DTIC ADA482529: Quantification of Discrete Oxide and Sulfur Layers on Sulfur-Passivated InAs by XPS

0.32 MB·English

by  Defense Technical Information Center

SURFACEANDINTERFACEANALYSIS
Surf.InterfaceAnal.2005;37:989–997
PublishedonlineinWileyInterScience(www.interscience.wiley.com).DOI:10.1002/sia.2095
Quantification of discrete oxide and sulfur layers
on sulfur-passivated InAs by XPS
D. Y. Petrovykh,1,2∗ J. M. Sullivan2,3 and L. J. Whitman2
1PhysicsDepartment,UniversityofMaryland,CollegePark,MD20742,USA
2NavalResearchLaboratory,Washington,DC20375-5342,USA
3DepartmentofChemistry,NorthwesternUniversity,Evanston,IL60208,USA
Received8November2004;Revised7April2005;Accepted7April2005
The initial quality and stability in air of InAs(001) surfaces passivated by a weakly-basic solution of
thioacetamide(CH CSNH )isexaminedbyXPS.TheS-passivatedInAs(001)surface canbemodeledas
3 2
a sulfur-indium-arsenic ‘layer-cake’ structure, such that characterization requires quantification of both
arsenicoxideandsulfurlayersthatareatmostafewmonolayersthick.Thisthicknessrangecomplicates
the quantitative analysis because neither standard submonolayer nor thick-film models are applicable.
Therefore,we developadiscrete-layermodelandvalidateitwithangle-resolvedXPSdataandelectron
attenuationlength(EAL)calculations.Wethenapplythismodeltoempiricallyquantifythearsenicoxide
andsulfurcoverageonthebasisofthecorrespondingXPSintensityratios.Copyright2005JohnWiley
&Sons,Ltd.
KEYWORDS:XPS;indiumarsenide;InAs;thioacetamide;oxide;passivation;oxidation;coverage;attenuation
INTRODUCTION theinitialcompositionofTAM-passivatedsurfacesandtheir
stability in air for up to 42days. The remarkably efficient
Surface passivation of III–V semiconductors1–5 has been a
TAMpassivationcreatesasurfacewith<1monolayer(ML)
subjectofintensiveinvestigationforoveradecade,primarily
of AsO –a coverage essentially undetectable in the As 3d
becauseofapplicationsofthesematerialsinmicroelectronics. x
regionmonitoredinourpreviouswork7(1MLD5.41ð1014
Morerecently,chemicalandbiologicalsensinghasemerged
atoms/cm2 forbulk-terminatedInAs(001)).Wehavefound
asanotherpotentialapplicationforIII–Vsemiconductors.6,7
that the more surface-sensitive As 2p photoelectrons are
Inherentdifferencesbetweensensingandelectronicdevices
required for characterization of such small amounts of
notwithstanding,bothcouldbenefitfromtheenhancements
AsO . In the passivation longevity study, the coverage of
provided by surface chemical passivation. For example, x
AsO and S extends from submonolayer to three MLs.
a standard treatment using ammonium sulfide solutions x
Becausetherangecannotbecoveredbyanysinglestandard
[(NH ) S ]2,3producesS-passivatedInAssurfaceswithwell-
4 2 x approximationusedforXPSanalysis,wehavedevelopeda
defined chemical and electronic properties, and short-term
discrete-layer(DL)model.Herewedescribethedetailsofthe
stability in air and aqueous solutions–a combination of
DLmodel,includingitsapplicationtodatafromelemental
properties which is required for sensing applications and
core levels (As 2p, In 3d, S 2p), and the corresponding
whichisalsoadvantageousformicroelectronicsprocessing
empiricalAsO andScoveragecalibrations.Theapplication
and fabrication.7 One particularly promising alternative to x
of the DL model for quantitative comparison between the
inorganic sulfide passivation is treatment with solutions
TAM passivation and the (NH ) S benchmark is reported
of thioacetamide (CH CSNH or TAM hereafter).8,9 For 4 2 x
3 2 elsewhere.11
InAs(110), basic aqueous solutions of this organic sulfide
havebeenreportedtoproducesmallerroughnessandmore
stabletunnelingcurrent(inscanningtunnelingmicroscopy EXPERIMENTAL
andspectroscopy)thanthe(NH ) S passivation.10
4 2 x
ThioacetamidepassivationofInAs(001)samples
To examine the benefits of the TAM organic sulfide
as a practical passivation approach for the technologically InAs(001) samples(³1cm2)weredicedfromacommercial
important InAs(001) surfaces, we use XPS to characterize single-side polished wafer (undoped, intrinsically n-type).
WeusedcommercialACSreagentgrade99.0%thioacetamide
(CH CSNH )andACSPLUSgrade29.7%aqueoussolution
ŁCorrespondenceto:D.Y.Petrovykh,Code6177,NavalResearch 3 2
ofNH OH.In-housetriple-distilledwaterwasusedtodilute
Laboratory,Washington,DC20375-5342,USA. 4
E-mail:[email protected] NH OH and rinse samples. We have explored a range of
4
Contract/grantsponsor:OfficeofNavalResearch. the TAM and NH OH concentrations and found that the
4
Contract/grantsponsor:AirForceOfficeofScientificResearch.
optimalpassivationwasprovidedbythefollowingsolution:
Contract/grantsponsor:DefenseAdvancedResearchProjects
Agency. 0.2gofTAMpowderdissolvedin15mlofthe1:9volume
Copyright2005JohnWiley&Sons,Ltd.
Report Documentation Page Form Approved
OMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it

does not display a currently valid OMB control number.
1. REPORT DATE 3. DATES COVERED
APR 2005 2. REPORT TYPE 00-00-2005 to 00-00-2005
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Quantification of discrete oxide and sulfur layers on sulfur-passivated
5b. GRANT NUMBER
InAs by XPS
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
Naval Research Laboratory,4555 Overlook Avenue REPORT NUMBER
SW,Washington,DC,20375
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)
11. SPONSOR/MONITOR’S REPORT
NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release; distribution unlimited
13. SUPPLEMENTARY NOTES
14. ABSTRACT

15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF
ABSTRACT OF PAGES RESPONSIBLE PERSON
a. REPORT b. ABSTRACT c. THIS PAGE Same as 10
unclassified unclassified unclassified Report (SAR)
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39-18
990 D.Y.Petrovykh,J.M.SullivanandL.J.Whitman
mixtureofthe29.7%NH OHstocksolutionandwater.This identification imply endorsement by the Naval Research
4
standardTAMsolutionwasusedtoprepareallthesamples Laboratory.) Nominal X-rayspot size and analyzerfield of
forthepassivationlongevityseries.Incontrasttopassivation viewwere(cid:1)1mm2.Thebindingenergies(BE)arereported
by(NH ) S ,wefoundthataddingelementalsulfur2,3,7 did with 0.1eV precision, on the basis of a two-point analyzer
4 2 x
notimprovetheefficiencyofpassivationbyTAM. energy calibration described in detail elsewhere.12 Two
Before passivation, InAs samples were degreased in types of normal-emission angle-integrated surveys were
acetoneandethanol(2minineachsolvent),rinsedinwater, used to monitor samples for presence of contaminants:
andblowndryunderflowingnitrogen.ThestandardTAM 0–1400eVBErange(1.8eVanalyzerresolution,1eVpoint
solutionswereheatedtojustbelowtheboilingpoint(³78°C) spacing),and0–550eVBErange(0.9eVanalyzerresolution,
in loosely capped glass vials placed in a waterbath. Upon 0.33eV point spacing). High-resolution normal-emission
heating,solutionsbecameslightlyyellow,andapHbetween angle-integrated scans were acquired for the As 3d, In 3d,
11.0and11.5wastypicallymeasured(usingpHpaper)after O 1s, C 1s, S 2p, and As 2p regions, with 0.36eV analyzer
completing the passivation. Following a 4-min passivation resolution (0.9eV for As 2p). In addition, high-resolution
in the standard TAM solution, each sample was rinsed for angle-resolved spectra were acquired at 35° and 65° (off-
2min in copious amounts of water and blown dry under normal)emissionanglesfortheAs3d,2p,andIn3dregions.
flowingnitrogen. The nominal acceptance angles were 4° and 30° (along the
The shortest exposure to air was ³5min for a TAM- energydispersiveandnondispersivedirectionsrespectively)
passivated sample that was transferred into an ultra-high for normal-emission angle-integrated measurements, and
vacuum (UHV) XPS chamber immediately after the final 5° for angle-resolved measurements. No specific effort
rinseinwater.Thelongevitydatawerecompiledfromthree was made to ensure a particular azimuthal alignment; all
separately prepared samples, each one measured two or sampleswerepositionedsuchthatthetiltinangle-resolved
threetimesatdifferenttimeintervalsupto42daysandstored measurements was approximately toward one of the easy
inacoveredplasticwafertraybetweenthemeasurements. cleavage directions, i.e. [110] or [110]. XPS measurements
were carried out at room temperature in a UHV chamber
XPSmeasurements withbasepressureof1ð10(cid:2)9 Torrwithoutanyadditional
XPS measurements were performed in a commercial XPS sampletreatment.
system (Thermo VG Scientific Escalab 220i-XL) equipped
withamonochromaticAlK˛source,ahemisphericalelectron XPSpeakfitting
energyanalyzer(58°anglebetweenthemonochromatorand The peaks in the elemental core-level spectra were fit
analyzer), and a magnetic electron lens. (Certain vendors using commercial XPS analysis software.13 A convolution
and commercial instruments are identified to adequately of Lorentzian and Gaussian line shapes was used to fit
specify the experimental procedure. In no case does such the individual peaks. A linear combination of Shirley and
Table1. Peakparametersfromfitstohigh-resolutionelementalXPSdataforanInAs(001)surfaceS-passivatedbytheTAM
treatment
FWHM(eV)
Emission Spin-orbit Spin-orbit
Peak Component angle(°) BEa(eV) intensityratio splitting(eV) Lorentzian Gaussian
S2p 0 161.6 0.52 1.17 0.12 1.00
In3d In-As 0 444.5 0.67 7.55 0.32 0.51
In-S 0 445.0 0.67 7.55 0.32 0.51
In-O 0 445.3 0.67 7.55 0.32 0.85
x
In-As 35 444.6 0.69 7.56 0.31 0.49
In-S 35 445.1 0.69 7.56 0.31 0.49
In-O 35 445.4 0.69 7.56 0.31 0.76
x
In-As 65 444.7 0.70 7.55 0.29 0.53
In-S 65 445.2 0.70 7.55 0.29 0.53
In-O 65 445.4 0.70 7.55 0.29 0.95
x
As3d As-In 0 40.7 0.68 0.71 0.13 0.57
As-In 35 40.8 0.69 0.70 0.12 0.54
As-In 65 40.9 0.68 0.70 0.23 0.46
As2p3/2 As-In 0 1323.1 0.5 1.3
As-O 0 1324.8 0.5 2.0
x
As-In 35 1323.4 0.6 1.2
As-O 35 1325.4 0.6 2.0
x
As-In 65 1323.4 0.5 1.2
As-O 65 1325.3 0.5 2.3
x
aForspin-orbitdoublets,theBEisgivenforthehigherintensitycomponent.
Copyright2005JohnWiley&Sons,Ltd. Surf.InterfaceAnal.2005;37:989–997
QuantificationofoxideandSlayersonS-passivatedInAsbyXPS 991
polynomial functions was used to model the inelastic As shown by the In 3d fits in Plate2(b), the In-S
5/2
electronbackground,withthecorrespondingcoefficientsfit component slowly decreases with increased air expo-
simultaneouslywiththepeaks.Polynomialtermsuptothe sure, while the In-O component increases. The increas-
x
secondorderwererequiredtofitthenonlinearbackground ing intensity of the high BE shoulder in the In 3d
5/2
in the In 3d, As 2p, and S 2p regions. The In-As, In-S, and data makes deconvolution of the In-S and In-O com-
x
In-O chemical components in the In 3d doublet were fit ponents ambiguous, but consistent fits that agree both
x
simultaneouslyforbothIn3d5/2andIn3d3/2peaks.Inthese with the increasing oxidation trend seen in the As 2p3/2
fits,thespin-orbitsplittingandintensityratiooftheIn3d5/2 data in Plate2(a) and the slowly decreasing total S cover-
andIn 3d3/2 peakswereleft asfreeparameters,but theBE age seen in Plate2(c) can be obtained. The S 2p data in
shifts and relative intensities of the chemical components Plate2(c) suggest that the loss of S from passivated sur-
were constrained to be identical between the In 3d5/2 and facesexposedtoairproceedsthroughformationofvolatile
In3d3/2 envelopes. Accordingly, only the fitting results for compounds, since S-Ox components do not appear in the
theIn3d5/2peakarepresentedinthepaper. S 2p region at any point in the passivation longevity
series.
XPSRESULTS
STRUCTUREMODELANDPEAK
We findthattheTAMpassivation treatmentisremarkably
ASSIGNMENTS
efficient in removing the native oxide and preventing
reoxidation–two key objectives of passivation.7 Efficient ‘Layer-cake’structuremodel
native oxide removal is crucial because InAs samples A simple ‘layer-cake’ structure model has been proposed
are placed directly into the TAM solution without a to describe the S-passivated InAs(001) surface, whereby a
preceding HCl-etch step typical for GaAs passivation.1,4,5 stack of alternating In and As atomic layers is passivated
The ability to prevent reoxidation and contamination in by a layer of S chemisorbed on the In-terminated surface
ambientdeterminesthetimeavailableforcarryingoutany (Fig.1(a)).7Themodelisconsistentwithchemicaldatafrom
additionalchemicalsteps. XPS,aswellasstructuralinformationfromcoaxial-impact-
ThedatainPlate1demonstratetheinitialhigh-qualityof collisionion-scatteringspectroscopyandelectron-diffraction
TAM-passivatedInAs(001).Thesurveyspectrum(Plate1(a)) observations.3,7,15,17 This is an unusually simple structure
is essentially identical to the one previously reported for a S-passivated III-V semiconductor. In contrast, on the
for (NH ) S -passivated InAs(001) (compare to Fig.1 in prototypicalS-passivatedGaAs(001)surface,bothGa-Sand
4 2 x
Ref.7). The intensities of the surface contamination C 1s As-Scomponentsareobserved,alongwithelementalAs.4,5,9
and O 1s peaks are comparable to those observed on a The nearly ideal In-S termination on InAs(001) is likely a
reference UHV-cleaved InAs(110) sample,14 demonstrating resultofsolubilitydifferencesbetweenIn-SandAs-Sinthe
how effectively the TAM-passivated surface resists both basicpassivatingsolutions.7
carbon and oxygen contamination in the ambient. The Qualitatively, the chemical information from XPS data
observedS2pBEof161.6eVandfull-widthhalf-maximum in Plate1 support the extension of the ‘layer-cake’ model
(FWHM) of 1.1eV (Plate1(b), Table1) are typical for a to TAM-passivated InAs(001). The BE and FWHM of
chemisorbedSlayeronS-passivatedInAs.15,16 FortheIn3d the S 2p peak (Plate1(b), Table1) are consistent with a
peaks in Plate1(d), the surface In-S and In-O components disordered chemisorbed S layer.15,16 Deconvolution of the
x
appearashighBEshoulders,theintensityofwhichincreases angle-resolved In 3d data indicates the presence of an In-S
in the off-normal spectra. The fit shown for the In 3d component (Plate1(d), Table1), while an As-S component
65° emission data (Plate1(d), Table1) is consistent with is not observed (Plate1(c) and 1(e), Table1). The structure
previously reported chemical shifts relative to the bulk model and consideration of potential oxidation pathways
In-As component: 0.45–0.50eV for In-S and 0.7–1.0eV for (Fig.1)suggestthatdevelopingamethodforquantification
In-O .7,15,16
x
Thesalientfeatureoftheangle-resolvedAs3dspectrais
S-Passivated InAs(001)
theabsenceofAs-O (BED44–45eV)andAs-Scomponents
x
(BE ³ 43eV),7,15,16 even in the 65° emission spectra where Layer-Cake Model Oxidation Pathways
surface sensitivity is enhanced. The absence of significant S S S S S O S S O
As-O confirmsthehighefficiencyoftheTAMpassivation. IInn IInn IInn IInn IInn IInn O In In In In In
x
The absence of an As-S component is also consistent with AAss AAss AAss AAss AAss As As As As O
exclusive S binding to In, in agreement with a previously IInn IInn IInn IInn IInn IInn In In In In In In
proposed ‘layer-cake’ structure model.7 In the surface- (a) (b)
sensitiveAs2pregion(Plate1(c)),asmallAs-O component
x
thatisassociatedwithsurfaceoxideisobserved,asindicated Figure1.SchematicofaS-passivatedInAs(001)surface.
bytheincreasingAs-O /As-Inintensityratioinoff-normal (a)Theidealized‘layer-cake’structuremodel:alternating
x
data. The same As-O /As-In intensity ratio in the As 2p InandAsatomiclayerswiththetopInlayerterminatedbythe
x
region can be used to track the long-term reoxidation passivatingSlayer.Notethattheidealizedstructureinvolves
following the TAM passivation, as shown in Plate2(a) by exclusivelyIn-SandnotAs-Sbonding.(b)Potentialoxidation
representative spectra covering air exposure from 5min to pathways:displacementofSbyO,oxidationofdefectsinthe
42days. topInlayer,oxygendiffusionthroughtheprotectivelayer.
Copyright2005JohnWiley&Sons,Ltd. Surf.InterfaceAnal.2005;37:989–997
992 D.Y.Petrovykh,J.M.SullivanandL.J.Whitman
of the AsO and S coverages is important for quantitative Surface-sensitiveAs2ppeaks:oxidationversus
x
analysisofthesurfacestructureasafunctionoftreatmentand sulfidizationsignatures
time.First,weconsidertheappropriatechoicesofreference TheAs2ppeaks(As2p showninPlates1(c)and2(a))are
3/2
peaks for such quantification in the following text. In the themostsurface-sensitiveintheAsspectrumduetothelow
‘Quantitative XPS Analysis’ section, we introduce a DL KEofAs2pphotoelectrons.Infact,thecorrespondingEAL
modelandthenuseittointerprettheexperimentalcore-level is comparable to the InAs lattice constant (see Appendix),
peakratiosandtoproduceempiricalAsOx andScoverage so thedetected As 2p photoelectrons originate almost
calibrations. exclusively from the top few atomic layers and are very
sensitive to surface composition. For example, the oxide
Peakassignmentsandproperties:As3d,S2p,andIn3d
component in Plate1 can be clearly observed in the As 2p
The As 3d peak offers three features desirable for use in region but not in the As 3d region, in agreement with a
characterization of S-passivated InAs: high intensity, nar- previous report for residual oxide layers on GaAs.18 As 2p
row FWHM, and large chemical shifts between the As-In, photoelectrons are also sensitive to surface contamination,
As-SandAs-Oxcomponents(Plate1(e),Table1).Thesewere e.g. note the lower As 2p intensity from the passivated
the reasons for previously choosing the As3d region for sample after 42days in air versus the native oxide control
characterization of reoxidation and band bending in our in Plate2(a) (two bottom spectra), the former sample had
(NH4)2Sx-passivated InAs(001) study.7 The superior effi- accumulated about three times more adventitious carbon,
ciency of the TAM passivation in preventing reoxidation, andthustheAs2psignalisstrongerattenuated.
however, results in As-Ox components close to the quanti- The interpretation of the high BE As 2p components
ficationlimitintheAs3dregion–evenafterafewdaysin warrants a special discussion. The native oxide control
air–andthusmakestheAs3dregionunsuitableforquanti- in Plate2(a) shows a component shifted by ¾4eV. A
tative analysis of reoxidation. The stability of the As 3d comparisonwiththecorrespondingAs3ddata(notshown)
signalactuallymakesitusefulasaninternalreference.As3d and literature values suggests that this component is a
photoelectronshavethehighestkineticenergy(KE)among mixtureofAs3CandAs5Coxides.18,19Thepassivatedsamples
elemental core-levels in our dataset, and the correspond- exposedtoairformorethanonedayshowthesame¾4eV
ing electron attenuation length (EAL) is about five times chemicalshift,indicatingsimilaroxidationstates(Plate2(a)).
largerthanthe0.606nmInAslatticeconstant(seeAppendix, However, the freshly passivated sample shows a high BE
TableA1),makingthebulkAscomponenttheleastaffected As2pcomponentwithonlya2eVshift(Plates1(c),2(a)and
byany changesin the top fewsurface layers. Accordingly, Table1),smallerthanthereported3eVshiftforAs3Coxide.19
allelementalspectrainPlates1and2arenormalizedtothe Theangle dependenceshownin Plate1(c) clearlyindicates
intensityofthecorrespondingbulkAs3dpeaks. thisisasurfaceAs2pcomponent,andweattributeittoasub-
The S 2p peak provides the most information about oxideassociatedwiththeinitiallimitedoxidation.ABEshift
the Slayer because different chemical states of S (i.e. withoxidecoveragereportedinapreviousstudyofresidual
physisorbed, chemisorbed, or oxidized) result in dramati- oxidesonGaAssupportsthesub-oxideinterpretationofthis
callydifferentS2pBEs.AsingleS2ppeak(Plates1(b),2(c)) As2pcomponent.18
indicates a narrow distribution of chemisorbed S states on For S-passivated InAs(001), we have to consider the
S-passivatedInAs(001).NotethatforchemisorbedSonInAs, possibility of an alternative assignment for the As 2p
theS2pspin-orbitsplittingistypically15,16notaspronounced component shifted by 2eV, as a 1.65eV shift has been
asinspectraofhighlyorderedSlayers(e.g.alkanethiolself- reported for As-S on GaAs(100) sulfidized in an aqueous
assembledmonolayersonAu),indicatingsomedisorderat (NH ) S solution (Fig.1(a) in Ref.20). Some As–S bonds
4 2
the In-S interface. Because the KE difference between the can be expected on S-passivated InAs(001) because of
S2pandAs3dpeaksis<10%,theS2p/As3dintensityratio imperfectionsinalayer-cakesurfacestructure(Fig.1).Since
is expected to be rather insensitive to any surface changes our energy resolution is insufficient to reliably separate
above the S layer, such as oxidation and contamination in such a putative As-S component from the As-O sub-
x
air,and,thus,canbeusedforthequantitativeanalysisofthe oxide component discussed above, we can only place a
timeevolutionofthetotalScoverage.ForS2ppeakfitting,it reasonable upper limit on the two coverages by assuming
isimportanttonotethat,forInAssamples,thebackground thatapproximatelyhalfoftheintensitycorrespondstoeach
intheS2pregionisdistinctlynonlinearand,thus,requires As-S and As-O , a limit consistent with the S coverage
x
inclusionofappropriatenonlineartermsinthebackground quantification.
fitting function (see the native oxide control spectrum in
Plate2(c)).
QUANTITATIVEXPSANALYSIS
TheIn3dpeakisthemostintensepeakinspectrafrom
InAs samples (Plate1(a)). For S-passivated surfaces, it also Standardoverlayer/bulkintensityratiomodels
containsthemostchemicalinformation,withbothIn-Sand Thefullexpressionforthenormal-emissionoverlayer/bulk
In-Ox components appearing along with the bulk In-As intensityratio,Iov/Ib,inthestandardXPSformalism(adapted
signal (Plate1(d)). The small difference in chemical shifts toincludeupdatedpracticaldefi(cid:1)nitionso(cid:2)fEALs(cid:3)21(cid:4),22)is:
between the In-S and In-O components (Plates1(d), 2(b), t
x 1(cid:2)exp (cid:2)
Table1) makes peak fitting ambiguous without additional Iov D Tov(cid:1)ovLQovNov (cid:2) Lo(cid:3)v (cid:2)1(cid:3)
constraints, as further discussed in the Scoverage analysis Ib Tb (cid:1)b LQb Nb exp (cid:2) t
section. Lb
ov
Copyright2005JohnWiley&Sons,Ltd. Surf.InterfaceAnal.2005;37:989–997
QuantificationofoxideandSlayersonS-passivatedInAsbyXPS
(a)
InAs(001) S 2p (b)
In 3d Ambient: TAM-passivation,
2 min DI water rinse,
5 min in air.
UHV: no treatment, RT.
O 1s 166164162160158
C 1s
dd
34
As 3p As In
As LMM
500 450 400 350 300 250 200 150 100 50 0
As 2p (c) In 3d (d) As 3d (e)
3/2 5/2
d)
3
s
A
ed to 0° noorAAss--SOX
aliz 0° 0°
m
or
n 35°
sity ( 35° 35°
n
e
S Int 65° 65°
XP 65°
132813241320 448 446 444 442 44 42 40 38
Binding Energy (eV)
Plate1.XPSdataforanInAs(001)surfaceimmediatelyaftertheTAMtreatment.(a)–(b)ThesurveyspectrumforInAs(001)
immediatelyafterTAMpassivationandahigh-resolutionclose-upoftheS2pregion.(c)–(e)High-resolutionangle-resolved
elementalcore-leveldata(emissionanglesasindicated).Deconvolutionoftheangle-resolvedspectrashowssurfacecomponents:
As-Ox(red)in(c),andIn-S(green),In-Ox(red)in(d).SymbolsDrawdata,thinlinesDfitcomponentsandbackgrounds,thicklinesD
fitsumcurves,thinlinesatbottomof(c)and(d)Dfitresiduals.
As 2p3/2 (a) In 3d5/2 (b) S 2p (c)
d)
3
As 5 min
o
d t
e
z
ali
m
or 4 days
n
y (
sit
n
e
S Int 42 days
P
X
native oxide
1329 1325 1321 447 444 168 165 162 159
Binding Energy (eV)
Plate2.High-resolutionelementalXPSdataforTAM-passivatedInAs(001)afterexposuretoair(normal-emission,angle-integrated,
timeinairasindicated).BottomspectraineachpanelarefromarigorouslydegreasedInAs(001)wafer(nativeoxidecontrol).(a)The
As-Oxcomponent(red)increasesrelativetoAs-In(blue)withincreasingexposuretoair.(b)ThefitresultsarefromthefixedIn-SBE
shiftmethod(seetext),andtheyshowanIn-Oxcomponent(red)increasingandanIn-Scomponent(green)decreasingwithtime.
(c)TheintensityoftheS2ppeakdecreaseswithtimeinair;theFWHMremainsconstant.NotetheabsenceofS-Oxfeaturesabove
164eV.SymbolsDrawdata,thinlinesDfitcomponentsandbackgrounds,thicklinesDfitsumcurves.
Copyright2005JohnWiley&Sons,Ltd. Surf.InterfaceAnal.2005;37
QuantificationofoxideandSlayersonS-passivatedInAsbyXPS 993
where T is the analyzer transmission function, (cid:1) is the
totalphotoelectriccross-section,23 Nistheelementalatomic
density, t is the overlayer thickness, LQ is the ‘‘EAL for
quantitative analysis’’ (QEAL) and L is the ‘‘average
ov
practicalEAL’’(PEAL).Adetaileddiscussion oftheQEAL
and PEAL definitions and calculations is presented in the
Appendix. Equation(1) assumes that the two signals are
acquired in parallel; thus, all the geometric factors and
the X-ray flux cancel out in the intensity ratio. For an
Figure2.Adiscrete-layermodelusedforquantificationof
oxide overlayer, the two signals in Eqn(1) correspond to
the two componentsof the same core-level peak separated AsOx andSoverlayers.Photoelectronspassingthrougha
byasmall chemical shift;therefore(cid:1) /(cid:1) D 1,T /T ³ 1, singleatomiclayerofthicknessaareexponentiallyattenuated
ox b ox b
Lb ³L DL andEqn(1)canbesimplified: withEAL(cid:5)byafactorRa.Formorethanonefullmonolayer,
ov ov ox
contributionsfromlowerlayersareattenuatedbyeach
(cid:1) (cid:2) (cid:3)(cid:4)
t overlayerorafraction(cid:4)thereof.
1(cid:2)exp (cid:2)
Iox D LQoxNox (cid:2) Lo(cid:3)x (cid:2)2(cid:3)
Ib LQb Nb exp (cid:2) t the total overlayer thickness tDL for use in the ‘thick-film’
Lox denominatorcanbeapproximatedas:
A typical limiting case for Eqn(2) is a thick-film oxide
t Da(cid:2)˛C(cid:4)(cid:3) (cid:2)6(cid:3)
DL
approximation (ThF), where the prefactor is cancelled by
assumingthattherespective oxideandbulkpropertiesare
Thecorrespondingsingle-layer(R )andtotal(R)attenua-
a
thesame:
(cid:1) (cid:2) (cid:3)(cid:4) tionfactorsaredefinedas:
t (cid:2) (cid:3)
IIobx D 1(cid:2)exepx(cid:2)p(cid:2)(cid:2)tLo(cid:3)x (cid:2)3(cid:3) RaDexp (cid:2)Laox andRD(cid:2)Ra(cid:3)˛C(cid:4) (cid:2)7(cid:3)
L
ox Thesubstratesignalintensityisgivenbythe‘thick-film’
denominatorofEqn(2):
Equation(3) relates the experimental intensity ratio X
andsignalattenuationintheoxideR:
I /LQN RDLQN (cid:2)R (cid:3)˛C(cid:4) (cid:2)8(cid:3)
(cid:2) (cid:3) b b b b b a
1(cid:2)R I t
XD , whereXD ox andRDexp (cid:2) (cid:2)4(cid:3)
R I L Theoxidesignalisgeneratedinagivendiscreteatomic
b ox
layer, and it is attenuated by a factor of R for each oxide
a
AnotherlimitingcaseforEqn(2)isasubmonolayeroxide layer above it. This model gives the discrete oxide signal
approximation(SubML),wherethesignalattenuationinthe intensity dependence on the number of full oxide layers ˛
oxide layer is assumed negligible (R D 1);LQbNb represents (Fig.2):
thetotalnumberofsubstrateatomscontributingtothesignal,
and LQoxNox is replaced by the total number nox of oxidized Iox(cid:2)˛D0(cid:3)/(cid:4),Iox(cid:2)˛D1(cid:3)/1C(cid:4)ÐRaand
atomsinthesubmonolayer:
I (cid:2)˛D2(cid:3)/1CR C(cid:4)Ð(cid:2)R (cid:3)2 (cid:2)9(cid:3)
ox a a
I n
ox D ox (cid:2)5(cid:3)
Ib LQbNb For up to three ML coverage, the oxide/bulk intensity
ratiocanthenbewritteninclosedformas:
(cid:1) (cid:4)
Discrete-layermodel
˛ ˛(cid:2)˛(cid:2)1(cid:3) 1
BthoethAscOhemanicdalSicnotuvietriaognesanadreobuertwpereionr1reasnudlt3sMsuLggonesmt tahnayt XD Iox D nML ˛C(cid:4)Ð(cid:2)Ra(cid:3)cos(cid:6) C 2 Ð (cid:2)Ra(cid:3)cos(cid:6) (cid:2)1
samplesxinthelongevityseriesofTAM-passivatedInAs(001). Ib LQbNb (cid:2)Ra(cid:3)c˛oCs(cid:4)(cid:6)
Forthiscoverageregime,however,neitherofthetwolimiting (cid:4) D(cid:4)C˛ (cid:2)10(cid:3)
DL
approximations described above (Eqns(3–5)) is strictly
applicable.Therefore,wedevelopamodifiedapproach–the where(cid:4) isthetotaloxidecoverage,the1/cos(cid:6)exponent
DL
discrete-layer(DL)modelshownschematicallyinFig.2.The appliedtotheR attenuationtermsaccountsfortheemission
a
generalapproachistoexplicitlyincorporatethegeneration angle(cid:6),n D5.41ð1014atoms/cm2 (cid:3)1MListhesurface
ML
of the oxide signal by a numberof discrete (full or partial) densityofbulk-terminatedInAs(001),andtheexpressionin
atomic layers by replacing the numerator in Eqn(2) by a thenumeratorgivestheappropriateresultslistedinEqn(9)
finite sum, while keeping the ‘thick-film’ approximation to for ˛ D 0,1,2. The closed-form expression in Eqn(10) can
describetheexponentialattenuationofthesubstratesignal. benumericallysolvedfor(cid:4),givenanexperimentalintensity
Itishelpfultodefineseveralmodelparametersindicated ratio X, angle (cid:6) and the appropriate choice of the discrete
inFig.2.Theoxideoverlayerisdescribedbyanumberoffull parameter ˛. The L (PEAL) and LQ (QEAL) values have
ox b
atomiclayers˛andafractionalcoverage(cid:4)fortheoutermost been calculated using the NIST SRD-82 software21,22 as
layer. If a is the thickness of one full atomic layer, then described in the Appendix and as listed in TableA1. For
Copyright2005JohnWiley&Sons,Ltd. Surf.InterfaceAnal.2005;37:989–997
994 D.Y.Petrovykh,J.M.SullivanandL.J.Whitman
quantificationofthebulkInandAssignals,thebulkatomic the ThF model (Eqns(3–4)). The SubML and ThF models
densityNb D1.80ð1022atoms/cm3 wasassumed(i.e.one- quantify the overlayer only in terms of the oxide coverage
halfoftheInAsbulkdensity).TheMLthicknessfortheoxide andthickness,respectively;theDLmodelresultsareshown
wasfixedataD0.3nm,whichisapproximatelyone-halfof using both metrics (related via Eqn(6)) for ease of cross-
theInAslatticeconstant,andisthevaluethatminimizesthe modelcomparison.TheDLmodelcoverages(Eqn(10))were
difference between the solutions of Eqn(10) for 0° and 65° calculated from normal and off-normal emission intensity
emissiondatafromoxidesofthickness>1ML. ratios;asthetwovaluesdifferedby<10%,theiraveragesare
ForAs2pphotoelectrons,the‘bulk’intensityoriginates
listed in Table2. Only the DL model results are presented
onlyinthetopfewsubstratelayers.Therefore,weexamined
inTable2fortheAs2p data;thecross-modeltrendswere
1/2
if a ThF denominator in Eqn(10) provides a sufficiently
essentiallythesameasthoseshownforAs2p .
3/2
accurate approximation. A full-discrete-layer (FDL) model
ThedatainTable2highlightseveralimportantfeaturesof
accounts for both the discrete nature of the As layers
themodels.Mostsignificantly,theDLandFDLmodelspro-
that generate the signal in the ‘layer-cake’ substrate and
videthebestapproximationforquantifyingtheoxideinthis
the DL attenuation of the substrate signal. For a semi-
range of coverage/thickness, and show quantitative agree-
infinite ‘layer-cake’ substrate, the sum of intensities from
ment with the SubML and ThF results for coverages <0.5
individual layers (effective attenuation length (cid:5) , layer
eff
spacing a? D aInAs/2 D 0.303nm) can be represented as a and>1MLrespectively.TheSubMLmodelbyconstruction
signalfromaneffectivenumberoflayersN : resultsinanestimateoftheoxidecoveragethatislinearwith
eff
thecorrespondingintensityratios,andthecombinationofthe
N D 1(cid:2) (cid:3),N n DLQN (cid:2)11(cid:3) cosinefactorandtheappropriateQEALproduceareasonable
eff 1(cid:2)exp (cid:2)a? eff ML b b agreementbetweennormalandoff-normalemissionSubML
(cid:5)eff results. The linear nature of the SubML model, however,
causesittoseverelyoverestimatecoverages½0.5ML.Con-
To obtain the FDL expression, the DL attenuation of
versely,theThFmodelunderestimatestheoxidecoveragefor
the substrate signal by the oxide must be combined with
values<1ML,ascanbeexpectedfromamodelthatassumes
Eqn(11);thus,theFDLdenominatorforEqn(10)becomes:
acompleteoverlayer.TheFDLcoveragesaresystematically
(cid:1) (cid:2) (cid:3)(cid:4)
a˛ ¾10%higherthantheDLvalues,indicatingthatinthisoxide
I DN n exp (cid:2)
b eff ML L cos(cid:6) thickness range the ThF denominator in Eqn(10) slightly
(cid:5) (cid:1) (cid:2) ox (cid:3) (cid:4)(cid:6)
underestimatesthesubstratesignalattenuationcomparedto
a
ð 1C(cid:4) exp (cid:2) (cid:2)1 (cid:2)12(cid:3) theFDLexpression(Eqn(12)).
L cos(cid:6)
ox
For practical applications of the DL and FDL models,
AsOx coverageanalysis we have calculated empirical As-Ox coverage calibration
Table2 presents the experimental I /I intensity ratios curves (Fig.3) on the basis of the model parameters used
ox b
measuredfortheAs2p3/2peakinnormaland65°emission, for the As 2p3/2 data in Table2. The DL model validation
and the corresponding coverages and thicknesses of the above (see also Table2 above and the Appendix) suggests
AsO overlayercalculatedusingtheSubMLmodel(Eqn(5)), an uncertainty estimate of about 20% for these calculated
x
theDLmodel(Eqn(10)),theFDLmodel(Eqns(10–12)),and coveragevalues.
Table2. OxideoverlayerquantificationfortheTAMpassivationlongevityseries
As2p3/2 As2p1/2
0° 65°
Time
inair (cid:4)0 (cid:4)65 (cid:4) (cid:4) t t (cid:4)
SbML SbML DL FDL DL ThF DL
(days) I /I (ML)a,b I /I (ML)a,b (ML)a,c (ML)a,d (nm)e (nm)f (ML)a,c
ox b ox b
0.004 0.184 0.42 0.413 0.39 0.40 0.44 0.120 0.076 0.39
0.017 0.198 0.45 0.619 0.58 0.47 0.52 0.141 0.081 0.47
0.729 0.462 1.06 1.060 0.99 0.72 0.77 0.217 0.170 0.80
3 0.459 1.06 1.243 1.16 0.75 0.79 0.225 0.169 0.84
4 0.694 1.60 1.526 1.43 0.90 0.92 0.269 0.235 1.01
33 1.137 2.61 2.219 2.07 1.18 1.20 0.353 0.339 1.18
42 1.263 2.90 2.494 2.33 1.26 1.31 0.378 0.365 1.37
aSurfacedensityofbulk-terminatedInAs(001)isusedtodefine1MLD5.41ð1014atoms/cm2.
bCoveragecalculatedusingthesubmonolayermodel,Eqn(5).
cCoveragecalculatedusingthediscrete-layermodel,Eqn(10).
dCoveragecalculatedusingthefull-discrete-layermodel,Eqns(10–12).
eThicknesscalculatedfromthediscrete-layermodelcoverage(cid:4) ,Eqn(6).
DL
fThicknesscalculatedusingthethick-filmmodel,Eqns(3–4).
Copyright2005JohnWiley&Sons,Ltd. Surf.InterfaceAnal.2005;37:989–997
QuantificationofoxideandSlayersonS-passivatedInAsbyXPS 995
3 boththeIn-SandIn-Oxcomponents.UsingEqn(10)andEAL
FDL
valuesfromTableA1,wecalculatedIn-SandIn-O coverages
α=0 x
ML) DL α=1 from components fit by the two methods: 1.5ML of In-S,
e ( 2 α=2 1.0 1.2MLofIn-Ox (fixedIn-S,Plate1(d),Table1),and1.9ML
ag 0.8 ofIn-S,0.5MLofIn-Ox(fixedIn-SandIn-Ox).Comparingthe
over 0.6 In3d5/2dataforincreasinglyoxidizedsamplesinPlate2(b),
C it appearsthat the In-O BE shift on the freshly passivated
Ox 1 0.4 x
s- 0.2 sample,whichcontainsprimarilysub-oxides,issmallerthan
A that observedon the native oxide control. Fixing the In-O
x
0.0
0.0 0.2 0.4 0.6 0.8 1 BE shift in a fit then would result in underestimating the
0
0 1 2 3 4 5 In-O contribution for samples with very low oxidation.
x
As-Ox/As-In Ratio in As 2p3/2 Conversely,theScoveragechangesonlybyaboutone-third
intheseries;thus,theIn-SBEshiftshouldchangeverylittle
Figure3.As-Oxcoveragecalibrationcalculatedusingthe
and fixing it improves the consistency of the fits without
discrete-layer(DL)andfull-discrete-layer(FDL)modelsfor
introducingartifacts(Plates1(d),2(b),Table1).
As-Ox/As-IncomponentintensityratiointheAs2p3/2region
Clearly, a different quantitative method, that does not
(normalemission).Symbolsindicatethenumberofcomplete
requiredeconvolutionoftheIn3dcomponents,isdesirable
oxidelayers(˛parameterinEqn(10));insetshowsaclose-up
fortrackingtheScoverageovertheentirelongevityseries.
ofthesubmonolayercoverageregion.
One approach is to modify Eqn(5) to include the element-
specificphotoelectriccross-sectionandtransmissionfunction
Scoverageanalysis ratios,(cid:1) /(cid:1) andT /T .TheScoverage(cid:4) isthengivenby:
S In S In S
TheScoveragefortheS-passivationlongevityseriescanbe
estimatedusingtwoindependentmethods.Thefirstmethod (cid:4) D TIn(cid:1)InLQInNb IS (cid:2)13(cid:3)
is to use the S 2p/As 3d intensity ratio (Table3). Because S TS (cid:1)S nML IIn
of the S 2p and As 3d peak properties discussed in the
where to produce the S coverage in ML, the numerical
previous section, this ratio is the best model-independent,
prefactor in front of the ratio of total S 2p and In 3d
semiquantitative measure of the S coverage,11 even for 5/2
intensitiesI /I is70.5.
S In
samples with unknown distribution of elements in the top
For the freshly passivated sample, Eqn(13) gives (cid:4) D
S
few atomic layers. The second method is to use In3d and
1.7ML (Table3), in good agreement with the DL result
S 2p peaks for a quantitative coverage estimate. However,
(1.5ML of In-S) and the upper limit of 0.2ML of As-S
this approach relies on deconvolution of In 3d chemical
obtained from the As 2p data. Note that the S coverage
components and assumptions about the surface structure
changes by ³35% in the longevity series (Table3). Given
and, thus, has more limited applicability than the first
thatEqn(13)isquantitativelyaccurateforthefirstsamplein
method.
theseries,thelinearnatureofEqn(13)suggestsitshouldbe
A freshly passivated surface is the sample for which
approximatelyvalidfortherestoftheseries.Equation(13)
estimatingtheabsoluteScoverageismostimportant(Plate1,
provides a better quantitative coverage estimate from
topspectrainPlate2).InordertousetheDLmodel,weneed
the S2p/In3d intensity ratio than Eqn(5) did in the
areliablemethodoffittingtheIn3dchemicalcomponents:
As-O /As-In2p case because the total In 3d signal is
x
In-As, In-S, and In-O (Plates1(d), 2(b)). We have tested
x contributed by both the bulk-dominated In-As component
two sets of fitting constraints: fixed BE shift for the In-S
(QEAL D 2.43nm) and the topmost In-S/In-O layers.
x
component only (Plates1(d), 2(b)) and fixed BE shifts for
However, it also means that the accuracy of this empirical
quantification reliesontheparticularsurface structureand
Table3. SulfurcoverageanalysisfortheTAMpassivation thusmaynotbegenerallyapplicable.
longevityseries
CONCLUSIONS
Timeinair
(days) S2p/AsAs(cid:2)In3da S2p/In3d5/2b (cid:4)S(ML)c Wehaveusedhigh-resolutionangle-resolvedXPSmeasure-
ments to quantitatively characterize the initial structure
0.004 0.27 0.024 1.7
and stability in air of InAs(001) surfaces passivated by
0.017 0.24 0.025 1.8
thioacetamide.We findthattheTAMtreatmentveryeffec-
3 0.20 0.018 1.3
tively removes native oxide and passivates InAs(001) with
4 0.23 0.020 1.4
a chemisorbed S layer against reoxidation and contamina-
33 0.18 0.016 1.1
tioninaqueoussolutionsandlaboratoryair.XPSelemental
42 0.17 0.016 1.1
analysisandreoxidationbehaviorforTAM-passivatedsam-
aExperimentalintensityratio:totalS2ptoAs-Incomponentof ples are consistent with the S/In/As ‘layer-cake’ structure
As3d. model proposed for other S-passivated InAs(001) surfaces.
bExperimentalintensityratio:totalS2ptoIn3d5/2. ThehighefficiencyoftheTAMpassivationresultedinvery
cScoveragecalculatedfromS2p/In3d5/2intensityratiosusing small amountsof As-Ox during theinitial reoxidation and,
Eqn(13),1MLD5.41ð 1014atoms/cm2. thus, required the use of oxide/bulk intensity ratios from
Copyright2005JohnWiley&Sons,Ltd. Surf.InterfaceAnal.2005;37:989–997
996 D.Y.Petrovykh,J.M.SullivanandL.J.Whitman
thesurface-sensitiveAs2pregion,andthedevelopmentofa fromthosefortheC3oxideby<5%;therefore,forsimplicity,
DLmodelforquantitativeanalysisofthemeasuredintensity weassumedtheC3oxidestoichiometry(In:As:OD1:1:3)
ratios.Inaddition,amethodwasdevelopedforquantifying fortheoverlayer.Inanothersimplification,theEALscalcu-
theScoverageonS-passivatedInAs(001)onthebasisofthe latedfortheoxidewerealsousedtoquantifytheSoverlayer.
analysis of S 2p and In 3d peak intensities. Quantitatively, In all cases, the EALs calculated for the overlayer differed
the following components were identified on an InAs(001) fromthosefortheInAssubstrateby<15%;thus,weexpect
surface freshly passivated bytheTAM treatment:³1.5ML thatthesimplifiedtreatmentoftheoverlayermaterialparam-
ofIn-S,³1.2MLofIn-O ,0.2–0.4MLofAs-O ,andpossibly etersdidnotsignificantlyaffecttheoveralluncertaintywhile
x x
up to 0.2ML of As-S. After 42days in air, both As-O and capturingtheprincipaleffectoftheoverlayerchemistryon
x
Scoverageswereslightly above1ML. In addition tooffer- theattenuationofphotoelectrons.
ingsuperiorpropertiespotentiallyusefulinpracticalsurface WeusedtheSRD-82softwaretocalculatetwoEALvalues
passivation,theTAMpassivationclearlyproducesexcellent for each elemental peak (TableA1): ‘‘EAL for quantitative
modelS-passivatedInAs(001)surfaces.ThequantitativeXPS analysis’’(QEAL)and‘‘averagepracticalEAL’’(PEAL).The
dataandanalysis,aswellastheDLmodelandempiricalcov- formal definitions of these two parameters are given in
erage calibration developed and validated on these model Ref.21,butthepracticalimplicationscanbesummarizedas
surfaceswillbeusefulasabenchmarkandquantitativeref- follows:QEALisessentiallyapropertyofasemi-infiniteslab
erence in studies of passivated InAs interfaces with more of material and thus is used in all pre-exponential factors,
complexstructures. whereas the PEAL relates to attenuation by the overlayer
and thus is used in the exponential attenuation terms. We
Acknowledgements actually use the ‘average’ PEAL, i.e. PEAL averaged over
D.Y.P.thanksDrJenniferC.Sullivan(NRL)forprovidingtheXPS a certain overlayer thickness (0.4nm). The average PEALs
dataofthenative-oxideInAs(001)controlandforhelpfuldiscussions calculated for a 1.5nm overlayer thickness differed from
ofsurfacechemistryandquantitativeXPSanalysisoffunctionalized
theseby<5%.Asareference,thebulkInAslatticeconstant
III-V semiconductors. The work was supported by the Office of
is0.606nm.
NavalResearch,theAirForceOfficeofScientificResearch,andthe
DefenseAdvancedResearchProjectsAgency. Two factors can potentially affect XPS data measured
using single-crystal InAs substrates: photoelectron diffrac-
tion and coherent inelastic scattering. These single-crystal
APPENDIX:EALCALCULATIONS
effectsprimarilyoccurforhigh-KEphotoelectronsthatprop-
agate through multiple ordered atomic layers before being
The EAL values used for quantitative XPS analysis were
calculated using the NIST SRD-82 software.21,22 The soft- emitted. Conversely, for the low-KE As 2p photoelectrons
warereliesontheTPP-2Mformula24forcalculatinginelastic thatoriginateinatopfewrelativelydisorderedatomiclayers,
meanfreepath(IMFP)values.Apreviousstudyhasshown anysingle-crystaleffectswillbeminimal.Similarly,allpho-
that the TPP-2M predictions for InAs were in good agree- toelectrons originating in the near-surface layers will not
mentwithIMFPsexperimentallymeasuredinInAs,25which be affected:forInAs-S this includes the S 2ppeak,and the
can be considered a good independent reliability test of As-Ox,In-Ox,In-Ssurfacecomponents.Thus,theonlypeaks
the TPP-2M-based EAL calculations for this material. The consideredinourstudythatcouldbeaffectedbytheuseof
uncertainty of the calculated EAL values is estimated by single-crystal substrates are the ‘bulk’ As-In and In-As 3d
theauthorsofthesoftwaretobe³20%.22Thevalidityofthe peaks.TheScofield-adjusted23 As/Inratioscalculatedfrom
TPP-2MpredictiveformulaandSRD-82EALcalculationshas these‘bulk’componentswere0.97š0.06(normalemission)
been established for photoelectrons with KE > 200eV,22,24 and1.00š0.07(65°emission)forsamplesinourstudy.The
and has been recently extended down to KE > 50eV
for elemental solids,26 but for the As 2p3/2 photoelectrons TableA1. CalculatedvaluesofPEALintheoxideoverlayer
(KE ³ 160eV) in InAs there could be some effects unac- andQEALinbulkInAs
countedforinthemodel.
The material parameters required by the software for KE Asymmetry Emission PEAL QEAL
theEALcalculationsincludethechemicalcomposition(stoi- Peak (eV) parameterˇ angle(°) Lox(nm) LQb (nm)
chiometry),density(cid:7),photoionizationasymmetryparameter
As3d 1445 1.055 0 2.26 3.11
ˇ, and band gap energy E . For the InAs substrate, we
g 35 2.26 3.09
assumed the bulk InAs 1:1 stoichiometry, and the stan-
65 2.27 3.02
dard values for (cid:7) D 5.68g/cm3 and E D 0.4eV. For the
g In3d5/2 1042 1.22 0 1.72 2.43
overlayers,theuncertaintyabouttheiractualstructureand
35 1.72 2.41
composition required some simplifying assumptions. For
65 1.73 2.35
overlayerdensity,thebulkInAsvalueisclosetotheaverage
As2p3/2 162 1.131 0 0.458 0.691
ofvaluesreportedfordifferentInandAsoxides(about7and
35 0.458 0.685
4g/cm3,respectively);therefore,weassumedthedensityto
65 0.476 0.664
be the same for the substrate and overlayer. Reported E
g As2p1/2 127 1.131 0 0.407 0.615
valuesforInandAsoxidesareprimarilywithinthe3–4eV
35 0.406 0.609
range;27therefore,weassumedE D3.5eVfortheoverlayer.
g 65 0.426 0.531
TheEALscalculatedfortheC5oxidestoichiometrydiffered
Copyright2005JohnWiley&Sons,Ltd. Surf.InterfaceAnal.2005;37:989–997