PREFACE TO THE FIRST EDITION
Capillary electrophoresis (CE) or high-performance CE (HPCE) is making the transition fro...
80 downloads
937 Views
22MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
PREFACE TO THE FIRST EDITION
Capillary electrophoresis (CE) or high-performance CE (HPCE) is making the transition from a laboratory curiosity to a maturing microseparations technique. Now used in almost 1000 laboratories worldwide, CE is employed in an ever-widening scope of applications covering both large and small molecules. The inspiration for this book arose from my popular American Chemical Society short course entitled, as is this text, "Practical Capillary Electrophoresis." During the first 18 months since its inception, nearly 500 students have enrolled in public and private sessions in the United States and Europe. I have been amazed at the diversity of the scientific backgrounds of my students. Represented in these courses were molecular biologists, protein chemists, analytical chemists, organic chemists, and analytical biochemists from industrial, academic, and government laboratories. Interestingly enough, CE provides the mechanism for members of this multidisciplinary group to actually talk with each other, a rare event in most organizations. But the diverse nature of the group provides teaching challenges as well. Most of the students are well versed in the art and science of liquid chromatography. However, CE is not chromatography (usually). It is electrophoresis, and it is governed by the art and science of electrophoresis. For those skilled in electrophoresis, CE offers additional separation opportunities that are not available in the slab-gel format. Furthermore, the intellectual process of methods development differs from that in either slab-gel electrophoresis or liquid chromatography. The key to grasping the fundamentals of CE is to develop an understanding of how ions move about in fluid solution under the influence of an applied electric field. With this background, it becomes painless to wander through the electrophoretic domain and explain the subdeties and permutations frequently illustrated on the electropherograms. Accordingly, a logical approach to methods
XIV
Preface
development evolves from this treatment. This is the goal of my course, and hopefully, I have translated this message into this text. Since I work independently, without academic or industrial affiliations, the writing of this text would have been impossible without the help of my friends and colleagues. In particular, 1 am grateful to Professor Ira Krull and his graduate student, Jeff Mazzeo, from Northeastern University for reviewing the entire manuscript; Dr. Michael Albin from Applied Biosystems, Inc., for providing his company's computerized bibliography on HPCE; and the Perkin-Elmer Corporation including Ralph Conlon, Franco Spoldi, and librarian Debra Kaufman and her staff for invaluable assistance. I am also thankful to my associates throughout the scientific instrumentation industry for providing information, intellectual challenges, hints, electropherograms, comments, etc., many of which are included in this text. Last, I thank my students for helping me continuously reshape this material to provide clear and concise explanations of electrophoretic phenomena. Finally, many of the figures in this text were produced by scanning the illustration in a journal article with subsequent graphic editing. While all efforts were made to preserve the integrity of the original data, subtle differences may appear in the figures produced in this book. Robert Weinberger Chappaqua, NY August 1992
PREFACE TO THE SECOND EDITION
It is hard to believe that seven years have passed since I wrote the first edition of this book. The time is ripe for a second edition. Not only has capillary electrophoresis matured, but my ability to articulate the field has improved as well. I have reorganized this book to better reflect usage in the field. There are now ten chapters instead of twelve. The material on isotachophoresis has been combined with the section on stacking, and the special topics chapter has been eliminated. With the exception of the introduction and the chapter on basic concepts, all of the other material has been extensively reorganized and rewritten. Emphasis has been placed on commercially available apparatus and reagents, although gaps in the commercial offerings are discussed as well. Note that micellar electrokinetic capillary chromatography (MECC) is considered as a variant of capillary zone electrophoresis (CZE) and is included in the chapter on secondary equilibrium. Cyclodextrins and chiral recognition are covered here as well. Many thanks to Dr. Bruce McCord, Mr. Ira Lurie, and Professor Ira KruU for reviewing some of the chapters in this second edition. The author gratefully acknowledges the support of Hewlett-Packard and in particular Dr. David Heiger. Much has been said about the ability of capillary electrophoresis (HPCE) to replace liquid chromatography (HPLC). Clearly it has not. As the first highperformance condensed phase technique, HPLC quickly replaced gas chromatography as the method of choice for separating polar molocules. As food for thought, imagine if capillary electrophoresis had a 25-year head start over HPLC. Then perhaps the chromatographers would be fighting the uphill battle of displacing HPCE. As noted in this text, HPCE is clearly superseding the slab gel, at least in the fields of DNA separations. Robert Weinberger Chappaqua, NY June 1,1999
MASTER SYMBOL LIST
A
Corrected peak area corr
r^
A
Raw peak area raw
r
a a a h C, c C C^ CLOD CMC %C D D, D
Fraction ionized Molar absorptivity Separation factor Detector optical pathlength Concentration Coefficient for resistance to mass transfer in the mobile phase Coefficient for resistance to mass transfer in the stationary phase Concentration limit of detection Critical micelle concentration Percentage of crosslinker in a gel Capillary diameter Diffusion coefficient
m
D^^ DR d AH Ap^ AP 6 ^ e E E E 8 8 8o
Solute diffusion in stagnant mobile phase Dynamic reserve Particle diameter, chromatography Height differential between capillary inlet and outlet Difference in mobility between two solutes Pressure drop Debye radius Zeta potential Charge per unit area Field strength Acceptable increase in H Detector efficiency Dielectric constant Molar absorptivity Permittivity of vacuum
XVi
Master Symbol List
/ g Y Y H dH/dt I If I k k' k' K, X K L L^ I^ Lf L^ ^^ L ^^^^^ L^ l.^. X m M MLOD N N n r\ P
Frictional force (Stoke's law) Gravitational constant Field enhancement factor Obstructive factor for diffusion, Van Deemter equation Height equivalent of a theoretical plate Rate of heat production Current Fluorescence intensity Excitation source intensity Conductivity Capacity factor Capacity factor in MECC Thermal conductivity Equilibrium constant Length of capillary Length of capillary to detector Length of the detector window Length of capillary from detector to fraction collector Length of the unpacked portion of a CEC capillary Length of the packed portion of a CEC capillary Total length of capillary Length of an injection plug Tortuosity factor, Van Deemter equation Mass Actual mass Mass limit of detection Number of segments in a polymer chain Number of theoretical plates Number of charges Viscosity Partition coefficient between water and micelle
wm
AP O O O Oj O* p p Q q R R R
Pressure drop Polymer concentration, size separations Quantum yield Overlap threshold Fluorescence quantum yield Entanglement threshold, size separations Density Resistivity Quantity of injected material Ionic net charge Resistance Peak ratio Displacement ratio
Master Symbol List
R
XVll
Resolution
s
r r S/N a a
Ionic radius (Stokes' law) Capillary radius Signal to noise ratio Peak variance Peak variance due to capillary wall effects
cap
a^
^
J
Peak variance due to the detector
det
a, „
Peak variance due to diffusion
diff
a^^ ^heat a
Peak variance due to electrodispersion P^ak variance due to Joule heating Peak variance due to injection
mj
a^ o
J
Peak variance in units of length Peak variance from all sources
tot
0/
T |JL TR |i^ %)T |Li^^ |Li^ V V 1) 1) D^ 1) eo
0)^ ^^ \) ^^^^^ W W.^ W^ W^ X. X^ Z Z
Time Absorption time to a stationary phase or wall Desorption time from a stationary phase or wall Lag time Migration time Migration time for a micellar aggregate Migration time for a neutral "unretained" solute Retention time Temperature Ionic mobility Transfer ratio Apparent mobility Percentage(measured) of monomer and crosslinker in a gel Electroosmotic mobility Electrophoretic mobility Partial molar volume of micelle Voltage Ionic velocity Mean linear velocity Electrophoretic velocity Electroosmotic velocity J
Solute velocity in the unpacked portion of a CEC capillary Solute velocity in the packed portion of a CEC capillary Power Width of an injection plug Spatial width of a sample zone Temporal width of a sample zone Intital length of an injection plug Zone length after stacking Number of valence electrons Charge
CHAPTER
1
Introduction 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.10
Electrophoresis Microchromatographic Separation Methods Capillary Electrophoresis Capillary Electrochromatography Micromachined Electrophoretic Devices Historical Perspective Generic HPCE Systems Instrumentation Sources of Information on HPCE Capillary Electrophoresis: A Family of Techniques References
1.1 ELECTROPHORESIS Electrophoresis is a process for separating charged molecules based on their movement through a fluid under the influence of an applied electric field. If two solutes have differing electrophoretic mobilities, then separation v^U usually occur. The separation is performed in a medium such as a semisolid slab-gel. Gels provide physical support and mechanical stabiUty for the fluidic buffer system. In some modes of electrophoresis, the gel participates in the mechanism of separation by serving as a molecular sieve. Nongel media such as paper or cellulose acetate are alternative supports. These media are less inert than gels, as they contain charged surface groups that may interact with the sample or the run buffer. A carrier electrolyte is also required for electrophoresis. Otherwise known as the background electrolyte (BGE), the carrier electrolyte, or simply the run buffer, this solution maintains the requisite pH and provides sufficient conductivity to allow the passage of current (ions), necessary for the separation. Frequently, additional materials are added to the BGE to adjust the resolution of the separation through the generation of secondary equilibria. Additives can also serve to maintain solubility and prevent the interaction of solutes or excipients with the gel matrix or, in the case of capillary electrophoresis, with the
^
Chapter 1
Introduction
capillary wall. The theory and practice of electrophoresis have been the subject of many textbooks and conference proceedings (1-9). Apparatus for conducting electrophoresis, such as that illustrated in Figure 1.1, is remarkably simple and low cost. The gel medium, which is supported on glass plates, is inserted into a Plexiglass chamber. Two buffer reservoirs make contact at each end of the gel. Electrodes immersed in the buffers complete the electrical circuit between the gel and power supply. Many samples can be separated simultaneously, since it is possible to use a multilane gel. One or two lanes are frequently reserved for standard mixtures to calibrate the electropherogram. Calibration is usually based on molecular size or, in isoelectric focusing, pi. Gels such as polyacrylamide or agarose serve several important functions: 1. they may contribute to the mechanism of separation; 2. they reduce the dispersive effects of diffusion and convection; and 3. they serve to physically stabilize the separation matrix. The gel composition is adjusted to define specific pore sizes, each for a nominal range of molecular sizes. This forms the basis for separations of macromolecules based on size. By proper calibration, extrapolation to molecular weight is straightforward. Reduction of convection and diffusion is an important function of the gel matrix. The production of heat by the applied field induces convective movement of the electrolyte. This movement results in band broadening that reduces the efficiency of the separation. The viscous gel media inhibits fluid movement in the electric field. Such a material is termed anticonvective. Since the gel is of high viscosity, molecular diffusion is reduced as well, further enhancing the efficiency of the separation.
BUFFER SOLUTION CATHODE
GEL ANODE
BUFFER SOLUTION
FIGURE 1.1
Drawing of an apparatus for slab-gel electrophoresis.
1.2
Microchromatographic Separation Methods
3
Finally, the gel must be sufficiently viscous to provide physical support. Low viscosity solutions or gels would flow if the plate is not held level. Immersion in detection reagents would be impossible, since handling or contact with fluid solutions would destroy the matrix and separation. In the capillary format, the gel is unnecessary since the walls of the capillary provide the mechanical stability for the separation. The basic procedure for performing gel electrophoresis is as follows: 1. 2. 3. 4. 5. 6. 7.
prepare, pour and polymerize the gel; apply the sample; run the separation; immerse the gel in a detection reagent; ^ destain the gel; preserve the gel; and photograph or scan the gel for a permanent record.^
These steps are extremely labor intensive. High performance capillary electrophoresis (HPCE) is the automated and instrumental version of slab-gel electrophoresis. In the DNA applications arena, the most important of which include DNA sequencing, human identification, and genetic analysis, HPCE is rapidly replacing the slab-gel as the separation method of choice. The separation of some polymerase chain reaction (PCR) products is shown in Figure 1.2. A restriction digest, used as a sizing standard, appears in the outer lanes. The middle three lanes of the gel show a triplicate run of a 500-mer double-stranded DNA PCR reaction. Quantitation for such a separation is difficult and often imprecise, but such information can be obtained with the aid of a gel scanner. Recoveries of material from the gel are performed using procedures such as the Southern blot (10). Sufficient material is recoverable for sequencing or other bioassays. Separations of the sizing standard and 500-mer PCR product by HPCE using a size selective polymer network are shown in Figure 1.3. Quantitation is readily performed using peak area comparison with the standard. However, fraction collection is difficult relative to the slab-gel, particularly for trace impurities, since only minuscule amounts of material are injected into the capillary.
1.2 MICROCHROMATOGRAPHIC SEPARATION METHODS The evolution of chromatographic methods over the last 40 years has produced a systematic and rational trend toward miniaturization. This is particularly true lOn-line detection is performed on an instrument such as an automated DNA sequencer. ^Automated gel scanners can be used in place of gel archiving or photography.
Chapter 1
Introduction
If
FIGURE 1.2 Slab-gel electrophoresis of a 500-mer double-stranded PCR reaction product in a 1.8% agarose ethidium bromide gel. Courtesy of Bio-Rad.
for gas chromatography, where the advantages of the open tubular capillary displaced the use of packed columns for most applications. Chromatographic separations all function via differential partitioning of a solute between a stationary phase and a mobile phase. A packed column offers solutes "a multiplicity of flow paths, some short, the majority of average length, and some long (11)." Solute molecules select various paths through the chromatographic maze. The detected peak suggests this distribution and is broadened. In the open tubular capillary, the choices for solute transport are limited, so that the solute elutes as a narrow band. In order for the open tubular capillary to function properly, its diameter must be quite small. Larger diameter capillaries present a problem, since solutes away from the walls do not sense the stationary phase in a timely fashion. However, a major problem with narrow inner diameter (i.d.) capillaries is loading capacity. Injection sizes must be kept small to avoid overloading the system. In gas chromatography (GC) this problem is overcome in part, since sensitive detectors such as the flame ionization detector (FID), electron capture detector (ECD), and mass spectrometer are easily interfaced. Improved efficiency is one of several advantages obtained through miniaturization. The most important of those is improved mass limits of detection
1.2
Microchromatographic Separation Methods
500
i
IL 303
10
1746
^^
15 TIME (min.)
20
FIGURE 1.3 Capillary gel electrophoresis of a 500-mer (top) double-stranded PCR reaction product and a low molecular weight sizing standard (bottom). Capillary: 50 cm x 50 [im i.d. Bio-Rad coated capillary; buffer: 100 mM tris-borate, pH 8.3, 2 mM EDTA with linear polymers; injection: electrokinetic, 8 kV, 8 sec; detection UV, 260 nm. Courtesy of Bio-Rad.
(MLOD). Since dilution of the solute is minimized in the miniaturized system, better MLODs are obtained than in large scale systems. This is particularly important when the available sample size is small, as sometimes happens in biomolecule separations. Miniaturization of GC has been exquisitely successful. These triumphs could not be directly transferred to liquid chromatography (LC) for several reasons. The most important is the lack of good detectors. Interface to the FID and ECD is not practical due to the incompatibility of the mobile phase with each detector. Pumping of the mobile phase at the low flow rates required by miniaturization is also more complex, particularly when gradient elution is required. Despite these problems, |I-LC systems are useful in sample-limited situations and for mass spectrometry where the reduced liquid flow rate is advantageous. Several books have been devoted to this important field (12-14).
6
Chapter 1
Introduction
Most of work with \i-LC employs 250 |im i.d. packed columns, and so the advantages enjoyed by open tubular GC are not realized in |Li-LC. The instrumental problems of injection and detection posed by open tubular LC have inhibited most people from using this technology
1.3 CAPILLARY ELECTROPHORESIS The arrival of HPCE solved many experimental problems of gels. Use of gels is unnecessary since the capillary walls provide mechanical support for the carrier electrolyte.3 The daunting task of automation for the slab-gel format is solved with HPCE. Sample introduction (injection) is performed in a repeatable manner. Detection is on-line, and the instrumental output resembles a chromatogram. The use of narrow diameter capillaries allows efficient heat dissipation. This permits the use of high voltage to drive the separation. Since the speed of electrophoresis is directly proportional to the field strength, separations by HPCE are faster than those in slab-gels. On the other hand, the relative speed of the slab-gel is enhanced, since multiple samples can be separated at once. HPCE is a serial technique; one sample is followed by another. This limitation has been overcome through the use of the capillary array for high throughput applications such as DNA sequencing (15,16) and serum protein analysis (17). Commercial instruments are now available for these applications. HPCE represents a merging of technologies derived from traditional electrophoresis and high performance liquid chromatography (HPLC). Both HPCE and HPLC employ on-line detection. Developments in on-column micro-LC detection have directly transferred over to capillary electrophoresis. One of the modes of HPCE, micellar electrokinetic capillary chromatography (Chapter 4), can be considered a chromatographic technique. Electrically driven separations through packed columns (Chapter 7) have been reported from many laboratories. While there is much in common between chromatography and electrophoresis, the fundamentals of HPCE are based on electrophoresis, not chromatography. Professor Richard Hartwick, formerly from the State University of New York at Binghamton, started many of his lectures on capillary electrophoresis with a discussion of transport processes in separations. While performing a separation, there are two major transport processes occurring: Separative transport arises from the free energy differences experienced by molecules with their physicochemical environment. The separation mechanism may be based on phase equilibria such as adsorption, extraction, or ion exchange. Alternatively, kinetic processes such as electrophoresis or dialysis provide the mechanism for separation. Whatever the mechanism for separation, each individual solute must have unique transport properties for a separation to occur. ^Gels are occasionally used in HPCE for running size separations. Pumpable polymer networks are preferred, since they can be changed for each run.
1.3
Capillary Electrophoresis
7
Dispersive transport, or band broadening, is the sum of processes of the dispersing zones about their center of gravities. Examples of dispersion processes are diffusion, convection, and restricted mass transfer. Even under conditions of excellent separative transport, dispersive transport, unless properly controlled, can merge peaks together. According to the late Professor Calvin Giddings as paraphrased by Hartwick, "separation is the art and science of maximizing separative transport relative to dispersive transport." In this regard, capillary electrophoresis is perhaps the finest example of optimizing both transport mechanisms to yield highly efficient separations. Figures 1.4 and 1.5 illustrate this concept, using a series of barbiturate separations to compare HPCE and HPLC. The mode of electrophoresis used in Figure 1.5 is micellar electrokinetic capillary chromatography (MECC), an electrophoretic technique that resembles reversed-phase LC. In the LC separation amobarbital and pentabarbital coelute, but they are resolved by HPCE. With some optimization work, amobarbital and pentabarbital can be separated by HPLC. But with HPCE, methods development often progresses rapidly because of the enormous peak capacity of the technique. Peak capacity simply describes the number of peaks can be separated per unit time. With a couple of hundred thousand theoretical plates,"^ many separations occur without extensive optimization efforts. In addition, peak symmetry is excellent using HPCE unless wall effects (Section 3.3) occur. With the absence of a stationary phase, many factors that contribute to peak broadening and tailing are minimized. It would be misleading to state that all separations are superior by HPCE or that methods development will always be straightforward. It is realistic, however, based on the experiences of many separation scientists skilled in the art of both techniques, to predict that HPCE will provide the requisite speed and resolution in the shortest possible run time with the least amount of methods development, under most circumstances. These same two figures illustrate an important limitation of HPCE, the concentration limit of detection (CLOD). In Figure 1.4, the LC separation requires a 1.25 |Lig/mL solution to give full scale peaks with 1-2% noise (the postcolumn reagent merely alkalized the mobile phase, permitting sensitive detection at 240 nm). The CLOD is approximately 30-fold better by HPLC. The MECC separation shown in Figure 1.5 required a solute concentration of 100 |ag/mL for a similar response, although the noise was lower (0.5%).^ On the other hand, the MLOD by capillary electrophoresis exceeds HPLC by a factor of 100. The ideal detector for HPCE will be mass sensitive and not depend on the narrow optical pathlength defined by the capillary itself. Descriptions, advantages, and limitations of many HPCE detectors can be found in Chapter 9. ^The theoretical plate (N) is a measure of the efficiency of a chromatographic of electrophoretic peak; N = 5.5'\(t^/Wiy, where t^ is the migration time and W is the peak width at half height. 5The CLOD can easily be improved through the use of stacking and/or extended pathlength flowcells.
8
Chapter 1
W^
wW
Introduction
u
TIME (MIN.) 11 FIGURE 1.4 Reversed-phase liquid chromatography of barbiturates. Column: Econosphere Cis, 25 cm X 4.6 mm i.d.; mobile phase: acetonitrile : water, 55/45 (v/v); injection size: 20 jxL; flow rate: 1.2 mL/min; postcolumn reagent: borate buffer, pH 10, 0.2 mL/min; detection: UV, 240 nm; solutes: (1) barbital, (2) butethel, (3) amobarbital and pentabarbital, (4) secobarbital; amount injected: 25 ng of each barbiturate from a 1.25 |Llg/mL solution.
The preceding comparison is significant since a |Li-separation technique is compared with conventional HPLC using a 4.6 mm i.d. column. Would it be better to compare HPCE with |i-LC? Perhaps so from an academic standpoint, but this would not reflect the current usage and thinking in the real world. Chemists are contemplating using HPCE to replace or augment conventional HPLC as well as |i-LC. Table 1.1 provides a comparison of slab-gel electrophoresis, |I-LC, HPLC, and HPCE. Two disadvantages of HPCE compared to conventional HPLC are sensitivity of detection and precision of analysis. These have prevented the most widespread use of HPCE. On the other hand, HPCE is replacing the slabgel for most high-throughput DNA applications. In this case, the ease of automation, precision and ruggedness of HPCE supercede the slab-gel.
1.3
C apillary Electrophoresis
X TIME (MIN.)
10
FIGURE 1.5 Micellar electrokinetic capillary chromatography of barbiturates. Capillary: 50 cm (length to detector) X 50 |lm i.d.; buffer: 110 mM SDS, 50 mM borate, pH 9.5; injection: 1 sec vacuum (5 nL); detection: UV, 240 nm; solutes: (1) phenobarbital, (2) butethel, (3) barbital, (4) amobarbital, (5) pentobarbital, (6) secobarbital; amount injected: 500 pg of each barbiturate from a 100 |lg/mL solution.
HPCE is a novel and alternative format for both liquid chromatography and electrophoresis. The unique properties of this technique include the use of: 1. 2. 3. 4. 5. 6.
capillary tubing in the range of 25-100 jim; high electric field strength; on-line detection in real time; only nanoliters of sample; limited quantities of mostly aqueous reagents; and inexpensive capillaries relative to HPLC columns.
The molecular weight range of analytes separable by HPCE is enormous. A search of the literature reveals applications covering small ions, small molecules,
10 TABLE 1.1
Chapter 1
Introduction
Comparison of Slab-Gel Electrophoresis, p-LC, Conventional LC, and HPCE Slab-Gel
p-LC
HPLC
HPCE
Speed
slow
moderate
moderate
fast
Intrumentation cost
low
high
moderate
moderate
CLOD
poor
poor
excellent
poor
MLOD
poor
good
poor
excellent
Efficiency
moderate
moderate
moderate
high
Automation
Htde
yes
yes
yes
Precision
poor
good
excellent
good
Quantitation
difficult
easy
easy
easy
Selectivity
moderate
moderate
moderate
high
Methods development
slow
moderate
moderate
rapid
Reagent consumption
low
low
high
minimal
Preparative mode
good
fair
excellent
poor
good
good
excellent
good
excellent excellent poor
fair good excellent
fair good excellent
excellent excellent excellent
Sensitivity
Ruggedness Separations DNA Proteins Small molecules
peptides, proteins, DNA, viruses, bacteria, blood cells, and colloidal particles. The molecular weight range of HPCE is easily from 3 for a lithium ion to 100,000,000 for a virus or particle.
1.4 CAPILLARY ELECTROCHROMATOGRAPHY A hybrid of chromatography and electrophoresis, capillary electrochromatography (CEC) employs the electrically driven electroosmotic flow (EOF) to pump a mobile phase through a packed capillary. The use of the EOF to generate flow solves some of the instrumental problems of pumping at nL flow rates. Capillary electrochromatography employs small diameter capillaries filled with a stationary phase. Reversed-phase packings are most often used, although an application with a cation-exchange material has been reported (18). An amazing efficiency 8 million plates per meter was reported in that paper, though the mechanism and reproducibility of the effect are still unclear.
1.6
Historical Perspective
11
Typically, 50 |im i.d. capillaries are used though larger diameter tubes can be employed at the expense of efficiency. Particle diameters of 3-5 |im porus material are most common, though it is possible to employ 1.5 |Lim pellicular packing. Since there is no pressure drop with an electrically pumped system, relatively long capillaries can be employed to generate hundred of thousands of theoretical plates. The reduction of eddy diffusion also contributes to the enhanced efficiency (19). The mobile phase is pumped using the EOF generated by both the wall of the capillary and the chromatographic packing. Formulation of the mobile phase is similar to conventional reversed-phase chromatography, except that a dilute buffer—for example, 1-10 mM tris, borate, or phosphate—is added to ensure sufficient electrical conductivity The capillary is usually pressurized to a few atmospheres to suppress bubble formation. The least mature of the electrically driven techniques, CEC capillaries and second generation instruments are now available. One promise for this technique is the ability to employ the vast existing chromatographic database to speed methods development.
1.5 MICROMACHINED ELECTROPHORETIC DEVICES Employing technology used in the fabrication of integrated circuits, it is now possible to create an electrophoretic apparatus on a chip (20-28). Designed for dedicated applications such as clinical analysis, genetic analysis, or DNA sequencing, chips can be manufactured at low cost in commercial quantities. These devices can form the basis of an automated laboratory, where the disposable chip serves as the separations device. A diagram of a simple micromachined HPCE chip is shown in Figure 1.6. The technological advantage of this device compared with a conventional capillary is its ability to perform extremely small injections (29). As a result, a shorter separation channel is required, again compared with the conventional capillary. Detection problems resulting from the small injection are solved through the use of laser-induced fluorescence (LIE). Micromachined electrophoretic devices are expected to have a huge impact in the DNA applications area.
1.6
HISTORICAL PERSPECTIVE
A century of development in electrophoresis and instrumentation has provided the foundation for HPCE. Reviews describing the history of electrophoresis were published by Vesterberg (30) and Compton and Brownlee (31). The highlights in the development of HPCE are given in Table 1.2.
12
Chapter 1
Introduction
Background Electrolyte A oil B o-
Sample
oSeparation Channel Detector Window FIGURE 1.6 Layout of the channels in a planar glass substrate. Channels are referred to by number and inlet points (reservoirs) as letters. Each channel is labeled with its content or its function. Overall dimensions are 14.8 cm x 3.9 cm x 1 cm thick. The location of one pair of platinum electrodes is shown; for clarity, the others are not. (A) BGE reservoir; (B) sample reservoir; (C) outlet reservoir. (1) BGE inlet; (2) sample inlet; (3) separation channel; (4) sample outlet. Injection is made where 4 crosses 3. Redrawn with permission from Anal. Chem., 64, 1926 (1992), copyright © Am. Chem. Soc.
A direct forerunner of modem CZE was developed by Hjerten in 1967 (32). To reduce the detrimental effects of convection caused by heat production, the 3 mm i.d. capillaries were rotated. While heat dissipation was unchanged, the rotating action caused mixing to occur within the capillary, smoothing out the convective gradients. In the 1970s, techniques using smaller i.d. capillaries were successfully developed (34). Superior heat dissipation permitted the use of higher field strength without the need for capillary rotation. In 1981, Jorgenson and Lukacs (35) solved the perplexing problems of injection and detection with 75 |Lim i.d. capillaries. Their advances clearly defined the start of the era of HPCE. Fluorescence detection was required at that time to record the electropherogram. The 1980s proved ripe for invention. Adaptation of gel electrophoresis (36) and isoelectric focusing (38) to the capillary format was successful. In 1984, Terabe et al. (37) described a new form of electrophoresis called micellar electrokinetic capillary chromatography (MECC). Chromatographic separations of small molecules, whether charged or neutral, were obtained by employing the micelle as a "pseudo-stationary" phase. Great advances in detection occurred during the 1980s to overcome, in part, the serious limitation of the short pathlength defined by narrow i.d. capillaries
1.6
u Pu
PC o U
u
W5
c/5 c« 'tob
a
^
D. c« u c:3 f^
O
^ TJ
s
o o fN
^
•
^
CD
I
W U
O O O in
>^
OH
P^nd
B T!UJ X 0
—1
d
00
rn
O
The Language of Electrophoresis
< I—I
2 o
1^
hs. ^
.-I
ON
in
(N
^ O
s
in rH
ON ON
u
u
=^0 ^ r ^ ON
5 o ^j
'
60 '
XT
^
^ •
O
.c
•
V
40
•
^
X T .
^
D
0
•^ •=-•
0 D
^ ^
D
V ^- u"
20 ¥
1 l±lll
I
^
D
^^o°
^ ^t°
^¥ ^ 0 ° ¥^^D
1
H
10 20 VOLTAGE (kV)
_—
1-
30
FIGURE 2.17 Ohm's law plots for capillary temperature control by air circulation. (A) no control; (B) 25°C; (C) 10°C; (D) 4°C. Redrawn with permission from J. High Res. Chromatogr., 14, 200 (1991), copyright © 1991 Dr. Alfred Heuthig Publishers.
2.7
49
Optimizing the Voltage and Temperature
A
100
' "V
1
•
• B •
80
V
•
1 LU
•V
•
•
^ C
60 ^
•
^
•
OL
-^
A
•
O
•
A
A
0
0
^ 0 „ «
°
40 • •
20
•
X •
-6. a -
o -S ,g QJ "w
a
PQ
CQ
2
>
XJ
-
w w w w w w u u u u u u
6 o
CO
> 1—t
IT! C ^ «5
U U
M
PU Cu
o o o GH
:3 N U
E
1o
<X)
o
3
^
U
t^ 1—I
X
^ u o ^
•—t
'DO
t/^
C
;3 u
'lO
X
o
^
CQ Q
W ^ O . ir5 ^•5 d E u 3
1 o(c« CJ
>
I
ci: £ • u 3 i^ o p-
-Ti u
>: w w w w O Q p. U U U U
^
(1^ Q
a 'S o
u 3 w
m
O
U^
71 00 ;3
(S
u
C Q>
O Pu
< Z fS
c: t3 >
w 5 --^
X o p o
u uH
<
"-r±r ^rr™«Llrr"t:^••r^r«-Il!::r:a--rrrff^illrZa ':s5s*«lll::zi±s»s^^
6
7
8
9
10
11
pH FIGURE 3.3 2.5 to 11.
Solution for the Henderson-Hasselbalch equation for bases of specified pK from pH
These calculations can be useful to correlate mobility with pH. Grossman et al. (6) developed an empirical relationship that linked mobility to a complex function, ln(q + l)/nO^^, where q is the charge and n is the number of amino acids. Deyl et al. (7, 8) and Rickard et al. (9) found a best fit correlation for mobility with q/M^'^, where M is the molecular weight. This is consistent with Offord's model, which describes mobility of large molecules. For small molecules, Ml/3 provides a better fit. Grossman's model (6) falls between these two values, as might be expected when separating peptides containing between 3 and 39 amino acids. The accuracy of the q/MW^/^ versus mobility model is illustrated in Figure 3.4 (9); in the figure, data from a series of peptides from two separate digests separated by CZE at three different pH values are plotted. If the pKa and molecular weight of a substance are known, the use of mobility calculations to select the initial experimental conditions can be a worthwhile undertaking. Although optimal separation conditions cannot be predicted using this model, the calculations are effective as a first approximation. The profound effect of pH on mobility is illustrated in Figure 3.5 for two peptides differing by one amino acid with sequences AFKAING and AFKADNG (10). At pH 2.5, the calculated charges on these two peptides are 1.41 and 1.36, respectively. At pH 4.0, the calculated charges become 1.02 and 0.46. It is expected and observed that greater resolution is found for the higher pH buffer.
78
Chapter 3
Capillary Zone Electrophoresis
3.000e-4' o
o S
a
CL O w O
UJ
,
1.000e-4'
a
^
o o n
> « ^
'V' -1.000e-4
»
o
o
no a
. ^ y^^CsL
.!)noOA^-
-0.04
•
*
•
1
-0.02
'
»
1[
»
-0.00
1
0.02
'
'»'
»
"
1
" »' •
0.04
1
»
' '1
o.oc
FIGURE 3.4 Correlation of mobility and q/MW^^ for a human growth hormone digest (separated at pH 2.35, 8.0, and 8.15) insulin-like growth factor 11 digest (separated at pH 2.35 and 8.15). Reprinted with permission from Ana/. Biochem., 197,197 (1991), copyright © 1991, Academic Press.
since the mobilities are better distinguished. While one would expect longer migration times as the pH is increased (the charge on the peptides is less positive) , the increase in the EOF partially negates this effect. During the course of methods development, peak broadening and/or peak tailing may be noted. This may be due the adherence of the solute to the capillary wall. If this occurs, the wall effects must be eliminated before any meaningful methods development can be completed.
3.3 SOLUTE-WALL INTERACTIONS A. THE PROBLEM OF WALL EFFECTS A key advantage of HPCE compared with HPLC is the absence of the chromatographic packing. The vast surface area of the packing material is responsible in part for irreversible adsorption of solutes, particularly proteins. The composition of the capillary surface, though, still provides opportunities for protein adsorption. Binding of solutes to the capillary wall leads to band broadening, tailing, and irreproducibility of separations. If the kinetics of adsorption/desorption are slow, broadened tailed peaks occur. Irreversible adsorption leads to modification of the capillary, altered EOF, and loss of resolution.
79
3.3 Solute-Wall Interactions
\ ^
10
10 TIME (min.)
FIGURE 3.5 Effect of buffer pH on the selectivity of peptide separations by CZE. Capillary length: 45 cm to detector (65 cm total) x 50 [im i.d.; BGE: citric acid, 20 mM, (A) pH 2.5, (B) pH 4.0; field strength: 277 V/cm; current: in A, 24 |lA, in B, 12 |lA; temperature: 30°C; detection: UV, 200 nm; peptides: (1) AFKAING, (2) AFKADNG. Reprinted with permission from Anal Chan., 61, 1186 (1989), copyright © 1989 Am. Chem. Soc.
Figure 3.6 illustrates the electrostatic binding of a protein to the capillary wall. At most pH values, the capillary wall has a negative charge due to silanol ionization. Separation of a protein at a pH below its pJ produces a cationic solute that ion-pairs to the capillary wall. Hydrophobic binding may occur as well between the epoxide moiety of fused silica and a hydrophobic solute. Since most separations occur in aqueous media, hydrophobic solutes are not well solvated, further enhancing this potential binding mechanism.
80
Chapter 3
Capillary Zone Electrophoresis
•MUMUttUiiriiiiiUiuifaUriiyd^^
FIGURE 3.6 Illustration depicting the ion-pair formation between a positively charged protein and the negatively charged capillary wall.
The problem of wall binding is most severe for large molecules. This is easily understood from the illustration in Figure 3.7. A small molecule can have but a single point of attachment to the capillary wall. A large molecule can lie
LARGE MOLECULE
SMALL MOLECULE
FIGURE 3.7 Ion-pair formation between large molecules results in multiple points of attachment with the capillary wall. This is not possible for small molecules.
3.3 Solute-Wall Interactions
81
down on the wall and ion-pair in many places. It then becomes for difficult for the large molecule to dislodge from the wall, since all points of attachment must simultaneously be broken. For small molecules, wall effects usually result in Gaussian band broadening and the effect is slight. This has been studied for lanthanide ions (11), where a coated capillary proved more efficient. Adsorption or retention in HPCE is determined by a solute's adsorption/desorption kinetics with the capillary wall. A first approximation of the impact of wall effects can be understood using a random walk model from chromatographic theory (12). Random walk theory considers a solute moving down the capillary in discrete steps. The peak variance is expressed as ^s = 2 ( - ^ ) ^ v , p t , L , 1 + fe
(3.3)
where t^ = the time for adsorption. Retention occurs whenever t^, the time for desorption, is greater than t^, and by definition, t^ = tjk\ The time for adsorption, t^, is a function of the diffusion coefficient, the capillary diameter, and the probability of binding to the capillary wall. Table 3.2 contains data showing the molecular weight and the diffusion coefficient for various molecules. The kinetics of mass transport to the capillary wall are slower for large molecules, and this in part indicates wall effects are more severe as the molecule weight increases. When solutes adhere to the wall, peak tailing may be observed since a desorbed solute does not return at once to the buffer solution. Retained solutes have a migration velocity of zero. Solutes in the buffer move at a rate determined by their migration velocity and, thus, move ahead of retained material. If we
Table 3.2
Diffusion Coefficients of Large and Small Molecules
Compound
Molecular Weight
D(cmV X 105)
j3-alanine
89
0.933^
Phenol
94
0.84^
Citric acid
192
0.66P
Cytochrome c
13,370
0.114^
^-Lactoglobulin
37,100
0.075'^
Catalase
247,500
0.04F
Myosin
524,800
O.OIF
Tobacco mosaic virus
590,000
0.0046^
^From Handbook of Physics and Chemistry, 46th ed., 1965, CRC, p. F46. ^From, B. L. Karger, L. R. Snyder and C. Horvath, An Introduction to Separation Science, 1973, John Wiley & Sons, p. 79. ''From A. L. Lehninger, Biochemistry, 1970, Worth Publishers, pp. 136-137.
82
Chapter 3
Capillary Zone Electrophoresis
solve Eq. (3.3) for the variance and plot the decrease in the number of theoretical plates versus k' (Figure 3.8), the dramatic impact of wall effects on efficiency is apparent. To achieve the theoretical efficiency of CZE, k' must approach zero. As the figure illustrates, even modest retention will lead to severe band broadening. In the worst case scenario, no elution occurs—the solute is completely bound to the capillary wall. This simple random walk model only estimates the impact of wall effects on efficiency. More sophisticated calculations have appeared in the hterature in 1995 (13, 14). In any event, the appropriate buffer additives or capillary coatings are required to minimize this form of band broadening. Using a clever experimental apparatus with multiple detectors, Towns and Regnier (15) were able to measure the binding of proteins to the capillary wall. Some of their data are reproduced in Table 3.3. Under their experimental conditions, all proteins showed some binding. As expected, the high-pl proteins bound most strongly, owing to their positive charge at pH 7. Wall effects on bare sihca have proved to be a difficult problem since the early days of HPCE (16). Since then, several solutions have been proposed including the use of 1. Extreme pH buffers 2. High-concentration buffers 3. Amine modifiers
-J r\ a (fl -I O
as
OS
s a
-d n3
"o "o PH f X
f5
e3 o
•32
X
0
o
in
^
r—1
(N
ex
1
Xi
o
2
ON
ex
ex ex ex
y^ O
W
o ,13 fX
I
•5 IS n3 O
B
§ p. X
vo ON
X ex
rH
o
s o
X
NON
ex
(N
3 rP
"So
^
ex
00 ON
in
o ex
I—I
1
ON ON
o
to
"^
I—I
O
ex
-—I
O
rtj
^
PH
3
;§
in 00
93
94
I o u
(-hi
N U u
3 PQ
St2
PC —1 "-H
>-i
H ,^
1/5
::3 TT: (y) LO
:7: ^
o
PLH
B
^ «
o>
c^
Xi
c
o u
w
XO
2OS
>
.2 '5C
3 TJ
O
< Ij
;3
S Z^ 2
'u 13
o
:;S ty-)
•—I
S B
a. U
o o 1—I
:s s
O
i-i
O -C O Xi CUXi
1—1
ss ss
•"^
CL,
b b
s so oi n o
LO
s
S
in
0 42
pH
1—I
0
0
"S
0 fN
'^ 0 0 P^ 0
0 r—1
1^
in
#
w H
!U
OS
3N n:3
1 5S
p^
0 (N
P-
S
,n
0
7^ a 0
3
0
' ^ D
(N
D. (N
u
]3 _c
to
8 a P3
5 l
u^ O ^y^
S B
X
ON
TS
I—I
-^
O
Of)
'2 Co
o o
_bjD
O
_o
TO
's t7^
OS
i'^ PH
r^
'o
fN
^
'2 i§ 3CO
u
Capillary Zone Electrophoresis
D.
K
B
o O
S S o
O
a
in
^
in
o
uS
T!
g
"C
3
H 4J
c/f —I
o
i n o o o r N r ^ h - o o o N O - — i c ^ ( N m
c in •—'
a^
(N
o" o (N
P
o
O
.—I
o
^
^ 1 — l i n o
o o rvj
o^
U
D
fc >' >' J
00
S B in
u
S S
T^ C
CT5 C/) CTJ )-l aj i->
C3
-^ -7^
CJ
•c rs
2 (/5
> ^fH* O
cn
2 ^^ .^
S
in -^ in
^
7-*
U
168
u
^ ^ ^
^ ^
< "3 en H
.2 t^
Pi O
p
1^1 " (in
S 2 2 B ^ B
.3 u P ^ S X
CO
11
iM It II
^^
^
CJ
O
N
3
PH
.
O
o
rH
o
•p
B
s ^-B
PH
P i Q
>o m
:^
O
u
PC
^
X
£
a O
(/)
o
s s o o (N
H
p
00
to
B I/)
^p g Q
bJO
V5
^p
p
P
C _C
c«
£
p
p
,—I
o o
^ O P . rH O to - ^ 'C ^ ;C p. H
P3
O )^
6 2
6 I "^
in
< ^ ^
^
rA (5 d
s
-^3
o
w
O
o
o ^ ^
00
^ ^
2 >^I N- ^_, in i^ S
!-4
Q
H 2 in
'(J
o in
f^
ON
1 '^
n ^ -^ c«
^
v!:/
d
^
^ '^ o ^ fi o m
B u O (N
B
i2
O o
PQ
Q
S
U
3O in X in X
B
a o^
^ ^ 3in I-X
u o
B
^ ^ O
o O
c
>>
g
E
O
^ ^ PS
O
O
3
>> 43 u o
r2
ON
(75
a
o
43
o
o
;-i
o
f
0
^ W)
^
^ s
o
^ H o xJ
2M
X3
^ B ^
P^
o -5
o ^ -r: o a *^
Chapter 5 Capillary Isoelectric Focusing
H
S w ^
>> >^
in in in 1^ -H ^ '
^
^ . — I
£ ^
U
—I
(N Pi
W
p
w u
s6
o
c>^
0
2
T3
o
u
.-4
4 .—I
o ^
g P
5 pi
velocity profll©
ed channel
(E
M
3
FIGURE 7.5 Diagram of eddy diffusion and flow velocity profiles in HPLC and CEC. Courtesy of Hewlett-Packard Corporation.
Using 5-|im particles and a 50-cm capillary, the CEC separation is approximately twice as efficient as capillary HPLC, in agreement with the theoretical prediction. When the particle size is reduced to 3 |im, the HPLC column length can be no longer than 25 cm because of the 400-bar pressure limit. This limits the plate count to 45,000 by HPLC, whereas the CEC separation using a 50-cm capillary yields 170,000 theoretical plates. When L5-|im material is used, the 50-cm CEC capillary yields 250,000 plates, compared with 30,000 plates on a 10-cm HPLC capillary
7.5 OPERATING CHARACTERISTICS OF PACKED CEC A. PREPARATION OF CEC CAPILLARIES While most users will purchase prepacked capillaries, the steps in column preparation are worth considering (27). These are illustrated in Figure 7.8 on p. 306 and described in the following: 1. Prepare a temporary outlit frit by sintering silica gel particles with a microflame torch.
304
chapter 7
Capillary Electrochromatography
PRESSURE
yjL O
10 ELECTRO
20
30
lyL 10
20 TIME (min)
30
40
FIGURE 7.6 Separation of polycyclic aromatic hydrocarbons on a drawn capillary packed with 3-|Lim Hypersil particles and derivatized in situ with octadecylsilane. Capillary: 90 cm (pressure), 80 cm (electro) x 30 |J.m i.d.; pressure (upper): 25 bar; voltage (lower): 320 V/cm; detection: fluorescence. Order of elution: naphthalene, 2-methylanthracene, fluorene, phenanthrene, anthracene, pyrene, and 9-methylanthracene. Reprinted with permission from Chromatographia, 32, 317 (1991) copyright © 1991 Vieweg.
2. Slurry pack the capillary with packing media dispersed in methanol at 350 bar for 3 h. 3. Prepare the permanent outlet frit using a thermal wire stripper. 4. Unpack the capillary by pumping from both ends. 5. Prepare the detector window. 6. Repack the capillary as in step 2. 7. Prepare the inlet frit by gently sintering the particles at the end of the capillary. While there are certainly many variations of this technique, the packing of these capillaries is an art form, and there is a high reject rate. The frit in particular is prone to problems. Frits must retain the chromatographic packing yet be sufficiently porous to allow the passage of solvent. Details of frit production have been described, along with recommendations (28). Frits prepared from
7.5 Operating Characteristics of Packed CEC
PRESSURE / / OraVEN \ / /
612
305
\\ \\
ELECTRICiUXY DRIVEN
600 DISTANCE (mm)
588
FIGURE 7.7 Expansion of the fluorene peak from Figure 7.6. Outer curve: pressure-driven chromatogram; inner curve: electrically driven chromatogram. Reprinted with permission from Chromatographia, 32, 317 (1991) copyright © 1991 Vieweg.
pure spherical silica gel appear best. Since production of frits by sintering destroys the polyimide, the capillary must be carefully handled. Packed capillaries are available from many sources, including HewlettPackard, Hypersil, Micro-Tech Scientific, Capital HPLC Ltd., Phase Separations, and Unimicro Technologies, Inc.
B. COLUMN EQUILIBRATION The capillary must be flushed with mobile phase prior to use. This can be done with an external HPLC pump, a manual syringe pump,^ or the instrument internal pressure, particularly if a high-pressure mode is available. It may take several hours to totally condition a capillary. When all air bubbles have left the capillary, it is best to further condition the capillary at low voltage. Alternatively, the capillary may be conditioned by pressurizing the inlet to 10 bar, running a voltage gradient to 25 kV over 30 min, and holding the voltage at 25 kV for an additional 30 min (29). ^Procedure given in Unimicro Technologies operating instructions. Table 7.3
Achievable Plate Numbers in Capillary HPLC and CEC CEC
Capillary HPLC Particle Size (pm)
Length (cm)
Plates/Column
Length (cm)
Plates/Column 115,000
5
50
55,000
50
3
25
45,000
50
170,000
1.5
10
30,000
50
250,000
306
Chapter 7
Capillary Electrochromatography
Temporary outlet frit
4
3
5
Outlet frit
inlet frit
Detector window FIGURE 7.8
C.
Illustration of the processes for frit production and packing of a CEC capillary.
INJECTION
Electrokinetic injection is usually employed in CEC. The mobility discrimination often seen in CZE is less significant in CEC, since the EOF is the driving force for injection. Hydrodynamic injection can also be employed if a high-pressure injection mode is available. Since only nanoliters of material must enter the capillary, 8 bar of pressure for 30 s provides a sufficient injection (30). Sample loading is improved by injecting the sample dissolved in a solvent containing more water than the mobile phase (31). In this case, the solutes are retained at the head of the packed capillary and then are eluted by the mobile phase. This increases the loading capacity of the system to the nanogram range.
D. CAPILLARY AND MOBILE-PHASE SELECTION Start this process by selecting packings and mobile phases similar to those that have been successful in HPLC. The data given in Table 7.2 should be useful in
7.5 Operating Characteristics of Packed CEC
307
ensuring that a packing material with good EOF characteristics is selected. In addition to C^g material, octyl, phenyl, cyano, and amino reversed-phase materials are available. If specialty phases are required and a packing apparatus is available, a purchased HPLC column will provide a lifetime supply of packing material. Selection of particle sizes of 3 |im or less will yield the most efficient separations. Using reversed-phase material, select a pH no less that 2.5 (to maintain EOF) nor greater than 9 (to prevent silica dissolution). At pH 2.5, the EOF is 5 to 10 times lower than at pH 7. Use buffers such as Tris, phosphate, borate, and acetate at concentrations from 5 to 25 mM to adjust pH and maintain conductivity. The buffer concentration should be optimized, as chromatographic efficiency often improves at higher (25 mM) concentrations. If the buffer concentration is too high. Joule heating may adversely affect the efficiency of the separation. To avoid the pH dependency of the EOF, cation- or anion-exchange packings can be used. A propylsulfonic acid, either by itself or mixed with CIQ material (mixed-mode packing), can provide sufficient EOF, even at acidic pH (30). Acetonitrile is best as a mobile-phase modifier, because unlike methanol, it does not lower the EOF That is always an important consideration when selecting the appropriate modifier. Typical acetonitrile concentrations are 20-80%, depending on the application. Acetonitrile also is more optically transparent than methanol, which can improve detectability Before selecting a solvent other than methanol or acetonitrile, ensure that the material is compatible with the instrument, vials, and vial closures. Methanol can also be used as a mobile-phase modifier when acetonitrile is not appropriate. When using low concentrations of methanol or acetonitrile, the wetting characteristics of the BGE are poor. This can lead to bubble formation, which interrupts the electric circuit. Pressurizing the system with 8-10 bar over both the inlet and outlet vials will minimize bubble formation. The use of low-conductivity buffers such as Tris and MES has also been reported to help suppress bubble formation (32). Another way of suppressing bubble formation is to add small amounts of surfactants to the system. When using submicellar concentrations of SDS (1-5 mM), bubble formation is suppressed and the EOF may be stabilized (33). In this case, 20% methanol was the modifier. Such a low concentration of modifier permits the SDS to bind to the packing material and modify the stationary phase. At higher modifier concentrations, such binding does not occur, since the organic solvent effectively keeps the SDS off the packing material. Yet another way to suppress bubble formation is to use supplementary |I-HPLC pressurized flow. The application of a voltage to a |I-HPLC separation clearly improves the chromatographic efficiency Another stated advantage is that the separation does not entirely rely on the packing to produce the EOF (34, 35). This approach has been named pressure-driven CEC or electro-HPLC. Since the
308
Chapter 7
Capillary Electrochromatography
capillary is coupled to an LC system, the formation of gradients is simpler as well. However, it should be noted that one of the previously cited papers showed superior separations of oligonucleotides using MECC (34). While instruments employing gradient elution are just being introduced, a step gradient can be designed with almost any commercial instrument (36). The separation begins with the weaker mobile phase, and after a period of time, the voltage is removed and the inlet and outlet vials replaced with the stronger eluent. After completion of the run, fresh weak solvent is used to reequilibrate the capillary
E. VOLTAGE Most instruments function at a maximum voltage of 30 kV Higher voltages will result in rapid analysis, provided Joule heating is insignificant. Use of a 20-cm capillary with 55-kV applied voltage provides an isocratic separation of 16 polycyclic aromatic hydrocarbons in 2 min (37).
F. DETECTION Postfrit detection is always employed when using UV absorbance detection, since the capillary packing is opaque. Extended path length capillaries can also be used to increase the sensitivity of detection (29). When using LIF detection, on-capillary detection is possible (37). This approach yields the highest chromatographic efficiency (700,000 plates/meter) and indicates that postfrit detection results in band broadening. The absence of micelles in CEC is advantageous with regard to mass spectrometry (31, 38, 39). Typical mobile phases contain acetonitrile, ammonium acetate (31), or trifluoroacetic (40). A sheath flow of a few microliters per minute is frequently used to provide a stable electrospray, since the flow rate through the column is quite small. Through the use of nanoelectrospray, the sheath fluid becomes unnecessary. Moving closer to cutting edge technology, open tubular CEC has been interfaced to a time-of-flight instrument via the nanoelectrospray. The advantage of the time-of-flight instrument is that it is nonscanning. In conjunction with an ion trap, which enriches the ions, limits of detection for peptides reach 10"^ M (41). CEC has also been interfaced to a nuclear magnetic resonance spectrometer (42).
G.
TEMPERATURE
As in all other forms of HPCE, temperature control is particularly important. When increasing the capillary temperature from 20°C to 60°C, the retention time %RSD was lowered from 7.3% to 4.3%, and the analysis time was short-
7.6 Applications
309
ened by 45% (43). To keep the current low and aid in the suppression of bubble formation, low-temperature operation at 15°C has been recommended (30).
7.6 APPLICATIONS A representative selection of applications and chromatographic conditions is given in Table 7.4. Some other applications of note are described in this section. The open tubular separation of sulfonic acids (Figure 7.9, p. 312) on a 10-|Xmi.d. capillary coated with 0.9% PS-264 or 10% OV-17 employs an ion-pairing reagent, tetrabutylammonium hydroxide, as a mobile-phase modifier (49). The use of a narrow-bore capillary improved the mass transfer problem in accordance with the second term of Eq. (7.3). Improvements over CZE and pressure-driven LC separations were demonstrated with this ion-pair CEC system. Thus, the chemistry employed in conventional HPLC separations can be used in CEC as well. Separations of the drug Isradipin and its by-products (Figure 7.10, on p. 313) (45) present a good example of a separation with small V values (0.17-0.90). The large number of plates per unit of time is consistent with the theory, which predicts optimal efficiency at small fe' values. In this early work, the retention time precision ranged from 1.6% to 2.2% (run/run) and 9% (capillary/capillary). Production of the narrow-bore packed capillaries proved problematic in this early work and still does. High-speed separations by CEC are best accomplished by maximizing the electric field strength. Since most instruments are incapable of producing greater that 30 k y it is necessary to use short capillaries to provide the high electric field. This is effective when the separations are not too complex. Figure 7.11, on page 314 shows the separation of some aromatic hydrocarbons using 1.5|lm nonporous ODS particles packed in a 6.5 cm (10 cm total length) X 100 |Xm capillary (37). With the voltage set at 28 kV, the field strength is 2800 V/cm. Above 28 ky arcing between the capillary and the capillary holder was observed. This becomes another limiting factor when using high voltage, since the grounding and shielding within the instrument become most critical. On commercial instrumentation, it is usually necessary to use the "short end" of the capillary to produce such a short capillary length (52). Another factor that permitted operation at such a high field strength was the low conductivity of the BGE. With 2 mM Tris in 70% acetonitrile, the current was 6.7 |LiA at 30 kV. An Ohm's law plot showed a 20% deviation from linearity at 20 ky so that this homemade system could use better heat dissipation. Combined with on-capillary LIE detection to minimize dispersion from the frits, the separation time for this simple mixture is less than 5 s. With a peak width of 0.2 s, it is necessary to set the detector time constant to 0.02-0.03 s to minimize band broadening, and the injection size must be kept quite small. Like other forms of HPCE, whenever the peak widths become narrow, all instrumental parameters must be carefully controlled to maintain the inherent efficiency of the separation.
310
u u
u>> u
o U U H u^
O
3- i n m T3 2 C O U pTi
r^
< ^
^ in
^
Capillary Electrochromatography
O ^
^I
K
of Q do
o :3
B 3-
2
B
PQ
ro
O
3.
(N
cy^ (J
J
o
(/5
Q
O w
X
B
o o
1—1
^ X
in 1^
in (N
B CJ
Q
z 595 nm. Key: (1) mesoporphyrin (dicarboxyl); (2) coproporphyrin (tetracarboxyl); (3) pentacarboxyl porphyrin; (4) hexacarboxylporphyrin positional isomers; (5) heptacarboxyl porphyrin; (6) uroporphyrin (octacarboxyl). Reprinted with permission from J. Chromatogr., 516, 271 (1991), copyright © 1991 Elsevier Science Publishers.
B. IONIC-STRENGTH-MEDIATED STACKING Differences between the conductivity of the injection zone and the BGE have an impact on the field strength that is distributed over each zone. As described in Section 2.1, this is simply a consequence of Ohm's law. Since a solute's electrophoretic velocity is proportional to the field strength, differing velocities can
337
8.6 Stacking and Trace Enrichment
TIME <miiiJ FIGURE 8.7 Impact of injection size on resolution and sensitivity. Injection buffer, 20 mM CAPS, pH 11. Approximate injection size: (A) 5 nL; (B) 10 nL; (C) 25 nL; (D) 50 nL; (E) 100 nL. Other conditions as per Figure 8.6. Reprinted with permission from J. Chromatogr., 516, 271 (1991), copyright © 1991 Elsevier Science Publishers.
338
Chapter 8
Injection
be realized within each compartment. The basis of stacking is to provide a high field strength over the injection zone. This is accomplished readily by injection of low-conductivity solutions. The phenomenological situation at low pH values is illustrated in Figure 8.8. If a low-conductivity injection solution is employed, the field strength must be higher over that zone than over the remainder of the capillary. When the positively charged solutes migrate out of the injection zone and encounter the BGE, the field strength abruptly drops, and thus, the solute's electrophoretic velocity slows. Meanwhile, the solutes at the middle to rear of the injection zone are still exposed to the high field strength and continue to move forward at "full" speed. As a result, the ions in the injection band continue to narrow, until all have migrated into the BGE. The negatively charged counterions stack up as well, but at the rear of the injection zone (not shown). At high pH values, as shown in Figure 8.9, a more complex situation occurs because of the presence of the strong EOF In this case, the EOF drives all solutes toward the cathode. The negatively charged anions migrate toward the anode and cross the boundary between the injection solution and the BGE at the rear of the injection zone. Despite these differences in the pH-mediated direction of electrophoresis, the net result is band compression, the degree of which is related by the ratio of
LOW FIELD
HIGH FIELD
ii)®(^©®
a\:^'
FIGURE 8.8
Illustration of stacking of cations in a low-pH buffer.
339
8.6 Stacking and Trace Enrichment
HIGH FIELD
LOW FIELD
ep +
FIGURE 8.9
Illustration of stacking of anions in a high-pH buffer.
conductivities of the injection and separation electrolytes. These differences can be quantitatively expressed as (32)^ El
(8.8) PI
where Ei and Ej are the field strengths over the injection zone and balance of the capillary, respectively, pi and p2 are the respective resistivities in these regions, Ci and Cj are the respective buffer concentrations, and / i s the field enhancement factor. Since the current or ion flux that passes through the capillary must be constant through each zone, the steady-state concentration of solutes [SJ and [S2] is inversely proportional to the field strength and, thus, the electrophoretic velocity; hence. [SJVi ^ [ S j v ^ '^The symbols describing the parameters given in (32) have been simplified for clarity.
(8.9)
340
Chapter 8 Injection
SO that
i ^ = l ^ = r, [SJ
(8.10)
[Q]
Since [SJ must increase, this can only be accomphshed by compression of the ions in the injection zone. Thus, X3 = ^ ,
(8.11)
7 where x^ is the effective sample plug length after stacking and x^ is the initial length of the injection plug. Based on Eq. (8.11), the sample should always be prepared in water. Unfortunately, this is not optimal because of the generation of electroosmotic pressures. The measured EOF is based on the average electroosmotic contribution of each zone adjusted by the zone length. Since fluids can be considered incompressible, a hydrodynamic component is introduced whenever localized zones contribute differently to the total EOF This effect generates band broadening by contributing hydrodynamically enhanced diffusion. Calculation and plotting of these effects leads to Figure 8.10, which shows optimal results for field enhancement factors ranging from 5 to 20 (32). When yis small, the peak variance is proportional to the injection time. At high y values, the laminar-flowinduced broadening becomes significant. These theoretical calculations are supported by experimental data shown in Figure 8.11 (32). The optimal procedure for stacking is to prepare the sample in an injection buffer that is 10-fold more dilute than the BGE. Based on Figure 8.7, it is possible to trade off some plates for sensitivity. Performed properly, ionic-strength-mediated stacking can improve sensitivity by a factor of 10. At low pH, the low-ionic-strength injection buffer may still linger within the capillary and cause problems when very large injections are made (33). Other effects can be observed when using large-volume stacking hydrodynamic injections. Because the field strength over the point of injection becomes enormous, labile samples may decompose due to the generated heat (34).
C. IONIC-STRENGTH-MEDIATED ANTISTACKING The antithesis of stacking is antistacking. The antistacking mechanism is illustrated in Figure 8.12 for cations at low pH. When a sample with a high ionic strength relative to the BGE is injected, the electric field over the injection zone declines as defined by Ohm's law. When a positive ion electrophoreses into the BGE, it becomes exposed to the high field strength over the BGE. As a result.
8.6 Stacking and Trace Enrichment
Field enhancement factor^ Y
341
too
FIGURE 8.10 Plot of peak variance versus the field enhancement factor, y for several gravitybased (15-cm) injection times using a 100 cm x 75 jim i.d. capillary. Reprinted with permission from Ana!. Oitm., 63, 2042 (1991), copyright © 1991 Am. Chem. Soc.
the cation accelerates away from those cations, still remaining in the injection zone. The result is substantial band broadening. Figure 8.13 shows the antistacking of anions at high pH. The anion crosses the boundary between the injection plug and the BGE at the rear of the zone. Now exposed to the high field strength, the anion accelerates toward the positive electrode. Ions remaining in the injection plug migrate more slowly toward the anode and, as a result, are pushed more rapidly by the EOF toward the cathode. Though the dispersion occurs at the opposite end of the injection plug than for cations, the ultimate result is the same. Because of the phenomenon of antistacking, only small injections can be made when the ionic strength of the sample is high relative to the BGE. This same problem occurs in the slab gel (35). In that case, it was suggested to allow a timeout step after sample application, prior to applying the voltage. The timeout allows the rapidly diffusing salt ions to disperse from the sample zone. In the gel, 30 min was required for the timeout, and no diffusion-related band broadening of DNA was observed. While not reported in HPCE, a timeout may be possible, though not for 30 min. There have been several proposed schemes
342
Chapter 8
Injection
6 minute minute Injection with pure water
c
3-
2
1
3. c
I SP
1'" < 4-
J
iljj
6 minute in
c 3
3-
€CO
1
1
J? 2 c
-p , o «0 1 X2
Ms and [S2] > [SJ.
357
8.6 Stacking and Trace Enrichment
The goal is to employ ITP only briefly during the injection process for enrichment, followed by separation via CZE. In doing so, CITP is employed only for a short time, and then the conditions for CZE are restored. In other words, a heterogeneous buffer (leader-terminator) system exists transiently, followed by restoration to the homogeneous environment of CZE. There are many approaches for meeting these requirements (64-75). First, a good CZE separation must be developed. The only new requirement is that the buffer co-ion must serve as a leader or a terminator. A series of ITP electrolyte systems is given in Tables 8.5 and 8.6.^ If a leader is selected, then fill the capillary with that buffer, inject 10-50% of the capillary with sample, and immerse the capillary in a terminating buffer. Activate the voltage for 30-90 s, and then return the inlet side of the capillary to the leading electrolyte and continue the run. In this example, after the focusing step, the leading ions quickly overtake the terminator and sample ions, and now a leader-sample-leader situation is regenerated, as required for CZE. In another mode, leader or terminator is added to the sample itself. If a sufficiently large injection is made, the conditions for ITP are fulfilled for a short time prior to the system automatically converting to CZE. If you are seeking to enrich trace components in the presence of a major component, the major component itself can serve as a leader or terminator. This is termed sample selfstacking (64, 65). Isotachophoresis can occur by accident if any of the foregoing conditions are met (30, 76). This is illustrated in Figure 8.24 using the urinary porphyrins applications described in Section 4.8. The injection is a urine sample from an ^Provided by Vladislav Dolnik from the CZE-ITP Internet Discussion Group.
Table 8.5
Buffers for Anionic ITP^
pH
Base
Terminator
3.6
P -Alanine
Glutamic, nicotinic, or pivalic acid
4.3
EACA^
Pivalic acid
4.9
Creatinine
IVIES'^
6.1
Histidine
IVIES'^
7.1
Imidazole
Hepes'^ + barium hydroxide
8.1
Tris
Glycine + barium hydroxide
9.5
Ethanolamine
EACA^ + barium hydroxide
^Use 10 mM hydrochloric acid (chloride as leader) with 20 mM base. ^'e-Amino-N-caproic acid. *^2-(N-Morpholino)ethanesulfonic acid. '^N-Cl-Hydroxyethyl) piperazine-N'-(2-ethanesulfonic acid).
358
Chapter 8
Injection
Table 8.6 Buffers for Cationic ITP^ pH
Base
4.7
Acetic acid
Acetic acid (proton as terminator)
6.1
MES^
Histidine
Terminator
7.2
MOPS^
Imidazole
8.1
Tricine
Tris or triethanolamine
8.7
Asparagine
Tris
9.8
Glycine
Ammonium hydroxide
10.3
/^-Alanine
Ammonium hydroxide or methylglucamine
^Use 10 mM potassium hydroxide (potassium as leader) with 20 mM acid. '^2-(N-Morpholino)ethanesulfonic acid. ''3-(N-Morpholino)propanesulfonic acid.
individual not having a porphyria. In Figure 8.24A, note the appearance of a spike for peak 6, uroporphyrin. This spike corresponds to milhons of theoretical plates. The spike was repeatable and thus not attributable to arcing or specks of material. The sample was next fortified with porphyrins, and again, only the uroporphyrin is focused (Figure 8.24B). Upon reduction of the injection size (Figures 8.24C and D) the uroporphyrin peak broadens, whereas the other peak widths remain the same. It is likely that endogenous chloride from the urine serves as the leading electrolyte and the mechanism of stacking is tITP Since only uroporphyrin is focused, the terminating ion (unknown, perhaps SDS or CAPS from the BGE) must have a mobility lower than that of uroporphyrin but greater than that of any of the other porphyrins. When a small injection is made, there is insufficient time for the chloride to set up as a transient leading zone prior to its migration toward the inlet (positive electrode) and out of the capillary While the solutes elute toward the negative electrode (by the EOF), electrophoresis of the anions is toward the anode, and this is anionic ITP Proof that tITP is occurring is obtained by increasing the injection size and observing a sharpening of the focused peak(s). J. PRACTICAL ADVICE Considering that tITP is an advanced procedure and not for the faint of heart, it is critical for this and all stacking techniques to carefully control the conductivity of the sample. With tITP, this is particularly important, since the injection size is so large. Regardless of which stacking technique is employed, smaller injections always provide for a more robust separation. When selecting a stacking procedure, it is advisable to start with the simpler methods, such as ionicstrength-mediated stacking.
359
8.6 Stacking and Trace Enrichment
MATRIX EFFECTS * UHtNi D
SPiKE, 2 s
C
SPIKE, 3s
W*WJ'w 'f»^
B
SPIKE, ,^J^^^0i/^^^^
MS^
16 TlfVIECmln.) FIGURE 8.24 Impact of injection time on the tITP of urinary porphyrins. Conditions and key as per Figure 8.6. Sample: (A) urine from a porphyria negative individual; (B-D) urine spiked with 300 pmol/mL porphyrins. Injection times as specified on figure. Reprinted with permission from J. Chromatogr., 516, 271 (1990), copyright © 1990 Elsevier Science Publishers.
The size of the injection is dictated by the requirements of the Umit of detection. It is often prudent to employ extended path length capillaries and offline sample preparation to help meet the required LOD. This eases the requirements for stacking and provides for more stable separation conditions.
K. MEMBRANE AND LC-BASED ENRICHMENT The development of a device for online trace enrichment has not been straightforward. A few years ago, a capillary containing a 1-2 mm plug of a polymeric
360
Chapter 8
Injection
reversed-phase packing was commercially available, but it no longer is (Jl). Large volumes of aqueous injection buffer can be loaded into the capillary by electrokinetic injection. Enrichment occurs through binding of hydrophobic solutes to the polymeric packing. After loading, a small volume of organic solvent is injected to elute the solutes into the capillary, after which CZE is performed in the usual manner. CLODs for peptides as low as 1 ng/mL have been reported, with migration time and peak area precision of better than 1.5%. There is ongoing research into the development of a membrane-based device to provide for online trace enrichment (78-84). These membranes allow injection sizes of 1 |LiL. Solutes are enriched by hydrophobic interaction with the membrane. Such a large sample injection would solve most sensitivity problems in HPCE, but these membranes are not commercially available. They remain a research tool.
REFERENCES l.Foret, E, Demi, M., Bocek, P. Capillary Zone Electrophoresis: Quantitative Study of the Effects of Some Dispersive Processes on the Separation Efficiency. J. Chromatogr, 1988; 452:601. l.Otsuka, K., Terabe, S. Extra-column Effects in High-Performance Capillary Electrophoresis. J. Chromatogr., 1989; 480:91. 3.Jones, H. K., Nguyen, N. T., Smith, R. D. Variance Contributions to Band Spread in Capillary Zone Electrophoresis. J. Chromatogr, 1990; 504:1. 4.Jacobson, S. C , Hergenroder, R., Koutny, L. B., Warmack, R. J., Ramsey, M.J. Effects of Injection Schemes and Column Geometry on the Performance of Microchip Electrophoresis Devices. AnalChem., 1994; 66:1107. 5. Englehardt, H., Cunat-Walter, M. A. Use of Plate Numbers Achieved in Capillary Electrophoretic Protein Separations for Characterization of Capillary Coatings. J. Chromatogr, A, 1995; 717:15. 6. Lloyd, D. K., Waetzig, H. Sodium Dodecyl Sulfate Is an Effective between-Run Rinse for Capillary Electrophoresis of Samples in Biological Matrixes. J. Chromatogr, B: Biomed. Appl, 1995; 663:400. 7. Huang, X., Coleman, W. E, Zare, R. N. Analysis of Factors Causing Peak Broadening in Capillary Zone Electrophoresis. J. Chromatogr, 1989; 480:95. 8.Dasgupta, P K., Surowiec, K. Quantitative Injection from a Microloop. Reproducible Volumetric Sample Introduction in Capillary Zone Electrophoresis. Anal Chem., 1996; 68:1164. 9. Evans, C. Direct Online Injection in Capillary Electrophoresis. Anal Chem., 1997; 69:2952. lO.Tsuda, T., Zare, R. N. Spht Injector for Capillary Zone Electrophoresis. J. Chromatogr, 1991; 559:103. 11. Knox, J. H., McCormack, K. A. Volume Expansion and Loss of Sample Due to Initial Self-Heating in Capillary Electroseparation (CES) Systems. Chromatographia, 1994; 38:279. 12. Tehrani, J., Macomber, R., Day, L. Capillary Electrophoresis: An Integrated System with a Unique Spht-Flow Sample Introduction Mechanism. HRC & CC, 1991; 14:10. 13. Rose, D. J., Jorgenson, J. W. Characterization and Automation of Sample Introduction Methods for Capillary Zone Electrophoresis. Anal Chem., 1988; 60:642. 14.Ermakov, S. V, Zhukov, M. Y., Capelli, L., Righetti, P G. Quantitative Studies of Different Injection Systems in Capillary Electrophoresis. Electrophoresis, 1994; 15:1158. 15. Huang, X., Gordon, M. J., Zare, R. N. Bias in Quantitative Capillary Zone Electrophoresis Caused by Electrokinetic Sample Injection. Anal Chem., 1988; 60:375. 16. Demorest, D., Dubrow, R. Factors Influencing the Resolution and Quantitation of Oligonucleotides Separated by Capillary Electrophoresis on a Gel-Filled Capillary. J. Chromatogr, 1991; 559:43.
References
361
17. Altria, K. D., Kelly, M. A., Clark, B. J. The Use of a Short-End Injection Procedure to Achieve Improved Performance in Capillary Electrophoresis. Chromatographia, 1996; 43:153. IS.Euerby, M. R., Johnson, C. M., Cikalo, M., Bartle, K. D. "Short-End Injection" Rapid Analysis Capillary Electrochromatography. Chromatographia, 1998; 47:135. 19. Dose, E. V, Guiochon, G. Problems of Quantitative Injection in Capillary Zone Electrophoresis. Anal Chem., 1992; 64:123. 20.Grushka, E., McCormick, R. M. Zone Broadening Due to Sample Injection in Capillary Zone Electrophoresis. J. Chromatogr, 1989; 471:421. 21.Fishman, H. A., Amudi, N. M., Lee, T. T., Scheller, R. H., Zare, R. N. Spontaneous Injection in Microcolumn Separations. Anal. Chem., 1994; 66:2318. 22. Fishman, H. A., Scheller, R. H., Zare, R. N. Microcolumn Sample Injection by Spontaneous Fluid Displacement. J. Chromatogr., A, 1994; 680:99. 23. Colyer, C. L. Unusual Peaks and Baseline Shifts in Capillary Electrophoresis. J. Capillary Electrophor, 1996; 3:131. 24. Colyer, C. L., Oldham, K. B. Emersion Peaks in Capillary Electrophoresis. J. Chromatogr, A, 1995; 716:3. 25.Guttman, A., Schwartz, H. E. Artifacts Related to Sample Introduction in Capillary Gel Electrophoresis Affecting Separation Performance and Quantitation. Anal. Chem., 1995; 67:2279. 26.Ermakov, S. V., Zhukov, M. Y., Capelli, L., Righetti, P. G. Experimental and Theoretical Study of Artifactual Peak Splitting in Capillary Electrophoresis. Anal. Chem., 1994; 66:4034. 27.Monson, R. S., Collins, T. S., Waterhouse, A. L. Artifactual Signal Splitting in the Capillary Electrophoresis Analysis of Organic Acids in Wine. Anal. Lett, 1997; 30:1753. 28.Revilla, A. L., Havel, J., Jandik, R Peak Splitting Observed during Capillary Electrophoresis of a- and/3-Naphthols in Borate Buffer. J. Chromatogr, A, 1996; 745:225. 29. Weinberger, R. Separations Solutions. Peak Splitting. Amer Lab., 1997; 29:24. 30. Weinberger, R., Sapp, E., Moring, S. Capillary Electrophoresis of Urinary Porphyrins with Absorbance and Fluorescence Detection. J. Chromatogr, 1990; 516:271. 31.Mikkers, E E. P., Everaerts, E M., Verheggen, T. P. E. M. High Performance Zone Electrophoresis. J. Chromatogr, 1979; 169:11. 32.Burgi, D., Chien, R.-L. Optimization in Sample Stacking for High-Performance Capillary Electrophoresis. Anal. Chem., 1991; 63:2042. 33. Chien, R.-L., Burgi, D. S. Sample Stacking of an Extremely Large Injection Volume in High-Performance Capillary Electrophoresis. Anal. Chem., 1992; 64:1046. 34.Vinther, A., Soeberg, H., Nielsen, L., Pedersen, J., Biedermann, K. Thermal Degradation of a Thermolabile Serratia marcescens Nuclease Using Capillary Electrophoresis with Stacking Conditions. Anal. Chem., 1992; 64:187. 35. Ha, W.-Y., Shaw, P-C, Wang, J. Improved Electrophoretic Resolution of DNA Fragments in Samples Containing High Concentrations of Salts. BioTechniques, 1999; 26:425. 36.Hjerten, S., Valtcheva, L., Li, Y.-M. A Simple Method for Desalting and Concentration of Microliter Volumes of Protein Solutions with Special Reference to Capillary Electrophoresis. J. Capillary Electrophor, 1994; 1:83. 37. Zhao, Y., McLaughlin, K., Lunte, C. E. On-Column Sample Preconcentration Using Sample Matrix Switching and Field Amphfication for Increased Sensitivity of Capillary Electrophoretic Analysis of Physiological Samples. Anal. Chem., 1998; 70:4578. 38. Zhang, R., Hjerten, S. A Simple Micromethod for Concentration and Desalting Utilizing a Hollow Fiber, with Special Reference to Capillary Electrophoresis. Anal. Chem., 1997; 69:1585. 39. Clarke, N. J., Tomlinson, A. J., Schomburg, G., Naylor, S. Capillary Isoelectric Focusing of Physiologically Derived Proteins with Online Desalting of Isotonic Salt Concentrations. Anal. Chem., 1997; 69:2786. 40. Clarke, N. J., Tomlinson, A. J., Naylor, S. Online Desalting of Physiologically Derived Fluids in Conjunction with Capillary Isoelectric Focusing-Mass Spectrometry. J. Am. Soc. Mass Spectrom., 1997; 8:743.
362
Chapter 8
Injection
41.Aebersold, R., Morrison, H. Analysis of Dilute Peptide Samples by Capillary Zone Electrophoresis. J. Chromatogr., 1990; 516:79. 42.Xiong, Y., Park, S.-R., Swerdlow, H. Base Stacking: pH-Mediated On-Column Sample Concentration for Capillary DNA Sequencing. Anal. Chem., 1998; 70:3605. 43.Shihabi, Z. K. Sample Stacking by Acetonitrile-Salt Mixtures. J. Capillary Electrophor., 1995; 2:267. 44. Shihabi, Z. K. Peptide Stacking by Acetonitrile-Salt Mixtures for Capillary Zone Electrophoresis. J. Chromatogr., A, 1996; 744:231. 45.Shihabi, Z. K., Friedberg, M. Insuhn Stacking for Capillary Electrophoresis. J. Chromatogr, A, 1998; 807:129. 46. Shihabi, Z. K. Serum Procainamide Analysis Based on Acetonitrile Stacking by Capillary Electrophoresis. Electrophoresis, 1998; 19:3008. 47.Quirino, J. R On-Line Concentration of Neutral Analytes for Micellar Electrokinetic Chromatography IV. Field-Enhanced Sample Injection. J. Chromatogr, A, 1998; 778:251. 48. Quirino, J. P, Terabe, S. Stacking of Neutral Solutes in Micellar Electrokinetic Chromatography. J. Capillary Electrophor, 1997; 4:233. 49. Quirino, J. P, Terabe, S. On-Line Concentration of Neutral Analytes for Micellar Electrokinetic Chromatography 3. Stacking with Reverse Migrating Micelles. Anal. Chem., 1998; 70:149. 50. Quirino, J. P, Terabe, S. On-Line Concentration of Neutral Analytes for Micellar Electrokinetic Chromatography I. Normal Stacking Mode. J. Chromatogr, A, 1997; 781:119. 51. Nielsen, K. R., Foley, J. P Zone Sharpening of Neutral Solutes in Micellar Electrokinetic Chromatography with Electrokinetic Injection. J. Chromatogr, A, 1994; 686:283. 52.Liu, Z., Sam, P, Sirimanne, S. R., McLure, P C , Grainger, J., Patterson, D. G., Jr. Field-Amplified Sample Stacking in Micellar Electrokinetic Chromatography for On-Column Sample Concentration of Neutral Molecules. J. Chromatogr, A, 1994; 673:125. 53. Quirino, J. P On-Line Concentration of Neutral Analytes for Micellar Electrokinetic Chromatography. 5. Field-Enhanced Sample Injection with Reverse Migrating Micelles. Anal. Chem., 1998; 70:1893. 54. Palmer, J., Munro, N. J., Landers, J. P High-Salt Sample Matrix-Induced Stacking of Neutral Analytes in MEKC. Anal. Chem., 1999; 71:1679. 55. Chien, R.-L., Burgi, D. S. Field Amphfied Sample Injection in High-Performance Capillary Electrophoresis. J. Chromatogr, 1991; 559:141. 56. Chien, R.-L., Burgi, D. S. Field-Amplified Polarity-Switching Sample Injection in High-Performance Capillary Electrophoresis. J. Chromatogr, 1991; 559:153. 57. Chien, R.-L. Mathematical Modeling of Field-Amplified Sample Injection in High-Performance Capillary Electrophoresis. Anal. Chem., 1991; 63:2866. 58. Chien, R.-L., Burgi, D. S. On-Column Sample Concentration Using Field Amphfication in CZE. Anal. Chem., 1992; 64:489. 59. Zhang, C.-X., Thormann, W. Head-Column Field-Amphfied Sample Stacking in Binary System Capillary Electrophoresis. 2. Optimization with a Pre-injection Plug and Application to Micellar Electrokinetic Chromatography Anal. Chem., 1998; 70:540. 60. Zhang, C.-X., Thormann, W. Head-Column Field-Amphfied Sample Stacking in Binary System Capillary Electrophoresis: A Robust Approach Providing Over 1000-Fold Sensitivity Enhancement. Anal. Chem., 1996; 68:2523. 61. Zhang, C.-X., Aebi, Y., Thormann, W. Microassay of Amiodarone and Desethylamiodarone in Serum by Capillary Electrophoresis with Head-Column Field-Amplified Sample Stacking. Clin. Chem., 1996; 42:1805. 62. Kaniansky D., Ivanyi, F, Onsuska, F I. On-Line Isotachophoretic Sample Pretreatment in Ultratrace Determination of Paraquat and Diquat in Water by Capillary Zone Electrophoresis. Anal. Chem., 1994; 66:1817. 63.Stegehuis, D. S., Irth, H., Tjaden, U. R., van der Greef, J. Isotachophoresis as an On-Line Concentration Pretreatment Technique in Capillary Electrophoresis. J. Chromatogr, 1991; 538:393.
References
363
64. Gebauer, E, Thormann, W, Bocek, P. Sample Self-Stacking in Zone Electrophoresis. Theoretical Description of the Zone Electrophoretic Separation of Minor Components in the Presence of Bulk Amounts of a Sample Component with High Mobility and Like Charge. J. Chromatogr., 1992; 608:47. 65. Gebauer, P., Thormann, W, Bocek, P. Sample Self-Stacking and Sample Stacking in Zone Electrophoresis with Major Sample Components of Like Charge: General Model and Scheme of Possible Modes. Electrophoresis, 1995; 16:2039. 66. Foret, E, Sustacek, V, Bocek, P On-Line Isotachophoretic Sample Preconcentration for Enhancement of Zone Detectability in Capillary Zone Electrophoresis. J. Microcolumn Sep., 1990; 2:229. 67. Foret, E, Szoko, E., Karger, B. L. On-Column Transient and Coupled Isotachophoretic Preconcentration of Protein Samples in Capillary Electrophoresis. J. Chromatogr, 1992; 608:3. 68. Foret, E, Szoko, E., Karger, B. L. Trace Analysis of Proteins by Capillary Zone Electrophoresis with On-Column Transient Isotachophoretic Preconcentration. Electrophoresis, 1993; 14:417. 69. van der Schans, M. J., Beckers, J. L., Moiling, M. C , Everaerts, E M. Intrinsic Isotachophoretic Preconcentration in Capillary Gel Electrophoresis of DNA Restriction Fragments. J. Chromatogr, A, 1995; 717:139. 70. Witte, D. T, Nagard, S., Larsson, M. Improved Sensitivity by Online Isotachophoretic Preconcentration in the Capillary Zone Electrophoretic Determination of Peptide-like Solutes. J. Chromatogr, A, 1994; 687:155. 71.Boden, J., Baechmann, K., Kotz, L., Fabry, L., Pahlke, S. Application of Capillary Zone Electrophoresis with an Isotachophoretic Initial State to Determine Anionic Impurities on as-Polished Silicon Wafer Surfaces. J. Chromatogr, A, 1995; 696:321. 72.Bergmann, J., Jaehde, U., Mazereeuw, M., Tjaden, U. R., Schunack, W. Potential of Online Isotachophoresis-Capillary Zone Electrophoresis with Hydrodynamic Counterflow in the Analysis of Various Proteins and Recombinant Human Interleukin-3. J. Chromatogr, A, 1996; 734:381. 73. Church, M. N., Spear, J. D., Russo, R. E., Klunder, G. L., Grant, P. M., Andresen, B. D. Transient Isotachophoretic-Electrophoretic Separations of Lanthanides with Indirect Laser-Induced Fluorescence Detection. Anal. Chem., 1998; 70:2475. 74.Enlund, A. M., Westerlund, D. Enhancing Detectability in CE Combining an Isotachophoretic Preconcentration with Capillary Zone Electrophoresis in a Single Capillary. Chromatographia, 1997; 46:315. 75.Krivankova, L., Vrana, A., Gebauer, E, Bocek, P Online Isotachophoresis-Capillary Zone Electrophoresis versus Sample Stacking Capillary Zone Electrophoresis. Analysis of Hippurate in Serum. J. Chromatogr, A, 1997; 772:283. 76.Janini, G. M., Muschik, G. M., Issaq, H. J. Sample Matrix Effects in Capillary Zone Electrophoresis. Effect of Chloride Ion on Nitrate and Nitrite. J. Capillary Electrophor, 1994; 1:116. 77.Swartz, M. E., Merion, M. On-Line Sample Preconcentration on a Packed-Inlet Capillary for Improving the Sensitivity of Capillary Electrophoretic Analysis of Pharmaceuticals. J. Chromatogr, 1993; 632:209. 78.Benson, L. M., Tomlinson, A. J., Mayeno, A. N., Gleich, G. J., Wells, D., Naylor, S. Membrane Preconcentration-Capillary Electrophoresis-Mass Spectrometry (mPC-CE-MS) Analysis of 3Phenylamino-l,2-propanediol (PAP) Metabolites. J. HighResolut. Chromatogr, 1996; 19:291. 79. Naylor, S., Tomlinson, A.J. Membrane Preconcentration-Capillary Electrophoresis-Mass Spectrometry in the Analysis of Biologically Derived Metabolites and Biopolymers. Biomed. Chromatogr, 1996; 10:325. 80. Naylor, S., Tomlinson, A.J. Membrane Preconcentration-Capillary Electrophoresis Tandem Mass Spectrometry (mPC-CE-MS/MS) in the Sequence Analysis of Biologically Derived Peptides. Talanta, 1998; 45:603. 81.Rohde, E., Tomlinson, A. J., Johnson, D. H., Naylor, S. Protein Analysis by Membrane Preconcentration-Capillary Electrophoresis: Systematic Evaluation of Parameters Affecting Preconcentration and Separation. J. Chromatogr, B: Biomed. Appl, 1998; 713:301. 82.Tomlinson, A. J., Benson, L. M., Braddock, W D., Oda, R. P Improved Online Membrane Preconcentration-Capillary Electrophoresis (mPC-CE).J. HighResolut. Chromatogr, 1995; 18:381.
364
Chapter 8
Injection
83.Tomlinson, A. J., Naylor, S. Systematic Development of On-Line Membrane Preconcentration-Capillary Electrophoresis-Mass Spectrometry for the Analysis of Peptide Mixtures. J. Capillary Electrophor., 1995; 2:225. 84. Tomlinson, A. J., Benson, L. M.Jameson, S., Naylor, S. Rapid Loading of Large Sample Volumes, Analyte Cleanup, and Modified Moving Boundary Transient Isotachophoresis Conditions for Membrane Preconcentration-Capillary Electrophoresis in Small Diameter Capillaries. Electrophoresis, 1996; 17:1801.
CHAPTER
9
Detection 9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9 9.10
On-Capillary Detection The Detection Problem Limits of Detection Detection Techniques Band Broadening Absorption Detection Fluorescence Detection Derivatization Mass Spectrometry Micropreparative Fraction Collection References
9.1 ON-CAPILLARY DETECTION The most common mode of detection in HPCE is on-capillary detection. Considering the minuscule dimensions of the capillary, this mode of detection has two advantages: 1. Since the capillary is contiguous between the electrodes, current leakage is eliminated. 2. Dilution of the eluting solutes either from dead volume in a flowcell or from the postcapillary reagent or sheath flow is eliminated as well. The characteristics of on-capillary detection differ dramatically from those of postcolumn detection in HPLC. In chromatography, solutes move through the chromatographic packing at velocities determined by the mobile-phase flow rate and the overall retention characteristics of each analyte. On the column, the peak velocities depend on each solute's retention characteristics. Once off the column, all solutes are swept past the detector by the mobile phase at identical flow rates. The detected peak widths are a function of chromatographic processes and are not related to the peak velocity past the flowcell. 365
366
Chapter 9
Detection
In HPCE, a different set of rules applies (1). The migration velocity of each solute through the capillary is a function of its electrophoretic mobility in conjunction with the EOE Since detection occurs on-capillary, these forces are operative as the solute is traversing the detection window. As a result, slower moving components spend more time migrating past the detector window than their more rapidly moving counterparts. Figure 9.1, top trace, illustrates the separation of a three-component mixture recorded directly from the detector output (1). The bottom trace gives the same separation, corrected for the zonal velocity. This correction is calculated by (9.1) where Wg is the spatial width of the sample in units of length, L^ is the effective capillary length, t^ is the migration time, w^ is the recorded temporal width in time units, and w^ is the spatial width of the detector window. Thus, two electropherograms can be defined. The spatial electrogram refers to the actual band-
-ii DETECTOR
INJECTOR
SPATIAL i ELECTROPHEROGRAM |
TEMPORAL ELECTROPHEROGRAM
TIME OR LENGTH FIGURE 9.1 Plots of the detector response (bottom trace) as a function of time and (top trace) as a function zone length within the capillary. Redrawn with permission from J. Chromatogr., 480, 95 (1989), copyright © 1989 Elsevier Science Publishers.
9.2 The Detection Problem
367
width on the capillary, whereas the temporal electrogram is defined by what the detector observes. This phenomenon has practical implications, since slower moving zones produce an increase in the peak area counts. When quantifying solutes by a response factor, a correction factor must be applied to normalize the peak area irrespective of the migration velocity: Aeo. = ^ ^ ,
(9.2)
Here, A^aw is the measured peak area, t^ is the migration time, and A^orr is the corrected peak area. Response factors are frequently employed when standards are unavailable. Applications involving oligonucleotides, impurity analyses of drugs, and chiral separations are typical examples. For these applications, it is assumed that all solutes have the same molar absorptivities. Whenever quantitation without a matching standard is used, it is necessary to normalize the peak areas. While the peak height is not related to the solute velocity as it passes through the optical window, quantitation is less accurate than peak area because of stacking/antistacking and electrodispersive effects. When standards are used, it is unnecessary to provide this correction, since it is assumed that the standard behaves identically to the solute, if the migration times are constant. If the migrations times vary slightly from run to run, area normalization does not improve precision (2). If the migration time precision is poor, it is best to correct that problem rather than normalizing the peak areas.
9.2 THE DETECTION PROBLEM Because of the minute amounts of material injected into the capillary, extremely high sensitivity detection is generally required for all forms of HPCE. The problem is exacerbated by the desire to dilute samples to rid separations of troublesome matrix effects. The instrumental problems for optical detection are twofold: (i) the short optical path length as defined by the capillary i.d.; and (ii) the poor optical surface of the cylindrical capillary. While square and rectangular capillaries have been around for some time, there have been no definitive studies showing any advantages in detection. An exception to this is viewing down the long end of a rectangular capillary (3). This approach resulted in substantial band broadening. In addition, proper alignment of square and rectangular capillaries in the optical window is critical. Commercial instruments use absorption detectors that are modifications of standard HPLC detectors. The absorption detectors are modestly sensitive, giving limits of detection of 10"^ M for solutes with very high molar absorptivities. While laser fluorescence detection can improve sensitivity down to 10"^^ M with commercially available equipment and approach single-molecule detection in
368
Chapter 9
Detection
more sophisticated apparatus, derivatization is usually required to tag a solute with an optimized fluorophore. Innovation is still required to solve the general detection problem in HPCE. Improvements of two to three orders of magnitude in absorption detection would solve many problems relating to matrix effects and Unear dynamic range. With the high resolving power of HPCE, less-selective detectors might prove useful, providing the sensitivity requirements are fulfilled.
9.3 LIMITS OF DETECTION There are two means of describing the limits of detection of a system: the concentration limit of detection (CLOD) and the mass limit of detection (MLOD). The CLOD relates to the concentration of the individual sample, whereas the MLOD describes what the instrument can measure. For example, the CLOD of many peptides is around 1 |ag/mL using absorption detection at 200 nm without stacking. If 10 nL of that material is injected and detected at three times the baseline noise, the MLOD is 10 pg. Another way of considering MLOD is based on the volume of the detector window If a l-|Xg/mL peptide solution is continuously aspirated through a capillary, then for a 1-nL detector window, the amount of material in the window at any given time is 1 pg. Thus, the MLOD can be manipulated by selecting the size of the detection window. It is frequently possible to improve the MLOD at the expense of the CLOD by compressing the detection window. For most analytical problems, the CLOD is the more important parameter, since it relates to the minimum detectable quantity of a solute in the sample of interest. In extreme cases where the amount of available sample is minuscule, the MLOD becomes the more important parameter describing the LOD. It is easily concluded the HPCE has excellent MLODs but poor CLODs, especially when compared with optical detection in HPLC. This is compensated for in part by the use of extended path length capillaries, online stacking, and offline trace enrichment.
9.4 DETECTION TECHNIQUES A tabulation of detectors that have been used in HPCE is given in Table 9.1. Most of these detection schemes are not available on commercial systems or are not practical for everyday use. Commercial systems utilize absorption, fluorescence, laser-induced fluorescence, or conductivity detection. Interfacing to various forms of mass spectrometry is quite mature, though problems with sensitivity remain. Conductivity detection and indirect detection have been covered in Section 3.6.
369
9.4 Detection Techniques Table 9.1
Detectors for HPCE
Technique
References
Absorbance Absorbance, diode array Absorbance, extended pathlength Absorbance, indirect Absorbance, photothermal Capillary vibration Chemiluminescence Circular dichroism Conductivity Conductivity, indirect Conductivity, suppressed Concentration gradient Electrochemical, ampeometric Electrochemical, indirect Fluorescence Fluorescence, indirect Fluorescence, laser-induced Fluorescence, microscopy Fluorescence, multiwavelength Inductively coupled plasma Inductively coupled plasma mass spectrometry Ion mobility spectrometry Mass spectrometry, electrospray Mass spectrometry, fast atom bombardment Mass spectrometry, ion trap Mass spectrometry, magnetic sector Mass spectrometry, tandem Mass spectrometry, time-of-flight
4-6 7-10 11-17 18-26 27-30 31-33 34-37 38 39-43 44 43,46 47-49 50-54 55 56-60 61-64 65-74 75-80 81-83 84-86 87-90 91 213,218-227 92-94 95-99 93,107, 108 100-106 97-98, 106 109-115 116-121 122 123 124-127 128-132 133,134 137-139 57, 135, 136 140-143
NMR Oscillometric detection Phosphorescence, sensitized Potentiometric (ion-selective) detection Radioactivity Raman Reaction detector, affinity Reaction detector, postcapillary Refractive index
More than 150 papers have appeared reporting on electrochemical detection. This mode of detection is targeted primarily for catecholamine detection in single nerve cells, though many other applications are possible. It is unlikely that this mode will become commercially available in the near future, due to the difficulty of fabricating microelectrodes. That is unfortunate, since the LOD can
370
Chapter 9
Detection
approach that of fluorescence for electroactive solutes. These detectors may become more practical with microfabricated devices (144-146). Detectors fall into one of two broad categories: 1. Bulk property detectors 2. Solute property detectors The bulk property detector measures a general property of matter. Refractive index and conductivity detection are the most important examples of the class. These detectors are not selective; they are universal. They are often less sensitive than solute property detectors. The sensitivity is enhanced if a solute's measured property is maximally differentiated from that of the BGE. This adds an additional constraint to methods development. Solute property detectors measure a physical property that is specific to the solute as compared with the BGE. These are represented by absorption, fluorescence, electrochemical, and radioactivity detectors. Fluorescence detection is far more sensitive and selective than absorption detection. All molecules that fluoresce must first absorb light. The converse is not true; most molecules do not fluoresce. The resultant selectivity is a doubleedged sword. Since the technique is selective, derivatization is frequently required to take advantage of the detector's inherent high sensitivity. Sensitivity of detection is high, since a low signal level above a very dark background is more easily measured than is a small difference between two high-intensity signals, as in absorption detection. With the use of the laser as the excitation source, detection problems are virtually eliminated because of the extreme sensitivity. However, derivatization is often required, since most solutes do not have native fluorescence. Information-rich detection is particularly important in HPCE. Since fraction collection is difficult, it is important to obtain additional information about a solute online. The mass spectrometer and multiwavelength absorption detectors are the most important examples of the group of detectors. Of the multiwavelength detectors, the diode array detector in particular is very useful during methods development, since it can aid in peak tracking. Another means of categorizing detectors is the nature of the response toward the solute. Most detectors used for HPCE are concentration-sensitive. They respond in proportion to the concentration of the solute as it traverses the detection window. The mass spectrometer is the most notable exception of the group. Sensitivity is based on the number of formed ions, so that this instrument responds to the mass of material that enters the source.
9.5 BAND BROADENING Two detector-related features can contribute to band broadening: the width of the detector window and the detector time constant. For all practical purposes,
9.5 Band Broadening
371
the detector-related contribution to peak variance is the same as the injection contribution (147, 148), and so
a. o DC
3
a
it -T
0
2
6 TIME
8
10
12
14
(ifiiti)
FIGURE 9.15 Separation of FMOC-derivatized amino acids. Capillary: 62 cm (40 cm to detector) X 50 |im i.d.; buffer: 20 mM borate, pH 9.5, 25 mM SDS; field strength: 416 V/cm; temperature: 30°C; detection: xenon arc fluorescence; excitation: 260 nm; emission: 305-nm-long wavepass filter. The CLOD is 10 ng/mL. Reprinted with permission from And. Chem., 263, 417 (1991), copyright © 1991 Am. Chem. Soc.
One can consider immunoassay a form of detection or even a biological derivatization (150, 185-190). There are basically two forms: competitive and direct (or noncompetitive) immunoassay. When a fluorescent-labeled antibody is available, the preferable noncompetitive immunoassay can be performed. Antibody and antigen are incubated together for a predetermined period of time, after which the free and bound forms are separated. Quantitation is either by the appearance of the immune complex or the disappearance of the labeled antibody In the competitive immunoassay, tagged antigen is mixed with untagged antibody and sample containing untagged antigen. The higher the concentration of
390
Chapter 9
Detection
untagged antigen, the lower the concentration of the tagged antigen-antibody complex. The technique yields nonlinear calibration curves and should only be used when tagged antibody is not available. LIF is particularly useful for detection, since the high sensitivity reduces the requirement for expensive labeled antibodies or antigens. The high specificity of fluorescence also reduces the potential for interference. If an antigen has native fluorescence, then a derivatized antibody is not required. DNA intercalators have revolutionized the fields of genetic analysis and human identification. While not a derivatization chemistry, these reagents are added to the BGE. The noncovalent interaction between the dye and DNA modifies mobility, and so this is in effect a form of secondary equilibrium. It can also be thought of as a noncovalent on-capillary derivatization technique. More important is the impact of the dye on detection. When the dye is not bound to DNA, its fluorescence is quenched via interaction (collision) with water. Once bound to DNA, fluorescence quenching is reduced, and the fluorescence quantum yields increase dramatically The classical DNA intercalator, ethidium bromide, is seldom used today, since its wavelength of absorption does not match the emission wavelength of the argon-ion laser. Dyes such as TOTO, YOYO, TO-PRO, and YO-PRO are usually selected because of their favorable wavelength of absorption. These applications are covered in more detail in Chapter 6. True on-capillary derivatization is possible as well. In this case, the front end of the capillary is used for the chemical reaction. As in the case of postcapillary reactions, the reaction kinetics need to be relatively rapid to minimize band broadening. This technique is ideal when sample volumes are extremely limited, as in the case of analysis of chemicals from a single cell (191). On-capillary chemistry can also be employed for enzyme assays (192-194). Known as enzyme-mediated microanalysis (EMMA), the technique employs reaction between an enzyme and a substrate. Given an excess of substrate, it is possible to amplify a reaction product to provide very high sensitivity. In another variant of on-column chemistry, PCR and subsequent size separations of amplified DNA have been integrated using microfabricated devices (195, 196). Postcapillary derivatization also works well with HPCE (57, 135, 155, 158, 197-199). This technique is less common than precapillary derivatization because of the lack of commercial equipment. The advantages of this approach are minimal sample handling and the ability to work with derivatives that have limited stability The principal requirements for postcapillary derivatization are as follows: 1. The derivatizing reagent is invisible to the detector. 2. Rapid reactions occur. 3. The reactor provides minimal band broadening. One such design is shown in Figure 9.16 (57). The basis for the function of this reactor is differential EOF The gap junction of this reactor is about 50 jam. With a separating capillary of 50 jim and a reactor capillary of 75 |lm, the
391
9.8 Derivatization
reagent is drawn into the reactor by differential EOF, since the volumetric requirements of the larger diameter capillary are not being fulfilled. Mixing is accomplished by convection. OPA (o-phthalaldehyde) is an ideal postcapillary reagent. The reagent does not fluoresce, it is stable, and it reacts quickly with primary amines. The optimal conditions are 3.7 mM OPA in run buffer, 0.5% mercaptoethanol, 2% methanol, and 40°C (57). The CLOD for OPA glycine is 60 ng/mL with xenon arc fluorescence, X = 350 nm, M > 400 nm. The run-to-run reproducibility was about 1% using peak areas. The sensitivity of peptide mapping can be greatly enhanced relative to absorption detection with the PCRS system, as shown in Figure 9.17 (57). The concentration of the analytes in the absorbance electropherogram was 40 times greater than in the fluorescence run. Among the other notable postcolumn reaction detectors is chemiluminescence detection (34, 36, 37, 200, 201). Chemistries such as peroxyoxylate, acridinium, luminol, and firefly luciferase have all been reported. Elimination of
Fluorescence Cell
Reactor Ceil to w a ^ reservoir
Exploded View of Buffer Junction
/ 50 \m gap i
l/t6x.007teftontut)«
FIGURE 9.16 Schematic of a liquid junction postcapillary reaction system. Reprinted with permission from Anal. Chem., 63, 417 (1991), copyright © 1991 Am. Chem. Soc.
392
Chapter 9
Detection
s^ "T"
0
"T" 8
2
TIME
10
12
14
(min)
FIGURE 9.17 CZE of a tryptic digest of/3-lactoglobulin with UV detection at 200 nm (top) and postcapillary derivatization with OPA (bottom). Capillary: 62 cm (40 cm to detector) x 50 |Lim i.d.; buffer: 20 mM borate, pH 9.5; field strength: 278 V/cm; postcapillary detection: xenon arc fluorescence; excitation: 390 nm; emission: 450-nm-long wavepass filter; sample concentration: absorption, 20 nmol/mL, fluorescence, 0,5 nmol/mL; injection: absorption, vacuum, 1 s, fluorescence, electrokinetic, 7 s at 5 kV Reprinted with permission from Anal. Chem., 63, 417 (1991), copyright © 1991 Am. Chem. Soc.
the excitation light source greatly diminishes the background light. As a result, the photomultiplier tube can be run at very high voltages, providing impressive sensitivity. Laserlike LODs on inexpensive instrumentation are possible; however, the simplicity of LIF and the lack of commercial postcapillary apparatus greatly limits application of this technique.
9.9 Mass Spectrometry
393
9.9 MASS SPECTROMETRY A. INTRODUCTION Coupling of HPCE to mass spectrometry (MS) is developing rapidly since first reported by Olivares et al. in 1987 (202). Looking back, the actual interfacing turned out not to be difficult, at least for electrospray ionization (ESI). Among the challenges were: 1. Providing an electrical contact in the absence of an outlet reservoir 2. Generation of sufficient fluid flow to maintain a stable spray 3. Finding compatible buffers and additives that are volatile and do not raise the ion currents 4. Injecting a sufficient quantity of material to ensure detectability 5. Compatibility of the speed of the separation with the scan speed of the mass spectrometer There are two compatible ionization techniques: electrospray and fast-atom bombardment (FAB). Virtually all work has been reported using electrospray techniques. Both techniques generally require makeup flows to elevate the total flow rate to 1-10 |iL/min, although nanoelectrospray operates without a makeup flow. A syringe pump is typically used for reagent delivery to avoid the pulsations characteristic of reciprocating pumps, particularly at low backpressures and low flow rates. Since detection is performed postcapillary and most of flow is provided from the makeup solution, the problems with peak area normalization described in Section 9.1 are eliminated. The HPCE instrumentation employed in mass spectrometry has several considerations. 1. The injection and capillary-filling mechanism must be pressure- rather than vacuum-driven. It is hard to imagine a simple means of connecting vacuum-driven equipment to any of the interfaces. 2. A design to safely route the capillary outside of the system is required. One such modification is shown for the Hewlett-Packard instrument in Figure 9.18. 3. To prevent siphoning, a lab-jack may be used to level the capillary inlet with the mass spectrometer. 4. The power supply should be capable of providing polarity switching and, in some cases, a negative electrode held at a potential other than ground. When the HP instrument is interfaced to an HP mass spectrometer (5989B MS Engine or 1100 LC/MSD), the negative electrode is grounded along with the electrospray. They must be linked with a grounding cable to be sure the are at the same ground. Important: There are several different power supply configurations that can be used to couple HPCE instruments with mass spectrometers. It is critical that
394
Chapter 9
Detection
Cassette
/
P^ ==1= ^ 6^
8x10^^
^Assumed that the average number of charges increases linearly v^ith Mr and the distribution is centered on mJz 1000. ''Peak width due to microheterogeneity typical of large biopolymers and contributions of impurities, solvent adducts, etc. '^ESl production before sampling losses assuming an 80% ionization efficiency. Detected ion intensities are 4-5 orders of magnitude lov^er due to inefficiencies arising from sampling, transmission, and detection. '^Peak width of 6 m/z units is too large for individual charge states to be resolved; a peak width of
"'•'""'"•'•
^
I
f
2
FIGURE 10.2 Impact of solute concentration on analytical figures of merit. Refer to Figure 10.1 for experimental conditions. Key: Log A = log peak area; Log HGT = log peak height; (MX 2) X 10~^ = scaled migration time (min); PW/3 = scaled peak width (s). Reprinted with permission from J. Liq. Chromatogr., 14, 953 (1991), copyright © 1991 Marcel Dekker.
all calculations. Results from an interlaboratory cross-validation study are given in Table 10.3c. The data indicate that linearity is reproducible in multiple laboratories. Increasing the linear range is possible with high-ionic-strength buffers (1). Figure 10.3 shows separations of some anti-inflammatory drugs at concentrations of 1 mg/mL and 250 |ig/mL using a low-ionic-strength buffer. Substantial losses in resolution are found at the higher solute concentrations. A similar separation in a high-ionic-strength buffer is shown in Figure 10.4. Almost no change in resolution is found between the run with high concentration and that with low concentration. Note that a shortened capillary was employed, since the EOF was substantially lowered by the high-ionic-strength buffer. The separations in these two figures were performed in 25-|im-i.d. capillaries to reduce the heating effects. The overall addressable concentration range of HPCE is illustrated in Figure 10.5 on p. 436. The use of the laser detector solves most of the compelling problems in HPCE. With this highly sensitive detector, it is possible to perform extreme dilutions of most samples. At high dilution, most ionic-strength-mediated effects from the solute and/or the sample matrix become insignificant. Unfortunately, derivatization is required for most laser-fluorescence-based applications.
434
Chapter 10
Putting It All Together
1.6
FIGURE 10.3 Band profile dependence on solute concentration and buffer ionic strength for low-ionic-strength buffer. Capillary: 65 cm (43 cm to detector) x 25 ^im i.d.; buffer: 20 mM SDS, 20 mM phosphate, pH 9.2; temperature: 30°C; injection: vacuum, 2 s; detection: UV, 230 nm. Key: (A) (1) suhndac, 1 mg/mL; (2) indomethacin, 1 mg/mL; (3) tolmetin, 1 mg/mL; (4) ibuprofen, 1 mg/mL; (5) naproxen, 100 |Llg/mL; (6) diflunisal, 500 |lg/mL. (B) 4x dilution of A. Reprinted with permission from J. Liq. Chromatogr., 14, 953 (1991), copyright © 1991 Marcel Dekker.
10.5 SAMPLE PREPARATION A. BASIC PRINCIPLES In HPCE, there are several important issues regarding sample preparation. Obviously, interfering components, if not separable, must be removed during the sample preparation process. This is true for all analytical techniques. In both chromatography and electrophoresis, the sample matrix can affect the resolution
435
10.5 Sample Preparation
L
FIGURE 10.4 Band profile dependence on solute concentration and buffer ionic strength for high-ionic-strength buffer. Conditions as per Figure 10.3 except: capillary: 20 cm to detector; buffer: 100 mM phosphate, 25 mM SDS, pH 7.0. Reprinted in part with permission from J. Liq. Chromatogr., 14, 953 (1991), copyright © 1991 Marcel Dekker.
436
Chapter 10
STATE-OF THE-ART
LASER FLUORESCENCE DETECTION
10""'
10"^^
10"''°
STACKING RANGE
10"^
Putting It All Together
NORMAL ANALYTIC RANGE
10"^
MICRO PREPARATIVE RANGE
10"^
10'^
CONCENTRATION (M) FIGURE 10.5
The dynamic ranges of capillary electrophoretic techniques.
of the separation. Unlike chromatographic techniques, in HPCE the sample matrix may have a profound impact on the amount of material that is injected into the capillary when electrokinetic injection is employed. The sample preparation process must deal with this problem. For samples containing high concentrations of solutes—for example, pharmaceutical dosage forms—simple dilution of the dosage form extract in the supporting electrolyte is sufficient, since generally only small injection volumes are required. Depending on the strength of the chromophore, final concentrations of 10"^-10"^ M provide adequate sensitivity. For complex samples or when high sensitivity is required, sample preparation to remove interferences and place the solute(s) in a CE-friendly solution is clearly indicated. Stacking techniques can be useful for improving sensitivity, but matrix effects and artifacts often interfere. There are hundreds of examples in the literature describing these techniques, some of which will be cited in the following discussion. Centrifugation or filtration to remove particulate matter is always good practice. Sonication to remove air is sometimes needed as well. In reversed-phase LC, it is generally bad practice to prepare the sample in a solvent with greater eluting power than the mobile phase, particularly if large injections are required. The fundamental requirement for HPCE is that the sample should never be prepared so as to have an ionic strength greater than that of the supporting electrolyte. This requirement can be loosened only when the injection volume can be kept small. It is usually good practice to desalt highionic-strength samples. For small molecules, the most useful forms of sample preparation include: 1. 2. 3. 4. 5. 6.
Liquid-liquid extraction (2-9) Solid-phase extraction (9-18) Supercritical fluid extraction (19) Protein precipitation (for blood serum or plasma) (20-26) Dialysisi (27-29) Ultrafiltration (30)
Un yivo microdialysis is also employed for sampling of neurochemicals. This subject is beyond the scope of this text.
10.5 Sample Preparation
437
For large molecules such as proteins and DNA, the following techniques are also applicable: 1. 2. 3. 4. 5. 6. 7. 8.
HPLC (31) Affinity LC (32) Ultrafiltration (33, 34) Solid-phase extraction (35-41) Dialysis (42, 43) Desalting (44-47) Sedimentation (48) Precipitation (49-54)
In the following sections, some of these modes of sample preparation are covered in greater detail.
B. DRUGS IN BIOLOGICAL FLUIDS 1.
Direct Injection
MECC is often preferred for separating small synthetic pharmaceuticals. Since surfactant solutions are utilized, direct injection of blood plasma or serum might be feasible, since a surfactant such as SDS binds strongly to and solubilizes serum proteins. Indeed, micellar liquid chromatography has been shown useful for direct injection. In this mode of HPLC, the surfactant solution serves as the mobile-phase modifier. Surfactant-bound serum proteins form an extremely large aggregate, which is excluded from the stationary phase. The protein bolus elutes on or about t^ in a relatively narrow band. The retained drug substance elutes some time later, producing clean chromatograms at the low microgramsper-milliliter level. In MECC, the protein-surfactant aggregate has a substantial net negative charge when SDS is used as the additive. The aggregate then elutes relatively late in the separation, leaving only a small window for interference-free monitoring of the drug substance. The advantage of MECC with direct injection over CZE with acetonitrile protein precipitation has been studied (55). The technique gave better interday precision (1.49% vs. 16.1%) than did CZE. Direct injection will become successful only under one of certain circumstances: 1. Selective detection is possible. 2. The drug substance is present in serum or plasma at high concentrations. Selective detection includes fluorescence and UV detection at wavelengths above 240 nm. When low UV detection is used, a CLOD of 5 jlg/mL is found (56). With liquid-liquid extraction, the CLOD drops to 1 |Llg/mL, and with solidphase extraction, a CLOD of 100 ng/mL is obtained. The baselines are always cleaner when some form of sample preparation is employed.
438
Chapter 10
Putting It All Together
Since low-UV detection is often required in HPCE, a typical separation is shown in Figure 10.6 for aspoxicillin at a concentration of 50 |Llg/mL (57). The problems of unambiguously assigning peak identity are immediately obvious. Nevertheless, direct injection is particularly useful when the sample size is extremely limited—for example, clinical determination of xanthines from premature infants (58).
y iw^
1 — 0
—r5
1 10
1 15
—r 20
TIME (min.) FIGURE 10.6 Direct plasma injection for the determination of aspoxicillin. Capillary: 65 cm (50 cm to detector) x 50 p,m i.d.; buffer: 50 mM SDS, 20 mM phosphate-borate, pH 8.5; voltage: 20 kV; detection: Uy 210 nm; temperature: ambient; solute concentration: 50 |J,g/mL. Reprinted with permission from J. Chromatogr., 515, 245 (1990), copyright © Elsevier Science Publishers.
10.3 Sample Preparation
2.
439
Protein Precipitation
In the simplest form of the procedure, proteins can be precipitated by adding 100 |iL of acetonitrile to 200 |LIL of blood plasma or serum that already contains an internal standard. The mixture is then vortex mixed for 30 s, allowed to stand for 5 min at room temperature, and centrifuged for 3 min at 9500g, and the supernatant is injected (56). The electropherograms are cleaner than for direct injection, but many endogenous components still appear. MECC is still advantageous, since some proteinaceous material may carry over into the supernatant. In conjunction with added salts to the sample, efficient stacking (Section 8.6) can be obtained (59-62). This is illustrated in Figure 10.7 (60). With the drugs dissolved in 67% acetonitrile-150 mM sodium chloride and 50% of the capillary filled with sample, efficient stacking is obtained (Figure 10.7a). The serum blank (Figure 10.7b) is relatively clean. The spiked serum sample is shown in Figure 10.7c. It is likely that tITP is the stacking mechanism at work and that chloride is the leading ion. 3.
Liquid-Liquid Extraction
Liquid-liquid extraction is useful for performing an offline trace enrichment. This is illustrated for the determination of thiopental in serum and plasma (6). Buffered serum (0.7 mL) containing an internal standard was extracted with 5 mL of pentane for 10 min and centrifuged. The upper organic layer was removed and evaporated to dryness. The residue was redissolved in 200 |lL of BGE, with separation by MECC. The electropherograms were free of endogenous sample peaks, and the results from patient samples correlated well with the HPLC assay. Should further enrichment and sensitivity be required, the pickup solvent volume can be reduced, and instead of using the BGE, a stacking electrolyte can be used. 4.
Solid-Phase Extraction
Solid-phase extraction is a widely used sample-preparation method for purifying drugs from biological fluids prior to HPLC. Wernly and Thormann (14) employ multistep solid-phase extraction to determine drugs of abuse such as barbiturates, hypnotics, amphetamines, opiods, benzodiazepines, and cocaine metabolites from a single urine specimen. In conjunction with multiwavelength detection, positive confirmation for drugs of abuse in screening urine samples is simple by MECC. The stepwise sample cleanup procedure is illustrated in Figure 10.8. In this threestep approach, methaqualone is eluted during the first step, morphine, codeine, and heroin during the second, and finally, benzoylecgonine in the third. There was some carryover between fractions that should be readily eliminated through fine-tuning. It is likely that this procedure can be easily adapted for the determination of a wide variety of drug substances in most biological fluid types.
440
Chapter 10
Putting It All Together
2
a
! '
-f
1 ' »i