Si(011)16×2 gas-source molecular beam epitaxy: Growth kinetics

Article abstract of DOI:10.1063/1.126495

The growth rates R_Si of Si layers deposited on Si(011) 16×2 by gas-source molecular beam epitaxy from Si₂H₆ were determined as a function of temperature T_s (400-975 °C) and Si₂H₆ flux J_Si2H6(5.0×10¹⁵-9.0×10¹⁶cm ^-2s^-1). R_Si ranges from 0.0015 μmh^-1 at T_s=400°C to 0.415 μm h^-1 at T_s=975 °C with J_Si2H6=2.2×10¹⁶cm^-2s^-1. In the surface-reaction-limited regime at T_s <725 °C, R_Si initially exhibits an exponential decrease with 1/T_s, then decreases at a slower rate at T_s≤550°C as an additional deposition pathway becomes operative. In the impingement-flux-limited regime. 725≤T_s≤900°C, R_Si is independent of T_s but increases linearly with J_Si2H6. At T_s>900°C, R_Si(T_s) increases with T_s due to surface roughening. Overall, R^·_Si(J_Si2H6,T_s) is well described at T_s≤900°C by a kinetic model incorporating two competing film growth mechanisms: (1) dissociative chemisorption of Si₂H₆ onto dangling bonds followed by fast surface dissociation steps and second-order H₂ desorption from the surface monohydride phase; and (2) Si₂H₆ insertion into Si-H surface bonds followed by second-order desorption of SiH₄.

Full text of DOI:10.1063/1.126495

Si(011) 16×2 gas-source molecular beam epitaxy: Growth kinetics
N. Taylor, H. Kim, P. Desjardins, Y. L. Foo, and J. E. Greene
Citation: Applied Physics Letters 76, 2853 (2000); doi: 10.1063/1.126495
View online: http://dx.doi.org/10.1063/1.126495
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/76/20?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Si 1−y C y / Si (001) gas-source molecular beam epitaxy from Si 2 H 6 and CH 3 SiH 3 : Surface reaction paths
and growth kinetics
J. Appl. Phys. 93, 3944 (2003); 10.1063/1.1555704
Temperature-modulated Si(001):As gas-source molecular beam epitaxy: Growth kinetics and As incorporation
Appl. Phys. Lett. 79, 3263 (2001); 10.1063/1.1415420
Arsenic incorporation during Si(001):As gas-source molecular-beam epitaxy from Si 2 H 6 and AsH 3 : Effects on
film-growth kinetics
J. Appl. Phys. 88, 7067 (2000); 10.1063/1.1324701
Arsenic-doped Si(001) gas-source molecular-beam epitaxy: Growth kinetics and transport properties
Appl. Phys. Lett. 74, 1290 (1999); 10.1063/1.123527
Growth of Si 1−x Ge x (011) on Si(011)16×2 by gas-source molecular beam epitaxy: Growth kinetics, Ge
incorporation, and surface phase transitions
J. Appl. Phys. 85, 501 (1999); 10.1063/1.369481
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.251.14.35 On: Wed, 17 Dec 2014 23:10:55

APPLIED PHYSICS LETTERS
VOLUME 76, NUMBER 20
15 MAY 2000
Si„011…16Ã2 gas-source molecular beam epitaxy: Growth kinetics
N. Taylor, H. Kim, P. Desjardins, Y. L. Foo, and J. E. Greenea)
Department of Materials Science, the Materials Research Laboratory, and the Coordinated Science
Laboratory, University of Illinois, Urbana, Illinois 61801
͑Received 19 November 1999; accepted for publication 20 March 2000͒
The growth rates RSi of Si layers deposited on Si͑011͒‘‘16ϫ2’’ by gas-source molecular beam
epitaxy from Si2H6 were determined as a function of temperature Ts ͑400–975 °C͒ and Si2H6 flux
JSi H (5.0ϫ1015–9.0ϫ1016 cmϪ2 sϪ1). RSi ranges from 0.0015 ␮m hϪ1 at Tsϭ400 °C to 0.415 ␮m
2
6
hϪ1 at Tsϭ975 °C with JSi H ϭ2.2ϫ1016 cmϪ2 sϪ1. In the surface-reaction-limited regime at Ts
2
6
Ͻ725 °C, RSi initially exhibits an exponential decrease with 1/Ts , then decreases at a slower rate
at Tsр550 °C as an additional deposition pathway becomes operative. In the
impingement-flux-limited regime, 725рTsр900 °C, RSi is independent of Ts but increases linearly
with JSi H . At TsϾ900 °C, RSi(Ts) increases with Ts due to surface roughening. Overall,
2
6
RSi(JSi H ,Ts) is well described at Tsр900 °C by a kinetic model incorporating two competing film
2
6
growth mechanisms: ͑1͒ dissociative chemisorption of Si2H6 onto dangling bonds followed by fast
surface dissociation steps and second-order H2 desorption from the surface monohydride phase; and
͑2͒ Si2H6 insertion into Si–H surface bonds followed by second-order desorption of SiH4. © 2000
American Institute of Physics. ͓S0003-6951͑00͒02320-2͔
Si1ϪxGex(011)
offers
potential
benefits
over
first reported investigation of the temperature-and flux-
dependent growth kinetics of Si͑011͒ gas-source molecular
beam epitaxy ͑GS-MBE͒ from Si2H6.
Si1ϪxGex(001) for certain types of device applications. The
maximum film/substrate conduction-band offset in strained
Si1ϪxGexSi͑011͒ heterostructures is predicted to be substan-
tially larger,1 the optical band gap decreases more rapidly
with increasing x,1 and optical selection rules allow hole–
intersubband transitions to be excited by light polarized par-
allel to Si1ϪxGex /Si͑011͒ multiple quantum well layers.2
A major difference between Si͑011͒ and Si͑001͒, which
has a significant effect on film growth kinetics and mecha-
nistic rate limitations during gas-source film growth, is the
considerably higher complexity of the Si͑011͒ surface unit
cell. Our previous results3,4 confirmed published reports
showing that the clean surface reconstruction is ‘‘16ϫ2.’’
However, this surface, while labeled 16ϫ2 based upon ob-
served diffraction pattern periodicities, actually has a nonor-
thogonal
All films were grown in a multichamber ultrahigh-
vacuum ͑UHV͒ system, described in detail in Ref. 3, with a
base pressure of 5ϫ10Ϫ11 Torr. The system is equipped with
provisions for TPD, reflection high-energy electron diffrac-
tion ͑RHEED͒, low-energy electron diffraction, and Auger
electron spectroscopy ͑AES͒. The film growth experiments
were carried out at temperatures Tsϭ400–975 °C using
Si2H6 precursor fluxes JSi H ϭ5.0ϫ1015–9.0ϫ1016 cmϪ2
2
6
s
Ϫ1. During deposition, the Si2H6 molecular beam is deliv-
ered to the substrate through a directed tubular doser. The
substrates were 0.5 mm thick B-doped Si͑011͒ wafers (3
ϫ1017 cmϪ3) with a miscut of 0.126° along ͗112͘.
Substrate cleaning consisted of solvent degreasing and
wet chemical oxidation/etch cycles followed by an ultravio-
let ozone treatment, UHV degassing, and oxide desorption.3
RHEED patterns from substrates subjected to this procedure
were sharp 16ϫ2. No residual C or O was detected by AES.
Deposited Si film thicknesses were determined using
secondary ion mass spectrometry and surface profilometry.
The Cameca IMS-5F spectrometer was operated with a
10 keV Oϩ2 primary ion beam to detect 11B from the lightly
doped substrates during sputter etching. Surface profilometry
was then used to measure the depth of the crater. The uncer-
tainty in film thickness is р 4 nm.
17
2
1
2
ͫ
ͬ
¯
¯
structure with surface vectors
¯
asϭ17
¯
aϩb and bsϭ2
¯a
3,4–6
¯
¯
¯
͓
ϩ2b, where
¯
aϭ1/2 011 and bϭ 100 .
The Si͑011͒
͓
͔
͔
surface reconstruction undergoes a gradual and reversible
transformation from 16ϫ2 to 1ϫ1 at temperatures between
700 and 770 °C.3,7,8 We showed, using D2 temperature pro-
grammed desorption ͑TPD͒, that the 16ϫ2 reconstruction
has a unit cell structure consisting of 16 adatom species and
8 ␲-bonded dimers, resulting in a maximum dangling bond
coverage of 0.5 ML referenced to the 1ϫ1 bulk surface.3
While the Si͑011͒ surface structure has received signifi-
cant attention, very little is known about the kinetics of
Si͑011͒ epitaxy. In this letter, we present the results of the
All Si͑011͒ layers exhibited sharp 16ϫ2 RHEED pat-
terns. D2 TPD spectra were obtained from Si͑011͒16ϫ2 sur-
faces dosed for successively longer times to provide normal-
ized coverages ␪D /␪D,sat , where ␪
is the saturation
D,sat
coverage. The spectra are very similar to those we obtained
4
from Si͑111͒ and consist of a low temperature desorption
a͒
peak centered at 430 °C and a high temperature feature at
Author to whom correspondence should be addressed; electronic mail:
greene@mrlxp2.mrl.uiuc.edu
550 °C. The primary difference between the Si͑011͒ and
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
0003-6951/2000/76(20)/2853/3/$17.00 2853 © 2000 American Institute of Physics
138.251.14.35 On: Wed, 17 Dec 2014 23:10:55

2854
Appl. Phys. Lett., Vol. 76, No. 20, 15 May 2000
Taylor et al.
FIG. 1. Fitted ͑solid and dashed lines͒ and measured ͑filled circles͒ D2TPD
spectra from a Si͑011͒ sample with ␪D /␪D,satϭ1.0. The linear heating rate in
all experiments was 5 °C sϪ1
.
Si͑111͒ TPD spectra is that the high-temperature Si͑011͒ fea-
ture contains a more obvious shoulder peak, split in this case
by 30 °C and centered at 520 °C. Following the procedure
described in Refs. 3 and 4, we have fit the data using stan-
dard Polanyi–Wigner analyses. An example is shown in Fig.
1. The ␤2 peak arises due to D2 desorption from the dideu-
FIG. 3. ͑a͒ Log–log plots of Si͑011͒ growth rates RSi as a function of the
incident Si2H6 flux JSi H at Tsϭ550 and 700 °C. ͑b͒ RSi͑cmϪ2 sϪ1͒ vs JSi H
2 6
at Tsϭ700 °C.
2
6
*
teride phase while the higher temperature ␤1 and ␤1 peaks
layer growth proceeds by the second-order dissociative ad-
sorption of Si2H6 onto dangling bonds whose coverage ␪db is
determined by the associative desorption rate of H2. The
surface structural phase transition from 1ϫ1 to 16ϫ2,
which for Si͑011͒ occurs near 700 °C, does not have a sig-
nificant effect on the shape of RSi(Ts).9 At TsϾ725 °C, the
steady-state hydrogen coverage on the 1ϫ1 reconstructed
Si͑011͒ surface is sufficiently low (␪HϽ0.19 ML) that RSi is
limited primarily by the Si2H6 impingement rate. With
are associated with desorption from adatoms and rest atoms
in the monohydride phase.3,4 All peaks are second order and
we have ignored the small feature near 375 °C, attributed to
desorption from trideuteride species, since it is not relevant
to the present Si͑011͒ growth kinetics experiments. Fre-
quency factors and desorption activation energies Ed are
v
1ϫ1015 sϪ1 and 2.22Ϯ0.02 eV for ␤2 , 2ϫ1015 sϪ1 and
2.55Ϯ0.02 eV for ␤1 , and 2ϫ1015 sϪ1 and 2.68Ϯ0.02 eV
*
for ␤1 .3,4
J
Si H ϭ2.2ϫ1016 cmϪ2 sϪ1, RSiӍ0.21 ␮m hϪ1 at 725рTs
2 6
The film growth rate RSi as a function of Ts with JSi H
2
6
р900 °C. The increase in RSi at TsϾ900 °C is due to sur-
face roughening as observed by RHEED, resulting in a
higher surface area and a change in the overall Si2H6 sticking
probability due to orientation dependence.
At Tsр550 °C, a new surface reaction pathway becomes
operative and RSi grows progressively larger than values pre-
dicted by a single exponential dependence on 1/Ts . As dis-
cussed below, we propose that this is due to RSi becoming
increasingly controlled by direct insertion of Si2H6 into Si–H
surface bonds.
ϭ2.2ϫ1016 cmϪ2 sϪ1 is shown in Fig. 2. Over the tempera-
ture range between 550 and 900 °C, RSi(Ts) follows the gen-
eral form expected for chemical vapor deposition in which
RSi decreases exponentially with 1/Ts at low temperatures in
a surface-reaction-limited mode, while at high temperatures
RSi tends to saturation in an impingement-flux-limited mode.
In the surface-reaction-limited growth mode, RSi increases
from 0.011 ␮m hϪ1 at 550 °C to 0.111 ␮m hϪ1 at 700 °C as
Figure 3͑a͒ is a plot of RSi as a function of Si2H6 flux
during film growth in the surface-reaction-limited regime
(Tsϭ550 °C) and in the transition region (Tsϭ700 °C). The
steady-state H coverage ␪H , obtained from using Eq. ͑2͒
below and noting that (␪dbϩ␪H)ϭ1 ML, is 0.87 ML at Ts
ϭ550 °C and only 0.26 at Tsϭ700 °C for JSi H ϭ2.2
2
6
ϫ1016 cmϪ2 sϪ1. RSi at Tsϭ550 °C remains constant at
0.011Ϯ0.001 ␮m hϪ1 and is not a function of JSi H since
2
6
film growth in this regime is controlled by the H2 desorption
rate rather than the precursor dosing rate. In contrast, RSi at
700 °C initially increases linearly with JSi H , typical of flux-
2
6
limited growth, before saturating at a value of 0.153
Ϯ0.006 ␮m hϪ1 with JSi H у3.3ϫ1016 cmϪ2 sϪ1. Thus, ␪
H
2
6
increased sufficiently at the higher fluxes, due to competition
between Si2H6 adsorption and H2 desorption, to drive the
FIG. 2. Experimental and calculated Si͑011͒ growth rates RSi as a function
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
of temperature Ts for an incident Si2H6 flux JSi H ϭ2.2ϫ1016 cmϪ2 sϪ1
.
growth kinetics into the surface-reaction-limited regime.
2
6
138.251.14.35 On: Wed, 17 Dec 2014 23:10:55

Appl. Phys. Lett., Vol. 76, No. 20, 15 May 2000
Taylor et al.
2855
We have previously demonstrated that RSi(JSi H ,Ts) for
way has been suggested by Kulkarni et al.11 and Boland12 to
contribute to low-temperature Si͑111͒ and Si͑001͒ growth,
respectively. Two of the three monohydride adatom back-
bonds on the Si͑011͒ 16ϫ2 surface are highly strained and
easily broken during H2 adsorption to form dihydride
species.3,4 By analogy, we propose that the backbonds are
also attacked by Si2H6, resulting in bond breakage and the
insertion of a SiH2 fragment accompanied by SiH4 desorp-
tion. This results in the addition of a second term to Eq. ͑3͒,
2
6
GS-MBE deposition from Si2H6 on the much simpler
Si͑001͒2ϫ1 ͑Ref. 10͒ surface can be well described by a
model, containing no fitting parameters, in which second-
order dissociative chemisorption of Si2H6 is followed by a
series of fast surface decomposition reactions with the final
step being first-order H2 desorption from the surface mono-
hydride phase. However, our TPD results for Si͑011͒ show
that Si2H6 adsorption and H2 desorption are both second
order.3,4 Thus, making the reasonable assumption that the
intermediate surface reaction steps are also fast on Si͑011͒,
Si–H
2
2J
S
␪
H
Si H
6
2JSi H SSi H
␪
db
Si H
2
6
2
6
2
6
2
RSiϭ
ϩ
.
͑4͒
is the Si2H6 reactive sticking probability via inser-
6
N
N
the overall change in ␪ during Si͑011͒ GS-MBE is given
db
Si–H
by the expression
S
Si H
2
d␪
tion, a thermally activated process with activation energy
ESi–H . Figure 2 shows that calculated results based on Eqs.
͑2͒ and ͑4͒ exhibit very good agreement with experimental
data over the entire growth temperature range until surface
roughening becomes significant at TsϾ900 °C. The best fit
db ϭϪ2JSi H ka␪ ϩkd 1Ϫ␪ ͒2.
͑1͒
2
db
͑
db
2
6
dt
The reaction rate constant ka for Si2H6 adsorption on
Si͑011͒ can be expressed as SSi H /Ns in which SSi H is the
2
6
2 6
zero-coverage reactive sticking probability and Ns is the bulk
surface site number density, 9.6ϫ1014 cmϪ2. kd in Eq. ͑1͒ is
the recombinative H2 desorption rate constant which can be
expressed as ␯•exp(Ed /kTs), where the frequency factor ␯
ϭ2ϫ1015 sϪ1 and the desorption activation energy Ed
ϭ2.68 eV are known from the ␤1 TPD results discussed
above. SSi H at 700 °C was determined from the slope of the
shown by the solid line in Fig. 2 gives ESi–Hϭ0.5 eV which
Si–H
yields S
values varying from 5.24ϫ10Ϫ4 at 550 °C to
Si H
2
6
1.09ϫ10Ϫ4 at 400 °C.
In conclusion, the RSi(Ts) data are well described by a
kinetic model incorporating two competing film growth
mechanisms: ͑1͒ dissociative chemisorption of Si2H6 onto
dangling bonds followed by fast surface dissociation steps
and second-order H2 desorption from the surface monohy-
dride phase; and ͑2͒ Si2H6 insertion into Si–H surface bonds
followed by second-order desorption of SiH4.
2
6
RSi vs JSi H data in Fig. 3͑b͒ to be 0.0075, approximately
2
6
20% of the Si͑001͒ value.10 We assume, as in the case for
Si͑001͒ GS-MBE from Si2H6,11 that SSi H for Si͑011͒ is not
2
6
strongly Ts dependent.
The steady-state dangling bond coverage obtained from
Eq. ͑1͒ is
The authors acknowledge the financial support of the
SRC and the Materials Science Division of the US Depart-
ment of Energy ͑DOE͒ under Award No. DEFG02-
ER9645439. P.D. is partially supported by NSERC of
Canada. The authors also appreciate the use of the Center for
Microanalysis of Materials at the University of Illinois,
which is partially supported by the DOE.
0.5 Ϫ1
2ISi H SSi H
2 6
2
6
␪ ϭ 1ϩ
͑2͒
ͫ
ͩ
ͪ
ͬ
db
N ␯ exp ϪE /kT ͒
͑
s
d
s
and the film growth rate is given by
2
db
2JSi H SSi H
␪
2
6
2
6
RSiϭ
,
͑3͒
1 C. W. Liu, J. C. Sturm, Y. R. J. Lacroix, M. L. W. Thewalt, and d. D.
Perovic, Appl. Phys. Lett. 65, 76 ͑1994͒.
N
2 T. L. Kreifels, R. L. Hengehold, Y. K. Yeo, P. E. Thomson, and D. S.
Simons, J. Vac. Sci. Technol. A 13, 636 ͑1995͒.
in which N is the bulk Si atom number density, 5
ϫ1022 cmϪ3. In the flux-limited regime where ␪ ap-
3 N. Taylor, H. Kim, T. Spila, J. A. Eades, G. Glass, P. Desjardins, and J. E.
Greene, J. Appl. Phys. 85, 501 ͑1999͒.
db
proaches its saturation value, Eq. ͑3͒ predicts RSi to be nearly
independent of Ts and to increase linearly with JSi H , as
4 H. Kim, N. Taylor, T. Spila, G. Glass, S. Y. Park, J. E. Greene, and J. R.
Abelson, Surf. Sci. 380, L496 ͑1997͒.
2
6
5 S. Miura, K. Kato, T. Ide, and T. Ichinokawa, Surf. Sci. 191, 259 ͑1987͒.
6 E. J. van Loenen, D. Dijkkamp, and A. J. Hoeven, J. Microsc. 152, 487
͑1988͒.
observed in Fig. 3͑b͒.
The dashed curve in Fig. 2 is a plot of RSi vs Ts based
upon Eqs. ͑2͒ and ͑3͒. While it provides a good fit to the data
for the growth temperature range 550ϽTsр900 °C, experi-
mental RSi values are higher than predicted at both higher
and lower temperatures. The difference at higher tempera-
tures is due to surface roughening.
The RSi(Ts) results at Tsр550 °C indicate that a new
reaction path for film growth becomes competitive with the
mechanism described by Eqs. ͑2͒ and ͑3͒. Dangling bond
coverages determined by TPD range from 0.133 ML at
550 °C to 0.002 ML at 400 °C. At these temperatures, where
7 G. Shimaoka, Appl. Surf. Sci. 65Õ66, 569 ͑1993͒.
8 Y. Yamamoto, T. Sueyoshi, T. Sato, and M. Iwatsuki, Jpn. J. Appl. Phys.,
Part 2 32, L532 ͑1993͒.
9 The effect of the surface phase transformation on RSi is difficult to observe
since it occurs over a narrow temperature range in the flux-limited to
surface-reaction-limited transition region. However, we know from TPD
results that ␪db(1ϫ1)ϭ2␪db(16ϫ2). Thus, the RSi(Ts) data suggest that
the Si2H6 reactive sticking probability is lower on the 1ϫ1 than on the
16ϫ2 reconstructed surface.
10 T. R. Bramblett, Q. Lu, T. Karasawa, M.-A. Hasan, S. K. Jo, and J. E.
Greene, J. Appl. Phys. 76, 1884 ͑1994͒.
11 S. K. Kulkarni, S. M. Gates, B. A. Scott, and H. H. Sawin, Surf. Sci. 239,
13 ͑1990͒; S. K. Kulkarni, S. M. Gates, C. M. Greenlief, and H. H. Sawin,
ibid. 239, 26 ͑1990͒.
␪
is high and approaching saturation, we propose that film
H
growth via the direct insertion of incident Si2H6 molecules
into Si–H surface bonds becomes significant. A similar path-
12 J. J. Boland, Phys. Rev. B 44, 1383 ͑1991͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
138.251.14.35 On: Wed, 17 Dec 2014 23:10:55