Characterization of the interactions of hydrogen, nitrogen and oxygen with silicon and III-V semiconductor surfaces

MISSING IMAGE

Material Information

Title:
Characterization of the interactions of hydrogen, nitrogen and oxygen with silicon and III-V semiconductor surfaces
Physical Description:
ix, 95 leaves : ill. ; 28 cm.
Language:
English
Creator:
Corallo, Cheryl Fox, 1959-
Publisher:
s.n.
Publication Date:

Subjects

Subjects / Keywords:
Electron-stimulated desorption   ( lcsh )
Thermal desorption   ( lcsh )
Silicon   ( lcsh )
Gases   ( lcsh )
Chemical Engineering thesis Ph. D
Dissertations, Academic -- Chemical Engineering -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1987.
Bibliography:
Bibliography: leaves 85-94.
Statement of Responsibility:
by Cheryl Fox Corallo.
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 - 001046277
oclc - 18461405
notis - AFC9225
sobekcm - AA00004836_00001
System ID:
AA00004836:00001

Full Text
















CHARACTERIZATION OF THE INTERACTIONS
OF HYDROGEN, NITROGEN AND OXYGEN WITH SILICON AND
III-V SEMICONDUCTOR SURFACES













By

CHERYL FOX CORALLO


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

UNIVERSITY OF FLORIDA


1987



















ACKNOWLEDGEMENTS

I would like to express my appreciation to Dr. Gar Hoflund for

his endless support and assistance. In his capacity as research

advisor he was outstanding. Beyond educational and financial support

he was instrumental in my growing appreciation of the "gourmet

experience." I would also like to thank Greg Corallo for the support

he provided and much, much more. Additionally, I would like to thank

my parents, Frank and Elaine Fox, for the myriad of things they have

done for me which go well beyond words.

Financial support for the research presented in this

dissertation was provided by the National Science Foundation, the

Petroleum Research Fund administered by the American Chemical

Society, the Air Force Office for Scientific Research, a Women

Entering Non-Traditional Fields fellowship administered by the

Graduate School and the Department of Chemcial Engineering. Thanks

go to Perkin-Elmer Physical Electronics in Eden Prairie, MN, at which

a portion of the data presented was obtained. I would also like to

thank Cerametals Inc. and Showa Denko K. K. for furnishing the

InP(111) samples through Sumitomo Electric U.S.A. Inc.

















TABLE OF CONTENTS

PAGE

ACKNOWLEDGEMENTS.............................................. ii

LIST OF FIGURES............................................................... v

ABSTRACT. .......................................................... viii

CHAPTERS

I GENERAL INTRODUCTION................................ 1

II AN ENERGY-RESOLVED, ELECTRON-STIMULATED
DESORPTION STUDY OF HYDROGEN FROM CLEANED
AND OXIDIZED Si(100)........................... 5

Introduction....................................... 5
Experimental.....................................* 7
Results and Discussion............................. 8
Conclusions ...................................... 22

III AN ION SCATTERING SPECTROSCOPY AND
TEMPERATURE-PROGRAMMED DESORPTION STUDY
OF THE INTERACTION OF N2 WITH Si(111)............ 23

Introduction....................................... 23
Experimental.......................... ............ 24
Results and Discussion............................. 26
Adsorption... .................................... 26
Desorption.......................... .......... 31
Conclusions........................................ 38

IV AN ELECTRON-STIMULATED DESORPTION STUDY OF
THE INTERACTION OF HYDROGEN, DEUTERIUM AND
OXYGEN WITH GaAs(100)............................ 40

Introduction..................................... 40
Experimental ...................................... 45
Results and Discussion............................. 48
Conclusions ...................................... 61













PAGE

V A CHARACTERIZATION STUDY OF THE NATIVE OXIDE
LAYER FORMED ON CHEMICALLY ETCHED InP(111)....... 63

Introduction....................................... 63
Experimental....................................... 66
Results and Discussion............................. 67
Conclusions....................................... 80

VI GENERAL CONCLUSIONS AND RECOMMENDATIONS FOR
FUTURE RESEARCH.................................. 82

REFERENCES................................................. 85

BIOGRAPHICAL SKETCH ................................ ...... 95
















LIST OF FIGURES


FIGURE PAGE

2-1 ISS spectra taken from (a) the cleaned sample
and (b) a sample exposed to 2000 L of 02 at
400 OC.............................................. 10

2-2 ESD mass spectrum taken from the cleaned
Si(100) surface corresponding to figure 2-
1(a) .................. ............................. 11

2-3 H energy distribution spectra taken from (a)
the cleaned sample and (b) the sample
following exposure to a 240 eV electron beam
for 20 minutes and a 400 eV beam for 15
minutes...................... ..... .............. .... 13

2-4 H+ energy distribution spectra taken from (a)
the electron-beam-exposed sample shown in
figure 2-3(b), and (b) the beam-exposed
sample which had been annealed at 650 OC for
15 minutes. The ordinates of (a) and (b) are
scaled relatively to one another................... 15

2-5 Angle-resolved, energy-resolved ESD spectra
obtained from a cleaned surface for (a) H +
desorbing normal to the surface and (b) H+
desorbing at 450 with respect to the sample
surface plane. ...................................... 17

2-6 H+ energy distribution spectra obtained from
(a) a sample exposed to 2000 L of 02 at 400
oC and (b) the cleaned sample (same as figure
2-3(a)). ............................................ 19

2-7 The upper portions of the H + energy
distribution spectra taken from (a) the
cleaned surface, (b) a surface exposed to a
240 eV electron beam for 30 minutes, (c) a
surface exposed as in (b) with an additional
exposure to a 400 eV beam for 25 minutes, (d)
the beam-exposed surface in (c) followed by a
400 OC anneal for 15 minutes and (e) the
beam-exposed surface followed by a 950 OC
anneal for 10 minutes. ............................. 21











PAGE


3-1 ISS spectrum taken from a Si(111) surface
which had been exposed to 75 L of N2 at room
temperature. For a scattering angle of 730
E1/E0 is 0.70 for N and 0.79 for Si. All ISS
data were taken at room temperature using 1
keV helium ions.................................... 27

3-2 Ratio of N ISS peak heights to that of Si as
a function of room temperature N2 exposure.......... 28

3-3 Typical TPD spectra for masses (a) 14, (b) 28
and (c) 29 for a sample which had been
exposed to 1000 L of N2 at room
temperature. The ordinates of (a), (b) and
(c) are not scaled relatively to one
another. The experimental peak area ratios
are (a(T1)):(a(T2)):(b):(c) 1:3.5:44.3:6.1........ 32

3-4 TPD spectra for a sample which had been
exposed to 1000 L of N2 at room
temperature. Spectra labeled (a) refer to
the desorption of H2 and (b) refers to the
desorption of H. The data in (A) were taken
from a sample with a different history than
the sample in (B) (see text)........................ 36

4-1 AES spectrum (a) of contaminated GaAs(100)
taken after solvent cleaning and (b) taken
after the IBA cleaning procedure..................... 47

4-2 ESD mass spectrum taken from the contaminated
surface corresponding to figure 4-1(a). All
ESD spectra were taken using a primary beam
energy of 140 eV.................................... 50

4-3 ESD mass spectrum taken after exposing the
clean GaAs surface corresponding to figure 4-
1(b) to a dose of 2000 L of D2 while heating
the sample at 200 OC................................ 53

4-4 ESD spectra corresponding to the mass
spectrum shown in figure 4-3. (a) Total-ion
(H+ and D+) energy distribution spectrum and
ESDIED spectra of (b) H+ and (c) D ................. 55

4-5 AES spectrum taken from GaAs(100) after a
dose of 12,000 L of 02 at room temperature.......... 57











PAGE


4-6 ESD spectra taken after exposing the oxidized
surface corresponding to figure 4-5 to 2000 L
of D2 while heating the sample at 200 OC. (a)
Total-ion (H+ and D ) energy distribution
spectrum and ESDIED spectra of (b) H+ and (c)
D ............................... ........... ........ 59

4-7 ESD mass spectra taken (a) at an ion kinetic
energy of 0.8 eV and (b) at an ion kinetic
energy of 3.5 eV.................................... 60

5-1 ISS spectrum of air-exposed InP(111)................ 68

5-2 Angle-integrated AES spectrum of the same
surface shown in figure 5-1......................... 70

5-3 AES spectra of the In M4 5N45N145 features
taken from the same surface shown in figure
5-2. Spectra (a) and (c) were taken at exit
angles of 100 and 850 respectively. Spectrum
(b) is angle integrated over all exit angles........ 71

5-4 XPS spectra of the (A) P 2p and (B) In 3d
peaks taken at exit angles of (a) 100 and
(b) 850............................................ 73

5-5 XPS spectra of the (A) C is and (B) 0 is
peaks taken at exit angles of (a) 100 and
(b) 850............................................ 74

5-6 Angle-resolved valence band XPS spectra taken
at exit angles of (a) 100 and (b) 850.............. 78

5-7 Angle-integrated AES compositional profile
obtained by argon-ion sputtering.................... 79

















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

CHARACTERIZATION OF THE INTERACTIONS
OF HYDROGEN, NITROGEN AND OXYGEN WITH SILICON AND
III-V SEMICONDUCTOR SURFACES

By

CHERYL FOX CORALLO

December, 1987

Chairman: Gar B. Hoflund
Major Department: Chemical Engineering

The interaction between gases and semiconductors has been

studied using ultrahigh vacuum (UHV) surface analytical techniques.

Electron-stimulated desorption (ESD) was used to study hydrogen

present on cleaned and oxidized Si(100). It is demonstrated by

kinetic energy analysis of the hydrogen ions emitted through ESD that

multiple states of adsorbed hydrogen are present. Electron beam

exposure, annealing and oxygen exposure dramatically alter the sites

or binding energy states of hydrogen present on the surface.

Ion scattering spectroscopy (ISS) and temperature-programmed

desorption (TPD) were used to investigate the interaction between N2

and a Si(111)(7x7) surface at room temperature. Adsorbed nitrogen is

detected using ISS for doses as low as 2 L (1L = 10-6 Torr*sec.) and

saturation occurs at 75 L. At least two chemisorbed states of

nitrogen are observed using TPD. Hydrogen, originating from the

sample bulk, associates with the surface nitrogen. The surface


viii












structure was monitored throughout the N2 adsorption experiments

using low energy electron diffraction (LEED).

The interaction of hydrogen with cleaned and oxidized GaAs(100)

was studied using ESD. Deuterium adsorbs on the cleaned and oxidized

surfaces for a radiantly heated sample in the range 200-400 OC.

Different adsorbed states of hydrogen are observed by kinetic energy

analysis of the desorbing ions.

The native oxide layer of a chemically polished and etched

InP(111) surface was characterized using Auger electron spectroscopy

(AES), angle-resolved X-ray photoelectron spectroscopy (ARXPS) and
0
ISS. The oxide layer is about 30A thick and forms a sharp interface

between the oxide and the bulk InP. The outermost atomic layer is In

rich, and the region just below this is P rich. InP and InPO3 exist

in approximately equal concentrations in the oxide layer. A small

amount of In203 is also detected. Carbon is present throughout the

oxide layer as graphite and/or hydrocarbons.
















CHAPTER I
GENERAL INTRODUCTION

The study of the interaction between gases and semiconductors is

very important for gaining fundamental knowledge of semiconductor

surface properties as well as in terms of understanding and

controlling the growth of epitaxial layers and passivating layers on

semiconductor surfaces. Numerous studies have been performed to

investigate these interactions. However, discrepancies continue to

exist and a considerable number of questions remain unanswered.

The focus of the experimental work which is presented here has

been to investigate the interaction of gases with Si and III-V (GaAs

and InP) semiconductor single crystals. In order to investigate

these interactions with minimal contamination effects, the studies

were done under ultra-high vacuum (UHV) conditions. The vacuum

system used in a majority of the studies presented here has been

described previously.1 (The study for which another vacuum system was

used is noted in the text.) Briefly, it consists of a stainless

steel chamber, typically at a base pressure of 10-10 Torr, with

capabilities for performing a wide variety of experimental

techniques. Techniques such as X-ray photoelectron spectroscopy

(XPS) (also called electron spectroscopy for chemical analysis or

ESCA), Auger electron spectroscopy (AES), ion scattering spectroscopy

(ISS) and electron-stimulated desorption (ESD) were performed using a

double-pass cylindrical mirror analyzer (CMA) equipped with an












internal coaxial electron gun and an angularly resolving movable

aperture. Low energy electron diffraction (LEED) patterns were

observed using a hemispherical grid system and were recorded by

photographing the pattern through a window opposite the hemisphere.

Temperature-programmed desorption (TPD) data were obtained using a

quadrupole mass spectrometer by recording the intensity of signals of

as many as four masses simultaneously as a function of temperature.

The disseration consists of five chapters in addition to the

general introduction. Chapters II and III are concerned with the

interaction of gases with the (100) and (111) surfaces of Si,

respectively. Chapter IV is a study of the interaction of gases with

a GaAs(100) surface, and Chapter V is an investigation of an InP(111)

surface. Each chapter is a complete individual study. The general

conclusions and recommendations for future research are presented in

Chapter VI.

More specifically, Chapter II presents results of an energy-

resolved ESD study of hydrogen from cleaned and oxidized Si(100).

One of the relatively few surface spectroscopies which detects hydro-

gen directly is ESD. Therefore, it is very useful in studying the

interaction between hydrogen and semiconductors. This interaction

has both scientific and technological importance particularly due to

the fact that hydrogen has been shown to reduce electrical activity

in a number of semiconductors.218 Using ESD, it was found that

cleaning the surface by ion bombardment and annealing does not com-

pletely remove hydrogen from a Si(100) surface. Kinetic energy











analysis of the hydrogen ions emitted through ESD is useful in

distinguishing between different adsorbed states of hydrogen.

Electron beam exposure, annealing and oxygen exposure dramatically

alter the kinetic energy distribution of the desorbing hydrogen

ions. Angle-resolved, energy-resolved ESD spectra of H+ from the

cleaned Si(100) surface are also presented which show an angular

dependence of the H+ energy distributions.

The interaction between nitrogen and silicon is also important,

particularly since silicon nitride is increasingly being used as an

insulatng layer in device production and nitrogen is sometimes

present as a background gas during device fabrication.19 In Chapter

III a study is presented in which ISS and TPD were used to study the

interaction between N2 and a Si(111)(7x7) surface at room

temperature. ISS provides an excellent means of monitoring the

amount of adsorbed species at the surface as a function of exposure

due to its high surface sensitivity. Using ISS, the adsorption of

nitrogen on Si(111) at room temperature was characterized and a

saturation dose was determined. The initial sticking probability was

also found. TPD results indicate that at least two chemisorbed

states of nitrogen exist. Hydrogen, which originates from the bulk

of the sample, associates with the surface nitrogen. LEED was used

to monitor surface structure throughout N2 adsorption and during

subsequent heating.

The interaction between GaAs and hydrogen is an important topic

which has been discussed in relatively few studies. The results of












an ESD study of the interaction of hydrogen and deuterium with

cleaned and oxidized GaAs(100) are presented in Chapter IV. Cleaning

the surface by ion bombardment and annealing does not completely

remove hydrogen from the surface so dosing was performed using

deuterium. It was verified that molecular deuterium does not adsorb

at significant rates at room temperature on the cleaned or oxidized

surface, but it does adsorb on both surfaces when the sample is

subjected to radiant heating in the range 200-400 OC with a tungsten

filament. ESD gives a semiquantitative relative measure of the

amount of adsorbed deuterium, and kinetic energy analysis of the

desorbing ions is useful in distinguishing between different

adsorbed states of hydrogen.

InP is becoming an important III-V semiconductor material for

the fabrication of semiconductor devices.20-23 Chapter V consists of

a characterization study of the native oxide layer formed on

chemically polished and etched InP(111) by exposure to air. ISS was

used to determine the composition of the outermost surface layer, and

angle-resolved XPS and depth profiling using Ar+ and AES give the

composition of the oxide layer. The thickness of the oxide layer and

the relative amount of the species present in this layer were also

determined.
















CHAPTER II
AN ENERGY RESOLVED, ELECTRON-STIMULATED DESORPTION
STUDY OF HYDROGEN FROM CLEANED AND OXIDIZED Si(100)

Introduction

An understanding of the interaction between hydrogen and

semiconductors has both technological and scientific importance.

Hydrogen has been shown to reduce electrical activity in silicon,2-12

germanium13 and gallium arsenide.14-18 For example, Benton et al.3

achieved hydrogen passivation of electrically active defects created

during laser annealing of crystalline silicon by exposure to atomic

hydrogen in a hydrogen plasma. Pearton and Tavendale showed that

both the deep donor and acceptor levels associated with gold are

passivated by exposing silicon samples to a hydrogen plasma. Mecha-

nisms responsible for hydrogen passivation in silicon have been

proposed by a number of authors.5-9,11,24,25

In addition to the interest in the passivation effects of

hydrogen, the interaction of hydrogen and silicon has been studied in

order to gain a better understanding of the surface geometry of

reconstructed clean silicon surfaces.26-31 The ease with which

hydrogen induces surface reconstructions on the low-index planes of

silicon has been used to evaluate the various models for

reconstruction of the clean surface. For example, Norton30 found

that changes in the diffraction pattern induced by adsorption of

hydrogen on Si(001) are consistent with the asymmetric dimer model as











opposed to the symmetric dimer model. Numerous calculations of

surface electronic properties have also been carried out for both

clean and hydrogen-exposed silicon surfaces,27,32-40 and a variety of

experimental studies of the silicon/hydrogen system have been

performed.26,29,31,41-66 These studies show that several surface

species (including mono-, di- and trihydrides) can exist depending

upon sample history and exposure conditions.

Relatively few surface analytical techniques detect hydrogen

directly due to its light mass, small cross section for many

processes and lack of core-level electrons. However, ESD is highly

sensitive to hydrogen because it is a mass spectrometric technique.

ESD is powerful in that it can yield information on surface bonding

structure and/or mechanisms of excitation and desorption. Few ESD

studies of the silicon/hydrogen system have been published. Madden

et al.5 correlated the threshold for electron- and photon-stimulated

desorption with a localized Auger final state for atomically adsorbed

hydrogen on Si(100). ESD was also used in a study by Kazmerski67 to

investigate the interaction of hydrogen ions at grain boundaries in

polycrystalline silicon.

In the present study ESD is used to examine the binding states

of hydrogen present on Si(100). Hydrogen is found to be present on

both the cleaned and oxygen-exposed Si(100) surfaces and energy

distributions of desorbing H+ from cleaned and oxygen-exposed

surfaces are presented. The effects of electron beam exposure and

annealing on the H+ energy distributions are also discussed.












Additionally, angle-resolved and energy-resolved ESD spectra of H+

from the cleaned Si surface are presented.



Experimental

A polished Si(100) single crystal was used in this study. The

sample was ultrasonically cleaned in electronic grade

trichloroethylene, acetone and methanol and then rinsed with

deionized water prior to insertion into the vacuum chamber. The Si

was heated radiantly to 700 OC by a tungsten filament placed directly

behind the sample, and heating above 700 OC was accomplished using

electron bombardment. The temperature of the sample was monitored

using an optical pyrometer. The sample was cleaned using cycles of

ion bombardment and annealing (IBA) at 950 oC until a distinct (2x1)

LEED pattern was obtained and the ISS oxygen signal was no longer

detected. The oxygen signal was below the AES detection limit before

it was below the ISS detection limit.

The vacuum system (base pressure of 2x10-10 Torr) used in these

experiments has been described previously.1 ESD, AES and ISS were

performed using a double-pass cylindrical mirror analyzer (CMA) (PHI

Model 15-255 GAR) equipped with an internal coaxial electron gun and

an angularly resolving movable aperture. The sample was positioned

at an angle of 450 with respect to the CMA axis. AES and ISS data

were taken by operating the CMA in the nonretarding mode. The

details of the data collection are given elsewhere.8 LEED patterns

were observed using a PHI Model 15-180 hemispherical grid system.












ESD data were taken using the CMA as a time-of-flight mass

spectrometer.69 Energy analysis of low-energy (less than 20 eV)

positive ions was accomplished by operating the CMA in the retarding

mode with a pass energy of 80 eV. The ions were first accelerated by

a negative potential placed on the second grid and inner cylinder.

The outer cylinder was maintained at a more positive potential (but

negative with respect to ground) than the inner cylinder. The

complete set of CMA operating voltages has been published pre-

viously.70 For mass analysis the electron beam was pulsed onto the

sample for 300 ns and pulses were counted for 300 ns after an

appropriate delay period (5.5 us for H+). A primary beam energy of

240 eV and a primary beam current of approximately 500 nA were

used. The mass spectrum of the desorbing ions was obtained by

selecting an ion kinetic energy and scanning the delay time. An

electron-stimulated desorption-ion energy distribution (ESDIED)

spectrum was obtained by scanning the ion kinetic energy with the

CMA. H+ was the only species found to desorb by ESD in this study.

Therefore, time gating was not necessary to obtain the energy

distributions of the hydrogen ions.



Results and Discussion

In ESD electrons impact a surface creating localized excitations

which decay causing desorption of ions, neutrals and metastable

species. These low energy ions can desorb only from the top atomic

layer because ions which are created beneath the top monolayer have a












high probability of being neutralized during the desorption

process. However, excitations created beneath the surface can emit

secondary electrons which may create excitations at the surface

resulting in emission of ions.71 Thus, ESD is highly surface sensi-

tive with respect to the emitted ionic signal but not as surface

specific with respect to the primary electron beam.

The ISS spectrum shown in figure 2-1(a) was taken from a sample

cleaned by IBA cycles followed by a 950 OC anneal. Only a silicon

peak is observed in this ISS spectrum. Also, an AES spectrum (not

shown) and a distinct (2x1) LEED pattern characteristic of the clean

Si(100) surface were obtained from this surface. The ESD mass

spectrum corresponding to figure 2-1(a) is shown in figure 2-2. It

is obvious from this spectrum that hydrogen is present on the Si(100)

surface which was cleaned by IBA cycles. The amount of hydrogen

present on this silicon surface is not known because the total ESD

cross sections for hydrogen on Si(100) are not known and an

independent means of measuring surface hydrogen concentration was not

available. It is difficult to determine the surface concentration of

hydrogen since relatively few techniques directly detect hydrogen and

these techniques are difficult to quantify. For example, Madden54

has shown that changes in the lineshape on the low energy side of the

Si L2,3VV AES peak are induced by hydride formation. However, the

AES sensitivity factor for hydrogen is not known so quantifying the

amount present is not possible using AES. Although the amount of

hydrogen present is not known, ESD provides considerable information

about the adsorbed states of hydrogen.













Hed ISS


(a)


(b)


0.6 0.7 0.8


Figure 2-1.


E/E0
ISS spectra taken from (a) the cleaned sample and
(b) a sample exposed to 2000 L of 02 at 400 "C.








































Figure 2-2.


H+
-- ~ ~1-------









I-


15
FLIGHT TIME (gs)
ESD mass spectrum taken from the cleaned Si(100)
surface corresponding to figure 2-1(a).












In most cases different peaks in an energy distribution spectrum

are due to different desorbing states. However, complexities such as

differing cross sections for desorption from different states or

desorption by more than one mechanism can make ESDIED spectra more

difficult to interpret. A review of ESDIED which covers the

experimental methods, ESD mechanisms and theory and addresses the

topics of ESD cross sections, desorption by more than one mechanism

and desorption from defect sites has been published by Hoflund.70

These topics are not discussed in detail in this paper.

An interesting effect on the energy distribution of the

desorbing ions was found to occur upon exposure of the sample to the

electron beam. The total ion (H') energy distributions taken from

(a) the cleaned sample and (b) the same sample following collection

of 33 ESD spectra (primary beam energy of 240 eV for 20 minutes) and

further exposure to a 400 eV beam for 15 minutes is shown in figure

2-3. Electron beam exposure results in a depopulation of the states

which desorb at higher kinetic energy implying that they have larger

desorption cross sections. This effect was also observed for less

severe beam exposures. However, the depopulation is less dramatic in

these cases. This implies that the features in the energy

distribution are due to desorption from different sites or binding

energy states with different desorption cross sections.

It was found that the higher energy states could be repopulated

with hydrogen by annealing. Energy distributions taken from (a) the

electron-beam-exposed sample shown in figure 2-3(b), and (b) the




















































0 10


Figure 2-3.


ION KINETIC ENERGY (eV)
H+ energy distribution spectra taken from (a)
the cleaned sample and (b) the sample following
exposure to a 240 eV electron beam for 20 minutes
and a 400 eV beam for 15 minutes.











beam-exposed sample following a 15 minute anneal at 650 oC are shown

in figure 2-4. The energy-resolved ESD spectrum following annealing

closely resembles spectrum 2-3(a) except for the small features

present on the top of the energy distribution spectra which are

discussed in more detail below. For less severe beam exposures and

correspondingly lessened depopulation of the higher energy states,

annealing temperatures as low as 400 OC are sufficient to repopulate

these states at reasonable rates.

An attempt was also made to adsorb deuterium on the Si(100)

surface under various conditions. In agreement with previous

studies,29,54 D2 does not adsorb appreciably at room temperature on a

clean Si(100) surface. Furthermore, D+ was not detected by ESD

following D2 exposures of 1000 to 3000 L (1L = 10-6 Torr*sec.) at

dosing temperatures of 400, 500 or 600 OC.

Both of the facts that annealing the silicon repopulates surface

hydrogen states and that D2 does not adsorb on Si(100) imply that the

source of the surface hydrogen detected by ESD is hydrogen in the

silicon bulk which diffuses to the surface during annealing. The

presence of molecular and atomic hydrogen in the Si lattice is

discussed in a paper by Corbett et al.72

Electron-stimulated desorption-ion angular distribution (ESDIAD)

experiments are often capable of yielding bonding structural

information,70 but it is usually not possible to obtain the energy

distributions of ions desorbing at particular angles using typical

ESDIAD experimental configurations. It is expected that ions which
















1 1 1 1





(a)


(b)















I I I I
0 10


Figure 2-4.


ION KINETIC ENERGY (eV)
H4 energy distribution spectra taken from (a) the
electron-beam-exposed sample shown in figure 2-3
(b), and (b) the beam-exposed sample which had
been annealed at 650 OC for 15 minutes. The
ordinates of (a) and (b) are scaled relatively
to one another.












desorb with different angular distributions from different sites or

binding states would also have different energy distributions. This

hypothesis was tested using the 120 aperture of the CMA to collect

ions which desorbed at different angles with respect to the

surface. The resulting angle-resolved, energy-resolved ESD spectra

are shown in figure 2-5. Spectrum (a) consists of H+ desorbing in

the direction normal to the surface, and spectrum (b) consists of

ions desorbing at approximately 450 with respect to the sample

surface plane (in the (171) direction relative to the bulk

silicon). The signal for the H+ desorbing normal to the surface is

similar in intensity to that of the H+ desorbing at an angle of

450. It is apparent from these spectra that H+ energy distributions

are angular dependent providing further evidence that multiple states

of adsorbed hydrogen are present on the cleaned Si(100) surface. In

fact, spectrum (a) appears to consist of at least two peaks while the

lower-energy feature is absent in spectrum (b). Hydrogen was

detected at all angles which were examined (approximately 150, 1450,

60o and 900 with respect to the sample surface plane). The signal

intensity for H+ desorbing at 600 is similar to that of H+ desorbing

at 450 and 900, but the signal intensity for H+ desorbing at 150 is

less than half of the signal intensity obtained at the other

angles. The energy distribution for hydrogen desorbing at 600 is

very similar to the distribution for normal desorption, while the

distribution for H+ desorbing at 150 resembles that of H* desorbing

at 450*








































II I
0 10
ION KINETIC ENERGY (eV)


Figure 2-5.


Angle-resolved, energy-resolved ESD
spectra obtained from a cleaned surface
for (a) H+ desorbing normal to the
surface and (b) H+ desorbing at 450 with
respect to the sample surface plane.












In order to investigate the effect that the presence of oxygen

has on the surface hydrogen, the sample was annealed at 950 OC for 5

minutes and then dosed with 2000 L of 02 at 400 oC. The ISS spectrum

obtained from this oxygen-exposed surface is shown in figure 2-

1(b). Following the oxygen exposure, the 0 ISS signal is 1.3 times

that of the Si signal. From scattering cross section estimates,

there are more than twice as many 0 atoms on this surface as Si

atoms. An ESD mass spectrum taken after the oxygen exposure reveals

that H+ is the only ionic species which desorbs from this surface

(i.e., no oxygen or hydroxyl ions are detected). The ESD ion energy

distribution spectrum obtained from this surface is shown in figure

2-6(a). The spectrum shown in figure 2-6(b) is the same as that

shown in figure 2-3(a) taken from the cleaned, annealed surface. The

hydrogen ions desorb from the oxygen-dosed surface with a much

narrower energy distribution than from the cleaned surface. The

higher energy states are depopulated for the oxygen-exposed surface;

even more so than for the electron-beam-exposed surface. It is

possible that this narrow energy distribution is due to desorption

from one well-defined state such as hydrogen bonded in the form of

hydroxyl groups. The formation of hydroxyl groups has been shown to

occur upon hydrogen-ion exposure at oxygen-rich grain boundaries in

polycrystalline Si by Kazmerski67 using ESD, SIMS and TPD. Using

high resolution electron energy loss spectroscopy (HREELS), Nishijima

et al.65 determined that SiH and SiOH species form on the

Si(111)(7x7) surface following exposure to H20 at 300 K. However,















I-I I


Figure 2-6.


ION KINETIC ENERGY (eV)
H energy distribution spectra obtained from (a)
a sample exposed to 2000 L of 02 at 400 *C and
(b) the cleaned sample (same as figure 2-3(a)).












other possibilities exist which could explain the ESD data. It may

be that the hydrogen desorbing from the oxygen-exposed surface is

bonded as on the cleaned surface but that the higher energy

desorption states are no longer able to desorb by ESD due to an

interaction with oxygen or that the higher energy states are no

longer present on the surface. More information is required to

specify the forms of the hydrogen at the Si(100) surface.

The presence of small features on the top of the ESDIED spectra

was observed during the course of this study. For a cleaned surface

which had been annealed at 950 oC, there are four primary features on

the ESDIED spectra and a shoulder on the high energy side (figure 2-

7(a)). Electron beam exposure results in slight shifts in the energy

of these features and a slight depopulation of the higher energy

states relative to the lowest energy state. This can be seen in

figure 2-7(b) for a surface exposed to a 0.5 uA, 240 eV electron beam

for 30 minutes. The results of additional beam exposure are

illustrated in figure 2-7(c). This spectrum was obtained following

exposure to a 240 'eV beam for 30 minutes and a 400 eV beam for 25

minutes. There is a further depopulation of the higher energy

states. Annealing the beam-exposed surface results in repopulation

of the higher energy states as has been discussed previously.

However, the primary small features are not well defined for anneal-

ing temperatures of 650 oC and below. Figure 2-7(d) was obtained

from a sample which had been annealed to 400 oC. Upon annealing to

950 C the well-defined small features are again present (figure 2-


















































Figure 2-7.


(a)


(b)






(c)






(d)




(e)


0 5
ION KINETIC ENERGY (eV)
The upper portion of the H+ energy distribution
spectra taken from (a) the cleaned surface, (b)
a surface exposed to a 240 eV electron beam for
30 minutes, (c) a surface exposed as in (b) with
an additional exposure to a 400 eV beam for 25
minutes, (d) the beam-exposed surface in (c)
followed by a 400 *C anneal for 15 minutes and
(e) the beam-exposed surface followed by a 950
OC anneal for 10 minutes.












7(e)). It is possible that the individual small features are due to

desorption from distinct adsorbed states of hydrogen on the surface.



Conclusions

ESD has been used to investigate the hydrogen adsorbed on a

cleaned and an oxidized Si(100) surface. IBA cleaning cycles do not

remove the surface hydrogen completely. The hydrogen detected using

ESD diffuses to the surface from the bulk of the Si sample during

annealing and is not due to adsorption of gas-phase hydrogen. Energy

distributions of the desorbing H+ provide information about the

adsorbed states of hydrogen on the surface. Electron beam exposure

results in a depopulation of the higher energy states which are

repopulated upon annealing. Exposure to oxygen results in a

narrowing of the energy distribution and a more marked depopulation

of the higher energy states. Angle-resolved, energy-resolved ESD

provides further evidence that multiple states of adsorbed hydrogen

are present on the cleaned Si(100) surface.
















CHAPTER III
AN ION SCATTERING SPECTROSCOPY AND
TEMPERATURE-PROGRAMMED DESORPTION
STUDY OF THE INTERACTION OF N2 WITH Si(111)

Introduction

The importance of very thin insulating layers in the production

of VLSI (very large scale integration) devices is evident. Silicon

nitride is increasingly being used instead of the oxide for this

purpose.19 As device down-scaling continues, a thorough

understanding of the chemical, physical and electronic properties of

insulating films becomes even more important. In addition, nitrogen

is sometimes present as a background gas during device fabrication.

For these reasons it is essential to understand the interaction

between nitrogen and silicon.

A number of studies have been done on films produced by chemical

vapor deposition (CVD) using SiH4 combined with NH3,73-76 NH3/N20,77

and NH3/NO.78'79 Relatively high pressure (0.1-760 Torr) nitridation

of Si using N2 or NH3 has also been studied. However, in order

to study the initial stages of growth of thin nitride films on Si

with minimal contamination effects, the study must be done under

ultrahigh vacuum (UHV). Investigations have been reported in which

Si is exposed to various nitriding gases such as: N,89-93 N8 ,9'

N2 95-97 NH3,92,95,98-100 NO97,101,102 and N20103 under UHV

conditions. The majority of these studies have focused on thermal

nitridiation and have shown that an (8x8) or a quadruplet surface is











formed on Si(111) depending upon the cleanliness and temperature of

the sample as well as the nitriding gas pressure. It has been

suggested that the quadruplet pattern is evidence of the formation of

a Si3N4 surface layer.89-91,95,98,99

Relatively few room temperature studies have been reported,

particularly concerning the interaction of Si with N90,93 or N2.95,97

Heckingbottom95 found only a slight deterioration of the Si(111)(7x7)

pattern with a room temperature N2 exposure at 10- Torr while

Wiggins et al.97 were unable to detect any adsorbed nitrogen using

AES for N2 exposures at 106 Torr for 5 minutes and sample

temperatures below 600 OC. However, it is demonstrated in the

present study using ISS and TPD that N2 does adsorb on Si(111) under

these conditions.



Experimental

A polished, n-type (P-doped) Si(111) single crystal was used in

this study. The resistivity of the sample was 5-30 ohm*cm. The

sample (17.5 x 9 x 0.5 mm) was ultrasonically cleaned in electronic

grade trichloroethylene, acetone and methanol and then rinsed with

deionized water prior to insertion into the vacuum chamber. The Si

was mounted in a tungsten frame and heated radiantly to 700 OC by a

tungsten filament placed directly behind the sample. Heating above

700 oC was accomplished using electron bombardment. The temperature

of the sample was monitored using an optical pyrometer. The sample

was cleaned using cycles of ion bombardment followed by annealing











(IBA) until a distinct (7x7) LEED pattern was obtained and the ISS

oxygen signal was minimized. The sample was cleaned by annealing at

920 OC for at least five minutes between the repeated N2 exposures

during TPD experiments.

The vacuum system (base pressure of 10-10 Torr) used in these

experiments has been described previously.1 ISS and AES were

performed using a double-pass cylindrical mirror analyzer (CMA) (PHI

Model 15-255 GAR) equipped with an internal coaxial electron gun and

an angularly resolving movable aperture. ISS data were taken using a

sputter ion gun (PHI Model 04-161) positioned at an angle of 600 with

respect to the sample surface plane. The sample was positioned at an

angle of 450 with respect to the CMA axis, and the angle resolving

aperture was used to select He+ ions scattered at an angle of 130

with respect to the sample plane for energy analysis. A primary beam

of He+ ions with an energy of 1 keV was used with the current to the

sample being 0.6-0.74 yA over an area with an approximate diameter of

2 mm. LEED patterns were observed using a PHI Model 15-180

hemispherical grid system and a primary beam energy of 90 eV. TPD

data were obtained using a quadrupole mass spectrometer (EAI Series

Quad 250) by recording the intensity of the signals of as many as

four masses simultaneously as a function of temperature.












Results and Discussion

Adsorption

ISS provides an excellent means of monitoring the amount of

adsorbed species at a surface as a function of exposure. Due to the

high surface sensitivity of ISS, the composition of the outermost one
1014
or two surface layers can be determined if the scattering cross

sections and ion neutralization probabilities are known.105 The

surface sensitivity of ISS is a result of the nearly 100% probability

of neutralization or inealastic scattering for an ion which

penetrates beneath the outermost layer of atoms. A potential problem

with ISS is surface alteration due to sputtering. This was minimized

in this study by using a relatively low primary beam flux and

scanning over the N and Si peaks only. An ISS spectrum was taken

over the entire range (E1/EO of 0 to 1) proceeding the N2 dosing to

check surface cleanliness. Prior-to N2 dosing the ISS 0 signal was

reduced to less than 5% of the Si peak height by IBA. This was the

minimum 0 signal obtainable and is below the AES detection limit. In

AES the noise level was typically less than 0.005 of the Si LVV (91

eV) peak intensity.

An ISS spectrum taken from a sample exposed to 75 L of N2 at

room temperature is shown in figure 3-1, and the ratio of N-to-Si ISS

peak heights as a function of N2 dose is shown in figure 3-2. The

adsorption of N2 on Si(111) under these conditions is characterized

by an initial rapid linear uptake, a region of more gradual growth

and saturation at approximately 75 L. A very important point is that
















He'





N Si


0.8


E /Eo


Figure 3-1.


ISS spectrum taken from a Si(111) surface which
had been exposed to 75 L of N2 at room tempera-
ture. For a scattering angle of 730 El/EO is
0.70 for N and 0.79 for Si. All ISS data were
taken at room temperature using 1 keV helium
ions.


OA



























OA
0.4 -





20 80 275 1000
DOSE (L)
Figure 3-2. Ratio of N ISS peak heights to that of Si as a
function of room temperature N2 exposure.











at saturation coverage the amount of nitrogen present is below the

detection limit of AES. This explains why Wiggins et al.97 did not

detect nitrogen adsorption using AES.

According to Parilis,106 assuming a potential of the screened

Coulomb Thomas-Fermi-Firsov form, a binary elastic collision and a

primary ion beam energy of approximately 1 keV or higher, the

differential cross section for scattering ( d( ) can be found from
dQ
the expression:


da(e) (m1 + m2)a z 1z2(13.68)(iT 9)
d m (z 1/2 z 12)/3E (2 e)2 92 sine





9 = 6 + arcsin (sin O/v)



where the subscript 1 refers to the incident ion and 2 refers to the
0
surface target atom, Z is the atomic number, ao = 0.468 A E (eV) is

the primary beam energy, and 9 and e are the scattering angles of

the incident ion in center of mass and laboratory coordinates

respectively. From equation (3-1) the differential scattering cross

section, for 6 = 73 is 1.2 x 10 for Si and 6.2 x 10 for N.

Then, the relative amounts of Si and N on the surface can be

estimated from ISS peak heights assuming that the neutralization

probabilities for He+ scattering off N and Si are the same. At

saturation the ratio of N-to-Si ISS peak heights is approximately 1.2

which yields a ratio of N-to-Si surface concentration of 2.2. This












implies that there are approximately 2 N atoms for each Si atom on

the surface or approximately 2/3 of a monolayer of nitrogen on the

Si(111) surface at saturation.

From the kinetic theory of gases, the rate of collision of N2

with the Si sample surface at room temperature and a N2 pressure of

107 Torr is 6.0 x 1013 collisions/sec. The sticking probability (s

= rate of adsorption/rate of collision) can be found from the

following expression:107


N -
o de
s(i) = (3-2)
vp dt


where 8 is the fractional coverage (number of molecules

adsorbed/maximum uptake), No is the number of surface sites, v is the

number of molecules colliding with the surface (number/area'time) at

room temperature, and p is the pressure. The initial sticking

probability (so) can be obtained from the initial slope in figure 3-2

assuming i is 1 at saturation and the number of surface sites

available for N2 adsorption is half the number of surface Si atoms (-

1/2 (7.8 x 1014)).107,108 In this study so is calculated to be i.7 x

10-2

The LEED pattern was monitored throughout the room temperature

N2 adsorption steps. For doses up to saturation (75 L), a distinct

(7x7) pattern remains. For much higher exposures (1000 3000 L),

the (7x7) pattern deteriorates slightly (all spots become less

sharply defined). These results are in agreement with the findings












of Heckingbottom.95 The sample which had been exposed to 3000 L of

N2 at room temperature was then heated. The slightly deteriorated

pattern remains unchanged for temperatures up to 570 OC. Upon

heating to 670 OC the 1/7th order spots decrease in intensity

relative to the (1x1) spots. This intensity change continues with

further heating to 800 OC. The distinct (7x7) pattern is recovered

by annealing the sample at 920 oC for 5 minutes. No features due to

the (8x8) nor the quadruplet pattern were observed.



Desorption

TPD was performed in order to gain a better understanding of the

adsorbed nitrogen. The experiments were carried out by exposing the

sample to 1000 L of N2 and then heating to approximately 770 OC. The

amount of each desorption product (assumed to be proportional to the

TPD peak area) varies with sample history. Nevertheless,

considerable information can be extracted from the TPD data.

Desorption peaks occur at only two temperatures 320(T1) and 510

oC(T2). The masses of species which desorb at T1 are 1, 2 and 14

amu. At T2 masses of 1, 2, 14, 15, 16, 28 and 29 amu desorb. The

TPD peaks have very distinct and repeatable shapes (see figure 3-3)

which depend upon the kinetics of desorption.109 The shape of the Ti

peak is indicative of first order kinetics while the shape of the T2

peak probably suggests more complex desorption kinetics. Assuming

that the rate of desorption can be described by an Arrhenius

expression and that the desorption is first order, the relationship















I I I I I I I


I I I I I I I


200


400


600


Figure 3-3.


TEMPERATURE (C)
Typical TPD spectra for masses (a) 14, (b)
28 and (c) 29 for a sample which had been
exposed to 1000 L of N at room temperature.
The ordinates of (a), ib) and (c) are not
scaled relatively to one another. The
experimental peak area ratios are (a(T,)):
(a(T2)):(b):(c) = 1:3.5:44.3:6.1.












between the TPD peak temperature (T p) and the desorption activation

energy (Ed) is given as:109


E -E
d v exp (33)
2 RT) (3-3)
RT p


where v is the frequency factor (assumed to be 1013 sec.-') and 8 is

dT/dt. Using this expression gives an Ed(T1) of 34.0 kcal/mol (1.5

eV/atom). This analysis probably does not apply to the T2 peak since

it probably is not first order.

The desorbing hydrogen is believed to come from the sample sur-

face or near-surface region and is present prior to the adsorption of

nitrogen. In fact, ESD has been used to verify the presence of hy-

drogen on the Si surface before and after exposure to nitrogen.110

The hydrogen detected on the surface is not believed to be due to

adsorption of background gases since attempts to adsorb D2 at room

temperature do not yeild TPD peaks for masses 3 or 4. Additionally,

it is commonly believed that at room temperature molecular hydrogen

does not adsorb appreciably on clean Si surfaces.29,54 Therefore, it

is most likely that hydrogen is present in the bulk of the Si and is

not removed by the cleaning procedure. Furthermore, the concen-

tration of surface hydrogen varies considerably with sample pretreat-

ment and history.

In order to interpret the TPD results the cracking patterns due

to the ionizer of the mass spectrometer and the isotope ratios must

be taken into account.111-113 These figures as well as average












experimental peak area ratios are given in table 3-1. With the aid

of this table the TPD features are assigned as follows. The 28 amu

peak is due to the desorption of N2 and not Si for the following

reasons: the experimental peak area ratio of the 28 and 29 amu peaks

is not the same as the Si isotope ratio (see table 3-1); no peak is

found for 30 amu corresponding to the 30 amu Si isotope; and most

importantly, no desorption of masses 28, 29 or 30 is observed from a

clean sample. The 29 amu peak is due to desorption of primarily N2H

since the ratio of the amount of 29 to 28 is too great for the 29

peak to be due mostly to desorption of N14N15 (see table 3-1). The

14 amu TPD peak is due to N desorption. The 15 amu peak is primarily

due to NH desorption since the peak area ratio of 14 to 15 amu does

not correspond with the ratio for the N isotopes (see table 3-1).

The 16 amu peak is believed to be due to NH2, and the 1 and 2 amu

peaks are due to H and H2, respectively. Species with masses greater

than 29 amu are not observed. Fragmentation or cracking of NxHy

species does occur in the ionizer, but the cracking patterns are not

known for most of these species.

The majority of the atomic hydrogen detected is not due to the

cracking of H2 since for a number of trials H desorbs only at T2

while H2 desorbs only at T1. At other times for the same dosing

conditions but a slightly different sample history, H2 desorbs at

both temperatures (see figure 3-4). Factors which may affect the

temperature at which H2 desorbs as well as the TPD peak areas include

the number of TPD runs done prior to the exposure to N2 and












TABLE 3-1
Isotope, Cracking and Experimental

Isotope Ratiosa) I


Species mass(amu)


relative amount(%)

92.21
4.70
3.09

99.63
0.37

99.985
0.015


Cracking Patternsb,c)


2.6
97.4


8.2
91.15
0.65


14 1.1
15 3.7
16 42.3
17 52.9


TPD Area Information


Avel
TPD

mass

14
15
16


28
29
30





14(T1
28(T2)

14(T2)
28(T2)


rage Experimental
Peak Area Ratios


relative amount(%)

80d)
7
13

78e)
22
0





6f)
94

88)
92


Referenced relative amounts were converted to percentages, a)
Reference 111, b) H2 and N2--reference 112, c) NH --reference 113, d)
Taken from 1 TPD run, e) Ranged from 74-88% mass 28, f) Ranged from
92-98% mass 28, g) Ranged from 92-92.7% mass 28.


H2:


N2:



NH3:














































































(suiun fAjejiqje) N


o0
o o 4
0 0
r-4 w 0
O-1) 1 a
-A 4-1 H

0 w "


LO rz 4
0 Cd0 0
m x a a IS

C ., 03
U CO 0

10 H (d C
uj ca
O C0 4-


0 C3 -H 0 1
o4 0)


:d3 0 4-1 4 C 41
0 M6 (4 ( a
0 U 4 0-H
DC a a,/ 4






s aj ,




h c o E-.

E4 0 4J 4
C r3C EH C
0 0 4-1








) 0 3
4 0



o o
o c
0 0
o H












subsequent TPD run, the heating and/or cooling rate for the 920 OC

annealing step between doses and possibly the presence of nitrogen in

the sample near-surface region following the annealing step. The

hydrogen detected could be a result of the cracking of NxHY

species. The facts that H and H2 do not desorb at temperatures other

than T1 or T2 and that their desorption peaks are the same shape as

the nitrogen peaks suggest that hydrogen is associated with or bonded

to the nitrogen on the surface.

TPD peak area comparisons reveal that the N desorbing at T2 may

be primarily due to the cracking of N2 (which only desorbs at T2) in

the mass spectrometer (see table 3-1). It is possible that the NH

and NH2, which desorb only at T2, are due to the cracking of NxHy

species. The presence of relatively more N2 on the surface than NxH

would explain the large amount of mass 14 detected with respect to

the amounts of masses 15 and 16. Mass 17 was not detected, and

considering the ratio of mass 15 to mass 16, it is not likely that

any NH3 desorbs (see table 3-1).

Following N2 dosing and a thermal desorption experiment it is

found using ISS that nitrogen remains on the Si surface. This may be

due to nitrogen which is bound in a state which does not thermally

desorb. In order to investigate the effects of the thermal

desorption process on the adsorbed nitrogen, a series of repeated TPD

runs were done without redosing between the runs. As the number of

runs increase, there is a tendency for the TPD areas of species

detected at T2 to decrease while the areas of T1 peaks either












slightly decrease or increase. This indicates that a portion of the

species adsorbed in state 2 (T2) may be populating state 1 (T1)

during TPD runs. For the case of hydrogen the analysis is not as

straightforward. The complexity in the data suggests the coexistence

of a number of competing processes which increase or decrease the

amount of hydrogen on the surface which is ultimately detected as H,

H2 or NxHy.



Conclusions

ISS and TPD have been used to study the interaction between N2

and the Si(111)(7x7) surface. Using ISS, the adsorption of nitrogen

as a function of dose was examined and a room temperature saturation

dose of 75 L is found. Following a saturation dose nitrogen cannot

be detected with AES. The initial sticking probability, so, of

nitrogen on Si(111) is 4.7 x 10-2 at room temperature.

A distinct (7x7) LEED pattern remains for the surface exposed to

75 L of N2 at room temperature. Additional exposures to N2 and

subsequent heating result in variations in this LEED pattern, but the

distinct (7x7) pattern is recovered by annealing at 920 OC for 5

minutes. Features due to the (8x8) or the quadruplet pattern are not

observed.

Using TPD, evidence of at least two chemisorbed states of

nitrogen is found. These two states desorb at 320 and 510 oC with

distinct and repeatable TPD peak shapes. Hydrogen also desorbs in

the forms of H, H2 and NxH at the same temperatures and with the







39




same peak shapes as the desorbing nitrogen. This indicates that the

hydrogen is associated with or bonded to the adsorbed nitrogen prior

to thermal desorption. The hydrogen is believed to originate in the

bulk of the Si sample.

















CHAPTER IV
AN ELECTRON-STIMULATED DESORPTION STUDY OF THE
INTERACTION OF HYDROGEN, DEUTERIUM AND OXYGEN
WITH GaAs (100)

Introduction

The interaction between GaAs and hydrogen is an important topic

which has been discussed in relatively few studies. Studies of this

interaction may provide information which is useful in the growth of

epitaxial films or passivating layers on semiconductor surfaces. It

has been shown that atomic hydrogen introduced by hydrogen plasma

exposure passivates certain electrically active deep-level defect

sites14,15 as well as certain shallow-level dopant sites16 in GaAs.

A similar effect is observed after incorporating hydrogen into GaAs

electrochemically in a phosphoric acid electrolyte. Reduction in

electrical activity due to hydrogen passivation has been observed in

silicon2-4 and germanium13 as well. Studies of this type are partic-

ularly important with regard to the growth of GaAs films using

molecular beam epitaxy (MBE)114 and chemical vapor deposition

(CVD).115-118 In the CVD process hydrogen is typically present

either as a carrier gas115 or in the compounds used to form the

GaAs;116-118 e.g. AsH3 or Ga(CH3)3. It has also been shown that

exposing GaAs surfaces to hydrogen plasmas provides an effective

means of removing contamination.119-121 For these reasons the role

of hydrogen in the formation and preparation of GaAs and its effects

on the properties of GaAs should be considered.












An early study by Pretzer and Hagstrum122 used ion neutral-

ization spectroscopy (INS) to study the (111), (111) and (110) GaAs

surfaces before and after a room temperature exposure to different

gases including 02, N2, H2 and CO. A 3 x 105 L exposure of a clean

GaAs surface to H2 causes no change in the INS spectrum. The authors

conclude that molecular hydrogen does not adsorb at significant rates

at room temperature on GaAs surfaces.

This finding was reaffirmed in the UV photoelectron spectroscopy

(UPS) study of Gregory and Spicer123 who investigated the adsorption

of 02, CO and H2 on n- and p-type, single crystal, cleaved GaAs(11O)

surfaces. They found that atomic hydrogen readily adsorbs on p-type

surfaces at room temperature causing a large change in the electron

energy distribution curve (EDC) between 1.6 and 5.5 eV below the

valence band maximum. For low exposures this adsorption is

reversible in that heating the sample to about 235 OC results in an

EDC which is quite similar to that of clean GaAs(11O). By comparing

these spectra with results of a study of Cs-covered GaAs (110),124

Gregory and Spicer assert that adsorbed atomic hydrogen behaves like

an alkali metal adsorbed on GaAs. Although room temperature

adsorption of molecular hydrogen is not observed, a 3 x 107 L

exposure to H2 with the sample heated to about 415 OC results in

adsorption. UPS shows that this adsorbed hydrogen is different than

the atomically adsorbed hydrogen. Heating the sample to about 700 OC

does not result in an EDC equivalent to an EDC from a clean

surface. Gregory and Spicer suggest that a chemical compound may

form or that hydrogen diffuses into the GaAs.












Mokwa et al.125 used TPD to investigate the interaction between

both molecular and atomic hydrogen and a cleaved GaAs(110) surface.

They found no interaction between H2 and the GaAs surface at 250 K.

However, atomic hydrogen adsorbs readily. Heating causes desorption

of AsH3 at 340 K, H2 at 450 K and Ga atoms above 700 K.

Luth and Matz used high resolution electron energy loss

spectroscopy (HREELS) to study atomically adsorbed hydrogen and

deuterium on both n- and p-type GaAs(110). They assigned loss

features to H and D bonded to both Ga and As atoms and were able to

construct Morse functions which adequately describe the electronic

potential energy curves and a rate expression which describes thermal

desorption of molecular hydrogen. Another HREELS study by Dubois and

Schwartz127 also identifies hydrogen atoms bonded to both Ga and As

atoms on a (100) surface of chromium-doped GaAs. A lengthy anneal

forms a Ga-rich surface, and the HREELS peak due to H bonded to As

disappears. Electron energy loss spectroscopy (ELS) studies by

128 129
Bartels et al. and Monch29 found that hydrogen atoms initially

bond to the As. Several studiesl232,6,,128,129 agree that a dipols

or depletion layer forms at these surfaces upon adsorption of atomic

hydrogen.

The polar (100) surface of GaAs is more complex than the (110)

cleavage surface because it reconstructs in numerous ways resulting

in a variable ratio of surface Ga-to-As atoms. The relationship

between structure and composition has been examined130,131 finding

that the As coverage varies from about 0.27 to 1.00 as the surface












reconstructs with LEED patterns: (4x6), (1x6), c(8x2), c(2x8) and

c(4x4). Bringans and Bachrach132-135 have used angle-resolved UPS to

study atomically adsorbed hydrogen on GaAs(100) and (111). They

found that this surface always exhibits the same LEED pattern

regardless of the original structure of the clean GaAs(100) and (111)

surfaces before adsorption. The hydrided structure of GaAs(100) lies

between the c(4x4) and c(2x8) structures, and annealing always yields

the c(2x8) pattern. The GaAs(lll) surface reverts to a (1x1)

symmetry after a saturation exposure to hydrogen. The authors

conclude that atomically adsorbed hydrogen on a GaAs(100) surface

results in an As-rich surface which does not agree with the results

of Dubois and Schwartz.127 The UPS results show that adsorption of

atomic hydrogen removes states from near the top of the valence band

and introduces new states 4.3 to 5.2 eV lower. A surface state

exists on this surface, and the fact that it does not disperse in

the 1 -* K direction suggests that it does not participate in bonding

with the hydrogen.

Friedel and Gourrier119 demonstrated that exposing a GaAs

surface to a hydrogen plasma is an effective way to remove carbon and

nitrogen contamination but is somewhat less effective for removing

oxygen. Chang and Darack120 have shown that a high frequency

hydrogen plasma can easily etch GaAs at a rate of about 20A/sec and

that this rate is proportional to both the hydrogen pressure and the

r.f. power of the discharge. A surface study of the effects of

exposing a GaAs(001) surface to a hydrogen plasma has been reported












by Friedel et al.121 In this study variations in the surface

structure, high lying core levels of As and Ga and valence band were

examined using angle-resolved UPS and reflective high energy electron

diffraction (RHEED) as a function of exposure to the plasma. By

curve resolving the As 3d and Ga 3d levels, they found that hydrogen

induces a new peak on the Ga spectrum and possibly two on the As

spectrum. They also believe that hydrogen initially bonds to the As

atoms and then to the Ga atoms at higher exposures. The Ga-to-As

ratio increases with plasma exposure, and the surface becomes more

disordered.

Many surface techniques including XPS, UPS, ELS and ISS are

relatively insensitive to hydrogen or can observe hydrogen only

indirectly through its influences on other spectral features. This

is usually due to its small cross section, light mass or lack of

core-level electrons. HREELS is an example of a technique in which

features due specifically to hydrogen are observed. Techniques which

rely upon a mass spectrometric determination of hydrogen are also

directly sensitive to hydrogen. This has been demonstrated in an

early study by Madey and Yates136 in which they found ESD of H+ from

a cleaved, p-type GaAs(11O) surface. H+ desorbs from both the

freshly cleaved surface and from a cesiated surface. They attributed

the presence of this hydrogen to adsorption of background hydrogen.

In the present study the usefulness of ESD in studying the

interaction of hydrogen with GaAs(100) and oxidized GaAs(100) was

investigated. This represents an initial attempt to use ESD to












examine hydrogen on GaAs surfaces in which the following questions

are considered. Under what conditions does hydrogen adsorb on GaAs

surfaces? Can ESD provide a measure of the amount of adsorbed

hydrogen? Is it possible to distinguish between different adsorbed

states of hydrogen using ESD? The preliminary results presented here

demonstrate that ESD is a useful technique for studying the

interaction of hydrogen with GaAs and, most probably, with other III-

V materials.



Experimental

A cut and polished, undoped single-crystal GaAs sample furnished

by MACOM was used in this study. It was produced using the liquid-

encapsulated Czochralski crystal growth method. A room temperature

chemical etch with a weak sodium hypochlorite solution was used to

produce a mirror finish of the (100) surface (20). After solvent

cleaning the sample in an ultrasonic bath of electronic grade

trichloroethylene followed by acetone and then methanol, the sample

(10 x 16 x 0.5 mm) was mechanically supported by a copper holder. A

0.5 mm diameter tungsten filament was mounted behind the sample for

radiant heating. The temperature of the sample was measured using an

iron-constantan thermocouple, which was mechanically pressed against

the sample, and an optical pyrometer was used over the range of 230

to 560 oC. After inserting the sample, the vacuum system was baked

for 24 hours at 200 OC and pumped down to a base pressure of 2 x 10-


10 Torr.


The sample was cleaned using cycles of room temperature,












argon ion bombardment followed by annealing (IBA). Specific

conditions used were an Ar ion sputter at a beam energy of 1.0 keV

and a beam current of 1 iA over an area of approximately 2 mm in

diameter for 15 minutes, a 5 minute anneal at 500 oC, an Ar ion

sputter at a beam energy of 1.0 keV and a current of 1.5 iA for 30

minutes concluding with a 500 OC anneal for 5 minutes. This

procedure results in a clean surface as determined by AES. AES

spectra taken from the contaminated and clean sample are shown in

figure 4-1, curves (a) and (b), respectively.

The vacuum system used in these experiments has been described

previously.1 AES and ESD were performed using a double-pass

cylindrical mirror analyzer (CMA)(PHI Model 15-255 GAR) equipped with

an internal coaxial electron gun and an angularly resolving movable

aperture. AES spectra were taken in the nonretarding mode using an

oscillation of 10 kHz and 0.5 V peak-to-peak, a primary beam energy

of 3 keV and a current of 25 pA over a spot size of approximately 0.1

mm in diameter.

ESD data were taken using the CMA as a time-of-flight mass

spectrometer as discussed by Traum and Woodruff.69 Electronic

control and data collection were performed using a PDP 11/02 computer

system and a computer-interfaced digital pulse counting circuit

developed by Gilbert et al.137 In these experiments energy analysis

of low energy (less than 20 eV) positive ions was accomplished by

operating the CMA in the retarding mode with a pass energy of 80

eV. The ions were first accelerated by a negative potential placed






47
















4-J
0 U 0


co



=- O H *







- J1




> 40
OCu




0
t4J3















- 0 o
0 v
Zao










"4












(sliun Ajejiiqje) 3p
NP











on the second grid and inner cylinder. The outer cylinder was

maintained at a more positive potential (but negative potential with

respect to ground) than the inner cylinder. The complete set of

operating voltages has been published previously.70 For mass

analysis the electron beam was pulsed onto the sample for 300 ns

using the electron gun deflector plates in the CMA. After an

appropriate delay period (14.2 is for H+, 5.8 us for D+ or 16.8us for

0 ) pulses were counted for a 300 ns period. A primary beam energy

of 140 eV and primary beam current of about 500 nA were used. The

sample was biased at +10.0 V. It was mounted at an angle of 450 with

respect to the electron beam, and ions desorbing in the direction

normal to the sample were selected with the movable aperture.

Three different types of ESD spectra were taken in this study.

A mass spectrum of the desorbing ions was obtained by selecting an

ion kinetic energy and scanning the delay time. An ESDIED spectrum

was obtained by selecting a delay time appropriate to the ion of

interest and scanning the kinetic energy with the CMA. The total-ion

energy distribution was obtained by scanning ion kinetic energy

without time gating.



Results and Discussion

In ESD electrons impact a surface creating localized excitations

which can decay causing desorption of ions, neutrals and metastable

species. Thus ESD is a destructive process in that it alters the

composition and structure of the surface. Beam exposures were











minimized in these experiments to reduce surface damage while still

obtaining acceptable signal-to-noise ratios. In many cases a second

spectrum was taken, and since in each case this duplicates the

initial spectrum, it is assumed that effects due to beam damage are

negligible.

ESD is a complicated technique in that it can be performed in

many different ways to obtain various types of information about

bonding at a surface and/or the mechanism of excitation and

desorption. These topics have been discussed in a recent review

article70 and are only considered when they are directly related to

the results presented here. Several points regarding ESD which do

relate to this study are its surface sensitivity, the isotope effect

and cross section determination.

In ESD a low energy ion can desorb only from the top mono-

layer. Ions which are created below the top monolayer have a nearly

100% probability of being neutralized during the desorption

process. However, excitations created beneath the surface can emit

secondary electrons which can create excitations at the surface

causing emission of ions.71 Thus, ESD is highly surface sensitive

(outermost monolayer) with respect to the emitted ionic signal but

not so surface specific with respect to the primary electron beam.

The ESD mass spectrum shown in figure 4-2 was taken from the

contaminated surface corresponding to figure 4-1(a). The peak at 4.2

us reveals the presence of hydrogen, but ESD peaks due to positive

ions containing carbon or oxygen do not appear although the surface


































5 15 25 35
FLIGHT TIME [pal


Figure 4-2.


ESD mass spectrum taken from the contaminated
surface corresponding to figure 4-1(a). All ESD
spectra were taken using a primary beam energy
of 140 eV.











is heavily contaminated with these species. After IBA cleaning the

ESD mass spectrum appears the same except that the total count due to

H + is reduced by a factor of 8 using the same data collection

parameters. This indicates that hydrogen is present as a surface

species even on a "clean" GaAs surface. Drawing an analogy between

GaAs and Si, which readily chemisorbs hydrogen and forms a hydride2-

4,72 may be useful in understanding the interaction between hydrogen

and GaAs. GaAs is both isoelectronic and isostructural with Si

except GaAs has a slightly larger lattice constant and its bonding

structure is less covalent. These facts suggest that a hydride

compound of GaAs may form and support the supposition of Gregory and

Spicer123 that hydrogen diffuses into GaAs forming a compound.

The surface concentration of hydrogen present in these

experiments is not known. In order to determine the surface

concentration of hydrogen using ESD, it is necessary to know the

total cross section for desorption. ESD cross sections of adsorbed

species vary over a large range but generally are orders of magnitude

smaller than the corresponding gas phase cross section due to the

high reneutralization probability of the escaping ions. However, an

independent measurement of the surface concentration is required to

determine the ESD cross section. As pointed out above, this is

difficult with hydrogen. An attempt was made to use TPD to estimate

the amount of hydrogen and deuterium present, but the sample mount

was too massive to perform high-quality TPD. A very crude estimate

from the TPD results suggests that tenths of a monolayer of deuterium












are adsorbed during the dosing procedure used. Efforts are being

made to improve the quantification of these results.

Regardless of the cleaning procedure used, hydrogen is always

present on the surface. It is possible that the surface hydrogen

originally present is not completely removed, that the surface

adsorbs background hydrogen during the cleaning process or that

hydrogen in the bulk diffuses to the surface during annealing. In

order to distinguish between surface hydrogen which is always present

and hydrogen which is adsorbed under particular conditions, deuterium

was used in these experiments. It was verified that molecular

deuterium does not adsorb on GaAs at significant rates at room

temperature. Heating the sample radiantly is known to dissociate D2

which effuses to the surface even when the filament is not in line-

of-sight of the surface.134 It is assumed in these studies that the

adsorbed deuterium is due to adsorption of dissociated deuterium but

the adsorption of some molecular deuterium cannot be ruled out. The

dose rate of deuterium atoms in these studies is only known in a

relative sense.

Figure 4-3 shows an ESD mass spectrum taken after exposing the

clean surface to a 2000 L dose of D2 while maintaining the sample at

200 OC. In addition to the H+ peak, which is the typical height from

a clean surface, a peak due to D+ appears at a flight time of 5.8

ps. Determining the relative amounts of adsorbed hydrogen and

deuterium from this data is not straightforward because ESD exhibits

an isotope effect which is particularly large for hydrogen. The





































Figure 4-3.


5 10 15 20
FLIGHT TIME [ps]
ESD mass spectrum taken after exposing the clean
GaAs surface corresponding to figure 4-1(b) to a
dose of 2000 L of D2 .while heating the sample at
200 C.












desorption cross section is proportional to exp (-cvM) where c is a

constant and M is the mass of the desorbing ion.138 An isotopic ion

of higher mass leaves the surface with approximately the same kinetic

energy as a lower mass isotopic ion. Thus, the heavier ion desorbs

with a lower velocity and has a greater probability for

reneutralization. This has been verified experimentally. The ratio

of the ionic desorption cross sections for H+ and D+ desorbing from a

tungsten surface is greater than 100:1.139-141 By this reasoning

figure 4-3 suggests that the surface concentration of deuterium is

greater than the surface concentration of hydrogen by a factor of

approximately 50.

The total-ion energy spectrum and ESDIED spectra of H+ and D+

are shown in figures 4-4(a), (b) and (c), respectively. Due to the

similarities of figures 4-4(b) and (c), it is believed that hydrogen

and deuterium are desorbing from the same adsorbed state by the same

mechanism. Unfortunately, it is not known if hydrogen desorbs from

states bound to Ga or As atoms or both. The spectra were taken in

the order (c), (b), (a). Comparing the high energy tails of these'

spectra suggests that with increased exposure to the primary beam, a

higher kinetic energy state may appear or that hydrogen is initially

present in another state. This is seen most readily by comparing the

widths of the H+ and D+ spectra with the width of the total-ion

spectrum. This second state would have a lower desorption cross

section than the more prominent state or it would be less densely

populated and, therefore, would require a larger beam exposure to

build up a significant count.
















a I C;2 ,
1
ro I'



4-4

-0

0u

1 ca











o o o c+

+0
.000 0






aa



Uo
I"" -4 '








zo
o 4=f
co z


0 1 4 4-
P- 0
^0 4- cd

U LA



I&0

o cf z





o 0
0 ^ 2 ada
-T I ---- i-- g -

^~~~ _u

a f Z os-
-? I ^ 0) *^











Following the 2,000 L D2 dose at 200 OC, the sample was heated

to 500 OC for 10 minutes. ESDIED spectra taken before and after

heating are quite similar in shape. However, after heating the D+

signal of the ESD mass spectrum is reduced by a factor of 2 with

respect to the H+ signal which essentially remains unchanged. This

result agrees with the results of Gregory and Spicer123 in which

adsorbed hydrogen could not be totally removed from the surface after

a 3 x 107 L exposure to molecular hydrogen with the sample heated at

415 C. As stated above, it is possible that both molecular and

atomic deuterium adsorb during dosing at elevated temperatures in

this study. It was also found that dosing the sample at 400 OC

rather than 200 OC results in spectra identical to those shown in

figures 4-3 and 4-4.

In order to examine the influence of surface oxygen on the

interaction between hydrogen and GaAs, the clean surface was oxidized

by dosing with 12,000 L of 02 at room temperature. The resulting AES

spectrum is shown in figure 4-5. Again, an ESD mass spectrum taken

after a 2,000 L dose of D2 at room temperature shows no evidence of

deuterium adsorption. However, heating the sample above 200 OC while

dosing causes adsorption of deuterium. The H+ and D+ counts are

about 50% greater from the oxidized surface than from the clean

surface, and the ratio of D+ to H+ is similar for both surfaces. It

was also found that deuterium adsorbs more readily on a freshly

oxidized surface than on a vacuum-aged (1 week at 10 Torr)

oxidized surface. The normalized H+ count from a vacuum-aged surface















Gall I ll I
SII I II IAs


(e
.0




















12,000 L of 02 at room temperature.











is 3 times as large as the count from a freshly oxidized surface.

Possible explanations for this behavior are that an oxidized surface

adsorbs water molecules during a long exposure to the background gas

in the UHV system or that the ion gauge tube filament dissociates

background hydrogen which then adsorbs at the surface.

ESDIED spectra of the oxidized surface after exposure to 2,000 L

of D2 while heating the surface at 200 oC are shown in figure 4-6.

The total-ion energy spectrum shown in figure 4-6(a) clearly reveals

the presence of a high kinetic energy shoulder. The presence of two

states is exibited by figures 4-6(b) and (c) for H+ and D+

respectively. However, the relative ratios of ions from the two

states are different for hydrogen and deuterium, and the ratios vary

as a function of beam exposure time. ESD mass spectra taken at ion

kinetic energies of 0.8 and 3.55 eV are shown in figures 4-7(a) and

(b) respectively. This figure shows that the deuterium-to-hydrogen

ratio is a function of ion kinetic energy. The difference in

concentration ratios is consistent with the interpretation that at

least two different binding states of hydrogen are present. The

precise nature of these states remains to be investigated.

Vasquez et al.142 have suggested that hydroxyl groups on GaAs

surfaces decompose according to the reaction


2Ga(OH)3 + Ga203 + 3H20


(4-1)





















IE














C 0

















o 0


-










o a a











I-1
SJ.Nflo 1YJ.OJ


00



0 cQ
0
0 a

44 CM



o co c

0
U 3















u M Q
c e4J

P 0)


i- '3 41 C






o ou M






X 414
oU M









z 1 0
W 0 0 ( Q






C -'e0
o < o -4 -H







w/ O U 4






60
-I
rgOzt4
*^ 4) s^

v6, ^
O r f

Z 0-
O B a-i-
l- I 4-





















































































SINno3 1YivLo.


0







*-I-
00



I-H

0 *r


r0


a






zoo

co
4-1 -1'











U. 0
to
L0




000
cri
t u
0 C






H


















Hr













at elevated temperature thereby reducing the oxygen content in the

surface region. This was considered in this study. It was found

that heating the deuterated, oxidized surface to 500 OC results in a

very large decrease in the ESD D+ signal. Also, monitoring the

desorbing species with a quadruple mass spectrometer while

increasing the sample temperature shows that D2, HD and H2 desorb and

that no H20 or other oxygen-containing species desorb. If hydroxyl

groups form under the dosing conditions used in this study, then

heating causes them to form molecular hydrogen leaving the oxygen on

the surface. This observation is also consistent with the fact that

surface oxygen on GaAs is more resistant to removal by a hydrogen

plasmal9 than other contaminants.



Conclusions

AES and ESD have been used to study the interaction between

deuterium and both clean and oxidized GaAs(100). Cleaning was

performed using IBA cycles, but it is not possible to remove the

surface hydrogen completely. Molecular deuterium does not adsorb on

either surface at significant rates at room temperature, but heating

the samples radiantly between 200 and 400 OC with a hot filament

causes adsorption of deuterium. A partially successful attempt to

use TPD to quantify the amounts of adsorbed deuterium indicates that

a few tenths of a monolayer probably adsorb during the dosing

period. Taking the ESD isotope effect into account, it is estimated

that the amount of adsorbed deuterium is about 50 times the amount of












residual hydrogen. Energy analysis of the desorbing ions indicates

that at least one and possibly two adsorbed states are present on the

clean GaAs(100) surface, but the nature of the adsorption states is

not known. Heating the sample at 500 OC reduces the amount of

adsorbed deuterium. It is presumed that most desorbs, but some may

diffuse into the bulk.

The ESD H+ and D+ signals from the oxidized surface are about

50% greater than those from the clean surface after dosing under the

same conditions. ESDIED clearly demonstrates that at least two

adsorption states of hydrogen are present on the oxidized surface.

Heating the sample to 500 oC causes desorption of D2, HD and H2, but

no oxygen or oxygen-containing species desorb from the deuterium-

dosed, oxidized GaAs(100) surface.

Although much remains to be done in characterizing the

interaction of hydrogen with GaAs surfaces, and more generally with

III-V semiconductor surfaces, the results presented in this chapter

demonstrate that ESD provides a very useful technique for this type

of study. It is now important to quantify the ESD results more

accurately and to characterize the adsorbed states of hydrogen. Both

are possible through the combined use of other surface techniques

with ESD.

















CHAPTER V
A CHARACTERIZATION STUDY OF THE NATIVE OXIDE
LAYER FORMED ON CHEMICALLY ETCHED InP(111)

Introduction

InP is becoming an important III-V semiconductor material for

fabrication of semiconductor devices.20-23 Small changes in the

composition of InP surfaces can significantly influence the

electronic structure at the surface. 143 Thus, the properties of

epitaxial layers and, consequently, of the final devices are

determined by the surface of the InP substrate and any oxide layer

which may be present. For this reason it is important to

characterize native oxide layers on InP substrates. Often the native

oxide layer is completely removed by polishing procedures or is

partially removed during chemical etching steps. Characterization of

the additional oxide formed upon exposure to air or formed by various

methods of oxidation is also important.

Numerous studies of the oxidation of InP have been performed in

order to investigate the growth mechanisms, surface compositions and

oxidation states of the surface species using various methods of

oxidation such as thermal oxidation,144,145 high-pressure thermal

oxidation,146 photoenhanced thermal oxidation,147 photoenhanced

oxidation,148 chemical oxidation149 and electron-stimulated

oxidation.150 The effects of etchants on InP surfaces and oxide

layers have also been investigated 51-154 as have the effects of

sputtering,155-157 annealingl58-160 and exposure to vacuum.143,161











XPS, SIMS and X-ray diffraction were used by Nelson et al.144 to

study the thermal oxides of p-type InP(100) formed in -1/3 atm of

02. They suggest that for a temperature in the range of 400 to 600

OC an outer layer of In203 and InPO4 is formed and an inner layer of

primarily InPO4 is formed. Elemental P is observed at the oxide/bulk

interface. For T>650 OC the oxide is composed of InPO4. Yamaguchi

and Ando145 report that a polycrystalline oxide composed of InPO4

(In203+P205) forms during a thermal oxidation (450-500 OC) of an

undoped InP(100) surface in dry oxygen (0.2-2 atm). At higher

temperatures the oxide is mostly In203. Infrared absorption and

electron diffraction data suggest that for T>620 OC the oxides become

In rich and lower oxides (possibly In20) form due to the loss of

phosphorous. Wilmsen et al.46 find that bulk oxides consisting of

InPO4, In203 and P205 form at elevated temperature (450 OC) and high

oxygen pressures (42-500 atm). Faithpour et al.147 used XPS and SEM

to investigate the laser-enhanced thermal oxidation of n-type, semi-

insulating InP(100) in a N20 atmosphere. They found that In203 and a

phosphate species form and that the growth rate is a function of the

light intensity, N20 pressure and temperature. The phosphate layer

thickness is believe to be limited by the outwardly diffusion of

phosphorus, which has also been suggested by Nelson et al.144 The

photon-stimulated oxidation of cleaved InP(110) studied by Koenders

et al.148 results in the formation of a depletion layer, but the fact

that no surface dipole is introduced suggests that oxygen is

incorporated into the lattice rather than adsorbed at the surface.












Exposure to oxygen with and without photon stimulation results in

depletion of phosphorus at the surface in addition to formation of

indium oxide. The chemical oxidation of n-type InP(100) in hot

nitric acid also reveals a depletion of surface phosphorus as

discussed by Korotchenkov et al.149 Annealing and exposure to vacuum

increases the amount of phosphorus in the surface region. Electron-

stimulated oxidation of n-type InP(100) was investigated by Oliver et

al.150 using AES, LEED and ELS. In203 and a volatile phosphorus

oxide form in the electron irradiated region. Heating above 500 OC

causes the formation of small droplets of elemental indium. Clark et

al.151 used angle-resolved XPS to investigate the oxide layers and

hydrocarbon contamination of InP surfaces after numerous

treatments. The as-received samples were found to be phosphorus

rich, but various treatments result in substantial differences in the

composition of the surface region. Bertrand153 concludes that the

use of different etchants causes the amount of oxidized phosphorus on

n-type InP(100) surfaces to vary. However, significant amounts of

oxidized indium were found only on a sample which was heavily

oxidized in air. Oxides formed in air after chemical etching with

Br-MeOH were studied by Wager et al.152 The as-etched surface is

only slightly oxidized and is primarily a phosphorus oxide.

Thus far, most of the InP studies have focused on the (100) and

(110) surfaces with only a few studies of the (111)

surface.145,154,162,163 This may be due to the difficulties

encountered in obtaining a mirror finish using conventional etching












procedures.154 In this study angle-resolved XPS, AES and ISS are

used to examine the oxide layer formed on chemically etched InP(111)

by exposure to air.



Experimental

An undoped InP(111) face cut to 10 was furnished by Sumitomo

Electric Industries, Ltd. for this study. The sample was grown by

the liquid encapsulated Czochralski method and was polished using a

mixture of Zr02 and NaOCI and then a mixture of Si02 and NaOH. The

sample was then etched with H3PO4:HBr (2:1) to produce a mirror

finish. The electrical properties and etch pit density are: carrier

concentration < 1016m-3 (n type), resistivity at room temperature =

1.8-1.9 x 101 ohm-cm, mobility = 3.9-4.0 x 103 cm2V-1 s-1 and

E.P.D. < 105cm2. The as-received sample was cut using a wafer saw

to 21 x. 15 x 0.35 mm and was ultrasonically cleaned for 5 minutes in

each of the following electronic-grade solvents: trichloroethylene,

acetone, methanol and then rinsed in deionized water.

A Perkin-Elmer PHI model 570 analytical system with a base

pressure of 2 x 1010 Torr was used in this study. The data was

collected at Perkin-Elmer Physical Electronics in Eden Prairie, MN.

A double-pass cylindrical mirror analyzer (CMA) with an internal

electron gun and angularly resolving movable aperture was used to

collect AES, XPS and ISS data. AES spectra were taken using a

primary beam energy of 5 keV and a beam current of 0.3 yA rastered

over an area of 2 x 2 mm. XPS spectra were taken using a Mg anode at












a voltage of 14 kV and power of 300 watts. High resolution XPS

spectra were obtained using a pass energy of 25 eV which resulted in

a total energy resolution of 0.75 eV. The high resolution XPS

spectra were taken in the angle-integrated mode and angle-resolved

mode at take-off angles of 100 and 850 with respect to the surface

plane. ISS spectra were obtained using a 1 keV beam of 3He a

scattering angle of 1440 and the movable aperture for improved

resolution. For all spectra taken, the sample was placed at an angle

of 500 with respect to the CMA normal.

In this study nondestructive techniques were used before

destructive techniques. Thus, the techniques were used in the

following order: XPS, angle-resolved XPS, AES, ISS and depth

profiling using argon-ion sputtering. The results are not presented

in this order in the following section.



Results and Discussion

Figure 5-1 shows the ISS spectrum taken from the air-exposed

InP(111) surface. Three peaks are discerned corresponding to In, P

and 0, and the In peak is much larger than the other two peaks. ISS

is essentially outermost layer sensitive for two reasons. The first

is that ions which penetrate beneath the top layer have a very high

probability of being neutralized, and the second is that even if they

are not neutralized, they usually scatter inelastically during escape

from the solid. In either case ions scattering off subsurface atoms

do not contribute to the ISS peaks. Inelastically scattered ions







68



































a.






0c 0



1
0 o






0


















*rI
/ ''""

/ *-<
o.( a
\ \-/
\ p-
\ ^
\ <0
I ~ ( T
1 I W


\ 0 0

ol ui S
'-' V ~ 0
\ ^ ^
\ T<


(Sl!un AjeiJ!qje) (3)N












comprise the background and, hence, the magnitude of the background

depends upon the electron mobility in the near-surface region. Based

on previous ISS studies, it appears that electrons in the surface

region of this oxide layer are not highly mobile since the background

is fairly large. The elastic scattering cross section increases as

the mass of the surface atom increases. Even though In is heavier

than P, the size of the In peak indicates that the outermost layer

consists predominantly (>80%) of In atoms with roughly equal amounts

of 0 and P atoms. No feature due to surface C appears.

Figure 5-2 shows an angle-integrated AES spectrum of the same

surface. An intensity analysis using standard methodsl64 yields an

In/P atomic concentration ratio which is slightly greater than 1 but

much less than the In/P ratio obtained in ISS. This is due to the

fact that AES is much less surface sensitive than ISS. The Auger

signal is an exponentially weighted average from electrons

originating within about three mean-free paths of the sample

surface. Since ISS reveals that the outermost layer is mostly In and

AES shows that the outermost few layers contain approximately equal

amounts of In and P, the one- or two-layer region below the outermost

region must be P rich. Both C and 0 appear in this spectrum in

approximately equal concentrations even though no C is detected by

ISS at the surface. High resolution AES spectra of the dominant In

peaks at a kinetic energy of about 400 eV are shown in figure 5-3.

Spectrum (a) was taken at an exit angle of 100 and spectrum (c) was

taken at an exit angle of 850. Spectrum (b) is angle integrated.






70






















iCi,






ul
2 O
o o
D 0










O <
o U'ca


*n





o N
60 -r4





0 -0






CI
0


P%4
3po4
(SII~0 Auu.iv NP











I I I II I I I I


(a)


(b)


I I I I


846


BINDING ENERGY (eV)


Figure 5-3.


AES spectra of the In M N 5N45 features taken from
the same surface shown i figure 5-2. Spectra (a)
and (c) were taken at exit angles of 10* and 85*
respectively. Spectrum (b) is angle integrated over
all exit angles.


864


SI I II


858


852











These data were taken in the pulse-counting mode and are

appropriately referenced so that the energy values are accurate. The

spectrum shown in (c) is quite similar in shape to that of In

metall65 and is taken to be representative of InP since most of these

electrons are originating deeply below the surface. The kinetic

energy of the M5N45N45 transition is 400.6 eV which is about 2.0 eV

smaller than values reported for In metal.165,166 This value is also

about 0.6 eV larger than values reported for In203 166 and agrees with

the value found by Clark et al.151 for InP. Spectrum (a) is much

broader and less well defined in comparison with spectrum (c). Based

on mean-free path estimates, most of these electrons originate within

10A of the surface since the exit angle is only 100. It appears that

another set of peaks due to another In species is contained in

spectrum (a). The kinetic energy of the M5N45N45 transition for this

species is about 397.1 eV. It is argued below that this species is a

phosphate and probably InPO3 rather than InP04. Even though the

outermost layer is mostly In, it is probably bonded as InPO3 which is

present in an amount almost as large as the amount of InP in the

near-surface region. Spectrum (a) is similar to the high-resolution

AES spectrum of the primary In features of an InP(100) surface which

was etched in a 20% solution of Br2/HBr/H20 (1:17:35).151

Figures 5-4 and 5-5 show high resolution XPS spectra of the P

2p, In 3d, C Is and 0 is peaks at exit angles of 100 and 850. A very

large difference is seen between spectra taken at 100 and 850 for the

P 2p peaks. The major P 2p feature observed at 129 eV in figure 5-






























































c-


- '---

-/-,


SliNn AUVE1S1yV (3)N


4-4
0
m
Q)


Cu


x
4
(U




Co


4d
Uc
(U



Cu
B


a
4-1


Cu

pq


0
cq
M










0J




00

o
u














4-1~
U
CMl









If
PLn













a)
*1-4
F
^^
4- ^






























C%4


co

-1



c. o


ua



z -1
J\ 'O n








z
4.4W





00
Z ca *
p a







1
(SI) 3V (











4A(b) is assigned as phosphorous in InP. This is in agreement with

other studies144,147,151,152 and the value assigned to P 2p in

GaP. At the surface-sensitive exit angle of 100 shown in figure

5-4A(a), another P 2p peak appears at a binding energy of 133.7 eV

which is nearly as large as the P peak due to InP. This is assigned

as a phosphate species and probably is InPO3 (or InPO4), but it is

difficult to rule out the presence of hydrogen-containing phosphate

species. The assignment as InPO3 seems somewhat more probable than

InP04 based on an analogy with NaPO3 or (NaPO3)3 and Na3PO4 which

have binding energies of 134, 133.5 and 132.2 eV respectively. A

study of the In-P-0 phase diagram by Schwartz et al.167 also suggests

that InPO3 forms at low temperatures while InP04 is stable at

elevated temperatures (>650 oC). Based on mean-free path arguments,

it is estimated that about 95% of the electrons collected are emitted

within 50A of the surface in the 850 exit-angle spectra and within

10A of the surface in the 100 exit-angle spectra. Thus, the spectra

shown in figure 5-4A imply that the near-surface region is composed

of InP and InPO3 in similar amounts and that the bulk is InP. The In

3d spectra shown in figure 5-4B support this hypothesis. The

spectrum shown in (b) is characteristic of In in InP at a binding

energy of 444.7 eV. The surface-sensitive spectrum shown in figure

5-4B(a) contains another In species which is assumed to be a

phosphate at a binding energy of about 445.5 eV.147 This is based on

the information about P contained in figure 5-4A(a). In in In203 has

a 3d binding energy which is about equal to that of In in InP so it












is not possible to determine if In203 is present from the spectra

shown in figure 5-4.

The C is and 0 is XPS features are shown in figure 5-5A and B

respectively for exit angles of (a) 100 and (b) 850. The C is peak

lies at a binding energy of 285.2 eV, in agreement with Hollinger et

al.168 and is most likely in the form of graphite or hydrocarbons.

It should be noted that this binding energy is about 0.6 eV higher

than the commonly reported reference values for these species.166

Therefore, selecting carbon as the reference for peak energy

assignments at a binding energy of 284.6 eV would shift all the

values reported here to lower binding energies by 0.6 eV. The C 1s

peak shown in figure 5-5A(b) is broader than that shown in (a) so

more than one form of C may be present. It is possible that

discrepancies in the C is peak position are due to charging

effects. However, these are believed to be minimal for the sample

used in this study due to its relatively high conductivity and the

thinness of the oxide layer.

XPS 0 is peaks are shown in figure 5-5B. These peaks are

discussed assuming that they are composed of a peak due to phosphate

at a binding energy of 532.0 eV and a peak due to In203 at a binding

energy of 530.5 eV. This seems reasonable with respect to both the

results of this analysis and previous studies.'144,147,152 The

surface-sensitive spectrum shown in figure 5-5B(a) is almost entirely

due to phosphate. However, the presence of a very small shoulder on

the low binding energy side, which is difficult to discern in the












figure, may indicate that a very small amount of In203 is present in

the near-surface region. The more bulk-sensitive spectrum shown in

figure 5-5B(b) shows a very slight shift to lower binding energy.

This may indicate that the concentration of In203 is greater with

respect to the phosphate at increasing depth below the surface, but

the majority of oxygen-containing species throughout the oxide layer

is phosphate.

Angle-resolved valence band XPS spectra are shown in figure 5-

6. The features of the bulk-sensitive spectrum shown in figure 5-

6(b) are in complete agreement with those shown in the spectrum

presented by Vesely and Kingston69 for clean InP. Vesely and

Kingston found good agreement between the experimental spectrum and

the theoretical results of Stukel et al.170 and Kramer et al.171 The

characters of the peaks are assigned as: 3 eV In 5s, 7 eV 0 2p

and 9.7 eV P 3p, and other peaks which are not shown are assigned

as: 14 eV P 3s, 18 eV In 4d, 23 eV 0 2s.166,172 The surface-

sensitive spectrum shown in figure 5-6(a) is quite similar to the

spectrum shown in figure 5-6(b) in the 5 to 12 eV region. However,

the broad feature at about 3 eV due to In 5s electrons is absent in

the surface-sensitive spectrum. This implies that In is bonded

differently in InP than in InPO 3.

Depth profiles for In, P, 0 and C are shown in figure 5-7.

These were obtained by performing angle-integrated AES while sputter

etching with argon ions. The atomic concentrations were determined

using Auger sensitivity factors. However, due to possible matrix





78










U)



I I I
>_ y(a)













10 0


BINDING ENERGY (eV)
Figure 5-6. Angle-resolved valence band XPS
spectra taken at exit angles of
(a) 100 and (b) 850.






79





















co
uC
4-1



0




Cu
(a















S -HC



0 0
0




h%
to 0




I e o




| OA








M NOUV81N30N*O OIWOIV












effects and sputtering artifacts, this figure shows only trends in

the concentrations. The profiles are shown as a function of sputter

time, but from the angle-resolved XPS results, it is estimated that

the sputtering rate was about 50A/min. The In/P ratio is greater

than 1 at the surface, but the large ratio found using ISS is not

observed since angle-integrated AES is less surface sensitive than

ISS. The P concentration increases rapidly suggesting that the

region below the outermost layer is P rich. The In, P, C and 0

concentrations all change rapidly in the oxide layer, and a sharp

interface is observed at a sputtering time of 0.5 minutes. Past this

interface only In and P appear. The layer beneath the oxide is P

rich, but the concentration of In increases while the concentration

of P decreases. Although the In and P concentrations change slowly

in this region, it appears that a composition corresponding to InP

may be approached.



Conclusions

The oxide layer which forms on a chemically polished and etched

InP(111) surface upon exposure to air is very thin (estimated to be

about 30A thick) and nonhoniogeneous. ISS shows that the outermost

layer is mostly In and AES combined with the ISS results suggests

that the region just below the surface layer is P rich. Indium is

believed to exist in the near-surface region as InPO3 and InP in

similar concentrations, but a very small amount of In203 is detected

in the subsurface region. Carbon is detected throughout the native







81




oxide layer, and XPS shows that it is present in graphitic or

hydrocarbon forms. Depth profiling using Ar+ and AES shows that the

oxide/InP interface is sharply defined. The region below the oxide

layer is P rich, but the composition appears to be approaching that

of InP as the sputtering time increases.
















CHAPTER VI
GENERAL CONCLUSIONS AND
RECOMMENDATIONS FOR FUTURE RESEARCH

Hydrogen is found to be present on cleaned and oxidized Si(100)

and GaAs(100) as well as cleaned and nitrogen-exposed Si(111). The

source of the surface hydrogen detected using ESD is the bulk of the

semiconductor which acquires hydrogen during growth, polishing and/or

cleaning processes. It would be helpful to substitute deuterium

where possible in the formation of the polished single crystal wafers

to determine the step(s) during which hydrogen is incorporated.

ESD is one of the relatively few surface analytical techniques

which detects hydrogen directly. Energy analysis of the desorbing

ions is very useful in distinguishing between different surface

binding sites of hydrogen. It was found that multiple states of

adsorbed hydrogen are present on cleaned and oxidized Si(100). The

ESD study of the Si(100) surface also revealed that the sample

treatment can dramatically alter the energy distribution of the

desorbing hydrogen ions. Electron beam exposure results in a

depopulation of higher energy desorption states which are repopulated

upon annealing. Oxygen exposure results in a more marked

depopulation of the higher energy desorption states. It was also

found from energy-resolved, angle-resolved ESD spectra that different

hydrogen bonding states desorb at different angles relative to the

surface plane with differing energy distributions. Further research












is necessary to more fully characterize the states of adsorbed

hydrogen on the cleaned and oxidized Si(100) surface. ESD threshold

studies would be helpful in this endeavor.

In the ISS and TPD study of the interaction between nitrogen and

a Si(111)(7x7) surface, ISS easily detects adsorbed nitrogen

following doses as low as 2 L and reveals that a saturation dose of

N2 at room temperature is 75 L. Using TPD, evidence of at least two

chemisorbed states of nitrogen is found. Hydrogen, which originates

from the bulk of the sample, is found to associate with the surface

nitrogen. No adsorbed nitrogen is detected using AES for doses as

high as 3000 L. Angle-resolved AES, a technique developed by

Hoflund173 which allows for greatly enhanced surface sensitivity

using AES, may be useful in determining whether the nitrogen is

present only on the surface or penetrates into the silicon.

From the ESD study of the GaAs(100), it was verified that

molecular deuterium does not adsorb at significant rates at room

temperature on the cleaned or oxidized surface, but that it does

adsorb on both surfaces when the sample is subjected to radiant

heating in the range 200-400 OC. There appears to be two states of

hydrogen on oxidized GaAs(100). There also appears to be two states

on the cleaned GaAs(100) surface, but the higher energy state has a

lower desorption cross section or is less densely populated. ESD

gives a relative measure of the amount of adsorbed deuterium. Again,

further research is needed to fully characterize the states of

hydrogen on the surface.












The native oxide layer present on chemicallly poslished and
0
etched InP(111) is approximately 30A thick. A sharp interface

exists between the oxide and the bulk InP. ISS reveals that the

outermost atomic layer is In rich, and AES combined with the ISS

shows that the region beneath this is P rich. InP and InPO3 are

present in approximately equal concentrations, and a small amount of

In203 is detected. Carbon is present throughout the oxide layer as

graphite and/or hydrocarbons. Similar studies of native oxide

layers formed under different polishing and/or etching conditions

would be useful. Additionally, it is suspected that hydrogen is

present in the oxide layer. It would be interesting to perform an

ESD study to determine whether or not this is true and to

characterize the hydrogen.

The studies performed have been successful in obtaining an

understanding of a number of interactions between gases and

semiconductors. Further research would be helpful in gaining a more

thorough understanding of these interactions. Ultimately, it is also

very important to correlate the information gained using surface

analytical techniques with a characterization of the electronic

properties of the systems studied.
















REFERENCES

1. G.B. Hoflund, D.F. Cox, G.L. Woodson and H.A. Laitinen, Thin
Solid Films 78, 357 (1981).

2. C.H. Seager and D.S. Ginley, Appl. Phys. Lett. 34, 337 (1979).

3. J.L. Benton, C.J. Doherty, S.D. Ferris, D.L. Flamm, L.C.
Kimerling and H.J. Leamy, Appl. Phys. Lett. 36, 670 (1980).

4. S.J. Pearton and A.J. Tavendale, Phys. Rev. B 26, 7105 (1982).

5. C.-T. Sah, J.Y.-C. Sun and J.J.-T. Tzou, Appl. Phys. Lett. 43,
204 (1983).

6. C.-T. Sah, J.Y.-C. Sun and J.J.-T. Tzou, J. Appl. Phys. 54,
5864 (1983).

7. J.I. Pankove, D.E. Carlson, J.E. Berkeyheiser and R.O. Wance,
Phys. Rev. Lett. 51, 2224 (1983).

8. W.L. Hansen, S.J. Pearton and E.E. Haller, Appl. Phys. Lett.
44, 606 (1984).

9. J.I.. Pankove, R.O. Wance and J.E. Berkeyheiser, Appl. Phys.
Lett. 45, 1100 (1984).

10.. N.M. Johnson, Phys. Rev. B 31, 5525 (1985).

11. J.I. Pankove, P.J. Zanzucchi, C.W. Magee and G. Lucovsky,
Appl. Phys. Lett. 46, 421 (1985).

12. N.M. Johnson and M.D. Moyer, Appl. Phys. Lett. 46, 787 (1985).

13. S.J. Pearton, Appl. Phys. Lett. 40, 253 (1982).

14. J. Lagowski, M. Kaminska, J.M. Parsey, H.C. Gatos and M.
Lichtensteiger, Appl. Phys. Lett. 41, 1078 (1982).

15. S.J. Pearton, J. Appl. Phys. 53, 4509 (1982).

16. J. Chevallier, W.C. Dautremont-Smith, C.W. Tu and S.J.
Pearton, Appl. Phys. Lett. 47, 108 (1985).

17. S.J. Pearton, W.C. Dautremont-Smith, J. Chevallier, C.W. Tu
and K.D. Cummings, J. Appl. Phys. 59, 2821 (1986).












18. Y.-C. Pao, D. Liu, W.S. Lee and J.S. Harris, Appl. Phys. Lett.
48, 1291 (1986).

19. M.M. Moslehi and K.C. Saraswat, IEEE Trans. Electron. Devices
ED-32, 106 (1985).

20. L. Messick, J. Appl. Phys. 47, 4949 (1976).

21. D. Fritsche, Electron Lett. 14, 51 (1978).

22. L. Meiners, J. Vac. Sci. Technol. 19, 373 (1981).

23. T. Kawakami and M. Okamura, Electron Lett. 15, 502 (1979).

24. J. Magarino, D. Kaplan, A. Friederich and A. Deneuville,
Philos. Mag. B 45, 285 (1982).

25. G.G. LeLeo and W.B. Fowler, J. Electron. Mater. 14a, 745
(1985).

26. T. Sakurai and H.D. Hagstrum, Phys. Rev. B 14, 1593 (1976).

27. K.C. Pandey, J. Vac. Sci. Technol. 15, 440 (1978).

28. W. M6nch P. Koke and S. Krueger, J. Vac. Sci. Technol. 19,
313 (1981).

29. K. Fujiwara, Phys. Rev. B 26, 2036 (1982).

30. N.G. Norton, Vacuum 33, 621 (1983).

31. H. Froitzheim, U. KOhler and H. Lammering, Surf. Sci. 149,
537 (1985).

32. N.D. Lang and A.R. Williams, Phys. Rev. Lett. 34, 531 (1975).

33. J.A. Appelbaum and D.R. Hamann, Phys. Rev. Lett. 34, 806
(1975).

34. I.P. Batra and S. Ciraci, Phys. Rev. Lett. 34, 1337 (1975).

35. K.C. Pandey, Phys. Rev. B 14, 1557 (1976).

36. I.P. Batra, S. Ciraci and I.B. Ortenburger, Solid State
Commun. 18, 563 (1976).

37. K.M. Ho, M.L. Cohen and M. SchlUter Phys. Rev. B 15, 3888
(1977).











38. J.A. Appelbaum, D.R. Hamann and K.H. Tasso, Phys. Rev. Lett.
39, 1487 (1977).

39. J.A. Appelbaum, G.A. Baraff, D.R. Hamann, H.D. Hagstrum and T.
Sakurai, Surf. Sci. 70, 654 (1978).

40. W.Y. Ching, D.J. Lam and C.C. Lin, Phys. Rev. B 21, 2378
(1980).

41. H. Ibach and J.E. Rowe, Surf. Sci. 43, 481 (1974).

42. J.E. Rowe, Surf. Sci. 53, 461 (1975).

43. T. Sakurai and H.D. Hagstrum, Phys. Rev. B 12, 5349 (1975).

44. K.C. Pandey, T. Sakurai and H.D. Hagstrum, Phys. Rev. Lett.
35, 1728 (1975).

45. H. Froitzheim, H. Ibach and S. Lehwald, Phys. Lett. 55A, 247
(1975).

46. T. Sakurai, K.C. Pandey and H.D. Hagstrum, Phys. Lett. 56A,
204 (1976).

47. T. Sakurai and H.D. Hagstrum, J. Vac. Sci. Technol. 13, 807
(1976).

48. S.J. White and D.P. Woodruff, J. Phys. C9, L451 (1976).

49. S.J. White and D.P. Woodruff, Surf. Sci. 63, 254 (1977).

50. T. Sakurai, T.T. Tsong and R.J. Culbertson, J. Vac. Sci.
Technol. 15, 647 (1978).

51. S.J. White, D.P. Woodruff and L. McDonnell, Surf. Sci. 72, 77
(1978).

52. S.J. White, D.P. Woodruff, B.W. Holland and R.S. Zimmer, Surf.
Sci. 74, 34 (1978).

53. L.C. Feldman, P.J. Silverman and I. Stensgaard, Nucl. Instr.
Methods 168, 589 (1980).

54. H.H. Madden, Surf. Sci. 105, 129 (1981).

55. H. Wagner, R. Butz, U. Backes and D. Bruchmann, Solid State
Commun. 38, 1155 (1981).

56. H.H. Madden, D.R. Jennison, M.M. Traum, G. Margaritondo and
N.G. Stoffel, Phys. Rev. B 26, 896 (1982).













57. S. Maruno, H. Iwasaki, K. Horioka, S.-T. Li and S. Nakamura,
Surf. Sci. 123, 18 (1982).

58. T.S. Shi, S.N. Sahu, G.S. Oehrlein, A. Hiraki and J.W.
Corbett, Phys. Stat. Sol. 74, 329 (1982).

59. S. Maruno, H. Iwasaki, K. Horioka, S.-T. Li and S. Nakamura,
Phys. Rev. B 27, 4110 (1983).

60. H.H. Madden, J. Vac. Sci. Technol. A 2, 961 (1984).

61. J.A. Schaefer, F. Stucki, J.A. Anderson, G.J. Lapeyre and W.
Gopel, Surf. Sci. 140, 207 (1984).

62. D. Schmeisser, Surf. Sci. 137, 197 (1984).

63. Y.J. Chabal, J. Vac. Sci. Technol A 3, 1448 (1985).

64. Y.J. Chabal and K. Raghavachari, Phys. Rev. Lett. 54, 1055
(1985).

65. M. Nishijima, K. Edamoto, Y. Kubota, H. Kobayashi and M.
Onchi, Surf. Sci. 158, 422 (1985).

66. C. Kleint, Vacuum 36, 267 (1986).

67. L.L. Kazmerski, in Eighteenth IEEE Photovoltaic Specialists
Conference (1985), p. 993.

68. C.F. Corallo and G.B. Hoflund, J. Vac. Sci. Technol. A 5, 713
(1987).

69. M.M. Traum and D.P. Woodruff, J. Vac. Sci. Technol. 17, 1202
(1980).

70. G.B. Hoflund, SEM J. 4, 1391 (1985).

71. R. Jaeger, J. St5hr and T. Kendelewicz, Phys. Rev. B 28,
1145 (1983).

72. J.W. Corbett, S.N. Sahu, T.S. Shi and L.C. Snyder, Phys. Lett.
93A, 303 (1983).

73. J.S. Johannessen and W.E. Spicer, Thin Solid Films 32, 311
(1976).

74. H.G. Maguire and P.D. Augustus, J. Electrochem. Soc. 119, 791
(1972).












75. R. Heckingbottom, G.W.B. Ashwell, P.A. Leigh, S. O'Hara and
C.J. Todd, Proc. Electrochem. Soc. 78, 419 (1978).

76. H.H. Madden and P.H. Holloway, J. Vac. Sci. Technol. 16, 618
(1979).

77. N.C. Tombs, F.A. Sewell, Jr. and J.J. Comer, J. Electrochem.
Soc. 116, 862 (1969).

78. M.J. Rand and J.F. Roberts, J. Electrochem. Soc. 120, 446
(1973).

79. D.M. Brown, P.V. Gray, F.K. Heumann, H.R. Philipp and E.A.
Taft, J. Electrochem. Soc. 115, 311 (1968).

80. A. Atkinson, A.J. Moulson and E.W. Roberts, J. Am. Ceram. Soc.
59, 285 (1976).

81. A. Atkinson, A.J. Moulson and E.W. Roberts, J. Mater. Sci. 10,
1242 (1975).

82. S.M. Hu, J. Electrochem. Soc. 113, 693 (1966).

83. R.G. Frieser, J. Electrochem. Soc. 115, 1902 (1968).

84. M.I. Kamchatka and B.F. Ormont, Russian J. Phys. Chem. 45,
1246 (1971).

85. T. Ito, T. Nozaki, T. Nakamura and H. Hishikawa, Appl. Phys.
Lett. 38, 370 (1981).

86. T. Ito, S. Hijiya, H. Arakawa, T. Nozaki, M. Shinoda and Y.
Fukukawa, J. Electrochem. Soc. 125, 449 (1978).

87. T. Ito, T. Nozaki, H. Arakawa and M. Shinoda, Appl. Phys.
Lett. 32, 330 (1978).

88. S.M. Murarka, C.C. Chang and A.C. Adams, J. Electrochem. Soc.
126, 996 (1979).

89. A.G. Schrott and S.C. Fain, Jr., Surf. Sci. 111, 39 (1981).

90. A.G. Schrott and S.C. Fain, Jr., Surf. Sci. 123, 204 (1982).

91. J.F. Delord, A.G. Schrott and S.C. Fain, Jr., J. Vac. Sci.
Technol. 17, 517 (1980).

92. C. Maillot, H. Roulet and G. Dufour, J. Vac. Sci. Technol. B
2, 316 (1984).











93. K. Edamoto, S. Tanaka, M. Onchi and M. Nishijima, Surf. Sci.
167, 285 (1986).

94. J.A. Taylor, G.M. Lancaster, A. Igniatiev and J.W. Rabalais,
J. Chem. Phys. 68, 1776 (1978).

95. R. Heckingbottom, The Structure and Chemistry of Solid
Surfaces, edited by G.A. Somorjai (Wiley, New York, 1969), p.
78-1.

96. R.B. Guthrie and F.L. Riley, Proc. Br. Ceram. Soc. 22, 275
(1973).

97. M.D. Wiggins, R.J. Baird and P. Wynblatt, J. Vac. Sci.
Technol. 18, 965 (1981).

98. R. Heckingbottom and P.R. Wood, Surf. Sci. 36, 594 (1973).

99. A.J. van Bommel and F. Meyer, Surf. Sci. 8, 381 (1967).

100. A. Glachant, D. Saidi and J.F. Delord, Surf. Sci. 168, 672
(1986).

101. M. Nishijima, K. Edamoto, Y. Kubota, H. Kobayashi and M.
Onchi, Surf. Sci. 158, 422 (1985).

102. D.-R. He and F.W. Smith, Surf. Sci. 154, 347 (1985).

103. E.G. Keim and A. van Silfhout, Surf. Sci. 152, 1096 (1985).

104. T.M. Buck, Methods of Surface Analysis: Methods and Phenomena
I, edited by A.W. Czanderna (Elsevier Scientific, Amsterdam,
1975), p. 75.

105. E.P.Th.M. Suurmeijer. and A.L. Boers, Surf. Sci. 43, 309
(1973).

106. E.S. Parilis in Proc. 7th Intern. Conf. Phenomena in Ionized
Gases, Belgrade (1965), p. 129.

107. J. Onsgaard, W. Heiland and E. Taglauer, Surf. Sci. 99, 112
(1980).

108. H.F. Dylla, J.G. King and M.J. Cardillo, Surf. Sci. 74, 141
(1978).

109. R.P.H. Gasser, An Introduction to Chemisorption and Catalysis
(Oxford Science, Oxford, 1985), Chpt. 3.

110. C.F. Corallo and G.B. Hoflund, unpublished manuscript.













111. R.C. Weast (ed.) CRC Handbook of Chemistry and Physics, (CRC
Press, Boca Raton, FL, 1979-80).

112. J.F. O'Hanlon, A User's Guide to Vacuum Technology (Wiley, New
York, 1980), Appendix E.

113. A. Cornu and R. Massot, Compilation of Mass Spectral Data
(Heyden & Sons Ltd., London, 1975), Vol. 2.

114. A.R. Calawa, Appl. Phys. Lett. 33, 1020 (1978).

115. H. Mori and S. Takagishi, Jap. J. Appl. Phys. 23, L877 (1984).

116. P.D. Dapkus, H.M. Manasevit and K.L. Hess, J. Cryst. Growth
55, 10 (1981).

117. E. Veuhoff, W. Pletsheu, P. Balk and H. Luth, J. Cryst.
Growth 55, 30 (1981).

118. L.M. Fraas, J. Appl. Phys. 52, 6939 (1981).

119. P. Friedel and S. Gourrier, Appl. Phys. Lett. 42, 509 (1983).

120. R.P.H. Chang and S. Darack, Appl. Phys. Lett. 38, 898 (1981).

121. P. Friedel, P.K. Larsen, A. Gourrier, J.P. Cabanie and W.M.
Gerits, J. Vac. Sci. Technol. B 2, 675 (1984).

122. D.D. Pretzer and H.D. Hagstrum, Surf. Sci. 4, 265 (1966).

123. P.E. Gregory and W.E. Spicer, Surf. Sci. 54, 229 (1976).

124. P.E. Gregory and W.E. Spicer, Phys. Rev. B 12, 2370 (1975).

125. W. Mokwa, D. Kohl and G. Heiland, Phys. Rev. B 29, 6709
(1984).

126. H. Luth and R. Matz, Phys. Rev. Lett. 46, 1652 (1981).

127. L.H. Dubois and G.P. Schwartz, Phys. Rev. B 26, 794 (1982).

128. F. Bartels, L. Surkamp, H.J. Clemens and W. Monch, J. Vac.
Sci. Technol. B 1, 756 (1983).

129. W. Monch, Thin Solid Films 104, 285 (1983).

130. R.Z. Bachrach, R.S. Bauer, P. Chiaradia and G.V. Hansson, J.
Vac. Sci. Technol. 18, 797 (1981).