DTIC ADA483102: Surface Reconstruction Phase Diagrams for InAs, AlSb, and GaSb

10 Pages·0.2 MB·English

by  Defense Technical Information Center

JournalofCrystalGrowth220(2000)384}392
Surface reconstruction phase diagrams for InAs, AlSb,
and GaSb
A.S. Bracker*, M.J. Yang, B.R. Bennett, J.C. Culbertson, W.J. Moore
ElectronicsScienceandTechnologyDivision,NavalResearchLaboratory,Code6876,4555OverlookAvenue,SW,
Washington,DC20375-5347,USA
Received19July2000;accepted15September2000
CommunicatedbyA.Y.Cho
Abstract
We presentexperimental#ux-temperature phase diagramsfor surface reconstructiontransitions on the 6.1A com-
s
poundsemiconductors.Thephasetransitionsoccurwithinorneartypicalsubstratetemperaturerangesforgrowthof
these materials by molecular beam epitaxy and therefore provide a convenient temperature standard for optimizing
growthconditions.PhaseboundariesforInAs(001)[(2]4)P(4]2)],AlSb(001)[c(4]4)P(1]3)],andGaSb(001)
[(2]5)P(1]3)]arepresentedasafunctionofsubstratetemperatureandGroupV-limitedgrowthrate(proportionalto
#ux),forbothcrackedanduncrackedGroupVspecies.Wediscussdi!erencesbetweenmaterialsintheslopesando!sets
ofthephaseboundariesforbothtypesof GroupVspecies. ( 2000ElsevierScienceB.V.Allrightsreserved.
PACS: 18.15.Hi;81.05.Ea;68.35.Bs;68.35.Rh
Keywords: Molecularbeamepitaxy;Galliumantimonide;Aluminumantimonide;Indiumarsenide;Surfacereconstruction
1. Introduction designs and improved performance that are di$-
cult to achieve with other compound semiconduc-
Thecompoundsemiconductorswith6.1Alattice tors. In order to realize these bene"ts, epitaxial
s
constant, including InAs, GaSb, and AlSb, have growth and device processing technologies for the
received increasing attention in recent years be- 6.1A compound semiconductors must advance to
s
cause of their potential for electronic and optical the level of more mature material systems such as
deviceapplications[1}6].Auniquecombinationof GaAs/AlGaAsand InP/InGaAs.
bando!set(brokengaptypeII,betweenInAsand In order to grow the highest quality compound
GaSb), high InAs electron mobility, and InAs sur- semiconductor thin "lms by molecular beam epi-
face Fermi level pinning permit novel device taxy(MBE),precursor#uxesandsubstratetemper-
ature should be optimized [7,8]. These growth
parameters a!ect the rates of adsorption, desorp-
tion, and surface di!usion of Group III and V
*Correspondingauthor.Fax:#[phone]. species,andthereforein#uencethequalityofcrys-
E-mailaddress:[email protected](A.S.Bracker). talgrowth.Activationbarriersandpreexponentials
0022-0248/00/$-seefrontmatter ( 2000ElsevierScienceB.V.Allrightsreserved.
PII: S0022-0248(00)00871-X
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
JUL 2000 2. REPORT TYPE 00-00-2000 to 00-00-2000
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Surface reconstruction phase diagrams for InAs, AlSb, and GaSb
5b. GRANT NUMBER
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,Electronics Science and Technology REPORT NUMBER
Division,4555 Overlook Avenue 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
see report
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 9
unclassified unclassified unclassified Report (SAR)
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39-18
A.S.Brackeretal./JournalofCrystalGrowth220(2000)384}392 385
for these processes depend on the structure and conditionsduringsurfacestabilizationbyaGroup
stoichiometryof the material surface(see e.g. Refs. V #ux, the system is not in strict equilibrium: the
[9,10]), which are in turn a!ected by temperature vapor and solid phases are at di!erent temper-
and beam #ux. To achieve the best growth condi- atures; the vapor velocity distribution is aniso-
tions, it is therefore desirable to know the recon- tropic;the incidentand desorbingspeciesmay not
struction phase boundaries over typical beam #ux be the same. Nevertheless, equilibrium thermo-
and substrate temperature ranges. dynamicsmaybe usedas anapproximatedescrip-
The #ux-temperature phase diagram is also tion of a Group V-stabilized surface [7,8]. We
convenient for estimating substrate temperature, adopt this view, although in Section 4 we discuss
particularly, when standard techniques for some evidence for non-equilibrium behavior.
temperaturemeasurementareunavailable.Remote
thermocouplemeasurementsinvacuumarenotori-
ously inaccurate. Furthermore, their calibration 2. Experimental method
can change during growth of small-bandgap
epilayers (e.g. InAs and GaSb), due to radiative The 6.1A semiconductors studied in this report
s
heating that increases with "lm thickness [11,12]. were grown by solid-source MBE. Growth rates
The accuracy of optical pyrometry is limited by and surface reconstruction symmetry were moni-
stray light from the substrate heater or source tored with re#ection high-energy electron di!rac-
ovens,coatingoftheviewport,changesinsubstrate tion(RHEED).A10kVelectronbeamwasincident
emissivity, and limited sensitivity below 5003C on the sample surface between 1.53 and 2.53. The
[13]. Transmission thermometry, which monitors beamhadbeenpreviouslyadjustedtofocusonthe
the shift in the optical transmission edge of the RHEED phosphor screen (&70cm from electron
substrate, is an e!ective alternative to pyrometry guntoscreen)withoutthesampleinplace.Allthree
[14}16].However,thistechniqueisnotpossiblefor 6.1A materials were grown on semi-insulating
s
InAssubstratesduetofreeelectronabsorption,and GaAs(001)substrates,andtheGaSbmeasurement
measurements for 6.1A heteroepitaxy on GaAs was also repeated on GaSb substrates, producing
s
substrates are not possible when a small-bandgap no shift of the phase boundary [16]. AlSb and
epilayer is thick enough to absorb most of GaSb layers on GaAs were roughly 1lm thick,
the source light (’1lm). When these or whileInAswasgrownona1lmAlSbbu!er.Thin
other techniques are not available, observation layers of InAs (1000A) were used in order to min-
s
of known surface reconstruction transitions imize absorption of radiation from the substrate
remains a convenient means to estimate substrate heater,whichisneededfortransmissionthermom-
temperatures. etry.Inordertocon"rmthatstraininthethinInAs
There have been several reports of reconstruc- layer (1.3% lattice mismatch with AlSb) does not
tionphasediagramsforIII}Vmaterials,especially signi"cantlyin#uencethereconstructiontransition
forGaAs.Earlyworkonreconstructiontransitions temperature, we measured this temperature
[17,18] includes measurements for all zincblende for InAs (001) [(2]4)P(4]2)] under a
III}V compounds at a "xed Group V #ux [18]. backgroundarsenic#uxforthicknessupto8000A.
s
Full phase diagrams have been reported for GaAs No dependence on the InAs layer thickness was
(001) [19}23], GaAs (111) [24], AlAs (001) observed.
[19,20,25], AlGaAs (001) [22], and for the 6.1A Substratetemperaturesweremeasuredbytrans-
s
materials[16,23,26,27],althoughnotallwithcalib- mission thermometry. This technique relies on the
rated#uxesofbothcrackedanduncrackedGroup changeinthe substratebandgap withtemperature
V species.Recentresults on GaSb fromour group [14}16]. Brie#y, light from the substrate heater is
werepresentedpreviously[16]andarediscussedin partially transmitted through the substrate and
this work as well. exits the chamberthrough a viewport. The lightis
Phase diagrams typically describe thermodyn- chopped (for phase-sensitive detection), focused
amic equilibrium conditions. However, for MBE into an optical "ber, dispersed by a grating
386 A.S.Brackeretal./JournalofCrystalGrowth220(2000)384}392
spectrometer, and "nally detected with a photo-
diode. The transmission spectrum of the substrate
is obtained by taking the ratio of the spectrum
obtained with the substrate in place to a back-
groundspectrumofthebaresubstrateheateremis-
sion. For GaSb substrates, the relation between
transmission edge and substrate temperature was
taken from Ref. [16]. In the appendix, we present
a new calibration for GaAs substrates, in order to
update de"ciencies in an earlier version.
EPI cracker models 500V-As and 175-Sb were
used to produce cracked Group V species. By
choosing the appropriate operating conditions for
thearsenicsource,it isstraightforwardtoproduce
Fig.1. TypicalGroupV-limitedgrowthratemeasurement,here
either arsenic dimers (As ) or tetramers (As ), but
the antimony cracker sou2rce (cracker tempe4rature for InAs with As2 (3.2]10~6Torr beam equivalentpressure).
Arsenic #ux limits growth at 0.802ML/s, yielding
9003C) produces a mixture of Sb2 and Sb atoms g2"4.0]10~6Torr/(ML/s).
due to the small Sb bond strength. A detailed
2
characterization(usingthe175-Sb)wasreportedin
Ref. [28] and showed that the source produces
AstheGroupIII#uxis increased,thegrowthrate
alarge#ux-dependentproportionofSbatomsun-
eventuallysaturates as growthbecomeslimitedby
deroperatingconditionstypicalforourwork.Anti-
the Group V #ux. An example of this behavior is
mony dimer beams discussed in later sections of
showninFig.1forInAs.DividingtheGroupV#ux
this paper should therefore be understood to in-
(beam-equivalent pressure units, torr) by the satu-
clude atoms as well. We use a separate antimony
ration growth rate (ML/s) gives the values g (for
source (operating temperatures (5003C) to pro- 2
dimers) and g (for tetramers) that are used to
duce tetramers. 4
rescale the vertical axes of the phase diagrams.
Although molecular beam #uxes are commonly
AGroupIII-richreconstructionRHEEDpatternis
reportedasiongaugepressurereadings,suchread-
oftenobservedduringtheV-limitedgrowthoscilla-
ingsareoftennotreproduciblebetweenlaborator-
tions.Wefoundthatg andg wereindependentof
ies due to di!erences in ion gauge geometry, 2 4
GroupV#ux,withtheonlyexceptionatthelowest
electrodevoltages,"lamentandelectrodecoatings,
#uxes of arsenic, where non-beam arsenic species
etc.Furthermore,theconversionfromaniongauge
may contribute a signi"cant fraction to the ion
readingtobeam#uxesrequiresquantitativevalues
gauge reading. The calibration was carried out at
forthespecies-dependentionizationcross-sections.
4203C,4303C,and5203CforInAs,GaSb,andAlSb,
Because of these shortcomings, we also present
respectively.Thesetemperaturesshouldberoughly
phase diagrams with the #ux expressed in units of
reproducedwhenrepeatingthecalibration,because
growth rate at a III/V #ux ratio for which the
the Group V incorporation coe$cients during
Group V #ux is limiting. This approach has been
growthon(001)surfacesbegintodecreaseathigh-
used previously in studies of epitaxial growth and
ersubstratetemperature(seee.g.Refs.[29,30]1for
surface reconstructions on GaAs (see e.g. Refs.
typical behavior).
[16,29]).
We used the following procedure to convert to
GroupV-limitedgrowthrateunits.Thesemeasure-
ments are made separately from the phase
1TheInAsphasediagraminRef.[23]forthe(2]4)P(4]2)
transitionmeasurements.Fora"xedGroupV#ux,
transition during growth should correspond roughly to the
thegrowthrateis measuredfromRHEEDoscilla-
temperaturedependenceofV-limitedgrowthasde"nedbygxin
tions as a function of Group III cell temperatures. thiswork.
A.S.Brackeretal./JournalofCrystalGrowth220(2000)384}392 387
Fig.2. SchematicrepresentationofRHEEDpatternevolution
assubstratetemperatureisrampedthroughphasetransitions.
Stars mark the points chosen as empirical de"nitions of the
phasetransitions.
3. Phase diagrams
Ourexperimentalde"nitionsforthe reconstruc-
tion phase transitionsare based on changes in the
RHEEDpattern(Fig.2).Ourprimaryintentionin
choosing these de"nitions was that the measure-
ments could be reproduced in other laboratories.
We are not concerned with whether the corre-
spondingchangesin surface structuresatisfy more
rigorouscriteriaforaphasetransition.Changesin
surface periodicity occur at di!erent temperatures
fordi!erentcrystaldirections,andmixedphasesor
domains may exist around the transition temper-
ature.Wesimplyrepresentthetransitionregionas
a line, while recognizing that additional surface
structures may exist.
Severalsourcesofexperimentalerrorcontribute
touncertaintyinthephasetransitiontemperatures.
The two most important errors in our experience
werethepositionoftheRHEEDelectronbeamon Fig.3. Reconstructionphasediagrams,withGroupV#uxex-
thesubstrate(duetolateraltemperaturegradients) pressedinbeam-equivalentpressureunits(s2 ands4 inTorr).
andunsteadyramprates.Abackgroundpressureof Horizontal axis is linear in reciprocal absolute temperature.
Solid lines are "ts using the expression and parameters in
undesired Group V species is likely to shift the
Table1.
transition temperature, especially for antimonides
whenarsenicisabackgroundspecies.Wekeptthe
arsenicsourceatidletemperatureduringantimon- these errors, repeated measurements rarely varied
idemeasurements,andviceversa,butunfortunate- by more than 103C.
ly this may not be possible during heterostructure Figs.3and4showthephasediagramsforsurface
growthsthatrequirebothGroupVspecies.Finally, reconstructionsonInAs,GaSb,andAlSb.Weused
there is some subjectivity in judging changes in asimpleexponentialformto"tthedatapoints,as
di!raction streaks as described in Fig. 2. Given describedinTable1.TheverticalaxesofFig.3are
388 A.S.Brackeretal./JournalofCrystalGrowth220(2000)384}392
Table1
Parameters used in the expression #ux"exp(!E/kB„#B)
for "ts in Figs. 3 and 4. BT033 and BML@4 give #ux in units of
beam-equivalentpressure (Torr) and Group V-limited growth
rate(ML/s),respectively
E(eV) BT033 BML@4
InAs As2 4.16 47.5 60.0
As4 2.95 31.5 43.5
AlSb Sb2 4.64 62.6 76.4
Sb4 3.10 39.0 52.5
GaSb Sb2 1.78 16.3 30.1
Sb4 1.72 15.7 29.2
aGroupV#uxofdimersortetramers,respectively.
This form of the phase diagram may be used as
a quick reference for estimating substrate temper-
ature, but these values may not compare well be-
tween laboratories, for the reasons discussed
earlier.
Fig. 4 shows the phase diagrams with vertical
axes rescaled in units of V-limited growth rate
(s /g and s /g ). The divisors g and g (for
2 2 4 4 2 4
growthwithdimersandtetramers,respectively)are
proportional to the incident Group V #ux (Torr)
during V-limited growth divided by the Group
V-limited growth rate in ML/s. They are
g (AlSb)"1.0]10~6, g (AlSb)"1.3]10~6,
2 4
g (GaSb)"1.0]10~6, g (GaSb)"1.3]10~6,
2 4
g (InAs)"3.9]10~6, g (InAs)"6.4]10~6, with
2 4
units of Torr/(ML/s). The units s /g are still pro-
x x
portionaltotheincidentGroupV#uxrequiredfor
stabilization, but they are more easily transferred
between laboratories than s , because ion gauge
x
sensitivitiesaredividedout.2Westressthatg was
x
measuredseparatelyandatonlyonesubstratetem-
Fig.4. Reconstructionphasediagrams,withGroupV#uxex- perature for purposes of rescaling each phase
pressedinunitsofGroupV-limitedgrowthrate(s2/g2ands4/g4
inML/s).Horizontalaxisislinearinreciprocalabsolutetemper-
ature.Solidlinesare"tsusingtheexpressionandparametersin
Table1.
2Whengxandsxarebothmeasuredinthetemperaturerange
wheregxisindependentoftemperature,thentheunitssx/gxare
expressedin the unitsof beam-equivalentpressure also equal to the rate at which Group V atoms would be
(Torr) as measured with an ion gauge, which is incorporated into the crystal during a V-limited growth
proportionalto the absolute Group V atomic #ux measurementfor#uxsx.Thisconditionisapproximatelytrue
forGaSbandAlSb.HoweverforInAs,thereconstructionphase
requiredtostabilizethesurfaceatthephasebound-
boundaries were measured at temperatures far higher than
ary. The plots on this scale are referred to sub-
wheregx wasmeasuredand inarangewheregx rises rapidly
sequently as s2 and s4, for stabilization under withtemperature.
A.S.Brackeretal./JournalofCrystalGrowth220(2000)384}392 389
boundary in Fig. 3 to produce Fig. 4. No growth the #ux scale in Fig. 4, should allow reproducible
occurred during the phase transition measure- measurement of the phase boundaries in other la-
ments. The slopes of the phase boundaries re#ect boratories.Becausethephaseboundariesfallinor
the thermodynamics and kinetics of stabilization near the usual growth temperature ranges for the
only. 6.1A materials, they should prove useful for tem-
s
The changes in the RHEED pattern used to peraturecalibrationwhen other techniquesare in-
identifythephasetransitionsaremarkedbystarsin accurate or unavailable. Although the AlSb (001)
Fig.2.Forallmeasurements,thesubstratetemper- [c(4]4)P(1]3)] transition falls considerably be-
ature was ramped upward at a rate of 23 per min- lowtheoptimaltemperaturerangeforAlSbgrowth
ute.ForAlSb(001)[c(4]4)P(1]3)],3wede"ned (500}6003C), most device structures containing
the transition temperature by equal intensities for AlSbalsocontainInAsandGaSb,whicharegrown
the 1/3 and 2/3 streaks of the 3]pattern and the at temperatures below 5003C. The GaSb (001)
1/2 streak of the 2]pattern [from c(4]4)], in the [(2]5)P(1]3)] transition falls within the com-
[110] direction. mongrowthtemperaturerangeforInAsandGaSb
TheInAs(001)[(2]4)P(4]2)]transitionwas (400}5003C), while the InAs (001) [(2]4)P
de"nedasthe"rstappearanceofthe4]di!raction (4]2)] transition lies above this range.
streaks in the [11 10] direction. Yamaguchi and We next discuss the signi"cant di!erences be-
Horikoshihavealsostudiedthistransitionindetail tween materials in the o!sets and slopes of the
forAs #ux[26].Theyusedaspecializedapparatus phase boundaries for cracked and uncracked
4
tomonitorchangesintheRHEEDspecularinten- Group V species.
sity and measured temperature and pressure with
pyrometry and an ion gauge, respectively. Their
measured phase boundary for As #ux occurred 4. Reactivity of Group V dimers and tetramers
2
roughly303C lower than ours, with #ux expressed
asan iongauge reading.This discrepancyis prob- The reactivity of uncracked (tetramer) and
ablyaconsequenceofadi!erentexperimentaldef- cracked(dimer5)GroupVspeciesonIII}Vsurfaces
inition for the transition, and possibly also due to haslongbeenanareaoffundamentalandpractical
inaccuracies associated with pyrometer or ion interest.Theformoftheincomingspecieswilla!ect
gauge measurements. theratesofadsorption,andquitepossiblydi!usion
The GaSb [(2]5)P(1]3)] phase diagram was and desorption as well. Adsorption of gas phase
recently reported by our group [16].4 The phase GroupVdimersmaybeassimpleasabondlength
diagramfromthatworkisreproducedhereinFigs. change and rearrangement on the surface to form
3and4.Thetransitionwasde"nedbytheappear- asurfacedimer.Incontrast,thetetrahedralGroup
anceofapseudo-(1]3)RHEEDpattern,asthe1/5 V tetramers must undergo a more complicated
and 4/5 streaks disappear, and before the 2/5 and electronic and nuclear rearrangement to produce
3/5streaksmoveapartintothetypicalpositionsfor surface dimers, resulting in slower rates. Some
the (1]3) pattern. experimental results for gallium arsenide growth
These de"nitions, combined with the experi- suggest that As chemisorption requires the
4
mental techniques described above for obtaining interaction of two neighboring physisorbed
3Recentwork(Ref.[40])hasshownthattheAlSb(1]3)isin 5For cracked antimony, this discussion assumes that the
fact a (4]3) structure, but the 4]periodicity is usually not monomercontentofthebeam,whichchangesasafunctionof
observedinRHEED. total#ux,doesnota!ecttheantimonyincorporationkinetics.
4TheGaSb‘(2]5)areconstructiondiscussedhereappearsin Ref.[28]hasshownthatdimersandatomshavenearlyidentical
scanning tunneling microscope images as (2]10) or c(2]10) incorporation probabilities during Group V-rich growth and
unitcells(seeRef.[41]).InRHEED,a(1]5)or(2]5)patternis proposedthatthisoccursbecausemostantimonyatomsrapidly
observed. formdimersonthesurface.
390 A.S.Brackeretal./JournalofCrystalGrowth220(2000)384}392
tetramers [31,32], and ab initio cluster model re- Ingeneral,theratioss /s andg /g neednotbe
4 2 4 2
sults support this mechanism [33]. In this mecha- equal, because they relate to di!erent processes.
nism,twoatomsfromeachtetramerarereleasedto Thevalueg involvesGroupVatomsthatperma-
x
thegasphase,sothattheincorporationcoe$cient nentlyincorporateintoagrowingcrystal.Thereare
isneverlargerthan50%.Incontrast,100%incor- frequent interactions between III and V species,
poration is possible with As . These stoichiomet- andalargefractionofGroupVspeciesareburied
2
ries have been widely assumed for III}V MBE by growth before they can desorb. The surface is
growth since the early investigations of GaAs highly disordered and has a high density of island
growth using modulated-beam mass spectrometry edge sites. For (001) surfaces, g increases weakly
x
[31,34], however, other studies have reported As with temperature under typical growth conditions
4
incorporation coe$cients greater than 50% [30],presumablyas moreGroupV speciesdesorb
[35}37]. Studies of antimonides report that Sb prior to incorporation. On the other hand, s re-
4 x
incorporation coe$cients are closer to 100% lates to the replacement of Group V atoms that
[38,39]. desorb from sites of the prevailing surface recon-
The vertical o!sets between the dimer- and tet- struction and increases strongly with temperature
ramer-phase boundaries in Fig. 4 (s /g vs. s /g ) (Figs.3and4).Thesurfaceis#atter,moreordered,
4 4 2 2
re#ect di!erences in the relative e$ciency6 of the andhasfewertypesofadsorptionsites.Thisstrong
two Group V species in stabilization vs. growth distinction between stabilization and growth ap-
(s /s vs. g /g ).7 For example, the GaSb phase pliestothoseprocessesforAlSbandGaSbthatare
4 2 4 2
boundaries for Sb and Sb are nearly coincident considered here.
4 2
(s /s +g /g ), suggesting that #uxes of dimers For InAs, there is a weaker distinction between
4 2 4 2
and tetramers have roughly the same relative ef- s and g , and a comparison is less meaningful.
x x
"cienciesinbothstabilizationandgrowthofGaSb. Here,botharsenicstabilizationandarsenic-limited
Incontrast,thelargeo!setbetweentheAlSbphase growth occur near the phase boundary between
boundariesimpliesthatproportionallymoreSb is As-andIn-richreconstructions[(2]4)and(4]2)].
4
needed for stabilization than for growth (s /s ’ In fact, the arsenic #ux corresponding to [g ]
4 2 x
g /g ). Unless the di!erent behaviors of AlSb and (indium #ux)] and the #ux s plotted in Fig. 3
4 2 x
GaSbaresolelyaconsequenceof stabilizationdif- represent points on two di!erent indium #ux iso-
ferences,thenthisresultimpliesthatoneorbothof bars(with"niteandzeroindium#ux,respectively)
thesematerialshasgrowthkineticsforSb andSb ofasingle#ux-temperaturephasediagramforInAs
2 4
that di!er from the model proposed for As and growth.Ref.[23]discussessuchphasediagramsfor
2
As on GaAs [31,32]. InAs.Thecomparisoniscomplicatedfurtherbythe
4
fact that the InAs phase boundariesin Figs. 3 and
4 were obtained for temperatures considerably
above where g was measured (&4203C). Measur-
6Here we de"ne e$ciency as inversely proportional to the x
ingbothgrowthandstabilizationaround4203Cis
#uxrequiredforagivenprocess.Thestabilization#uxsxranges
impractical,becauseextremelysmallarsenic#uxes
fromzero,wherenoGroupVspeciesdesorbfromthesurface
(andthusdonotneedreplacing),to in"nity,wherenoGroup would be needed for stabilization, while at higher
V species stick to the surface. The corresponding e$ciency temperatures(’5003C)theisobarsmerge,and the
ranges from in"nity to zero, respectively. The values gx distinctionbetweengrowthandstabilizationislost.
[Torr/(ML/s)]areproportionaltotheGroupV#uxrequiredfor
Theslopesofthephaseboundariesre#ectkinetic
V-limitedgrowth,perunitofgrowthrate,andrangefromone
di!erencesintheabilityofdimersandtetramersto
(whenGroupsVandIIIatomic#uxesareequal)toin"nity.The
analogouse$ciencyrangesbetweenoneandzero,respectively. stabilize the surface at the phase boundary. For
7Theratioss4/s2 org4/g2 bythemselvesarenotmeaningful a surface in equilibrium with vapor, the composi-
numbers,becausetheycontainunknown(anddi!erent)ioniz- tions of incident and desorbing #uxes would be
atione$cienciesfor dimersand tetramers.In the comparison
equivalent, and the slope of a log pressure vs. in-
s4/s2%g4/g2,however,theionizatione$cienciescancelouton
verse temperature plot is proportional to the free
leftandright,sothatwemaycomparetheratiosforstabilization
vs.growth. energy of desorption or adsorption. The situation
A.S.Brackeretal./JournalofCrystalGrowth220(2000)384}392 391
in this work is better described as steady state , forGaSbunderdimerandtetramerantimony#ux
‘ a
because the incident Group V composition(dimer have similar slopes and little o!set, in contrast to
andtetramerconcentrations)thatisimposedbythe those for AlSb, suggesting fundamental di!erences
experimentwill generallynot match the desorbing inthekineticsofantimonystabilizationofthetwo
composition. In this case, the slopes and o!sets of material surfaces.
the reconstruction phase boundaries may depend
on the energies and widths of various kinetic bar-
riers.ForInAsandAlSb,thetwolineshavedi!er- Appendix
ent slopes, implyingthat the kinetics of dimer and
tetramer incorporation are di!erent. The parallel In ordertoapplythetransmissionthermometry
lines for GaSb suggest that there is a common techniquetoagivensubstratematerial,aquantitat-
rate-limitingincorporationstepfordimersandtet- ive calibration curve is required. This curve gives
ramers. One possibility is that GaSb has only the in#ection point of the optical transmission
a small barrier for breakup of surface Sb into spectrum as a function of the substrate temper-
4
dimers, so that dimer and tetramer kinetics are ature.Becausetheroutinelyusedcalibrationcurve
dominated by the barrier for subsequent dimer [14]forGaAssubstratesisknowntobeinaccurate
incorporation. byasmuchas303C,wepresenthereupdatedcalib-
The adsorption of Group V species onto III}V ration curves for semi-insulating and n‘-doped
surfaces is in#uenced by the surface structure and GaAs.
stoichiometry. Unlike the antimonides, InAs has The technique used to obtain the calibration
a III-rich reconstruction on the high temperature curves is described in detail in Ref. [16]. Brie#y,
sideofthetransition,soitisnotsurprisingthatthe avacuumfurnacewas usedto heatthe GaAs sub-
qualitativefeaturesofitsphaseboundariesaredif- strate, and a spring-loaded thermocouple was
ferent than at least one of the antimonides. How- placedindirectcontactwiththewafer.Transmission
ever, the similar dimer and tetramer phase spectra were recorded for a series of substrate tem-
boundary slopes for GaSb stand out compared to peratures.InFig.5,thein#ectionpointofthetrans-
AlSb,consideringthatthetwomaterialshavesim- mission edge is plotted as a function of substrate
ilar surfaces. Above the phase transition, both ex-
hibit a (4]3) bilayer-Sb reconstruction with
a small fraction of Group III atoms ([40], see
footnote 3). Below the phase transition, AlSb has
ac(4]4)reconstruction,whileGaSbshows(2]5)-
like reconstructions. The (2]5) surfaces have the
highest Group V content of any III}V zincblende
material and are conducting rather than semicon-
ducting [41]. These unusual features may contrib-
ute to the di!erence in AlSb and GaSb reactivity
with the two types of antimony.
5. Conclusion
The surface reconstruction phase diagrams re-
portedhere should serveas a guide for optimizing
Fig. 5. Calibration curve for transmission thermometry of
growth conditions and as a means for calibrating
semi-insulating and n‘-doped (n"1018cm~3) GaAs. The
substratetemperatureforMBEgrowthofthe6.1A
s dashedcurverepresentsthepreviouscalibrationfromRef.[14]
compound semiconductors. The phase boundaries forbothtypesofsubstrates.
392 A.S.Brackeretal./JournalofCrystalGrowth220(2000)384}392
temperature.The"tsarefourth-orderpolynomials nologyEnhancementsforMBEProductionofMIMICS,
with the form „"a#bE#cE2#dE3#eE4, MIMICPhaseIIIContractDAAL01-89-C-0907,Varian
with „ in 3C and E in cm~1. The coe$cients are Associates,PaloAlto,CA,1991.
a"!9070.5, b"4.8124, c"!8.1385]10~4, [15] E.S. Hellman, J.S. Harris, J. Crystal Growth 81 (1987)
39.
d"5.8157]10~8, e"!1.5514]10~12 for semi-
[16] M.J. Yang, W.J. Moore, C.H. Yang, R.A. Wilson,
insulating GaAs and a"!2.6712]104, b" B.R.Bennett, B.V.Shanabrook, J. Appl. Phys. 85 (1999)
11.759, c"!1.8343]10~3, d"1.2436]10~7, 6632.
e"!3.1536]10~12 for n‘ GaAs.For these cal- [17] J. Piao, R. Beresford, W.I. Wang, J. Vac. Sci. Technol.
ibrations, wafer thickness was 470$15lm B8(1990)276.
[18] M.Yano,H.Yokose,Y.Iwai,M.Inoue,J.CrystalGrowth
(SI}GaAs) and 440$15lm (n‘-GaAs). The
111(1991)609.
n‘ doping level was (1}1.5)]1018cm~3. [19] K. Reginski, J. Muszalski, V.V. Preobrazhenskii, D.I.
Lubyshev,ThinSolidFilms267(1995)54.
[20] K. Reginski, J. Muszalski, V.V. Preobrazhenskii, D.I.
Lubyshev,ThinSolidFilms267(1995)51.
Acknowledgements
[21] C. Chatillon, J.C. Harmand, F. Alexandre, J. Crystal
Growth130(1993)451.
The authors would like to thank B.V. Shanab- [22] L.DaKweritz,R.Hey,Surf.Sci.236(1990)15.
rook for helpful discussions. This work was sup- [23] S.M.Newstead,R.A.A.Kubiak,E.H.C.Parker,J.Crystal
ported by the O$ce of Naval Research and Growth81(1987)49.
[24] K.Yang,L.J.Schowalter,Appl.Phys.Lett.60(1992)1851.
DARPAundertheVirtualIntegratedPrototyping
[25] A.M.Dabiran,P.I.Cohen,J. CrystalGrowth150(1995)
Initiative. ASB was supported by a National Re-
23.
search Council NRL Research Associateship. [26] H.Yamaguchi,Y.Horikoshi,Appl.Phys.Lett.64(1994)
2572.
[27] P.M.Thibado,B.R.Bennett,B.V.Shanabrook,L.J.Whit-
man,J.Crystal.Growth175(1997)317.
References
[28] P.D.Brewer,D.H.Chow,R.H.Miles,J.Vac.Sci.Technol.
B14(1996)2335.
[1] J.B.Boosetal.,IEEETrans.Electron.Devices45(1998) [29] E.S.Tok,J.H.Neave,F.E.Allegretti,J.Zhang,T.S.Jones,
1869. B.A.Joyce,Surf.Sci.371(1997)277.
[2] D.H.Chowetal.,IEEEElectron.DeviceLett.17(1996)69. [30] E.S. Tok, T.S. Jones, J.H. Neave, J. Zhang, B.A. Joyce,
[3] F.Fuchsetal.,Appl.Phys.Lett.71(1997)3252. Appl.Phys.Lett.71(1997)3278.
[4] W.W.Bewleyetal.,Appl.Phys.Lett.74(1999)1075. [31] C.T.Foxon,B.A.Joyce,Surf.Sci.50(1975)434.
[5] P.R.Hammaretal.,Phys.Rev.Lett.83(1999)203. [32] E.S.Tok,J.H.Neave,J.Zhang,B.A.Joyce,T.S.Jones,Surf.
[6] B.R.Bennett,B.V.Shanabrook,Molecularbeamepitaxy Sci.374(1997)397.
ofSb-basedsemiconductors,in:A.W.K.Liu,M.B.Santos [33] Y.Fukunishi,H.Nakatsuji,Surf.Sci.316(1994)168.
(Eds.),ThinFilms:HeteroepitaxialSystems,WorldScient- [34] C.T.Foxon,B.A.Joyce,Surf.Sci.64(1977)293.
i"c,London,1999,p.401. [35] C.E.C.Wood,C.R.Stanley,G.W.Wicks,M.B.Esi,J.Appl.
[7] J.Y. Tsao, Materials fundamentals of Molecular Beam Phys.54(1983)1868.
Epitaxy,AcademicPress,Boston,1993. [36] C.Webb,D.Liu,J.N.Eckstein,Appl.Phys.Lett.60(1992)
[8] M.A.Herman,H.Sitter,MolecularBeamEpitaxy:Funda- 571.
mentalsand CurrentStatus,2ndEdition,Springer,New [37] A.K.Ott,S.M.Casey,S.R.Leone,J.CrystalGrowth181
York,1996. (1997)326.
[9] T.Ito,K.Shiraishi,Jpn.J.Appl.Phys.37(1998)4234. [38] K.R. Evans, C.E. Stutz, P.W. Yu, C.R. Wie, J. Vac. Sci.
[10] K.Shiraishi,ThinSoldFilms272(1996)345. Technol.B8(1990)271.
[11] B.V.Shanabrook,J.R.Waterman,J.L.Davis,R.J.Wagner, [39] J.R.Waterman,B.V.Shanabrook,R.J.Wagner,J.Vac.Sci.
Appl.Phys.Lett.61(1992)2338. Technol.B10(1992)895.
[12] B.V.Shanabrook,J.R.Waterman,J.L.Davis,R.J.Wagner, [40] W.Barvosa-Carter,A.S.Bracker,B.R.Bennett,J.C.Cul-
D.S.Katzer,J.Vac.Sci.Technol.B11(1993)994. bertson,B.Z.Nosho,B.V.Shanabrook,L.J.Whitman,N.
[13] D.S.Katzer,B.V.Shanabrook,J.Vac.Sci.Technol.B11 Modine, H. Kim, E. Kaxiras, Phys. Rev. Lett. 84 (2000)
(1993)1003. 4649.
[14] R.Powell,D.Kirillov,Improvedtemperatureuniformity [41] L.J. Whitman, P.M. Thibado, S.C. Erwin, B.R. Bennett,
andmeasurements,in:T.Cooper(Ed.),ProcessandTech- B.V.Shanabrook,Phys.Rev.Lett.79(1997)693.