Ge chemical vapor deposition on GaAs for low resistivity contacts

Article abstract of DOI:10.1149/1.3528273

This paper reports the two-dimensional epitaxial growth of thin intrinsic Ge and in-situ doped n-Ge on GaAs substrates by atmospheric pressure chemical vapor deposition. High quality Ge growth on GaAs is activated almost instantly by an optimized pregrowth procedure. Ga autodoping is found to occur within the first Ge monolayers close to the Ge/GaAs interface. The introduction of high P H₃ flows during the Ge growth yields the in-situ n-Ge growth on GaAs with an electrically activated dopant concentration of 3× 10¹⁹ cm^-3. Additionally, n-Ge selective growth on the source/drain areas of an n-GaAs metal oxide semiconductor structure, followed subsequently by the NiGe formation, has been demonstrated to yield an ohmic contact with a contact resistance of 0.13 ωcm.

Full text of DOI:10.1149/1.3528273

Journal of The Electrochemical Society, 158 ͑3͒ H203-H207 ͑2011͒
H203
0013-4651/2010/158͑3͒/H203/5/$28.00 © The Electrochemical Society
Ge Chemical Vapor Deposition on GaAs for Low Resistivity
Contacts
B. Vincent,a,z A. Firrincieli,a,b W.-E. Wang,c N. Waldron,a A. Franquet,a
B. Douhard,a W. Vandervorst,a,d T. Clarysse,a G. Brammertz,a R. Loo,a,
*
J. Dekoster,a M. Meuris,a and M. Caymaxa,
*
aIMEC, Kapeldreef 75, 3001 Leuven, Belgium
bESAT, K. U. Leuven, Leuven, Belgium
cAssignee at IMEC, Leuven, Belgium
dInstituut voor Kern- en Stralingsfysica, K. U. Leuven, Leuven, Belgium
This paper reports the two-dimensional epitaxial growth of thin intrinsic Ge and in-situ doped n-Ge on GaAs substrates by
atmospheric pressure chemical vapor deposition. High quality Ge growth on GaAs is activated almost instantly by an optimized
pregrowth procedure. Ga autodoping is found to occur within the first Ge monolayers close to the Ge/GaAs interface. The
introduction of high PH3 flows during the Ge growth yields the in-situ n-Ge growth on GaAs with an electrically activated dopant
concentration of 3 ϫ 1019 cm−3. Additionally, n-Ge selective growth on the source/drain areas of an n-GaAs metal oxide semi-
conductor structure, followed subsequently by the NiGe formation, has been demonstrated to yield an ohmic contact with a contact
resistance of 0.13 ⍀ cm.
© 2010 The Electrochemical Society. ͓DOI: 10.1149/1.3528273͔ All rights reserved.
Manuscript submitted November 10, 2010; revised manuscript received November 29, 2010. Published December 29, 2010.
Group III–V materials which are used to replace the Si channel
for high mobility n-channel field-effect transistors ͑nFETs͒ in
CMOS ͑complementary metal oxide semiconductor͒ technology,
have recently attracted many interests.1-3 However, various III–V
related processing issues, such as the required high quality III–V
channel growth inside shallow trench isolation,4 the need of passi-
vating the heavily Fermi-level pinned III–V/high-k interfaces,5 and
the source/drain engineering for lowering contact resistance,6 are the
key challenges that remain to be solved before the implementation
of III–V onto very-large-scale integrated ͑VLSI͒ silicon platform.
GaAs, because of its strong Fermi level pinning with most
metals,7 retards the modulation of the Schottky barrier height at the
metal/GaAs interface and, hence, unlikely to form a good contact
with low resistivity. Standard ohmic contacts on GaAs with low
resistivity have been obtained only with AuGeNi alloyed structures.8
In order to render the GaAs contacts Au-free VLSI compatible, PtGe
contacts have been demonstrated but without any possible self-
aligned approach.9 Recently, selective Ge epitaxial growth by ultra-
high vacuum chemical vapor deposition ͑UHV-CVD͒ on GaAs
source/drain after Ni germanide formation has been proposed by
Luo et al.,10 as a successful Au-free self-aligned approach that takes
advantage of Ge and GaAs being nearly lattice matched.
by UHV-CVD.10,11 In this work, both undoped and n-type doped Ge
were grown by an AP-CVD using a 200 mm ASM Epsilon reactor.
GeH4 ͑1% diluted in H2͒ and PH3 ͑0.1% diluted in H2͒ were used as
Ge and P ͑for in-situ n-type growth͒ precursors, respectively. For all
the experiments, H2 was used as the carrier gas and GeH4 flow was
fixed at 200 sccm.
The Ge growth processes were developed on 2 in. bulk GaAs
semi-insulating substrates. Semi-insulating substrates allow the elec-
trically active dopant concentration to be determined by micro four
point probe measurement,15 four point probe measurement, or Hall
measurement ͑these three methods gave identical sheet resistance
values͒. Ge thickness was determined by cross-section scanning
electron microscopy ͑XSEM͒ and x-ray diffraction ͑XRD͒. In addi-
tion to cross section transmission electron microscopy ͑XTEM͒,
XRD was applied to assess the quality of the layers grown. Surface
morphology was inspected by a Nomarski microscope. Ge/GaAs
interface and Ge–Ga, Ge–As interdiffusion were investigated by
secondary ions mass spectroscopy ͑SIMS͒ and time-of-flight SIMS
͑TOF-SIMS͒. TOF-SIMS analyses were performed with an
ION-TOF IV apparatus. Positive and negative depth profiles were
obtained in a dual beam mode with a Ga beam ͑15 kV͒ for analysis
and a Xe beam ͑350 eV͒ for sputtering.
Results from Luo et al.10 showed that an undesirable Stranski–
Krastanov mode Ge islanding is formed on the GaAs during growth.
A rather thick ͑ Ͼ 100 nm͒ Ge layer is required for a continuous
layer to form after all the islands are collapsed. It is worth noting
that a continuous Ge layer on GaAs is mandatory to avoid any direct
Ni contact to the GaAs. This implies that Luo’s approach does not
allow the implementation of thin NiGe contacts. Recent results from
Zhang et al.11 showed that the islanding issues can be somewhat
lessened by introducing additional Si into the Ge growth. However,
a minimum of 44 nm thick Ge layer is still needed.
This paper reports drastic improvements for growing ultrathin
͑down to 7 nm͒ continuous Ge films on ͑100͒ GaAs substrates by
using the atmospheric pressure CVD ͑AP-CVD͒. Our experimental
protocol is first described, followed by undoped Ge growth on GaAs
and the in-situ growth of highly n-doped Ge, and then concluded
with Ni germanide formation on selective n-Ge grown on n-GaAs.
Circular transmission line method ͑CTLM͒ structures were used
to evaluate the contact resistance of the most heavily n-doped Ge
grown on the semi-insulating GaAs substrates.
Ge AP-CVD on (100) GaAs.— In the following sections, pre-
growth bake conditions are described first and followed by growth
conditions for intrinsic and in-situ n-doped Ge on 2 in. semi-
insulating GaAs substrates.
Pre-growth bake.— Figure 1 compares both the surface mor-
phologies that are characterized by a Nomarski microscope and
͑004͒ ␻-2␪ XRD scans of three Ge layers grown identically at
425°C under atmospheric pressure on semi-insulating GaAs sub-
strates. Each substrate has received a wet-chemical HCl cleaning ͑in
order to remove As and Ga native oxide10͒ and was loaded in the
reactor at 200°C under a H2 environment. It is worth noting that the
HCl cleaning process provides
a chlorine-terminated GaAs
surface.16 Three different bake conditions were tested on those
samples: ͑1͒ temperature ramp-up from 200 to 425°C under a H2
environment, ͑2͒ temperature ramp-up from 200 to 425°C followed
by a H2 bake at 425°C for 10 min, and ͑3͒ temperature ramp-up
from 200 to 425°C followed by an AsH3 bake at 425°C for 10 min.
The perfectly fitted ͑004͒ ␪-2-␪ results from XRD show that Ge
layers ͑peak fringes͒ grown on GaAs are clearly monocrystalline
Experimental
Various undoped Ge growths on GaAs substrates have been de-
scribed in the literature by molecular beam epitaxy12-14 and recently
*
Electrochemical Society Active Member.
z E-mail: benjamin.vincent@imec.be

H204
Journal of The Electrochemical Society, 158 ͑3͒ H203-H207 ͑2011͒
Figure 2. ͑Color online͒ Ge thickness, measured by XSEM, as a function of
deposition time for Ge grown on GaAs at atmospheric pressure, 375, 400,
and 425°C. No incubation time is observed whatever the growth tempera-
ture.
Intrinsic Ge growth on GaAs.— In the papers mentioned
earlier10,11 concerning Ge growth on GaAs surface by UHV-CVD
techniques, long incubation times up to 20–40 min have been re-
ported before starting the Ge growth. It is suspected that this incu-
bation time depends strongly on the As/Ga ratio at the GaAs top
surface, i.e., the richer in As, the longer the incubation.10 For our
AP-CVD growth conditions, Fig. 2 details the Ge thickness depos-
ited at 375, 400, and 425°C as a function of the deposition time. We
note that no incubation time exists with our process. After our pre-
growth procedure and starting the growth at atmospheric pressure, a
Ga rich monolayer is generated immediately on the GaAs surface.
We note, however, that for 425°C, the growth rate stabilizes only
after 10 nm Ge grown at 4.9 nm/min. Slower growth rates for Ge
thinner than 10 nm might be attributed to As or Ga segregation at
the top Ge surface ͑discussed later͒ impacting Ge growth kinetics.
Figure 3 compares the Ge growth rates at 375, 400, and 425°C
for 30 min deposition time. An activation energy of 1.05 eV is ob-
tained. In the inset of Fig. 3 XTEM done on the thinnest ͑7 nm͒ Ge
layers grown on GaAs at 375 and 425°C are compared. It is evident
that these ultrathin Ge layers are continuous without any islanding
formation. Moreover, the Ge layers grown are of high epitaxial qual-
ity because no defects are observed. For 425°C, the fairly rough Ge
layers and the presence of an irregular black contrast at the
Ge/GaAs interface are observed, which is suspected due to residues
remaining from the pregrowth temperature ramp-up. We will show
later that those residuals are not oxygen clusters. Additionally, these
residues do not seem to impact the quality of the Ge layer grown.
A completely different growth mechanism is then highlighted by
comparing our AP-CVD approach with UHV-CVD ones described
previously in the literature: Ge grows instantaneously and continu-
ously on GaAs without islanding or incubation time.
Figure 1. ͑Color online͒ Nomarski microscope surface picture and ͑004͒
␻ − 2␪ x-ray diffraction scan on Ge layers grown on GaAs, which received
different pregrowth baking treatments: ͑a͒ temperature ramp-up under H2
from 200 to 425°C, ͑b͒ ramp-up under a H2 environment from 200 to 425°C
followed by a H2 bake at 425°C during 10 min, and ͑c͒ ramp-up under H2
from 200 to 425°C followed by an AsH3 bake at 425°C during 10 min.
independent of the pregrowth baking conditions. The surface mor-
phology yielded from the H2 ramp-and-bake ͑Fig. 1b͒ or from the
H2-ramp-AsH3-bake ͑Fig. 1c͒ is rougher than that of a sole H2-
ramp ͑Fig. 1a͒. We also note that an AsH3 bake reduces the Ge
growth rate by a factor of 2 ͑75 nm grown instead of 150 nm for the
same deposition time, as determined by XRD spectra fit and con-
firmed by XSEM͒. We know, indeed, that an AsH3 bake induces an
As-rich GaAs surface, while Ge growth on GaAs is initiated by the
formation of Ge–Ga bonds,17 because a rich surface is then expected
to slow down the growth as observed by Luo et al. and this study.
A straight ramp-up bake from 200°C ͑loading temperature͒ to
the growth temperature was identified to be the best pregrowth
preparation ͑after HCl cleaning͒ yielding the smoothest starting sur-
face. This pregrowth baking condition was chosen for the rest of the
work.
More effort was performed on the Ge/GaAs layers characteriza-
tion in order to determine any eventual O contamination at the
Ge/GaAs interface and eventual Ge–GaAs intermixing. Low energy
͑500 eV͒ SIMS was performed using a Cs beam and a 45° sputter
angle on 7 nm Ge grown at 375 and 425°C, respectively. Figure 4
shows that an O bump at the Ge/GaAs interface is present for the Ge
layer grown at 375°C, whereas it disappears when the growth is
done at 425°C. We then conclude that O will not desorb from the

Journal of The Electrochemical Society, 158 ͑3͒ H203-H207 ͑2011͒
H205
Figure 3. ͑Color online͒ Growth rate as function of the inverse temperature
for a deposition time of 30 min. Both ultrathin ͑7 nm͒ Ge layers grown at
375 and 425°C are continuous on GaAs, 2D grown with high epitaxial
quality.
GaAs surface if the temperature ͑during the ramp-up bake or during
the growth͒ is not high enough. However, 425°C permits to remove
this O interface contamination. The black spots observed by XTEM
on the Ge layer grown at 425°C are then not linked to any O con-
tamination. Moreover, the SIMS profiles from Fig. 4 show that both
Ga and As are detected in the Ge layer as a surface peak, associated
with long tails. In order to deconvolute the real As and Ga signals
from those of SIMS artifacts, such as strong signal overlapping
Figure 5. ͑Color online͒ ͑a͒ Positive and ͑b͒ negative TOF-SIMS done on
7 nm Ge grown on GaAs at 425°C. As and Ga are detected within the first
Ge top surface monolayers. Ga within the Ge layer is also highlighted.
among the main constituents ͑layer and substrate͒, their isotopes,
and various clusters, additional TOF-SIMS measurements were per-
formed.
Figures 5a and 5b present a positive and negative TOF-SIMS
profile performed on 7 nm Ge grown on GaAs at 425°C, respec-
tively. As and Ga surface peaks are observed again. The mass spec-
trum reconstructed between sputter times 0 and 50 s shows that the
ratio between 74GeH and 76GeH is equal to the theoretical isotope
ratio. No 75As signal should be present at mass 75. However, be-
cause the Ge signal is around 10ϫ higher than that of As. The As
signal should have only a very limited impact on the isotope ratio.
The mass spectrum reconstruction was also carried out for the nega-
tive detection mode in between 0 and 50 s. From isotope ratio cal-
culation, we validate that the signals at mass 75 correspond not only
to the contribution of 74GeH but also to As presence within the first
few Ge monolayers at the surface.
Concerning Ga, the identification of Ga presence in Ge is a bit
more complicated because a 69Ga+ ion beam is used for TOF-SIMS
analysis. Based on the evolution ͑between sputter times 0 and 50 s͒
of both Ga isotopes, clearly different from those of the Ge-clusters,
Ga is also likely present within the first Ge monolayers that are close
Figure 4. ͑Color online͒ Low energy SIMS done on 7 nm Ge layers grown
on GaAs at 375 and 425°C. O bump at the Ge/GaAs interface disappeared
when growth temperature was 425°C.

H206
Journal of The Electrochemical Society, 158 ͑3͒ H203-H207 ͑2011͒
Figure 6. n-Type carrier concentration as a function of the PH3 partial pres-
sure in the chamber for in situ n-Ge growth on GaAs at 425°C.
to the surface, as already observed for As. In the Ge layer itself
͑50–150 s͒, Ga isotope signals were compared with measurements
done directly on a Ge bulk wafer ͑without GaAs in order to deter-
mine the Ga measurement background͒. The signal is higher than
the background level ͑Ϸ101 counts for positive TOF-SIMS͒, which
means that Ga is also present in the Ge layer itself. The presence of
Ga in Ge was further confirmed by Hall measurement on the Ge
layer grown on a semi-insulating GaAs sample. Electrical active
p-type concentration as high as ϫ19 cm−3 was measured in the Ge
layer as the result of autodoping from the GaAs substrate. This
strong Ga up-diffusion in Ge from a Ga rich GaAs surface has also
been observed in the literature.18
Ge growth on the GaAs substrate was then achieved with high
epitaxial growth quality ͑no crystalline defect observed by TEM͒
and without islanding formation in the first growth stage. 2D ͑two-
dimensional͒ layers, as thin as 7 nm, were achieved with no incu-
bation time. Additionally, As and Ga segregations from GaAs sur-
face to Ge layer were highlighted by SIMS/TOF-SIMS. Both As and
Ga are present within the first few monolayers that are close to the
Ge surface. Additionally, Ga segregation induces a p-type autodop-
ing within the whole Ge layer, as thin as 7 nm. This p-type autodop-
ing in the first Ge monolayers close to the Ge/GaAs interface is
especially problematic for contacting n-GaAs ͓for n-metal-oxide-
semiconductor field-effect transistor ͑nMOSFETs͔͒ because a p-n
junction is then formed at the Ge/GaAs interface. To tackle this
issue, growth of in-situ n-type Ge on GaAs has been developed.
Figure 7. ͑Color online͒ ͑a͒ I–V characteristics on NiGe/n–GaAs MOS ca-
pacitor source/drain highlighting a successful ohmic contact to n-GaAs. ͑b͒
Resistance value as a function of the CTLM spacing on NiGe/n–GaAs 2 in.
substrates: 0.13 ⍀ cm was extracted for RC.
from 3.7 ϫ 10−3 to 3.7 mTorr. We note that the dopant concentra-
tion is already very high ͑7 ϫ 1018 cm−3͒ even with a PH3 partial
pressure, as low as 3.7 ϫ 10−4 mTorr: P incorporation is very fast
for low Ge growth rates. The highest P concentration obtained
reaches 3 ϫ 1019 cm−3. This value is in agreement with the highest
n-type level obtained for in-situ n-doped Ge growth processes using
GeH4 and PH3 already developed on Si or Ge. We suppose the
n-type concentration to be uniform within the Ge layer ͑such as in
the case of n-Ge growth on Ge substrates͒ except within the first
7–10 nm of Ge close to the Ge/GaAs interface where Ga still might
have segregated ͑such as in the case of undoped Ge growth on
GaAs͒ and might counterdope the P doping.
The highly doped Ge growth process on GaAs giving the elec-
trical active P concentration of 3 ϫ 1019 cm−3 was then further used
to test contacts on n-GaAs pattern substrates. This process using
GeH4 as the Ge precursor and H2 as the carrier gas is totally selec-
tive toward oxide at growth temperature of 425°C.
In situ n-type Ge growth on GaAs.— In-situ n-type Ge growth
on GaAs was developed using the same procedure as detailed above
and by adding PH3. AsH3 as n-type precursor was also tried but the
Ge growth is totally impeded due to the formation of a continuous
As rich surface during the growth.
Growth temperature was kept at 425°C in order to favor oxygen
desorption from the initial GaAs surface, as explained earlier. Ge
was grown for 30 min with different PH3 partial pressure. We mea-
sured a constant growth rate ͑ ϳ 5 nm/min͒ independent of the PH3
partial pressures. However, the in-situ n-Ge layers grown were much
thicker than those of 7 nm undoped Ge layers.
Figure 6 shows the evolution of the Ge electrical active doping
concentration for different PH3 partial pressures. The electrical ac-
tive P concentration increases linearly with PH3 partial pressures

Journal of The Electrochemical Society, 158 ͑3͒ H203-H207 ͑2011͒
H207
Resistivity measurements.— The in-situ n-Ge growth process de-
veloped on GaAs giving the highest P electrically active concentra-
tion ͑3 ϫ 1019 cm−3͒ was used to form ohmic contacts to two types
of pattern n-GaAs structures.
support in the DualLogic project no. 214579. Furthermore, we thank
the IMEC core partners within the IMEC’s Industrial Affiliation Pro-
gram on Logic.
IMEC assisted in meeting the publication costs of this article.
First on a CMOS GaAs capacitor structure, 40 nm n-Ge ͑no
growth rate change has been observed from pattern versus blanket
substrates͒ was grown selectively on source/drain regions, as shown
on the inset of Fig. 7a. 100 nm of Ni was deposited on n-Ge, and
further standard Ni germanidation process was performed.19 I–V
curve, shown in Fig. 7, on those contacts highlight an ohmic behav-
ior, as expected.
Next, in order to extract contact resistance ͑RC͒, CTLM struc-
tures were processed on 2 in. n-GaAs bulk wafers: a SiO2 hard mask
was used for the selective growth of n-Ge. After an identical ger-
manidation process that was used on the CMOS capacitor, Fig. 7b
shows the resistance value as a function of the CTLM spacing for a
diameter of 240 ␮m and spacing ranging from 256 to 8 ␮m. From
this measurement, a RC value of 0.13 ⍀ cm was extracted using the
CTLM theory.20
References
1. Y. Sun, E. W. Kiewra, J. P. de Souza, J. J. Bucchignano, K. E. Fogel, D. K. Sadana,
and G. G. Shahidi, IEEE Electron Device Lett., 30, 5 ͑2009͒.
2. M. Radosavljevic, B. Chu-Kung, S. Corcoran, G. Dewey, M. K. Hudait, J. M.
Fastenau, J. Kavalieros, W. K. Liu, D. Lubyshev, M. Metz, et al., Tech. Dig. - Int.
Electron Devices Meet. ͑2009͒.
3. Y. Q. Wu, M. Xu, R. S. Wang, O. Koybasi, and P. D. Ye, Tech. Dig. - Int. Electron
Devices Meet. ͑2009͒.
4. N. D. Nguyen, G. Wang, G. Brammertz, M. Leys, N. Waldron, G. Winderickx, K.
Lismont, J. Dekoster, R. Loo, M. Meuris, et al., ECS Trans., 33͑6͒, 933 ͑2010͒.
5. Y. C. Chang, C. Merckling, J. Penaud, C. Y. Lu, W.-E. Wang, J. Dekoster, M.
Meuris, M. Caymax, M. Heyns, J. Kwo, et al., Appl. Phys. Lett., 97, 112901
͑2010͒.
6. U. Singisetti, M. A. Wistey, G. J. Burek, A. K. Baraskar, B. J. Thibeault, A. C.
Gossard, M. J. W. Rodwell, B. Shin, E. J. Kim, P. C. McIntyre, et al., IEEE
Electron Device Lett., 30, 11 ͑2009͒.
7. W. E. Spicer, I. Lindau, P. Skeath, and C. Y. Su, J. Vac. Sci. Technol., 17, 666
͑1980͒.
8. N. Braslau, J. Vac. Sci. Technol., 19, 803 ͑1981͒.
Conclusion
9. E. D. Marshall, B. Zhang, L. C. Wang, P. F. Jiao, W. X. Chen, T. Sawada, S. S. Lau,
K. L. Kavanagh, and T. F. Kuech, J. Appl. Phys., 62, 3 ͑1987͒.
10. G.-L. Luo, Z.-Y. Han, C.-H. Chien, C.-H. Ko, C. H. Wann, H.-Y. Lin, Y.-L. Shen,
C.-T. Chung, S.-C. Huang, C.-C. Cheng, et al., J. Electrochem. Soc., 157, H27
͑2010͒.
11. X. Zhang, H. Guo, C.-H. Ko, C. H. Wann, C.-C. Cheng, H.-Y. Lin, H.-C. Chin, X.
Gong, P. S. YaLim, G.-L. Luo, et al., Dig. Tech. Pap. - Symp. VLSI Technol. 233,
233 ͑2010͒.
12. R. S. Bauer and J. C. Mikkelsen, J. Vac. Sci. Technol., 21, 2 ͑1982͒.
13. C.-A. Chang and T.-S. Kuan, J. Vac. Sci. Technol. B, 1, 2 ͑1983͒.
14. A. D. Katnani, P. Chiaradia, H. W. Sang, P. Zurcher, and R. S. Bauer, Phys. Rev. B,
31, 4 ͑1985͒.
15. D. H. Petersen, O. Hansen, T. B. Hansen, P. Boggild, R. Lin, D. Kjaer, P. F.
Nielsen, T. Clarysse, W. Vandervorst, E. Rosseel, et al., J. Vac. Sci. Technol. B, 28,
C1C27 ͑2010͒.
Ohmic contact on n-GaAs was obtained by the use of selective
growth of smooth and continuous n-type Ge followed by NiGe for-
mation on source/drain areas of a metal oxide semiconductor ͑MOS͒
capacitor. The AP-CVD process developed permits Ge to grow on
GaAs without islanding and without incubation time. With the pro-
cedure proposed, Ga auto-doping within the first Ge monolayers
close to the Ge/GaAs interface was highlighted and was compen-
sated by adding PH3 during the growth to achieve an n-Ge/n-GaAs
hetero-junction. The good RC value ͑with respect to GaAs͒ obtained
͑0.13 ⍀ cm͒ with the self-aligned NiGe formation approach pro-
vides a path to VLSI compatible processing for III–V material chan-
nels.
16. S. Osakabe and S. Adachi, Jpn. J. Appl. Phys., 36, 7119 ͑1997͒.
17. Y. Bai, K. E. Lee, C. W. Cheng, M. L. Lee, and E. A. Fitzgerald, J. Appl. Phys.,
104, 084518 ͑2008͒.
18. Y. G. Sadof-ev, Tech. Phys. Lett., 19, 5 ͑1993͒.
Acknowledgment
19. D. P. Brunco, B. De Jaeger, G. Eneman, J. Mitard, G. Hellings, A. Satta, V. Terz-
ieva, L. Souriau, F. E. Leys, G. Pourtios, et al., J. Electrochem. Soc., 155, H552
͑2008͒.
20. J. H. Klootwijk and C. E. Timmering, in Proceedings of the IEEE 2004 Interna-
tional Conference on Microelectronic Test Structures ͑2004͒.
The authors thank Wendy Vanherle, Dirk Rondas, and Gillis
Winderickx for their support on the pregrowth cleaning and SEM
inspection. We acknowledge the European Commission for financial