Semiconductors and Semimetals, Vol. 17: C W Beam Processing of Silicon and Other Semiconductors

SEMICONDUCTORS AND SEMIMETALS VOLUME 17 CW Beam Processing of Silicon and Other Semiconductors Semiconductors and Sem...

Author: R. K. Willardson

20 downloads 638 Views 19MB 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

DOWNLOAD PDF

SEMICONDUCTORS AND SEMIMETALS VOLUME 17 CW Beam Processing of Silicon and Other Semiconductors

Semiconductors and Semimetals A Treatise

Edited by R . K . WILLARDSON

ALBERT C . BEER

WILLARDSON CONSULTING

BATlFLLE COLUMBUS LABORATORIES

SPOKANE, WASHINGTON

COLUMBUS, OHIO

SEMICONDUCTORS AND SEMIMETALS VOLUME 17 CW Beam Processing of Silicon and Other Semiconductors

Volume Editor JAMES F. GIBBONS DEPARTMENT OF ELECTRICAL ENGINEERING STANFORD ELECTRONICS LABORATORIES STANFORD UNIVERSITY STANFORD, CALIFORNIA

1984

ACADEMIC PRESS, INC. (Harcourt Brace Jovanovich, Publishers)

Orlando San Diego New York London Toronto Montreal Sydney Tokyo

Academic Press Rapid Manuscript Reproduction

COPYRIGHT@ 1984. BY ACADEMICPRESS, INC. ALL RIGHTS RESERVED. NO PARTOFTHIS PUBLICATION MAY BE REPRODUCEDOR TRANSMITTED IN ANY FORM OR BY ANY MEANS. ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOW. RECORDING.OR ANY INFORMATION SORAGE AND RETRIEVAL SYSTEM. WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

ACADEMIC PRESS,INC. Orlando. Florida 32887

United Kin dom Editionpublished by ACADEM~CPRESS. INC. (LONDON)LTD. 24/28 Oval Road, London NWl

7DX

Library of Congress Cataloging in Publication Data Ilaln entry under t i t l e :

Semiconductors and semlmtals. Vol. 13 edited by K. Zanlo; v. 17 edited by James F. Gibbons. Includes bibllographlcal references and Indexes. 1. Sealconduetors--Collected w r k r . 2. S a l m t a l s - Collected works. I. Ylllardson, Robart I(. 11. Beer, Albert C. QCb10.9eS47 537.6'22 65-26048 ISM 0-12-752117-8 (v. 17)

PRIhTED INTHECSlTEDS~AlESOPAMERlCA

84858687

9 8 7 6 3 4 3 2 1

Contents LISTOF CONTRIBUTORS . . . . . . . . . . . . . . . . . . . . . . PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . .

ix xi

Chapter 1 Beam Processing of Silicon James F . Gibbons 1.1 1.2 1.3 1.4 1.5 1.6

Central Features of Pulsed Beam Annealing of Ion Implanted Silicon Central Features of CW Beam Annealing of Ion Implanted Silicon . Annealing with Large Diameter Scanning CW Sources . . . . CW Beam Recrystallization of Polysilicon Films . . . . . . Silicide Formation with Scanning Beams . . . . . . . . Rapid Thermal Processing with Stationary Incoherent Sources . . References . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . .

2 30 40 46 51 55

66

Chapter 2 Temperature Distributions and Solid Phase Reaction Rates Produced by Scanning CW Beams Arto Lietoila. Richard B . Gold. James F. Gibbons. and Lee A . Christel 2.1 2.2 2.3 2.4 2.5 2.6

Introduction . . . . . . . . . . . . . . . . . . . The Kirchhoff Transform . . . . . . . . . . . . . . . Temperature Calculations for CW Lasers . . . . . . . . . Temperature Calculations for Wide Area Optical Sources: The Arc Lamp Temperature Calculations for Scanning CW Electron Beams . . . Reaction Rate Calculations . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . .

. . . . . . .

. . . .

71 72 74 89 93 101 105

Chapter 3 Applications of CW Beam Processing to Ion Implanted Crystalline Silicon Arto Lietoila and James F. Gibbons 3.1 Introduction .

. . .

. . . V

107

vi 3.2 3.3 3.4 3.5 3.6 3.7

CONTENTS

Experimental Methods . . . . . . . . . . . . . CW Laser Induced Crystallization of Amorphous Silicon . Annealing of “As Implanted Si: C < C.. . . . . . . Annealing of “As Implanted Si: C > C,, . . . . . . High Resolution Selective E-Beam Annealing of As Implanted Annealing of B Implanted Si . . . . . . . . . . . . . . . . . . . . . . . . References .

. . . . . . . .

. . . .

Silicon

. . . .

. . . .

. . . .

. . . .

. 108 . . 122 . . 137 . . 147

.

.

.

.

.

.

165

. . . . . . . . . 166 . . . . . . . .

173

Chapter 4 Electronic Defects in CW Transient

Thermal Processed Silicon N . M . Johnson 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8

Introduction . . . . . . . . . . . . . Device Processing and Measurement Techniques . CW Beam Processed Bulk Silicon . . . . . Lateral Nonuniformities in Scanned-Beam Annealing Isothermal Transient Annealing . . . . . . Post-Recrystallization Processing . . . . . Beam-Crystallized Silicon Thin Films . . . . Summary and Conclusions . . . . . . . References . . . . . . . . . . . . .

. . . . . . . . . . . 177 . . . . . . . . . . . 179 . . . . . . . . . . . . 184

.

. . . . . . . . . . . 195 . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . 200 . . . 204 . . . 211 . . . 220 . . 223

Chapter 5 Beam Recrystallized Polycrystalline Silicon:

Properties. Applications. and Techniques K . F. Lee . T. J . Stultz. and James F . Gibbons . I Recrystallization of Thin Polycrystalline Films with a Scanning CW Laser . .2 Resistivity Reduction in Beam-Recrystallized Polysilicon Films . . . . .3 Electronic Properties of Laser Recrystallized Films Intended for MOS Device Fabrication . . . . . . . . . . . . . . . . . . . . . 5.4 Characteristics of MOS Devices and Integrated Circuits Fabricated on Laser Recrystallized Polysilicon Films . . . . . . . . . . . . . . 5.5 Improvements in the Recrystallization Process . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . .

. . . . 228 . . . . 247 . . .

250

. .

. . . 260 . . . 278 . . . 335

Chapter 6 Metal-Silicon Reactions and Silicide

Formation T. Shibatu. A . Wakita. 7: W. Sigrrioii. and James F. Gibbons 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 341 6 . 2 Experimental Techniques . . . . . . . . . . . . . . . . . . . . 344 6.3 Experimental Results: Laser Processing . . . . . . . . . . . . . . . . 357

vii

CONTENTS

6.4 Experimental Results: Electron Beam Processing . . . . . . . . . . . . . 371 6.5 Thermal Stability and Oxidation Properties of CW Beam Reacted Silicides . . . . . 375 6.6 Laser Processing of Nb3X Superconductors . . . . . . . . . . . . . . 385 References . . . . . . . . . . . . . . . . . . . . . . . . 393

Chapter 7 CW Beam Processing of Gallium Arsenide Yves I . Nissim and James F . Gibbons 7.1 7.2 7.3 7.4 7.5

Introduction . . . . . . . . . . . . . . . CW Laser Annealing of Ion Implanted GaAs . . . . CW Laser Processing of Thin Films Deposited on GaAs . Rapid Thermal Processing . . . . . . . . . Novel Applications of Rapid Thermal Processing in GaAs References . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

INDEX . . . . . . . . . . . . . . . . . . . . . CONTENTS OF PREVIOUS VOLUMES . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

.

397

. . 398 . . 411 . . 430 . . 436 . 443

. . . 447 . . . . 453

This Page Intentionally Left Blank

List of Contributors Numbers in parentheses indicate the pages on which the authors' contributions begin.

LEEA. CHRISTEL',Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (71) JAMES F. GIBBONS, Department of Electrical Engineering, Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (1, 71, 107, 227, 341, 397) RICHARDB . GOLD', Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (71) N. M. JOHNSON,Xerox Corporation, Palo Alto Research Center, Palo Alto, California 94304 (177) K . F. LEE3, Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (227) ARTOLIETOILA4, Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (71, 107) YvES I. NISSIM', Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (397) T. SHIBATA', Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (341) T. W. SIGMON, Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (341) T. J. STULTZ', Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (227) A. WAKITA,Stanford Electronics Laboratories, Stanford University, Stanford, California 94305 (341) 'Present address: SERA Solar Corporation, Santa Clara, California 95054. 'Present address: Adams-Russell Company, Burlington, Massachusetts 01803. 3Present address: AT&T Bell Laboratories, Holmdel, New Jersey 07733. 4Present address: Micronas, Inc., 00101 Helsinki, Finland. *Present address: CNET, 92220 Bagneux, France. 6Present address: VLSI Research Center, Toshiba Research and Development Center, Kawasaki City, Kanagawa, 201 Japan. 'Present address: TS Associates, San Jose, California 95128.

ix


Preface This book had its beginnings at the 1979 winter meeting of the American Physical Society, at which I was asked to review the basic features of cw laser processing of ion implanted silicon. The review covered techniques for calculating laser-induced temperatures, experimental results on the annealing of ion implanted layers in silicon, and enough material on structural (TEM) and electronic (DLTS) defects to suggest that the process was reasonably well behaved and perhaps had a future in materials preparation and processing. A1 Beer, who was attending the meeting, thought the subject might be suitable for a short research monograph that would be useful to those entering the field. We agreed that, as things seemed reasonably tidy, the project could easily be finished within a year. However, the field was just then entering a period of very rapid growth. The fabrication of MOSFETS on polysilicon films recrystallized by a cw scanning laser, reported in July 1979, opened up the field of silicon-on-insulators (SOI). There followed a succession of significant contributions to this technology at a number of laboratories, dealing with (1) such things as thermal profile management, beam shaping, seeding, and zone recrystallization on the materials side and (2) a number of new device configurations, including the general technique of three-dimensional integration, on the applications side. Simultaneously, the use of a cw scanning laser beam to react metals with silicon to form single-phase silicide layers was being explored; and its potential consequences for materials, device, and integrated circuit technology required that it be included if the book was to offer a broad, balanced view of the field. Finally, as these areas began to settle down enough to be reviewed with some reasonable perspective, the rapid thermal processing technique was reported in Japan. Its natural relationship to the cw scanning laser process and its “throughput” advantage from a manufacturing point of view then required a further modification of many chapters of the book. The result is that 5 years have passed since A1 Beer and I shook hands on the project. The cw beam technology as it relates to semiconductor processing is now much more mature and better understood than it was in 1979-1980, though its applications in the semiconductor industry are very likely still in their infancy. It is on this hopeful note that my coauthors and I now offer this volume. Chapter 1 provides an overview of the general field of beam processing, with a brief discussion of the main topics to be developed in depth in the succeeding chapters. Chapter 2 deals with the calculation of cw beam-induced temperature

xi

xii

PREFACE

profiles in semiconductor substrates under a variety of conditions that are found in practical applications. A review of the use of cw beam processing for annealing ion implanted silicon is presented in Chapter 3, and the question of electronic defects remaining after various implantation and cw beam annealing cycles is reviewed in Chapter 4. In Chapter 5 we discuss the recrystallization of thin polysilicon films by cw beam systems and summarize its applications in device fabrication. The use of a cw beam to react a thin metal film with a silicon (crystalline or polycrystalline) substrate to form a silicide is reviewed in Chapter 6 , together with oxidation characteristics and other properties of the silicide layers that are important for practical applications. Finally, in Chapter 7 we review the current state of the art regarding the use of cw beams for processing of GaAs. It will be apparent from this chapter that much remains to be done in the beam processing of compound semiconductors. To conclude this preface, I would like to express my most sincere appreciation to the authors of these chapters. We were all privileged to share the experience and excitement of working in a field during the early stages of its development, both as colleagues at Stanford and as members of the larger community of “beam annealers” whose pioneering work is described here in terms that we hope will reflect the respect we have for all of them. My coauthors and I would also like to acknowledge our enormous indebtedness to Ms. Mary Cloutier, both for her extraordinary skill in preparing this manuscript and for the patience and unfailing good humor that she maintained through its several revisions. Finally, 1 want to thank Dick Reynolds and Sven Roosild at DARPA, Ben Wilcox at NSF, and Horst Wittman at ARO for the support that enabled the Stanford group to participate in the development of the field.

JAMES F. GIBBONS

CHAPTER 1

Beam Processing of Silicon James F. Gibbons STANFORD ELECTRONICS LABORATORIES STANFORD UNIVERSITY STANFORD, CALIFORNIA

1.1 CENTRAL FEATURES OF PULSED BEAM

ANNEALINGOF ION IMPLANTED SILICON . . . . . . . . . . . . . . . 1.2 CENTRAL FEATURES OF C W BEAMANNEALING OF ION IMPLANTED SILICON

..........................

1.3 ANNEALING WITH LARGEDIAMETER SCANNING Cw SOURCES ............................. 1.4 CW BEAM RECRYSTALLIZATION OF POLYSILICON FILMS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 SILICIDE FORMATION WITH SCANNING BEAMS . . . . . . . . . . . 1.6 RAPID THERMAL PROCESSING WITH STATIONARY INCOHERENT SOURCES

...................

REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 30 40 46 51

55 66

High power lasers have been used i n t h e semiconductor indust r y f o r many y e a r s , f o r a p p l i c a t i o n s such as wafer s c r i b i n g , r e s i s t o r trimming and t h e d r i l l i n g of ceramics, where t h e c l e a n removal of a s p e c i f i c amount of material i s r e q u i r e d ; and f o r c o n t a c t a l l o y i n g , where permanent, - r e l a t i v e l y e x t e n s i v e chemical D e s p i t e t h e mature development of t h e s e changes are sought [1.1]. a p p l i c a t i o n s , however, lasers and e l e c t r o n beams have only rec e n t l y been considered f o r t h e p r o c e s s i n g o p e r a t i o n s t h a t are required during s i l i c o n device f a b r i c a t i o n . The e a r l i e s t work w i t h Q-switched lasers w a s performed by Russian workers [1.2-1.41, who showed t h a t i o n implanted s i l i c o n could be annealed s u c c e s s f u l l y w i t h laser p u l s e s of s u f f i c i e n t energy and d u r a t i o n . The importance of t h i s work was q u i c k l y a p p r e c i a t e d by workers a t s e v e r a l d i f f e r e n t l a b o r a t o r i e s i n Europe [1.5-1.71 and t h e United S t a t e s 11.8-1.101, and a large volume of published work has now appeared on t h e use of both Qswitched lasers and pulsed e l e c t r o n beams f1.111 f o r semiconductor A t t h e same t i m e , t h e use of scanning cw laser and processing. e l e c t r o n beams f o r a n n e a l i n g i o n implanted s i l i c o n was being e x p l o r e d a t S t a n f o r d [1.12-1.141, w i t h a p a r t i c u l a r f o c u s on i t s a p p l i c a t i o n t o semiconductor device p r o c e s s i n g . More r e c e n t l y , both scanning and s t a t i o n a r y i n c o h e r e n t l i g h t s o u r c e s have been used t o a c h i e v e a n n e a l i n g of i o n implanted s i l i c o n w i t h a wafer throughput t h a t i s compatible w i t h t h e needs of t h e semiconductor industry. SEMICONDUCTORS AND SEMIMETALS, VOL 17

1

Copynght 0 1984 by Academic Press, Inc All nghts of reproductlon m any form reserved ISBN 0-12-752117-8

2

J . F. GIBBONS

A s a r e s u l t of t h e s e e f f o r t s , beam a n n e a l i n g of i o n implanted s i l i c o n has now been r a t h e r thoroughly s t u d i e d and shown t o possess s e v e r a l p o t e n t i a l advantages over t h e f u r n a c e a n n e a l i n g process. I n a d d i t i o n , l a s e r s , e l e c t r o n beams and a r c sources have been used t o r e c r y s t a l l i z e vapor d e p o s i t e d p o l y s i l i c o n , w i t h s u b s t a n t i a l improvements i n i t s e l e c t r o n i c p r o p e r t i e s [1.15,1.16] ; t o f a c i l i t a t e t h e formation of metal s i l i c i d e s [1.17-1.201; and t o perform a number of o t h e r p r o c e s s i n g f u n c t i o n s t h a t are of i n c r e a s i n g importance i n t h e f a b r i c a t i o n of f i n e geometry i n t e g r a t e d c i r c u i t s and h i g h speed d e v i c e s . I n t h i s c h a p t e r w e w i l l review t h e b a s i c mechanisms by which beam a n n e a l i n g proceeds and summari z e t h e b a s i c m e t a l l u r g i c a l and e l e c t r o n i c p r o p e r t i e s of beamannealed material.

Two d i s t i n c t l y d i f f e r e n t beam a n n e a l i n g mechanisms have been i d e n t i f i e d , depending on t h e d u r a t i o n of t h e beam exposure. For Q-switched l a s e r s o r pulsed e l e c t r o n b e a m s , exposure times a r e t y p i c a l l y i n t h e range of 5 ns - 500 ns and t h e a n n e a l i n g p r o c e s s then i n v o l v e s t h e formation of a t h i n s u r f a c e l a y e r of molten s i l i c o n t h a t r e c r y s t a l l i z e s on t h e u n d e r l y i n g s u b s t r a t e when t h e r a d i a t i o n i s removed. If t h e i r r a d i a t e d sample i s an i o n implant e d s i n g l e c r y s t a l and t h e depth of t h e molten l a y e r i s s u f f i c i e n t t o envelop t h e i m p l a n t a t i o n damaged region, t h e molten l a y e r regrows by a l i q u i d phase e p i t a x i a l process on t h e c r y s t a l l i n e s u b s t r a t e t o produce material w i t h a very high degree of s t r u c t u r a l p e r f e c t i o n and very s u p e r i o r e l e c t r o n i c p r o p e r t i e s . For cw systems, on t h e o t h e r hand, t h e s i l i c o n s u r f a c e i s t y p i c a l l y exposed t o t h e beam f o r 0.1-10 ms and, i n some c a s e s , f o r d u r a t i o n s of s e v e r a l t e n s of seconds. The a n n e a l i n g of i o n implanted m a t e r i a l can then proceed by a s o l i d phase e p i t a x i a l regrowth process a t temperatures t h a t a r e w e l l below t h e m e l t i n g p o i n t . A s i n t h e pulsed beam case, a very h i g h degree of cryst a l l i n e p e r f e c t i o n and very s u p e r i o r e l e c t r o n i c p r o p e r t i e s can be obtained under a p p r o p r i a t e a n n e a l i n g c o n d i t i o n s . However, t h e absence of m e l t i n g proves t o be of some i n t e r e s t s i n c e no r e d i s t r i b u t i o n of t h e implanted impurity p r o f i l e then occurs during t h e a n n e a l i n g p r o c e s s , whereas s i g n i f i c a n t impurity r e d i s t r i b u t i o n occurs when a n n e a l i n g i s e f f e c t e d by a p u l s e d laser o r e l e c t r o n beam. These and o t h e r d i f f e r e n c e s i n t h e two a n n e a l i n g In processes make i t convenient t o d i s c u s s them s e p a r a t e l y . what f o l l o w s w e f i r s t c o n s i d e r t h e pulsed beam a n n e a l i n g p r o c e s s and then t a k e up t h e cw a l t e r n a t i v e .

1.1

CENTRAL SILICON

FEATURES OF PULSED

BEAM ANNEALING OF ION IMPLANTED

W e w i l l begin w i t h a b r i e f review of experimental d a t a t h a t c h a r a c t e r i z e t h e a n n e a l i n g of i o n implanted s i l i c o n u s i n g a Qswitched l a s e r and then t u r n t o a b r i e f mathematical a n a l y s i s of the process.

1. BEAM PROCESSING OF SILICON

3

Laser a n n e a l i n g has been demonstrated u s i n g a v a r i e t y of s o u r c e s and p u l s e d u r a t i o n s . I n most of t h e experiments r e p o r t e d s o f a r , implanted samples have been annealed d i r e c t l y i n t h e l a b o r a t o r y ambient; i.e., no s p e c i a l p r e c a u t i o n s have been taken t o immerse t h e wafer i n an i n e r t environment d u r i n g annealing. The area i r r a d i a t e d by t h e beam i s , i n most cases, l a r g e compared t o b o t h t h e d i f f u s i o n l e n g t h f o r h e a t i n t h e s o l i d and t h e t h i c k n e s s of t h e w a f e r , s o t h e p r o c e s s can be t r e a t e d as being b a s i c a l l y one dimensional.

Experimental r e s u l t s from v a r i o u s l a b o r a t o r i e s are reasonably c o n s i s t e n t and are summarized i n Table 1.1. For convenience, t h e d a t a are s e l e c t e d t o i l l u s t r a t e t h e a n n e a l i n g of As+-implanted s i l i c o n w i t h t h e As+ i m p l a n t a t i o n c o n d i t i o n s chosen t o provide amorphous r e g i o n s of d i f f e r i n g t h i c k n e s s e s a t t h e sample s u r f a c e . For The p r i n c i p a l f e a t u r e s of t h e d a t a are as follows. p u l s e s of 25-100 ns d u r a t i o n ( t y p i c a l of a Q-switched l a s e r ) :

1. Surface m e l t i n g is i n i t i a t e d at a t h r e s h o l d energy of 0.2-3.5 J/cm2, depending on t h e laser wavelength and t h e t h i c k n e s s of t h e amorphous l a y e r . 2. F u l l a n n e a l i n g of a n i o n implanted l a y e r ( a s judged by c r y s t a l recovery and e l e c t r i c a l a c t i v i t y ) r e q u i r e s a d d i t i o n a l energy i n an amount t h a t a l s o depends on both t h e laser wavelength and t h e t h i c k n e s s of t h e amorphous l a y e r . TABLE 1.1. of As+-Implanted

Laser

Experimental R e s u l t s f o r Pulsed Laser Annealing Silicon AS+

d'

Melt Initiated

Imp. P a r . Ruby 3 50 n s

5x1Ol5, 400 keV

43008

Ruby 9 60 n s

1.4x1Ol6, 100

17008

Fully Recovered

2 J/cm2 0.64

Ref.

[5]

1.4

Nd :YAG ,

110 ns

8x1Ol5, 100 keV

40 ns

130 keV

468

3.5

4.5

[211

30 keV

468

0.2

0.8

[211

Nd:YAG, Doubled 40 n s

-15008

6 J/cm2

[7]

4

J. F. GIBBONS

E a r l y evidence s u g g e s t i n g t h e s u r f a c e m e l t i n g t h e o r y of pulsed laser a n n e a l i n g w a s provided by Auston e t a l . [1.21], who measured t h e o p t i c a l r e f l e c t i v i t y of samples d u r i n g t h e a n n e a l i n g c y c l e w i t h a low power (2mW) continuous He-Ne laser. F i g u r e 1.1 shows a t y p i c a l t i m e r e s o l v e d s u r f a c e r e f l e c t i v i t y measurement . with a when a Nd:YAG l a s e r o p e r a t e d a t a wavelength of 0.53 m p u l s e w i d t h of 30 n s and a t o t a l energy d e n s i t y energy of 2.75 J/cm2 i s used t o i r r a d i a t e a s i l i c o n sample implanted w i t h As' a t 30 keV t o a dose of 1015/cm2. A s t h e a n n e a l i n g l a s e r t u r n s on, t h e i n i t i a l r e f l e c t i v i t y of t h e amorphous l a y e r , Ra, rises w i t h i n c r e a s i n g sample t e m p e r a t u r e t o a v a l u e It;, where s u r f a c e The r e f l e c t i v i t y t h e n rises a b r u p t l y t o melting i s i n i t i a t e d . t h e r e f l e c t i v i t y of molten s i l i c o n and remains a t t h i s v a l u e a s t h e l i q u i d - s o l i d i n t e r f a c e p e n e t r a t e s more deeply i n t o t h e sample. The p e n e t r a t i o n ceases a t a t i m e a f t e r t h e a n n e a l i n g laser p u l s e has t e r m i n a t e d t h a t depends on t h e t o t a l energy absorbed d u r i n g the pulse.

A s t h e l i q u i d - s o l i d i n t e r f a c e r e v e r s e s i t s d i r e c t i o n and proceeds back toward t h e s u r f a c e , t h e molten zone regrows on t h e u n d e r l y i n g s i l i c o n c r y s t a l . The r e f l e c t i v i t y u l t i m a t e l y drops t o RG, t h e r e f l e c t i v i t y of t h e h o t s o l i d . The s u r f a c e is m e l t e d The f a l l t i m e , T f , i s t h e time r e q u i r e d f o r a t o t a l time 5. f o r t h e regrowth f r o n t t o move through one o p t i c a l s k i n d e p t h and i s a measure of t h e regrowth v e l o c i t y . The r e f l e c t i v i t y d e c r e a s e s t o i t s room t e m p e r a t u r e v a l u e f o r c r y s t a l l i n e s i l i c o n , Rc, as t h e s u r f a c e cools.

100 nrldiv. FIGURE 1.1. Ref. 1.21).

T i m e r e s o l v e d s u r f a c e r e f l e c t i v i t y trace ( a f t e r

1. BEAM PROCESSING OF SILICON 1.1.1

5

Qualitative Analysis of the Melt Threshold

The initial interaction of a light beam with a material is always with its electrons. Energy absorbed by the electrons is ultimately shared with the atoms of the material as the electronic excitation is transformed into heat. Under most conditions the transfer of electronic excitation to heat is accomplished with relaxation times on the order of 1 ps or less and may therefore be considered to be instantaneous. The basic process of pulsed beam annealing then consists of the absorption of sufficient energy to melt a layer of silicon having a thickness at least equal to the thickness of the implantation-damaged region, followed by liquid phase epitaxial regrowth of the melted layer. The laser energy required to bring the surface of the sample to its melting temperature is therefore a critical parameter in the process. To estimate the threshold energy required to produce surface melting, we consider the sample geometry shown in Fig. 1.2 [1.22]. The sample consists of an implantation-damaged layer of thickness xd and optical absorption coefficient ad resting on a crystalline solid with optical absorption coefficient aC. A laser pulse of intensity 1, and duration T~ illuminates the front surface of the sample at normal incidence. The reflection coefficient at this surface is R. The laser intensity is assumed to be sufficiently low that ordinary one-photon optical absorption processes are dominant.

LASERPULSE ABSORBING MEDIUM

Xd

FIGURE 1.2. Intensity and temperature profiles in laserirradiate silicon. (a) Penetration depth of light small compared (b) Penetration depth of light large compared to (DT )If2. to (DT;)~’~ (after Ref. 1.22).

J . F. GIBBONS

6

The r e l e v a n t thermal p r o p e r t i e s of t h e material are i t s s p e c i f i c h e a t C v , i t s mass d e n s i t y p and i t s t h e r m a l c o n d u c t i v i t y K. The h e a t d i f f u s i v i t y i s t h e n g i v e by D = K/pCv and d e t e r T h i s l e n g t h g i v e s an mines a c h a r a c t e r i s t i c l e n g t h (DT ) P e s t i m a t e of t h e d i s t a n c e o v e r which t h e t e m p e r a t u r e p r o f i l e i s s p r e a d by h e a t d i f f u s i o n d u r i n g t h e laser p u l s e .

”.

To estimate t h e s u r f a c e t e m p e r a t u r e r i s e produced i n a l i m i t i n g c a s e of p r a c t i c a l i n t e r e s t , we suppose t h a t t h e laser e n e r g y i s absorbed i n a the diffusion length f o r h e a t [a-l < Under t h e s e c o n d i t i o n s [ i l l u s t r a t e d i n Fig. 1.2(a)] can be c o n s i d e r e d t o h e a t a s l a b of t h i c k n e s s ( D T ~ ) The c o r r e s p o n d i n g temperat u r e r i s e AT i s t h e n o b t a i n e d from

.

( I o ~ p ) ( l - R ) = PC,AT(DT~)1 / 2

(1.1)

T h i s formula s u g g e s t s t h a t t h e s u r f a c e t e m p e r a t u r e i s p r o p o r t i o n a l t o t h e energy i n t h e laser p u l s e (Io T ~ and ) predicts a t h r e s h o l d e n e r g y of

a-1 6 ( D T ~ ) ~ / Z

(1.2)

where Tm i s t h e m e l t i n g t e m p e r a t u r e and To i s t h e ambient temperature. The extreme t e m p e r a t u r e change e n v i s a g e d (- 14OOOC) of c o u r s e makes i t n e c e s s a r y t o a c c o u n t f o r t h e t e m p e r a t u r e dependence of t h e t h e r m a l and o p t i c a l p a r a m e t e r s a p p e a r i n g i n E q . (1.2) i n o r d e r t o a r r i v e a t a meaningful estimate of E& f o r p r a c t i c a l u s e . The p a r a m e t e r s t h a t a r e c u s t o m a r i l y used f o r p r o c e s s a n a l y s i s are g i v e n i n Table 1.2. The t h e r m a l c o n d u c t i v i t y i s an e s p e c i a l l y s e n s i t i v e f u n c t i o n of t e m p e r a t u r e and i t s t e m p e r a t u r e dependence must be t a k e n i n t o account t o o b t a i n a c c u r a t e t h e o r e t i c a l p r e d i c t i o n s . For n u m e r i c a l e s t i m a t e s w e u s e t h e a v e r a g e v a l u e of K c a l c u l a t e d from T

(1.3a) LO

w i t h K(T) f o r s i l i c o n g i v e n by

(W/ cm°K)

(1.3b)

1.

BEAM PROCESSING OF SILICON

7

TABLE 1.2. Thermal and Optical Parameters for Solid Silicon. The optical absorption coefficients for implantation-amorphized silicon have not been measured; values quoted for ad are for sputtered silicon. Parameter Cv, J/gm-OK P , gm/cm3 K, W/cm-OK R ad ( h = 0.69 ad ( h = 1.06 ac (1 = 0.69 ac ( h = 1.06

Value at T = 300°K

pm) pm) pm) pm)

Value at T = 1685'K

0.65 2.33 1.45 0.35 3x104 3-6~10~ 3x10 10

Ref.

0.95 2.33 0.25 0.7

Substitution of numerical values from Table 1.2 leads to values of E& that are in satisfactory agreement with experiment for short wavelength irradiation such as that from ruby or frequency-doubled Nd:YAG lasers, where the basic assumption of heat transport made in t e analysis is at least approximately valid The validity of this assumption is [i.e., a-1 < (DT )lh]. improved if an Bmplantation-amorphized sample is irradiated, of course, since the optical absorption coefficient is then increased. For Q-switched Nd:YAG lasers operated at the fundamental frequency, however, the optical absorption depth (a-l) in crystalline silicon is substantially greater than the diffusion length for heatZ An order of magnitude estimate of the melt threshold energy Em can be made for this case by assuming that the thickness lab that is heated by the radiation is (a-l). Replacing a-1 in Eq. (1.2), we obtain

with

a-1

exp

8

The steady state temperature distributions can now be readily obtained by first solving Eq. (2.4) €or 8 , and then applying the inverse Kirchhoff transform Eq. (2.7) to obtain the true temperature. The normal boundary Conditions for T are

ART0 LIETOILA ET AL..

74 T (z=L)

=

To

(2.9)

where z i s the depth coordinate, and L is the sample thickness. TABLE 2.1. The Fitting Parameters, A and TK, for the Thermal Conductivity of Various Semiconductors [ K = A/(T-T,].

Material

A(W/cm)

si

299

GaAs InP

sic

91 115.5 430

TK("K>

99 90.9 130.5 320

Temperature Range

> > > >

300°K 250°K 300°K 500°K

Refs. 2.2,2.4,2.6 2.2,2.3,2.6 2.3, 2 . 6 2.5, 2.6

Condition (2.8) follows from the fact that the heat losses through the sample front surface (radiation, convection) are assumed to be negligible. Condition (2.9) means that the sample is in perfect thermal contact with a heat sink having a temperature Those conditions imply for 8 : To. (2.10) 0 (z=L) =

0

(2.11)

In the case of a thermally isolated sample, radiation losses dominate and these boundary conditions become invalid. This case will be discussed in Section 2.4. In Section 2.3 we develop solutions to Eq. ( 2 . 4 ) using the boundary conditions Eqs. (2.10) and (2.11). 2.3

TEMPERATURE CALCULATIONS FOR CW LASERS

The majority of scanning beam annealing experiments to date have been performed using cw lasers. We therefore begin by presenting, in this section, temperature calculations for cw laser processing. As was mentioned in Section 2.1, cw laser (and circular electron) beams are, in annealing applications, typically so tightly focussed that the beam dimensions are clearly much less than the thickness of the semiconductor wafers. Therefore, both the vertical and lateral heat flows must be considered, and the temperature calculations have to be three-dimensional. On the other hand the actual layers which are subject to the beam

2. TEMPERATURE DISTRIBUTIONS AND REACTION RATES

75

processing (e.g., implanted layers or deposited thin films) are usually less than a micron thick, and are therefore thin compared to the typical beam diameters (20-100 pm). Therefore, the vertical temperature drop across the layer to be processed is negligible, and the surface temperature can be assumed to prevail across the entire layer. (An exception to this is the use of a modified SEM for beam annealing, in which case the beam dimensions can be comparable to the thickness of the layer.) We will begin by showing how the maximum surface temperatures are obtained, and we will demonstrate that in practical applications the temperature can be assumed to be uniform in depth throughout the the layers to be processed. The effects of the wafer thickness, absorption coefficient, beam shape (elliptical, circular), and scanning speed are then studied. Finally, we will present a calibration procedure to facilitate the determination of the processing temperatures in practical annealing experiments. The Maximum Temperature Obtained with a Scanning CW Laser

2.3.1

The maximum temperature is achieved at the sample surface at the center of the beam under the following conditions:

1.

The beam is circular.

2.

The beam is stationary.

3.

The wafer is semi-infinite.

4. The absorption depth can be assumed to be zero. In most annealing experiments, these conditions are practically met because: spherical lasers are used, the scanning speeds are slow, the wafer thickness is usually several times the beam diameter, and e.g., for the Ar+ laser, the absorption depth is only about 0.7 pm. In what follows we present the temperature calculations assuming that the above conditions 1-4 prevail. The effects of deviations from these conditions are studied in later sections. The high power cw lasers used for beam annealing are usually designed to operate in the TEMoo mode, in which case the lateral beam intensity dietribution can adequately be approximated by a Gaussian distribution:

I = I~ exp (-r2/2)

(2.12)

76

ART0 LIETOILA ET AL.

where r i s t h e d i s t a n c e from t h e beam c e n t e r and w i s d e f i n e d as t h e beam r a d i u s . We w i l l f o l l o w t h i s d e f i n i t i o n of w ( l / e - p o i n t s i n i n t e n s i t y ) throughout t h i s volume. However, o t h e r d e f i n i t i o n s have a l s o been used i n t h e l i t e r a t u r e . C l i n e and Anthony [2.7], and N i s s i m e t a l . (2.21 d e f i n e t h e beam r a d i u s by t h e l/&-points i n i n t e n s i t y , which g i v e s a r a d i u s a f a c t o r of fi smaller t h a n o u r s . On t h e o t h e r hand, t h o s e doing r e s e a r c h on lasers p e r se, sometimes d e f i n e t h e beam r a d i u s by t h e l / e p o i n t i n t h e e l e c t r i c f i e l d s t r e n g t h , which corresponds t o t h e l / e 2 - p o i n t s i n i n t e n larger s i t y . This g i v e s a beam r a d i u s which i s a f a c t o r of than t h a t determined a c c o r d i n g t o t h e d e f i n i t i o n i n Eq. (2.12). The d i f f e r e n t d e f i n i t i o n s are compared i n Table 2.2. TABLE 2.2. Different Definitions Gaussian Laser Beam (TEMoo Mode)

Defining v a l u e of i n t e n s i t y

Symbols used

of

the

Relative value f o r a g i v e n beam

I/&

R, r

1/n

1/e l/e2

W

1

W

fi

for a

Beam, Radius

Refs.

2.2, 2.8, 2.17

2.6, 2.9

2.7

Assuming a Gaussian beam shape and c o n d i t i o n s 1-4, above, Lax [2.8] h a s shown t h a t t h e maximum l i n e a r s u r f a c e t e m p e r a t u r e a t t h e beam c e n t e r i s g i v e n by (2.13)

W e define the quantity within brackets absorbed power p e r u n i t r a d i u s , Pa:

Pa =

P (1-R) w

i n Eq.

(2.13)

as

the

(2.14)

.

The maximum t r u e t e m p e r a t u r e v s Pa f o r d i f f e r e n t semiconthe inverse d u c t o r s can now be o b t a i n e d u s i n g E q s . (2.13-2.14), Kirchhoff t r a n s f o r m , and t h e e x p r e s s i o n s f o r t h e t h e r m a l conducThe r e s u l t s are g i v e n i n F i g . 2.1 f o r S i , Sic, t i v i t y , Eq. ( 2 . 5 ) . GaAs and I n P u s i n g d i f f e r e n t s u b s t r a t e backsurface t e m p e r a t u r e s T o . That f i g u r e shows- t h a t u s e of h i g h e r v a l u e s of To r e s u l t s i n a s u b s t a n t i a l r e d u c t i o n of laser power f o r a g i v e n s u r f a c e t e m p e r a t u r e . This i s due t o t h e f a c t t h a t t h e thermal c o n d u c t i v i t y f o r a l l of t h e s e materials d e c r e a s e s w i t h i n c r e a s i n g t e m p e r a t u r e . It i s f o r t h e same reason t h a t t h e T v s . P curves are n o n l i n e a r and concave up.

2500

-

0,

2000

-

5

1500

-

u

a Si C

I

500

1000 Pa --

1400 -

To =550°C 35OoC

1500

2000

2500

500

.

I

I

.

1500 (1-R)P

(W/cm)

15OoC

.

1000

paZw

I

2000

.

~

2500

.

1

.

3000

(W/crn)

25OC

C InP

100

200

300

400

500

600

700

800

FIGURE 2.1. Maximum surface temperature a t the beam center as a f u n c t i o n of t h e absorbed power p e r u n i t radius f o r ( a ) s i l i c o n , (b) s i l i c o n carbide, ( c ) gallium arsenide and ( d ) indium phosphide.

78


2.3.2

Vertical and Lateral Temperature Distributions

I f the maximum temperature vs. depth must be calculated, the following formula, derived by Lax [2.8], can be used for the linear temperature at the beam center:

where Cmx(Z=O) is given by Eq. (2.13) and Z = z/w. We plot in Fig. 2.2 the ratio G(z)/Gmx, which we call the normalized linear temperature nz, as a function of the dimensionless depth coordinate Z = z/w. From Fig. 2.2(a), it is evident that in most cases the surface temperature can be used throughout the layers of interest. For example, using a typical beam radius of 40 pm, the temperature within the first 1 pm of the sample does not vary by more than 4 % .

b

x

2

-? N

Y

0

t

0.92 0901

0

"

0.01

'

1

0.02

'

z

' 003

'

1

.

1 I

0.04 0.05

0

1

2

z

3

4

5

FIGURE 2.2. Normalized linear temperature G(Z)/Gmx as a function of normalized depth Z = z/w at the beam center (x=y=O) for (a) small and (b) large values of 2. The linear temperature at the sample surface at a distance r from the beam center can be expressed L2.81 as a function of the dimensionless distance R = r/w: O(R) , =,O

(R=O) exp

(- F)I, (5)

(2.16)

where here I, is the modified Bessel function of order zero. We plot in Fig. 2.3 the ratio O(R)/Q,,(R=O) as a function of R and compare this to the normalized laser intensity distribution I(R)/Io. That figure or Eq. (2.16) together with Eqs. (2.7) and (2.13), can be used to obtain the true surface temperature as a function of distance from the beam center.

2.

TEMPERATURE DISTRIBUTIONS AND REACTION RATES

79

R = r/w

FIGURE 2.3. Normalized linear temperature O(R)/Omax as a function of normalized lateral distance R = r/w from the beam center at the sample surface (PO; r2=x2+y2). For comparison, the dashed line shows the profile of laser intensity. 2.3.3

Effect of Sample Thickness

The formulas given above for 0, Eqs. (2.13-2.15), were derived assuming that the sample is infinitely thick compared to the beam dimensions. This is not always the case: use of very powerful lasers and long focal length lenses may result in beam diameters in excess of 100 pm and the sample thickness must then be taken into account. A quick check of the validity of the infinite sample assumption can be performed by using Eq. (2.15) to calculate what the linear temperature rise would be at a depth equal to the actual sample thickness in an infinitely thick wafer. If the result is close to zero, no correction is required. Otherwise, the formulas given above, and Fig. 2.1, give temperatures that are too high, and an appropriate correction must be carried out. We will now present this correction for the linear temperature 8. Equation (2.7) can then be applied to obtain the true surface temperatures in wafers of finite thickness.

It has been shown using the method of images f2.91 that the linear surface temperature, k, in a sample whose thickness is L can be expressed as the following series: m

4,=

Omax [l + C (-l)k 2exp(4k2L2/w2) erfc (2kL/w)J k=l

(2.17)


80

Figure 2.4 shows the ratio of OL/&~ = r& as a function of We can see that for w/L < 0.3, the infinite sample approximation is good. On the other hand, for w/L > 2 , the temperature approaches the values obtained from one-dimensional solutions indicated by the dashed line in Fig. 2.4.

w/L.

1.oc

'-ONE \

DIMENSIONAL APPROXI MATlON

-

0" Q- P/w w 3 1/e -points in intensity

0.1L BEAM RADIUS/SAMPLE THICKNESS, w / L

FIGURE 2.4. Dependence of the linear temperature 0 on the ratio w/L of the beam radius to the sample thickness. The value of 0 is normalized to the maximum value, , O, obtained for a semiinfinite sample. The dashed line is valid for the one-dimensional approximation. (After Ref. 2 . 9 ) 01982 Elsevier Science Publishing Co.

For intermediate wafer thickness, the correct temperatures can be obtained in two ways. If the deviation of the infinite sample approximation is significant (i.e., w/L = l), one has to OL is then obtained using first calculate ,,Q using E q . ( 2 . 1 3 ) . Eq. (2.17) or Fig. 2.4, and the true temperature is calculated using Eq. ( 2 . 7 ) . However, if the change in 0 due to the wafer thickness is small, a much simpler procedure can be used. Differentiation of E q s . (2.6,2.7) leads to the following expression for the change in the real temperature: (2.18) is the maximum real temperature achieved for infinite where,,T samples (see Fig. 2.1). [Note that temperatures in Eq. (2.18) The relative change in G, have to be expressed in Kelvin.] AO/Omax, is obtained from Fig. 2.4.

2. 2.3.4

TEMPERATURE DISTRIBUTIONS AND REACTION RATES

81

Effect of Absorption Coefficient

Lax [2.81 has shown that for a finite absorption coefficient ,,the maximum linear temperature is reduced. To account for this, , depends on we calculate the linear normalized temperature ‘Iwhich the dimensionless parameter W = aw: (2.19) where Omax (W=m) is given by Eq. (2.13). na can be expressed as 2

Lax i2.81 has shown that

(2.20)

Figure 2.5 gives the normalized linear temperature na as a function of W. The figure shows that the assumption of an infinite absorption coefficient is enerally good. For example, for the Ar+ laser, a is 1 . 2 ~ 1 0 4 cm-5 at 300°K for Si, and a typical value for w would be 20 um. This gives W = 24, and na = 0.97. However, if a Kr laser is used, the absorption coefficient is substantially lower, namely about 4000 cm-l at 300’K. The lower power output of that laser may necessitate the use of a smaller spot to obtain high enough temperatures, and the effect of the absorption coefficient may then have to be taken into account. The situation will be further complicated by the fact that the absorption coefficient also depends on temperature.

0.8

10

20

30

40

50

W:aw

FIGURE 2.5. Normalized linear surface temperature ‘I, = OmX(W)/ O,(W=m) at the beam center (x=y=z=O) as a function of the product, W, of the absorption coefficient, a, and the beam radius, w. [After Ref. 2.9.1

82

ART0 LIETOILA E 7 AL.

2.3.5

E f f e c t of Scanning Speed

C l i n e and Anthony [2.7] have shown t h a t t h e e f f e c t of t h e scanning speed on t h e l i n e a r temperature depends on t h e dimensionless parameter V:

v =

pcpwv 2

(2.21)

K(T~)

where v is t h e scanning speed. Again, t h e maximum l i n e a r temperaN i s s i m e t a l . 12.21 g i v e v a l u e s t u r e i s reduced by a f a c t o r , nv. of nv f o r a wide range of scanning speeds and f o r v a r i o u s a s p e c t r a t i o s 8 f o r e l l i p t i c a l beams ( s e e S e c t i o n 2.3.7). For low scanning speeds, nv can be approximated by t h e f o l l o w i n g expression:

nv

= 1.0

-

Figure 2.6 g i v e s

(2.22)

0.72 V

nv f o r large v a l u e s of V.

a W

a

I 06

a a w

2

5 0.4

0 W

1 '!

;; 0 2 z

a

0

z

"

1

3

2

4

v FIGURE 2.6. Normalized l i n e a r s u r f a c e temperature n = 6 ( V ) / Omax a t t h e beam c e n t e r (x=y=z=O) as a f u n c t i o n of t h e normalized scan speed V [ s e e Eq. (2.21)] f o r a c i r c u l a r beam (8=1) and e l l i p t i c a l beams of v a r i o u s a s p e c t r a t i o s 8 (see S e c t i o n 2.3.7). It should be emphasized t h a t normally, V 5 nm) can be obtained by Transmission Electron Microscopy (TEM). The depth distribution and electronic properties of point defects are detected by Deep Level Transient Spectroscopy (DLTS)

.

In this section, very brief reviews of VdP, SIMS and RBS measurement techniques are given. The DLTS technique and its applications are treated in Chapter 4 . With regard to TEM, we refer to standard textbooks; e.g., Ref. [3.10]. 3.2.1

Van der Pauw Measurements

Van der Pauw measurements of a doped layer can be performed in two ways. The sheet measurements give the true sheet resistivity of the layer and an estimate, Ns, for the active sheet carrier concentration; a weighed average mobility v, can then be calculated. Differential Van der Pauw measurements, in turn, give the carrier concentration and mobility as a function of depth. In the following, the two types of measurements are explained. The details of the anodic sectioning for stripping measurements are then given. Some of the error sources are also discussed. 3.2.1.1

Sheet Van der Pauw Measurements

.

The geometrical configuration for the measurements has been presented by Van der Pauw [3.11] These measurements give directly the sheet resistivity Rs and an estimated sheet carrier concentration Ns of the measured layer. The following formulas are valid for Rs and N, [3.12] :

0

ART0 LIETOILA AND JAMES F. GIBBONS

110

where d is the layer thickness, p is carrier concentration, and u mobility. The Hall scattering factor rH is defined as

where T is the carrier momentum relaxation time. electrons, rH = 1 [3.13].

For degenerate

The weighed average mobility, II, can be calculated from Rs and Ns: 1 (3.4)

!J=

4 Rs Ns

Use of formulas (3.1) and (3.2) yields

If the mobility were constant throughout the doped layer, the sheet measurements would give correct values for the active sheet concentration and mobility. However, the carrier mobility is a monotonically decreasing function of carrier concentration. In this case, the measured Ns will be lower than the real active dose. 'Ihe difference can be up to 30% for a heavy n-type dose, since p varies rapidly when the electron concentration is greater than 1019 cm-3. If the distribution of impurities is known, the above formulas (3.1, 3.2, 3.5) can be used to calculate the expected values for R,, u , and Ns. The measured values are then compared with the calculations, and if the agreement is good, the annealing is considered to be successful. This method is especially suitable for cw beam annealing, since the impurity redistribution is negligible during annealing, and the doping profiles can be accurately predicted. 3.2.1.2

Differential Van der Pauw Measurements

If information is needed about the carrier concentration and mobility vs. depth in the material, differential Van der Pauw measurements must be performed. In this method, thin sections of material are removed and the sheet measurements are performed after removal of each successive layer. This then allows the calculation of the sheet resistivity and sheet carrier concentration

3.

ION IMPLANTED CRYSTALLINE SILICON

111

(per unit area) in each layer. In other words, the limits of integrals in Eqs. (3.1, 3.2) are replaced by the limits of each layer. The weighed average mobility in each layer can also be calculated: the expression for it is Eq. (3.5) with appropriately modified integration limits. Providing that the thickness of each layer is known, the mean carrier concentration in it can be calculated, and concentration and mobility can be plotted as a function of depth. 3.2.1.3

Anodic Sectioning

The most widely used method to perform sectioning for differential Van der Pauw measurements employs anodic oxidation and oxide stripping. This is an elaborate procedure, but it is far superior to etching, since the thickness of the grown oxide can be measured after each step, allowing accurate determination of of the layer thicknesses. The standard anodization solution is prepared from [3.14] 1 gal. ethylene glycol, 15.4 g KNO3 (0.04 N solution), 90 ml H20 (2.5 vol - X ) , 6 g Al(N03)3*9H20. It is customary to illuminate the sample through the front surface during the oxidation. This is not essential for n-type layers on p-substrate. However, for p on n, the light is necessary to get enough current through the reverse-biased pn-junction (as the name of the method suggests, the Si substrate is positively biased in anodic oxidation). The thickness of the anodic oxide layer depends on the voltage prevailing across the oxide when this process is stopped, and calibration curves exist in literature [3.141. However, the main reason for using the anodization method is a desire to obtain accurate values for layer thicknesses. Therefore, the oxide thicknesses should be measured by ellipsometry or some other suitable method. Once the oxide thickness is known, it then has to be converted to the thickness of removed silicon. For thick oxide layers (> 1000 a), the ratio of silicon thickness to oxide thickness is expected to be about the same as for thick thermal oxide, namely 0.44. However, this ratio is larger when the oxide gets thinner. The ratio for about 45 nm thick oxide has been measured to be 0.56 f 0.01 [3.151. 3.2.1.4

Discussion of Error Sources

As was pointed out in Section 3.2.1.1., sheet Van der Pauw measurements give values that are systematically too low for the

112


sheet carrier concentration, unless the mobility is constant throughout the whole layer. This is true also for differential Van der Pauw measurements: this measurement yields the sheet concentration in each layer and the total active dose is obtained by summing up the sheet concentrations of each layer. However, since the individual layers are thin, the variation of mobility is minute within each of them. Therefore, the differential measurements yield a much more accurate value for the active dose. The inaccuracy in the determination of the thickness of the removed layers does not cause any error in the value of the total active dose, since the differential measurements give the total number of carriers in each layer. However, when the volume concentration versus depth is calculated, a possible error in the thickness measurement causes errors in the shape of the carrier profile. The Van der Pauw method gives correct results only if the four contacts are point contacts and are at the edge of the sample. To achieve this, various etched "Van der Pauw structures" have been proposed [3.16]. Such structures are unnecessary, however, if sufficiently large square samples can be used. The contacts are formed by placing tungsten probes directly onto the sample surface. l'hese probes can be regarded as point contacts, but of course they cannot be placed exactly at the sample edge. Probably the probes can be placed within 0.2 mm of the sample edge. Van der Pauw [3.11] has calculated that the relative error caused by this in sheet carrier concentration is about that fraction of the sample dimension by which the probes are separated from the edge. Thus if 5x5 mm samples are used, the error should be less than -5%. Additional uncertainty in differential Van der Pauw measurements results if only a small fraction ((5%) of the active carriers is removed with each step. [In this case, two numbers which are very close to each other will be subtracted.] This problem usually occurs for the first point of measurement because the surface concentration of typical implants is quite low unless a very low energy is used. However, from the second layer on, a sufficient fraction of the remaining dopants is usually removed at each step to avoid this type of error.

3.2.2.

S IMS Technique

The popularity of secondary ion mass spectrometry (SIMS) as a technique for detecting impurities in semiconductors stems from the fact that, under appropriate conditions, it has the sensitivity to measure depth profiles of impurities down to concentrations on the order of 1 x 1015 cm-3 I3.17-3.191. In this section we shall give a brief overview of the general technique and will

3.


113

show how SIMS may be applied to measure depth profiles of common impurities in semiconductors. 3.2.2.1

General Description

The traditional method of secondary ion mass spectrometry is schematically illustrated in Fig. 3.1 i3.201. A primary source of low energy ions is generated, which collide with the atoms of the surface of a specimen. This induces a beam of sputtered secondary ions which may then be mass analyzed. Monitoring the concentration of constituents in this secondary ion beam as a function of time gives a depth profile of each of the host and impurity atoms.

ION MICROPROBE 'ON

TO MASS SPECTROMETER/

FIGURE 3.1. Artist's conception of the secondary ion production process (after Ref. 3.20). The efficiency of ionization of the secondary ions determines the sensitivity of SIMS to these elements. There are two processes by which these atoms become ionized, (a) kinetic ionization and (b) chemical ionization. In the first of these, the incident ion must impart enough kinetic energy through electronic stopping losses to actually ionize the secondary sputtered ion. Chemical ionization is the process whereby the primary ions chemically ionize the surface atoms and is statistically the most favorable. It is therefore desirable to use a chemically reactive ion such as Oz, 0-, or Cs+ as the primary beam. Other elements such as Ar+ are used, but they tend to produce lower yields of either positive or negative secondary ions because they induce only kinetic ionization.

114 3.2.2.2


Depth P r o f i l i n g of I m p u r i t i e s

Ion implanted i m p u r i t i e s i n S i a r e u s u a l l y d i s t r i b u t e d a t depths l e s s than 300 nm, r e q u i r i n g a depth p r o f i l i n g technique w i t h a depth r e s o l u t i o n 5 nm. One method of a c h i e v i n g t h i s The primary i o n high depth r e s o l u t i o n i s i l l u s t r a t e d i n Fig. 3.2. beam i s r a s t e r scanned i n o r d e r t o c r e a t e a wide flat-bottomed c r a t e r . An e l e c t r o n i c g a t i n g technique i s a p p l i e d such t h a t t h e d e t e c t o r only a c c e p t s i o n s w h i l e t h e beam i s scanning i n t h e In t h i s way, only t h e f l a t c e n t e r r e g i o n of t h e c r a t e r bottom. p o r t i o n of t h e c r a t e r c o n t r i b u t e s t o t h e s i g n a l r e c e i v e d . N e u t r a l primary i o n s , however, can n o t be r a s t e r scanned and a r e t h e r e f o r e a b l e t o s p u t t e r any p o r t i o n of t h e specimen a t any t i m e . Elect r o n i c a p e r t u r i n g i s of course unable t o reduce t h e n o i s e produced by n e u t r a l i o n s s t r i k i n g t h e c r a t e r s i d e w a l l s and t h e sample s u r f a c e . This l i m i t s t h e r e s o l u t i o n and s e n s i t i v i t y of SIMS, p a r t i c u l a r l y f o r t h e c a s e of an i o n implanted p r o f i l e where t h e s i d e w a l l impurity c o n c e n t r a t i o n may be s e v e r a l o r d e r s of magnitude h i g h e r than t h a t on t h e c r a t e r f l o o r .

-i

BEAM

1-26v

i SAMPLE

I,

SAMPLE

I

I

CRATER PRODUCED BY RASTERED PRIMARY BEAM

FIGURE 3.2. I l l u s t r a t i o n of craters produced by a r a s t e r e d and u n r a s t e r e d beam ( a f t e r Ref. 3.60). F o r c e r t a i n atoms such a s s u l f u r , t h e r e i s a n o t h e r l i m i t a t i o n t o t h e u l t i m a t e system d e t e c t a b i l i t y [3.211. Diatomic oxygen h a s t h e same atomic mass number a s monatomic s u l f u r and i s t h e r e f o r e A high background of r e s i d u a l oxygen analyzed simultaneously. w i l l t h e r e f o r e produce a f a l s e b a s e l i n e on t h e SIMS output. E f f e c t s such a s t h i s due t o molecular "feedthrough" can u s u a l l y

3.


115

be minimized by improving the ultimate vacuum in the analysis chamber. An example of a SIMS output is given in Fig. 3.3. for a Te implant in GaAs. The horizontal axis represents the number of Tecounts in the detector during a detecting time-interval, and the horizontal axis is the sputtering time. This data is converted to impurity concentration vs. depth through suitable calibration procedures.

lx10’4 Te/cm2 120 keV

-0 o 1 SECONDS

FIGURE 3 . 3 . SIMS output from a Te implanted GaAs sample (after Ref. 3.60). The depth calibration is conveniently performed by measuring the depth of the sputtered crater by a step profiler. Since the total sputtering time is known, it is then easy to convert the sputtering time to depth. This procedure of course assumes that the sputtering rate is constant, which is usually the case for silicon. When converting counts/interval to impurity concentration, the background count rate is first subtracted. That rate is obtained from the end of the SIMS output. For example, in Fig. 3 . 3 , the background rate (for time > 200s) is about 0.5 countsfs. Final calibration is then performed by requiring that the integrated impurity dose be the same as the implanted dose. Another way to calibrate concentrations is to use known calibration samples, which should be analyzed in the same run together with the unknown samples.

116


Thus t h e SIMS technique cannot provide a b s o l u t e q u a n t i t a t i v e i n f o r m a t i o n about t h e number of i m p u r i t i e s i n t h e sample i n quest i o n . If t h a t i n f o r m a t i o n i s needed (e.g., i f t h e implanted dose i s n o t a c c u r a t e l y known), RBS a n a l y s i s must be performed. This method i s e x p l a i n e d i n t h e f o l l o w i n g s e c t i o n .

3.2.3

RBS and MeV Ion Channeling Analysis

A s w a s i n d i c a t e d above, t h e Rutherford b a c k s c a t t e r i n g and channeling techniques can be used t o determine a b s o l u t e numbers of i m p u r i t i e s i n semiconductors, t h e i r l a t t i c e l o c a t i o n s Unfortunately t h e and c r y s t a l l i n e q u a l i t y of t h e h o s t l a t t i c e . s t u d y of i m p u r i t i e s must, i n p r a c t i c e , be l i m i t e d t o atoms which In a r e h e a v i e r t h a n t h e h o s t atoms, a s w i l l be e x p l a i n e d below. t h e f o l l o w i n g w e w i l l g i v e a s h o r t d e s c r i p t i o n of t h i s technique i n g e n e r a l , and t h u s show how i t can be used f o r t h e t h r e e abovementioned a p p l i c a t i o n s .

This s e c t i o n i s aimed t o h e l p a r e a d e r u n f a m i l i a r w i t h t h i s method t o b e t t e r understand t h e r e s u l t s p r e s e n t e d l a t e r in t h i s c h a p t e r . Those wishing a n in-depth d e s c r i p t i o n of t h e backscatt e r i n g technique should c o n s u l t Ref. 3 . 2 2 .

3.2.3.1

General D e s c r i p t i o n

A schematic diagram of a t y p i c a l b a c k s c a t t e r i n g s p e c t r o A Van de Graaff a c c e l e r a t o r metry system i s g i v e n i n Fig. 3.4. Momentum g e n e r a t e s a m n o e n e r g e t i c MeV i o n beam of He i o n s . a n a l y s i s i s made by a magnet t o o b t a i n a beam of 4He n u c l e i a t a w e l l d e f i n e d energy. This beam i s then c o l l i m a t e d by a s e t of s l i t s , e n t e r s t h e t a r g e t chamber and impinges upon t h e sample t o be analyzed. The i n c i d e n t p a r t i c l e s c o l l i d e w i t h t h e l a t t i c e and impurity atoms of t h e sample and some of them are b a c k s c a t t e r e d from t h e near s u r f a c e r e g i o n through a s i n g l e e l a s t i c c o l l i s i o n . A s o l i d s t a t e d e t e c t o r i s used t o c o l l e c t and a n a l y z e t h e energy d i s t r i b u t i o n of b a c k s c a t t e r e d p a r t i c l e s f o r a given d i r e c t i o n . For a t a r g e t s u r f a c e p e r p e n d i c u l a r t o t h e i n c i d e n t beam, any d i r e c t i o n l o c a t e d a t an angle g r e a t e r than 90" from t h e incoming beam can be s e l e c t e d t o c o l l e c t t h e b a c k s c a t t e r e d 4He p a r t i c l e s . A common choice i s t o p l a c e t h e d e t e c t o r a t 170' where t h e backs c a t t e r e d 4He p a r t i c l e s can be considered as having passed through t h e same d e p t h of analyzed m a t e r i a l on t h e i r way i n and o u t of t h e sample. D e t e c t o r s may a l s o be placed a t g r a z i n g a n g l e s when b e t t e r depth r e s o l u t i o n i s needed i 3 . 2 3 1 . The d e t e c t o r i s t y p i c a l l y d e v i c e which i s n o t s e n s i t i v e t o p a r t i c l e . The e l e c t r i c a l s i g n a l detector is proportional t o the

a solid s t a t e surface barrier t h e charge s t a t e of t h e c o l l e c t e d generated i n t h e s u r f a c e b a r r i e r i n c i d e n t p a r t i c l e energy. This

IMASS SELECTION VAN DE GRAAFF ACCELERATOR (MODEL K )

3 MeV MFLECTION

MULTI- CHANNEL ANALYSER

U

BY'\'. 9

'+' I

GRAZING DETECTOR ( ~ - 8 3 0 )

/ /

SCHEMATIC DIAGRAM OF THE BACKSCATTERING SPECTROMETRY SYSTEM

11s

A R T 0 LIETOILA AND JAMES F. GIBBONS

signal is amplified at several stages and then stored in a multichannel analyzer. The final form of the data shows the number of backscattered events per channel as a function of energy. This spectrum can then be displayed on a plotter or printer. The sample to be analyzed is held on a precision goniometer (typically 0.01" accuracy). Several constructions exist, with the typical system providing the following three degrees of freedom: a 360" rotation perpendicular to the incident beam, -+ 90' sample tilt from its normal to the beam direction, and vertical translation. 3.2.3.2

Basic Physical Process: Shape of a Spectrum

A s indicated above, the output of anRBS system is the number of backscattering events vs energy of the backscattered particles. The spectrum is digitized in the multi-channel analyzer to give the number of particles in each channel (typically 3-5 keV wide).

There are three factors governing the shape of an RBS spectrum for a given incident particle energy and flux. 'he kinematic factor, €$,, determines the fraction of energy the projectile retains in the elastic collision. The probability of collision yielding backscattering to the detector angle is proportional to the scattering cross section, u. Finally, the projectile loses energy when it travels inside the target material; this energy loss is described by the stopping cross section, E. The kinematic factor depends on the backscattering angle and the mass ratio of the projectile and the target atom. For a constant projectile mass, the kinematic factor increases with increasing target mass. This is why trace impurities in Si can only be detected if their mass is larger than that of Si; in this case the signal from the impurities will be well separated from the silicon signal. The scattering cross section for 4He particles is proportional to the square of the target atom number. This is another reason why heavier impurities are more easily detected. The stopping cross section, which determines how much energy the projectile loses while traveling in the target material, increases when the projectile energy decreases. This causes nonlinearity in depth vs energy calibration. (The concentration of implanted impurities in Si is always so low that the energy loss of the projectile is determined by Si material only.) An example of the backscattering spectrum of 2.0 MeV 4He+ The sample analyzed was crystalions is given in Fig. 3.5 L3.241. line Si implanted with a nominal dose of 1.2 x 1015 As atomsfcm2

3.


119

at 250 keV. The plot gives, point by point, the number of counts in each channel (5.0 keV channel width). The silicon signal is a step with leading edge at 1.13 MeV. The particles scattered from inside the sample have a lower energy since energy is lost in going through the material. The As signal (plotted on amplified scale) appears at higher energies (1.5-1.6 MeV), since the kinematic factor is higher for As (mass = 75 amu) than for Si (mass = 28 amu). The As-signal is relatively higher than the Si-signal because the scattering cross-section is about a factor of 6 higher for As.

Energy ( M e V )

FIGURE 3.5. Ener spectrum of 2 MeV 4He+ ions backscattered from Si implanted to aY5As dose of 1 . 2 ~ 1 0c 2~ ~ at 250 keV. The ' m vertical arrows indicate the energies of particles scattered from surface atoms of 28Si and 75As (from Ref. 3.24). 3.2.3.3

Total Number of Impurities

The total number of impurity atoms in the Si sample (i.e., the implantation dose) can be determined directly by RBS techniques. The impurity dose, Ni, is given by (Ref. 3.22, p.139):

where Ai is the total number of counts (area) of the impurity signal and HSi is the height of the Si-signal at the edge; &is the width of a channel in the multi-channel analyzer; osi and oi are

120

ART0 LIETOILA AND JAMES E GIBBONS

the scattering cross sections for Si and the impurity atoms, respectively; and [ E ~ I S is ~ the stopping cross section factor for the He particles in Si. HSi is a machine parameter, and us$, ui, and [ E ~ ] Sare ~ tabulated in Ref. 3.22. The RBS technique thus yields an absolute value for Ni. Unfortunately, this method is applicable only to impurities of higher mass than Si. 3.2.3.4

Channeling Technique: Thickness of Amorphous Layer, Crystalline Quality and Lattice Location of Impurity At oms

The high resolution goniometers of a backscattering system can be used to align the principal crystalline orientation of the sample with the direction of the incident 4He+ beam. If the sample has good crystalline quality, the yield will drop to 3-42 of the random value, since most of the host atoms lie in rows that terminate in a surface atom, and they are therefore hidden from the beam by it. This feature can be used to study the crystalline quality of the sample. An example of a series of channeling spectra is given in Fig. 3 . 6 . The Si sample was first amorphized with implantation and then subjected to low temperature thermal annealing for increasing times. The annealing caused the amorphous layer to regrow via solid phase epitaxy (SPE; see Section 3.3.2 below). The spectrum for the as-implanted sample has a high yield at energies from 1.13 MeV down to about 0.93 MeV, corresponding to particles scattered from the amorphous layer. Particles which go through the amorphous layer to the underlying crystalline material will partly channel, and the yield is markedly lowered. However, low angle scattering in the amorphous layer causes the beam to lose coherence, and the yield is higher than for a completely crystalline sample. The energy scale of the particles scattered from silicon can be translated to a depth scale (Ref. [ 3 . 2 2 1 , p. 59), as is shown in Fig. 3.6. Thus the thickness of the amorphous layer can be determined. Figure 3.6 shows the trailing edge of the amorphous Si signal moving to higher energies with increasing annealing time, corresponding to movement of the crystalline-amorphous interface closer to sample surface. Simultaneously, the yield from the underlying crystalline material is decreased because a thinner amorphous layer causes less incoherence in the helium beam. Figure 3.6 shows that the crystalline structure of the sample was completely restored after 60 minutes of annealing. The quality of the crystal can now be judged by comparing the

3.


121

yield of the crystalline sample (taken just behind the surface peak) to the amorphous yield. This ratio is called the minimum yield, xmin. For a perfect Si sample of orientation, )cdn = 3.4%. The sample illustrated in Fig. 3.6 has a minimum yield of 6%; this indicates that some volume damage exists in the sample. However, a part of the high yield is also due to the high surface peak characteristic of a damaged surface layer in regrown samples. 4000 I

Omin.'

2000

I

I

I

depth

1

I5minx 30mina 45min" 0

.-......

x x

60min.

(A)

-

%

.&

*-wo 09

I

I

10

1.1

1.2

Energy (MeV)

FIGURE 3.6. Aligned spectra of 2 MeV 4He ions for Si. The sample was self implanted to a total dose of 8x1015 cm-2 at multiple energies ranging from 50 to 250 keV, and then annealed at 5OO0C for different times (0-45 min) (from Ref. 3.25). Finally, if the sample were implanted with impurities, such as As, their lattice location can be determined by the channeling technique. If the impurities are substitutional in the crystal, they are "hidden" behind host atoms, and little or no signal i s then obtained from the impurities. It is customary to calculate the ratio of the total impurity yield (signal area) to the implanted dose. If all impurities are substitutional, that ratio has the value of x d n (Ref. 3.22, p.269); anything above that indicates presence of interstitial impurities.

122 3.3 3.3.1


CW LASER INDUCED CRYSTALLIZATION OF AMORPHOUS SILICON Introduction

It has been well established that when implantation amorphized silicon is furnace annealed at low temperatures (4006OO0C), the annealing mechanism is solid phase epitaxy (SPE) which proceeds from the amorphous crystalline interface towards the sample surface f3.251. However, simultaneously with the epitaxial regrowth, nucleation and growth of polycrystallites can take place in the amorphous layer [3.26]. When the polysilicon formation has proceeded far enough, the epitaxial regrowth will stop. Thus there is a limit on the thickness of the amorphous layer that can be regrown to single crystal silicon [3.27,3.28]. Since the early days of laser processing, the experiments have suggested that the mechanism of cw laser annealing of amorphous layers is SPE. However, due to the extreme rapidity of the phenomenon (annealing typically takes place in milliseconds or less), a clear proof for this has only recently been developed. Olson et al. [3.29] were able to observe in situ the cw laser induced regrowth of amorphous silicon. Their measurement technique utilizes interference effects within the amorphous layer, which has different optical properties than the underlying crystalline material. The regrowth can be observed by monitoring the reflectivity of the sample which is being annealed with a stationary beam. In addition to SPE, the nucleation and growth of polycrystalline silicon has also been observed in laser processed amorphous layers. Gat [3.27] has shown that an attempt to anneal a deep 4000 a thick) self implantation amorphized layer with a Kr+ cw laser leads always to formation of a polycrystalline layer (of unknown thickness) on top of the sample. Hess et al. i3.281 have found that while thin ( - 1200 a thick) UW-evaporated amorphous layers could be epitaxially regrown (in solid phase), 5000 a thick layers showed development of polycrystalline material. Little further characterization of the polycrystalline layer was performed in either of the above experiments. (13

Studies on the cw laser induced regrowth of amorphous silicon are the subject of this section. First we will give a short review of the kinetics of SPE as observed in low temperature furnace processing. m e n we report on the observation of SPE and measurement of regrowth rates under typical laser annealing conditions. Finally, we give results on regrowth and nucleation of polycrystalline silicon in deep amorphous layers.

3. 3.3.2


123

Low Temperature Regrowth Kinetics

The kinetics of SPE in the temperature range of 450-600°C have been extensively studied by Csepregi and coworkers [3.25, 3.30, 3.311. The most immportant features are as follows. The regrowth proceeds at a constant rate from the amorphouscrystalline interface towards the sample surface for all crystalline orientations except . The latter exhibits two distinct growth velocities, the first 1000 A growing considerably more slowly than the rest of the layer t3.311. The growth rate vg is thermally activated, obeying the formula vg

=

vgo exp (-Ea/kT)

(3.7)

The activation energy, Ea, was measured to be, for all crystal orientations, 13.311: Ea

=

2.35

f

0.1 eV

(3-8)

The pre-exponential terms for the most interesting crystal orientations are 13.311:

f

= 1.4

1014 A i s

a10>

=

0.2

1014 a i s

a i i > , slow

=

0.35x 1014

ais

a i l > , fast

(3.9)

A recent measurement on Si gave the values of 2.85 0.1 eV for Ea and 3.68~10~’A/s for vgo.

The growth rate can be greatly enhanced by doping the amorphous layer with electrically active impurities, such as As, B, or P [3.30]. The n-type dopants do not change the activation energy, whereas B, while enhancing the regrowth still more than P and A s , appears to decrease the activation energy. Table 3.1 gives the enhancement of the regrowth caused by P-doping at different concentrations. 3.3.3

Observation of cw Laser Induced SPE

Williams et al. [3.2] were the first to find evidence that the mechanism of cw beam induced crystallization of amorphous Si is SPE. These authors used glancing angle ion channeling measurements to measure amorphous layer thickness in 75As-implanted and cw laser annealed silicon samples. They found that under suitable


124

c o n d i t i o n s , t h e amorphous-crystalline i n t e r f a c e had moved a s h o r t d i s t a n c e towards t h e sample s u r f a c e , w h i l e a h i g h e r l a s e r power r e s u l t e d i n complete regrowth of t h e amorphous l a y e r ; t h i s beh a v i o r i s ' i l l u s t r a t e d i n Fig. 3.7. TABLE 3.1. The E f f e c t of Phosphorus Doping C o n c e n t r a t i o n on t h e Rate of S o l i d Phase E p i t a x i a l Regrowth of S i l i c o n a t 475°C ( a f t e r Ref. 3.30).

Enhancement factor

Regrowth rate (Urnin)

Doping concentration ( 1020 0 - 3

-

3.25

0.93

8.3

1.10 1.60 2.20 2.80

9.1 12.5 1 7 .O 19.3

j

30keV As-SI

1

DOSE 3 r 1 0 ' 5 ~ sCm-*

(100)

1 9 MeV He'

CWARGON LASER ANNEALED

i w 2

2.6 2.8 3.8 5.2 5.9

.

('gp)

+!z-

.

SOLID STATE DETECTOR

( A ! St REGROWTt

280

300

320

340

360

380

CHANNEL NO ( 4 . 2 keV/CHANNEL!

FIGURE 3.7. Glancing a n g l e RBS random and a l i g n e d s p e c t r a f o r s i l i c o n which w a s implanted w i t h A s at 30 keV and annealed w i t h a s c a n n i n g A r + laser. The a l i g n e d s p e c t r a a r e f o r ( a ) asimplanted, (b) p a r t i a l l y a n n e a l e d , and ( c ) f u l l y annealed samples (from Ref. 3 . 2 ) .

3.


125

While Williams e t a l . 13.21 found only one p o i n t of i n t e r mediate s t a g e i n t h e regrowth p r o c e s s , t h e technique introduced by Olson e t a l . [3.4] allowed i n s i t u o b s e r v a t i o n of t h e growth of t h e e n t i r e l a y e r . T h e i r technique i s based on t h e i n t e r f e r e n c e e f f e c t s w i t h i n t h e amorphous l a y e r which has d i f f e r e n t o p t i c a l p r o p e r t i e s t h a n t h e underlying c r y s t a l l i n e material. That i n t e r f e r e n c e makes t h e sample r e f l e c t i v i t y depend p e r i o d i c a l l y on t h e t h i c k n e s s of t h e amorphous l a y e r , as i l l u s t r a t e d i n F i g . 3.8. Thus, by monitoring t h e r e f l e c t i v i t y of a sample being annealed w i t h a s t a t i o n a r y laser beam, i t i s p o s s i b l e t o observe t h e regrowth. Figure 3.9 shows an example of a r e f l e c t i v i t y vs t i m e p l o t measured by Olson e t a l . 13.41, which c l e a r l y demonstrates t h a t t h e regrowth is proceeding from t h e o r i g i n a l i n t e r f a c e towards t h e sample s u r f a c e .

0.490 0.470 I>

-5

0.450

Iu

0.430 w

a 0.410 0.390 0.370 L 4QM)

1

3000

2000

loo0

0

AMORPHOUS FILM THICKNESS, A

FIGURE 3.8. Calculated r e f l e c t i v i t y ( a t 0.6328 um) of amorphous Si f i l m on c r y s t a l l i n e S i s u b s t r a t e a s a f u n c t i o n of f i l m t h i c k n e s s . (Note: Thickness scale is reversed t o s i m p l i f y comparison w i t h experimental d a t a , F i g . 3.9 .) The r e f r a c t i v e f o r amorphous S i i n d i c e s used i n c a l c u l a t i o n s are 4.85-0.6121 01981 Elsevier and 4.16-0.0181 f o r c r y s t a l l i n e S i (from Ref. 3 A ) . Science Publishing Co.

3.3.4

Rate of CW Laser Induced Regrowth

L i e t o i l a e t a l . 13.321 have measured t h e e p i t a x i a l regrowth rates under a c t u a l cw laser a n n e a l i n g c o n d i t i o n s , u s i n g a scanning beam and l a s e r powers t h a t y i e l d a maximum s u r f a c e temperature c l o s e t o 1000°C. This was accomplished by determining t h e s c a n speed of a s i n g l e l a s e r s c a n r e q u i r e d t o regrow a known amorphous l a y e r a t t h e c e n t e r of t h e beam. The regrowth r a t e s can be d e t e r mined by u s i n g t h e formalism p r e s e n t e d i n Chapter 2 which g i v e s t h e e f f e c t i v e dwell t i m e f o r a scanning Gaussian beam. S p e c i a l


126

attention was paid to the determination of the annealing temperature: one of the precautions was to leave the laser power unchanged after melting power calibration and regulate the annealing temperature by changing the sample holder temperature. This experiment is described below. 0.49

7-rc '

0.47

- 0.45 1 $ w

0.43

3

U

0.41 0.39 0.37

o,35i

I

i

0

_10

i . . 5

~

-.._

15

-

20

d i

A 25

TIME, sec

FIGURE 3.9. The reflectivity of a HeNe probe vs time for a Si sample which is being annealed with a stationary Ar' laser turned on at t = 0). The sample is Si implanted to a j 5 A s dose of 5x1014 at 158keV. The laser power is 7.5W and substrate temperature To = 460°C (from Ref. 3.4). 9981 Elsevier Science Publishing Co.

3.3.4.1

Sample Preparation

The samples were of polished Si. The amorphous layer was formed by self implantation to avoid impurity effects. The implantation energy and dose were 40 keV and l ~ l O ~ ~ c r nrespec-~, tively. The samples were held at LN2 temperature during implantation to prevent any self annealing. The thickness of the as implanted amorphous layer was measured by using 2.2 MeV4He+ channeling techniques (see Section 3 . 2 ) , and was found to be 960 8 . Prior t o laser annealing,the samples were cleaned in methanol and D.I. water, the purpose of which is mainly to remove major dust particles. Annealing was performed in the laboratory atmosphere.

3. 3.3.4.2


127

Measurement of the Laser Beam Diameter

The cw laser scanning system described in Chapter I was used with a 136 mm focusing lens. The focussed beam radius was measured by determining the power required to reach surface melting of Si with the substrate temperature To = 332°C. A scanning speed of 6 cm/s was used, which is slow enough to allow the steady state temperature calculations (Chapter 11) to be used. The effect of scanning speed on temperature will be discussed later. The incident power (discounting the losses in the beam shaping devices) required to cause surface melting of crystalline Si was found to be 5.0 W. The parameter pa [see Eqs. (2.13,2.14)] has to be 1210 W/cm to reach the maximum temperature of 1412OC for To = 332°C [Eqs. (2.7,2.13)]. 'Ihe reflectivity of crystalline Si was measured to be 38% for the Ar+ laser radiation; thus w = 26 pm is obtained. We would like to stress that the beam radius determined above is used only to compute the dwell time of the spot. The calculation of the annealing temperature is based solely on the calibration against the melting power. 3.3.4.3

Procedure to Measure the Regrowth Rates

After the laser power required to reach melting was determined and the beam radius was calculated, the laser was kept at that power. The annealing temperature was then controlled by changing the substrate temperature, To. This was done because the beam radius i s known to vary with laser power and a constant beam radius is essential for accurate determination of the annealing temperature. For each substrate temperature To, two scanning speeds, v1 and v2, were determined so that the sample region underneath the center of the beam was just regrown at vl, whereas v2 was too fast to allow full regrowth. The results of the scan were judged optically: the regrown area appears dark against the lighter amorphous background [3.3]. The values of To as well as v1 and v2 are given in the first two columns of Table 3.2. 3.3.4.4

Analysis of Results

The peak annealing temperatures for each substrate temperature, To, were obtained using the calculation presented in Chapter 11. The scan speeds were quite low suggesting the use of Eqs. (2.7,2.13 ) derived for a stationary beam. However, first order corrections to the temperature due to scan speed were made by

128


c a l c u l a t i n g t h e l i n e a r temperature 0 f o r t h e scanning beam [Eqs. (2.21,2.12)] and applying t h e i n v e r s e Kirchhoff t r a n s f o r m [Eq. (2.7)1 * The Kirchhoff transform i s not q u i t e e x a c t l y v a l i d i f V ( a parameter r e l a t e d t o t h e scanning v e l o c i t y : see Chapter 11) d e v i a t e s from z e r o . However, V g e t s q u i t e small i n t h i s case: t h e change i n t h e temperature due t o scanning w i l l be a t most about 20°C, s o t h i s i s not a c r i t i c a l p o i n t . The peak a n n e a l i n g temperatures are now determined a s f o l lows. Since t h e melt temperature c a l i b r a t i o n w a s performed on c r y s t a l l i n e s i l i c o n , w h i l e t h e a c t u a l samples were amorphous, t h e m e l t i n g v a l u e , 1210 W/cm, f o r p has t o be m u l t i p l i e d by t h e r a t i o of a b s o r p t i v i t i e s (1-R) of amorphous and c r y s t a l l i n e s i l i con. The r e f l e c t i v i t y of t h e samples used h e r e w a s measured t o be 4 5 % , s o t h e e f f e c t i v e p i s 1073 W/cm f o r t h e s e samples. For each s u b s t r a t e temperature and scanning speed, t h e parameter V i s c a l c u l a t e d u s i n g Eq (2.21). Equation (2.22) g i v e s t h e c o e f f i c i e n t s n , which determine how much t h e l i n e a r maximum temper,, i s reduced from t h e value f o r a s t a t i o n a r y b e a m . ature @ Equation (2.13), with p = 1073 W/cm, is used t o c a l c u l a t e t h e Om, stat i s t h e n d i v i d e d by t h e v a l u e of s t a t i o n a r y OmX,Stat. n W l t f o r t h e m e l t i n g c a l i b r a t i o n scan (.986, see Table 3 J), and m u l t i p l i e d by t h e corresponding value of n f o r each s c a n . The r e s u l t i n g v a l u e of 0 i s f i n a l l y converted t o r e a l temperature The r e s u l t s are g i v e n i n Table 3.2. using Eq. ( 2 . 7 ) .

.

Next t h e e f f e c t i v e dwell times are determined u s i n g t h e f o r m a l i s m p r e s e n t e d i n S e c t i o n 2.5. For any thermally a c t i v a t e d process t h e e f f e c t of a s i n g l e laser s c a n a t t h e c e n t e r of t h e scan l i n e can be d e s c r i b e d u s i n g a c o n s t a n t temperature, equiv a l e n t t o t h e peak temperature, Tmx, and a reduced dwell t i m e , teff : ( 3 .lo) The dwell time r e d u c t i o n f a c t o r f w a s shown i n S e c t i o n 2.5 t o depend mainly, but s t i l l n o t s t r o n g l y , on t h e s u b s t r a t e temperat u r e and t h e a c t i v a t i o n energy of t h e p r o c e s s . Values of f a r e given i n F i g . 2.14. I n t h i s case, where t h e s u b s t r a t e temperaW e used t h e curve f o r E, = t u r e s a r e around 200°C, f = 0.26. 2.35 e V i n F i g . 2.14; t h i s c h o i c e is n o t c r i t i c a l , however, s i n c e The e f f e c t i v e dwell-times c a l c u l a t e d f depends only weakly on Ea. f o r each s c a n a r e given i n Table 3.2. The l a s t s t e p i n t h e d a t a r e d u c t i o n i s t h e c a l c u l a t i o n of t h e regrowth r a t e s u s i n g t h e known t h i c k n e s s of t h e amorphous l a y e r and t h e e f f e c t i v e dwell t i m e s . For each s u b s t r a t e temperat u r e we o b t a i n a lower and a n upper l i m i t f o r t h e regrowth rate,

TABLE 3.2.

a)

Results of the Study of the CW Laser Induced SPE Regrowth of Si

174

1.5 2.3

0.86

0.003 0.005

0.998 0.996

357.0 356.3

797 795

0.88 0.57

1.1 1.7

184

1.5 3.0

0.84

0.003 0.007

0.998 0.995

367.3 366.2

825 822

0.88 0.44

1.1 2.2

193

3.0 4.5

0.81

0.007 0.010

0.995 0.993

375.3 374.6

847 845

0.44 0.29

2.2 3.3

202

3.0 6.0

0.80

0.007 0.014

0.995 0.990

384.6 382.6

872 866

0.44 0.22

2.2 4.4

211

6.0 9.0

0.78

0.014 0.021

0.990 0.985

391.8 389.6

89 1 886

0.22 0.15

4.4 6.4

332

6.0

0.59

0.019

0.986

578.1

1412

---

---

v1 is the velocity at which complete regrowth was observed at the center of the beam; for v2, the crystalline interface did not reach the sample surface.

130


corresponding to scan velocities v1 and v2, respectively. The lower and upper limits for the regrowth rate vs temperature are given in Table 3.2 and in Fig. 3.10. In the figure a bar is drawn to connect the limits obtained at each substrate temperature. For comparison, Fig. 310 also shows the low temperature data given by Csepregi et al. (3.25, 3.30, 3.311 and in Ref. 3.59. Figure 3.10 shows that the regrowth rates induced by laser annealing are one to two orders of magnitude higher than the rates extrapolated from furnace annealing data. This difference is too large to be explained by the uncertainties in the observed regrowth velocities (a factor of 1.5 - 2 ) or in the calculated temperatures (-- 2 20°C). (The observed data are, however, too scattered to make any firm conclusions regarding the activation energy.) Thus, these measurements would indicate that the epitaxial regrowth is enhanced by the laser radiation. It should be emphasized however, that the extrapolation of growth rates from the low temperature data (see Fig. 3.10) is subject to considerable uncertainty, since the measurements performed in Refs. 3.25, 3.30, 3.31, and 3.59 were limited to a narrow temperature range. 3.3.5

Obstructed Regrowth of Thick Amorphous Layers

Hess et al. [3.28] have observed that under their experimental conditions, UHV deposited 0.5 um thick amorphous Si layers could not be regrown with a scanning cw laser. 'Ihis was shown to be a problem stemming from the layer thickness rather than layer preparation, since 0.1 um thick, identically deposited layers grew fully with laser scanning. Gat also observed 13.271 that implantation amorphized, 0.4 pm thick Si layers did not fully regrow but turned at least partly polycrystalline. Hess et al. [3.28] postulated that spontaneous nucleation and growth of polycrystalline silicon would stop the epitaxial regrowth in thick layers. They suggested that doping of the amorphous material would make it possible to regrow thicker layers, since doping is known to enhance the epitaxial regrowth rate [3.30]. The latter hypothesis is very speculative: the growth of polycrystalline silicon may also be enhanced by doping. Also, it is not known whether the nucleation of polycrystallites originates at the sample surface or in the bulk of the amorphous material. In the former case, the ratio of the epitaxially regrown and polycrystalline layer thicknesses should be about the same if the thickness of the original amorphous layer is varied. But if the nucleation takes place in the bulk, the regrown thickness should stay constant, and the thickness of the

3


TEMPERATURE ("C) 700 600

1000 900 800 I

I

I

llkT

131

500

(d')

FIGURE 3.10. The rate of cw laser induced solid phase epitaxial regrowth in (100) Si (after Ref. 3.32). For comparison, the low temperature growth data given in Refs. 3.25 and 3.59 are also given.

remaining polycrystalline layer should vary with the original amorphous layer thickness. Lietoila 13-33] has conducted a study which sheds some light on the above questions. Different amorphous layer thicknesses, doped and undoped, were used to determine where the spontaneous nucleation originates and the effect of doping. Results of this study are reported in the following.

3.3.5.1

Sample Preparation, Experimental Techniques

Silicon wafers of orientation were amorphized by ion implantation. Three different implantation schedules were carried


132

out to form two different thicknesses of undoped amorphous silicon and one P-doped layer. The implantation schedules are given in Table 3.3. The wafers were held at LN2 temperature during implantation; and all implants were performed in the same implanter. The laser annealing was performed using the system described in Chapter I with a 160 mm focusing lens. 'he substrate temperature was held constant, To = 332OC, and the processing temperature was controlled by changing the laser power. The beam radius for the different laser powers was determined by measuring the power required to melt the sample surface for elevated substrate temperatures (up to To = 530°C). This procedure is described in Section 3.3.4.2. and it was carried out to make the determination of the annealing temperature more accurate. The scan rate was kept constant, v = 2.7 cm/s, and the distance between adjacent scan lines was 10 um in all cases.

"he thicknesses of the amorphous layers were measured by 2.2 MeV 4He+ channeling techniques as described in Section 3.2. For calculation of the layer thicknesses, the symmetrical mean energy approximation was used. 3.3.5.2

Undoped Amorphous Layers

The effect of laser power on the regrowth thickness was studied using samples subjected to the implantation schedule I (see Table 3.3). nose implants result in a continuous amorphous layer of 3740 8, as measured by channeling; the channeling spectrum is given in Fig. 3.11. TABLE 3.3. Implantation Schedules for the Study of Laser Induced Epitaxial Regrowth. All implants were performed at LN2 temperature (after Ref. 3.32). ~

Schedule

~

Ion

Energy (keV)

Dose (cm2)

I

si

25 50 90 170

3.0~10~~ 6.0~1014 1.1~1015 2.8~1015

I1

si

100

l.OXlOl6

25

3.0~1014 6.0~1014 1 .iX1015 2.8~1015

I11

P

50

90 170

3.


133

The laser powers used, the resulting maximum annealing temperatures, and results of regrowth are given in Table 3.6. The channeling spectra for a few cases are also given in Fig. 3.11. The fact that the channdJing yield behind the random peak drops clearly as the amorphous layer gets thinner means that the regrown layer has good crystalline quality. DEPTH ( p m )

0.8

0.4

1

0

~

SCHEDULE

1

~

1

I Tmax

990OC 117OoC 1320O C

0.8

1.0

1.2

1.4

6

ENERGY (MeV)

FIGURE 3.11. Channeling spectra for Si samples implanted according to Schedule I (see Table 3.3) and laser annealed at powers resulting in different maximum temperatures TmX. The solid line is the as-implanted spectrum (from Ref. 3.33).

In Table 3.4 are also included the calculated maximum thickness of regrowth, obtained using the extrapolated low temperature regrowth kinetics measured by Csepregi et al. [3.31] The thickness was calculated at the beam center, including the contributions of the two adjacent scans; the same dwell time, 0.6 m s , was assumed for each. This is not quite correct, since the contributions of the adjacent scans come from slightly off the beam center. However, since the overlapping is tight (1. 83%), the error is not significant.

.

Table 3.4 shows that throughout the power range used, the thickness of regrowth is almost constant. The thickness of the remaining polysilicon layer decreases slightly with increasing laser power, but even for the highest power, more than 800 a of Si has turned polycrystalline. That power resulted in a heavy network of sliplines on the sample surface, which is why no higher power was employed.


134

TABLE 3.4. Results of Laser Annealing of the Deep Siimplanted Amorphous Layer (Schedule I) (after Ref. 3.33).

Amorphous/ PolySi Layer(A)

-

-

0.65 0.71 0.78 0.84 0.91

990°C 1070°C 1170°C 1240°C 1320°C

Regrown Layer(A)

-

3740 1440 1300 1230 1120 830

2300 2440 2510 2620 2910

Calculated Regrowth(A) Cspregi et al. [3.25,3.31] Ref. 13.391

-

180

680 2890 6430 15700

1810 9000 5 .2x104 1.4~105 4.1~105

The question as to whether the formation of polycrystalline silicon originates at the sample surface or in the bulk of the amorphous material was studied by laser scanning wafers implanted according to Schedule I1 (see Table 3.3). This implant results in a continuous amorphous layer which is only 2570 a thick or about the same as the regrowth for the above samples of Schedule 1. If the spontaneous crystallization originated at the sample surface, one would expect the ratio of the random and regrown layers to be the same as for the above thicker amorphous layers. But, if the nucleation started homogeneously in the bulk of the amorphized material, the regrowth of the thinner amorphous layers (Schedule 11) should be complete. Table 3.5, and Fig. 3.12, show that the regrowth of the samples, which originally had 2570 a of amorphous material, is almost complete. The remaining layer is too thin to be measured by channeling, since the FWHM of the surface peak is about the same as the system resolution (7 channels or 25 keV). However, we can say that it is at most 200 A , most likely less. TABLE 3.5. Channeling Results of As-implanted and Laser Annealed Wafers with the Shallower Si-implanted Amorphous Layer Schedule (11). Results for a virgin sample are included for comparison (after Ref. 3.33). P/Pme1t

TmaX

- Virgin Sample 0.78 0.84

1170°C 1240°C

Amorphous/ Polysilicon Layer(A)

-

2570 S i h a s regrown, whereas t h e maximum t h i c k n e s s of regrowth f o r < l o o > S i w a s shown i n S e c t i o n 3.3.5.2 t o be almost 3000 8. The r e s u l t s p r e s e n t e d above c l e a r l y i n d i c a t e t h a t , i n p r a c t i c a l cases, only S i can be s u c c e s s f u l l y annealed w i t h a cw l a s e r , i f t h e implant dose i s heavy enough t o c r e a t e an amorphous s u r f a c e l a y e r . But i f t h e dose i s so l i g h t t h a t no amorphous l a y e r is formed, both and (111) Si can be cw l a s e r annealed.

3.4.3

Use of E l e c t r o n Beam Sources

Two t y p e s of cw e-beam s o u r c e s have been used f o r anneali n g purposes, namely an e-beam welder [3.35] and a modified SEM ( 3 . 3 6 , 3.371. The d i s t i n c t i v e d i f f e r e n c e between t h e s e two

3.


143

systems is in the operating power: the welder is designed to yield currents high enough to melt sizable metal structures, whereas a SEM operates at very low currents. Thus, the beam diameter of the welder is typically about 0.1 nun or more, whereas the SEM beam has to be focused down to a few pm. The tight focusing in the SEM makes it possible to selectively anneal very small structures. In what follows, annealing results obtained with the two types of e-beam sources are reported. 3.4.3.1

The E-beam Welder

Regolini et al., [3.35] have used an e-beam welder [Hamilton Model EBW (7.5)] to anneal ion implanted Si. The samples were p-type Si, which were 75As implanted at 100 keV to dose of 1.5~105 cm-2. The wafers were held at LN2 temperature during implants. Typical e-beam parameters used for annealing were 30 keV at 0.5 mA with the beam focused to a spot approximately 300 pm in diameter. The scanning speed was 2.5 cmls, and the temperature of the sample holder remained below 5OoC during the anneal. Control samples were thermally annealed at 575OC for 30 min. Results of channeling studies for as-implanted and annealed samples are given in Fig. 3.18. The as-implanted spectrum was used to determine the implanted dose (1.5~1015 cm-2) and the thickness of the amorphous layer. The channeling spectrum for the e-beam annealed sample shows that a large number of As atoms (-25%) are on interstitial sites. 'he minimum Si yield for that sample is about 5%, indicating some residual damage in the lattice. The results for the thermally annealed sample are similar: 70% As substitutionality and 5% minimum Si yield. The modest results for the control samples can be explained by the low furnace annealing temperature, 575OC. Figure 3.19 shows the results of differential Van der Pauw measurements for the e-beam annealed samples. Plotted in that figure is also the calculated Pearson impurity distribution, from which it is apparent that no diffusion of impurities has taken place during annealing. However, the electrical activation is incomplete at the peak of the concentration profile. The total active dose is 1.02~1015 cm-2, which is 67% of the implanted dose. This number agrees well with the result from channeling, 75% substitutionality.

-

144


RP-~’A; IMPLANT

ENERGY (MeV)

FIGURE 3.18. Aligned channeling spectra for 1.5 MeV 4He ions incident on Si implanted with 75As to a dose of 1.5~1015 cm-2 at 100 keV and then subjected to different anneals (from Ref. 3.35).

Regolini et al., [3.35] also performed TEM studies on the e-beam annealed samples, with the results given in Fig. 3.20. Selected area diffraction pattern (left insert) indicates that complete recrystallization of the amorphous layer has occurred. However, the bright field micrograph indicates that residual defects in the form of small dislocation loops (40-100 a image diameter) remain. The average loop concentration was estimated to be -1014 cm-2, which is significantly more than for the cw laser annealed samples studied in section 3.4.2.1. Removal of 280 8, of Si by anodic oxidation and etching reduced the loop density by about an order of magnitude, indicating that these loops are mostly associated with surface damage. After removal of the high concentration of loops, a distribution of rod-like structures -500 a in average length was observed. It is possible that those rods serve as precipitation sites for the As atoms. 3.4.3.2

The Modified SEX

The SEM was first applied by Ratnakumar et al., l3.371 to anneal ion implanted Si. They used a Cambridge Stereoscan S-4 scanning electron microscope which was modified f o r high current

3.


145

loz1

I. 1.5~10"'5A3/cm'f

100 keV

(UNANNEALED)

FWHM.460

A

?- 10

z

w

u

-

A

L

I

I

200 400

A

I

I

I

I

6 0 0 SO; 1000 1200 DEPTH ( A )

J

FIGURE 3.19. Results of differential Van der Pauw measurements for a sample which was implanted to a 75As dose of 1 . 5 ~ 1 0 ~ ~ at 100 keV and then e-beam annealed. The solid line shows the calculated LSS profile for the 75As implant (from Ref. 3.35). operation. Even in the modified version, the current was limited to a range of 15-30 PA, the voltage being 10-30 kV. The principal results of Ratnakumar et al., r3.371 were that complete recrystallization and good electrical activity was obtained for silicon implanted to an 75As dose of 4 ~ 1 0 ~ cm-2 5 at 150 keV

.

The SEM described above was also used by Re olini et al., silicon samples which were 55As implanted to a dose of 1 . 1 ~ 1 0 ~ ~ at 100 keV. The SEM annealing conditions in this case were 50-70 M at 20 kV, with a scanning speed of 10 cmls. The estimated beam radius was 6 pm. 13.361 to anneal

Results of differential Van der Pauw measurements for the SEM annealed samples are shown in Fig. 3.21. The total active dose obtained from stripping measurement is 9 . 6 ~ 1 0 1 ~cm-2, which means that 14% of As is inactive. The partial inactivity of As was also confirmed by channeling, (see Fig. 3.22): the fraction of As atoms on non-interstitial sites was found to be 12%. Thus the activation obtained with the SEM is better than that achieved by the welder (67% activity, see Section 3.4.3.1).

146


FIGURE 3.20. TEM micrograph of an e-beam (welder) annealed, 75As implanted sample (from Ref. 3.35).

The crystalline quality of the SEM annealed samples was studied by channeling and "EM. The minimum silicon yield (see Fig. 3.22) for the SEM annealed samples was found to be the same as for a virgin control sample, 4%. This indicates a perfect recovery of the crystalline structure. And indeed, the TEM analysis, with the results given in Fig. 3.23 shows that the annealed layer is completely crystalline with few remaining defects. 'Ihe inset in Fig. 3.22(b) shows the electron-diffraction pattern for the dark-field image; the sharpness of the Kikuchi line confirms the good crystalline quality of the layer. The bright-field micrograph [Fig. 3.22(a)] reveals, however, the presence of some dislocation loops. The largest loops are < 200 A in diameter, and the estimated area density of the loops is 1010 This is four orders of magnitude less than the defect density produced by the e-beam welder (Section 3.4.3.1). However, the residual defect density in the cw laser annealed samples is still lower (Section 3.4.2.1).

3.


147

FIGURE 3.21. Electron concentration and mobility vs. depth in a 75As implanted, SEM-annealed sample. The solid lines show the calculated (LSS) 75 As distribution of the implant and the Irvin mobility calculated assuming complete activation of the asimplanted As distribution (from Ref. 3.36). 3.5 3.5.1

ANNEALING OF 75As IMPLANTED Si: C>Css Introduction

Numerous authors i3.39-3.441 have reported that pulsed laser annealing of ion implanted semiconductors can incorporate impurities on substitutional lattice sites in concentrations which greatly exceed the solid solubilities. Some authors have also observed that the supersaturated impurity concentrations are electrically active and thermally unstable [3.39,3.42-3.441. The enhanced solubilities are commonly explained by "solute trapping" of impurities [3.45]. This explanation relies on the melting hypothesis for pulsed laser annealing: when the molten layer resolidifies via liquid phase epitaxy at high speed, impurities would be trapped on substitutional sites. Since cw beam annealing of ion implanted Si takes place in solid phase, the formation of supersaturated impurity concentrations seems to be a great deal less likely. However, if the

148


-DEPTH 4000 2000

(A) 0

1 1 x 1015 A;/cm2/100

0

+

kV

RANDOM SEM ANNEALED UNIMPLANTED SINGLE CRYSTAL

ENERGY (MeV)

FIGURE 3.22. Aligned spectra of 2.2 MeV 4He ions for Si which was implanted with As and then e-beam (SEM) Annealed (from Ref. 3.36).

FIGURE 3.23. TEM micrographs for 75As implanted (100 keV, 1.1~1015 cm-z), SEM annealed Si: (a) bright-field image and (b) weak beam dark-field image (g, 3g, s+ conditions) and

electron diffraction pattern with spot sequence 000 (large spot), 220, 440 and 660 (from Ref. 3.36).

3.


149

implant produces an amorphous layer, the circumstances are more favorable. The mechanism of SPE is broadly similar to liquid phase epitaxy : in both processes, there is an interface betweeen two phases which proceeds towards the sample surface.

In Section 3.3, it was shown that the rate of SPE under typical laser annealing conditions is quite high. The dwell time of the laser is typically short ( 700°C. However, at the edge of a scan line (labelled '' unscanned region" in Fig. 3.44), we see small dislocation loops whose concentration increases with increasing distance from the center of the scan. This means that the laser induced temperature in the edge region was such as to allow some growth of the defects during the scan, but was too low to produce a defect-free crystalline structure. 3.7.2

Use of a Scanning Low Resolution E-Beam

Yep et-al. [3.58] have used a broadly rastered, low resolution electron beam to anneal 1lB-implanted diode structures. The

170


FIGURE 3.43. Transmission electron micrograph of a Bimplanted sample which was thermally annealed at 1000°C for 30 min. (from Ref. 3.57). device properties are beyond the scope of this chapter, but since the experimental apparatus of Yep et al. incorporates some interesting features, we report on their experiments. Diodes were fabricated on (100) n-t e Si wafers (2" diameat 50 keV. The ter) by implanting B to a dose of 1xlOl' diode areas were defined to be 16 mil (0.41 mm) diameter dots by a 0.63 pm thick Si02 layer. The electron beam annealing was performed using an accelerating voltage of 5.2 keV and a current of 52 mA. The annealing system is illustrated in Fig. 3.45. The electron beam was deflected electrostatically k1.5" in the x-direction and 0.75" in the y-direction. In addition, the sample stage was mechanically scanned at 0.13 inchis in the y-direction. This arrangement was effectively a 1.5"~ 3" bar of electrons annealing the wafer. The annealing temperature was measured t o be 850 +2OoC by an optical pyrometer, and the exposure time for a given point was 3-4 s. Annealing was completed by a single pass across the whole wafer, which only took about 20 s. Control wafers were annealed at 850OC and 1000°C for 30 min in flowing nitrogen. The B doping profiles in the annealed samples were determined by SIMS, with the results given in Fig. 3.46. While both furnace

3.


171

FIGURE 3 . 4 4 . Transmission electron micrograph of B-implanted, laser annealed Si. The area labelled "laser scanned" is at the center of the scan, and the "unscanned region" refers to the edge of a single scan (from Ref. 3 . 5 7 ) .

X A N D Y SCANNED 5.2 keV E . B E A M

SPEED 0.13 IN/SEC

\\-

FIGURE 3 . 4 5 . Experimental setup f o r the low resolution ebeam annealing system (from Ref. 3.58). 9 9 8 1 Elsevier Science Publishing CO.


172

anneals have broadened t h e B p r o f i l e by 1200-1600 8 , t h e p r o f i l e has s h i f t e d only about 450 8 d u r i n g e-beam annealing. Figure 3.46 a l s o shows t h e e l e c t r i c a l l y a c t i v e B p r o f i l e , as determined by s p r e a d i n g r e s i s t a n c e measurements. The measured area under t h i s curve shows t h a t t h e a c t i v a t i o n of boron by e-beam a n n e a l i n g i s about 7 5 % .

t FIGURE 3.46. SIMS l l B p r o f i l e f o r t h e sample as-implanted, a f t e r e-beam a n n e a l and a f t e r f u r a n c e a n n e a l a t 1000°C f o r 30 min. For comparison, the e l e c t r i c a l l y a c t i v e l l B p r o f i l e obtained by s p r e a d i n g r e s i s t a n c e a f t e r e-beam a n n e a l i s shown (Ref. 3 . 5 8 ) . 01981 Elsevier Science Publishing Co.

I n conclusion,

the

e-beam

a n n e a l i n g method

by Yep e t a l .

(3.581 d i f f e r s s i g n i f i c a n t l y from those p r e s e n t e d i n S e c t i o n s 3 . 4 . 3 , 3.5.6 and 3.6. The s c a n frequency w a s 550 Hz i n t h e xd i r e c t i o n and 330 Hz i n the y - d i r e c t i o n . While t h e l a t t e r systems f e a t u r e a narrow, low power beam, l e a d i n g t o s h o r t dwell t i m e s but long p r o c e s s i n g t i m e f o r a wafer, Yep e t a l . u t i l i z e a broad beam w i t h a long exposure t i m e ( 4 s ) and f a s t wafer proc e s s i n g . The long dwell t i m e permits the temperature t o be Even though t h e r e s u l t s of Yep e t r e l a t i v e l y low, about 850°C. a l . [ 3 . 5 8 ] a r e obviously n o t y e t o p t i m a l ( t h e e l e c t r i c a l a c t i v a 7 5 x 1 , t h e i r experiment demonstrates t h e p o t e n t i a l t i o n was only of a wide beam t o combine d i f f u s i o n f r e e a n n e a l i n g and a high throughput c a p a c i t y .

-

-

3


173

REFERENCES 3.1 3.2 3.3 3.4

3.5 3.6 3.7 3.8 3.9 3.10

Gat, A., Gibbons, J. F., Magee, T. J., Peng, J., Deline, V. R., Williams, P., and Evans, C. A., Jr., Appl. Phys. Lett. 32, 276 (1978). Williams, J.S., Brown, W. L., Leamy, H. J., Poate, J. M., Rodgers, J. W., Rousseau, D., Rozgonyi, G. A., Shelnutt, J. A., and Sheng, T. T., Appl. Phys. Lett. 33, 542 (1978). Gat, A., Lietoila, A., and Gibbons, J. F., J. Appl. Phys. 50, 2926 (1979). Olson, G. L., kkorowski, S. A., Roth, J. A., and Hess, L. D., in "Laser and Electron-Beam Solid Interactions and Materials Processing" (J. F. Gibbons, L. D. Hess and T. W. Sigmon, eds.), p. 125. North-Holland (1981). Gat, A., Gibbons, J. F., Magee, T. J., Peng, J., Williams, P., Deline, V., and Evans, C. A., Jr., Appl. Phys. Lett. 33, 389 (1978). Lietoila, A . , Gibbons, J. F. , Magee, T. J. , Peng, J., and Hong, J. D., Appl. Phys. Lett. 35, 532 (1979). Lietoila, A . , Gibbons, J. F., and Sigmon, T. W., Appl. Phys. Lett. 36, 765 (1980). Regolini, J. L., Sigmon, T. W., and Gibbons, J. F., Appl. Phys. Lett. 35, 114 (1979). See, e.g., White, C. W., Christie, W. H., Appleton, B. R., Wilson, S. R., and Pronko, P. P., Appl. Phys. Lett. 33, 662 (1978). Thomas, G., "Transmission Electron Microscopy of Materials , Wiley (1971). Van der Pauw, L. J., Philips Technical Review 20, 220 (1958/1959). Johansson, N. G. E., Mayer, J. W., and Marsh, 0. J., Solid State Electron. 13, 317 (1970). Bube, R.H., "Electronic Properties of Crystalline Solids," p. 379, Academic Press (1974). Barber, H. D., Lo, H. B., and Jones, J.E., J. Electrochem. SOC. 123, 1404 (1976). Regolini, J. L., unpublished. Dearhaley, G., Freeman, J. H . , Nelson, R. S., and Stephen, J., "Ion Implantation," p. 539, North-Holland (1973). Hofker, W.K., Werner, H.W., Oosthoek, D.P., and de Grefte, H. A. M., Appl. Phys. 2, 265 (1973). Williams, P., Lewis, R. K., Evans, C. A., Jr., and Harley, P. R., Analytical Chem. 49, 1399 (1977). Williams, P., and Evans, C. A., Jr., Appl. Phys. Lett. 30, 559 (1977). Dobrott, R. D., NBS Special Publication 400-23, ARPA/NBS Workshop IV, Surface Analysis for Silicon Devices, held at NBS, Gaithersburg, Md., 23-24 April 1975, p. 31. Ibid, p. 45 Chu, W.-K., Mayer, J.W., and Nicolet, M.-A,, "Backscattering Spectrometry," Academic Press (1978). "

3.11 3.12 3.13 3.14 3.15 3.16. 3.17 3.18 3.19 3.20 3.21 3.22

174 3.23 3.24 3.25 3.26 3.27 3.28

3.29 3.30 3.31 3.32 3.33 3.34

3.35

3.36

3.37

3.38 3.39 3.40

3.41

A R T 0 LIETOILA AND JAMES F. GIBBONS

Nuclear I n s t r u m e n t s and Methods 126, 205 Williams, J . S . , (1975). Chu, W.-K., Muller, H., and Mayer, J. W., Sigmon, T. W., J. Appl. Phys. 5, 347 (1975). C s e p r e g i , L., Mayer, J. W., and Sigmon, T. W., Phys. L e t t . 54A, 157 (1975). German, P., Squelard, S . , Bourgoin, J. C., Zellama, K., and Thomas, P. A,, J. Appl. Phys. 50, 6995 (1979). Gat, A., Ph.D. T h e s i s , S t a n f o r d U n i v e r s i t y , Department of E l e c t r i c a l Engineering (1979). Roth, J. A., Anderson, C. L., and Dunlap, Hess, L. D., H. L., "Laser-Solid I n t e r a c t i o n s and Laser P r o c e s s i n g 1978," ( S . D. F e r r i s , H. J. Leamy, and J. H. P o a t e , eds.) A I P Conf. Proc. 50, 496, American I n s t i t u t e of P h y s i c s (1979). Olson, G. L., Kokorowski, S. A., McFarlane, R. A., and Hess, L. D., Appl. Phys. L e t t . 37, 1019 (1980). Kennedy, E. F., G a l l a g h e r , T. J., Mayer, Csepregi, L., J . W., and Sigmon, T. W., J. Appl. Phys. 48, 4234 (1977). Kennedy, E. F., Mayer, J. W., and Sigmon, Csepregi, L., T. W., 3. Appl. Phys. 49, 3906 (1978). L i e t o i l a , A., Gold, R. B., and Gibbons, J . F., Appl. Phys. L e t t . 53, 1169 (1982). Ph.D. T h e s i s , S t a n f o r d U n i v e r s i t y , DepartL i e t o i l a , A., ment of Applied Physics (19811, Chapter IV. Auston, D. H., Golovchenko, J. A . , Smith, P. R., Surko, C. M., and Venkatesen, T. N. C., Appl. Phys. Lett. 33, 539 (1978). Gibbons, J. F., Sigmon, T. W., Pease, R e g o l i n i , J. L., R. F. W., Magee, T. J., and Peng, J., Appl. Phys. Lett. 34, 410 (1979). Johnson, N. M., S i n c l a i r , R., Sigmon, R e g o l i n i , J. L., T. W., and Gibbons, J. F., i n "Laser and Electron-Beam P r o c e s s i n g of Materials," (C. W. White and P. S . Peercy, eds.), p. 297, Academic P r e s s (1980). Pease, R. F. W., B a r t e l i n k , D. J., Ratnakumar, K. N., Johnson, N. M., and Meindl, J.D., Appl. Phys. L e t t . 35, 463 (1979). and Dutton, R. W., S t a n f o r d D'Avanzo, D. C., Rung, R. D., 5013-2 Electronics Laboratories T e c h n i c a l Report No. (1977). Narayan, J., and Young, R. T., AIP ConferWhite, C. W., ence Proceedings 50, 275 (1979). Grob, J. J., Grob, A., S i f f e r t , Fogarassy, E., Stuck, R., P., i n "Laser and Electron-Beam P r o c e s s i n g of Materials," (C. W. White and P. S. Peercy, eds.), p. 117, Academic P r e s s (1980). i n "Laser and E l e c t r o n B e a m P r o c e s s i n g of Rimini, E., E l e c t r o n i c Materials," ECS Proc. 80, 270, E l e c t r o c h e m i c a l S o c i e t y (1980).

3 - ION IMPLANTED CRYSTALLINE SILICON 3.42 3.43 3.44

3.45 3.46 3.47 3.48 3.49 3.50 3.51 3.52 3.53 3.54

3.55 3.56 3.57 3.58

3.59 3.60 3.61

175

Tamura, M., Natsuaki, N., and Tokuyama, T., in "Laser and Electron-Beam Processing of Materials," (C. W. White and p. 247, Academic Press (1980). P. S. Peercy, eds.), Chu, W. K., Mader, S . R., and Rimini, E., in "Laser and Electron-Beam Processing of Materials," (C. W. White and P. S . Peercy, eds.) p. 253, Academic Press (1980). Pianetta, P., Amano, J., Woolhouse, G., and Stolte, C. A., "Laser and Electron-Beam Solid Interactions and Materials Processing," (J. F. Gibbons, L. D. Hess and T. W. Sigmon, eds.), p. 239, North Holland (1981). Jackson, K. A., Gilmer, G. H., and Leamy, H. J., in "Laser and Electron-Beam Processing of Materials," (C. W. White and P. S . Peercy, eds.), p. 104, Academic Press (1980). Lietoila, A,, Ph.D. Thesis, Stanford University, Department of Applied Physics (1981). Chapter V. Lietoila, A., Gibbons, J. F., Magee, T. J., Peng, J., and Hong, J . D., Appl. Phys. Lett. 35, 532 (1979). Lietoila, A,, Gibbons, J. F., and Sigmon, T. W., Appl. Phys. Lett. 36, 765 (1980). Lietoila, A., Gibbons, J. F., Sigmon, T. W., Magee, T. J., Peng, and Hong, J . D., [Ref. 3.411, p. 350. Sze, S . M., "Physics of Semiconductor Devices," p. 43, Wiley (1969). Trumbore, F. A., Bell Syst. Tech. J. 39, 205 (1960). Tsai, M.Y., Morehead, F.F., Baglin, J.E.E., and Michael, A. E., J. Appl. Phys. 51, 3230 (1980). Fair, R. B., J. Electrochem. SOC. 125, 323 (1978). Chu, W. K., and Masters, B. J., "Laser-Solid Interactions and Laser Processing - 1978," ( S . D. Ferris, H. J. Leamy, and J. H. Poate, eds.) AIP Conf. Proc. 50, 305, American Institute of Physics (1979). See, e.g., Bennerman, K. H., Phys. Rev. 137A, 1497 (1965); and Van Vechten, J. A., and Turmond, C.D., Phys. Rev. B14, 3551 (1976). Regolini, J. L., Sigmon, T. W., and Gibbons, J. F., Appl. Phys. Lett. 35, 114 (1979). Gat, A., Gibbons, J. F., Magee, T. J., Peng, J., Williams, P., Deline, V., and Evans, C. A,, Jr., Appl. Phys. Lett. 33, 389 (1978). Yep, T. O., Fulks, R. T., and Powell, R. A., in "Laser and Electron-Beam Solid Interactions and Materials Processing" (J. F. Gibbons, L. D. Hess and T. W. Sigmon, eds.) p. 345, North-Holland ( 1981) Lietoila, A., Wakita, A., Sigmon, T. W., and Gibbons, J.F. J. Appl. Phys. 53, 4399 (1982). Lidow, A., Ph.D. Thesis, Stanford University, Dept. of Applied Physics (1977). Nissim, Y. I., Ph.D. Thesis, Stanford University, Dept. of Electrical Engineering (1981).

.


CHAPTER 4

Electronic Defects in CW Transient Thermal Processed Silicon N . M . Johnson XEROX CORPORATION PALO ALTO RESEARCH CENTER PALO ALTO, CALIFORNIA

4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DEVICE PROCESSING AND MEASUREMENT TECHNIQUES ....................................... CW BEAMPROCESSED BULKSILICON. . . . . . . . . . . . . . . . . LATERAL NONUNIFORMITIES IN SCANNED-BEAM ANNEALING. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ISOTHERMAL TRANSIENT ANNEALING. . . . . . . . . . . . . . . . . . POST-RECRYSTALLIZATION PROCESSING. . . . . . . . . . . . . . . . BEAM-CRYSTALLIZED SILICONTHINFILMS. . . . . . . . . . . . . . SUMMARY AND CONCLUSIONS

........................

REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.1

177 179 184 195 200 204 21 1 220 223

INTRODUCTION

Directed energy sources such as lasers and electron beams and incoherent light sources have been proposed as alternatives to conventional furnace annealing for recrystallizing the amorphous layer created by high-dose ion implantation in singlecrystal silicon. These energy sources have also been used to crystallize silicon thin films on insulating amorphous substrates in order to obtain semiconducting material for electronic device fabrication. In both applications the purpose of transient thermal processing is to produce single-crystal material of high crystalline perfection. With cw energy sources, recrystallization of implanted amorphous layers occurs by solid-phase epitaxy, with lower densities of extended defects (e.g., stacking faults) than can be achieved by conventional furnace annealing. However, materials studies with techniques such as capacitance transient spectroscopy and luminescence reveal high densities of residual defects in and near the recrystallized layers, which give rise to deep levels in the silicon forbidden-energy band. This chapter reviews the current state of knowledge in the ongoing study of residual defects in recrystallized bulk silicon and introduces the subject of electronic defect evaluation in crystallizedsilicon thin films. The high degree of perfection and control which has been achieved in silicon integrated-circuit fabrication must be retained while realizing the potential advantages of beam recrysSEMICONDUCTORS AND SEMIMETALS, VOL. 17

177

Copyright 0 1984 by Academic Press, Inc. All rights of reproduction in any form reserved. ISBN 0-12-752117-8

178

N.M. JOHNSON

tallization. Only then will the unique features of spatial and temporal localization of energy be appreciated. The stringency of this requirement becomes evident when it is noted that one of the major achievements underlying the success of present-day silicon integrated-circuit processing is the minimization and control of electrically active defects in single-crystal silicon. Impurities such as oxygen and carbon can be tolerated at concentrations of the order of 10l8 ~ m - ~which , is typical for Czochralski-grown silicon, because they can be rendered electrically inactive. On the other hand, a metallic impurity such as gold can significantly alter the electrical pro erties of a device at concentrations of less than loll cm-' through the introduction of deep traps and recombination centers. For transient thermal processing of silicon, the issue centers on the removal of electronic defects associated with lattice damage, such as that created by ion implantation. This chapter first examines the problem of residual defects which result from the use of cw transient thermal sources to recrystallize the amorphous layer created by high-dose ionimplantation in silicon. This topic is of central importance in silicon integrated-circuit processing due to the extensive use of ion implantation to introduce dopant impurities (e.g., As, P, and B) for spatially controlling the free-carrier concentration. The displacement damage created by ion implantation is conventionally removed by furnace annealing, which also electrically activates the implanted impurities through their placement on substitutional lattice sites. As discussed in Chapter 3, directed energy sources such as laser and electron beams can be used for this purpose. The potential advantages of energy beams are derived from their spatial and temporal features. For example, regions of silicon can be selectively annealed thereby preventing possible electrical degradation of other regions not requiring thermal cycling. It has been shown that implanted layers can be recrystallized with low densities of extended defects and with complete electrical activation of the dopant. Cw transient thermal sources, including incoherent light sources, offer the additional feature that the implanted amorphous layer is recrystallized without dopant redistribution. This is a consequence of recrystallizing the amorphous layer by solid-phase epitaxial regrowth on a time scale of seconds or less, which precludes significant dopant diffusion. In addition, it is generally found that free-carrier mobilities in recrystallized layers equal their values in conventionally processed bulk single-crystal silicon. The above features of cw transient thermal processing are particularly applicable in silicon submicron device fabrication and have motivated numerous studies of residual electronic defects and their dependence on transient annealing conditions. Beam-crystallized silicon thin films are of technological interest as a materials/processing system for fabricating highperformance thin-film transistors (TFT) and circuits. As discus-

4. ELECTRONIC DEFECTS IN SILICON

179

sed in Chapter 5 , such devices have been fabricated on insulating layers over single-crystal silicon wafers and could lead to vertical integration in silicon integrated-circuit technology. High-performance TFTs have also been fabricated on bulk glass substrates and may be applicable for switching and logic circuitry in large-area displays. For a TFT technology, crystalline silicon thin films offer the significant feature of utilizing conventional silicon microelectronic processing techniques in the fabrication of thin-film devices and circuits. The potential impact of these rapidly emerging device technologies underscores the need for comprehensive evaluation of the electronic properties of crystallized-silicon thin films, and in particular, characterization of electronic defects and their correlation with materials processing and device operation. The chapter is organized as follows: general experimental considerations of device fabrication and measurements of elecElectronic defects tronic defects are discussed in Section 4.2. in cw beam annealed bulk silicon are reviewed in Section 4 . 3 . This section is concerned both with residual defects arising from the incomplete removal of ion-implantation damage by beam annealing and with the generation of defects under such annealing conditions. Laterally distributed defects are examined in Section 4.4, which includes consideration of the effect of patterned dielectric overlayers in scanned beam annealing. Section 4.5 presents results from deep level studies in isothermal transient annealed silicon. Section 4.6 reviews post-recrystallization processing techniques for further removal of residual defects. The emerging topic of beam-crystallized silicon thin films is discussed in Section 4.7 with results for electronic defects in silicon layers on both thin film and bulk amorphous substrates. A summary with conclusions is presented in Section 4.8. 4.2

DEVICE PROCESSING AND MEASUREMENT TECHNIQUES

In bulk single-crystal silicon, several forms of energy beams have been used to recrystallize the amorphous layer created by high-dose ion implantation. The features of cw beam processing include spatial selection of annealed regions, negligible dopant redistribution, and retention of majority-carrier transport properties. As discussed in Chapter 1, these features have been demonstrated with scanned cw lasers, scanned electron beams, incoherent light sources, and line-source (shaped) electron beams. However, from electrical measurements that are specifically sensitive to minority-carrier transport and from direct measurements of deep levels it has been found that all of the above forms of beam annealing leave residual electronic defects, with densities in excess of those obtained by conventional furnace annealing. This section reviews general considerations in the design and fabrication of test devices for the characterization of electronic defects in transient thermal annealed silicon.

N. M.JOHNSON

4.2.1

CW Transient Thermal Processing

The principal diagnostic technique that has been used for electronic defect evaluation is capacitance transient spectroscopy performed on current-rectifying devices. Both Schottkybarrier and p-n junction diodes have been used €or this purpose. Schematic cross-sectional diagrams of these structures are shown in Fig. 4.1, with the amorphized layer resulting from high-dose ion implantation represented by the cross-hatch patterns. The Schottky diode (Fig. 4.la) is the simplest device to fabricate, requiring only the deposition of a metallic rectifying contact onto the silicon surface. However, the Schottky contact cannot be used to directly evaluate implanted dopants because the high doses required to amorphize silicon (>1014 produce a degenerately doped surface layer which destroys the current rectifying feature of the Schottky barrier. To utilize this structure, self implantation may be used to create lattice displacement damage, of the form accompanying implantation of the commonly used dopants in silicon, without altering the shallow dopant concentration. Alternatively, the Schottky diode is well suited for studying induced defects in the absence of implantation. Dopant implants are conveniently studied with p-n junction test devices in which the heavily doped surface layer permits the fabrication of structures which closely approximate one-sided step junctions. The p-n junction structure can be realized i n two configurations: the mesa diode (Fig. 4.lb) and the oxidepassivated planar diode (Fig. 4.1~). As test devices, these two configurations differ solely in the form of the junction perimeter. And while the planar p-n junction is the more technologically relevant configuration, the mesa structure is often the more convenient to fabricate for defect studies. For capacitance transient spectroscopy it is essential to have back electrical contacts which remain ohmic at low temperatures, since the measurement is performed over a range of temperatures typically from 78 K to above room temperature. This i s readily achieved through the use of semiconducting epilayers on degenerately doped substrates. On bulk single-crystal silicon the back surface may be implantedwith the substrate dopant, which is then electrically activated with a furnace anneal (e.g., 800°C, 30 min, N2), in order to produce a degenerately doped layer for ohmic contact. To further facilitate electrical connection to the back contact, a metallic thin film (e.g., aluminum) can be deposited over the back surface of the wafer and then sintered In studies of scanned(at, e.g., 450°C, 30 min, forming gas). beam annealing, preparation of back contacts can be completed prior to front surface processing. Transient thermal processing is used to recrystallize the amorphous layer created by ion implantation and to electrically


n -type Si

METAL

RECRYSTALLIZED LAYER

-’ p-type Si

181

\-RECRYSTALLIZED LAYER

I

n-type si METALLURGICAL JUNCTION

METAL

LL

RECRYSTALLIZED J’ LAYER

n-type Si METALLURGICAL JUNCTION

p-type Si

FIGURE 4.1. Schematic diagrams of two-terminal current rectifying devices for electronic defect evaluation in recrystallized bulk silicon: (a) Schottky-barrier diode, (b) p-n junction (mesa) diode, and (c) oxide-passivated planar p-n junction diode. The cross-hatch areas represent material driven amorphous by ion implantation and subsequently recrystallized. activate implanted dopants. Cw beam processing is performed with a focused laser or electron beam by scanning the energy beam over the silicon surface in a raster pattern. Energy beams shaped to form a scanned line source permit recrystallization over large areas in a single pass. In scanned beam processing, the silicon wafer is locally heated with dwell times of the deposited energy typically of the order of milliseconds. With isothermal transient annealing (Section 4 . 5 ) , the entirewafer is annealed at an essentially constant temperature €or times on the order of seconds.

N. M.JOHNSON

182

T e s t d e v i c e s are completed by vacuum e v a p o r a t i n g m e t a l l i c t h i n f i l m s o v e r t h e f r o n t s u r f a c e of t h e wafer and u s i n g conFor p-n ventional photolithography t o define the electrodes j u n c t i o n d i o d e s , plasma e t c h i n g o r w e t chemical e t c h i n g i s used t o form t h e mesa s t r u c t u r e , and on p l a n a r diodes t h e j u n c t i o n i s defined by chemically e t c h e d windows i n t h e s i l i c o n - d i o x i d e passivation layer. For c o n t r o l d e v i c e s , conventional f u r n a c e a n n e a l i n g r e p l a c e s t r a n s i e n t thermal p r o c e s s i n g i n t h e f a b r i c a t i o n of Schottky o r p-n j u n c t i o n test d e v i c e s .

.

4.2.2

Measurement Techniques

S e v e r a l e l e c t r i c a l and p h y s i c a l measurement t e c h n i q u e s have been used t o e v a l u a t e r e s i d u a l e l e c t r o n i c d e f e c t s , and t h e i r possible c o r r e l a t i o n with s t r u c t u r a l imperfections, i n t r a n s i e n t thermal r e c r y s t a l l i z e d s i l i c o n . Both c a p a c i t a n c e t r a n s i e n t spectroscopy and luminescence have been used t o d i r e c t l y probe deep l e v e l s i n t h e s i l i c o n bandgap. However, i t s high s e n s i t i v i t y and d i r e c t e l e c t r i c a l d e t e c t i o n of r e s i d u a l deep l e v e l s have rendered c a p a c i t a n c e - t r a n s i e n t spectroscopy t h e most widely used diagnostic technique f o r electronic-defect e v a l u a t i o n . I n d i r e c t o b s e r v a t i o n s of r e s i d u a l d e f e c t s , by d e t e c t i o n of m i n o r i t y - c a r r i e r recombination, have been made w i t h current/photocurrent-voltage measurements and w i t h electron-beam-induced c u r r e n t s . S t r u c t u r a l information f o r c o r r e l a t i o n w i t h e l e c t r i c a l measurements has been obtained from Nomarski-interference microscopy, t r a n s m i s s i o n e l e c t r o n microscopy, and x-ray topography. Most of t h e r e s u l t s reviewed i n t h i s c h a p t e r were o b t a i n e d by c a p a c i t a n c e t r a n s i e n t s p e c t r o s c o p y . The b a s i c method is termed deep-level t r a n s i e n t spectroscopy (DLTS) [ 4 .l] and y i e l d s t h e energy l e v e l s and d e n s i t i e s of e l e c t r o n i c d e f e c t s . Many of t h e r e s u l t s p r e s e n t e d h e r e were o b t a i n e d from DLTS measurements performed i n t h e constant-capacitance mode, which i s p a r t i c u l a r l y a p p l i c a b l e when measuring t r a p d e n s i t i e s comparable t o t h e shallow dopant c o n c e n t r a t i o n . A d e t a i l e d a n a l y s i s of constant-capacitance DLTS (CC-DLTS) measurements of bulk-semiconductor d e f e c t s h a s I n a d d i t i o n , v a r i o u s DLTS techbeen p r e s e n t e d elsewhere [ 4 . 2 ] niques have been developed t o o b t a i n s p a t i a l depth p r o f i l e s of d e f e c t l e v e l s [4.1-4.31

.

.

The e s s e n t i a l f e a t u r e s of t h e DLTS technique are i l l u s t r a t e d i n F i g . 4.2 w i t h an energy-band diagram f o r a S c h o t t k y - b a r r i e r s t r u c t u r e w i t h a s i n g l e t r a p l e v e l a t energy Et i n t h e bandgap of an n-type semiconductor. The DLTS measurement i n v o l v e s f i r s t t h e a p p l i c a t i o n of a r e v e r s e b i a s VR t o e s t a b l i s h a d e p l e t i o n l a y e r of width WD. The t r a p l e v e l i n t e r s e c t s t h e quasi-Fermi energy EFS a t XD. Traps l o c a t e d below EFS are f i l l e d w i t h elect r o n s and t h o s e above a r e empty. Voltage p u l s e s p e r i o d i c a l l y


Metal

183

Semiconductor

FIGURE 4.2. Schematic energy-band diagram for a Schottkybarrier structure with a single discrete defect level at an energy Et in the silicon bandgap. The distance WD is the depletion width for a reverse bias VR, and XD is the depth below the silicon surface at which the defect level intersects the quasi-Fermi level. The distance X is the depth at which the trap level intersects the quasi-Fermf energy during a pulse bias Vp. reduce the depletion width in order to populate additional traps. After returning to the quiescent bias, these trapped electrons are thermally emitted to the conduction band and swept out of the depletion layer. This gives rise t o a capacitance transient in the transient-capacitance mode or a voltage transient in the constant-capacitance mode of measurement. Analysis of the DLTS signal provides a characterization of deep levels in the semiconductor. The DLTS signal is obtained by forming the difference of the transient response (either capacitance or voltage transient) measured at two delay times ti and t2 after a charging pulse. For an exponential transient, the delay times define an emission rate constant eo which is given by the expression [4.1]

184

N. M.JOHNSON

A DLTS spectrum is obtained by recording the DLTS signal over a range of temperatures. An emission peak appears at that temperature for which the emission rate of a given trap, en, equals eo. The emission rate constant can be varied to obtain the emission rate over a range of temperatures. Then the activation energy for thermal emission is obtained from an Arrhenius analysis of the emission rate, which from considerations of detailed balance may be expressed as

where un is the electron capture cross section, vn is the mean thermal velocity for electrons, Nc is the effective density of states in the semiconductor conduction band, Ec is the conductionband minimum, k is Boltzmann's constant, and T is the absolute temperature. The DLTS analysis thus yields the activation energy for thermal emission of charge carriers and the cross-section for capturing free carriers. In addition, the amplitude of the DLTS signal is proportional to the trap concentration and can be used to measure the spatial depth profile of deep levels.

4.3

4.3.1

CW BEAM PROCESSED BULK SILICON Recrystallized Silicon

Electronic defects remaining after beam recrystallization of a n implanted amorphous layer may be considered to arise from either of two origins. Ion implantation introduces displacement damage to depths beyond the amorphized surface layer. A beam annealing schedule can be sufficient to recrystallize the amorphous layer by solid phase epitaxy while only partially removing this more deeply penetrating lattice damage. On the other hand, beam irradiation can generate both electronic and extended structural defects as a consequence of the high thermal gradients and rapid quenching which are consequences of the spatial and temporal features of beam recrystallization. A clear distinction between these two origins can be obscured by defect interactions; that is, the form, distribution, and density of induced defects may be altered by the presence of displacement damage and impurities. Residual electronic defects in cw beam-recrystallized silicon were first evaluated in Schottky-barrier diodes which were processed with a scanned cw Ar-ion laser i4.41. Epitaxial silicon


185

wafers with n-type conductivity were implanted at room temperature with SiH' to create displacement damage, of the form accompanying the implantatlon of dopants, without altering the shallow dopant concentration. The ions were implanted at 80 keV to a dose of 2x1015 cm'2, which is sufficient to drive the silicon amorphous to a depth of -120 nm. The im lanted species was chosen to avoid nitrogen contamination (N24) which can occur during 28Si+ implantation. CC-DLTS emission spectra are presented in Fig. 4 . 3 for an as-laser-annealed diode. The two emission spectra were recorded with different combinations of reverse and pulse biases, which define different spatial observation windows for trap detection. The bounds of the spatial intervals quoted in Fig. 4 . 3 (i.e., AWD) are the steady-state depletion depths under reverse and pulse biases, which were computed from the measured device I

2a

-

I

I

Epitaxial Silicon N-Type, (100) Self Implanted en = 347 sec-'

I

I

(- 0.49 eV)

->-E 15 ID

5

iij v)

5

9 0

10

0

5

0 -1

I t 100

I 150

I

I

I

200

250

300

0

T(K) FIGURE 4 . 3 . CC-DLTS spectra for electron emission in selfimplanted cw laser-annealed silicon and from a furnace-annealed control device.

N. M.JOHNSON

186

capacitance. The i n t e r v a l s i n d i c a t e t h e r e l a t i v e d e p t h s of t h e The e l e c t r o n e m i s s i o n o b s e r v a t i o n windows f o r t h e two s p e c t r a . s p e c t r a are dominated by two d e f e c t l e v e l s which are l o c a t e d n e a r t h e middle of t h e s i l i c o n band gap. The e m i s s i o n s p e c t r a recorded w i t h d i f f e r e n t o b s e r v a t i o n windows i n d i c a t e t h a t t h e r a t i o of t h e d e n s i t i e s of t h e two d e f e c t l e v e l s v a r i e s r a p i d l y w i t h depth. The a c t i v a t i o n e n e r g i e s were o b t a i n e d from measurements over s p a t i a l i n t e r v a l s i n which a s i n g l e e m i s s i o n peak dominated t h e spectrum. The a c t i v a t i o n energy f o r t h e hight e m p e r a t u r e peak i s 0.56 eV. The s p a t i a l s e p a r a t i o n of t h e two t r a p s i s p r i m a r i l y due t o a r a p i d d e c r e a s e w i t h d i s t a n c e i n t h e The s p a t i a l d e n s i t y of t h e s h a l l o w e r of t h e two midgap l e v e l s . s e p a r a t i o n i s less complete f o r t h i s s h a l l o w e r l e v e l , g i v i n g rise t o an u n c e r t a i n t y i n t h e a c t i v a t i o n energy which i s found A broad emission s i g n a l of r e v e r s e p o l a r i t y t o be -0.49 e V . a t i n t e r m e d i a t e t e m p e r a t u r e s i s a s c r i b e d t o a h i g h t r a p concent r a t i o n n e a r t h e s i l i c o n s u r f a c e which c o n t r i b u t e s t o a h i g h A t low t e m p e r a t u r e s a t h i r d degree of charge compensation. e m i s s i o n peak of low i n t e n s i t y i s d e t e c t a b l e , w i t h a n a c t i v a t i o n energy of 0.19 eV; t h i s l e v e l is removed by a forming-gas a n n e a l as d i s c u s s e d below. (See S e c t i o n 4.3.2 f o r further discussion of t h i s l e v e l . ) Also shown i n Fig. 4.3 i s t h e f e a t u r e l e s s emiss i o n spectrum f o r a c o n t r o l d i o d e which was f u r n a c e annealed a t 1000°C (30 min, N 2 ) . CC-DLTS emission s p e c t r a are shown i n Fig. 4.4 f o r a diode which r e c e i v e d a forming-gas a n n e a l (15% H2, 85% Np, 450 C , 30 min.). These s p e c t r a are dominated by t h e same two midgap d e f e c t l e v e l s a s found above, w i t h the l e v e l a t -0.49 eV decaying more r a p i d l y t h a n t h e 0.56 e V l e v e l w i t h d e p t h i n t o t h e s i l i c o n subs t r a t e . However, t h e r e is an a d d i t i o n a l e m i s s i o n peak w i t h an a c t i v a t i o n energy of 0.28 e V which i s comparable i n magnitude t o This l e v e l i s w e l l r e s o l v e d i n t h e e m i s s i o n t h e midgap peaks. spectrum and can be unambiguously p r o f i l e d by t h e double-correlat i o n CC-DLTS technique. The d e f e c t - d e n s i t y d i s t r i b u t i o n i s shown i n Fig. 4.5. The h o r i z o n t a l b a r s mark t h e s p a t i a l i n t e r v a l s over which average d e f e c t d e n s i t i e s were measured, and t h e v e r t i c a l b a r s d e n o t e t h e u n c e r t a i n t i e s i n t h e d e n s i t i e s due t o u n c e r t a i n t y i n t h e measurement of t h e n e t e m i s s i o n s i g n a l . The d e f e c t d e n s i t y d e c r e a s e s monotonically w i t h d e p t h below t h e l a s e r - i r r a d i a t e d s u r f a c e , over t h e i n v e s t i g a t e d i n t e r v a l . Also shown i s t h e p r o j e c t e d This approximates t h e range f o r Si' implanted a t 77 keV [4.5]. p r o j e c t e d range of s i l i c o n f o r t h e a c t u a l l y implanted s p e c i e s , SiH+ i f i t is assumed t h a t t h e i o n i z e d molecule d i s s o c i a t e s i n t o s i l i c o n and hydrogen atoms a t t h e s i l i c o n s u r f a c e , w i t h equa p a r t i c l e v e l o c i t i e s a t t h e i n s t a n t of s e p a r a t i o n .

ing and

High d e n s i t i e s of deep l e v e l s remain a f t e r cw laser anneal. ntroduced a n n e a l e f f e c t i v e l y removed t h e l e v e l a t 0.19 eV The a n o t h e r l e v e l a t 0.28 e V . The d e n s i t i e s of a l l 450-C


187

25

20

Epitaxial Silicon N-TYw, (100) Self Implanted CW Laser Annealed (1,= 350 C) ' Furnace Annealed: 450 C, 30 rnin. eo = 347 sec-'

I

> E

1 15 6 i7j v)

< io

I-

V

0

5

C

T(K)

FIGURE 4.4. CC-DLTS emission s p e c t r a f o r a self-implanted laser-annealed diode which a l s o r e c e i v e d a 45OoC f u r n a c e anneal. l e v e l s d e c r e a s e w i t h depth i n t o t h e s i l i c o n s u b s t r a t e from > IOl5 i n t h e near-surface r e g i o n , as shown f o r t h e l e v e l a t 0.28 e V i n Fig. 4.5. The p r o j e c t e d range f o r Si' r e v e a l s t h a t t h e meas u r e d d e f e c t s r e s i d e i n material t h a t was n o t d r i v e n amorphous by i o n i m p l a n t a t i o n . Diodes f a b r i c a t e d w i t h cw laser-processed Czochralski-grown s i l i c o n produce t h e same r e s u l t s as shown above f o r e p i t a x i a l material [4.61. E l e c t r o n i c d e f e c t l e v e l s i n scanned electron-beam annealed The test d e v i c e s w e r p-n (SEBA) s i l i c o n are shown i n Fig. 4.6. j u n c t i o n d i o d e s f a b r i c a t e d by i m p l a n t i n g As' (100 keV, 5x101 Z

(B,1-3 a-cm) s i l i c o n . The i o n i n t o Czochralski-grown p-type energy and dose were s u f f i c i e n t t o d r i v e t h e s i l i c o n s u r f a c e amorScanned electron-beam a n n e a l i n g phous t o a depth of -100 nm. was performed w i t h both a commercial electron-beam welder i4.7 1 and a scanning e l e c t r o n microscopy 14.81, w i t h t h e s u b s t r a t e s a t room temperature. The p a j u n c t i o n diodes were completed by d e p o s i t i n g aluminum over t h e r e c r y s t a l l i z e d s u f a c e , u s i n g photol i t h o g r a p h y and plasma e t c h i n g t o d e f i n e t h e mesa s t r u c t u r e , and CC-DLTS s p e c t r a i n Fig. s i n t e r i n g (45OoC, 30 min. forming g a s ) . 4.6 are f o r h o l e emission i n t h e d e p l e t i o n l a y e r of t h e p-type s u b s t r a t e . The h o l e t r a p a t 0.28 e V dominates t h e emission

N. M.JOHNSON

188

Epitaxial Silicon N Type. (100) Self Implanted CW L a w A n d & (T, = 350 Cl Furnace Anne* 450 C, 30 min.

1075

t

\

3 -1

Depth (10' \ )

E,-0.28

FIGURE 4.5. S p a t i a l depth p r o f i l e of t h e d e f e c t l e v e l a t eV i n self-implanted laser-annealed s i l i c o n .

s p e c t r a , w i t h an a d d i t i o n a l peak a t 0.36 e V i n t h e SEM-annealed device. These are a c t i v a t i o n e n e r g i e s f o r thermal emission of h o l e s t o t h e s i l i c o n v a l e n c e band. Neither l e v e l i s detectable i n t h e furnace-annealed c o n t r o l diode. The d e f e c t d e n s i t i e s were found t o be i n t h e 1013 cm-3 range t o g r e a t e r d e p t h s t h a n a micron below t h e m e t a l l u r g i c a l j u n c t i o n [4.9]. R e s u l t s f o r cw Ar-ion l a s e r - f a b r i c a t e d p-n j u n c t i o n s are shown i n Fig. 4.7 [4.10]. Boron-doped s i l i c o n wafers were implanted w i t h As' (100 keV, 4 ~ 1 0 1 ~ cw laser annealed w i t h t h e s u b s t r a t e a t 350 C , and f a b r i c a t e d i n t o mesa d i o d e s ; I n Fig. 4.7 f o u r h o l e t r a p s are t h e d i o d e s were n o t s i n t e r e d . d e t e c t a b l e w i t h a c t i v a t i o n e n e r g i e s as f o l l o w s : 0.10 e V , 0.20 e V , 0.28 e V , and 0.45 e V ; t h e s e are denoted as H(0.10), H(0.20), H(0.28), and H(0.451, respectively. The H(0.28) l e v e l is t h e same l e v e l shown i n Fig. 4 . 6 f o r s i n t e r e d d i o d e s and w a s not p r e s e n t when t h e s u b s t r a t e t e m p e r a t u r e was s e t t o 250 C d u r i n g l a s e r annealing. The dominant h o l e t r a p a t 0.45 e V w a s found t o be u n s t a b l e a t room temperature. The d e f e c t d e n s i t y decays w i t h t i m e , w i t h t h e l e v e l t r a n s m u t i n g t o t h e s h a l l o w e r l e v e l a t 0.10 eV. By comparing these d e f e c t s i n laser-annealed s i l i c o n w i t h

" 4. ELECTRONIC DEFECTS IN SILICON

Cz Silicon P-Type, (100) As+ Implanted

(0.28 eV)

n

-> E

-m

1

189

3-

SEBA (SEM)

P-N Junction Hole Traps eo = 347 ~ c - l

-

SEBA (Welder)

C

en iij 2 v)

Furnace Annealed (1000 C, 30 min, N,

4 0 0

1

1-

80 100

150

200

250

300

330

FIGURE 4.6. CC-DLTS spectra for hole emission in As+implanted scanned electron-beam annealed silicon. The spectrum for a furnace-annealed p-n junction diode is also shown. defects produced by furnace annealing followed by rapid cooling, and with other published results, the laser induced defects were identified as interstitial Fe and Fe-B pairs. A comparison of electronic defect levels in laser-annealed and scanned-electron-beam-annealed silicon is shown in Fi 4.8. t4.111. The test devices were As+-implanted (100 keV, 4~101~'cm-~) p-n junction diodes, fabricated with the mesa structure, which were annealed with either a cw Ar-ion laser at a substrate temperature of 250' C or with a scanning electron microscope. Figure 4.8 shows a DLTS spectrum immediately after diode preparation. For cw laser annealing, the hole emission spectrum is dominated by a level at 0.45 eV, as seen previously. As shown with spectra (a) and (b), the density of this level increases rapidly with laser power P, where PM is the power required to melt the silicon.

N. M. JOHNSON

190 1

1

I

I

I

1

1

I

1

I

H(0.281

Si:B (100) 10-20S1-crn As+ 100 KeV 4 x

lo” ~ r n - ~

CW Ar+ LASER ANNEALED

~ = 3 rn 9s BIAS PULSE 0

-

10 V

H(0.45)

H(0.I) I

t

3 5 0 75

1

100

I

125

1

150

I

175

I

XK, TEMPERATURE (K)

t

I

I

225

250

275

FIGURE 4.7. Typical capacitance transient spectrum of ptype si implanted with As+ and annealed by cw Ar-ion laser. (N. H. Sheng and J. L. Merz, Ref. 4.10.) These laser powers are below that required to generate slip When the laser dislocations as discussed in Section 4.3.3. power i s sufficiently high to produce slip dislocation, the dominant defect level is found t o depend on substrate temperature and appears at 0.43 eV for a temperature of 250° C. The SEBA results for slip-free annealing reveal only a weak signal at E,++0.40 eV, which does not depend strongly on electron-beam power. A s discussed above, the 0.45 eV level i s not stable at room temperature and i s associated with interstitial iron. 4.3.2

Ream-Induced Defects

Recrystallization of implanted amorphous layers on singlecrystal silicon involves the epitaxial regrowth of the amorphous layers, with electrical activation of the implanted dopant, and removal of displacement damage in the near-surface region which remains crystalline during ion implantation. It has also been observed that beam irradiation of single-crystal silicon can induce or quench in electronic defects [ 4 . 9 , 4 . 1 2 , 4 . 1 3 ] . This is illustrated in Fig. 4.9 with the electron-emission spectrum for an electron-beam annealed Schottky-barrier diode. The diode was not implanted and received the same beam-annealing schedule used to recrystallize the As+-implanted layer in Fig. 4 . 6 . The electron-emission spectrum for a SEBA-annealed diode is compared


I

151

I

100

I

I

125

I

I

150

I

I

1

I

I

1

200 225 TEMPERATURE (K) 175

I

I

250

191

I

I

275

I

FIGURE 4.8. Capacitance transient spectrum of p-type Si implanted with As+ and annealed by either a cw Ar-ion laser (LA), using three different values of laser power, or a scanning elecTS is the substrate temperature during beam tron beam (SEBA). annealing. (N. H. Sheng et al., Ref. 4.11.)

to that for unannealed material. The spectrum for beam-annealed silicon is dominated by two emission peaks with activation energies of 0.19 and 0.44 eV. Shoulders on both peaks indicate additional unresolved emission centers, In the unannealed control, no emission peaks are detectable on the indicated scale of sensitivity. Similar quenched-in deep levels have been observed in cw laser irradiated silicon, where it was further determined that the low-temperature emission peak has an electric-field dependent activation energy with a zero-field value of 0.22 eV [4.14]. The 0.44 eV level has been ascribed to the phosphorus-vacancy complex [4.9,4.13-4.151. The spatial distributions of the quenched-in defects differ in electron-beam irradiated t4.91 and cw laserirradiated silicon t4.131, but in both cases the defects are located in the near-surface region of the silicon. The essential observation is that even without displacement damage from ion implantation the localized heating and rapid quenching, which are inherent features of beam recrystallization, can provide the stimulus for defect formation during laser or electron-beam annealing. However, it has been further observed, with cw laser annealing, that beam-induced damage can be strongly influenced by the crystalline quality of the substrate before laser annealing t4.121. In addition, near the point of melting,

N. M.JOHNSON

192 1.1

I

I

I

I

I

1

I

Schottky Diode Electron Traps h

> E

v

SEBA (Welder)

Unannealed

80 100

150

200

250

300 330

T (K) FIGURE 4 . 9 . CC-DLTS spectrum for electron emission in unimplanted scanned electron-beam annealed silicon. Also shown is the spectrum for the unannealed material. beam-induced slip dislocations are generated as discussed in the next section. 4.3.3

Minimization of Beam Recrystallization Defects

Critical optimization and control of annealing conditions To illustrate, are required for successful beam processing. with a scanned electron beam too low a beam current (for a given accelerating voltage) yields incomplete recrystallization of the implanted amorphous layer, while beam currents near that required for melting generate slip dislocations, which can be readily viewed in an optical microscope as a cross grid of slip lines over the annealed area. These extremes have also been documented in scanned laser annealed silicon. The photoluminescence efficiency of the recrystallized material varies dramatically depending upon the creation or suppression of dislocations. For example, beyond a critical exposure time, at a constant power near that required

4.

ELECTRONIC DEFECTS IN SILICON

193

-400 \

c: Q

-300

2

I-

-200

?? ;

t; W -100

01

I

I

I

I

(x6

0.7

0.8

a9

zj

10

ANNEALING LASER POWER, PIP,

FIGURE 4.1 0. Dependence of short-circuit photocurrent I hot0 in arbitrary units and sheet resistivity p on laser power P !or a diode implanted with 4x1Ol5 As+/cm2 and annealed with a 100 urn laser spot and 6 pm step. Laser power is normalized to the power required for melting, Po. (M. Muzita et al., Ref. 4.17.) for melting, slip appears and the luminescence is dominated by Such studies have dislocation-related defect levels [4.16]. further revealed that the range of laser power over which good quality annealed material can be achieved is limited [4.11,4.17], as further discussed below. Finally, electronic deep levels associated with dlslocations in deformed silicon have been detected by capacitance transient spectroscopy 14.181. Optimum annealing conditions for a cw Ar-ion laser are shown in Fig. 4.10 [4.17,4.19]. Single-crystal silicon with ptype conductivity was implanted with As+ at 100 keV to a dose of 4x1015 The laser annealing conditions were selected €or laterally uniform recrystallization, with laser power a variable. After annealing, the sheet resistance p and shortcircuit photocurrent Iphoto of mesa p-n junction diodes were measured over a range of laser powers P up to that required for melting, Po. The sheet resistance depends on majority-carrier properties, while the photocurrent is sensitive to those of the minority carriers. As shown in Fig. 4.10, for P > O.68Po the sheet resistivity achieves a minimum value of 45 O/square, corresponding to 100% electrical activation of the implanted arsenic.

N. M.JOHNSON

On the other hand, the relative magnitude of the photocurrent increases with laser power to a peak value at 0.76Po and then decreases at higher laser powers. The effect of laser power on electronic-defect density has been directly evaluated with DLTS i4.111. In Fig. 4.11 are plotted defect concentrations versus beam power for scanned-laser and scanned-electron beam annealing. The defect concentration of scanned-electron-beam-annealed diodes is lower than that of laserannealed diodes and is comparatively insensitive to electron-beam power. However, for laser-annealed diodes, the defect concentration increases more than one order of magnitude with increasing laser power. For high-power laser-annealed devices, the defect concentration is comparable to the substrate dopant Concentration. Results from EBIC are also shown in Fig. 4.11. The EBIC chargecollection efficiency decreases as the defect concentration increases. It was found in this study [4.111 that laser-induced defects can extend several microns below the implanted junction. I

L

o

t

I

I

A'"

I 4

V

z

\ A

\

W

w

-0.6 L L

LL W

-

= I

w U w

"

>-

\

2

0

k V

0 LA Ts=250"c H(0.45)

+-04

I

0.6

I

2

R

LA T,~250"c H(0.43) A SEBA Ts:25"c H(0.40) A EBlC ACCELERATING VOLTAGE : 30 KeV 0

I

a8 ANNEALING BEAM POWER, PIP,

I

I .o

FIGURE 4.11. Defect concentration measured by DLTS, as a function of annealing beam power. The broken curve displays the EBIC charge-collection efficiency as a function of laser power. [ N . H. Sheng et al., Ref. 4.11.1

4. 4.4


195

LATERAL NONUNIFORMITIES IN SCANNED-BEAM ANNEALING This section discusses some practical limits in optimizing

cw beam annealing conditions for minimum densities of residual

defects in the recrystallized silicon layer. First, the specific issue of lateral variations in defect density due to scan line separation is discussed with results from luminescence and electronbeam induced current measurements. Then the topic of scanned-beam annealing of silicon wafers with patterned dielectric overlayers is addressed with results from transmission electron microscopy. 4.4.1

Scan Line Separation

Defect luminescence has been used to evaluate cw laserannealed silicon [4.16,4.20,4.21]. In particular, it has been found that when the lateral separation between adjacent laser-scan lines is reduced, the luminescence intensity decreases [4.201. This is demonstrated in Fig. 4.12 with luminescence spectra for n-type Czochralski-grown silicon which was implanted with SiH+ at 80 keV to a fluence of 2x1015 cm-2 and recrystallized at a substrate temperature of 350' C with the scanned beam of a cw Ar-ion laser. Sample A was annealed at 5.2 W with a 40 pm diameter spot scanned at 12 cm/sec and with a 15 pm separation between lines. The two lower spectra correspond to samples laser annealed at 11.5 W, with an 80 pm diameter spot and a scan speed of 5 cm/sec. For sample B the separation of the scan lines was 30 um and for sample C it was 15 pm. Although the spectra are similar, the defect luminescence intensity varies by a factor of 30, with sample C giving the weakest signal. Inspection of the samples in an optical microscope showed that A and B have discernible lateral nonuniformity consisting of stripes between adjacent scan lines, which is not present in sample C . These samples were also evaluated with electron spin resonance. In each sample the characteristic dangling bond resonance of amorphous silicon [4.211 was observed with volume densities (calculated with an estimated thickness of 180 nm for the initial1 amorphized layer) of 4.I x ~ O ~1.~lx1017, , and 0.4x101x1 O1 cm-s for samples A, B, and C , respectively. Thus, as the lateral separation between adjacent scan lines was reduced, the luminescence intensity and spin density decreased and the nonuniformity of the surface was also reduced. These results indicate that the defects are associated with an imperfectly crystallized interface between adjacent laser annealed zones. It was suggested that the luminescence is possible due to extended defects such as dislocations or grain boundaries. Lateral nonuniformities in cw laser annealed silicon have been directly demonstrated with electron beam-induced currents

196

N . M. JOHNSON

FIGURE 4 . 1 2 . Luminescence s p e c t r a of n-type s i l i c o n , showi n g the e f f e c t of changing the l a s e r a n n e a l i n g c o n d i t i o n s . Details of t h e p r e p a r a t i o n of samples A, B y and C are g i v e n i n t h e text.

f 4 . 1 7 1 . A t y p i c a l EBIC image i s shown i n Fig. 4 . 1 3 . The sample was implanted w i t h As' a t 100 keV t o a f l u e n c e of 6x1Ol4 and annealed wich an Ar-ion l a s e r beam w i t h a 40 v m diameter s p o t A f t e r a n n e a l i n g , mesa diodes were f a b r i c a t e d and 30 vm s t e p . by s t a n d a r d p h o t o l i t h o g r a p h i c t e c h n i q u e s , and ohmic metal c o n t a c t s were formed on both s i d e s of t h e diode. The m e t a l l u r g i c a l p-n j u n c t i o n w a s l o c a t i o n 0.2-0.3 u m below t h e implanted s u r f a c e . The E B I C measurement was performed w i t h a 5 keV a c c e l e r a t i n g The E B I C v o l t a g e which r e s u l t e d i n a n e l e c t r o n range of -0.3 vm. image arises from t h e d i f f u s i o n of beam-generated m i n o r i t y c a r riers t o t h e j u n c t i o n where they d r i f t i n t h e e l e c t r i c f i e l d of t h e d e p l e t i o n l a y e r , y i e l d i n g an induced c u r r e n t i n t h e e x t e r n a l c i r c u i t . This c u r r e n t is monitored as the e l e c t r o n beam i s r a s t e r e d over t h e specimen t o produce an image which i s d i s p l a y e d on a cathode r a y tube. The upper and lower p o r t i o n s of Fig. 4 . 1 3 d i s p l a y t h e s t a n d a r d secondary e l e c t r o n image of a diode. In t h i s mode t h e laser-annealed s u r f a c e appears t o be smooth and defect free. However, when t h e E B I C mode i s switched on ( c e n t e r


197

FIGURE 4.13. EBIC micrograph ( c e n t e r ) and secondary emiss i o n image ( t o p and bottom) e x c i t e d by 5 keV beam. The sample w a s implanted w i t h 6x1014 As+/cm2 and annealed w i t h 40 pm s p o t I n t h e E B I C mode, dark regions correspond t o and 30 pm s t e p . Ref. 4.17 .] low charge c o l l e c t i o n . [M. Mizuta e t a1

.,

of F i g . 4.13), h i g h - c o n t r a s t dark l i n e s appear which are p a r a l l e l t o t h e d i r e c t i o n of t h e laser s c a n . The EBIC dark-line s p a c i n g i s of t h e o r d e r of t h e l a s e r - s c a n s t e p ; t h e d e v i a t i o n s from cons t a n t s p a c i n g are considered t o be a s s o c i a t e d w i t h complex o v e r l a p e f f e c t s [4.171. F u r t h e r , i t was found t h a t improved r e s u l t s were o b t a i n e d f o r a large-diameter beam (100 pm), small s c a n s t e p (6 pm), and slow scan speed (6 cm/sec), which i s c o n s i s t e n t w i t h t h e photoluminescence r e s u l t s d e s c r i b e d above. The d a r k - s t r i p e f e a t u r e s have a l s o been observed i n x-ray topographs of cw Ar-ion l a s e r - a n n e a l e d , s l i p - f r e e s i l i c o n , w i t h a one-to-one correspondence between t h e dark s t r i p e s i n t h e EBIC image and i n t h e x-ray topograph [4.22]. With TEM i t has been determined t h a t t h e d a r k - s t r i p e r e g i o n s c o n t a i n a h i g h d e n s i t y of It h a s been d i s l o c a t i o n loops due t o i n t e r s t i t i a l atoms [4.23]. f u r t h e r determined t h a t C02 laser a n n e a l i n g (at 10.6 pm wavel e n g t h ) does n o t d i f f e r s i g n i f i c a n t l y from cw Ar-ion laser anneali n g (0.514 pm wavelength) i n t h a t t h e u s e of a scanned h o t s p o t t o a n n e a l i m p l a n t a t i o n damage seems always t o be accompanied by t h e p r o d u c t i o n of r e s i d u a l i n t e r s t i t i a l d i s l o c a t i o n l o o p s , which act t o d e c r e a s e m i n o r i t y - c a r r i e r l i f e t i m e [4.24]; t h i s i s s u e i s f u r t h e r d i s c u s s e d i n S e c t i o n 4.4.2. The above i d e n t i f i e d l i m i t a t i o n s of raster-scanned p o i n t s o u r c e s can be overcome by u s i n g scanned-line s o u r c e s , where t h e l e n g t h of t h e l i n e s o u r c e i s g r e a t e r t h a n t h e maximum l a t e r a l dimension of t h e s u r f a c e area t o be r e c r y s t a l l i z e d .

198 4.4.2

N. M. JOHNSON

P a t t e r n e d Dielectric Overlayers

Most s t u d i e s of e l e c t r o n i c d e f e c t s i n cw beam-recrystall i z e d s i l i c o n have been conducted on material which was uniformly amorphized by i o n i m p l a n t a t i o n and r e c r y s t a l l i z e d over l a t e r a l dimensions l a r g e compared t o t h e examined area. Isolating the r e c r y s t a l l i z a t i o n process i n t h i s way s e r v e s t o f o c u s a t t e n t i o n on b a s i c m a t e r i a l s i s s u e s . However, t h e a c t u a l a p p l i c a t i o n t o integrated-circuit fabrication introduces additional constraints which must be addressed i f beam r e c r y s t a l l i z a t i o n i s t o r e p l a c e conventional furnace annealing. For example, i n i n t e g r a t e d c i r c u i t p r o c e s s i n g only small well-defined areas of a s i l i c o n wafer a r e i o n implanted t o h i g h doses, such as t h e s o u r c e and d r a i n c o n t a c t s i n a metal-oxide-semiconductor f i e l d - e f f e c t t r a n s i s t o r (MOSFET). The implanted area i s o f t e n d e f i n e d by a window etched i n a d i e l e c t r i c t h i n - f i l m o v e r l a y e r . An example i s shown i n Fig. 4.l(c). The p l a n a r p n j u n c t i o n diode i s bounded by a s i l i c o n - d i o x i d e l a y e r , which e l e c t r i c a l l y p a s s i v a t e s t h e s i l i c o n s u r f a c e . This is a more complex s t r u c t u r e f o r beam r e c r y s t a l l i z a t i o n s i n c e t h e oxide s t e p a t t h e edge of t h e implanted amorphous l a y e r a l t e r s both t h e s p a t i a l d i s t r i b u t i o n of energy d e p o s i t e d by t h e scanned beam and t h e temperature p r o f i l e ( o r heat: d i s s i p a t i o n ) d u r i n g r e c r y s t a l l i z a t i o n . Oxide-passivated p - n j u n c t i o n diodes have been examined a f t e r scanned electron-beam a n n e a l i n g w i t h a modified SEM [4.25]. In F i g . 4.14 i s shown a Nomarski i n t e r f e r e n c e o p t i c a l micrograph f o r an SEM-annealed r e g i o n of a n implanted amorphous l a y e r bounded by a s i l i c o n - d i o x i d e o v e r l a y e r . The s t a r t i n g material was Czochralski grown -oriented s i l i c o n w i t h p-type c o n d u c t i v i t y ( B , 1+3 Q-cm). A s i l i c o n - d i o x i d e l a y e r w a s thermally grown t o a t h i c k n e s s of 260 nm and p h o t o l i t h o g r a p h i c a l l y p a t t e r n e d and chemically e t c h e d t o expose areas of s i l i c o n f o r i o n i m p l a n t a t i o n . The e n t i r e wafer w a s t h e n implanted w i t h As+ at 100 keV t o a dose of lX1015 The SEM a n n e a l i n g c o n d i t i o n s were as f o l l o w s : beam v o l t a g e of 20 kV, l i n e - s c a n r a t e of 1 0 msec/line, frame-scan rate of 40 sec/frame, r a s t e r e d area of 1.10x1.15 mm, beam d i a m e t e r of 12+14 pm, s u b s t r a t e temperature of 25OC, and a beam c u r r e n t of 78 PA. The beam c u r r e n t w a s -90% of t h a t r e q u i r e d t o m e l t t h e exposed s i l i c o n . I n F i g . 4.14 t h e SEM-recrystallized r e g i o n of t h e implanted s i l i c o n is d i s t i n g u i s h e d from t h e amorphized a r e a by t h e d i f f e r e n c e i n r e f l e c t i v i t y . I n F i g . 4.15 i s shown a b r i g h t - f i e l d TEU micrograph of t h e boundary between a r e g i o n of implanted and r e c r y s t a l l i z e d s i l i c o n and an o x i d i z e d r e g i o n ; t h e r a s t e r e d e l e c t r o n beam overlapped t h e oxide l a y e r , and t h e oxide was removed f o r TEM. The dark band running d i a g o n a l l y a c r o s s t h e micrograph l o c a t e s t h e boundary between t h e r e c r y s t a l l i z e d and o x i d e - p a s s i v a t e d r e g i o n s and w a s i d e n t i f i e d a s a zone of amorphous material.

4.


199

FIGURE 4.14. Nomarski interference optical micrograph of an oxide-passivated SEM-annealed p-n junction in silicon.

In samples annealed at higher beam currents, the boundary contained a high density of dislocations. In a MOSFET such a defective zone would be particularly detrimental to device operation since it would be situated at the critical interface between the source or drain and the channel. Diodes fabricated in the above material were also evaluated by DLTS, a spectrum from which is The dominant hole trap has an activation shown in Fig. 4.16. ener y of 0.28 eV, as discussed in Section 4.3, with a density of -l0lf cm-3 at a depth of 1 pm below the metallurgical junction. No hole emission peak is detectable on the same scale of sensitivity in the furnace-annealed diode. The results presented above illustrate the added difficulties which can be encountered in the practical application of beam recrystallization in silicon integrated-circuit processing. It is not anticipated that nonuniform recrystallization arising from patterned dielectric overlayers will be alleviated with scannedline sources. However, this is an area of cw beam processing requiring further investigation.

200

N. M . JOHNSON

FIGURE 4 . 1 5 . Bright-field TEM micrograph of the boundary in single-crystal silicon between a region of exposed silicon ~ ~ and a region which was implanted with As+ (100 keV, I x ~ OcmV2) which was covered with a thermally grown layer of silicon dioxide (260 nm thick) during implantation and SEM recrystallization. The silicon-dioxide layer was removed for ‘ E M . 4.5

ISOTHERMAL TRANSIENT ANNEALING

A s reviewed in Chapter 1 , transient isothermal annealing techniques have recently been introduced to activate ion implanted dopants through solid-phase regrowth. The implanted silicon wafer is irradiated by an incoherent light source (e.g., tungsten halogen lamp, arc lamp, or graphite heater), and the damage is annealed on a time scale of seconds. The combination of largearea illumination and long dwell times (as compared to fast pulse or scanned laser annealing) insure that the wafer is heated essentially uniformly. The technique results in minimum redistribution of implanted dopants, complete electrical activation, and uniform annealing across the wafer. Residual electronic defects after transient isothermal annealing have also been investigated r 4 . 2 6 ,

4.271.

4.

I

I


I

I

I

E, = 0.28 eV (N,) = 1.6 X 1014 cm-3 (7100A < X, < 11,200A)

201

I

Cz Silicon P T W e *(loo) As+ Implanted Oxide-Passivated P-N Junctia Hole Traps e, = 347 sec-'

Furnace Anneal

(1000C, 30 min., N,)

I

1

0 100

I

150

I

200

I

250

I 300

1

3:

T (K)

FIGURE 4 . 1 6 . CC-DLTS spectrum for hole emission in an oxide-passivated SEM-annealed p-n junction diode. Also shown is the spectrum for a furnace-annealed control diode. A comprehensive study of electronic defects in silicon after transient isothermal annealing has recently been reported which includes a comparison of different annealing systems and clearly demonstrates the importance of quench rate on the density of residual defects [4.271. A typical trap-level spectrum for a boron-doped epitaxial silicon wafer after transient thermal annealing is shown in Fig. 4.17; the wafer was not ion implanted. After an anneal at 900°C for 20 seconds, a large DLTS peak appears at a temperature of -1 7 K. From the DLTS analysis, the trap concentration NT = 10" ~ m - ~the , trap ionization energy E E + 300 meV, and and the capture cross-section u = 2x10ZlS cg2; this appears to be the same as the % + 0.28 eq level discussed in Section 4.3. The defect level is not detectable in the starting (reference) material, but is present in low concentration after a conventional furnace anneal. In unimplanted Czochralskigrown and float zone grown silicon, the level was also not detectable after the transient thermal annealing (i.e., NT < 1x10'' 100 keV) p-n while in arsenic-implanted ( 5 ~ 1 0 1As+ ~ junction diodes fabricated in Czochralski-grown silicon, the level is resent after transient annealing at a concentration of 5x1Ol2 e m ' . Transient thermal annealing of n-type (phosphorus-doped)

N. M.JOHNSON

202

Epi -Silicon (111) PF

p = 2 Qcm

0.9

*

(3

-

0.7

-

NT= 1 x 1 015cm-3 ET = E, +300me V

cn

0.5

5 0

ITA : T=9OO0C,t = 2 0 s

-

-J

a

DLTS : tl =lms t, = 2 m s

p -Type / Boron

furnace anneaiing (900O C , 30 min)

0.3 0.1

0

Reference

I

1 3

l

r

i

160

l

r

r

l

r

l

l

r

l

200 240 TEMPERATURE

l

l

l

280

l

I 1

K

FIGURE 4.17. DLTS spectrum f o r a boron-doped e p i t a x i a l Also s i l i c o n sample a f t e r t r a n s i e n t thermal annealing ( I T A ) . shown a r e s p e c t r a f o r a f u r n a c e annealed c o n t r o l sample and f o r t h e unannealed ( r e f e r e n c e ) m a t e r i a l . No s i g n i f i c a n t DLTS s i g n a l s i l i c o n wafers was a l s o i n v e s t i g a t e d . w a s d e t e c t e d between 50 K and 300 K, from which i t w a s concluded t h a t no e l e c t r i c a l l y a c t i v e t r a p l e v e l s w i t h c o n c e n t r a t i o n s cm-3 a r e induced i n t h e upper h a l f of t h e g r e a t e r than s i l i c o n bandgap under i s o t h e r m a l t r a n s i e n t a n n e a l i n g c o n d i t i o n s . The dependence of t r a p c o n c e n t r a t i o n on t h e p r i n c i p a l In Fig. 4.18 annealing parameters was a l s o i n v e s t i g a t e d [4.27]. i s shown t h e e f f e c t of a n n e a l i n g temperature on t h e t r a p concent r a t i o n f o r t h r e e d i f f e r e n t a n n e a l i n g systems. The samples i n t h e SEL and AG 210 annealing systems were cooled a t a r a t e of -lOO°C/sec; t h e cooling rate i n t h e Varian I A 200 system was not c o n t r o l l a b l e . A l l t h r e e curves peak at a temperature of -lOOO°C. The r e l a t i v e displacements of t h e t h r e e curves i s considered t o be p r i m a r i l y due t o d i f f e r i n g c o o l i n g rates f o r samples i n t h e t h r e e systems, even f o r nominally t h e same s p e c i f i e d r a t e . This i s f u r t h e r d i s c u s s e d below. The t r a p c o n c e n t r a t i o n s induced a f t e r i s o t h e r m a l t r a n s i e n t annealing i n a s i n g l e system w i t h d i f f e r e n t ramp down rates is

4.

cni3

I-

I

Z

0

SEL

ITA: -

X'

-

2 oz

A G 2'0

I-

z

,,"O\

w

\

0

d

z

t =20s fast ramp down (50 - 150 'C/S)

iSEL AG 210 1

ramp down not controllable

VA IAN IA200

\

# ,\'

0

203

Epi-Silicon (111)

-

Z


\ \

1

\

0 0

\

\

\ 1

-

n 4 oz

-

I-

-

VARIAN I A 2 0 0 4

\ \

1 \ \

\ \

13

0

FIGURE 4.18. Trap c o n c e n t r a t i o n of t h e 300 meV l e v e l v e r s u s a n n e a l i n g t e m p e r a t u r e . The t h r e e curves are measured w i t h d i f f e r e n t systems as i n d i c a t e d . shown i n F i g . 4.19. The rates of sample cooling are not w e l l d e f i n e d because of problems i n t h e measurement of sample temp e r a t u r e under t r a n s i e n t c o n d i t i o n s . The dashed curve ( s o l i d circles) and s o l i d curve (open c i r c l e s ) were taken on s e p a r a t e samples from t h e same w a f e r . The open circles r e p r e s e n t measurement p o i n t s taken a t a slow ramp down of 5OC/sec, and t h e s o l i d circles are p o i n t s t a k e n at a t e n times f a s t e r r a t e . Additiona l l y , f u r n a c e a n n e a l s were performed f o r comparison. The r e s u l t s r e v e a l t h a t t h e production of t r a p s i s s t r o n g l y reduced by a slow ramp down of t h e temperature, independent of t h e a n n e a l method. F u r t h e r o b s e r v a t i o n s on t h e behavior of t h e 300 meV l e v e l are as f o l l o w s [4.27] : The d e f e c t c o n c e n t r a t i o n d e c r e a s e s w i t h i n c r e a s i n g d i s t a n c e from t h e s u r f a c e , which may be caused by t h e temperature g r a d i e n t d u r i n g c o o l down o r may i n d i c a t e t h a t t h e s u r f a c e i s a p o s s i b l e s o u r c e of t h e d e f e c t . The l e v e l can be removed by a f u r n a c e a n n e a l a t temperatures above 400OC. The t r a p l e v e l i s induced by i s o t h e r m a l t r a n s i e n t a n n e a l i n g a f t e r

N . M. JOHNSON

204

Epi-Siiicon (111) p -Type/ Boron p = 2 Qcm

I-

z z

n: t 520s * = fast ram down (SO0 C I S ! o = slow ramp down

F

-

0

(5"C/S)

I-

-E

I

I

1

Polycrystaliine Silicon

I

-10

-

N

400 -CW Laser Crystallized MOS Capacitor, P-Type

I

300-
cD1. A s s e e n i n F i g . 4.29, t h e r e s u l t s are ( 1 ) a lower peak h e i g h t and ( 2 ) t h e peak s h i f t s t o a lower t e m p e r a t u r e (i.e., TD2 < T D ~ ) . I n c r e a s i n g CD reduces t h e magnitude of t h e s i l i c o n s u r f a c e p o t e n t i a l w i t h c o r r e s p o n d i n g d e c r e a s e s i n t h e width of t h e s i l i c o n space-charge l a y e r and i n t h e magnitude of t h e e l e c t r i c - f i e l d i n t e n s i t y a t every p o i n t


219

within this depletion layer 14.601. If the emission peak arose from a discrete level spatially distributed in the bulk of the silicon, then the reduced depletion width would mean fewer defect centers contributing to the emission signal and consequently a lower peak height, as is observed. However, this would not cause the temperature of the emission peak to shift. For hole emission from a bulk discrete level, the change in the electric field would have to mediate the shift of the peak temperature when CD is changed. This follows from the fact that the trap filling conditions were identical for the two spectra and from the experimental determination that the DLTS signals which comprise the spectra were saturated with respect to both gate-voltage pulse amplitude and pulse width. Thus, it is the emission cycle, rather than the capture cycle, that is being affected by changing CD. For hole emission from a discrete level, the peak appears at a temperature T which is related to the DLTS delay times and trap parameters as follows: (4.3) where e-l is the hole emission time, y is the attempt-to-escape P frequency for trapped holes, EA is t\e activation energy for thermal emission, and k is the Boltzmann constant. Since the right-hand side of Eq. (4.3), which is commonly termed the DLTS emission rate window (as discussed in Section 4.2.2), is a constant, a change in Tp must be compensated by a change in EA and/or up. Furthermore, since Tp depends logarithmically on yp, a strong functional dependence of yp on field would be required, which is not generally observed for discrete deep levels. However, electric-field modulation of the activation energy is a welldocumented phenomenon. Specifically, Poole-Frenkel emission from a neutral center, which for hole emission would be an acceptorlike defect, yields the following field-dependent activation energy (for a one-dimensional coulombic defect potential) 14.611:

where F is the magnitude of the electric field, Eo is the zerofield activation energy, q is electronic charge, and E~ is the silicon permittivity. (A defect center that is neutral when empty and chargedwhen occupied will not experience this effect, as discussed In Section 4.3.2.) Equation (4.4) predicts that the activation energy increases with decreasing electric field. From Eq. (4.31, this requires that the peak temperature increase, rather than decrease, with an increase in CD. Another mechanism which would yield the same qualitative conclusion is high-f ield tunnel emission which has been documented for several point defects in 111-V semiconductors (see, €or example, Ref. 4.62). Thus, the

N. M . JOHNSON experimental r e s u l t s are i n c o n s i s t e n t w i t h f i e l d - m o d u l a t e d emission from a d i s c r e t e l e v e l .

hole

The dependence on CD of t h e high-temperature emission peak i n Fig. 4.29 i s c o n s i s t e n t w i t h t h e e f f e c t of s u r f a c e g e n e r a t i o n on t h e DLTS measurement of i n t e r f a c e s t a t e s n e a r t h e s i l i c o n mid-gap [4.55]. In t h i s c a s e , a t t e m p e r a t u r e s corresponding t o t h e DLTS d e t e c t i o n of m a j o r i t y - c a r r i e r e m i s s i o n ( i . e . , h o l e s ) from i n t e r f a c e states n e a r midgap, m i n o r i t y - c a r r i e r emission from t h o s e s t a t e s becomes a c o m p e t i t i v e p r o c e s s and can i n t r o d u c e a peak i n a DLTS spectrum due t o s u r f a c e g e n e r a t i o n of charge through a slowly v a r y i n g continuous deep-level d i s t r i b u t i o n . The dependence on CD i s due t o t h e change of t h e s u r f a c e p o t e n t i a l w i t h gate voltage. The s u r f a c e p o t e n t i a l d e t e r m i n e s t h e i n t e r s e c t i o n of t h e Fermi l e v e l w i t h t h e i n t e r f a c e which s p e c i f i e s t h e e q u i l i b rium o c c u p a t i o n of t h e i n t e r f a c e s t a t e s ; s t a t e s above t h e Fermi l e v e l do not change occupancy d u r i n g t h e c a p a c i t a n c e t r a n s i e n t ( i n p-type c a p a c i t o r s ) and t h e r e f o r e cannot c o n t r i b u t e t o t h e DLTS s i g n a l . Reducing t h e magnitude of t h e s u r f a c e p o t e n t i a l , by i n c r e a s i n g t h e e q u i l i b r i u m d e p l e t i o n c a p a c i t a n c e , pinches o f f t h e s u r f a c e g e n e r a t i o n p r o c e s s which both a t t e n u a t e s t h e e m i s s i o n peak and s h i f t s i t t o lower t e m p e r a t u r e s , a s d i s p l a y e d i n Fig. 4.29. Based on t h e arguments p r e s e n t e d above, t h e e n t i r e DLTS spectrum i n Fig. 4.29 may be a s c r i b e d t o a continuous d i s t r i b u t i o n Then t h e i n t e r f a c e - s t a t e of deep l e v e l s a t t h e Si-SiO2 i n t e r f a c e . d e n s i t y can be e s t i m a t e d w i t h a s t a n d a r d a n a l y s i s from t h e DLTS s i g n a l a t t e m p e r a t u r e s below t h e emission peak where h o l e emission dominates 14. 5 51 . For example, t h e DLTS s i g n a l j u s t below t h e peak ( i . e . , n e a r 210 K) y i e l d s a v a l u e of approximately 6x1010 eV-l cm-2 f o r t h e d e n s i t y of i n t e r f a c e s t a t e s a t e n e r g i e s n e a r t h e s i l i c o n midgap. Against t h i s background s i g n a l from i n t e r f a c e s t a t e s , t h e d e n s i t y of a s p a t i a l l y uniform d i s c r e t e l e v e l would have t o be > lx1014 cm-3 t o be d e t e c t a b l e i n t h e c a p a c i t o r s examined i n t h i s study. To conclude, t h e r e s u l t s of t h i s s t u d y provide a d e f e c t - s p e c i f i c v e r i f i c a t i o n of t h e device-grade q u a l i t y of C02 l a s e r - c r y s t a l l i z e d s i l i c o n t h i n f i l m s on f u s e d s i l i c a .

4.8

SUMMARY AND CONCLUSIONS

From t h e e x p e r i m e n t a l s t u d i e s reviewed i n S e c t i o n s 4.34.6, t h e f o l l o w i n g g e n e r a l o b s e r v a t i o n s may be drawn r e g a r d i n g e l e c t r o n i c d e f e c t s i n cw t r a n s i e n t thermal annealed bulk s i n g l e crystal silicon: 1.

R e s i d u a l e l e c t r o n i c d e f e c t s remain a f t e r t r a n s i e n t t h e r m a l a n n e a l i n g w i t h a l l i n v e s t i g a t e d forms of energy sources. The d e f e c t d e n s i t i e s are i n e x c e s s of t h o s e o b t a i n e d by c o n v e n t i o n a l f u r n a c e a n n e a l i n g .


221

2.

These defects are located in the near-surface region. In recrystallized silicon the defects extend with decreasing density into material not driven amorphous by ion implantation.

3.

Even in material which has not been ion implanted, electronic defects are quenched in during transient thermal annealing.

4.

Beam-annealing conditions (e.g., beam power and scan line separation) can be optimized for recrystallizing implanted amorphous layers with a minimum density of residual defects. However, the use of a raster-scanned point source to anneal implantation damage seems always to be accompanied by the production of laterally-distributed residual defects.

5.

Laterally nonuniform recrystallization can also result from wafer topography (e.g., patterned dielectric overlayers), which introduces lateral variations in energy deposition and heat dissipation during beam recrystallization.

6.

Post-recrystallization processing can be used to reduce the density of residual defects. Furnace anneals in the range of 400-700°C and hydrogen passivation at temperatures of 100-3OO0C have been shown to be effective in removing deep levels.

From the standpoint of electronic defects, of the several cw transient thermal processing techniques that have now been investigated, isothermal transient annealing is the most promising as a replacement for conventional furnace annealing in silicon integrated-circuit technology. Of course, this technique most closely approaches a furnace anneal in its thermal treatment of a silicon wafer. This is further reflected in the similarity of the deep levels which result from these two annealing procedures and in the convergence of their residual defect densities as the quench rate in isothermal transient annealing is reduced. On the other hand, directed energy sources, such as scanned laser and electron beams, may offer unique advantages for materials modifications other than recrystallization as well as provide a controlled means of introducing defects in silicon for fundamental studies. For beam-crystallized silicon thin films on amorphous substrates, the results presented in Section 4.7 indicate that the Si-SiO2 interface is the dominant source of residual electronicdefects in this materials system. Since the measurements were performed on MOS devices in which the interface was a source of

222

N. M.JOHNSON

d e t e c t a b l e d e f e c t s , i n t e r f a c e s t a t e s e s t a b l i s h e d a h i g h background s i g n a l f o r d e t e c t i n g d e f e c t s i n the bulk of t h e s i l i c o n t h i n f i l m s . This background r e q u i r e d t h a t t h e d e n s i t y of a s p a t i a l l y uniform bulk d e f e c t be g r e a t e r t h a n t y p i c a l l y lx1014 cm-3 t o be d e t e c t a b l e . T h i s e f f e c t i v e s e n s i t i v i t y t o bulk d e f e c t s i s r a t h e r low f o r s t u d i e s d i r e c t e d toward t h e c o r r e l a t i o n of r e s i d u a l Novel d e f e c t s w i t h materials p r o c e s s i n g and device o p e r a t i o n . deep-level s p e c t r o s c o p i c t e c h n i q u e s can be combined w i t h m u l t i p l e t r a n s i s t o r s ) t o a c h i e v e high s e n s i t e r m i n a l MOS d e v i c e s (e.g., t i v i t y f o r bulk d e f e c t d e t e c t i o n , i n p a r t by e x c l u d i n g any conIt may be a n t i c i p a t e d t h a t t r i b u t i o n from t h e i n t e r f a c e 1 4 . 6 3 1 . such t e c h n i q u e s and d e v i c e s w i l l be extended t o e l e c t r o n i c - d e f e c t c h a r a c t e r i z a t i o n i n beam-crystallized s i l i c o n t h i n f i l m s .


223

REFERENCES 4.1 4.2

4.3

4.4 4.5

4.6

4.7

4.8 4.9

4.10 4.11 4.12

4.13 4.14

4.15 4.16 4.17 4.18 4.19

J. Appl. Phys. 45, 3014 and 3023 (1974). Lang, D. V., Johnson, N. M., B a r t e l i n k , D. J., Gold, R. B., and J . F. Gibbons, J. Appl. Phys. 50, (1979). L e f e v r e , H. and Schulz, H., Appl. Phys. 12, 45 (1977). Johnson, N . M., Gold, R. B., and J. F. Gibbons, Appl. Phys. L e t t . 34, 704 (1979). Gibbons, J. F., Johnson, W. S., and Mylroie, W. W., "Proj e c t e d Range S t a t i s t i c s , Semiconductors and R e l a t e d Mate r i a l s , " 2nd ed. Halstead, New York (1975). Gold, R. B., L i e t o i l a , A . , and Gibbons, Johnson, N. M., J. F., "Laser-Solid I n t e r a c t i o n s and Laser P r o c e s s i n g 1978," (S. D. F e r r i s , H. J. Leamy, and J. M. Poate, eds.) AIP Conference Proceedings, pp. 550-555. American I n s t i t u t e of P h y s i c s , New,York (1979). R e g o l i n i , J. F., Gibbona, J. F., Sigmon, T. W., Pease, R. F. W., Magee, T. J., and Peng, J., Appl. Phys. L e t t . 34, 410 (1979). B a r t e l i n k , D. J., Ratnakumar, K. N . , Pease, R. F. W., Johnson, N . M., and Meiqdl, J. D., Appl. Phys. L e t t . 35, 463 (1979). Johnson, N. M., R e g o l i n i , J. L., B a r t e l i n k , D. J., Gibbons, J. F., and Ratnakumar, K. N.,. Appl. Phys. L e t t . 36, 425 (1980). Sheng, N. M. and Merz, J. L., "Laser and E l e c t r o n Beam Int e r a c t i o n s w i t h S o l i d s , " (B. R. Appleton and G. K. Celler, eds.) pp. 313-318. E l s e v i e r , New York (1982). Sheng, N. H., Mizuta, M., and Merz, J. L., Appl. Phys. L e t t . 4 0 , 68 (1982). Mizuta, M., and Merz, J. L., "Laser and Sheng, N. H., Electron-Beam S o l i d I n t e r a c t i o n s and Materials Processi n g , " ( J . F. Gibbons, I,. D. Hess, and T. W. Sigmon, eds.) pp. 155-162. E l s e v i e r , New York (1981). Chantre, A . , Kechouane, M., and Bois, D., "Laser and Electron-Beam I n t e r a c t i o n s w i t h S o l i d s , " (B. R. Appleton and G. K. Celler, eds.) pp. 325-330. E l s e v i e r , New York (1982). Chantre, A., Kechouane, and Boise, L, "Defects i n Semiconductors 11," (S. Mahajan and J. W. C o r b e t t , eds.), pp. 547-551. E l s e v i e r , New York (1983). IEEE Trans. Nucl. S c i . NS-23, 1497 Kimerling, L. C., ( 1976). Uebbing, R. H., Wagner, P., Baumgart, H., and Q u e i s s e r , H. J., Appl. Phys. L e t t . 37, 1078 (1980). Mizuta, M., Sheng, N. H., Merz, J. L., L l e t o i l a , A., Gold, R. B., and Gibbons, J. F., Appl. Phys. L e t t . 37, 154 (1980). Klmerling, L. C. and P a t e l , Appl. Phys. L e t t . 34, 73 (1979). Sheng, N. H., and Merz, J. L., Appl. Phys. Mizuta, M., L e t t . 38, 453 (1981).

224 4.20

4.21

4.22

4.23

4.24 4.25 4.26

4.27

4.28 4.29

4.30

4.31 4.32 4.33 4.34

N. M . JOHNSON S t r e e t , R. A , , Johnson, N. M., and L i e t o i o l a , A., "Laser and Electron-Beam P r o c e s s i n g of Materials," (C. W. White Academic P r e s s , New York and P. S. Peercy, eds.) p. 435. (1980). S t r e e t , R. A., Johnson, N. M., and Gibbons, J. F., J. Appl. Phys. 50, 8201 (1979). and P h i l l i p p , F., "Laser Rozgonyi, G. A., Baumgart, H., and Electron-Beam S o l i d I n t e r a c t i o n s and M a t e r i a l s Processi n g , " ( J . F. Gibbons, L. D . Hess, and T. W. Sigmon, e d s . ) , pp. 193-199. E l s e v i e r , New York (1981 ) Baumgart, H., Hildebrand, O., P h i l l i p p , F a , and Rozgonyi, G . A , , "Proceedings of 2nd Oxford Conference on Microscopy of Semiconductor Materials." I n s t i t u t e of P h y s i c s , London (1981). Baumgart, H., P h i l l i p p , F., and Leamy, H. J., "Laser and Electron-Beam I n t e r a c t i o n s w i t h S o l i d s , " (B. R. Appleton E l s e v i e r , New York (1982). and G. K. Celler) pp. 355-360. Johnson, N. M., S i n c l a i r , R., and Moyer, M. C., IEEE Trans. Elec. Dev. ED-27, 2199 (1980). Benton, J . L., C e l l e r , G. K., Jacobson, A. C., Kimerling, L. C., Lischner, D. J., Miller, G. L., and Robinson, McC., "Laser and Electron-Beam I n t e r a c t i o n s w i t h S o l i d s , " (R. R. Appleton and G. K. C e l l e r , eds.) pp 765-770. E l s e v i e r , New York (1982). P e n s l , G., S c h u l z , M., Johnson, N. M., Gibbons, J. F., and Hoyt , J., "Energy Beam-Solid I n t e r a c t i o n s and T r a n s i e n t Thermal P r o c e s s i n g , " (J. C. C. Fan and N. M. Johnson, eds.) i n p r e s s . E l s e v i e r , New York (1984). "Radiation E f f e c t s i n Semiconductors Kimerling, L. C., 1979." I n s t . Phys. Conf. S e r . 31, B r i s t o l (1977). Johnson, N. M., B a r t e l l i n k , D . J., Moyer, M. D., Gibbons, J. F., L i e t o i l a , A., Ratnakumar, K. N., and R e g o l i n i , J . L., "Laser and Electron-Beam P r o c e s s i n g of Materials," (C. W. White and P. S . Peercy, eds.) pp. 423-429. Academic Press, New York (1980). Benton, J. L., Doherty, C. J., F e r r i s , S . D., Kimerling, L. C., Leamy, H. J., and Celler, G. K., "Laser and E l e c t r o n Beam P r o c e s s i n g of Materials," (C. W. White and P. s. Peercy, eds.) pp. 430-434. Academic P r e s s , New York (1980). Kimerling, L. C. and Benton, J. L., "Laser and E l e c t r o n B e a m P r o c e s s i n g of Materials,'* (C. W. White and P. s. Peercy, eds.) pp. 385-396. Academic Press, New York (1980). Benton, J. L., Doherty, C. J., F e r r i s , S . D., Flamm, D. L., Kimerling, L. C., and Leamy, H. J., Appl. Phys. L e t t . 36, 670 (1980). Johnson, N. M., B i e g e l s e n , D . K., and Moyer, M. D., Appl. Phys. L e t t . 38, 900 (1981). Maby, E. W., Atwater, H. A., Keigler, A. L., and Johnson, N. M., Appl. Phys. Lett. 43, 482 (1983).

.

4. 4.35 4.36

4.37

4.38

4.39 4.40

4.41

4.42 4.43

4.44 4.45 4.46

4.47

4.48 4.49 4.50 4.51

4.52


225

Kamins, T. I. and Von Herzen, B. P., IEEE Elec. Dev. L e t t . EDL-2, 313 ( 1 9 8 1 ) . Ng, K. K., Celler, G. K., P o v i l o n i s , E. I., Frye, R. C., Leamy, H. J., and Sze, S. M., IEEE Elec. Dev. L e t t . EDL-2, 316, ( 1 9 8 1 ) . Frye, R. C. and Ng, K. K., "Grain Boundaries i n Semicond u c t o r s , " (C. H. Seager, G. E. P i k e , and H. J . Leamy, eds.) E l s e v i e r , New York ( 1 9 8 2 ) . pp. 275-286. B i e g e l s e n , D. K., Johnson, N. M., Nemanich, R. J., Moyer, M. D., and Fennel, L. E., "Laser and Electron-Beam I n t e r a c t i o n s w i t h S o l i d s , " (B. R. Appleton and G. K. Celler, eds.) pp. 331-336. E l s e v i e r , New York ( 1 9 8 2 ) . Fan, J. C. C., Geis, M. W., and Tsaur, B.-Y., Appl. Phys. L e t t . 3 8 , 365 ( 1 9 8 1 ) . Maby, E. W., Geis, M. W., LeCoz, L. Y., S i l v e r s m i t h , D. J. Mountain, R. W., and A n t o n i a d i s , D. A., IEEE Elec. Dev. Lett. EDL-2, 241 ( 1 9 8 1 ) . Geis, M. W., Smith, H. I., Tsaur, B.-Y., Fan, J. C. C., Maby, E. W., and A n t o n i a d i s , D. A . , Appl. Phys. L e t t . 4 0 , 158 ( 1 9 8 1 ) . Hawkins, W. G., Black, J. G., and G r i f f i t h s , C. H., Appl. Phys. L e t t . 4 0 , 319 ( 1 9 8 2 ) . F e n n e l l , L. E., Moyer, M. D., B i e g e l s e n , D. K., Chiang, A., and Johnson, N. M., "Energy Beam-Solid I n t e r a c t i o n s and T r a n s i e n t Thermal P r o c e s s i n g , " (J. C. C. Fan and N. M. Johnson, eds.). E l s e v i e r , New York ( 1 9 8 4 ) i n p r e s s . A t w a t e r , H. A., Thompson, C. V., Smith, H. I., and Geis, M. W., Appl. Phys. L e t t . 4 3 , 112 ( 1 9 8 3 ) . Johnson, N. M., Moyer, M. D., and F e n n e l l , L. E., Appl. Phys. L e t t . 4 1 , 562 ( 1 9 8 2 ) . Johnson, N. M., Moyer, M. D., F e n n e l l , L. E., Maby, E. W., and A t w a t e r , H., "Laser-Solid ' I n t e r a c t i o n s and T r a n s i e n t Thermal P r o c e s s i n g of Materials," ( J . Narayan, W. L. Brown, E l s e v i e r , New York and R. A. Lemons, eds.) pp. 491-497. (1983). Johnson, N. M. and Moyer, M. D., "Comparison of Thin Film T r a n s i s t o r s and SO1 Technologies," (H. W. Lam and M. J. Thompson, eds.) E l s e v i e r , New York ( 1 9 8 4 ) i n p r e s s . Wiley, J. D. and Miller, G. L., IEEE Trans. Elec. Dev. EDL-22, 265 ( 1 9 7 5 ) . Kamins, T. I., Lee, K. F., and Gibbons, J. F., IEEE Elec. Dev. L e t t . EDL-1, 5 ( 1 9 8 0 ) . Seager, C. H., P i k e , G. E., and Ginley, D. S., Phys. Rev. L e t t . 4 3 , 532 ( 1 9 7 9 ) . Werner, J., J a n t s c h , W., Froehner, K. H., and Q u i e s s e r , H. J., "Grain Boundaries i n Semiconductors, " (C. H. Seager , G. E. P i k e , and H. J. Leamy, eds.). E l s e v i e r , New York (1982). N i c o l l i a n , E. H. and B r e w s , J. R., "MOS Physics and Technology." Wiley, New York ( 1 9 8 2 ) .

N. M.JOHNSON

4.53 4.54

4.55 4.56 4.57 4.58

4.59

4.60 4.61 4.62

4.63

G a t , A . , Gerzberg, L., Gibbons, J. F., Magee, T. J., Peng, Appl. Phys. L e t t . 33, 775 (1978). and Hong, J. D., Johnson, N. M., B i e g e l s e n , D. K., and Moyer, M. D., "Laser and Electron-Beam S o l i d I n t e r a c t i o n s and Materials Processi n g , " (J. F. Gibbons, L. D. Hess, and T. W. Sigmon, eds.) pp. 463-470. E l s e v i e r , New York (1981). J. Vac. Sci. Techn. 2 1 , 303 (1982). Johnson, N. M., Johnson, N. M., B i e g e l s e n , D. K., and Moyer, M. D., J. Vac. Sci. Techn. 1 9 , 390 (1981). P i n i z z o t t o , R. F., Lam, H. W., and Vaandrager, B. L., Appl. Phys. Lett. 40, 388 (1982). Johnson, N. M., B i e g e l s e n , D. K., Tuan, H. D., Moyer, M. D., and F e n n e l l , L. E., I E E E Elec. Dev. L e t t . EDL-3, 369 (1982). Chiang, A . , Z a r z y c k i , M. H., Meuli, W. P., and Johnson, N. M., "Energy B e a r S o l i d I n t e r a c t i o n s and T r a n s i e n t Thermal ElP r o c e s s i n g , " (J. C. C. Fan and N. M. Johnson, eds.). s e v i e r , New York (1984). "Physics of Semiconductor Devices," 2nd ed. Sze, S . M., Chap. 7. Wiley, New York (1981). Milnes, A. G., "Deep I m p u r i t i e s i n Semiconductors," Chap. 5. Wiley, New York (1973). Makram-Ebeid, S., "Defects i n Semiconductors," (J. Narayan, E l s e v i e r , New York and T. Y. Tan, eds.) pp. 495-501. (1981). C o l l e t , M. G., S o l i d - s t a t e Elec. 1 8 , 1077 (1975).

CHAPTER 5

Beam Recrystallized Polycrystalline Silicon: Properties, Applications, and Techniques K. F. Lee", T. J . Stultzt and James F. Gibbons STANFORD ELECTRONICS LABORATORIES STANFORD UNIVERSITY STANFORD. CALIFORNIA

5.1 RECRYSTALLIZATION OF THINPOLYCRYSTALLINE FILMSWITH A SCANNING cw LASER . . . . . . . . . . . . . . . . . . 5 . 2 RESISTIVITYREDUCTIONIN BEAM-RECRYSTALLIZED

228

.................................

247

PROPERTIES OF LASERRECRYSTALLIZED 5.3 ELECTRONIC FILMSINTENDED FOR MOS DEVICE FABRICATION ...................................... AND 5.4 CHARACTERISTICS OF MOS DEVICES INTEGRATED CIRCUITS FABRICATED ON LASER

250

POLYSILICON FILMS

RECRYSTALLIZEDPOLYSILICON FILMS. . . . . . . . . . . . . . . . . .

260

5.5 IMPROVEMENTS IN THE RECRYSTALLIZATION

PROCESS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

278 335

Heavily doped polycrystalline silicon (polysilicon) is a material that is widely used in present day silicon integrated circuit technology for gates and interconnection lines in MOS integrated circuits. Initial interest in the beam annealing of polysilicon arose because of its potential as a process that could be used to obtain resistivity reduction in as-deposited films. Experiments undertaken to explore that potential are described in Sec. 5.2. These experiments showed that beamrecrystallized polysilicon films have electronic properties that closely approximate those of single crystal material. This result ultimately led to successful attempts to fabricate MOS transistors and integrated circuits directly in beam recrystallized polysilicon 15.1-5.41. A substantial interest now exists in the potential of beam recrystallized silicon-on-insulators (SOI) as a substrate for integrated circuit fabrication. Our development of this subject will begin with a discussion in Sec. 5.1 of the mechanisms by which thin films of polycrystalline Si are recrystallized by a scanning circular cw laser beam. In Sec. 5.2 we will consider the basic majority carrier electrical properties of the recrystallized material and establish the limits *Present address: AT&T Bell Laboratories. Holmdel, New Jersey 07733 ?Present address: TS Associates, San Jose, California 95128. SEMICONDUCTORS AND SEMIMETALS, VOL. 17

227

Copynght 0 1984 by Academic Press, Inc. All rights of reproduction in any form reserved. ISBN 0-12-752117-8

K . F. LEE ET AL. In on r e s i s t i v i t y r e d u c t i o n t h a t are p o s s i b l e w i t h t h i s p r o c e s s . Sec. 5.3 w e w i l l d e s c r i b e t h e b a s i c e l e c t r o n i c p r o p e r t i e s of rec r y s t a l l i z e d f i l m s i n t e n d e d f o r u s e as MOSFET s u b s t r a t e s , l e a d i n g t o a d i s c u s s i o n of t h e e l e c t r i c a l c h a r a c t e r i s t i c s of d e v i c e s fabWe conclude t h e r i c a t e d d i r e c t l y i n t h e s e f i l m s i n Sec. 5.4. c h a p t e r w i t h a d i s c u s s i o n of a v a r i e t y of improvements i n t h e r e c r y s t a l l i z a t i o n p r o c e s s t h a t have been developed t o eliminate v a r i o u s d e f e c t s i n d e v i c e and f i l m p r o p e r t i e s t h a t are o b t a i n e d when a c i r c u l a r beam i s used f o r t h e r e c r y s t a l l i z a t i o n p r o c e s s .

5.1

RECRYSTALLIZATION SCANNING CW LASER

OF

THIN

POLYCRYSTALLINE

FILMS

WITH A

I n t h i s s e c t i o n w e d e s c r i b e f i r s t t h e a n n e a l i n g of t h i n polys i l i c o n f i l m s i r r a d i a t e d by a c w argon l a s e r , w i t h emphasis on t h e r e l a t i o n between l a s e r power and t h e s u r f a c e morphology, c r y s t a l s t r u c t u r e , dopant d i s t r i b u t i o n and m a j o r i t y c a r r i e r e l e c t r i c a l p r o p e r t i e s of t h e annealed f i l m s . Three laser power l e v e l s (low, medium, and h i g h ) were i d e n t i f i e d i n e a r l y s t u d i e s d i r e c t e d toward u n d e r s t a n d i n g t h e r e c r y s t a l l i z a t i o n mechanisms i n p o l y s i l i c o n f i l m s . The g e n e r a l e f f e c t s It can be s e e n from t h i s f i g u r e t h a t a r e d e s c r i b e d i n Fig. 5.1. complete dopant a c t i v a t i o n o c c u r s a t low power l e v e l s , s i m i l a r t o t h o s e l e a d i n g t o complete dopant a c t i v a t i o n i n ion-implanted s i n CONDITIONS

OPTICAL

TEM

SIMS

LOW POWER

I

NO CHANGE IN SURFACE MORPHOLOGY

MEDIUM POWER

I I

HIGH POWER

INCREASE IN SURFACE ROUGHNESS

SURFACE SMOOTHING

I I / I NO INCREASE IN GRAIN

SLIGHT INCREASE IN GRAIN SIZE AT THE

DRASTIC INCREASE IN

REDISTRIBUTION NO DOPANT

DOPANT REDISTRIBUTION AT T H E SURFACE

UNIFORM DOPANT REDISTRlBUTlON

COMPLETE DOPANT LOW MOBILITY

I

COMPLETE DOPANT ACTIVATION WITH INCREASED MOBILITY

I

COMPLETE DOPANT ACTIVATION WITH HIGH MOBILITY

FIGURE 5.1. Summary of t h e e f f e c t s of l a s e r power on t h e ann e a l i n g of p o l y s i l i c o n and t h e corresponding measurement methods.

5.

BEAM RECRYSTALLIZED POLYCRYSTALLINE SILICON

229

gle-crystal silicon. A moderate increase in grain size occurs with dopant redistribution at somewhat higher irradiation intensities. In the medium power range, partial melting causes a growth of surface crystallites and significant dopant redistribution in the molten region. Annealing at the highest power levels results in complete melting of the (thin) polysilicon layer, with an attendant drastic increase in the grain size and uniformity of the doping distribution across the film. A sharp reduction in the electrical resistivity of the film is observed in this high power recrystallization stage. In what follows we provide experimental details on this latter process and also describe features that are observed where a single line scan is compared with an overlapped scan frame. 5.1.1

Sample Preparation

The polysilicon films that are typically used for these studies are 0.5-0.6 pm thick layers deposited by low pressure chemical vapor deposition (LPCVD) on a layer of thermally grown silicon dioxide In some cases the silicon dioxide layer is replaced with a silicon nitride layer. After deposition, the wafers may be doped by either ion implantation or thermal diffusion and then subjected to the beam processing operations. For the laser processing experiments to be discussed below, a cw argon laser was operated in the multiline mode with a laser beam width on the sample surface of 40 pm. This circular spot was scanned across the surface at a rate of 12.5 cm/sec in all cases. The variables employed were substrate temperature, laser power, and the amount of overlap between adjacent scan lines. For reasons that are suggested in Chapter 2 it is advantageous so that relato use subtantial substrate thermal bias (350'C) tively small increments in laser power will produce substantial increments in the surface temperature of the irradiated film. For the experiments reported below, laser power was increased in 1 W increments from 5 W to 13 W, corresponding to a range of surface temperatures from 600' to 1500'C.

It should be emphasized that, while these conditions can produce very desirable changes in the properties of the films, they are certainly not the only set of conditions that would achieve these results; a different laser spot size, scan rate, laser power and substrate temperature can be chosen to yield very similar results. 5.1.2

Growth Mechanisms [5.5]

To study the mechanisms of crystallite growth, Gerzberg et al. L5.53 implanted arsenic and boron ions into separate samples to serve both as dopants and markers which could be used to investigate the recrystallization process. SEM was used to monitor

230

K. F. LEE ETAL.

changes in surface texture as a function of laser power (Fig. 5.2). For the arsenic implanted samples annealed at low and medium laser power levels (5 and 9 W respectively), no significant change in surface morphologywas observed, as shown in Figs. 5.2(a) and ( b ) . Above 10 W, however, distinct lines were apparent in the nonoverlapped sample with the microstructure inside the line pointing in the direction of the scan. Figure 5.2(c) is a SEM micrograph of an annealed line obtained in this high power region, showing the directionality in the line, the smoothing of the original fine surface roughness, and the formation of large surface structures. The grain size obtained at the high power level was analyzed through transmission electron microscopy ('JXM) and diffraction analysis. Samples were jet thinned from the back surface before presents T E M micrographs of boronexamination. Figure 5.3 implanted polysilicon samples irradiated with both single and overlapping lines. The long crystallites in the single line scan are characteristic of the high power region. These crystallites are found to be continuous, columnar structures extending from the Si02 interface to the surface of the film i5.61. The difference between the single and overlapped scanning conditions will be explained below in terms of thermal conductivity variations.

CONTROL AND 5 w

*P

-

9w

12

w

-

2P FIGURE 5.2. SEM photomicrographs of laser processed polysilicon at power levels of 5 W, 9 W and 1 2 W.

5.


231

FIGURE 5.3. TEM photomicrographs of single and overlapping lines on boron-implanted polysilicon.

Figure 5.4 is a series of TEM photomicrographs of arsenicimplanted polysilicon, laser annealed at low-, medium-, and highpower levels. The as-deposited samples display fine grain structure with an average grain size of -550 A Laser processing at 7 and 9 W produces grain growth but no significant change in surface morphology compared to the as-deposited samples. However, a sharp change is observed at power levels of 10 W and above, as evidenced by the 15 to 25 pm grain size in the sample annealed with 11 W laser power. The mechanism for this dramatic growth in crystallite size and its implications are discussed in Sec. 5.1.5.

.

232

K. F. LEE ETAL.

FIGURE 5.4. TEM photomicrographs of single and overlapping lines on arsenic-implanted polysilicon.

A global description of the recrystallization process can be gained from plots of the average grain size vs. laser power for the boron- and arsenic-implanted samples shown in Figs. 5.5 and 5.6. In Fig. 5. 5, the grain size of the boron-implanted samples is seen to increase monotonically in the medium power range ( 6 to 8 W). The transition to the high power level occurs in the single line scan when the power is greater than 9 W. In a narrow power range from 8 to 9 W, grain size increases from 1 pm to 10 vm and, at a laser power level of 11 W, the grain size (length) increases t o 20 urn.

Figure 5.6 is a plot of grain size for the arsenic-implanted polysilicon films after laser processing. A s was the case for the boron-implanted samples, grain size increases monotonically with laser power from 6 to 9 W. The transition to the high power level occurs between 9 and 10 W, above which a 20 vm grain size is commonly observed.

5.


ipm)

I

I

I

1

I

I

1

233 I

B IMPLANTED POLY AND L.A. n OVERLAPPING LINES NON-OVERLAPPING SCAN w

-

100-

N

/*-•

cn 2

-

10-

2

./ ./.

1-

(3

W

0

2

0.1 -

>

-

e n ,-,-,

w

a

/*

0.01 -

-

0.001

I

I

I

I

I

I

I

1

As IMPLANTED POLY AND L.A.

a

wN v)

100-

za

10 -

(3

1-

U

0

OVERLAPPING LINES

1

w

0

2

0.1 -

w

>

a

0.01 -

Average grain size of arsenic-implanted laserFIGURE 5.6. annealed polysilicon as a function of annealing power levels for single and overlapping lines.

234 5.1.3

K. F. LEE E T A L .

Dopant Distribution

Secondary ion-mass spectrometry (SIMS) was used to obtain dopant-profile measurements corresponding to the various power levels described in the previous section. Oxygen primary ionbombardment and positive secondary ion-mass spectrometry (B’ and A s + ) were employed, using the CAMECA IMS-3f ion microanalyzer. The boron profiles were determined through standard depth profiling; however, sensitive measurements of the A s profiles required secondar ion initial kinetic discrimination to reduce the amount of (30Si$9 Si. 1 60)+ detected at the same nominal mass as 75As+. This analysis was sufficient to provide an A s detection limit of 3 to 5 x10l8 atoms/cm3. A s can be seen in Fig. 5.7, the high oxygen content of Si02 and minor surface-charging effects produce an anornolously high mass-75 intensity that misrepresent the true arsenic content in the oxide region. This problem, however, did not hinder the measurements in the polysilicon layer.

9w

11w

0.3~

1OZ1

-

5P

6W /

3

:

2

0 + a

lozo

10”

Z W

0

60

lOl8

0 z W v)

a:

k

0'

\

I

I

NON-LASER ANNEALED

lo+ I-

L-

/#

,,4'e

cn

--0-

-----e--

20-

; '"

-

+-

PULSED LASER ANNEALED

\

t

W W

P-DOPED POLY Si THERMAL ANNEAL: 450%, N,

I

cn

I

I

I

1

I

A

I

FIGURE 5.14. Sheet resistivity of laser-annealed ( 0 ) and unannealed ( 0 ) poly Si as a function of subsequent thermal annealing time (45OOC); P-diffused poly Si. In this section we summarize experiments that were performed to elucidate the mechanism of sheet resistivity reduction in heavily doped polysilicon by laser annealing and its thermal stability during post laser heat treatment. A summary of both Q-switched Nd:YAG and cw argon laser annealed films will be presented, leading to conclusions on the minimum resistivity that can be achieved for phosphorus doped, laser recrystallized polysilicon films.

K. F. LEE ET AL. 5.2.1.

Electrical Properties

Figure 5.15 shows the resistivity changes that occur in ion implanted ( 3 ~ 1 0P+~ ~ions/cm2), laser-recrystallized polysilicon films (1800 A thick) during thermal annealing at 1000°C. Immediately following laser annealing, the cw laser-annealed poly Si (CA poly) exhibited a measured resistivity of 14 Q/O , and the pulsed laser annealed poly Si (PA poly), 17 Q/O In each case, the resistivity is smaller than that of the thermal annealed control sample (31 Q / O ). However, it can be noted that the resistivity of both the CA poly and the PA poly increase rapidly in the first stage of thermal annealing. The stable resistivity of the PA poly is 40 Q / O , and is much higher than that of the thermally annealed poly. On the other hand, the CA poly rises to only 20 Q / o with the thermal treatment, and remains lower than the control sample. The reason for this large variation is related to dopant precipitation and has been thoroughly discussed by Shibata et al. L5.181.

.

-

*_--

m

L

;

i

0

Ew-LASERANNEALED

a

ieooA POLY si ON NITRIDE 3 x 10'6cm2, P*-IMPLANTED (90 keV) THERMAL ANNEAL: 1000°C, N2

0

10 20 THERMAL ANNEAL TIME (min)

30

FIGURE 5.15. Sheet resistivity of cw-laser-annealed (0) and pulsed-laser-annealed ( 0 ) Si as a function of thermal annealing cm-2p+ implanted with time. The samples are 1800 A Si, 3 x 90 keV. The thermal-annealed controls ( X ) are also shown. The thermal annealing was performed at 1000°C in N2 ambient.

5. 5.2.2


249

Limits of Resistivity Reduction

The dramatic decrease in sheet resistivity obtainable by cw beam processing is accompanied by a nearly 100% activation of the implanted impurities. The highest carrier concentration in aslaser-processed samples is in excess of 1021/cm3, which exceeds the solid solubility limit of phosphorus in single crystal silicon ( 4 x 1020/cm3 i5.191). However, activation of impurities at levels in excess of the solid solubility produces a sheet resistance which is thermally unstable, leading to the precipitation of dopants during subsequent heat treatment. Precipitation then occurs at grain boundaries as well as in the grain crystallites. A s a result of precipitation, the lowest resistivity obtainable in laser processed phosphorus do ed polysilicon is achieved at an average doping level of 5 x 10'0/cm3.

It is instructive to plot the mobility data as a function of the active impurity concentration. The results are shown in Fig. 5.16. The solid line indicates the mobility calculated from the Irvin curve assuming 100% activation of the dopant. VOLUME CARRIER CONCENTRATION (crn-3)

0

al

c \

-

-IRVIN MOBILITY

"I.,.

0 .CW A A PULSED x THERMAL ONLY ----- CALCULATED

100

20

-

L.0.33~

A

0

lot5

I

1

I

I

8

1

. 1

2 4 6 8 10' 2 SHEET CARRIER CONCENTRATION

0

4

FIGURE 5.16. Hall mobility of Si films plotted as a funcand the volume carrier tion of sheet carrier concentration (cm'2) concentration (cm-3). Open (0,A) and solid (*,A) symbols refer to the values before and after the subsequent thermal anneal, respectively. The solid line represents single-crystal mobility calculated from the Irvin curve. Calculated mobilities of S i for a variety of grain sizes are also shown by dotted lines.

250

K. F. LEE ET AL.

It is seen that the majority carrier mobility in the CA poly is very nearly equal to the single crystal values obtained by Irvin, while that in the PA poly or the TA poly is substantially smaller. It is interesting to observe that mobility values for doping concentrations that exceed the solid solubility are obtained by a relatively straightforward extrapolation of the Irvin curve at lower doping concentrations. 5.3

ELECTRONIC PROPERTIES OF LASER RECRYSTALLIZED FILMS INTENDED FOR MOS DEVICE FABRICATION

The possibility of fabricating devices directly in beam recrystallized polysilicon films is of considerable interest in integrated circuit technology. The characteristics of devices fabricated in these films will be strongly dependent on both the electronic properties in the film and the properties of the interfacial layers between the polysilicon and various insulators that are grown during device fabrication. In conventional silicon-gate MOS applications, the polysilicon is very heavily doped, similar to that discussed in Section 5.2, so that the polysilicon film is never depleted even at the insulator boundary. However, when an active device is to be placed in a laser recrystallized polysilicon film, the film will be only moderately doped, s o that depletion and inversion layers can easily be formed by residual charges that may reside in the insulator or at the polysilicon/ insulator interface. The possibility of using beam recrystallized polysilicon films for practical devices depends on keeping these residual charges at sufficiently low levels that they do not dominate the device behavior. The utility of the films for integrated circuit fabrication also depends critically on the carrier velocity-electric field characteristic of the films, and particularly on whether this characteristic is affected by the large number of grain boundaries that may exist within the film. In this section we describe experiments that were undertaken t o explore these questions. 5.3.1

Velocity-Field Characteristics

To assess the potential of these films for MOSFET applications, it is useful to know at least empirically how grain boundary scattering will affect the carrier velocity-vs-electric field characteristics, especially in the high electric field region. Measurements of electron velocity in both laser recrystallized and silicon-on-sapphire films made by Cook et al. 15.201 are shown in Fig. 5.17 where we also include for comparison measurements made on single crystal material by Norris and Gibbons [5.21]. As can be seen, the electron velocity in

5.


ELECTRON VELOCITY

25 1

18pm

c40pm

-I

ESENT WORK:

6/U

r 55 -

E

4-

321-

/ I

0

1

10

I

I

20

I

I

1

30

1

.

40

1

I

50

I

I

60

E (hV/cm)

FIGURE 5.17. Electron velocity vs. electric field in Si, SOS and laser-recrystallized-polysilicon. heavily doped, laser recrystallized polysilicon films is actually larger than that in SOS films, even though the SOS films are more lightly doped and nominally "single crystal". Neither of the films gives the sharp saturation and high terminal velocity that are characteristic of single crystal material, though adequate velocity is obtained for device purposes. 5.3.2

Oxidation Characteristics [5-8 1

Kamins et al. [5.8] have conducted a set of basic experiments to determine the oxide thickness grown on laser-recrystallized polysilicon under conditions which might be used in an integratedcircuit process. Only polysilicon films deposited onto silicon nitride with no subsequent nitrogen anneal were studied. After laser recrystallization, some of the wafers were doped by a 95OoC, POCl3 predeposition which produced a sheet resistance of 10 Q/u in single crystal silicon. Both undoped and doped films were then oxidized. Two different oxidation cycles were used. A 125 min, 1000°C TCE/O2 oxidation, which forms 1000 8, of oxide on (100)-oriented, single crystal silicon, was used in some cases. A 210 min, 800°C steam oxidation was also investigated in order to emphasize the effects of the silicon orientation and structure. The oxide

K. F. LEE E T A L .

252

thicknesses were measured on the polysilicon and single crystal control wafers with a W spectrophotometer [5.22]. After the oxide thicknesses were measured, the oxidation cycles were repeated without stripping the oxide so that thicker oxides, which could be measured more easily, were grown. The oxide thicknesses after the first oxidation cycle are The oxide thicknesses grown on the undoped shown in Table 5.3. polysilicon with the 1000°C TCE/O2 oxidation fall between those on the (100)- and (lll)oriented, single crystal silicon oxidized simultaneously, being about 6% greater than that on (100)-oriented silicon after the first oxidation. After the second oxidation the oxide thicknesses continued to fall between those of the two orientations of single crystal silicon. Oxidation under these conditions is influenced by diffusion of oxygen through the already-formed oxide, as well as by surface reaction, so the effect of the silicon structure and orientation is small. TABLE 5.3.

Oxide Thickness on Laser-Recrystallized Polysilicon Oxide Thickness

Doping

Oxidation Temperature ('C)

Ox id$zing

(81 Single-Crystal 5

Poly-Silicon

Pnbient

Recrystallized

Fine-Grain

-2100,

c111,

VndDped

IOM)

TCf/O*

iosa

1060

980

1100

Undoped

800

Steam

19m

1830

1140

1570

PhQSphOWS

800

Steam

6030

6000

Wed

I

I

4830

The 800°C steam oxidation is controlled primarily by the surface reaction so the effects of orientation and structure should be more significant, as observed, with the oxide thickness on (111)-oriented, single crystal silicon 46% greater than that on (100)-silicon. The oxide thickness on the fine-grain polysilicon appears to be slightly greater than that on (110)-silicon, with the oxide on the laser-annealed regions probably even slightly thicker. Thus, the oxide thickness grown on undoped, laserrecrystallized polysilicon under surface-reaction-controlled conditions differs only slightly from that on fine-grain polysilicon and is similar to that on the fast-oxidizing orientations of single crystal silicon. The influence of the laser recrystallization on heavily phosphorus doped polysilicon is more dramatic. Under these conditions, the oxidation rate is controlled by the dopant concentration at the surface, and the effect of crystal orientation

5.


253

is less significant, as can be seen by comparing the oxide thicknesses on the two orientations of doped, single crystal silicon in Table 5.3. The oxide thickness on the doped, fine-grain polysilicon is markedly less than that on the single crystal silicon or on the laser-recrystallized polysilicon, while that on the laser-recrystallized polysilicon is close to that on the single crystal silicon. Thinner oxides on heavily doped, fine-grained polysilicon than on equivalently doped single crystal silicon have been observed before and have been related to the lower active dopant concentration near the surface of the polysilicon [5.221. During the doping cycle, the dopant can diffuse farther into the polysilicon than into the single crystal silicon so that the average surface concentration is lower, and the oxide subsequently grown is thinner. In addition, even if the dopant concentration were the same, the lower electrical activity in the polysilicon would keep the Fermi level closer to midgap so that the dopant-enhanced oxidation would not be as significant. Similar reasoning can account for the thicker oxide grown on the laser-recrystallized polysilicon than on fine-grain polysilicon. The dopant appears to diffuse away from the surface of the laser-recrystallized polysilicon at about the same rate as in single crystal silicon and less rapidly than in fine-grained polysilicon since grain boundary diffusion is less important; consequently, the oxide grown is thicker than on fine-grain polysilicon

.

We conclude from these results that, while fine-grained polysilicon films exhibit a strong (110) preferred orientation, the preference for a particular orientation is less pronounced in laser-recrystallized polysilicon, although there is some preference for (111) orientation. The oxide thickness grown on lightly doped laser-recrystallized polysilicon is not greatly different from that on fine-grained polysilicon and is probably dominated by the fast-oxidizing. orientations under surfacereaction-controlled conditions. The oxide thickness on heavily doped, laser-recrystallized polysilicon is similar to that on single crystal silicon and is much greater than that on finegrained polysilicon. 5.3.3

Charges at a Laser Recrystallized Polysilicon/Insulator Interface [5.23]

Kamins et al. [5.23] investigated the behavior of the oxide/ polysilicon interfaces by fabricating capacitor structures and making capacitance-voltage measurements with the depletion region extending into both the film and in some cases the underlying

K. F. LEE ET AL.

254

s u b s t r a t e . Both t h e r m a l l y grown s i l i c o n d i o x i d e and low p r e s s u r e CVD s i l i c o n n i t r i d e were i n v e s t i g a t e d as t h e i n s u l a t i n g l a y e r .

A l l of t h e s t r u c t u r e s s t u d i e d i n t h e s e experiments (Fig. 5.18) were f a b r i c a t e d on 0.01 Sl-cm n-type s i l i c o n w a f e r s which subsequently served a s t h e g a t e e l e c t r o d e . A 1000 A t h i c k , TCE/O2 g a t e oxide w a s grown on some w a f e r s a t lOOO"C, w h i l e o t h e r s were covered w i t h 1000 A of low p r e s s u r e CVD s i l i c o n n i t r i d e . A 5500 A t h i c k f i l m of LPCVD p o l y s i l i c o n was t h e n d e p o s i t e d a t 625°C on a l l wafers. The wafers which contained an oxide l a y e r were next annealed a t 1100°C i n N2 f o r 1 hour t o ease c o n t r o l of t h e Wafers c o n t a i n i n g subsequent laser r e c r y s t a l l i z a t i o n process. a n i t r i d e l a y e r d i d n o t r e q u i r e t h i s thermal a n n e a l i n g s t e p . A l l p o l y s i l i c o n f i l m s were implanted w i t h 1 0 l 2 boron ions/cm2 a t 100 keV. P o r t i o n s of each wafer were t h e n r e c r y s t a l l i z e d w i t h a scanThe subning c w argon l a s e r , w i t h a beam width of about 70 um. s t r a t e temperature was h e l d a t 35OoC, and a power l e v e l was chosen which r e s u l t e d i n t h e formation of the long columnar c r y s t a l l i t e s d e s c r i b e d i n S e c t i o n 5.2. Aluminum was t h e n d e p o s i t e d and d e f i n e d i n t o s q u a r e s 300 pm o r 900 pm on a s i d e , t h e aluminum s e r v i n g a s a contact t o t h e p o l y s i l i c o n "substrate". The p o l y s i l i c o n was plasma e t c h e d u s i n g t h e metal a s a mask. A f t e r a 450°C H 2 a n n e a l , t h e high frequency ( 1 MHz) c a p a c i t a n c e v o l t a g e c h a r a c t e r i s t i c s were measured, w i t h t h e d e p l e t i o n r e g i o n s e x t e n d i n g i n t o t h e polys i l i c o n . To f a c i l i t a t e d i s c u s s i o n a l l g a t e v o l t a g e s mentioned below a r e those a p p l i e d t o t h e n' s i n g l e c r y s t a l wafer which served a s t h e g a t e e l e c t r o d e . 5.3.3.1

R e c r y s t a l l i z e d Films on Si02

The s t r u c t u r e s w i t h s i l i c o n d i o x i d e beneath t h e p o l y s i l i c o n w i l l be considered f i r s t . Laser powers of 14-16 W produced o v e r l a p p i n g s c a n s and t h e d e s i r e d l a r g e - g r a i n s t r u c t u r e i n t h e p o l y s i l i c o n . I n t h e l a s e r - r e c r y s t a l l i z e d areas well-defined accumulation and i n v e r s i o n r e g i o n s were observed i n t h e capacitanceThe c a p a c i t a n c e w a s very c l o s e t o voltage c h a r a c t e r i s t i c s . When t h e t h a t expected f o r t h e 1000 8, t a r g e t oxide t h i c k n e s s . opposite p o l a r i t y voltage is applied, t h e capacitance decreases

NEUTRAL REGION DEPLETION REG1

-

PO LY S I L I C0 N

1 SiOz

or SijN,

n* SILICON WAFER

FIGURE 5.18. Cross s e c t i o n of t e s t c a p a c i t o r s t r u c t u r e , showing d e p l e t i o n region e x t e n d i n g i n t o t h e p o l y s i l i c o n when a t h e "gate" v o l t a g e i s a p p l i e d t o t h e n+ wafer. O1980 IEEE

5.


255

t o a value i n d i c a t i n g a depletion region approximately 2000 A wide i n the polysilicon. This p o l a r i t y of voltage a l s o tends t o d e p l e t e the n+ g a t e wafer, but t h e heavy doping i n t h i s wafer i n c r e a s e s t h e voltage necessary f o r s i g n i f i c a n t d e p l e t i o n and a l s o l i m i t s t h e width of t h e depletion region so t h a t t h e e f f e c t of t h e depletion i n t o t h e n+ wafer can be neglected i n t h e following discussion. The depletion region width i n t h e p o l y s i l i c o n c o r r e s onds t o a dopant concentration of approximately (2.2 - 2.6) x lop6 cm-3, confirming t h a t v i r t u a l l y a l l t h e dopant i s a c t i v a t e d by t h e l a s e r r e c r y s t a l l i z a t i o n . The doping concentration i s somewhat higher than t h e average do a n t concentration of 1.8 x 1 0 l 6 expected i f a l l the 1 x lo1! ~ r n - boron ~ atoms implanted were a c t i v e and were uniformly d i s t r i b u t e d through t h e 5500 A-thick p o l y s i l i c o n f i l m . (Laser r e c r y s t a l l i z a t i o n does not appear t o change t h e average f i l m thickness .) This anomalously high apparent conc e n t r a t i o n may possibly be r e l a t e d t o dopant segregation near t h e bottom of the p o l y s i l i c o n f i l m a s i t r e c r y s t a l l i z e s o r t o r e s i d u a l d e f e c t s t a t e s which l i m i t t h e depletion-region width. I n the regions of the same wafer not a f f e c t e d by t h e l a s e r , t h e minimum capacitance i s considerably less than i n t h e recryst a l l i z e d regions, corresponding t o a maximum depletion-region width of about 4500 A, close t o t h e thickness of t h e p o l y s i l i c o n film, suggesting t h a t t h e high r e s i s t a n c e of the unaffected polys i l i c o n may cause the e n t i r e thickness of the f i l m t o a c t as a d i e l e c t r i c when attempts a r e made t o d e p l e t e t h e s u r f a c e . (Aluminum from the deposited e l e c t r o d e may have penetrated i n t o t h e p o l y s i l i c o n s l i g h t l y , causing the apparent maximum depletionregion width t o be s l i g h t l y smaller than the p o l y s i l i c o n f i l m thickness .) Samples which received n e i t h e r the l l O O ° C thermal anneal nor the laser processing showed very l i t t l e dependence of t h e capacitance OR t h e g a t e voltage. A f t e r a negative bias-temperature stress t o i n s u r e t h a t any p o s i t i v e mobile ions present did not influence t h e r e s u l t s , t h e f l a t b a n d voltage i n t h e l a s e r - r e c r y s t a l l i z e d regions was found t o If QHS i s taken t o be -0.90 V, be i n t h e range -2.9 t o -3.4 V . t h e fixed-charge density is c a l c u l a t e d t o be about 4 x 1011 cm-2, which is somewhat higher than t h a t expected i n s i n g l e c r y s t a l s i l i c o n but not unreasonable s i n c e t h e processing was not optimized f o r t h i s unconventional s t r u c t u r e . (X-ray measurements i n d i c a t e a c r y s t a l s t r u c t u r e containing g r a i n s of various orient a t i o n s , with a weak preference f o r (111) o r i e n t a t i o n under t h e conditions employed here 15.81, s o t h a t t h e minimum f ixed-charge density would be higher than t h a t expected f o r (100) s i l i c o n . ) The voltage between the measured flatband and inversion points of the C-V c h a r a c t e r i s t i c was g r e a t e r than calculated,

K. F. LEE ETM.

however, suggesting the presence of some fast states or lateral nonuniformities. If this distortion in the curve were entirely related to fast states, their density between flatband and inversion would be about 2 x loll cm-2. A portion of the distortion in the C-V characteristic may be related to nonuniformities since a region treated under conditions which produced incompletely overlapped recrystallized regions showed distorted C-V characteristics corresponding to a parallel combination of the recrystallized and unaffected curves. In the regions which were not affected by the laser, the magnitude of the flatband voltage varied significantly and was generally greater than suggesting an effective charge density greater than 1012tm'2. Thus, the properties of the silicon-dioxidefpolysilicon interface under a layer of laser-recrystallized polysilicon resemble, but are inferior to, those at the interface between a thermally grown oxide and single-crystal silicon. This interface is, however, of much higher quality than most semiconductor-insulator interfaces not formed by thermal oxidation. For example, the interface between single crystal silicon and a deposited oxide layer is generally unstable and can, at best, be characterized by very high fixed-charge and interface state densities. A more recent detailed study by Le and Lam i5.241 of electrical characteristics at the interface between laser-recrystallized polysilicon and an underlying thermally grown oxide has shown that the lowest fixed-oxide charge density is obtained when the oxide layer is thermally grown in an oxygen/HCl ambient. A doublelayered encapsulating structure obtained by first depositing a 10 nm plasma CVD oxide on the polysilicon film and then a 6 nm LPCVI) nitride film on the oxide prior to laser recrystallization was also found to be effective in reducing charge at the back interface. Oxide charge densities in the range of 3 ~ 1 0 ~ ~ / were cm~ obtained, consistent with the earlier work of Kamins et al.

5.3.3.2

Polysilicon Films on Si3N4

Capacitors containing an insulating layer of silicon nitride were also tested. The polysilicon deposited on the silicon nitride contained a wide range of crystal structure after laser recrystallization depending on the laser power used, which varied from 14 to 17 W. No long grains were seen at the lowest power, while the highest power produced totally overlapped recrystallized regions containing long grains. The capacitance-voltage characteristics were also quite different in regions annealed at different powers.

5.


257

I n all cases, the maximum capacitance corresponded to a silicon nitride layer with a relative permittivity of 7.0 for the 1000 a thickness deposited. The behavior of the capacitancevoltage characteristics in depletion was markedly different in the differently processed regions (Fig. 5.19). The minimum value of capacitance was lowest in the region recrystallized at the lowest power (approximately 44% of the nitride Capacitance); the minimum capacitance increased with increasing laser power to about 60% of the nitride capacitance for the highest laser power. I n addition, at large positive gate voltages 1(> 30 V, in the direction tending to invert the polysilicon) , the capacitance again increased toward its maximum value, suggesting the presence of a source of minority carriers to charge the inversion layer rapidly. In the region processed at the lowest laser power, there was no flat portion of the C-V curve at the minimum capacitance. As the laser power and the minimum capacitance increased, a flat region developed in the C-V characteristic. (Rapid minority-carrier generation was also seen in samples with a silicon-dioxide insulator at high positive gate voltages).

This behavior of the minimum capacitance would not be consistent with defects witkiln the polysilicon film, since a higher defect concentration at higher lases powers would have to be postulated, in contrast to the reduced defect density expected as the long grain structure becomes more fully developed.

High frequency capacitance-voltage characterFIGURE 5.19. istics of structure containing a silicon nitride insulator under polysilicon films recrystallized at three different laser powers. O1980 IEEE

K. E LEE ET AL. No meaningful value of the fixed charge or fast-state density can be extracted from the capacitance voltage measurements because of electron trapping in the insulator. Even before a bias temperature stress, the flatband voltage changes significantly when large voltage ramps are applied. (In contrast, no change was seen in samples containing a silicon-dioxide insulator when tested with similar, high-magnitude voltages.) Limited tests indicated no significant improvement by laser annealing the nitride before the polysiliconwas deposited, in addition to recrystallizing the polysilicon with the laser after the polysilicon deposition. The results obtained from the structures containing silicon nitride show that the polysiliconlsiliconnitride interface is not well behaved, a situation similar to that of a silicon-nitridelsingle crystal-silicon interface. 5.3.3.3

Summary of Interface Properties

Capacitance-voltage characteristics have been measured to determine the interface properties at the back surface of a layer of laser-recrystallized polysilicon. The interface between the recrystallized polysilicon and an underlying oxide layer can be characterized by an effective fixed-charge density and a faststate density, both in the low-to-middle-lO1l cm-2 range. The polysiliconlsilicon nitride interface is not as well behaved. The minimum capacitance depends on the laser power used to recrystallized the polysilicon, and charge trapping at the interface precludes the determination of a meaningful value of interface charge. 5.3.4

Effect of Laser Recrystallization of Polysilicon on an Underlying Crystalline Silicon Substrate 15.251

Gibbons and Lee [5.261 first showed that three-dimensional integrated circuits could be made by utilizing MOS transistors fabricated in both an underlying crystalline silicon substrate and (simultaneously) in layers of recrystallized polysilicon lying above the substrate. Since then a number of authors have developed other three-dimensional structures, a topic which we will discuss further in Sec. 5.4. For structures of this type to be of maximum utility, the properties of the devices fabricated in the underlying single crystal silicon must not be degraded by the recrystallization of the polysilicon layer. Kamins [5.27] has shown that the threshold voltage and surface mobility of 59 randomly chosen MOS transistors fabricated in a 64 mm single crystal wafer were basically unchanged following recrystallization of an overlying polysilicon layer (separated from the substrate by a 0.1 mm

5.


259

oxide layer). Hence, the process is compatible with typical IC fabrication. Later, Kamins and Drowley r5.251 studied the effect of recrystallization on substrate minority carrier lifetime, which is a more exacting test of the process. These authors showed that the recrystallization process produces significant degradation of the generation lifetime in the underlying substrate as measured by MOS deep depletion techniques. However, a subsequent furnace treatment ( 3 0 min. at 95OoC and 60 min. at 8OO0C in nitrogen) increased the generation lifetime to its original value except when the substrate was visibly damaged by melting. Measurements were also carried out on junction diode reverse leakage current for diodes made underneath polysilicon layers that were subjected to the recrystallization process, with similar results. The work of Kamins and Drowley, together with the earlierwork of Kamins, thus suggests that the point defects in the substrate produced during laser recrystallization of an overlapping polysilicon film are substantially eliminated by subsequent heat treatment, and that recrystallization is therefore compatible with the fabrication of high quality devices in the substrate. Further evidence on this point is presented in Chapter 4 . 5.3.5

Properties of Laser Substrates

Recrystallized Films

on

Quartz

The recrystallization experiments and results described previously have been carried out exclusively with a silicon substrate. There are a number of applications, however, in which it would be desirable to recrystallize a polysilicon film on a nonsilicon substrate, especially a transparent substrate such as quartz (for imaging applications). Suitable substrate materials must be compatible with silicon processing techniques and temperatures. In addition, the thermal expansion coefficient of the substrate should ideally match that of silicon. Quartz and sapphire both satisfy the conditions of compatibility with silicon processing, and sapphire also has a thermal expansion coefficient near that of silicon. Indeed, the fabrication of recrystallized silicon films on sapphire both by beamactivated solid phase epitaxy and by laser recrystallization offer some opportunities that have not been explored. In contrast, several experiments have been performed to study the electronic properties of recrystallized films on quartz. Kamins and Pianetta [5.28] showed that device worthy films could be prepared despite the large mismatch in thermal expansion coefficients by etching the film into islands either prior to or following the recrystallization. Surface channel mobilities in these first experiments were found to be low, however, in the range of 250-300 cm2/V-sec, and sometimes as low as 60-80 cmz/V-sec.

260

K. F. LEE ETAL..

In subsequent work, Tsaur et al. [5.29] showed that the stress in polysilicon films recrystallized on fused silicon substrates can in fact produce larger surface channel mobilities than are obtained in bulk silicon. Tsaur et al. correctly attributed this increase to increased electron population in high mobility valleys in the Brillouin zone, corresponding to tensile stress in (100)-oriented films. Johnson et al. [5.30] obtained similar results using a CO2 laser to recrystallize polysilicon films that were patterned into islands with an hour-glass shape to promote the growth of the preferred (100)-orientation while at the same time suppressing the formation of microcracks that arise on account of the thermal mismatch. Johnson et al. achieved surface channel mobilities in excess of 1000 cm2/V-sec. Leakage effects were obtained from "back-channel" effects in these devices, suggesting that the recrystallization process at the quartz/polysilicon interface leads to material with surface-related defects. However, the leakage was not significantly different from that obtained when an oxidized silicon substrate is used as the base for polysilicon deposition and recrystallization. These experiments thus show that high quality recrystallized films can be prepared on appropriate nonsilicon substrates when precautions are taken to suppress microcrack formation arising from the thermal mismatch problem. 5.4

CHARACTERISTICS OF MOS DEVICES AND INTEGRATED CIRCUITS FABRICATED ON LASER RECRYSTALLIZED POLYSILICON FILMS

The electronic properties of laser annealed polycrystalline films have strongly suggested from the outset that these films can be used directly for the fabrication of active devices. A number of studies, some already referred to, have been performed to evaluate this possibility. Most of these evaluations have been performed using MOS field effect transistor structures, since these devices utilize majority carrier properties and should be minimally sensitive to lifetime and grain boundary effects that could dominate the characteristics of bipolar devices. Channel lengths for MOS transistors that are on the order of, or smaller than, the grain size in the laser recrystallized films may be expected to yield device performance that is comparable to that of devices fabricated in single crystal silicon, whereas the properties of grain boundaries have been found to dominate device behavior when devices are fabricated on fine grain polysilicon films [5.31]. In what follows we discuss the electrical characteristics of MOSFETs fabricated on such films under a variety of different conditions.

5. 5.4.1


261

MOSFETs F a b r i c a t e d on Polysilicon/Si3N4 S u b s t r a t e s

The f i r s t MOSFET devices f a b r i c a t e d i n laser r e c r y s t a l l i z e d The polyp o l y s i l i c o n f i l m s were f a b r i c a t e d by Lee e t a l . [5.1]. s i l i c o n samples used were 5500 A t h i c k , d e p o s i t e d by low-pressure chemical vapor d e p o s i t i o n (LPCVD). The s u b s t r a t e s were s i n g l e c r y s t a l s i l i c o n onto which a 1000 A l a y e r of Si3N4 had been deposited. Phos horus was implanted a t an energy of 100 keV t o a dose of 3 x 10p2/cm2 t o form t h e channel f o r t h e depletion-mode d e v i c e s . Boron was implanted a t an energy of 100 keV w i t h a dose of 3 x 1011/cm2 f o r t h e enhancement mode devices. The w a f e r s were then annealed by an argon cw scanning laser ( c i r c u l a r beam) so t h a t long g r a i n s were formed. The subsequent p r o c e s s i n g s t e p s a r e desc r i b e d i n Ref. 5.1. A photograph of t h e f i n a l depletion-mode d e v i c e s t r u c t u r e i s shown i n Fig. 5.20. F a b r i c a t i o n of t h e enhancement mode devices w a s similar except t h a t s t e p s r e l a t e d t o t h e mesa formation were omitted s i n c e no i s o l a t i o n i s r e q u i r e d "between devices. Devices w i t h channel l e n g t h s of 50 p m w e r e f a b r i c a t e d ; t h e channel widths were 250 pm f o r t h e d e p l e t i o n mode devices and 270 pm f o r t h e enhancement mode devices. Since t h e g r a i n s formed by laser a n n e a l i n g tend t o a l i g n themselves w i t h t h e laser s c a n d i r e c t i o n s , channels were f a b r i c a t e d both p a r a l l e l and perpendic u l a r t o t h e l a s e r s c a n d i r e c t i o n t o t e s t t h e importance of such alignment. The source-drain I-V c h a r a c t e r i s t i c s of both t h e enhancement and d e p l e t i o n mode devices are shown i n Fig. 5.21. Since t h e c u r r e n t i n a d e p l e t i o n mode t r a n s i s t o r flows through t h e e n t i r e t h i c k n e s s of t h e conducting l a y e r , f i l m p r o p e r t i e s can be c a l c u l a t e d from t h e transconductance and d r a i n c u r r e n t . In t h e l i n e a r r e g i o n t h e transconductance i s given by [5.32]:

FIGURE 5.20.

Photograph of a deep depletion-mode device.

262

K . F. LEE ETAL..

FIGURE 5.21. (a) Source-drain I-V characteristics for a deep depletion-mode device (VG = 0 to -12 V). (b) Source drain I-V characteristics for an enhancement-mode device.

From the transistor channel geometry (L = 50 pm, W = 250pm), the mobility is calculated to be 450 cm2/V-sec, compared to a mobility of 750 cm2/V sec in single crystal silicon at a dopant

5.

BEAM RECRYSTALLIZED PQLYCRYSTALLINE SILICON

263

Concentration of 6 x 10l6 ~ m - ~corresponding , to the dose implanted into the polysilicon film. (Because the film melted during laser annealing, the phosphorus can be assumed to be uniformly distributed through the thickness of the film, and, as will be shown below, most of the implanted phosphorus contributed electrically active carriers. Therefore, the entire implanted dose can be used to calculate the average dopant concentration.) Because of the high oxidation temperature, the gate oxide thickness can be assumed to be the same as that on single crystal silican (1000 A ) . The carrier concentration in the channel can also be calculated from the geometry af the device and the flatband voltage t 5 -323,

If most o f the dopant in the polysilicon I s active, h s is equal to that of single crystal silicon (-0.21 U), and the gate oxide thickness may again by assumed to be the same as that on si gle crystal silicon. Since Nf/q was found to be 1 x 10l1 cm-' on enhancement mode transistors fabricated on (100)-oriented single crystal wafers in the present experiment, and a value three times as hi h would be expected on (1112-oriented silicon, Nf/q = 2 x 10" is used in the calculation of Vm for the polysilicon device to account for the random orientation of the crystallites. Assuming the third term to be negligible, a flatband voltage of -1.1 V is obtained with an uncertainty of 0.5 V because of possible deviations of Nf/q from the value used.

*

v-5 = VFB

From the resistance at concentration of 5 x 1OI6 cm

= -1.1 V, an average carrier can be calculated from

where d = 0.5 UIU is the thickness of the film. Since this value is comparable to the average dopant concentration of 6 x 10l6 cm-3 added by ion implantation, it is seen that a high percentage of the dopant is electrically active, and little has been lost during processing. The film properties observed in these laser-annealed polysilicon films are compared to the bulk properties of single crystal silicon and also to those of fine grained polysilicon at a dopant concentration of 6 x 10l6 cm-3 in Table 5.4. For the enhancement mode device the field effect mobility is similarly calculated from the source drain characteristics in the linear region and the device geometry (L = 50 pm, W = 270 Urn) to

K. F. LEE ETAL.

264

TABLE 5.4. Electrical Properties of Polysilicon and SingleCrystal Silicon at an Average Dopant Concentration of 6x1Ol6 ~ m - ~ . Laserannealed Polysilicon Carrier conc. (cm-3 Mobility (cm2/Vsec) Resistivity ( Q cm)

5x1016 450 0.28

Singlecrystal Silicon

Fine-Grained Polysilicon*

6x1016 750

-1x 1012

0.14

-1 05

-60

*J. Y. W. Seto, J . Appl. Phys. 4 6 , 5247 (1975), for holes. be 340 cm2/V sec.

This value may be compared to the value of 630 cm2/V sec expected for the field effect mobility in single crystal silicon of the same dopant concentration (50% of the bulk mobility [5.33]). The threshold voltage (defined to be the gate voltage which induces a drain current of 1 a)is measured to be +2-5 V. Again, assuming Nf/q = 2 x 10l1 crnq2, the threshold voltage is calculated to be -0.2 V in the absence of defect levels in the polysilicon. The difference between the measured and calculated threshold voltages may be attributed to the charging of defect levels before the surface can be inverted, as well as to uncertainties in the value of Nf/q used in the calculations. This difference represents a significant improvement over previous polysilicon MOSFETs where differences of about one order of magnitude larger were observed 15.311. 5.4.2

MOS Devices and Integrated Circuits on Polysilicon/Si02 Substrates

The experiments described in the previous section show that MOS devices can be fabricated on laser recrystallized films with electrical properties that are far superior to those obtained in as deposited material, and comparable to similar device characteristics obtained when single crystal silicon substrates are employed. However, as suggested in Sec. 5.3, the nitride layer is not an ideal choice as an insulating layer because of the high density of surface states at the nitride-silicon interface. For integrated circuit applications, a thick oxide layer i s a far better choice as an insulating material. Hence most of the work done in this field since the original work of Lee et al. i5.11 has employed oxide layers.

5. 5.4.2.1


265

MUSFETs [5.21

The f i r s t MOSFET d e v i c e s on SiO2/Si s u b s t r a t e s were t h o s e of Tasch e t a l . [5.2]. Enhancement mode and d e p l e t i o n mode t r a n s i s W/L r a t i o n s of 2515, 25/10 and 25/25 (pm/um) were t o r s havf a b r i c a t e d on b o t h annealed and unannealed a r e a s of t h e wafers t o provide a d i r e c t comparison of device type and annealed a g a i n s t unannealed p o l y s i l i c o n on t h e same wafer.

The I-V c h a r a c t e r i s t i c s of enhancement (boron-implanted) and d e p l e t i o n (undoped p o l y s i l i c o n ) d e v i c e s , f a b r i c a t e d on laser anThe t h r e s h o l d v o l t a g e s (VT) n e a l e d areas, a r e shown i n Fig. 5.22.

FIGUaE 5.22. Source-drain c h a r a c t e r i s t i c s of t r a n s i s t o r s The f a b r i c a t e d i n p o l y s i l i c o n on oxide (W = 25 pm, L = 5 urn). scales i n both photographs a r e : 100 PA per v e r t i c a l d i v i s i o n , 1 V per h o r i z o n t a l d i v i s i o n , 0.5 V p e r s t e p , and gm per d i v i s i o n = 200 1.1s. ( a ) Enhancement mode (boron i m p l a n t ) , VG = 0-5V; ( b ) Depletion mode (undoped), V G = 0-3.5V.

K. E LEE ETAL.. determined from measurements of IDSa g a i n s t VG were 0.35-0.45 V f o r t h e enhancement d e v i c e s , and -0.5 t o -0.7 V f o r t h e d e p l e t i o n devices. In t h e unannealed areas of t h e wafer, t h e measured t h r e s h o l d v o l t a g e s of t h e p o l y s i l i c o n t r a n s i s t o r s ranged from 7 t o 1 0 V , w i t h l i t t l e i f any d i f f e r e n c e between implanted and unimplanted d e v i c e s . The m o b i l i t i e s w e r e determined from t h e measured s l o p e of the drain-source c u r r e n t (IDS) a g a i n s t g a t e v o l t a g e (VG) c u r v e s a t low d r a i n v o l t a g e (VD = 0.1 V) u s i n g t h e w e l l known expression (5.3) where P i s t h e s u r f a c e m o b i l i t y and Co i s t h e g a t e oxide capaciThe measured s u r f a c e m o b i l i t i e s were 170 tance p e r u n i t area. and 215 c m 2 / V s , r e s p e c t i v e l y , f o r t h e N-channel enhancement and depletion devices. These v a l u e s compare f a v o r a b l y w i t h 400 500 c m 2 / V s and 600 - 650 c m 2 / V s which a r e o b t a i n e d i n p r e s e n t N-channel SOS d e v i c e s ( u s i n g 0.5 Prn s i l i c o n l a y e r s ) and bulk NMOS d e v i c e s , r e s p e c t i v e l y . The measured s u r f a c e m o b i l i t i e s on p o l y s i l i c o n d e v i c e s ( a r e a s of wafer which d i d n o t r e c e i v e laser a n n e a l i n g ) ranged from 23 - 29 c m 2 / V s , which i s i n agreement w i t h previous r e p o r t e d r e s u l t s r5.311. Assuming t h e metal semiconductor work f u n c t i o n d i f f e r e n c e $MS t o be -0.8 t o -0.9 V f o r a h e a v i l y doped N+ s i l i c o n g a t e , and assuming t h e f i x e d charge a t t h e SiO2-Si i n t e r f a c e t o be Qss/q = 2.6 x 1O1O ernm2 ( t y p i c a l of t h e p r o c e s s used t o f a b r i c a t e t h e d e v i c e s i n t h i s s t u d y ) then from t h e measured VT of -0.5 t o -0.7, a v a l u e of 1 0 l 2 - l 0 l 3 12m-~ i s i n f e r r e d f o r t h e average doping l e v e l of t h e undoped laser annealed p o l y s i l i c o n , u s i n g t h e followi n g e q u a t i o n f o r t h e t h r e s h o l d v o l t a g e 15.341:

where

In t h i s e q u a t i o n , 4~ is t h e Fermi l e v e l i n t h e s i l i c o n f i l m , NA is t h e average doping i n t h e f i l m and t S i i s t h e s i l i c o n - f i l m t h i c k f o r t h e average doping of t h e n e s s . The v a l u e of 10l2 - 1013 undoped p o l y s i l i c o n i s c o n s i s t e n t w i t h measured r e s i s t a n c e s of load r e s i s t o r s f a b r i c a t e d on undoped p o l y s i l i c o n d e p o s i t e d by t h e same process. The above e q u a t i o n can a l s o be a p p l i e d t o determine t h e average carrier c o n c e n t r a t i o n i n t h e s i l i c o n f i l m f r t h enhanceand t h e ment d e v i c e s . Using +MS = -0.9 V, Nf/q = 4 x 1 0 l 8 em-',

5.


267

measured VT v a l u e of +0.4 V , a c o n c e n t r a t i o n of 8 x 1015 cm-3 i s i s obtained. This i s i n good agreement w i t h t h e average v a l u e of 1 x 1 0 l 6 cm-3 c a l c u l a t e d from t h e boron i m p l a n t a t i o n dose and energy, when account i s t a k e n of p a r t i a l d i f f u s i o n of boron i n t o t h e Si02 i n subsequent processing. A more a c c u r a t e p r e d i c t i o n of t h e enhancement d e v i c e t h r e s h o l d v o l t a g e , based on t h e implantat i o n and subsequent process c o n d i t i o n s , can be made w i t h t h e a i d of t h e SUPREM process modelling program [5.35]. For t h e i o n i m p l a n t a t i o n and f a b r i c a t i o n c o n d i t i o n s (temperatures and t i m e s ) d e s c r i b e d e a r l i e r , SUPREM p r e d i c t s a t h r e s h o l d v o l t a g e of 0.38 V. Also, t h e f i n a l c a l c u l a t e d boron p r o f i l e drops o f f r a p i d l y beyond 300 nm and is down t o 5 x 1013 cmq3 a t 500 nm. Therefore, f o r t h e 500 nm s i l i c o n f i l m on Si02, t h e SUPREM p r e d i c t i o n i s expected t o be q u i t e v a l i d , s i n c e a n e g l i g i b l e amount of d i f f u s i o n of boron a c r o s s t h e underlying SiO2-Si i n t e r f a c e i s involved. The leakage c u r r e n t between t h e s o u r c e and d r a i n i n t h e o f f c o n d i t i o n was examined, i n view of t h e d i f f i c u l t y i n a c h i e v i n g low Source-drain leakage c u r r e n t s were mealeakage i n SOS devices. sured w i t h t h e g a t e and source grounded and 5 V on t h e d r a i n . The The r e s u l t s correspond t o devices had channel widths of 25 pm. leakages of 1 - 6 pA per micrometer of channel width, and match t h e b e s t r e p o r t e d v a l u e s f o r SOS. 5.4.2.2

Ring O s c i l l a t o r s [5.41

Ring oscillators have a l s o been f a b r i c a t e d on l a s e r recryst a l l i z e d f i l m s d e p o s i t e d on t h i c k oxide l a y e r s . The i n d i v i d u a l i n v e r t e r s t a g e s i n t h e s e r i n g o s c i l l a t o r c i r c u i t s c o n s i s t e d of 6 pm channel enhancement mode d r i v e r devices w i t h a deep deplet i o n d e v i c e a s load. Each s t a g e was c o n s t r u c t e d w i t h a f a n o u t of t h r e e , w i t h a two s t a g e source follower output on each r i n g o s c i l l a t o r a s a b u f f e r . Both seven and e l e v e n s t a g e r i n g o s c i l l a t o r s were f a b r i c a t e d , both being f u n c t i o n a l and of very s i m i l a r performance.

In Fig. 5.23, t h e d e l a y p e r s t a g e and t h e power delay product of an e l e v e n s t a g e r i n g o s c i l l a t o r a r e p l o t t e d as a f u n c t i o n of t h e supply v o l t a g e (VDD). These r e s u l t s were obtained f o r t h e r i n g o s c i l l a t o r o p e r a t i n g i n t h e l a r g e s i g n a l regime, namely, The minimum propagation d e l a y i n v e r t i n g between VDD and ground. o b t a i n a b l e w a s 44 n s e c per s t a g e a t 1 0 v o l t s VDD and t h e minimum power delay product f o r s u s t a i n i n g o s c i l l a t i o n was 4 . 1 pJ. A t 5 v o l t s VDD, t h e propagation delay and t h e power delay product are 57.5 n s e c and 7 pJ, r e s p e c t i v e l y . I s o l a t e d enhancement mode d e v i c e s , which have t h e same geometry a s t h e d r i v e r i n t h e r i n g o s c i l l a t o r c i r c u i t , were a l s o c h a r a c t e r i z e d . A s u r f a c e e l e c t r o n m o b i l i t y of about 300 cm2/

K. F. LEE ETAL..

268

0 0

-"

O O

-0

I

0

a

0

0

c

0

SUBSTRATE C R O U N O t O

0

W

W

0 0

VI +

0

d

a

0

W

0

* 5

5 0 .-

0

0

0

Y

O

0

W

L

0

0

t

u

0-

t

0 0

40

1 RING OSCILLATOR POWER SUPPLY (VOLTS)

FIGURE 5.23. Switching d e l a y and product p e r s t a t e of an e l e v e n s t a g e r i n g o s c i l l a t o r w i t h t h e s u b s t r a t e grounded. The channel l e n g t h of t h e d r i v e r t r a n s i s t o r s i s 6 pm. O1980 IEEE V-sec (compared t o 640 cm2/V-sec measured i n d e v i c e s f a b r i c a t e d i n bulk s i l i c o n s l i c e s ) , a t h r e s h o l d v o l t a g e of about -2 V and a sub-threshold leakage c u r r e n t of about 1 0 nA p e r micron channel width w i t h -5 V g a t e v o l t a g e s and 5 V supply v o l t a g e were measured. I s o l a t e d d e p l e t i o n mode d e v i c e s having t h e same geometry a s t h e load d e v i c e i n t h e r i n g o s c i l l a t o r e x h i b i t e d a t y p i c a l t h r e s h o l d v o l t a g e of -3.6 V and a d r a i n c u r r e n t of t y p i c a l l y The s u b t h r e s h o l d 196 PA a t 5 V supply and w i t h t h e g a t e grounded. c u r r e n t decreased by about 3-112 decades p e r v o l t g a t e v o l t a g e . The shape of t h e s u b t h r e s h o l d I-V c h a r a c t e r i s t i c s and t h e f a c t t h a t t h e leakage c u r r e n t d i d n o t s c a l e w i t h t h e channel width i n d i c a t e t h a t t h e edge-related leakage c u r r e n t dominates, a r e s u l t As a not s u r p r i s i n g s i n c e no channel s t o p implant was used. r e f e r e n c e , t h e propagation d e l a y i n s i m i l a r r i n g o s c i l l a t o r s fabr i c a t e d i n bulk s i l i c o n , w i t h a d e p l e t i o n width of 3 pm between t h e n+ r e g i o n and t h e s u b s t r a t e (and t h u s t h e same n+ t o s u b s t r a t e c a p a c i t a n c e compared t o t h a t provided by t h e 1 pm t h i c k oxide l a y e r ) and o p e r a t i n g i n t h e l a r g e s i g n a l mode i s t y p i c a l l y 36 ns a t 5 V VDD.

5. 5.4.2.3


269

SOI/CMOS C i r c u i t s t5.361

The most comprehensive c i r c u i t evaluation of beam r e c r y s t a l l i z e d p o l y s i l i c o n f i l m s on oxidized s i l i c o n s u b s t r a t e s published These authors f a b r i c a t e d t o d a t e is t h a t of Tsaur e t a l . t5.331. a CMOS t e s t c i r c u i t chip containing s i x a r r a y s of 360 t o 533 para l l e l t r a n s i s t o r s , two 3 l s t a g e r i n g o s c i l l a t o r s and two i n v e r t e r chains. The SO1 s t r u c t u r e s consisted of a O.5pm thick S i f i l m deposited (CVD) on a l p m t h i c k l a y e r of Si02 grown on a 2" S i wafer. The f i l m was capped with 2pm of CVD Si02 and 30 nm of s p u t t e r e d Si3N4 before r e c r y s t a l l i z a t i o n . The r e c r y s t a l l i z a t i o n process employed a g r a p h i t e s t r i p h e a t e r i n place of a cw scanning l a s e r ( d e t a i l s of t h i s r e c r y s t a l l i z a t i o n process a r e described i n Sec. 5.5). The y i e l d of useable devices and c i r c u i t s was i n t h e range of 83%, with l o c a l m e t a l l i z a t i o n d e f e c t s accounti n g f o r approximately 10% of t h e f a i l u r e s . The r i n g o s c i l l a t o r s employed t r a n s i s t o r s with a gate length of 5pm and exhibited t y p i c a l switching delay t i m e s of -2ns and power-delay products of 0.2-0.3 pV a t a supply voltage of 5V. This work suggests t h a t high q u a l i t y c i r c u i t s with y i e l d s comparable t o those of s i m i l a r c i r c u i t s on bulk c r y s t a l l i n e s i l i c o n can be f a b r i c a t e d on 2 i n c h SO1 wafers. Current attempts t o extend t h e wafer s i z e t o 100 and 125mm have l e d t o r e c r y s t a l l i z e d f i l m s t h a t t o d a t e exceed t h e tolerances f o r wafer f l a t n e s s t h a t are acceptable f o r VLSI lithography. 5.4.2.4

Dynamic RAM Cells

.

A promising a p p l i c a t i o n of beam r e c r y s t a l l i z a t i o n € o r memory device f a b r i c a t i o n has been proposed by J o l l y e t a l . i5.371 These authors have developed a dynamic RAM c e l l i n beam r e c r y s t a l l i z e d p o l y s i l i c o n t h a t provides s i g n i f i c a n t advantages over i t s s i n g l e c r y s t a l counterpart. The s t r u c t u r e i s i l l u s t r a t e d schematically i n F i g . 5.24. By placing t h i n oxides both above and below t h e

B. L.

w. L. 0

THIN OXIDE 2

1

n+

FIGURE 5.24.

Dynamic RAM c e l l .

01983 IEEE

*

I...

OXIDE 1

270

K. F. LEE ETAL..

storage region, Jolly et al. were able to double the storage capacitance of the cell. Since the principal property of importance in such a cell is the ratio of storage capacitance to bit line capacitance, doubling the storage capacitance represents a significant improvement in circuit performance. In addition a thick oxide underneath the bit line reduces the bit line capacitance, representing a further improvement of the cell over its singlecrystal counterpart. The complete isolation of the storage region by oxides also reduces the susceptibility of the cell to soft errors resulting from the collection of charges by junction leakage or alpha particle bombardment. The authors point out that long storage times are feasible, limited only by the leakage on the lower surface of the recrystallized film in which the access transistor is fabricated. Techniques for reducing this leakage are discussed in Sec. 5.4.5. 5.4.3

NMOS Logic Circuits in COq Laser Recrystallized Silicon

on Quartz [5.38]

In Sec. 5.3.5 we discussed the work of Johnson et al. [5.30], who showed that high quality single crystal islands could be grown on fused quartz substrates. High performance depletion mode and enhancement mode n-channel MOSFeTs were fabricated in this material, with electron mobilities in excess of 900 cm2/V-sec and leakage currents of less than 1 pA/um of channel width. Building on these basic results, Chiang et al. [5.38] have optimized processing steps involving ion implantation and high temperature annealing cycles with the aid of the SUPREM simulation program [5.351 to achieve simultaneously low leakage currents and voltage thresholds appropriate for NMOS logic circuits. The threshold voltage variation achieved from this work was 0.3 volts and the yield of useable discrete devices was 98%. NMOS ring oscillators with 3 ns propagation delaylstage have been fabricated using this technology. Operational inverters and two-phase dynamic shift registers have also been demonstrated, leading to the prospect of integrating logic circuits and image-sensing arrays of thin film transistors on fused quartz substrates. 5.4.4

Minimum Feature Size Considerations I5.391

All of the devices described to this point have employed geometries that were large enough to include many individual grains in the underlying film. Under these circumstances the properties of the devices are largely independent of the direction of current flow in the channel with respect to the direction of travel of the laser beam during recrystallization. However, this feature is not maintained as the device dimensions shrink.

5.


271

The r o l e of t h e g r a i n boundaries i n t h e l a s e r r e c r y s t a l l i z e d material on t h e performance of f i n e geometry MOSFETS w a s f i r s t addressed by Ng et a l . [5.39] These authors f a b r i c a t e d NOSFETS of various dimensions and c o r r e l a t e d the e l e c t r i c a l characteri s t i c s with channel length and channel width.

.

The output c h a r a c t e r i s t i c s of MOSFETS with d i f f e r e n t channel lengths a r e shown i n Fig. 5.25(a). The f a m i l i a r "kink" e f f e c t i s observed due t o the f l o a t i n g s u b s t r a t e , a c h a r a c t e r i s t i c of For channel lengths of silicon-on-insulator s t r u c t u r e s i5.401 less than 3 pm, t h e r e is a leakage current from source t o d r a i n which is independent of t h e g a t e b i a s . This leakage c u r r e n t is found t o be weakly temperature dependent down t o 77'K. I n order t o determine t h e o r i g i n of t h i s leakage current, it w a s measured a s a function of channel length f o r a f i x e d d r a i n b i a s a t a l a r g e negative g a t e voltage (-3V). The r e s u l t s are shown i n Fig. 5.25 ( b ) . A unique c h a r a c t e r i s t i c of these d a t a i s t h a t between chann e l lengths of 2.3 and 3.3 pm, t h e leakage current drops by more than s i x orders of magnitude. The presence of t h i s "threshold" channel length suggests t h a t t h e leakage is not due t o any mechanism t h a t is proportional ( o r n e a r l y so) t o l / L , such as leakage through g r a i n boundaries o r leakage through the bottom p o l y s i l i c o n /SiOz i n t e r f a c e . For leakage c u r r e n t i n short-channel devices, t h e p o s s i b i l i t y of punch-through e f f e c t s is a l s o eliminated due Ng e t a l . t h e r e f o r e t o i t s d r a i n voltage dependence i5.411. i n t e r p r e t t h e s e r e s u l t s a s a r i s i n g from g r a i n boundary d i f f u s i o n of As from t h e source and d r a i n regions i n t o t h e channels during device f a b r i c a t i o n . After t h e source and d r a i n implantation, t h e wafers were subjected t o a t o t a l of 90 min. a t 900°C, which is a s u f f i c i e n t time-temperature cycle t o d i f f u s e As f a r a s 1.5 pm i n t o t h e g r a i n boundaries [5.42]. This d i f f u s i o n would e x p l a i n t h e abrupt rise i n leakage current below a channel length of = 3 pm. Subsequent work by Ng e t a l . r5.431, i n which source and d r a i n A s implants were annealed by a r a p i d thermal annealing process, enabled g a t e lengths t o be reduced below 1 pm with no s i g n i f i cant i n c r e a s e i n leakage. The annealing cycle i n t h i s case cons i s t e d of a 10 second anneal a t lOOO"C, f o r which A s d i f f u s i o n along t h e g r a i n boundaries would be n e g l i g i b l e . Nineteen s t a g e r i n g o s c i l l a t o r s f a b r i c a t e d by t h i s technique show a propagation delay of 118 ps/stage, which i s t h e f a s t e s t reported t o d a t e f o r SO1 s t r u c t u r e s . A photomicrograph of t h e device and t h e measured o s c i l l a t i o n behavior i s shown i n F i g . 5.26.

.

I n summary, c h a r a c t e r i s t i c s of MOSFETS f a b r i c a t e d i n l a s e r c r y s t a l l i z e d p o l y s i l i c o n show an i n t e r e s t i n g dependence on chann e l l e n g t h . A sharp i n c r e a s e i n source-to-drain leakage current i s observed f o r channel lengths below = 3 pm when long thermal cycles a r e used t o anneal t h e source and d r a i n implants. The e f f e c t i v e e l e c t r o n surface mobility is found t o i n c r e a s e with decreasing channel length, and t h e highest values, compared to

272

K. F. LEE ETM.

FIGURE 5.25(a). Output characteristics of MOSFETS with 30 Dm channel width and varying channel lengths. 01981 IEEE

FIGURE 5.25(b). Source-to-drain leakage current as a function of channel width. Channel width = 120 pm. vds = 0.1 v. O1981 IEEE

5.


273

FIGURE 5.26. MOSFET circuits in Si-on-insulator by laser crystallization and rapid thermal annealing. 01984 Elsevier Science Publishing Co.

those from similar crystallization technique, are obtained. However, in order to capitalize on the increased mobility, enhanced diffusion through grain boundaries must be avoided, for example by using a rapid thermal annealing cycle for annealing. 5.4.5

Three-dimensional Devices on Recrystallized Polysilicon

The experiments described in the previous subsections show that MOSFETs and ring oscillators can be fabricated on the "free" surface of a recrystallized polysilicon film to obtain devices with characteristics very similar to those that would be obtained with single crystal silicon. The interface charge measurements discussed in Sec. 5 . 3 . 3 also indicate that Nf values at the SiQ2/ recrystallized Holysilicon interface can be kept at or below the mid-lOll/cm level. Additional measurements [5.27] indicate that Nf values at the crystalline silicon/Si02 interface can be kept at or below the mid -l0l1/cm2 level. These results suggest that devices can be made in which the bulk silicon is used for one device and the bottom of the recrystallized polysilicon film is used for a second device. A structure that

K . F. LEE E T A L

274

would b e n e f i t d i r e c t l y from t h i s p o s s i b i l i t y is a one-gate-wide CMOS i n v e r t e r i n which the bulk s i l i c o n is used f o r t h e p channel d e v i c e and t h e r e c r y s t a l l i z e d p o l y s i l i c o n f i l m f o r i t s n channel complement. The b a s i c d e v i c e s t r u c t u r e w a s f i r s t c o n s t r u c t e d by Gibbons The j o i n t use of a and L e e [5.26] and i s shown i n Fig. 5.27. s i n g l e g a t e t o d r i v e b o t h the n- and p-channel d e v i c e s l e d K l e i t man 15.441 t o s u g g e s t t h e term JMOS t o d e s c r i b e t h i s s t r u c t u r e . F a b r i c a t i o n d e t a i l s f o r t h i s s t r u c t u r e are d e s c r i b e d by Gibbons and Lee I5.271. I n what f o l l o w s w e d i s c u s s b r i e f l y the f a b r i c a t i o n and b a s i c e l e c t r i c a l c h a r a c t e r i s t i c s of JMOS s t r u c t u r e s We then c o n s i d e r t h a t were made t o e x p l o r e t h e c e n t r a l i d e a . subsequent i n v e s t i g a t i o n s of t h e JMOS s t r u c t u r e as a h i g h packing d e n s i t y form of CMOS, as w e l l as o t h e r three dimensional d e v i c e s t r u c t u r e s t h a t are r e l a t e d t o i t .

,,

LASER-RECRYSTALLISED CVD POLYSILICON

Ai

CON

P+

1

\ n-TYPE (100)

FIGURE 5.27. 01980 IEEE

p+ \GATE OXIDE FOR BOTTOM LEVEL DEVICE

Schematic of a "high-rise"

CMOS s t r u c t u r e .

5.


275

The drain characteristics of devices obtained from the fabrication schedule outlined [5.33] are shown in Figs. 5.27(b) and (c). An analysis of these characteristics shows threshold voltages and carrier surface mobilities comparable to those reported in the work previously described. A similar three dimensional structure with the p channel device on top has been reported by Colinge et al. 15.451, and a self-aligned fabrication technique leading to reduction of overlap capacitance and an improved fabrication schedule has also been published by Goeloe et al. 15.461.

Alternatives to the JMOS form of three dimensional device integration have also been published by a number of authors. Kamins [5.47] has fabricated a CMOS structure with the p-channel devices in a layer of recrystallized polysilicon and the n-channel devices in adjacent, laterally displayed regions of the underlying single-crystal silicon. A schematic illustration of the transistor pair is shown in Fig. 5.28. This process proposed by Kamins allows the use of existing circuit layouts with only minor modification, which is an important feature. Kawamura et al. [5.48] have fabricated a three dimensional CMOS integrated circuit in which one type of transistor is fabricated directly above a transistor of the opposite type, each transistor having its own gate. The structure is shown schematically in Fig. 5.29. Seven stage ring oscillators were fabricated with this device, with a propagation delay of 8.2 ns per stage. One of the most interesting features of the construction proposed by Kawamura et al. is that it can be repeated vertically to stack several transistors with separate gates and proper insulation between each device, using only minor modifications of existing technology. P-CHANNEL TRANSISTOR IN LASER-RECRYSTALLIZED POLY-SILICON

77

N-CHANNEL TRANSISTOR IN SINGLE-CRYSTAL SILICON

p-Si

FIGURE 5.28. Schematic cross section of transistor pair, showing a p-channel transistor in recrystallized polysilicon and an n-channel transistor in the single-crystal substrate. 01982 IEEE

K. E LEE E T A L .

276

'Jss VDD

"OUT

0

?

PSG

sio, n

1

Si

I

FIGURE 5.29. Schematic cross section of the 3-D CMOS IC, showing an n-channel transistor in recrystallized silicon and a pchannel transistor in the single silicon substrate. O l g a 3 IEEE

5.4.6

Memory Applications of Three Dimensional Integration

In the preceding section we have described a representative set of the efforts that have been made to study the fabrication Studies of the and properties of three dimensional devices. speed of such configurations by SPICE computer circuit simulation have shown that, for the configurations so far envisaged at least, planar S O 1 devices are faster than their three dimensional counterparts (Gibbons et al. 15.491). Hence circuit speed is not likely to be a feature of commercially important three dimensional circuits in the near future. However, the significant improvement in device packing density obtainable in three dimensional integration does offer promise of early application. Two types of three dimensional memory circuits have been reported. A stacked CMOS static RAM cell has been described by For maximum packing density, the JMOS conChen et al. [5.50). figuration of Colinge et al. referred to earlier was employed by these authors, in which the n channel devices are fabricated in the single crystal silicon. A schematic illustration of their device is shown in Fig. 5.30. Interest in this circuit arises because it could provide a significant improvement in the noise immunity of the circuit compared to its n-MOS counterpart which uses high value polysilicon resistors as load elements.

5.


POLY P+

FIGURE 5.30.

POLY /

Stacked CMOS structure.

'1983

277

N-

IEEE

A second potential contribution of three dimensional device integration to the construction of memory circuits is that of Sturm et al. [5.51], who have modified the DRAM of Jolly et al. described earlier to obtain significant improvements in the leakage of the access transistor. A schematic of the structure is shown in Fig. 5.31, which can be obtained from the earlier circuit of Jolly et al. by simply folding the plate and a portion of the polysilicon film into the region of thick oxide underneath the access transistor.

The procedure for fabricating this structure is described by Sturm et al. f5.511 and is relatively straightforward. A large reduction of the leakage on the lower surface of the recrystallized film in which the access transistor is fabricated is obtained when sufficient bias is applied to the field plate. Storage times of several hundred seconds are obtained with this improvement, leading to device properties that rival those of DRAMS made in single crystal material.

5.4.7

Stacked MOSFETs in a Single Film of Laser-Recrystallized Polysilicon t5.521

All of the devices envisaged in the work referenced above are based on fabricating only one device in each beam-recrystallized polysilicon film. However, one can imagine multiply-stacked structures in which both sides of a given recrystallized polysilicon film are driven simultaneously by independent gates. Such a structure, in which two transistors are fabricated in the same

K . F. LEE ET A L

278

P

-V

FIGURE 5.31. DRAM.

'1984

Schematic cross section of the folded SO1

IEEE

laser-recrystallized polysilicon film was fabricated for exploratory purposes by Gibbons et al. [5.52]. For convenience in the initial fabrication, the p-channel transistor employed a bulk silicon substrate as a gate and has its channel on the lower surface of the recrystallized polysilicon film. The n channel device employed the upper surface of the laser recrystallized film, which was modulated by an overlying polysilicon gate that was fabricated conventionally. The source and drain areas of the two devices were implanted in a cross-shaped island, as shown in Fig. 5.32. The shape of this configuration suggests the name cross-MOS for this type of 3D integration. While a structure utilizing two transistors in a single device region has been reported in bulk silicon i5.531, the present silicon-on-insulator structure allows for the independent operation of the separate gates. 5.5

IMPROVEMENTS IN THE RECRYSTALLIZATION PROCESS

We now describe a number of experiments which have been performed to improve the crystallographic quality of beam recrystallized films. These improvements include attempts to control the nucleation and growth processes in order to (1) increase the average grain size of the polycrystalline films, (2) form large areas of single crystal material and/or (3) control the crystallographic orientation of the recrystallized material. To understand and appreciate the principles behind the experiments to be described, it is first necessary to understand those factors that influence the crystallographic features of the recrystallized films. We therefore begin this section with a discussion of the nature and origin of the features observed in continuous thin

5.


1

I

n+

i

1 @ 0 p

I

n+

I 1-

I

I50,um

I 1

L-L P+

FIGURE 5.32.

279

-

Schematic top view of cross-MOS structure. O1982 IEEE

films which have been recrystallized using a standard circular laser beam. Subsequent sections will then describe modifications to the film and/or recrystallization procedure which can lead to significant improvements in the quality of the processed films. 5.5.1

Effect of Temperature Gradient

It is observed that the large crystallites which are formed during laser recrystallization with a circular beam are developed at an angle to the laser scan boundaries, producing a chevron structure in the direction of the laser scan i5.61. The direction and curvature of these crystallites can be understood as follows: Let "a" be the radius formed by the laser spot where the polysilicon film is completely molten, as in Fig. 5.33. "a" then defines the radius of the region where long grains are formed. A radial temperature gradient is produced. With the spot stationary and the power of the laser abruptly reduced, the polysilicon film will start cooling fromthe edge toward the center. The direction of recrystallization, being in the direction of the temperature gradient, will then result in the formation of radial grains.

280

K. F. LEE ET AL.

m R e s u l t a n t g r a i n growth

FIGURE 5.33. Schematic illustrating effect of temperature gradient on grain growth. Scan direction is from left to right. Consider now a scanning spot. The part of the spot that falls behind the scan cools o f f and recrystallizes. Since the isotherm is curved, the direction of the thermal gradient is not constant. A s the spot moves away, recrystallization that starts out at a part of the recrystallization front close to the edge of the scan line continues at a lower part of the front close to the center of the line. A curvature in the recrystallization direction is thus produced. One can compute the shape of the recrystallization direction if precise information about temperature profile and recrystallization velocity is known. For slow enough scan velocities, such that steady state temperature calculations apply (an estimate of the error in true temperature due to this assumption gives a value of about 5% at a scan speed of 12 cm/sec), the instantaneous temperature gradient is the same as that of a stationary beam. In particular, if the recrystallization front catches up with the molten spot fast enough, so that the super-cooled melt recrystallizes as soon as the spot is displaced infinitesimally, then the

5.


28 1

direction of the temperature gradient is given by the instantaneous radial vector. Mathematically, from Fig. 5.33, we can write (5.5)

For the initial condition y = a, x = 0, the solution is a + x = - a’-y‘+aln

a 2 - y2 Y

(5.6)

Initially, at the edge of the spot, the grains grow normal to the scan direction. A s they continue to grow, they curve more towards the direction of the scan until at the center the recrystallization direction is along the scan direction. One can see that the recrystallization speed at various parts of the recrystallization front is related to the recrystallization speed at the center by a geometric factor. The recrystallization speed at the center must be equal to the scan speed. 5.5.2

Beam Shaping F5.541

Since the shape of the trailing edge liquid-solid interface strongly influences the shape of the resulting grain structure, it is reasonable to assume that a modification of this interface would result in an altered structure. For instance, if the interface was concave with respect to the solidification front, crystallites would tend to propagate toward the outer edges of the laser scan. Consequently, grains which originate near the outer edges of the front would be consumed at the scan boundaries rather than growing inward in a competing manner as described for the circular case. In the same way, grains which originate near the center of the front would grow outwards towards the scan boundaries, resulting in wider grains. Experiments using this beam shape have been performed and have verified the anticipated behavior. The resulting grain structure was a reverse chevron pattern, fanning out in the direction of the scan. When the liquid-solid interface is curved, as in both of the above cases, grain growth will not be unidirectional. For large area recrystallization, where it is necessary to have many overlapping laser scans, it would be beneficial if unidirectional grain growth were occurring, such that grains grown on a particular scan might be extended in length by subsequent scans. Using a slanted liquid-solid interface, it is possible to accomplish this type of grain growth behavior.

K. F. LEE ET AL.

282

SCAN

+

D I RECTION

D I RECT 1 ON (a 1

Ib) SCAN

D I RECTl ON FIGURE 5.34. Laser beam shapes and grain growth tendencies: (a) circular, (b) crescent, and (c) half moon. With a slanted interface, grains tend to grow from one side of the scanned region to the other. This configuration results in unidirectional grain growth with grains generally rectangular in shape. Subsequent overlapping scans can then extend the length of previously grown grains since the grain growth direction is generally constant. In Fig. 5.34 we schematically summarize the grain growth tendencies resulting from convex, concave and slanted liquid-solid interfaces. Experiments were performed by Stultz and Gibbons [5.54] to study the possibilities of beam shaping. The laser used for these experiments was a 18 W cw argon ion laser operated in the multiline mode. The laser beam was expanded and collimated from about 1 mm to 2 cm in diameter and then focused onto the sample using a 180 mm focal length lens. A spatial filter was used to improve the beam profile. Beam shaping was accomplished by placing thin metal masks of appropriate shapes into the collimated portion of the beam. The samples for recrystallization were mounted on a heated stage (35OOC) which was translated through the fixed laser beam at a rate of approximately 1 cm/s. The scan overlap was nominally 20% of the scan width. The samples used were 0.5 pm thick LPCVD silicon deposited The deposition temperature was 625OC and the deposition was performed using pure silane. The samples intended for electrical characterization were implanted with phosphorus to a dose of lxlOI4 cm-2 at 100 keV. These samples were then capped with 0.5 pm of CVD Si02 to prevent cracking. Contact openings in the oxide were made and the contact areas were implanted with phosphorus to 5x1Ol5 cm-2. Laser recrystallization was performed through the oxide and all electrical measurements were made with the oxide cap in place. on quartz substrates.

The results of recrystallizing the silicon films using a slanted interface were quite dramatic. Single scanned regions

5.


283

comprised p r i n c i p a l l y of r e c t a n g u l a r g r a i n s w e r e observed, w i t h g r a i n boundaries running a t n e a r r i g h t a n g l e s t o t h e d i r e c t i o n of t h e s l a n t e d l i q u i d s o l i d i n t e r f a c e . The m a j o r i t y of t h e g r a i n s were a s long a s t h e s c a n width and about h a l f a s wide i n t h e d i r e c t i o n of t h e scan. Figure 5.35 i s a photomicrograph of a s i n g l e scanned r e g i o n which has been Secco etched [5.551 t o d e l i n e a t e t h e g r a i n boundaries. The s c a n width i s about 65 pm and t h e average As shown, t h e s l a n t e d g r a i n s i z e i s about 65 pm x 25 um. i n t e r f a c e n o t only induced u n i d i r e c t i o n a l g r a i n growth but a l s o s i g n i f i c a n t l y i n c r e a s e d t h e average g r a i n s i z e of t h e r e c r y s t a l l i z e d f i l m , a s compared t o t h a t achieved u s i n g a c i r c u l a r beam. I n o r d e r f o r a r e c r y s t a l l i z e d f i l m t o be compatible w i t h semiconductor p r o c e s s i n g t e c h n i q u e s , t h e s u r f a c e must be smooth and t h e f i l m t h i c k n e s s uniform. The s u r f a c e morphology of r e c r y s t a l l i z e d poly using a c i r c u l a r beam g e n e r a l l y degrades w i t h i n c r e a s e d beam diameter. For diameters g r e a t e r than 100 pm, t h e chevron p a t t e r n begins t o be less d i s t i n c t and random g r a i n growth becomes observable. To compare t h e e f f e c t of t h e s l a n t e d i n t e r f a c e t o t h e convex i n t e r f a c e on t h e s u r f a c e morphology, r e c r y s t a l l i z a t i o n was c a r r i e d out by performing a series of scans u s i n g t h e masked p o r t i o n of t h e beam a s t h e t r a i l i n g edge The followed by a series of scans i n t h e o p p o s i t e d i r e c t i o n . scan width w a s expanded t o about 150 pm. A l l o t h e r experimental The r e s u l t s a r e shown i n Fig. c o n d i t i o n s were h e l d c o n s t a n t . 5.36. The photomicrograph u s i n g Nomarski o p t i c a l microscopy shows t h a t t h e s u r f a c e i s s i g n i f i c a n t l y smoother f o r t h e p o l y s i l i c o n r e c r y s t a l l i z e d with the s l a n t e d i n t e r f a c e .

FIGURE 5.35. Photomicrograph of shaped beam laser r e c r y s t a l l i z e d p o l y s i l i c o n u s i n g a s l a n t e d i n t e r f a c e . The sample was Secco e t c h e d t o d e l i n e a t e g r a i n boundaries.

284

K. F. LEE ETAL.

FIGURE 5.36. Comparison of surface morphology of laser recrystallized polysilicon using slanted versus convex interface.

Previous studies on the structure of recrystallized polysilicon using a circular laser beam have shown that the grains had a mixture of crystal orientation (Sec. 5.1.6). Since films recrystallized using the slanted interface undergo unidirectional solidification, it was felt that a preferred orientation or textured structure might be present. Using a conventional X-ray diffractometer, as-deposited, circular beam recrystallized and shaped beam recrystallized films were analyzed for grain texture. I n Fig. 5.37 we present the results of the diffractometer runs. All data has been corrected for finite film thickness and normalized to a common integrated intensity for direct comparison. A diffractometer trace constructed from powder diffraction data representing an ideal specimen with totally random grain orientation is also presented for comparison. A s shown, the as-deposited and circular beam recrystallized films show no pronounced preferred orientation, while the shaped beam recrystallized film is clearly oriented material. Electrical characterization of the recrystallized films was made using van der Pauw-Hall measurements. Samples of 5 mm x 5 mm were prepared as described above. Circular and shaped beam recrystallization was performed through the oxide cap. The cap was used to suppress film cracking which results from the large difference between the thermal expansion coefficients of the

5.


SHAPED B E A M

285

-

A

CIRCULAR @EAM

A

A

AS DEPOSITED

RANDOM (IDEALIZED)

[I111 I

30

[220]I

[311]

40 50 20 (degrees)

60

(4001 70

FIGURE 5.37. Normalized diffractometer traces of shaped beam recrystallized, circular beam recrystallized, as-deposited and idealized random polycrystalline silicon. silicon film and the underlying quartz substrate. Because of the shaped-beam-induced unidirectional crystal growth, the grain boundaries in the recrystallized film are nearly parallel to one another. In addition, each subsequent overlapping scan tends to extend the length of a grain grown by a previous scan. In this manner, grains with lengths in excess of 250 pm were grown. These grains, however, have large aspect ratios since the grain width is in general not increased by this process. As a result, the measured current-voltage relationship of a film taken parallel to the grain growth direction exhibits a much lower resistivity than that measured across the grain boundaries in the narrow direction. This results in large anisotropies in resistance when measuring the sheet resistivity using the van der Pauw-Hall method. In Table 5.5 we present electrical data obtained to compare shaped with circular beam recrystallized films. The measured carrier concentrations using the Hall technique are consistent with the nominal inpurity implant dose. The second column in Table 5.5 compares the resistance anisotropy observed using shaped versus circular beams. A s expected, the shaped beam gives a higher anisotropy ratio resulting from the aligned grain boundaries. The remaining data in Table 5.5 should be considered for comparison only. It should be realized that because of the abrupt discontinuities in the recrystallized film due to grain

K . F. LEE E T A L .

286

TABLE 5.5. Electrical Characteristics of Shaped Beam and Circular Beam Recrystallized Silicon. Recrystallization Carrier Conc. Conditions

Shaped Beam (6W) Circ-ular Beam (6W)

Resistance Sheet Hall Anisotropy Resistance Mobility (Q/O1 (cm2/V-s)

6.5~101~ 4.1 7 . 1 ~ 1 0 ~ ~ 1.1:l

311 1258

307 69

boundaries and possible microcracks, a literal interpretation of the sheet resistance and Hall mobility cannot be made. However, it is worth noting that the mobility expected from crystalline material with the same impurity concentration is about 300 em2/ V - s , which is the value obtained using a shaped beam. 5.5.3

Effect of High Scan Speed i5.561

The experiments described above were performed at a scan speed where a steady state approximation is valid. Schott [5.561 has examined the regime between pulsed laser annealing and conventional cw laser annealing which is achievable by increasing the scan speed of the laser. A brief discussion of this technique follows

.

Thermal etching of the polysilicon films has been a problem from the beginning of laser recrystallization studies. For typical scan speeds in the 10-50 cm/s range, small laser instabilities or irregularities in the sample can initiate the stripping of the film off the insulator. The problem is less severe for nitride than for oxide due to the better wetting of nitride by molten silicon. Once initiated, thermal etching tends to propagate due to the increased thermal isolation of the remaining material. However, increasing the laser scan speed to 100 cm/s reduces the tendency to runaway etching 15.571, and increasing the scan speed t o 250 cm/s virtually eliminates the problem. Decoration etching studies show no significant change in crystalline quality of the films resulting from the increased scan speed. A related problem involves the use of a capping layer during recrystallization to prevent film movement or to influence crystalline texture. Here again there is a tendency of the molten film to de-wet, creating voids in the recrystallized film. Faster scan speeds (up to 500 cm/s) can minimize this problem.

5.


287

The problems associated with the recrystallization of an MOS poly gatelinterconnect level for reduced resistivity are compounded by the fact that the film is deposited over a patterned oxide. The variations in oxide thickness, and hence, thermal conductivity to the substrate, greatly increase the tendency to thermal etching, as shown in Fig. 5.38. By increasing the scan speed to 250 cm/s, etching can again be eliminated as shown in Fig. 5.39. Melting and large grain recrystallization is confined to the more thermally isolated region over the thicker field oxide and minimal effect is observed over the thin gate oxide. This is acceptable since the desired effect is in long interconnect lines over the field oxide. Even though discrete device performance is not affected, film movement during anneal can degrade circuit performance. If the film is patterned before anneal, however, and very fast scanning is used, the deposited energy density can be reduced t o confine melting to small structures (e.g., narrow interconnect lines) on the field oxide, with minimal film movement and no problems in circuit behavior. For vertical integration, one may wish to recrystallize a poly layer which is deposited over underlying device structures. Thermal pre-anneals are highly undesirable and the nonuniform recrystallization due to variations in thermal conductivity through the underlying substructure is a serious problem. Figure 5.40 shows that increasing the scan speed even further (to 500 cm/s) allows a reasonably uniform anneal over both thick and thin oxide layers simultaneously.

Thermal etching of polysilicon over patterned FIGURE 5.38. oxide laser annealed at 50 cm/s (2OOX). O1982 Elsevier Science Publishing CO.

288

K. F. LEE ET AL.

Elimination of etching for 250 cm/s scan speed. F I G U R E 5.39. Recrystallization of poly over thick (field) oxide but not over thin (gate) oxide (200x1. "1982 Elsevier Science Publishing Co.

F I G U R E 5.40. Uniform recrystallization of poly over both "1982 Elsevier Science thick and thin oxide at 500 cm/s (200x1. Publishing Co.

5.


289

The fact that fast scanning reduces the sensitivity of recrystallization to variations in thermal isolation, has implications for local lateral seeding. Here one tries to obtain epitaxial regrowth of polysilicon in contact with the substrate which acts as a seed for the recrystallization of poly over oxide in adjacent areas (see Sec. 5.5.5 and Ref. [5.58]). Due to the difference in thermal conductivity to the substrate, it is difficult to get conditions of good epi growth in the contact and good recrystallization on oxide, simultaneously. If the oxide layer introduces a step over which the crystallization must proceed, it has been found that extended lateral overgrowth is even more difficult to achieve. Figure 5.41 shows a cross sectional view of an unannealed poly layer going up over a high oxide step of about 0.5 pm. Figure 5.42 shows a failure to induce single-crystal overgrowth for a scan speed of 50 cm/s. Note that there is also some thermal etching. Figure 5.43 shows that if the scan speed is increased to 500 cm/s, significant single-crystal lateral overgrowth occurs. The average extent of the overgrowth is 10 vm. Recently explosive crystallization phenomena have been investigated in amorphous Si films on glass substrates [5.59]. Explosive crystallization was only observed above a certain threshold scan speed. Increasing the scan speed allowed the crystallization front to travel further and thereby increased the period of the characteristic arc-like features of explosive crystallization. Figure 5.44 shows a similar phenomenon occurring in the amorphous layer created by ion implantation. It is first observed (and then often sporadically) at 100 cm/s. A s the scan speed is increased, no change in the period of the features (approximately 2 pm) is observed, however, up to a scan speed of 500 cm/s. The central portion of the track shows melting, while the outer portion does not.

step. co.

FIGURE 5.41. Unannealed poly layer over 0.5 pm field oxide (Marker equals 1 pm.1 "1982 Elsevier Science Publishing

290

K. F. LEE ET AL.

FIGURE 5.42. No l a t e r a l overgrowth a t 50 crn/s. Laser scanned r i g h t t o l e f t , s t e p p e d t o p t o bottom. (Secco e t c h e d , long marker e q u a l s 10 pm.1 01982 E l s e v i e r Science P u b l i s h i n g CO.

FIGURE 5 . 4 3 . S i g n i f i c a n t l a t e r a l overgrowth a t 500 cm/s. Laser scanned t o p t o bottom, s t e p p e d l e f t t o r i g h t . (Secco "1982 E l s e v i e r S c i e n c e e t c h e d , long marker e q u a l s 10 urn.) P u b l i s h i n g Co.

5.


291

FIGURE 5.44. Explosive crystallization in implant amorphized layer in single crystal wafer for scan speeds greater than 100 cm/s (lOOOX). el982 Elsevier Science Publishing Co. 5.5.4

Beam Recrystallization of Amorphous Substrates

Patterned Silicon Films on

We now describe experiments performed to study whether single orientation, single-crystal silicon could be prepared directly on an amorphous substrate (Si02 or Si3Nq) by patterning the polycrystalline film prior to laser processing. This possibility is suggested by the work of Smith et al. [5.60], who found that single-crystal KC1 could be grown by depositing the KC1 on very finely engraved quartz substrates. It should be emphasized, however, that in the present experiments the substrates were engraved or specially prepared in any way. Rather, the polycrystalline film was simply etched to define islands of various sizes.

=

5.5.4.1

Polysilicon Islands on Si3N4/Si Substrates

The possibility that polysilicon islands on amorphous substrates could be recrystallized into single crystal material was first proven by Gibbons et al. [5.61]. The polysilicon samples used for this study were 0.55 um thick, deposited by LPCVD on crystalline Si onto which had been deposited a 1000 W layer of Si3N4 (LPCVD). Islands of polycrystalline Si were then formed by standard photolithographic techniques. The islands ranged in size from 2 x 20 vm to 20 x 160 pm. The 2x20 pm islands were arranged with the long dimension both parallel and perpendicular to the laser scan direction.

292

K. F. LEE ET AL

A f t e r formation of t h e i s l a n d s , some of t h e wafers were m u l t i p l y implanted w i t h phosphorus according t o t h e f o l l o w i n g schedule: 3 x 1014/cm2 a t 50 keV, 6 x 1014/cm2 a t 100 keV, cm2 a t 150 keV, and 6 x 1015/cm2 a t 100 keV. The samples were then i r r a d i a t e d w i t h a scanning c i r c u l a r beam. The samples were h e l d t o a h e a t e d b r a s s sample h o l d e r by a vacuum chuck d u r i n g a n n e a l i n g . S u b s t r a t e t e m p e r a t u r e s of 3 5 O O C were used f o r a l l experiments r e p o r t e d h e r e . The l a s e r was b o laser power/scan rate condifocused i n t o a 40 u m s p o t . 9 W a t -0.15 cm/sec and 11 W a t 1 2 cm/sec. t i o n s were used: Nomarski o p t i c a l o b s e r v a t i o n of t h e annealed f i l m s r e v e a l e d smooth f e a t u r e l e s s s u r f a c e s i n a l l 2 x 20 pm i s l a n d s , independent of t h e i r o r i e n t a t i o n w i t h r e s p e c t t o t h e scanning d i r e c t i o n . F e a t u r e l e s s s u r f a c e s were a l s o observed on 12 x 30 u m L shaped i s l a n d s . However, t h e 20 x 160 u m i s l a n d s show s u r f a c e s t r u c t u r e s i m i l a r t o t h a t o b t a i n e d on continuous f i l m s , independent of t h e o r i e n t a t i o n of t h e i s l a n d w i t h r e s p e c t t o t h e beam. I n Fig. 5.45(a) a r e p r e s e n t a t i v e scanning e l e c t r o n micrograph of a p a t t e r n e d p o l y s i l i c o n s t r i p e ( 2 x 20 Vm) a r r a y a f t e r scanning laser a n n e a l i n g i s shown. No s i g n i f i c a n t a l t e r a t i o n s i n t h e morphology or geometric f e a t u r e s of t h e s t r i p e p a t t e r n s were observed i n any of t h e i n d i v i d u a l i s l a n d s t r u c t u r e s examined a f t e r laser a n n e a l i n g f o r e i t h e r of the l a s e r power/scan r a t e c o n d i t i o n s employed. I n a d d i t i o n , t h e s u r f a c e s of t h e s t r i p e s shown i n

FIGURE 5.45. E l e c t r o n micrographs of laser-annealed i s l a n d arrays. ( a ) Scanning e l e c t r o n micrograph of laser-annealed i s l a n d s i n Si3N4 f i l m s ; ( b ) b r i g h t - f i e l d t r a n s m i s s i o n e l e c t r o n micrograph of laser-annealed i s l a n d s t r u c t u r e and s e l e c t e d a r e a d i f f r a c t i o n p a t t e r n ( r i g h t i n s e t ) t y p i c a l of r e g i o n .

5.


293

F i g . 5.45(a) appear smooth and s t r u c t u r e l e s s with an apparent absence of s u r f ace roughness t y p i c a l l y noted i n as-deposited polycrystalline films.

To examine t h e i n t e r n a l m i c r o s t r u c t u r e and c r y s t a l l i n i t y of t h e i s l a n d s t r u c t u r e s , samples f o r TEM/TED a n a l y s i s u s i n g a j e t t h i n n i n g technique were p r e p a r e d . I n a l l cases a t h i n (300 A ) n i t r i d e l a y e r remained suspended a c r o s s a c e n t r a l h o l e w i t h t h e p o l y s i l i c o n i s l a n d s a v a i l a b l e f o r observation i n the e l e c t r o n microscope. The as-deposited i s l a n d s t r u c t u r e s were composed of t h e p o l y c r y s t a l l i n e g r a i n s ranging from 200 t o 500 8, i n average s i z e . In a l l cases s e l e c t e d a r e a d i f f r a c t i o n p a t t e r n s showed continuous r i n g p a t t e r n s t h a t a r e t y p i c a l l y of ( f i n e g r a i n e d ) In c o n t r a s t , the i o n p o l y c r y s t a l l i n e f i l m s ( s e e F i g . 5.45(b). implanted s t r u c t u r e s showed an absence of s t r u c t u r e and t h e recorded d i f f r a c t i o n p l a t e s e x h i b i t t h e expected amorphous p a t t e r n s . A f t e r laser annealing, t h e 2 x 20 pm i s l a n d s show an absence of f i n e g r a i n e d o r amorphous s t r u c t u r e and are completely r e c r y s t a l l i z e d t o form s i n g l e c r y s t a l s t r i p e s on t h e amorphous Si3N4 I d e n t i c a l r e s u l t s were obtained i n both s u b s t r a t e [ F i g . 5.45(b)] t h e as-deposited ( p o l y c r y s t a l l i n e ) and i o n implanted (amorphous) s t r i p e s , s u g g e s t i n g t h a t t h e formation of r e c r y s t a l l i z e d s i n g l e c r y s t a l i s l a n d s i s r e l a t i v e l y independent of t h e i n i t i a l c r y s t a l l i n e state of t h e s t a r t i n g m a t e r i a l f o r t h e experimental condit i o n s used i n t h i s s t u d y .

.

Throughout t h e e n t i r e l e n g t h of t h e s t r i p e shown i n F i g . 5.45(b) t h e r e a r e no microscopic d e f e c t s , and s e l e c t e d area d i f f r a c t i o n p a t t e r n s ( r i g h t i n s e t ) i n d i c a t e t h a t t h e i s l a n d is s i n g l e c r y s t a l and of (100) o r i e n t a t i o n . Of a d d i t i o n a l i n t e r e s t is the f a c t that t h e r e c r y s t a l l i z a t i o n i s not dependent on t h e r e l a t i v e p o s i t i o n i n g of t h e i s l a n d s t r u c t u r e with r e s p e c t t o laser s c a n d i r e c t i o n . Experiments conducted on samples i n which t h e laser s c a n d i r e c t i o n was p a r a l l e l o r perpendicular t o t h e long a x i s of t h e s t r i p e x h i b i t i d e n t i c a l r e s u l t s w i t h no a p p a r e n t differences i n structure. To f u r t h e r i n v e s t i g a t e t h e c r y s t a l l i n i t y of t h e i s l a n d s t r u c t u r e s approximately one h a l f of t h e t h i c k n e s s of t h e i s l a n d s t r u c t u r e s w a s removed by chemical t h i n n i n g , l e a v i n g t h e Si3N4 f i l m i n t a c t a s a support membrane. S e l e c t e d area d i f f r a c t i o n a g a i n showed only s i n g l e c r y s t a l (100) p a t t e r n s , confirming t h a t r e c r y s t a l l i z a t i o n occurred throughout t h e e n t i r e t h i c k n e s s of t h e island. Examination of t h e l a r g e r (20 p m x 160 pm) i s l a n d s t r u c t u r e s and "t-shaped" a r r a y s (12 pm x 30 pm) showed t h a t laser a n n e a l i n g r e s u l t e d i n t h e formation of l a r g e p o l y c r y s t a l l i n e g r a i n s of 10 um maximum l e n g t h , t y p i c a l l y arranged i n a chevron p a t t e r n

K . F LEEETAL

294

p o i n t i n g i n t h e d i r e c t i o n of t h e l a s e r scan. The occurrence of s t a c k i n g f a u l t s w a s a l s o noted i n some s t r u c t u r e s , independent of i s l a n d dimensions. The n u c l e a t i o n of s t a c k i n g f a u l t s i s n o t thought t o be e s s e n t i a l f o r regrowth s i n c e smaller s i n g l e c r y s t a l i s l a n d a r r a y s and l a r g e columnar s t r u c t u r e s a r e observed c o n t a i n i n g no s t a c k i n g f a u l t s o r o t h e r microscopic d e f e c t s . From t h e d a t a o b t a i n e d w e can conclude t h a t d e f e c t f r e e s i n g l e c r y s t a l S i s t r i p e s of (100) o r i e n t a t i o n can be formed on amorphous Si3N4 f i l m s by laser r e c r y s t a l l i z a t i o n .

5.5.4.2

C o n t r o l of

Lattice Heat Flow [5.62]

A thorough i n v e s t i g a t i o n of t h e c o n d i t i o n s n e c e s s a r y f o r producing c r y s t a l l i n e o r n e a r l y c r y s t a l l i n e i s l a n d s on amorphous These s u b s t r a t e s has been performed by B i e g e l s e n e t a l . i5.621. a u t h o r s demonstrated how s e l f - a l i g n i n g techniques coupled w i t h choice of i s l a n d shape and s w e p t zone m e l t i n g can l e a d t o c o n t r o l of n u c l e a t i o n and growth. In t h e i r experiments, t h e TEMoo o u t p u t of a cw argon i o n l a s e r was formed i n t o a c o l l i m a t e d e l l i p t i c a l beam (by a c y l i n d r i c a l t e l e s c o p e ) and focused t o a s p o t -3Ox90pm. This s p o t was scanned (normal t o i t s wide dimension) a t 1 cm/sec over t h e c e n t r a l axes of t h e 0.5-pm-thick chemical-vapordeposited (CVD) i s l a n d s . The blackbody r a d i a t i o n from t h e h e a t e d zone was imaged o n t o a v i d i c o n e i t h e r i n t r a n s m i s s i o n when t r a n s p a r e n t s u b s t r a t e s (e.g. fused s i l i c a ) were used, o r i n r e f l e c t i o n , A microscope when opaque s u b s t r a t e s (e.g., s i l i c o n ) were used. lamp i s a l s o used t o i l l u m i n a t e t h e n o n r a d i a t i n g f e a t u r e s . F i g u r e 5.46(a) shows a s t a t i o n a r y molten zone d w e l l i n g i n a 10-vm-wide i s l a n d d e p o s i t e d on s i l i c a . The necking down of t h e zone a t t h e edges c l e a r l y i n d i c a t e s edge cooling. This i s a r e s u l t of t h e f a c t t h a t only t h e i s l a n d i s absorbing. The t r a n s p a r e n t s u b s t r a t e remains c o l d i n t h e "sea" around t h e i s l a n d and (The d a r k e r r e g i o n i n t h e c e n t e r of a c t s as a l a t e r a l h e a t s i n k . t h e s p o t a r i s e s from a reduced c o l l e c t i o n e f f i c i e n c y caused by An example of t h e s u r f a c e f l u c t u a t i o n s i n the molten region.) o p p o s i t e s i t u a t i o n i s shown i n Fig. 5.46(b). Here t h e i s l a n d s r e s i d e on p r e d e p o s i t e d l a y e r s of CVD p o l y s i l i c o n (1 ym t h i c k ) and t h e r m a l l y grown oxide (88 nm). The oxide l a y e r i s maximal a n t i r e f l e c t i o n t h i c k n e s s a t room temperature. (Even a t h i g h e r t e m p e r a t u r e s , a s t h e o p t i c a l p r o p e r t i e s change, t h e a b s o r p t i o n i n t h e s i l i c o n l a y e r i s always g r e a t e r than t h a t w i t h no addiIn t h i s c o n f i g u r a t i o n more power i s d i e l e c t r i c tional layer.) absorbed by t h e sea than by t h e uncoated i s l a n d s , so t h a t a n e t h e a t i n f l u x t o t h e i s l a n d r e s u l t s ( o r , more a c c u r a t e l y , t h e r e i s less h e a t l o s s t o t h e s u b s t r a t e a t t h e i s l a n d p e r i m e t e r t h a n a t i t s c e n t e r ) . The r e s u l t i n g edge h e a t i n g h e r e may be u n d e s i r a b l y strong. To o p t i m i z e t h e l a t e r a l thermal p r o f i l e (and t o u s e t h e

5.


295

FIGURE 5.46. Video monitor display of blackbody radiation from silicon islands in various configurations: (a) deposited on bare Si02, (b) continuous polysilicon layer and antireflection dielectric layer interposed, (c) continuous polylayer and dual layers interposed, (d) moat in continuous surround. 01981 Elaevier Science Publishing Co.

minimum necessary heating of the substrate, thus minimizing substrate damage) one can utilize other configurations as indicated in Figs. 5.46(c) and 5.46(d). Figure 5.46(c) is the same as Fig. 5.46(b) with the addition of 64 nm of CVD silicon nitride over the oxide below the islands. (However, this nitride can also encapsulate the islands - see later.) This dual dielectric coating is a maximally reflective coating, and the absorption in the sea is intermediate between situations 5.46(a) and 5.46(b). A configuration which is much simpler from a processing point of view and which can be easily and spatially nonuniformly tailored is shown in Fig. 5.46(d). This is the same as case 5.47(a) except that instead of photolithographically stripping away the entire sea, only a moat (here -4 pm wide) is removed. A demonstration of the results of heat flow control is provided by TEM. At relatively low laser powers, for the case of edge cooling, large grains grow in the island centers and only small grains at the edges. Conversely, for edge enhanced heating, one can observe small-grain development at the island center. In both cases at higher laser powers the small grains no longer form; however, for the regions in which small grains would have grown at low power, there still exists a higher probability of competitive nucleation. This edge enhanced heating explains why in the fortuitously chosen earliest embodiments of this technology (i.e., polysilicon islands on silicon nitride coated crystal silicon substrates) single-crystal islands could be grown, which were larger than any grains grown under identical conditions (i.e., on the same wafer, in the same laser scan) in continu.ous polyregions [5.54].

296

K. F. LEE ETAL.

FIGURE 5.47. SEM photographs of laser annealed i s l a n d s on d u a l d i e l e c t r i c on c r y s t a l s i l i c o n s u b s t r a t e s : ( a ) unencapsulated, ( b ) 20 nm Si3N4 e n c a p s u l a t i o n . 01981 Elsevier Science Publishing Co. For t h e slow s c a n s rates used, t h e molten zone i n a b a r e i s l a n d i s a convex, surface-tension-determined h i l l . Shock d r i v e n r i p p l e s quench i n t o t h e s o l i d i f y i n g topography and mass t r a n s p o r t o c c u r s t o t h e t r a i l i n g edge of t h e i s l a n d . F i g u r e 5.47(a) i s a scanning e l e c t r o n microscopy photograph of two such i s l a n d s viewed a t a n e a r g l a n c i n g a n g l e (-10') t o enhance t o p o g r a p h i c s e n s i t i v ity. (The l e f t i s l a n d w a s scanned away from t h e viewer, t h e r i g h t It i s found t h a t (20 nm of an e n c a p s u l a t i n g toward t h e viewer.) l a y e r , e.g. low-pressure CVD s i l i c o n n i t r i d e , r e d u c e s t h e above mentioned e f f e c t s of s u r f a c e t e n s i o n on a f r e e s u r f a c e . F i g u r e 5.47(b) shows an e n c a p s u l a t e d i s l a n d annealed under c o n d i t i o n s i d e n t i c a l t o t h o s e of Fig. 5.47(a). In both cases complete m e l t ing h a s occurred. This i s a w o r s t case comparison. Encapsulation i n c r e a s e s t h e o p t i c a l a b s o r p t i o n i n t h e i s l a n d s and t h e r e f o r e l e a d s t o a lower v i s c o s i t y of t h e s i l i c o n . T h e r e f o r e one would expect g r e a t e r mass t r a n s p o r t were i t n o t f o r t h e p h y s i c a l encaps u l a t i o n . Moreover, a t no power d i d melted unencapsulated i s l a n d s n o t show mass t r a n s p o r t . ) F l a t i s l a n d s can t h u s be achieved w i t h n e g l i g i b l e mass t r a n s p o r t . A further considerable benefit of e n c a p s u l a t i o n arises from i t s s u p p r e s s i o n of t h e n u c l e a t i o n of a b l a t i o n , i.e., s u r f a c e f l u c t u a t i o n s do n o t punch through and change t h e topology. One can t h u s i r r a d i a t e w i t h h i g h e r power a n d / o r f o r c e t h e molten f i l m t o h i g h e r t e m p e r a t u r e s t o improve c r y s t a l growth. Another c r u c i a l s t e p i n t h e growth of s i n g l e - c r y s t a l i s l a n d s i s t h e a b i l i t y t o f o r c e n u c l e a t i o n a t a s i n g l e p o i n t . The techn i q u e used h e r e i s i s l a n d p e r i m e t e r d e f i n i t i o n , i.e., a t a p e r a t t h e l e a d i n g edge of t h e i s l a n d . As t h e zone p a s s e s , t h e " p o i n t " c o o l s f i r s t and a c t s a s a seed f o r a l l subsequent growth on t h e

5.


297

i s l a n d . (Other t e c h n i q u e s , e.g. r e c t a n g u l a r i s l a n d s w i t h t a p e r e d moat width might a l s o provide t h e same r e s u l t . ) For i l l u s t r a t i o n Fig. 5.48 shows TEM b r i g h t f i e l d images of two i s l a n d s scanned simultaneously from l e f t t o r i g h t . The i s l a n d w i t h t h e l e a d i n g t a p e r i s s i n g l e c r y s t a l l i n e ( a s shown by s e l e c t e d a r e a d i f f r a c t i o n ) w i t h a maximum width of 20 um, whereas t h e i s l a n d w i t h a f l a t l e a d i n g edge has m u l t i p l y nucleated. The curving dark bands ( l a b e l e d A i n Fig. 5.48) a r e e x t i n c t i o n and bend contours [5.63]. These a r e a s s o c i a t e d w i t h s i n g l e - d i f f r a c t i o n d i r e c t i o n s This example w a s chosen and are continuous a c r o s s t h e i s l a n d . because i t a l s o i l l u s t r a t e s t h e dominant s t r u c t u r a l d e f e c t i n The s t r a i g h t bordered dark areas ( l a b e l e d t h i s growth method. B ) are s t a c k i n g f a u l t s . The lower i s l a n d i n Fig. 5.48 i s s i n g l e c r y s t a l l i n e i n t h e sense t h a t t h e r e a r e no g r a i n boundaries, and s e l e c t e d a r e a d i f f r a c t i o n i s everywhere t h e same. Frequently twins a r e a l s o formed t o r e l a x t h e s t r a i n s developed e i t h e r i n t h e growth process or a r i s i n g from d i f f e r e n t i a l c o n t r a c t i o n r e l a t i v e t o t h e s u b s t r a t e o r encapsulant d u r i n g cool down.

TEM photographs of l a s e r annealed i s l a n d s : unFIGURE 5.48. encapsulated on d u a l d i e l e c t r i c and on c r y s t a l s i l i c o n s u b s t r a t e . 01981 Elsevier Science Publishing Cow

K . F. LEE ET AL.

298

I s l a n d s 20 v m wide have been grown w i t h no l a r g e d e f e c t s and a low d e n s i t y of small d i s l o c a t i o n l o o p s . However, c o n t r o l of stress and d e f e c t f o r m a t i o n remains as a n i m p o r t a n t problem. No attempt h a s been made y e t t o c o n t r o l t h e c r y s t a l a x i s normal t o the substrate.

5.5.4.3

Use of A n t i r e f l e c t i n g S t r i p e s t o C o n t r o l Heat Flow

An i n t e r e s t i n g technique f o r c o n t r o l l i n g t h e h e a t f l o w based on r e a d i l y a v a i l a b l e s i l i c o n f a b r i c a t i o n technology h a s been These a u t h o r s u s e a n t i published by Colinge e t a l . i5.641. r e f l e c t i n g ( A R ) s t r i p e s ( d e p o s i t e d by LPCVD on t h e p o l y s i l i c o n f i l m p r i o r t o r e c r y s t a l l i z a t i o n and t h e n p a t t e r n e d by s t a n d a r d p h o t o l i t h o g r a p h i c t e c h n i q u e s ) t o a s s u r e t h a t more laser energy i s coupled i n t o t h e f i l m below t h e s t r i p than i n t h e open area between s t r i p e s (see F i g . 5.49). The s t r i p e s are t y p i c a l l y 1020 pm wide and a r e s e p a r a t e d by 30-50 v m s p a c e s . The laser b e a m i s scanned p a r a l l e l t o t h e s t r i p e s . S i n g l e c r y s t a l material i s propagated between t h e s t r i p e s (and p a r a l l e l t o them) n e a r t h e c e n t e r of t h e beam. Grain boundaries t e n d t o form a t t h e edges of t h e open area and t o accumulate underneath t h e AR' s t r i p e s s i n c e t h i s i s t h e material t h a t is h o t t e s t and r e c r y s t a l l i z e s last.

S i 3 N 4 AR STRIPES POLYSiLlCON

FILM

-

J -OXIDE

SUEST R AT E

FIGURE 5.49. Use of a n t i r e f l e c t i n g s t r i p e s t o c o n t r o l h e a t flow ( a f t e r C o l i n g e , e t a l . L5.641).

5. 5.5.5

5.5.5.1


299

Seeded Growth F i r s t Experiment

I n t h e previous s e c t i o n s of t h i s c h a p t e r w e have d i s c u s s e d a v a r i e t y of approaches t h a t have been used t o r e c r y s t a l l i z e f i l m s i n t o s i n g l e ( o r " n e a r l y - s i n g l e " ) c r y s t a l m a t e r i a l , i n which t h e f i l m i s d e p o s i t e d d i r e c t l y on an amorphous s u b s t r a t e p r i o r t o r e c r y s t a l l i z a t i o n . However, i f c r y s t a l l i n e s i l i c o n s u b s t r a t e s are employed, o r i e n t e d s i n g l e c r y s t a l s i l i c o n - o n - i n s u l a t o r f i l m s To accomplish t h i s end, a p o l y s i l i c o n f i l m i s can be grown. d e p o s i t e d on an oxided s i l i c o n s u b s t r a t e i n which a c e r t a i n r e g i o n of t h e oxide has been etched away t o expose t h e s i n g l e crystal substrate. By scanning a focused cw argon laser beam o n t o t h e a r e a where t h e p o l y c r y s t a l l i n e s i l i c o n i s deposited d i r e c t l y on t h e exposed s i l i c o n s u b s t r a t e , t h e p o l y c r y s t a l l i n e s i l i c o n i s converted i n t o an e p i t a x i a l l a y e r by a l i q u i d phase process. By scanning t h e laser beam from t h e e p i t a x i a l r e g i o n t o t h e region where t h e p o l y c r y s t a l l i n e s i l i c o n i s d e p o s i t e d on t h e s i l i c o n dioxide l a y e r , t h e polycrystalline silicon is converted i n t o a s i n g l e c r y s t a l through a a seeded growth p r o c e s s , where t h e p r e v i o u s l y formed e p i t a x i a l l a y e r i s used as the seed. This process i s named " l a t e r a l seeding" [5.58] and is shown scheLam et a l . [5.58] f i r s t explored t h i s m a t i c a l l y i n Fig. 5.50. technique as a method f o r o b t a i n i n g l a r g e area, c o n t r o l l e d o r i e n t a t i o n , s i n g l e c r y s t a l s i l i c o n films.

LASER BEAM

INDUCED EPITAXY LAYER

FIGURE 5.50. Schematic r e p r e s e n t a t i o n of t h e l a t e r a l seedi n g process. Reprinted by permission of t h e p u b l i s h e r , t h e Electrochemical S o c i e t y , Inc.

300

K . F. LEE ETAL.

S i n g l e c r y s t a l s i l i c o n wafers of (100) o r i e n t a t i o n were used by Lam e t a l . as t h e s u b s t r a t e . A 140 nm t h i c k l a y e r of CVD s i l i c o n n i t r i d e was d e p o s i t e d o n t o t h e s u r f a c e of t h e w a f e r s . P h o t o r e s i s t w a s a p p l i e d t o t h e s u r f a c e of t h e n i t r i d e l a y e r and r e c t a n g u l a r p a t t e r n s of v a r i o u s s i z e s were d e f i n e d p h o t o l i t h o g r a p h i c a l l y . After developing and baking t h e p h o t o r e s i s t , t h e exposed n i t r i d e r e g i o n s were removed by a plasma e t c h i n g techn i q u e . A 0.55 pm t h i c k l a y e r of t h e t h e n exposed s i l i c o n w a s removed by a n a n i s o t r o p i c plasma e t c h . After removing t h e photor e s i s t , t h e w a f e r s were placed i n an o x i d a t i o n f u r n a c e set a t 1000°C w i t h a steam ambient, timed t o grow a one micron t h i c k l a y e r of o x i d e , s o t h a t t h e o x i d e s u r f a c e w a s almost c o p l a n a r w i t h t h e o r i g i n a l s i l i c o n s u r f a c e . After e t c h i n g t h e wafers i n a 10%HF s o l u t i o n f o r 90 sec, t h e n i t r i d e l a y e r was removed by e t c h i n g i n h o t phosphoric a c i d . The w a f e r s w e r e t h e n e t c h e d i n a 10% HF s o l u t i o n f o r 30 sec, d r i e d and loaded immediately i n t o a low p r e s s u r e CVD r e a c t o r , where a 0.5 v m t h i c k l a y e r of p o l y s i l i c o n w a s d e p o s i t e d a t 620°C. A thermal a n n e a l i n n i t r o g e n a t l l O O ° C f o r 1 hour completed t h e p r e p a r a t i o n . It was shown t h a t t h e l l O O ° C a n n e a l s i g n i f i c a n t l y reduces t h e "etching" (removing t h e p o l y s i l i c o n from t h e s u r f a c e of t h e i n s u l a t o r ) of t h e p o l y s i l i c o n Without t h i s d u r i n g t h e l a s e r r e c r y s t a l l i z a t i o n p r o c e s s I5.21 thermal a n n e a l i n g s t e p , t h e proper beam power window f o r t h e r e c r y s t a l l i z a t i o n of t h e polysilicon-on-oxide r e g i o n i s t y p i c a l l y 20.5W. Furthermore, t h e beam power f o r proper l i q u i d phase regrowth of t h e p o l y s i l i c o n o n - s i l i c o n r e g i o n i s h i g h e r t h a n t h e beam power f o r t h e r e c r y s t a l l i z a t i o n of t h e polysilicon-on-oxide r e g i o n due t o t h e h i g h e r thermal c o n d u c t i v i t y of t h e p o l y s i l i c o n on-silicon r e g i o n . Consequently, a beam power l e v e l chosen t o induce e p i t a x i a l regrowth i n t h e p o l y s i l i c o n - o n - s i l i c o n region w i l l be e x c e s s i v e f o r t h e r e c r y s t a l l i z a t i o n i n t h e p o l y s i l i c o n on-oxide r e g i o n and w i l l u s u a l l y r e s u l t i n t h e removal of t h e polys i l i c o n from t h e oxide s u r f a c e . The thermal a n n e a l i n g widens t h e window of beam power f o r proper r e c r y s t a l l i z a t i o n of t h e polysilicon-on-oxide area t o +3W. The e x a c t mechanism is not pres e n t l y u n d e r s t o o d . This thermal a n n e a l i n g s t e p a l l o w s t h e u t i l i z a t i o n of a s i n g l e beam power l e v e l t o be used f o r both r e g i o n s s i m u l t a n e o u s l y . The f i n i s h e d s t r u c t u r e is shown i n F i g . 5.51.

.

The laser-induced r e c r y s t a l l i z a t i o n was performed w i t h a scanning cw argon i o n laser system. Two d i f f e r e n t l e n s systems, a s i n g l e 190 mm f o c a l l e n g t h s p h e r i c a l l e n s and a combination of a 300 and a 100 mm f o c a l l e n g t h c y l i n d r i c a l l e n s e s , were u s e d . Most of t h e experiments d i s c u s s e d were performed w i t h t h e s p h e r i cal lens. The raster l i n e s c a n speed was 10.4 c m p e r sec and t h e l i n e - t o - l i n e s t e p s i z e v a r i e d from 5 t o 20 pm. The w a f e r s w e r e mounted such t h a t t h e l i n e s were scanned p a r a l l e l and perp e n d i c u l a r t o t h e oxide and s i n g l e c r y s t a l s u b s t r a t e i n t e r f a c e . The wafers were h e l d w i t h a vacuum chuck and were h e a t e d t o a temperature as high as 500°C.

5.


301

FIGURE 5.51. Cross-sectional electron micrograph of the finished structure. The oxide layer is 0.96 pm thick. The polysilicon layer is 0.5 pm thick. The lines below the oxide region are artifacts of the sample preparation. Reprinted by permission of the publisher, the Electrochemical Society, Inc.

The wafers were studied by optical microscopy, using a Secco etch [5.14], to decorate the grain boundaries and defects. Electron channeling was used to study the crystallographic orientation of the laser recrystallized silicon-on-oxide material. An optical interference contrast micrograph of a typical laterally seeded silicon-on-oxide area is shown in Fig. 5.52. It is generally observed that when the silicon-on-oxide area is small, such as the smaller squares at the bottom of the picture, no grain boundaries are observed, indicating that the silicon layer is a single crystal. For the silicon-on-oxide areas that are larger in size, large single crystal regions that are free of grain boundaries are obtained immediately adjacent to the epitaxial regions, but the entire silicon-on-oxide region is not of one orientation.

302

K . F. LEE ETAL

FIGURE 5.52. An o p t i c a l i n t e r f e r e n c e c o n t r a s t micrograph of a l a t e r a l s e e d i n g area a f t e r e t c h i n g i n a Secco e t c h f o r 1 0 s e c . The two l a r g e s q u a r e silicon-on-oxide areas are 70 x 70 pm2 i n s i z e . The 1 3 W laser beam scanned from l e f t t o r i g h t and t h e l i n e s s t e p p e d from t o p t o bottom. R e p r i n t e d by p e r m i s s i o n of t h e p u b l i s h e r , t h e E l e c t r o c h e m i c a l S o c i e t y , Inc. It is The l a s e r s c a n l i n e s are a l s o e v i d e n t i n F i g . 5 3 2 . i n t e r e s t i n g t o n o t e t h a t t h e s i n g l e c r y s t a l r e g i o n extended o v e r t h r e e successive scan l i n e s , i n d i c a t i n g t h a t t h e s i n g l e c r y s t a l silicon-on-oxide a r e a formed d u r i n g t h e f i r s t s c a n s e r v e d a s t h e seed f o r t h e growth d u r i n g t h e subsequent s c a n s . T h e r e f o r e , t h e s i n g l e c r y s t a l area indeed propagated l a t e r a l l y , u s i n g t h e prev i o u s l y formed s i n g l e c r y s t a l as t h e s e e d .

5.5.5.2

Seeded Growth w i t h a P a t t e r n e d A n t i - R e f l e c t i o n Coati n g [5.651.

W e d i s c u s s e d i n Sec. 5.5.4.3 a technique f o r employing a patterned anti-reflection c o a t i n g t o manage t h e l a t e r a l h e a t flow problem i n a r e c r y s t a l l i z i n g f i l m . Drowley e t al. [5.65] have combined t h i s i d e a w i t h t h e i d e a of seeded growth t o produce r e c r y s t a l l i z e d m a t e r i a l t h a t i s completely f r e e of g r a i n boundari e s except underneath t h e AR c o a t i n g . The AR c o a t i n g p a t t e r n used by Drowley e t a l . c o n s i s t e d of a series of p a r a l l e l s t r i p e s A 50 v m t e r m i n a t i n g i n s e e d i n g windows, as shown i n F i g . 5.53. x 250 v m e l l i p t i c a l laser beam was used f o r r e c r y s t a l l i z a t i o n . The beam w a s scanned p e r p e n d i c u l a r t o t h e s t r i p e s , w i t h t h e l o n g a x i s of t h e beam p a r a l l e l t o t h e s c a n d i r e c t i o n . The beam was s t e p p e d by 1-2 urn between s u c c e s s i v e s c a n s t o promote c r y s t a l growth. The g r a i n boundaries were completely confined t o t h e r e g i o n beneath t h e AR s t r i p e s .

5.


SEED W I Z W

303

POLYSILICON ONOXIDE

I-

.-

. AR COATING STRIPES

LASER BEAM (DASHED1 MELT FRONT (SOLID1

a)

'

-

I

SCAN STEP DIRECTION AN0 DIRECTloN~ CRYSTAL GROWTH DIRECTION

AR COATING 0.5 prn POLYSILICON

7 b)

1.2 prn SiOI

L Si SUBSTRATE

FIGURE 5.53. (a) The recrystallization scheme. (b) Crosssection through the seeded SO1 structure. O1984 E l s e v i e r Science Publishing CO. Propagation distances of up to several millimeters can be obtained for a structure containing 10 pm wide AR stripes and 10 pmwide spaces. With spaces greater than 20 pm between the AR stripes, single crystal propagation is limited to 30-50 pm from the seed edge; at that point new (typically low angle) grain boundaries nucleate and propagate in the space, similar to the behavior seen in lateral laser epitaxy without the patterned AR coating. 5.5.5.3

Experiments in Seeded Island Growth

Trimble et al. [5.66] have demonstrated lateral seeding in two types of structures. For the first structure, continuous Si02 sheets with narrow "via" holes filled with crystalline Si were fabricated by local oxidation of (100) Si wafers. Polycrystalline Si films were then deposited and patterned into rectangular pads by a second local oxidation process. The two patterns were

304

K. F. LEE ETAL.

matched s o t h a t t h e v i a s were w i t h i n t h e r e c t a n g u l a r Si i s l a n d s and near one of t h e i r edges. A schematic c r o s s - s e c t i o n of a t y p i c a l i s l a n d i s shown i n F i g . 5.54. The f i r s t l o c a l o x i d a t i o n process ensured t h a t t h e r e was no t o p o l o g i c a l o r thermal disconThe second t i n u i t y of t h e d e p o s i t e d Si f i l m i n t h e v i a r e g i o n . l o c a l o x i d a t i o n sequence e f f e c t i v e l y provided Si02 c r u c i b l e s o r tubs surrounding S i i s l a n d s from a l l but one ( t o p ) s i d e . The second s t r u c t u r e was s i m i l a r t o t h a t i n v e s t i g a t e d by Lam e t a l . i5.581 and c o n s i s t e d of r e c t a n g u l a r SiO2 pads r e c e s s e d i n t o c r y s t a l l i n e (100) S i wafers and then covered w i t h a continuous s h e e t of d e p o s i t e d p o l y c r y s t a l l i n e S i ( F i g . 5.53).

I n both s t r u c t u r e s 1 pm of thermal, steam grown Si02 insuS i f i l m s of l a t e d the d e p o s i t e d S i from t h e bulk S i s u b s t r a t e . ~ 0 . 6 urn t h i c k n e s s were prepared by low p r e s s u r e chemical vapor d e p o s i t i o n (LF'CVD) a t ~650°C. T h e i r m i c r o s t r u c t u r e w a s t h a t of f i n e g r a i n e d S i w i t h a l a r g e amorphous component.

A.

Lsio, (100)Si SUBSTRATE

/-CVD

B.

POCYSILICON

(100) Si SUBSTRATE

FIGURE 5.54. Schematic c r o s s - s e c t i o n s of t h e two s t r u c t u r e s s t u d i e d . ( a ) Si i s l a n d s recessed i n t o SiOz " t u b s " , w i t h v i a l i n k s t o t h e s u b s t r a t e ; ( b ) continuous S i d e p o s i t e d over r e c e s s e d Si02 pads. 01982 E l s e v i e r Science P u b l i s h i n g Co.

5.


305

The c r y s t a l l i z a t i o n and e p i t a x i a l overgrowth of t h e d e p o s i t e d S i f i l m s were performed w i t h a m u l t i l i n e cw A r + l a s e r . An i n t e n s e , 10-20W, laser beam was u s u a l l y focused t o a50 pm s p o t on t h e S i samples. The samples were vacuum clamped t o a r e s i s t i v e l y

h e a t e d s t a g e a t 4OO0C, which was moved i n a d e s i r e d f a s h i o n by a computer c o n t r o l l e d x-y t a b l e . I n some c a s e s , t h e beam was shaped w i t h 2 c y l i n d r i c a l l e n s e s i n t o a n e l o n g a t e d e l l i p s e , and scanning m i r r o r s could be used t o superimpose motion of t h e beam o v e r t h e t a b l e movement. Before d e s c r i b i n g t h e r e s u l t s of r e c r y s t a l l i z a t i o n e x p e r i ments, t h e temperature p r o f i l e s expected f o r uniform i r r a d i a t i o n of d i f f e r e n t semiconductor s t r u c t u r e s w i l l be b r i e f l y d i s c u s s e d . Fastow e t a l . 15.671 pointed o u t t h a t S i pads on o x i d i z e d S i are h o t t e r n e a r t h e edges, c o n t r a r y t o t h e conventional c r y s t a l growth experience and t o t h e case of S i i s l a n d s on t r a n s p a r e n t s u b s t r a t e s . The two c h a r a c t e r i s t i c s i t u a t i o n s are presented i n F i g . 5.55(a) and 5.55(b) w i t h t h e corresponding temperature p r o f i l e s . The second case, w i t h t h e lowest temperature a t t h e c e n t e r , i s h i g h l y advantageous f o r transforming t h e whole i s l a n d i n t o a s i n g l e c r y s t a l l i t e , as t h e n u c l e a t i o n would s t a r t a t t h e single point i n the center. HEAT FLOW

- 50pm

FIGURE 5.55. Temperature p r o f i l e s of Si-on-insulator i s l a n d s t r u c t u r e s d u r i n g uniform, argon laser i r r a d i a t i o n . The d i r e c t i o n of h e a t flow i s shown w i t h arrows, and t h e l i g h t l y d o t t e d r e g i o n s are d i r e c t l y h e a t e d by i r r a d i a t i o n . 01982 E l s e v i e r Science Publishing Co.

306


The structure of Fig. 5.54(a) is similar to the lower part of Fig. 5.55(b). Here, a via hole link between the deposited and bulk Si acts always as a heat sink since the thermal conductivity of Si is -100 times that of SiOp. Most early attempts to induce epitaxial overgrowth with laser beams failed, due to large thermal mismatch between the seeding and the overgrowth regions. The problem was usually compounded by the presence of a -1 pm step where the Si film crossed the edge of the Si02 layer. The local oxidation procedure, utilized in both structures shown in Fig. 5.54, planarized the surface before Si deposition, thus eliminating the topographic discontinuity. Vias were made intentionally very narrow, 1.5 um in width, to increase the thermal impedance. Their placement at the edges of Si islands, which would otherwise be the hottest areas, reduced the ratio of temperatures between the seeding and the overgrowth regions to acceptable levels. In Fig. 5.56, we show optical micrographs of two 20 pm x 100 pm Si islands after a single traverse of a laser beam from left to right. The surface of the islands was etched to delineate grain boundaries. It is well known that unless the Si film is melted through its whole thickness, the regrowth is seeded from the remaining crystallites and submicron grains result. This allows us to estimate the temperature profile in the upper island in Fig. 5.56 which was swept with a 11W beam. It is clear that incomplete melting occurred at the vias on both ends of the rectangular pad, and also in the center, in good agreement with the analysis of Fig. 5.54(b). The lower island, which was swept with a 13W laser

FIGURE 5.56. Optical micrographs of laser recrystallized via islands, after chemical delineation of grain boundaries. (a) 11W, and (b) 13W laser beam was swept from left to right. s1982 Elsevier Science Publishing C o .

5.


307

l a s e r beam, was completely melted and r e c r y s t a l l i z e d from t h e S e l e c t e d area e l e c t r o n channeling conv i a a t t h e l e f t edge. firmed t h a t t h e c r y s t a l l o g r a p h i c o r i e n t a t i o n was t h e same a s t h e s u b s t r a t e ; i.e., (100) p l a n e w i t h axes a l i g n e d w i t h t h e edges of t h e i s l a n d . A c r o s s s e c t i o n of a t y p i c a l r e c r y s t a l l i z e d i s l a n d i s shown i n Fig.5.57. The oxide has been etched away t o enhance c o n t r a s t , followed by l i g h t e t c h i n g of t h e S i t o show t h e g r a i n boundaries. The r e c r y s t a l l i z e d S i is smooth, but some t h i c k n e s s v a r i a t i o n i s n o t i c e a b l e , which i s l i k e l y caused by mass r e d i s t r i b u t i o n i n t h e wake of t h e laser beam. A " b i r d ' s beak" d i s t o r t i o n from p l a n a r i t y of t h e f i l m n e a r t h e v i a i s c h a r a c t e r i s t i c of t h e l o c a l o x i d a t i o n p r o c e s s . The i r r e g u l a r i t y of t h e s u r f a c e d i d n o t , however, a f f e c t t h e c r y s t a l l i z a t i o n process. A c l o s e r i n s p e c t i o n of F i g . 5.56 r e v e a l s t h a t t h e s i n g l e c r y s t a l l i n e r e g i o n breaks up i n t o smaller g r a i n s a t a d i s t a n c e of * 30 pm from t h e seeding p o i n t . Others have r e p o r t e d s i m i l a r The problems i n d i f f e r e n t l a t e r a l overgrowth s t r u c t u r e s [5.58] n a t u r e of t h i s l i m i t a t i o n of t h e seeded overgrowth is n o t f u l l y understood. It w a s suggested by Lam e t a1 [5.58] t h a t t h e stress buildup i n t h e r e c r y s t a l l i z i n g f i l m , caused by t h e d i f f e r i n g t h e r mal expansion c o e f f i c i e n t s of S i and Si02, produces a progress i v e l y i n c r e a s i n g d e f e c t d e n s i t y , w i t h t h e e v e n t u a l breakup of t h e c r y s t a l l i n e l a t t i c e by f e a t h e r shaped r e g i o n s f i l l e d w i t h d i s l o c a t i o n s , and then g r a i n boundaries are formed. Lam e t a 1 have a l s o n o t i c e d t h a t t h e e x t e n t of a s u c c e s s f u l overgrowth i n c r e a s e s w i t h t h e i n c r e a s i n g s u b s t r a t e temperature, which is compatible w i t h t h e e x c e s s i v e stress h y p o t h e s i s . A different e x p l a n a t i o n was r e c e n t l y proposed by Leamy [5.591. He argues t h a t

.

.

.

SEM micrograph of an i s l a n d FIGURE 5.57. Cross-sectional w i t h a v i a . SiOq was etched away t o improve c o n t r a s t ( a f t e r Ref. 5.57). @1982 E l s e v i e r Science P u b l i s h i n g Co.

K . F. LEE ETAL..

zone r e f i n i n g of i m p u r i t i e s contaminating t h e d e p o s i t e d f i l m would l e a d t o a c e l l u l a r s t r u c t u r e of c r y s t a l l i n e S i , surrounded The stress by a s u p e r s a t u r a t e d s o l u t i o n of t h e i m p u r i t i e s . e f f e c t s a s s o c i a t e d w i t h t h e c e l l s t r u c t u r e would then cause formation of low a n g l e g r a i n boundaries. By modifying t h e laser i r r a d i a t i o n procedure, t h e problem of These r e s u l t s do n o t p e r m i t , l i m i t e d overgrowth w a s e l i m i n a t e d . however, d e t e r m i n a t i o n of t h e predominant f a c t o r r e s p o n s i b l e f o r t h e l i m i t e d e x t e n t of s i n g l e c r y s t a l l i n e growth i n t h e convent i o n a l configuration. The f i r s t s u c c e s s f u l m o d i f i c a t i o n of t h e c r y s t a l l i z a t i o n p r o c e s s involved r a p i d l y s c a n n i n g t h e laser beam i n 2 o r t h o g o n a l d i r e c t i o n s such t h a t a c r e s c e n t shaped L i s s a j o u s f i g u r e r e s u l t e d Sweeping t h e i n s t e a d of a c o n v e n t i o n a l c i r c u l a r molten zone. a l t e r e d molten zone a l o n g t h e l o n g i t u d i n a l a x i s of t h e r e c t a n g u l a r i s l a n d s produced seeded s i n g l e c r y s t a l l i n e r e g i o n s as l a r g e as 50 x 100 urn, shown i n Fig. 5.58. In Fig. 5.59(a), an a l t e r n a t i v e r e c r y s t a l l i z a t i o n procedure i s shown s c h e m a t i c a l l y . Here t h e laser beam w a s scanned a c r o s s t h e i s l a n d s w i t h an i n c r e m e n t a l motion i n t h e overgrowth d i r e c t i o n of only 1 0 vm p e r scan. Each t i m e t h e beam c r o s s e s t h e i s l a n d , i t remelts most of t h e S i r e c r y s t a l l i z e d i n t h e p r e c e d i n g scan. S i n c e t h e c e n t e r of t h e i s l a n d i s c o o l e r t h a n t h e edges, new There i s of course c r y s t a l l i n e growth always s t a r t s t h e r e . competing c r y s t a l growth from t h e s i d e of t h e molten zone remote from t h e v i a s . This random growth i s e l i m i n a t e d , however, by

FIGURE 5.58. O p t i c a l micrograph of s i n g l e - c r y s t a l l i n e v i a i s l a n d produced by crescent-shaped beam ( a f t e r Ref. 5.57). el982 E l s e v i e r Science P u b l i s h i n g Co.

5.


309

FIGURE 5.59. Transverse scanning procedure. ( a ) Schematic; ( b ) o p t i c a l micrograph of t r a n s v e r s l y scanned, s i n g l e c r y s t a l l i n e O1982 Elsevier Science Fublishing CO. V i a i s l a n d s ( a f t e r Ref 5 .57)

.

.

t h e subsequent s c a n s . In F i g . 5.59(b), r e s u l t s of t r a n s v e r s e laser scans are p r e s e n t e d . S i n g l e c r y s t a l l i n e areas as l a r g e as 50 x 500pm have been formed i n t h i s manner. 5.5.5.4

Seeded L a t e r a l Epitaxy of Thick Films

The work reviewed in t h e previous s e c t i o n of t h i s c h a p t e r h a s been concerned e n t i r e l y w i t h t h e r e c r y s t a l l i z a t i o n of thin f i l m s on i n s u l a t i n g s u b s t r a t e s (with o r without s e e d i n g ) , i n which a scanning laser ( o r e l e c t r o n ) beam i s used as a h e a t s o u r c e . With a p p r o p r i a t e a t t e n t i o n t o h e a t flow c o n s i d e r a t i o n s , t h e m a t e r i a l produced i n t h i s way i s s i n g l e c r y s t a l (100) s i l i c o n w i t h e l e c t r i c a l p r o p e r t i e s t h a t approach those of bulk s i l i c o n . However, t h e material c o n t a i n s low a n g l e g r a i n boundaries (subboundaries) t h a t could be important i n l i m i t i n g t h e a p p l i c a b i l i t y of t h e f i l m s f o r b i p o l a r t r a n s i s t o r f a b r i c a t i o n .

.

Celler e t a1 t5.681 have r e c e n t l y d e s c r i b e d a r e c r y s t a l l i z a 10-15 pm) t i o n process based on uniform m e l t i n g of t h i c k (e.g., s i l i c o n l a y e r s d e p o s i t e d over amorphous i n s u l a t o r s , such as Si02, w i t h a n extended r a d i a t i v e h e a t s o u r c e . P e r i o d i c openings i n t h e i n s u l a t o r serve both as s e e d i n g r e g i o n s f o r e p i t a x i a l r e c r y s t a l l i z a t i o n and t o impose t h e l a t e r a l temperature g r a d i e n t s necessary f o r propagation of t h e s o l i d - l i q u i d i n t e r f a c e . Typically S i over a n e n t i r e 3-inch wafer i s melted simultaneously i n a few seconds, a f t e r which r e c r y s t a l l i z a t i o n proceeds over a 10-100 second p e r i o d . The r e s u l t a n t f i l m s are s i n g l e c r y s t a l l i n e , f r e e of any g r a i n boundaries and c o n t a i n only moderate d e n s i t i e s of d i s l o c a t i o n s and s t a c k i n g f a u l t s .

310

K. F. LEE ET AL.

FIGURE 5.60. O p t i c a l micrographs of two wafer c r o s s - s e c t i o n s a f t e r 10 sec Schimmel e t c h i n g ; ( a ) as d e p o s i t e d sample, ( b ) a f t e r lamp m e l t i n g and r e c r y s t a l l i z a t i o n . 01983 Elsevier Science Publishing CO. The sample geometry employed by Celler e t a l . i s shown i n Fig. 5.60. Sample p r e p a r a t i o n i n v o l v e s o x i d a t i o n , p h o t o l i t h o graphy and wet e t c h i n g , S i d e p o s i t i o n (APCVD) and e n c a p s u l a t i o n w i t h a second oxide f i l m ( 2 vm of LPCVD oxide). The system used f o r r a p i d h e a t i n g is shown i n Fig. 5.61. It c o n s i s t s of two r e c t a n g u l a r chambers w i t h l a t e r a l dimensions 10 x 12.5 i n c h e s , p o s i t i o n e d one above t h e o t h e r . The upper chamber c o n t a i n s a bank of air-cooled t u n g s t e n halogen lamps suspended Two q u a r t z windows s e p a r a t e t h e below a gold p l a t e d r e f l e c t o r . lamps from t h e wafer chamber. The wafers a r e placed on q u a r t z The lamps p i n s -0.5 i n c h above a water cooled aluminum base. provide uniform h e a t i n g of t h e t o p s u r f a c e of t h e wafers. Equally uniform r a d i a t i v e c o o l i n g of t h e back s u r f a c e causes a which allows conv e r t i c a l temperature g r a d i e n t of -lOO°C/cm, t r o l l e d m e l t i n g of t h e f i l m s d e p o s i t e d on t h e S i wafers. A t -78 W/cm2 i n c i d e n t power l e v e l , t h e e n t i r e d e p o s i t e d Crystalf i l m and some of t h e u n d e r l y i n g s u b s t r a t e is melted. l i z a t i o n of t h e d e p o s i t e d s i l i c o n begins a t t h e openings i n t h e o x i d e and proceeds l a t e r a l l y over t h e oxide. Films w i t h a t h i c k n e s s of 10-15 urn and s u r f a c e dimensions of -1 mm on a s i d e

5.


311

can be recrystallized in this manner, making the process competitive with standard dielectric isolation techniques for fabrication of bipolar transistors with high collector breakdown voltages.

REFLECTOR TUNGSTEN HALOGEN LAMPS

~oooooooooooooo*

r

I

OUARTZ PLATE

I

QUARTZ PLATE

1

WAFER

FIGURE 5.61. 5.5.6

Schematic cross-section of the RTA apparatus.

Large Area Recrystallization

One of the principal limitations evident in the early work on laser recrystallization is the small processing area which results from the use of the highly focused beam. This small area limits the area and quality of the recrystallized films, as discussed earlier. As a result, large area recrystallization techniques have been developed whereby a scanning narrow molten zone spanning the entire width of the sample, in some cases an entire wafer, has been used to process the material. This process is known as zone recrystallization and was first successfully demonstrated in silicon by Fan et al. [5.69] using a graphite strip heater. Geis et al. [5.70] have published a review of zone recrystallization efforts as they have developed since the early work of Fan et al. I n the early work of Fan, the sample was placed on a flat graphite heater which was used to raise the temperature of the entire sample to near the melting point. A narrow graphite strip located above the sample was then resistively heated and used to create a narrow molten zone in the thin polycrystalline film. The molten zone was then scanned across the film by translating the narrow strip across the sample as shown schematically

K. F. LEE E T A L .

312

LARGE G R A I N RECRY STALL1

FINE GRAIN 1 pm T H E R M A L SiOz

0.5 vim LPCVD Si

MOVABLE UPPER STRIP-HEATER

---*

I I

'ER

SAMPLE

FIGURE 5.62. (a) Schematic drawing of sample used for zone recrystallization experiments and (b) schematic diagram showing zone recrystallization using a grapite strip heater. in Fig. 5.62(a). This technique gave large grain films (single crystal films when seeding was used) with excellent electronic properties. Subsequent to this work, other radiant sources such as focused high pressure mercury arc lamps [5.71] have been used in the zone recrystallization process. In the following sections the structure of zone recrystallized films and the origin of the characteristic low angle grain boundaries observed in these films will be discussed. 5.5.6.1

Sample Preparation and Experimental Conditions

The samples used for the zone recrystallization experiments are silicon thin films which have been deposited on thermally oxidized single crystal silicon wafers. The silicon films are deposited by low pressure chemical vapor deposition ( L P C V D ) . The thermal oxide and LPCVD silicon are nominally 1 micron and 0.5

5.


313

microns thick, respectively. Before recrystallization, the silicon layer is encapsulated with 2 microns of CVD silicon dioxide. This encapsulating layer is found to be necessary to prevent agglomeration of the molten silicon during the recrystallization process. Further, although encapsulating layers thinner than 2 microns were found to prevent agglomeration, the thicker encapsulant provided smoother recrystallized surfaces and induced a predominantly texture in the recrystallized film. A schematic illustration of the sample structure used in this work is shown in Fig. 5.62(a), in which a focussed heat source is suggested as an alternative to the graphite strip geometry. The deposited surface layer can be visualized as performing two functions; (1) reducing the total surface free energy of the molten layer by providing a second surface for the silicon to wet and ( 2 ) mechanically restraining the film from balling up or agglomerating. However, it is also found that the encapsulating layer improves the surface smoothness and induces a texture in the recrystallized layer. The efficacy of the encapsulant on these two additional features was also found to improve with increasing thickness. These two effects will be discussed in more detail in later sections. The process of zone recrystallization as it is performed with a scanning arc lamp system is as follows. The sample is placed on the substrate heater located in the reaction chamber. The chamber is evacuated and backfilled with argon gas. The substrate heater is then used to raise the sample temperature to near its melting point, between 1000°C and 135OOC. The sample temperature is monitored using an optical pyrometer. After the sample reaches the desired temperature, the arc lamp or strip heater is turned on and scanned across the film. This creates a narrow molten zone, 0.2 cm - 0.3 cm in width, which is translated through the film at rates typically less than 0.5 cm/s. Although a variety of initial sample temperatures, scan rates and power combinations have been used to create a stable molten zone, in general the best results are obtained using a sample temperature of approximately 125OOC and a scan rate of 0.2 cm/s. Unless otherwise stated, the results presented in the following sections were obtained using these conditions. 5.5.6.2

Structure of Zone Recrystallized Films

Although the zone recrystallized silicon films are in general very smooth and virtually featureless when viewed under conventional optical microscopy, some linear features or striations are faintly visible using Nomarski differential interference microscopy. To delineate the linear features, samples are etched using the Secco defect etchant 15.551. After etching, the samples

314

K. F. LEE ETAL.

FIGURE 5 . 6 3 . ( a ) SEM photomicrograph of Secco e t c h e d zone r e c r y s t a l l i z e d f i l m showing " h a i r p i n " shaped low a n g l e boundar( b ) "EM b r i g h t f i e l d micrograph of h a i r p i n f e a t u r e . ies. e x h i b i t e d a r e g u l a r a r r a y of h a i r p i n - l i k e f e a t u r e s which tended t o be a l i g n e d i n t h e zone s c a n d i r e c t i o n , as shown i n Fig. 5.63. The s p a c i n g of t h e s e f e a t u r e s is e s s e n t i a l l y uniform over a g i v e n sample. However, from sample t o sample, t h e s p a c i n g r a n g e s from 25 microns t o over 100 microns and i s found t o be dependent

5.


315

on the recrystallization conditions, film thickness and impurity concentrations of the film. The dependence of the spacing on these various parameters will be discussed below. Transmission electron microscopy (TEM) was used to characterize the nature of the linear defects. Figure 5.63(b) is a TEM brightfield photomicrograph of one of the hairpin features. As shown, the linear features run nearly parallel to the direction and the material between them has an in-plane orientation of of (100). Further microscopic analysis revealed that the linear features were in fact low angle grain boundaries and the material between them subgrains, all in general having a azimuthal orientation near and a texture of in the zone scan direction. The subgrains were further analyzed using a scanning electron microscope (SEM) in the channeling contrast mode. In this mode, contrast is obtained due to variations in the crystallographic alignment of the material being irradiated [5.72]. Thus variations in the subgrain azimuthal orientation will lead to contrast under these conditions. Figure 5.64 is a SEM photomicrograph showing such contrast and thus indicating that there is indeed such a variation between subgrains in the film. To quantify the subgrain orientation variation, TEM Kikuchi pattern analyses were performed. By measuring the shift of the Kikuchi patterns obtained on either side of the low angle boundary, relative to the diffraction pattern, a very accurate determination

FIGURE 5.64. SEM photomicrograph of zone recrystallized film taken under channeling contrast conditions showing variation in azimuthal subgrain orientation.

316

K. F. LEE E T A L .

of t h e degree of m i s o r i e n t a t i o n can be made [5.72]. A TEM photomicrograph of a t y p i c a l low a n g l e boundary and t h e Kikuchi patt e r n s o b t a i n e d from e i t h e r s i d e i s shown i n F i g . 5.65. Analyses of t h e s e type f i g u r e s i n d i c a t e d t h a t t h e m i s o r i e n t a t i o n s ranged from a few t e n t h s of a degree t o a couple of degrees throughout the film. Low a n g l e boundaries are g e n e r a l l y j u s t a r r a y s of polygonized To f u r t h e r c h a r a c t e r i z e t h e low a n g l e bound i s l o c a t i o n s [5.731. d a r i e s , TEM a n a l y s e s of t h e d i s l o c a t i o n networks forming t h e s e boundaries were made. It w a s determined t h a t t h e d i s l o c a t i o n s r a n i n t h e d i r e c t i o n a t 60 degrees t o t h e i r Burger's v e c t o r , 112 (110). T h i s type of d i s l o c a t i o n i s t y p i c a l i n c o v a l e n t c r y s t a l s w i t h t h e diamond s t r u c t u r e such as germanium and s i l i c o n [ 5 . 7 4 1 . A TEM photomicrograph of a low a n g l e boundary d i s l o c a t i o n a r r a y and a s t e r e o g r a p h i c p r o j e c t i o n showing t h e r e l a t i v e o r i e n t a t i o n s of t h e boundary, t h e d i s l o c a t i o n s and t h e Burger's v e c t o r a r e shown i n F i g . 5.66. Although microscopic a n a l y s e s of t h e r e c r y s t a l l i z e d f i l m provide some very p r e c i s e i n f o r m a t i o n concerning t h e n a t u r e of t h e small d e f e c t s i n t h e m a t e r i a l , i t is very d i f f i c u l t t o o b t a i n

FIGURE 5.65. TEM photomicrograph of low a n g l e boundary w i t h i n s e t Kikuchi p a t t e r n s from e i t h e r s i d e . Subgrains have t i l t miso r i e n t a t i o n s ranging from a few t e n t h s of a degree t o a couple of degrees.

5.


317

DISLOCATION LINES

L O W ANGLE BOUNDARY WITH 60' DISLOCATIONS

LAB LAB

LOW ANGLE BOUNDARY

BV D

BURGER'S VECTOR DISLOCATION

STEREOGRAPHIC

PROJECTION

FIGURE 5.66. TEM photomicrograph of dislocation array forming low angle boundary. Alsb shown are a model of dislocation array and a stereographic projection of low angle boundary, dislocation and Burger's vector. information about the macroscopic features of the film using these techniques. This is especially true when the features of interest are much larger than the probe or beam used to evaluate the material. There is however, an ingenious technique proposed by Smith for evaluating the crystallographic orientations in large grain polycrystalline films using an array of etch pits [5.75]. This technique provides a means of determining texture, azimuthal orientation and the location of large qngle grain boundaries in large grain polycrystalline silicon. With this technique, the silicon film to be analyzed is coated with a Si02 film of 200 nm or more. (In these experiments, the 2 micron thick Si02 encapsulant was used.) The Si02 is then coated with photoresist and a grid array of circular holes is patterned into

318

K . F. LEE ET AL

the resist using s t a n d a r d p h o t o l i t h o g r a p h i c techniques. The h o l e s a r e nominally 25 microns i n diameter w i t h a spacing of 50 microns between openings. Following t h e development of t h e p h o t o r e s i s t , the p a t t e r n of h o l e s is t r a n s f e r r e d t o t h e Si02 u s i n g a b u f f e r e d HF oxide e t c h a n t . Finally, the s i l i c o n film which has been exposed by t h e oxide e t c h i n g is e t c h e d i n a 40% by weight s o l u t i o n of KOH i n D I H20 a t 80°C f o r 6-8 minutes. KOH i s a very a n i s o t r o p i c e t c h a n t i n s i l i c o n , e t c h i n g much more Conr a p i d l y i n t h e d i r e c t i o n t h a n i n t h e i5.761. s e q u e n t l y t h e symmetry and alignment of t h e e t c h p i t s can be r e l a t e d t o t h e c r y s t a l l o g r a p h i c n a t u r e and o r i e n t a t i o n of t h e s i l i c o n being etched. Figure 5.67 s c h e m a t i c a l l y i l l u s t r a t e s t h e f e a t u r e s of an e t c h p i t which would be formed i n a t h i n f i l m of s i l i c o n having (100) o r i e n t a t i o n p a r a l l e l t o t h e p l a n e of the film. A s shown, t h e p i t s are t r u n c a t e d pyramids w i t h (111) f a c e t s o r s i d e s and & f o l d r o t a t i o n a l symmetry. The d i a g o n a l s Similarly, i f the p i t s of t h e p i t s a r e i n t h e (100) d i r e c t i o n s . were formed i n a f i l m w i t h a (111) o r i e n t a t i o n t h e y would exh i b i t 3-fold or 6-fold r o t a t i o n a l symmetry. C l e a r l y t h e symmetry and d i a g o n a l alignment of t h e p i t s can be used t o determine t h e in-plane o r i e n t a t i o n and t e x t u r e of t h e f i l m being e t c h e d . Figure 5.68 i s an o p t i c a l photomicrograph of a zone r e c r y s t a l l i z e d f i l m which has been e t c h e d a s d e s c r i b e d above. The &-fold r o t a t i o n a l symmetry and d i a g o n a l alignment of t h e e t c h p i t s indicate a {loo] in-plane o r i e n t a t i o n and t e x t u r e i n t h e zone s c a n d i r e c t i o n , r e s p e c t i v e l y . Although t h e s e d a t a c o r r o b o r a t e t h e TEM a n a l y s e s d e s c r i b e d above, t h e r e a l u t i l i t y of t h i s technique i s t h a t t h i s i n f o r m a t i o n h a s been o b t a i n e d over a very l a r g e f i l m a r e a , o r d e r s of magnitude l a r g e r than t h e a r e a sampled u s i n g t h e TEM. Thus u s i n g a s t a n d a r d o p t i c a l microscope, r e l a t i v e l y expans i v e a r e a s of t h e f i l m can be q u i c k l y analysed f o r t h i s type of c r y s t a l l o g r a p h i c information.

A second important f e a t u r e of t h e e t c h p i t t e c h n i q u e i s t h a t t h e l o c a t i o n of l a r g e a n g l e g r a i n boundaries i s very s t r a i g h t forward. As shown i n t h e photomicrograph i n Fig. 5.69 a g r a i n boundary w i l l r e s u l t i n a misalignment of t h e d i a g o n a l s of t h e e t c h p i t s i n t h e g r a i n s on e i t h e r s i d e of t h e boundary, c o r r e s ponding t o t h e c r y s t a l l o g r a p h i c m i s o r i e n t a t i o n of t h e g r a i n s themselves. F u r t h e r , a s a l s o shown, e t c h p i t s which happen t o f a l l on t h e g r a i n boundary i t s e l f w i l l have d i s t o r t e d f e a t u r e s . Using t h e e t c h p i t technique d e s c r i b e d above, t h e zone r e c r y s t a l l i z e d f i l m s were found t o be composed of l a r g e g r a i n s , t y p i c a l l y 0.05 c m t o 0 . 2 c m i n w i d t h and up t o s e v e r a l c e n t i m e t e r s long. For t h e most p a r t , t h e g r a i n s e x h i b i t e d (100) t e x t u r e and alignment as d i s c u s s e d e a r l i e r ; however , i n c e r t a i n cases ( 1 1 11 o r i e n t a t i o n s were observed. The f i l m s which had t h e {111) g r a i n s were f i l m s which had been r e c r y s t a l l i z e d under very r a p i d s c a n Under t h e s e c o n d i t i o n s , t h e f r e e z i n g c o n d i t i o n s , > 0.4 cm/s.

5.


319

TOP VIEW

“1001

FIGURE 5.67. Drawing showing features of anisotropic etch pit formed in a thin silicon film with (100) orientation.

320

K. F. LEE ET AL

F I G U R E 5.68. Photomicrograph of zone recrystallized film which has been etched using etch pit technique. Sample contains no grain boundaries.

F I G U R E 5.69. Photomicrograph of zone recrystallized sample which has been etched using etch pit technique. Sample has a grain boundary running down center of figure.

5.


321

rate was slower than the scan rate. Consequently the solidification front would lag far behind the actual lamp position, creating a large extended molten zone. In time, the molten zone would become supercooled and unconstrained dendritic growth would occur. This type of growth leads to a very faceted film with random crystallite orientations. An example of a film grown under these conditions is shown in Fig. 5.70. A s stated layers, e.g., surfaces, even such a faceted

5.5.6.3

earlier, films grown with thin encapsulating less than 0.5 microns, tended to have faceted in the absence of dendritic growth. A film with surface is shown in Fig. 5.71.

Origin of Structure in Zone Recrystallized Films

Crystallization is simply the long range ordering of atoms in a periodic solid phase structure or lattice. The final configuration of the crystalline (or polycrystalline) state is determined by a compromise between the thermodynamic driving

FIGURE 5.70. Photomicrograph of sample which was zone recrystallized under very rapid scan conditions leading to dendrltic growth.

322

K. E LEE ET AL.

forces and the kinetics of the reaction, and is the lowest free energy state which can be achieved under the constraints of the kinetics involved in the process. In the zone recrystallization of thin films, minimization of the free energy, associated with the crystalline bonding configuration, the grain boundary density and the interfacial area is the primary thermodynamic driving force which influences this final state. The kinetic limitations are principally atomic mobility, nucleation, and adatom attachment rates. In zone recrystallization, linear growth rates of several millimeters per second are achieved. This is extremely rapid when compared to other crystal growing techniques such as Czochralski or Bridgman, where linear rates of tenths of millimeter per minutes are common [ 5 . 7 7 ] . Consequently it is likely that the nucleation rate is the dominant growth rate limiting mechanism in the zone recrystallization process. Although there are several comprehensive theoretical treatments on the nucleation and growth of thin films [5.78-5.801, quantitative evaluation of thin film zone recrystallization process is impeded by the lack of consistent and experimentally verified materials data, even in the well studied silicon-silicon dioxide system. For example, the possible role of oxygen as an impurity which induces constitutional supercooling during the recrystallization process is of considerable interest, as will

FIGURE 5.71. Photomicrograph of sample with a faceted surface resulting from zone recrystallizing with a thin encapsulating layer.

5.


323

be discussed below. However, the necessary parameters for making a quantitative assessment of this role are either not available or the published values vary sufficiently that virtually any argument can be reasonably supported. As a consequence only qualitative and/or semi-quantitative arguments can be forwarded to explain the various observed phenomena. The following discussions on the observed features are presented with this in mind. There are three striking features of the zone recrystallized film which are of interest: (1) the prevalence of the azimuthal orientation, (2) the growth parallel to the zone scan direction and ( 3 ) the origin and spacing of the low angle boundaries. The predominance of the (100) in-plane orientation is a feature which is extremely fortuitous but not well understood. It is fortunate in that fabrication of MOS integrated circuits is best performed in {loo) silicon material. Unfortunately there is no well-founded explanation for this growth behavior. It has been argued that the interfacial free energy between silicon and silicon dioxide is minimum for the (100) planes I5.811. However, there exists to date no data on the interfacial free energy versus orientation for this system. Another possible explanation takes into account the anisotropic stress-strain characteristics of silicon. Since the silicon film is sandwiched between two layers of silicon dioxide, it will experience stress from differential thermal expansion as the sample is heated up. In addition there will be a contribution to the stress field resulting from volumetric contraction of the silicon upon melting. Consequently grains of different orientation will experience different densities of elastic strain energy. If these grains are in fact the seed grains for the film growth, then a textured growth from the grain with a preferred orientation might occur r5.82-5.841. Another mechanism which is known to affect grain growth tendencies is the formation and mobility of dislocations i5.851. Since this is indeed a polycrystalline system composed of many low angle boundaries as described above, growth accommodation by dislocation motion may play an important role in the final crystal orientation.

A final explanation for the {loo} orientation has to do with the match between the planar atomic densities of (100) silicon and In the amorphous (or microcrystalline) silicon dioxide 15.861. cited reference, an analysis of the orientation dependent oxidation rate of silicon based on planar atomic densities is given. It has been suggested that an extension of this analysis might account for the observed (100) texture [5.87]. Specifically, if the {loo) planes in silicon offer the best atomic match-up with the Si02 surface, then growth with this orientation would result

K. F. LEE ET AL.

in the least number of dangling and/or strained bonds and thus would be energetically favorable. Unfortunately some experimentally unverified assumptions on the nature and composition of the Si02 surface must be made to support this argument. These assumptions have to do with the possible existence of microcrystalline phases of 8-crystoballite and the nature of the bonding configuration in the transition region between the Si02 and silicon regions. One of the experimentally observed features in the zone recrystallized films which is not necessarily in accord with the above explanations is the fact that films which have been recrystallized on Si3N4 also have the {loo) orientation. Thus for example, unless the bonding configuration and surface free energy relations between the silicon crystal faces and both silicon nitride and silicon dioxide are very similar, this added observation is in conflict with the interface controlled models. It is clear from these remarks that more work must be done in this area to obtain a satisfactory explanation for the in-plane {loo) film orientation. A s described earlier, the zone recrystallized films tend to grow with the direction parallel to the scan direction. It has also been observed that the solidification front during the zone processing is faceted rather than planar f5.841. Lemons [5.88] has argued that this growth is a result of constitutional supercooling with oxygen as the most likely impurity leading to this effect. Although constitutional supercooling can lead to cellular and dendritic growth, the basic experimentally observed trends in the zone recrystallized films are not entirely in accord with this mechanism. Further, the materials parameters used to support this argument vary sufficiently in the literature that an equally strong argument against constitutional supercooling can be made.

Constitutional supercooling is a situation where interface instability can result from the existence of a zone of "constitutionally supercooled" material ahead of the liquid solid interface. This zone is termed supercooled because its equilibrium melting temperature is depressed below the nominal melting temperature of the material due to an increase in the concentration of an impurity in that region. This increase in concentration is a result of the segregation of the impurity during the solidification process. (An excellent review of this phenomena is given in Ref. 5.85, Chapter 9.) It has been shown that the existence of a zone of constitutional supercooling ahead of a smooth planar interface is given by the following condition

5.


325

where G is the temperature gradient in the liquid at the interface, V is the freezing velocity, mL is the liquidus slope of the phase diagram at the impurity concentration of interest, Cs(0) is impurity concentration in the solid at the interface, DL is liquid diffusion coefficient and ko is equilibrium segregation coefficient. To determine under what growth conditions constitutional supercooling will exist, it is necessary to obtain values for ko, mL, C,(O) and k . The published values of the segregation coefficient for oxygen in silicon range from 0.5 [5.89] to 1.25 [5.90]. Thus it is not even clear whether oxygen is rejected from or incorporated into the liquid zone. Published values for the solid solubility limit of oxygen in silicon range from to atomic fraction and the liquid diffusion coefficient ranges from cm2/s to cm2/s. Finally, the liquidus slope can be calculated from thermodynamics if the heat of fusion and segregation coefficient are known. If Eq. (5.7) is evaluated using k, = 0.5, mL = 225, DL = Cs(0) = and V = 0.2 cm/s, it is found that a thermal gradient greater than 4500°C/cm is required t o prevent constitutional supercooling, which is a very unlikely situation under these experimental conditions. Thus one could conclude with Lemons that the film was indeed constitutionally supercooled. On the other hang, if Eq. (5.7f is evaluated using ko = 1.25, mL = 100, CL = 10- , C (0) = 10- and V = 0.2 cm/s, then a thermal gradient of only 0.!'C/cm would prevent constitutional supercooling. Clearly the issue of whether or not the film is constitutionally supercooled during recrystallization cannot be made on the basis of available data in the literature. In the following, an alternative model based on absolute supercooling and nucleation limited kinetics is presented. This model adequately explains the faceted growth behavior and is consistent with the observed trends in low angle boundary spacing. It is well known that "pure" materials with moderately high entropies of fusion and anisotropic growth rates tend to grow with a faceted interface. Silicon falls into this category, and thus even in the absence of impurities, might be expected to grow in this manner. However, another more intuitive way of viewing this is that if the nucleation and growth rates on different crystallographic planes vary significantly, then a growing crystal would be expected to be bounded by the slowest growing planes. In silicon, the (111) planes are the slowest, as evidenced by its habit form as well as by experimental observations during the growth of crystals [5.92]. Indeed, consistent with this description, the interface facets observed during the zone recrystallization of silicon thin films have been shown to be closely aligned with the (111) direction [5.841.

326

K. F. LEE ET AL.

Now, f o r n u c l e a t i o n of new p l a n e t o o c c u r , a c e r t a i n amount of thermodynamic d r i v i n g f o r c e i n t h e form of u n d e r c o o l i n g from t h e e q u i l i b r i u m melt t e m p e r a t u r e , Tm, must be p r e s e n t . I f we c a l l t h e n u c l e a t i o n t e m p e r a t u r e Tn, t h e n t h e r e w i l l e x i s t a r e g i o n i n t h e molten zone bounded by Tm and Tn where l i q u i d and s o l i d can c o e x i s t . The w i d t h of t h i s r e g i o n i s set by t h e thermal g r a d i e n t Under s t e a d y s t a t e c o n d i t i o n s , t h e shape of t h e i n the m e l t . s o l i d i f i c a t i o n f r o n t w i l l be one which p e r m i t s t h e f a s t e s t growth r a t e w h i l e a t t h e same t i m e being bounded by t h e (111) planes. A p l a n a r i n t e r f a c e a l i g n e d w i t h t h e Tn i s o t h e r m o f f e r s t h e minimum s u r f a c e area f o r n u c l e a t i o n and growth w h i l e a t t h e same time l e a v i n g t h e zone of undercooled m e l t i n f r o n t of i t unoccupied by s o l i d . Thus t h i s i s an u n l i k e l y c o n f i g u r a t i o n f o r i t t o assume. On t h e o t h e r hand, a f a c e t e d i n t e r f a c e o f f e r s t h e maxiumum a r e a f o r n u c l e a t i o n w h i l e a t t h e same time having l i q u i d and s o l i d coe x i s t i n g i n t h e undercooled r e g i o n . The a n g l e s between t h e f a c e t s are s e t by t h e c r y s t a l l o g r a p h i c r e l a t i o n between t h e p l a n e s forming the facets. The l e n g t h of t h e f a c e t s i s determined by t h e thermal g r a d i e n t i n t h e m e l t ; a s t e e p e r g r a d i e n t reducing t h e d i s t a n c e between Tm and Tn and t h u s reducing t h e p e n e t r a t i o n d e p t h and l e n g t h of t h e f a c e t . The growth of a f i l m w i t h a f a c e t e d i n t e r f a c e a c t u a l l y has two growth d i r e c t i o n s ; one microscopic and one macroscopic. Assuming t h a t t h e n u c l e a t i o n and growth rates on each of t h e f a c e t p l a n e s are approximately e q u a l , and imposing t h e r e s t r i c t i o n t h a t t h e in-plane o r i e n t a t i o n must be (100) , as has been observed, then by symmetry t h e macroscopic growth d i r e c t i o n must be in t h e (100) d i r e c t i o n . The microscopic growth d i r e c t i o n would be t h e p r o j e c t i o n of t h e (111) d i r e c t i o n s o n t o t h e (100) f i l m p l a n e , which are v e c t o r s p o i n t i n g a t 45 d e g r e e s t o t h e macroscopic growth d i r e c t i o n . A schematic of t h i s growth s t r u c t u r e is shown in Fig. 5.72. Assuming a growth behavior as d e s c r i b e d above, t h e o r i g i n A s t h e (111) of t h e low a n g l e boundaries i s e a s i l y understood. f a c e t s grow and converge upon one a n o t h e r , any microscopic m i s o r i e n t a t i o n i n any c r y s t a l l o g r a p h i c d i r e c t i o n w i l l n e c e s s a r i l y T h i s mismatch l e a d t o a mismatch a t t h e p o i n t of convergence. w i l l be accommodated by t h e g e n e r a t i o n of a d i s l o c a t i o n . As growth c o n t i n u e s , rows of d i s l o c a t i o n s w i l l be formed a l o n g a l i n e running p a r a l l e l w i t h t h e macroscopic growth d i r e c t i o n , . These rows of d i s l o c a t i o n s are t h e observed low a n g l e g r a i n boundaries. Thus f a r a model has been d e s c r i b e d which e x p l a i n s t h e f a c e t e d growth and t h e o r i g i n of t h e low a n g l e boundaries based on t h e a n i s o t r o p i c n u c l e a t i o n and growth rates i n s i l i c o n . This model s u g g e s t s t h a t l a y e r can occur. The f a c e t s i z e and t h u s low a n g l e boundary s p a c i n g i s dependent on t h e thermal g r a d i e n t

5.


r l

AT FOR NUCLEATION

{ 1 1 1 } FACET PLANES

MICROSCOPIC GROWTH DIRECTION

I

327

SCAN AND MACROSCOPIC GROWTH DIRECTION

T; TO,

LOW ANGLE BOUNDARIES

FIGURE 5.72. Schematic drawing of faceted liquid-solid interface showing origin of low angle boundaries. Tmo is the melting temperature of pure silicon and Tno is the temperature at which nucleation occurs.

in the molten zone and not on the impurity concentration. Experiments conducted to investigate this conclusion are described below. 5.5.6.4

Effect of Scan Rate

In the first series of experiments, the sub-boundary spacing was determined as a function of the zone scan rate. With all other parameters held constant, an increase in scan rate will produce a decrease in the thermal gradient. Thus an increase in scan rate should result in an increase in low angle boundary spacing. Over the limited range of scan rates under which the zone recrystallization process could be conducted without introducing other growth mechanisms; e.g. dendritic, this indeed was found to be the case. Scan rates from 0.03 cm/s to 0.35 cm/s produced boundary spacings from 25 microns to over 100 microns. Although variations of up to a factor of 2 in boundary spacing were found on a given sample, the overall trend clearly indicated an increase in spacing with increasing speed. However, these results do not conclusively rule out constitutional supercooling. In the classical case, the cell spacing in constitutionally supercooled material is inversely proportional to the product of the growth rate and the thermal gradient [5.93]. Since an independent control of the growth rate and temperature gradient is not available in the recrystallization system, it cannot be determined whether the increase in the scan rate was larger or smaller than the decrease in the temperature gradient.

328 5.5.6.5

K.F. LEE ETAL. E f f e c t of Oxygen

A second experiment was performed i n l i g h t of t h e s u g g e s t i o n of Lemons 15.881 t h a t oxygen would be t h e most l i k e l y impurity c r e a t i n g t h e c o n s t i t u t i o n a l l y supercooled c o n d i t i o n . A s e r i e s of 2 . 5 ~ 1 0 ~ ~ s i l i c o n f i l m s were doped w i t h oxygen t o 2 . 5 ~ 1 0 ~ ’ cm-3 and 2 . 5 ~ 1 0 ~ ’ p r i o r t o zone r e c r y s t a l l i z a t i o n . These i m p u r i t y c o n c e n t r a t i o n s correspond t o oxygen atomic c o n c e n t r a t i o n s After of approximately O.OOl%, 0.01%, and 0.1%, r e s p e c t i v e l y . r e c r y s t a l l i z i n g t h e f i l m s , they were Secco etched t o d e l i n e a t e Two i n t e r e s t i n g f e a t u r e s w e r e t h e g r a i n boundaries 15.551. observed i n t h e r e c r y s t a l l i z e d and e t c h e d f i l m s . F i r s t , t h e subboundary s p a c i n g was found t o i n c r e a s e w i t h i n c r e a s i n g oxygen c o n c e n t r a t i o n . Second, whereas the f i l m w i t h 0.001% oxygen appeared s i m i l a r t o undoped f i l m s which had been p r e v i o u s l y r e c r y s t a l l i z e d and e t c h e d , t h e o t h e r series of samples were cove r e d w i t h small e t c h p i t s . F u r t h e r , t h e sample w i t h 0.1% oxygen was uniformly covered w i t h t h e s e p i t s whereas t h e p i t s on t h e sample w i t h 0.01% were l o c a t e d i n bands running p a r a l l e l t o t h e zone scan d i r e c t i o n . The bands of e t c h p i t s on t h a t sample were uniformly spaced w i t h a c h a r a c t e r i s t i c s p a c i n g approximately 3 times l a r g e r than t h e sub-boundary spacing. Photomicrographs of t h e s e samples a r e shown i n Fig. 5.73. The e t c h p i t s are i n t e r p r e t e d a s r e s u l t i n g from oxygen induced d e f e c t s and/or oxygen p r e c i p i t a t e s which e t c h more r a p i d l y than c r y s t a l l i n e s i l i c o n .

5.


329

Photomicrographs of zone recrystallized silicon FIGURE 5.73. containing (a) .001% oxygen, (b) .01% oxygen and (c) .l% oxygen. Samples have been Secco etched.

K.E LEE E T A L . The banding is interpreted as an indication of the onset of true constitutional supercooling. In this case the interface has broken down into a cellular structure and the oxygen has segregated to the cell boundaries. However, superimposed on this structure is the faceted interface which still generates the low angle boundaries. This structure and interface is schematically shown in Fig. 5.74. The absence of banding in the 0.1% sample can be interpreted as a case where the oxygen concentration might be so high that the background concentration in the film after segregation is still very large and consequently pitting will occur in all areas of the film. Although these experiments and observations still do not conclusively demonstrate whether the faceted growth is a result of constitutional supercooling or simply from anisotropic growth and nucleation limited kinetics, they do indicate the direction that future experiments must take to help unravel this question. 5.5.6.6

Other Defects in Zone Recrystallized Films

In Sections 5.5.6.2 and 5.5.6.3, the nature and origin of crystallographic defects which are present in zone recrystallized films were discussed. These defects are observed in otherwise high quality films. However, one also observes gross imperfections which are related to either the zone recrystallization conditions or the properties of the deposited film and/or encapsulant.

Tm (EQUILIBRIUM)

LOW ANGLE BOUNDARIES

T" (EQUI LI B R l UM

FIGURE 5.74. Schematic drawing of model for constitutionally supercooled interface with superimposed faceted growth.

5.


331

The principal defects observed in the zone recrystallized films other than the ones discussed previously are: (1) substrate slip and warpage , (2) substrate melt-through, ( 3 ) protrusions along sub-boundaries, ( 4 ) thinned out regions or film voids, and (5) film agglomeration or balling up. Substrate slip and warp damage may or may not affect the electronic quality of the recrystallized film; however the nonplanarity of the resulting sample precludes its use in modern integrated circuit processing equipment such as contact mask aligners. This defect is a result of the large thermal stresses the sample can experience during the recrystallization process. It can be minimized by insuring that there are no lateral thermal gradients introduced by the substrate heater and by using relatively slow thermal cycle rates during the preheat and cool down steps. Substrate-melt through is a situation when the molten zone (or an area of it) becomes too hot and melting of the underlying substrate occurs. Under this condition the thermal oxide between the two molten areas can rupture and allow the film and substrate to fuse together. The resulting melt-through pit has a symmetry characteristic of the underlying substrate. Sub-boundary protrusions are usually a result of film impurities. During the solidification process these impurities will segregate to the boundaries and locally depress the equilibrium melting temperature of the film. Consequently that region of the film will solidify after the material surrounding it. Since silicon undergoes a volumetric expansion during solidification, these small molten regions which are surrounded by solid silicon must expand upward upon freezing. Clean materials, steep thermal gradients and reduced scan velocities can effectively prevent this defect. Regions where the recrystallized film has thinned out even to the point of exposing the underlying thermal oxide have also been observed in the recrystallized films. These nonuniformities in the film thickness can obviously pose a serious problem in device processing. Although it is believed that these "voids" may be a result of localized impurities or defects at the lower interface, no experimental evidence of this fact has been obtained. It has, however, been observed that the density of these defects decreases with increasing substrate bias temperature. This is reasonable since the surface free energy of a molten layer generally decreases with increasing temperature and thus the overall wetting properties of the film will be improved. Film agglomeration or balling up is perhaps the most serious problem encountered during the zone recrystallization process.

332

K . F. LEE ETM.

It results in large areas of agglomerated silicon surrounded by the exposed thermal oxide. Although the mechanism involved in this defect is easily understood, the reliable control of it is not, at least as of this writing. The agglomeration is simply a result of trying to minimize the total free energy of the film by reducing the surface to volume ratio. The obvious solution to the problem is to reduce the interface free energy between silicon and silicon dioxide, i.e. improve the wetting characteristics of the film. The conventional approach to improving wetting characteristics is to introduce an impurity at the interface or change substrates. However, impurities are in general undesirable in semiconductor device grade silicon and the silicon-silicon dioxide system has electronic and processing properties which are very attractive. Consequently, an alternative approach is needed. As stated, encapsulating layers have been shown to be effective in preventing film agglomeration. It is generally believed that the role of the encapsulant i s structural in that it mechanically suppresses film balling up. It also plays a role in reducing the surface free energy of the film by providing a second surface for the film to wet. However, since the silicon-silicon dioxide wetting characteristics are rather poor, this is only a small contribution. It has been observed that silicon wets silicon nitride better than silicon dioxide; however, the electronic properties of the silicon-silicon nitride interface are inferior to the silicon-silicon dioxide interface. Many investigators have found that a dual layer encapsulant is more effective than the single layer of Si02 f5.13, 5.941. However, it has also been observed by the same investigators that the efficacy of the second layer is highly dependent on the deposition technique and parameters. Alternatively, it has been observed that the tendency for agglomeration is affected by the silicon film deposition technique and conditions. In fact it has been found that films which are not prone to agglomeration are relatively insensitive to the encapsulant. With these films the encapsulant type, deposition technique and thickness serve only to modify surface features of the recrystallized film. For example, silicon dioxide layers less than 0.5 microns have been successfully used in the recrystallization process. However, as shown earlier, the resulting films had very faceted surfaces and in some cases had a large number of grains which were not (100) oriented. Attempts to recrystallize silicon films which have been deposited by sputtering and atmospheric pressure chemical vapor deposition (APCVD) have also been made. However the only films which were successfully zone recrystallized without agglomeration were deposited using LF'CVD. Unfortunately LPCVD is not the simple answer, since it has been found that whereas films deposited in one LPCVD reactor gave good results, films deposited in another reactor may not. Chemical analyses of the films using

5.


333

Auger and Secondary Ion Mass Spectrometry did not reveal any systematic differences between the deposited films. A variety of encapsulating layers was used on the films which severely agglomerated. However for those films, none of the encapsulants were successful in preventing agglomeration, including the dual layer cap described above. Clearly this problem is the most perplexing and demanding, and its understanding and solution must be obtained before the thin film zone recrystallization process can become a universally useful technique. 5.5.6.7

Geometrical Techniques for Improving the Crystallinity of Zone Recrystallized Films

Thus far our development of the zone recrystallization process has concentrated entirely on attempts to crystallize thin, large area silicon films deposited directly on SiO2/Si substrates. In the earlier work on laser recrystallization, both the seeding process and the introduction of diamond-shaped islands were found to improve the crystallinity and lead in some cases to material that was free of both grain and subgrain boundaries. These same techniques have been employed in the hopes of improving the quality of zone recrystallized films. In particular, Lam et al. [5.95] have used a seeding technique in which the seed was an annular ring of thickness -1 mm on the outer edge of a 3 inch silicon wafer, the center of which was covered with a thermal oxide. High quality regrowth has been obtained with this geometry using a graphite strip heater, though surface flatness is still somewhat too large for the most demanding (VLSI) applications. The seeded films also display the hairpin structures that are characteristic of the unseeded films. An interesting structure has also been developed by Atwater et al. r5.961 to limit and perhaps ultimately eliminate the formation of subgrain boundaries. The structure is similar to the use of the pointed island structures of Biegelsen et al. in that the film is patterned to provide a series of periodic, planar constrictions placed in such a relation to each other that only one grain can propagate through the pattern. A schematic illustration of the concept is shown in Figure 5.75. Atwater et al. have succeeded in recrystallizing films without sub-boundaries using this technique, but to date the area of the films is limited to a few square centimeters.

334

K. F. LEE ETAL.

FIGURE 5.75. (a) Hourglass shaped pattern for elimination of sub-boundaries. (b) Selection of a single orientation from t w o initial grains by recrystallization through the neck of an hourglass structure.

5.


335

REFERENCES 5.1 5.2 5.3 5.4 5.5 5.6 5.7

5.8 5.9 5.10 5.11 5.12

5.13

5.14 5.15 5.16

5.17 5.18 5.19 5.20 5.21 5.22 5.23

L e e , K. F., Gibbons, J. F., Saraswat, K. C., and Kamins, T. I., Appl. Phys. L e t t . 35, 173 (1979). Tasch, A. F., Jr., Holloway, T. C., Lee, K. F., and Gibbons, J. F., Elec. Letts. 15, 435 (1979). Kamins, T. I., Lee, K. F., Gibbons, J. F., and Saraswat, K. C., IEEE Trans. Elec. Devices ED-27, 290 (1980). Lam, H. W., Tasch, A. F., Jr., Holloway, Lee, K. F., and Gibbons, J. F., IEEE Elec. Dev. L e t t . EDL-1, 99 (1980). Gerzberg, L., G a t , A., L e e , K. F., Gibbons, J. F., Peng, J., Magee, T. J., D e l i n e , V. R., and Evans, C. A., Jr., J . Electrochem. SOC., Extended A b s t r a c t s , 80-2, 1053 (1980). Gat, A., Gerzberg, L., Gibbons, J . F., Magee, T. J., Peng, J., and Hong, J. D., Appl. Phys. L e t t . 33, 775 (1978). Gibbons, J. F., Johnson, W. S., and Mylroie, S. W., "Proj e c t e d Range S t a t i s t i c s i n Semiconductors". Dowden, Hutchi n s o n , and Ross (1975). Kamins, T. I., Lee, K. F., and Gibbons, J. F., Appl. Phys. L e t t . 36, 7 (1980). Kamins, T. I., Mandurah, M. M., and Saraswat, K. C., J. Electrochem. SOC. 125, 927 (1978). Kodera, H i r o s h i , Jap. J. Appl. Phys. 2, 212 (1963). "Physics of Semiconductor Devices", p. 68. Sze, S. M., Wiley (1981). Kamins, T. I., Monoliu, J., and Tucker, R. N., J. Appl. Phys. 43, 83 (1972); a l s o Tsukamoto, K. e t a l . , J. Appl. Phys. 48, 1815 (1977). Ohkura, M., Kusukawa, K., Yoshida, I., Miyao, M. and Tokuyama, T., Extended A b s t r a c t s of 1 5 t h Conf. on S o l i d S t a t e Devices and Materials, Japan S o c i e t y of Applied P h y s i c s (August 1983), pp. 43-46. Kamins, T. I., J. Electrochem. SOC., 128, 1824 (1981). Lasky, J. B., J . App. Phys. 53, 9038 (1982). Drowley, C. and Kamins, T. I. i n "Laser-Solid I n t e r a c t i o n s and T r a n s i e n t Thermal P r o c e s s i n g of Materials" (J. Narayan, W. L. Brown and R. A. Lemons, eds.), 13, 511-516, North Holland (1983). S h i b a t a , T., Izuka, H., Kohyama, S., and Gibbons, J. F., Appl. Phys. L e t t . 35, 2 1 (1979). S h i b a t a , T., Lee, K. F., Gibbons, J. F., Magee, T. J., Peng, J., and Hong, J. D., J. Appl. Phys. 52, 3625 (1981). Bell Syst. Tech. J. 39, 205 (1960). Trumbore, F. A., Cook, R. K., Frey, J., Lee, K. F., and Gibbons, J . F., unpublished N o r r i s , C. B., Jr., and Gibbons, J.F., IEEE Trans. on Elec. Dev. ED-14, 1, 38 (1967). Kamins, T. I., J. Electrochem. SOC. 126, 838 (1979). Kamins, T. I., Lee, K. F., and Gibbons, J. F., IEEE Elec. Dev. L e t t . EDL-1, 5 (1980).

.

336 5.24 5.25 5.26 5.27 5.28 5.29 5.30 5.31 5.32 5.33 5.34 5.35 5.36 5.37 5.38

5.39 5.40 5.41 5.42 5.43

5.44 5.45 5.46

K. F. LEE E T A L . Ley H. P. and Lam, H. W., IEEE Elec. Dev. Lett. EDL-3, 6 (1982). Kamins, T. I. and Drowley, C. I., IEEE Elec. Dev. Lett. EDL-3, 12 (1982). Gibbons, J. F., and Lee, K. F., IEEE Elec. Dev. Lett. EDL-1, 117 (1980). Kamins, T. I., IEEE Elec. Dev. Lett. EDL-3, 11 (1982). Kamins, T. I. and Pianetta, P. A., IEEE Elec. Dev. Lett. EDL-1, 10 (1980). Tsaur, B. Y., Fan, J. C. C. and Geis, M. W., Appl. Phys. Lett. 40, 322 (1982). Johnson, N. M., Biegelsen, D. K., Tuan, H. C., Moyer, M. D. and Fennell, L. E., IEEE Elec. Dev. Lett. EDL-3, 12 (1982). Kamins, T. I., Solid St. Electr. 15, 789 (1972). Muller, R. S., and Kamins, T. I., "Device Electronics for Integrated Circuits". Wiley (1977). Leistiko, O., Grove, A. S., and Sah, C. T., IEEE Trans. Elec. Dev. ED-12, 248 (1965). Grove, A. S., "Physics and Technology of Semiconductor Devices", Chap. 11. Wiley (1967). Antoniadis, D. A,, Hansen, S. E., and Dutton, R. W. , "SUPREM I1 - A Program for IC Process Modelling and Simulation", Technical Report 5019-2, Stanford Electronics Labs (1978). Tsaur, B. Y., Fan, J. C. C., Chapman, R. L., Geis, M. W., Silversmith, D. J. and Mountain, R. W., IEEE Elec. Dev. Lett. EDL-3, 12, 398 (1982). Jolly, R.D., Kamins, T. I. and McCharles, R. H., IEEE Elec. Dev. Lett, EDL-4, 1, 8 (1983). Chiang, A., Zarzycki, M. H., Meuli, W. P. and Johnson, N . M. in "Laser-Solid Interactions and Transient Thermal Processing of Materials" (J. C. C. Fan and N . M. Johnson, eds.), North Holland (1984) in press. Ng, K. K., Celler, G. K., Povilonis, E. I., Frye, R. C., Leamy, H. J. and Sze, S. Z., IEEE Elec. Dev. Lett. EDL-2, 316 (1981). Lepselter, M. P., IEDM Tech. Digest, 42 (1980). Tihanyi, J. and Schlotterer, H., IEEE Trans. Elec. Dev. ED-22, 11, 1017 (1975). Sze, S. M., "Physics of Semiconductor Devices", 2nd ed. Wiley (1981). Ng, K. K. , Aller, G. K. , Povilonis, E. I., Trimble, L. E. and Sze, S. M., in "Laser-Solid Interactions and Transient Thermal Processing of Materials" (J. C. C. Fan and N . M. Johnson, eds.) North Holland (1984) in press. Kleitman, D., private communication. Colinge, J. P. and Demoulin, E., IEDM Technical Digest, 557 (1981). Goeloe, G. T., Maby, E. W., Silversmith, D. J., Mountain, R. W. and Antoniadis, D. A . , IEDM Technical Digest, 554, (1981).

5. 5.47 5.48 5.49 5.50 5.51 5.52 5.53 5.54 5.55 5.56 5.57 5.58 5.59

5.60 5.61 5.62

5.63 5.64 5.65

5.66

5.67


337

Kamins, T. I., IEEE Elec. Dev. L e t t . EDL-3, 11, 341 (1982). Kawamura, S., S a s a k i , N. I w a i , T., Nakano, M. and Takagi, M., IEEE Elec. Dev. L e t t . EDL-4, 10, 366 (1983). Gibbons, J. F., Giles, M. D., Lee, K. F. and Walker, J. F., SPIE Conference, Los Angeles (1983). Lam, H. W., Malhi, S. D. S. and P i n i z z o t t o , Chen, C.-E., R. F., IEEE Elec. Dev. L e t t . EDL-4, 8, 272 (1983). Sturm, J., G i l e s , M. D. and Gibbons, J. F., IEEE E l e c . Dev. L e t t . EDL-5, 5, (1984). Gibbons, J. F., Lee, K. F., Wu, F. C., and Eggermont, G. E. J., IEEE Elec. Dev. L e t t . 3, 8 (1982). Elmasry, M. I., and El-Mansy, Y. A., I E E E Hamdy, E. Z., Trans. Elec. Dev. ED-28, 322 (1981). S t u l t z , T. J. and Gibbons, J. F., Appl. Phys. Lett. 39, 499 (1981). d'Argona, F. Secco, J., J. Electrochem. SOC. 119, 948 (1972). S c h o t t , J. T., i n "Laser and Electron-Beam I n t e r a c t i o n s w i t h S o l i d s " , p. 517. E l s e v i e r (1982). G a t , A. and Moore, J., p r i v a t e communication. Lam, H. W., P i n i z z o t t o , R. F. and Tasch Jr., A. F., J. Electrochem. SOC. 128, 1981 (1981). Auvert, G., Bensahel, A. Georges, Nguyen, V. T., Henoc, P. Morin, F., and C o i s s a r d , P., Appl. Phys. L e t t . 38, 613 (1981 1. Gibbons, J. F. Lee, K. F., Magee, T. J., Peng, J. and Ormond, R., Appl. Phys. L e t t . 34, 1 2 , 831 (1979). Smith, H. I. and F l a n d e r s , D. C., Appl. Phys. Lett. 32, 349 (1978). Johnson, N. M., B a r t e l i n k , D. J. and B i e g e l s e n , D. K., i n "Laser and Electron-Beam S o l i d I n t e r Moyer, M. D., a c t i o n s and Materials Processing" (J. F. Gibbons, L. D. Hess and T. W. Sigmon, eds.), pg. 487. North Holland (1981). Thomas, G., "Transmission E l e c t r o n Microscopy of Metals". Wiley (1966). Colinge, J. P., Demoulin, E., Bensahel, D., and Auvert, G. Appl. Phys. Lett. 41, 346 (1982). Drowley, C. I., Zorabedian, P. and Kamins, T. I., "LaserS o l i d I n t e r a c t i o n s and T r a n s i e n t Thermal Processing" (J. C. C. Fan and N. M. Johnson, e d s . ) , North Holland (1984) i n press. Trimble, L. E., Celler, G. K., Ng, K. K., Baumgart, H. and Leamy, H. J., i n "Laser and Electron-Beam I n t e r a c t i o n s w i t h S o l i d s , (B. R. Appleton and G. K. Celler, eds.), pp. 505510. North Holland (1982). Fastow, R., Leamy, H. J., C e l l e r , G. K., Wong, Y. H. and Doherty, C. J., i n "Laser and E l e c t r o n Beam S o l i d I n t e r a c t i o n s and Laser Processing", (J. F. Gibbons, L. D. Hess and T. W. Sigmon, eds.), p. 495. North Holland (1981).


5.68

5.69 5.70

5.71 5.72 5.73 5.74 5.75 5.76 5.77 5.78 5.79 5.80 5.81

5.82 5.83 5.84

5.85 5.86 5.87 5.88 5.89 5.90 5.91 5.92

Celler, G. K. Robinson, McD., L i s c h n e r , D. J. and Sheng, T. T., i n "Laser-Solid I n t e r a c t i o n s and T r a n s i e n t Thermal P r o c e s s i n g of Materials", (J. Narayan, W. L. Brown and R. A. Lemons, eds.), pp. 575-580. North Holland (1983). Appl. Phys. Fan, J. C. C., Geis, M. W. and Tsaur, B-Y., Lett. 38, 365 (1981). Geis, M. W., Smith, H. I., Tsaur, B-Y., Fan, J. C. C., S i l v e r s m t t h , D. J., Mountain, R. W. and Chapman, R. L., i n " Laser-Solid I n t e r a c t i o n s and T r a n s i e n t Thermal Proc e s s i n g of Materials, (J. Narayan, w. L. Brown and R. A. Lemons, e d s . ) , pp. 477-489. North Holland (1983). S t u l t z , T. J . and Gibbons, J. F., Appl. Phys. Lett., 4 1 , 824 (1982). Bowen, D. K. and Hall, C. R., "Microscopy of Materials." Cambridge P r e s s (1973). K e l l y , A. and Groves, G. W., " C r y s t a l l o g r a p h y and C r y s t a l D e f e c t , " , pp. 345-356. Addison-Wesley (1970). I b i d , pp. 254-255. B e z j i a n , K. A , , Smith, H. I., Carter, J. M. and G e i s , M. W., J . Electrochem. S O C . 129, 1848 (1982). Kendall, D. L., Annual Review of Materials S c i e n c e , 9 , 373 (Huggins, Bube, Vermilyea, e d s . ) , (1979). Brandle, C. C., " C r y s t a l Growth," (Pamplin, B. R., ed.), pp. 275-301. Pergamon P r e s s (1980). S a t o , H., Annual Review of Materials Science 2 , 217 (1972). J o y c e , B. A., Rep. Prog. Phys. 37, 363 (1974). Chapera, K. L., "Thin Film Phenomena," Chap. 4. McGrawH i l l (1969). Leamy, H. J., "Laser and E l e c t r o n Beam I n t e r a c t i o n s w i t h North S o l i d s " , (B. R. Appleton and G. K. Celler e d s . ) . Holland (1982). Vook, R. W. and W i t t , F., J. Appl. Phys. 36, 2169 (1965). Vook, R. W. and W i t t , F., J. Vac S c i . Tech. 2, 243 (1965). Geis, M. W., Smith, H. I., Tsaur, B-Y, Fan, J. C. C., S i l v e r s m i t h , D. J. and Mountain, R. W., J. Electrochem. SOC. 129, 2812 (1982). " P h y s i c a l Metallurgy" (Cahn, ed.), Chap. 19. Cahn, R. W., North-Holland (1970). T i l l e r , W. A., J. Electrochem. Soc. 128, 689 (1981). T i l l e r , W. A., p r i v a t e communication. Lemons, R., "Laser-Solid I n t e r a c t i o n s and T r a n s i e n t Thermal P r o c e s s i n g of Materials", ( J . Narayan, W. L. Brown and R. A. Lemons, eds.). Norch-Holland (1983). Trumbore, F. A., BSTI 39, 205 (1960). Y a t s u r u g i , Y., Akiyama, N. and Endo, Y., J. Electrochem. S O C . 120, 975 (1973). " C r y s t a l Growth", (Bardsley, Hurle and Jackson, K. A., Mullin, e d s . ) , Chap. 5. North-Holland (1979). Abe, T., Kikuchi, K. S h i r a i , S. and Muraoka, S., "Semicond u c t o r S i l i c o n " , Electrochem. Soc. (1981).

5. 5.93 5.94 5.95 5.96


339

Flemings, M. C., "Solidification Processing", Chap. 3. McGraw-Hill (1974). Maby, E., Geis, M., LeCoz, Y., Silversmith, D., Mountain, R. and Antoniadis, D., IEEE Elect. Dev. Lett. EDL-2, 214 (1981 ) Lam, H. W., personal communication. Atwater, H., Smith, H. I. and Geis, M. W., Appl. Phys. Lett. 41, 747 (1982).

.


CHAPTER 6

Metal-Silicon Reactions and Silicide Formation T. ShibataY: A . Wakita, T. W. Sigmon, and James F. Gibbons STANFORD ELECXRONICS LABORATORIES STANFORD UNIVERSITY STANFORD. CALIFORNIA

6.1 6.2 6.3 6.4

INTRODUCTION

.....................................

EXPERIMENTAL TECHNIQUES ......................... EXPERIMENTAL RESULTS: LASERPROCESSING. . . . . . . . . . . EXPERIMENTAL RESULTS:ELECTRONBEAM PROCESSING

.......................................

6.5 THERMAL STABILITY AND OXIDATION PROPERTIES OF cw BEAMREACTEDSILICIDES . . . . . . . . . . 6 . 6 LASERPROCESSING OF Nb3X SUPERCONDUCTORS ................................. REFERENCES

6.1

.......................................

341 344 357

371 375

385 393

IN'IRODUCTION

The reaction of thin metal films with both single crystal and amorphous silicon to form silicide compounds is a subject that is currently receiving considerable attention. 'his interest arises from the fact that future trends in silicon integrated circuit technology, particularly the push toward higher device packing density and decreased dimensions, will require low resistance, thin film materials with high temperature properties that are compatible with other silicon processing operations. Such films will find important applications as gates for MOS transistors, device interconnects and ohmic contacts [6.1-6.31. Presently there are a number of silicide compounds that look promising for fulfilling these needs. An excellent review of the thermal equilibrium interdiffusion and reaction of thin metal films on silicon, including silicide formation, is given in Ref. [ 6 . 4 ] . Here both the deposition and furnace annealing processes are discussed in detail, along with useful measurement techniques for determining reacted film composition and quality. Within the last two years, the use of laser, electron and ion beams to react metal/silicon films have also been actively investigated [6.5-6.121. New and unique features provided by these forms of beam processing include shorter reaction times, localized heat treatment and rapid heating and cooling rates. As in the *Present address: VLSI Research Center, Toshiba Research and Development Center, Kawasaki City, Kanagawa, 201 Japan. SEMICONDUCTORS AND SEMIMETALS, VOL. 17

341


342

T. SHIBATA ET A L

case of beam annealing of ion-implanted semiconductors, both the basic mechanisms and the experimental results depend on whether pulsed or scanned cw beams are used to promote the reactions. For both cases experimental results are found in many cases to differ sharply from those obtained with standard furnace processing. Since the use of pulsed laser and electron beams to form silicides is thoroughly reviewed elsewhere [6.131, only a brief description of the most important results will be given here for comparison with the cw case. Following this brief review, we will concentrate on the experimental results obtained when scanned cw laser and electron beams are employed to produce reactions in a metal/silicon system. 6.1.1

Pulsed Laser Silicide Formation

The most important results obtainedwhen a laser beam is used to react a thin metal filmwith a silicon substrate are summarized in Table 6.1. As of this writing there is no published data on the use of pulsed e-beams for silicide formation. Since the principal results obtained with the cw-scanned e-beam [6.8] are similar to those obtained with the cw laser, they are not included in Table 6.1; however, important differences are discussed in Sec. 6.4 of this chapter. It can be seen in Table 6.1 that the essential difference between the pulsed and cw scanned lasers is the length of time during which the laser energy is deposited in the film (20-200 ns as TABLE 6.1 Summary of Silicide Formation Results for Pulsed and Scanned CW-Laser Processes. Pulsed Laser

CV Laser

Laser Irradiation Time

20-200 nsec

msec

Hechanism of silicide formation

Melting. mixing in a molten phase and rapid solidification

Solid phase diffusion (basically) Nucleation controlled reaction and eutectic melting can also be involved.

Results : Silicide phase formed

Hixed phase with Si precipitation

Single phase (generally) with no Si precipitation.

Composition of the film

Continuously changes by increasing powers

Selected by power levels.

Structure of the film

Cellular structure

Homogeneous layer

Silicidefsilicon interface

Deep metal penetration into Si substrate

Uell-defined interface

Application to refractory metals on Si

Unable to react thicker films

Possible with multiple laser scans

Formation of bubble structure

6.

METAL-SILICON REACTIONS AND SILICIDEFORMATION

343

-

opposed to 1 ms). It will become clear that this exposure time and the associated heating and cooling rates play a decisive role in the physical processes that occur during silicide formation.

As is the case for the pulsed laser annealing of ion-implanted, elemental semiconductors, melting and rapid resolidification occurs in metal/silicon films that are reacted using pulsed lasers [6.5]. The rapid melting is typically followed by the nucleation and growth of a number of metalfsilicon species in the molten phase, after which a very rapid resolidification occurs due to the extremely high cooling rates involved I6.141. The principal result is that the reacted film has frozen in a mixed phase compound that consists of several silicide compounds and also silicon and metal precipitation [6.151. The composition of the reacted layer can be changed by adjustment of the laser power level. It is also possible to obtain compositions inaccessible by conventional furnace annealing. However, for the formation of uniform, large area, single phase compounds, pulsed laser processing is not presently useful. The structure of the films, as revealed by transmission electron microscopy, indicates that they consist of small cellular silicide clusters on the order of 0.1 urn in diameter, interspersed with silicon precipitates that are separated from each other by metal-rich walls [6.16]. The simultaneous formation of large cells, on the order of 1 micron in diameter is also observed; and deep penetration of the metal into the silicon substrate has been verified [6.16]. The formation of this mixed-phase cellular structure would be expected from the rapid melting-resolidification mechanism for the reaction. These results are in sharp contrast to those obtained with the scanned cw laser, where uniform, large area, single-phase silicide compounds are produced. 6.1.2

Scanned cw-Laser and E-Beam Silicide Formation

Recent work in the application of scanned cw-lasers and ebeams for the formation of silicide compounds has led to a process that appears to be superior in many respects to that obtained with pulsed lasers. As in the annealing of ion-implanted amorphous layers in single crystal, elemental semiconductors, a solid phase process is responsible for the metal-silicon reactions obtained when the scanning cw beam is used at sufficiently low power levels. In the remainder of this chapter we will discuss the mechanisms by which single phase silicide compounds are formed when a scanned cw laser or electron beam is used. A brief overview of the data to be presented is provided below for convenient comparison with the pulsed case.

344

T. SHIBATA ET AL.

B a s i c a l l y i t i s found t h a t s i n g l e phase s i l i c i d e compounds w i t h l i t t l e o r no Si p r e c i p i t a t i o n a r e o b t a i n e d when a scanned cw beam i s used t o promote t h e m e t a l - s i l i c o n r e a c t i o n . The phase of t h e f i l m (e.g., Pd2Si o r PdSi) can be s e l e c t e d by a p p r o p r i a t e choice of t h e power l e v e l i n t h e scanning beam. In those s p e c i a l c a s e s where more t h a n one phase a p p e a r s , t h e r a t i o of t h e phases can be determined i n some c a s e s from t h e power/(beam r a d i u s ) f a c t o r of t h e l a s e r . The f i l m s a r e g e n e r a l l y found t o be homogeneous w i t h a s h a r p l y d e f i n e d s i l i c o n / s i l i c i d e i n t e r f a c e 16.171. The formation of r e f r a c t o r y metal s i l i c i d e s h a s a l s o been accomp l i s h e d , w i t h t h e s u c c e s s f u l formation of MoSi2, NbSi2 and WSi2 f i l m s r e p o r t e d i n t h e l i t e r a t u r e i6.181. Finally, i t is possible t o e s t a b l i s h r e a c t i o n k i n e t i c s by m u l t i p l e scanning i n a manner t o be d e s c r i b e d below. Although t h e m a j o r i t y of m e t a l / s i l i c o n r e a c t i o n s s t u d i e d depend on a s o l i d phase i n t e r d i f f u s i o n mechanism, c e r t a i n anomalies a r e observed. For t h e s p e c i a l case of PdSi formation, a n u c l e a t i o n - c o n t r o l l e d r e a c t i o n i s b e l i e v e d t o dominate i6.71, w h i l e f o r P t S i evidence of e u t e c t i c m e l t i n g and r a p i d r e s o l i d i f i c a t i o n i s o b t a i n e d [6.191.

6.2

EXPERIMENTAL TECHNIQUES

I n t h i s s e c t i o n w e f i r s t d i s c u s s t h e sample s t r u c t u r e s t h a t are normally used t o s t u d y t h e formation of s i l i c i d e s by beam processing. W e t h e n d e s c r i b e the p r i n c i p a l a n a l y t i c a l methods t h a t a r e used t o s t u d y t h e p h y s i c s of t h e beam a n n e a l i n g process. These i n c l u d e Rutherford b a c k s c a t t e r i n g , t h e X-ray Read camera, and o p t i c a l microscopy. 6.2.1

Sample P r e p a r a t i o n

I n Fig. 6.1 w e show a schematic i l l u s t r a t i o n of t h e sample s t r u c t u r e s u t i l i z e d i n t h e s e experiments. To o b t a i n well-controll e d r e a c t i o n s , t h e metal f i l m s were d e p o s i t e d by e l e c t r o n beam e v a p o r a t i o n ( o n t o p-type s i l i c o n s u b s t r a t e s ) i n a h i g h vacuum, o i l - f r e e system. S p u t t e r d e p o s i t i o n of t h e metal f i l m s w a s avoided s i n c e i t normally l e a d s t o s u f f i c i e n t A r i n c l u s i o n t o i n h i b i t the the desired reaction. A s suggested i n Fig. 6 . l ( a ) , samples used f o r e-beam processi n g c o n s i s t e d o n l y of t h e s i l i c o n s i n g l e c r y s t a l s u b s t r a t e and t h e metal f i l m . When t h e cw l a s e r was employed t o promote s i l i c i d e formation, i t was n e c e s s a r y t o add a t h i n (200 A ) l a y e r of s i l i con on t h e metal s u r f a c e t o provide an a n t i r e f l e c t i o n c o a t i n g Fig. 6 . l ( b ) ] . Without t h i s s i l i c o n o v e r c o a t i n g , a high laser

6.

METAL-SILICON REACTIONS AND SILICIDE FORMATION

345

SCANNED C W e-BEAM

Ill

METAL F I L M S e- BEAM DEPOSITED Pd - - Nb

---

M o - - - \ 500

w

---)

i

Si SUBSTRATE

CW LASER BEAM

(100) Si SUBSTRATE

-1

FIGURE 6.1. Schematic of sample structure and film thicknesses utilized in the (a) electron-beam and (b) cw laser reacted silicide experiments. power is required to compensate for the substantial amount of laser power that is reflected from the metal surface. Then, once the silicide begins to form, the lower reflectance of the silicide compound results in an increased power absorption in the film during the reaction, leading to difficulty in controlling the reaction. This problem is easily avoided by use of the thin silicon overcoat. Since the reflectance does not change significantly for the electron beam, this overcoat is not required for the scanning electron beam experiments. In Table 6.2 we list the principal metal/silicon systems to be discussed in this chapter. Also listed are some of the basic properties that characterize each sample, such as the metal layer thickness and whether or not a silicon overcoat is used.

346

T. SHIBATA ET AL.

TABLE 6.2 Description of Samples used for Scanned cw-Laser and Electron-Beam Silicide Formation. Beam Process

Si-Overcoating

Metal Layer

Si(200 A )

pd(1300 Pt(1000 Mo( 530 W ( 440 Nb(1100

Scanned-Laser Beam

Scanned-Electron Beam 6.2.2

None

Substrate

a)

A) A) A) A)

P-type Si

Pd(1400 A ) Pt(1100 A ) Nb(1250 A )

Typical LaserlElectron Beam Annealing Conditions

The laser processing systemused for all experiments reported here is the cw argon system described in Chapter 1. The laser was operated in the multi-line mode. The beam was focussed by a 135 mm lens to obtain a spot on the sample surface approximately 50 pm in diameter. 'Jke spot was scanned across the sample surface at a rate of approximately 12 cm/sec. The substrate was held at a fixed temperature during the laser irradiation. Substrate temperatures used for the various reactions studied are listed in Table 6.3. The laser beam irradiation produces a temperature distribution that is Gaussian-like. The peak temperature in the center of the beam is given by the formula derived in Chapter 2 [Eq. 2.461:

and A are constants where To is the substrate temperature and in the empirical expression for the temperature-dependent thermal conductivity. w is the beam radius, P is the laser output power and R is the reflectivity of the sample. To determine the laser power required t o achieve a specific temperature, it is necessary to know R and w. R can be experimentally measured before each scan, and w is determined from Eq. (6.1) by measuring the power at which the surfaqe of a clean silicon wafer starts to melt (T = 1412OC). Since t k depth distribution of temperature is constant within a few micrometers from the surface, the entire thickness of the metal film (0.1-0.5 pm) can be considered to be heated to a uniform temperature during laser annealing. The error in calculating Tmax that arises by using the power required to melt crystalline silicon as a calibration is estimated to be f 10% for silicon [6.20]. Use of this calculated temperature to predict growth rates for silicides will be shown later in

6.


347

t h i s c h a p t e r t o provide very s a t i s f a c t o r y agreement w i t h t h e o r y . However, complications can occur when t h e sample r e f l e c t i v i t y changes d u r i n g a n n e a l i n g . The u n c e r t a i n t y of t h e a n n e a l i n g t e m p e r a t u r e can then be s u b s t a n t i a l l y l a r g e r than t h e e s t i m a t i o n based on t h e s i l i c o n m e l t i n g p o i n t . For t h e scanning e l e c t r o n beam experiments, a Hamilton Standard e l e c t r o n beam welder (Model EBW 7.5) was employed a t a worki n g vacuum of T o r r . The samples were mounted on a copper h e a t s i n k w i t h Dow Corning 340 h e a t s i n k compound. The h e a t s i n k was maintained n e a r room temperature d u r i n g t h e r e a c t i o n . Typical e l e c t r o n beam parameters used f o r t h e r e a c t i o n were: a s u b s t r a t e temperature of < 50°C; e l e c t r o n beam energy 31 keV; beam c u r r e n t , 0.1 - 0.3 mA; beam s p o t s i z e ( a s measured by t h e r e a c t e d p a t h ) 100 pm; s c a n r a t e 13 cm/sec; and a s t e p s i z e of less t h a n 100 pm. 6.2.3

A n a l y t i c a l Methods:

Rutherford B a c k s c a t t e r i n g

Analysis of both t h e r e a c t i o n k i n e t i c s and t h e average f i l m composition of t h e s i l i c i d e s produced by scanned l a s e r and e l e c t r o n beam p r o c e s s i n g can be conveniently c a r r i e d out w i t h Rutherford b a c k s c a t t e r i n g (RBS) t e c h n i q u e s . When combined w i t h a s t r u c t u r a l a n a l y s i s technique, such as g l a n c i n g a n g l e X-ray d i f f r a c t i o n i n a Read camera, Rutherford b a c k s c a t t e r i n g can prov i d e a r a p i d , h i g h l y q u a n t i t a t i v e a n a l y s i s of most t h i n f i l m RBS proves t o be p a r t i c u l a r l y u s e f u l metal-silicon r e a c t i o n s . f o r a n a l y s i s of beam-reacted f i l m s , as they tend t o be uniform and of reasonably l a r g e area s o t h a t a n a l y s i s by mm2 s i z e 4He+ probing beams i s e a s i l y accomplished. The a p p l i c a t i o n of Rutherford b a c k s c a t t e r i n g t o t h e measurement of t h i n f i l m p r o p e r t i e s has been d i s c u s s e d elsewhere [6.4] and w i l l n o t be d e a l t w i t h i n d e t a i l here. However, a b r i e f des c r i p t i o n w i l l be u s e f u l t o t h e r e a d e r who i s u n f a m i l i a r w i t h t h i s technique. The review given below w i l l be based on t h e a n a l y s i s of a t h i n metal f i l m on a s i l i c o n s u b s t r a t e as i t appears immedia t e l y a f t e r d e p o s i t i o n and a f t e r a s i l i c i d e - f o r m i n g r e a c t i o n . To a n a l y z e material p r o p e r t i e s u s i n g MeV helium p a r t i c l e s , t h e s u r f a c e of t h e sample t o be analyzed is i r r a d i a t e d w i t h a w e l l c o l l i m a t e d monoenergetic beam of helium atoms i n t h e MeV energy r a n g e . A s t h e probing beam p e n e t r a t e s the material, i t l o s e s energy by two processes: interactions with the electrons i n t h e s o l i d and i n t e r a c t i o n s w i t h t h e h o s t atom i n n u c l e a r scattering events. Only t h o s e p a r t i c l e s t h a t are s c a t t e r e d by c l o s e range n u c l e a r encounters i n t h e m a t e r i a l y i e l d t h e d e s i r e d information on t h e material p r o p e r t i e s .

A 4He i o n t h a t e x p e r i e n c e s a c l o s e range c o l l i s i o n w i t h a t a r g e t atom w i l l have i t s momentum e s s e n t i a l l y reversed and can

348

T. SHIBATA ET AL.

thus be backscattered into a solid state detector, as shown in Fig. 6.2(a). Such a particle will create in the detector a number of electron-hole pairs that is proportional to its energy.

ENERGY ( M e V )

ENERGY (MeV)

FIGURE 6.2. Rutherford backscattering spectra taken using 2.2 MeV 4He particles for (a) as deposited Si/Nb/Si sample and (b) identical sample with cw laser reaction (21 scans at p = 0.9).

6. METAL-SILICON REACTIONS AND SILICIDE FORMATION

349

This in turn creates a current that is proportional to the particle energy. This current is then used to measure the collected particle's energy. The function of the electronics following the detector is to convert this current (and energy) into a digital signal and store a count in the system memory at a location corresponding to the measured energy. Since we are primarily interested in depth profiles and compositional analysis of reacted and unreacted thin metal films on Si substrates, we shall utilize an experimental backscattering spectrum to illustrate the RBS analysis technique. In Fig. 6.2 we show.backscattering data taken on unreacted [Fig. 6.2(a)] and reacted [Fig. 6.2(b)] Nb/Si films. The initial sample structure was chosen for laser processing and consists of a Si(200 i%)/Nb (11001) / Si sandwich. Although the large majority of 4He particles incident on such a sample will penetrate many microns into the substrate before stopping, a small fraction (approximately one in 106) will experience close range (i.e., Rutherford) collisions and be backscattered out of the sample. A particle scattered from a Si atom on the surface of the sample of Fig. 6.2(a) will have an energy given by E = E(MEo

(6.2a)

where E(M, the kinematic factor, is given by [6.21]: m cos o

+ fM2 - m2 sin2 o m+M

I'

(6.2b)

In this equation, m is the mass of the 4He atom (4), M is the mass of the target atom, and 0 is the scattering angle in laboratory coordinates. For the geometry used in the RBS system used to take the data of Fig. 6.2, 0 = 170'. Using Eq. (6.2b), we find KSi = 0.5647 for Si and & = 0.8429 for Nb. Hence the resultant energy detected for a 2.2 MeV 4He particle scattered through 170' by a collision with a Si surface atom will be 1.242 MeV. This energy is referred to as the "Bi edge" energy and is indicated by an arrow at this energy in Fig. 6.2. The atomic composition at the surface of a sample can then (in principle) be measured by analysis of the associated backscattered energies. Helium particles that are not backscattered from the target surface will enter the film, where they lose energy to electronic interactions. The stopping cross section E is defined as [6.21]

1dE €=Ndx

350

T. SHIBATA ET AL.

where dE/dx i s t h e energy loss r a t e and N i s t h e atomic d e n s i t y of A s a 2.2 MeV 4He p a r t i c l e p e n e t r a t e s t h e t h i n S i the material. o v e r l a y e r i t l o s e s energy a t a rate of about 23.5 eV /A . Thus a 4He p a r t i c l e t h a t p e n e t r a t e s through t h e 200 A S i o v e r l a y e r w i l l have an energy of

0

a t the Si/W interface. In t h e "near s u r f a c e " approximation I6.211 t h e t e r m dE/dx i s considered a c o n s t a n t , having a v a l u e c a l c u l a t e d from t h e i n i t i a l beam energy, i.e.,

For t h e c a s e under c o n s i d e r a t i o n = 23.5 ev/ A , x1 I

EO

E(x1) = 2.2 MeV

200 8 , so

-

4.7 keV = 2.195 MeV.

I f t h e 4 He p a r t i c l e s u f f e r s a b a c k s c a t t e r i n g c o l l i s i o n a t x l , i t w i l l a r r i v e a t t h e d e t e c t o r w i t h energy E g i v e n by

Here t h e s u b s c r i p t i has been used s i n c e a p a r t i c l e a t t h e Si/Nb i n t e r f a c e may b a c k s c a t t e r from e i t h e r a S i o r Nb atom. The d i f f e r e n c e between t h e e n e r g i e s of t h e i n c i d e n t and emergent beams can be w r i t t e n i n terms of t h e energy l o s s [ S ] o r s t o p p i n g c r o s s s e c t i o n f a c t o r [ E ] , and i s AE = AxfS] = NAx[c]

(6.7)

u s i n g the near s u r f a c e approximation. The v a l u e s of [ E ~ ] and [So] are t a b u l a t e d as a f u n c t i o n of Eo and 0 w i t h t h e energy E

6.


approximated as &Eo AE is AE = (200 8)(47

[6.21].

35 1

For scattering from a Si atom at xi,

eV/8) = 9.4 keV.

(6.10)

Here the tabulated value for [ S o ] at 2.2 MeV has been used. By extension of this analysis we find that 2.2 MeV 4He particles backscattered from Si atoms in the Si overlayer will arrive at the detector with energies in the range 1.242 to 1.233 MeV. Under ideal conditions these particles would produce a rectangular backscattering spectrum, though as seen in Fig. 6.2(a), for very thin films the system response will broaden the rectangle into a bell-shaped curve. The area under the backscattering spectrum associated with the thin silicon layer is n

AE A = > - C i = H X i=1

(6.11)

L,

where n represents the number of channels, Ci is the number of counts in the ith channel, and 6~ is the system resolution in K.eV/channel. This area can be related to the physical parameters of the measurement through do A = NAx

(6.12)

QAQ

where Q is the total charge of 4He particles incident on the substrate, da/dQ is the differential Rutherford cross-section and A 0 is the solid angle subtended by the detector. For a 4He particle that reaches the Si/Nb interface and is scattered from a Nb atom the energy of the 4He particle arriving at the detector will be r

"I

4.71 -

EO

lx

ENb

5.2 = 1.845 MeV.

(6.13a) (6.13b)

On the spectrum, Fig. 6.2(a), ENb is the energy at which the Nb signal is at "half-height". Although it is close to the "Nb edge" (1.854 MeV), defined as the energy of the 4He backscattered from a surface Nb atom, it does not define the Nb edge since the Nb in this case is not located at the sample surface. Particles can

T. SHIBATA ET AL

scatter throughout the Nb film, producing the signal shown in Fig. 6.2(a). The area of this signal is proportional to the number of Nb atoms/cm2 in the film. One can calculate the energy corresponding to the lower half height of the Nb signal from

To perform this calculation we must assume a value for the density of the Nb film. By providing an independent measurement of the Nb film thickness, and measuring AENb experimentally an estimate of [So] and a corresponding estimate of film density can be obtained.

By a similar analysis one can find that 4He particles backscattered from the silicon atoms at the interface c between the niobium and the silicon substrate will reach the detector with an energy around 1.1 MeV; an exact calculation requiring a knowledge of the density of the Nb film. Particles that penetrate the silicon substrate before being backscattered will give rise to the low energy portion of the spectrum ( < 1.10 MeV) shown in Fig. 6.2(a). Let us now consider the changes in the RBS spectrum that occur when the silicon-metal-silicon sample undergoes a cw laser-induced silicide-forming reaction. In Fig. 6.2(b) we plot the backscattering yield vs particle energy for the Nb/Si couple after cw laser reaction. The surface Si film reacts with the underlying Nb, but the energy resolution of the backscattering system ( m 17 keV) is insufficient to detect a large change in the backscattered particle signal over that of the unreacted case. However, close examination of the Nb signal shows that the leading edge has moved up in energy to 1.854 MeV corresponding to backscattering of particles from Nb atoms at the sample surface. There is also a large change in the spectrum at the Nb/Si substrate interface. Here both the Si and Nb signals have spread out, with a step appearing in both signals. The observed increase in width of the Nb and Si signals is due to Si atoms in the Nb film causing an increase in 4He particle energy loss and vice versa. The composition of the silicide layer formed in this region + 1.18 MeV; 1.65 + 1.8 MeV) can be determined from the ratio of the spectrum heights from Eqs. 6.11 and 6.12. (-- 1.05

(6.14)

In the near-surface, thin-film approximation, this equation can be written as

6.

METAI,-SILICON REACTIONS AND SILICIDE FORMATION

353

(6.15) [Esi] have where the relationships (dU/dGIM a (ZMI2, and [EN,,] been assumed. For the 21 laser scan reaction of Fig. 6.2(b) this ratio is found to be n 0.5. Continued laser scanning of this film would result in complete mixing of the Nb and Si atoms with the Si signal extending from the substrate to 1.24 MeV. Such a reaction is shown in Fig. 6.8 for the case of Si/Pd/Si couples. For film thicknesses on the order of 0.1 + 0.2 pm, these approximations allow measurements to be made with an accuracy of 5 + 10%.

The composition of the film can be determined not only from the height of the spectrum components, but also (and more accurately) by integration of the areas underneath the metal and silicon peaks corresponding to the intermixed layer. By assuming an atomic density for either the metal or the silicide film, a depth scale can also be generated which then allows measurement of the rate at which the reaction front proceeds through the film.

I

I

I

I

I

I

ENERGY (MeV)

FIGURE 6.3. MeV 4~ ion backscattering spectra for Si(ZOOA)/ Nb (1100 R)/Si samples with multiple laser scans. bser irradiation was done at p = 0.9, where P is defined in Sec. 6.3.1. Reprinted by permission of the publisher, the Electrochemical Society, Inc. Publishing Co.

O1981 Elsevier Science

354

T. SHIBATA ET A L

Data on the formation kinetics of the reaction can also be obtained from the RBS spectra. For this purpose we observe from the backscattering spectrum in Fig. 6 . 3 that the reaction clearly did not go to completion during the first scan. However, as the number of scans of the laser is increased from 1 to 21, the thickness of the reacted film is seen to expand into the pure niobium metal. In Fig. 6 . 4 we plot the thickness of the niobium silicide formed by the laser as a function of the number of laser scans. We will return to a thermodynamic analysis of this thickness relation below. The composition determined from the Rutherford backscattering spectrum is an areal composition in atoms/cm2. lhis composition is most accurately described as the average composition within the film. In order to determine the metallurgical phase(s) of the reacted material, other techniques must be used. Typically glancing angle X-ray diffraction is used for this separate determination of film composition. The X-ray diffraction measurement, to be described in the next section, will indicate whether the reacted film is single phase, multiple phases, or single phase with metal and/or silicon precipitates. For the majority of the experiments to be discussed, glancing angle X-ray diffraction was employed to determine the phase composition of the films and Rutherford backscattering was used to indicate whether or not the entire film was reacted to form the X-ray measured phase.

NORMALIZED LASER POWER p : 0 9

'

0

4

r

-

NUMBER OF SCANS

FIGURE 6.4. lhe thickness of NbSi2 determined from backscattering spectra as a function of the number of laser scans. The data shows the parabolic growth for NbSi2 formation since the number of scans is proportional to the annealing time. 01981 Elsevier Science P u b l i s h i n g C O -

6.


355

The uniformity of the reacted films can also be qualitatively understood from the Rutherford backscattering spectrum. Again if we refer to Fig. 6.3, we see that for the as-deposited niobium on silicon, a sharp edge exists at the niobium-silicon interface. This sharp edge is displayed as a rapid decrease in the niobium signal at an energy of about 1.70 MeV. However, a rapid rise in the silicon concentration in a nonuniform manner at the metal/ silicon interface will cause these spectrum edges to become diffuse. As can be seen in the 21-scan reacted niobium silicon layer, the interface between the niobium metal and the NbSi2 formed (solid dots near 1.75 MeV) is still quite steep, indicating a uniform interfacial layer between the NbSi2 and metallic niobium. 6.2.4

~

Analytical Methods:

Read Camera

As we mentioned in the previous section, Rutherford backscattering measurements provide only information on the average composition of a reacted film. If we take a completely reacted NbSi2 film as an example, RBS can verify that the Nb and Si are distributed throughout the film in exactly a 1:2 ratio, but it cannot determine whether the compound NbSi2 has actually been formed, or whether the film is composed of a uniform mixture of other Nb,Siy compounds whose average composition is NbSi2. A unique determination of the phases present in the reacted film can be obtained from X-ray diffraction analysis. Several techniques are candidates for such experiments, including the recording X-ray diffractometer [6.22], the Seemann-Bohlin camera Although contrast in the Read [6.23] and the Read camera [6.24]. camera is limited and great accuracy is hard to achieve, exposure times are short (a few hours) and the instrument is therefore well suited to rapid compositional surveys of thin films. The Read camera is a wide film cylindrical instrument in which the sample is held with its surface at a small angle (typically 13") to the X-ray beam. The sample is both parallel to and along the cylindrical axis of the instrument. The camera geometry is illustrated in Fig. 6.5. Commercial instruments will accept samples up to approximately 1" x 1" with a suitable sample holder. The princ'iple of operation of the camera follows directly from the Bragg Law, 2d sin 0 = X

(6.16)

It is clear from this formula that the angle of reflection is related to the spacing between lattice planes d for a given X-ray wavelength X (Cu-Ka radiation is typically used). If a

356

T.SHIBATA ET AL.

L

PHOTOGRAPHIC

FILM

FIGURE 6.5 Schematic diagram of the Read camera illustrating the geometry used for the development in the text. silicide film consists of small, randomly-oriented polycrystals of a given composition, then in general we expect to find reflections at the angles O(), e(), etc., corresponding to the lattice plane spacings d(), d(). The reflections produce lines of film exposure as suggested in Fig. 6.5. Comparison of these lines with ASRl powder diffraction files [6.25] then permits positive identification of the compositional phase of the set of crystallites that are responsible for the separate reflections. The accuracy of identification is improved dramatically by careful analysis of the exposure lines that correspond to large angle reflections. For this case, we have, by differentiating the Bragg Law, Ad -

d = -

A@ tan G

(6.17)

For 0 + 180", tan Q, 0 and then small changes in the spacing of the lattice planes will give rise to large changes in the reflection angle. The Read camera image obtained for the Nb/Si film that was used as an example in the previous section on Rutherford backscattering is shown in Fig. 6.6. This pattern was obtained with the sample that had been given 21 scan frames of laser processing. The open circles in the pattern identify reflection lines (and therefore lattice parameter spacings) appropriate to crystallites of NbSi having a variety of orientations. %he solid circles identify

6.


357

X-RAY DIFFRACTION PATTERN FROM READ CAMERA

FIGURE 6.6. Glancing angle X-ray diffraction pattern for the Nb/Si structure laser annealed at p = 0.9 for 21 frames. Formation of the NbSi2 phase ( 0 ) is evident. reflection lines arising from pure Nb metal. To within the accuracy of the Read camera these are the only two phases present in the film. (The spots in the pattern correspond to X-ray reflections from the underlying silicon single crystal). We conclude from data of this type and the corresponding RBS spectra that the reacted layer was exactly NbSi2 for each of the laser scans employed. 6.3

EXPERIMENTAL RESULTS: LASER PROCESSING

In this section we will discuss the experimental results obtained on samples processed with scanning laser beams. For convenience we divide the discussion into the reaction of nearnoble metals on silicon, including Pd and Pt, and then we consider the reaction of refractory metals such as Mo, W and Nb on silicon. In Section 6.4 we present data obtained with electron beam processing; and in Section 6.5 we discuss the formation of the superconducting compounds Nb3A1 and Nb3Si using cw laser reaction. We summarize the beam process parameters and results of reactions that are described in detail in the following section in Table 6.3. 6.3.1

Near Noble Metals on Silicon: Pd/Si

In Fig. 6.7 we show an optical micrograph of a sample consisting of Si (200 A)/Pd (1300 A)/Si that was laser beam scanned as a function of l a s e r power level. It can be seen from this figure that the reaction of the Si/Pd/Si couple is initiated

358

T. SHLBATA ET AL.

TABLE 6 . 3 Summary of Beam Process Parameters and Resultant Reactions for the Silicides Discussed in this Work. METAL FILM

REACTED PHASE

RERCTED THICKXESS*

Id) Pd

PdZSi PdSl

Pt

Pt

Mo

1450

u

MoSi2 WSi2

Sb

NbSiZ

1350

,I+

+if

1930

2600

--

--

1200

F (NORMRLIZED LASER POWER1

POWER PARAMETER" (Watts/ucml

0.71 1.4 0.69

0.9

0.154 0.2 0.3

T,,b (OC)

50(%l.5)

0.185

0.m 0.25 I10 s c a n s l 0.9

0.093 0.087

37 31

---

0.14R

0.86

RESISTIVITY ( U S 'cm)

350(?1.0)

190 110

--

0.089

(21 scans)

Pd

w2si

Pt

PdSr PtSi

2280 2730 1670

Nb

NbSi2

12002

0.112 0.136 0.093 0.173

50

39 20

28 --

C a l c u l a t e d from b a c k s c a t t e r i n g s p e c t r a u s i n g t h e n e a r s u r f a c e a p p r o x i m a t i o n f o r t h e e n e r g y loss

**

C a l c u l a t e d v a l u e s of absorbed p o w e r f i e a m

radius:

(l-R)P/w

f A mixed phase compound i n c l u d i n g m e t a s t a b l e P t Z S i j

# T h i c k n e s s of a p a r t i a l l y r e a c t e d f i l m

at about 4 W of laser power. The laser-induced reaction is easily detected by the unaided eye through a surface color change. For the reaction initiated at a power level of 4 W, the surface color was observed to change from blue (unreacted film) to yellow (reacted film, low power phase). The low power phase was found to be stable up to a threshold power of about 7 W, above which a different high power phase appears in the center of the beam scan line. At this power level, the surface changes to a gray color that is characteristic of the high power phase. This high power phase was stable up to the power at which surface oblation was clearly evident. To obtain uniformly annealed areas for subsequent characterization, overlapped scans were performed with each scan line overlapped by at least 40%.

To simplify the discussion it is convenient to normalize the output power level of the laser (P) to the power at which the surface of a clean silicon wafer just begins to melt (Po). The normalized power (p = P/Po) is experimentally reproducible, thereby minimizing the dependence on the laser system and/or other experimental conditions. For typical experimental configurations, the values of Po were 9.0 and 5.0 W for 50" and 35OoC substrate temperatures, respectively. The normalized laser power is also related to a critical parameter of the cw beam process; i.e., the ratio of beam power to spot radius (P/w). This relation

6.


359

FIGURE 6.7. Optical micrograph of a sample consisting of Si(200 A)/Pd(1300 a>/Si after single laser scans at various power levels. The low power phase is formed between 4 and 7 W and the high power phase at the above 8 W. Reprinted by permission of the publisher. the Electrochemical Society, Inc.

is given by P/w (watts/micron) = 0.337 p or 0.191 p for Tsub = 5OoC or 35OoC respectively, using the formulation of Chapter 2. To summarize the argument presented in Chapter 2, we recall that the beam radius, w, is defined by the l/e point of the beam intensity assuming a Gaussian distribution. Since the actual power absorbed by the sample is reduced by the reflection, the essential power parameter is (1 - R)P/w, where R denotes the reflectance of the sample. This parameter is listed in Table 6.3 and has been calculated using experimentally determined values for R. Backscattering spectra for the Pd/Si samples which were laser annealed at p = 0.71 (low power phase) o r p = 1.1 (high power phase) are shown in Fig. 6.8. lbe spectrum for the as deposited sample is shown by the solid line. It is clearly seen that the entire metal layer was reacted with the Si after the laser irradiation. The average composition of the low power phase and high power phase was'determined to be Pd2Si and PdSi, respectively, using the near surface approximation for [ S ] . It should be noted that the spectrum of PdSi did not change significantly with different laser powers ranging from p = 0.89 up to p = 1.4. These results are quite different from those obtained with pulsed laser annealing, where the average composition of the silicide layer changes continuously with increasing laser power [6.141. Figure 6.9 shows the X-ray diffraction pattern obtained from The diffraction lines from the high power phase sample (p = 1.1). PdSi are indicated by solid circles with the most intense line from

360

T. SHlBATA ET AL.

-A S

DEPOSITED

p = 1 . 1 ( P d SI)

ENERGY (MeV)

FIGURE 6.8. Backscattering spectra for Si/Pd/Si laser annealed samples for various normalized laser power levels. The solid line (-1 represents data taken on unreacted samples, the circles ( 0 0 0 0 ) on samples reacted at low powers and the closed circles ( J-) samples reacted a t high power levels.

FIGURE 6 . 9 .

Glancing angle X-ray diffraction pattern from

a Read camera on the sample reacted at high power as shown in

The line indicated by 0 is the most intense diffraction Fig. 6.8. line from Pd2Si. Reprinted by permission of the publfsher, the Electcocheaical Society. Inc.

6.

METAL,-SILICON REACTIONS AND SILICIDE FORMATION

361

Pd2Si by an open circle. It can be concluded that the compound is essentially single phase PdSi with a trace amount of Pd2Si. A similar result was obtained for the low power phase sample; however, here the compound formed was mainly Pd2Si including a trace amount of PdSi. 6.3.2

Growth Mechanisms for NbSi2 and Pd2Si

To investigate the mechanisms by which NbSi2 and PdqSi grow, it is useful to first consider what the growth rate would be if the cw beam process were equivalent to furnace annealing. To explore this possibility, we first note that the data for the formation of silicides in a furnace are well characterized by the relation Ax or

AX)^

At exp

(- 2)

(6.18)

Here, Ax is the thickness of the reacted layer, A a material constant, t the annealing time, Ea the activation energy, and T the annealing temperature. The selection of Ax or AX)^ is made according to the formation kinetics; i.e., linear or parabolic growth, respectively. If the laser induced reaction is due to a solid phase mechanism, then the analytical treatment developed in Chapter 2, can be used to characterize the silicide formation. According to this formulation, the thickness of the reacted layer due to cw laser annealing can be obtained from Eq. (6.18) simply by replacing t and T by their effective values: teff and Teff. Teff is the peak value of the temperature distribution produced by the laser beam (Tmax), which is calculated in terms of the absorbed power divided by beam radius: (1 - R)P/w. The beam radius, w, is defined by the l/e point of the beam intensity assuming a Gaussian cross section, P the laser beam power, and R is the reflectivity. An example of the temperature calculations is shown in Fig. 6.10. The expressions for teff differ according to the laser scan mode, i.e., single scan or multiple overlapped scan, and are discussed in Chapter 2 . The data presented in Fig. 6.4 show that NbSi2 formed by laser annealing is characterized by a parabolic growth rate. As detailed growth kinetics data does not exist in the parabolic growth regime for NbSi2, we have used laser annealing data to determine A and Ea for NbSi2. The nealed at substrate ted usin of [So]&i2

Si(200 A)/Nb(1100 A)/Si samples were laser anvarious power levels with multiple (20) scans with the held at 350%. The reacted layer thickness was calculathe near surface 4He energy loss factor €or 2.2 MeV 4He = 83.7 eV/ A. The results for AX)^ are plotted

362

T. SHIBATA ET AL..

FIGURE 6.10. Example of r e a c t i o n temperature . c a l c u l a t i o n as a f u n c t i o n of absorbed powerfradius f o r a g i v e n (35OOC) s u b s t r a t e temperature.

a s a f u n c t i o n of t h e c a l c u l a t e d r e c i p r o c a l temperature i n Fig. It can be s e e n t h a t t h e d a t a l i e on a s t r a i g h t l i n e . A 6.11. l e a s t s q u a r e s f i t of t h e d a t a r e s u l t s i n t h e a c t i v a t i o n energy Ea = 1.43 e V f o r t h e p r o c e s s . S i n c e t e f f i s dependent upon b o t h Ea and Tmx, an i t e r a t i o n procedure w a s employed t o o b t a i n a more For each p o i n t , t e f f w a s c a l c u l a t e d u s i n g a c c u r a t e v a l u e f o r Ea. t h e m u l t i p l e o v e r l a p s c a n model w i t h Ea = 1.43 e V , t h e n a l e a s t This s q u a r e s f i t w a s performed t o o b t a i n (Ax) 2Iteff vs. 1/T. The n e x t i t e r a t i o n r e s u l t e d i n an a c t i v a t i o n energy Ea = 1.47 e V . d i d not a l t e r t h e v a l u e of Ea, g i v i n g t h e f i n a l r e s u l t s En = 1.47 e V and A = 1.08 x 1014 A2/sec. This a c t i v a t i o n energy i s d i f f e r e n t from t h a t (Ea = 2.1 - 2.7 eV) r e p o r t e d i n Ref. t6.261; however, i n t h a t work t h e l i n e a r , e a r l y s t a g e of growth was used t o d e t e r mine t h e r a t e s . F i g u r e 6.12 demonstrates t h e dependence of Tmax, t e f f and t h e t h i c k n e s s of t h e NbSi2 l a y e r a s a f u n c t i o n of t h e normalized l a s e r power. F i g u r e 6.12 a l s o shows t h a t t h e e x p e r i m e n t a l d a t a This is cona r e w e l l d e s c r i b e d by a s o l i d phase r e a c t i o n model. s i s t e n t s i n c e t h e maximum temperature value Tmax o b t a i n e d i n This v a l u e i s w e l l t h e experiment i s about 1100°C ( p = 0.96). below t h e published e u t e c t i c temperature (- 1300°C) f o r t h e Nb-Si system f6.271. S i m i l a r experiments were performed t o s t u d y t h e r e a c t i o n k i n e t i c s f o r t h e Si(200 A)/Pd(1300 A)/Si system. The samp l e s w e r e l a s e r annealed under c o n d i t i o n s t h a t form t h e Pd2Si phase. For t h i s c a s e o n l y one laser s c a n was used, and t h e r e f o r e t h e s i n g l e s c a n model w a s used f o r t h e c a l c u l a t i o n of t e f f .

6.


363

20 LASER SCANS teff

'0i.O

- 7msec 7.5

8.0

8.

FIGURE 6.11. Arrhenius plot of the square of the reacted thickness (NbSi2) as a function of calculated temperature [6.10] for a constant number of laser scan frames (21) and velocity. The thickness of the Pd2Si layer determined frombackscattering measurements is plotted as a function of the normalized laser power in Fig. 6.13. Also plotted is the calculated Pd2Si thickness vs laser power using the furnace anneal data of Ref. [6.28]. A systematic discrepancy between the experimental data and the calculation is apparent, the fit cannot be improved by adjustment of the constant A. However, using the recently reported furnace annealing data for the activation energy for Pd2Si formation from samples similar to ours (Ea = 0.93 eV) [6.29], the calculation was repeated, giving the results shown by the solid line in Fig. 6.13. The constant A was calculated from the best fit of the curve to the data. The agreement of this calculation with the experimental data indicates that the laser induced Pd2Si formation occurs by a mechanism similar to that discussed in Ref. [6.29]. 6.3.3

Formation Mechanism for PdSi

For the formation of PdSi, furnace annealing experiments have produced only inhomogeneous layers [6.30]. Recently, the formation of homogeneous single phase PdSi layers has been achieved by ion beam mixing [6.12] and cw-laser [6.7] or electron beam [6.8] processing. The formation of PdSi by cw-laser annealing cannot be explained by a solid phase or grain boundary diffusion mechanism.

364

T. SHIBATA ET AL.

h

Tmox 1-

1000 900

' 9

u

6.5 EFFECTIVE ANNEAL TIME :

c c

0 h

2ooo

w

0.8

P

0.9

iXiER:MLNi

Y

.-N

I .o

'1;

v)

9

2

v) v)

W

z Y 0

E

O 0

L

L

0.8 0.9 I .o p (normarized laser power)

FIGURE 6.12. Effective temperature, effective anneal time and NbSi2 thickness calculated as a function of the normalized laser power (p).

Eutectic melting provides one possible mechanism for the process, supported by the following observations: (a)

the surface of the PdSi phase shows a laminar-like morphology which suggests that some sort of melting could be occurring; (b) we have been unable to detect a gradual transition from Pd2Si to PdSi; and (c) the transition from Pd2Si to PdSi occurs abruptly above a threshold power (Pth > 0.8 Po).

6.


365

p : NORMALIZED LASER POWER

Thickness of Pd2Si as a function of normalized FIGURE 6.13. laser power. Calculated results are shown for two different activation energies. The calculated temperature at this threshold power is about 83OoC - 86OoC,which is close to the eutectic temperature for 58% Si in the Pd-Si system. However, if this eutectic mixture is formed during laser annealing, the excess Si must be rejected during the formation of the PdSi. So far, we have been unable to observe the excess Si either by backscattering or X-ray diffraction. However, we cannot rule out the possibility that the excess Si is recrystallized epitaxially back onto the Si substrate. The above observations can be alternatively interpreted in terms of the nucleation controlled reaction model proposed in Ref. [6.31]. We believe that this model is compatible with the concept of a moving hot spot produced by the scanning beam causing the propagation of a nucleation reaction from a nucleation site. Evidence for such a model is provided in Fig. 6.14, which is a Nomarsky interference micrograph of the Pd/Si sample after laser irradiation at p = 0.95. Here two single line scans were performed from two different directions. The center region of each scan line is seen to consist of light and dark regions. The darker region corresponds to the PdSi phase, while the lighter region consists of a nonuniform mixture of Pd and Si with no definite crystallite structure. It is especially interesting to note that the PdSi phase starts to form when the scanning beam passes over either the scratch (first scan) or the already existing

366

T.SHIBATA ET AL.

FIGURE 6.14 Nomarsky contrast optical micrograph of a Pd/Si sample laser scanned from two different directions. The PdSi phase is the darker center region which forms when the scanning beam passes over the scratch (1st scan) or the already existing PdSi phase (2nd Scan). O1981 Elsevfer Science Publishing CO.

PdSi phase (second scan). If the process is begun at the scribed edge of the wafer, a uniform PdSi layer can be formed over the entire area of the wafer. These observations strongly suggest that the formation of PdSi by laser process is a nucleation-controlled reaction, as suggested by Ref. [6.31]. However, the calculated temperature at the threshold power of the PdSi formation is close to the eutectic melting temperature, and we cannot rule out this mechanism. 6.3.4

Near Noble Metals on Si:

Pt/Si

The cw laser annealing of si(200 A>/pt(lOOO A)/si samples resulted in surface changes similar to those found in the Pd/Si sample (Fig. 6.7) when observed under an optical microscope. For this system the reaction started at a normalized power around p c 0.5, with a transition from a low power phase to a higher power phase occurring at p % 0.74. In contrast to this similarity however, the backscattering and X-ray diffraction results were quite different from those obtained for the Pd-silid.de films. The average composition of the low power phase calculated from the backscattering spectrum was close to PtZSi, but it was

6.


367

found fromthe X-ray diffraction analysis that the layer consisted mainly of Pt3Si and PtSi with a small amount of Ptl2Si5 and Pt2Si. The average composition found for the higher power phase was approximately PtSi2, a phase which does not exist in the equilibrium phase diagram. The X-ray diffraction pattern was identified as a mixture of PtSi, PtuSi5 and Pt3Si. However there existed three definite lines which could not be fitted to any data for Pt-Si compounds listed in the Powder Diffraction File. It should be noted that no Si precipitation or unreacted Pt was found in either of the above samples. We will return to a discussion of this data in Sec. 6.4 where similar reactions promoted by an electron beam were found to produce substantially different results. 6.3.5

Formation of Pt2Si3

In Fig. 6.15, we show a typical X-ray diffraction pattern found for a Si(200 a)/Pt(lOOO a)/Si sample laser annealed at p = 0.89. The backscattering spectrum for this sample indicated an average composition of PtSi2. However, X-ray analysis shows that a mixed phase silicide has been formed. Several unidentifiable lines are well explained by the metastable Pt2Si3 phase recently discussed in Ref. [6.32]. Further verification

Glancing angle X-ray diffraction pattern for FIGURE 6.15. the Pt/Si structure laser annealed at p = 0.86. Formation of superconducting Pt2Si3 phase is evident.

368

T. SHIBATA ET AL..

of this phase is provided by the fact that the layer exhibited superconducting behavior with a transition temperature around 4OK. The transition to this mixed phase Pt-silicide can be understood as arising from eutectic melting since it occurs at p > 0.75 with the calculated temperature for this power being about 840°C. 'his temperature is slightly higher than the lowest eutectic temperature in the Pt-Si system (83OOC). For this case the heating and cooling rates achieved by the scanning beam are of primary importance for rapid quenching of the metastable phase. 6.3.6

Refractory Metals on Si: (Mo, W, Nb)/Si

For pulsed laser annealing, it was reported that only a limited amount of reaction was observed in the Mo/Si system and no reaction whatsoever was observed in the Nb/Si system. For comparison, Fig. 6.16 shows the backscattering spectra obtained before and after cw laser annealing on Si(200 A)/Mo(530 8 ) Si samples. It is clear from these data that the entire film of Mo was completely reacted after laser annealing at p = 0.88. The average composition of the silicide was calculated to be MoSi2 from the spectrum (thickness = 1450 8 ) . The X-ray diffraction pattern shown in Fig. 6.17 indicates that the reacted layer is single phase MoSi2, free of unreacted Mo or Si precipitation. For the Si(200 A)/W(440 a)/Si(xtl) and Si(200 8)/Nb(1100 8 ) / Si(xt1) systems, laser annealing at p = 0.85 and 0.9, respectively, produced only a limited amount of reaction at the metal-Si interfaces. Therefore, multiple frame laser scans were performed for these two systems. The results of the backscattering analysis for W/Si and Nb/Si are given in Fig. 6.18 and Fig. 6.19, respectively. Figure 6.18 shows that the W layer of about 440 A was completely reacted to form a WSi2 film of about 1200 8 . That the film was single phase WSi2 was also verified by X-ray diffraction It should be pointed out that ten complete analysis (Fig. 6.20). scan frames were required to fully react this film. From the backscattering data one can conclude that the reacted film is quite uniform with a well defined silicide/silicon interface. In Fig. 6.19 the backscattering data clearly shows the advancement of the silicon niobium interface, occurring for a constant laser power, with the number of laser scan frames. The reacted film composition was calculated to be NbSi2 and verified by X-ray diffraction analysis (Fig. 6.21). The analysis of this film has been discussed in the section on Growth Mechanisms (Sec. 6 . 3 . 2 ) and will not be discussed further in this section.

6.

METALAILICON REACTIONS AND SILICIDEFORMATION

369

2.2 MeV 'He+ (1

41

w /. . n

-

u o o o AS DEPOSITED . I p I 0.88,1 SCAN (MOS12)

0 0

0

0

!?&J 0

0

+ .

:

;

o

O

- O 0 1.8

0

41.9

2.0

FIGURE 6.16. Backscattering s p e c t r a f o r Si(200 A)/Mo(500 A ) / Si s t r u c t u r e s before ( 0 0 0 ) and a f t e r ( 0 0 0 ) laser reaction. A l a y e r of MoSi2 of about 1450 thickness was obtained a f t e r 1 laser scan a t p = 0.88. Reprinted by permission of t h e publisher, t h e Electrochemical Society, Inc.

a

X-ray d i f f r a c t i o n p a t t e r n obtained from t h e FIGURE 6.17. Mo/Si sample laser annealed a t p = 0.88. Reprinted by permission of t h e p u b l i s h e r , t h e Electrochemical Society, Inc.

370

T. SHIBATA ETAL.

I

1

’

- 18 l n -16

-2

3

2 MeV ‘Hef

0

0

0

-

0

P

3

ENERGY (MeV)

FIGURE 6.18. Backscattering spectra for Si(200 8 , ) / W ( 4 4 0 a/ Si structures. About 400 8, of W (000) is completely reacted to form a WSi2 film ( 0 0 0 ) of about 1200 8, thickness after 10 laser Scans at p = 0.85. Reprinted by permission of the publisher, the Electrochemical Society, Inc. I

I

I

ENERGY ( M e V )

FIGURE 6.19 Backscattering spectra for Si(200 8,)/Nb(1100 a)/ Si samples with multiple laser scans. The thickness of the silicide (NbSi2) increases with the number of scans.

6.


6.4 EXPERIMENTAL RESULTS:

371

ELECTRON BEAM PROCESSING

In this section we discuss the formation of silicides by the use of a scanned cw electron beam. Although all experimental results to be discussed utilized the electron beam system described in Sec. 6.2.2 to promote the reactions, it should be pointed out that similar results have been obtained with a modified SEM [6.33]. In Table 6.4 and Fig. 6.la the materials used and the sample configuration for the electron-beam experiments are summarized. Table 6.4 lists the important experimental parameters and results achieved for these electron-beam reacted silicides.

6.4.1

Near Noble Metals on Si : Pd/Si

The formation of Pd2Si and PdSi by scanning electron-beam reaction proceeds in a manner analogous to the scanned laser. Again the choice of the phase is determined by selection of the e-beam P/r value. Figure 6.22 shows backscattering data taken on a 1400 A layer of Pd on Si both before and after scanned e-beam processing. 'Jbo distinct phases are clearly discernible in this data. Glancing angle X-ray analysis confirms the phases calculated from the backscattering spectra. Again, similar to the scanned laser results, essentially single phases are found with only a trace amount of the other phase present. TABLE 6.4. List of Properties of Scanned Electron-Beam Reacted Silicide Thin Films

UNREACTED THICKNESS

REACTED THICKNESS

Ptsi

-

1100 A

-

1670 k

0.93 kW/m

24OoC

27.9

Fine g r a i n ( 1 U) orange p e e l

PdzSi

-

1400 A

-

2280

A

1 . 1 2 kW/m

290'C

38.5

F U ~grain ( 2 PI orange p e e 1

2730 8

1.36 kW/m

375%

20.2

1250 R

-

METAL FILM

REACTED PHASE

Pt

w

PdSi

Nb

NbSiz

-

P/I ( E g = 3 1 keV1

ESTIMATED SURFACE TEMP. ( T s d = 50-C)

RESISTIVITY (Un-cm)

SURFACE TEXTURE

Laminar

like 510 A

1.36 kw/cm

375'C

780 k

1 . 5 5 kW/m

450oc

__ -_

A

1 . 7 3 kW/m

525'C

--

1200

UnlrnOYn

372

T.SHIBATA ET AL.

X-ray diffraction pattern obtained from the FIGURE 6.20. Si/W/Si structure laser annealed at p = 0.85 for 10 scans. Reprinted by permlesion of the publieher. the Electrochemical Society, Inc.

X-RAY

DIFFRACTION PATTERN FROM READ CAMERA

FIGURE 6.21. X-ray diffraction pattern obtained from the Si/Nb/Si structure after laser annealing.

6.

-cn

I

I-

-

-


SCANNED CW e-BEAM REACTED

Si(100) 3

1Pd

f k . .

'

-1400i

373

pd ?I

2 2 MeV 4He+

* * * AS DEPOSITED

-

00

PdpSi (P/r = 0.112 W / p ) ; t t 2280

-

A A

PdSi (p/r = 0.136 W / p ) ;

a

~27308, I - 4

A 0 ' A0 A 0 A 0

a

A

.

B 4

ENERGY ( M e V ) FIGURE 6 . 2 2 Backscattering spectra for Pd/Si before (0.0) and after reaction by a scanning electron beam; ( 0 0 0 ) Pd2Si and (AAA) PdSi.

The e-beam-reacted PdSi exhibits a laminar like surface structure along the direction of the scan. In Fig. 6.23 we show optical Nomarsky micrographs of both the Pd2Si and PdSi films. A point of potential interest for device fabrication is that the PdSi film was found to have a very low resistivity (20 pn-cm) as compared to the other films discussed in this chapter (see Table 6 . 4 ) . The theoretical development of Chapter 2 can be used to study scanned e-beam reactions by using the beam voltage and current to calculate the beam power (P). The reflectivity is set to zero and the beam diameter w is measured experimentally. 6.4.2

Near Noble Metals on Si : Pt/Si

In Fig. 6.24 we show backscattering data taken on electronbeam reacted Pt with Si. The data clearly shows that a uniform reaction of the film has taken place at a P/r value of 0.093 WID. The average composition of the film calculated from the spectrum is PtSi. X-ray diffraction analysis verifies that the film is single phase PtSi.

374

T.SHIBATA ET AL.

FIGURE 6.23.

Normarski micrographs of electron beam reacted

SCANNED CW e-BEAM REACTED 16

-

11ooA a

14 -

___ __

12 -

s1

- (r3)i'':

pt

*** AAA

6 4-

-

2 2 MeV 4He'

AS DEPOSITED PtSl (P/r=0.093W//.L) ( t = 1670 A)

A A.

6

A A A'

A

A . A A -

4 0

4

2 -

0-

FIGURE 6.24. Backscattering spectrum shown for Pt/Si film reacted by a scanning electron beam. The closed circles (.*) represent the spectrum from the As-deposited film while the triangles ( A A A ) are for the e-beam reacted. Reprinted by permission of the publisher, the Electrochemical Society, Inc.

6.


375

This result is in sharp contrast to that reported for the laser-beam reacted PtSi in Sec. 6.3.4, where for similar P/r values and scan speeds, a mixed phase silicide resulted. This difference can be caused by two phenomena: the basic physical differences between the laser and e-beam experiments are sample structure and beam energy deposition profile. Although we cannot rule out the possibility that interfacial effects such as bond breaking at the Pt/Si interface by the penetrating electrons may help this reaction proceed, recent experimental evidence suggests that the difference is due to the presence of the thin silicon overcoating present on the laser samples. 6.4.3

Refractory Metals on Si : Nb/Si

The formation of NbSi2 from Nb/Si by use of a scanned e-beam has also been reported 16.181. As in the case of the laser reacted films progression of the reaction interface has been observed by backscattering measurements. The existence of a single phase (NbSi2) compound was verified by using glancing angle X-ray analysis. 6.5

THERMAL STABILITY AND OXIDATION PROPERTIES OF CW BEAM REACTED SILICIDES

The thermal stability and oxidation characteristics of silicide compounds are important properties for the application of these materials in semiconductor technology. Future integrated circuits will require both gate and interconnect metallizations that have low sheet resistivity and can withstand various wet and dry processing steps, including high temperature furnace processing and oxidation. The capability of oxidizing a silicide film without any phase change and degradation of its electrical properties will be especially important prerequisites for the use of these materials. To date a number of refractory metal silicides have been proposed as alternatives to the polysilicon gate material and electrical interconnections, because of their lower bulk resistivities and compatability with present and future MOS process technologies. However, the extremely low bulk resistivity and unique properties afforded by such noble metal silicides as PdSi and PtSi2 also makes these materials promising candidates for possible device application. Thermal oxidation studies have been reported for several metal silicides, including MoSi2, WSi2, TaSi2, NiSi2, and TiSi2 [6.34-6.431. However, the reported oxidation properties have been found to vary widely, depending on the technique used to

376

T. SHIBATA ET

a.

deposit the films. For tungsten-based films, a tungsten-rich silicide phase (W5Si3) is found after steam oxidation of sputtered WSi2 films [6.38]. This is in contrast to the behavior of coevaporated WSi2 films, where no change in the film stoichiometry during oxidation has been reported [6.37]. Also, it has been observed that dry oxidation of WSi2 results in films of poor surface quality i6.38). Furthermore we would expect that MoSi2 films on silicon would behave similarly to WSi2 because MoSi2 is chemically and structurally alike to WSi2. However, when these refractory silicides are deposited on oxide substrates, decomposition of the film to metal-rich phases are observed [6.34, 6.36-6.381. Data on the oxidation properties or thermal stability of PdSi films have been reported because of the difficulty of forming these films by conventional furnace techniques. In the following sections we present studies of the thermal stability and oxidation properties of WSi2, MoSi2, and PdSi as formed by scanned cw laser beam reaction, and compare the results with those reported for both sputtered and coevaporated films. 6.5.1

Refractory Metal Silicides : WSi2, MoSi2

In Fig. 6.25 we show Rutherford backscattering data taken on oxidized, laser-reacted WSi2 films as a function of oxidation time. Ihe data shown are for samples oxidized in steam at 900'C. Also shown is the spectrum of an unoxidized WSi2 structure. It can be seen from Fig. 6.25 that as the oxidation process proceeds, the tungsten peak is shifted to lower energies. Ihis shift arises from the energy loss of the 4He atoms in the SiOz layer being grown on top of the WSi2 film. Conversion of this energy loss into a corresponding thickness allows measurement of the oxidation rate to be made. At the silicon edge of the spectrum it is also seen that the interface between the silicon crystal and the WSi2 (occurring at about 1.1 MeV for the unoxidized sample) shifts to lower energies following the thermal oxidation. A new peak at around 0.8 MeV, superimposed on this silicon spectrum, also appears due to 4He particles backscattered from the oxygen atoms in the Si02 film. If one further examines the tungsten peaks in this figure, a decrease in the slope of the leading edges of the peaks is seen to occur with increasing oxidation time. This causes the peak height to decrease with increasing oxide thickness. This result indicates that the oxidation of the WSiz film is somewhat nonuniform, as the areas under these peaks remain constant. The constant area also indlcates that no tungsten I s lost from the film during the oxidation. By using glancing angle X-ray diffraction analysis, the stoichiometry of the WSi2 films after oxidation was studied. Diffraction lines corresponding only to single phase WSi2 films in steam at 900°C for similar periods of time [6.38]. Although it has been

6.


377

ENERGY ( M e V )

FIGURE 6.25. 2.2 MeV 4He backscattering spectra for laser represents the reacted and oxidized WSi2. The solid line (-) unoxidized control WSi2 layer. Spectra are shown for steam oxidation at 900°C for 20 min. ( - * * ) , 60 min. ( o o o ) , and 120 min (xxx) oxidation times. reported that dry oxidation of sputter deposited WSi2 films results in films of poor quality, laser-reacted WSi2 films subjected to dry oxidation at 1000°C appear to have smooth surfaces, suggesting that the Si02 film grows on the WSi2 without decomposition of the silicide. Also the X-ray diffraction pattern obtained from laser reacted films subjected to a 60 min. dry oxidation at 1000°C shows no diffraction lines present other than those of the single phase WSi2 compound. Furthermore, no indication of wO3 formation was found and surface morphology was good after the process. One can conclude from these observations that the oxidation mechanism for the laser reacted films is similar to that proposed for coevaporated WSi2 films where it is believed the silicon atoms are supplied only from the silicon substrate by solid state diffusion through the silicide film. Similar to WSi2, the dry oxidation of laser-formed MoSi2 on silicon proceeds without any change in phase nor any indication of Moo3 formation. However, as in sputtered MoSi2 on oxides, MoSi2 transforms to the metal rich silicide once the underlying silicon is completely consumed or silicon transport from the silicon substrate to the silicide surface is impeded by impurities

378

T. SHIBATA ET AL..

at the silicon-silicide interface. By assuming local equilibrium at the growing oxide-silicide interface and favorable formation kinetics, the phases observed at this interface can be predicted from the ternary phase diagram for all oxidation conditions. A simplified ternary phase diagram was constructed from available thermodynamics data (Fig. 6.26). Only the stable tie lines between two phases have been determined. Solubility regions in the single and two phase regions, miscibility gaps, and ternary phases have been neglected. Details on the construction can be found in several references [6.44, 6.451. Although the phase diagram has been oversimplified, the diagram allows us to explain the co-existence of certain phases and the absence of metal oxides. For Si02 growth to continue and yet maintain the Most2 phase, a silicon flux to the MoSi2-Si02 interface is required. Therefore, an interface silicon concentration is established at this interface so that the overall composition at this interface lies 0

Mo

5\

+!’ FIGURE 6.26.

f +

+!”

c;’”.

+!

Mo-Si-0 ternary phase diagram.

Si

6.


379

along t h e MoSi2-Si02 t i e l i n e . However, i f s i l i c o n t r a n s p o r t i s impeded d u r i n g o x i d a t i o n by i m p u r i t i e s , t h e s i l i c o n c o n c e n t r a t i o n d e c r e a s e s and t h e o v e r a l l i n t e r f a c e composition may l i e i n r e g i o n s 11, 111, I V of t h e t e r n a r y phase diagram o r along any of t h e t i e l i n e s , such as Mo5Si3-Si02 o r Mo3Si-SiO2 (Fig. 6.26). From t h e thermodynamics w e can p r e d i c t t h a t only Si02 w i l l form a s long a s t h e o v e r a l l i n t e r f a c e composition l i e s i n t h e s i l i c o n r i c h r e g i o n of t h e phase diagram. Phases corresponding t o r e g i o n s 11, 111 and IV have been observed when oxygen o r Si02 i s p r e s e n t a t t h e MoSi2-silicon i n t e r f a c e p r i o r t o o x i d a t i o n [6.46]. F i g u r e 6.27 shows t h e b a c k s c a t t e r i n g s p e c t r a of MoSi2 f i l m s Although a f t e r 15 hours of on p o l y s i l i c o n oxidized a t 9 0 9 O C . o x i d a t i o n i n dry 02 t h e molybdenum peak has decreased and t h e r a t i o of t h e s i l i c o n t o molybdenum peak has i n c r e a s e d beyond 2, g l a n c i n g a n g l e X-ray d i f f r a c t i o n i n d i c a t e s MOSi2 as t h e only molybdenum phase p r e s e n t . Since less than 0.5% of molybdenum was d e t e c t e d a t t h e MoSi2-oxide i n t e r f a c e and t h e molybdenum peak a r e a s remain e s s e n t i a l l y c o n s t a n t d u r i n g o x i d a t i o n , t h e s e r e s u l t s

b

JJ -2j2 -.MV .

2000 BEFORE OXlOATlON

AFIER OXIDATION IN O W AT 909C

OXYGEN

1500

cn

F

5

1000

0

0

500

0

-.-

5Hr

,. -.

---I~HP I I I

0.4

0.6

0.8

I

I

I

1.2 1.4 1.0 ENERGY ( M e V )

I

I

1.6

1.8

FIGURE 6.27. 4He b a c k s c a t t e r i n g s p e c t r a f o r l a s e r r e a c t e d MoSi2 on p o l y s i l i c o n oxidized f o r 0 h r s . ( s o l i d ) , 2 h r s . ( d o t s ) , The s h i f t i n 5 h r s . (dash-dots), and 1 5 h r s . (dash) a t 909OC. t h e molybdenum peak i s due t o t h e growth of Si02 on t h e s u r f a c e .

380

T.SHIBATA ET AL.

indicate very little loss of molybdenum from the film. Furthermore, the decrease in the slopes of the leading and trailing edges suggest that the oxidation is nonuniform. Therefore, the decrease in the molybdenum peak height possibly results from the nonuniformity of the oxidation and/or internal oxidation along grain boundaries. The TEM micrograph from the cross-sectional view of the oxidized film confirms that the interface has become nonuniform from long oxidations (Fig. 6.28). Grains of MoSi2 suspended in the Si02 also suggest that oxidation along grain boundaries has occurred. For similar oxide thicknesses grown in wet 02, but with shorter times, the molybdenum peak height remains constant and the interface nonuniformity is comparable to that observed for WSi2 oxidized in steam. The use of these silicides in integrated circuits also requires that the film retains its conductivity. Table 6 . 5 summarizes the resistivities of 100 nm of MoSi2 oxidized for 0.5, 4 , and 10 hours in dry 0 2 . Only after 10 hours of oxidation does the resistivity of the film degrade from 12 to 27 Cl/a This degradation corresponds to the observed increase in the silicon to molybdenum ratio.

.

FIGURE 6.28. Cross-sectional TEM micrograph of MoSi2 on polysilicon oxidized at 900°C for 15 hours.

6.


381

TABLE 6.5. R e s i s t i v i t i e s of 1000 A of MoSi2 on P o l y s i l i c o n Oxidized a t Various T i m e s a t 1007°C.

Dry Oxidation Condition

Resistivity

0 hr. 0.5 4 10

a

of 1000 Doped P o l y s i l i c o n

N

a/a

2.1 2.2 3.0 3.9

a/o n/o n/n 27 a10 12 10 12

- 100 n/o

Resistivity pil-cm

120 200 120 270

10-3 a cm

The oxide t h i c k n e s s e s were measured by Rutherford backscatt e r i n g f o r f i l m s o x i d i z e d i n dry 02 a t temperatures of 800 t o 1100°C f o r times t h a t ranged from 0.5 t o 42 hours, and i n w e t 0 2 a t 800 t o 1000°C f o r 0.5 t o 2 hours. For both w e t and dry oxidat i o n , t h e growth of Si02 on t h e s i l i c i d e s u r f a c e i s p a r a b o l i c w i t h t i m e . By making an Arrhenius p l o t of -EaJkT xo2 = Ae where A i s t h e pre-exponential f a c t o r , k i s t h e Boltzmann cons t a n t , and Ea i s t h e a c t i v a t i o n energy; t h e a c t i v a t i o n e n e r g i e s f o r t h e d r y and w e t o x i d a t i o n processes a r e determined t o be 1.6 and 1.3 5 0.2 e V , r e s p e c t i v e l y ( F i g . 6.29). 6.5.2

Near Noble Metal S i l i c i d e s :

Pd/Si

Annealing and o x i d a t i o n s t u d i e s on PdSi f i l m s formed' by scanned cw l a s e r beam r e a c t i o n have been performed over a temperature In F i g . 6.30 w e show b a c k s c a t t e r i n g range from 750°C t o 900°C. s p e c t r a taken u s i n g 2.2 MeV 4He on PdSi f i l m s t h a t were isochron a l l y annealed i n an N 2 ambient a t temperatures from 750" t o 900°C f o r one hour. Also shown f o r comparison on t h i s f i g u r e i s a I t i s c l e a r from spectrum of a laser r e a c t e d unannealed f i l m . Fig. 6.30 t h a t t h e PdSi f i l m has completely decomposed d u r i n g t h e 900°C one hour a n n e a l , w i t h deep d i f f u s i o n of Pd i n t o t h e s i l i c o n s u b s t r a t e r e a d i l y observable. The spectrum i s shown i n t o t h e The d i f f e r e n c e i n t h e h e i g h t of t h e s i l i c o n edge (- 1.24 M e V ) . s i l i c o n spectrum f o r t h e c a s e of t h e 900°C a n n e a l compared t o t h a t of t h e unannealed sample i s due t o d i l u t i o n of t h e s i l i c o n s u b s t r a t e by t h e Pd and t h e r e s u l t a n t d i f f e r e n c e i n t h e s t o p p i n g

T. SHIBATA ET M .

382

108:

J

1 107-!

i06-: WET

lo5-[ Y

io4-;

lo3-

6

1

7

8 IOOO/T

10

9

11

K"

FIGURE 6.29. Determination of the activation energy from the Arrhenius plot of the oxide growth on MoSI2.

6.


383

UNAN N EALED 750°C,1 HR. * * * * * * 8 5 O o C , 1 HR.

7

ooooo

ENERGY I MeV)

FIGURE 6.30. Backscattering spectra for N2 annealing of laser reacted PdSi as a function of temperature. Also shown is the as-reacted sample for reference. power. Nomarsky microscopy indicates that the surface morphology of the 900°C annealed sample is bad with visible surface damage present. In contrast to the sample annealed at 900°C the backscattering spectrum for the sample annealed at 85OoC for one hour shows very little change compared to the spectrum of the unannealed sample. A very slight change is observed at the Si/PdSi interface (- 1.65 MeV), and a trace amount of Pd2Si is found to appear in the X-ray diffraction pattern of this sample. However, a complete phase change is observed for the sample annealed at 75OOC for one hour, as seen in Fig. 6 . 3 0 . Both the backscattering and X-ray diffraction analysis confirm that the entire film decomposed from PdSi into Pd2Si with no trace of PdSi or polycrystalline silicon evident. The data thus suggest that the excess silicon in the PdSi film has deposited at the silicon Pd2Si interface epitaxially. In Table 6.6 a summary of the film properties obtained during one hour anneal in flowing N2 are listed.

384

T. SHIBATA ET AL.

TABLE 6.6. ments. Temperature (“C)

Time

Phase Changes of PdSi Under Various Heat Treat-

Oxide Thickness (A)

Results of F i l m Characterization X-ray D i f f r a c t i o n

RBS

-__

PdzSi ( N O PdSI)

850

_--

PdSi (Trace o f PdzSi)

900

_-_

PdzSi (Trace o f PdSi)

Deep metal p e n e t r a t i o n into S i s u b s t r a t e

560

Pd2Si (NO PdSi)

Well-defined s i l i c i d e s i l i c o n interface

440

PdSi (Trace o f PdZSi)

750

7 50

1 hour

1 hour

850 7 50

1 hour

1750

PdZSi ( N o PdSi)

850

2410

PdSi > PdzSi (?tixed Phase)

900

1610

PdzSi (Trace o f PdSi)

Well-defined s i l i c i d e s i l i c o n interface

Gradual i n t e r f a c e

Deep metal p e n e t r a t i o n

I n Table 6.6 w e a l s o summarize t h e e f f e c t of dry o x i d a t i o n on l a s e r - r e a c t e d PdSi f i l m s . From t h e d a t a l i s t e d i n t h i s t a b l e i t can be s e e n t h a t t h e behavior of t h e PdSi a t 750 and 850°C i n t h e oxygen ambient i s similar t o t h a t observed d u r i n g t h e n i t r o g e n ambient a n n e a l s , except f o r t h e growth of an Si02 o v e r l a y e r on the film surface. Since t h e oxides formed d u r i n g dry o x i d a t i o n were found t o be q u i t e t h i n (400-600 A ) , steam o x i d a t i o n of t h e PdSi f i l m was i n v e s t i g a t e d a s a technique f o r i n c r e a s i n g t h e oxide f i l m t h i c k n e s s without i n c r e a s i n g t h e time-temperature product. We show t h e r e s u l t s of t h e steam o x i d a t i o n a t 750 and 850°C f o r one hour i n Fig. 6.31 t h e r e s u l t a n t d a t a i s summarized a g a i n i n Table 6.6. The b a c k s c a t t e r i n g spectrum i n Fig. 6.31 c l e a r l y shows t h a t t h e PdSi phase remains s t a b l e d u r i n g t h e 850°C o x i d a t i o n but changes t o Pd2Si f o r t h e 750°C o x i d a t i o n . We can a l s o see t h a t t h e i n t e r f a c e between t h e PdSi f i l m and t h e s i l i c o n s u b s t r a t e i s no l o n g e r w e l l d e f i n e d a f t e r t h e o x i d a t i o n but i s s p r e a d o u t i n energy i n d i c a t i n g i n t e r m i x i n g and p o s s i b l e deep metal p e n e t r a t i o n i n t o t h e s u b s t r a t e . This r e s u l t is i n c o n t r a s t t o t h a t observed f o r t h e s e t e m p e r a t u r e s f o r both t h e N2 and dry 02 a n n e a l s . X-ray d i f f r a c t i o n a n a l y s i s on t h e sample s u b j e c t e d t o w e t o x i d a t i o n a t 850°C shows t h e e x i s t e n c e of e s s e n t i a l l y t h e two phases PdZSi and PdSi.

6.

-

0 43

s


-?TI-

385

STEAM OXIDATION OF PdSi

=Si

2.2 MeV 4H$

m

-

0

6.6

PdSi

\'-;

; I

LASER PROCESSING OF Nb3X SUPERCONDUCTORS*

The s y n t h e s i s of m e t a s t a b l e superconducting compounds w i t h t h e A15 c r y s t a l s t r u c t u r e , such as Nb3Ge, Nb3A1, and Nb3Si, has r e c e i v e d c o n s i d e r a b l e a t t e n t i o n because of t h e w e l l known potent i a l of t h e s e materials f o r h i g h temperature s u p e r c o n d u c t i v i t y (Tc > 20'K). The d i f f i c u l t i e s t h a t have been encountered i n t h e s y n t h e s i s are w e l l i l l u s t r a t e d by Nb3A1, which i s t h e least unNb@1 i s known t o e x i s t as s t a b l e member of t h e group [6.47]. a niobium r i c h A15 compound over a l a r g e temperature range i n However, t h e A15 t h e e q u i l i b r i u m phase diagram [6.48-6.491 phase i s b e l i e v e d t o approach s t o i c h i o m e t r y only a t r e l a t i v e l y h i g h temperatures (1700-1960OC) t6.48-6.491 Hence, even though t h e exact d e t a i l s of t h e phase diagram a t high temperatures remain u n s e t t l e d , i t is t o be expected t h a t a n e a r l y s t o i c h i o metric A15 phase w i l l be d i s p r o p o r t i o n a t e a t low temperatures unless i t is cooled very r a p i d l y . Rapid cooling, on t h e o t h e r

. .

. .

*The work r e p o r t e d i n t h i s s e c t i o n was done i n c o l l a b o r a t i o n w i t h J Kwo, R D Feldman, and T H Geballe and R H Hammond [6.47 , 6.511

. . . .

..

386

T. SHIBATA ET AL.

hand, t e n d s t o f r e e z e i n atomic d i s o r d e r and s t r u c t u r a l d e f e c t s . Much of t h e e f f o r t i n t h e s y n t h e s i s of a high-Tc Nb3A1 has t h e r e f o r e been d i r e c t e d toward f i n d i n g a s u i t a b l e compromise between s t o i c h i o m e t r y and o r d e r i n g . Even g r e a t e r d i f f i c u l t y is encountered i n t h e s y n t h e s i s of NbjSi, s i n c e a s t o i c h i o m e t r i c A15 phase a p p a r e n t l y does n o t even A s a r e s u l t , mae x i s t i n t h e e q u i l i b r i u m phase diagram [6.50]. t e r i a l s s y n t h e s i s procedures t h a t a r e known t o be capable of formi n g m e t a s t a b l e phases are very e s s e n t i a l t o t h e s y n t h e s i s of a high-Tc form of Nb3Si [6.51]. A s i n d i c a t e d i n t h e p r e v i o u s s e c t i o n , d e p o s i t i o n of concent r a t e d energy w i t h t h e cw laser beam h a s been s u c c e s s f u l l y a p p l i e d t o t h e m o d i f i c a t i o n of e l e c t r o n i c materials, i n c l u d i n g t h e formaI n p a r t i c u l a r , metastable t i o n of m e t a s t a b l e a l l o y s [6.7, 6.181. s i l i c i d e s , which have so f a r proven d i f f i c u l t o r impossible t o form by e i t h e r conventional f u r n a c e a n n e a l i n g o r p u l s e d beam a n n e a l i n g , can r e a d i l y be o b t a i n e d w i t h cw b e a m a n n e a l i n g . I n e f f e c t , t h e a n n e a l i n g p r o c e s s u s i n g a scanning cw beam can be With such d e s c r i b e d as a "high temperature, short-time f u r n a c e " a p r o c e s s , t h e a n n e a l i n g temperature can e a s i l y be r a i s e d t o above 2000°C by a p r o p e r choice of t h e laser power; t h e a n n e a l i n g t i m e can be i n c r e a s e d by employing m u l t i p l e laser s c a n s ; and h e a t i n g and c o o l i n g r a t e s can be c o n t r o l l e d by changing t h e beam scan 106"C/sec>. speed (up t o a m a x i m u m c o o l i n g r a t e of

.

-

These f e a t u r e s are q u i t e a t t r a c t i v e f o r t h e s y n t h e s i s of Nb-A1 and Nb-Si A15 compounds. For example, it i s r e a s o n a b l e t o expect t h a t w i t h cw beam a n n e a l i n g , n e a r l y s t o i c h i o m e t r i c NbyU can be quenched from a high temperature A15 s t a t e t o room t e m p e r a t u r e w i t h o u t decomposition i n t o low temperature phases. In a d d i t i o n , t h e formation of t h e m e t a s t a b l e A15 Nb3Si phase might be favored by t h i s p r o c e s s . I n t h i s s e c t i o n w e d e s c r i b e r e c e n t a t t e m p t s t o apply cw laser a n n e a l i n g t o t h e s y n t h e s i s of Nb3A1 and Nb3Si, and w e d i s c u s s t h e p r e l i m i n a r y r e s u l t s and t h e potent i a l a p p l i c a t i o n of cw beam a n n e a l i n g f o r forming m e t a s t a b l e phases of t h e s e m a t e r i a l s .

6.6.1

Sample P r e p a r a t i o n

a)

were preThin f i l m s of Nb-A1 (5000 4 ) and Nb-Si (1000 pared by t h e electron-beam coevaporation p r e s s u r e of ( 1 .O + 3 - 0 ) x T o r r . T y p i c a l d e p o s i t i o n c o n d i t i o n s are 3 0 A/sec dep300°C s u b s t r a t e temperature f o r t h e Nbyll, and o s i t i o n rate, 50 A/sec d e p o s i t i o n r a t e , 5OO0C s u b s t r a t e temperature f o r t h e Nb3Si. F u r t h e r d e t a i l s on t h e t h i n - f i l m d e p o s i t i o n are g i v e n i n These f i l m s were d e p o s i t e d o n t o a R e f s . [6 -47, 6.51, 6.521. s i n g l e c r y s t a l S i s u b s t r a t e c o a t e d w i t h a t h i n l a y e r of Si3N4

-

-

-

-

6.

METAL -SILICON REACTIONS AND SILICIDE FORMATION

387

or Si02 as a buffer. This substrate is used because (a) single crystal silicon is a good heat conductor at high temperatures and ( b ) analytical calculations for the laser-induced temperature rise [lo] are well established for a silicon substrate. 6.6.2

Results and Discussion

Figure 6.32 shows a typical x-ray diffraction pattern obtained from as-deposited samples of Nb-A1. The principal lines shown were identified as arising from the bcc structure. Some weak lines from the A15 phase were also observed. A variety of laser annealing experiments were performed with calculated laserannealing temperatures (Tmax) in the range 1300°C to 1800°C. Figure 6.32 also shows the x-ray diffraction pattern of a sample (of 24.6 at. % Al) which was laser-annealed at 1400°C with 10 30% in each laser scan frames. Scan iines were overlapped by frame. Formation of the A15 structure after laser annealing is clearly observed, with some weak diffraction lines from the bcc phase. All laser annealed Nb3A1 samples examined exhibited

-

FIGURE 6.32. Read-Camera diffraction pattern for a Nb-A1 sample (24.6 at.%Al); (a) as-deposited (b) after laser anneal 10.

388

T. SHIBATA ET AL.

similar diffraction patterns, where both the A15 and bcc phases exist. The relative abundance of the A15 to the bcc phase varied substantially with the composition of the films and laser annealing conditions (Tmx and number of laser scan frames). One important point is that diffraction lines from the tetragonal Nb2A1 phase, which is the second phase typically obtained at lower temperatures were not observed. This is particularly important since the phase boundary between the A15 and Nb2A1 phase is less than 22 at. X A1 for temperatures below about l5OO0C [6.47-6.481. This result therefore shows that the cw laser processing succeeded in quenching the high temperature phase to room temperature without decomposition into lower temperature phases that accompanies the conventional thermal-quench method. In Fig. 6.33 Tc is plotted as a function of the number of laser scan frames. The laser annealing temperature is about i 4 O O ' C for all samples. Data are shown for two sets of %-A1 samples with different compositions (21.0 and 24.6 at. X Al). Tc is seen to increase linearly with the number of scan frames up to 10 frames, after which it remains essentially constant. X-ray diffraction analysis showed that this Tc enhancement was correlated with the growth of A15 phase relative to the bcc phase. Therefore, ten laser scans are probably sufficient to carry the

z 20 I

I

1

I

1 I I l l ,

I

I

I

I l l l l

Y

Nb-AP

i

[Laser Anneal Temperature (Tmax): 1400°C

G

(L

t

W

a

z

w

z

0

21.0 at 0 z -

O/o

AP

0 I

10

I 00

NUMBER OF LASER SCANS FIGURE 6 . 3 3 . Superconducting transition temperatures (Tc) as a function of high temperature laser scans at 14OO0C, for two sets of NbAl samples (of 2.10 and 24.6 at. % Al). 01981 Elsevier Science E'ublfshing Co.

6. METAL-SILICON REACTIONS AND SILICIDE FORMATION

389

reaction to completion. It is also shown in Fig. 6.33 that Tc is consistently higher for films with higher A1 composition. Figure 6 . 3 4 shows the dependence of Tc on laser annealing temperature for the same two sets of NbAl samples discussed above. Ten laser scans were employed at each temperature. It is seen that higher anilealing temperature reduces Tc, particular for the low A1 composition (21.0 at. X Al) samples. X-ray diffraction analysis showed that the samples annealed at high temperatures had less A15 phase (and more bcc phase) than the samples annealed at lower temperatures for both A1 compositions. These observations are qualitatively consistent with the equilibrium phase diagrams of Nb-A1 [ 6 . 3 8 - 6 . 3 9 1 , considering the uncertainty in temperature determinations. Briefly, the compositional range of the bcc solid solution is more Al-rich (> 21 at. X Al) at high temperatures ( > 18OO0C). Hence the increased formation of the bcc phase at high temperatures, particularly for the sample of 21.0 at. X Al, leads to the lower values of Tc.

z 0 I-

0 0

I500

2000

LASER ANNEAL TEMPERATURE Tmox (" C )

Superconducting transition temperature (Tc) as FIGURE 6 . 3 4 . a function of low temperature laser scans at 800°C.

T. SHIBATA ET AL.

Low Temperature Anneal From t h e d i s c u s s i o n above one can conclude t h a t t h e h i g h temperature s t a t e i n the phase diagram can be quenched t o room t e m p e r a t u r e by s c a n n i n g cw l a s e r annealing, without a corresponding phase change. However, atomic d i s o r d e r i s u s u a l l y observed f o l l o w i n g a r a p i d quench t o room temperature. Low temperature f u r n a c e a n n e a l i n g a t 700-800°C can of c o u r s e improve t h e o r d e r i n g , b u t such a n n e a l i n g u s u a l l y l e a d s t o decomposition of t h e quenched i n phase t o t h e unwanted low-temperature phases. A s d i s c u s s e d e a r l i e r , a s i n g l e l a s e r beam s c a n i s e q u i v a l e n t t o f u r n a c e a n n e a l i n g f o r very s h o r t times (- msec) w i t h very s h o r t h e a t i n g and c o o l i n g c y c l e s . By r e p e a t e d s c a n n i n g , one can t h e r e f o r e extend t h e e f f e c t i v e a n n e a l i n g t i m e w i t h o u t changing c o o l i n g and h e a t i n g r a t e s . This i s e x a c t l y what i s r e q u i r e d f o r anneali n g Nb-A1 samples. Using such a p r o c e d u r e , t h r e e d i f f e r e n t samp l e s were s u b j e c t e d t o 800°C l a s e r a n n e a l i n g f o r 100 s c a n frames.

Nb - At : Tc Enhancement by Low Temperature Loser Anneal

14Oo0C

800 O C (100 scans)

(I0 scans) LASER A N N E A L

FIGURE 6.35. Superconducting t r a n s i t i o n t e m p e r a t u r e s (Tc) as a f u n c t i o n of laser a n n e a l i n g t e m p e r a t u r e f o r t h e same two sets of Nb-A1 samples.

6.


39 1

Values of Tc before and after this process are shown in Fig. 6.35. A T, enhancement of 0.5-1.5'K is observed for all samples. There is no significant change in the relative amounts of the A15 to the bcc phase, and the second phase Nb2Alwas not observed in any of these cases. We tentatively attribute this Tc enhancement to the atomic ordering obtained by low-temperature laser annealing. 6.6.3

Experimental Results for Nb-Si

X-ray diffraction analysis showecT that the as-deposited films of Nb-Si (1000 a) were highly disordered or amorphous. Laser 50 cmlsec. The annealing was performed with a scan speed of crystal structures obtained after a single laser scan are summarized in Table 6.7.

-

TABLE 6.7. Summary of Various Crystalline Structures of NbSi Films Produced by Laser Annealing.

Composition of NbSi Film

19.9 at X Si

9 w

A15 (single phase)

Ti3P type

>>

11

20.5 at % Si

w

Ti3P type

A15

22.8 at % Si Tetragonal (CrgSi3 type)

+

Ti3P type (no A15)

+

Tetragonal (Cr5B3 type)

Figure 6.36 shows the x-ray diffraction pattern obtained from the 19.9 at. % Si sample after laser annealing at 9 W. Formation of the single phase A15 compound is clearly observed. T, measurements performed on this sample, both resistively and inductively, gave values of 4°K.

-

From Table 6.7 it is seen that the Ti3F structure was obtained at higher laser annealing temperatures (11 W). Table 6.7 also shows that the A15 structure becomes more difficult to form when the composition is close to stoichiometric Nb3Si. It should be noted that laser annealing can crystallize 19.9 at. X Si Nb-Si samples to the single phase A15 structure without melting, while melting is clearly involved in case of the splat growth of a stoichiometric A15 phase. An amorphous Nb-Si film

392

T. SHIBATA ET AL.

FIGURE 6 . 3 6 . Read-Camera diffraction pattern for a Nb-Si sample (19.9 at. % S i ) (a) as deposited ( b ) after laser anneal.

consisting of layers in which the composition changes gradually from 19.9 to 25 at. % would be interesting for laser annealing, since a single laser scan could possibly crystallize the 19.9 at. % Si layer into the A15 phase, which might then propagate into the layer of 25 at. % Si.

6.

METAL.-SILICON REACTIONS AND SILICIDE FORMATION

393

REFERENCES

6.1 6.2 6.3 6.4

6.5 6.6 6.7 6.8

6.9 6.10 6.11 6.12 6.13 6.14 6.15 6.16

6.17

6.18

6.19

6.20

Crowder, B. L., and Z i r i n s k y , S . , IEEE Trans. on Elec. Dev. ED-26, 369 (1979). Shah, P. L., IEEE Trans. on Elec. Dev. ED-26, 631 (1979). Yu, H. N., Reisman, A., Osburn, C. M., and Critchlow, D. L., (1979). IEEE Trans. on Elec. Dev. ED-26, 318. "Thin Films I n t e r d i f f u s i o n and Reactions" (J. M. Poate, K. N. Tu and J . W. Mayer, eds.) Wiley I n t e r s c i e n c e , New York (1978). Poate, J. M., Leamy, H. J., Sheng, T. T., and Celler, G. K., Appl. Phys. L e t t . 33, 918 (1978). Liau, Z. L., Tsau, B. Y., andMayer, J. W., Appl. Phys. L e t t . 34, 221 (1979). S h i b a t a , T., Gibbons, J. F., and Sigmon, T. W., Appl. Phys. L e t t . 36, 566 (1980). Sigmon, T. W., R e g o l i n i , R. L. and Gibbons, J. F., "Symposium on Laser and E l e c t r o n Beam P r o c e s s i n g of E l e c t r o n i c Mater i a l s , " Los Angeles, p. 350 (1979). Tsau, B. Y., Liau, Z. L., and Mayer, J. W., Appl. Phys. L e t t . 34, 1968 (1979). Tsau, B. Y . , L i a u , Z. L., and Mayer, J. W., Phys. L e t t . 71A, 270 (1979). Kanayama, T., Tanoue, H., and Tsurushima, T., Appl. Phys. L e t t . 35, 222 (1979). Tsau, B. Y., Lau, S . S . , and Mayer, J. W., Appl. Phys. L e t t . 35, 225 (1979). "Laser and E l e c t r o n Beam P r o c e s s i n g of Semiconductor S t r u c t u r e s " (J. W. Mayer and J. M. Poate, eds.) t o be published. von Allmen, M., and W i t t m e r , M., Appl. Phys. L e t t . 34, 68 (1979). W i t t m e r , M., and von Allmen, M., J. Appl. Phys. 50, 4786 (1979). van Gurp, G. J., Eggermont, G. E. J., Tamminga, Y., S t a c y , W. T., and G i j s b e r s , J. R. M., Appl. Phys. L e t t . 35, 273 (1979). Lau, S. S . , Mayer, J. W., Tsau, B. Y., and von Allmen, M., i n "Laser and E l e c t r o n B e a m P r o c e s s i n g of Materials" (C. W. White and P. S . Peercy, eds.) p. 511, Academic P r e s s , New York (1980). S h i b a t a , T., Sigmon, T. W., and Gibbons, J. F. i n "Proceedi n g s of t h e Symposium on Thin Film I n t e r f a c e s and I n t e r a c t i o n s " (J. E. E. B a g l i n and J. M. P o a t e , eds.) E l e c t r o chem. Soc., p. 458, P r i n c e t o n (1980). S h i b a t a , T., Sigmon, T. W., and Gibbons, J. F., i n "Laser and E l e c t r o n Beam P r o c e s s i n g of Materials" (C. W. White and P. S. Peercy, eds.) p. 530, Academic Press, NewYork (1980). L i e t o i l a , A., Ph.D. t h e s i s , S t a n f o r d U n i v e r s i t y , Dept. of Applied P h y s i c s (1981). See a l s o Sec. 2.3.6 of t h i s volume.

394 6.21 6.22 6.23 6.24 6.25 6.26

6.27 6.28 6.29

T. SHIBATA ET AL.

Chu, W. K., Mayer, J. W., and N i c o l e t , M. A . , " B a c k s c a t t e r i n g Spectrometry", Academic Press, New York (1978). A z a r o f f , L. V . , "Elements of X-Ray C r y s t a l l o g r a p h y , " 383, M c G r a w - H i l l , New York (1968). Feder, R., and B e r r y , B. S., J. Appl. C r y s t . 3 , 372 (1970). Read, M. H . , and H a n s l e r , D. H., Thin S o l i d Films 1 0 , 123 (1972). American S o c i e t y f o r T r e s t i n g and M a t e r i a l s , Powder D i f f r a c t i o n Data F i l e s , P h i l a d e l p h i a , PA. Wagner, R. J . , Lau, S. S., and Mayer, J. W., "Proceedings of t h e Symposium on Thin Film Phenomena - I n t e r f a c e s and I n t e r a c t i o n s " , 59 (1978). " C o n s t i t u t i o n of Binary A l l o y s , " M c G r a w - H i l l , Hansen, M., New York, 1016 (1959). Sigurd, D., and S c o t t , R. E., S o l i d S t a t e Bower, R. W., E l e c t r o n . 1 6 , 1461 (1973). Cheung, N., Lau, S. S., N i c o l e t , M. A., Mayer, J . W., and i n "Proceedings of t h e Symposium on Thin Film Sheng, T. T. I n t e r f a c e s and I n t e r a c t i o n s " (J. E. E. B a g l i n and J. M. P o a t e , e d s . ) 80-2, p. 494, Electrochem. SOC., P r i n c e t o n

(1980). 6.30 6.31 6.32

6.33

6.34 6.35 6.36 6.37 6.38 6.39 6.40 6.41 6.42

Hutchins, G . A. and Shepola, A . , Thin S o l i d Films 18, 343 (1973). Anderson, R . , B a g l i n , J . , Dempsey, J . , Hammer, W., d ' H e u r l e , F., and P e t e r s o n , S., Appl. Phys. L e t t . 35, 285 (1979). Tsaur, B. Y., Mayer, J . W . , and Tu, K. W., i n "Proceedings of t h e Symposium on Thin Film I n t e r f a c e s and I n t e r a c t i o n s " ( J . E . E. B a g l i n and J. M. P o a t e , eds.) 80-2, p. 264, Electrochem. SOC., P r i n c e t o n (1980). Johnson, N., Sigmon, T., Nemanich, R., Mayer, D., and Lau, S. S ., i n "Proceedings of M a t e r i a l s Research S o c i e t y Symposium on Laser and E l e c t r o n Beam P r o c e s s i n g " , Boston (1980). Inoue, T., and Koike, K., Appl. Phys. L e t t . 33, 826 (1978). Mochizuki, T., S h i b a t a , K., Inoue, T., and Ohuchi, K., Jpn. J . Appl. Phys. Suppl. 17-1, 37 (1977). Majni, G., Nava, T., and O t t a v i a n i , G., i n "Proceedings of M a t e r i a l s Research S o c i e t y Symposium on Thin Film I n t e r f a c e s and I n t e r a c t i o n s , " Boston (1979). Z i r i n s k y , S., Hammer, W., d ' H e u r l e , F., and B a g l i n , J . , Appl. Phys. L e t t . 33, 76 (1968). Mohammadi, F. , Saraswat, K. C., and Meindl, J . D., Appl. Phys. L e t t . 35, 529 (1979). S h i b a t a , T., Wakita, A., Sigmon, T. W., and Gibbons, J. F . , Appl. Phys. L e t t . 40, 77 (1982). Lindenberger, W. S., and Muraka, S. P., F r a s e r , D. B., Sinha, A. K., J. Appl. Phys. 51, 3241 (1980). Razouk, R., Thomas, M., and P r e s s a c c o , S . , J . Appl. Phys. 53, 5342 (1982). Appl. Phys. L e t t . 40, 175 B a r t u r , M., and N i c o l e t , M-A., (1982).

6. 6.43 6.44 6.45 6.46 6.47


395

Chen, J., Houng, M-P., Hsiung, S-K., and Liu, Y-C., Appl. Phys. Lett. 37, 824 ( 1 9 8 0 ) . Schwartz, G.P., Sunder, W. A., Griffiths, J. E. and Gualtieri, G. J., Thin Solid Films, 9 4 , 205-212 ( 1 9 8 2 ) . Beyers, R., submitted for publication. Wakita, A. S., Sigmon, T. W., and Gibbons, J. F., submitted for publication. Kwo, J., Hammond, R. H., Geballe, T. H., J. Appl. Phys. 5 1 ( 3 ) , 1726 (1980).

6.48 6.49 6.50

6.51 6.52

Lundin, C. E., and Yamamoto, A. S., Trans. Am. Inst. Mech. Eng. 2 3 6 , 863 ( 1 9 6 6 ) . Svechinkov, V. N., Pan, V. M., and Latzshev, V. I., Metallafigik 3 2 , 28 ( 1 9 7 0 ) . Pan, V. M., Pet'kov, V. V., Kulik, 0. G., in "Physics and Metallurgy of Superconductors", Moscow, USSR, 1965-1966, E. H. Savitskii and V. V. Baron, eds. (Consultants Bureau, New York, 1 9 7 0 ) . Feldman, R. D., Hammond, R. H., and Geballe, T. H., Appl. Phys. Lett. 3 5 , 818 ( 1 9 7 9 ) , and references therein. Gat, A . and Gibbons, J. F., Appl. Phys. Lett. 3 2 ( 3 ) , 142 February 1 , 1 9 7 8 ) .


CHAPTER 7

CW Beam Processing of Gallium Arsenide Yves I . Nissim" and James F. Gibbons STANFORD ELECTRONICS LABORATORIES STANFORD UNIVERSITY STANFORD. CALIFORNIA

7 . 1 INTROD~JCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 . 2 cw LASERANNEALING OF ION IMPLANTED GaAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3 cw LASERPROCESSING OF THIN FILMS DEPOSITED ON GaAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4 RAPID THERMAL PROCESSING ........................ 7.5 NOVELAPPLICATIONS OF RAPID THERMAL PROCESSING IN GaAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . REFERENCES

7.1

.......................................

397 398 41 1 430

436 443

INTRODUCTION

As is suggested in the previous chapters, the use of cw beams to process silicon can provide significant advantages in both material and devices properties. Compound semiconductors, and in particular GaAs, have from the beginning been potential candidates for the application of the beam annealing technology. In particular the conventional thermal annealing of GaAs often gives unsatisfactory results in both activation of dopants and removal of damage induced by ion implantation. Furthermore, encapsulation or controlled arsenic overpressure are required to prevent the substrate from surface decomposition (As evaporation) at elevated temperatures. A short, localized heat treatment under beam irradiation can be expected to suppress decomposition and thereby possibly improve the electrical properties of the annealed layers. However, the results to date on cw beam annealing of implanted layers in G a b have not shown substantial improvements as compared to thermal annealing. Major difficulties arising from the fragility and dissociation of the material under laser irradiation are responsible for the limited success achieved in early attempts to anneal implanted GaAs with a cw laser system. However, the temperature calculations presented in Chapter 2 demonstrate that a laser beam with an elliptical cross section produces a more gradually distributed temperature gradient at the surface of a semiconductor than a beam with a circular cross section. The combination of 'Present address: CNET, 92220 Bagneaux, France. SEMICONDUCTORS AND SEMIMETALS, VOL. 17

397


398

YVES I. NISSIM AND JAMES F. GIBBONS

the elliptical geometry and an appropriate set of annealing parameters in different environments does result in efficient electrical activation (Section 7.2). Multiply scanned electron beam can also be used as an alternative to laser, giving results that also will be presented in the same section. The mechanism that could produce annealing of ion implanted amorphous layers in GaAs at low temperatures is solid phase epitaxial regrowth. While this process is well characterized in silicon, it is much less well understood in GaAs. Recently, however, solid phase epitaxial recrystallization in GaAs has been observed at different laboratories using a well controlled, precisely defined ion implantation schedule. These results are presented in Section 7.2 with their limitations. Cw laser processing of GaAs has made a more signiflcant contribution as a means of reacting deposited thin films with the underlying substrate. Laser alloying of thin metallic film to n-GaAs has resulted in alloyed ohmic contacts that are significantly better than conventional thermally alloyed contacts. It has been also demonstrated that a scanned cw laser can assist the diffusion and activation of tin in semi-insulating GaAs from + an Sn02/Si02 thin film source. This process produces thin n layers that display excellent ohmic behavior directly after metal evaporation, (i.e., without an alloying step). This nonalloyed ohmic contact technology is well suited to device technology since the required laser treatment operates at incident power levels that are well below the laser induced damage threshold of the GaAs substrate. This subject is discussed in Section 7 . 3 . More recently, rapid thermal processing has been used to anneal implanted layers in GaAs, InP and HgCdTe; and to provide a technique for studying impurity diffusion. A novel process employing controlled evaporation of plated Zn doping source has also been used to produce significant improvements in Schottky barrier heights. We will discuss these topics in Sections 7.47 ‘6 below. 7.2

CW LASER ANNEALING OF ION IMPLANTED GaAs

Among the different doping techniques for GaAs, ion implantation is used extensively to control both the doping levels and junction depth. A reproducible annealing sequence that is capable o f activating the implanted dopants (on substitutional sites) and removing the ion implantation induced damage is required. In what follows we discuss attempts to anneal ion implanted GaAs with a scanning beam. 7.2.1

The Elliptical Beam Approach

The temperature induced in a semiconductor by a scanning elliptical beam has been calculated in Chapter 2. As can be

7.

CW BEAM PROCESSING OF GALLIUM ARSENIDE

399

s e e n from Fig. 2.7, a laser beam w i t h an e l l i p t i c a l c r o s s s e c t i o n narrow i n t h e d i r e c t i o n of t h e s c a n and l a r g e i n t h e d i r e c t i o n p e r p e n d i c u l a r t o t h e s c a n , produces a more uniformly d i s t r i b u t e d temperature g r a d i e n t at t h e s u r f a c e of t h e i r r a d i a t e d semicond u c t o r . This p r o p e r t y is of i n t e r e s t f o r processing b r i t t l e materials such as G a A s because i t reduces t h e thermal shock t h a t a c i r c u l a r beam would c r e a t e . This beam geometry a l s o p e r m i t s a l a r g e area t o be annealed i n a s i n g l e scan, w i t h s u b s t a n t i a l l y less power than a c i r c u l a r beam of l a r g e r a d i u s would r e q u i r e . The s i z e of t h e annealed a r e a i s n e v e r t h e l e s s l i m i t e d by t h e amount This f i g u r e of power a v a i l a b l e , as can be s e e n i n F i g . 7.1. compares t h e maximum temperature o b t a i n e d i n G a A s f o r d i f f e r e n t value of t h e a s p e c t r a t i o 6 as a f u n c t i o n of t h e absorbed power p e r u n i t r a d i u s . For i n s t a n c e , the maximum temperature of 1000°C i s o b t a i n e d i n GaAs f o r f3 = wy/wx = 20 w i t h seven t i m e s t h e = 1. Since t h e power r e q u i r e d i n t h e case where fi = wx/wx thermal c o n d u c t i v i t y of G a A s i s small compared t o t h e thermal c o n d u c t i v i t y of s i l i c o n , t h i s technique w i l l be more a p p l i c a b l e i n GaAs u s i n g a cw laser of moderate output power (- 10 w a t t s ) . S i g n i f i c a n t e l e c t r i c a l a c t i v i t y a f t e r cw laser a n n e a l i n g of i o n implanted l a y e r s i n GaAs was f i r s t r e p o r t e d by Fan e t a l . , I7.21. A cw Nd:YAG laser ( o p e r a t i n g a t 1.06 vm) and a set of unconventional a n n e a l i n g parameters as compared t o t h o s e u t i l i z e d f o r s i l i c o n were used. The thermal shock induced by t h e laser w a s reduced by a combination of a h i g h s u b s t r a t e temperature (58OOC) and a laser beam w i t h an e l l i p t i c a l c r o s s s e c t i o n . Scanning speeds between one and two o r d e r s of magnitude lower t h a n t h o s e r e p o r t e d f o r s i l i c o n were used. T y p i c a l l y , speeds between 1 cm/ sec and 0.5 m/sec r e s u l t e d i n reasonable electrical a c t i v a t i o n . The s u b s t r a t e w a s i n a forming gas environment d u r i n g i r r a d i a t i o n . The samples s t u d i e d were implanted w i t h lxlOI4 Se+/cm2 at an TBACK= 550'C

350'C

15OoC

25'C

GALLIUM ARSENIDE

1

2 3 (1-RIP Pa = wx ( k w l c m )

4

5

FIGURE 7 .l. The " t r u e " maximum temperature induced i n Ga As f o r d i f f e r e n t v a l u e s of t h e a s p e c t r a t i o f3 = 1 and B = 20).

400


energy of 400 keV. The implant temperature was 30OOC i n t h e c a s e of h i g h e s t a c t i v i t y . Values of 272 f o r e l e c t r i c a l a c t i v i t y w i t h a Hall m o b i l i t y of 2100 cm2/V.sec f o r a s i n g l e s c a n and 59% w i t h a m o b i l i t y of 1400 cm2/V.sec f o r t e n c o n s e c u t i v e s c a n s were t h e b e s t r e s u l t s r e p o r t e d . It was observed t h a t s u r f a c e d e t e r i o r a t i o n ( s l i p l i n e s ) occurred when t h e i n c i d e n t power and/or t h e The formation of s l i p l i n e s scanning speed were i n c r e a s e d . i m p l i e s a l i m i t on t h e maximum temperature a t t h e s u r f a c e of t h e i r r a d i a t e d sample. The r e s u l t s p r e s e n t e d i n Ref. 7.2 demonstrate t h a t t h e h i g h e s t e l e c t r i c a l a c t i v i t y i s o b t a i n e d when s l i p l i n e s j u s t s t a r t t o appear; however, t h e s l i p l i n e s a l s o l e a d t o a It i s important t o n o t e t h a t encapsulalower s u r f a c e m o b i l i t y . t i o n of t h e implanted samples p r i o r t o laser a n n e a l i n g enhances s u r f a c e d e t e r i o r a t i o n , probably due t o i n c r e a s e d s t r e s s a t t h e semiconductor-encapsulant i n t e r f a c e . These r e s u l t s , o b t a i n e d f o r medium implanted d o s e s , become poorer f o r h i g h e r doses; and t h e technique w a s found t o be u n s u c c e s s f u l f o r l o w implant doses [ 7.31 ( t y p i c a l l y 1 O1 i o n s /em2). I f t h e i n c i d e n t power i s i n c r e a s e d above t h e t h r e s h o l d f o r s l i p l i n e formation, t h e GaAs s u b s t r a t e decomposes due t o As e v a p o r a t i o n . When t h e s u b s t r a t e i s h e l d i n a forming gas environment as d e s c r i b e d i n t h e previous paragraph, t h e s u b s t r a t e decomposition i s observed by an excess of G a i n t h e form of a grey powder a t t h e s u r f a c e of t h e i r r a d i a t e d sample. However i f t h e s u b s t r a t e i s h e l d i n a l a b o r a t o r y environment o r an oxygen environment, an oxide i s formed when decomposition occurs f7.41. This oxide has been i d e n t i f i e d , u s i n g X-ray d i f f r a c t i o n , a s t h e g a l l i u m oxide B-Ga2O3. S p u t t e r Auger p r o f i l i n g has been used t o study t h e composition of t h i s oxide. The r e s u l t i n g p r o f i l e i s shown i n Fig. 7.2. It can be s e e n t h a t t h e g a l l i u m oxide i s completely free of A s ( w i t h i n t h e s e n s i t i v i t y of t h e measurements) and i s about 1200 A t h i c k , corresponding t o an observed blue c o l o r . The i n t e r f a c e w i t h t h e u n d e r l y i n g s u b s t r a t e i s r e l a t i v e l y In t h i s experiment an argon i o n l a s e r w a s s h a r p (2100 A). focused by a c y l i n d r i c a l l e n s onto a GaAs s u b s t r a t e heated t o 550°C. A scanning speed of 0.5 mm/sec was used. The r e s u l t i n g temperature p r o f i l e s and t h e d i f f e r e n t working a r e a s a r e shown i n The e l e c t r i c a l a c t i v i t y obtained a f t e r oxide growth Fig. 7.3. on low dose implanted l a y e r s ( 1 ~ 1 0%+/em2) ~ ~ i n semii n s u l a t i n g GaAs w a s s t u d i e d . The a n n e a l i n g a r e a i s c h a r a c t e r i z e d by a low e l e c t r i c a l a c t i v i t y (10 t o 13%) and h i g h m o b i l i t y (2500 cm2/V.sec) i n t h e lower power regime and h i g h e r a c t i v a t i o n (30%) w i t h reduced m o b i l i t y ( = l o 0 cm2/V.sec) accompanied by t h e format i o n of s l i p l i n e s i n t h e h i g h e r power regime. When t h e oxide i s formed, h i g h e r temperatures can be induced, r e s u l t i n g i n v e r y h i g h dopant a c t i v a t i o n . The e l e c t r i c a l a c t i v i t y a s judged by s h e e t e l e c t r i c a l measurements was found t o be 100% when t h e i m p u r i t y Hall m o b i l i t i e s i n c o n t e n t i n t h e oxide l a y e r i s s u b t r a c t e d . 0 ~ ~ were r e p o r t e d t h e range of 1800 cm2/V.sec ( f o r 1 ~ 1 e-/cm2)

7.


401

SPUTTERING RATE -i lOO%Mll

X

I

'**.* X

X

AS X

00

oooooooooo

o

o

ooooooo

0

Po x

x

I

I

10

20

I

30

SPUTTERING TIME (MIN.)

FIGURE 7.2. oxide. 01981 Elsevier

Sputtering Auger profile of the laser grown Science Publishing Co.

when the laser parameters were chosen to obtain a uniform oxide layer. The overlap between the annealing and oxide growth areas shown in Fig. 7.3 indicates that the oxide can grow at lower powers if the growth is initiated by an enhanced absorption (edge or scratch on the wafer). While scanning over an already grown layer, the oxide growth will continue. The decomposition of the surface of the GaAs and the formation of 8-Ga2O3 acts as a stress release and thus prevents the slip lines from forming. An alternate method to obtain a uniform heating at the surface of a semiconductor has been reported using a multiply-scanned electron beam 17.51. The sample is thermally isolated and exposed to multiple randomly interlaced raster scans of the beam. This results in an uniform exposure over the scanned area. Exposure times on the order of 5 sec have resulted in high induced temperatures and successful annealing of low dose Si implants (6x1012/ cm2) in GaAs [7.6]. The resulting annealed profile is presented in Fig. 7.4 and compared to an as implanted LSS profile. Activation of 50 to 60% of the implanted dopant were reported with a mobility of 3800 cm2/V.sec. This value of mobility for an active dopant concentration of 1017/cm3 is the highest obtained by beam annealing techniques so far and comparable to the best thermal annealing.

YVES I . NISSIM AND JAMES F. GIBBONS

402

1200

1000

-

G 800 0

w (L

$

600

(r

f

&-

400

2 00

0

FIGURE 7.3. Temperature induced in GaAs by a laser focused on an elliptical spot of aspect ratio B = 20. The maximum temperature at the center, and the temperature at the edge of the beam are plotted. 01981 Elsevier Science Publishing Co. 7.2.2

Solid Phase Epitaxy in GaAs:

Low Temperature Annealing

Solid phase epitaxial recrystallization of ion implanted amorphous layers has been observed to occur in silicon at temperatures on the order of 500°C i7.71. In this process, the crystallinity of the material is recovered and most of the dopant impurities are incorporated on substitutional sites. As presented in Chapter 3 , cw laser annealing of amorphous layers in S i proceed via the same mechanism. The difficulties presented in the previous section to obtain good crystal recovery and electrical activation following ion implantation in GaAs would be overcome if a complete epitaxial recrystallization process in solid phase could be induced by laser irradiation. Unfortunately this process is more complex in GaAs and has only recently been demonstrated. We begin our discussion of this topic by reviewing furnace annealing studies.

7.


403

/-\

I

/

0.2

' ,'

04

0.6 dpplh

,

0.8

pm

FIGURE 7.4. Carrier c o n c e n t r a t i o n p r o f i l e of e l e c t r o n beam annealed G a A s implanted w i t h s i l i c o n ( 6 ~ 1 0 1 ~ a t 360 keV).

7 .2.2.1

Furnace Annealing

Gamo e t a l . r7.81 r e p o r t e d experiments performed on Zn implanted G a A s s u b s t r a t e s f o r a range of annealing temperatures between 200 and 600°C. Incomplete o r h i g h l y d e f e c t i v e c r y s t a l recbvery was observed f o r temperatures up t o 500°C and complete e p i t a x i a l regrowth was obtained a t 600°C. Local s t o i c h i o m e t r i c imbalance produced during t h e i m p l a n t a t i o n c y c l e w a s proposed as a l i m i t i n g f a c t o r . More r e c e n t l y , Williams and Austin [7.91 have shown t h a t c r y s t a l l i n i t y recovery could occur a t temperatures a s low as 180°C. The i s o c h r o n a l a n n e a l i n g of an amorphous l a y e r o b t a i n e d by a n implant of 5 ~ 1 0 Ar+/cm2 1 ~ at 100 keV with t h e subs t r a t e h e l d a t l i q u i d n i t r o g e n d u r i n g t h e implant is shown i n F i g . 7.5a. The i m p l a n t a t i o n s c h e d u l e i s j u s t s u f f i c i e n t t o create an amorphous l a y e r continuous t o t h e s u r f a c e . The e p i t a x i a l growth i s s t a r t e d a t 130°C and i s almost complete a t 180°C. I f t h e i n c i d e n t dose is i n c r e a s e d t o 2x1014 Ar+/cm2 i t is then necessary t o reach 600°C before t h e e p i t a x y is complete as s e e n i n F i g . 7 .5b. A s t h e implant dose i n c r e a s e s , t h e q u a l i t y of t h e

404


a) 5 X 1 0 1 3 A r h (1OO)GaAs

t-*

300A

280

300

320 CHANNEL NUMBER

340

FIGURE 7.5. Isochronal (10 min) annealing of (a) 5x10l3 The dotted Ar+/cm2 and (b) 2x1014 Ar+/cm2 implanted in GaAs. curve in (a) is for unimplanted GaAs. The GaAs was encapsulated with AP metal for the 600°C anneal in (b).

regrown material degrades and the regrowth rate decreases at constant annealing temperature. This illustrates the strong dose dependence of this mechanism and precludes any meaningful regrowth kinetic calculations. Further studies by Williams and Harrison [7.10] have shown that the furnace annealing of ion implanted amorphous layers in GaAs at doses well above amorphization occurs in stages. This is illustrated in Fig. 7.6 where the normalized disorder as measured by channeling for 8x1013 Ar+/ cm2 implant at 100 keV in GaAs is plotted as a function of annealing temperature. It can be seen that the crystalline to amorphous transition occurs at very low temperatures (125-230'C) but 50% disorder is still observed after the transition. Only in a second stage (400-600'C) does the annealing of extended defects occur and the crystalline quality is recovered. TM analysis by Kular et al. 17.111 has shown that the annealed layers are a highly twinned single crystal in the first stage of annealing (low temperature)

7.


405

8~10~~Arcm-~-(100)GaAs 100

zb

.!? D

-7

?

annealing of extended defects

50-

s amorphous-t~crystalline transition (epitaxy)

I

0

I

100

I

I

I

I

500 Anneal Temperature (OC)

200

300

400

\3

600

FIGURE 7.6. Normalized d i s o r d e r as measured by channeling (100 keV) implanted i n GaAs p l o t t e d as a f o r 8 ~ 1 0 Ar+/cm2 1 ~ f u n c t i o n of furnace a n n e a l temperature (15 min. a n n e a l s ) i n d i c a t i n g two a n n e a l i n g s t a g e s , O1981 Elsevier Science Publishing Co. and a s i n g l e c r y s t a l w i t h d i s l o c a t i o n loops a t h i g h e r temperature. An even h i g h e r temperature r e s u l t s i n t h e r e d u c t i o n of t h e dens i t y of loops. The p r o c e s s of s o l i d phase e p i t a x i a l regrowth i n G a A s has In contrast been e x t e n s i v e l y s t u d i e d by Grimaldi e t a l . [7.12]. t o t h e work r e p o r t e d above, t h e regrowth has been observed t o depend s t r o n g l y on t h e i n i t i a l t h i c k n e s s of t h e amorphous l a y e r . Complete recovery a f t e r f u r n a c e a n n e a l i n g was obtained f o r t h i c k A model i n which t h e growing c r y s t a l f r o n t n e s s e s < 400 8 . accumulates l o c a l d e f e c t s has been proposed. The t h i c k n e s s of 400 A would t h e n be t h e maximum d i s t a n c e t h a t t h e i n t e r f a c e can move b e f o r e t h e d e f e c t s begin t o p r e c i p i t a t e i n c l u s t e r s . These c l u s t e r s i n t u r n could then be removed only i f t h e a n n e a l i n g t e m p e r t u r e i s e l e v a t e d t o a range between 700 and 8OOOC. Below t h i s c r i t i c a l t h i c k n e s s t h e i n c i d e n t dose, type of i n c i d e n t atoms and s u b s t r a t e temperature during i m p l a n t a t i o n become less c r i t i c a l f o r t h e regrowth. I n t h e l i g h t of t h e s e r e s u l t s , a number of d i f f e r e n t e x p e r i mental choices were made by N i s s i m e t a l . l7.131 t o s t u d y t h i s mechanism. The work r e p o r t e d i n Ref. 7.9 shows t h a t a v e r y f a s t

406


regrowth rate is observed along the direction of GaAs. In an attempt to control the process, the (511) orientation in GaAs was chosen (it has been observed in silicon [7.14] that the (511) orientation had a slower rate than the direction). The implantations were carried out with the substrate held either at ice water (for the highest doses) or liquid nitrogen temperature, to obtain a sharp crystalline amorphous interface and reduce the necessary incident dose of amorphization. The contribution of dopant atoms to the regrowth rate was minimized by the implantation of As' ions. Annealing was carried out at a range of temperatures between 400 and 500'C. When high doses are used to amorphize the GaAs (typically above 1015 As+/cm2), epitaxial growth from the crystalline amorphous interface is observed, but it terminates before reaching the surface (40% of the initial amorphous layer consistently recrystallized). The recrystallized region has a high concentration of defects as revealed by a high backscattering yield. When the annealing temperature reaches 5OO0C, a slow annealing of defects accompanies the regrowth. This is characterized by a reduction in the backscattering yield with an increase in the annealing time without evolution of the damaged layer. These results have been supported by TEM analysis as shown in Fig. 7.7. After annealing, polycrystallites of GaAs are detected within the implanted region as shown in Fig. 7.7(b). Careful examination of the diffraction pattern inset indicates the possible existence of residual dislocation line structure beneath the polycrystalline layer. Using controlled chemical stripping, 750 A of material was etched from the surface of the sample and specimens were again prepared for TEM examination. Figure 7.7(c) shows that all the polycrystalline layers have been removed, and a complex array of dislocation nesting is observed. The diffraction pattern indicates the presence of an imperfectly regrown single crystal layer. The formation of this polycrystalline region is interpreted as a competitive mechanism to the regrowth that proceeds from the surface down to the crystallineamorphous interface and meets the recrystallizing front. In order to understand the large defect concentration in the regrown material, damage distribution and net stoichiometric imbalance resulting from the implantation were calculated using a Boltzmann transport approach [7.15,7.16]. The result of this calculation is shown in Fig. 7.8 for an implant of 1 . 1 ~ 1 0As+/cm2 ~~ at 145 keV. The measured thickness of the amorphous layer and recrystallized layer after annealing are indicated in this plot. A minimum damage density of 2 . 8 ~ 1 0keV/cm3 ~~ is required to obtain an amorphous layer, and the damage density at which the regrowth stopped is 9x1021 keV/cm3. This upper limit is not well established since the recrystallized region was observed to be highly damaged. A series of As+ implants (3x1Ol4 As+/cm2 at 145 keV followed by lx1014 As+/cm2 at 80 keV), whose doses and energies were selected to achieve a damage density of ". 3x1021 keV/cm3, and to ensure that this value was maintained in the near surface region, led

FIGURE 7.7. Representative bright field transmission electron micrograph obtained from samples implanted with 180 keV As ions to a dose of 4.6~1015 cm-2 in GaAs. (a) As implanted; (b) implanted and annealed at 475'C for 10 min; (c) implanted, annealed and 750 A stripped from the surface. Insets show selected area diffraction patterns from regions. to complete recrystallization. This is illustrated in the channeling spectra of Fig. 7.9. The nature of the microstructure in the as-implanted and annealed samples in this experiment was studied by TEM (Fig. 7.10). After implantation [Fig. 7.lO(a)] an'absence of microstructure and the presence of an amorphous layer is observed. After annealing complete recrystallization is obtained and well defined single crystal diffraction patterns are detected. Within the implanted region, residual damage is observed in the form of dislocation loops of 150-200 A average diameter [Fig. 7.10(b)] at a concentration of = lOl1/cm2. It should be noted that the same implant schedule performed in GaAs led to complete recrystallization with some discontinuous areas at the surface of amorphous material. From the results obtained in these different experiments, sufficient conditions to obtain solid phase epitaxial regrowth of ion implanted layers in GaAs can be summarized in the following way:

YVES I. NlSSIM AND JAMES F. GIBBONS

408

4MORPYOUS AFTER INNEAL 4MORPYOUS

- ~

4

--I

1

~ 1 1 . 1 0 ' ' i ~ ~A~* +

--

I145XIV

L5 IMPLANTED PROFILE OMETRIC bNCE ING FROM 11

10'"

-

ns-

0

19

10

2

P W

U

0

u

I8

10

-

TOTAL STOCHIOMETRIC IMBALANCE 14LING INTO ACCOUNT A S * I IMPLANTEQ SPECIE I

\ \\.

17

10

0

1000

500 DEPTH

1500

[i)

FIGURE 7.8. Net stoichiometric imbalance and damage distribution in GaAs produced by an implantation of 1 . 1 ~ 1 0As~'/c~ m2 at 145 keV (calculation).

The damage density induced by the implantation must be below a critical threshold but above the level of amorphization. Such windows can eventually be found for different incident ions by holding the substrate at liquid nitrogen temperature during the implantation. Polycrystalline growth from the surface must be suppressed with a continuous damage density through the implanted layer. An alternative to these conditions can be obtained by keeping the amorphous layer thickness below 400 A.

7.2.2.2

CW Laser Annealing

As shown in the previous section, the mechanism of solid phase epitaxial growth in GaAs induced by furnace annealing is not fully understood. For this reason very little work on CW

7.


409

-It-

3 ~ 1 Aslcrn' 0 ~ ~AT 145 heV +1 *lo" Aslcm2 AT 80 keV

ANNEALED -e& 10 MIN AT 475°C

00

ALIGNED

SINGLE

41

I

1.25

I

1.5

1.75

ENERGY (MeV )

FIGURE 7.9. Channeling spectra illustrating the complete epitaxial regrowth at 475'C for 3x1Ol4 As+/cm2, 145 keV plus lx1014 As+/cm2, 80 keV implanted into GaAs. beam annealing has been reported. Since the regrowth can be initiated at low temperature, backsurface substrate heating cannot be used here. Therefore, formation of slip lines will occur at low temperatures, preventing sufficient power increase to achieve high temperatures. Williams et al. [7.17] have compared the behavior of the regrowth when induced by a furnace or a beam irradiation. A cw argon laser was used and annealing parameters were set to be just below the threshold of slip line formation with a dwell time of 15 msec. At doses below amorphization the cw laser induces a complete recrystallization similar to the 180'C furnace annealing. Above the amorphization level, only partial regrowth is observed. These results indicate that the two types of annealing give essentially similar results at low temperatures. The annealing of extended defects observed at higher temperature (600-900'C) furnace annealing is prevented by slip line formation in the case of laser annealing.


410

FIGURE 7.10. Representative bright field transmission electron micrographs obtained from Sam les implanted with 3x1OI4 As+/ em2 at 145 keV plus lxlOI4 As+/cm at 80 keV. (a) as implanted; ( b ) implanted and annealed at 475°C. Selected area diffraction patterns are shown in insets.

P

7.2.2.3

Electrical Activation

During the recrystallization of the lattice, one would expect to be able to incorporate dopant atoms onto lattice sites that

would result in electrical activity. Unfortunately, different laboratories have reported unsuccessful dopant activation during the epitaxial growth for ranges of temperatures between 400-600°C. The best implantation schedules as determined in Section 7.2.2.1 were carried out with the dopants either introduced at low dose within the amorphous layer or with the dopants being the amorphizing species, but in both cases very little or no dopant activation was observed after the low temperature anneal. A further anneal at a higher temperature with an encapsulated sample is required

7.


411

to obtain reasonable electrical activity. Compensation of the dopants after low temperature anneals can be speculated to come from the remaining dislocation loops or the redistribution of background impurities during epitaxial regrowth. 7.3

CW LASER PROCESSING OF THIN FILMS DEPOSITED ON GaAs

In this section we consider the doping of GaAs that may be obtained when a cw beam is used to react a thin deposited film with the underlying substrate. Both alloyed and nonalloyed ohmic contacts to n.GaAs have been obtained from irradiated thin films deposited on GaAs. The performance and reliability of a number of GaAs microwave, logic and optoelectronic devlces are determined in large part by the properties of their ohmic contacts. In microwave MESFET's, for example, parasitic source resistance due to the ohmic contact is a major contributor to noise in state-of-the-art devices. The electrical performance of a typical FET (1 pm gate-length) degrades significa tly fo values of specific contact resistance P, above the midlo-' a-cm range. As device geometries become smaller, the demands will be even more stringent. The two different approaches reported here can result in the formation of low specific contact resistance ohmic contacts well suited to the small geometry device processing requirements.

1

7.3.1

CW Laser Alloying of Au-Ge/GaAs

The conventional technique for the formation of ohmic contacts to n-GaAs consists of the evaporation and subsequent thermal alloying of a layer of eutectic-composition Au-Ge [7.18]. When heated to a temperature typically 100 to 15OoC above the Au-Ge eutectic point of 356OC, the contact metal melts and dissolves a portion of the GaAs. Upon cooling, the epitaxially regrown GaAs layer incorporates Ge as a substitutional dopant. Provided the Ge is located primarily on Ga sites, it acts as a donor, and a degenerate n+ layer is formed. The lowered energy barrier at the metal-semiconductor interface then permits ohmic conduction [7.19]. There are many limitations to this technique. Although a Ni or Pt overlayer i s almost always included to improve the wetting of the molten Au-Ge, the alloyed contacts are characterized by significant surface roughness and poor edge definition, by microprecipitates of nonuniform composition, and by uneven penetration into the underlying GaAs [7.20, 7.211. The alloy temperature and time, moreover are constrained by the amphoteric nature of Ge and the resulting need to prevent the loss of the extremely volatile As dissolved in the Au-Ge melt. It has been shown that rapid heating rates and short times lead to

412


improved contact resistance and morphology [7.221, but there is a practical limit (on the order of 50°C/sec) to the rate which can be achieved with conventional furnace processing. The ability of cw beam irradiation to produce high surface temperatures for short periods of time has the potential to eliminate many of the problems just discussed. The first attempt to form laser-alloyed contacts was reported by Pounds et al., [7.23] in 1974 using a Q-switched laser. Although the process was not well controlled (melting of few microns of substrate resulted from irradiation) and the resulting specific contact resistance were relatively high R-cm’), the basic feasibility of the technique was demonstrated. Gold et al., [7.24] reported the use of lasers (free running ruby laser and cw argon laser) to alloy Au-Gelstructures resulting in improved surface morphology as compared with furnace alloying. This is illustrated in Fig. 7.11 where the micrographs of furnace, pulsed and cw laser, alloyed contacts morphologies are compared. To further quantify the improvement a stylus profiler was used to characterize the surfaces of both furnace and laser alloyed samples [7.251. These profile are shown in Fig. 7.12; the laser-alloyed surface (pulse laser shown here) is seen to be significantly smoother. A similar improvement was observed on alloyed samples whose contact metal had been stripped to expose the metal-semiconductor interface. The metallization system reported by Gold j7.251 was designed specifically to be compatible with laser processing. The deposition sequence was Au (200 A ) followed by Ge (100 a); some samples also had a thin layer of Ni, nominally 20 A , at the Au/GaAs interface. The presence of Ge as the top layer, rather than Au o r eutectic composition Au-Ge resulted in high light absorption (reflectivity at X =5145 A was measured to be only 31%) and allowed alloying to take place at laser powers well below that at which damage occurred. This basic technique of self limiting

FIGURE 7.11. Microscopic appearance of alloyed 115 um-square Au-Ge contacts to n-GaAs. The alloyed metal are Au (200 A ) followed by Ge (100 8 ) (a) thermal anneal 45OoC, 60 sec; (b) ruby laser 15 J/cm2, 1 msec; (c) scanned argon laser anneal 2.5 W.

7.


413

. U

E

c

"i

FURNACE

P

W

>

I-

LASER

1 , , , , , , , , , , , 1 HORIZONTAL

(50pmldiv.)

FIGURE 7.12. Surface profilometer characterization of pulsed alloyed Ge-Au/GaAs ohmic contacts. Laser energy was 15 J/cm2; furnace alloy was 45OoC, 10 sec. \ absorption has also been successfully applied to the cw laser processing of metal-silicides (Chapter 6). The scanning cw argon ion laser system described in Chapter 1 was focused on a 40 pm spot and scanned at a speed of 12.5 cm/sec. An incident power of 2.5 W resulted in the surface morphology shown in Fig. 7.13(a). When Au-Ge layers are used, the metal tends to pull back from the edge of the contact, due most likely to the surface tension of molten Au-Ge. However if a Ni underlayer i s utilized this behavior is not seen [Fig. 7.13(b)] which can be explained by the

FIGURE 7.13. Nomarski photomicrographs of cw argon laser alloyed contacts on GaAs. (a) Ge-Au (b) Ge-Au-Ni. Size of contact squares is 115 pm.

414


improved a d h e s i o n of t h e metal. E x c e l l e n t adhesion w a s observed i n any c o n f i g u r a t i o n w i t h a f r e e running ruby laser i r r a d i a t i o n . I n t h i s c a s e e n e r g i e s i n t h e v i c i n i t y of 15 J / c m 2 were used ~ i n the l e a d i n g t o a s p e c i f i c c o n t a c t r e s i s t a n c e of 2 ~ 1 0 -Q-cm2 b e s t case, b u t w i t h a poor r e p r o d u c i b i l i t y . An e x t e n s i v e comparison of d i f f e r e n t beam i r r a d i a t i o n technique t o a l l o y Au:Ge based c o n t a c t s on G a A s w a s r e p o r t e d by Eckhardt f7.261. The b e s t r e s u l t s (lowest p c , f i r m e s t adhesion between c o n t a c t and G a A s and h i g h e s t r e p r o d u c i b i l i t y ) were obt a i n e d w i t h t h e s i n g l e mode (TEM,) cw Ar-ion laser. The summary of t h e r e s u l t s , and e x p e r i m e n t a l c o n d i t i o n s i s p r e s e n t e d i n Table 7.1. A l l t h e samples used i n t h i s s t u d y c o n s i s t e d of s e m i i n s u l a t i n g G a A s implanted w i t h 4x1Ol2 Si+/cm2 a t 1 0 0 keV ( n 2 1017 a t t h e sample s u r f a c e ) . The laser beam w a s always focused o n t o t h e sample i n a 66 Dm s p o t s i z e .

A number of 1 Pm g a t e GaAs MESFET's were prepared w i t h a l a s e r a l l o y e d Au:Ge based c o n t a c t and compared w i t h f u r n a c e a l The laser annealed c o n t a c t s loyed c o n t a c t s from t h e same wafer. were found t o be s u p e r i o r o r a t l e a s t as good i n e v e r y r e s p e c t . T h e i r r e s i s t a n c e was lower by 10-20% and t h e dc c h a r a c t e r i s t i c s were e x c e l l e n t t7.271. Average v a l u e s of g a i n , and n o i s e measured a t 1 4 GHz were somewhat b e t t e r f o r t h e laser annealed devices. In a l l cases t h e s u r f a c e morphology and edge d e f i n i t i o n of t h e laser annealed c o n t a c t s w a s f a r s u p e r i o r t o t h a t of t h e i r t h e r m a l l y annealed c o u n t e r p a r t s . Furthermore, because of t h e s h o r t n e s s of t h e l a s e r - a n n e a l i n g p e r i o d (1-100 msec), i n t e r d i f f u s i o n of semiconductor and metal c o n s t i t u e n t s i s reduced ( v e r i f i e d by Auger spectroscopy s t u d i e s i7.261) and microscopic phase s e p a r a t i o n due t o m e t a l flow i s prevented. A number of o t h e r l a b o r a t o r i e s have r e p o r t e d t h e formation c o n t a c t r e s i s t a n c e a l l o y e d ohmic c o n t a c t s u s i n g pulsed laser (9-switched o r f r e e r u n n i n g ) o r p u l s e d e l e c t r o n beam. But

of low

TABLE 7.1. Laser.

Contact Metal

Au :Ge-Ni-Au Au:Ge-Pt-Au Au:Ge-Ag-Au Au :Ge-Ti-Au In-Au :G e

B e s t Ohmic C o n t a c t s Formed on G a A s by CW Ar-ion

S p e c i f i c Contact R e s i s t a n c e (Q-cm)

4 . 8 ~ 10-6 I .5x10-5 2. oX10-4 1 3 x 1 0-5 1.3~1 0-6

Laser C h a r a c t e r i s t i c s Power Scan V e l o c i t y (W) (cmls) 4.0

3.8 4.1 3.8 3.5

0.43 0.43 0.43 0.2

0.43

7.


415

i n a l l cases r e p o r t e d , no r e l i a b i l i t y o r l i f e t i m e d a t a was obtained. I f t h e beam a l l o y i n g technique has brought an important improvement i n t h e formation of ohmic c o n t a c t s t o G a A s , t h e d e g r a d a t i o n of t h e s e c o n t a c t a t e l e v a t e d temperatures c o n t i n u e s The t o be a l i m i t i n g f a c t o r t o t h e l i f e t i m e of Ga As devices. n e x t s e c t i o n w i l l d e s c r i b e a d i f f e r e n t technique t o form ohmic c o n t a c t t o nGaAs t h a t are s t a b l e a t e l e v a t e d temperatures. 7.3.2

Cw Laser A s s i s t e d D i f f u s i o n and A c t i v a t i o n of Sn i n GaAs

Among t h e d i f f e r e n t doping techniques a v a i l a b l e , i o n implant a t i o n and e p i t a x y ( l i q u i d , vapor phase o r molecular beams) a r e D i f f u s i o n processes i n GaAs a r e d i f f i c u l t preferred f o r GaAs. due t o t h e high vapor p r e s s u r e of a r s e n i c a t d i f f u s i o n temperat u r e s . They u s u a l l y r e q u i r e s o p h i s t i c a t e d equipment such a s s e a l e d ampoules o r c o n t r o l l e d environments. An open tube d i f f u s i o n technique i s presented h e r e u s i n g t h i n f i l m s o l i d sources. It r e p r e s e n t s a low-cost, r e p r o d u c i b l e method t o o b t a i n t h i n Thermal t r e a t h e a v i l y doped n l a y e r s on semi-insulating GaAs. ments o r a combination of thermal and cw l a s e r p r o c e s s i n g can be used t o c o n t r o l t h e d i f f u s i o n process. The t h i n f i l m s o u r c e i s a mixture of Sn02 and Si02 i n s o l u t i o n deposited using a In a d d i t i o n t o t h e d i f f u s i o n , an i n t e r f a c e conventional spinner. A d e t a i l e d a n a l y s i s of t h i s r e a c t i o n i s observed t o occur. r e a c t i o n and t h e r o l e of t h e d i f f e r e n t atoms involved i n t h e The combination of t h e i n t e r f a c e complete process i s presented. l a y e r and Sn doped G a A s l a y e r s w i l l r e s u l t i n t h e formation of low c o n t a c t r e s i s t a n c e non-alloyed ohmic c o n t a c t s t o n-type m a te r ia1

.

7.3.2.1

Source and Sample P r e p a r a t i o n

The sample p r e p a r a t i o n s t e p s a r e s c h e m a t i c a l l y d e s c r i b e d i n Fig. 7.14. Semi-insulating chrome-doped GaAs s u b s t r a t e s , o r i e n t e d along t h e d i r e c t i o n a r e used. The s o u r c e c o n s i s t s of a t i n s i l i c a spin-on f i l m obtained from Emulsitone 17.281. The s o l u t i o n is a mixture of Sn02 and Si02 i n e t h y l a l c o h o l . A p h o t o r e s i s t s p i n n e r i s used t o s p i n t h e f i l m on. R o t a t i o n speeds of 3000 t o 5000 rmp a r e s e l e c t e d depending on t h e s i z e of t h e s u b s t r a t e A processed. The f i l m s obtained a r e t y p i c a l l y 0.3 pm t h i c k . thermal t r e a t m e n t of 2OO0C f o r 15 min. removes t h e b i n d e r ( e t h y l a l c o h o l ) l e a v i n g behind a l a y e r of Sn02/Si02 w i t h a molecular c o n c e n t r a t i o n of about l : l O , according t o t h e vendor.

A s an encapsulant f o r f u r t h e r thermal t r e a t m e n t s , a l a y e r The of CVD Si02 ( S i l o x ) i s deposited on t h e Sn02/Si02 film. The t h i c k n e s s of t h i s temperature of d e p o s i t i o n i s 45OOC. second l a y e r h a s been found t o be c r i t i c a l t o avoid t h e thermal

YVES 1. NISSIM AND JAMES F. GIBBONS

416

SI

Cr.

o2

(SILOX)

DOPED

GaAs

I

1 -

S P I N . ON A SnOl 1 5 0 , / ETHYL ALCOHOL SOLUTION

2

BAKE

~

FIGURE 7.14. diffusion i n GaAs.

2OO0C

/15min TO REMOVE BINDER

Schematic of t h e sample p r e p a r a t i o n f o r t i n

stress o c c u r r i n g a t h i g h temperature caused by thermal expansion between t h e s u b s t r a t e and t h e It was found t h a t t h e oxide t h i c k n e s s should 0.8 um. A t h i c k n e s s of 0.5 pm i s chosen t o condition. 7.3.2.2

the difference i n deposited l a y e r s . be no more than be t h e s t a n d a r d

Thermal D i f f u s i o n

Sn i s an element i n c o l u m I V of t h e p e r i o d i c t a b l e , and t h e r e f o r e i s a p o s s i b l e amphoteric dopant f o r t h e 1 1 1 - V compound GaAs. I n t h i s system, Sn has a s t r o n g tendency t o occupy Ga s i t e s and then is f r e q u e n t l y used a s a n-type dopant. The d i f f u s i o n of t h i s i m p u r i t y i n GaAs from Sn doped s i l i c o n d i o x i d e h a s been r e p o r t e d from v a r i o u s t y p e s of t h i n f i l m d e p o s i t i o n [7.29-7.321. The s t u d y of t h e thermal d i f f u s i o n from t h e spin-on s o l u t i o n presented h e r e shows s i g n i f i c a n t l y d i f f e r e n t chemical and e l e c t r i c a l behavior of Sn compared t o t h e doped s i l i c o n d i o x i d e s o u r c e s r e p o r t e d before. Any a b r u p t changes i n temperature of t h e double l a y e r "source-cap" (SnO2/SiO2-SiO2) s t r u c t u r e d e s c r i b e d i n t h e sample p r e p a r a t i o n produce stresses t h a t cause t h e f i l m s t o peel. F o r t h i s r e a s o n , a slow thermal ramp i s r e q u i r e d b e f o r e any subsequent thermal t r e a t m e n t . Since t h i s procedure i n i t i a t e s t h e d i f f u s i o n of Sn from t h e s o u r c e , a v e r y p r e c i s e ramping c y c l e i s performed t o i n s u r e r e p r o d u c i b i l i t y of t h e experiments. The t o t a l c y c l e i s done i n 15 min i n a N2 ambient. A s a n example, a d e t a i l e d temperature ramp from room temperature t o 9OO0C i n 15 min i s d e s c r i b e d below:

7. (a) (b) (c) (d) (e) (f)


Room temperature t o 700°C t o 85OOC i n 3 85OoC t o 875OC i n 1 875OC t o 89OoC i n 1 89OoC t o 900°C i n 1 900°C f o r 2 min

417

7OO0C i n 7 min ( = 100°C per min) min (5OoC p e r min) min min min

I f t h e f i n a l t e m p e r a t u r e i s not 900°C t h e n t h e temperature reached i n s t e p (a) i s changed i n o r d e r t o f o l l o w t h e same ramp from s t e p ( b ) t o ( f ) . A Van d e r Pauw technique i s used t o measure s h e e t resistiv i t y , H a l l m o b i l i t y and s h e e t carrier c o n c e n t r a t i o n a f t e r t h e These ramping c y c l e . The r e s u l t s are p r e s e n t e d i n Table 7.2. d a t a show t h a t t h i s sequence i s r e s p o n s i b l e f o r both d i f f u s i o n and e l e c t r i c a l a c t i v a t i o n of Sn. Measurable a c t i v i t y starts when t h e f i n a l temperature of 8 O O O C i s reached w i t h an a c t i v e dose of 4 . 7 ~ 1 0atoms/cm2. ~ ~ This v a l u e can r e a c h 4 . 4 ~ 1 0atoms/ ~ ~ cm2 f o r a ramp t o 950OC. The low v a l u e of the m o b i l i t y o b t a i n e d a t 8OOOC i n d i c a t e s t h a t t h e d i f f u s i o n has b a r e l y s t a r t e d and i s very nonuniform. Expected v a l u e s of H a l l m o b i l i t y are o b t a i n e d as t h e f i n a l temperature i s i n c r e a s e d . TABLE 7.2 Electrical Characterization Layers Obtained A f t e r a Ramping Cycle Only.

pH(cm2/V.sec)

750 8 00 850 9 00 950

I

974 2 24 122 64

of

t h e Diffused

Ns (

)

-

--

1353 2255 2098 2034

4.7 x 1012 1.2 1013 2.4 1013 4.4 1013

A comparison of t h e chemical p r o f i l e and t h e e l e c t r i c a l l y a c t i v e one i s o b t a i n e d by a combination of SIMS and d i f f e r e n t i a l Van d e r Pauw a n a l y s i s . Data f o r t h e two measurements are preIt s e n t e d i n Fig. 7.15 f o r a sample ramped t o 900°C i n 1 5 min. can be seen i n t h i s f i u r e t h a t only 3 t o 4% of t h e t o t a l conas measured by RBS) i s e l e c t r i c a l l y c e n t r a t i o n of Sn ( 7 ~ 1 0 ~ ' a c t i v e . The t o t a l d i f f u s i o n d e p t h i s = 1600 A. The l a r g e d i f f e r e n c e between t h e two p r o f i l e s i s l o c a t e d i n the f i r s t 300 a of t h e s u r f a c e where t h e t o t a l amount of Sn r e a c h e s a v a l u e above 1020/cm3. Following t h i s near s u r f a c e r e g i o n , most of t h e i m p u r i t i e s are e l e c t r i c a l l y a c t i v e . Further s t u d i e s , reported l a t e r , i n d i c a t e t h e formation of a Sn-As compound i n t h i s r e g i o n , e x p l a i n i n g t h e h i g h Sn c o n c e n t r a t i o n . Consequently t h e y i e l d of

418

YVES 1. NISSIM AND JAMES E GIBBONS

IV

THERMAL RAMP TO 9 0 0 ° C IN 15 M I N U T E S

-SIMS PROFILE ('"Sn) --*--

\

\.

-0-

N~ 1, VAN DER PAUV MOBILITY( PROFILE

lo*

. >

n

5

-0

lo3

>

e

I

-

-J

: 500 DEPTH

1000

m

o

5

1500

(A)

Fig. 7.15. SIMS and Van der Pauw stripping profiles for a sample ramped to 900°C in 15 min. The Hall mobility is also shown. Sn in this matrix does not correspond exactly to the scale shown in Fig. 7.14 since the calibration was performed for Sn in GaAs. A dotted SIMS profile in the near surface region is plotted for this reason. The variation of mobility in the diffused layer is also shown in Fig. 7.15. These measured values agree to within 10% of those expected from Irvin-Sze f7.331. Extensive studies of the thermal diffusion following the initial thermal ramp was reported by Nissim et al., i7.341. They have shown that if further diffusion is carried out with the "source-cap" (Sn02/Si02: SiOp), a carrier concentration of 2x1014 cm-2 with sheet resistivity as low as 20 .Q/o can be achieved. Due to the large amount of Sn introduced into the substrate during the ramp, this cycle can be viewed as a conventional predeposition step. The equivalent of a drive in diffusion was performed by removal of the "source-cap", deposition of a new encapsulant (As-doped CVD SiOp), and thermal annealing. The sheet electrical characteristics of the diffused layer thermally annealed at BOO, 850 and 900°C for 30 min are shown in Fig. 7.16 as a function of ramping (predeposition) temperature. It can be seen that here again very low sheet resistivity (20/0) can be achieved. Electrical depth profiles were performed for the 85OoC/30 min anneal using a differential Van der Pauw technique and chemical stripping. The results are shown in Fig. 7.17.

7.


- I ?

THERMAL ANNEALING WITHOUT "SOURCE CAP" o 30min.at 8OO0C\

A 30 min. at 850 OC x 30 min. at 900°C

419

/

r2int

z W

V 2

0 V

LL

w

LL LL

a

V IW W

I

Y,

,

THERMAL ANNEALING WITHOUT "SOURCE CAP"

300.

.

0

3 0 min. at 8 0 0 ° C

x

30 min. a t 9 0 0 ° C

0

-c> 2

200-

I-

?! cn W a c W W

I

rp,

too.

FIGURE 7.16. Evolution tion) conditions of the sheet diffused layer when thermally of different temperatures (a) sheet resistivity.

for different ramping (predeposielectrical characteristics of the annealed without the "source-cap'' sheet carrier concentration (b)

420


0

0.2

0.4 0.6

0.8

1.0

1.2

4

DEPTH (,urn)

FIGURE 7.17. Electrical profiles for ramping cycles to 600, 800 and 9 0 0 ° C followed by an anneal at 850°C/30 min without the "source-cap". A SIMS profile for the same anneal and a ramping temperature of 9 5 O o C is also shown.

7.3.2.3

%-Laser

Assisted Diffusion and Activation

Following an initial thermal ramp, the diffusion of Sn impurities can also be assisted using a scanning cw laser beam l7.351. The characteristic features of this process are presented here. The ramping cycle prior to laser treatment is necessary to prevent the peeling of the double layer "source-cap" during irradiation and apparently to break down an interfacial barrier to start the diffusion process. It has been impossible to start the diffusion of Sn with the scanning laser alone. The thermal ramp gives an electrically active Sn concentration of 2 to 3 x101*/cm3 to a depth of about 1600 in the GaAs substrate as seen in the previous section. The wafers were then laser scanned using a cw argon ion laser. The beam was focused by a 136 mm lens, leading to a 50 pm beam diameter at the focal plane. The spot was then scanned with 15 um spacing

7.


421

between a d j a c e n t l i n e s a t a speed of 1 2 cm/sec. In order t o reduce t h e thermal stress c r e a t e d by t h e i n c i d e n t beam a t t h e s u r f a c e of t h e wafer, t h e s u b s t r a t e was h e l d a t 35OoC d u r i n g i r r a d i a t i o n . These parameters were found t o be optimum and were k e p t c o n s t a n t . The i n v e s t i g a t i o n of d i f f e r e n t temperature regimes w a s accomplished by changing o n l y t h e i n c i d e n t power l e v e l . The r e f l e c t i v i t y of t h e s u b s t r a t e w i t h i t s double l a y e r "source-cap" was d i r e c t l y measured and found t o be 24%. Using t h e c a l c u l a t e d curves of laser induced t e m p e r a t u r e i n GaAs [7.1] and t h e above parameters , a maximum s u r f a c e temperature was a s s o c i a t e d w i t h each i n c i d e n t laser power. The d i f f u s i o n of Sn w a s f i r s t s t u d i e d i n a regime where t h e laser power w a s k e p t below t h e l e v e l r e q u i r e d t o m e l t t h e subs t r a t e . This working window was determined by observing t h e laser power r e q u i r e d t o j u s t produce v i s i b l e thermal e t c h i n g and t h e n r e d u c i n g t h e s e t t i n g by 5%. A power l e v e l of P = 0.61W, l e a d i n g t o a maximum induced temperature of about 8OO0C, was o b t a i n e d . A series of 1, 3 and 5 scan frames were performed w i t h t h e double l a y e r " s o u r c e cap" remaining on t h e s u b s t r a t e . The Van d e r Pauw technique was used t o c h a r a c t e r i z e t h e r e s u l t i n g s h e e t r e s i s t i v i t y , Hall m o b i l i t y and s h e e t c a r r i e r c o n c e n t r a t i o n . The r e s u l t s are summarized -in Table 7.3. The i n c r e a s e of t h e ~~ accoms h e e t carrier c o n c e n t r a t i o n from 2.44 t o 3 . 0 1 ~ 1 0 cm-*, panied by a d e c r e a s e of s h e e t r e s i s t i v i t y w i t h t h e l a r g e r number of s c a n s , i l l u s t r a t e s e i t h e r d i f f u s i o n from t h e source and/or a c t i v a t i o n of Sn i m p u r i t i e s induced d u r i n g t h e thermal c y c l e .

SIMS a n a l y s i s was performed on a sample t h a t was t h e r m a l l y ramped only, and on a sample scanned f i v e times a f t e r t h e ramp. The r e s u l t i n g p r o f i l e s shown i n Fig. 7.18 i l l u s t r a t e an i n c r e a s e i n Sn c o n c e n t r a t i o n and an i n - d i f f u s i o n of about 150 8 . To i n v e s t i g a t e t h e c o n t r i b u t i o n of t h e i r r a d i a t i o n t o t h e d i f f u s i o n from t h e "source-cap", compared t o t h e a c t i v a t i o n of t h e i m p u r i t y p r e s e n t i n t h e s u b s t r a t e a f t e r t h e ramp, t h e "source-cap" was removed b e f o r e laser scanning. The r e f l e c t i v i t y of t h e b a r e TABLE 7.3 Sheet E l e c t r i c a l C h a r a c t e r i z a t i o n of t h e LaserD i f f u s e d Layers With t h e "Source-Cap'' On.

Thermal ramp: 900'C i n 15 min.

+ + +

+

0 1 3 5

scan scan scans scans

p(n/o)

122 118 117 107

I-IH(cm2 /V.sec) 2098 201 7 1838 1940

N, ( cme2 ) 2.44 2.63 2.91 3.01

1013 1013 1013 1013

422


THERMAL RAMP 0 TO 9 0 0 ° C IN 15 MIN:-,

1

\ I

SIMS PROFILE

t\ i

THERMAL RAMP: --SCANS \,ON TOP OF "SOURCE-CAP"

'\, +5 LASER

I

I

1

I

800

400

DEPTH

FIGURE 7.18. SIMS p r o f i l e s ramped + l a s e r scanned samples.

I

I

1200

(i) of

Sn obtained

on

ramped

and

s u b s t r a t e ( a f t e r t h e thermal ramp) was measured and t h e i n c i d e n t l a s e r power a d j u s t e d t o reach t h e same maximum temperature (2 800°C) as before. The wafer w a s scanned f i v e t i m e s and t h e s h e e t e l e c t r i c a l measurements were c a r r i e d o u t . The r e s u l t s , on, when compared w i t h those obtained w i t h the "source-cap" suggest t h a t t h e t o t a l i n c r e a s e i n a c t i v e i m p u r i t i e s due t o l a s e r i r r a d i a t i o n i s 77% from t h e s o u r c e and 23% from t h e a c t i v a t i o n of Sn i n t r o d u c e d d u r i n g t h e ramp. The a c t i o n of t h e laser i s found t o be more s i g n i f i c a n t i f t h e sample i s l e f t " a t temperature" f o r a s h o r t t i m e a f t e r t h e ramp. For t h i s reason t h e above experiment w a s r e p e a t e d a f t e r thermally ramping t h e s u b s t r a t e and i t s source t o 900°C and l e a v i n g i t f o r 5 min l o n g e r a t 900°C. A s e r i e s of 5 s c a n s inducing a maximum temperature of about 800°C was then performed. A net i n c r e a s e i n c a r r i e r c o n c e n t r a t i o n from 3 t o 8 x 10l8 is observed a f t e r t h e s c a n s . The t o t a l Sn p r o f i l e was o b t a i n e d u s i n g RBS. The a n a l y s i s was done w i t h 2.2 MeV helium atoms a f t e r removal of the source. The b a c k s c a t t e r e d p a r t i c l e s p e c t r a obt a i n e d f o r random o r i e n t a t i o n a r e presented i n Fig. 7.19. A sample having received a thermal treatment only, a n o t h e r one r e c e i v i n g a d d i t i o n a l laser s c a n s (Tmx = 800°C) and t h i r d scanned a t h i g h e r l a s e r power (TMx = 850"C), were analyzed. The Sn

7.


423

Ramp * 5 m i n at QOOOC

-Ramp o Ramp

5 min at 9 0 0 ° C * 5 LASER SCANS AT P.0.61 W 5 min at QOO°C *3 LASER SCANS AT P ~ 0 . 6 6 W

ENERGY ( M e V )

FIGURE 7.19. RBS spectrum of the Sn profile in the laser diffused layers for different incident laser power. profile is characterized by the large peak in the surface which increased after laser scan, fusion of Sn from the source. An increase in profile after irradiation is also observed. the diffusion of Sn in GaAs.

concentration at showing the difthe depth of the This illustrates

The temperature regime where the incident power is above the melting threshold of the GaAs substrate has also been investigated. Melting occurs at a power of about P = 1W. Different experiments were tried in the range of 1 to 2W. When melting occurs, the surface of the substrate becomes very smooth, and depending on the incident power level, different colors appear in the scanned area. This is illustrated in Fig. 7.20 for incident powers of 1.2 and 1.9W. The behavior of this system is very similar to the silicon-metal (silicide) reactions induced by a scanning laser (Chapter 6 ) . The formation of a series of compounds involving Sn and As is believed to occur. However those formed during the high power scanning could not be identified since the substrate cracks during an overlapping scan (P > 1.4W) as shown in Fig. 7.20(b). The compound formed by the lower power scan, has been identified and will be discussed in the following section. Electrical characterization of these diffused

424


FIGURE 7.20. Nomarski o p t i c a l micrograph of s i n g l e laser s c a n a t power above t h e m e l t i n g t h r e s h o l d of t h e s u b s t r a t e ( a ) 1.2 W; ( b ) 1.9 W. l a y e r s showed high carrier c o n c e n t r a t i o n s (3 x 1014/cm2) but anomalously low m o b i l i t i e s as compared t o t h o s e obtained i n t h e s o l i d phase regime. 7.3.2.4

Surface Reaction Analysis

The chemical p r o f i l e s o b t a i n e d u s i n g t h e RBS o r SIMS techniques show a n anomalously h i g h c o n c e n t r a t i o n of Sn c l o s e t o t h e However t h e r e i s s u r f a c e i n a l l t h e d i f f u s e d l a y e r s analyzed. no corresponding h i g h s u r f ace c o n c e n t r a t i o n of e l e c t r o n s ( s e e F i g . 7.15 f o r example). This o b s e r v a t i o n s u g g e s t s t h a t a chemical reaction takes place i n addition t o the diffusion process. To s t u d y t h i s p o s s i b i l i t y , a sample ramped t o 900°C i n 15 min. and a sample ramped and l a s e r scanned f i v e times a t a power inducing a maximum temperature of 8 O O O C were prepared f o r t r a n s mission e l e c t r o n microscopy and d i f f r a c t i o n a n a l y s i s u s i n g conB r i g h t f i e l d ( F i g . 7.21) and ventional jet thinning techniques. dark f i e l d t r a n s m i s s i o n e l e c t r o n micrographs i n d i c a t e d p r e c i p i t a t i o n a f t e r both t h e thermal ramp and laser p r o c e s s i n g . S e l e c t e d a r e a e l e c t r o n d i f f r a c t i o n p a t t e r n s revealed t h a t t h e p r e c i p i t a t e s were composed of a t i n - a r s e n i d e compound (As2Sn3). A d d i t i o n a l l y , an amorphous g a l l i u m oxide w a s d e t e c t e d i n t h e sample (B-Ga203). After laser i r r a d i a t i o n , an i n c r e a s e of t h e amount of s u r f a c e I n a l l cases coverage by B-GagOg c r y s t a l l i t e s was observed. t h e s u r f a c e o x i d e f i l m appeared as a l a t e r a l l y d i s c o n t i n u o u s f i l m . A f t e r a thermal ramp t o 900°C i n 15 min, t h e compound l a y e r is 300 8, t h i c k (from SIMS p r o f i l e ) . T h i s i n d i c a t e s t h a t t h e r e is a l a r g e e x c e s s of G a atoms f r e e o r involved i n t h e formation of t h e o x i d e . To i n v e s t i g a t e t h e p o s s i b l e d i f f u s i o n of t h e s p e c i e s o u t of t h e i n t e r f a c e r e g i o n , RBS a n a l y s i s was performed b e f o r e and a f t e r t h e thermal ramp w i t h t h e double l a y e r "source-cap" remaini n g on t h e s u b s t r a t e . The r e s u l t s of t h i s experiment are shown i n

7.


425

FIGURE 7.21. Bright field transmission electron micrograph showing formation of As2Sn3 plates. In the inset are the rings associated with As2Sng and B-Ga2O3. Fig. 7.22. After the ramp, diffusion of Sn in the GaAs substrate and in the Si02 is observed. A slight shift of the GaAs edge indicates that part of the encapsulant was lost during the cycle. But the most striking feature is the formation of an extra peak after the ramp. The energy of this peak corresponds to a gallium signal at the surface. This demonstrates that Ga is dissolved from the GaAs by the Si02 layers and diffuses toward the outermost surface. These analyses have shown that the different components in this system have a precise behavior resulting in the formation of the doped layers: Sn diffuses into the GaAs substrate to create an n+-type layer, and reacts at the interface with As to form As2Sn3. The extra Ga diffuses through the "source-cap'' and forms B-Ga2O3 oxide. 7.3.2.4

Nonalloyed Ohmic Contact Formation

Relatively high carrier concentration can be achieved in GaAs when Sn impurities are diffused from a spin-on Sn02/Si02 source. The thermal cycle initiating the diffusion of impurities has been optimized'for the best values of contact resistance with two main criteria: a high value of active carrier concentration and short diffusion length. These conditions are met after a thermal ramp from room temperature to 900°C in 15 minutes. The

426

YVES I . NISSIM AND JAMES E GIBBONS

SPIN -ON SOURCE

SPIN-ON

SnO, /S!O1

. -

-

o

AS DEPOSITED AFTER RAMP TO 9 0 0 ° C IN 15 MIN

0

ENERGY ( M e V )

FIGURE 7.22. RBS spectra of a GaAs sample with its “sourcecap” on before and after a thermal ramp. resulting profile studied in the previous section is shown i n Fig. 7.15. It corresponds to a relative1 flat active carrier ~~ to a depth for about concentration profile at 2 - 3 ~ 1 0atomsfcmr 1600 4 . In addition t o the diffusion, an interface reaction occurs, leading to the formation of a tin-arsenide compound (Sn3As2). This reaction takes place i n the first 300 of the diffused layer. Subsequent metallization of these layers shows ohmic behavior without an alloying cycle. It has been seen that following the thermal ramp, cw laser scans can assist the diffusion and activation of Sn. Nonalloyed ohmic contact formation has therefore been investigated on the thermally ramped and laser scanned layers. The laser irradiation parameters described previously were kept constant, and contact resistance of these layers was determined as a function of incident power I7.361. Values of specific contact resistance were measured using the Transmission Line Model (TIM) [7.37-7.381. The pattern is isolated by a mesa etch and the metal is patterned using a standard photolithographic lift-off. The spacing between metal pads was 6 , 4 , 2, 4 and 6 pm. The symmetry of this structure is intended to eliminate any systematic errors due to lateral gradients in sheet resistance. Following diffusion, the “sourcecap” is removed and the TIM patterns are prepared. The values of specific contact resistance and sheet resistivity obtained from these measurements are plotted as a function of incident laser power in Fig. 7.23. All the samples were ramped to 900°C in 15

7.


427

t-

cn

100 5 w LL

60

L

20

2

w w

FIGURE 7.23. Effect of laser power on the properties of the diffused n+ layers and nonalloyed Tc-Pt-Au contacts. minutes before laser irradiation. Immediately after the ramp, the of the nonalloyed contacts is specific contact resistance (p,) about loW4Q-cm2. A dramatic decrease of two orders of magnitude is observed when the wafer is scanned at incident laser powers between 0.4W and 0.6W. Specific contact resistance between 1 and 2 ~ 1 0 -Q-cm2 ~ are obtained for this processing window while the GaAs surface remains smooth and free of visible damage. Above 0.65W a slight increase of specific contact resistance is observed, probably due to deterioration of the surface. The bars on the pc data represent the range of values obtained for various runs. The temperature calculation developed in Chapter 2 has been applied in this experiment to define the temperature limits of the different regions. The results are shown in Fig. 7.24. The three distinct regions that appear on this figure are: (i) The processing window, that leaves a mirror finish on the surface and corresponds to induced temperature between 600 and 800OC. This window is large enough to insure very reproducible specific contact resistance in the low S1-cm2 range. It has been experimentally observed that five consecutive laser scans of the sample are optimum in achieving the lowest value of pc.

(ii) The region where the induced temperature is between 800°C and the melting point of the GaAs substrate. Slip lines and surface decomposition occur in this window and the specific contact resistance of the resulting layers is increased.


428

1600

GALLIUM ARSENIM

1400 L

-v

BEAM DIAM: 5 0 p m SCAN SPEED : 12c.m/sec

1200

g 1000 3

800 w

n

5 600 400

200 0

0.2

0

0.4

0.6 0.8 1.0 INCIDENT POWER ( W )

1.2

1.4

FIGURE 7.24. Temperature induced i n t h e GaAs d u r i n g laser i r r a d i a t i o n as a f u n c t i o n of i n c i d e n t power. The d i f f e r e n t regimes f o r ohmic c o n t a c t f o r m a t i o n are shown. ( i i i ) Above t h e m e l t i n g p o i n t of GaAs, t h e s u r f a c e becomes s h i n y again. This regime has been d i s c u s s e d i n t h e p r e v i o u s s e c t i o n . The Hall m o b i l i t y of t h e d i f f u ed l a y e r i s a f f e c t e d but v a l u e s of pc as low as 2 . 6 ~ 1 0 - ~Q-c$ have been measured. I n t h e remainder of t h i s s e c t i o n t h e p r o p e r t i e s and o b t a i n e d A s t a n d a r d nonw i t h i n t h e p r o c e s s i n g window are considered. a l l o y e d ohmic c o n t a c t f a b r i c a t i o n i s e s t a b l i s h e d by a ramp t o 900°C i n 15 minutes followed by laser s c a n s a t P = 0.54W. For a g i v e n metal semiconductor b a r r i e r h e i g h t (0.7 eV i n t h i s system) t h e o r e t i c a l c a l c u l a t i o n s of s p e c i f i c c o n t a c t resistance as a f u n c t i o n of carrier c o n c e n t r a t i o n are r e p o r t e d i n Ref. 7.19. F om t h se c a l c u l a t i o s i t can be s e e n t h a t t h e v a l u e s of pc = lo-' Q-cm and p, = lo-' Q-cm 2 are o b t a i n e d f o r c a r r i e r con-

5

e s~p e, c t i v e l y . The carrier c e n t r a t i o n of 8x1Ol8 and 5 ~ 1 0 ~ ~ / c rm c o n c e n t r a t i o n i n t h e d i f f u s e d l a y e r o b t a i n e d a f t e r a thermal ramp t o 900°C ( 3 ~ 1 0 ~ ~ c mwould - ~ ) r e s u l t i n v a l u e s of s p e c i f i c c o n t a c t Q-cm2. r e s i s t a n c e i n t h e v i c i n i t y of But i t h a s been imposs i b l e t o measure a y va e of c a r r i e r c o n c e n t r a t i o n t h a t would Furthermore, Fig. 7.23 r e s u l t i n pC = lo-' Q-cn? as observed. shows t h a t a d e c r e a s e of two o r d e r s of magnitude of p c i s o b t a i n e d a t i n c i d e n t laser powers (0.4 t o 0.6 W) where t h e s h e e t r e s i s t i v i t y Only above of t h e d i f f u s e d l a y e r s remains c o n s t a n t a t 80 n / o 0.6 W i s t h e r e a s i g n i f i c a n t l a s e r - a s s i s t e d d i f f u s i o n and/or

.

7.


429

activation of tin. These two observations suggest that a contribution other than the doping level is responsible for low values of pc. A very distinct characteristic of the diffused profile is the presence of the tin-arsenide compound (Sn3As2) in the first 300 A. The phase diagram of the tin-arsenic system [7.39] is plotted in Fig. 7.25. The melting point of Sn3As2 is shown to be 596'C. It is striking to notice that this temperature corresponds to the laser-induced threshold temperature at which the specific contact resistance drops by 2 orders of magnitude (from the temperature calculation of Fig. 7.24). The action of the laser scan can then be interpreted as causing the melting of isolated Sn3As2 plates nucleated during the thermal ramp, after which they regrow into a more uniform film. This argument is supported by the fact that when the sample was scanned at laser powers inducing temperatures very close to the melting point of the compound, multiple scans were necessary (typically 5) to obtain the lowest pc. This compound would then form an interface layer that reduces the barrier between the metal and the semiconductor. In view of these results, the low value of specific contact resistance is believed to arise from t o co tributions. The high active carrier concentration ( 3 x 10r8/cm2) in the diffused layer provides values of pc in the n-cm2 range. The second contribution comes from the interface layer (SnjAs2) that red ces the metal-semiconductor barrier and brings pc into the 10-'Q-cm2 range. To check the stability of these nonalloyed contacts, aging experiments were carried out at elevated temperatures. The degradation of contact resistance, normalized to 1 mm of contact I

I

I

I

I

1 (@825' )

800

I-

a cc W

a

5

400

I-

200

-

I

- -

I

Y

*

I I I

FIGURE 7.25.

I I

1

I

Phase diagram of the Sn-As system.


430

width, is plotted for two different temperatures: 320 and 400°C in Fig. 7.26. One thousand hours at 32OOC produces an increase in resistance of less than 50%. Even at 400°C the slope of the resistance increase is still small. These results are compared to the behavior at elevated temperature of an alloyed Au-Ge ohmic contact [7.40-7.411. As can be seen in Fig. 7.26 the contact resistance is three times smaller for the case of the nonalloyed diffused contacts and they sustain temperatures that would destroy a conventional ohmic contact. These characteristics may make this diffused contact technology very attractive in the improvement of the reliability of GaAs MESFETS. Application of these techniques to the fabrication of such devices is described in Ref. 7.42. 7.4

R A P I D THERMAL PROCESSING

Sealy and Surridge [ 7 . 4 3 ] first showed that rapid thermal processing in the 1 to 1 0 0 second time scale could result in good electrical activation and high mobility for ion implanted n-type layers in GaAs. Since then a number of studies have been performed in an effort to establish the annealing mechanisms and the range of system operating parameters required to obtain results equivalent to or better than those obtained in furnace annealing. At present the results are mixed, but there is clear promise for the future. In what follows we review the main results of the research to date.

u m

E

8

REF 17 421

0

FIGURE 7.26. contacts.

High temperature stability of planar nonalloyed

7. 7.4.1


431

Resume of Furnace R e c r y s t a l l i z a t i o n

W i l l i a m s 17.441 has c h a r a c t e r i z e d t h e f u r n a c e a n n e a l i n g of i o n implanted G a A s as o c c u r r i n g i n t h r e e s t a g e s . A b a s i c amorp h o u s - t o - c r y s t a l l i n e t r a n s i t i o n , l a b e l l e d as s t a g e I, o c c u r s a t temperatures i n t h e range of 150-200°C. This t r a n s i t i o n i s c h a r a c t e r i z e d by a major r e d u c t i o n i n damage as measured by He' i o n c h a n n e l i n g and/or TEM. However, u n l i k e s i l i c o n , t h i s t r a n s i t i o n l e a d s t o a highly d e f e c t i v e surface l a y e r containing a h i g h d e n s i t y of twins and o t h e r , more complex, p o l y c r y s t a l l i n e structures. A second, broad a n n e a l i n g s t a g e , l a b e l l e d as s t a g e 11, occurs 400°C t o 800°C and i s c h a r a c t e r i n t h e temperature range from i z e d by t h e removal of extended c r y s t a l l i n e d e f e c t s . However, removal of t h e s e d e f e c t s does n o t g u a r a n t e e e l e c t r i c a l a c t i v i t y . I n s t e a d , f o r n-type dopants e s p e c i a l l y , a t h i r d a n n e a l i g stage l e a d i n g t o dopant a c t i v a t i o n ( s t a g e 111) occurs a t temperatures i n t h e range 85O0C-15O0"C. As suggested e a r l i e r , i t i s important t o p r o t e c t t h e s u r f a c e a g a i n s t d i s s o c i a t i o n and a r s e n i c l o s s , by using e i t h e r encapsulating l a y e r s o r A s overpressure, a t annealing t e m p e r a t u r e s i n excess of about 45OoC. N

7.4.2

Annealing of Si-implanted G a A s and i t s P o t e n t i a l f o r MESFETS

Arai e t a l . [7.45] f i r s t r e p o r t e d u s i n g halogen lamps t o For " e n c a p s u l a t i o n " , t h e G a A s w a s p l a c e d f a c e down anneal GaAs. on a S i wafer and annealed f o r 5 seconds a t a temperature of It i s b e l i e v e d t h a t under t h e s e circumstances o n l y t h a t 950°C. amount of A s (- one monolayer) w i l l e v a p o r a t e t h a t i s r e q u i r e d t o e s t a b l i s h an e q u i l i b r i u m p r e s s u r e of A s over GaAs i n t h e small space between t h e wafers. Arai e t a l . o b t a i n e d n e a r l y 100% a c t i v a t i o n of a 3 ~ 1 0 ' ~ / c m ~ Kuzuhara e t a l . [7.46], u s i n g a s i m i l a r S i implant a t 7 0 keV. system, o b t a i n e d 75% a c t i v i t y f o r a 150 keV, 5x101*/cm2 S i implant a t a peak temperature of 950'C and a n a n n e a l i n g t i m e of 2 seconds. Longer a n n e a l i n g times r e s u l t e d i n a r s e n i c l o s s and t h e formation of g a l l i u m p i t s on t h e s u r f a c e , c o n s i s t e n t w i t h t h e d a t a of Rose e t a l . [7.47].

-

Davies e t a l . 17.481 have used f i l a m e n t lamps focussed by e l l i t i c a l m i r r o r s t o a n n e a l G a A s implanted w i t h S i t o a dose of 4x10P4/cm2 at an energy of 200 keV. The samples were capped w i t h a 900 8 l a y e r of Si3N4 d e p o s i t e d by CVD a t 400°C. They obtained a maximum of 50% e l e c t r i c a l a c t i v i t y f o l l o w i n g an anneal a t a peak temperature e s t i m a t e d t o be 1000°C. The peak temperature i s reached i n about 1 second, w i t h t h e lamps being s h u t o f f 2-3


432

seconds a f t e r t u r n on. The i m p u r i t y p r o f i l e o b t a i n e d f o r a t o t a l The maximum e l a p s e d t i m e of 2-112 seconds is shown i n Fig. 7.27. carrier c o n c e n t r a t i o n of ,-6x1Ol8 /em3 r e p r e s e n t s h i g h e l e c t r i c a l a c t i v i t y . The f l a t t o p is a t t r i b u t e d by t h e a u t h o r s t o s o l i d solubility effects. However, t h e complete temperature h i s t o r y i n c l u d e s a long f a l l time due t o t h e n e a r l y l o s s l e s s o p t i c a l c a v i t y used, and t h i s may a l s o have a n important b e a r i n g on t h e c a r r i e r c o n c e n t r a t i o n p r o f i l e s they o b t a i n . F i n a l l y , Tabatabaie-Alavi e t a l . [7.491 have used an a r c lamp a n n e a l i n g system t o a c t i v a t e a GaAs sample implanted w i t h S i t o a dose of 4 ~ 1 0 1 ~ / c m a t~ an energy of 200 keV. These a u t h o r s comment t h a t t h e i r system i s capable of uniform i l l u m i n a t i o n over a 100 nnn d i a m e t e r wafer and a s such i s a p p r o p r i a t e f o r I C processing. The d a t a they o b t a i n f o r a 3 second exposure a t 116OOC and exposure at llOO°C a r e shown i n Fig. 7.28. The

a 10 second

-

I

1

1

\

\

I -

\ \

I I

i/

.i

2000 ; V V

vv

qvvopvv 0.1

0.2

W

\

>" w'

0.3

E V

I

0.4

Y

1000 ri

5

D E P T H (prn)

FIGURE 7.27. E l e c t r o n and S i i m p u r i t y d i s t r i b u t i o n s i n i o n implanted, f i l a m e n t lamp annealed GaAs (Davies e t a l . (7.481. O1982 IEEE

7.


433

200 k e V 1 4 x 10'4Cm-2

o4 V

al

5

(Y

o3 >5

-

t

=!

m

0

r

O2

i DEPTH ( p m )

FIGURE 7.28. Electron and Si impurity distributions in ion implanted GaAs, annealed with argon ac lamp (Tabatabaie-Alavi et al. 17.491. significant difference in the profile qhapes is apparent. It is interesting to compare these profile shapes with calculations of Christel and Gibbons L7.501 on the stoichiometry imbalance produced by recoil effects in such an implant. The result of the calculation is given in Fig. 7.29; when laid over the 3 second data of Tabatabaie-Alavi et al., it is apparent that for short enough anneal times the stoichiometric imbalance is a critical factor in the electrical activity obtained. This imbalance is evidently modified upon further annealing, though the use of an Si02 cap and the dissolution of Ga from the sample must also be included in a thorough analysis of the problem.

7.4.3

FET Fabrication Using Rapid Annealing

Kuzuhara et al. [7.461 have reported on the properties of FETs fabricated using Si-implanted channels with annealing carried out in both furnace and halogen lamp systems. The test devices

434


0 2 '-0 -1 -

~

Recoiled Go OJ1d

AS

FIGURE 7 .29. S t o i c h i o m e t r i c d i s t u r b a n c e s produced by r e c o i l e f f e c t s i n Si-implanted GaAs.

had g a t e l e n g t h s of 1 pm and gate widths of 300 urn. The gm o b t a i n e d f o r t h e d e v i c e s annealed i n t h e halogen lamp system was 100 mS p e r mm of g a t e w i d t h compared t o 80 mS p e r mm of g a t e w i d t h f o r d e v i c e s s u b j e c t e d t o a f u r n a c e a n n e a l i n g c y c l e . There t h u s appears t o be an improvement a r i s i n g from t h e u s e of r a p i d thermal a n n e a l i n g . However, t h e t r a n s c o n d u c t a n c e o b t a i n e d w i t h r a p i d devices, thermal a n n e a l i n g i s n o t as good as i n s t a t e - o f - t h e - a r t where a similar geometry d e v i c e made w i t h an implanted channel and a c a p l e s s f u r n a c e a n n e a l i n a gaseous environment c o n t a i n i n g Ga and A s a t 850°C f o r 20 minutes g i v e s gm of 140-160 mS p e r mm. Devices made w i t h e p i t a x i a l l y grown channels a l s o have gm of 160 mS/mm. Even mre d r a m a t i c improvements are p o s s i b l e by u s i n g molecular beam e p i t a x y t o produce a n undoped GaAs l a y e r f o r t h e FET channel and a doped GaAlAs l a y e r t o supply e l e c t r o n s t o i t ( s o - c a l l e d High E l e c t r o n M o b i l i t y T r a n s i s t o r , o r HEMT). Hence, as w i t h t h e microwave S i b i p o l a r t r a n s i s t o r , t h e c u r r e n t s t a t e o f - t h e - a r t i n GaAs s t i l l r e p r e s e n t s something of a c h a l l e n g e f o r r a p i d a n n e a l i n g , and t h e d i r e c t i o n of t h e technology toward MBE might a l s o r e p r e s e n t a c h a l l e n g e f o r any i o n i m p l a n t a t i o n p r o c e s s .

7. 7.4.4


435

Rapid Thermal Annealing of Be and Zn Implants in GaAs

Be and Zn are generally implanted into GaAs to form p regions in pn junction devices, including both diodes and JFETs. Early work on rapid thermal annealing of these dopants [7.51-7.541 showed that halogen lamps, graphite strip heaters, incandescent lamps and dc A r ' arc lamps could all be used to obtain reasonable majority carrier properties in the annealed layer. In addition, Tabatabaie-Alavi et al. [7.55] have extended their work to the fabrication and characterization of pin diodes made by Be implants followed by annealing with a water-walled argon arc lamp. The hole density and mobility profiles obtained by TabatabaiAlavi et al. [7.55] following a double implant ( 4 . 4 ~ 1 0 ~ ~ / c atm ~ 50 keV followed by 5 . 1 ~ 1 0 ~ ~ / c atm ~150 keV) and anneal at 95OoC for 10 seconds are shown in Fig. 7.30 (approximately 5 seconds are used to reach the final temperature, after which annealing is

I

lo'%

012

oi4

0!6 0.8 ; . 1

1.2

Depth ( p m )

FIGURE 7.30. Room-temperature carrier concentration and mobility profiles measured by differential Hall technique on Beimplanted (4.4~10~~ em2, 50 keV and 5 . 1 ~ 1 0 ~ ~ 150 keV) GaAs samples arc lamp annealed for 10 s to 95OoC.

436


carried out for an additional 5 seconds). The implant was performed into a 14 pm thick, undoped (%-NA 10 15/cm3 ) layer grown on an n+ substrate by vapor phase epitaxial techniques.

-

Residual damage in such structures usually manifests-itself as excess leakage current, poor diode ideality factor (n) and low electric field at breakdown. The lamp annealed diodes exhibited an electric field at breakdown that was within about 20% of the theoretical value, which is superior to results obtained with furnace annealing. Ideality factors TI as low as 1.6 (1 bein ideal) were measured, with leakage current densities of 1.4~101 K A/cm2, which is difficult to achieve with furnace annealing. The process therefore appears to offer some advantage over furnace processing with respect to residual defects following the annealing step. 7.4.5

Effects Due to Extended Hold Time in Annealing of ZnImplanted GaAs

Suzuki et al. t7.561 have discovered a previously unreported effect in the annealing of moderately high dose Zn implants into Cr:GaAs. Zn was implanted to a dose of 5x1Ol4 ions/cm2 at 150 keV and annealed at a variety of annealing temperatures, temperature hold times and heating rates. With no encapsulant and no As vapor pressure, an activation of over 90% was achieved by annealing at 800°C with zero hold time. Prolonging the hold time and/or decreasing the heating rate reduces the activation to values in the range of 70%. Suzuki et al. find from He backscattering measurements that the degree of crystallogaphic recovery of implantation-induced damage is similar to the dependence of electrical activation on heating rate. In prolonged hold time experiments, the electrical profiles show a double peak, implying a depletion of Zn atoms around R (or the point of maximum damage). Suzuki et al. attribute this effect to gettering of Zn atoms in the implantation-induced damage layer. Further measurements are required to clarify this process. 7.5

NOVEL APPLICATIONS OF RAPID THERMAL PROCESSING IN GaAs

To conclude the discussion of GaAs, we describe two novel applications that are presently being explored. The applications represent attempts to dope GaAs and control its surface properties without using ion implantation, thus avoiding the stoichiometric disturbances associated with the implantation process.

7. CW BEAM PROCESSING OF GALLIUM ARSENIDE 7.5.1

437

Controlled S i Diffusion

G r e i n e r and Gibbons 17.571 have observed t h a t S i d i f f u s i o n from a t h i n e l e m e n t a l source can be c o n t r o l l e d by t h e use of a t h i n Si02 encapsulant d e p o s i t e d by CVD a t 400°C. The experimental d a t a s u g g e s t t h a t S i d i f f u s e s r a p i d l y when two S i atoms are l o c a t e d on n e a r e s t neighbor Ga-As s i t e s , making a n e u t r a l donoracceptor pair. The samples used f o r t h e s e experiments were Cr-doped etched i n s t a n d a r d s o l u t i o n s f7.571. Immediately following t h e c l e a n i n g c y c l e , t h e samples were loaded i n t o an electron-beam e v a p o r a t o r where they received After the S i 100 8, of S i at a base p r e s s u r e of 2 ~ 1 0 ' ~Torr. d e p o s i t i o n , t h e samples were encapsulated w i t h e i t h e r Si3N4 o r Si02 o r l e f t uncapped. Si02 was deposited by chemical vapor 1000 8, of s i l i d e p o s i t i o n a t 420°C t o a t h i c k n e s s of 0.5pm. con n i t r i d e was deposited by plasma enhanced chemical vapor d e p o s i t i o n a t 20OoC. During t h e anneal, each GaAs sample was placed f a c e up on a S i sample h o l d e r and a n o t h e r S i cover wafer was placed on top i n c o n t a c t w i t h t h e sample t o i n c r e a s e t h e Silicon thermal mass of t h e sample f o r feedback s t a b i l i z a t i o n . was d i f f u s e d a t temperatures of 850°C t o 1O5O0C f o r times from R i s e times from 200°C were 7-8 sec. After the 3 t o 300 sec. a n n e a l , samples were etched i n HF t o remove t h e encapsulant. Residual S i from t h e elemental source was removed i n a CF4 plasma, l e a v i n g a smooth GaAs s u r f a c e . GaAs wafers t h a t were cleaned and

Secondary Ion Mass Spectroscopy (SIMS) was used t o measure t h e chemical p r o f i l e s of t h e d i f f u s e d Si. The e l e c t r i c a l prof i l e s were determined u s i n g t h e d i f f e r e n t i a l van d e r Pauw technique and chemical s t r i p p i n g . Annealing of both S i proximity and Si3Nq-encapsulated samples However, r e s u l t e d i n no measurable e l e c t r i c a l c o n d u c t i v i t y . samples encapsulated w i t h Si02 showed c o n s i d e r a b l e c o n d u c t i v i t y These r e s u l t s w i t h s h e e t r e s i s t a n c e s a s low a s 50 ohmslsquare. a r e summarized i n Table 7.3. A similar encapsulant e f f e c t has Electribeen observed f o r low dose S i implants i n G a A s 17.581. c a l p r o f i l i n g of t h e Si02 encapsulated samples showed an abru t j u n c t i o n and a maximum e l e c t r o n c o n c e n t r a t i o n of 5-6x1018/cm The s u r f a c e c o n c e n t r a t i o n of S i a s measured by SIMS was 2x1020/ cm3 and a sharp d i f f u s i o n f r o n t was observed. The SIMS and e l e c t r i c a l p r o f i l e s f o r a 3 sec anneal at 105OoC a r e shown i n Fig. 7.31. These measurements i n d i c a t e n e a r l y complete compenWe w i l l u s e t h i s amphoteric c h a r a c t e r of S i i n s a t i o n of S i . t h e d i f f u s i o n model.

3.


438

TABLE 7.3. Summary of E l e c t r i c a l Measurements For Encapsulated S i D i f f u s i o n Using Rapid Thermal P r o c e s s i n g

Diffusion

850 900 9 50 1000 1050

30 0 20.5 15.5 10.5 10 .o

Si02

van d e r Pauw

60.4 169.3 97.1 70.7 51.3

1119 954 884 848 777

9.3 3.9 7.3 10.4 15.7

DEPTH ( p m ) FIGURE 7.31. SIMS and e l e c t r i c a l (van d e r Pauw) p r o f i l e s f o r S i d i f f u s e d i n t o G a A s a t 105OoC f o r 3 sec under 0.5 Urn Si02 encapsulation. Mechanisms f o r S i d i f f u s i o n t h a t i n v o l v e motion of i s o l a t e d s i l i c o n atoms through Ga and A s v a c a n c i e s r e q u i r e changes i n vacancy t y p e and charge s t a t e t h a t would presumably produce v e r y low d i f f u s i o n c o e f f i c i e n t s . However, t h e d i f f u s i o n of a n e a r e s t neighbor ( S i - S i ) p a i r could be much s i m p l e r . P a i r e d S i atoms can move s u b s t i t u t i o n a l l y by exchanging s i t e s w i t h e i t h e r a G a o r A s vacancy. This is e x p r e s s e d by

7.


439

w i t h a s i m i l a r e x p r e s s i o n f o r t h e exchange w i t h an A s vacancy. This mechanism conserves both charge and vacancy type. G r e i n e r and Gibbons model t h e d i f f u s i o n of S i i n G a A s by assuming t h a t S i - S i n e a r e s t neighbor p a i r s ( r e f e r r e d t o a s S i complexes) a r e t h e main d i f f u s i n g s p e c i e s i n h e a v i l y Si-doped GaA s. I f e q u i l i b r i u m e x i s t s between s i n g l e and complexed S i , t h e f r a c t i o n of S i e x i s t i n g as complexes can be obtained from The mass-action r e l a t i o n between p a i r e d t h e l a w of mass-action. and s i n g l e S i atoms i s

where t h e s u b s c r i p t concentration.

indicates

site location

and

[

] denotes

E l e c t r i c a l and SIMS measurements i n d i c a t e t h a t heavy S i dopitpj (l(j20/cm3) r e s u l t s i n almost complete compensation ( n = 5x10 /cm ). Using t h e approximation t h a t [Sica] = [ s i i s ] , t h e c o n c e n t r a t i o n of S i complexes i n terms of t h e t o t a l S i concentrat i o n , [ S i ] can be expressed a s

+

[sic+a-siisl

= 1/2

Here w e have assumed t h a t only s i n g l e and complexed s u b s t i t u t i o n a l S i exists. For [ S i ] = K, 27% of t h e S i is p r e d i c t e d t o be complexed.

The e q u i l i b r i u m c o n s t a n t f o r S i complex formation, K, can be e s t i m a t e d by c a l c u l a t i n g t h e p r o b a b i l i t y of f i n d i n g S i atoms r e s i d i n g on a d j a c e n t l a t t i c e s i t e s w i t h an added term c o n t a i n i n g t h e energy of formation of t h e complex I7.591. The energy of coulombic i n t e r a c t i o n from an i o n i z e d S i donor and a c c e p t o r on a d j a c e n t s i t e s i s taken as t h e energy of complex formation. A The r e s u l t i s n e a r e s t neighbor spacing of 2.45 fi was used [7.59]. K = N/Z exp(-.56eV/kT)

where N i s t h e d e n s i t y of t h e Ga o r As s u b l a t t i c e and Z i s t h e number of c o n f i g u r a t i o n s of t h e p a i r (Z = 4 ) . A t 1000°C, K = 8 . 3 ~ 1 O ~ ~ / c[7.57]. m~ Assuming t h a t S i complexes a r e t h e main d i f f u s i n g s p e c i e s , t h e equation f o r t h e t o t a l S i f l u x s i m p l i f i e s t o FluxSi = 2

D~~~~~~~

(a[si-sii/a[sii)

(a[sil/at)

440


The equation for the flux explicitly in terms of the total Si concentration is FiuxSi = Deff

(a[sii/at)

This effective diffusion coefficient is proportional to [Si] for [Si]K, representing the high and low concentration regimes. The diffusion equation was solved numerically assuming that (a) solid solubility of Si at the surface is maintained, and (b) the diffusion coefficient for the Si complexes, Dcom lex, is constant. The results of the calculations are shown in fig. 7 .32 along with the SIMS data for a 1050°C, 3 sec diffusion.

To demonstrate the diffusion of amphoteric pairs in GaAs, silicon and germanium were co-diffused from a thin film Si-Ge alloy. A Si.gGe.4 alloy was substituted for the thin Si layer used previously and diffused under identical conditions.

'""I

A SIMS

DEPTH

(i)

FIGURE 7 3 2 . Calculated Si profile using the complexed Si diffusion model and fitted to the SIMS data.

7.


441

The SIMS profiles for Si and Ge co-diffused at 105OoC for 100 sec are shown in Fig. 7.33. The separate profiles are seen to be the same within the resolution of SIMS. This is the expected result based on the formation and diffusion of Si-Ge pairs. For comparison, the co-diffusion of Si and Ge indicates a diffusion coefficient ten times lower than that for Si diffusion alone. 7.5.2

Schottky Barrier Modifications

Rapid thermal processing can also be used to modify the properties of Schottky barrier contacts in a way that is of potential importance in GaAs devices and integrated circuits. The experimental procedure builds on an idea that has been used to improve GaAs liquid junction solar cells. Parkinsen et al. [7.60] have shown that hole lifetime at a GaAs surface can be improved by dipping the sample into a Ru Cl3 solution. The Ru is thought to bond preferentially to As surface atoms atoms leaving Ga surface atoms open to further surface chemistry. This then provides the possibility of plating Zn onto the GaAs with the Zn atoms on the surface bonding to Ga

Ge Si

1 0.1 0.2 0.3 0.4 C 5

1015

DEPTH

FIGURE 7.33. 105OoC for 10 sec.

(pm)

SIMS profiles for Si and Ge co-diffused at


442

atoms. Experimentally this technique is found to provide a means of plating a uniform 100 A layer of Zn onto the GaAs. Highly nonuniform plating results if the Ru treatment is omitted. If this sample is then heated to 500-600°C for 25-50 seconds in an evacuated chamber, all of the Zn appears to evaporate except for about one monolayer bonded to the surface. A subsequent diffusion of the residual Zn into the GaAs permits carrier density profiling experiments which show the total residual Zn doping to be constant at a level of 4-8x1014/cm2, corresponding approximately to 1/2 monolayer of Zn. If the samples are removed from the annealer prior to the diffusion cycle, Schottky barriers can be made with a standard A1 evaporation. The properties of such barriers are of course modified by the presence of the thin, Zn-doped surface layer. Barrier height measurements and diode ideality factors are shown in Table 7.4 for barriers made in this way. The results show that the Zn layer increases the Schottky barrier height by nearly 0.3 volt. FETs made with such improved barrier heights would provide significant improvement of the noise margins in GaAs digital integrated circuits.

7.5.3

Summary

Rapid thermal processing can provide efficient annealing of ion implanted layers in GaAs with properties that are at least promising for device applications. The technique can also be used to provide controlled diffusion from oxide-encapsulated, surface-deposited dopants and, €or elements with high vapor pressure, a technique for monolayer-thickness doping. The properties of such surfaces may be of importance in device fabrication in situations where modified Schottky barrier properties are required or implantation doping is to be avoided. TABLE 7.4. Surfaces Plating Time (sec) 0 120 20 20

Schottky Barrier Properties for Zn-treated GaAs

Evaporation Time (sec) 0 25 25

45

Evap. Temp."C 0 580 500 5 00

4 ,v (C-V)

+(14) ,V

0.66 0.97 0.95 1.08

0.72 0.92 0.89 0.97

n 1.o

1.17 1.2 1.2

7.


443

REFERENCES 7.1

Immorlica, A. A., and Eisen, F. H., Appl. Phys. Lett. 29, 94 (1976).

7.2

7.3

7.4

7.5 7.6 7.7 7.8 7.9

Fan, J. C. C., Donnelly, J. P., Bozler, C. O., and Chapman, R. L., Proc. 7th Intern. Symposium on GaAs and Related Institute of Physics, London (1979). Compounds, p. 472. Olson, G. L., Anderson, C. L., Dunlap, H. L., Hess, L. D., McFarlane, R. A., and Vaidyanathan, K. V., in "Laser and Electron Beam Processing of Electronic Materials", (C. L. Anderson, G. K. Celler and G. A. Rozgoni, eds.) p. 467. ECS, Princeton (1980). Nissim, Y. I., and Gibbons, J. F., in "Laser and Electron Beam Solid Interactions and Materials Processing" (J. F. Gibbons, L. D. Hess and T. W. Sigmon, eds.) p. 275. North Holland (1981). MacMahon, R. A., Ahmed, H., Dobson, R. M., and Speight, J. D., Elect. Lett. 16, 295 (1980). Shah, N. J., Ahmed, H., Sanders, I. R., and Singleton, J. F., Elect. Lett. 16, 433 (1980). Csepregi, L., Mayer, J. W., and Sigmon, T. W., Phys. Lett. 54A, No. 2, 157 (1975). Gamo, K., Ineda, T., Mayer, J. W., Eisen, F. H., and Rhodes, C. G., Rad. Effects 33, 85 (1977). Williams, J. S . , and Austin, M. W., Appl. Phys. Lett. 36, 994 (1980).

7.10

7.11 7.12 7.13

Williams, J. S. and Harrison, H. B., in "Laser and Electron Beam Solid Interaction and Materials Processing" (J. F. Gibbons, L. D. Hess and T. W. Sigmon, eds.) p. 209. North Holland (1981). Kular, S . S . , Sealy, B. J., Stephens, K. G., Sadana, D. K., and Pooker, G. R., Solid State Elec. 23, 831 (1980). Grimaldi, M. G., Paine, B. M., Nicolet, M. A., and Sadana, D. K., J. Appl. Phys. 52, 4038 (1981). Nissim, Y. I., Christel, L. A., Sigmon, T. W., Gibbons, J. F., Magee, T. J., and Ormond, R., Appl. Phys. Lett. 39, 598 (1981).

7.14 7.15 7.16

Csepregi, L., Kennedy, E. F., Mayer, J. W., Sigmon, T. W., J. Appl. Phys. 49, 3906 (1978). Christel, L. A., Gibbons, J. F. and Mylroie, S . , J. Appl. Phys. 51, 6176 (1980). Christel, L. A. and Gibbons, J. F., J. Appl. Phys. 52, 5050 ( 1980).

7.17

7.18

Williams, J. S . , Austin, M. W., and Harrison, H. B., in "Thin Film Interfaces and Interactions" (J. E. E. Baglin ECS, Princeton (1980). and J. M. Poate, eds.) p. 187. Braslaw, N., Gunn, J. B., and Staples, J. L., Solid State Electron. 10, 3 8 1 (1967).

444 7.19 7.20 7.21 7.22 7.23 7.24

7.25 7.26

7.27

7.28 7.29 7.30 7.31 7.32 7.33 7.34 7.35 7.36 7.37 7.38 7.39 7.40 7.41 7.42

YVES I. NISSlM AND JAMES F. GIBBONS

Pang, Y. K., and Sze, S . M., S o l i d S t a t e Chang, C. Y., E l e c t r o n . 1 4 , 541 (1971). Robinson, G. Y., S o l i d S t a t e E l e c t r o n . 18, 331 (1975). Solid S t a t e Electron. 22, 141 (1979). C h r i s t o u , A., Yokoyama, N., Ohkawa, S . , and Ishikawa, H., Jap. J . Appl. Phys. 14, 1071 (1975). Pounds, R. S., S a i f i , M. A , , and Hahn, W. C., Jr., S o l i d S t . E l e c t r o n . 17, 245 (1974). Gold, R. B., Powell, R. A., and Gibbons, J. F., i n "LaserS o l i d I n t e r a c t i o n s and Laser P r o c e s s i n g " ( S . D. F e r r i s , H. J . Leamy, and J. M. P o a t e , eds.) p. 635. AIP, New York, (1979). Gold, R. B., Ph.D. T h e s i s , S t a n f o r d U n i v e r s i t y , Department of E l e c t r i c a l Engineering (1980). i n "Laser and E l e c t r o n B e a m P r o c e s s i n g of Eckhardt, G., Materials" (C. W. White and P. S . Peercy, eds.) p. 467. Academic Press (1980). and Krumm, C. Eckhardt, G., Anderson, C. L., Hess, L.D., F., i n "Laser-Solid I n t e r a c t i o n s and Laser P r o c e s s i n g " ( S . D. F e r r i s , H. J. Leamy, and J. M. P o a t e , eds.) p. 641. AIP, New York (1979). Emulsitone CO., 19 L e s l i e Court, Whippany, N. J . Van Much, W., IEM J . R e s . Dev. 10, 438 (1966). Gibbon, C. F., and Ketchow, D. R., J. Electrochem. SOC. 118, 915 (1971). Fujimoto, M., and Kudo, K., Yamazaki, H., Kowasaki, Y., Jap. J. Appl. Phys. 1 4 , 7 1 7 (1975). Arnold, N. Daemkes, H., and H e i m e , K., Jap. J. Appl. Phys. 19, Supplement 19.1, 361 (1.980). S z e , S. M., "Physics of Semiconductors Devices", p. 38. Wiley, New York (1969). N i s s i m , Y. I., Gibbons, J. F., Evans, C. A., Jr., D e l i n e , V. R., and Norberg, J. C., Appl. Phys. Lett. 37, 89 (1980). N i s s i m , Y. I., Gibbons, J. F., Magee, T. J., and Ormond, R., J. Appl, Phys. 52, 227, (1981). N i s s i m , Y. I., Gibbons, J. F., and Gold, R. B., IEEE Trans. E l e c t r o n . Dev. ED-28, 607 (1981). IEEE Trans. E l e c t r o n . Dev. Murrmann, H. and Widmann, D., ED-16, 1022 (1969). H., Solid St. Electron. 15, 145 (1972). Berger, H. " C o n s t i t u t i o n of Binary Alloys", p. 181. Hansen, H., M c G r a w - H i l l (1958). C h r i s t o u , A., S l e g e r , K., Proc. 6 t h B i e n n i a l C o r n e l l , E l e c . Engr. Conf., 169 (1977). Macksey, H. M., i n "Gallium Arsenide and R e l a t e d Compounds" (L. F. Eastman, ed.) p. 254, I n s t . of Phys. Conf. S e r . 33b, London (1977). N i s s i m , Y. I., Gibbons, J. F., Gold, R. B., and Dobkin, D. M., ECS Meeting Hollywood, F l o r i d a (1980).

7. 7.43

7.44

7.45 7.46 7.47

7.48 7.49 7.50


445

Sealy, B. J. and Surridge, R. K., IBBM-78, Conference Proceedings (J. Gyulai, T. Lohner, and E. Pgsztor, eds.) pp. 487-495. Budapest ( 1 9 7 9 ) . Williams, J. S., "Laser-Solid Interactions and Transient Thermal Processing of Materials," (J. Narayan, W. L. Brown and R. A. Lemons, eds.), p. 621. North-Holland ( 1 9 8 3 ) . Arai, M., Nisibiyama, K., and Watanabe, N., Jap. Int. Appl. Phys. 20, L 124 ( 1 9 8 1 ) . Kuzuhara, M., Kohzu, H., and Takayama, Y., Appl. Phys. Lett. 4 1 , 755 ( 1 9 8 2 ) . Rose, A., Pollak, J. T. A., Scott, M. D., Adams, F. M., Williams, J. S . , and Lawson, E. M., '"'Laser-Solid Interactions and Transient Thermal Processing of Materials," (J. Narayan, W. L. Brown, and R. A. Lemons, eds.), pp. 633-639. North-Holland ( 1 9 8 3 ) . Davies, D. E., McNally, P. J., Lorenzo, J. P., and Julian, M., IEEE Elec. Dev. Lett. EDL-3, 102 ( 1 9 8 2 ) . Tabatabaie-Alavi, K., Masum Choudhury, A. N. M., Fonstad, C. G., and Gelpey, J. C., Appl. Phys. Lett. 4 3 , 5 ( 1 9 8 3 ) . Christel, L. A., and Gibbons, J. F., J. Appl. Phys. 5 2 , 8 (1981).

7.51 7.52 7.53 7.54 7.55

Kuzuhara, M., Kohzu, H., and Takayama, Y., Appl. Phys. Lett. 4 1 , 755 ( 1 9 8 2 ) . Davies, D. E., McNally, P. J., Lorenzo, J. P., and Julian, M., IEEE Elec. Dev. Lett. EDL-3, 102 ( 1 9 8 2 ) . Chapman, R. L., Fan, J. C. C., Donnelly, J. P., and Tsaur, B-Y., Appl. Phys. Lett. 4 0 , 805 ( 1 9 8 2 ) . Tabatabaie-Alavi, K., Masum Choudhury, A. N. M., Fonstad, C. G., and Gelpey, J. C., Appl. Phys. Lett. 4 3 , 505 ( 1 9 8 3 ) . Tabatabaie-Alavi, K., Masum Choudhury, A. N. M., and Kanbe, H., Fonstad, and Gelpey, J. C., Appl. Phys. Lett. 4 3 , 647 (1983).

7.56

Suzuki, T., Sakurai, H., and Arai, M., Appl. Phys. Lett. 43, 951 (1983).

7.57 7.58

Greiner, M. E., and Gibbons, J. F., Appl. Phys. Lett., 15 April 1983 (to be published). Onuma, T., Hirao, T., and Sugawa, T., J. Elec. SOC. 129, 4 , 837 ( 1 9 8 2 ) .

7.59 7.60

Kroger, F. A., "Chemistry of Imperfect Crystals," p. 275. North Holland ( 1 9 6 4 ) . Parkinsen, B., Heller, A., and Miller, B., Appl. Phys. Lett. 3 3 , 6 , 521 ( 1 9 7 8 ) .


Index As+-implanted Si, C>CSs ,

Beam recrystallization effect of oxygen, 328 effect of scan rate, 327 geometric techniques for improving crystallinity, 333 large area, 311 miscellaneous defects, 330 origin of structures in zone recrystallized films, 321 seeded thick films, 309 strip heater for recrystallization of polysilicon films, 312 structure of zone recrystallized films, 313 Carrier mobility in laser recrystallized polysilicon films, 249 CMOS circuits on beam recrystallized polysilicon films, 269 CW beam annealing basic mechanisms o f , 34 impurity profiles in, 35 i o n implanted silicon, central features of, 30 CW beam processing reaction rate calculations, 1 01 CW beam recrystallization of polysilicon films basic process, 47 graphite strip heater, 49 recrystallization process improvements, 48 three dimensional integration, 50 CW laser alloying of Au-GefGaAs contacts,.411 annealing. - _ 1 07ff As+-implanted Si, C

Semiconductors and Semimetals, Vol. 17: C W Beam Processing of Silicon and Other Semiconductors

Semiconductors and Semimetals Volume 72 Silicon Epitaxy (Semiconductors and Semimetals)

Semiconductors and Semimetals Volume 72 Silicon Epitaxy (Semiconductors and Semimetals)

CW beam processing of silicon and other semiconductors

SEMICONDUCTORS and SEMIMETALS V23 Pulsed laser processing of semiconductors

Semiconductors and Semimetals, Volume 21, Part A: Hydrogenated Amorphous Silicon : Preparation and Structure (Semiconductors and Semimetals)

SEMICONDUCTORS & SEMIMETALS V38, Volume 38 (Semiconductors and Semimetals)

Semiconductors and Semimetals: Defects, HgCd, Se, HgCdO and Te (Semiconductors and Semimetals, Vol. 16)

Semiconductors and Semimetals, Volume 67: Ultrafast Physical Processes in Semiconductors (Semiconductors and Semimetals)

Semiconductors and semimetals: Indentification of defects in semiconductors

The Spectroscopy of Semiconductors: Volume 36 (Semiconductors and Semimetals)

Semiconductors and Semimetals v. 4

Semiconductors and Semimetals Infrared Detectors

Semiconductors and Semimetals. Volume 25: Diluted Magnetic Semiconductors (Semiconductors and Semimetals)

Semiconductors and Semimetals, Volume 67: Ultrafast Physical Processes in Semiconductors (Semiconductors and Semimetals)

Semiconductors and Semimetals, Volume 75 Laser Crystallization of Silicon

Semiconductors and Semimetals, Volume 51A Identification of Defects in Semiconductors

SEMICONDUCTORS & SEMIMETALS

SEMICONDUCTORS & SEMIMETALS

Semiconductors and Semimetals (Vol. 21: Hydrogenated Amorphous Silicon, Part C: Electronic and Transport Properties)

SEMICONDUCTORS & SEMIMETALS

SEMICONDUCTORS & SEMIMETALS

Germanium Silicon: Physics & Materials (Semiconductors and Semimetals, Volume 56)

Oxygen in Silicon, Volume 42 (Semiconductors and Semimetals)

Processing and Properties of Compound Semiconductors (Semiconductors and Semimetals, Volume 73)

Germanium Silicon: Physics & Materials (Semiconductors and Semimetals, Volume 56)

Hydrogen in Semiconductors II, Volume 61 (Semiconductors and Semimetals)

Semiconductors and Semimetals, Volume 59 Nonlinear Optics in Semiconductors II

Chemical Mechanical Polishing in Silicon Processing, Volume 63 (Semiconductors and Semimetals) (Vol 63)

Hydrogen in Semiconductors (Semiconductors and Semimetals, Volume 34)

Conducting Organic Materials and Devices, Volume 81 (Semiconductors and Semimetals) (Semiconductors and Semimetals)

Semiconductors and Semimetals, Vol. 17: C W Beam Processing of Silicon and Other Semiconductors

Semiconductors and Semimetals Volume 72 Silicon Epitaxy (Semiconductors and Semimetals)

Semiconductors and Semimetals Volume 72 Silicon Epitaxy (Semiconductors and Semimetals)

CW beam processing of silicon and other semiconductors

SEMICONDUCTORS and SEMIMETALS V23 Pulsed laser processing of semiconductors

Semiconductors and Semimetals, Volume 21, Part A: Hydrogenated Amorphous Silicon : Preparation and Structure (Semiconductors and Semimetals)

SEMICONDUCTORS & SEMIMETALS V38, Volume 38 (Semiconductors and Semimetals)

Semiconductors and Semimetals: Defects, HgCd, Se, HgCdO and Te (Semiconductors and Semimetals, Vol. 16)

Semiconductors and Semimetals, Volume 67: Ultrafast Physical Processes in Semiconductors (Semiconductors and Semimetals)

Semiconductors and semimetals: Indentification of defects in semiconductors

The Spectroscopy of Semiconductors: Volume 36 (Semiconductors and Semimetals)

Semiconductors and Semimetals v. 4

Semiconductors and Semimetals Infrared Detectors

Semiconductors and Semimetals. Volume 25: Diluted Magnetic Semiconductors (Semiconductors and Semimetals)

Semiconductors and Semimetals, Volume 67: Ultrafast Physical Processes in Semiconductors (Semiconductors and Semimetals)

Semiconductors and Semimetals, Volume 75 Laser Crystallization of Silicon

Semiconductors and Semimetals, Volume 51A Identification of Defects in Semiconductors

SEMICONDUCTORS & SEMIMETALS

SEMICONDUCTORS & SEMIMETALS

Semiconductors and Semimetals (Vol. 21: Hydrogenated Amorphous Silicon, Part C: Electronic and Transport Properties)

SEMICONDUCTORS & SEMIMETALS

SEMICONDUCTORS & SEMIMETALS

Germanium Silicon: Physics & Materials (Semiconductors and Semimetals, Volume 56)

Oxygen in Silicon, Volume 42 (Semiconductors and Semimetals)

Processing and Properties of Compound Semiconductors (Semiconductors and Semimetals, Volume 73)

Germanium Silicon: Physics & Materials (Semiconductors and Semimetals, Volume 56)

Hydrogen in Semiconductors II, Volume 61 (Semiconductors and Semimetals)

Semiconductors and Semimetals, Volume 59 Nonlinear Optics in Semiconductors II

Chemical Mechanical Polishing in Silicon Processing, Volume 63 (Semiconductors and Semimetals) (Vol 63)

Hydrogen in Semiconductors (Semiconductors and Semimetals, Volume 34)

Conducting Organic Materials and Devices, Volume 81 (Semiconductors and Semimetals) (Semiconductors and Semimetals)

Recommend Documents