Dislocations at lattice mismatched widegap II-VI/GaAs heterointerfaces as laser light scatterers

MISSING IMAGE

Material Information

Title:
Dislocations at lattice mismatched widegap II-VI/GaAs heterointerfaces as laser light scatterers experiment and theory
Physical Description:
xv, 83 leaves : ill. ; 29 cm.
Language:
English
Creator:
Rouleau, Christopher M., 1966-
Publication Date:

Subjects

Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1994.
Bibliography:
Includes bibliographical references (leaves 79-82).
Statement of Responsibility:
by Christopher M. Rouleau.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002018859
notis - AKK6295
oclc - 32769129
System ID:
AA00003224:00001

Full Text







DISLOCATIONS AT LATTICE MISMATCHED WIDEGAP II-VI/GaAs
HETEROINTERFACES AS LASER LIGHT SCATTERERS:
EXPERIMENT AND THEORY













By

CHRISTOPHER M. ROULEAU


DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY





























Dedicated to my father












ACKNOWLEDGEMENTS


I would like to express my sincere appreciation to Professor Robert M.


Park, my supervisory


committee chairman and dissertation advisor, for his


constant


professional


guidance,


support,


encouragement


during


research


studies.


The


knowledge


skills


that


shared


greatly


enhanced my learning experience.

I would also like to thank Drs. Paul H. Holloway, Joseph H. Simmons,


Kevin S. Jones, and David B.


Tanner for serving as my doctoral committee.


Their assistance and interest in my work are gratefully appreciated.


I am also indebted to Christopher


Santana,


colleague,


whose


expertise


transmission


electron


microscopy


sample


preparation


analysis was instrumental in developing this dissertation.

Many thanks are also due to Dr. James V. Masi, professor of Electrical


Engineering at


Western New


England


College,


who, as my undergraduate


professor and mentor, encouraged me always to "keep the spirit and curious

mind."







A special word of thanks is also extended to my best friend and wife,


Mary


who


provided


support


encouragement throughout my


graduate


studies.














TABLE OF CONTENTS


ACKNOWLEDGEMENTS

LIST OF TABLES.....

LIST OF FIGURES

ABSTRACT


. S C S S S S S S S S a a S S S S S S S


a. S S S S S S S S S S S 9 5 S a S S S S S S 1jj


CHAPTERS


INTRODUCTION


Background and Motivation .
Outline of Dissertation Contents


* . . . a a a
* * . . S .


2 GaAs SUBSTRATE CLEANING FOR HETEROEPITAXY


Background.
Experimental
MBE S)
Charact
UHV C
Precision
In Situ GaAs i


Apparatus 7
stem Configuration . . . . . 7
erization of Optical System . . .. . 10
compatible rf Plasma Source ............... 12
n Substrate Temperature Calibration ......... 21
Surface Cleaning . . . . . 25


Conclusions


1 TA OTD T T/-TAT'T


D8~Ze


C1~ ~ r~eb T~T/1 I


~~m~hr~


.


.








33


Experimental Apparatus
MBE Growth Chambel
Laser Probe Apparatus
Heteroepitaxy
Particle Contamination
Ex Situ Substrate Prep
Laser Light Scattering


Laser Light Scattering
Structural Characterization.
Lattice Mismatch Sensitivity
Conclusions . . .


* C . . 34
r....................... 34

37
Considerations . . 37
aration . 40
from Typical Epilayers .. 41
from Very Thick Epilayers .... 47
.51
C C a . . . 579
. . . . . . 59


4 THEORY


The Role of Electric Microfields


Deformation Potential . . . . ... 61
Charged Dislocation Potential . . . 64
Piezoelectric Potential . . . .. . . 66
The Linear Electro-optic Effect. . . . . 71
Conclusions .. 73


5 CONCLUSIONS . . . . . . . .



BIOGRAPHICAL SKETCH . . . ... . . .


Background.












LIST OF TABLES


IaWle


uage


ZnSe


material


parameters


necessary


evaluate


coefficient of the microelectric field induced by the deformation


potential (Eq.


S. 0 0 65


ZnSe


material


coefficient
dislocation


parameters


of the m
potential


necessary


licroelectric


(Eq.


field
GaAs


evaluate


induced
material


the
the (


scalar
chargedd


parameters


used in some cases (denoted by a superscript a) since values for
ZnSe were unavailable.. .


scalar












LIST OF FIGURES


Figure


Dage


Schematic


custom-designed


molecular


beam


epitaxy


system showing conventional effusion sources, reflection high


energy


electron


diffraction


system,


ultra-high


vacuum


compatible rf plasma source fitted with a precision leak valve
and gas line for hydrogen. Also illustrated is a schematic of the
experimental setup used to characterize the surface morphology


of GaAs during
light (or lack 1


geometry


cleaning by recording the scattered HeNe laser


thereof)


indicated


during the


in the


figure


cleaning
provides


procedures.


The


simultaneous


reflection high energy electron diffraction analysis as shown.


. 9


Schematic


apparatus


employed


characterize


camera-based detection system transfer function at 632.8nm.


. 1


Camera-based


detection


system


transfer


function


recorded


632.8nm.


Schematic of the ultra-high vacuum compatible rf plasma source


used to generate a flux of atomic hydrogen.


conditions were 320W


Typical operating


forward power (<2W reflected) with an


equilibrium background pressure of H in the
maintained at 5xlO"mbar.


growth chamber


Schematic of the experimental setup used to record the optical
emission spectrum associated with the hydrogen plasma created
in the discharge region of the ultra-high vacuum compatible rf


nl ncm cnnrrp indilata, in ;irra


1 arnA A


u








characteristic optical


emission


spectrum recorded


from


hydrogen plasma region of the ultra-high vacuum compatible rf
plasma source illustrating emission lines characteristic of atomic


hydrogen.


The rf power used to create the discharge was 320W


reflected) while an equilibrium


hydrogen


in the


molecular


beam


background pressure of


epitaxy


growth


chamber


5xlOmbar was maintained.


Inset shows an emission spectrum


from an ECR plasma source (from Sugata et al.1).


Evolution
recorded


vacuum


Spectrum


characteristic


from


compatible


was


hydrogen


plasma


recorded


optical


plasma


source


days


emission


region


as a function


after


that


spectrum
ultra-high


time.


Fig.


Spectrum (b) was recorded after evacuating the H feed line with


a pump for


minutes and then back-filling the line with H.


Spectrum (c) was recorded 24 hours later and spectrum (d) was


recorded


after


several times.


evacuating


back-filling


feed


line


Note: the source was on during recording of the


spectra and off otherwise.


. . . 20


Schematic of the experimental setup used to calibrate precisely
the substrate temperature by recording the change in scattered
HeNe laser light corresponding to the formation of a eutectic


compound from selected binary alloys.


22


plot


temperature
formation.


substrate surface
showing the


temperature


points


versus


associated


thermocouple


with


eutectic


Inset indicates that there was no more then +1/-4%


error


(normalized


regime explored.


T/C


temperature)


over


temperature


S24


A plot of the intensity of laser light scattering versus substrate


temperature (ramp rate


100C/minute).


The data was recorded


during conventional thermal treatment and during a combined


thermal/H-atom
resnectivelv


treatment


(001)GaAs


wafers,


1/;








Representative


reflection


high


energy


electron


patterns recorded from atomically clean GaAs surfaces resulting


from


conventional


thermal


treatment


thermal/H-atom treatment and after removal of the H-atom flux,
(a), (b), and (c), respectively. .. .. . . . .


Schematic of the experimental setup used to record elastically
scattered laser light (ELLS) and specularly reflected laser light


during


molecular


beam


VI/GaAs heterostructures.


epitaxial


growth


widegap


4 a S S 4 4 5 5 5 4 4 4 5 5 35


Dark-field 1
densities res
continuous (
respectively.
micrographs.


optical
sulting


micrographs


from


seconds)


cleaving,


flow


indicating


burst


cleaning,


typical


particle


cleaning
(b). and


Note: the cleave line runs horizontally through the
S. 39


Intensity
supplied


scattering
American


occurring


Xtal


near


Technology)


ZnSe/GaAs (GaAs
heterointerface as a


function of ZnSe epitaxial deposition time and layer thickness.
Also included in the figure is a laser reflection interferogram to
compare signal phasing. . . . . . . .


Intensity


scattering


occurrmng


near the


ZnSe/GaAs


(GaAs


supplied by Sumitomo Electric) heterointerface as a function of


ZnSe epitaxial
included in the


deposition
figure is


tunme
a laser


nd layer
reflection


thickness.


Also


interferogram


compare signal phasing.


A ray model illustrating the employment of a HeNe laser beam


to probe
deposition.


ZnSe/GaAs


Defect evolution


heterointerface


gives


during


a scattered


epitaxial


signal


which was detected and quantified using the apparatus shown


in Fig.


C . S C S C S 4 4 C S S S S C S S C S C


a aa 4.aa -


*uFlll Ez* *w* I' S ** Ell *r .rln .a u n n .E f'ffUnlf 'V fff a i* ar. a n S .


1r~*4n~


dif~f~raction


combined


n.. nrnr


/n


rr~luu -








thick


30pm)


film.


Also


included


in the


figure


is a laser


reflection interferogram.
this case.


Note that the signals are in phase for
. . a a a a a a a a a a .


Plan-view


scanning


electron


micrographs


ZnSe/GaAs)


indicating the surface features present on a thin (-60nm), thick


-840nm)


very


thick


30pm)


film,


respectively.


. a a a a a a a a a


matrix


plan-view


transmission


(70,000X) comparing and contrasting the defects present near
the ZnSe/GaAs heterointerface as a function of film thickness


micrographs


(row


through


= 60,


120,


300,


360,


respectively) and substrate type (column


600,


Xtal Technology and Sumitomo Electric, respectively).

Recompilation of laser scattering data presented in Figs.


840nm,


= ~Amei


52

14 and


15 which compares and contrasts the laser scattering from an
epilayer grown on an American Xtal Technology GaAs wafer


that


wafer.


an epilayer


grown


on a Sumitomo


Epilayers of the thicknesses


indicated


Electric


by the


arrows were targeted for structural characterization.


GaAs


vertical


difference in the detectable onset (in terms of layer thickness)


of scattering.


S* S S a a a S S a a a a a a .. . 54


representative


resolution


selected


cross-sectional


area


diffraction


transmission


pattern


electron


and high-
micrograph


which are indicative of the crystallinity and defect type present
in the epilayers represented by the micrographs shown in Figs.


19(g)-(i).


. a a a a a a a a S S S S S S S S S a a a a S S


Elastically scattered laser light data recorded in real-time during
the deposition of ZnSe/GaAs and ZnTe/GaAs heterostructures.


The


room


lattice-mismatches


temperature


indicated


lattice-mismatches.


figure


Note


correspond


the difference


detectable onset (in terms of laver thicknlma nf crnttprin


electron


and 2


Note the


CI1








cartesian


showing


mesh


plot


electric


function


field


intensity


presented in Eq. 1
to the deformation


potential as a function of position around the dislocation core.
The dislocation core lies along the electric field intensity axis.


cartesian


mesh


plot


function


presented


showing the electric field intensity due to a charged dislocation


as a function


position


around


dislocation


core.


The


dislocation core lies along the electric field intensity axis.


cartesian


showing


mesh


plot


piezoelectric


function


field


position around the dislocation core.
along the electric field intensity axis.


mtensity


presented


function


The dislocation core lies
. S S S .












Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


DISLOCATIONS AT LATTICE MISMATCHED


WIDEGAP II-VI/GaAs


HETEROINTERFACES AS LASER LIGHT SCATTERERS:


EXPERIMENT


Christopher M.


August,


Chairperson: Robert M.


Park


AND THEORY


Rouleau


1994


Major Department: Materials Science and Engineering

This work concerned an investigation using real-time in situ electron

and laser beam probes of GaAs substrate surfaces intended for epitaxy and


strain


relief


process


associated


with


growing


ZnmS cGaAs


heterostructures by the molecular beam epitaxy technique.


GaAs


surface


cleaning


process


was


developed


using


remote H atom flux supplied by an ultra-high vacuum compatible rf plasma


source.


GaAs surface cleanliness and morphology were monitored in situ


and in ralnt-time nqino rtflprtinn hioh i npirov plpstrnn diffimtinn (RHFr.FTn








From RHEED observations, GaAs surfaces were found to clean readily


at temperatures near


350C using the


H atom treatment as opposed to the


conventional


thermal


treatment,


which requires


temperatures


near


6000C.


The


atomically


clean


GaAs


surfaces


were


found


specular


ELLSS


observations)


when


prepared


using


atom


treatment,


hydrogen


stabilization of the surface is suggested from the RHEED observations.

In terms ofheteroepitaxial growth, a near-normal incidence HeNe laser


probe was


employed


that generated both a specularly


reflected laser light


signal


an ELLS


signal


beyond


a critical


epilayer thickness.


was


determined that the ELLS


signal


was


generated locally at the ZnSe/GaAs


heterointerface, as opposed to the free-surface, based on the observation of


a it phase


difference


between


two


signals,


which


were


monitored


simultaneously.


postulated,


based


theoretical


model,


that


dislocations arising at the ZnSe/GaAs heterointerface can act as laser light

scatterers due to strong microelectric fields that surround the defects and that

perturb the refractive index in a localized manner.

Postgrowth, ex situ transmission electron microscopy was performed

on a selection of heterostructures, and a high degree of correlation was found







defect structure present near the heterointerface,


which strongly supports the


theoretical model.












CHAPTER 1
INTRODUCTION


Background and Motivation


The quality


of a heteroepitaxial structure,


which for the purposes of


work


is a system


comprising


an interface


between


two


dissimilar


materials,


namely


a film


material


a substrate


material,


critically


dependent upon the nature of the substrate surface prior to epitaxy and the


degree


of lattice


parameter


thermal


expansion


coefficient


mismatch


between the two materials.


Photonic devices are particularly sensitive to the


presence


structural


defects


that


arnse


as a consequence


nonideal


substrate surface preparation and/or significant lattice parameter and thermal

expansion coefficient mismatches.

Traditionally, the results of mismatched heteroepitaxy and/or nonideal

surface preparation are revealed by performing postgrowth characterization

of suitably processed samples using techniques such as transmission electron


mimrrnornr, QTIu nk hihronl'rulinn v_ rlA4rnnaa Car naInnna a.., rae a


U~~1Mt~U









characterization


techniques


cannot readily


distinguish


between


sources


defects (substrate surface irregularities, lattice mismatch strain,


etc.).


The


work of this dissertation focused on addressing independently the nature of


a substrate


surface and


issue


of plastic


deformation


a mismatched


heterointerface.


The


subject


material


system


concerned


ZnSe/GaAs


heterostructures.

The primary motivation of the project was to develop in situ optical

probes that could be used to assess wafer quality and to study strain relief


in real-time during epitaxial growth.


An objective was to correlate optical


data recorded during film growth with structural defect analysis data obtained


postgrowth.


Another issue of major concern in this project was to develop


an in situ cleaning technique for GaAs that would provide an ideal surface


epitaxy


namely


a contamination-free,


stoichiometric,


atomically


smooth surface.


Again, the idea was to develop an in situ probe that would


be capable of providing useful information


on the nature of the substrate


surface during in situ processing.


Real-time,


in situ monitoring


techniques,


general,


represent very










energy


electron


diffraction


(RHEED)


employed


on a routine


basis


determine


surface


reconstruction


assess


quality


substrate


surfaces and subsequent epilayers.


Recently, however,


optical


techniques


have become increasing popular due to their noninvasive nature and their


ability


provide


information


on properties


other


than


crystal


structure.


Ellipsometry, for instance, has far-reaching capabilities that include real-time,


monitoring


control


alloy


composition,


growth


rate,


substrate


temperature


during


MBE


deposition."-3


Also,


real-time,


momtormg


free


carter


concentration


doped


ZnSe


films


quantitative cathodoluminescence analysis has


been reported.4


In addition,


Lavoie et al.5 have reported real-time, in situ monitoring of the GaAs oxide


desorption process prior to epitaxial


growth


by measuring the intensity


elastically scattered laser light at the GaAs surface and have concluded that


there is


an associated roughening


of the GaAs


surface.


Also along these


lines, Pidduck et al.6'7 and Robbins et al.' have performed similar elastic laser


light scattering measurements


during


Si MBE


SiGe/Si


heteroepitaxy,


respectively


and have concluded


that changes


surface morphology


r








4
As will be described in the following chapters, we have developed our


own


optical


probing


methods


study


real-time


ZnSe/GaAs


heteroepitaxial


system,


will


demonstrated


that


heterointerface


defects,


rather


than


surface


irregularities,


result


laser


light


scattering


following


plastic


deformation


particular


material


observation of heterointerface laser light scattering is novel.


Outline of Dissertation Contents


system.


The


The dissertation is composed of 5 chapters.


Chapter


presents the


background and motivation for the project.


Chapter


pertains


GaAs


substrate


cleaning


epitaxy


subdivided


into


three


main


sections


describing


experimental


apparatus,


conventional


thermal and combined


thermal/H-atom


cleaning studies,


relevant conclusions.

Chapter 3 pertains to ZnSe/GaAs heteroepitaxy and is subdivided into


three


main


sections


describing


experimental


apparatus,


results


optical


monitoring


during


heteroepitaxy


under


various


conditions,


relevant









Chapter


presents


a theoretical model


account


for the observed


laser


light


scattering


during


ZnSe/GaAs


epilayer


growth,


chapter


summarizes the important findings of the research effort.













CHAPTER 2
GaAs SUBSTRATE CLEANING FOR HETEROEPITAXY


Background


The conventional procedure employed in cleaning GaAs substrates for


epitaxy


which involves wet chemical processing followed by in situ heating


prior to MBE growth, has been extensively studied, for example, by Contour


et al.9


and Vasquez, Lewis and Grunthaner,'o


and by authors cited in these


references.


Problems with this technique, however, include the requirement


of a processing temperature in the vicinity


of 6000C and the fact that the


resultant GaAs surface can be rough on the atomic level.


In recent years,


various novel cleaning procedures have been reported,


including in situ chemical etching of GaAs by gaseous HC1,"


photochemical


processing of GaAs,n and preliminary work on


the removal


of the native


oxides


from the GaAs


surface


using various


hydrogen plasmas.13"


These


plasma


techniques


have


been


applied


with


varymg


degrees


success


Aaiandin a


hrno


nl an mt


a. r*Co j.


UUIIUII~~~~~~~~~lIIIE~~an naea 4..l .a na. C AIIi iII ilI~ Iili:Ir r irnIIIFl


,,,,:,d


~b~YIII~UI)I~M









GaAs


particles


ions


generated


m some


plasmas


degrade surface electronic properties.1'


In an effort to explore new soft procedures, Schaefer et al.,


19 Petit et


20 and Petit and Houzay21 recently reported on the interaction of atomic


hydrogen,


generated by


dissociating H2


via a hot filament,


with


the GaAs


native oxides, concluding that the oxides can be reduced to a large extent by


such a


technique whilst limiting As desorption,


GaAs decomposition,


dopant deactivation.

With these reports in mind, one of the objectives of this work was to


investigate


the use


of an


ultra-high


vacuum


(UHV)


compatible rf plasma


source to generate an atomic hydrogen flux for GaAs oxide reduction.


had previously developed such an rf plasma source for doping ZnSe p-type

via an atomic nitrogen beam.2


Experimental Apparatus


MBE System Confieuration


Growth


chamber.


The


experiments


were


carried


min a custom-








8
UHV compatible rfplasma source manufactured by Oxford Applied Research

(Oxfordshire, UK), which was previously used as a source of nitrogen atoms


for doping ZnSe p-type.


In addition to these conventional components, the


MBE chamber was also configured with a laser probe apparatus as illustrated


in Fig.


Laser probe apparatus.


The apparatus consisted of a imW HeNe laser


(X=632.8nm) mounted


outside


MBE


growth


chamber,


laser beam


being directed through a viewport towards the substrate, which was mounted


on a heated


substrate


holder.


The


detector/amplifier


stage


apparatus consisted of a color, charged coupled device (CCD) video camera


(with a light intensity sensitivity of 2.51ux) and a color video monitor.


The


camera was mounted outside the system on its own viewport, its optical axis

lying along the sample normal, which was 100 away from the optical axis of


the incident laser beam.


This particular geometry provided for simultaneous


RHEED analysis as indicated in the figure.


The camera was focused on the


substrate surface, and the visible spot arising from elastically scattered laser

light (ELLS) due to surface morphology changes (roughening) was displayed








rf Plasma Source


Phosphor
Screen


Movable
Ion
Gauge


Mirror


H Atoms
",


UHV


CCDC


Heated
GaAs
Substrate


Zjnc


HeNe
Laser


Electron
Gun
(RHEED)


Monitor


Bias Network


ELLS Signal
to Recorder


Si photodiode

Figure 1 Schematic of the custom-designed molecular beam epitaxy system
showing conventional effusion sources, reflection high energy electron
diffraction system, and ultra-high vacuum compatible rf plasma source fitted
with a nrecisinn leak valve and oan lin few IhvgrnmC n Aln l ;liitratl ; a


H2


v \J









to the monitor'


cathode ray tube (CRT) over the image of the spot, and the


output voltage from its respective bias network was recorded as a function


of time.


this experiment.


Characterization of Optical System


It should be noted that the specular reflection was not utilized in


Due


the unorthodox nature


of the


detection


system


employed in


quantifying the intensity of elastically


scattered laser light, it was deemed


important to characterize the transfer function of the camera-based detection

system, and the apparatus illustrated schematically in Fig. 2 was employed


to perform this task.


tungsten lamp,


lens,


diffuser,


and laser line filter


combination was used to provide a monochromatic (X=632.8nm) input signal


whose intensity could be varied over a wide dynamic range.


To quantify the


intensity of light entering the camera system from the source, a beam splitter

was placed in front of the camera lens to direct a portion of the input light


towards a photodiode.


The output voltage from the bias network associated


with this photodiode was then proportional to the light intensity directed into








W Lamp



Lens


DiffUser


Laser Line Filter (632.8nm)


Glass
Slide

Neutral
Densityr
Filter


- I-.


Bias Networks


XY Recorder

to Computer


Lens


Si Photodiodes


CCD
Camera


Monitor


~L--LI


I








12
the input voltage, both intensities being recorded simultaneously with an XY

recorder as indicated in the figure.


illustrated


detection


system


exhibited a


threshold


beyond


which


linear


regime


was


observed.


input


intensity


increased,


system


responded


an increasingly


nonlinear fashion


indicated


figure.


This


was


most


likely


automatic


gain


control


circuitry


both


CCD


camera


monitor.


further


increases in input intensity


the system response was clamped, presumably


m response


saturation


of the


CCD


array.


Using this


information,


attempts were made throughout the experiments to keep the detection system


in a regime


where


clamping


occur.


The


system


was,


however,


allowed to operate in the linear/nonlinear regimes so that a reasonable degree

of dynamic range could be achieved.


UHV


Compatible rf Plasma Source


work,


plasma


source


was


fitted


with


an additional


precision leak valve and hydrogen gas line (see Fig.


which allowed for







































Input Intensity (arb.


units)


Figure
632.8n


Camera-based


detection


system


transfer


function


recorded










gas introduced into the discharge chamber.


Figure 4 shows a schematic of


the rf plasma source as configured for hydrogen atom production.

Coupled rf (13.56MHz) energy is manually adjusted with a capacitive


power


matching


unit


so that


reflected


power


is usually


than


approximately


Gas


flow


into


pyrolytic


boron


nitride


(PBN)


discharge chamber is controlled via a UHV


compatible precision leak valve.


water-cooled rf induction


surrounds


the discharge chamber,


silicon-diode-based optical emission monitor provides a measure of plasma


intensity


inside


discharge


chamber.


PBN


multi-aperture


disk


positioned at the end of the discharge chamber allows the hydrogen species


to exit into the growth chamber as shown.


The source of hydrogen was a


conventional high pressure (-2000psig) cylinder ofultra-high-purity (99.999%


H, and <3ppm H20) hydrogen,


which was regulated to


1 Opsig before being


introduced into the precision leak valve.


Characteristic


hydrogen


spectrum.


examine


species


plasma,


optical


emission


spectroscopy


(OES)


was


performed


using


apparatus


shown


Fig.


Following


removal


optical


emission


-


--























4),
0

ro


C
Pca




S





'3


o


a


a

i
at'-

o
.-
-~: 0J
4)


a
'9
14


OI~


$%o


,~rS~6

\ O
trf~$


trS1










UHV\
D Plasma
Discharge


Ziuc


16





Movable
rfPlasma Source Ion
7 Gauge


UHV
SH Atoms
\I


Heated
Substrate
Holder


SFocusing Lens


Grating


IR intensified
Linear Diode Array


Monoc


bromnator


(0.3 m)


Multichannmel
AnMlyzer


I


I


Minor








17
(150 grooves/mm) contained in the monochromator dispersed the light onto


a linear array of 760 IR-intensified silicon diodes.


A multichannel analyzer


converted the signals from the diode array into an emission spectrum.

Figure 6 illustrates a typical optical emission spectrum recorded from


hydrogen


plasma


region


under


standard


operating


conditions


source.


The rf power used to create the discharge


was


320W


while the


equilibrium background pressure of hydrogen in the MBE growth chamber


was 5x10"6mbar.


The emission spectrum shown in Fig.


6 is similar to that


reported


cyclotron


Sugata


resonance


(see


(ECR)


6 inset)


microwave


who


discharge


employed


source,


an electron


although


present spectrum only atomic hydrogen lines are evident as suggested by the


featureless


baseline


strong


emission


lines


commonly


associated


with


electronic


transitions


neutral


hydrogen


atom.


The


emission


lines


designated H, Hp, and I-, represent the first three lines in the characteristic


visible spectrum (Balmer series) of the hydrogen atom,2


and consequently,


atomic hydrogen would seem to compose a significant fraction of the flux

emanating from the rf plasma source.











H (656)


ICl


,
U,
4)


I I
400 500 600

Wavelength (nm)


HW(486)





34)


I. I I


a a a U I


. I


- I


& I - I a
- . . K p t a a p a a


400


450


500


550


600


650


700


750


800


Wavelength (nm)
Figure 6 A acteristic optical emission spectrum recorded from the
hydrogen plasma region of the ultra-high vacuum compatible rf plasma


r


I


I










production,


time-dependent OES


was


performed over a


period of 14 days


following


attainment


hydrogen


plasma


shown


More


specifically


plasma


was


ignited


on selected


days


stabilized,


optical emission spectrum was recorded, and then the rf plasma source was


shut down.


Figure 7 is a compilation of those results, and it should be noted


that all spectra are the same scale.


As indicated in Fig.


7(a), the spectrum recorded on day


indicated


a lack of the characteristic H7 and Hp lines and a broadening of the H, line.


New


features


were


also


present


near


580,


780nm.


After


mechanical pump was used to evacuate the hydrogen feed line for 5 minutes


and the line was back-filled with hydrogen, Fig.


7(b) indicated a return of


the characteristic and Hn lines and a corresponding narrowing of the H,


line.


The features near 580 and 755nm, however,


were not present.


After


24 hours, however,


the features near 580 and 755nm returned as indicated


figure


(see


7(c)).


The


optimum


spectrum,


as indicated


characteristic


hydrogen


emission


lines


featureless


baseline,


represented by Fig.


7(d) and was obtained by evacuating and back-filling the








20

SH H Ha

-\ (a)


0


-(b)







-*





-






400 450 500 550 600 650 700 750 800


Wavelength (nm)
Figure 7 Evolution of the characteristic optical emission spectrum recorded
from the hydrogen plasma region of the ultra-high vacuum compatible rf
plasma source as a function of time. Spectrum (a) was recorded 13 days
after that of Fig.6. Spectrum (b) was recorded after evacuating the H feed
I- ....AL 1-... L_ r a .4 4 44 ,a V .4








21
adopted as a standard operating procedure for all future experiments in which

the rf plasma source was employed.


It is


interesting to


note that the


positions


of the


additional


features


present in Figs.


7(a) and (c) correspond well with emission band locations


associated with active nitrogen,


which is not surprising since the source had


previously


been


used


with


nitrogen24


therefore


likely


most


probable contaminant along with air in the feed line.


Precision Substrate


Temperature Calibration


Before the oxide desorption experiments,


was employed


the laser scattering apparatus


to calibrate precisely the temperature at the surface of the


substrate


monitoring


formation


of a


eutectic


from


selected


binary


alloys.


The melting temperature of In was also used as a low temperature


calibration point.


shown


in Fig.


the heated substrate


holder, having


an integral


thermocouple (T/C) that was displaced from the substrate, was tilted such

that the angle of incidence of the impinging laser beam was 700 since it was










Phosphor
Screen


Eutectic
Alloys


Movable
Ion
Gauge
c-

|


UHV


Mirror -

CCD Ca



HeNe
Laser


















Si photodiode


Zjac


Heated
lubstrate
Holder


Electron
Gun
(RHEED)


Monitor


Bias Network


ELLS Signal
to Recorder


Figure 8 Schematic of the experimental setup used to calibrate precisely the
substrate temperature by recording the change in scattered HeNe laser light









formation.


The specular reflection


was


found to be


least sensitive to this


event, hence the use of the scattered laser light signal.


Temperature


reference


standards


were


formed


electron


beam


evaporation


selected


metals


onto


appropnate


substrates.


Specifically


1000A of Al was deposited onto Si and 1000A of Au was deposited onto Ge

since the eutectic formation temperatures of these binary alloys are 577 and


356C


, respectively.


A third reference standard was formed by attaching a


piece


wire


GaAs


which


upon


reaching


156C


melting


temperature of In,


would


wilt over.


Prior to


being mounted


MBE


growth


chamber, all three references were soldered to a single


Mo holder


with


99.9999%


pure


then


immediately


transferred


a load-lock


mounted on the MBE growth chamber.


The calibration curve shown in Fig.


increasing the


9 was arrived at by very slowly


T/C temperature from room temperature to 6500C and then


ascribing the reference temperatures to events (i.e.,


wilting of the In wire,


Au/Ge eutectic


formation, Al/Si


eutectic formation) as they


occurred.


indicated in the inset of Fig. 9, the erro


r (normalized with respect to the


T/C










600
Teu(AI/Si)



500


400-

T,,(Au/Ge)


300-




200 -

Tm(In)

100 -



0-


100


200


300


400


500


600


T/C Temperature (oC)









250-5000C.


increases


This


with


is a reasonable


increasing


variation


temperature


because


while


the emissivity


thermal


of Mo


conductivity


decreases with increasing temperature.


It should be noted that the calibration


was used in all future experiments to convert monitored, or


T/C, temperature


actual


substrate temperature so reported


temperatures


would


be system


unspecific allowing the present results to be compared to the literature.


In Situ GaAs Surface Cleaning


The


substrates


used


cleaning


study


were


epi-ready,


vertical


gradient freeze (VGF) grown GaAs(001) oriented 20 towards [110] supplied


American Xtal Technology (Dublin,


CA).


Prior to being mounted in the


MBE


growth


chamber,


substrates were


soldered


a Mo


holder with


99.9999% pure In, and then immediately transferred to a load-lock mounted

on the MBE growth chamber.


The


laser


light


scattering


data recorded


from


GaAs


wafers


treated


thermally and using a combined thermal/H-atom treatment are compared and


contrasted in Fig. 10.


A 10C/minute temperature ramp, which was achieved











































1 \
II


--r


0
u,


o
0-



o
0




0
-o-


Sac

B

'C
va


0
- 0O
CO


oW W
o^a
g's
Os
o85'-
III








27
temperature from room temperature to the maximum operating temperature

(6300C) of the heater station as shown in the figure.


indicated in Fig.


10(a),


conventional


thermal


treatment of GaAs


resulted in an associated surface roughening upon oxide desorption, the oxide


desorption


process


being


observed


simultaneous


monitoring


RHEED pattern.


et al.


This result is in agreement with the observations of Lavoie


In contrast, however, surface roughening did not accompany the oxide


desorption process in the case of the combined thermal/H-atom treatment as


evidenced


lack


of laser


scatterng


oxide


desorption regime


indicated


Fig.


10O(b).


Also,


oxide


desorption


temperature


significantly


lower in the case of the combined thermal/H-atom


treatment


than in the case of conventional thermal treatment, 350-3600C versus 610-


6200C,


respectively.


Similar


oxide


desorption


temperatures


have


been


reported in the literature whenever activated hydrogen has been employed to


clean


GaAs. 15-17,19-21


cases,


reduction


in temperature


been


attributed to the fact that the oxygen in the mixed As/Ga oxide (As2O3 and

GaO03) readily reacts with hydrogen at reduced temperatures, more so in the









to be reduced at a very slow rate,


which can be greatly increased by heating


the substrate during processing.


be seen


from Fig.


10(b),


surface


roughening was not observed using the combined thermal/H-atom treatment;


apparently


such


an observation


would


require


operation


substrate


heater beyond its maximum rated power level.

As indicated above, simultaneous monitoring of the RHEED pattern


during


each


treatment


indicated


when


oxide


desorption


process


was


completed.


Upon completion of the oxide desorption process, representative


RHEED patterns of the atomically clean GaAs surfaces were recorded, and


these patterns are shown in Fig.


As indicated in Fig.


11(a), the pattern


resulting from conventional thermal treatment of GaAs is characteristic of a

reconstructed surface, a result that is not surprising because of the presence


of dangling bonds at the free-surface upon oxide desorption.


In contrast, the


RHEED pattern shown in Fig. 11 l(b), which was recorded from a surface that

experienced the thermal/H-atom treatment, is indicative of an unreconstructed


surface, a surface where the atoms maintain


their bulk-like positions.


removing the H-atom flux and raising the temperature to 5500C,


however,









29









~~ Il


V'0







a
0.



9.4


tat'.5
6 U-
Cu
a- V
0B

'a
0J
a

CIta





h rl BC
Vrd .
iffs:










thermally


cleaned surfaces.


It is interesting to note that upon lowering the


temperature


1500C


reapplying


H-atom


flux,


reconstruction


reversed, as indicated by a RHEED pattern that returned to that of Fig. 1 l(b),


suggesting


that


atoms


were


terminating


dangling


bonds.


reconstructed surface was again achieved by removing the H-atom flux and


raising the temperature to 5500C,


is reversible.


which suggests that the termination effect


Finally, exposure to unexcited hydrogen gas (H,) had no effect


on the surface reconstruction,


which clearly illustrates the enhanced reaction


between


atomic


GaAs


surface.


These


observations


(i.e.,


effectiveness of H, reconstruction reversibility, and ineffectiveness of H) are


in agreement with


those


of O'Keeffe


a1.26


and it would


seem


that


impinging hydrogen atoms terminate the dangling bonds, thus allowing the


surface atoms to maintain their bulk-like positions.


Although not specifically


determined in this case, it can be speculated that the GaAs surface is As-rich


following


oxide


desorption


during


treatment


because


of the


similitude


between the present results and those of O'Keeffe et al.? Specifically


were able to compare a deliberately


they


As-stabilized GaAs surface obtained at









temperature corresponded to a two-folded structure (As-stable 2


at low temperature.


1 pattern)


Therefore, they concluded that the reconstructed GaAs


surface following H treatment must be an As-stable 2 X


of the similarity they observed between RHEED patterns.


surface because

It then follows,


based on this final conclusion, that the GaAs surface prior to reconstruction


(during H treatment) must necessarily be As-rich.


It should be noted that the


ZnSe


epilayers


considered


following


chapter


were


grown


unreconstructed GaAs surfaces (i.e.


GaAs surfaces immediately following


H treatment).


Conclusions


GaAs


surfaces


readily


cleaned


temperatures


or below


350C


prior to


MBE


growth


when


prepared


using


combined


thermal/H-atom


treatment,


H-atom


flux


being


derived


from


a UHV


compatible


plasma


source,


opposed


conventional


thermal


treatment,


which requires temperatures in the vicinity of 6000C.


A further


benefit associated with


atom


treatment is


that the atomically


clean









than conventional thermally treated GaAs surfaces,


which are considerably


rougher, as evidenced by laser light scattering observations.












CHAPTER 3
LASER LIGHT SCATTERING DURING ZnSe/GaAs HETEROEPITAXY


Background


Following breakthroughs in the ZnSe p-type doping area24,27


emphasis


has shifted recently in the wide-bandgap II-VI semiconductor research field


toward


provision


lattice-matched


widegap


II-VI/GaAs


epitaxial


structures


blue/green


diode


laser


application.28


Although


first


blue/green


diode


lasers


represented


considerable


achievement29'30


structures employed in the fabrication of these devices were not completely


lattice-matched and, consequently


devices exhibited short lifetimes even at


reduced temperatures.

It is generally accepted that perfect lattice-matching between widegap

II-VI epitaxial materials and GaAs substrates will be a primary requirement

with regard to the development of long-lived (at room temperature) ZnSe-


based blue/green diode lasers.


At present, however, time consuming ex situ


character nation


tec~hninne~c


llrh


ac hiorh


rsnlirtinf n


V 1"t'Ol


AiF~FFFFFFFFFFFFFFFFra~n tn








34
cross-sectional transmission electron microscopy are applied to determine the


extent of lattice-matching in widegap


II-VI/GaAs heterostructures.


The primary objectives of this portion of the work were to develop an

in situ optical probe which could be used to study the strain relief process


associated


with


mismatched


epitaxy


real-time


during


growth


attempt through theoretical and experimental approaches to correlate optical


information


obtained


in situ with major


structural


defects


associated


with


plastic deformation.


Experimental Apparatus


MBE Growth Chamber


As in chapter 2, the experiments were carried out in a custom-designed

molecular beam epitaxy system equipped with conventional effusion sources

for Zn, Se, and Te, a reflection high energy electron diffraction system for

assessing substrate surface quality and subsequent epilayer quality, and an

ultra-high vacuum compatible rf plasma source which was used to remove


GaAs


oxide


prior


epitaxy.


addition


these


conventional








Bias Network


LRI Signal
to Recorder


H2,
Srf Plasma Source


Si I
Photodiode


Phosphor
Screen


Movable
Ion
Gauge

UHV


Mirror


ESe,


CCD C


GaAs
Substrate


HeNe
Laser


zinc


ZnSe
Epilayer


Electron
Gun
(RHEED)


Monitor


ELLS Signal
to Recorder


Bias Network


Si photodiode


~clcL~









Laser Probe Apparatus


The apparatus (slightly


different than that of chapter


2) consisted of


a 1mW HeNe laser (X=632.8nm) mounted outside the MBE growth chamber,

the laser beam being directed through a viewport towards the substrate which


was mounted on a heated Mo substrate holder.


The detector/amplifier stage


of the apparatus consisted of a color CCD camera and a video monitor.


The


camera was mounted outside the system on its own viewport, its optical axis


lying


away from


the optical axis of the incident laser beam,


the laser


beam


having


near-normal


incidence.


The


camera


was


focused


on the


substrate


surface


visible


spot


arising


from


ELLS


defect


evolution was displayed on the video monitor.


The intensity of the scattered


laser light was quantified by attaching a Si photodiode to the monitor's CRT

over the image of the spot and the output voltage from its respective bias

network was recorded as a function of time.


The


specularly


reflected


beam


was


utilized


experiment


monitor film


thickness in situ by employing laser reflection interferometry


(LRI).4


As can


be seen from the figure,


the HeNe


laser was employed to








37
intensity of the reflected signal varied sinusoidally with time during growth

due to interference between the reflection occurring at the growing ZnSe free


surface and the reflection occurring at the ZnSe/GaAs heterointerface.


film


thickness


measurements


performed


on calibration


samples


using


cross-sectional scanning electron microscopy (XSEM) revealed actual film


thicknesses


which


allowed


a correlation


made


between


oscillation


period and film thickness.


Specifically, an oscillation period was determined


to correspond to -120nm of deposited material.


This oscillation period/film


thickness calibration


was


used to deduce the thickness


of the


ZnSe/GaAs


films in real-time during deposition.


It should be noted that the intensities


of both the scattered laser light and the specularly reflected laser light were

recorded simultaneously so that signal phasing could be compared.


Heteroepitaxy


Particle Con


nation Considerations


It was


very


important that prior to


heteroepitaxy,


wafer surface


remain smooth and specular such that any subsequent scattering


could be


could be









promoted such a condition.


This condition, however,


was only


guaranteed


provided the wafer enter the MBE chamber in a smooth and specular state

(i.e., free of surface defects, particulate matter, etc.).


an effort to characterized


the state


of the


wafer


surface prior to


transfer into the load-lock, a full 2"


diameter as-received wafer was scanned


with a HeNe laser and it was found that scattering could not be detected.

Upon cleaving (a practice generally used to economize material), however,

the state of the surface was dramatically altered.


As indicated in Fig.


13(a), a number of particles were generated in the


vicinity of the cleave line, their number density being -1700cm"2


.In an effort


remove


particulate


matter,


short


burst


compressed


chlorodifluoromethane gas from a non-residue dust remover (ultra-filtered to

<0.2pm) was directed at the wafer surface which reduced the particle count


to -1100cm


, a reduction of -35% (see Fig.


13(b)).


As shown in Fig. 13(c),


further exposure to the gas stream (-5 seconds) had a detrimental effect as


evidenced by a substantial increase in particle density,


-10600 to -35800cm-2


depending


on the count location.


The most probable explanation


for the









39
'C5


i-a
dC


CS
01
le
a
we








*a C
rdU
n



^ ~In








40
stream to be accelerated towards the wafer by the impinging gas molecules.

Electrostatic forces then bind the particles to the wafer surface making them


very


difficult to remove.


Based


on this study,


it was decided


that full 2"


diameter as-received wafers would be used for the heteroepitaxy


study to


ensure that scattering could be attributed to heteroepitaxy.


Ex Situ Substrate Preparation


Two types of substrates were used in the heteroepitaxy study.


The


first type was epi-ready,


vertical gradient freeze (VGF) grown


GaAs(001)


oriented 2 towards [110] whereas the second type was liquid encapsulated


Czochralski (LEC) grown GaAs(001) also oriented 20 towards


[110]; their


respective manufacturers were American Xtal Technology (AXT) of Dublin,


CA and Sumitomo Electric of Hyogo, Japan.


Prior to being soldered to a


Mo holder with 99.9999% pure In, the Sumitomo wafers were first cleaned

and oxidized by placing them in an ultra-violet (UV) ozone cleaning system


manufactured by UVOCS Incorporated for 6 minutes.


The combination of


UV (X=254 and 185nm) radiation and ozone generated by these systems have










alternative to thermal and air formed oxides.32 The


AXT


wafers, however,


were untreated as they were epi-ready and required no pre-cleaning before


being


soldered


underwent


holder.


any wet


chemical


a result,


processing and


neither


following


substrate


material


mounting,


they


were immediately transferred to a load-lock mounted on the


MBE growth


chamber.


Laser Light Scattering from


Typical Epilayers


illustrated in Fig.


the case of heteroepitaxy


on the


AXT


wafers,


laser light scattering was not detectable (by the detection system)


upon ZnSe growth initiation until approximately 390nm of material had been


deposited


whereupon


increasingly


intense


scattering


was


detected


with


increasing film


thickness.


The waveform also had an


oscillatory form as


shown


figure.


should


noted


that


order


observe


scattering upon growth initiation, it was imperative that the GaAs surface be


specular following oxide removal as discussed in chapter 2.


The growth was


terminated near 840nm to avoid saturating the detection system as discussed










(s!un


qJe) AIssuesui leu6!s iuI


SE -


C


SJ


o




0





-o
- 0






a
-0






-0


" CN~
E
S

U) "
1=O
Cs)

E C


CO




0
--



t

0
-0


- *









heteroepitaxy on the Sumitomo wafers as indicated in Fig.


Observation


of the RHEED pattern following deposition in each case indicated a smooth


ZnSe


free-surface


which


suggested


that the


source


observed


laser


scattering was at a location other than the free-surface.


The most striking feature of Figs.


14 and 15 is the t phase difference


between the ELLS signal and the LRI signal which, using optical ray tracing,

can be explained by placing an emitter at the heterointerface and considering


intense scattering at the heterointerface as opposed to the free surface.


explain further, the qualitative model illustrated in Fig.


16 was developed.


indicated in Fig.


laser beam


impinges


on the


sample at


near-normal


incidence


a portion


of the


beam


transmitted


into


epilayer and proceeds towards the heterointerface.


It is considered, that in


addition to specular reflection off the heterointerface, a small portion of the


beam


scattered


into


other (nonspecular)


angles


presence of


defects which occur during epitaxy.


For simplicity


only that angle which


points in the direction of the detection system as indicated in the figure is


considered.


The scattered


beam


traverses


the epilayer and


in addition










(si!un qJe) A!suelul leu6gs iui


-s t US


h '3


020






0


0
o





0
O
0
CD





rO
-0




o
- 0


6II


0
- o
cO,


:2'


I!

























CI'-I
Sin





CCI


4'

.-I-


strr


BU


vi



ig
I 3
'-a


'N


ti~;~cta~P2


r: n









free-surface.


Since the index step from epilayer to vacuum is negative, no


phase shift is introduced upon reflection.


The reflected portion of the beam


again traverses the epilayer, reflects off the heterointerface, and then makes

a final pass through the epilayer to emerge on the vacuum side of the free-


surface.


At this point, a x+28 phase shift has been introduced,


7i resulting


from prior reflection off the heterointerface and 28 resulting from traversing


the epilayer twice.


The phase difference, AELS, between the newly emergent


beam and the original portion of the scattered beam is then 28 + x which is

necessarily a function of the wavelength of the laser, K, the refractive index


epilayer,


thickness


epilayer,


angle


transmittance within the epilayer,


In a similar manner, it can be easily


shown


that A


S28 since 69L ELS


due to the small angles involved.


Finally


AELLS" LRI


= 7i and hence the i phase difference between the ELLS


signal and the LRI signal as observed in Figs.


14 and 15.


It is interesting to compare the present results with those of Olson and

Kibbler" who employed a similar technique to monitor scattering from GaP


during


MOCVD


GaP/Si.


The


GaP/Si


system


very


similar









under similar conditions, similar results may be achieved.


Comparing sets


of data, it is apparent that their data resembles the LRI data in Figs.


14 and


15 which would place the source of their scattering outside the epilayer or


more appropriately,


at the free-surface.


They do indeed attribute their ELLS


signal


surface


scattering


quite


easy,


usmg


corollaries


arguments above, to show the lack of a it phase difference between a ELLS

signal and a LRI signal for such a case.


Laser Light Scattering from


Very


Thick Epilayers


To experimentally


demonstrate the case of surface scattering, a very


thick


(>30pm)


ZnSe


film


was


grown


on an AXT


wafer


since


we had


observed in the past via RHEED observations that the surfaces of very thick


ZnSe


layers


appear to


be rather rough.


Upon initiating


growth,


data


resembled that of Fig.


14 in that the ELLS and LRI signal phasing was n.


Very much later on, however, the ELLS and LRI signal phasing shifted to,


remained


as can


observed


Fig.


Observation


RHEED pattern following deposition suggested a rough surface as indicated









(sv!un


qje) Aii!sualui leu!6is I-


0
N
r)


r
-~

r


cr


0





r
_o





0+


U


oo
a r*

* ,M


O *
*S




'pg




4 ag
.0 *0.u




^*rt


's


'0
u~iz


- 0j


'4-4
aO


- mpg'


Sc~


-0
- o
0o









surface had degraded as expected.


Scanning electron microscopy (SEM) was


employed to explore the nature of the degradation and the micrographs in


compare


contrast the


surfaces


of a


very


uki


(60nm)


film,


typical film (840nm), and the very thick (~


that all the films presented in Fig.


30pm) film.


18 were grown on AX


The very thin and typical film, Figs.


It should be noted

[T wafers.


18(a) and (b), respectively


were


essentially featureless except for some debris which aided in focusing the


scanning electron microscope.


As suspected, however,


the very thick film


presented a radical departure from a typical specular surface as evidenced by


the presence of a large number of features as shown in Fig.


18(c).


As can


be observed in the figure, the surface was covered with large faceted features


13800cm-2)


smaller


interspersed


features


7000cm-2)


which


could


possibly


features,


serve


as seeds


because


their


larger


size,


features.


shape,


These


distribution,


large


small


were


clearly


responsible for the observed scattering and substantiate, to a large extent, the

premise that a it phase shift is indicative of heterointerface scattering while


a 0 phase shift is indicative of surface scattering.


Consequently, the present
























I



it


#~',

Ih,* fl

t~~~t -.
~:


~'I,i


- 10pm










Structural Characterization


Many


of the characteristics of the optical data presented in Figs.


, such as the amplitude of the modulation and the


"DC component"


ELLS


signal,


dependent


optical


properties


heterointerface as well as its morphological properties, these characteristics


being


controlled by the multiplication and


propagation of defects near the


heterointerface while the film is growing.


mvestigate


defect


structure


near


heterointerface


function


of film


thickness


with


respect


both


AXT


Sumitomo


wafers


, two series of 6 epilayers were grown to thicknesses of 60,


,300,


360, 600, and 840nm.


Plan-view transmission electron micrographs (PTEM)


of these epilayers are compared and contrasted in Fig.


As can be observed from Fig.


there is a general trend in defect


density which mirrors the increase in optical scattering data presented in Fig.


20, a compilation of Figs.


14 and


Furthermore, the optical data in Fig.


suggests


there


was


a difference


epilayer


quality


which


was


dependent on the tvoe of substrate used.


Specifically.


the detectable onset
































































lFnrP 1 A WnatixY nf nta.-vieTw trmnTmihqirn slrtrn tin mmr'ronhe


~


vll















































-cC


114
* 1"1
IE rr


















































Is.


"orE

II n


*6 P II
0I ~c I



i-fl.


I"I




Cs)

dl.h


w 'a


I









280nm (see Fig. 20(b)).


This difference can most likely be attributed to the


different defect structure present near the heterointerface for layer thicknesses


ranging


from


300nm


(compare


Figs.


19(a)-(c)


Figs.


19(g)-(i)).


Specifically, the micrographs of epilayers grown on AXT


wafers indicate a


typical defect evolution from nearly dislocation free material (see Figs.


and (b)) to dislocated material (see Fig.


19(a)


19(c)) as a result of exceeding the


critical thickness (~


180nm)34 for ZnSe/GaAs.


Those micrographs associated


with


epilayers


grown


on Sumitomo


wafers


(see


Figs.


19(g)-(i)),


however,


indicate some other defect structure near the heterointerface whose density


increased


with


increasing


film


thickness


300nm.


High


resolution


transmission electron microscopy (HRTEM) was performed on these films


despite


presence


of these


defects,


epilayers


remained


single


crystal


excellent


registry


was


maintained


across


ZnSe/GaAs


heterointerface.


A representative selected area diffraction pattern (SADP)


high


resolution


cross-sectional


transmission


electron


micrograph


(HRXTEM) are shown in Figs.


21(a) and (b), respectively


to corroborate


these observations.









































































V~p

I.


It V ./
1 ,. ;-


~g


'awrsP s.. p








57
density which correlates well with the similitude of the optical data following

500nm of film deposition (see Fig. 20).


Lattice Mismatch Sensitivity


The laser probe technique was also applied during the MBE growth


of ZnTe/GaAs epilayers in an effort to explore the technique'


lattice-mismatch.


sensitivity to


The scattering data recorded from the ZnTe/GaAs system


is compared and contrasted to that observed from the ZnSe/GaAs system in


It should be noted that these epilayers were grown on Sumitomo


wafers in an effort to minimize consumption of AXT


wafers.


As can be seen from the figure, the detectable onset of scattering from


the ZnTe/GaAs material system occurred at


-140nm whereas the detectable


onset of scattering from the ZnSe/GaAs material system occurred at -280nm,


much later in terms of layer thickness.


Furthermore,


the rate of change of


scattering


(with


respect


layer


thickness)


was


greater


case


ZnTe/GaAs.


Both of these observations seem to be highly correlated with


the degree of lattice-mismatch as indicated in the figure which would suggest
































4)a

.0


n


a~ n


I 1










Conclusions


conclusion,


in situ


optical


probe


described


above


is entirely


capable of detecting plastic deformation in lattice-mismatched heteroepitaxial


systems,


such


ZnSe/GaAs


real-time


during


epitaxial


deposition.


Furthermore, the technique by its real-time nature is appropriate for defect

evolution monitoring.

It is further concluded that the observation of a x phase shift between

the LRI signal and the ELLS signal is indicative of heterointerface scattering

rather than surface scattering in the ZnSe/GaAs heterostructure case.

Finally, should perfectly lattice-matched structures be grown on "ideal"


substrates,


laser


light


scattering reported


here


should not


detected


which


could


have


important


consequences


provision


lattice-


matched structures for widegap II-VI/GaAs diode lasers, for instance.













CHAPTER


THEORY


The Role of Electric Microfields


general


trend


been


observed


that


band-edge


absorption


in a semiconductor


shifted


lower


energies


upon


plastic


deformation of the crystal, the magnitude of the energy shift being dependent


upon the extent of the deformation."35


An interesting explanation which has


been


proposed


account


absorption-edge


shifting


phenomenon


concerns


presence


strong


electric


fields


around


resultant


dislocations.3 It has been argued, in fact, that the similitude of the absorption

behavior in strong electric fields (Franz-Keldysh effect) and following plastic


deformation is striking.3"


Also, arguments have been made which suggest that


in order to present a unified theory of exponential absorption edges in both

ionic and covalent materials, electric microfields must be considered.36 Based


on the


aforementioned


evidence


possibility


nn f'*, IA ^4a..a 4... ^


considered


that


. *r


electric

l nd4,na


rr nAA A: nCrr~


~ A EA nlrr








61
mismatch strain relief, are responsible for the observed laser light scattering

local to the ZnSe/GaAs heterointerface.


The


following


discussion


pertains


consideration


electric


microfields.


The


field distribution around a dislocation is


first considered


with regard to three different potentials,


namely the deformation potential,


the charged dislocation potential, and the piezoelectric potential.


Second, the


magnitudes of these fields are considered and subsequently the electro-optic


effect


evoked


order


argue


that


a refractive


index


perturbation


sufficient to scatter


light


could result in


present material


system


as a


consequence of electric microfields.


Deformation Potential


As a result of the


strain field surrounding the core of a dislocation


line, strong local distortions of the crystal potential occur in the vicinity of


the dislocation's core.


Farvacque and Lenglart37


originally


formulated the


Fourier


transform


Farvacque38


deformation


calculated


inverse


potential


Fourier transform


later


which


Vignaud


results











E.r)=


ZeNb (1


sin(e)I(r/A)


16t 4EoL(1


where


I(r/XG)


IK
oK


+(r/XG)2


cos(RK/r)JC(K)dK


1.03 +0.283(r/AG) +0.7198(r/IX)


and where r and 6 are cylindrical coordinates, Z is the number of valence


electrons (Z=4 on average for zincblende structures),


e is the charge on an


electron, N is the number density of atoms, b0 is the edge component of the


Burgers


vector,


v is Poisson's


ratio,


is a screening


length,


permittivity of free space, se

the core of the central ion


is the dielectric constant, R is the radius around

where its pseudopotential is taken as zero (for


R=0 in practice the approximation above follows), and Jj(x) is the first order


Bessel function of the first kind.


The magnitude of Eq.


was plotted above


the x-y plane as shown in Fig. 23 for values of x ranging from


to+1









































Figure 23 A catesian mesh plot of the function presented in Eq. 1 showing
the electric field intensity due to the deformation potential as a fimetinn of









The scalar coefficient in the Eq.


was


found to be -600 kV/cm for


ZnSe,


this value being derived by


considering the parameters


presented in


table


Charged Dislocation Potential


potential


developed


presence


dangling


unsatisfied


bonds


the core


of a


dislocation.


Specifically,


if a


bond


considered to be occupied by one ground state electron, it can either accept


another


electron


or it


can


donate the


ground


state electron


the crystal.


This process can result in a net negative or positive line charge depending


upon


degree


which


dislocation


donates


or accepts


electrons.


Schroter"


first modeled


electric


field


resulting


from


charged


carners


trapped on a dislocation line and later,

a refined form of the model as shown ii


~ecjr)


Vignaud and Farvacque3s presented

i Ea. 2.


KI(r/AG)


2neo Lb. G





















Table


ZnSe


material


parameters


necessary


evaluate


scalar


coefficient of the microelectric field induced by the deformation
potential (Eq. 1).


"for an ideal 600 dislocation


b8~n2


where n=2.434 for ZnSe at 632.8nm


Parameter Value

N (m"3) 4.39x 102
b, (m)i) '3.47xl10
v ~--0.33
EL b5.92










Again, Eq. 2 was plotted above the x-y plane as shown in Fig.


24 for values


of x and y ranging as before.


before,


the scalar component


of Eq.


was


found to equal


-120


kV/cm for ZnSe, this value being derived from the parameter values shown

in table 2.


Piezoelectric Potential


non-centrosymmetnc


compounds


such


materials


having


zincblende


crystal


structure/'"


long range


strain


fields


associated


with


dislocations in this class of compounds induces an electric polarity within the


crystalline lattice via the piezoelectric effect.


The expression of the Fourier


transform


of the total piezo-potential


was


first calculated by


Vignaud and


Farvacque41


later,


expression


was


refined


same


authors3"


which allowed the piezoelectric field to be expressed in real space as shown
A#


in Eq.


F (A =


f2(O) G(rIXG f)4()r/XG)


e,,b,













U4


0

a-







21

U)


40.0



30.0



20.0"



10.0



0.0


-1.0


4


Figure 24 A cartesian mesh plot of the fimunction presented in
the electric field intensity due to a charged dislncation as


Eq. 2 showing
a fimctinn of




















Table 2


ZnSe


material


parameters


necessary


evaluate


scalar


coefficient
dislocation


of the


potential


microelectric


(Eq.


field
GaAs


induced
material


charged


parameters


used in some cases (denoted by a superscript a) since values for
ZnSe were unavailable.


"Vignaud and Farvacque"


bfor an ideal 600


dislocation


Parameter Value

f (electron/site) -'0.1
Ao (m) -'lOOxlO-10
b (m) b4.o01x10-1









where


1 -


i-iu


f2(0)---sin(20)+ --cos(20)
2


K2 (rK)
0K2 +G(r//2


a1+bi(rI/,) +cI(r/XG)2


f4(O)


= isin(4) + cos(40)
4 A


and where e14 is the piezo-modulus, a2=a4


, b2=0.04, b4=0.


c2=0.001


c4=0.064.


As before, Eq.


3 was plotted as shown in Fig. 25 over the same


domain as before.


Lastly


the scalar coefficient in Eq.


3 was found to be -60 kV/cm for


ZnSe, this value being derived with el4 assumed to be -0.1


C/m


(ref 38).


In summary,


the magnitudes of the electric microfields near the core


I,(rlX )
















U

0


'0
0


* -
.



U
0
0
0
N3
0
-


40.0



30.0'



20.0


10.0


0.0


-1.0


Figure 25 A cesian mesh plot of the function pi
the piezoelectric field intensity as a function


eseted in Eq. 3 showing
of position around the


PI









dislocation'


core (see


Figs.


22-24).


Furthermore,


the deformation electric


field is


also


largest in magnitude and extent (compare


Figs.


23-25) which


may


make


deformation


potential


largely


responsible


for the


observed


scattering.


It is interesting to note that Moriya,


42 has theoretically predicted


experimentally


crystals


confirmed


could be sufficiently


scattering


explained


dislocations


sapphire


considering the changes


in the


dielectric


constant


consequence


strain


fields


around


dislocations,


which would tend to support the present hypothesis.


As a final note, it should be mentioned that, since in some cases (as

noted) parameter values for ZnSe were not available, known GaAs parameter

values were used in order that our calculations could be performed.


The Linear Electro-optic Effect


semiconductors


(such


ZnSe)


which


are normally


optically


isotropic, the presence of an electric field induces biaxial birefringence which

means that the refractive index in the direction of the field is unaltered while


the refractive indices in the transverse directions are changed by


43 This










An3=
An(E) rE
2


where n is the refractive index of the material (2.434 for ZnSe at 632.8nm),


the electro-optic coefficient (-2x10"7


cm/kV


ZnSe),"44


and E is


magnitude of the electric field which, in this case, can range from


kV/cm as stated earlier.


10 to 300


It is interesting to note that the magnitudes of these


fields


comparable


those


employed


during


electroabsorption


electro-optic


modulation


ZnSe/ZnS0oSeo0,


waveguides44


(30-40kV/cm)


which would tend to support the hypothesis.

Using the values above, An is found to range from -1.4x105 to -4.3xl04,

which represents a significant perturbation of the refractive index local to the


dislocation's


core.


Recalling


that


field


extends


approximately


(which


on the


order


100A),


a dislocation


could


modeled


as a


filament with radius ~-A and refractive index n +


A mass ensemble of


such


objects (i.e.,


the dislocation substructure) could


then account for the


observed


scattering


of laser


light


near


heterointerface


as postulated










Conclusions


conclude,


refractive


on account


present


index perturbations required to


theoretical


produce the


analysis,


observed laser


light


scattering at ZnSe/GaAs heterointerfaces could adequately be a consequence

of microelectric fields present around the heterointerface defects.












CHAPTER 5
CONCLUSIONS


GaAs


surface


cleaning


process


was


developed


using


remote H atom flux supplied by a UHV


compatible rf plasma source.


GaAs


surface cleanliness and morphology were monitored in situ and in real-time

using RHEED and a laser light scattering technique, respectively.

From RHEED observations, GaAs surfaces were found to readily clean


at temperatures near


3500C using the


H atom treatment as opposed to the


conventional thermal treatment which requires temperatures near 6000C.


The


atomically clean GaAs surfaces were found to be specular, as evidenced by


laser


light


scattering


observations,


when


prepared


using


atom


treatment.


RHEED observations also suggested that hydrogen stabilization


of the


GaAs


surface


occurred,


stabilization


effect


being


reversible


evidenced


observation


a reversible


surface


reconstruction


RHEED.


Although


not investigated,


suggested


that


future


work


performed to elucidate the influence of this phenomena on interface states,








75
the key aspect of this portion of the work was that the provision of smooth,


specular GaAs


surfaces was a necessary prerequisite


for subsequent laser


scattering measurements during heteroepitaxy since it was important that any

scattering be attributed to epitaxy and not a rough GaAs surface.

The laser light scattering (at X=632.8nm) characteristics of undoped


ZnSe/(100)GaAs


epilayers


have


been


studied


situ,


in real-time


during


MBE


growth


it has


been


shown


a laser


light


scattering


signal


originates near the


heterointerface in these


films


, as opposed to


their free


surfaces, when the film thickness exceeds the critical thickness.


Specifically,


a near-normal incidence


HeNe


laser probe was employed during


epitaxial


deposition


ZnSe/GaAs


epilayers,


laser


probe


generating


both


specularly reflected signal as well


as nonspecular signals.


By monitoring


both the intensity of laser light scattered into a particular direction (100 away

from the incident/reflected beam) and the intensity of the specularly reflected


laser light simultaneously,


it was observed that the relative temporal phasing


between


signals


was x.


Such an


observation suggested


(based


on an


optical ray tracing model) that the source of scattering was located within the









observations revealed smooth,


featureless film surfaces which


would seem


to rule out the possibility of significant surface scattering.

Similar experiments were carried out on a very thick ZnSe/(100)GaAs

epilayer and it was observed that the relative signal phasing in this case was

0, rather than a7, which is consistent with predictions derived from the optical


tracing


model


when


surface


scattering


considered.


Thus,


concluded that a relative temporal phasing of 7t between the two signals is

indicative of a source of scattering within the epilayer itself.


From


theoretical


considerations,


postulated,


fact,


that


dislocations arising at the ZnSe/(100)GaAs heterointerface as a consequence

of strain relief act as laser light scatterers and these defects are therefore the


source of the observed scattering.


The theory


question suggested


that


strong


microelectric


fields


surround


resultant


defects


perturb


refractive


index


(via


the electro-optic effect) in a localized manner.


It is


postulated that the collective ensemble of such perturbations is sufficient to


scatter


laser


light


incident


ZnSe/(100)GaAs


heterointerface.


Calculations indicated that the field intensities near typical dislocations are








77

To further support the theoretical model, however, post-growth, ex situ

structural characterization (transmission electron microscopy) was performed


on a selection of ZnSe/(100)GaAs heterostructures


in an effort to correlate


optical


data


with


defect


structure


near


ZnSe/(100)GaAs


heterointerface.


In general, the structural data mirrored the optical data in


that both the amount of defects (evaluated qualitatively) and the intensity of

scattering increased with increasing film thickness, leveling off after 500nm


of ZnSe had been deposited.


Furthermore, the optical data was also sensitive


to the type of defect structure occurring near the heterointerface, the type of


defect


structure


being


dependent


on the


source


GaAs


substrate


material.


Thus,


conclusion,


high


degree


correlation


observed


between


laser


light


scattering


data


defect


structure


near the


heterointerface


supports


theoretical findings


concerning the


laser


light


scattering potential of misfit related defects in the ZnSe/(100)GaAs material

system.


further


concluded


that


given


conditions


where


a x phase


difference


present,


optical


probe


indeed


monitoring









heterointerface) in such films.


Furthermore, it can be concluded that should


perfectly lattice-matched structures be grown (i.e., defect-free material), laser


light


scattering


should


observed


during


growth


which


could


have


important consequences


for the provision of lattice-matched


structures


widegap II-VI/GaAs diode lasers, for instance.












REFERENCE LIST


R.H. Hartley, M.A. Folkard, D. Can, P.J. Orders, D. Rees, I.K. Varga,
V. Kumar, G. Shen, T.A. Steele, H. Buskes and J.B. Lee, J. Vac. Sci.
Technol. B 10, 1410 (1992).


G.N. Maracas, J.L. Edwards, K.
Jobs and J.A. Woolam, J. Vac.


Shiralagi, K.Y. Choi, R. Droopad, B.
Sci. Technol. A 10, 1832 (1992).


W.E. Quinn, D.E. Aspnes,
Schwarz, M.C. Tamargo, S.
Technol. B 10, 759 (1992).


M.J.S.P.
Gregory


Brasil, M.A.A.
and R.E. Nahoc


1*


Pudensi,
r. J. Vac.


S.A.
Sci.


C.M. Rouleau and R.M. Park, Appl. Phys. Lett.


60, 2723 (1992).


C. Lavoie,
Buuren, J.


S.R. Johnson, J.A. Mackenzie, T.
Vac. Sci. Technol. A 10, 930 (1992).


A.J. Pidduck, D.J. Robbins,
Glasper, J. Electrochem. Soc.


A.J. Pidduck, D.J. Robbins,
Glasper, J. Electrochem. Soc.


A.G. Cullis, D.B.
136, 3083 (1989).


D.B. Gasson, C.
136, 3088 (1989).


Tiedje


Gasson


Pickering


D.J. Robbins, A.G.
9, 2048 (1991).


Cullis and A.J. Pidduck, J.


Vac.


Technol. B


J.P. Contour, J. Massies, A. Saletes and P. Staib, Appl. Phys. A 38,
(1985).


R.P. Vasauez. B.F. Lewis and FR


Gnmthaner Annl Phv T .ett


n










Contour, J.
Rochette, J.


Hirose, S.


Massies, A.


Vac.


Saletes, M.


Technol. B 5,


Yokoyama and Y


Outrequin, F.
730 (1987).


Yamakage, J.


Vac.


Simondet and


Technol. B


1445 (1985).


R.P.H. Chang, C.C. Chang and S. Dirack, J.


Vac. Sci.


Technol. 20,


(1982).

A. Takamori, S. Sugata, K. Asakawa, E. Miyauchi and H. Hashimoto,
Jpn. J. Appl. Phys. 26, L142 (1987).


S. Sugata, A.
Hashimoto, J.


Takamori, N.


Vac.


Takado, K.


Technol. B 6,


Asakawa, E. Miyauchi and H.


1087


(1988).


P. Friedel and J. Landesman, Philos. Mag. B 55,


Lu, M.T


(1987).


Schmidt, D. Chen and R.M. Osgood Jr., Appl. Phys. Lett.


1143 (1991).


Cherchour,


Proix and


Sebenne,


Rev.


Phys.


Appl.


22, 285


(1987).


J.A.


Schaefer,


Europhys. Lett.


r. Persch, S. Stock,
12, 563 (1990).


Allinger and A.


Goldmann,


Petit, F.


Houzay


and J.M. Moison, J.


Vac.


Technol.


2172 (1992).


E.J. Petit and F. Houzay, J.


Vac. Sci.


Technol. B


12, 547


(1994).


R.M. Park, J.


B.W.


Vac. Sci.


Shore and D.H.


Technol. A 10,


Menzel, Principles of


(1992).


A tomic Spectra (Wiley,


m










T.B.


Massalski,


., Binary


Materials Park,


Okamoto,


A lloy


OH,


Phase
1990).


P.R.


Subramanian


Diagrams,


(ASM


Kacprzak,


International,


P. O'Keeffe


Den,


O'Moram, S.


. Morikawa and


Komuro, M.


Ishii,


Aoyagi, Jpn.


Meguro,


Appl.


. Iimura,


Phys.


3301


(1992).


Ohkawa, T


Karasawa and T


Mitsuyu, Jpn. J


Appl.


Phys. 30, L152


(1991).


Okuyama,


Akimoto, Electron.


Miyajima,


Lett. 28,


r. Morinaga, F
1798 (1992).


Hiei, M.


Ozawa and K.


M.A. Haase, J
1272 (1991).


Qiu, J.M. DePuydt and H. Cheng, Appl. Phys. Lett. 59,


H. Jeon, J.


Ding,


Patterson, A.


Nurmikko,


D.C.


Grillo,


Kobayashi and R.L.


Gunshor, Appl.


Phys.


Lett. 59


, 3619 (1991).


J.R.


Vig, J


.Vac.


Technol. A 3


1027


(1985).


Ingrey,


W.M.


Lau and N.S.


McIntyre,


. Vac.


Technol.


984 (1986).


J.M.


Olson and A.


Kibbler, J.


Cryst.


Growth


182 (1986).


J. Kleiman, R.M. Park and S.B.


Qadri, J


Appl. Phys.


,2067 (1987).


Vignaud and J.L.


Farvacque, J


Appl.


Phys.


1261


(1989).


J.D.


Dow, Phys. Rev. B 5, 594 (1972).


Farvacque and P


Lenglart, Phys. Status Solidi B 80,


(1977).










M.E.


Lines


A.M.


Glass,


Principles


and


Applications


Ferroelectrics and Related Matenals (Clarendon Press, Oxford, 1977).


Vignaud and J.L.


Farvacque, Phys.


Status Solidi B 125,


785 (1984).


Moriya, Philos.


Mag. B 64,


425 (1991).


J.I. Pankove, Optical Processes in Semiconductors (Dover, New
1971).


York,


M.H.
Quan.


Jupina,


E.M.


Elect. 28,


Garmire,


Shibata and


embutsu,


IEEE J.


663 (1992).












BIOGRAPHICAL SKETCH


Christopher M. Rouleau was born on March 23,


1966, in Springfield,


Massachusetts.


attended


school


Hampden,


Massachusetts,


graduated


from


Minnechaug


Regional


High


School


(Wilbraham,


Massachusetts) in


1984.


He entered


Western


New


England


College (WNEC) in


Springfield,


Massachusetts,


that same year to pursue a degree in electrical engineering


and qualified for the dean's list every semester.


While attending WNEC, he


received


a number


awards,


including


1986


Sophomore


Academic


Award, the


1986 Kenneth A. Macleod Scholarship Award, and the 1987 and


1988


Electrical Engineering Departmental Awards.


Christopher graduated


in 1988 from


Western New England College with a Bachelor of Science in


Electrical


Engineering


degree.


supenor


academic


achievement


was


recognized at graduation as he was not only awarded the honor of summa


cum


laude but was also given special recognition for attaining the highest


cumulative grade Doint average among sraduatine seniors









In the fall of 1988


, Christopher came to the University


of Florida to


pursue a master's degree in materials science and engineering.


1991 he


earned


that


Robert


degree;


Park,


thesis,


was


titled


conducted


Growth


under the


Molecular


direction


Beam


of Professor


Epitaxy


Electrical Characterization of Cl-doped ZnSe/(100)GaAs Epitaxial Layers.


While at the University


of Florida,


Christopher continued his pursuit


excellence.


was


awarded


first


prize


1992


Student


Poster


Competition at the 21st Annual Symposium on Applied Vacuum Science and


Technology, the


1991


Outstanding Bent Award (Tau Beta Pi,


Florida alpha


chapter),


prestigious


1993


Materials


Research


Society


Graduate


Student Award.


Christopher


currently


an active


member


four


honor


societies


(Sigma Beta


Tau,


Tau Beta Pi, Phi Kappa Phi, and Alpha Sigma Mu) and


professional


societies


(American


Vacuum


Society


Materials


Research Society).


He has also authored or coauthored


publications to


date and has presented papers at three national meetings.

Christopher plans to pursue a postdoctoral appointment at Oak Ridge










MBE)


technique


with


view


towards


fabrication


high-efficiency


photovoltaic cells.