The integration of high-k dielectric on two-dimensional crystals by atomic layer deposition

Article abstract of DOI:10.1063/1.3703595

We investigate the integration of Al₂O₃ high-k dielectric on two-dimensional (2D) crystals of boron nitride (BN) and molybdenum disulfide (MoS₂) by atomic layer deposition (ALD). We demonstrate the feasibility of direct AL

Full text of DOI:10.1063/1.3703595

The integration of high-k dielectric on two-dimensional crystals by atomic layer
deposition
Han Liu, Kun Xu, Xujie Zhang, and Peide D. Ye
Citation: Applied Physics Letters 100, 152115 (2012); doi: 10.1063/1.3703595
View online: http://dx.doi.org/10.1063/1.3703595
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
Investigation of atomic layer deposited beryllium oxide material properties for high-k dielectric applications
J. Vac. Sci. Technol. B 32, 03D117 (2014); 10.1116/1.4867436
Comparative band alignment of plasma-enhanced atomic layer deposited high-k dielectrics on gallium nitride
J. Appl. Phys. 112, 053710 (2012); 10.1063/1.4749268
Seeding atomic layer deposition of high-k dielectric on graphene with ultrathin poly(4-vinylphenol) layer for
enhanced device performance and reliability
Appl. Phys. Lett. 101, 033507 (2012); 10.1063/1.4737645
Low temperature growth of high-k Hf–La oxides by remote-plasma atomic layer deposition: Morphology,
stoichiometry, and dielectric properties
J. Vac. Sci. Technol. A 30, 01A147 (2012); 10.1116/1.3665419
Characteristics of high-k Al 2 O 3 dielectric using ozone-based atomic layer deposition for dual-gated graphene
devices
Appl. Phys. Lett. 97, 043107 (2010); 10.1063/1.3467454
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:
128.114.34.22 On: Tue, 25 Nov 2014 05:37:52

APPLIED PHYSICS LETTERS 100, 152115 (2012)
The integration of high-k dielectric on two-dimensional crystals by atomic
layer deposition
Han Liu, Kun Xu, Xujie Zhang, and Peide D. Yea)
School of Electrical and Computer Engineering and Birck Nanotechnology Center, West Lafayette,
Indiana 47906, USA
(Received 7 February 2012; accepted 28 March 2012; published online 13 April 2012)
We investigate the integration of Al2O3 high-k dielectric on two-dimensional (2D) crystals of
boron nitride (BN) and molybdenum disulfide (MoS2) by atomic layer deposition (ALD). We
demonstrate the feasibility of direct ALD growth with trimethylaluminum and water as precursors
on both 2D crystals. Through theoretical and experimental studies, we found that the initial ALD
cycles play the critical role, during which physical adsorption dominates precursor adsorption at
the crystal surface. We model the initial ALD growth stages at the 2D surface by analyzing
Lennard-Jones potentials, which could guide future optimization of the ALD process on 2D
C
V
crystals. 2012 American Institute of Physics. [http://dx.doi.org/10.1063/1.3703595]
The application of atomic layer deposition (ALD) tech-
niques to metal gates and high-k dielectrics in the past dec-
ade has triumphantly extended Moore’s law for the
continued scaling down of silicon based CMOS devices.1 In
addition, the integration of high-k materials on other semi-
conductors, such as Ge, GaAs, InGaAs, GaSb, etc., has also
been comprehensively studied in the pursuit of alternative
channel materials to replace silicon at the 10 nm node and
beyond.2–5 In 2004, graphene, a fascinating material labeled
as a perfect two dimensional (2D) crystal with an electron
mobility approaching 200 000 cm2/Vs at room temperature,
By using the Lennard-Jones potential model, we propose
several ways to optimize ALD growth on these 2D crystals,
which are different from the methods used on 3D bulk
crystals.
BN and MoS2 2D crystals were thinned from bulk crys-
tals by mechanical exfoliation,6 and then transferred to 300-
nm SiO2 covered Si substrates. The samples were then
soaked in acetone for 6 h, followed by the acetone, methanol,
and isopropanol rinse for 30, 15, and 30 s, respectively, to
clean the tape residue on SiO2 substrate. After that, the sam-
ples were loaded into an ASM F-120 ALD system. TMA and
water were used as precursors. Pulse times of 0.8 and 1.2 s
were used for TMA and water, respectively, with a purge
time of 6 s for both. Al2O3 was deposited with a range of
substrate temperatures from 200 ꢀC to 400 ꢀC by 50 ꢀC steps.
Figures 1(a)–1(f) show selected atomic force microscopy
(AFM) images on BN and MoS2 surfaces after 111 ALD
cycles at 200 ꢀC, 300 ꢀC, and 400 ꢀC, with an expected
Al2O3 thickness of ꢁ10 nm. The Al2O3 growth rate on SiO2
substrates did not have significant temperature dependence;
however, its growth on BN and MoS2 flakes was strongly
temperature dependent. We observed a uniform Al2O3 layer
formed at 200 ꢀC on both BN and MoS2 substrates. Our pre-
vious study showed that the leakage current density was
was realized and has shown promise as
a silicon
replacement.6–8 Furthermore, following research has
unveiled other similar materials that exist as layered
2D-materials, including boron nitride (BN), Bi2Te3, Bi2Se3,
MoS2, etc. These other materials can also be isolated to a
single atomic layer via mechanical exfoliation.6,9–11
However, researchers have noticed that the deposition of
high-k dielectrics onto 2D crystals, such as graphene, is not
as easy as deposition onto Ge or III-V bulk materials. A typi-
cal example is the failure of Al2O3 deposition on graphene
basal plane with trimethylaluminum (TMA) and water as
ALD precursors, which is the most reliable ALD process
with a wide process window. This failure has been under-
stood to be caused by the difficulty of forming chemical
bonds on the graphene basal plane due to existing global sp2-
hybridation.12,13 Despite several attempts to integrate high-k
dielectrics onto 2D systems,11,14,15 the integration of high-k
dielectric onto such 2D crystals has not been thoroughly
studied. In this letter, we focus on the growth of ALD Al2O3
on two typical 2D materials: boron nitride, a sister material
of graphene and previously used as a graphene dielectric;16
and MoS2, a promising layer-structured semiconducting ma-
terial with a satisfying band gap. Our results show that the
initial ALD growth on such 2D materials is determined by
physical adsorption of the precursors, and therefore is very
sensitive to growth temperature. As a result, the ALD pro-
cess window for 2D crystals would be consequently reduced.
FIG. 1. AFM images of BN or MoS2 surface after 111 cycles of ALD Al2O3
at 200 ꢀC, 300 ꢀC and 400 ꢀC. All images are taken in a 2 lm by 2 lm region
with a scale bar of 500 nm.
a)Author to whom correspondence should be addressed. Electronic mail:
yep@purdue.edu.
C
V
0003-6951/2012/100(15)/152115/4/$30.00
100, 152115-1
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:
128.114.34.22 On: Tue, 25 Nov 2014 05:37:52

152115-2
Liu et al.
Appl. Phys. Lett. 100, 152115 (2012)
TABLE I. Binding energy (Eads ¼ Ea þ Eb ꢃ Eab) in kcal/mol.
Adsorption
Eads
BN þ H2O!BN-H2O
4.1
12.8
11.1
33.0
BN þ TMA!BN-TMA
MoS2 þ H2O!MoS2-H2O
MoS2 þ TMA!MoS2-TMA
ratio on MoS2 than that on BN at higher temperatures. Such
temperature dependent growth indicates that the growth is
controlled by physical adsorption of the precursors at the
substrate surface and will be further discussed later.
In order to achieve insight in the understanding of the
interactions at the substrate surfaces and hence understand
the initial ALD cycles for 2D crystals, density function
theory (DFT) studies were performed by using the M06-2ꢂ
method19 with basis sets 3-21G(d) for Mo and 6-311Gþ(d,p)
for H, O, C, B, N, and Al. Table I shows the calculated
adsorption energies of the two ALD processes. It can be seen
that the binding energy of TMA on BN is 8.7 kcal/mol
greater than that of H2O on BN and the binding energy of
TMA on MoS2 is 21.9 kcal/mol greater than that of H2O on
MoS2. This implies that TMA is more easily physically
absorbed on both types of crystals. Figure 3 shows the four
adsorption structural models. For the BN system, the O atom
of H2O is adsorbed at the B atom while the Al atom of TMA
is adsorbed at the N atom. The calculated distances of O-B
FIG. 2. Al2O3 coverage estimation from MATLAB analysis by Otsu method.
Error bars show standard deviation of 5-6 measurements taken on the differ-
ent areas of the same sample.
relatively small (ꢁ2 ꢂ 10ꢃ4A/cm2 under 1 V gate bias) for
MoS2 based metal-oxide-semiconductor structure, suggest-
ing that the ALD Al2O3 thin film on MoS2 was of good qual-
ity.17 With elevated growth temperatures, it was obvious that
the Al2O3 film was not uniform on both BN and MoS2 sub-
strates. When the growth temperature was increased to
250 ꢀC, pinhole defects started to appear at the 2D surface.
With further increase of growth temperatures, these pinholes
tended to expand and finally connect with each other, leaving
island like Al2O3 clusters on the 2D basal plane. In contrast
to the growth on basal plane, the growth on edges remains
constant at the range between 200 ꢀC to 400 ꢀC, due to the
existence of dangling bonds at the basal edges.12,13 We ana-
lyzed several AFM images with a MATLAB script to quantify
the Al2O3 coverage and used this as a metric for the ease of
ALD growth, although the coverage percentage may have
evident run-to-run variance due to fluctuations of chamber
pressure, which has a significant impact to the surface
adsorption. The Otsu method was applied to distinguish the
boundary between the regions on BN or MoS2 flakes “with”
or “without” Al2O3 growth.18 As shown in Figure 2, the
Al2O3 coverage was monotonically decreasing with
increased temperature and had a slightly increased coverage
˚
and Al-N are 2.769 and 2.604 A, respectively. These distan-
ces are greater than the corresponding covalence bonds.
Therefore, the adsorption energies include the front molecu-
lar orbital interaction and van der Waals contribution. For
the MoS2 system, TMA is adsorbed at the S atom with the
˚
Al-S distance of 2.653 A. However, H2O is located at the tri-
gonal center formed by three S atoms and the O atom is over
˚
the Mo atom with the O-Mo distance of 4.092 A. Apparently,
the distances of Al-S is shorter than that of O-S, though the
atomic volume of Al is bigger than that of O. Comparing to
the H2O adsorption by adsorption energy, the TMA adsorp-
tion is more stable.
Table II lists the atomic charges, polarizabilities, and
frontier molecular orbital levels of the interaction atoms in
˚
FIG. 3. Binding structure models of H2O and TMA on BN and MoS2. Bond length is in A.
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:
128.114.34.22 On: Tue, 25 Nov 2014 05:37:52

152115-3
Liu et al.
Appl. Phys. Lett. 100, 152115 (2012)
TABLE II. Atomic charges, polarizabilities, and frontier molecular orbital
levels of the interaction atoms in H2O, TMA, BN, and MoS2 in atomic unit.
H2O
TMA
BN
MoS2
ei(LUMO)
ei(HOMO)
Pi
0.1933
ꢃ0.4009
6.140
ꢃ0.0068
ꢃ0.3210
48.996
0.0363
ꢃ0.2972
34.666
ꢃ0.1672
ꢃ0.2823
112.316
O
H
Al
C
B
N
Mo
S
Qa
ꢃ0.582 0.291 1.337 ꢃ0.481 0.991 ꢃ0.924 2.176 ꢃ1.088
H2O, TMA, BN, and MoS2. It can be seen that O atoms with
negative charges would have electrostatic interactions with
the positively charged B and Mo atoms, while Al atoms with
positive charges would be interacting with the negatively
charged N and S atoms. Also, one can see that the polariz-
ability of TMA is much greater than that of H2O, while the
polarizability of MoS2 is much greater than that of BN. This
implies that the interactions of TMA-MoS2 would have the
largest dispersion energy and the interactions of H2O-BN
would have the least. In addition to this van der Waals inter-
action, the frontier molecular orbitals of these model mole-
cules may take an important role in the combination. From
Table II, we see that the gaps between the lowest unoccupied
molecular orbital (LUMO) level and the highest occupied
molecular orbital (HOMO) level are 0.5942 au for H2O,
0.3142 au for TMA, 0.3335 au for BN, and 0.1151 au for
MoS2. Thus, the orbital interactions of H2O with BN and
MoS2 would be less than that of TMA with BN and MoS2,
respectively. This analysis supports the predicted result of
adsorption energies.
FIG. 4. An illustrative Lennard-Jones potential Model for physical adsorp-
tion at 2D crystal surfaces. Strong, intermediate and weak adsorptions are
qualitatively presented here. The depth of the potential well indicates the
adsorption energy, noted as Eads. These curves are not scaled by calculated
values.
mined by the polarizability of the substrate and molecules.
Using BN as an example, nitrogen serves as a positive
charge center while boron serves as a negative charge center;
while graphene has no polarization due to perfect symmetry,
the interaction between the BN substrate and ALD precur-
sors would be stronger than that of graphene and ALD pre-
cursors. That is to say, the depth of the potential well in the
BN system will be larger than that in the graphene counter-
part. The other reason is the growth temperature, which is
correlated with the thermal energy of the precursor mole-
cules. At lower temperatures, the thermal energy is small
that the molecules are trapped in the potential well despite
nitrogen purge, while at higher temperatures where the ther-
mal energy of the molecule is greater than the depth of the
potential well, the excited molecule can escape, thus undoing
ALD cycles.
In classical ALD theories, deposition with precise thick-
ness control is determined by self-limited precursor adsorp-
tion at substrate surfaces and is classified into two types:
physical and chemical adsorption surface. During the initial
ALD cycles on 2D crystals, excepting a few chemically
active materials such as the topological insulators Bi2Te3
and Bi2Se3, which are easily oxidized at growth temperatures
and hence facilitate the formation of chemical bonds for pre-
cursors at the surface,14 chemical adsorption is rarely
observed. This is due to the absence of dangling bonds at
their basal planes. Consequently, physical adsorption is the
dominant adsorption method at the 2D surface. This view is
also supported by the result that such deposition is strongly
temperature dependent. It is interesting to note that ALD
Al2O3 can be deposited on BN at 200 ꢀC, while Al2O3 can
only grow at graphene’s edges, even though BN is extremely
structurally similar to graphene. Such a difference between
Al2O3 deposition on graphene and BN can be explained
using the framework of the Lennard-Jones potential model,
which has been generally used to model the molecular
adsorption on graphene and carbon nanotube surfaces.20–22
As shown in Figure 4, for each ALD pulse-purge cycle, the
pulse action pushes the precursor molecules to the vicinity of
substrate, where the molecule has the lowest potential
energy; while the purge action pushes the molecule away
from the substrate, to the x-axis infinity, where the molecules
encounter an energy barrier. There are two factors that deter-
mine the ultimate molecular state: One is the depth of the
potential well, shown as the adsorption energy and deter-
Therefore, the ALD window for deposition on 2D crys-
tals is different from previous studies on bulk materials. For
bulk substrates, the lower temperature limit of the ALD
growth window is determined by precursor condensation and
incomplete reaction at lower temperatures, and the high tem-
perature limit is determined by precursor decomposition as
well as desorption.23 For 2D substrates, the low temperature
limit still remains similar as it is only related to the precur-
sors, regardless of the substrate material. However, the high
temperature limit, since desorption is much easier at 2D
surfaces, is at a dramatically lower temperature. This creates
a large challenge to dielectric integration for high perform-
ance devices, such as threshold shifts observed in our previ-
ous study on MoS2 top-gated MOSFETs.17 Given our
discussion above, we can clearly see that the first several
ALD cycles is critical, not only for properties related to
interface quality but to allowing further deposition as Al2O3
can provide dangling bonds for the chemical adsorption of
the precursors. One way to optimize the ALD process is to
change the pulse and purge times to better control the surface
adsorption/desorption at the initial stages of deposition.
Alternatively, a seeding layer, such as an ultrathin Al
film24,25 or ALD TMA and O3 process at a low tempera-
ture,26 can also provide a solution for high quality dielectric
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:
128.114.34.22 On: Tue, 25 Nov 2014 05:37:52

152115-4
Liu et al.
Appl. Phys. Lett. 100, 152115 (2012)
4A. Nainani, T. Irisawa, Z. Yuan, Y. Sun, T. Krishnamohan, M. Reason, B.
R. Bennett, J. B. Boos, M. G. Ancona, Y. Nishi, K. C. Saraswat, Tech.
Dig. - Int. Electron Devices Meet. 2010, 138.
growth as played with graphene. Another related question is
the “interface” issue, as it has been generally accepted that
the origin of interface states is attributed to unpassivated
dangling bonds.27 In the absence of dangling bonds at sur-
face of 2D crystals, the definition of “interface states”
between 2D crystals and dielectrics may need to be reconsid-
ered, and such an issue may need further investigation. This
research is still on-going.
5M. Xu, R. Wang, and P. D. Ye, IEEE Electron Devices Lett. 32, 488
(2011).
6K. S. Novoselov, D. Jiang, F. Schedin, T. J. Booth, V. V. Khotkevich, S.
V. Morozov, and A. K. Geim, Proc. Natl. Acad. Sci. U.S.A. 102, 10451
(2005).
7K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, M. I. Katsnelson,
I. V. Grigorieva, S. V. Dubonos, and A. A. Firsov, Nature 438, 197
(2005).
In summary, we have demonstrated direct ALD growth
of Al2O3 with TMA and water as precursors on 2D crystals
of BN and MoS2. We have also performed a DFT study on
surface adsorption of 2D crystals at different geometric sub-
strate locations. Both experimental and theoretical results
show that the ALD growth on 2D crystals is determined by
physical absorption and is enhanced by in-plane polarization
of the substrate. Our results have provided insight into
growth mechanisms and will allow better solutions for high-
quality dielectric integration to be found and provide a big
step forward for electronic or photonic devices based on 2D
crystals in the future.
8Y. Zhang, J. W. Tan, H. L. Stormer, and P. Kim, Nature 438, 201 (2005).
9C. Jin, F. Lin, K. Suenaga, and S. Iijima, Phys. Rev. Lett. 102, 195505
(2009).
10H. Zhang, C. X. Liu, X. L. Qi, X. Dai, Z. Fang, and S. C. Zhang, Nat.
Phys. 5, 438 (2009).
11B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti, and A. Kis, Nat.
Nanotechnol. 6, 147 (2011).
12Y. Xuan, Y. Q. Wu, T. Shen, M. Qi, M. A. Capano, J. A. Cooper, and P.
D. Ye, Appl. Phys. Lett. 92, 013101 (2008).
13X. Wang, S. M. Tabakman, and H. Dai, J. Am. Chem. Soc. 130, 8152
(2008).
14H. Liu and P. D. Ye, Appl. Phys. Lett. 99, 052108 (2011).
15G. Lee, B. Lee, J. Kim, and K. Cho, J. Phys. Chem. C 113, 14225 (2009).
16C. R. Dean, A. F. Young, I. Meric, C. Lee, L. Wang, S. Sorgenfrei, K.
Watanabe, T. Taniguchi, P. Kim, K. L. Shepard, and J. Hone, Nat. Nano-
technol. 5, 722 (2010).
The authors would like to thank G. Q. Xu, H. B. Lu, N.
J. Conrad, and P. Kim for their valuable discussions.
17H. Liu and P. D. Ye, IEEE Electron Device Lett. 33, 546 (2012).
18H. F. Ng, Pattern Recogn. Lett. 27, 1644 (2006).
19Y. Zhao, N. E. Schultz, and D. G. Truhlar, J. Chem. Theory Comput. 2,
364 (2006).
1K. Mistry, C. Allen, C. Auth, B. Beattie, D. Bergstrom, M. Bost, M.
Brazier, M. Buehler, A. Cappellani, R. Chau, C.-H. Choi, G. Ding, K.
Fischer, T. Ghani, R. Grover, W. Han, D. Hanken, M. Hattendorf, J. He, J.
Hicks, R. Heussner, D. Ingerly, P. Jain, R. James, L. Jong, S. Joshi, C.
Kenyon, K. Kuhn, K. Lee, H. Liu, J. Maiz, B. McIntyre, P. Moon, J.
Neirynck, S. Pae, C. Parker, D. Parsons, C. Prasad, L. Pipes, M. Prince, P.
Ranade, T. Reynolds, J. Sandford, L. Shifren, J. Sebastian, J. Seiple, D.
Simon, S. Sivakumar, P. Smith, C. Thomas, T. Troeger, P. Vandervoorn,
S. Williams, and K. Zawadzki, Tech. Dig. - Int. Electron Devices Meet.
2007, 247.
20T. Heine, L. Zhechkov, and G. Seifert, Phys. Chem. Chem. Phys. 6, 980
(2004).
21S. Patchkovski, J. S. Tse, S. N. Yurchenko, L. Zhechkov, T. Heine, and
G. Seifert, Proc. Natl. Acad. Sci. U.S.A. 102, 10439 (2005).
22A. Ambrosetti and P. L. Silvestrelli, J. Phys. Chem. C 115, 3695 (2011).
23S. M. George, Chem. Rev. 110, 111 (2010).
24S. Kim, J. Nah, I. Jo, D. Shahrjerdi, L. Colombo, Z. Yao, E. Tutuc, and S.
K. Banerjee, Appl. Phys. Lett. 94, 062107 (2009).
25T. Shen, J. J. Gu, M. Xu, Y. Q. Wu, M. L. Bolen, M. A. Capano, L. W.
Engel, and P. D. Ye, Appl. Phys. Lett. 95, 172105 (2009).
26B. Lee, S.-Y. Park, H.-C. Kim, K. J. Cho, E. M. Vogel, M. J. Kim, R. M.
Wallace, and J. Kim, Appl. Phys. Lett. 92, 203102 (2008).
27D. K. Schroder, Semiconductor Material and Device Charaterization, 3rd
ed. (John Wiley and Sons, Inc., 2006), pp 342.
2C. O. Chui, S. Ramanathan, B. B. Triplett, P. C. McIntyre, and K. C.
Saraswat, IEEE Electron Devices Lett. 23, 473 (2008).
3Y. Xuan, Y. Q. Wu, and P. D. Ye, IEEE Electron Devices Lett. 29, 294
(2008).
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:
128.114.34.22 On: Tue, 25 Nov 2014 05:37:52