Full text of DOI:10.1007/BF02665847

6Jo3ur2nal of ELECTRONIC MATERIALS, Vol. 30, No. 6, 2001
Special Issue Paper
Martinka, Almeida, Benson, and Dinan
Characterization of Cross-Hatch Morphology of MBE
(211) HgCdTe
M. MARTINKA, L.A. ALMEIDA, J.D. BENSON, and J.H. DINAN
U.S. ArmyCECOMNightVision&ElectronicSensorsDirectorate, 10215BurbeckRoad, Ft. Belvoir,
VA 22060-5677
We present the results of a detailed study of the nature and origin of cross-hatch
patterns commonly observed on (211) HgCdTe epilayers deposited by molecular
beam epitaxy. Cross-hatch patterns were examined using x-ray topography as
well as Nomarski, interferometric, and atomic force microscopies. Cross-hatch
–
patterns were–generally comprised of three sets of lines, parallel to the [231],
–
–
[213], and [011] directions. The lines parallel to the [011] direction exhibited
–
–
distinct properties compared to the two sets of lines parallel to [231] and [213].
Under growth conditions characterized by excessive Hg flux (low temperature),
–
lines parallel to [011] were periodic and tended to dominate the cross-hatch
pattern. In some cases, bands of dislocations, 10–100 mm in width, formed
–
parallel to [011]. Under optimized growth conditions, on very closely lattice-
–
matched substrates, (dislocation densitie–s <105 cm––2) lines parallel to [011]
vanished entirely, and lines parallel to [231] and [213] became sparse. The
remaining lines were typically fragments terminated by either a single disloca-
tion, a cluster of dislocations (micro-void), or the wafer’s edge. The density of
these line fragments tended to decrease as the dislocation density decreased.
Under the best growth conditions on very closely lattice-matched substrates we
have achieved dislocation densities of 5 ¥ 104 cm–2, which is comparable to the
dislocation density of the CdZnTe substrate.
Key words: Cross-hatch, HgCdTe, CdZnTe, lattice-mismatch, defects
–
INTRODUCTION
[011] directions. The three sets of cross-hatch lines
–
were not equivalent; the lines parallel to the [011]
For nearly lattice-matched heteroepitaxy, cross-
hatch patterns have been observed in numerous ma-
terial systems and have been attributed to strain-
induced modulation of the surface. In the case of
(111)B HgCdTe/CdZnTe the nature and origin of
these patterns have been reported.1,2 Patterns pos-
sess three-fold symmetry and consist of <110> slip
directions, defined as the intersections of a (111)
growth plane with three other {111} planes. A corre-
lation was found between the appearance or absence
of cross-hatch and the degree of lattice matching and
dislocation density. For (211)B HgCdTe, several ob-
servations of cross-hatch and its relation to lattice
mismatch have been reported.3-6 In this work we
present a detailed account of the cross-hatch patterns
observed in (211)B HgCdTe deposited by molecular
beam epitaxy (MBE). These patterns were comprised
direction exhibited distinct properties compared to
–
–
the two sets of lines parallel to [231] and [213].
EXPERIMENTAL DETAILS
Hg1–xCdxTeMBEisachievedusingamodifiedFisons
VG-80 chamber, equipped with spectroscopic
ellipsometry (SE) for composition control. Standard
CdTe and Te effusion cells and a customized Hg cell
deliver source materials. Epitaxy was carried out on
bulk Cd1–yZnyTe (211)B (0.02 < y < 0.035) substrates,
with etch pit densities (EPD) < 5 ¥ 104 cm–2. The
nominal growth temperature was 185∞C. Substrates
were mounted on graphite susceptors using colloidal
graphite suspended in isopropanol. Hg1–xCdxTe
epilayers studied here were all single layers with
constant compositional profiles (Dx < 0.02) as mea-
sured by SE; no buffer layers were employed. Our
MBE process was well controlled both for epilayer
composition and crystalline quality. Epilayers rou-
–
–
of three sets of lines, parallel to the [231], [213], and
(Received November 21, 2000; accepted January 16, 2001)
632

Characterization of Cross-Hatch
Morphology of MBE (211) HgCdTe
633
a
b
Fig. 1. (a) Schematic representation–of geometric relationship between cross-hatch lines and {011} cleavage planes. (b) Nomarski micrograph of
–
cross-hatch lines and relation to (101) and (101) cleaves. Black lines added to clarify directions.
tinely exhibited EPD values <1 ¥ 106 cm–2, while more
than 50% of the epilayers exhibited EPD values < 5 ¥
105 cm–2.Whilecross-hatchpatternswereobservedon
Hg1–xCdxTe epilayers of various compositions
(0.22 < x < 0.50), the majority of results presented
below were for epilayers with 0.3 < x < 0.45. Epilayer
thickness ranged from 4 mm to 8 mm.
geometricrelationshipsbetweenthecross-hatchlines
and the {110} cleavage planes, and Fig. 1b displays a
Nomarskimicrographshowingcross-hatchlinesofan
as-grown Hg1–xCdxTe epilayer and their relation to
–
–
the (101) and (110) cleaves. As shown in Fig. 1–, we
–
have defined a as the angle between [231] and [213]
–
and have defined b as the angle between [011] and
–
[231].Theseanglesweremeasuredonseveralsamples
and were found to be a = 44.4 ± 0.3∞ (6 samples, 20
measurements) and b = 67.9 ± 0.5∞ (4 samples, 8
measurements), in excellent agreement with pre-
dicted values. The correct assignment of the cleavage
planes was confirmed by measurement of angles
between cleavage planes and the (211) surface.
Detailed microscopy was performed on a Hg1–
Cross-hatch patterns of selected Hg1–xCdxTe
epilayers were observed using x-ray topography, in-
terferometric microscopy, and atomic force micros-
copy (AFM). Nomarski microscopy was used exten-
sively in this study. Dislocation densities of
Hg1–xCdxTe epilayers were determined using stan-
darddefectdecorationetching7 andarereportedhere-
afterasetchpitdensities(EPD).Epilayerx-valueand
thickness were determined using Fourier transform
infrared (FTIR) transmission. Lattice mismatch was
measuredbysubstrateandepilayerpeakseparations
in x-ray diffraction q-2q and double crystal rocking
curves (DCRC). Compositions of selected substrates
were estimated using FTIR.
xCdxTe epilayer grown in tension (afilm–asub/asub
∫
Da/asub~ –0.032%) on a bulk Cd1–yZnyTe substrate
(nominal y = 0.035). This epilayer displayed the
following characteristics: x = 0.228, 5.9 mm thick,
DCRC = 68 arcsec, EPD = 7 ¥ 105 cm–2. The three
sets of cross-hatch lines were–clearly vis–ible using
Nomarski microscopy. The [231] and [213] lines
were sharply defined in x-ray topographs, while
the [01–1] lines exhibited weaker contrast. Atomic
force microscopy (AFM) confirmed there were mor-
phological features –on the Hg1–xCdxTe surface cor-
RESULTS
Orientation and Topography of Cross-Hatch Pattern
As was observed for (111) Hg1–xCdxTe, a (211)B
cross-hatch pattern is generally comprised of three
sets of lines, associated with the intersection of {111}
planes with the (211)B growth plane. The pattern has
mirror –symmetry, and the lines are parallel to the
–
–
responding to [231] [213] [011] directions with depth
variations of 5 Å to 10 Å. Interferometric micros-
copy showed surface modulations similar to those
observed by AFM. Spatial frequency fast Fourier
transform analysis of interferometric data was per-
for–med. This analysis revealed that the
–
–
[231], [213], and [011] directions. These {110} and
{321}planesareparalleltothe[111]zoneaxes.Angles
and azimuthal orientations of the cross-hatch lines
relative to the {110} cleavage planes agree with these
assigned directions. Fig. 1a schematically details the
[011] lines were locally periodic, with a spatial
–
period of 9.4 mm. Similar analysis of the [231] and
–
[213] directions showed no periodicity.

634
Martinka, Almeida, Benson, and Dinan
a
b
c
Fig. 2. Nomarski micrographs of three defect etched HgCdTe epilayers grown consecutively at higher substrate temperatures (a) 188∞C, (b) 189∞C,
and (c) 191∞C.
Fig. 3. Cross-hatch line density as a function of EPD.
Dislocations Related to Growth Conditions
Fig. 4. Nomarski micrograph of defect decoration etched HgCdTe
–
–
–
Distinction between the [011] and two equivalent
epilayer. Only [231] and [213] cross-hatch lines are visible. Arrows
indicate lines terminated by etch pit.
–
[231]directionswasobservedonHg1–xCdxTeepilayers
grown under conditions characterized by excessive
Hg flux (low temperature). Figure 2 shows Nomarski
micrographs of three defect-etched epilayers grown
sequentially on substrates with a nominal y-value of
0.035. The purpose of this sequence was optimization
of substrate temperature to achieve low EPD values.
The nominal substrate temperature and resulting
EPD values for each epilayer were a) 188∞C EPD
banding, b)189∞C EPD=1.3¥ 106 cm–2, c)191∞C EPD
withlargevoidsand/orareasofpolycrystallinegrowth.
Such epilayers also did not exhibit cross-hatch pat-
terns. Growthconditio–nsseemedtoaffecttheappear-
ance and density of [011] lines. However, this correla-
tion was difficult to quantify, since epilayers which
typically exhibited this correlation (i.e., Fig. 2a and b)
–
had high EPD and [011] line den–sities. On such
samples accurately determining [011] line densities
was typically not possible with optical microscopy. As
explained in the following section, optimized growth
conditions were necessary to discern the effects of
lattice matching on EPD and cross-hatch behavior.
= 2.0 ¥ 105 cm–2. Figure 2a and b (to a lesser degree)
–
shows bands of high EPD running parallel to [011]
with areas between these bands exhibiting low EPD.
The epilayer shown in Fig. 2c exhibited low EPD and
had no bands of dislocations. For this epilayer the
–
–
–
SUPPRESSION OF [011] CROSS-HATCH LINE
[231] and [213] li–nes were prominent and dense com-
pared to the [011] lines. Figure 2 demonstrates the
critical role of growth conditions on EPD values and
distributions.
Under optimized growth conditions, on closely lat-
tice-matched substrates, (epilayer dislocation densi-
ties ~105 cm–2) the lines parallel to [–231] and [–213]
dominatedthecross-hatchpattern.Forsuchas-grown
epilayers, the cross-hatch lines were typically not
visible by Nomarski microscopy but could be revealed
by defect decoration etching.7 It has been observed
that the density of cross-hatch lines tends to decrease
as the dislocation density decreases for epilayers
deposited under optimized conditions. Figure 3 plots
the densities of both [01–1] and [–231] sets of cross-
Some correlations between growth conditions and
cross-hatch patterns were observed. Reasonably well
controlled growth conditions were necessary to ob-
serve cross-hatch patterns. On occasion, epilayers
were grown w–hich exhibited “needle” defects oriented
parallelto[011]. Theseepilayersdidnotexhibitcross-
hatch patterns. Growth conditions characterized by
deficient Hg (high temperature) resulted in epilayers

Characterization of Cross-Hatch
Morphology of MBE (211) HgCdTe
635
a
b
c
Fig. 5. Nomarski micrographs of successive defect decoration etchings on a single piece of HgCdTe originally 5.2 mm thick. (a) etch depth 1.9 mm,
(b) etch depth 3.7 mm, (c) etch depth 4.7 mm. No additional cross-hatch lines and few additional etch pits are revealed by successive etches.
a
b
c
d
Fig. 6. (a and b) Nomarski micrographs of as-grown and (c and d) defect-etched surfaces of Hg1–xCdxTe (x = 0.38) epilayers deposited
simultaneously on Cd1–yZnyTe substrates of different compositions (a and c) y = 0.02, (b and d) y = 0.03.
hatch lines as a function of EPD for 15 Hg1–xCdxTe
epilayers. These epilayers had a variety of composi-
tions (0.31 < x < 0.49) and were grown on bulk Cd1–
yZnyTe substrate (0.03 < y < 0.035). Observation of
cross-hatch line density required defect decoration
etching for nearly all samples. Figure 3 indicates that
to depths of 1.9 mm, 3.7 mm, and 4.7 mm of the original
5.2 mm epilayer. With each successive etch the line
fragments became more sharply defined. No addi-
tional cross-hatch lines and only a few additional etch
pits were revealed by successive etches which con-
-
firmed that the slip planes associated with the [011]
cross-hatchlineswereinactivethroughoutthegrowth
of the epilayer.
the[01–1]setofcross-hatchlinesvanishedforepilayers
–
with EPD < 105 cm–2. The density of the [231] lines
tended to decrease as dislocation density decreased.
Simultaneous Growth on Substrates with
Different Lattice Parameters
Two samples indicated in Fig. 3 exhibited a very low
–
density of short [231] line fragments in small areas of
the wafer, while one sample exhibited no cross-hatch
lines. Presumably, these samples were very nearly
To better understand the role of lattice-mismatch
on EPD and cross-hatch line density, epitaxy was
carried out simultaneously on two substrates with
differing lattice parameters. Hg1–xCdxTe epilayers (x
= 0.38, t = 3.9 mm) were deposited in tension on bulk
Cd1–yZnyTe substrates with nominal compositions,
y = 0.02 and y = 0.03 (designated samples A and B,
respectively). Figure 6 shows Nomarski micrographs
of as-grown and defect-etched surfaces of the result-
ing epilayers. Epilayer A (y = 0.02) faintly displayed
cross-hatch lines as-grown, whereas epilayer B dis-
played no cross-hatch lines on the as-grown surface.
Both samples showed distinct cross-hatch lines when
etched. EpilayerBhadasignificantlylowerdensityof
lattice-matched.
–
We have observed that the [231] sets of cross-hatch
lines on Hg1–xCdxTe epilayers grown under optimized
growth conditions (EPD < 105 cm–2) were typically
fragmentsterminatedbyeitherasingledislocation, a
cluster of dislocations (micro-void), or the wafer’s
edge. Figure 4 is a Nomarski micrograph of a defect-
etched Hg1–xCdxTe epilayer (x = 0.47, t = 5.2 mm) with
an EPD of 4.7 ¥ 104 cm–2, grown on bulk Cd1–yZnyTe
substrate (nominal y = 0.03). For this sample no lines
–
parallel to– [011] were observed. Lines parallel to
–
[231] and [213] had spatial frequencies of ~30 mm–1.
Some of the cross-hatch lines in Fig. 4 (indicated by
arrows)areterminatedbyasingledislocation.Apiece
of the epilayer depicted in Fig. 4 was defect-etched
repeatedly to determine the behavior of the etch pits
and line fragments with depth. The micrographs of
Fig. 5 show the results of three consecutive etchings
–
–
[231] cross-hatch lines and virtually no [011] lines
–
(one [011] line fragment is visible in Fig. 6b). The
EPDs of these samples also showed the influence of
lattice-mismatch. Epilayer A had an EPD of 3.7
¥ 105 cm–2, while the more closely lattice matched
epilayerBhadanEPDof7.3¥ 104 cm–2. SampleBalso

636
Martinka, Almeida, Benson, and Dinan
exhibited an extremely narrow DCRC full width at
half maximum (FWHM) of 17 arcsec.
elastic correction factor C = 0.5 yielded the relaxed
lattice mismatch contributions given above.
Lattice mismatch of these samples was determined
by measuring the separation of the 422 x-ray diffrac-
tionpeaksfromthesubstrateandepilayer.Forsample
A, simultaneous q-2q measurements of epilayer and
substrate determined the lattice constant of the sub-
strate (ymeas = 0.020) and epilayer/substrate peak
separation (230 arcsec). For sample B, DCRC was
required to measure epilayer/substrate peak separa-
tion (56 arcsec). An elastic correction factor C was
estimated from the simultaneous q-2q measurement
of epilayer and substrate (Sample A) and a calculated
relaxed epilayer lattice constant based on FTIR x-
values (xA = 0.385 and xB = 0.382). Using this correc-
tion factor the lattice mismatch of sample B was
determined from the peak separation of the DCRC.
The relaxed lattice mismatch for the two samples
determined from this analysis was Da/asub = –0.07%
for sample A and Da/asub ~ –0.02% for sample B.
These slightly mismatched layers were in tension
and not fully relaxed. Layers that exhibited cross-
hatch and those that did not may have retained
residual strain. The mismatch and residual strain
both contributed to the observed separation of the x-
ray peaks. To distinguish these contributions, the
normal lattice constants of the epilayer and substrate
were measured simultaneously and used together
withanestimatedvalueoftheepilayerrelaxedlattice
constant obtained by applying Vegard’s law to the x-
values measured by FTIR. Even though these biaxi-
allystressedepilayerswithnormalexpansionoftheir
free surface may exhibit changes quite different from
the small, 10 meV/kbar, energy gap dependence on
hydrostatic pressure,8 no significant differences in x-
value were observed in these simultaneously grown
films. Thiswasseeninthesimilarvaluesreportedfor
xA and xB above. Alternatively, significant changes in
x-ray peak separation were observed. Scaling the
observed DCRC peak separation with the resulting
CONCLUSIONS
We have made detailed observations of the cross-
hatch patterns of (211)B HgCdTe grown by MBE.
These observations have revealed a distinction be-
–
tween the cross-hatch lines parallel to [011] and the
–
–
two sets of equivalent lines parallel to [231] and [213].
Under optimized growth conditions on closely lattice
matched substrates (Da/–asub ~ 0.02%) the slip system
associated with t–he [011] lines is inactive, and the
generation of [011] cross-hatch lines is suppressed.
Withtheseconditionssatisfieditispossibletoachieve
HgCdTe epilayers with dislocation densities ~5
¥ 104 cm–2, which is comparable to the dislocation
density of the CdZnTe substrate.
ACKNOWLEDGEMENTS
The authors wish to thank the following colleagues
for their helpful support during this study: Donna
Advena and Phil Boyd of the Army Research Labora-
tory, David Rhiger of Raytheon.
REFERENCES
1. S.P. Tobin, F.T.J. Smith, P.W. Norton, J. Wu, M. Dudley, D.
DiMarzio, and L.G. Casagrande, J. Electron. Mater. 24, 1189
(1995).
2. D.R. Rhiger, S. Sen, J.M. Peterson, H. Chung, and M.
Dudley, J. Electron. Mater. 26, 515 (1997).
3. D.R. Rhiger, S. Sen, and E.E. Gordon, J. Electron. Mater. 29,
669 (2000).
4. D.D. Edwall, M. Zandian, A.C. Chen, and J.M. Arias, J.
Electron. Mater. 26, 493 (1997).
5. T. Skauli, T. Colin, R. Sjolie, and S. Lovold, J. Electron.
Mater. 29, 687 (2000).
6. T.T. Lam, C.D. Moore, R.L. Forrest, M.S. Goorsky, S.M.
Johnson, D.B. Leonard, T.A. Strand, T.J. deLyon, and M.D.
Gorwitz, J. Electron. Mater. 29, 804 (2000).
7. J.H. Tregilgas, T.L. Polgreen, and M.C. Chen, J. Cryst.
Growth 86, 460 (1988).
8. R. Dornhaus and G. Nimtz, Springer Tracts in Modern
Physics 98, 161 (1983).