J. Hazard. Mater. 142 (2007) 1–53. [5] X.H. Feng, L.M. Zhai, W.F. Tan, F. Liu, J.Z. He, Adsorption and redox reactions of heavy metals on synthesized Mn oxide minerals, Environ. Pollut. 147 (2007) 366–373. [6] R.N. Dai, J. Liu, C.Y. Yu, R. Sun, Y.Q. Lan, J.D. Mao, A comparative study of oxidation of Cr(III) in aqueous ions, complex ions and insoluble compounds by manganese-bearing mineral (birnessite), Chemosphere 76 (2009) 536–541. [7] Y.T Meng, Y.M. Zheng, L.M. Zhang, J.Z. He, Biogenic Mn oxides for effective adsorption of Cd from aquatic environment, Environ. Pollut. 157 (2009) 2577–2583. [8] X.J. Li, C.S. Liu, F.B. Li, Y.T. Li, L.J. Zhang, C.P. Liu, Y.Z. Zhou, The oxidative transformation of sodium arsenite at the interface of ␦-MnO2 and water, J. Hazard. Mater. 173 (2010) 675–681. [9] S.B. Lee, J.S. An, Y.J. Kim, K. Nam, Binding strength-associated toxicity reduction by birnessite and hydroxyapatite in Pb and Cd contaminated sediments, J. Hazard. Mater. 186 (2011) 2117–2122. [10] R.G. Burns, V.M. Burns, The mineralogy and crystal chemistry of deep-sea manganese nodules—a polymetallic resource of the twenty-first century, Philos. Trans. R. Soc. London Ser. A286 (1977) 283–301. [11] R.M. Taylor, R.M. Mckenzie, The association of trace elements with manganese minerals in Australian soils, Aust. J. Soil Res. 4 (1966) 29–39. [12] R.M. Mckenzie, The reaction of cobalt with manganese dioxide minerals, Aust. J. Soil Res. 8 (1970) 97–106. [13] R.M. Mckenzie, The adsorption of lead and other heavy metals on oxides of manganese and iron, Aust. J. Soil Res. 18 (1980) 61–73. [14] R.G. Burns, The uptake of cobalt into ferromanganese nodules, soils, and synthetic manganese(IV) oxides, Geochim, Cosmochim. Acta 40 (1976) 95–102.

326

H. Yin et al. / Journal of Hazardous Materials 196 (2011) 318–326

[15] R.G. Burns, Mineralogical Applications of Crystal Field Theory, 2nd ed., Cambridge University Press, Cambridge, 1993. [16] C.P. Lienemann, M. Taillefert, D. Perret, J.F. Gaillard, Association of cobalt and manganese in aquatic systems: chemical and microscopic evidence, Geochim. Cosmochim. Acta 61 (1997) 1437–1446. [17] Y. Takahashi, A. Manceau, N. Geoffroy, M.A. Marcus, A. Usui, Chemical and structural control of the partitioning of Co, Ce, and Pb in marine ferromanganese oxides, Geochim. Cosmochim. Acta 71 (2007) 984–1008. [18] J.W. Murray, J.G. Dillard, The oxidation of cobalt(II) adsorbed on manganese dioxide, Geochim. Cosmochim. Acta 43 (1979) 781–787. [19] D.L. Crowther, J.G. Dillard, J.W. Murray, The mechanism of Co(II) oxidation on synthetic birnessite, Geochim. Cosmochim. Acta 47 (1983) 1399–1403. [20] S. Fendorf, P.M. Jardine, R.R. Patterson, D.L. Taylor, S.C. Brooks, Pyrolusite surface transformations measured in real-time during the reactive transport of Co(II)EDTA2− , Geochim. Cosmochim. Acta 63 (1999) 3049–3057. [21] A. Manceau, V.A. Drits, E. Silvester, C.L. Bartoli, B. Lanson, Structural mechanism of Co2+ oxidation by the phyllomanganate buserite, Am. Mineral. 82 (1997) 1150–1175. [22] S.H. Lee, T.W. Kim, D.H. Park, J.H. Choy, S.J. Hwang, Single-step synthesis, characterization, and application of nanostructured Kx Mn1−y Coy O2−ı with controllable chemical compositions and crystal structures, Chem. Mater. 19 (2007) 5010–5017. [23] D.R. Lide, W.M. Mickey Haynes, Handbook of Chemistry and Physics, 19th ed., CRC Press/Taylor and Francis, Florida, 2010. [24] H. Yin, X.H. Feng, G.H. Qiu, W.F. Tan, F. Liu, Characterization of Co-doped birnessites and application for removal of lead and arsenite, J. Hazard. Mater. 188 (2011) 341–349. [25] R.M. McKenzie, The synthesis of birnessite, cryptomelane, and some other oxides and hydroxides of manganese, Miner. Mag. 38 (1971) 493–503. [26] N. Kijima, H. Yasuda, T. Sato, Y. Yoshimura, Preparation and characterization of open tunnel oxide ␣-MnO2 precipitated by ozone oxidation, J. Solid State Chem. 159 (2001) 94–102. [27] D.J. Miller, M.C. Biesinger, N.S. McIntyre, Interactions of CO2 and CO at fractional atmosphere pressures with iron and iron oxide surfaces: one possible mechanism for surface contamination? Surf. Interface Anal. 33 (2002) 299–305. [28] M.C. Biesinger, B.P. Payne, A.P. Grosvenor, L.W.M. Lau, A.R. Gerson, R.S.C. Smart, Resolving surface chemical states in XPS analysis of first row transition metals, oxides and hydroxides: Cr, Mn, Fe, Co and Ni, Appl. Surf. Sci. 257 (2011) 2717–2730. [29] D.W. Oscarson, P.M. Huang, W. Liaw, The oxidation of arsenite by aquatic sediments, J. Environ. Qual. 9 (1980) 700–703. [30] M. Villalobos, B. Toner, J. Bargar, G. Sposito, Characterization of the manganese oxide produced by Pseudomonas putida strain MnB1, Geochim. Cosmochim. Acta 67 (2003) 2649–2662. [31] M. Villalobos, B. Lanson, A. Manceau, B. Toner, G. Sposito, Structural model for the biogenic Mn oxide produced by Pseudomonas putida, Am. Mineral. 91 (2006) 489–502. [32] J.B. Kim, J.B. Dixon, C.C. Chusuei, Y.J. Deng, Oxidation of chromium(III) to (VI) by manganese oxides, Soil Sci. Soc. Am. J. 66 (2002) 306–315. [33] R.M. Potter, G.R. Rossman, The tetravalent manganese oxides: identification, hydration, and structural relationships by infrared spectroscopy, Am. Mineral. 64 (1979) 1199–1218. [34] C.M. Julien, M. Massot, C. Poinsignon, Lattice vibrations of manganese oxides: Part I. Periodic structures, Spectrochim. Acta A: Mol. Biomol. Spectrosc. 60 (2004) 689–700. [35] A. Chahi, S. Petit, A. Decarreau, Infrared evidence of dioctahedral–trioctahedral site occupancy in palygorskite, Clays Clay Miner. 50 (2002) 306–313. [36] R.D. Shannon, Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides, Acta Crystallogr. A 32 (1976) 751–767. [37] J. Portier, G. Campet, J. Etourneau, B. Tanguy, A simple model for the estimation of electronegativities of cations in different electronic states and coordinations, J. Alloys. Compd. 209 (1994) 285–289. [38] S. Martinez-Alonso, J.R. Rustad, A.F.H. Goetz, Ab initio quantum mechanical modelling of infrared vibrational frequencies of the OH group in dioctahedral phyllosilicates. Part II: main physical factors governing the OH vibrations, Am. Mineral. 87 (2002) 1224–1234. [39] H.W. Nesbitt, G.W. Canning, G.M. Bancroft, XPS study of reductive dissolution of ˚ by H3 AsO3 , with constraints on reaction mechanism, Geochim. 7 A-birnessite Cosmochim. Acta 62 (1998) 2097–2110.

[40] V.A. Drits, E.J. Silvester, A.I. Gorshkov, A. Manceau, Structure of synthetic monoclinic Na-rich birnessite and hexagonal birnessite: I. Results from Xray diffraction and selected-area electron diffraction, Am. Mineral. 82 (1997) 946–961. [41] B. Lanson, V.A. Drits, E.J. Silvester, A. Manceau, Structure of H-exchanged hexagonal birnessite and its mechanism of formation from Na-rich monoclinic buserite at low pH, Am. Mineral. 85 (2000) 826–838. [42] S.M. Webb, G.J. Dick, J.R. Bargar, B.M. Tebo, Evidence for the presence of Mn(III) intermediates in the bacterial oxidation of Mn(II), Proc. Natl. Acad. Sci. U.S.A. 102 (2005) 5558–5563. [43] J.J. Morgan, Manganese in natural waters and Earth’s crust: its availability to organisms, in: A. Sigel, H. Sigel (Eds.), Metal Ions in Biological Systems, Manganese and its Role in Biological Processes, Marcel Dekker, New York, 2000. [44] S.W. Knipe, J.R. Mycroft, A.R. Pratt, H.W. Nesbitt, G.M. Bancroft, X-ray photoelectron spectroscopic study of water adsorption on iron sulphide minerals, Geochim. Cosmochim. Acta 59 (1995) 1079–1090. [45] A.S. Özcan, Ö. Gök, A. Özcan, Adsorption of lead(II) ions onto 8hydroxy quinoline-immobilized bentonite, J. Hazard. Mater. 161 (2009) 499– 509. [46] S.M. Maliyekkal, K.P. Lisha, T. Pradeep, A novel cellulose-manganese oxide hybrid material by in situ soft chemical synthesis and its application for the removal of Pb(II) from water, J. Hazard. Mater. 181 (2010) 986– 995. [47] C.H. Giles, T.H. MacEwan, S.N. Nakhwa, D. Smith, Studies in adsorption. Part XI. A system of classification of solution adsorption isotherms and its use in diagnosis of adsorption mechanisms and in measurement of specific surface area of solids, J. Chem. Soc. 3 (1960) 3973–3993. [48] D.G. Kinniburgh, General purpose adsorption isotherms, Environ. Sci. Technol. 20 (1986) 895–904. [49] R.P. Han, W.H. Zou, H.K. Li, Y.H. Li, J. Shi, Copper(II) and lead(II) removal from aqueous solution in fixed-bed columns by manganese oxide coated zeolite, J. Hazard. Mater. B137 (2006) 934–942. [50] E. Eren, B. Afsin, Y. Onal, Removal of lead ions by acid activated and manganese oxide-coated bentonite, J. Hazard. Mater. 161 (2009) 677–685. [51] K.D. Kwon, K. Refson, G. Sposito, Surface complexation of Pb(II) by hexagonal birnessite nanoparticles, Geochim. Cosmochim. Acta 74 (2010) 6731– 6740. [52] W. Zhao, Q.Q. Wang, F. Liu, G.H. Qiu, W.F. Tan, X.H. Feng, Pb2+ adsorption on birnessite affected by Zn2+ and Mn2+ pretreatments, J. Soil Sediment. 10 (2010) 870–878. [53] M. Villalobos, J. Bargar, G. Sposito, Mechanisms of Pb(II) sorption on a biogenic manganese oxide, Environ. Sci. Technol. 39 (2005) 569–576. [54] M.G. Keizer, W.H. Van Riemsdijk, ECOSAT: Equilibrium Calculation of Speciation and Transport, User Manual Version 4.7, Wageningen Agricultural University, The Netherlands, 1999. [55] A. Amirbahman, D.B. Kent, G.P. Curtis, J.A. Davis, Kinetics of sorption and abiotic oxidation of arsenic(III) by aquifer materials, Geochim. Cosmochim. Acta 70 (2006) 533–547. [56] R.M. Mckenzie, The influence of cobalt on the reactivity of manganese dioxide, Aust. J. Soil Res. 9 (1971) 55–58. [57] B.J. Lafferty, M. Ginder-Vogel, M.Q. Zhu, K.J.T. Livi, D.L. Sparks, Arsenite oxidation by a poorly crystalline manganese-oxide. 2. Results from X-ray absorption spectroscopy and X-ray diffraction, Environ. Sci. Technol. 44 (2010) 8467– 8472. [58] C. Tournassat, L. Charlet, D. Bosbach, A. Manceau, Arsenic(III) oxidation by birnessite and precipitation of manganese(II) arsenate, Environ. Sci. Technol. 36 (2002) 493–500. [59] M.Q. Zhu, K.W. Paul, J.D. Kubicki, D.L. Sparks, Quantum chemical study of arsenic(III,V) adsorption on Mn-Oxides: implications for arsenic(III) oxidation, Environ. Sci. Technol. 43 (2009) 6655–6661. [60] J.G. Speight, Lange’s Handbook of Chemistry, 16th ed., McGraw-Hill Press, New York, 2005. [61] B.A. Manning, S.E. Fendorf, B. Bostick, D.L. Suarez, Arsenic(III) oxidation and arsenic(V) adsorption reactions on synthetic birnessite, Environ. Sci. Technol. 36 (2002) 976–981. [62] X.H. Feng, W.F. Tan, F. Liu, H.D. Ruan, J.Z. He, Oxidation of AsIII by several manganese oxide minerals in absence and presence of goethite, Acta Geol. Sin. (Engl.) 80 (2006) 249–256.

Journal of Hazardous Materials 196 (2011) 327–334

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Adsorption behavior of some aromatic compounds on hydrophobic magnetite for magnetic separation Takahiro Sasaki ∗ , Shunitz Tanaka Division of Environmental Material Science, Graduate School of Environmental Science, Hokkaido University, Sapporo, Hokkaido, 060-0810, Japan

a r t i c l e

i n f o

Article history: Received 4 March 2011 Received in revised form 8 September 2011 Accepted 9 September 2011 Available online 16 September 2011 Keywords: Adsorption behavior Aromatic compounds Hydrophobic magnetite Hydrophobic interaction ␲-electron interaction

a b s t r a c t In this study, a hydrophobic magnetite coated with an alkyl chain or a phenyl group on the surface was prepared and used as an adsorbent to investigate the adsorption behavior of aromatic compounds having various values of log Pow (phenol 1.46, benzonitrile 1.56, nitrobenzene 1.86, benzene 2.13, toluene 2.73, chlorobenzene 2.84 and o-dichlorobenzene 3.38) onto hydrophobic magnetite. The hydrophobic magnetites were modified with stearic acid and phenyltrimethoxysilane, and the modification amounts were 9.84 × 10−3 and 4.17 × 10−2 mmol/g, respectively. The aromatic compounds used in this study were divided into 3 groups depending on the log Pow : 1 < log Pow < 2, 2 < log Pow < 3 and 3 < log Pow . The adsorption amounts of above each group on the magnetite at an initial concentration of 100 ppm were 3.62 × 10−3 (nitrobenzene), 1.92 × 10−2 (phenol), 1.13 × 10−1 (chlorobenzene), 2.42 × 10−1 (benzene), and 3.10 × 10−1 mmol/g (dichlorobenzene), respectively. This indicates that the adsorption behaviors depend on the strength of hydrophobicity of aromatic compounds. The adsorption mechanism for 2 < log Pow < 3 and 3 < log Pow is hydrophobic interaction and that for 1 < log Pow < 2 is ␲-electron interaction. The quantitative relationship between the amount of adsorbed compounds and modified functional groups and the fitting for adsorption isotherm models suggested that this adsorption might form a multi-layer adsorption in the most cases. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Global attention to water pollution from harmful chemicals has increased in recent decades. Many kinds of pollutants have been discovered in aquatic environments such as rivers, ponds and seas. These contaminants originated from industrial and domestic wastewater, sometimes from accidental spills. Various mono and polycyclic aromatic compounds have been found in aquatic environments. These aromatic compounds must be removed before the water is discharged or consumed. In order to treat harmful organic compounds in effluent, two types of technology are currently available. The first is a decomposition technology where hazardous organic compounds are converted to more environmentally friendly compounds. Technologies such as chemical oxidation [1], electrolysis [2], photo oxidation [3], and ozonation [4] are included in this category. The second type is a separating technology, where harmful organic pollutants are separated from the effluent by various methods.

∗ Corresponding author at: N10W5 kita-ku, Sapporo, Hokkaido 060-0810, Japan. Tel.: +81 11 706 2219, fax: +81 11 706 2219. E-mail addresses: [email protected] (T. Sasaki), [email protected] (S. Tanaka). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.033

Technologies such as membrane separation [5], adsorption [6], and solvent extraction [7] belong to the second category. Among these types of technologies, adsorption is one of the simplest and most effective processes. Adsorption is fast, economic and widely applicable techniques. Using adsorption is applicable for various pollutants such as organic compounds and heavy metals by selecting the type of adsorbent and adsorption conditions. In addition, recently treatment methods for wastewater using lowcost adsorbents such as by-products or waste materials have been reported [8,9]. Gupta and co-workers have reviewed the details of treatment methods for various pollutants in water using low-cost adsorbents [10,11]. Activated carbon is the most widely used adsorbent in various cases because of a large capacity and a wide variety of adsorbates. However, there are some limitations, particularly in regeneration [12]. There is poor mechanical rigidity and low selectivity when activated carbon is applied to real environmental pollution. Furthermore, it is difficult to collect activated carbon powder that has been widely diffused into the environment. If it is not collected, the adsorbent that is used to adsorb harmful pollution could become a secondary source of pollution. Magnetic separation has been applied recently in various fields such as analytical biochemistry [13], medical science [14] and biotechnology [15]. From an environmental point of view, magnetic separation offers advantages due to the easy recovery of the

328

T. Sasaki, S. Tanaka / Journal of Hazardous Materials 196 (2011) 327–334

1 < log Pow < 2 OH

1.46 a)

1.56 a)

Phenol

Benzonitrile

NO2

NO2

CN

1.62 d)

1.86 a)

Nitrocyclohexane

Nitrobenzene

2 < log Pow < 3 Cl

CH3

2.13 a)

2.73 b)

Benzene

Toluene

2.84 a) Chlorobenzene

3 < log Pow Cl Cl

3.38 c)

o-Dichlorobenzene Fig. 1. Structures and log Pow s of aromatic compounds and nitrocyclohexane. a) the values of log Pow was referred to Ref. [25] b) the values of log Pow was referred to Ref. [26] c) the values of log Pow was referred to Ref. [27] d) the values of log Pow was referred to Ref. [28].

adsorbent without filtration or centrifugation. Several studies have reported magnetic separation using modified magnetite (Fe3 O4 ) as an environmentally friendly approach to remove heavy metal ions [16,17] and organic pollutants [18,19]. The removal of harmful organic compounds with adsorbents is in most cases based on the hydrophobic interaction between an adsorbent and its target compounds. This hydrophobic interaction has been applied to not only the removal of harmful substances by adsorbents but also to the preconcentration of analytes by using solvents [7] and solid extraction [20]. A hydrophobic adsorbent can be prepared by using a hydrophobizing agent on the surface of materials such as magnetite and silica beads [18,21]. One of the most popular techniques for hydrophobizing involves the use of silane coupling agents. Silane coupling agents are known as surface modifiers that can add an organic property to the surface of an inorganic material [22,23]. Therefore, a silane coupling agent is important when forming a hybrid inorganic-organic material. Another technique is the use of an ionic surfactant. An ionic group of the surfactant will turn toward the surface of mineral oxides such as alumina, silica, titanium dioxide and ferric oxide, and then the alkyl chain, the hydrophobic group of the surfactant, will orient to the outside. As a result, the surface of the material becomes hydrophobic [24]. The strength of the hydrophobic interaction depends on the degree of hydrophobicity of both the adsorbent and the adsorbate. The hydrophobicity of an organic compound can be varied by the structure and functional group of the compound. The octanol–water partition coefficient (Pow ) is a well-known indicator of the hydrophobicity of an organic compound. A higher hydrophobicity compound will have a larger Pow . Therefore, the hydrophobicity, or Pow , of an organic compound is very important in the prediction of the adsorptive behavior of some organic compounds in water. The aim of the present study was to clarify the adsorption behaviors of organic compounds on hydrophobic magnetite and to evaluate the possibility of magnetic separation for the removal of organic compounds dissolved in water. Organic compounds

with low Pow s, or relatively weak hydrophobicities, were used in this study. These organic compounds were selected on the basis of the value of log Pow . The selected compounds were divided into 3 groups according to log Pow : 1 < log Pow < 2, 2 < log Pow < 3, and 3 < log Pow . The Pow s of phenol, benzonitrile and nitrobenzene fell into the 1 < log Pow < 2 group. Those of benzene, toluene and chlorobenzene were in the 2 < log Pow < 3 group. o-Dichlorobenzene was in the 3 < log Pow group [25–29]. The individual values of log Pow s of aromatic compounds used in the adsorption experiments are summarized in Fig. 1. The hydrophobic magnetite was prepared by hydrophobizing the surface of a magnetite particle. Stearic acid and phenyltrimethoxysilane were used to hydrophobize the surface of magnetite. By using two different types of hydrophobic magnetite, the difference in adsorption behaviors of the various aromatic compounds was investigated. 2. Experiments 2.1. Materials Magnetite where average size was 0.3 ␮m (the data was provided from Kishida Chemical.), was purchased from Kishida Chemical Co., Ltd. (Osaka, Japan) and used as the adsorbent carrier. The modifying reagents, stearic acid and phenyltrimethoxysilane were purchased from MP biomedicals Japan K. K. (Tokyo, Japan) and Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan), respectively. The organic compounds used as adsorbates, phenol, nitrobenzene and benzene were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Toluene, chlorobenzene and o-dichlorobenzene were purchased from Nacalai Tesque, Inc. (Kyoto, Japan). Benzonitrile was purchased from Kanto Chemical Co., Inc. (Tokyo, Japan). Nitrocyclohexane was purchased from Tokyo Chemical Industry Co., Ltd. In analysis for fatty acid, a boron trifluoride solution was used as an esterification agent, anhydrous sodium sulfate was used as a dehydration agent, and methylene chloride was used as a solvent, and all were purchased from Wako Pure Chemical Industries, Ltd. The reagents used in Si analysis, (hydrochloric acid, nitric acid,

T. Sasaki, S. Tanaka / Journal of Hazardous Materials 196 (2011) 327–334

sulfuric acid, hydrofluoric acid, boric acid, hexaammonium heptamolybdate tetrahydrate and sodium chloride) were purchased from Wako Pure Chemical Industries, Ltd. All regents were of analytical grade and used without further purification. 2.2. Preparation of hydrophobic magnetite Two kinds of hydrophobic adsorbents were prepared: stearic acid modified magnetite (SA-mag) and phenyl group modified magnetite (Ph-mag) [29,30]. SA-mag was prepared in the following way. 1.0 g of magnetite was added in 50 ml of methanol for distribution. Then 10 mg of stearic acid was added to the suspension with stirring to dry out the methanol. The residue was washed twice with methanol and dried in an oven at 50 ◦ C. Ph-mag was prepared in the following way. 1.0 g of magnetite was added in 40 ml of ethanol and distributed with stirring. Then, 0.1 ml of phenyltrimethoxysilane was added to the suspension. After phenyltrimethoxysilane was sufficiently dissolved in the solvent, 0.058 ml of H2 O and 0.025 ml of 1 M HCl were added to the suspension. The suspension was then heated at 50 ◦ C with continuous stirring until the solvent was dry. The obtained magnetite was heated in a muffle furnace at 120 ◦ C for 1 h. After heating, the modified magnetite was washed twice with ethanol and dried in an oven at 50 ◦ C. 2.3. Characterization of hydrophobic magnetite 2.3.1. Surface areas of hydrophobic magnetite The surface areas of SA-mag and Ph-mag were determined using a N2 BET analysis (AutoSorb6 YUASA, Japan). The samples were pretreated by degassing at 80 ◦ C for 6 h. 2.3.2. Amount of modified stearic acid on magnetite For desorption of modified stearic acid from SA-mag, 30 mg of SA-mag was washed with 20 ml of ethanol for 1 h with sonication. The washing procedure was repeated 3 times. And then, the collected ethanol solution that included stearic acid was evaporated under vacuum. The obtained sample was dissolved in 3 ml of methanol and 1 ml of methanolic solution containing lauric acid (1 mg/l) as the surrogate standard. The mixture was transferred to a test tube and 2 ml of 14% BF3 methanolic solution was added. The sample was placed in a water bath at 80 ◦ C for 3 min and then 1 ml of water was added to stop the reaction. The fatty acid esters were extracted twice by 1 ml of methylene chloride each time. The collected organic phase was dehydrated with anhydrous sodium sulfate and placed in a 10 ml volumetric flask after filtration. The adjusted 10 ml of sample solution was measured using a GC-17A (Shimadzu, Japan) equipped with a GCMS-QP5050A (Shimadzu, Japan). A DB-5 ms (30 m × 0.25 mm × 0.25 ␮m) column (Agilent, USA) was used. 1 ␮l of sample solution was injected. The injections were performed in splitless mode. The carrier gass was helium (Air Water, Japan) at a constant flow of 1.5 ml/min. The injection port was heated to 200 ◦ C. The oven temperature was set at 75 ◦ C for 2 min, then increased 30 ◦ C/min to 270 ◦ C, and the final temperature was held for 1 min. The temperature at the detector was 280 ◦ C. All mass spectra were acquired in the electron impact (EI) mode as the ionization source with a quadrupole mass filter. The analysis was carried out in SIM mode, and the selected ions of the compound were m/z 55, 74 and 87. The concentrations of the fatty acids were calculated using the internal standard method [31]. 2.3.3. Amount of modified phenyl group on magnetite The amount of modified phenyltrimethoxysilane on the magnetite surface was determined from the measurement of silica dioxide as a decomposition component of Ph-mag in the following way [32]. First, 1 g of Ph-mag was decomposed using 40 ml of the mixed acid contained 6 M hydrochloric acid and 6 M nitric acid

329

covering a watch glass with heating at 80 ◦ C until the black precipitate disappeared. The obtained precipitated silicate was filtrated by a membrane filter and washed with water several times. Second, the filter paper carrying the precipitation was moved into a PTFE beaker and then 5 ml of 4% (w/v) NaCl solution and 3 ml of 60% hydrofluoric acid were added to the beaker. The mixture was heated until dry here in a water bath. After dissolving the obtained residue with 15 ml of water, 10 ml of saturated boric acid solution was added to the mixture and then was heated almost to the boiling point ( toluene > chlorobenzene. This result was not consistent with the order of log Pow . That is, the order of the adsorption ability did not always depend on the log Pow of the compounds in this group. The adsorption mechanism in this group was also that of multi-layer adsorption based on the hydrophobic interaction. Therefore, the adsorption amount depended on the strength of the interaction and the size of the molecule forming the multi-layer. Benzene was adsorbed most because there was no substituent group and it had the smallest molecular size [33]. On the other hand, toluene and chlorobenzene are more bulky than benzene because of a substituent group. The methyl group, a substituent group of toluene, supplies electrons to the benzene ring because

A 0.25 qe (mmol/g)

qe (mmol/g)

A 0.40

331

Benzene Toluene Chlorobenzene

0.20 0.15 0.10 0.05 0

0

20

40

60

40

60

Ce (ppm)

0.40

B 0.25

Ph-mag

0.30

qe (mmol/g)

qe (mmol/g)

SA-mag

0.20 0.10 0

0.20 0.15 0.10 0.05

0

5

10

15

Ce (ppm) Fig. 6. The adsorption amounts of o-dichlorobenzene onto SA-mag and Ph-mag in various initial concentrations (10–100 ppm).

0

0

20

Ce (ppm)

Fig. 7. The adsorption amounts of aromatic compounds in 2 < log Pow < 3 onto (A) SA-mag and (B) Ph-mag in various initial concentrations (10–100 ppm).

332

T. Sasaki, S. Tanaka / Journal of Hazardous Materials 196 (2011) 327–334

the functional group is the electron-donating group. By contrast, chlorine, a substituent group of chlorobenzene, is an electronwithdrawing group and withdraws electrons from the benzene ring. The strength of the ␲-electron interaction depends on the richness of the ␲-electron in the benzene ring, that is, the interaction works well in a ␲-electron-rich state. The ␲-electron interaction between aromatic compounds seems to be important for the formation of a multi-layer in these cases. Thus, toluene adsorbed more than chlorobenzene. 3.3.3. 1 < log Pow < 2 Phenol, benzonitrile, nitrobenzene and nitrocyclohexane had the lowest hydrophobicity (1 < log Pow < 2) in this study. Fig. 8 shows the results of these adsorption experiments. The adsorption amounts of these compounds in this group were small by comparison with other groups. This happened because these compounds have strong polar groups such as hydroxyl, cyano and nitro groups and can be hydrated with water molecules around their polar groups. Therefore, the hydrophobic interaction between these compounds and the modified magnetite becomes weak. In the same way, the interaction between their compounds also becomes weak. According to Fig. 8, phenol, benzonitrile and nitrobenzene were more selectively adsorbed onto Ph-mag than onto SA-mag. These results indicate that the functional group on the magnetite surface significantly affects adsorption behavior and ␲-electron interactions between the compounds and the phenyl group on Ph-mag works stronger rather than hydrophobic interaction in these cases. When nitrocyclohexane, having no aromatic ring, was scarcely adsorbed onto Ph-mag, this suggested that the adsorption mechanism of this group was based on ␲-electron interaction. The order of the adsorption amount of these aromatic compounds was as follows: phenol > benzonitrile > nitrobenzene. These results did not agree with the order of log Pow (phenol 1.46, benzonitrile 1.56, nitrobenzene 1.86). Phenol has a hydroxyl group, an electrondonating group, and the ␲-electron interaction between phenol

qe (mmol/g)

A

0.03

Phenol Benzonitrile Nitrobenzene Nitrocyclohexane

0.02

and Ph-mag was strongest among them. However, the substituent groups of benzonitrile and nitrobenzene are electron-withdrawing groups that make their aromatic rings relatively electron-poor. The electron-withdrawing property of benzonitrile might be weaker than that of nitrobenzene because the dipole moment of benzonitrile, 4.05, is smaller than that of nitrobenzene, 4.22–4.91 [34,35]. The aromatic ring of benzonitrile is more electron-rich than that of nitrobenzene. The ␲-electron interaction between the compound and the phenyl group on the magnetite surface was dominant in this adsorption mechanism and so the adsorption amount of benzonitrile on Ph-mag was more than that of nitrobenzene. According to Fig. 8A, phenol was adsorbed on SA-mag, and the amount of adsorbed phenol was close to the modified amount of stearic acid on magnetite. However, this does not mean that adsorption is a one-to-one relationship between phenol and stearic acid like a mono-layer adsorption because of the shape of the adsorption isotherm. A feature of the adsorption isotherm is that the adsorption amount immediately increases with a high concentration of adsorbate during multi-layer adsorption. Moreover, the dipole moment of phenol, 1.53 [36], was the lowest and hydration for phenol was the weakest. Thus, a small amount of phenol could be adsorbed onto SA-mag by hydrophobic interaction. However, the adsorption behaviors of this group were not clear in the initial concentration range (10–100 ppm). Fig. 8 shows that the adsorption amounts of the compounds are smaller compared with the amounts of the modified functional groups, particularly in the case of a phenyl group with an initial concentration of 100 ppm. Therefore, the adsorption experiment for this group should be carried out with a higher initial concentration. 3.4. Adsorption isotherm models This section describes an investigation into the adsorption isotherms of these aromatic compounds onto two adsorbents (in Fig. 5) by application of the adsorption isotherm models. The multilayer adsorption model in the gas phase is known as the BET model. Many researchers have reported the use of the-multi layer adsorption model in the liquid phase based on the BET model, but this model has not yet been established [37]. Thus, in this study, adsorption behaviors were studied by fitting these adsorption data into the Langmuir and Freundlich models. The Langmuir and Freundlich models are as follows: 1 1 1 = + qe KL qm Ce qm

0.01

1/n

qe = Kf Ce

0.00

B

0.03

qe (mmol/g)

0

0.02

20

40

60 Ce (ppm)

80

100

120

20

40

60 80 Ce (ppm)

100

120

0.01

0.00 0

Fig. 8. The adsorption amounts of aromatic compounds and nitrocyclohexane in 1 < log Pow < 2 onto (A) SA-mag and (B) Ph-mag in various initial concentrations (10–100 ppm).

(1) (2)

where qe is the equilibrium adsorption capacity, Ce represents the solute concentration in equilibrium, qm (mg/g) is the maximal sorption capacity, KL (L/g) is a binding constant, and Kf and n are the Freundlich constants to be determined. Table 1 shows the fitting results for the Langmuir and Freundlich models. When applying the Langmuir model, most correlation coefficients are not improved. The Langmuir model is well known as a mono-layer adsorption model and most of the adsorption behaviors seem not to be of the mono-layer adsorption type, because of the low correlation coefficients. On the other hand, when applying the Freundlich model, most correlation coefficients are better than that of the Langmuir model. However, most of the correlation coefficients for the Freundlich model were not so good. The Freundlich model does not provide insight into adsorption behavior, because of an empirical formula. However, the Freundlich model is well known to be a better fit for adsorption into a porous material such as activated carbon. Thus, these adsorption behaviors seemed not to be porous adsorption. This result also agrees with Fig. 4. These fitting results for the models were not definitive as to whether the

T. Sasaki, S. Tanaka / Journal of Hazardous Materials 196 (2011) 327–334

333

Table 1 Isotherm parameters of aromatic compounds onto SA-mag and Ph-mag. Adsorbent

Adsorbate

Freundlich model

Langmuir model

Kf

n

R2

qm (mmol/L)

KL (L/mg)

R2

SA-mag

Phenol Benzonitrile Nitrobenzene Benzene Toluene Chlorobenzene o-Dichlorobenzene

3.46e−5 2.70e−4 – 3.83 1.76 2.80 16.4

0.453 0.647 – 2.52 1.80 1.45 2.97

0.807 0.988 – 0.984 0.962 0.930 0.883

1.17e−3 9.52e−4 – 0.234 0.203 0.0690 0.248

0.0101 9.86e−3 – 0.135 0.0543 0.933 1.90

0.213 0.991 – 0.960 0.911 0.423 0.979

Ph-mag

Phenol Benzonitrile Nitrobenzene Benzene Toluene Chlorobenzene o-Dichlorobenzene

0.528 0.0383 0.115 3.78 1.53 1.56 16.1

3.63 1.26 1.83 3.07 1.70 0.885 2.82

0.932 0.978 0.995 0.951 0.935 0.986 0.853

0.0189 0.0185 0.0134 0.135 0.199 0.0907 0.292

0.135 0.0151 0.0310 0.508 0.0534 0.224 1.19

0.819 0.937 0.908 0.812 0.927 0.724 0.944

0.05 Ph-mag

qe (mmol/g)

0.04

SA-mag

0.03 0.02 0.01 0

0

2

4

6

8

10

Conc. of Salt (w/v%) Fig. 9. Influence of salt in adsorption of nitrobenzene onto SA-mag and Ph-mag.

adsorption behaviors in 1 < log Pow < 2 were mono or multi-layer adsorption. 3.5. Nitrobenzene adsorption in various salt concentrations Shown in Fig. 9 are nitrobenzene adsorption isotherms for various concentrations of salt on SA-mag and Ph-mag. This experiment was performed to validate whether hydration with water inhibits the adsorption of aromatic compounds in 1 < log Pow < 2 on hydrophobic magnetite. The dehydration effects of the addition of salt were investigated. The addition of salt deprives hydrated adsorbate of water molecules. If the adsorption amount of nitrobenzene is increased when the adsorption experiment is carried out under dehydration conditions, then hydration would be an inhibiting factor in its adsorption onto hydrophobic magnetite. As shown in Fig. 9, the adsorption amounts of nitrobenzene on SA-mag and Ph-mag increased with increasing salt concentration. Furthermore, nitrobenzene was hardly adsorbed onto SA-mag without the addition of salt, but was adsorbed onto SA-mag in the presence of salt. The nitrobenzene adsorption onto SA-mag was by hydrophobic interaction, which seems to indicate that the hydrophobicity of nitrobenzene was enhanced by dehydration. This result indicates that hydration is one of the inhibitors of the adsorption of nitrobenzene onto hydrophobic magnetite. 4. Conclusion Aromatic compounds with various log Pow s were investigated for adsorption onto hydrophobic magnetite coated alkyl chains and phenyl groups. The adsorption behaviors of each compound were divided into 3 groups depending on the log Pow : 1 < log Pow < 2, 2 < log Pow < 3 and 3 < log Pow . The adsorption amounts generally

increased as the log Pow of each group increased. However, the adsorption amounts for the compounds in each group did not depend on the values of log Pow . In the 2 < log Pow < 3 and 3 < log Pow groups, the adsorption mechanism was mainly a hydrophobic interaction between aromatic compounds and the surface of hydrophobic magnetite. The adsorption behaviors did not depend on the difference in modified functional groups on the magnetite surface. The adsorption behavior of hydrophobic magnetite in these groups seemed to form a multi-layer on the hydrophobic magnetite because of the quantitative relationship between the amount of adsorbed aromatic compound and the amount of modified functional groups on the magnetite, and there was a poor fitting for the adsorption isotherm models. However, the adsorption behavior of the 1 < log Pow < 2 group was sensitive to the modified functional group on the hydrophobic magnetite. The main adsorption mechanism for this group was the ␲-electron interaction between the compounds and the phenyl group on Ph-mag rather than the hydrophobic interaction. The adsorption behavior of this group could not be demonstrated for either mono or multilayer adsorption under these adsorption conditions. The results of adsorption experiments under dehydration conditions indicated that an inhibiting factor for nitrobenzene adsorption is hydration from water molecules. The simple system of the present study, which used magnetite without porosity, enabled clarification of the factors that determine adsorption behavior. Thus, the results were significant with regard to the selection of an optimal surface modifier for adsorption or solid-phase extraction as well as magnetite separation, which could be valuable in the design of a novel high-performance adsorbent. Acknowledgements The N2 BET analysis in this work was carried out with Autsorb6 at the OPEN FACILITY, Hokkaido University Sousei Hall. References [1] E. Ferrarese, G. Andreottola, I.A. Oprea, Remediation of PAH-contaminated sediments by chemical oxidation, J. Hazard. Mater. 152 (2008) 128–139. [2] Z.M. Shen, D. Wu, J. Yang, T. Yuan, W.H. Wang, J.P. Jia, Methods to improve electrochemical treatment effect of dye wastewater, J. Hazard. Mater. 131 (2006) 90–97. [3] X. Shen, L. Zhu, G. Liu, H. Yu, H. Tang, Enhanced photocatalytic degradation and selective removal of nitrophenols by using surface molecular imprinted titania, Environ. Sci. Techol. 42 (2008) 1687–1692. [4] F.J. Beltrán, G. Ovejero, J.M. Encinar, J. Rivas, Oxidation of polynuclear aromatic hydrocarbons in water. 1. Ozonation, Ind. Eng. Chem. Res. 34 (1995) 1596–1606. [5] U.K. Ghosh, N.C. Pradhan, B. Adhikari, Separation of water and o-chlorophenol by pervaporation using HTPB-based polyurethaneurea membranes and application of modified Maxwell-Stefan equation, J. Membr. Sci. 272 (2006) 93–102.

334

T. Sasaki, S. Tanaka / Journal of Hazardous Materials 196 (2011) 327–334

[6] D.M. Nevskaia, E. Castillejos-Lopez, V. Munoz, A. Guerrero-Ruiz, Adsorption of aromatic compounds from water by treated carbon materials, Environ. Sci. Technol. 38 (2004) 5786–5796. [7] M.R. Burkhardt, S.D. Zaugg, T.L. Burbank, M.C. Olson, J.L. Iverson, Pressurized liquid extraction using water/isopropanol coupled with solid-phase extraction cleanup for semivolatile organic compounds polycyclic aromatic hydrocarbons [PAH], and alkylated PAH homolog groups in sediment, Anal. Chim. Acta 549 (2005) 104–116. [8] V.K. Gupta, I. Ali, Removal of endosulfan and methoxychlor from water on carbon slurry, Environ. Sci. Technol. 42 (2008) 766–770. [9] V.K. Gupta, A. Mittal, V. Gajbe, J. Mittal, Adsorption of basic fuchsin using waste materials-bottom ash and deoiled soya-as adsorbents, J. Colloid Interface Sci. 319 (2008) 30–39. [10] V.K. Gupta, P.J.M. Carrott, M.M.M. Ribeiro Carrott, Suhas, Low cost adsorbents: growing approach to wastewater treatment – a review, Crit. Rev. Environ. Sci. Technol. 39 (2009) 783–842. [11] I. Ali, V.K. Gupta, Advances in water treatment by adsorption technology, Nat. Protoc. 1 (6) (2007) 2661–2667. [12] H. Zhan, Regeneration of exhausted activated carbon by electrochemical method, Chem. Eng. J. 85 (2002) 81–85. [13] J.I. Taylor, C.D. Hurst, M.J. Davies, N. Sachsinger, I.J. Bruce, Application of magnetite and silica-magnetite composites to the isolation of genomic DNA, J. Chromatogr. A 890 (2000) 159–166. [14] D.S. Jiang, S.Y. Iong, J. Huang, H.Y. Xiao, J.Y. Zhou, Immobilization of Pycnoporus sanguineus laccase on magnetic chitosan microspheres, Biochem. Eng. J. 25 (2005) 15–23. [15] J. Kim, J.E. Lee, J. Lee, J.H. Yu, B.C. Kim, K. An, Y. Hwang, C.H. Shin, J.G. Park, J. Kim, T. Hyeon, Magnetic fluorescent delivery vehicle using uniform mesoporous silica spheres embedded with monodisperse magnetic and semiconductor nanocrystals, J. Am. Chem. Soc. 128 (2006) 688–689. [16] J.F. Liu, Z.S. Zhao, G.B. Jiang, Coating Fe3 O4 magnetic nanoparticles with humic acid for high efficient removal of heavy metals in water, Environ. Sci. Technol. 42 (2008) 6949–6954. [17] S. Shin, J. Jang, Thiol containing polymer encapsulated magnetic nanoparticles as reusable and efficiently separable adsorbent for heavy metal ions, Chem. Commun. (2007) 4230–4232. [18] X. Zhao, Y. Shi, Y. Cai, S. Mou, Cetyltrimethylammonium bromide-coated magnetic nanoparticles for the preconcentration of phenolic compounds from environmental water samples, Environ. Sci. Technol. 42 (2008) 1201–1206. [19] G. Shan, H. Liu, J. Xing, G. Zhang, K. Wang, Separation of polycyclic aromatic compounds from model gasoline by magnetic alumina sorbent based on ␲complexation, Ind. Eng. Chem. Res. 43 (2004) 758–761. [20] X. Zhang, H. Niu, Y. Pan, Y. Shi, Y. Cai, Chitosan-coated octadecyl-functionalized magnetite nanoparticles: preparation and application in extraction of trace pollutants from environmental water samples, Anal. Chem. 82 (2010) 2363–2371. [21] P. Yu, Q. Wang, X. Zhang, X. Zhang, S. Shen, Y. Wang, Development of superparamagnetic high-magnetization C18-functionalized magnetic silica

[22]

[23]

[24]

[25] [26]

[27]

[28] [29]

[30]

[31]

[32]

[33] [34] [35]

[36]

[37]

nanoparticles as sorbents for enrichment and determination of methylprednisolone in rat plasma by high performance liquid chromatography, Anal. Chim. Acta 678 (2010) 50–55. M. Mikhaylova, D.K. Kim, C.C. Berry, A. Zagorodni, M. Toprak, A.S.G. Curtis, M. Muhammed, BSA immobilization on amine-functionalized superparamagnetic iron oxide nanoparticles, Chem. Mater. 16 (2004) 2344–2354. Y. Fu, R. Yuan, L. Xu, Y. Chai, Y. Liu, D. Tang, Y. Zhang, Electrochemical impedance behavior of DNA biosensor based on colloidal Ag and bilayer two-dimensional sol–gel as matrices, J. Biochem. Biophys. Methods 62 (2005) 163–174. N.P. Hankins, J.H. O’Haver, J.H. Harwell, Modeling effects of pH and counterions on surfactant adsorption at the oxide/water interface, Ind. Eng. Chem. Res. 35 (1996) 2844–2855. M.G. Khaledi, E.D. Breyer, Quantitation of hydrophobicity with micellar liquid chromatography, Anal. Chem. 61 (1989) 1040–1047. X. Ruan, L. Zhu, B. Chen, Adsorptive characteristics of the siloxane surfaces of reduced-charge bentonites saturated with tetramethylammonium, Environ. Sci. Technol. 42 (2008) 7911–7917. P. Ruelle, The n-octanol and n-hexane/water partition coefficient of environmentally relevant chemicals predicted from the mobile order and disorder [MOD] thermodynamics, Chemosphere 40 (2000) 457–512. S.B. Haderleln, R.P. Schwarzenbach, Adsorption of substituted nitrobenzenes and nitrophenols to mineral surface, Environ. Sci. Technol. 27 (1993) 316–326. K. Sugiyama, K. Fukunaga, Y. Kuramochi, A. Iwata, S. Nishijima, S. Takeda, A. Nakahira, Oil spill recovery method using magnetic material by magnetic separation, JP 2000-176306. (In Japanese). K. Sever, M. Sarikanat, Y. Seki, V. Cecen, I.H. Tavman, Effect of fiber surface treatments on mechanical properties of epoxy composites reinforced with glass fabric, J. Mater. Sci. 43 (2008) 4666–4672. A.G. Casado, E.J.A. Hernández, P. Espinosa, J.L. Vílchez, Determination of total fatty acids (C8 –C22 ) in sludges by gas chromatography-mass spectrometry, J. Chromatogr. A 826 (1998) 49–56. E.F. Joy, A.J. Barnard Jr., Standard Methods of Chemical Analysis, in: F.J. Welcher (Ed.), Commercials Acids and Bases, vol. 2, 6th ed., D. Van Nostrand, New Jersey, 1963, pp. 580–581. A.F.M. Barton, CRC Handbook of Solubility Parameters and Other Cohesion Parameters, CRC Press, Inc., Boca Raton, FL, 1983. F.D. Lewis, B. Holman, Singlet states of benzonitrile and pdimethylaminobenzonitrile, J. Phys. Chem. 84 (1980) 2326–2328. I. Zilberberg, M. Ilchenko, O. Isayev, L. Gorb, J. Leszczynski, Modeling the gasphase reduction of nitrobenzene to nitrosobenzene by iron monoxide: a density functional theory study, J. Phys. Chem. A 108 (2004) 4878–4886. X. Ruan, L. Zhu, B. Chen, Adsorption characteristics of the siloxane surface of reduced-charge bentonites saturated with tetramethylammonium cation, Environ. Sci. Technol. 42 (2008) 7911–7917. A. Ebadi, J.S. Soltan Mohammadzadeh, A. Khudiev, What is the correct form of BET isotherm for modeling liquid phase adsorption? Adsorption 15 (2009) 65–73.

Journal of Hazardous Materials 196 (2011) 335–341

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Zinc induces chemokine and inflammatory cytokine release from human promonocytes Tsui-Chun Tsou a,∗ , How-Ran Chao b , Szu-Ching Yeh a , Feng-Yuan Tsai a , Ho-Jane Lin a a b

Division of Environmental Health and Occupational Medicine, National Health Research Institutes, Zhunan, Miaoli 350, Taiwan Department of Environmental Science and Engineering, National Pingtung University of Science and Technology, Neipu, Pingtung 912, Taiwan

a r t i c l e

i n f o

Article history: Received 18 May 2011 Received in revised form 8 September 2011 Accepted 9 September 2011 Available online 19 September 2011 Keywords: Zinc Promonocytes Chemokines Inflammatory cytokines

a b s t r a c t Our previous studies found that zinc oxide (ZnO) particles induced expression of intercellular adhesion molecule-1 (ICAM-1) protein in vascular endothelial cells via NF-␬B and that zinc ions dissolved from ZnO particles might play the major role in the process. This study aimed to determine if zinc ions could cause inflammatory responses in a human promonocytic leukemia cell line HL-CZ. Conditioned media from the zinc-treated HL-CZ cells induced ICAM-1 protein expression in human umbilical vein endothelial cells (HUVEC). Zinc treatment induced chemokine and inflammatory cytokine release from HL-CZ cells. Inhibition of NF␬B activity by over-expression of I␬B␣ in HL-CZ cells did not block the conditioned medium-induced ICAM-1 protein expression in HUVEC cells. Zinc treatment induced activation of multiple immune response-related transcription factors in HL-CZ cells. These results clearly show that zinc ions induce chemokine and inflammatory cytokine release from human promonocytes, accompanied with activation of multiple immune response-related transcription factors. Our in vitro evidence in the zinc-induced inflammatory responses of vascular cells provides a critical linkage between zinc exposure and pathogenesis of those inflammatory vascular diseases. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Epidemiologic studies indicate that exposure to fine particulate matter (PM) in air pollution is associated with systemic inflammatory markers [1] as well as incidence of cardiovascular morbidity and mortality [2–4]. However, mechanisms behind this correlation remain largely unknown. When inhaled into the respiratory tract, small particles tend to rest in the deeper part of the lungs. Ultrafine particles [5–7] and, of particular relevance to the present study, their dissolved chemicals such as metal ions can penetrate the deepest part of the lungs and cross the pulmonary epithelial barrier into the bloodstream, directly exposing the vascular cells, such as monocytes and endothelial cells, to pollutants. Analysis of zinc levels in suspended PM in air revealed that 82–93% of zinc was in the small PM10 particles [8]. Studies in mothers subjected to cigarette smoking or air pollution showed that both cigarette smoking and air pollution contributed to the increased levels of placental zinc [9]. A previous study in ambient air zinc levels and health care utilization for asthma revealed the association between elevated ambient air zinc and increased

∗ Corresponding author. Tel.: +886 37 246 166x36511; fax: +886 37 587 406. E-mail address: [email protected] (T.-C. Tsou). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.035

pediatric asthma morbidity [10]. ZnO particles induce cytotoxicity and apoptosis in mammalian cells [11,12] and the dissolved zinc ions seem to play the critical role in toxic effect of ZnO particles [13]. In our previous studies, ZnO particles induced ICAM-1 expression in vascular endothelial cells via an NF␬B dependent pathway [14] and zinc ions alone were sufficient to induce similar levels of ICAM-1 expression as ZnO particles, suggesting that dissolved zinc ions might play the major role in inflammatory effect of ZnO particles on vascular endothelial cells [15]. These studies suggest that metal particle composition, or its dissolved metal ions, may determine the capability of metal oxide nanoparticles to induce inflammation in vascular endothelial cells. Increasing evidence indicates that chronic obstructive pulmonary disease, asthma, and atherosclerosis are associated with systemic inflammatory cytokine changes. Various pathophysiological stimulators induce cytokine release, including modified LDL [16,17], free radicals [18], hemodynamic stress [19,20], and hypertension [21]. On the basis of our previous findings in vascular endothelial cells using ZnO particles, the present study aimed to determine if zinc could cause inflammatory responses in other vascular cells. We found that zinc induces chemokine and inflammatory cytokine release from human promonocytes possibly via activation of multiple immune response-related transcription factors.

336

T.-C. Tsou et al. / Journal of Hazardous Materials 196 (2011) 335–341

2. Materials and methods 2.1. Materials Zn(CH3 COO)2 (#370080250) was obtained from ACROS Organics (Geel, Belgium). Zinc preparation in this study was tested to be endotoxin-free by using an endotoxin inhibitor, polymyxin B, as previously described [22]. RayBio Human Cytokine Antibody Array 3 (#AAH-CYT-3) was obtained from RayBiotech, Inc. (Norcross, GA, USA). Rabbit polyclonal antibodies against ICAM-1 (sc-7891) and I␬B␣ (sc-847) and a goat polyclonal antibody against p-I␬B␣ (sc-7977) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). A mouse monoclonal antibody against actin (MAB1501) was purchased from Chemicon Int. Inc. (Temecula, CA, USA). Endothelial cell growth supplement (ECGS) was obtained from Sigma–Aldrich (St. Louis, MO, USA). Dulbecco’s phosphatebuffered saline (D-PBS), M199 medium, and RPMI 1640 medium were obtained from Life Technologies (Grand Island, NY, USA). Fetal bovine serum (FBS) was obtained from HyClone (Logan, UT, USA). Penicillin (10,000 units/ml)/streptomycin (10,000 ␮g/ml) solution was obtained from Invitrogen Corp. (Carlsbad, CA, USA). Gentamycin sulfate was purchased from Biological Industries (Kibbutz Beit Haemek, Israel). 2.2. Construction of recombinant adenoviruses Construction of recombinant AdEasy-GFP, AdEasy-I␬B␣, and AdV-NF␬B-Luc has been previously described [14]. For construction of AdV-AP-1-Luc, a 2360-bp DNA fragment containing seven copies of the AP-1 response element, a TATA box, and a firefly luciferase gene in pAP-1-Luc (Stratagene, La Jolla, CA, USA), was amplified by PCR using pfu DNA polymerase. The amplified DNA fragment was digested by KpnI and SalI. After separation by agarose gel electrophoresis, the purified KpnI/SalI-digested DNA fragment was cloned into the KpnI/SalI-digested pACCMV.pLpA vector [23]. The function of the CMV promoter of the pACCMV.pLpA vector used here had been abolished. The recombinant adenovirus AdVAP-1-Luc was generated by homologous recombination between the pJM17 plasmid [24] and the pACCMV.pLpA vector in 293 human embryo kidney cells. Construction of other recombinant adenoviruses (AdEasy-C/EBP-Luc, AdEasy-CRE-Luc, AdEasy-NFATLuc, AdEasy-SRE-Luc, and AdEasy-STAT-Luc) was generated using the AdEasyTM Adenoviral Vector System (Stratagene, La Jolla, CA, USA) (see Supplementary material in detail). The recombinant adenoviruses were purified and concentrated according to the manufacturer’s instructions. General information of these immune response-related transcription factor-mediated luciferase reporter adenoviruses is summarized in Table 1. 2.3. Cells and treatments Human promonocytic leukemia cell line HL-CZ (BCRC-60043), originally established by Dr. Wu-Tse Liu (National Yang-Ming University, Taipei, Taiwan) [25], were purchased from Bioresource Collection and Research Center (BCRC, Hsinchu, Taiwan) and were routinely cultured in RPMI 1640 medium. HUVEC cells were obtained by using collagenase digestion of umbilical veins [26] and were routinely cultured in M199 medium as previously described [27]. HL-CZ cells (8.5 × 106 cells per 100-mm dish) were left untreated or treated with Zn(CH3 COO)2 as indicated for 6 h. Following treatments, conditioned media were dialyzed against D-PBS with stirring at 4 ◦ C for 42 h, sterilized with a 0.45-␮m syringe filter, and then was added with M199 medium (with 20% FBS and 30 ␮g/ml ECGS) at 1/1 ratio (v/v). HUVEC cells were treated with this conditioned medium/M199 mixtures for different time periods

as indicated. Following the treatments, cell lysates were collected for immunoblot analyses. In some cases requiring adenovirus infection, HL-CZ cells (3 × 106 cells per 100-mm dish) were first infected with AdEasyGFP or AdEasy-I␬B␣ at a multiplicity of infection (MOI) of 50 pfu/cell for 24 h. The infected cells were replaced with fresh RPMI 1640 medium and cultured for another 24 h for recovery. Hereafter, the cells were ready for the zinc treatments as just described. 2.4. Immunoblot analysis Following treatments, cells were lysed in ice-cold RIPA buffer (50 mM Tris–HCl, pH 7.5, 5 mM EDTA, 1 mM EGTA, 1% Triton X-100, 0.25% sodium deoxycholate) containing PMSF, (2 mM), aprotinin (2 ␮g/ml), leupeptin (2 ␮g/ml), NaF (2 mM), Na3 VO4 (2 mM), and ␤-glycerophosphate (0.2 mM). The cell lysates were subjected to SDS–PAGE and immunoblot analysis, as described previously [28]. The blots were probed with a primary antibody against ICAM-1, phosphor-I␬B␣ (p-I␬B␣), I␬B␣, or actin. HRP-conjugated secondary antibodies. Protein bands in the membrane were visualized in an Xray film by using Western Lightning Chemiluminescence Reagent Plus (PerkinElmer Life Sciences, Boston, MA, USA). The protein band intensity was quantified by densitometry scanning of X-ray films. 2.5. Analysis of cytokines in conditioned media from HL-CZ cells Following zinc treatments of HL-CZ cells, conditioned media were dialyzed against D-PBS with stirring at 4 ◦ C for 42 h and sterilized with a 0.45-␮m syringe filter. Cytokines in conditioned medium were analyzed with the RayBio Human Cytokine Antibody Array 3 according to the manufacturer’s instructions (see Supplementary material in detail). The kit provides a simple array format, and highly sensitive approach to simultaneously detect 42 cytokine expression levels from conditioned media. 2.6. Immune response-related transcription factor-mediated luciferase reporter assay To determine the activation of those immune response-related transcription factors, HL-CZ cells (1.0 × 104 cells per well in 96-well plates) were infected with one of the recombinant adenoviruses (AdV-AP-1-Luc, AdV-NF␬B-Luc, AdEasy-SRE-Luc, AdEasy-NFATLuc, AdEasy-CRE-Luc, AdEasy-C/EBP-Luc, and AdEasy-STAT-Luc) (Table 1) at a MOI of 1 pfu/cell for 24 h. Following the adenovirus infection, the infected cells were replaced with fresh RPMI 1640 medium and cultured for another 24 h for recovery. Then, the infected cells were left untreated or treated with 150 ␮M Zn(CH3 COO)2 for 6 h. Luciferase activity of each sample was determined using the Luciferase Assay System (Promega, Madison, WI), according to the manufacturer’s instructions. 2.7. Statistics Each experiment was performed independently at least three times. The statistical analysis was expressed using the mean ± standard deviation (SD) from each independent experiment. Induction of ICAM-1 protein expression in HUVEC cells by conditioned medium were examined by Student’s t-tests with 2000 bootstrap samples. One-sample t-tests were used to determine the significant differences in induction of chemokine and inflammatory cytokine release from HL-CZ cells between the zinc-treated and untreated groups (test value = 1). Differences were considered statistically significant when p < 0.05. Analyses were carried out using the Statistical Package for Social Sciences (SPSS) version 12.0 (SPSS Inc., Chicago, IL, USA).

T.-C. Tsou et al. / Journal of Hazardous Materials 196 (2011) 335–341

337

Table 1 Immune response-related transcription factor-mediated luciferase reporter adenoviruses. Recombinant adenoviruses

Transcription factors

Response element sequences (RES) a direction (5 → 3 )

AdV-AP-1-Luc

AP-1

TGACTAATGACTAATGACTAATGACTAA TGACTAATGACTAATGACTAA

AdEasy-C/EBP-Luc

C/EBP

ATTGCGCAATATTGCGCAATATTGCGCAAT

AdEasy-CRE-Luc

CREBP

AGCCTGACGTCAGAGAGCCTGACGTCAGAG AGCCTGACGTCAGAGAGCCTGACGTCAGAG AGCCTGACGTCAGAGAGCCTGACGTCAGAG AGCCTGACGTCAGAG

AdEasy-NFAT-Luc

NFAT

ACGCCTTCTGTATGAAACAGTTTTTCCTCC ACGCCTTCTGTATGAAACAGTTTTTCCTCC ACGCCTTCTGTATGAAACAGTTTTTCCTCC

AdV-NF␬B-Luc

NF␬B

GGGGACTTTCCGCTTGGGGACTTTCCGCT GGGGACTTTCCGCTGGGGACTTTCCGCT GGGGACTTTCCGC

AdEasy-SRE-Luc

Elk1/SRF

CCATATTAGGACATCTAGGATGT CCATATTAGGACATCTAGGATGTCCATATTAGG AGCTAGCCCATATTAGGACATGCTAGGATGT CCATATTAGGAGCATCTAGGATGTCCCATATTAGG AC

AdEasy-STAT-Luc

STAT

GGTTCCCGTAAATGCATCAGGTTCCCGTAAA TGCATCAGGTTCCCGTAAATGCATCAGG TTCCCGTAAATG

a

RES-driven luciferase constructs

The response elements are marked by shading or underlining.

3. Results 3.1. Conditioned media from the zinc-treated HL-CZ cells induce ICAM-1 protein expression in HUVEC cells Our previous study showed that zinc ions alone are sufficient to induce similar levels of ICAM-1 expression as ZnO particles, suggesting that the dissolved zinc ions play the major role in inflammatory effect of ZnO particles on vascular endothelial cells. In this study, we used HL-CZ cells, a human promonocytic leukemia cell line, and a soluble zinc compound Zn(CH3 COO)2 to determine if zinc ions could cause inflammatory responses in this human promonocyte cell line. HL-CZ cells were treated with different concentrations of Zn(CH3 COO)2 (0, 30, 50, 100, and 150 ␮M) for 6 h and then the conditioned media were collected. Then, HUVEC cells were treated with the conditioned media for 24 h. Following treatments, HUVEC cell lysates were collected for analysis of ICAM-1 protein expression with immunoblot. Results in Fig. 1A showed that levels of ICAM-1 protein expression in HUVEC cells were positively correlated with the Zn(CH3 COO)2 concentrations used in HL-CZ treatments. The time-dependent ICAM-1 induction in HUVEC cells by the conditioned media from HL-CZ cells treated with 150 ␮M Zn(CH3 COO)2 was also observed (Fig. 1B). The ICAM-1 induction could be up to 6–7 folds. These results suggested that zinc treatments might cause release of inflammatory cytokines from HL-CZ cells into culture medium and the released inflammatory cytokines were able to activate ICAM-1 expression in HUVEC cells. 3.2. Zinc treatments induce chemokine and inflammatory cytokine release from HL-CZ cells Because a large number of cytokines have been characterized, it was complicated that how to effectively identify the expression profiles of multiple cytokines in conditioned medium. By using the RayBio Human Cytokine Antibody Array 3 for detection of secreted/active cytokines, we were able to simultaneously detect 42 cytokine levels in conditioned media. Results in Fig. 1 showed

that conditioned media from HL-CZ cells treated with 150 ␮M Zn(CH3 COO)2 caused the maximum level of ICAM-1 expression in HUVEC cells. Therefore, conditioned media by such zinc treatments were collected for cytokine analysis with the cytokine antibody array. As shown in Table 2, the zinc treatment induced significant releases of GRO-␣, IL-6, IL-7, IL-8, and IL-10 by 3.98, 1.92, 1.72, 1.34, and 1.46 folds, respectively. Although, the array detects only 42 cytokines, the results clearly show that the zinc treatment causes significant releases of chemokines (e.g., GRO-␣ and IL-8), pro-inflammatory cytokines (e.g., IL-6 and IL-7), and antiinflammatory cytokines (e.g., IL-10) from HL-CZ cells. 3.3. Inhibition of NFB activity by over-expression of IB˛ in HL-CZ cells does not block the conditioned medium-induced ICAM-1 expression in HUVEC cells Because NF␬B plays the major role in regulating the zincinduced ICAM-1 expression in HUVEC cells [14], it was of importance to further ask if NF␬B also mediates the zincinduced inflammatory cytokine release from HL-CZ cells. By over-expression of I␬B␣ in HL-CZ cells using an adenovirusmediated expression system, we investigated whether the zinc treatment was able to induce I␬B␣ phosphorylation in HL-CZ cells and whether overexpression of I␬B␣ in HL-CZ cells could block the conditioned medium-induced ICAM-1 expression in HUVEC cells. As shown in Fig. 2, in the un-infected and the AdEasyGFP-infected controls, treatment of HL-CZ cells with 150 ␮M Zn(CH3 COO)2 for 6 h induced degradation of endogenous I␬B␣ by 63%; I␬B␣ phosphorylation was barely detectable most likely due to the rapid polyubiquitination and subsequent degradation of phosphorylated I␬B␣ by the 26S proteasome [29]. In the adenovirus-mediated overexpression of I␬B␣ experiments, results indicated that inhibition of NF␬B activity by over-expression of I␬B␣ in HL-CZ cells did not block the conditioned medium-induced ICAM-1 protein expression in HUVEC cells; meanwhile, the zinc treatments did enhance phosphorylation of exogenous I␬B␣ in HLCZ cells. Because of the abundant I␬B␣ expression by adenovirus

338

T.-C. Tsou et al. / Journal of Hazardous Materials 196 (2011) 335–341

Table 2 Analysis of zinc effect on cytokine release from HL-CZ cells with RayBio Human Cytokine Antibody Array 3. Cytokines

ENA-78 GCSF GM-CSF GRO GRO-␣ I-309 IL-1␣ IL-1␤ IL-2 IL-3 IL-4 IL-5 IL-6 IL-7 IL-8 IL-10 IL-12 p40p70 IL-13 IL-15 INF-␥ MCP-1 MCP-2 MCP-3 MCSF MDC MIG MIP-1␦ RANTES SCF SDF-1 TARC TGF-␤1 TNF-␣ TNF-␤ EGF IGF-1 Angiogenin Oncostatin M Thrombopoietin VEGF PDGF BB Leptin * **

Induction fold (treated/untreated)

1st

2nd

3rd

0.540 1.027 0.978 1.854 3.454 1.054 0.799 0.992 1.018 0.550 0.764 0.812 1.930 1.427 1.239 1.631 0.613 0.892 0.708 0.904 1.449 1.005 0.540 0.962 1.176 0.699 0.716 1.031 0.951 0.678 0.724 0.813 1.008 0.781 0.827 0.533 1.014 0.886 1.012 1.361 0.890 0.998

1.059 1.009 1.484 2.479 5.212 1.186 1.271 1.052 1.447 0.987 2.004 1.735 2.028 1.868 1.348 1.313 0.987 1.060 1.198 1.001 1.223 1.461 1.503 1.168 0.906 1.099 1.032 0.997 0.864 0.918 0.988 1.017 0.847 1.009 0.920 1.249 0.965 0.976 1.249 1.817 0.837 1.026

0.914 0.797 0.712 1.634 3.276 0.871 0.836 0.975 1.009 1.046 1.014 0.795 1.787 1.849 1.431 1.434 0.913 0.902 0.998 0.979 1.013 1.121 0.749 1.058 0.872 0.863 0.820 0.862 0.794 0.844 1.011 0.970 0.900 1.028 0.894 0.698 0.690 0.761 0.984 1.243 0.725 0.922

Mean

SD

p value for one-sample t-test (test value = 1)

0.838 0.944 1.058 1.989 3.981 1.037 0.969 1.006 1.158 0.861 1.261 1.114 1.915 1.715 1.339 1.459 0.838 0.951 0.968 0.961 1.228 1.196 0.931 1.063 0.985 0.887 0.856 0.963 0.870 0.813 0.908 0.933 0.918 0.939 0.880 0.827 0.890 0.874 1.082 1.474 0.817 0.982

0.268 0.128 0.392 0.438 1.070 0.158 0.262 0.040 0.250 0.271 0.656 0.538 0.121 0.249 0.096 0.161 0.198 0.094 0.246 0.051 0.218 0.237 0.507 0.103 0.167 0.201 0.161 0.089 0.079 0.123 0.159 0.107 0.082 0.137 0.048 0.375 0.175 0.108 0.146 0.303 0.084 0.054

0.404 0.530 0.822 0.060 0.040* 0.725 0.855 0.812 0.388 0.468 0.562 0.749 0.006** 0.038* 0.026* 0.038* 0.292 0.465 0.843 0.319 0.211 0.289 0.835 0.403 0.888 0.433 0.262 0.551 0.103 0.119 0.422 0.393 0.227 0.524 0.050 0.507 0.388 0.181 0.434 0.114 0.064 0.621

p < 0.05. p < 0.01.

system, we were able to detect the I␬B␣ phosphorylation. These results suggest that inhibition of NF␬B alone is not sufficient to completely block the inflammatory cytokine release from HL-CZ cells. 3.4. Zinc treatment induces activation of multiple immune response-related transcription factors in HL-CZ cells On the basis of our present results, it was suggested that, in addition to NF␬B, multiple immune response-related transcription factors might be involved in the zinc-induced cytokine release from HL-CZ cells. To verify this hypothesis, seven recombinant adenoviruses carrying a response element-driven luciferase reporter gene were established (Table 1). These immune response-related transcription factors include AP-1, C/EBP, CREBP, NFAT, NF␬B, SRF, and STAT [30–32]; the activated transcription factors mediate luciferase expression via binding to their respective response elements. HL-CZ cells were infected with one of the recombinant adenoviruses and then were treated with 150 ␮M Zn(CH3 COO)2 for 6 h. As shown in Fig. 3, the zinc treatment induced activation of AP1, C/EBP, CREBP, NFAT, NF␬B, SRF, and STAT by 1.18, 2.90, 2.46, 1.64, 4.27, 1.42, and 1.42 folds, respectively, in HL-CZ cells. Among them,

C/EBP, CREBP, NFAT, NF␬B, and SRF were significantly activated by the zinc treatment. 4. Discussion Zinc is an essential trace element for animals and play important roles in regulation of immune function in humans [33]. However, excess zinc may also deregulate the homeostasis of immune system. Epidemiological studies revealed the association between elevated ambient air zinc and increased pediatric asthma morbidity [10]. Animal studies indicated that the ambient PM2.5 samples with higher levels of metals, such as zinc, caused increases in the allergic respiratory disease in mice [34]. Our previous in vitro evidence revealed an important role for ZnO particle, or its dissolved zinc ions, in modulating inflammatory responses of vascular endothelial cells [14,15]. The present study further demonstrates that zinc ions induce chemokine and inflammatory cytokine release from vascular promonocytes, possibly, via activating multiple immune response-related transcription factors. In the previous study, we evaluated vascular endothelial dysfunction by using ICAM-1 expression, an indicator for inflammatory response. ICAM-1, continuously present in low concentrations in

T.-C. Tsou et al. / Journal of Hazardous Materials 196 (2011) 335–341

339

Fig. 2. Inhibition of NF␬B activity by over-expression of I␬B␣ in HL-CZ cells does not block the conditioned medium-induced ICAM-1 protein expression in HUVEC cells. HL-CZ cells were left uninfected (−) or infected with a recombinant adenovirus (ADV), AdEasy-I␬B␣ (I␬B␣) or AdEasy-GFP (GFP). Cells were treated with 150 ␮M of Zn(CH3 COO)2 for 6 h. Following treatments, both HL-CZ cell lysates and the conditioned media were collected. Then, HUVEC cells were treated with the conditioned media for 24 h and HUVEC cell lysates were collected. By using immunoblot analysis, I␬B␣ expression (I␬B␣) and phosphorylation (p-I␬B␣) in HL-CZ cells and protein levels of ICAM-1 and actin in HUVEC cells were determined.

Fig. 1. Conditioned medium (CM) from the zinc-treated HL-CZ cells induces ICAM1 protein expression in HUVEC cells. (A) HL-CZ cells were treated with different concentrations of Zn(CH3 COO)2 (0, 30, 50, 100, and 150 ␮M) for 6 h and then the conditioned medium was collected. HUVEC cells were treated with the conditioned medium for 24 h. (B) HUVEC cells were also treated with the conditioned medium for 6, 12, and 24 h; the conditioned medium was collected from HL-CZ cells treated with 150 ␮M Zn(CH3 COO)2 for 6 h. For C2 controls, HUVEC cells were treated with the conditioned medium for 24 h; the 6-h HL-CZ cultured medium with no zinc treatment were used as the conditioned medium. For C1 controls, HUVEC cells were cultured in M199 medium for 24 h. Following treatments, HUVEC cell lysates were analyzed for ICAM-1 and actin protein levels by immunoblot analysis. Representative immunoblots are shown (inserts). The protein levels of ICAM-1 and actin were quantified. The experiment was repeated three times. The data are expressed as relative ICAM-1 protein levels compared to that of the untreated control (C1) and are presented as means ± SD. Differences in ICAM-1 expression were examined by using Student’s t-tests with 2000 bootstrap samples (for A, 0 vs. C1 and the other CM-treated groups vs. 0; for B, the other CM-treated groups vs. C2). *p < 0.05.

the membranes of leukocytes and endothelial cells, is a ligand for LFA-1 (integrin), a receptor found on leukocytes [35]. Here, we showed that the conditioned medium from HL-CZ cells treated with 150 ␮M Zn(CH3 COO)2 for 6 h caused a maximum level of ICAM1 induction (Fig. 1). Therefore, we collected these conditioned media for analysis of cytokine secretion using the RayBio Human Cytokine Antibody Array 3. The kit detects secreted cytokines in conditioned medium that presents a more accurate reflection of active cytokine levels. Cytokines or other proinflammatory mediators induce adhesion molecule expression and facilitate leukocyte attachment to vascular endothelium via ICAM-1/LFA-1 binding [36,37]. The adhesion of monocytes to the arterial wall and their subsequent infiltration and differentiation into macrophages is a crucial step in the development of atherosclerosis. Analysis of cytokines in conditioned media clearly showed that the zinc treatment induced chemokine and inflammatory cytokine release from HL-CZ cells. With the Cytokine Antibody Array, the GRO antibody detects CXCL1, CXCL2, and CXCL3; the

GRO-␣ antibody detects only CXCL1. The zinc treatment induced marked releases of GRO (CXCL1, CXCL2, and CXCL3) and GRO-␣ (CXCL1) by 1.99 and 3.98 folds, respectively (Table 2). Because GRO activity involves three CXC chemokines, it is suggested that GRO␣ could be the major CXCL chemokine secreted by HL-CZ cells. GRO-␣, initially isolated and characterized by its growth stimulatory activity on malignant melanoma cells, is a chemoattractant for neutrophils [38,39]. Recently, many new functions of GRO-␣ have been discovered and associated with atherosclerosis, angiogenesis, and many inflammatory conditions [40]. Moreover, the zinc treatment also induced significant releases of inflammatory cytokines, including IL-6, IL-7, IL-8, and IL-10. In addition to its known role in mediating the systemic acutephase response, IL-6 plays multiple roles in initiating and sustaining vascular inflammation [41]. IL-7 has many roles in T cells, dendritic cells, and bone biology in humans and is involved in chronic inflammation linking stroma and adaptive immunity [42]. IL-8, or CXCL8, is also a CXC chemokine and bears the primary responsibility for the recruitment of monocytes and neutrophils, the signature

Fig. 3. Zinc treatment induces activation of multiple immune response-related transcription factors in HL-CZ cells. HL-CZ cells were infected with one recombinant virus carrying a transcription factor-mediated luciferase reporter gene. The infected cells were treated with 150 ␮M Zn(CH3 COO)2 for 6 h. Following the treatments, luciferase activity of each sample was determined. The experiment was repeated three times. The data are expressed as relative reporter activity as compared to that of the untreated control and are presented as means ± SD. Differences in reporter activity (the zinc-treated vs. the zinc-untreated) of each transcription factor were examined by using one-sample t-tests. *p < 0.05, **p < 0.01, and ***p < 0.001.

340

T.-C. Tsou et al. / Journal of Hazardous Materials 196 (2011) 335–341

cells of acute inflammatory response [43]. IL-10, an important cytokine with anti-inflammatory properties, is produced by activated immune cells, in particular monocytes/macrophages and T cell subsets including Tr1, Treg, and Th1 cells [44]. On the basis of this information, the present results demonstrate the potential impacts of zinc exposure on disturbing homeostasis of inflammation via these inflammatory cytokines. The major role of NF␬B in regulation of zinc-induced ICAM-1 expression in vascular endothelial cells has been demonstrated [14]. However, results in Fig. 2 suggest that, in addition to NF␬B, other immune response-related transcription factors may be also involved in zinc-induced chemokine/inflammatory cytokine release from HL-CZ cells. Indeed, among the seven immune response-related transcription factors tested, C/EBP, CREBP, NFAT, NF␬B, and SRF were significantly activated (Fig. 3). Therefore, the zinc-induced inflammatory responses in promonocytes involve a more sophisticated signaling regulation than that in vascular endothelial cells. Moreover, the 6-h zinc treatment conditions were used in cytokine analyses and immune response-related transcription factor-mediated luciferase reporter assay; this short-term treatment was designed to avoid the potential secondary inflammatory responses. However, we could not rule out the possibility in the meantime. Thus, deciphering the time-course activation of inflammation-related signaling molecules and transcription factors in detail is needed in the following study. The present study mainly dealt with the potential impacts on homeostasis of vascular immune system by those excess zinc exposures, especially from ambient particulate pollutants. In addition to vascular endothelial cells [14], here we further demonstrated that zinc induces chemokine and inflammatory cytokine release from human promonocytes. On the basis of this and our previous studies [14,15], the possible scenario of inflammatory responses induced by zinc from ambient particulate pollutants is described. First, fine particles tend to be trapped in the deeper pulmonary alveoli through inhalation. Second, in situ decomposition of the trapped particles results in a local increase of metal ions (such as zinc and nickel) and thus may activate pulmonary inflammation. Third, ultrafine particles and dissolved metal ions are able to cross the pulmonary epithelial barrier and then into the bloodstream. Finally, direct exposure of vascular cells, including endothelial cells and circulating blood cells, to those proinflammatory metals ions, such as zinc ions in this study, elicits vascular inflammation. These studies provide new insight for understanding the mechanisms of those inflammatory diseases induced by ambient particulate pollutants. 5. Conclusions Using human HL-CZ promonocytes as an in vitro system, this study reveals two important findings. Zinc treatment induces chemokine and inflammatory cytokine release from HL-CZ cells. The process involves activation of multiple immune responserelated transcription factors, including C/EBP, CREBP, NFAT, NF␬B, and SRF. Conflict of interest The authors declare that there are no conflicts of interest. Acknowledgements This work was supported by grants from the National Health Research Institutes (EO-099-PP-03, EO-100-PP-03) and the National Science Council (NSC97-2314-B-400–003-MY3) in Taiwan. We are grateful to Dr. Shu-Ching Hsu (Vaccine Research and

Development Center, National Health Research Institutes, Taiwan) for helpful suggestion. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.09.035. References [1] B. Hoffmann, S. Moebus, N. Dragano, A. Stang, S. Mohlenkamp, A. Schmermund, M. Memmesheimer, M. Brocker-Preuss, K. Mann, R. Erbel, K.H. Jockel, Chronic residential exposure to particulate matter air pollution and systemic inflammatory markers, Environ. Health Perspect. 117 (2009) 1302–1308. [2] J.M. Samet, F. Dominici, F.C. Curriero, I. Coursac, S.L. Zeger, Fine particulate air pollution and mortality in 20 U.S. cities 1987–1994, N. Engl. J. Med. 343 (2000) 1742–1749. [3] A. Peters, D.W. Dockery, J.E. Muller, M.A. Mittleman, Increased particulate air pollution and the triggering of myocardial infarction, Circulation 103 (2001) 2810–2815. [4] C.A. Pope, R.T. 3rd, G.D. Burnett, M.J. Thurston, E.E. Thun, D. Calle, J.J. Krewski, Godleski, Cardiovascular mortality and long-term exposure to particulate air pollution: epidemiological evidence of general pathophysiological pathways of disease, Circulation 109 (2004) 71–77. [5] W.G. Kreyling, M. Semmler, F. Erbe, P. Mayer, S. Takenaka, H. Schulz, G. Oberdorster, A. Ziesenis, Translocation of ultrafine insoluble iridium particles from lung epithelium to extrapulmonary organs is size dependent but very low, J. Toxicol. Environ. Health A 65 (2002) 1513–1530. [6] A. Nemmar, P.H. Hoet, D. Dinsdale, J. Vermylen, M.F. Hoylaerts, B. Nemery, Diesel exhaust particles in lung acutely enhance experimental peripheral thrombosis, Circulation 107 (2003) 1202–1208. [7] A. Nemmar, P.H. Hoet, B. Vanquickenborne, D. Dinsdale, M. Thomeer, M.F. Hoylaerts, H. Vanbilloen, L. Mortelmans, B. Nemery, Passage of inhaled particles into the blood circulation in humans, Circulation 105 (2002) 411–414. [8] V. Vadjic, S. Zuzul, G. Pehnec, Zinc levels in suspended particulate matter in Zagreb air, Bull. Environ. Contam. Toxicol. 85 (2010) 628–631. [9] H.C. Sorkun, F. Bir, M. Akbulut, U. Divrikli, G. Erken, H. Demirhan, E. Duzcan, L. Elci, I. Celik, U. Yozgatli, The effects of air pollution and smoking on placental cadmium, zinc concentration and metallothionein expression, Toxicology 238 (2007) 15–22. [10] J.M. Hirshon, M. Shardell, S. Alles, J.L. Powell, K. Squibb, J. Ondov, C.J. Blaisdell, Elevated ambient air zinc increases pediatric asthma morbidity, Environ. Health Perspect. 116 (2008) 826–831. [11] A. Gojova, B. Guo, R.S. Kota, J.C. Rutledge, I.M. Kennedy, A.I. Barakat, Induction of inflammation in vascular endothelial cells by metal oxide nanoparticles: effect of particle composition, Environ. Health Perspect. 115 (2007) 403–409. [12] T. Xia, M. Kovochich, M. Liong, L. Madler, B. Gilbert, H. Shi, J.I. Yeh, J.I. Zink, A.E. Nel, Comparison of the mechanism of toxicity of zinc oxide and cerium oxide nanoparticles based on dissolution and oxidative stress properties, ACS Nano 2 (2008) 2121–2134. [13] W. Song, J. Zhang, J. Guo, J. Zhang, F. Ding, L. Li, Z. Sun, Role of the dissolved zinc ion and reactive oxygen species in cytotoxicity of ZnO nanoparticles, Toxicol. Lett. 199 (2010) 389–397. [14] T.C. Tsou, S.C. Yeh, F.Y. Tsai, H.J. Lin, T.J. Cheng, H.R. Chao, L.A. Tai, Zinc oxide particles induce inflammatory responses in vascular endothelial cells via NF-␬B signaling, J. Hazard. Mater. 183 (2010) 182–188. [15] S.C. Yeh, F.Y. Tsai, H.R. Chao, T.C. Tsou, Zinc ions induce inflammatory responses in vascular endothelial cells, Bull. Environ. Contam. Toxicol. 87 (2011) 113–116. [16] J.A. Berliner, D.S. Schwartz, M.C. Territo, A. Andalibi, L. Almada, A.J. Lusis, D. Quismorio, Z.P. Fang, A.M. Fogelman, Induction of chemotactic cytokines by minimally oxidized LDL, Adv. Exp. Med. Biol. 351 (1993) 13–18. [17] J. Hulthe, B. Fagerberg, Circulating oxidized LDL is associated with subclinical atherosclerosis development and inflammatory cytokines (AIR Study), Arterioscler. Thromb. Vasc. Biol. 22 (2002) 1162–1167. [18] Y. Xu, M. Rojkind, M.J. Czaja, Regulation of monocyte chemoattractant protein 1 by cytokines and oxygen free radicals in rat hepatic fat-storing cells, Gastroenterology 110 (1996) 1870–1877. [19] A.V. Sterpetti, A. Cucina, A.R. Morena, S. Di Donna, L.S. D’Angelo, A. Cavalarro, S. Stipa, Shear stress increases the release of interleukin-1 and interleukin-6 by aortic endothelial cells, Surgery 114 (1993) 911–914. [20] K. Sakai, M. Mohtai, J. Shida, K. Harimaya, S. Benvenuti, M.L. Brandi, T. Kukita, Y. Iwamoto, Fluid shear stress increases interleukin-11 expression in human osteoblast-like cells: its role in osteoclast induction, J. Bone Miner. Res. 14 (1999) 2089–2098. [21] M. Humbert, G. Monti, F. Brenot, O. Sitbon, A. Portier, L. Grangeot-Keros, P. Duroux, P. Galanaud, G. Simonneau, D. Emilie, Increased interleukin-1 and interleukin-6 serum concentrations in severe primary pulmonary hypertension, Am. J. Respir. Crit. Care Med. 151 (1995) 1628–1631. [22] L. Cardoso, M. Araujo, A. Goes, L. Pacifico, R. Oliveira, S. Oliveira, Polymyxin, B as inhibitor of LPS contamination of Schistosoma mansoni recombinant proteins in human cytokine analysis, Microbial Cell Factories 6 (2007) 1. [23] A.M. Gomez-Foix, W.S. Coats, S. Baque, T. Alam, R.D. Gerard, C.B. Newgard, Adenovirus-mediated transfer of the muscle glycogen phosphorylase gene into

T.-C. Tsou et al. / Journal of Hazardous Materials 196 (2011) 335–341

[24]

[25] [26]

[27]

[28]

[29] [30] [31] [32]

[33]

hepatocytes confers altered regulation of glycogen metabolism, J. Biol. Chem. 267 (1992) 25129–25134. W.J. McGrory, D.S. Bautista, F.L. Graham, A simple technique for the rescue of early region I mutations into infectious human adenovirus type 5, Virology 163 (1988) 614–617. W.T. Liu, C.L. Chen, S.S. Lee, C.C. Chan, F.L. Lo, Y.C. Ko, Isolation of dengue virus with a human promonocyte cell line, Am. J. Trop. Med. Hyg. 44 (1991) 494–499. T.C. Tsou, S.C. Yeh, E.M. Tsai, F.Y. Tsai, H.R. Chao, L.W. Chang, Arsenite enhances tumor necrosis factor-␣-induced expression of vascular cell adhesion molecule-1, Toxicol. Appl. Pharmacol. 209 (2005) 10–18. T.C. Tsou, S.C. Yeh, F.Y. Tsai, J.W. Chen, H.C. Chiang, Glutathione regulation of redox-sensitive signals in tumor necrosis factor-alpha-induced vascular endothelial dysfunction, Toxicol. Appl. Pharmacol. 221 (2007) 168–178. T.C. Tsou, S.C. Yeh, F.Y. Tsai, L.W. Chang, The protective role of intracellular GSH status in the arsenite-induced vascular endothelial dysfunction, Chem. Res. Toxicol. 17 (2004) 208–217. D. Krappmann, C. Scheidereit, A pervasive role of ubiquitin conjugation in activation and termination of I␬B kinase pathways, EMBO Rep. 6 (2005) 321–326. I. Rahman, W. MacNee, Role of transcription factors in inflammatory lung diseases, Thorax 53 (1998) 601–612. V. Poli, The role of C/EBP isoforms in the control of inflammatory and native immunity functions, J. Biol. Chem. 273 (1998) 29279–29282. A. Kasza, P. Wyrzykowska, I. Horwacik, P. Tymoszuk, D. Mizgalska, K. Palmer, H. Rokita, A.D. Sharrocks, J. Jura, Transcription factors Elk-1 and SRF are engaged in IL1-dependent regulation of ZC3H12A expression, BMC Mol. Biol. 11 (2010) 14. A.S. Prasad, Zinc in human health: effect of zinc on immune cells, Mol. Med. 14 (2008) 353–357.

341

[34] S.H. Gavett, N. Haykal-Coates, L.B. Copeland, J. Heinrich, M.I. Gilmour, Metal composition of ambient PM2.5 influences severity of allergic airways disease in mice, Environ. Health Perspect. 111 (2003) 1471–1477. [35] R. Rothlein, M.L. Dustin, S.D. Marlin, T.A. Springer, A human intercellular adhesion molecule (ICAM-1) distinct from LFA-1, J. Immunol. 137 (1986) 1270–1274. [36] T. Collins, Endothelial nuclear factor-␬B and the initiation of the atherosclerotic lesion, Lab. Invest. 68 (1993) 499–508. [37] L. Yang, R.M. Froio, T.E. Sciuto, A.M. Dvorak, R. Alon, F.W. Luscinskas, ICAM-1 regulates neutrophil adhesion and transcellular migration of TNF-␣-activated vascular endothelium under flow, Blood 106 (2005) 584–592. [38] B. Moser, I. Clark-Lewis, R. Zwahlen, M. Baggiolini, Neutrophil-activating properties of the melanoma growth-stimulatory activity, J. Exp. Med. 171 (1990) 1797–1802. [39] C. Schumacher, I. Clark-Lewis, M. Baggiolini, B. Moser, High- and low-affinity binding of GRO ␣ and neutrophil-activating peptide 2 to interleukin 8 receptors on human neutrophils, Proc. Natl. Acad. Sci. U.S.A. 89 (1992) 10542–10546. [40] C. Bechara, H. Chai, P.H. Lin, Q. Yao, C. Chen, Growth related oncogene-alpha (GRO-alpha): roles in atherosclerosis, angiogenesis and other inflammatory conditions, Med. Sci. Monit. 13 (2007) RA87–RA90. [41] A.R. Brasier, The nuclear factor-␬B-interleukin-6 signalling pathway mediating vascular inflammation, Cardiovasc. Res. 86 (2010) 211–218. [42] S.M. Churchman, F. Ponchel, Interleukin-7 in rheumatoid arthritis, Rheumatology (Oxford) 47 (2008) 753–759. [43] S. Apostolakis, K. Vogiatzi, V. Amanatidou, D.A. Spandidos, Interleukin 8 and cardiovascular disease, Cardiovasc. Res. 84 (2009) 353–360. [44] R. Sabat, G. Grutz, K. Warszawska, S. Kirsch, E. Witte, K. Wolk, J. Geginat, Biology of interleukin-10, Cytokine Growth Factor Rev. 21 (2010) 331–344.

Journal of Hazardous Materials 196 (2011) 342–349

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Preparation of nanocrystalline Fe3−x Lax O4 ferrite and their adsorption capability for Congo red Lixia Wang a,b , Jianchen Li a , Yingqi Wang a , Lijun Zhao a,∗ a b

Key Laboratory of Automobile Materials (Jilin University), Ministry of Education and School of Materials Science and Engineering, Jilin University, Changchun 130022, China School of Mechanical Science and Engineering, Northeast Petroleum University, Daqing 163318, China

a r t i c l e

i n f o

Article history: Received 2 June 2011 Received in revised form 8 September 2011 Accepted 9 September 2011 Available online 16 September 2011 Keywords: La3+ -doped magnetite Adsorption Desorption Wastewater treatment

a b s t r a c t This investigation was to increase the adsorption capacity of magnetite for Congo red (CR) by adulterating a small quantity of La3+ ions into it. The adsorption capability of nanocrystalline Fe3−x Lax O4 (x = 0, 0.01, 0.05, 0.10) ferrite to remove CR from aqueous solution was evaluated carefully. Compared with undoped magnetite, the adsorption values were increased from 37.4 to 79.1 mg g−1 . The experimental results prove that it is effectual to increase the adsorption capacity of magnetite by doped La3+ ions. Among the La3+ -doped magnetite, Fe2.95 La0.05 O4 nanoparticles exhibit the highest saturation magnetization and the maximum adsorption capability. The desorption ability of La3+ -doped magnetite nanoparticles loaded by CR can reach 92% after the treatment of acetone. Furthermore, the Fe3−x Lax O4 nanoparticles exhibited a clearly ferromagnetic behavior under applied magnetic field, which allowed their high-efficient magnetic separation from wastewater. It is found that high magnetism facilitates to improve their adsorption capacity for the similar products. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Dyes and pigments are widely used as the coloring agents. Colored organic effluent is produced in industries such as textiles, paper, plastics, leather, food and cosmetic, etc. The total dye consumption in textile industry worldwide is more than 10,000 tonnes/year and approximately 100 tonnes of dyes were discharged into waste streams by the textile industry every year [1]. It was reported that nearly 40,000 dyes and pigments are listed, which consist of more than 7000 different chemical structures [2]. Such colored effluent can affect photosynthetic processes of aquatic plants, reducing oxygen levels in water and, in severe cases, resulting in the suffocation of aquatic flora and fauna [3]. Dye effluents are the pollutants that contain chemicals that exhibit toxic effect towards microbial populations and can be toxic and carcinogenic to organisms and human beings. Congo red (CR) (sodium salt of benzidinediazobis-1naphthylamine-4-sulfonic acid) is metabolized to benzidine, a known human carcinogen and exposure to this dye can cause some allergic responses [4]. The treatment of contaminated CR in wastewater is difficult because the dye is generally present in sodium salt form giving it very good water solubility. Due to their chemical structures, dyes resist fade when exposed to light, water and many chemicals and therefore it was difficult to be

∗ Corresponding author. Tel.: +86 431 85095878; fax: +86 431 85095876. E-mail address: [email protected] (L. Zhao). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.032

decolorized once dyes were released into the aquatic environment. Synthetic dyes are difficult to biodegrade due to their complex aromatic structures, which provide them physico-chemical, thermal and optical stability. Also, the high stability of its structure makes it difficult to biodegrade and photodegrade [5]. To remove dyes and other colored contaminants from wastewaters, several physical, chemical, physico-chemical and biological methods (e.g., adsorption, coagulation-flocculation [6], biodegradation, ion-exchange, chemical oxidation [7], ozonation [8], nanofiltration [9], micellar enhanced ultrafiltration [10] and electrochemical methods have been developed. A number of adsorbents, such as activated carbon [11], orange peel [12], sawdust [13], montmorillonite [14], wheat bran and rice bran [15], and mesoporous Fe2 O3 [16], have been used for the removal of CR from aqueous solutions. But the adsorption capacity of these adsorbents is not large. Adsorbent-grade activated carbon is cost-prohibitive and both regeneration and disposal of the used carbon are often very difficult [17]. Widespread application of some of these adsorbents is restricted due to high cost, difficult disposal and regeneration. One of the new developments for removing dyes from water or wastewater in recent years is to use ferrite as adsorbents [18,19]. However, there are still some practical problems to be solved, such as the incompatible relation between the magnetic properties and the sizes. As far as we know, the decrease of the particle size will increase the surface disorder of nanoparticles. Thus, the surface energy will increase with the decreasing particle sizes. However, the saturation magnetization of magnetic powders is decreased with the decrease of the particle sizes, which is a disadvantage

L. Wang et al. / Journal of Hazardous Materials 196 (2011) 342–349

343

high-resolution transmission electron microscope (HRTEM, JEOL3010, 300 kV) and Energy dispersive X-ray spectrum (EDX, Oxford Instruments INCA Energy TEM 200, 300 kV) were used to characterized the microstructure. The hysteresis loops were measured on a VSM-7300 vibrating sample magnetometer (VSM) (Lakeshore, USA) in room temperature. IR spectra of the samples were characterized using a FTIR spectrophotometer (NEXUS, 670) in KBr pellets. A UV–vis spectrophotometer was used for determination of CR concentration in the solutions.

Scheme 1. Structure of Congo red molecule.

2.4. Adsorption experiments for the magnetic separation after the wastewater treatment. To solve this problem, in this contribution, the high surface activity is trying to be obtained by deforming the crystal structure which can be proved by the change of the lattice constant after substitution. Moreover, we have done an investigation on the relation between the adsorption activity and magnetic properties of Fe3−x Lax O4 nanoparticles. It is found that the higher magnetic properties facilitate to improve the adsorption capacity. 2. Materials and methods

The stock solution of CR (1 g L−1 ) was prepared in deionized water and desired concentrations of the dye were obtained by diluting the same with water. The calibration curve of CR was prepared by measuring the absorbance of different predetermined concentrations of the samples at max = 497 nm using UV–vis spectrophotometer (CR has a maximum absorbency at wavelength 497 nm on a UV–vis spectrophotometer). The amount of adsorbed CR (mg g−1 ) was calculated based on a mass balance equation as given below:

2.1. Adsorbate

qe =

Congo red [CR, chemical formula = C32 H22 N6 Na2 O6 S2 , FW = 696.68, max = 497 nm] is a benzidine-based anionic disazo dye, i.e., a dye with two azo groups. The structure is as illustrated in Scheme 1. An accurately weighed quantity of the dye was dissolved in double-distilled water to prepare stock solution (1 g L−1 ).

where qe is the equilibrium adsorption capacity per gram dry weight of the adsorbent, mg g−1 ; C0 is the initial concentration of CR in the solution, mg dm−3 ; Ce is the final or equilibrium concentration of CR in the solution, mg dm−3 ; V is the volume of the solution, dm3 ; and W is the dry weight of the hydrogel beads, g. Take one adsorption of CR for example. Standard solution with initial concentrations of 30 mg L−1 was prepared. Then, 15 mg of Fe3−x Lax O4 nanoparticles was added to 50 mL of the above solution under stirring. After a specified time, the solid and liquid were separated by magnet and UV–vis adsorption spectra was used to measure the CR concentration in the remaining solutions. A standard curve, which was used to convert absorbance data into concentrations for kinetic and equilibrium studies, was drawn to calculate the concentration of each experiment.

2.2. Synthesis of nanocrystalline Fe3−x Lax O4 ferrite In a typical experiment, FeSO4 ·7H2 O and LaCl3 ·7H2 O were dissolved in 20 mL of ethylene glycol (EG) by intensive stirring, accordingly a homogeneous solution was obtained, and then 1.5 g of NaOH was added to the solution at room temperature with simultaneous vigorous agitation. The mixtures were stirred vigorously for 30 min, and then sealed in a Teflon-lined stainless-steel autoclave and maintained at 200 ◦ C for 8 h. After the completion of the reaction, the solid product was collected by magnetic filtration and washed several times with deionized water and absolute ethanol respectively. The final product was dried in a vacuum oven at 100 ◦ C for 6 h. Black powders were obtained and characterized as Fe3−x Lax O4 (x = 0, 0.01, 0.05, 0.10). Detailed experimental parameters are listed in Table 1 (from S1 to S4). Furthermore, the experimental works were carried out in winter, so the room temperature was lower about 13 ◦ C.

(C0 − Ce ) × V W

3. Results and discussion 3.1. Characterization of Fe3−x Lax O4 Fig. 1 shows the XRD patterns of S1 and S3. All the diffraction peaks in Fig. 1a can be indexed to the face-centered cubic structure of magnetite according to JCPDS card no. 19-0629, and the lattice ˚ The diffraction peaks of S3 show the constant of S1 is 8.40014 A. same structure with S1, and no impurities can be detected from

2.3. Characterization The phases were identified by means of X-ray diffraction (XRD) with a Rigaku D/max 2500pc X-ray diffractometer with Cu K␣ radiation () 1.54156 (Å) at a scan rate of 0.02◦ /1(s), morphologies were characterized by a JEOL JSM-6700F field emission scanning electron microscopy (FESEM) operated at an acceleration voltage of 8.0 kV. Transmission electron microscope (TEM, Philips Tecnai 20, 200 kV), Table 1 Summary of the experimental parameters. Samples

FeSO4 ·7H2 O (g)

S1 (Fe3 O4 ) S2 (Fe2.99 La0.01 O4 ) S3 (Fe2.95 La0.05 O4 ) S4 (Fe2.90 La0.10 O4 )

0.8341 0.8313 0.8202 0.8063

± ± ± ±

0.0002 0.0002 0.0002 0.0002

LaCl3 ·7H2 O (g) 0.0000 0.0037 0.0185 0.0371

± ± ± ±

0.0002 0.0002 0.0002 0.0002

NaOH (g) 1.5000 1.5000 1.5000 1.5000

± ± ± ±

0.0002 0.0002 0.0002 0.0002

(1)

Fig. 1. XRD patterns of: (a) S1 and (b) S3.

344

L. Wang et al. / Journal of Hazardous Materials 196 (2011) 342–349

adsorption capacity (mg g-1 )

80

d c

60

b 40

a

20

0 0

20

40

60

80

100

time (min) Fig. 2. Adsorption capacity of: (a) S1; (b) S2; (c) S4 and (d) S3. (Adsorption conditions for CR: 50 mL of 100 mg L−1 of dye, adsorbent dosage 0.015 g, natural pH, temperature: 13 ◦ C.)

˚ The Fig. 1b. Furthermore, the lattice constant of S3 is 8.39244 A. incorporation of La ions may reside on the boundaries of the magnetite, which make the shortening of Fe–O bond length, therefore, the lattice constant of S3 is smaller than S1. Similar results have been reported in work done by Zhao and El-Bahy [20,21]. The strong and sharp peaks indicate that the S1 and S3 are well crystallized. By comparison of the lattice constants between S1 and S3, we can confirm that the crystal structure of magnetite was deformed little by the doping of La3+ ions. 3.2. Effect of La3+ -doped amount on the adsorption capacity of magnetite After we ascertain the pure phase Fe3−x Lax O4 ferrites, a series adsorption experiments were carried out. The adsorption capacity of S1 to S4 for CR was shown in Fig. 2. Their adsorption values

for CR are 37.4,48.6,79.1 and 63.1 mg g−1 , respectively. An exciting experimental result is obtained that the doping of La3+ ions favors increasing the adsorption capacity of magnetite for CR. Especially, S3 exhibits the maximum adsorption capacity. Furthermore, at the beginning of the contact time about 5 min, a rapid removal of CR was observed. After 90 min, the adsorption for CR almost reaches saturation. In order to study the effect of morphologies or particle sizes of the Fe3−x Lax O4 on the adsorption capacity of CR in the aqueous solution, SEM photos were shown in Fig. 3a. S1 is composed by octahedral nanoparticles with edge length about 10–30 nm. Uniform nanoparticles with particle sizes around 20 nm are observed from Fig. 3b. However, irregular shapes and broad size distribution are appeared with the increasing doped contents of La ions for S3 and S4 (Fig. 3c and d). Their particle sizes are in the range of 60–200 nm and 80–300 nm, respectively. To our best knowledge, the adsorption capacity of nanopowders increased with the decrease of particle sizes (namely the increase of surface areas). However, in this experiment, it is found that the adsorption capacity could be improved by doped La3+ ions, without accompanying the decrease of the particle sizes. Further insight into the nanostructure of samples was gained using TEM and HRTEM. Take S3 as example. Fig. 4a shows the TEM image of S3, which are in agreement with the above SEM findings. The HRTEM image (Fig. 4b) and the corresponding fastFourier-transform (FFT) pattern-which is framed in Fig. 4b with a square-is shown in Fig. 4c, it represents a face-centered cubic diffraction spots pattern. The clear lattice fringes can prove the high crystallinity of the as-prepared S3. Further, the dominantly exposed planes of S3 are {1 1 1}. The lattice spacing between two adjacent fringes we can observe is corresponding to the set of (1 1 1) planes with a lattice space of 0.5 nm. The lattice fringes are parallel throughout, which prove the single-crystalline nature of S3. Besides, EDX analysis (Fig. 4d) exhibited that S3 was essentially composed of Fe, La and O elements, indicating that La ions was introduced into the magnetite.

Fig. 3. SEM images of (a) S1, (b) S2, (c) S3 and (d) S4.

L. Wang et al. / Journal of Hazardous Materials 196 (2011) 342–349

345

Fig. 4. (a) SEM image, (b) HRTEM image, (c) FFT pattern framed in (b), and (d) EDX pattern of S3.

By comparing the S1 with other samples, it is certain that the proper doped La3+ ions can effectively increase the adsorption capacity. Among the S2, S3 and S4, it can be concluded that the concentration of the doped La3+ ions and the particle sizes exhibit a combination influence on the adsorption capacity of magnetite. The proper doped amount and particle sizes lead to the maximum adsorption capacity of magnetite. The ionic radius (r) of La3+ is ˚ rFe 3+ = 0.64 Å). Such much larger than that of Fe3+ (rLa 3+ = 1.06 A, a substitution makes the change of lattice constant. Compared S1 with S3, the lattice constant of S3 is smaller than that of S1. The lattice constant is corresponding to the distortion height of octahedral site ([MeO6 ]) for magnetite with face-center cubic structure. The decrease of lattice constant makes the increase of distortion degree, hence, the magnetite substituted by La3+ ions is more unstable. Meanwhile, the dopant would lead the imperfect coordination and produce surface defects of magnetite. Finally, the unstable state might lead to the increase of surface energy. To decrease the surface energy of magnetite, it is prone to the adsorption of CR on its surface. Moreover, the structural mismatch caused by doped La3+ ions should change the surface charge of Fe3 O4 [23], which may also conduce to the adsorption of CR on the surface based on the electrostatic adherence principle. 3.4. Effect of initial dye concentrations and contact time on adsorption In order to know the effect of initial dye concentration and contact time on the removal of CR, four different concentrations (0.030, 0.050, 0.080, and 0.100 g L−1 ) are selected to investigate the adsorption of CR on the surface of Fe2.95 La0.05 O4 . With the increase of initial CR concentrations from 0.030 to 0.100 g L−1 , the amount of

CR removal was increased from 29.2 to 79.11 mg g−1 as shown in Fig. 5. Very rapid adsorption is observed at previous 2–5 min, and thereafter a gradual increase occurs with increasing contact time up to 20–30 min depending on the initial dye concentration. Then, the adsorption keeps a weak increase during the following time. Therefore, the adsorption equilibrium almost happen at 40 min. Similar results have been reported for the adsorption of CR on calcium-rich fly ash in work done by Acemio˘glu [24]. 3.5. Adsorption isotherms Analysis of adsorption isotherm is of fundamental importance to describe how adsorbate molecules interact with the adsorbent surface. To simulate the adsorption isotherm, two commonly used 80

adsorption capacity (mg g-1 )

3.3. Adsorption mechanism

60

40

100mg/l 80mg/l 50mg/l 30mg/l

20

0

0

20

40

60

80

100

time (min) Fig. 5. Effect of initial dye concentration on CR removal from S3. (Conditions: 50 mL of CR, adsorbent dosage 0.015 g, natural pH, temperature: 13 ◦ C.)

346

L. Wang et al. / Journal of Hazardous Materials 196 (2011) 342–349

Table 2 Adsorption parameter obtained from adsorption isotherm for S3. Freundich

Langmuir

KF

n

rF 2

qmax (mg g−1 )

KL

rL 2

RL

9.8084 ± 0.0002

2.14 ± 0.01

0.9898 ± 0.0002

107.64 ± 0.01

0.0279 ± 0.002

0.9976 ± 0.0001

0.5444 0.4175 0.3094 0.2639

models, the Freundlich [25] and Langmuir [26] isotherms, were selected to explicate dye–ferrite interaction. The Freundlich adsorption isotherm can be expressed as: log qe = log KF +

1 log Ce n

where KF and n are the Freundlich adsorption isotherm constants, being indicative of the extent of the adsorption and the degree of nonlinearity between solution concentration and adsorption, respectively. KF and 1/n values can be calculated from intercept and slope of the linear plot between log Ce and log qe . The Langmuir isotherm is expressed as:

log(q1 − qt ) = log q1e −

where KL and C0 are the same as defined before. The value of RL calculated from the above expression. The nature of the adsorption process to be either unfavorable (RL > 1), linear (RL = 1), favorable (0 < RL < 1) or irreversible (RL = 0). The Freundlich isotherm was employed to describe heterogeneous systems and reversible adsorption, which does not restrict to the monolayer formations. Unlike the Freundlich isotherm, the Langmuir isotherm is based on the assumption that a structure of adsorbent is homogeneous, where all sorption sites are identical and energetically equivalent. Fig. 6 represents the plot of the experimental data based on Freundlich and Langmuir isotherms model, respectively. Table 2 shows the calculated values of Freundlich and Langmuir model’s parameters. The comparison of correlation coefficients (r2 ) of the linearized

(7)

b y=0.3334x+0.0093

0.025

-1

1/qe (g mg )

logqe(mgg-1)

(6)

The h, q2e and K2 can be obtained by linear plot of t/qt versus t. Fig. 7 is the plots of the pseudo-first order and second order kinetics of CR adsorption on S3. The calculated kinetic parameters are given in Table 3. The correlation coefficient for the pseudo-first-order model is relatively lower (r1 2 = 0.8527), the calculated qe value (q1e ) obtained

2

r =0.9976

0.020

1.6

0.015

1.5 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0

0.010 0.01

logCe(mgL-1)

(5)

h = K2 q2e 2

0.030

1.7

0.1 0.1 0.1 0.1

where K2 is the pseudo-second order rate constant (g mg−1 min−1 ). The initial adsorption rate, h (mg g−1 min−1 ) at t → 0 is defined as:

a

1.8

K1 t 2.303

t 1 t = + qt q2e K2 q2e 2

2.0 1.9

± ± ± ±

where qt is the amount of dye adsorbed per unit of adsorbent (mg g−1 ) at time t, K1 is the pseudo-first order rate constant (min−1 ). The adsorption rate constant (K1 ) were calculated from the plot of log(q1e − qt ) against t. Ho and McKay [28] presented the pseudo-second order kinetic as:

(4)

y=0.4668x+0.9916 r2=0.9898

30.0 50.0 80.0 100.0

The adsorption kinetic models were applied to interpret the experimental data to determine the controlling mechanism of dye adsorptions from aqueous solution. Here, Pseudo-first-order, pseudo-second-order and the intraparticle diffusion model were used to test dynamical experimental data. The pseudo-first order kinetic model of Lagergren [27] is given by:

(3)

1 (1 + KL C0 )

0.0002 0.0002 0.0002 0.0002

3.6. Adsorption kinetics

where qmax is the maximum amount of adsorption with complete monolayer coverage on the adsorbent surface (mg g−1 ), and KL is the Langmuir constant related to the energy of adsorption (L mg−1 ). The Langmuir constants KL and qmax can be determined from the linear plot of 1/Ce versus 1/qe . The essential characteristics of Langmuir isotherm can be expressed by a dimensionless constant called equilibrium parameter RL that is defined by the following equation: RL =

± ± ± ±

form of both equations indicates that the Langmuir model yields a better fit for the experimental equilibrium adsorption data than the Freundlich model. This suggests the monolayer coverage of the surface of S3 by CR molecules. The maximum adsorption capacity (qmax ) of the S3 beads for CR was 107.64 mg g−1 (Table 2). Here, RL -values obtained are listed in Table 2. All the RL -values for the adsorption of CR onto S3 are in the range of 0.5444–0.2639, indicating that the adsorption process is favorable.

(2)

1 1 1 1 = + qe qmax KL qmax Ce

C0 (mg L−1 )

0.02

0.03

0.04

0.05

0.06

-1

1/ce (L mg )

Fig. 6. Adsorption isotherms for adsorption of CR on S3 (15 mg of adsorbent) (a) Freundlich and (b) Langmuir.

L. Wang et al. / Journal of Hazardous Materials 196 (2011) 342–349

347

Fig. 7. Adsorption kinetic for adsorption of CR on S3 (15 mg of adsorbent, initial dye concentration 100 mg L−1 , natural pH, test-temperature: 13 ◦ C) (a) pseudo-first order and (b) pseudo-second order. Inset in (b) in turn is CR solution, mixing with the magnetic absorbents and separation of the adsorbent from solution with a magnet after reaction (2) and 30 min, respectively.

Table 3 Adsorption parameters obtained from Fig. 5.

79.11 ± 0.01

Pseudo-first-order

Pseudo-second-order

K1 (min−1 )

q1e (mg g−1 )

r1 2

K2 (g mg−1 min−1 )

q2e (mg g−1 )

h (mg g−1 min−1 )

r2 2

0.0246 ± 0.0002

12.22 ± 0.01

0.8527 ± 0.0002

0.0085 ± 0.0002

79.43 ± 0.01

53.62 ± 0.01

0.9994 ± 0.0001

from this equation does not give reasonable value (Table 3), which is much lower than experimental data (qe,exp ). This result suggests that the adsorption process does not follow the pseudo-first-order kinetic model, which is similar to the result reported for adsorption of CR onto Australian clay materials [29]. On the contrary, the results present an ideal fit to the second order kinetic for adsorbent with the extremely high r2 2 = 0.9994 (Fig. 7b). A good agreement with this adsorption model is confirmed by the similar values of calculated q2e and the experimental ones for adsorbent. The best fit to the pseudo-second order kinetics indicates that the adsorption mechanism depends on the adsorbate and adsorbent. CR is an acidic dye with negative charge because of the existence of sulphonated group (–SO3 − Na+ ). Here, the higher adsorption capacity of the CR for S3 is probably because of the dopant of La ions, which may increase the surface positive charges of magnetite, so we speculate that an electrostatic attraction may the main adsorption mechanism. Inset in Fig. 7b represented the photograph of adsorption and magnetic separation behavior. A light pink solution was observed after 2 min of adsorption. Further prolonging the adsorption time to 30 min, a colorless solution was gained. More importantly, simple and rapid separation of CR-loaded magnetite adsorbent from treated water can be achieved via an external magnetic field.

at 1050 cm−1 corresponding to C–N bond only appears in Fig. 9b, which discloses that CR was loaded on the surface of S3. This also serves as another evidence of physical adsorption because of in a physical adsorption at the mineral–water interface; an oxyanion 100 80

Desorption(%)

qe,exp (mg g−1 )

60 40 20 0 0

20

40

60

80

100

time(min) Fig. 8. Desorption ratio of loaded magnetite nanoparticles with time.

3.7. Desorption Desorption is also a key role for the practical application of magnetic powders to water treatment. A facile desorption method and high-efficient desorption can facilitate to reduce cost, because the spent adsorbent and the CR can obtain recycling chance. Desorption process was conducted by mixing 5 mg of CR-loaded modified S3 with 30 mL of acetone solutions and shaking for different time. Fig. 8 is the desorption percentage with the time. The desorption efficiency calculated as Eq. (8) was 92%. Therefore, the CR could be desorbed from the loaded nanoparticles by acetone solutions. Desorption ratio (%) =

Amount of desorbed CR × 100 Amount of adsorbed CR

(8)

FT-IR analysis was also performed to reveal the surface nature of S3, as shown in Fig. 9. The spectra display a broad band at 580 cm−1 , which is believed to be associated with the stretching vibrations of the tetrahedral groups (Fe3+ –O2− ) for S3. However, the band

Fig. 9. FT-IR spectra of (a) as-prepared, (b) CR-absorbed and (c) CR-desorbed S3.

348

L. Wang et al. / Journal of Hazardous Materials 196 (2011) 342–349

Table 4 Adsorption capacities of CR dye on various adsorbents.

S3 [3] [15] [15] [16] [29] [30] [31] [32] [32] [32] [33] [34] [35] [36] [37] [38] Present study

will retain its hydration shell and will not form a direct chemical bond with the oxide surface [22]. Moreover, it is an evident proof that CR is removed sufficiently from the surface of S3 by acetone, because Fig. 9a and c exhibit the same FT-IR spectra. 3.8. Performance evaluation The maximum adsorption capacity (qmax ) for S3 nanoparticles to CR calculated from the Langmuir isotherm model is listed in Table 4 with literature values of qmax of other adsorbents for CR adsorption [3,15,16,29–38]. All of the adsorbents used for CR adsorption have considerably lower qmax values than S3 used in this study, except chitosan hydrogel beads impregnated with carbon nanotubes CS/CNT [3], maghemite nanoparticles [16] and CTAB modified chitosan beads [37,38]. However, the simplicity of the preparation method and magnetic separation of S3 nanoparticles makes them better adsorbent than the others for CR adsorption. 3.9. Magnetic properties The magnetic properties of magnetic absorbents directly influence the callback efficiency. Hence, excellent magnetic performance is also a key role for the magnetic material as magnetic absorbent. Here, the magnetism of S1 to S4 is evaluated. It is exciting to find that the magnetic properties of magnetic absorbents do have influence on their adsorption ability. The room-temperature hysteresis loops of S1 to S4 were shown in Fig. 10. Furthermore, the magnetic parameters of samples obtained from hysteresis loops were listed in Table 5. The test results show that the doped-La3+ ions make the Ms values of magnetite decreased to some extent. It must acknowledge that magnetite still keep the high Ms values after La3+ ions were doped into them, even if the lowest Ms value is 81.4 emu/g for S2. We amazingly find that the adsorption abilities of La3+ -doped magnetite are proportional to their Ms values and independent of their particle sizes. This is an important proof that excellent magnetism facilitates to increase the adsorption capacity for the similar mag-

S4

40

S1

0 Magnetization (emu/g)

Chitosan hydrogel beads impregnated with carbon nanotubesCS/CNT 450.40 Wheat bran 22.73 Rice bran 14.63 208.33 Maghemite nanoparticles Cattail root 38.79 4.43 Sugar cane bagasse 35.70 Jute stick powder Bentonite 19.90 Kaolin 5.60 Zeolite 4.30 66.23 Palm kernel seed coat 7.08 Activated red mud 22.62 Anilinepropylsilica xerogel 71.46 Marine alga 352.50 CTAB modified chitosan beads 373.29 CS/CTAB beads 107.64 Fe2.95 La0.05 O4

S2

80

Reference

Magnetization (emu/g)

qmax (mg g−1 )

Type of adsorbent

-40 -80 -10000

-5000

40 20 0 -20 -40 -200

0

-100 0 100 Field (Oe)

5000

200

10000

Field (Oe) Fig. 10. Magnetization curve measured at room temperature for the Fe3−x Lax O4 ferrites.

netic products. Both the heavy-metal ions and organic matters show feeble paramagnetism or antiferromagnetism, so the magnetic powders which possess high Ms will be beneficial to the adsorption of heavy-metal ions and organic matters. Moreover, the magnetic materials with high Ms are help to the finally magnetic separation. Therefore, it is very meaningful to both keep the magnetic properties almost constant and increase the adsorption capacity. 4. Conclusions Fe3−x Lax O4 (x = 0, 0.01, 0.05, 0.10) ferrite nanoparticles were successfully synthesized by a facile one-step solvothermal synthesis. Compared with the pure magnetite, La3+ -doped magnetite exhibit more excellent adsorption ability. Furthermore, among the La3+ -doped products, the sample (Fe2.95 La0.05 O4 ) possessing the biggest Ms value owes the strongest adsorption capacity. The adsorption capacity of magnetite for CR is improved not by increasing the specific area but by deforming the crystal structure via doped- La3+ ions. By comparison with many other adsorbents, Fe3−x Lax O4 nanoparticles have higher adsorption capacities for CR. It would be a good method to increase adsorption efficiency of magnetite for the CR removal in a wastewater treatment process by doping La3+ ions. Analysis of adsorption isotherm shows that our adsorption experiment accord with Langmuir model. Again, adsorption kinetic model indicates that the adsorption mechanism depends on the adsorbate and adsorbent. In a word, the Fe3−x Lax O4 nanoparticles were a kind of excellent absorbent because of their high adsorption, desorption and recovery efficiency. Acknowledgements The financial supports from the Natural Science Foundation of Jilin Province (20101542) of China and the National Foundation of Doctoral Station (grant No. 20100061110019) are acknowledged. References

Table 5 Magnetic parameters obtained from hysteresis loops. Samples

Ms (emu/g)

S1 S2 S3 S4

90.5 81.4 86.2 82.2

± ± ± ±

0.1 0.1 0.1 0.1

Mr (emu/g)

Hc (Oe)

± ± ± ±

156.6 116.5 115.9 105.9

15.6 11.2 14.8 16.4

0.1 0.1 0.1 0.1

± ± ± ±

0.1 0.1 0.1 0.1

[1] S.A. Saad, K.Md. Isa, R. Bahari, Chemically modified sugarcane bagasse as a potentially low-cost biosorbent for dye removal, Desalination 264 (2010) 123–128. [2] A. Demirbas, Agricultural based activated carbon for the removal of dyes from aqueous solutions: a review, J. Hazard. Mater. 167 (2009) 1–9. [3] S. Chatterjee, D.S. Lee, M.W. Lee, S.H. Wooa, Congo red adsorption from aqueous solutions by using chitosan hydrogel beads impregnated with nonionic or anionic surfactant, Bioresour. Technol. 101 (2010) 1800–1806.

L. Wang et al. / Journal of Hazardous Materials 196 (2011) 342–349 [4] S. Chatterjee, M.W. Lee, S.H. Wooa, Adsorption of Congo red by chitosan hydrogel beads impregnated with carbon nanotubes, Bioresour. Technol. 100 (2009) 3862–3868. [5] I.D. Mall, V.C. Srivastava, N.K. Agarwal, I.M. Mishra, Removal of Congo red from aqueous solution by bagasse fly ash and activated carbon: kinetic study and equilibrium isotherm analyses, Chemosphere 61 (2005) 492–501. [6] M.L. Marechal, Y.M. Slokar, T. Taufer, Decolouration of chlorotriazine reactive azo dyes with H2 O2 /UV, Dyes Pigments 33 (1997) 181–298. [7] M. Pollock, Neutralizing dye-housewastes with flue gases and decolorizing with fly ash, Am. Dyestuff Rep. 62 (1973) 21–23. [8] G. Reynolds, N. Graham, R. Perry, R.G. Rice, Aqueous ozonation of pesticides: a review, Ozone Sci. Eng. 11 (1989) 339–382. [9] S. Chakraborty, M.K. Purkait, S. Dasgupta, S. De, J.K. Basu, Nanofiltration of textile plant effluent for color removal and reduction in COD, Sep. Purif. Technol. 31 (2003) 141–151. [10] M.K. Purkait, S. DasGupta, S. De, Removal of dye from wastewater using micellar-enhanced ultrafiltration and recovery of surfactant, Sep. Purif. Technol. 37 (2004) 81–92. [11] M.K. Purkait, A. Maiti, S. DasGupta, S. De, Removal of Congo red using activated carbon and its regeneration, J. Hazard. Mater. 145 (2007) 287–295. [12] M. Arami, N.Y. Limaee, N.M. Mahmoodi, N.S. Tabrizi, Removal of dyes from colored textile wastewater by orange peel adsorbent: equilibrium and kinetic studies, J. Colloid Interface Sci. 288 (2005) 371–376. [13] R. Jain, S. Sikarwar, Removal of hazardous dye Congo red from waste material, J. Hazard. Mater. 152 (2008) 942–948. [14] Z. Yermiyahu, I. Lapides, S. Yariv, Visible absorption spectroscopy study of the adsorption of Congo red by montmorillonite, Clay Miner. 38 (2003) 483–500. [15] X.S. Wang, J.P. Chen, Biosorption of Congo red from aqueous solution using Wheat Bran and Rice Bran: batch studies, Sep. Sci. Technol. 44 (2009) 1452–1466. [16] A. Afkhami, R. Moosavi, Adsorptive removal of Congo red, a carcinogenic textile dye, from aqueous solutions by maghemite nanoparticles, J. Hazard. Mater. 174 (2010) 398–403. [17] X.S. Wang, J.P. Chen, Removal of the azo dye Congo red from aqueous solutions by the Marine Alga Porphyra yezoensis Ueda, Clean 37 (2009) 793–798. [18] X.Y. Hou, J. Feng, X.H. Liu, Y.M. Ren, Z.J. Fan, M.L. Zhang, Magnetic and high rate adsorption properties of porous Mn1−x Znx Fe2 O4 (0 ≤ x≤ 0.8) adsorbents, J. Colloid Interface Sci. 353 (2011) 524–529. [19] S.X. Zhang, H.Y. Niu, Y.Q. Cai, X. Zhao, Y.L. Shi, Arsenite and arsenate adsorption on coprecipitated bimetal oxide magnetic nanomaterials: MnFe2 O4 and CoFe2 O4 , Chem. Eng. J. 158 (2010) 599–607. [20] L.J. Zhao, H. Yang, L.X. Yu, W. Sun, Y.M. Cui, Y. Yan, S.H. Feng, Structure and magnetic properties of Ni0.7 Mn0.3 Fe2 O4 nanoparticles doped with La2 O3 , Phys. Status Solidi 201 (2004) 3121–3128.

349

[21] Z.M. El-Bahy, A.A. Ismaila, R.M. Mohamed, Enhancement of titania by doping rare earth for photodegradation of organic dye (Direct Blue), J. Hazard. Mater. 166 (2009) 138–143. [22] F.A. Petit, H. Debontride, M. Lenglet, G. Juhel, D. Verchere, Contribution of spectrometric methods to the study of the constituents of chromating layers, Appl. Spectrosc. 49 (1995) 207–210. [23] Y.N. Shen, H.L. Zhao, X.T. Liu, N.S. Xu, Preparation and electrical properties of Ca-doped La2 NiO4+ı cathode materials for IT-SOFC, Phys. Chem. Chem. Phys. 12 (2010) 15124–15131. [24] B. Acemio˘glu, Adsorption of Congo red from aqueous solution onto calcium-rich fly ash, J. Colloid Interface Sci. 274 (2004) 371–379. [25] H.M.F. Freundlich, Uber die adsorption in losungen, Z. Phys. Chem. 57 (1906) 385–470. [26] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and platinum, J. Am. Chem. Soc. 40 (1918) 1361–1403. [27] S. Lagergren, About the theory of so-called adsorption of soluble substances, Kungliga Svenska Vetenskapsakademiens, Handlingar 24 (1898) 1–39. [28] Y.S. Ho, G. McKay, Pseudo-second order model for sorption processes, Process Biochem. 34 (1999) 451–465. [29] V. Vimonsesa, S. Lei, B. Jin, C.W.K. Chow, C. Saint, Kinetic study and equilibrium isotherm analysis of Congo red adsorption by clay materials, Chem. Eng. J. 148 (2009) 354–364. [30] N.A. Oladoja, A.K. Akinlabi, Congo red biosoption on palm kernel seed coat, Ind. Eng. Chem. Res. 48 (2009) 6188–6196. [31] A.S. Raymundo, R. Zanarotto, M. Belisário, M.G. Pereira, J.N. Ribeiro1, A.V.F.N. Ribeiro, Evaluation of sugar-cane bagasse as bioadsorbent in the textile wastewater treatment contaminated with carcinogenic Congo red dye, Braz. Arch. Biol. Technol. 53 (2010) 931–938. [32] G.C. Panda1, S.K. Das1, A.K. Guha, Jute stick powder as a potential biomass for the removal of Congo red and rhodamine B from their aqueous solution, J. Hazard. Mater. 164 (2009) 374–379. [33] T. Tuutijarvi, J. Lu, M. Sillanpaa, G. Chen, As (V) adsorption on maghemite nanoparticles, J. Hazard. Mater. 166 (2009) 1415–1420. [34] Z. Hu, H. Chen, F. Ji, S. Yuan, Removal of Congo red from aqueous solution by cattail root, J. Hazard. Mater. 173 (2010) 292–297. [35] A. Tor, Y. Cengeloglu, Removal of Congo red from aqueous solution by adsorption onto acid activated red mud, J. Hazard. Mater. 138 (2006) 409–415. [36] L. Wang, A.Q. Wang, Adsorption characteristics of Congo red onto the chitosan/montmorillonite nanocomposite, J. Hazard. Mater. 147 (2007) 979–985. [37] L. Wang, A.Q. Wang, Removal of Congo red from aqueous solution using a chitosan/organomontmorillonite nanocomposite, J. Chem. Technol. Biotechnol. 82 (2007) 711–720. [38] S. Chatterjee, M.W. Lee, S.H. Woo, Influence of impregnation of chitosan beads with cetyl trimethyl ammonium bromide on their structure and adsorption of Congo red from aqueous solutions, Chem. Eng. J. 155 (2009) 254–259.

Journal of Hazardous Materials 196 (2011) 350–359

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Organo/layered double hydroxide nanohybrids used to remove non ionic pesticides D. Chaara a,1 , F. Bruna b , M.A. Ulibarri a , K. Draoui c , C. Barriga a,∗ , I. Pavlovic a a Dpto de Química Inorgánica e Ingeniería Química, Instituto Universitario de Química Fina y Nanoquímica (IUQFN),Universidad de Córdoba, Campus de Rabanales, Campus de Excelencia Internacional Agroalimentario, Ceia3, Edificio Marie Curie, 14071 Córdoba, Spain b Instituto de Recursos Naturales y Agrobiología de Sevilla (IRNAS), CSIC, Avenida Reina Mercedes 10, Apartado 1052, 41080, Sevilla, Spain c Department de Chimie, Laboratoire LPCIE, Faculté des Sciences, BP 2121, Tetouan, Morocco

a r t i c l e

i n f o

Article history: Received 15 July 2011 Received in revised form 7 September 2011 Accepted 9 September 2011 Available online 16 September 2011 Keywords: Organo/layered double hydroxide Nanohybrid Pesticide Adsorption Controlled release

a b s t r a c t The preparation and characterization of organo/layered double hydroxide nanohybrids with dodecylsulfate and sebacate as interlayer anion were studied in detail. The aim of the modification of the layered double hydroxides (LDHs) was to change the hydrophilic character of the interlayer to hydrophobic to improve the ability of the nanohybrids to adsorb non-ionic pesticides such as alachlor and metolachlor from water. Adsorption tests were conducted on organo/LDHs using variable pH values, contact times and initial pesticide concentrations (adsorption isotherms) in order to identify the optimum conditions for the intended purpose. Adsorbents and adsorption products were characterized several physicochemical techniques. The adsorption test showed that a noticeable increase of the adsorption of the non-ionic herbicides was produced. Based on the results, the organo/LDHs could be good adsorbents to remove alachlor and metolachlor from water. Different organo/LDHs complexes were prepared by a mechanical mixture and by adsorption. The results show that HTSEB-based complex displays controlled release properties that reduce metolachlor leaching in soil columns compared to a technical product and the other formulations. The release was dependent on the nature of the adsorbent used to prepare the complexes. Thus, it can be concluded that organo/LDHs might act as suitable supports for the design of pesticide slow release formulations with the aim of reducing the adverse effects derived from rapid transport losses of the chemical once applied to soils. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Agricultural pesticides are often detected in natural waters, and this has raised concerns regarding the protection of health and the environment. They are an important group of organic pollutants which production and uses are still increasing but must be controlled to minimize contamination problems [1,2] The research of new adsorbent is a strategy to remediate the contamination of water produced by an increase in the use of pesticides to improve agriculture production. Layered double hydroxides (LDHs), also known as hydrotalcitelike compounds, consist of brucite-like layers which contain the hydroxides of divalent (MII ) and trivalent (MIII ) metal ions and have an overall positive charge balanced by hydrated anion

∗ Corresponding author. Tel.: +34 957 218648; fax: +34 957 218621. E-mail address: [email protected] (C. Barriga). 1 Permanent address: Department de Chimie, Laboratoire LPCIE, Faculté des Sciences, BP 2121, Tetouan, Morocco. 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.034

between layers. These compounds present a general formula [MII (1−x) MIII x (OH)2 ]x+ (An− )x/n ·mH2 O where An− is the intercalated anion. Because of the strong hydration of these inorganic ions the interlayer spaces have hydrophilic nature they resemble clay minerals. As a result, the natural clay minerals as well as LDHs show rather weak affinity to most of the non ionic organic compounds and are seldom by use as sorbents for organic compounds [3,4] Under suitable conditions, the inorganic ions on clay minerals and layered double hydroxides can be replaced by organic ions which make the interlayer spaces become hydrophobic, [5–9]. Therefore, the adsorption ability of clay minerals and LDHs for organic pollutants can be significantly improved by modifying the interlayer. These organoclays and organo/LDHs have applications in a wide range of organic pollution control fields [10–14] To date, there are few studies of LDHs modified with organic anions and in particular of their adsorption properties. The number of publications of organoclay (more than 2000) is much higher than organo/LDH (approx. 50) over the last 10 years (source: Citation Report of Web of Science in June 2011). However, recently the scientific interest in organo/LDH is increasing. This gives rise

D. Chaara et al. / Journal of Hazardous Materials 196 (2011) 350–359

to an emerging research field of the hydrotalcites since the results obtained regarding its application as adsorbents of pesticides are very promising. The organo/LDHs with suitable anion can provide interlayer spaces between 2 and 4 nm and might be considered as nanohybrids with hydrophobic characteristics in the interlayer space and external surface. The aim of this work was to prepare different nanohybrid intercalated hydrotalcites with two organic anions: dodecylsulfate (DDS) and sebacate (SEB) and with different ratio Mg/Al = 3 and 2, to increase the charge density of the layers and consequently to increase the content of interlayer sebacate anions. These nanohybrids were used to assess the removal of alachlor and metolachlor from water, two widely used herbicides with hydrophobic characteristic. Additionally, the metolachlor has been selected to explore the release behavior from the formulation with organo/LDH nanohybrids. 2. Materials and methods 2.1. Organic anions, pesticides and soil The organic anions used for the preparation of the organo/LDH nanohybrids containing DDS and SEB, were supplied as soluble sodium salt and acid form, respectively, by Sigma–Aldrich. The log Kow values are 5.4 and 1.86 for dodecylsulfate and sebacate acid respectively. (2-chloro-2 ,6 -diethyl-N-(methoxymethyl)) Alachlor and metolachlor (2-chloro-N-(6-ethyl-o-tolyl)-N-[(1RS)-2-methoxy-1methylethyl]acetamide) are selective pre-emergence herbicides which belong to aniline herbicides. Analytical standard alachlor and metolachlor was purchased from Sigma–Aldrich. The water solubility values of alachlor and metolachlor at 25 ◦ C are 0.110 g/L and 0.120 g/L and their log Kow values are 2.9 and 3.45 respectively (data obtained from Scifinder Scholar). The molecular structures of the herbicides and organic anions used are shown in Fig. 1. The soil used in the leaching experiments was fluvisol from the terraces of the Guadalquivir River, Córdoba (Spain). The soil was sampled (0–20 cm), air-dried, and sieved (2 mm) prior to use. It had 660 mg/kg sand, 150 mg/kg silt, 190 mg/kg clay, and 3.6 mg/kg organic matter. Soil pH was 8.7 in a 1:2 (w:w) soil:deionized water mixture.

351

for alachlor and metolachlor respectively. The amount of herbicide adsorbed (Cs ) was calculated from the difference between the initial (C0 ) and equilibrium (Ce ) solution concentrations. Desorption was realized immediately after adsorption from the highest equilibrium point of the adsorption isotherm and was repeated three times. Adsorption–desorption data were fitted to the Langmuir equation: Ce = Cs

C   1  e Cm

+

Cm L

(1)

and the logarithmic form of Freundlich equation: logCs = logKf + nf logCe

(2)

where Cm is the maximum adsorption capacity at the monolayer coverage (mmol/g), L (L/mmol) is a constant related to the adsorption energy and Kf (mmol 1−nf Lnf g−1 ) and nf are the Freundlich constants. 2.4. Characterization of the adsorbents and adsorption products The adsorbents HTSDS1, HTSDS2 and HTSEB and the adsorption products were characterized by different physical chemical techniques. Powder X-ray diffraction (PXRD) patterns were recorded on powder samples at room temperature under air conditions, using a Siemens D-5000 instrument with Cu K␣ radiation. FT-IR spectra were recorded by using the KBr disk method on a Perkin Elmer Spectrum One spectrophotometer and the ATR-FT-IR method was used for alachlor. Elemental chemical analyses for Mg and Al were carried out by atomic absorption spectrometry on a Perkin Elmer AA-3100 instrument. DDS and SEB amounts were calculated from elemental analysis of S and C respectively, carried out on Elemental Analyse Eurovector EA 3000 instrument. The interlayer water amount was obtained from TG-curves recorded on a Setaram Setsys Evolution 16/18 apparatus, in air at the heating rate of 5 ◦ C/min. Scanning electron microscopy (SEM) micrographs were obtained using a JEOL JSM 6300 instrument; the samples were prepared by deposition of a drop of sample suspension on a Cu sample holder and covered with an Au layer by sputtering in a Baltec SCD005 apparatus.

2.2. Synthesis of the organo/LDH nanohybrids

2.5. Preparation of organo/LDH– metolachlor complexes

The organo/LDHs containing DDS and SEB anions and Mg/Al = 3 and 2 respectively, were obtained by the coprecipitation method [15] using N2 atmosphere and CO2 -free water. The samples were figured as HTDDS1 and HTSEB. For comparison purposes another organo/LDH with DDS was obtained under the same experimental conditions but without washing the precipitate. After removing the supernatant the tube containing the solid was dried. This organo/LDH was named HTDDS2. A carbonate–Mg/Al hydrotalcite (HTCO3 ) was also prepared by the coprecipitation method [16] for the same purpose.

Four complexes were prepared with HTDDS1 and HTSEB and metolachlor. Two of them were based on the adsorption isotherms, figured as HTDDS1–MetoAds and HTSEB–MetoAds , and loaded with a 3% of the herbicides. The other two were prepared by mechanical mixture of the components by soft grinding of the adsorbents and the herbicides (in the same proportion as the adsorption complexes) dissolved in acetone and then were let the solvent evaporate. These figured as HTDDS1–MetoM and HTSEB–MetoM . 2.6. Bath release experiments

2.3. Adsorption and desorption experiments Alachlor and metolachlor adsorption isotherms on HTDDS1, HTDDS2 and HTSEB were obtained by the batch equilibration procedure. Triplicate 20 mg adsorbent samples were equilibrated through shaking for 24 h at room temperature with 30 mL of herbicide solutions with initial herbicide concentrations (C0 ) ranging between 0.1 and 0.35 mmol/L. After equilibration, the supernatants were centrifuged and separated to determine the concentration of herbicides by UV–visible spectrophotometry at 265 nm and 220 nm

The release of metolachlor into water from the organo/LDH–herbicide complexes was compared with the release of the herbicides as a free (technical) product. For this purpose, 0.18 mg of metolachlor as organo/LDHs–herbicide complexes or as a technical product was added to 500 mL of distilled water. The experiments were conducted as described by Bruna et al. [17]. The herbicide concentration was determined by HPLC using a Waters 1525 chromatograph coupled to a Waters 2996 diode-array detector and UV detection at 220 nm for metolachlor.

352

D. Chaara et al. / Journal of Hazardous Materials 196 (2011) 350–359

Fig. 1. The molecular structure of organic anions and herbicides.

2.7. Soil-column leaching experiments Leaching experiments were conducted as described by Bruna et al. [17]. The calculated pore volume of the columns after saturation was 68 ± 2 mL. The leaching experiment was conducted in triplicate and the concentrations of metolachlor were analysed by HPLC. 3. Results and discussion

spacing d(0 0 3) for the organo/LDH adsorbents are also included in Table 1. As expected, the positions of the basal reflections of all the prepared organo/LDHs are shifted to smaller 2 reflection angles regarding carbonate hydrotalcite, which reveals expansion in the interlayer distances, moreover several harmonic are also observed indicating well ordered structure. The van der Waals end-to-end length of dodecylsulfate anion estimated was 2.08 nm taking into account that the LDH layer thickness is 0.48 nm so the dodecylsulfate chains can fit perfectly in the perpendicular direction and

3.1. Characterization of organo/LDH nanohybrid sorbents 3.1.1. Elemental analysis The results of elemental analysis for the nanohybrides used as adsorbents are shown in Table 1 together with other characteristics. The organic anion was determined from the ratio S/Al for the samples HTDDS1 and HTDDS2, and from C/Al for the HTSEB sample. The content of S indicates that the layer charge in HTDDS1 is not balanced only by the organic anion. The exchange percentage in the product was 92% based upon anion exchange capacity (AEC). However, in HTDDS2 it was 100% and the amount of S was higher than required to compensate the layer charge suggesting the precipitation of sodium dodecylsulfate, as it is indicated in Table 1. The elemental analysis of C for the HTSEB sample indicated a small excess of sebacate to compensate the layer charge which could be considered as precipitated salt as has been shown in Table 1. The proposed formulae were obtained from the elemental analysis, assuming that all the positive charge is compensated by the maximum amount possible of dodecylsulfate and sebacate anions for HTDDS and HTSEB respectively. The amount of water was attained from TG data (not included) and metal content analysis. 3.1.2. X-ray diffraction The PXRD patterns of the organo/LDHs included in Fig. 2 together with HTCO3 show that they are typical hydrotalcitelike compounds. The reflections for HTCO3 were indexed on the basis of a hexagonal unit cell with a and c parameters 0.304 nm and 2.34 nm respectively. The corresponding values of the basal

Fig. 2. PXRD patterns for the samples HTDDS1, HTDDS2, HTSEB and HTCO3 .

D. Chaara et al. / Journal of Hazardous Materials 196 (2011) 350–359

353

Table 1 Chemical composition of adsorbents, structural data and proposed formulae. Sample

HTDDS1 HTDDS2 HTSEB HTCO3

wt.%

Atomic ratio

Mg

Al

N

S

C

Mg/Al

S/Al

C/Al

11.1 9.0 13.5

3.9 3.3 18.1

– 0.35 – –

4.2 5.9 –‘ –

27.1 31.3 18.9

3.2 3.1 1.9 2.7

0.9 1.5 – –

15.7 21.7 5.3 –

d0 0 3 (nm)

d1 1 0 (nm)

Proposed formulae

2.58 3.84 1.48 0.78

0.152 0.152 0.150 0.152

[Mg0.76 Al0.24 (OH)2 ](C12 H25 SO4 )0.22 (CO3 )0.01 ·0.83H2 O [Mg0.76 Al0.24 (OH)2 ](C12 H25 SO4 )0.24 ·0.79H2 Oa [Mg0.65 Al0.35 (OH)2 ](C10 H16 O4 )0.175 ·1.36H2 Ob [Mg0.73 Al0.27 (OH)2 ](CO3 )0.135 ·0.64H2 O

The excess of S corresponds to a 0.122NaC12 H25 SO4 per FW and the wt.% of N to 0.05NaNO3 . b The same for the sample HTSEB.

the chains could be in an all-trans conformation [6] in the HTDDS1 sample with a basal spacing of 2.58 nm (Table 1). However, for the HTDDS2 sample a basal spacing of 3.84 nm was observed, similar to those obtained in previous studies [18,19], which could suggest that there is a bilayer arrangement with a slant angle ˛ = 60.4◦ between the chain of DDS and the surface of the layer [20,21]. A schematic arrangement of the anions on modified hydrotalcites (nanohybrids) is included in Fig. 3. For the HTSEB sample a basal spacing 1.48 nm was observed. This lead to a gallery height of 1.0 nm so the anions could be considered as monolayer in a vertical position and titled towards the brucite-like layers according to the length of sebacate (≈1.3 nm) [17], the slant angle required is sin−1 (1.0/1.3) = 50.3◦ . The basal spacing was lower than that obtained for an organohydrotalcite with a ratio Mg/Al = 3.2, d0 0 3 = 1.58 nm by Bruna et al. [17]. This can be due to the difference of the layer charge and/or the dried treatment in each case [18]. The value of spacing corresponding to (1 1 0) reflection, included in Table 1, agrees with the metal ratio, which increases slightly as charge density decreases. 3.1.3. FT-IR spectroscopy Fig. 4 shows the FT-IR spectra of the organo/LDH samples (HTDDS1 is not included because it is almost equal to HTDDS2). The data reveal the LDH-like structure of all adsorbents with the corresponding interlayer anions (DDS and SEB) as reported previously [22–25] 3.1.4. Scanning electron microscopy The SEM images, included in Fig. 5, showed that the HTCO3 sample consisted of thin plate-like crystal with a irregular shape and size 1 wt%). It can be suggested that under N2 flow the particles of metallic oxide need higher temperatures to spread up to the carbon surface in the comparison of CO2 , which can react from temperatures moderated to remove hetero atoms (O and H) of the wood. Finally, it can be seen from Table 8, that the higher the temperature of activation or pyrolysis the more basic is the surface of the carbon. It can be inferred from the pHPZC that surface of carbons evolutes from a soft acid surface to a basic one in agreement with the changes in the oxygenated functional groups detected by FTIR and XPS discussed above. In other words, the higher the temperature of preparation the more hydrophobic behavior of carbons is expected. 3.3.2. SEM Fig. 5 shows the SEM images of selected AC prepared under CO2 and N2 flow. SEM images show some white dots that have been

Table 6 Summary of functional carbon groups detected from XPS analysis in the O1s region for the AC prepared under CO2 flow by 1 h. Temperature (◦ C)

Metallic oxides

350 450 600 700 800 900

× × √ √ √ √

Carbonyl √ √ √ √ √ √

Phenol and ethers √ √ √ √ √ √

Quimisorbed oxygen or water × × × × √ √

Table 7 Summary of oxide carbon groups detected from XPS analysis in the O1s region for the AC prepared under N2 flow by 1 h. Temperature (◦ C)

Metallic oxides

Carbonyl

350 450 600 700 800 900

× × × × √ √

× × × √ √ √

Phenol and ethers √ √ √ √ √ √

Quimisorbed oxygen or water × × × × √ √

J. Matos et al. / Journal of Hazardous Materials 196 (2011) 360–369

365

Table 8 BET surface area (SBET ) and pHPZC of carbons prepared from Algarroba (Alg) under CO2 and N2 flow by 1 h and a comparison against those obtained from Apamate (Apa). T (◦ C)

SBET CO2 (m2 /g)

Alg-350 Alg-450 Alg-600 Alg-700 Alg-800 Alg-900 Apa-450a Apa-600a Apa-700a Apa-800a Apa-900a

92 350 870 1038 1167 752 352 426 570 770 548

a

± ± ± ± ± ± ± ± ± ± ±

1 7 17 28 31 20 5 13 14 16 21

pHPZC ACCO2

SBET N2 (m2 /g)

5.9 6.1 7.0 7.8 8.3 8.9 6.3 7.2 8.0 8.5 9.1

34 220 497 527 549 471 31 360 388 519 590

± ± ± ± ± ± ± ± ± ± ±

1 6 1 2 2 12 5 12 13 15 13

Results obtained from the Apamate wood at the same experimental conditions [16].

Fig. 5. SEM images of selected activated carbons. (A) CO2 , 350 ◦ C, (B) N2 , 350 ◦ C, (C) CO2 , 600 ◦ C, (D) N2 , 600 ◦ C, (E) CO2 , 800 ◦ C and (F) N2 , 800 ◦ C.

pHPZC ACN2 5.8 5.9 6.9 7.7 7.8 8.3 6.1 7.1 7.9 8.5 8.9

366

J. Matos et al. / Journal of Hazardous Materials 196 (2011) 360–369

Fig. 6. Adsorption–desorption N2 isotherm of AC prepared at 800 ◦ C by 1 h. (A) Under CO2 flow and (B) under N2 flow.

associated with the inorganic composition of wood. AC particles are in a micrometer scale in agreement with sieving performed. Independently of the gas and activation temperature, AC showed a cellular fibrous morphology. SEM images suggest that the present activated carbons are constituted by an interconnected channel framework in concordance with the fact that the precursor is constituted by a fibrous structure. For example, Fig. 5C indicates that under CO2 flow, even at moderate temperatures (600 ◦ C), an incipient activation occur in spite of this temperature is lower than that commonly considered as the critical temperature for the spontaneous activation under CO2 flow (about 690 ◦ C) [29]. Similar tendencies of an interconnected porous system as a function of temperature have been reported for the case of activated fibers obtained from rayon fibers [40] and for activated carbons obtained from almond shells [41]. 3.3.3. Texture of AC For the materials prepared at temperatures higher than 600 ◦ C, the adsorption–desorption N2 isotherms of carbon prepared under CO2 and N2 flow, showed very similar trends characteristic of a micropore framework as suggest Fig. 6 and figures in supplementary material. Fig. 6 shows the analysis for two selected carbons prepared at 800 ◦ C by 1 h under CO2 and N2 flow, denoted ACCO2 and ACN2 , respectively. Both isotherms correspond to a type I indicating that the framework is mainly composed by micropores. A summary of BET surface areas (SBET ) of AC prepared from Algarroba (hard wood) is compiled in Table 8 and results obtained from Apamate (soft wood) at the same experimental conditions [16] are also shown in Table 8 for comparative purposes. In general, results obtained from Algarroba and Apamate followed very similar trends.

As expected, BET surface area of AC prepared under CO2 flow was higher than under N2 . It must be noted that a maxima in the BET surface area was obtained at 800 ◦ C, both under CO2 and N2 flow. This temperature is the same than that we have found before [16] for the preparation of AC from the sawdust of Apamate [16] and for the activation of carbon foams obtained from the controlled pyrolysis of saccharose under CO2 or N2 flow [30]. Porosimetry parameters such as micropore area (porearea ), micropore volume (porevolume ), total volume or pore (Vtot ) and pore diameter (Wpore ) are showed in Tables 9 and 10 for the AC obtained under CO2 and N2 flow, respectively. It can be seen that the higher the activation temperature the higher the micropores volume (porevolume ) and the higher the total volume of pore of AC. For the micropore area (porearea ), a maximum is reached at 800 ◦ C in agreement with the BET surface area. In most of cases, the microporous area contributes with about 90% of the total surface area. In addition, it can be seen from Tables 9 and 10 that the higher the final temperature of activation the lower the mean width of pore (Wpore ). For temperatures between 350 and 450 ◦ C macroporous and mesoporous carbons were obtained, respectively; whereas between 600 and 800 ◦ C microporous were obtained. A carefully analysis of mesopore volume compiled in Tables 9 and 10 showed that in spite of the framework of the carbon materials is mainly micropore, the samples prepared by gasification with CO2 showed higher contribution in the mesopore range than samples prepared under N2 flow. This was expected because as indicated above, CO2 reacts effectively with the carbon at temperatures higher than 600 ◦ C. In general, the mean pore diameter decreases monotonically with the increase of the activation or pyrolysis temperature. This could be the consequence of a pseudo-graphitization of the graphene sheets, in spite the present maxima temperature is 900 ◦ C, clearly lower than that require for this phenomena. The mean pore diameter was lower for the AC obtained under N2 flow than those obtained under CO2 atmosphere in any of the temperatures studied. For example, at 350 ◦ C, the mean macropore diameter on N2 flow was about the half ˚ This trend is the same than that on CO2 flow (922 against 1815 A). for the temperature where mesoporous were obtained (450 ◦ C), and also in the range (600–900 ◦ C) where micropore AC were obtained. It should be pointed out that at 900 ◦ C the lowest mean width of pore of about 6.7 A˚ (Table 9) and 5.3 A˚ (Table 10) were obtained under CO2 and N2 flows, respectively. These AC can be classify as an ultramicroporous AC which have shown several potential applications such as a double layer capacitor and electrode material [42], as a separation membrane [43], as catalytic support for the hydrogen production from the dry methane reforming [20,21,44,45], and as an efficient adsorbent for the hydrogen uptake and storage [46]. 3.4. General discussion We present here a first part of a major research showing some insights about the design of both textural and functional groups on the surface of AC. It can be summarized that in this work, a hydrophobic and ultramicroporous AC can be prepared by controlling the temperature and atmosphere of the thermal degradation of waste biomass. The potential of these AC, mainly in the industry and in environmental green chemistry applications by cataand photocatalytic heterogeneous reactions will be presented in the next two works. It should be pointed out that carbon derived materials obtained from sawdust of wood contain lower ash content that those prepared from other lignocellulosic materials such as agroindustrial bio-wastes and clearly much lower than those obtained from petroleum precursors. This is one of the reasons why activated carbon materials prepared from sawdust of wood has been employed successfully in catalytic heterogeneous reactions [1]. For example, Laine et al. [11,12] have showed that pore volume of activated carbon supports play a synergistic role upon

J. Matos et al. / Journal of Hazardous Materials 196 (2011) 360–369

367

Table 9 Micropore area (porearea ), micropore volume (porevolume ), mesopore volume (mporevolume ), total pore volume (Vtot ) and mean width of pore (Wpore ) of AC prepared by activation under CO2 flow by 1 h. T (◦ C) 350 450 600 700 800 900

– 326.7 624.1 892.0 1003 645.7

porevolume (cm3 /g)b

mporevolume (cm3 /g)b

Vtot (cm3 /g)c

˚ c Wpore (A)

d

d

0.030 0.107 0.336 0.478 0.538 0.667

1815 54.5 20.6 20.0 18.6 6.72

– 0.100 0.241 0.411 0.462 0.573

– 0.007 0.095 0.067 0.076 0.094

Obtained by HJ method. Obtained by HJ method. Obtained by HK method. Not estimated.

the activity and selectivity of NiMo catalysts the in thiophene hydrodesulphurization. Also, our group has showed that pore size distribution clearly influence the catalytic activity of Ni and NiMo catalysts in the ethylene hydrogenation [9] and the kinetics of coke deposition [10]. Also, our group have showed in different works about synthesis of activated carbons by physical activation or by pyrolysis [16], and by chemical activation [17] that the pore size distribution and surface area of activated carbons remarkably affects the photoactivity of TiO2 in the photocatalytic detoxification of 4-chlorophenol. In addition, we have showed that textural properties of activated carbon clearly influence the selectivity of main intermediate products detected during the aromatic molecules as phenol and 2,4-dichlorophenoxiacetic acid [13,14] and more recently on 4-chlorophenol [47,48] and 2-propanol [49] photooxidations. In this sense, Fig. 7 shows the influence of the two carbons prepared at 800 ◦ C by 1 h upon the phenol adsorption and on the photo efficiency of TiO2 in the phenol photo detoxification under UV-irradiation. It can be seen from Fig. 7 that any of two TiO2 –AC binary materials adsorbed higher phenol (after 15 min adsorption in the dark). This enhancement in phenol molecules around photoactive TiO2 enhances the photo efficiency of the semiconductor as can also be seen in Fig. 7. This enhances has been attributed to the presence of a common contact interface between TiO2 and AC that make possible a continuous transfer of the species from the AC to the TiO2 surface [49]. Our present enforces are aimed to prepare hierarchically macro–meso–micro porous carbon materials to study the influence of pore size distribution on the selectivity of NiMo catalysts in hydrocracking reactions and to verify the presence of confinement effects on the selectivity of hydrocracking consequence of specific pore size an pore volume of the support such as in the case of zeolites [50,51]. We do believe that the present results regarding the pyrolysis of sawdust of a hard wood as Algarroba consists of essentially 3 different zones is a very important finding and deserve to be studied carefully. A better explanation for the present results where the reaction rates rather decreased with the increasing temperature from 350 to 600 ◦ C could be due to the

7

7

Phads 6

kapp

6

5

5

4

4

3

3

2

2

1

1

0

kapp (min-1)x10-3

c d

d

Phads (micromols)

a b

porearea (m2 /g)a

0

TiO2

TiO2 + AC-CO2

TiO2 + AC-N2

Photocatalysts Fig. 7. Summary of kinetic results of phenol adsorption in the dark (Phads ) and first-order apparent rate-constants (kapp ) of phenol photodegradation under UVirradiation.

presence of different chemical structures of the starting materials at zero holding time after heating up to the set temperatures (i.e., 350, 400 and 600 ◦ C). Therefore, the estimation of kinetic parameters from data of modulated thermogravimetric analysis is required to better understand the correlation of the influence of each pyrolysis zone on the pore size distribution and pore volume of carbon materials. In addition, the influence of ashes of lignocellulosic carbon precursors and the influence of additives (chemical activators) that can play the role of catalysts to improve the textural properties are necessary to clarify the significance of the present results. In this way, our groups have already reported preliminary studies regarding the influence of the pyrolysis atmosphere [52] and the effect of

Table 10 Micropore area (porearea ), micropore volume (porevolume ), mesopore volume (mporevolume ), total pore volume (Vtot ) and mean width of pore (Wpore ) of AC prepared by activation under N2 flow by 1 h. T (◦ C) 350 450 600 700 800 900 a b c d

porearea (m2 /g)a d

– 201.7 449.1 494.0 515.4 454.8

Obtained by HJ method. Obtained by HJ method. Obtained by HK method. Not estimated.

porevolume (cm3 /g)b

mporevolume (cm3 /g)b

Vtot (cm3 /g)c

˚ c Wpore (A)

d

d

0.016 0.111 0.190 0.201 0.208 0.231

922 43.3 16.3 15.8 15.8 5.33

– 0.101 0.172 0.188 0.195 0.223

– 0.010 0.018 0.013 0.013 0.008

368

J. Matos et al. / Journal of Hazardous Materials 196 (2011) 360–369

chemical additives [53] on the topological organization of carbon materials obtained from the controlled pyrolysis of saccharose.

[18]

4. Conclusions [19]

AC was prepared from the sawdust of wood by physical activation and pyrolysis under CO2 and N2 flow, respectively. Maxima BET surface area was obtained at 800 ◦ C, both under CO2 and N2 atmospheres to then decrease at higher temperatures of activation. IR and XPS suggest that the higher the activation temperature the more basic is the functional groups on surface of carbons. Porosimetry showed that the higher the activation temperature the higher the micropores volume and the higher the micropore surface of AC. The higher the activation temperature the lower the pore diameter obtaining an ultramicroporous activated carbon at 900 ◦ C. It can be concluded that mean pore width and functionalization on the surface of AC can be easily controlled and this feature permits to think in waste biomass as a potential source for the synthesis of carbon materials with different and potential modern applications.

[26] [27]

Acknowledgement

[28]

J. Matos thanks to the Ministry of Science and Technology for financial support.

[29]

[20]

[21]

[22] [23] [24]

[25]

[30]

Appendix A. Supplementary data [31]

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.09.046.

[32]

References [33] [1] F. Rodríguez-Reinoso, M. Molina-Sabio, Activated carbons from lignocellulosic materials by chemical and/or physical activation: an overview, Carbon 30 (1992) 1111–1118. [2] K. Gergova, S. Eser, Effects of activation method on the pore structure of activated carbons from apricot stones, Carbon 34 (1996) 879–888. [3] F. Rodríguez-Reinoso, The role of carbon materials in heterogeneous catalysis, Carbon 36 (1998) 159–175. [4] M. Jagtoyen, F.F. Derbyshire, Activated carbons from yellow poplar and white oak by H3 PO4 activation, Carbon 36 (1998) 1085–1097. [5] N. Kannan, G. Rengasamy, Comparison of cadmium ion adsorption on various carbons, Fresenius Environ. Bull. 11 (2002) 160–164. [6] N. Tancredi, N. Medero, F. Möller, J. Piriz, C. Plada, T. Cordero, Phenol adsorption onto powdered and granular activated carbon, prepared from Eucalyptus wood, J. Colloid Interface Sci. 279 (2004) 357–363. [7] N. Kannan, G. Rengasamy, Comparison of cadmium ion adsorption on various activated carbons, Water Air Soil Pollut. 163 (2005) 185–201. [8] T. Tay, S. Ucar, S. Karagöz, Preparation and characterization of activated carbon from waste biomass, J. Hazard. Mater. 165 (2009) 481–485. [9] J. Matos, J. Laine, J.L. Brito, Activated carbon supported Ni–Mo: effects of pretreatments and composition on catalyst reducibility and on ethylene conversion, Appl. Catal. A: Gen. 152 (1997) 27–42. [10] J. Matos, J. Laine, Ethylene conversion on activated carbon supported NiMo catalysts: effect of the support, Appl. Catal. A: Gen. 241 (2003) 25–38. [11] J. Laine, F. Severino, M. Labady, J. Gallardo, The synergistic participation of the support in sulfided Ni–Mo/C hydrodesulfurization catalysts, J. Catal. 138 (1992) 145–149. [12] J. Laine, F. Severino, M. Labady, Optimum Ni composition in sulfided Ni–Mo hydrodesulfurization catalysts: effect of the support, J. Catal. 147 (1994) 355–357. [13] J. Matos, J. Laine, J.-M. Herrmann, Synergy effect in the photocatalytic degradation of phenol on a suspended mixture of titania and activated carbon, Appl. Catal. B: Environ. 18 (1998) 281–291. [14] J. Matos, J. Laine, J.-M. Herrmann, Effect of the type of activated carbons on the photocatalytic degradation of aqueous organic pollutants by UV-irradiated titania, J. Catal. 200 (2001) 10–20. [15] J. Matos, J. Laine, J.-M. Herrmann, D. Uzcategui, J.L. Brito, Influence of activated carbon upon titania on aqueous photocatalytic consecutive runs of phenol photomineralization, Appl. Catal. B: Environ. 70 (2007) 461–469. [16] T. Cordero, C. Duchamp, J.-M. Chovelon, C. Ferronato, J. Matos, Surface nanoaggregation and photocatalytic activity of TiO2 on H-type activated carbons, Appl. Catal. B: Environ. 73 (2007) 227–235. [17] T. Cordero, C. Duchamp, J.-M. Chovelon, C. Ferronato, J. Matos, Influence of L-type activated carbons on photocatalytic activity of TiO2 in 4-

[34] [35]

[36]

[37]

[38] [39] [40]

[41]

[42] [43] [44]

[45]

[46]

[47]

[48]

chlorophenol photodegradation, J. Photochem. Photobiol. A: Chem. 191 (2007) 122–131. A. Dufour, A. Celzard, V. Fierro, E. Martin, F. Broust, A. Zoulalian, Catalytic decomposition of methane over a wood char concurrently activated by a pyrolysis gas, Appl. Catal. A: Gen. 346 (2008) 164–173. A. Dufour, A. Celzard, B. Ouartassi, F. Broust, V. Fierro, A. Zoulalian, Effect of the micropores diffusion on kinetics of CH4 decomposition over a wood-derived carbon catalyst, Appl. Catal. A: Gen. 360 (2009) 120–125. J. Matos, K. Díaz, V. García, T. Cordero, J.L. Brito, Methane transformation in presence of carbon dioxide on activated carbon supported nickel–calcium catalysts, Catal. Lett. 109 (2006) 163–169. K. Díaz, V. García, J. Matos, Activated carbon supported Ni–Ca: influence of reaction parameters on activity and stability of catalyst on methane reformation, Fuel 86 (2007) 1337–1344. D. Mohan, C.U. Pittman, P.H. Steele, Pyrolysis of wood/biomass for bio-oil: a critical review, Energy Fuels 20 (2006) 848–889. C. Di Blasi, C. Branca, Kinetics of primary product formation from wood pyrolysis, Ind. Eng. Chem. Res. 40 (2001) 5547–5556. C. Branca, C. Di Blasi, Kinetics of the isothermal degradation of wood in the temperature range 528–708K, J. Anal. Appl. Pyrolysis 67 (2003) 207–219. E. Schröder, K. Thomauske, C. Weber, A. Hornung, V. Tumiatti, Experiments on the generation of activated carbon from biomass, J. Anal. Appl. Pyrolysis 79 (2007) 106–111. J. Matos, C. Nahas, L. Rojas, M. Rosales, Open Environ. Eng. J. 4 (2011) 1–10. J. Laine, A. Calafat, LabadyF M., Preparation and characterization of activated carbons from coconut shell impregnated with phosphoric acid, Carbon 27 (1989) 191–195. M. Sefain, S. El-Kalyoubi, N. Shukry, Thermal behavior of holo- and hemicellulose obtained from rice straw and bagasse, J. Polym. Sci.: Polym. Chem. 23 (1985) 1569–1577. T. Wigmans, Industrial aspects of production and use of activated carbons, Carbon 27 (1989) 13–22. J. Matos, M. Labady, A. Albornoz, J. Laine, J.L. Brito, Topological organization and textural changes of carbon macro-networks submitted to activation with N2 and CO2 , J. Mater. Sci. 39 (2004) 3705–3716. F.E. Huggins, G.P. Huffman, N. Shah, F.W. Lytle, R.B. Greegor, R.G. Jenkins, In Situ XAFS studies of catalyzed pyrolysis and gasification reactions of lignite, Phys. B: Condens. Matter 158 (1989) 178–179. S.G. Chen, R.T. Yang, Unified mechanism of alkali and alkaline earth catalyzed gasification reactions of carbon by CO2 and H2 O, Energy Fuels 11 (1997) 421–427. J. Rodriguez, F. Ruette, J. Laine, Molecular modeling of micropores in activated carbon, Carbon 32 (1994) 1536–1537. N.L. Owen, D.W. Thomas, Infrared studies of hard and soft woods, Appl. Spectrosc. 43 (1989) 451–455. E. Fuente, J.A. Menéndez, M.A. Diez, D. Suárez, M.A. Montes-Morán, Infrared spectroscopy of carbon materials: a quantum chemical study of model compounds, J. Phys. Chem. B 107 (2003) 6350–6359. A. Swiatkowski, H. Grajek, M. Pakula, S. Biniak, Z. Witkiewicz, Voltammetric studies of the gradual thermal decomposition of activated carbon surface oxygen complexes, Colloids Surf. A: Phys. Eng. Aspects 208 (2002) 313–320. A. Terzyk, The influence of activated carbon surface chemical composition on the adsorption of acetaminophen (paracetamol) in vitro. Part II. TG. FTIR, and XPS analysis of carbons and the temperature dependence of adsorption kinetics at the neutral Ph, Colloids Surf. A: Phys. Eng. Aspects 177 (2001) 23–45. H. Valdés, M. Sánchez-Polo, J. Rivera-Utrilla, C.S. Zaror, Effect of ozone treatment on surface properties of activated carbon, Langmuir 18 (2002) 2111–2116. T.L. Barr, Recent advances in X-ray photoelectron spectroscopy studies of oxides, J. Vac. Sci. Technol. A 9 (1991) 1793–1805. C. Basquet, B. Rousseau, H. Estrade-Szwarckopt, P. Le Cloirec, Observation of activated carbon fibres with SEM and AFM correlation with adsorption data in aqueous solution, Carbon 38 (2000) 407–422. A. Marcilla, S. García-García, M. Asensio, J.A. Conesa, Influence of thermal treatment regime on the density and reactivity of activated carbons from almond shells, Carbon 38 (2000) 429–440. J. Koresh, A. Soffer, Double layer capacitance and charging rate of ultramicroporous carbon electrodes, J. Electrochem. Soc. 124 (1977) 1379–1385. C.W. Jones, W.J. Koros, Characterization of ultramicroporous carbon membranes with humidified feed, Ind. Eng. Chem. Res. 34 (1995) 158–163. J. Matos, M. Rosales, C. Urbina de Navarro, G. González, Medusa-like carbon nanotubes formed during dry methane reforming on activated carbon-supported Ni–Ca catalyst, Acta Microsc. 18 (2009) 322–332. J. Matos, M. Rosales, G. González, C. Urbina de Navarro, Changes on texture and crystalline phase of activated carbon-supported Ni–Ca catalyst during dry methane reforming, Open Mater. Sci. J. 4 (2010) 125–132. V.V. Bhata, C.I. Contescu, N.C. Gallego, F.S. Baker, A typical hydrogen uptake on chemically-activated, ultramicroporous carbon, Carbon 48 (2010) 1331–1340. J. Matos, A. García, T. Cordero, J.-M. Chovelon, C. Ferronato, Eco-friendly TiO2 –AC photocatalyst for the selective photooxidation of 4-chlorophenol, Catal. Lett. 130 (2009) 568–574. J. Matos, A. García, P.S. Poon, Environmental green chemistry applications of nanoporous carbons, J. Mater. Sci. 45 (2010) 4934–4944.

J. Matos et al. / Journal of Hazardous Materials 196 (2011) 360–369 [49] J. Matos, E. García-López, L. Palmisano, A. García, G. Marci, Influence of activated carbon in TiO2 and ZnO mediated photo-assisted degradation of 2-propanol in gas–solid regime, Appl. Catal. B: Environ. 99 (2010) 170–180. [50] E. Benazzi, L. Leite, N. Marchal-George, H. Toulhoat, P. Raybaud, New insights into parameters controlling the selectivity in hydrocracking reactions, J. Catal. 217 (2003) 376–387.

369

[51] H. Toulhoat, P. Raybaud, E. Benazzi, Effect of confinement on the selectivity of hydrocracking, J. Catal. 221 (2004) 500–509. [52] J. Matos, J. Laine, A carbon macro-network from the controlled pyrolysis of saccharose, J. Mater. Sci. Lett. 17 (1998) 649–651. [53] J. Matos, M. Labady, A. Albornoz, J. Laine, J.L. Brito, Catalytic effect of KOH on textural changes of carbon macronetworks by physical activation, J. Mol. Catal. A: Chem. 228 (2005) 189–194.

Journal of Hazardous Materials 196 (2011) 370–379

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Crystallization evolution, microstructure and properties of sewage sludge-based glass–ceramics prepared by microwave heating Yu Tian a,b,∗ , Wei Zuo a , Dongdong Chen a a b

School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology (SKLUWRE, HIT), Harbin 150090, China

a r t i c l e

i n f o

Article history: Received 19 April 2011 Received in revised form 7 September 2011 Accepted 10 September 2011 Available online 16 September 2011 Keywords: Sewage sludge Glass–ceramics Microwave Double-layer reactor

a b s t r a c t A Microwave Melting Reactor (MMR) was designed in this study which improved the microwave adsorption of sewage sludge to prepare glass–ceramics. Differential scanning calorimetry (DSC), X-ray diffraction (XRD), and scanning electron microscopy (SEM) were used for the study of crystallization behavior and microstructure of the developed glass–ceramics. DSC and XRD analysis revealed that crystallization of the nucleated specimen in the region of 900–1000 ◦ C resulted in the formation of two crystalline phases: anorthite and wollastonite. When the crystallization temperature increased from 900 to 1000 ◦ C, the tetragonal wollastonite grains were subjected to tensile microstresses, causing the cracking of crystal. Al ions substituted partially Si ions and occupied tetrahedral sites, giving rise to the formation of anorthite. The relationship between microwave irradiation and crystal growth was studied and the result indicated that the microwave selective heating suppressed the crystal growth, giving apparent improvements in the properties of the glass–ceramics. The glass–ceramics products exhibited bending strength of 86.5–93.4 MPa, Vickers microhardness of 6.12–6.54 GPa and thermal expansion coefficient of 5.29–5.75 × 10−6 /◦ C. The best chemical durability in acid and alkali solutions was 1.32–1.61 and 0.41–0.58 mg/cm2 , respectively, showing excellent durability in alkali solution. © 2011 Elsevier B.V. All rights reserved.

1. Introduction One of the main environmental problems is the safe disposal of the huge amount of sewage sludge that is produced every day in wastewater treatment plants [1]. Among the methods of the treatment of sewage sludge, glass–ceramics preparation seems to be a promising one for converting sewage sludge into novel materials that possess attractive mechanical and chemical properties [2]. Sewage sludge containing large amounts of CaO, SiO2 , and Al2 O3 can be a good raw material for glass–ceramics production. By controlling the initial composition and by suitable heat treatment, a variety of crystalline phases will be obtained [3]. They exhibit bending strength, Vickers microhardness, fracture toughness, chemical durability and thermal shock resistance superior to those of glass, and in some cases traditional ceramics [4,5]. It should be noted that the chemical energy of the organic components in sewage sludge could be recovered during the prepared procedure of glass–ceramics as an auxiliary energy source. The

∗ Corresponding author at: School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China. Tel.: +86 451 8608 3077/13804589869; fax: +86 451 8628 3077. E-mail address: [email protected] (Y. Tian). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.045

chemical energy in sewage sludge can be recovered during the prepared procedure of glass–ceramics as an auxiliary energy source, reducing the emission of CO2 which is favorable to Kyoto Protocol [6]. Other advantages of this technology are the possibility of immobilizing heavy metal ions (held in the framework of glass or encapsulated into the crystallization phase) [7], the large reduction of volume (vary between 40 and 90%), and the flexibility of treatment procedure (which may accept different types of sewage sludge, either municipal or industrial) [8]. Preparing glass–ceramics by the conventional technology is an energy-intensive process, with the process temperature as high as about 1300 ◦ C and the process time required as several hours [9]. Economic analysis of a glass–ceramics preparation system which can process 0.5–1.0 ton of sewage sludge per hour showed that the operating costs of this unit ranged from US$100–420 per ton, including labor, fuel and maintenance [10]. Another critical point in glass–ceramics preparation is the difficulty in controlling the size and the type distributions of the crystals due to the thermal inertia of the conventional heating [11]. To overcome these drawbacks, microwave heating has been developed as an alternative technology for the preparation of dense structural glass–ceramics, which is characterized by shorter reaction time, reduced energy consumption, and suppressed crystal size. It was found that the treatment temperature was decreased

Y. Tian et al. / Journal of Hazardous Materials 196 (2011) 370–379

371

Fig. 1. A schematic of the microwave preparation reactor assembly: (1) microwave cavity; (2) Microwave Melting Reactor (MMR); (3) waveguide; (4) magnetron; (5) PC with fuzzy logic algorithm; (6) power governor; (7) infrared radiation thermometer.

from 1300 ◦ C to 1000 ◦ C when a glass–ceramics was sintered from barium aluminosilicate glass in microwave field [12]. It was also reported that an abrasion resistant glass–ceramics was developed from the MgO–Al2 O3 –TiO2 system in 20 min by microwave heating [13]. Moreover, more uniform and strong bonding was observed in the glass–ceramics prepared by microwave, indicating that microwave energy suppressed the grain growth in crystal phase due to a fast heating rate and apparent low-temperature crystallization [8]. Sewage sludge is a poor receptor of microwave energy to achieve the temperature necessary for preparing glass–ceramics. It has been proved that microwave-induced preparation is possible, if an effective receptor is added into the raw sludge. The temperature of sewage sludge can achieve 1200 ◦ C in microwave field when it was homogeneously blended with microwave receptor, such as graphite and char [9]. However, there are fundamental disadvantages of this method when it is applied in glass–ceramics preparation. The chemical composition of the samples shows uncontrollable changes in virtue of adding microwave receptor, leading to the poor properties of the products. In addition, the microwave receptor could not be recovered due to the encapsulation of silicate matrix in the glass–ceramics, increasing the operating cost of the procedure. Attempts termed as “hybrid microwave sintering” were also made to set around the sample directly to initially heat the material at room temperature [14]. However, the temperature of sewage sludge could not reach high enough owing to the significant reflection loss on the interface between microwave receptor layer and the air surrounding it. To solve these problems, a new Microwave Melting Reactor (MMR) was designed in this study for preparing glass–ceramics

from sewage sludge. In MMR, microwave absorption of sewage sludge can be improved by the double-layer structure and the required temperature can be achieved in a very short of time, usually in a few minutes. A wave-transparent layer was introduced into the MMR system to decrease the reflection coefficient of the interface between the air and the MMR. Another important property of the powder was the low thermal conductivity which could give the sample a good heat insulation quality. The double-layer structure in MMR provides the even distribution of temperature and electromagnetic field in the samples, favoring the production of glass–ceramics with desired qualities. Further researches presented in this paper were focused on: (1) investigating the influence of heat-treatment schedule on the crystallization behavior and microstructure of the microwave-prepared glass–ceramics, (2) defining the evolution of crystallization in microwave field which were hardly found by applying the conventional procedures, and (3) gaining an insight into the chemical and physical properties of the glass–ceramics prepared by microwave in comparison to that obtained from conventional process. 2. Experimental 2.1. The design of MMR The 2.45 GHz microwave furnace, which consisted of a rectangular multimode cavity, a continually adjustable power supply (0.50–2.7 kW), a temperature controlling system, and a Microwave Melting Reactor, was used for microwave heating experiment. As shown in Fig. 1, the Microwave Melting Reactor (MMR) consisted of a wave-absorbing layer and a wave-transparent layer. The

372

Y. Tian et al. / Journal of Hazardous Materials 196 (2011) 370–379

wave-transparent layer is the surface layer which plays an important role in avoiding the reflection loss of the incident wave on the front surface between the reactor and air. The wave-absorbing layer beneath it absorbed the incident wave transmitted through the wave-transparent layer and transformed the electromagnetic energy to thermal energy. In terms of optimizing the MMR performance, the material properties and the thickness of each layer are most important parameters to design the reactor structure. Active carbon, a well known microwave receptor, was filled in the microwave-absorption layer. The material filled in microwavetransparent layer was selected according to the expression for reflection coefficient as given in Eq. (1) [15]:

   2 − 1    +

R=

2

(1)

Proximate analysis (wt.%) Aa

Va

Caa , b

Oc

Ha , b

Na , b

Sa , b

24.50

75.50

39.40

24.43

5.71

4.75

1.18

Heavy metal content in dry sewage sludge (ppm) Cr

Cd

Cu

Pb

Zn

Fe

Ni

143

4.75

138

59.9

700

10,200

38.7

A: ash content; V: volatile matter content. a Dry base. b Ash free basis. c Calculated by difference.

1

where R is the reflection coefficient, 1 and 2 are the characteristic impedances of the air and the material filled in microwavetransparent layer, respectively. It is clear that the value of 2 should be close to the value of 1 for decreasing the reflection coefficient of the incident wave. Based on the results of our preliminary experiments, ferric oxides mixed with aluminum oxides (Fe2 O3 /Al2 O3 = 1:1) were adopted as the materials filled in the microwave-transparent layer. Before filled into the microwavetransparent layer, the Fe2 O3 and Al2 O3 grains were ground using a mill to obtain a mixed powder with particle size ≤75.0 ␮m. The mixture powder had the properties of both lower characteristic impedance and higher microwave transmission rate, decreasing the reflection coefficient of the interface between the air and the MMR. Another important property of the powder is the low thermal conductivity which gives the sample a good heat insulation quality. The thicknesses of microwave-transmission layer and microwave-absorption layer were determined according to the penetration depth (DE , depth of the microwave energy penetrates into a material). DE can be calculated by the Fresnel formula [8]: DE =

Table 1 Chemical characteristics of sewage sludge.

0 √  εr tgı

(2)

where 0 is the length of electromagnetic wave in vacuum, εr is the material dielectric constant, and tgı is the dielectric loss tangent. According to the calculation of Eq. (2), both the optimal thicknesses of microwave-transmission layer and microwave-absorption layer were determined as 2 mm. The glass preparation from sewage sludge has been carried out to test the behavior of MMR in microwave heating. It was observed that the temperature of the specimen required for preparing glass was reached (1300 ◦ C) and parent glass was prepared successfully in this reactor. 2.2. Parent glass production Sewage sludge used in experiments was collected from urban wastewater treatment plants in Harbin, China. Selected chemical characteristics and the heavy metal contents of this sludge are given in Table 1. The dehydrated sewage sludge (moisture content was 79.8%) which contained small hard particles were crushed in a mortar and then heated at 1000 ◦ C until the sludge samples reached constant weight to remove the volatile components. Sewage sludge must be mixed with additives to lower the melting temperature from its melting around 1500 ◦ C. In our experiments, CaO and waste glass were used as effective additives. The chemical compositions of the sewage sludge, waste glass and raw materials were examined by X-ray fluorescence spectroscopy (XRF) and the results are shown in Table 2, indicating that the formed glass should be in the SiO2 –CaO–Al2 O3 ternary phase system. Fig. 2 shows the phase diagram of the CaO–Al2 O3 –SiO2 system. The chemical compositions of raw materials for preparing glass–ceramics could be located in the

wollastonite–anorthite subsystem (the region marked by hatching in Fig. 2). The batch composition, prepared by mixing 52.0 wt.% of the raw sludge with 21.0 wt.% of CaO and 21.0 wt.% of waste glass, was chosen on the basis of the eutectic composition (CaO 38.0, Al2 O3 20.0 and SiO2 42.0 mass%) [16]. Additionally, 6.0 wt.% TiO2 were added to the base glass composition as nucleating agents. Mixtures obtained above were melted by microwave processing and conventional processing respectively. In the microwave process, glasses were prepared by melting sewage sludge in a corundum crucible at 2000 W microwave power for 10 min and then cooled naturally to room temperature. In the conventional process, glasses were prepared by melting the mixture in an alumina crucible at 1450 ◦ C for 2 h after which melts were preheated at 600 ◦ C to reduce thermal shock. The results of XRD analysis for sewage sludge, parent glasses obtained from these two different heating processes are shown in Fig. 3. 2.3. Glass–ceramics production It is important to determine nucleation and crystal growth temperatures precisely for effective conversion of glasses to glass–ceramics. Differential scanning calorimetry analysis (DSC) is performed using a calorimeter (STA449C, NETZSCH) with ␣Al2 O3 as standard sample. The glass powders are heated from room temperature to 1100 ◦ C with the rate of 10 ◦ C/min in order to detect the nucleation and crystallization temperatures. According to the results of DSC, heat-treatment schedule for the microwaveproduced parent glass should include a nucleation stage at 760 ◦ C for 30 min followed by a crystal growth stage at different temperatures (900 ◦ C, 950 ◦ C and 1000 ◦ C) for 60 min in microwave irradiation. For conventional glass, sample should be held at nucleation temperature (820 ◦ C) for 90 min and then heated to crystallization temperature (1000 ◦ C) for 120 min in electric furnace. Fig. 4 shows the processes of glass–ceramics preparation by microwave and conventional methods. The types of the crystalline phases were characterized by X-ray with Cu K␣ radiation (XRD: P|max-␥␤, Rigaku, Japan). The step length was 0.02◦ with scanning speed of 5◦ /min in the range of 10–90◦ (Cu Ka = 1.5418 A). The schemes of glass–ceramics preparation by microwave and conventional heating are shown in Fig. 4. 2.4. Methods to evaluate glass–ceramics properties Several techniques were used to evaluate the properties of glasses and glass–ceramics. The morphology of the crystalline phases made in thermal glass treated was investigated using a scanning electron microscope (SEM, S-4700, HITACHI). Archimedes’ method was employed to measure the apparent density of the glass–ceramics. Hardness and fracture toughness were measured by an indentation method using the Vickers indenter. Vickers hardness was measured with loads of 100–1000 g with loading

Y. Tian et al. / Journal of Hazardous Materials 196 (2011) 370–379

373

Fig. 2. Phase diagram of the CaO–Al2 O3 –SiO2 system. The chemical compositions of sewage sludge correspond to the eutectic point marked by hatching (CS, wollastonite; CAS2 anorthite).

times of 10 s. Bending strength was obtained from a four-point method with spans of 20 and 40 mm at a cross-head speed of 100 mm/min, as designated by American Society of Testing Materials (ASTM) E855-90 [17]. The thermal expansion coefficient (20–400 ◦ C) was measured by TMA with a heating rate of 10 ◦ C/mm in atmosphere. Chemical durability was measured following the designation of American Society of Testing Materials (ASTM) C27988 [18]. First, powdered specimens were prepared in particle sizes of 4.75–6.75 mm. 20.0 g specimen powder was then immersed into 100 ml of 1 mass% H2 SO4 (about 0.10 mol/l) or 1 mass% NaOH (0.25 mol/l) and boiled on a hot plate for 48 h. The specimens were dehydrated and acid/alkali durability was estimated by measuring the weight loss of powders. 2.5. Methods of heavy metal leaching tests Leaching tests of sewage sludge and glass–ceramics were subjected to the toxicity characteristic leaching procedure (TCLP) method according to the US Environmental Protection Agency [19]. The sludge and glass–ceramics samples were manually crushed ( E > F  A > B under the same SIE. 3.1.1. Plasma conversion of HCHO Generally, VOCs can be removed by discharge plasma via three pathways, i.e., direct electron impacts, gas-phase radical attacks and ion collisions. Results presented in Fig. 3 show that HCHO can be removed not only in the plasma zone (system A), but also in the post-plasma cylinder (system B). For an SIE of 20 J/L, the conversion of HCHO was 36% and 29% for systems A and B, respectively. Considering that unstable plasma species, such as energetic electrons and some gas-phase radicals, cannot reach the postplasma reactor because of their millisecond lifetimes [23,24], the removal of HCHO in system B can only be attributed to relatively stable (not including O3 ) and/or metastable plasma species (e.g. N2 metastable states). In fact, it has been reported that for plasma removal of HCHO, N2 metastable states may be more important than electrons owing to their longer lifetime [25]. These excited states of N2 contribute to the removal of HCHO via two possible pathways: direct attacks towards HCHO molecules and indirect reactions through the O2 dissociation processes, as shown in Eqs. (5)–(9) [25,26]. N2 (A3 + u ) + HCHO → H + HCO + N2 

Fig. 3. Effects of specific input energy on the HCHO conversion.

(5)

N2 (a ) + HCHO → H + HCO + N2

(6)

3 3 N2 (A3 + u ) + O2 → O( P) + O( P) + N2

(7)

N2 (a ) + O2 → O(3 P) + O(3 P) + N2

(8)

O + HCHO → OH + HCO

(9)

X. Fan et al. / Journal of Hazardous Materials 196 (2011) 380–385

383

where N2 (a ) denotes the three states of a1 , a1  and ω1 , with a mean energy of 8.4 eV; while the energy of N2 (A3 + u ) is 6.2 eV for the = 0 level. In the case of system A in this study, besides the efficient removal of HCHO in the discharge zone, further removal of the unreacted HCHO can be expected in the downstream cylinder too, due to the residual long-living plasma-excited species.

3.1.2. Plasma-catalytic conversion of HCHO As shown in Fig. 3, HCHO can be more efficiently removed in the plasma–catalyst hybrid systems (D–F) when compared with the plasma alone processes (systems A and B). For an SIE of 20 J/L, the conversion of HCHO was 87%, 76% and 72% for systems D, E and F, respectively. This result suggests that the downstream MnOx /Al2 O3 catalyst can be effectively activated by long-living plasma species for HCHO conversion at room temperature and the heterogeneous oxidation reactions over the catalyst surface are much more important for HCHO removal than the homogeneous reactions in the discharge zone. As stated previously, not only short-living unstable reactive species are produced in plasma discharges, a fraction recombines to form more stable species such as O3 [21,27]. Comparison of the HCHO conversion in systems C and F shows that although O3 does not oxidize HCHO in the gas phase, it does initiate the removal of HCHO over the MnOx /Al2 O3 catalyst. O3 catalytic decomposition mechanism research shows that O3 decomposition over manganese oxide catalyst produces atomic oxygen and peroxide as the intermediate species [28,29]. These highly active oxygen species should be mainly responsible for the catalytic oxidation of HCHO in system F. Besides O3 , other long-living plasma-excited species, which account for the HCHO removal in system B, can also exist in the second cylinder of system E. These species may trigger both homogeneous and heterogeneous reactions of HCHO, resulting in higher HCHO conversion in system E than in system F under the same conditions. The test results in Fig. 3 also show that the treatment system D behaves the best in terms of HCHO conversion in this study, which can be easily attributed to the best use of the plasma-generated active species in a two-stage HCHO destruction process: firstly, HCHO was attacked by energetic electrons and reactive species in the discharge zone; secondly, unreacted HCHO from the discharge zone was further removed in the post-plasma stage mainly via catalytic processes initiated by O3 and also other long-living active species. Comparison of the HCHO conversion in systems D, E and F shows that the O3 initiated catalytic oxidation reactions play a significant role in the plasma-catalytic removal of HCHO. Moreover, it should be noticed that for the three hybrid systems, the difference in HCHO conversion is not significant for an SIE lower than 3 J/L. We may have to consider that this phenomenon occurs for the following reasons. On the one hand, the production of high energy long-living species, such as N2 (A3 + u ) and N2 (a ), is very limited in low-energy discharge plasma. Therefore, the conversion of HCHO in system E may only result from the catalytic ozonation process just as that in system F. On the other hand, it can be seen from Fig. 3 that the heterogeneous reactions are much more important than the homogeneous reactions towards HCHO conversion, especially in the case of low SIE. This probably explains the small difference in HCHO conversion between systems D and E. Nevertheless, the difference in HCHO conversion among the three hybrid systems becomes remarkable at higher SIE, with the increasing production of high-energy long-living species and also higher HCHO conversion in the discharge zone.

Fig. 4. Effects of specific input energy on the energy efficiency concerning HCHO conversion.

In summary, HCHO can be removed not only by short-living active species in the discharge zone, but also by long-living species except O3 downstream the plasma reactor. Compared with the plasma alone processes, the tandem plasma–catalyst hybrid systems perform much better in HCHO conversion, mainly arising from the O3 initiated heterogeneous destruction of HCHO over the post-plasma MnOx /Al2 O3 catalyst. 3.2. Energy efficiency Fig. 4 presents the energy efficiency concerning HCHO conversion as a function of SIE for treatment systems A, B, and D–F. It can be seen that the energy efficiency decreased with the increase of SIE for all the five treatment systems. Obviously, the concentration of HCHO in the gas stream decreased with the increase of HCHO conversion, resulting in lower collision probability between HCHO molecules and active species and hence lower removed amount of HCHO per kWh of energy consumption at higher SIE. Meanwhile, more of the energy in plasma was converted into heat, photons, and used for byproduct formation (as shown in Fig. 5) with the increase of SIE. Higher SIE favors the complete removal of HCHO (Fig. 3) but causes serious energy inefficiency of the process. Therefore, maximum available value for input power in the plasma reactor will be determined not only by HCHO conversion but also by the

Fig. 5. Effects of specific input energy on the emission of O3 .

384

X. Fan et al. / Journal of Hazardous Materials 196 (2011) 380–385

energy efficiency. Nevertheless, compared with the plasma alone processes (systems A and B), the energy efficiency was greatly improved by introducing post-plasma MnOx /Al2 O3 catalyst (systems D–F), indicating that the plasma-catalysis hybrid processes have higher HCHO removal capability and are more promising for practical applications. The maximum energy efficiency was 3.1, 3.1 and 2.5 g/kWh for systems D, E and F, respectively, compared to 0.9 g/kWh for system A and 0.3 g/kWh for system B. 3.3. Byproduct formation 3.3.1. Ozone As long as the NTP process is operated in air-like mixtures, the formation of O3 , a hazardous discharge byproduct, is unavoidable. Fig. 5 shows the O3 outlet concentration as a function of SIE for treatment systems A and D. In fact, O3 outlet concentrations were measured in both the presence and absence of HCHO in air in this study for all treatment systems. Results prove that low levels of HCHO in the gas stream as well as the introduction of a buffer flask hardly influence the O3 outlet concentration. As seen from Fig. 5, the O3 outlet concentration increased with the increase of SIE for both plasma alone and plasma-catalysis hybrid processes. Compared with plasma alone, however, the presence of post-plasma MnOx /Al2 O3 catalyst significantly reduced the O3 emission. For an SIE of 20 J/L, the O3 outlet concentration decreased from 57.2 ppm for system A to 13.9 ppm for system D. It is clear that mainly O3 induced by gas discharge was decomposed catalytically over the MnOx /Al2 O3 surface, producing highly active oxygen species which play a key role in the enhanced removal of HCHO in the plasma-catalysis hybrid processes (Fig. 3). Nevertheless, it should be noticed that even in the presence of MnOx /Al2 O3 catalyst, the O3 emission (13.9 ppm for an SIE of 20 J/L) is still high. In future research, it will be tested if the simultaneous catalytic removal of HCHO and O3 can be further improved by introducing catalysts that are more reactive towards O3 decomposition. 3.3.2. Formic acid Formic acid (HCOOH) is a common intermediate produced during the HCHO oxidation process [5,22]. Outlet concentrations of HCOOH were measured for treatment systems A and D in this study to investigate the influence of downstream MnOx /Al2 O3 catalyst on the formation of decomposition byproducts. In order to obtain measurable concentrations of HCOOH, a higher initial HCHO concentration of 40.9 ± 0.5 ppm was used. Fig. 6a and b shows the outlet concentration and the yield of HCOOH as functions of SIE, respectively. As seen from Fig. 6a, the HCOOH concentration at the outlet of system A increased with the increase of SIE, indicating that HCOOH was indeed produced as a byproduct in the plasma decomposition of HCHO and the absolute production of HCOOH increased with the increase of HCHO removed at higher SIE. On the contrary, the HCOOH outlet concentration of system D linearly decreased with the increase of SIE and was much lower than that of system A under the same SIE. For an SIE of 80 J/L, the HCOOH outlet concentration was 2.0 and 0.1 ppm for systems A and D, respectively. The difference in HCOOH production between the two systems indicates that HCOOH produced in the discharge zone can be effectively removed over the downstream MnOx /Al2 O3 catalyst, especially at higher SIE. The presence of post-plasma MnOx /Al2 O3 catalyst does not only significantly enhance the conversion of HCHO (Fig. 3), but also favors the elimination of organic byproducts. In addition, results presented in Fig. 6b show that the HCOOH yield decreased with the increase of SIE for both systems A and D. Since the absolute removal of HCHO increased with the increase of SIE, the decreasing HCOOH yield in system D can be easily attributed to its decreasing production of HCOOH, as shown in

Fig. 6. Effects of specific input energy on the production of HCOOH: (a) HCOOH concentration and (b) HCOOH yield.

Fig. 6a. On the other hand, although the production of HCOOH increased with the increase of SIE in system A (Fig. 6a), the HCOOH yield decreased monotonously, suggesting that more of the removed HCHO tends to undergo further oxidation reactions in higher-energy discharge plasma to form end products such as CO2 and CO.

4. Conclusions The roles of various plasma species in the plasma and plasma-catalytic removal of low-concentration HCHO in air were experimentally studied in this work. The main findings can be summarized as follows:

(1) Both short- and long-living plasma species (other than O3 ) contribute to HCHO removal in the gas phase. (2) O3 does not initiate HCHO removal in the gas phase but does trigger heterogeneous destruction of HCHO over the MnOx /Al2 O3 catalyst, well explaining the greatly enhanced HCHO conversion by combining plasma with the MnOx /Al2 O3 catalyst in series. (3) The best use of plasma-generated active species for HCHO destruction can be achieved in a plasma–catalyst hybrid system where HCHO is introduced through the discharge zone and then the catalyst bed, leading to the highest energy efficiency concerning HCHO conversion. (4) The introduction of MnOx /Al2 O3 catalyst after the plasma reactor significantly reduces the emission of discharge byproducts (O3 ) and organic intermediates (HCOOH), showing great potential for indoor VOCs’ purification.

X. Fan et al. / Journal of Hazardous Materials 196 (2011) 380–385

Acknowledgement The authors thank the National High Technology Research and Development Program (863) of China (No. 2010AA064904) for the financial support of this work. References [1] T. Salthammer, S. Mentese, R. Marutzky, Formaldehyde in the indoor environment, Chem. Rev. 110 (2010) 2536–2572. [2] D.A. Missia, E. Demetriou, N. Michael, E.I. Tolis, J.G. Bartzis, Indoor exposure from building materials: a field study, Atmos. Environ. 44 (2010) 4388–4395. [3] X.J. Tang, Y. Bai, A. Duong, M.T. Smith, L.Y. Li, L.P. Zhang, Formaldehyde in China: production, consumption, exposure levels, and health effects, Environ. Int. 35 (2009) 1210–1224. [4] S. Tanada, N. Kawasaki, T. Nakamura, M. Araki, M. Isomura, Removal of formaldehyde by activated carbons containing amino groups, J. Colloid Interface Sci. 214 (1999) 106–108. [5] F. Shiraishi, D. Ohkubo, K. Toyoda, S. Yamaguchi, Decomposition of gaseous formaldehyde in a photocatalytic reactor with a parallel array of light sources. 1. Fundamental experiment for reactor design, Chem. Eng. J. 114 (2005) 153–159. [6] C.B. Zhang, H. He, K.I. Tanaka, Perfect catalytic oxidation of formaldehyde over a Pt/TiO2 catalyst at room temperature, Catal. Commun. 6 (2005) 211–214. [7] X.F. Tang, J.L. Chen, Y.G. Li, Y. Li, Y.D. Xu, W.J. Shen, Complete oxidation of formaldehyde over Ag/MnOx –CeO2 catalysts, Chem. Eng. J. 118 (2006) 119–125. [8] C.Y. Li, Y.N. Shen, M. Jia, S.S. Sheng, M.O. Adebajo, H.Y. Zhu, Catalytic combustion of formaldehyde on gold/iron-oxide catalysts, Catal. Commun. 9 (2008) 355–361. [9] R.H. Wang, J.H. Li, OMS-2 catalysts for formaldehyde oxidation: effects of Ce and Pt on structure and performance of the catalysts, Catal. Lett. 131 (2009) 500–505. [10] J. Van Durme, J. Dewulf, W. Sysmans, C. Leys, H. Van Langenhove, Efficient toluene abatement in indoor air by a plasma catalytic hybrid system, Appl. Catal. B: Environ. 74 (2007) 161–169. [11] M. Schiorlin, E. Marotta, M. Rea, C. Paradisi, Comparison of toluene removal in air at atmospheric conditions by different corona discharges, Environ. Sci. Technol. 43 (2009) 9386–9392. [12] E. Marotta, A. Callea, M. Rea, C. Paradisi, DC corona electric discharges for air pollution control. Part 1. Efficiency and products of hydrocarbon processing, Environ. Sci. Technol. 41 (2007) 5862–5868. [13] M.B. Chang, C.C. Lee, Destruction of formaldehyde with dielectric barrier discharge plasmas, Environ. Sci. Technol. 29 (1995) 181–186. [14] A. Ogata, H. Einaga, H. Kabashima, S. Futamura, S. Kushiyama, H. Kim, Effective combination of nonthermal plasma and catalysts for decomposition of benzene in air, Appl. Catal. B: Environ. 46 (2003) 87–95.

385

[15] H.X. Ding, A.M. Zhu, F.G. Lu, Y. Xu, J. Zhang, X.F. Yang, Low-temperature plasmacatalytic oxidation of formaldehyde in atmospheric pressure gas streams, J. Phys. D: Appl. Phys. 39 (2006) 3603–3608. [16] Y.F. Guo, X.B. Liao, J.H. He, W.J. Ou, D.Q. Ye, Effect of manganese oxide catalyst on the dielectric barrier discharge decomposition of toluene, Catal. Today 153 (2010) 176–183. [17] X. Fan, T.L. Zhu, M.Y. Wang, X.M. Li, Removal of low-concentration BTX in air using a combined plasma catalysis system, Chemosphere 75 (2009) 1301–1306. [18] X. Fan, T.L. Zhu, Y.J. Wan, X. Yan, Effects of humidity on the plasmacatalytic removal of low-concentration BTX in air, J. Hazard. Mater. 180 (2010) 616–621. [19] H.L. Chen, H.M. Lee, S.H. Chen, M.B. Chang, S.J. Yu, S.N. Li, Removal of volatile organic compounds by single-stage and two-stage plasma catalysis systems: a review of the performance enhancement mechanisms, current status, and suitable applications, Environ. Sci. Technol. 43 (2009) 2216–2227. [20] A.M. Harling, D.J. Glover, J.C. Whitehead, K. Zhang, The role of ozone in the plasma-catalytic destruction of environmental pollutants, Appl. Catal. B: Environ. 90 (2009) 157–161. [21] W.J. Liang, J. Li, J.X. Li, T. Zhu, Y.Q. Jin, Formaldehyde removal from gas streams by means of NaNO2 dielectric barrier discharge plasma, J. Hazard. Mater. 175 (2010) 1090–1095. [22] K. Hensel, J. Pawlat, K. Takashima, A. Mizuno, Possibilities of formaldehyde removal by discharge plasma, in: Proceedings of the International Joint Power Generation Conference, Atlanta, GA, USA, June 16–19, 2003, pp. 803–807. [23] K. Urashima, J.S. Chang, Removal of volatile organic compounds from air streams and industrial flue gases by non-thermal plasma technology, IEEE Trans. Dielectr. Electr. Insul. 7 (2000) 602–614. [24] N.Q. Yan, Z.C. Wu, T.E. Tan, Modeling of formaldehyde destruction under pulsed discharge plasma, J. Environ. Sci. Health A 35 (2000) 1951–1964. [25] N. Blin-Simiand, S. Pasquiers, F. Jorand, C. Postel, J.R. Vacher, Removal of formaldehyde in nitrogen and in dry air by a DBD: importance of temperature and role of nitrogen metastable states, J. Phys. D: Appl. Phys. 42 (2009), doi:10.1088/0022-3727/42/12/122003. [26] H.X. Ding, A.M. Zhu, X.F. Yang, C.H. Li, Y. Xu, Removal of formaldehyde from gas streams via packed-bed dielectric barrier discharge plasmas, J. Phys. D: Appl. Phys. 38 (2005) 4160–4167. [27] J. Van Durme, J. Dewulf, C. Leys, H. Van Langenhove, Combining non-thermal plasma with heterogeneous catalysis in waste gas treatment: a review, Appl. Catal. B: Environ. 78 (2008) 324–333. [28] W. Li, S.T. Oyama, Mechanism of ozone decomposition on a manganese oxide catalyst. 2. Steady-state and transient kinetic studies, J. Am. Chem. Soc. 120 (1998) 9047–9052. [29] W. Li, G.V. Gibbs, S.T. Oyama, Mechanism of ozone decomposition on a manganese oxide catalyst. 1. In situ Raman spectroscopy and ab initio molecular orbital calculations, J. Am. Chem. Soc. 120 (1998) 9041–9046.

Journal of Hazardous Materials 196 (2011) 386–394

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Biodegradation of chlorobenzoic acids by ligninolytic fungi ˇ Milan Muzikáˇr a,b , Zdena Kˇresinová a,c , Kateˇrina Svobodová a , Alena Filipová a , Monika Cvanˇ carová a,c , a a,∗ Kamila Cajthamlová , Tomáˇs Cajthaml a b c

Institute of Microbiology, Academy of Sciences of the Czech Republic, v.v.i., Vídenská 1083, CZ-142 20 Prague 4, Czech Republic ˇ Institute of Chemical Technology Prague, Faculty of Food and Biochemical Technology, Technická 5, CZ-160 28 Prague 6, Czech Republic Institute of Environmental Studies, Faculty of Science, Charles University, Benátská 2, CZ-128 01 Prague 2, Czech Republic

a r t i c l e

i n f o

Article history: Received 27 May 2011 Received in revised form 23 August 2011 Accepted 10 September 2011 Available online 16 September 2011 Keywords: Chlorobenzoic acid Polychlorinated biphenyls Biodegradation White rot fungi Irpex lacteus

a b s t r a c t We investigated the abilities of several perspective ligninolytic fungal strains to degrade 12 mono-, diand trichloro representatives of chlorobenzoic acids (CBAs) under model liquid conditions and in contaminated soil. Attention was also paid to toxicity changes during the degradation, estimated using two luminescent assay variations with Vibrio fischeri. The results show that almost all the fungi were able to efficiently degrade CBAs in liquid media, where Irpex lacteus, Pycnoporus cinnabarinus and Dichomitus squalens appeared to be the most effective in the main factors: degradation and toxicity removal. Analysis of the degradation products revealed that methoxy and hydroxy derivatives were produced together with reduced forms of the original acids. The findings suggest that probably more than one mechanism is involved in the process. Generally, the tested fungal strains were able to degrade CBAs in soil in the 85–99% range within 60 days. Analysis of ergosterol showed that active colonization is an important factor for degradation of CBAs by fungi. The most efficient strains in terms of degradation were I. lacteus, Pleurotus ostreatus, Bjerkandera adusta in soil, which were also able to actively colonize the soil. However, in contrast to P. ostreatus and I. lacteus, B. adusta was not able to significantly reduce the measured toxicity. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Chlorinated organic pollutants are a class of serious environmental contaminants because of their environmental persistence and ecotoxicity. Chlorinated benzoic acids (CBAs) are widespread environmental pollutants resulting primarily from microbial biodegradation of polychlorinated biphenyls (PCBs), reviewed, e.g., in Field and Alvarez [1], and some herbicides [2]. CBAs are significantly more soluble than their parent compounds and can therefore enter into the aqueous phase from the contaminated soil of polluted sites. Some mono-, di-, and tri-CBAs have been shown to cause genomic damage to tobacco plants [3], and to be toxic to aquatic organisms such as ciliate, Daphnia, algae and fish [4–6]. Several mono, di and trichlorinated isomers were also found to possess estrogenic-disrupting activity [7]. CBAs represent crucial recalcitrant metabolites on the biphenyl pathway during bacterial PCB transformation. Although it was found that CBAs are not very toxic toward bacteria, substantial negative effects of their presence on the bacterial transformation of PCBs have been reported

∗ Corresponding author. Tel.: +420 241062498; fax: +420 241062384. E-mail address: [email protected] (T. Cajthaml). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.041

[8,9]. Moreover, soil bacteria that co-metabolize PCBs via the main biphenyl upper pathway tend to accumulate CBAs as dead-end products because they are generally unable to further transform these substrates [10]. Another great limitation of organopollutant bacterial biodegradation is the fact that bacterial degrading enzymes are usually intracellular and the transfer of the pollutant into the bacterial cell represents an important limiting step. On the other hand, ligninolytic fungi, with their extracellular low-substrate-specificity enzymes, represent a promising alternative for biodegradation of various aromatic pollutants [11]. The ligninolytic system consists of three major peroxidases: lignin peroxidase (LiP), manganese peroxidase (MnP), versatile peroxidases and laccase, which belong among phenoloxidases. Their degradative abilities have been documented e.g., for chlorophenols, polycyclic aromatic hydrocarbons, PCBs, dioxins, furans, endocrine disrupters and others [12–15]. Moreover, the fungi were shown to be capable of splitting the aromatic rings of various persistent pollutants [14,16]. In contrast to a number of articles dealing with bacterial CBA degradation, only few papers have been published describing potential degradation of these compounds by fungi. Kamei et al. identified 4-CBA acid after degradation of 4,4 dichlorobiphenyl by Phanerochaete sp. MZ 142 and suggested its further transformation via a reductive pathway [14]. Other authors showed that ortho and meta mono-CBAs and benzoic acid (BA)

M. Muzikáˇr et al. / Journal of Hazardous Materials 196 (2011) 386–394

significantly induced the activities of cytochrom P-450 from Phanerochaete chrysosporium [17]. BA was then proven to be transformed by microsomes containing P-450 from the same fungus. The role of the monooxygenases of P-450 was clarified earlier by other authors, when the enzyme was heterologically expressed and hydroxyl benzoate and protocatechuic acid were detected as the degradation products of benzoate [18]. Generally, the data published in the literature document was efficiency of fungi to degrade PCBs and suggest possible transformation of PCBs to CBAs. CBAs are critical metabolites on the bacterial degradation pathway mainly, due to high specificity of individual bacterial enzymes. Therefore it is reasonable to investigate also CBA degradation abilities of ligninolytic fungi, especially when these organisms represent a promising alternative to bacterial PCB degradation applications. The aim of this work was to investigate the abilities of several promising ligninolytic fungal strains to transform 12 representatives of CBAs with various degree of chlorination (mono-, di-, tri-CBAs). The degradation performance was tested in model liquid nutrient media, where the production of CBA degradation products, the activities of ligninolytic enzymes and changes in the acute toxicity were also monitored. Moreover, the applicability of the fungi was also tested in an artificially contaminated soil, where toxicity was also monitored. 2. Materials and methods 2.1. Materials Standards and chemicals. 2-CBA; 2,3-CBA; 3,4-CBA; 3,5CBA; 2,3,5-CBA; 2,4,6-CBA and HPLC internal standard 2,3dichlorophenol were obtained from Sigma–Aldrich (Steinheim, Germany). 3-CBA; 4-CBA; 2,4-CBA; 2,5-CBA and 2,6-CBA were from Merck (Darmstadt, Germany). 2,3,6-CBA was purchased from Supelco (Steinheim, Germany). All the compounds were employed without further purification to prepare stock solutions in dimethyl formamide as described below. All the solvents were purchased from Merck, Germany or Chromservis (Prague, Czech Republic) and were of p.a. quality, trace analysis quality or gradient grade. All the chemicals used for the biochemical studies were from Sigma–Aldrich (Steinheim, Germany). 2.2. Microorganisms, inocula preparation and enzyme activities measurement Fungal cultures, inocula preparation and degradation experiments. All of the ligninolytic fungal strains used in this study (Irpex lacteus 617/93, Bjerkandera adusta 606/93, Phanerochaete chrysosporium ME 446, Phanerochaete magnoliae CCBAS 134/I, Pleurotus ostreatus 3004 CCBAS 278, Trametes versicolor 167/93, Pycnoporus cinnabarinus CCBAS 595, Dichomitus squalens CCBAS 750) were obtained from the Culture Collection of Basidiomycetes of the Academy of Science, Prague. Fungal inocula were grown under stationary conditions for 7 d at 28 ◦ C in 250 mL Erlenmeyer flasks containing 20 mL of either complex malt extract-glucose (MEG) medium or low-nitrogen mineral medium (LNMM). MEG medium (pH 5.5) contained 5 g malt extract broth (Oxoid, UK) and 10 g glucose per liter of distillated water and LNMM contained 2.4 mM diammonium tartrate [19]. The cultures were then homogenized with the Ultraturrax-T25 (IKA-Labortechnik, Staufen, Germany) and this suspension was used for inoculation in the degradation experiments. Enzyme determination. LiP (E.C. 1.11.1.14) was assayed with veratryl alcohol as the substrate [20] and MnP (E.C. 1.11.1.13) was determined with 2,6-dimethoxyphenol [21]. Laccase (Lac, E.C. 1.10.3.2) was estimated with

387

2,2-azinobis-3-ethylbenzo-thiazoline-6-sulfonic acid as the substrate [22]. Manganese-independent peroxidase (MIP) was calculated from the peroxidase activity of MnP assay detected in the absence of Mn2+ ions. One unit of enzyme produced 1 ␮mol of the reaction product per minute under the assay conditions at room temperature. 2.3. Degradation of CBAs in liquid media The degradation experiments in the liquid media were performed as static cultures, incubated in 250 mL Erlenmeyer flasks in five parallel experiments at 28 ◦ C. Twenty milliliters of the respective medium (MEG or LNMM) was inoculated with a 5% suspension of homogenized pre-inocula (1 mL) of the respective fungal strain. The cultures were immediately spiked with a solution of the CBAs in dimethyl formamide (100 ␮L). The final amount of each CBA was 200 ␮g per flask. The heat-killed controls were performed with oneweek growth of fungal cultures, which were killed in an autoclave before addition of the CBA solution. All of the cultures were incubated in the darkness at 28 ◦ C and harvested after 7, 14 and 21 days. 2.4. Fungal treatment of contaminated soil For preparation of the soil degradation experiment, 1.0 mL aliquots of a mycelia suspension of each fungal strain were added to 16 cm × 3.5 cm test-tubes containing 10 g of commercial straw pellets (ATEA Praha, Prague, Czech Republic), the moisture contents of which had been previously adjusted to 70% (w/w) and subsequently sterilized by autoclaving (121 ◦ C, 45 min). After inoculation, the cultures were closed with cotton–wool stoppers and then grown for 14 d at 28 ◦ C [23] The colonized substrate was then covered with a layer of soil (20 g), which had been previously artificially spiked with a mixture of CBAs in acetone. The relevant controls were prepared in the same way, however, without fungal inoculation. Main properties of the used sandy–loamy soil were as follows: total organic carbon 0.8%, total organics 1.4%, pH 5.3, water-holding capacity 31% and the granulometric composition was: sand 50.9%, fine sand 31.2%, silt 10.8%, clay 7.1%. The soil was air-dried and sieved through a 2-mm mesh before contamination and the final concentration of each CBA in the soil after contamination was 10 ␮g/g. The soil samples were then moistened to 15% humidity. The tubes were incubated at 28 ◦ C and the samples were harvested after 30 and 60 d. All the respective controls and samples were performed in five replicates. 2.5. Extraction and quantitative analyses of CBAs The whole content of each liquid culture was homogenized with Ultraturrax and acidified to approximately pH 2. It was then extracted with five 10 mL portions of ethyl acetate, the extracts were dried with sodium sulfate and concentrated using a rotary evaporator to a final volume of 10 mL. The extraction recoveries of all the CBAs were better than 95%. To enable HPLC analysis, an aliquot of the ethyl acetate extract was mixed with acetonitrile in a ratio of 1:10 (v/v), and the mixture was used for injection [16]. The soil samples were submitted to extraction using a Dionex 200 ASE extractor (Palaiseau, France). The soil samples (3 g) were mixed with sodium sulfate before extraction (v/v) and the extraction conditions were: 3 cycles; 150 ◦ C; 10.34 MPa; solvent system hexane–acetone, 1% acetic acid [24]. To avoid CBA volatilization, 500 ␮L of DMSO was added to the extracts as a solvent stopper and the extracts were concentrated using a vacuum rotary evaporator (60 kPa, 40 ◦ C) to approximately 1.5 mL. 50 ␮L of internal standard (IS, 2,3-dichlorophenol 0.9 mg/mL in ACN) was added to

388

M. Muzikáˇr et al. / Journal of Hazardous Materials 196 (2011) 386–394

each sample and the IS was used to calculate the volume extracts. The mixture was then directly injected into the HPLC. The quantitative analyses were performed using the Alliance Waters system (Prague, Czech Republic) equipped with a PDA detector and Empower software was used for data processing. Separation of the CBA mixture was performed on an XBridge C18 column (250 mm × 4.6 mm I.D., particle size 3.5 ␮m) from Waters (Prague, Czech Republic). The separation was carried out with a gradient (v/v) of acetonitrile (B) and water solution (A) of 0.1% TFA. The gradient program was as follows (min/%B): 0/17; 30/17; 60/34; 70/50. The applied flow rate was 0.8 mL min−1 and temperature was 35 ◦ C [24]. 2.6. Qualitative analyses of CBA degradation products Qualitative analysis of the CBA degradation products was performed in that the degradation intermediates were separated and characterized or identified by gas chromatography–mass spectrometry (GC–MS; 450-GC, 240-MS ion trap detector, Varian, Walnut Creek, CA). The ethyl acetate extracts were injected both directly without any derivatization and also after trimethylsilylation with N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA, Merck, Germany) and methylation with diazomethane [25]. The GC instrument was equipped with a split/splitless injector maintained at 240 ◦ C. DB-5MS column (Agilent, Prague, Czech Republic) was used for the separations (30 m, 0.25 mm I.D., 0.25 mm film thickness). The temperature program was started at 60 ◦ C and was held for 1 min in the splitless mode. Then the splitter was opened with ratio 1:50. The oven was heated to 120 ◦ C at a rate of 25 ◦ C/min with a subsequent temperature ramp up to 240 ◦ C at a rate of 2.5 ◦ C/min, where this temperature was maintained for 20 min. The solvent delay time was set at 5 min and the transfer line temperature was set at 240 ◦ C. The mass spectra were recorded at 3 scans s−1 under electron impact at 70 eV and mass range 50–450 amu. The excitation potential for the MS/MS product ion mode employed was 0.2 V and was increased to 0.8 V for more stable ions. Acetonitrile was used as the medium for chemical ionization (CI), where the ionization maximum time was 2000 and 40 ␮s for the reaction. 2.7. Analyses of ergosterol The total ergosterol was extracted and analyzed as described previously [23]. Briefly, samples (0.5 g) were sonicated with 3 mL of 10% KOH in methanol at 70 ◦ C for 90 min. Distilled water (1 mL) was added and the samples were extracted three times with 2 mL of cyclohexane, evaporated under nitrogen, redissolved in methanol and analyzed isocratically using a Waters Alliance HPLC system (Waters Milford, MA) equipped with a LiChroCart column filled with LiChrosphere® 100 RP-18e (250 × 4.0 mm; particle size 5 ␮m; ˚ equilibrated with 100% methanol at a flow rate pore size 100 A) of 1 mL min−1 . Ergosterol was detected at 282 nm and quantified with a 5-point calibration curve over a linear range from 0.5 to 50.0 ␮g/mL. 2.8. Toxicity assay The luminescent bacteria Vibrio fischeri (strain NRRL-B-11177), which was used for all the toxicity tests, were purchased freeze-dried from the supplier Ing. Musial (Czech Republic). The freeze-dried bacteria were rehydrated and stabilized in 2% (w/v) NaCl solution at 15 ◦ C for 1 h according to the standard procedure ISO 2007 [26]. An acute toxicity test of samples after degradation in liquid media was performed using the corresponding ethyl acetate extracts. Aliquots of the extracts (0.5 mL) were evaporated to dryness and dissolved again in dimethyl sulfoxide, which was directly applied to the test (2% of DMSO in the reaction mixture).

The amount of dimethyl sulfoxide varied between media, due to different sensitivities of the test toward the media matrix (see below). Three replicates for each sample were used to carry out the ecotoxicity test. The luminescence readings were obtained with a Lumino M90a luminometer (ZD Dolní Újezd, Czech Republic) at a temperature of 15 ± 0.2 ◦ C. The inhibition of bioluminescence was recorded after 15-min exposure. The toxicity of soil samples was measured by a kinetic Flash assay using the luminescent bacterium [27,28]. The samples were prepared by weighing 1.5 g dried soil and 6 mL 2% (w/v) NaCl solution. The sample suspension was mixed continuously and 0.5 mL was placed into the measuring cuvette. The contents of the measuring cuvette were mixed continuously by adapted luminometer LUMINO M90a and 0.5 mL of the bacterial solution was dispensed into the sample. The signal was recorded permanently for 60 s. The light inhibition was calculated as the difference between the height of the peak that was observed immediately after addition of the bacteria to the sample and the luminescence intensity after a contact time of 60 s.

3. Results and discussion 3.1. Degradation of CBAs in liquid cultures The representatives of mono, di and tri-CBAs that were tested in this study were employed at a relatively high concentration of 10 ␮g/mL. The fact that the compounds are partially soluble in water and their acute toxic properties were confirmed by the observation of fungal biomass development. Generally, the fungal strains were partly affected by CBAs and their biomass reached about 50–70% compared to non toxic controls (data not shown). This finding is in agreement with the observation of Dittmann et al. who tested the development of the mycelia of fungal strains in two liquid nutrient media after the addition of various concentrations of 3-CBA [29]. In contrast to this observation, the fungal strains in our study were very efficient in degradation of CBAs in the liquid media. The time course of the individual CBA degradation in both media is presented in Tables 1 and 2. The results clearly demonstrate that all of the tested strains were at least partially able to transform CBAs. I. lacteus, P. cinnabarinus and D. squalens were found to be the most efficient degraders in complex MEG media. P. cinnabarinus and D. squalens were able to degrade about 78% and 73% of total CBA, respectively, while I. lacteus degraded 92% of total CBAs in complex media compared to the heat-killed controls. Particularly I. lacteus removed all of the CBAs from the media except 2,6-CBA and 2,3,6-CBA, which were degraded to approximately 50% of original amount, while P. cinnabarinus did not significantly transform 2,6-CBA, 2,3,6-CBA, 2,4,6-CBA and D. squalens did not significantly degrade 2,3,6-CBA (ANOVA, P = 0.05). B. adusta exhibited the poorest degradation ability in both the liquid media. The most efficient strains in the LNNM media were once again found to be I. lacteus, P. cinnabarinus and D. squalens. All of the three most efficient strains were able to transform CBAs with percent removals ranging from 76% to 77%. 2,6-CBA, 2,3,6-CBA and 2,4,6-CBA again appeared to be the most recalcitrant compounds where D. squalens did not significantly degrade 2,6-CBA and 2,4,6-CBA; P. cinnabarinus – 2,6-CBA, 2,3,6-CBA and 2,4,6-CBA; I. lacteus – 2,6-CBA and 2,3,6-CBA. These results suggest a possible connection between the substituted ortho and para positions and persistency toward the fungal degradation mechanism. Only a few publications in journals deal with transformation of CBAs by fungi. The above-mentioned work of Dittmann et al. also included the degradation of 3-CBA by P. chrysosporium, P. ostreatus, Heterobasidion annosum and two other ectomycorrhizal fungi [29]. However, in contrast to our results, the authors observed only limited degradation of the compound in the range of several

M. Muzikáˇr et al. / Journal of Hazardous Materials 196 (2011) 386–394

389

Table 1 Residual CBA amounts in heat-killed controls and after incubation of the tested fungal strains in LNNM media (ND: not detected). Amount of CBA (␮g per flask) LNNM – 7 days Control B. adusta D. squalens I. lacteus P. chrysosporium P. magnoliae P. ostreatus P. cinnabarinus T. versicolor LNNM – 14 days Control B. adusta D. squalens I. lacteus P. chrysosporium P. magnoliae P. ostreatus P. cinnabarinus T. versicolor LNNM – 21 days Control B. adusta D. squalens I. lacteus P. chrysosporium P. magnoliae P. ostreatus P. cinnabarinus T. versicolor

2-CBA

3-CBA

4-CBA

2,3-CBA

2,4-CBA

2,5-CBA

2,6-CBA

3,4-CBA

3,5-CBA

2,3,5-CBA

2,3,6-CBA

2.4.6-CBA

196 ± 4 156 ± 5 ND ND ND 27 ± 2 156 ± 9 ND 73 ± 39

207 ± 4 53 ± 12 33 ± 4 19 ± 6 ND ND 28 ± 3 ND 25 ± 1

207 ± 2 41 ± 34 ND ND ND ND ND ND ND

195 ± 3 174 ± 5 ND 45 ± 7 94 ± 14 133 ± 2 162 ± 8 ND 25 ± 3

207 ± 2 156 ± 31 ND 66 ± 2 ND 74 ± 4 105 ± 17 ND 104 ± 8

193 ± 3 141 ± 11 ND 85 ± 5 92 ± 15 126 ± 6 167 ± 8 ND 46 ± 3

205 ± 2 197 ± 8 211 ± 15 179 ± 18 196 ± 5 189 ± 3 195 ± 5 190 ± 8 242 ± 11

182 ± 8 64 ± 19 ND ND ND ND 87 ± 60 ND ND

196 ± 1 77 ± 28 ND ND ND 165 ± 2 55 ± 1 ND ND

165 ± 2 161 ± 2 ND 153 ± 2 136 ± 10 ND 124 ± 46 ND ND

199 ± 1 185 ± 7 ND 176 ± 21 177 ± 7 184 ± 8 172 ± 14 176 ± 12 206 ± 14

185 ± 0 176 ± 8 207 ± 6 128 ± 11 171 ± 6 180 ± 4 163 ± 13 173 ± 7 140 ± 4

170 ± 15 124 ± 5 ND ND ND ND 105 ± 1 ND 66 ± 13

173 ± 14 45 ± 2 56 ± 4 ND ND ND ND ND 47 ± 4

184 ± 21 ND ND ND ND ND ND ND ND

168 ± 19 157 ± 22 126 ± 8 13 ± 2 25 ± 1 100 ± 11 132 ± 1 ND ND

208 ± 17 85 ± 5 ND ND ND 60 ± 19 15 ± 2 ND ND

167 ± 18 114 ± 7 ND ND 40 ± 1 77 ± 13 159 ± 3 ND ND

181 ± 16 198 ± 14 189 ± 11 173 ± 7 188 ± 12 182 ± 15 193 ± 11 184 ± 6 186 ± 34

156 ± 17 46 ± 0 ND ND ND ND ND ND ND

165 ± 19 ND ND ND ND 136 ± 23 ND ND ND

138 ± 19 145 ± 5 ND 103 ± 3 92 ± 0 ND 114 ± 6 ND ND

169 ± 17 187 ± 7 ND 171 ± 9 175 ± 12 153 ± 32 188 ± 5 163 ± 8 183 ± 3

159 ± 21 179 ± 10 175 ± 11 95 ± 4 165 ± 8 151 ± 27 161 ± 13 136 ± 41 136 ± 3

196 ± 5 112 ± 10 ND ND ND ND 70 ± 13 ND 76 ± 5

195 ± 9 54 ± 5 71 ± 1 ND ND ND ND ND 32 ± 15

204 ± 8 ND ND ND ND ND ND ND ND

189 ± 3 167 ± 8 105 ± 3 19 ± 2 20 ± 2 24 ± 5 101 ± 19 ND ND

208 ± 6 89 ± 2 ND 22 ± 0 ND 35 ± 5 ND ND ND

191 ± 5 106 ± 5 ND ND 40 ± 6 20 ± 2 130 ± 25 ND ND

203 ± 6 201 ± 14 186 ± 5 170 ± 1 185 ± 24 176 ± 9 182 ± 36 196 ± 7 195 ± 13

180 ± 15 ND ND ND ND ND ND ND ND

188 ± 5 ND ND ND ND ND ND ND ND

170 ± 7 130 ± 8 ND 67 ± 7 98 ± 4 129 ± 7 105 ± 18 ND ND

195 ± 7 190 ± 13 ND 183 ± 1 180 ± 18 152 ± 12 179 ± 39 175 ± 9 194 ± 8

183 ± 7 182 ± 12 191 ± 4 71 ± 6 168 ± 13 153 ± 11 157 ± 30 176 ± 8 134 ± 11

percent, even though, in one case, the authors employed a similar concentration (15.6 mg/L) to that used in our study (20 mg/L). In order to employ the toxicity test, we diluted the samples from the two media in different ways. The theoretical (original) concentrations of the individual CBAs in the reaction mixture for MEG and

LNNM media samples were 0.5 and 0.25 ␮g/mL, respectively. Since we detected only a decrease in the toxicity in preliminary tests, the dilution of the samples was set to reach about 90% inhibition for the controls. The evaluation of the acute toxicity test with V. fischeri was performed by comparison of the inhibition of the sample

Table 2 Residual CBA amounts in heat-killed controls and after incubation of the tested fungal strains in MEG media (ND: not detected). Amount of CBA (␮g per flask) MEG – 7 days Control B. adusta D. squalens I. lacteus P. chrysosporium P. magnoliae P. ostreatus P. cinnabarinus T. versicolor MEG – 14 days Control B. adusta D. squalens I. lacteus P. chrysosporium P. magnoliae P. ostreatus P. cinnabarinus T. versicolor MEG – 21 days Control B. adusta D. squalens I. lacteus P. chrysosporium P. magnoliae P. ostreatus P. cinnabarinus T. versicolor

2-CBA

3-CBA

4-CBA

2,3-CBA

2,4-CBA

2,5-CBA

2,6-CBA

3,4-CBA

3,5-CBA

2,3,5-CBA

2,3,6-CBA

2,4,6-CBA

211 ± 4 201 ± 20 ND ND ND ND 71 ± 9 ND 90 ± 22

195 ± 4 138 ± 64 97 ± 6 ND ND 44 ± 4 ND ND ND

209 ± 2 162 ± 69 ND ND ND ND ND ND ND

197 ± 2 200 ± 8 ND ND 52 ± 5 56 ± 7 127 ± 7 ND 42 ± 2

207 ± 6 192 ± 25 131 ± 11 46 ± 2 ND ND 66 ± 5 ND 89 ± 2

195 ± 6 189 ± 17 ND 70 ± 27 ND 47 ± 10 145 ± 9 ND 39 ± 1

206 ± 4 200 ± 9 205 ± 20 204 ± 4 209 ± 7 213 ± 7 214 ± 14 201 ± 12 202 ± 7

195 ± 5 154 ± 39 ND ND ND ND ND ND ND

199 ± 4 175 ± 32 ND ND ND ND ND ND ND

175 ± 2 166 ± 13 ND 86 ± 5 ND 76 ± 6 132 ± 10 ND ND

172 ± 6 177 ± 8 151 ± 1 151 ± 4 167 ± 8 172 ± 7 156 ± 11 167 ± 18 154 ± 2

193 ± 5 193 ± 7 160 ± 13 114 ± 11 185 ± 6 188 ± 5 150 ± 14 190 ± 15 174 ± 3

210 ± 30 208 ± 3 ND ND ND ND 37 ± 9 ND 84 ± 16

186 ± 32 191 ± 2 ND ND ND 34 ± 1 ND ND ND

197 ± 24 211 ± 1 213 ± 3 ND ND ND ND ND ND

199 ± 27 162 ± 51 101 ± 6 ND ND 40 ± 8 91 ± 19 ND ND

206 ± 26 203 ± 7 124 ± 6 ND ND 39 ± 5 58 ± 3 ND ND

195 ± 21 194 ± 1 ND ND ND 60 ± 15 117 ± 19 ND ND

212 ± 25 202 ± 12 ND 98 ± 10 226 ± 4 221 ± 2 231 ± 11 188 ± 17 197 ± 5

176 ± 26 175 ± 3 ND ND ND ND ND ND ND

181 ± 25 186 ± 1 ND ND ND ND ND ND ND

164 ± 28 151 ± 10 ND ND ND 86 ± 17 115 ± 15 ND ND

178 ± 26 148 ± 28 145 ± 3 107 ± 12 161 ± 5 162 ± 4 170 ± 7 151 ± 25 150 ± 7

193 ± 24 192 ± 1 166 ± 1 ND 181 ± 2 189 ± 11 150 ± 4 170 ± 23 169 ± 6

209 ± 15 140 ± 53 ND ND ND ND 22 ± 1 ND 121 ± 30

175 ± 17 47 ± 2 ND ND ND ND ND ND ND

188 ± 11 42 ± 9 197 ± 5 ND 93 ± 19 29 ± 8 ND ND 136 ± 22

186 ± 13 173 ± 26 ND ND ND 55 ± 9 86 ± 6 ND ND

193 ± 9 133 ± 46 74 ± 14 ND ND 46 ± 1 61 ± 4 ND ND

182 ± 8 138 ± 54 54 ± 9 ND ND 52 ± 3 126 ± 14 ND ND

213 ± 9 200 ± 6 ND 88 ± 20 194 ± 23 236 ± 3 200 ± 13 187 ± 27 206 ± 11

172 ± 10 80 ± 13 ND ND 59 ± 5 56 ± 0 ND ND ND

177 ± 9 105 ± 10 ND ND 75 ± 4 ND ND ND ND

157 ± 6 137 ± 15 ND ND ND 82 ± 1 119 ± 12 ND ND

166 ± 12 170 ± 4 140 ± 2 96 ± 9 142 ± 19 174 ± 3 187 ± 23 150 ± 25 160 ± 5

181 ± 8 183 ± 7 151 ± 14 ND 167 ± 14 195 ± 2 171 ± 18= 162 ± 34 172 ± 6

390

M. Muzikáˇr et al. / Journal of Hazardous Materials 196 (2011) 386–394

Table 3 Luminescence inhibition in heat-killed controls and after incubation of the tested fungal strains in MEG and LNNM media after 21 days of incubation. Luminescence inhibition (%)

LNMM

Control B. adusta D. squalens I. lacteus P. chrysosporium P. magnoliae P. ostreatus P. cinnabarinus T. versicolor

89.4 72.1 11.9 27.2 44.9 41.9 27.5 11.3 14.5

± ± ± ± ± ± ± ± ±

MEG 8.2 4.3 8.5 13.9 15.4 10.1 8.3 2.7 4.0

94.5 85.9 35.0 7.4 38.2 52.8 34 .8 20.8 36.3

COOCH3 COOCH3

COOH ± ± ± ± ± ± ± ± ±

1.0 12.5 9.4 3.0 30.7 17.7 2.8 5.6 7.5

Clx COOH

Clx

CHO HO luminescence with their respective controls (Table 3). As mentioned above, the toxicity test revealed that the tested fungi were generally able to decrease the measured acute toxicity, suggesting that the degradation products of CBAs were either not accumulated or they were less toxic than the original CBAs.

HO

Clx COOCH3

CH3O

Clx

Clx

Clx

CHO

CH2OH 3.2. Detection of CBA degradation products The CBA metabolites were analyzed with GC–MS and their structures were suggested by comparing the mass spectra with the data in the NIST 08 library and independently by interpreting the fragmentation pattern. Additionally, unknown structures of metabolites were explored using MS/MS (product ion scan) to clarify the fragmentation sequence. The mass spectral characteristics of the detected CBA degradation products are listed in Table 4. Some of the metabolites were detected after trimethylsilylation (e.g., chlorobenzyl alcohols) and several of them were confirmed by comparison with the available chemical standards. All of the intermediates were detected at only trace levels, suggesting that none of them were accumulated during degradation. The group of the detected intermediates includes chlorobenzaldehydes, chlorobenzyl alcohols, chlorobenzoic acid methyl esters and the methoxy or hydroxy derivatives of these structures. The metabolites were found in various fungal strain cultures when representatives of monochloro and dichloro benzaldehydes and alcohols were found in all cultures, as well as methyl esters of di-CBA. Methyl ester representative of tri-CBA was detected only in the culture of I. lacteus, however, methoxy derivatives of tri-CBA and di-CBA methyl esters were found in all fungi. Trichlorinated hydroxybenzyl alcohols were detected in all fungal cultures too. A possible scheme of CBA fungal degradation pathway constructed of the detected metabolites is shown in Fig. 1. The results generally correspond to the results of Kamei et al. [14], who studied the transformation of 4,4 -dichlorobiphenyl by Phanerochaete sp. MZ 142, where these authors detected the formation of 4-CBA, the methyl ester of 4-CBA and further reduced transformation products: 4-chlorobenzyl alcohol and 4-chlorobenzaldehyde. Such a reduction mechanism could be explained by the action of an intracellular aryl alcohol oxidase system [30]. Matsuzaki et al. showed that the enzymes that are probably involved in the transformation, i.e. aryl alcohol dehydrogenase, aryl aldehyde dehydrogenase and also cytochrome P-450 of P. chrysosporium were up-regulated after the addition of BA to the fungal culture [31]. The other types of transformation products, i.e. hydroxyl and methoxy derivatives, which were found in our study, have already been described by Matsuzaki and Wariishi following transformation of BA by P. chrysosporium [18]. The detected metabolites include 4-hydroxy, 2-hydroxy and 4-hydroxy-2methoxy derivatives. In another work, the authors demonstrated that heterologously expressed P-450 cytochromes from the CYP53 family of P. chrysosporium, Aspergilus niger and Rhodotorula minuta were able to hydroxylate BA at the 4-position [32]. P-450-mediated

CH3O

Clx

CH2OH

Clx

CH2OH HO

Clx

CH3O

Clx

Fig. 1. Proposed pathway of CBA degradation by ligninolytic fungi.

hydroxylation of BA at other positions in fungi has not been reported to date. Moreover, the authors employed quantitative PCR to demonstrate that the expression of the cytochrome is regulated by the presence of BA at the transcription level. The induction of cytochrome P-450 by BA and also by 3 and 4-CBA has been published elsewhere [17]. The measurement of fungal extracellular ligninolytic activities in this study demonstrated that most of the activities were suppressed or the maxima activity peaks were delayed during cultivation by the presence of CBAs. Only rare cases when the situation was different were recorded for the activities of MnP and laccase of T. versicolor, which were significantly induced in MEG and LNNM media, respectively. Particularly the laccase activity increased from 20 U/L to 230 U/L. These findings indirectly confirm that ligninolytic enzymes need not play a key role in the degradation of CBAs. 3.3. Degradation of CBAs in soil The soil degradation experiment was monitored after 30 and 60 days and the residual concentrations after the application of the fungal strains are depicted in Fig. 2. The results show that, except for strains P. cinnabarinus and T. versicolor, which degraded only 30% and 39% of the total CBA, respectively, within 60 days of incubation, all the other strains under study were able to substantially remove CBAs from soil in the range of 85–99% of total CBA. The results are partially contrary to the experiments in liquid cultures, because P. cinnabarinus belonged among the most degrading strains in both the liquid media. On the other hand, B. adusta appeared to be effective in soil while this strain belonged among the less degrading in the liquid media. I. lacteus was found to be the most efficient

Table 4 Retention data and electron impact mass spectral characteristics of CBA metabolites. MW according to CI

m/z of fragment ions (relative intensity)

Structural suggestion

5.431 5.494 5.582 7.603 7.724 7.913 8.329 8.532 8.875 9.445 9.685 10.883 11.203 11.729 11.894 12.308

140 140 140 174 174 174 174 174 214 214 214 204 204 204 204 204

o-Chlorobenzaldehyde m-Chlorobenzaldehyde p-Chlorobenzaldehyde 3,5-Dichlorobenzaldehyde 2,4-Dichlorobenzaldehyde 2,5-Dichlorobenzaldehyde 2,3-Dichlorobenzaldehyde 3,4-Dichlorobenzaldehyde TMS p-chlorobenzyl alcohol TMS m-chlorobenzyl alcohol TMS o-chlorobenzyl alcohol 2,6-Dichlorobenzoic acid methyl ester 3,5-Dichlorobenzoic acid methyl ester 2,4-Dichlorobenzoic acid methyl ester 2,5-Dichlorobenzoic acid methyl ester 3,4-Dichlorobenzoic acid methyl ester

12.651

204

12.945 13.139 13.27 13.604 14.091

170 248 248 248 238

14.288 15 15.801

248 248 238

16.391

238

18.625

310

18.699

298

18.741

310

19.695

234

19.82 20.46 20.592

186 298 310

20.688 21.057 21.526 22.693 23.481

298 298 234 298 234

27.313 29.207

268 312

142 (23.8), 141 (36.6), 140 (73.7), 139 (99.9), 111 (55.0), 75 (32.0), 51 (19.6), 50 (29.8) 142 (20.6), 141 (36.1), 140 (66.6), 139 (99.9), 113 (18.4), 77 (22.7), 75 (33.5), 74 (19.1) 142 (16.0), 141 (37.1), 140 (49.4), 139 (99.9), 113 (16.8), 111 (49.6), 77 (15.1), 74 (16.8) 176 (62.4), 174 (70.3), 173 (99.9), 145 (47.0), 139 (54.2), 111 (50.1), 75 (61.1), 74 (52.6) 176 (39.5), 175 (70.2), 174 (61.4), 173 (99.9), 147 (16.9), 145 (25.9), 75 (18.0), 74 (15.0) 176 (38.2), 175 (68.4), 174 (61.3), 173 (99.9), 111 (25.0), 75 (61.0), 74 (45.9), 50 (25.3) 176 (37.5), 175 (69.0), 174 (62.8), 173 (99.9), 147 (21.9), 145 (37.4), 75 (36.4), 74 (26.3) 176 (38.9), 175 (69.5), 174 (64.9), 173 (99.9), 147 (28.1), 145 (43.0), 75 (29.1), 74 (24.9) 201 (34.0), 199 (99.9), 179 (18.5), 163 (30.7), 127 (25.4), 125 (82.5), 89 (25.1), 73 (18.6) 201 (31.8), 199 (90.9), 179 (33.4), 171 (19.9), 169 (60.1), 127 (30.2), 125 (99.9), 89 (32.7) 201 (20.8), 199 (58.7), 179 (24.2), 169 (20.5), 127 (31.9), 125 (99.9), 89 (24.2), 73 (12.5) 206 (32.5), 204 (43.4), 177 (9.9), 175 (63.4), 173 (100), 147 (9.6), 145 (7.9), 109 (7.8), 75 (33.9) 208 (2.0), 206 (14.1), 204 (20.5), 177 (8.9), 175 (63.1), 173 (100), 147 (21.1), 145 (33.2), 109 (17.8), 75 (16.0) 208 (5.4), 206 (12.9), 204 (20.7), 177 (9.9), 175 (61.1), 173 (100), 147 (15.5), 145 (29.0), 109 (16.2), 75 (95.6) 208 (3.7), 206 (18.3), 204 (25.7), 177 (11.9), 175 (62.1), 173 (93.3), 147 (6.4), 145 (19.3), 109 (17.1), 75 (100) 208 (2.7), 206 (19.8), 204 (32.4), 177 (10.3), 175 (64.6), 173 (100), 147 (19.7), 145 (35.1), 109 (22.9), 74 (27.6) 208 (1.3), 206 (16.3), 204 (19.0), 177 (19.0), 175 (56.9), 173 (100), 149 (29.4), 147 (47.7), 145 (19.0), 109 (14.4), 75 (97.4) 172 (19.4), 171 (35.0), 170 (56.3), 169 (100), 141 (6.8), 126 (13.6), 111 (11.7), 77 (15.5) 235 (67.6), 233 (99.9), 205 (18.5), 203 (25.6), 161 (41.4), 159 (64.2), 123 (18.4), 103 (18.9) 235 (58.0), 233 (84.2), 161 (61.1), 159 (99.9), 123 (13.8), 103 (29.1), 73 (12.7) 235 (70.9), 233 (99.9), 205 (32.3), 203 (45.8), 161 (58.5), 159 (89.7), 147 (27.7), 123 (21.1) 242 (6.4), 240 (19.3), 238 (19.7), 211 (29.3), 209 (95.0), 207 (100), 183 (3.4), 181 (14.1), 179 (14.5), 143 (11.0), 109 (12.8), 74 (12.9) 235 (69.6), 233 (99.9), 205 (17.8), 203 (25.3), 161 (46.3), 159 (71.1), 123 (17.5), 103 (22.8) 235 (68.5), 233 (94.6), 203 (16.7), 161 (67.3), 159 (99.9), 75 (27.5), 73 (28.5), 59 (34.2) 242 (7.1), 240 (25.7), 238 (25.1), 211 (31.5), 209 (98.1), 207 (100), 183 (5.7), 181 (12.6), 179 (14.5), 143 (10.5), 109 (11.7), 74 (14.1) 242 (9.3), 240 (29.4), 238 (29.5), 211 (29.7), 209 (95.4), 207 (100), 183 (6.8), 181 (21.6), 179 (20.7), 143 (14.2), 109 (15.6), 74 (14.5) 271 (32.4), 269 (100), 267 (76.8), 241 (15.8), 239 (45.5), 237 (42.8), 197 (13.7), 195 (46.6), 193 (44.2), 157 (14.2), 125 (5.5), 123 (15.8), 93 (15.9) 285 (33.2), 283 (100), 281 (99.5),239 (49.1), 237 (50.6), 209 (80.6), 207 (83.3), 205 (2.8), 165 (13.3), 167 (33.3) 271 (32.8), 269 (100), 267 (87.2), 241 (13.0), 239 (38.3), 237 (37.2), 197 (22.0), 195 (68.5), 193 (65.9), 157 (14.6), 125 (5.6), 123 (15.7), 93 (28.1) 238 (4.3), 236 (41.1), 234 (67.2), 207 (15.4), 205 (84.6), 203 (100), 162 (14.1), 160 (18.6), 111 (15.0), 97 (31.9) 188 (7.2), 186 (25.8), 157 (31.6), 155 (100), 127 (17.8), 99 (13.7) 285 (31.5), 283 (100), 281 (98.0), 239 (73.8), 237 (36.1), 209 (68.2), 207 (68.5), 167 (12.6),165 (33.8) 271 (35.5), 269 (100), 267 (99.1), 241 (10.0), 239 (31.1), 237 (31.1), 197 (26.8), 195 (81.5), 193 (82.1), 157 (16.7), 125 (6.4), 123 (18.0), 93 (39.2) 285 (31.6), 283 (100), 281 (95.3), 239 (73.8), 237 (34.6), 209 (47.7), 207 (43.3), 167 (14.8),165 (33.0) 285 (34.4), 283 (100), 281 (98.5), 239 (36.8), 237 (32.4), 209 (51.5), 207 (54.4), 167 (13.2),165 (35.6) 238 (3.0), 236 (22.3), 234 (30.4), 207 (10.2), 205 (68.4), 203 (100), 162 (5.3), 160 (8.3), 111 (11.2), 97 (17.0) 285 (26.1), 283 (100), 281 (90.9), 239 (52.1), 237 (49.7), 209 (74.2), 207 (64.2), 167 (19.7),165 (48.1) 238 (5.1), 236 (21.3), 234 (29.5), 207 (11.5), 205 (70.5), 203 (100), 162 (10.9), 160 (11.9), 111 (15.4), 97 (19.9) 270 (15.1), 268 (15.3), 241 (28.3), 239 (95.8), 237 (100) 312 (12.3), 314 (13.6), 301 (42.6), 299 (100), 297 (97,3), 271 (24.3), 269 (76.7), 267 (73.4), 227 (31.8), 225 (75.6), 223 (88.5)

Type of derivatization

Trimethylsilylation Trimethylsilylation Trimethylsilylation Trimethylsilylation

2,3-Dichlorobenzoic acid methyl ester ?-Chloro-?-methoxybenzaldehyde TMS 2,5-dichlorobenzyl alcohol TMS 2,4-dichlorobenzyl alcohol TMS 3,5-dichlorobenzyl alcohol 2,4,6-Trichlorobenzoic acid methyl ester

Trimethylsilylation Trimethylsilylation Trimethylsilylation Trimethylsilylation

TMS 2,3-dichlorobenzyl alcohol TMS 3,4-dichlorobenzyl alcohol 2,3,6-Trichlorobenzoic acid methyl ester

Trimethylsilylation Trimethylsilylation Trimethylsilylation

2,3,5-Trichlorobenzoic acid methyl ester TMS ?,?,?-trichlorobenzyl alcohol

Trimethylsilylation

TMS ?,?,?-trichloro-?-hydroxybenzyl alcohol

Trimethylsilylation

TMS ?,?,?-trichlorobenzyl alcohol

Trimethylsilylation

M. Muzikáˇr et al. / Journal of Hazardous Materials 196 (2011) 386–394

tR (min)

?,?-Dichloro-?-methoxybenzoic acid methyl ester ?-Chloro-?-hydroxybenzoic acid methyl ester TMS ?,?,?-trichloro-?-hydroxybenzyl alcohol TMS ?,?,?-trichlorobenzyl alcohol

Methylation Trimethylsilylation Trimethylsilylation

TMS ?,?,?-trichloro-?-hydroxybenzyl alcohol TMS ?,?,?-trichloro-?-hydroxybenzyl alcohol ?,?-Dichloro-?-methoxybenzoic acid methyl ester TMS?,?,?-trichloro-?-hydroxybenzyl alcohol ?,?-Dichloro-?-methoxybenzoic acid methyl ester

Trimethylsilylation Trimethylsilylation

?,?,?-Trichloro-?-methoxybenzoic acid methyl ester TMS ?,?,?-trichloro-?-methoxybenzyl alcohol

Trimethylsilylation

Trimethylsilylation

391

392

M. Muzikáˇr et al. / Journal of Hazardous Materials 196 (2011) 386–394

Fig. 2. Residual concentrations of CBAs in contaminated soil after incubation with the tested fungal strains: A – 30 days; B – 60 days.

strain in soil, where this fungus had already depleted 98% of the CBA within 30 days. One probable explanation for the discrepancy between these results and model conditions in the liquid media and the soil degradation experiment lies in the different abilities of fungi to penetrate into contaminated soil [33]. Therefore, we tried to estimate the relative amount of fungal biomass using the analysis of ergosterol in soil samples with CBAs and also without the addition of pollutants (Fig. 3). Despite the high variability of the data, the results indicate that the fungi that showed the highest CBA depletion (I. lacteus, P. ostreatus, and B. adusta), were also the strains that were able to significantly colonize the contaminated soil. The only exception was P. magnolia, where we detected a

significantly lower amount of ergosterol despite the high removal of CBAs (99% within 60 days). The parameters of kinetic Flash toxicity assay were adjusted to also include recording of a possible increase in toxicity. The data obtained from this test are depicted in Fig. 4. The best results, in terms of inhibition reduction, were obtained with strains I. lacteus and P. ostreatus, corresponding to their CBA degradation efficiencies. On the other hand, unexpected results were observed with strains B. adusta and P. magnolia where, in spite of their high degradation rate, the detected residual toxicity was not significantly different from the controls (t-test, P < 0.05). These results of Flash assay are in agreement with the results from toxicity estimation in

Fig. 3. Ergosterol concentrations in non-contaminated soil and in soil contaminated by CBAs after 30 and 60 days of incubation.

M. Muzikáˇr et al. / Journal of Hazardous Materials 196 (2011) 386–394

393

Fig. 4. Luminescence inhibition of the Flash assay in contaminated soil (control) and in soil with the tested fungal strains after 60 days of incubation.

the liquid cultures, where a residual toxicity in these fungal cultures was also detected. This could possibly be explained by the formation and accumulation of toxic metabolites and, probably for the same reason, significantly elevated toxicity was observed for T. versicolor. 4. Conclusion To the best of our knowledge, this is the first paper providing a general description of the ability of ligninolytic fungi to biodegrade CBAs that represent crucial toxic and highly persistent metabolites on bacterial biodegradation pathways of polychlorinated biphenyls. The ability of the fungi has been examined under liquid conditions and also verified in contaminated soil. The tested fungal strains were able to degrade CBAs in soil in the 85–99% range within 60 days when I. lacteus was found to be the most efficient degrading strain under both of the tested conditions. Several new degradation products have been identified when mainly methoxy and hydroxy derivatives were produced together with reduced forms of the original acids. The results show that the fungi are probably able to transform CBAs via several pathways with significant reduction of toxicity during the process. The promising degradation results from this study emphasize the need for further research, especially to identify the participation of different enzymatic machineries, in order to improve the understanding of the degradation mechanisms. The results for the liquid media and from the consequent soil experiment show that the presence of a bioremediative organism is of key importance; however, in soil, i.e. under conditions with limited pollutant bioavailability, active colonization of the soil is of equal importance. Acknowledgments This work was supported by grants no. 2B06156 of the Ministry of Education, Youth and Sports of the Czech Republic and no. 525/09/1058 of the Science Foundation of the Czech Republic and by Institutional Research Concept No. AV0Z50200510. References [1] J.A. Field, R.S. Alvarez, Microbial transformation and degradation of polychlorinated biphenyls, Environ. Pollut. 155 (2008) 1–12.

[2] S.R. Sorensen, M.S. Holtze, A. Simonsen, J. Aamand, Degradation and mineralization of nanomolar concentrations of the herbicide dichlobenil and its persistent metabolite 2,6-dichlorobenzamide by Aminobacter spp. isolated from dichlobenil-treated soils, Appl. Environ. Microb. 73 (2007) 399–406. [3] T. Gichner, P. Lovecká, B. Vrchotová, Genomic damage induced in tobacco plants by chlorobenzoic acids—metabolic products of polychlorinated biphenyls, Mutat. Res. 657 (2008) 140–145. [4] M. Muccini, A.C. Layton, G.S. Sayler, T.W. Schultz, Aquatic toxicities of halogenated benzoic acids to Tetrahymena pyriformis, B. Environ. Contam. Tox. 62 (1999) 616–622. [5] Y.H. Zhao, G.D. Ji, M.T.D. Cronin, J.C. Dearden, QSAR study of the toxicity of benzoic acids to Vibrio fischeri, Daphnia magna and carp, Sci. Total Environ. 216 (1998) 205–215. [6] P.Y. Lee, C.Y. Chen, Toxicity and quantitative structure–activity relationships of benzoic acids to Pseudokirchneriella subcapitata, J. Hazard. Mater. 165 (2009) 156–161. [7] K. Svobodová, M. Plaˇcková, V. Novotná, T. Cajthaml, Estrogenic and androgenic activity of PCBs, their chlorinated metabolites and other endocrine disruptors estimated with two in vitro yeast assays, Sci. Total. Environ. 407 (2009) 5921–5925. [8] D.H. Pieper, M. Seeger, Bacterial metabolism of polychlorinated biphenyls, J. Mol. Microbial. Biotech. 15 (2008) 121–138. [9] S.A. Adebusoye, F.W. Picardal, M.O. Ilori, O.O. Amund, Influence of chlorobenzoic acids on the growth and degradation potentials of PCB-degrading microorganisms, World J. Microb. Biot. 24 (2008) 1203–1208. [10] K. Furukawa, H. Fujihara, Microbial degradation of polychlorinated biphenyls: biochemical and molecular features, J. Biosci. Bioeng. 105 (2008) 433–449. [11] D.P. Barr, S.D. Aust, Mechanisms white rot fungi use to degrade pollutants, Environ. Sci. Technol. 28 (1994) 78A–87A. [12] C. Pinedo-Rivilla, J. Aleu, I.G. Collado, Pollutants biodegradation by fungi, Curr. Org. Chem. 13 (2009) 1194–1214. [13] N. Kasai, S. Ikushiro, R. Shinkyo, K. Yasuda, S. Hirosue, Metabolism of monoand dichloro-dibenzo-p-dioxins by Phanerochaete chrysosporium cytochromes P450, Appl. Microbiol. Biot. 86 (2010) 773–780. [14] L. Kamei, R. Kogura, R. Kondo, Metabolism of 4,4 -dichlorobiphenyl by white-rot fungi Phanerochaete chrysosporium and Phanerochaete sp. MZ142, Appl. Microbiol. Biot. 72 (2006) 566–575. [15] T. Cajthaml, Z. Kˇresinová, K. Svobodová, M. Möder, Biodegradation of endocrine-disrupting compounds and suppression of estrogenic activity by ligninolytic fungi, Chemosphere 7 (2009) 745–750. ˇ sek, M. Möder, Breakdown products on metabolic [16] T. Cajthaml, P. Erbanová, V. Saˇ pathway of degradation of benz[a]anthracene by a ligninolytic fungus, Chemosphere 64 (2006) 560–564. [17] D. Ning, H. Wang, Y. Zhuang, Induction of functional cytochrome P450 and its involvement in degradation of benzoic acid by Phanerochaete chrysosporium, Biodegradation 21 (2010) 297–308. [18] F. Matsuzaki, H. Wariishi, Functional diversity of cytochrome P450s of the white-rot fungus Phanerochaete chrysosporium, Biochem. Bioph. Res. Co. 324 (2004) 387–393. ˇ Novotny, ˇ sek, Degradation of PAHs ´ V. Saˇ [19] T. Cajthaml, P. Erbanová, A. Kollmann, C. by ligninolytic enzymes of Irpex lacteus, Folia Microbiol. 53 (2008) 289–294. ˇ sek, M. Matucha, Degradation of anthracene by [20] B.R.M. Vyas, S. Bakowski, V. Saˇ selected white rot fungi, FEMS Microbiol. Ecol. 14 (1994) 65–70. [21] E. De Jong, A.E. Cazemier, J.A. Field, J.A. De Bont, Physiological role of chlorinated aryl alcohols biosynthesized de novo by the white rot fungus Bjerkandera sp. strain BOS55, Appl. Environ. Microbiol. 60 (1994) 271–277.

394

M. Muzikáˇr et al. / Journal of Hazardous Materials 196 (2011) 386–394

[22] E. Matsumura, E. Yamamoto, A. Numata, T. Kawano, T. Shin, S. Murao, Structure of the laccase-catalyzed oxidation products of hydroxybenzoic acids in the presence of ABTS [2,2-azino-di-(3-ethylbenzothiazoline-6-sulfonic acid)], Agric. Biol. Chem. 50 (1986) 1355–1357. ˇ [23] S. Covino, M. Cvanˇ carová, M. Muzikáˇr, K. Svobodová, A. D‘annibale, M. Petruccioli, F. Federici, Z. Kˇresinová, T. Cajthaml, An efficient PAH-degrading Lentinus (Panus) tigrinus strain: effect of inoculum formulation and pollutant bioavailability in solid matrices, J. Hazard. Mater. 183 (2010) 669–676. [24] Z. Kˇresinová, M. Muzikáˇr, J. Olˇsovská, T. Cajthaml, Determination of 15 isomers of chlorobenzoic acid in soil samples using accelerated sample extraction followed by liquid chromatography, Talanta 84 (2011) 1141–1147. ˇ sek, P. Popp, Study of fungal degradation [25] T. Cajthaml, M. Möder, P. Kaˇcer, V. Saˇ products of polycyclic aromatic hydrocarbons using gas chromatography with ion trap mass spectrometry detection, J. Chromatogr. A 974 (2002) 213–222. [26] ISO 11348-3: 2007 Water quality – Determination of the inhibitory effect of water samples on the light emission of Vibrio fischeri (Luminescent bacteria test) – Part 3: Method using freeze-dried bacteria. [27] J. Lappalainen, R. Juvonen, K. Vaajasaari, M. Karp, A new flash method for measuring the toxicity of solid and colored samples, Chemosphere 38 (1999) 1069–1083.

[28] J. Lappalainen, R. Juvonen, J. Nurmi, M. Karp, Automated color correction method for Vibrio fischeri toxicity test. Comparison of standard and kinetic assays, Chemosphere 45 (2001) 635–641. [29] J. Dittmann, W. Heyser, H. Bücking, Biodegradation of aromatic compounds by white rot and ectomycorrhizal fungal species and the accumulation of chlorinated benzoic acid in ectomycorrhizal pine seedlings, Chemosphere 49 (2002) 297–306. [30] F. Guillen, A.T. Martinez, M.J. Martinez, C.S. Evans, Hydrogen-peroxideproducing system of Pleurotus eryngii involving the extracellular enzyme aryl-alcohol oxidase, Appl. Microbiol. Biot. 41 (1994) 465–470. [31] F. Matsuzaki, M. Shimizu, H. Wariishi, Proteomic and metabolomic analyses of the white-rot fungus Phanerochaete chrysosporium exposed to exogenous benzoic acid, J. Proteome Res. 7 (2008) 2342–2350. [32] F. Matsuzaki, H. Wariishi, Molecular characterization of cytochrome P450 catalyzing hydroxylation of benzoates from the white-rot fungus Phanerochaete chrysosporium, Biochem. Biophys. Res. Commun. 334 (2005) 1184–1190. [33] P. Baldrian, Wood-inhabiting ligninolytic basidiomycetes in soils: ecology and constraints for applicability in bioremediation, Fungal Ecol. 1 (2008) 4–12.

Journal of Hazardous Materials 196 (2011) 395–401

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Comparison of o-toluidine degradation by Fenton, electro-Fenton and photoelectro-Fenton processes Jin Anotai a , Somporn Singhadech b , Chia-Chi Su c , Ming-Chun Lu c,∗ a National Center of Excellence for Environmental and Hazardous Waste Management, Department of Environmental Engineering, Faculty of Engineering, King Mongkut’s University of Technology Thonburi, Bangkok 10140,Thailand b Department of International Postgraduate Programs in Environmental Management (Hazardous Waste Management), Chulalongkorn University, Bangkok 10330, Thailand c Department of Environmental Resources Management, Chia-Nan University of Pharmacy and Science, Tainan 717, Taiwan

a r t i c l e

i n f o

Article history: Received 27 May 2011 Received in revised form 5 September 2011 Accepted 11 September 2011 Available online 22 September 2011 Keywords: Box–Behnken design Electro-Fenton process Hydroxyl radicals Photoelectro-Fenton process o-Toluidine

a b s t r a c t A Box–Behnken design (BBD) statistical experimental design was used to investigate the degradation of o-toluidine by the electro-Fenton process. This method can be used to determine the optimal conditions in multivariable systems. Fe2+ concentration (0.2–1.0 mM), H2 O2 concentration (1–5 mM), pH (2–4), and current (1–4 A) were selected as independent variables. The removal efficiencies for o-toluidine and chemical oxygen demand (COD) were represented by the response function. Result by 2-level factorial design show that the pH and the Fe2+ and H2 O2 concentrations were the principal parameters. Among the main parameters, the removal efficiencies for o-toluidine and COD were significantly affected by pH and Fe2+ concentration. From the Box–Behnken design predictions, the optimal conditions in the electro-Fenton process for removing 90.8% of o-toluidine and 40.9% of COD were found to be 1 mM of Fe2+ and 4.85 mM of H2 O2 at pH 2. Under these optimal conditions, the experimental data showed that the removal efficiencies for o-toluidine and COD in the electro-Fenton process and the photoelectro-Fenton process were more than 91% and 43%, respectively, after 60 min of reaction. The removal efficiencies for o-toluidine and COD in the Fenton process are 56% and 27%, respectively. © 2011 Elsevier B.V. All rights reserved.

1. Introduction o-Toluidine is an important aromatic amine that is used in the dyestuffs and rubber industry. However, short-term exposure to otoluidine may induce methaemoglobinaemia, whereas long-term or repeated exposure to o-toluidine could be possibly carcinogenic to humans [1]; it may cause bladder cancer [2]. It is difficult to completely treat wastewater containing o-toluidine because of its resistance to biodegradation. Presently, advanced oxidation processes (AOPs) have been used for wastewater treatment, particularly in cases where the contaminant species are difficult to remove by biological or physicochemical processes [3–7]. AOPs are based on the generation of a powerful oxidant, the hydroxyl radical (• OH), which can react with most organic pollutants and then degrade them [8,9]. The Fenton process is one of the most widely used AOPs because of its low investment cost [10]. The Fenton reaction is shown below: Fe2+ + H2 O2 → • OH + Fe3+ + OH−

∗ Corresponding author. Tel.: +886 6 266 4911; fax: +886 6 266 3411. E-mail address: [email protected] (M.-C. Lu). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.043

(1)

However, the iron sludge produced from the Fenton process requires further treatment and disposal, and this is a major disadvantage of this process. This major drawback can be resolved by coupling the Fenton process with electric discharge, the so-called “electro-Fenton (EF) process”. The advantage of the electrochemical Fenton process is that it produces much less iron sludge than the traditional Fenton process. In this process, ferric ions (Fe3+ ) are effectively electroregenerated to ferrous ions (Fe2+ ), as shown in Eq. (2); this can be expressed in terms of current efficiency. Fe3+ + e− → Fe2+

(2)

The capability of the electro-Fenton process has been confirmed by Harrington and Pletcher [11], with more than 90% chemical oxygen demand (COD) removal with current efficiencies higher than 50% and acceptable energy consumptions. The efficiency of the electro-Fenton process can be improved by using UV or visible light illumination in a process known as the photoelectro-Fenton (PEF) process. This improvement is due to the higher production rate of • OH from the photoreduction of Fe(OH)2+ (Eq. (3)) and the photodecomposition of complexes from Fe3+ reactions (Eq. (4)) [12–15] Fe(OH)2+ + hv → Fe2+ + • OH R(CO2 )-Fe

3+

+ hv →

R(• CO

2)

(3) + Fe(II) →

•R

+ CO2

(4)

396

J. Anotai et al. / Journal of Hazardous Materials 196 (2011) 395–401

Normally, Fenton-type processes [16–18] are affected by the pH and initial Fe2+ and H2 O2 concentrations. To determine the optimal conditions for o-toluidine degradation and the effect of variables on the electro-Fenton process, the Box–Behnken design (BBD) was used in this investigation. The BBD can be used to find the optimal conditions in multivariable systems [19]. The statistical design of an experiment reduces the number of experiments that must be performed and corresponding time spent and can be used to optimize the operating parameters in multivariable systems. Few studies have used BBD for the degradation of azo-dyes and organic contaminants by the photo-Fenton process [19,20]. However, there have been no studies published on the use of BBD for the degradation of o-toluidine by the electro-Fenton process. In this study, the optimal conditions for o-toluidine degradation and the effect of four variables (pH, Fe2+ concentration, H2 O2 concentration and current) on the electron-Fenton process were investigated using BBD. As only several significant factors were involved in optimization, response surface methodology (RSM) was applied. Additionally, the o-toluidine degradation performances of ordinary Fenton, electroFenton and photoelectro-Fenton processes were also compared. 2. Materials and methods 2.1. Material and reactor o-Toluidine (99.5%, Merck), hydrogen peroxide (H2 O2 , 35%, Merck), and ferrous sulfate hepta-hydrated (FeSO4 ·7H2 O, Merck) were reagent grade and used without further purification. Fig. 1 shows the three kinds of reactors. The Fenton reactor was cylindrical stainless steel (diameter: 13 cm; height: 35 cm). The total volume of the reactor was 3.5 L. The electro-Fenton reactor, a cylindrical reactor, was operated in constant current mode. The anode was titanium net coated with RuO2 /IrO2 (DSA), and the cathode was stainless steel. The DSA anode with an inside diameter of 7 cm and height of 35 cm, and the cathode had an inside diameter of 2 cm and height of 35 cm. The electrodes were connected with direct current (DC) power. In the photoelectro-Fenton reactor, a set of 6 UV lamps fixed inside a cylindrical Pyrex tube (allowing wavelengths  > 320 nm to penetrate) were used as the irradiation source. The UV lamps were connected to the power supply E-safe, 2003, Switching Power Supply (Max. 300 W), Model: LC-B300AT. 2.2. Analysis method In the photoelectro-Fenton experiment, synthetic wastewater containing 1 mM o-toluidine was prepared and then initial pH was adjusted with perchloric acid (HClO4 ). After pH adjustment, a predetermined amount of catalytic ferrous sulfate was added into the solution and then the UV lights were turned on. H2 O2 was also added in the same time to start the reaction. Additionally, in the electro-Fenton experiment, solution with 1 mM o-toluidine was prepared and then ferrous ions were added after the pH was adjusted to the desired value. In the meantime, the power supply was turned on, and hydrogen peroxide was added to initiate the reaction. Samples (1 mL) were taken at predetermined time intervals and were immediately injected into a tube containing sodium hydroxide solution to quench the Fenton reaction by increasing the pH to 11. The sample was then filtered (0.45 ␮m filter) to remove precipitates and kept for 12 h before COD analysis. This process was used to avoid quantifying the effect of the H2 O2 concentration on the COD value. COD was analyzed by a closed reflux titrimetric method based on the standard methods [21]. The Fe2+ concentration was determined using the 1,10-phenanthroline method [22]. Total organic carbon was measured with an Elementar liquid TOC analyzer. The concentration of o-toluidine

was determined using high performance liquid chromatography (HPLC) with a Spectra system model SN4000 pump and Asahipak ODP-506D column (150 mm × 6 mm × 5 ␮m). The detection limit of o-toluidine was 0.005 mM or 0.535 ppm. Organic acids were analyzed using a Dionex DX-120 ion chromatograph with an Ion Pac AS11 anion column at 30 ◦ C. 2.3. Experimental design BBD are a class of rotatable or nearly rotatable second-order designs based on three-level incomplete factorial designs. Among all the RSM designs, BBD requires fewer runs [23]. The DesignExpert software version 7.0 (Stat-Ease, Inc., Minneapolis, USA) was used to find the optimal conditions of o-toluidine degradation by the electro-Fenton process. The effects of the significant factors were determined by BBD. The significant factors and the appropriate studied ranges were pH: 2–4, Fe2+ concentration: 0.2–1.0 mM, and H2 O2 concentration: 1–5 mM. The concentration of o-toluidine was fixed at 1 mM for all experiments. 3. Results and discussion 3.1. Effect of various parameters on o-toluidine removal efficiency The Fe2+ concentration, H2 O2 concentration, pH and current were selected as experimental conditions for the BBD. The removal efficiencies for o-toluidine and COD were represented by a response function. Table 1 shows the two levels of the four factors on BBD. The values of variables, the experimental data and the results are presented in Table 2. The maximum removal rate of o-toluidine was 100% and the minimum was 23% (Table 2). When 1.0 mM of Fe2+ , 5.0 mM of H2 O2 and 1.0 A of current at pH 2 were applied, the otoluidine removal was 94.4% (run 2). However, as current increased from 1.0 A to 4.0 A, the removal of o-toluidine slightly increased to 100% (run 6). It was found that the amount of current was not sensitive in the applied range and therefore could be neglected. The correlation of o-toluidine and COD removal efficiencies obtained from BBD is shown in Table 1, where a higher correlation means that the parameter has a higher effect on o-toluidine and COD. The correlation can be as high as 1 or low as −1. The result indicates that the current has a slight effect on o-toluidine, with a correlation of only 0.098 (Table 1). The degradation of o-toluidine depended on the initial concentration of Fe2+ and H2 O2 , showing a high correlation in o-toluidine removal efficiency about 0.617 and 0.278 for Fe2+ and H2 O2 concentration, respectively. The same trend was also observed in COD removal efficiency. This correlation indicates that the Fe2+ and H2 O2 concentrations have a positive effect on the removal efficiencies for o-toluidine and COD, indicating that increasing Fe2+ and H2 O2 concentrations increased the removal efficiencies for o-toluidine and COD. The pH has a negative effect on o-toluidine and COD removal, and so the removal efficiencies for otoluidine and COD decreased with increasing pH of solution. From the correlation values, initial pH and the Fe2+ and H2 O2 concentrations were the most significant factors that affected o-toluidine and COD removal. Table 3 shows the levels of significant factors of o-toluidine and COD removal efficiencies. Results from the experiment revealed that the maximum removal of o-toluidine was 91.4% and that of COD was 42% (run 1) (Table 4). The correlation values indicate that pH had the most pronounced effect on o-toluidine and COD removal (−0.725 for o-toluidine and −0.593 for COD) (Table 3). The Fe2+ concentration had a comparable effect on these responses, and H2 O2 concentration had a greater effect on COD removal than the degradation of o-toluidine. Fig. 2 shows the response surface plot of the effect of pH and Fe2+ concentration on the o-toluidine and COD removal efficiencies. This

J. Anotai et al. / Journal of Hazardous Materials 196 (2011) 395–401

397

Fig. 1. The experimental reactors.

Table 1 The two levels of variables and the value of correlation on o-toluidine and COD removal efficiency from Box–Behnken statistical design. Variables

Symbol

pH Fe2+ (mM) H2 O2 (mM) Current (A)

A B C D

Variable level

Correlation

Low

High

o-Toluidine

COD

2 0.2 1 1

4 1 5 4

−0.628 0.617 0.278 0.098

−0.517 0.633 0.274 0.259

Table 2 o-Toluidine and COD removal from the two levels of variables in electro-Fenton process with 1 mM of o-toluidine designed by the BBD. Run number

pH

Fe2+ (mM)

H2 O2 (mM)

Current (A)

o-Toluidine removal (%)

COD removal (%)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

4.0 2.0 4.0 4.0 2.0 2.0 2.0 2.0 4.0 4.0 2.0 2.0 4.0 4.0 2.0 4.0

0.2 1.0 0.2 1.0 0.2 1.0 1.0 1.0 1.0 0.2 0.2 0.2 0.2 1.0 0.2 1.0

5.0 5.0 5.0 5.0 5.0 5.0 1.0 1.0 1.0 1.0 5.0 1.0 1.0 5.0 1.0 1.0

4.0 1.0 1.0 4.0 1.0 4.0 1.0 4.0 4.0 1.0 4.0 1.0 4.0 1.0 4.0 1.0

17.0 94.4 32.0 56.0 42.0 100 66.4 68.5 45.0 23.0 74.0 40.0 26.5 61.4 56.4 48.0

23.0 57.0 22.0 33.02 8.0 59.5 33.0 46.0 34.0 19.0 38.4 22.5 27.0 36.0 34.6 31.0

plot shows the negative effect of pH on the removal efficiencies. The o-toluidine and COD removals decreased as the initial pH of the solution increased from 2.0 to 4.0 because the oxidation potential of hydroxyl radicals (• OH) and the dissolved fraction of iron species decreased [24,25]. The results also show that increasing the Fe2+ concentration can enhance o-toluidine and COD removal efficiencies because more Fe2+ reacts H2 O2 producing more • OH. Analysis of variance (ANOVA) tests for o-toluidine and COD removal were conducted to determine the suitability of the

response function and the significance of the effects of independent variables on the response function (Table 5). ANOVA indicates that the predictability of the model is at the 95% confidence level. Values of “Prob > F” less than 0.05 indicates a significant effect of the corresponding variable on the response. The result shows that the F-values of o-toluidine and COD removal were 11.10 and 8.06, respectively; imply that the model is significant. There are only 0.15% and 0.43% chances for o-toluidine and COD removal, respectively; that the model’s F-values this large could occur due to noise.

Table 3 The levels of significant factors and the value of correlation on o-toluidine and COD removal efficiency from BBD. Significant factor

Symbol

Variable level

Correlation

Low

Center

High

o-Toluidine

COD

pH Fe2+ (mM) H2 O2 (mM)

A B C

2 0.2 1

3 0.6 3

4 1 5

−0.725 0.484 0.294

−0.593 0.455 0.408

398

J. Anotai et al. / Journal of Hazardous Materials 196 (2011) 395–401

Table 4 o-Toluidine and COD removal from the levels of significant factors in electro-Fenton process with 1 mM of o-toluidine and 1 A designed by the BBD. Run number

pH

Fe2+ (mM)

H2 O2 (mM)

o-Toluidine removal (%)

COD removal (%)

1 2 3 4 5 6 7 8 9 10 11 12 13

2.0 4.0 2.0 2.0 2.0 4.0 3.0 40 3.0 3.0 3.0 4.0 3.0

0.6 0.6 1.0 0.2 0.6 0.6 0.2 0.2 0.6 1.0 0.2 1.0 1.0

5.0 5.0 3.0 3.0 1.0 1.0 5.0 3.0 3.0 1.0 1.0 3.0 5.0

91.4 31.2 75.0 47.4 63.4 34.0 51.0 15.0 60.0 52.0 42.4 49.4 78.0

42.0 17.6 36.0 16.4 30.0 16.0 24.0 17.0 32.0 21.0 16.4 21.2 36.0

In this case, pH and Fe2+ were significant model terms affecting percent o-toluidine and COD removal. 3.2. The prediction of optimal conditions of o-toluidine degradation by BBD The goal of this part was to establish the optimal conditions for maximum removal of o-toluidine and COD by the electro-Fenton process. The BBD can provide an empirical relationship between the response function and the variables. The mathematical relationship between the removal of o-toluidine and the three significant variables can be approximated by quadratic polynomial equation, and the equations for the removal of o-toluidine and COD by the electro-Fenton process are shown below: o − toluidine removal (%) = 64.22 − 11.02A + 12.36B + 3.25C + 2.73AB − 1.02AC + 8.58BC − 1.17ABC

(5)

COD removal (%) = 30.16 − 2.31A + 7.47B + 0.84C + 0.91AB − 2.81AC + 8.03BC − 1.81ABC

of o-toluidine and COD at each value of pH, Fe2+ concentration and H2 O2 concentration. On the basis of the coefficients in Eqs. (5) and (6), it indicates that pH (A) and Fe2+ (B) concentration have negative and positive effects on o-toluidine and COD removal efficiencies, respectively. In other words, removal of o-toluidine and COD decreased with the pH (A) while increasing with Fe2+ (B) and H2 O2 (C) doses. Fe2+ dose had a more profound effect on o-toluidine and COD removal as compared to H2 O2 . The electro-Fenton process utilizes electrochemical generation of ferrous ions from ferric ions and ferric complexes. Ferrous ions were continuously recycled electrochemically, and therefore they were not depleted during the degradation of o-toluidine. Fe2+ concentration has the greatest effect on removal of o-toluidine with the largest coefficient (12.36). In this study, the removal of o-toluidine and COD were selected as “maximize” and then, Fe2+ and H2 O2 concentrations and pH were used as “within the range.” Consequently, these individual goals were combined into an overall desirability function by the software to find the best optimal conditions. The final equation relationship between the response function (o-toluidine and COD removal) and the key parameters can be determined by Eqs. (7) and (8).

(6)

Fe2+

where A, B and C are pH, concentration and H2 O2 concentration, respectively. The equations are used to calculate the removal

o − toluidine removal (%) = 78.74 − 18.45 × pH + 30.81 × [Fe2+ ] + 3.74 × [H2 O2 ]

(7)

Table 5 ANOVA tests for o-toluidine and COD removal by BBD. Source

Sum of squares

df

Mean squares

F-value

p-value Prob > F

o-Toluidine removal Model A (pH) B (Fe2+ ) C (H2 O2 ) AB AC BC ABC Residual Cor Total

7790.65 3387.24 3271.84 663.06 1.56 190.44 262.44 14.06 802.07 8592.72

7 1 1 1 1 1 1 1 8 15

112.95 3387.24 3271.84 663.06 1.56 190.44 262.44 14.06 100.26

11.10 33.78 32.63 6.61 0.016 1.90 2.62 0.14

0.0015 0.0004 0.0004 0.0330 0.9037 0.2055 0.1443 0.7178

Significant Significant Significant

COD removal Model A (pH) B (Fe2+ ) C (H2 O2 ) AB AC BC ABC Residual Cor Total

1808.81 552.25 826.56 155.00 52.56 119.90 68.89 33.64 256.41 2065.22

7 1 1 1 1 1 1 1 8 15

258.40 552.25 816.56 155.00 52.56 119.90 68.89 33.64 32.05

8.06 17.23 25.79 4.84 1.64 3.74 2.15 1.05

0.0043 0.0032 0.0010 0.0591 0.2362 0.0891 0.1808 0.3356

Significant Significant Significant

J. Anotai et al. / Journal of Hazardous Materials 196 (2011) 395–401

399

Fig. 2. Three-dimensional representation of the response surface plot of the effect of pH and Fe2+ concentration on (a) o-toluidine and (b) COD removal efficiency.

COD removal (%) = 30.41 − 6.58 × pH + 12.63 × [Fe2+ ] + 2.27 × [H2 O2 ]

(8)

According to Eqs. (7) and (8), at the optimal conditions of pH 2, 1 mM of Fe2+ and 4.85 mM of H2 O2 , the maximum removals of o-toluidine and COD were 90.8% and 40.9%, respectively. 3.3. Comparison between various processes The optimal conditions were used to investigate the removal efficiencies for o-toluidine and COD in the Fenton process, electroFenton process and photoelectro-Fenton process. The results are shown in Fig. 3. Fig. 3(a) shows that the removal efficiency for otoluidine in the three processes was almost the same in the first 2 min. After 2 min, the removal of o-toluidine in the Fenton process was slightly increased. The removal efficiency for o-toluidine was approximately 56% after 60 min of reaction. However, the removal efficiencies for o-toluidine in the electro-Fenton process and the photoelectro-Fenton process were 91% and 99%, respectively, after 60 min of reaction. The removal of o-toluidine was due to the formation of • OH via Eq. (1). Moreover, Fe3+ in the solution was able to regenerate inside the reactor when electric discharge and UV irradiation were used, allowing numerous Fe2+ react with H2 O2 to generate • OH. Ferrous ions are not depleted during the oxidation reaction, as shown in Eqs. (2)–(4). Therefore, the electro-Fenton process and the photoelectro-Fenton process can enhance the oxidation rate of o-toluidine. The same result was found for COD removal efficiency, as shown in Fig. 3(b). The removal efficiencies for COD in the three processes were similar in the first 2 min. However, the removal efficiency of COD was significantly different between the various processes after two min of reaction. The removal efficiencies for COD were 27% for the Fenton process, 45% for the electro-Fenton process and 43% for the photoelectro-Fenton process after 60 min of reaction. Fig. 4 shows that maleic and oxalic acids were identified as the intermediates from the oxidation of o-toluidine. Maleic and oxalic acid were found in the electro-Fenton and the photoelectroFenton processes after 1 min of reaction and the concentrations increased with time. The decrease in maleic acid occurred in the electro-Fenton and the photoelectro-Fenton processes after 45 and 10 min, respectively (Fig. 4(a)). The decrease in oxalic acid occurred in the electro-Fenton and the photoelectro-Fenton processes after 30 and 45 min, respectively (Fig. 4(b)). However, maleic and oxalic acids were found after 10 min of reaction in the Fenton process and their concentrations increased with time until the end of the reaction. The concentration of maleic and oxalic acid were generated

quickly because of the increased concentration of • OH and degradation of o-toluidine. This result indicates that the electro-Fenton and the photoelectro-Fenton processes have higher efficiencies to degrade o-toluidine than the traditional Fenton process. The accumulation of intermediates in the photoelectro-Fenton process was lower than in the electro-Fenton process (Fig. 4(a) and (b)). Moreover, the removal efficiency for TOC in the electro-Fenton and the photoelectro-Fenton processes were 12% and 31%, respectively (Fig. 4(c)). These phenomena show that the intermediates were efficiently mineralized by the action of UV light in the photoelectroFenton process.

Fig. 3. Comparison between various processes on (a) o-toluidine removal and (b) COD removal efficiency. Experimental conditions: 1 mM of o-toluidine, 1 mM of Fe2+ and 4.85 mM of H2 O2 at pH 2. Each data has twice samplings.

400

J. Anotai et al. / Journal of Hazardous Materials 196 (2011) 395–401

results with 91% of o-toluidine removal and of 45% COD removal in the electro-Fenton process, indicating the reliability of the methodology used. Both electric discharge and UV irradiation could significantly enhance the degradation of o-toluidine. Therefore, more intermediates, such as maleic and oxalic acids, were derived during the beginning of the electro-Fenton and the photoelectroFenton processes than in the traditional Fenton process. Acknowledgements The authors would like to thank the National Science Council of Taiwan, for financially supporting this research under Contract No. NSC 96-2628-E-041-001-MY3. References

Fig. 4. The concentrations of (a) maleic acid, (b) oxalic acid and (c) TOC during the degradation of o-toluidine. Each data has twice samplings.

4. Conclusions This study investigated the optimization of o-toluidine treatment by the electro-Fenton process applying the Box–Behnken experimental design methodology. The results showed that pH and Fe2+ concentrations were important factors in the removal efficiencies for both o-toluidine and COD. The removal efficiencies for o-toluidine and COD increase with decreasing pH and increasing Fe2+ concentration. The optimal conditions for the maximum removal of o-toluidine and COD (90.8% and 40.9%, respectively, from prediction) were 1 mM of Fe2+ and 4.85 mM of H2 O2 at pH 2. Obviously, the calculating o-toluidine and COD removals applying the predicted conditions approaches the experimental

[1] G. Korinth, L. Lüersen, K.H. Schaller, J. Angerer, H. Drexler, Enhancement of percutaneous penetration of aniline and o-toluidine in vitro using skin barrier creams, Toxicol. In Vitro 22 (3) (2008) 812–818. [2] E. Richter, K. Gaber, U.A. Harréus, C. Matthias, N. Kleinsasser, o-Toluidine adducts in human bladder DNA and hemoglobin by the local anesthetic prilocaine, Toxicol. Lett. 164S (2006) S255. [3] J.H. Sun, S.P. Sun, M.H. Fan, H.Q. Guo, L.P. Qiao, R.X. Sun, A kinetic study on the degradation of p-nitroaniline by Fenton oxidation process, J. Hazard. Mater. 148 (2007) 172–177. [4] E. Brillas, I. Sirés, M.A. Oturan, Electro-Fenton process and related electrochemical technologies based on Fenton’s reaction chemistry, Chem. Rev. 109 (2009) 6570–6631. [5] J. Virkutyte, E. Rokhina, V. Jegatheesan, Optimisation of electro-Fenton denitrification of a model wastewater using a response surface methodology, Bioresour. Technol. 101 (2010) 1440–1446. [6] S. Mohajeri, H.A. Aziz, M.H. Isa, M.A. Zahed, M.N. Adlan, Statistical optimization of process parameters for landfill leachate treatment using electro-Fenton technique, J. Hazard. Mater. 176 (2010) 749–758. [7] L.C. Almeida, S. Garcia-Segura, N. Bocchi, E. Brillas, Solar photoelectroFenton degradation of paracetamol using a flow plant with a Pt/air-diffusion cell coupled with a compound parabolic collector: process optimization by response surface methodology, Appl. Catal. B: Environ. 103 (2011) 21–30. [8] W.H. Glaze, J.W. Kang, R.H. Chapin, The chemistry of water treatment processes involving ozone, hydrogen peroxide and ultraviolet radiation, Ozone Sci. Eng. 9 (4) (1987) 335–352. [9] M.J. Liou, M.L. Lu, Catalytic degradation of nitroaromatic explosives with Fenton’s reagent, J. Mol. Catal. A: Chem. 277 (2007) 155–163. [10] S. Esplugas, J. Gimenez, S. Contreras, E. Pascual, M. Rodriguez, Comparison of different advanced oxidation processes for phenol degradation, Water Res. 36 (4) (2002) 1034–1043. [11] T. Harrington, D. Pletcher, The removal of low levels of organics from aqueous solutions using Fe(II) and hydrogen peroxide formed in situ at gas diffusion electrodes, J. Electrochem. Soc. 146 (1999) 2983–2989. [12] H.J. Benkelberg, P. Warneck, 3-Chlorophenol elimination upon excitation of dilute iron(III) solution: evidence for the only involvement of Fe(OH)2+ , J. Phys. Chem. 99 (1995) 5214–5221. [13] R. Andreozzi, V. Caprio, R. Marotta, Iron(III) (hydr) oxide-mediated photooxidation of 2-aminophenol in aqueous solution: a kinetic study, Water Res. 37 (2003) 3682–3688. [14] M. Fukushima, K. Tatsumi, K. Morimoto, The fate of aniline after a photo-Fenton reaction in an aqueous system containing iron(III), humic acid, and hydrogen peroxide, Environ. Sci. Technol. 34 (2000) 2006–2013. [15] C.L. Hsueh, Y.H. Huang, C.Y. Chen, Novel activated alumina-supported iron oxide-composite as a heterogeneous catalyst for photooxidative degradation of reactive black 5, J. Hazard. Mater. B 129 (2006) 228–233. [16] H. Zhang, C. Fei, D. Zhang, F. Tang, Degradation of 4-nitrophenol in aqueous medium by electro-Fenton method, J. Hazard. Mater. 145 (2007) 227–232. [17] N. Masomboon, C. Ratanatamskul, M.C. Lu, Chemical oxidation of 2, 6dimethylaniline in the Fenton process, Environ. Sci. Technol. 43 (2009) 8629–8634. [18] J. Anotai, C.C. Su, Y.C. Tsai, M.C. Lu, Effect of hydrogen peroxide on aniline oxidation by electro-Fenton and fluidized-bed Fenton processes, J. Hazard. Mater. 183 (2010) 888–893. [19] F. Ay, E.C. Catalkaya, F. Kargi, A statistical experiment design approach for advanced oxidation of Direct Red azo-dye by photo-Fenton process, J. Hazard. Mater. 162 (2009) 230–236. [20] N. Masomboon, C.W. Chen, J. Anotai, M.C. Lu, A statistical experimental design to determine o-toluidine degradation by the photo-Fenton process, Chem. Eng. J. 159 (2010) 116–122. [21] APHA, Standard Methods for the Examination of Water and Wastewater, 18th Ed., American Public Health Association, Washington, DC, 1992.

J. Anotai et al. / Journal of Hazardous Materials 196 (2011) 395–401 [22] H. Tamura, K. Goto, T. Yotsuyanagi, M. Nagayama, Spectrophotometric determination of iron(II) with 1,10-phenanthroline in the presence of large amounts of iron(III), Talanta 21 (1974) 314–318. [23] S.L.C. Ferreira, R.E. Bruns, H.S. Ferreira, G.D. Matos, J.M. David, G.C. Brandão, E.G.P. da Silva, L.A. Portugal, P.S. dos Reis, Box–Behnken design: and alternative for the optimization of analytical methods, Anal. Chim. Acta 597 (2007) 179–186.

401

[24] W.P. Ting, M.C. Lu, Y.H. Huang, Kinetics of 2, 6-dimethylaniline degradation by electro-Fenton process, J. Hazard. Mater. 161 (2009) 1484–1490. [25] M. Panizza, G. Cerisola, Electro-Fenton degradation of synthetic dyes, Water Res. 43 (2009) 339–344.

Journal of Hazardous Materials 196 (2011) 402–411

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Reconstruction of metal pollution and recent sedimentation processes in Havana Bay (Cuba): A tool for coastal ecosystem management M. Díaz-Asencio a,∗ , J.A. Corcho Alvarado b , C. Alonso-Hernández a , A. Quejido-Cabezas c , A.C. Ruiz-Fernández d, M. Sanchez-Sanchez c, M.B. Gómez-Mancebo c, P. Froidevaux b, J.A. Sanchez-Cabeza e a

Centro de Estudios Ambientales de Cienfuegos, Carretera Castillo de Jagua, Cienfuegos, CITMA-Cienfuegos, Cuba Institute of Radiation Physics (IRA), University Hospital and University of Lausanne, Rue du Grand-Pré 1, 1007 Lausanne, Switzerland c Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas (CIEMAT), Madrid, Spain d Universidad Nacional Autónoma de México. ICMyL, Mazatlán, Mexico e Institute of Environmental Science and Technology, and Physics Department, Universitat Autónoma de Barcelona, 08193 Bellaterra, Barcelona, Spain b

a r t i c l e

i n f o

Article history: Received 20 June 2011 Received in revised form 12 September 2011 Accepted 12 September 2011 Available online 16 September 2011 Keywords: Pollutants 210 Pb 239,240 Pu 137 Cs Sediment dating Havana Bay

a b s t r a c t Since 1998 the highly polluted Havana Bay ecosystem has been the subject of a mitigation program. In order to determine whether pollution-reduction strategies were effective, we have evaluated the historical trends of pollution recorded in sediments of the Bay. A sediment core was dated radiometrically using natural and artificial fallout radionuclides. An irregularity in the 210 Pb record was caused by an episode of accelerated sedimentation. This episode was dated to occur in 1982, a year coincident with the heaviest rains reported in Havana over the XX century. Peaks of mass accumulation rates (MAR) were associated with hurricanes and intensive rains. In the past 60 years, these maxima are related to ˜ periods, which are known to increase rainfall in the north Caribbean region. We observed a strong El Nino steady increase of pollution (mainly Pb, Zn, Sn, and Hg) since the beginning of the century to the mid 90s, with enrichment factors as high as 6. MAR and pollution decreased rapidly after the mid 90s, although some trace metal levels remain high. This reduction was due to the integrated coastal zone management program introduced in the late 90s, which dismissed catchment erosion and pollution. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Havana Bay is one of the largest and most important estuaries in the Cuban Island. The economic, commercial and recreational values of the bay have been, however, threatened by pollution and the reduction of water depth due to infilling [1]. The environmental degradation of the bay ecosystem has been intensified in the past decades due to the fast economical growth of Havana City. This one has become the most contaminated bay in the island [2]. In order to rehabilitate this marine ecosystem, several pollution-reduction measures have been implemented over the past decade. Reliable information about the input of pollutants to Havana Bay is however required for evaluating the impact of the environmental management practices. In the absence of long-term monitoring data, sedimentary records can provide retrospective information about the past inputs of pollutants into aquatic environments. Pollutants such as heavy metals often have a strong affinity for particle

∗ Corresponding author. E-mail address: [email protected] (M. Díaz-Asencio). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.09.037

surfaces and, therefore, accumulate in the sediments. Hence, dated sediment profiles of major and trace elements can be used to obtain reliable information about the extent and history of pollution and sedimentary conditions [3–7]. The quantitative reconstruction of a contaminant input into an aquatic system requires a good sediment chronology. The most widely used method for dating recent marine and lacustrine sediments is based on the examination of 210 Pb profiles. The natural radionuclide 210 Pb (T1/2 = 22.23 yr) enters the aquatic environment mainly by atmospheric deposition; however it can be produced in situ, in the water column and the sediments, by decay of its precursor radionuclide 226 Ra (T1/2 = 1600 yr). The 210 Pb dating methods are based on the radioactive disequilibrium between the 210 Pb and 226 Ra [8,9]. 210 Pb has shown to be an ideal tracer for dating aquatic sediments deposited during the last 100–150 years, a period of time with significant environmental changes due to industrialization and population growth. 210 Pb dating should be always corroborated by an additional chronostratigraphic marker in the same sediment core [10,11]. Among the most commonly used time markers we find anthropogenic fallout radionuclides such as 137 Cs, 239,240 Pu or 241 Am

M. Díaz-Asencio et al. / Journal of Hazardous Materials 196 (2011) 402–411

403

the ships arriving to the country. Agriculture and intensive forest exploitation in the bay catchment increased soil erosion and, therefore, sediment input to the Bay. Industrial activities began in the 1850s with the construction of oil refineries, electric power plants and gas production [2,16]. The area of Havana experienced a fast economical growth during the 20th century, with a high diversity of industries and commercial activities, and a large population growth that required massive urbanization (from 250,000 inhabitants in 1899 to 2.2 million inhabitants in 2001). The anarchic growth of many activities over the past 400 years has caused severe damages to the natural resources and facilities of the Havana Bay. The damages have been intensified by the lack of waste treatment facilities [2,17,18]. The bay receives contaminants from numerous sources such as an oil refinery, power stations, urban and industrial wastewaters, three shipyards, riverine and stream discharges, and atmospheric fallout [1]. 3. Sampling and laboratory methods 3.1. Sampling

Fig. 1. Map of Havana Bay, with sampling station.

[13]. The onset of anthropogenic radionuclides, originating from the atmospheric nuclear weapon tests (early 1950s) and their peak value in 1963 [12] have been widely used as time markers in numerous marine and lacustrine studies [13–15]. In this work we reconstructed the historical trends of pollutants entering the Havana Bay by analyzing their sediment concentration profiles. The chronology of the sediment core was based on the 210 Pb dating method. Due to the low levels of 137 Cs found in the sediments, the chronology was further validated with 239,240 Pu and 241 Am fallout radionuclides. Enrichment factors and fluxes of pollutants were used to describe the history of pollutions in this aquatic ecosystem. A full geochemical analysis was undertaken to look at possible impacts of changing catchment sources, diagenesis, and atmospheric contamination. The potential origin of the most important pollutants and the impact of the pollution-reduction measures taken to protect the Havana Bay ecosystem are also discussed. Despite the large number of studies of pollution in coastal environments, only a few have been conducted in the Caribbean region. Hence, this study provides important information about pollution trends in a coastal ecosystem of this tropical region.

In February 2008, sediment cores were collected with an UWITEC corer avoiding dredged areas of the bay (Fig. 1). In order to optimize analytical time and resources, we chose the cores with the best likelihood to show good temporal record (section 1 of Supporting information). We sampled three sediment cores from the station B (23◦ 08.107 N 82◦ 20.043 ) at a water depth of 8 m (Fig. 1): one core for radionuclides, metals and grain size analysis; one for organic pollutants (not reported here) and one was kept frozen for future analysis. The sediment core was vertically extruded and sliced into 1 cm sections. Each section was dried at 45 ◦ C, sieved through a 1 cm sieve and homogenized. The mud content in the sediments showed a slight decreasing trend from 15 cm depth to the surface. The sediments also showed a strong change in color at about 15 cm depth. 3.2. Laboratory analyses Grain size was determined by standard methods of sieving and pipetting analysis [19]. Content of organic matter for each section was estimated by the loss on ignition method (LOI: 450 ◦ C, for 8 h). The content of total carbon and nitrogen was measured by using a CHN analyzer (LECO TRUSPEC). Total carbon was measured as CO2 with an infrared detector. N2 was measured by using a thermal conductivity detector. Inorganic carbon was quantified by using an infrared detector (SSM-500, Shimadzu) after sample acidification with phosphoric acid and heating (200◦ C). Major and trace elements were measured by Wavelength Dispersion X-Ray Fluorescence Spectrometry (WDXRF) using a Panalytical system (AXIOS) with Rhodium tube. Total mercury concentrations were determined by using an Advanced Mercury Analyser (LECO AMA254, detection limit of 0.01 ng Hg).

2. Site description

3.3. Sediment dating

Havana Bay (NW Cuba) is located aside Havana City, is a typical enclosed bay with a catchment area of about 68 km2 . It is characterized by a mean depth of 10 m, an area of 5.2 km2 and a water mean residence time of 7–9 days [1]. The bay is an estuary with deltaic systems in the fluvial discharge zones of the Luyano and Martin Perez rivers, and the Tadeo, Matadero, Agua Dulce and San Nicolas streams (Fig. 1). The population density, commercial and harbors activities in Havana City increased significantly since 1850. The city became a key transshipment point between Europe and America in the 19th century. Nowadays, the port of Havana receives about 50% of

The chronology of the sediment cores was determined by the method (section 2.3 of Supporting information). Sediment samples were placed in sealed plastic containers and stored for at least three weeks in order to allow 226 Ra to reach equilibrium with its daughter nuclides. 226 Ra was then analysed by high-resolution gamma spectrometry, using a low-background intrinsic Ge coaxial detector ORTEC model GX10022. 226 Ra was determined via the 352 keV emitted by its daughter nuclide 214 Pb in equilibrium. Supported 210 Pbsup was derived from the assumption of equilibrium with 226 Ra. The total 210 Pb activity was determined by high-resolution ␣ spectrometry of its decay product 210 Po, assumed 210 Pb

404

M. Díaz-Asencio et al. / Journal of Hazardous Materials 196 (2011) 402–411

Fig. 2. Profiles in core B of (a) particle size distribution, (b) carbonates and aluminisilicates distribution, and (c) organic (Corg ) and inorganic (Cinorg ) carbon, phosphorous (P) and nitrogen (N).

to be at equilibrium. Aliquots (0.5 g) of dry sediment were spiked with 209 Po as a yield tracer and dissolved by adding a mixture of 1:1:0.5 HNO3 + HCl + HF using an analytical microwave system [20]. 210 Po was electrodeposited onto silver discs [21,22] and counting was done in an integrated Camberra alpha spectrometer with ion implanted planar silicon detectors (active area of 450 mm2 ; and 18 keV of nominal resolution). The 210 Pb in excess (210 Pbex ) to the 210 Pb supported by 226 Ra (210 Pb sup ) was determined by subtracting the 210 Pbsup from the total activity of 210 Pb measured in each layer. 210 Pb ex was then introduced in the models in order to obtain the sedimentation rate (section 2.3 of Supporting information). Measurements of 137 Cs, 239,240 Pu and 241 Am were used to validate the 210 Pb dating models. 137 Cs was measured via its emission at 662 keV by high-resolution gamma spectrometry. The sediment samples were then crushed and ashed at 550 ◦ C for 48 h prior to the radiochemical analyses of 239,240 Pu and 241 Am. Composite samples were prepared by mixing layers. The method combines high-pressure microwave digestion for the dissolution of the sample and the highly selective extraction chromatographic resins TEVA and DGA (Triskem International, France) for the separation and purification of Pu and Am [23]. The alpha sources were

prepared by electrodeposition onto stainless steel discs [24]. Highresolution ␣-spectrometry was performed on a ␣-spectrometer with PIPS detectors (Alpha Analyst, Canberra Electronic).

4. Results 4.1. Characteristic of the sediments The sediments are predominantly fine and displayed small variations in grain size in the overall samples (Fig. 2a). The sediments consisted mainly of clay (4 ␮m, 15–42%) sized particles. In the upper 5 cm, the percent of silt and very fine to coarse sand sized particles increased slightly (Fig. 2a). The sediments were mainly composed of carbonates (11–45%) and aluminosilicates (40–80%). The mineral composition of the sediments did not change significantly from the bottom of the core up to 20 cm depth (Fig. 2b). However, from 20 cm depth up to the surface large variations were observed with the amount of carbonates correlating negatively to the amount of alumino-silicates (Fig. 2b).

M. Díaz-Asencio et al. / Journal of Hazardous Materials 196 (2011) 402–411

Fig. 3. Profiles of Si, Ca, Fe, Al and Mn in core B. Uncertainties were below 2% for all elements. The dashed line indicates the slump.

The amount of carbonates showed a general trend to decrease towards the surface, but it increased rapidly in the top 4 cm (Fig. 2b). The content of inorganic carbon (Cinorg ) in the sediments was almost constant along the whole core; but the organic carbon (Corg ) showed a surface maximum and then decreased with depth depicting two zones of rapid change (at 2–3 cm, and at 16–17 cm depth; Fig. 2c). The large percentage of Corg in the top 2–3 cm may be related to a change in the sources of organic matter, more complex and less biodegradable, typical of industrial organic wastes. However, the increase of organic matter may be also related to the reduction of particles observed in the Bay over the past decades. In the segment 3–16 cm, Corg was nearly constant around 4%; then, below 16 cm depth, Corg decayed to about 1.5% (Fig. 2c). Similarly to the pattern of Corg , nitrogen (N) and phosphorous (P) profiles showed increasing trends towards the surface, with nearly constant concentrations between 3 and 16 cm depth (Fig. 2c). 4.2. Major and tracer elements in the sediment core The concentration profiles of Al, Fe, Si, and Mn showed a slight decrease in the two uppermost layers, but no significant changes below 3 cm depth (Fig. 3). The Ca content showed an opposite trend

405

to Al (Fe, Si), with slightly higher concentrations at the surface (Fig. 3). In fact, Ca, which is probably in the form of biogenic carbonates, acts as a dilute of the trace metal concentrations in the bulk sediments. A slight change in the profiles is observed at 15–20 cm, which probably indicates the input of sediments with high content of Ca (Fig. 3). The concentration profiles of Pb, Zn, S, Hg, Sn and Cr in the sediment core showed similar increasing trends towards surface (Fig. 4). Maximum concentrations for Zn (450 mg kg−1 ), Pb (123 mg kg−1 ) and S (1.4 mg kg−1 ) were observed at the core surface, while for Hg (1.4 mg kg−1 ), Sn (18.6 mg kg−1 ) and Cr (365 mg kg−1 ) the maximum concentrations were found at depths between 5 and 15 cm. The maximum concentrations are comparable to the concentrations found in other polluted coastal sediments such as Porto Marghera in Italy [25], Halifax Harbor in Nova Scotia [26] and Barcelona in Spain [27]. The similarities in the Pb, Zn and Sn profiles (linear correlation R2 > 0.9 and p < 0.01 for each combination) suggest that these elements possibly originated from the same source and/or that the geochemical affinities to the sediment particles are similar. Pb, Zn, Hg, Sn, Cr and S profiles showed a plateau between 3 and 15 cm depth suggesting a similar time of deposition for the whole segment (e.g. an episodic event).

4.3. Radionuclide profiles and sediment chronology The 210 Pbex profile has non-monotonic features suggesting irregularities in the process of sediment accumulation (Fig. 5a). The surface 210 Pbex activities were around to 230 Bq kg−1 (Fig. 5a), relatively high compared to activities found in other studies from Havana Bay [28] and from other Cuban coastal locations [4,29]. A detailed analysis of the 210 Pbex distribution with depth suggests that the record can be divided into three distinct segments. At the top (0–3 cm) and bottom (15–35 cm) sections of the sediment core, 210 Pb ex declined exponentially with depth, indicating regular sedimentation. However, 210 Pbex was almost constant throughout the 3–15 cm segment of the core. A flattening of 210 Pbex indicates either a dilution of the 210 Pb atmospheric flux by sediment mixing, accelerating sedimentation and/or the occurrence of slumps due to, for example, heavy rains (typical in this tropical region). The trends observed in the profiles of 210 Pbex (Fig. 5a), Corg , P, N (Fig. 2c) and some major and trace elements (Figs. 3 and 4) suggest that most probably the section from 3 to 15 cm was instantly deposited as a result of an episodic event or slump. This is also supported by the color change of the sediment core at about 15 cm

Fig. 4. Profiles of (a) Pb, Zn, Sn and Cr; and b) S and Hg in core B. The uncertainties were: Pb (