Appl. Phys. Lett. 102, 072105 (2013); https://doi.org/10.1063/1.4793483 102, 072105
© 2013 American Institute of Physics.
Interface charge engineering at atomic layerdeposited dielectric/III-nitride interfacesCite as: Appl. Phys. Lett. 102, 072105 (2013); https://doi.org/10.1063/1.4793483Submitted: 28 November 2012 . Accepted: 12 February 2013 . Published Online: 22 February 2013
Ting-Hsiang Hung, Sriram Krishnamoorthy, Michele Esposto, Digbijoy Neelim Nath, Pil Sung Park, andSiddharth Rajan
ARTICLES YOU MAY BE INTERESTED IN
Electrical properties of atomic layer deposited aluminum oxide on gallium nitrideApplied Physics Letters 99, 133503 (2011); https://doi.org/10.1063/1.3645616
Presence and origin of interface charges at atomic-layer deposited Al2O3/III-nitride
heterojunctionsApplied Physics Letters 99, 193504 (2011); https://doi.org/10.1063/1.3658450
Two-dimensional electron gases induced by spontaneous and piezoelectric polarizationcharges in N- and Ga-face AlGaN/GaN heterostructuresJournal of Applied Physics 85, 3222 (1999); https://doi.org/10.1063/1.369664
Interface charge engineering at atomic layer deposited dielectric/III-nitrideinterfaces
Ting-Hsiang Hung,a) Sriram Krishnamoorthy, Michele Esposto, Digbijoy Neelim Nath,Pil Sung Park, and Siddharth RajanDepartment of Electrical and Computer Engineering The Ohio State University Columbus, Ohio 43210, USA
(Received 28 November 2012; accepted 12 February 2013; published online 22 February 2013)
Interface charges at atomic layer deposited Al2O3/III-nitride interfaces were investigated for
III-nitride layers of different polarity. A large positive sheet charge density is induced at the
Al2O3/III-nitride interface on all the orientations of GaN and Ga-polar AlGaN, and this sheet
charge can be significantly altered using post-metallization anneals. It is proposed that the charges
are caused by interfacial defects that can be passivated and neutralized through a H2 based anneal.
Tailoring of the interface charge density described here can be used to improve critical device
characteristics such as gate leakage and electron transport, and for lateral electrostatic engineering.VC 2013 American Institute of Physics. [http://dx.doi.org/10.1063/1.4793483]
III-nitride-based high electron mobility transistors
(HEMTs) are outstanding systems for applications of high
power switching1–6 and high frequency devices7 due to the
combination of large band gap and good electronic transport
properties. Dielectric layers inserted between the gate and
the semiconductor can suppress gate leakage power switch-
ing devices and in scaled high frequency devices. Metal-in-
sulator-semiconductor HEMTs (MISHEMTs)8–19 structure
can efficiently suppress gate leakage in transistors scaled to
achieve higher frequency operation and could also be used
for applications in GaN power switching.1,3
The interface trap density and related dispersion in atomic
layer deposited (ALD) dielectric/III-nitride interfaces are sig-
nificantly lower than in III-As semiconductors,20,21 but recent
work has shown that a high density of fixed charges of the
order of 1 lC/cm2 is induced at ALD-grown Al2O3/GaN and
Al2O3/AlN structures,22–24 an effect that is unique among
previously investigated semiconductor material systems.
While this charge is not modulated by gate voltage and does
not lead to hysteresis under normal device operation, it does
significantly modify the electrostatics in the system. The high
interfacial fixed charges may degrade the mobility of the
2-dimensional electron gas (2DEG)25 and induce electric
fields in the oxide leading to increased tunneling-related
leakage currents. This fixed-charge-induced electric field also
causes leakage through gate dielectrics in N-polar III-nitride
HEMTs.26 Following previous work on post-metallization
annealing (PMA) to passivate interface states in silicon
technology,27,28 in this work, we investigated the effect of
post-metallization anneal on Al2O3/GaN interface charge
density for different polarities of GaN.
Ga-polar GaN samples were grown using a Veeco
GEN 930 plasma molecular beam epitaxy system
(PAMBE) on Fe-doped semi-insulating GaN/sapphire
templates.29 The N-polar samples were grown by PAMBE
on n-doped free standing GaN template.29 The epitaxial
layer consisted of 200 nm unintentionally doped GaN
followed by 100 nm silicon-doped GaN ([Si]� 1� 1018
cm3]. As-received bulk m-plane GaN samples ([Si]
�6� 1016 cm3) were used.30 A 29 nm Al0.3Ga0.7N/1 nm
AlN/GaN HEMT31 sample on Si substrate with 2DEG sheet
carrier density of 1.1� 1013 cm2 (as-received) was used in
the study. The Al0.3Ga0.7N layer was then recessed to 9 nm.
Three Al2O3 layers of nominal thickness 6 nm, 12 nm, and
18 nm were deposited by ALD on each Ga-polar, N-polar
GaN, and m-plane GaN sample at 300 �C, using trimethyla-
luminum (TMA) and H2O as precursors. For the case of
Al2O3 on AlGaN, 17 nm Al2O3 was deposited by ALD. Six
H2O pulses (0.1 s for each) were used at the beginning of
deposition and then followed by TMA precursor. The dif-
ference between estimated and measured Al2O3 thickness
was less than 0.3 nm as confirmed by ellipsometry. Post-
deposition anneal (PDA) consisted of 700 �C in forming gas
for 1 min. Gate patterns were defined by optical contact
aligner, and a Ni/Au/Ni (30/200/30 nm) stack was deposited
using an e-beam evaporator. Large contacts for ohmic con-
tact were patterned using contact lithography, and buffered
oxide etch (BOE) 10:1 was used to locally remove the
oxide layer for large features in order to get ohmic contacts.
After gate metallization, PMA was carried out in the form-
ing gas (5% H2, 95% N2) in a rapid thermal annealing
system at temperatures varying from 400 �C to 550 �C.
A quantitative analysis of the Ni/Al2O3/GaN capacitors
was done to determine conduction band discontinuity, elec-
tric field in the dielectric layer, and interface fixed charge
for each polarity. C-V measurements were performed using
an Agilent B1500 semiconductor device analyzer equipped
with medium power source/monitor units (MPSMUs) and
multi frequency capacitance measurement unit (MFCMU).
Figure 1(a) shows the C-V measurement result for three
different thicknesses of Al2O3 on Ga-polar GaN before
PMA treatment. The flat band voltage (VFB) was derived
from capacitance voltage profiles and was found to be
�2.8 V,�1.7 V, and �0.5 V for 18 nm, 12 nm, and 6 nm
Al2O3/GaN, respectively. The conduction band diagram of
Ni/Al2O3/GaN MIS structure is shown in the inset of
Figure 1(a).
a)Author to whom correspondence should be addressed. Electronic mail:
[email protected]. Tel.: þ1-614-688-8458.
0003-6951/2013/102(7)/072105/4/$30.00 VC 2013 American Institute of Physics102, 072105-1
APPLIED PHYSICS LETTERS 102, 072105 (2013)
A simple analytical expression relating the applied flat
band gate voltage to the interfacial parameters can be derived
as
qVgi ¼ ðub � DEc � usÞ � qFoxtoxi; (1)
where ub is the Ni/Al2O3 conduction band barrier height, Fox
is the electric field in the oxide, us is the energy difference
between conduction band and Fermi level in GaN, DEc is the
conduction band offset between Al2O3 and GaN, and Vgi is
the flat band gate bias for Al2O3 thickness toxi.
From C-V measurements, the VFB extracted was found
to vary linearly with oxide thickness, confirming the validity
of Eq. (1) (Figure 1(b)) and indicating that the charge was
distributed as a sheet, rather than throughout the oxide. Since
the flat band voltage dependence on Al2O3 thickness is lin-
ear, the electric field in the oxide can be assumed to be inde-
pendent of the oxide thickness, indicating the absence of a
significant density of bulk oxide charges. From Eq. (1), the
slope of this curve gives the oxide field, while the intercept
can be used to determine the conduction band offset. us is
estimated as 18 meV using doping density in GaN, and the
Ni/Al2O3 conduction band barrier height ub was estimated to
be 3 eV from internal photoemission measurements.32 Using
the slope and intercept of the fitted line, Fox was determined
to be 1.92 MV/cm before PMA. Using Fox at the Al2O3/GaN
interface, we estimate a net sheet charge density of
þ9.5� 1012 cm�2. The total interface fixed charge (rfix),
derived by adding the interface net charge (rnet) and GaN
spontaneous polarization charge (rsp_GaN¼ 1.81� 1013
cm�2) in the Ga-polar case, is 2.7� 1013cm�2. Figure 1(b)
shows the flat band voltage (VFB) vs. oxide thickness curves
for 5-min post-metallization anneals at 400 �C, 450 �C, and
500 �C. The slope of the curve in Figure 1(b) decreases as
the PMA temperature is increased, indicating a reduction in
the electric field and the net interface charge density, as
shown in Table I. After 500 �C PMA, the interface net posi-
tive charges were reduced to 1.1� 1012 cm�2, almost 95%
TABLE I. Electric field in the oxide and interface charge density for different polarities of GaN after PMA.
PMA conditionGa-polar No PMA 400 �C 5 min 450 �C 5 min 500 �C 5 min
Fox (MV/cm) 1.92 0.83 0.45 0.22
rnet (cm�2) þ9.5� 1012 þ4.1� 1012 þ2.2� 1012 þ1.1� 1012
PMA conditionN-polar 400 �C 5 min 450 �C 5 min 500 �C 5 min 550 �C 5 min
Fox (MV/cm) 1.67 1.08 0.51 0.16
rnet (cm�2) þ9.2� 1012 þ5.9� 1012 þ2.8� 1012 þ9.0� 1011
PMA conditionnon-polar No PMA 400 �C 5 min 450 �C 5 min
Fox (MV/cm) 1.42 0.46 �0
rnet (cm�2) þ7.8� 1012 þ2.5� 1012 �0
FIG. 1. (a) C-V characteristics for Al2O3/
Ga-polar GaN before PMA. (b) Flat-band
voltage versus oxide thickness for the
Ga-polar GaN for different PMA
temperatures.
FIG. 2. (a) C-V profile for Al2O3/N-polar
GaN after 400 �C PMA. (b) Flat-band
voltage versus oxide thickness for Al2O3/
N-polar GaN at different temperature
anneals.
072105-2 Hung et al. Appl. Phys. Lett. 102, 072105 (2013)
lower than the pre-anneal charge density, and the total rfix
was reduced to 1.9� 1013 cm�2. The intercept of the curves
was used to determine the conduction band offset, which
was found to be 2.1 eV before PMA. After PMA, DEc was
found to approximately the same (2.3 eV), showing that the
annealing does not change the band line-ups significantly.
While the N-polar samples have opposite surface
termination and spontaneous polarization to the Ga-polar
samples, our results indicated that the electrostatics of the
ALD/GaN interface is similar to the Ga-polar surface. After
ALD deposition, a net positive charge density was found at
the N-polar GaN surface. Figure 2(a) shows the C-V
measurement results of the Al2O3/N-polar GaN after differ-
ent PMA treatments. We extracted flat band voltage from
C-V results, and the relation between VFB and oxide thick-
ness for N-polar case is shown in Figure 2(b). The conduc-
tion band offset for the N-polar case was found to be
2.1 eV. The electric field in the oxide was found to reduce
from 1.7 MV/cm to 0.2 MV/cm, indicating that the inter-
face net charges were reduced to 9.0� 1011 cm�2 as the
PMA temperature was increased from 400 �C to 550 �C.
This indicates that, in N-polar case, positive GaN spontane-
ous polarization charges were neutralized after higher tem-
perature of PMA. The oxide field Fox and net interface
charge density rnet are listed in Table I.
To determine the effects of polarization on the interfa-
cial properties, a series of PMA experiments were done on
non-polar surfaces using Al2O3/m-plane GaN structures.
Figures 3(a) and 3(b) show the C-V measurement and VFB
verses oxide thickness plot for this structure. Before PMA,
positive interface fixed charges (7.8� 1012 cm�2) were
found to exist at the Al2O3/m-plane GaN interface, causing
an electric field in the oxide under flat band conditions in the
GaN. The conduction band offset as determined from the
measurements was 2.1 eV. PMA was found to remove the
interface charges completely, similar to the Ga- and N-polar
samples.
The case of Al2O3 on AlGaN is of special technologi-
cal interest for AlGaN/GaN HEMTs. Figure 4 shows C-V
profile before and after PMA treatment at different temper-
atures for Al2O3/AlGaN structures. We find that there is a
higher positive charge at the Al2O3/AlGaN interface than at
the Al2O3/GaN interface, as predicted from previous
FIG. 3. (a) C-V profile for Al2O3/m-plane
GaN without PMA. (b) Flat-band voltage
versus oxide thickness for the m-plane GaN
for different PMA temperatures.
FIG. 4. C-V profile for Al2O3/Al0.3Ga0.7N/AlN/ GaN before and after
400 �C, 450 �C, 500 �C PMA.
FIG. 5. Gate leakage current in Al2O3/Ga-polar GaN after different PMA temperatures showing suppression of leakage with increased PMA.
072105-3 Hung et al. Appl. Phys. Lett. 102, 072105 (2013)
work.24 The flat band voltage shifts in the positive direction
after higher temperature PMA, similar to the case of Al2O3
on GaN, showing that the positive charge density has been
reduced.
Based on the energy band diagrams, the bare n-type
GaN surface and the AlGaN surface without ALD dielectric
have net negative surface charges (equal to the positive
depletion layer charge). Our results show that after ALD
deposition, net positive charges are induced at the ALD/GaN
interface for all polarities, showing that rnet changes from
negative to positive after depositing ALD Al2O3. We also
find that PMA treatment can efficiently reduce the interface
positive net charges in Al2O3 on Ga-polar, N-polar, m-plane
GaN and AlGaN. The origin of this net positive charge is not
known—it could be attributed to either the increase of posi-
tive interface fixed charges or a reduction in the number of
previously existent negative surface states after Al2O3 depo-
sition. The polarity-independent result indicates that the
source of the Al2O3/III-nitride interface fixed charge does
not originate from intrinsic polarization-charge inversion22
but rather may be due to defects at the surface.33 We
hypothesize that the gate metal assists the absorption of
forming gas at the Al2O3/III-nitride interface and compen-
sate the spontaneous polarization charge. The absorption of
hydrogen atoms becomes more efficient when we increased
the PMA temperature, and the interface net charge is reduced
substantially.
The use of PMA to control the Al2O3/III-nitride in inter-
face net charges is important for design of AlGaN/GaN
HEMTs for various applications. The positive interface
charge also affects the dielectric leakage characteristics sig-
nificantly. In Figure 5, the gate leakage current for the
Al2O3/Ga-polar GaN structures discussed earlier in this letter
is shown. As expected, the decrease in the electric field in
oxide reduces the leakage currents by suppressing field-
assisted tunneling mechanisms (Fowler-Nordheim tunnel-
ing). The high interface fixed can act as remote scattering
centers and decrease electron mobility especially when gate
oxide thickness is scaled,25 and the channel is close (several
nm) from the fixed charges. The results described here pro-
vide a method to reduce the interface charge density, thereby
eliminating mobility degradation due to remote ionized im-
purity scattering. Perhaps unique to the III-nitride system,
the ability to tune high dielectric/semiconductor charge
densities (of the order of 1013) can provide a new way to
engineer lateral band structures and charge density in semi-
conductor devices. For example, selective patterning of
metals can be used to create different band profiles in differ-
ent regions of a device, giving unprecedented flexibility in
lateral band structure design.
In conclusion, that post metallization anneal in a hydro-
gen containing ambient can effectively reduce the Al2O3/III-
nitride interface net charges in Ga-polar, N-polar, non-polar
GaN, and AlGaN cases. The investigation gives further
insight of the origin of the interface fixed charges at oxide/
III-nitride interface and provides a method to engineer inter-
face charge density. The suppression of the interface charges
using PMA is shown to reduce the gate leakage in reverse
bias and could be critical for mitigating remote ionized im-
purity scattering in GaN-based MISFETs.
This work was funded by the ONR DEFINE MURI
(N00014-10-1-0937) program (Program manager: Dr. Daniel S.
Green).
1W. Huang, T. P. Chou, Y. Niiyama, T. Nomura, and S. Yoshida, IEEE
Electron Device Lett. 30(10), 1018–1020 (2009).2Y. Uemoto, M. Hikita, H. Ueno, H. Matsuo, H. Ishida, M. Yanagihara, T.
Ueda, T. Tanaka, and D. Ueda, IEEE Trans. Electron Devices 54(12),
3393–3399 (2007).3M. Kanamura, T. Ohki, T. Kikkawa, K. Imanishi, T. Imada, A. Yamada,
and N. Hara, IEEE Electron Device Lett. 31(3), 189–191 (2010).4Y. Cai, Y. Zhou, K. M. Lau, and K. J. Chen, IEEE Trans. Electron
Devices 53(9), 2207–2215 (2006).5R. Chu, A. Corrion, M. Chen, R. Li, D. Wong, D. Zehnder, B. Hughes,
and K. Boutros, IEEE Electron Device Lett. 32(5), 632–634 (2011).6T. Palacios, A. Chakraborty, S. Rajan, C. Poblenz, S. Keller, S. P.
DenBaars, J. S. Speck, and U. K. Mishra, IEEE Electron Device Lett.
26(11), 781–783 (2005).7Y.-F. Wu, B. P. Keller, S. Keller, D. Kapolnek, P. Kozodoy, S. P.
Denbaars, and U. K. Mishra, Appl. Phys. Lett. 69, 1438 (1996).8P. D. Ye, B. Yang, K. K. Ng, J. Bude, G. D. Wilk, S. Halder, and J. C. M.
Hwang, Appl. Phys. Lett. 86, 063501 (2005).9Y. Q. Wu, P. D. Ye, G. D. Wilk, and B. Yang, Mater. Sci. Eng., B 135,
282 (2006).10D. Gregusova, R. Stoklas, K. Cico, T. Lalinsky, and P. Kordos, Semicond.
Sci. Technol. 22, 947 (2007).11P. Kordos, D. Gregusova, R. Stoklas, S. Gazi, and J. Novak, Solid-State
Electron. 52, 973 (2008).12Y.-Z. Yue, Y. Hao, and J.-C. Zhang, in IEEE Proceedings of the
Compound Semiconductor Integrated Circuit Symposium CSIC, Monterey,
CA, USA, 12-15 October 2008.13Q. Feng, Y. Hao, and Y.-Z. Yue, Semicond. Sci. Technol. 24, 025030 (2009).14O. I. Saadat, J. W. Chung, E. L. Piner, and T. Palacios, IEEE Electron
Device Lett. 30(12), 1254 (2009).15Z. H. Liu, G. I. Ng, S. Arulkumaran, Y. K. T. Maung, K. L. Teo, S. C.
Foo, and V. Sahmuganathan, Appl. Phys. Lett. 95, 223501 (2009).16X. Xin, J. Shi, L. Liu, J. Edwards, K. Swaminathan, M. Pabisz, M.
Murphy, L. F. Eastman, and M. Pophristic, IEEE Electron Device Lett.
30(10), 1027 (2009).17J. Shi, L. F. Eastman, X. Xin, and M. Pophristic, Appl. Phys. Lett. 95,
042103 (2009).18J. Shi and L. F. Eastman, IEEE Electron Device Lett. 32(3), 312 (2011).19Z. H. Liu, G. I. Ng, S. Arulkumaran, Y. K. T. Maung, and H. Zhou, Appl.
Phys. Lett. 98, 163501 (2011).20R. D. Long and P. C. McIntyre, Materials 5, 1297 (2012).21S. Stemmer, V. Chobpattana, and S. Rajan, Appl. Phys. Lett. 100, 233510
(2012).22M. Esposto, S. Krishnamoorthy, D. Nath, S. Bajaj, T.-H. Hung and
S. Rajan, Appl. Phys. Lett. 99, 133503 (2011).23J. Son, V. Chobpattana, B. M. McSkimming, and S. Stemmer, Appl. Phys.
Lett. 101, 102905 (2012).24S. Ganguly, J. Verma, G. Li, T. Zimmermann, H. Xing, and D. Jena, Appl.
Phys. Lett. 99, 193504 (2011).25T.-H. Hung, M. Esposto, and S. Rajan, Appl. Phys. Lett. 99, 162104 (2011).26S. Rajan, A. Chini, M. Wong, J. S. Speck, and U. K. Mishra, J. Appl.
Phys. 102, 044501 (2007).27P. Lundgren, M. O. Anderson, and K. R. Farmer, J. Appl. Phys. 74, 4780
(1993).28I. C. Kizilyalli, J. W. Lyding, and K. Hess, IEEE Electron Device Lett.
18(3), 81 (1997).29Saint-Gobain Crystals-Photonic Materials, 33 Powers St., Milford, NH
03055, USA.30Kyma Technologies, 8829 Midway West Rd., Raleigh, NC 27617, USA.31NTT Advanced Technology Corporation, Shinjuku Mitsui Bldg. 2-1-1,
Nishi-shinjuku, Shinjuku Ku, Tokyo, 163-0431, Japan.32Y.-C. Yeo, T.-J. King, and C. Hu, J. Appl. Phys. 92, 7266 (2002).33G. J. Burek, Y. Hwang, A. D. Carter, V. Chobpattana, J. J. M. Law, W. J.
Mitchell, B. Thibeault, S. Stemmer, and M. J. W. Rodwell, J. Vac. Sci.
Technol. B 29, 040603 (2011).
072105-4 Hung et al. Appl. Phys. Lett. 102, 072105 (2013)