In situ surface pre-treatment study of GaAs and In0.53Ga 0.47As

Article abstract of DOI:10.1063/1.3702885

The impact of using multiple cycles of trimethyl-aluminum (TMA) prior to Al₂O₃ deposition on the properties of (NH ₄)₂S treated In_0.53Ga_0.47As and GaAs substrates was investigated by in sit

Full text of DOI:10.1063/1.3702885

In situ surface pre-treatment study of GaAs and In0.53Ga0.47As
B. Brennan, D. M. Zhernokletov, H. Dong, C. L. Hinkle, J. Kim, and R. M. Wallace
Citation: Applied Physics Letters 100, 151603 (2012); doi: 10.1063/1.3702885
View online: http://dx.doi.org/10.1063/1.3702885
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/100/15?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Scaled ZrO2 dielectrics for In0.53Ga0.47As gate stacks with low interface trap densities
Appl. Phys. Lett. 104, 182912 (2014); 10.1063/1.4875977
Diffusion of In0.53Ga0.47As elements through hafnium oxide during post deposition annealing
Appl. Phys. Lett. 104, 011601 (2014); 10.1063/1.4860960
Passivation of In0.53Ga0.47As/ZrO2 interfaces by AlN atomic layer deposition process
J. Appl. Phys. 114, 034107 (2013); 10.1063/1.4815934
Studies on atomic layer deposition Al2O3/In0.53Ga0.47As interface formation mechanism based on air-gap
capacitance-voltage method
Appl. Phys. Lett. 101, 122102 (2012); 10.1063/1.4753927
The influences of surface treatment and gas annealing conditions on the inversion behaviors of the atomic-layer-
deposition Al 2 O 3 / n-In 0.53 Ga 0.47 As metal-oxide-semiconductor capacitor
Appl. Phys. Lett. 97, 042903 (2010); 10.1063/1.3467813
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:
129.24.51.181 On: Mon, 01 Dec 2014 02:37:01

APPLIED PHYSICS LETTERS 100, 151603 (2012)
In situ surface pre-treatment study of GaAs and In0.53Ga0.47As
B. Brennan,a) D. M. Zhernokletov, H. Dong, C. L. Hinkle, J. Kim, and R. M. Wallaceb)
Department of Materials Science and Engineering, University of Texas at Dallas, Richardson,
Texas 75080, USA
(Received 23 February 2012; accepted 26 March 2012; published online 10 April 2012)
The impact of using multiple cycles of trimethyl-aluminum (TMA) prior to Al2O3 deposition on
the properties of (NH4)2S treated In0.53Ga0.47As and GaAs substrates was investigated by in situ
x-ray photoelectron spectroscopy. Increasing the number of TMA cycles prior to Al2O3 atomic
layer deposition (ALD) was seen to decrease the concentration of As-As detected at the
oxide-semiconductor interface. The impact of annealing the (NH4)2S treated GaAs surface in situ
C
V
prior to ALD, in various environments, was also investigated.
Physics. [http://dx.doi.org/10.1063/1.3702885]
2012 American Institute of
In order to keep up with scaling requirements of the
semiconductor industry, the implementation of high mobility
channels in metal oxide field effect transistors (MOSFETs)
is expected to be necessary by future generations of semicon-
ductor devices.1 However, one of the major challenges in
realizing this is engineering the interface between the chan-
nel material and the gate oxide.2,3 The principal channel
materials under investigation at the moment for n-type devi-
ces are arsenide based III-V materials, due to their high elec-
tron mobility, which is seen to scale with increasing indium
content.1 The native oxides of these materials have been
shown to have poor electrical performance and so it is neces-
sary to incorporate alternative high permittivity (high-k) ox-
ide materials. Oxygen interaction at this interface has been
predicted and shown to generate defects in the form of As-
As bonding (e.g., dimers) and Ga dangling bonds at this
interface.4,5 As a result, methods to passivate the III-V sur-
face have been utilized in order to reduce the oxygen interac-
tion and prevent defects from being generated, prior and
during high-k oxide growth.6–9 However, high interfacial
defect densities (Dit) are still seen with the majority of these
processes, and as such significant refinement of surface prep-
aration procedures is still needed.10 Recent studies have
shown improved device characteristics by incorporating mul-
tiple cycles of trimethyl-aluminum (TMA) prior to atomic
layer deposition (ALD) (Refs. 11 and 12); however, the justi-
fication for this has not been thoroughly explored.
sistor13 characteristics. The samples were then rinsed in
flowing deionized water for 10 s. The InGaAs and GaAs
samples were then cleaved in two, with one piece of each
material mounted as “companions” onto two separate sample
holders (i.e., each holder with one InGaAs and one GaAs
sample companion set). These sample sets were then loaded
into an ultra-high vacuum (UHV) deposition and analysis
tool, described elsewhere,14 within 7 min of removal from
the (NH4)2S solution.
One sample set would undergo a single pulse of TMA
(0.1 s pulse time in 200 sccm of N2 carrier gas) prior to
Al2O3 deposition, while the other would receive ten pulses
of TMA. XPS scans of the As 2p3/2, Ga 2p3/2, O 1s, In 3d5/2
,
C 1s, As 3d, In 4d, and Ga 3d core levels were taken, after
initially loading the samples to UHV, upon exposure to the
ALD reactor for 30 min at 300 ꢁC (to replicate the conditions
experienced by the samples during ALD), after TMA-only
pre-treatments, and after 10 full cycles (TMA þ H2O) of
Al2O3 deposition.
A third set of GaAs samples underwent the same
(NH4)2S chemical treatment, loaded into the UHV system
and, subsequently, annealed in either UHV (<1 ꢀ 10ꢂ8
mbar), N2 (10 mbar), or forming gas (10 mbar with 10% H2)
at 500 ꢁC for 30 min, again without exposing the samples to
atmosphere. XPS scans were taken before and after anneal-
ing. XPS was carried out using a monochromated Al Ka
x-ray source (hꢀ ¼ 1486.7 eV) which is connected in situ to
the ALD reactor though an UHV transfer tube maintained at
a pressure of <5 ꢀ 10ꢂ10 mbar. XPS peak fitting was carried
out using AANALYZER software,15 with the detailed fitting pro-
cedure described elsewhere.16 The ALD was carried out in a
commercial Picosun ALD reactor, at a pressure of ꢃ9 mbar,
with TMA and H2O pulse times of 0.1 s and high purity N2
purge time of 4s.
This study investigates the effect of varying the number
of trimethyl-aluminum pretreatment cycles prior to ALD of
Al2O3, using in situ x-ray photoelectron spectroscopy (XPS)
to determine what changes are taking place at the GaAs and
In0.53Ga0.47As surfaces. The impact of annealing the surfaces
in different conditions prior to ALD is also explored.
The wafers used in this study were n-type In0.53Ga0.47As
(S doped 4 ꢀ 1017) and GaAs (Si doped 2 ꢀ 1018) and were
initially degreased for 1 min each in acetone, methanol, and
isopropanol followed immediately by an optimized ammo-
nium sulfide treatment in 10% (NH4)2 S for 20 min at room
temperature.7 Such optimized sulphide treatments have
recently been correlated with improved capacitor8 and tran-
The As 2p3/2 spectra from the GaAs and InGaAs sam-
ples, after exposure to the ALD reactor at 300 ꢁC and with ei-
ther one cycle or ten cycles of TMA, are shown in Figure 1.
The (NH4)2S treatment was seen to significantly decrease the
concentration of arsenic oxides present on the surface when
compared to a native oxide sample—with only trace amounts
of As2O5, As2O3, and a sub oxide—labeled As 1þ and attrib-
uted to As2O, detected on all samples. Upon introduction to
the ALD reactor at 300 ꢁC, As2O5 was reduced below the
a)Electronic mail: barry.brennan@utdallas.edu.
b)Electronic mail: rmwallace@utdallas.edu.
C
V
0003-6951/2012/100(15)/151603/4/$30.00
100, 151603-1
2012 American Institute of Physics
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:
129.24.51.181 On: Mon, 01 Dec 2014 02:37:01

151603-2
Brennan et al.
Appl. Phys. Lett. 100, 151603 (2012)
detection limit and the concentration of As2O3 was seen to
significantly decrease. Upon exposure to the TMA pre-
treatments, the As2O3 peak was decreased to within XPS
detection limits on all four samples, confirming previous
results which showed that the first cycle of the ALD deposi-
tion process was able to fully remove the As2O3 from the
GaAs and InGaAs surfaces.17,18 A small residual As 1þ
peak was seen to persist even after the TMA pretreatments,
which could be further evidence of the ligand exchange
mechanism between the trivalent TMA molecule and As2O3
to form Al2O3 being the preferable reaction mechanism.19
The As2O state has also previously been reported to be ther-
mally very stable in comparison to other As oxidation states,
when it was present at the InGaAs surface.16
increased concentration with 10 cycles of TMA cycles and at
a similar level on both GaAs and InGaAs. This component
has been previously detected in synchrotron-based photo-
electron emission studies.20 It is possible that this is an indi-
cation of breaking of an As-As dimer at the surface, creating
an arsenic dangling bond,21 or also possibly As-Al bond
formation.22
The corresponding Ga 2p3/2 and In 3d5/2 spectra from
the GaAs and InGaAs samples, after introduction to the
ALD reactor at 300 ꢁC and after TMA pre-treatments, are
shown in Figure 3. Upon annealing in the reactor, the gal-
lium and indium oxides were seen to increase, due to an oxy-
gen transfer from the reducing arsenic oxides.16 The total
oxygen concentration from the O 1s spectra was seen to
decrease upon annealing, possibly due to the removal of
some of these arsenic oxide species from the samples, as
well as hydroxide related species. Upon TMA pre-treatment,
the signal from the Ga2O3 is decreased to within detection
limits of XPS, as well as a decrease in the concentration of
the Ga 1þ state, without full removal.6 This Ga 1þ (In 1þ)
state is likely due to a combination of Ga2O (In2O) and Ga-S
(In-S) states that persist throughout the treatments here, with
similar levels of sulfur detected on all four samples, based
on the peak area of the S 2p peak (not shown). Sulfur is
expected to preferentially bond with the group three ele-
ments.23 The In 3d5/2 spectra in Figures 3(e) and 3(f) follow
a similar trend to that of the Ga peaks, with no significant
However, a peak attributed to As-As bonding was also
seen to persist after TMA pretreatments and at an elevated
level on the GaAs samples in comparison to the InGaAs sam-
ples, as seen in the ratio of the peak areas of the peaks to that
of the bulk peaks (Figure 2). This difference in concentration
could be due to increased reactivity of the GaAs surface,
reflected in an initially higher level of oxide present after
(NH4)2S treatment in comparison to the InGaAs surface,
based on the peak area of the O 1s spectra (not shown). Previ-
ous studies have proposed that these As-As bonds form as a
result of the oxidation of the surface, principally forming gal-
lium oxides due to Ga having a higher affinity for oxygen,
leaving the interface locally arsenic rich and generating As-
As anti-bonding sites.5 The higher gallium concentration in
GaAs compared to InGaAs could then also explain the
increased As-As concentration seen here on the GaAs sample.
A variation in the concentration of As-As present when
either one cycle or ten cycles of TMA are employed in pre-
treating the surface was also detected. With ten cycles of
TMA, there is an overall decrease in As-As concentration
relative to that observed after one cycle of TMA. However,
there is also a concomitant emergence of a lower binding
energy peak, labeled “Asꢂ” in the spectra, detected at an
FIG. 1. XPS core level spectra for the As 2p3/2 peaks from GaAs [(a) and
(b)] and In0.53Ga0.47As [(c) and (d)] after exposure to the ALD reactor at
300 ꢁC for 30 min, and with 1 cycle [(a) and (c)] or 10 cycles [(b) and (d)] of
trimethyl-aluminum.
FIG. 2. Ratio of the fitted area of the As-As and As- components from the
As 2p3/2 spectra for the GaAs (a) and In0.53Ga0.47As (b) samples at different
stages during sample processing, showing a decrease in the As-As peak area
with increased number of TMA cycles.
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:
129.24.51.181 On: Mon, 01 Dec 2014 02:37:01

151603-3
Brennan et al.
Appl. Phys. Lett. 100, 151603 (2012)
FIG. 4. XPS spectra after (NH4)2S treatment of GaAs and subsequent
anneals in UHV, forming gas, and N2, from the (a) As 2p, (c) Ga 2p, (d) O
1s, and (e) S 2p core levels. The ratios of the fitted As-As and As-features to
the As 2p bulk peak area are shown in (b), and (d) shows the S 2p to Ga 3s
fitted peak area ratio.
FIG. 3. Ga 2p3/2 (a)-(d) and In 3d5/2 (e) and (f) core level spectra from the
GaAs and InGaAs samples after exposure to the ALD reactor at 300 ꢁC for
30 min and after either one cycle [(a), (c), and (e)] or ten cycles [(b), (d), and
(f)] of TMA.
uum and N2 annealed surfaces show no evidence of As-As;
however, As- states are seen to emerge, particularly on the
vacuum annealed surface. This could correlate to the absence
of oxygen on the surface, with the increased level of sulfur
remaining on the N2 annealed surface mitigating the forma-
tion of As- states. On the FGA sample, there is still possible
evidence of As-As states; however, the concentration is close
to the detection limit associated with XPS peak fitting.
Previous reports have suggested that the As-As state
forms as a result of growth of the native oxides on the
(In)GaAs surface.4 Therefore, it may be possible to further
diminish the impact of these states by minimizing the atmos-
pheric exposure time of the semiconductor samples after
(NH4)2S treatment and prior to ALD pre-treatment, as
recently reported.25 The results presented here suggest that,
for GaAs, in situ annealing in N2 at 500 ꢁC prior to ALD
may be able to fully remove As-As states from the surface
and still retain an appreciable S coverage, which can sup-
press the formation of arsenic dangling bonds when com-
pared to a vacuum anneal.
changes taking place with increased number of TMA cycles.
Fitting of the In 3d5/2 peak is complicated by an apparent
asymmetric lineshape for this core level which impacts on
the accuracy of the peak deconvolution.24
The effect of annealing the (NH4)2S treated GaAs sur-
face, in situ in different environments, was also investigated.
Figure 4 shows the (a) As 2p, (c) Ga 2p, (d) O 1s, and (e) S
2p spectra from (NH4)2S treated surfaces with subsequent
anneal in UHV, forming gas, and N2 at 500 ꢁC. All anneals
are seen to remove the As oxides to within detection limits
of XPS, as well as decrease the states associated with the gal-
lium oxide states. In fact, from the O 1s spectra after the vac-
uum and N2 anneal, the concentration of oxygen is below
XPS detection limits. This suggests the residual states seen
in the Ga 2p spectra on these samples are due solely to Ga-S
bonding. This is also reflected in the S 2p spectra, where the
lowest concentration of sulfur is seen on the vacuum
annealed surface, highlighted further by the ratio of the cal-
culated peak area from the fitted S 2p and Ga 3s peaks, seen
in Figure 4(f). Due to the similarity of electronegativites, it
is usually difficult to discern Ga-S from the Ga2O state, and
as such for the (NH4)2S treated and forming gas anneal
(FGA) samples in Figure 4(c), where an oxygen signal is
seen in the O 1s spectra, the peak labeled Ga2O is due to a
combination of both of these states.
In conclusion, we have shown that by utilizing multiple
cycles of TMA prior to Al2O3 deposition, it was possible to
decrease the concentration of As-As states that were detected
at the Al2O3/(In)GaAs interface. This suggests that pre-
treatments with TMA, prior to high-k dielectric deposition,
may be an effective method to improve InGaAs device per-
formance,11,25,26 which are possibly correlated to the concen-
tration of As-related defects.4,5 In situ annealing of the
(NH4)2S treated GaAs surface prior to oxide deposition has
The As-As and As- states in the As 2p spectra in Figure
4(a) are also seen to vary as a function of annealing condi-
tion, with the ratio of their fitted peak areas to that of the
bulk peak showing this more clearly in Figure 4(b). The vac-
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:
129.24.51.181 On: Mon, 01 Dec 2014 02:37:01

151603-4
Brennan et al.
Appl. Phys. Lett. 100, 151603 (2012)
10B. Shin, J. R. Weber, R. P. Long, P. K. Hurley, C. G. Van de Walle, and P.
C. McIntyre, Appl. Phys. Lett. 96, 152908 (2010).
also been shown to decrease the concentration of As-As
states to within detection limits of XPS.
11A. D. Carter, W. J. Mitchell, B. J. Thibeault, J. J. M. Law, and M. J. W.
Rodwell, Appl. Phys. Express 4, 091102 (2011).
The authors thank Dr. P. Hurley (Tyndall) and Professor
E. Vogel (Georgia Tech) for useful discussions. This work
was supported by the Semiconductor Research Corporation
Focus Center on Materials Structures and Devices (MSD),
by the Nanoelectronics Research Initiative (NRI) with the
National Institute for Standards and Technology (NIST),
through the Midwest Institute for Nanoelectronics Discovery
(MIND), and the National Science Foundation (NSF) under
ECCS Award No. 0925844.
12H. D. Trinh, E. Y. Chang, P. W. Wu, Y. Y. Wong, C. T. Chang, Y. F.
Hsieh, C. C. Yu, H. Q. Nguyen, Y. C. Lin, K. L. Lin, and M. K. Hudait,
Appl. Phys. Lett. 97, 042903 (2010).
13J. J. Gu, A. T. Neal, and P. D. Ye, Appl. Phys. Lett. 99, 152113 (2011).
14R. M. Wallace, ECS Trans. 16, 255 (2008).
15A. Herrera-Go´mez, P. Pianetta, D. Marshall, E. Nelson, and W. E. Spicer,
Phys. Rev. B 61, 12988 (2000); AANALYZER is software for XPS peak
deconvolution and is available at http://qro.cinvestav.mx/~aanalyzer/.
16B. Brennan and G. Hughes, J. Appl. Phys. 108, 053516 (2010).
17P. D. Ye, G. D. Wilk, B. Yang, J. Kwo, S. N. G. Chu, S. Nakahara, H.-J.
L. Gossmann, J. P. Mannaerts, M. Hong, K. K. Ng, and J. Bude, Appl.
Phys. Lett. 83, 180 (2003).
18C. L. Hinkle, A. M. Sonnet, E. M. Vogel, S. McDonnell, G. J. Hughes, M.
Milojevic, B. Lee, F. S. Aguirre-Tostado, K. J. Choi, H. C. Kim, J. Kim,
and R. M. Wallace, Appl. Phys. Lett. 92, 071901 (2008).
1J. del Alamo, Nature (London) 479, 317 (2011).
2V. V. Afanas’ev, M. Badylevich, A. Stesmans, G. Brammertz, A. Delabie,
S. Sionke, A. O’Mahony, I. M. Povey, M. E. Pemble, E. O’Connor, P. K.
Hurley, and S. B. Newcomb, Appl. Phys. Lett. 93, 212104 (2008).
3C. L. Hinkle, B. Brennan, S. McDonnell, M. Milojevic, A. M. Sonnet, D.
M. Zhernokletov, R. V. Galatage, E. M. Vogel, and R. M. Wallace, ECS
Trans. 35, 403 (2011).
19B. Brennan, M. Milojevic, H. C. Kim, P. K. Hurley, J. Kim, G. Hughes,
and R. M. Wallace, Electrochem. Solid-State Lett. 12, H205 (2009).
20M. V. Lebedev, D. Ensling, R. Hunger, T. Mayer, and W. Jaegermann,
Appl. Surf. Sci. 229, 226 (2004).
21M. V. Lebedev, E. Mankel, T. Mayer, and W. Jaegermann, J. Phys. Chem. C
114, 21385 (2010).
4W. Wang, C. L. Hinkle, E. M. Vogel, K. Cho, and R. M. Wallace, Micro-
electron. Eng. 88, 1061 (2011).
22B. Brennan, M. Milojevic, R. Contreras-Guerrero, H. C. Kim, M. Lopez-
Lopez, J. Kim, and R. M. Wallace, “Investigation of interfacial oxidation
control using sacrificial metallic Al and La passivation layers on InGaAs,”
J. Vac. Sci. Technol. B (submitted).
5J. Robertson, Appl. Phys. Lett. 94, 152104 (2009).
6C. L. Hinkle, M. Milojevic, B. Brennan, A. M. Sonnet, F. S. Aguirre-Tos-
tado, G. J. Hughes, E. M. Vogel, and R. M. Wallace, Appl. Phys. Lett. 94,
162101 (2009).
23P. Moriarty, B. Murphy, L. Roberts, A. A. Cafolla, L. Koenders, P. Bailey,
and G. Hughes, Phys. Rev. B 50, 14237 (1994).
7B. Brennan, M. Milojevic, C. L. Hinkle, F. S. Aguirre-Tostado, G. Hughes,
and R. M. Wallace, Appl. Surf. Sci. 257, 4082 (2011).
´
E. O’Connor, B. Brennan, V. Djara, K. Cherkaoui, S. Monaghan, S. B.
24C. L. Hinkle, M. Milojevic, E. M. Vogel, and R. M. Wallace, Appl. Phys.
Lett. 95, 151905 (2009).
8
Newcomb, R. Contreras, M. Milojevic, G. Hughes, M. E. Pemble, R. M.
Wallace, and P. K. Hurley, J. Appl. Phys. 109, 024101 (2011).
9A. Alian, G. Brammertz, C. Merckling, A. Firrincieli, W.-E. Wang, H. C.
Lin, M. Caymax, M. Meuris, K. De Meyer, and M. Heyns, Appl. Phys.
Lett. 99, 112114 (2011).
25
´
E. O’Connor, S. Monaghan, K. Cherkaoui, I. M. Povey, and P. K. Hurley,
Appl. Phys. Lett. 99, 212901 (2011).
26B. Shin, J. B. Clemens, M. A. Kelly, A. C. Kummel, and P. C. McIntyre,
Appl. Phys. Lett. 96, 252907 (2010).
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:
129.24.51.181 On: Mon, 01 Dec 2014 02:37:01