Impact of postdeposition annealing upon film properties of atomic layer
deposition-grown Al2O3 on GaN
, Andre Wachowiak, and Paul Matthias Jordan
Annett Winzer,a) Nadine Szabo
NaMLab gGmbH, N€
othnitzer Str. 64, 01187 Dresden, Germany
Johannes Heitmann
Institute of Applied Physics (IAP), TU Bergakademie Freiberg, Leipziger Str. 23, 09599 Freiberg, Germany
Thomas Mikolajick
NaMLab gGmbH, N€
othnitzer Str. 64, 01187 Dresden, Germany and Institute of Semiconductor
and Microsystems Technology (IHM), TU Dresden, N€
othnitzer Str. 64, 01162 Dresden, Germany
(Received 30 September 2014; accepted 12 December 2014; published 29 December 2014)
Atomic layer deposition-grown Al2O3 thin films are grown on n-type GaN and annealed at 300 or
500 C in various atmospheres. Metal–insulator–semiconductor capacitors (MISCAPs) are used as
simplified test structures for AlGaN/GaN heterostructure field effect transistors with an Al2O3 gate
dielectric. Electrical characterization of the unannealed MISCAPs reveals a low leakage current
density of 1.4 109 A/cm2 at 2 MV/cm. Annealing at 500 C in N2 or a forming gas results in
a degradation of this leakage level by more than one order of magnitude, whereas the leakage current of the Al2O3 films annealed at 500 C in O2 is increased to 5.2 109 A/cm2 at 2 MV/cm.
The photoassisted capacitance–voltage technique, the conductance method, and border trap analysis are used to study the influence of the annealing ambient atmosphere upon the Al2O3/GaN interface. For all atmospheres, thermal treatments at 500 C marginally affects the border oxide trap
density, but the forming gas anneal at 500 C passivates the interface traps most efficiently. While
the O2 thermal treatment reduces the interface trap density in the Al2O3/GaN system, the N2 anneal
creates interface trap states, indicating the formation of an oxygen deficient defect level at the
C 2014 American Vacuum Society.
Al2O3/GaN interface during N2 annealing. V
[http://dx.doi.org/10.1116/1.4904968]
I. INTRODUCTION
Gallium nitride-based high electron mobility heterostructure field effect transistors (HFETs) are promising device
concepts for high-power and high-frequency applications.
To improve their electrical characteristics such as gate leakage current, breakdown voltage, and current collapse, the
Schottky metal gate can be replaced by a metal–insulator–
semiconductor (MIS) contact with a high-j (high dielectric
constant) gate dielectric. Here, a convenient band alignment
of the gate dielectric toward the GaN is one basic criterion to
form an efficient barrier for both electrons and holes and to
achieve low gate leakage currents.1 In the past, suitable
high-j dielectric materials have been developed and their
fabrication has been transferred from Si-based technology to
that of GaN.2 Compared to other high-j gate dielectric materials such as ZrO2 (Ref. 2) and HfO2, Al2O3 offers a larger
bandgap with high conduction and valence band offsets to
the GaN (Ref. 1) and, in addition, its dielectric constant of
9 allows for good gate coupling.3 The atomic layer deposition (ALD) of Al2O3 is a well-established and widely studied
deposition technique, and its sequential self-terminating
gas–solid reactions on the substrate surface result in a precise film thickness control and the deposition of highly conformal as well as homogeneous thin films.
In this work, we focus on the integration of ALD-grown
Al2O3 in GaN MIS-capacitors, which represents a simplified
a)
Electronic mail: [email protected]
01A106-1 J. Vac. Sci. Technol. B 33(1), Jan/Feb 2015
test structure for future HFETs with MIS-gate contacts. To
achieve electrical properties adapted to these requirements,
we investigate the influence of a postdeposition annealing
treatment at different temperatures and in various atmospheres upon the leakage current and interface trap density. In
terms of the interface characterization, we apply photoassisted C–V measurements,4,5 the conductance method6 and
border trap analysis7 to study the distribution of the diverse
trap states at the Al2O3/GaN interface.
II. EXPERIMENT
For device fabrication, a 1 lm-thick n-type GaN epitaxial
layer (Si donor density, nD, 1019 cm3) was grown on cplane (0 0 0 1) sapphire by metal organic vapor phase epitaxy. Details of the GaN growth on sapphire are described
elsewhere.8 After a wet chemical clean with 1% HF for 30 s,
samples were loaded into the ALD reactor within 15 min.
The Al2O3 thin films with thicknesses of 9, 14, and 20 nm
were grown by thermal ALD with trimethylaluminium
(TMA) as the metal precursor and H2O as the oxidant. The
ALD processes were run at 300 C substrate temperature and
at a chamber pressure of 0.04–0.07 mbar. The ALD
sequence was initiated with the TMA injection and consisted
of a sequence comprising a 20 ms TMA pulse/5 s N2 purge/
20 ms H2O pulse/6 s N2 purge. The Al2O3 film thickness was
determined by x-ray reflectometry (XRR). Postdeposition
annealing was performed for 10 min at 300 or 500 C in either N2, O2, or a forming gas (10% H2/N2) at an ambient
2166-2746/2015/33(1)/01A106/6/$30.00
C 2014 American Vacuum Society
V
01A106-1
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 87.172.47.218 On: Mon, 05 Jan 2015 13:13:33
01A106-2 Winzer et al.: Impact of postdeposition annealing upon film properties
atmosphere (1 atm). For subsequent electrical characterization,
the Al2O3 stacks received the deposition of nonalloyed ohmic
contacts and top metal electrodes (Fig. 1), wherein, after a
local Al2O3 removal with 1% HF, Ti (10 nm)/Al (200 nm)
ohmic contacts were evaporated through a shadow mask. The
metal–insulator–semiconductor capacitors were completed
with the deposition of e-beam-evaporated Pt (50 nm) top electrodes with 100 lm diameters upon the Al2O3. Electrical measurements were performed with a Keithley SCS 4200
parameter analyzer in combination with a preamplifier (detection limit ¼ 0.1 pA), allowing the measurement of very low
leakage currents. The current–voltage (I-V) and capacitance–voltage (C–V) measurements were performed to determine the
leakage current density and the dielectric constant, respectively. For the photoassisted C–V, as has been reported by
Swenson and Mishra,4 we used a xenon light source with a
monochromatic filter of 350 nm wavelength. Both the photoassisted C–V technique and the border trap analysis were performed at 1 MHz, 50 mV AC amplitude and 42 ms/V bias
voltage sweep rate. For the conductance method, the Al2O3GaN system was probed from 10 kHz to 2 MHz.
III. METHODOLOGY
Owing to its wide band gap, GaN exhibits very large time
constants for electron emission from deep trap levels and an
extremely low generation of minority charge carriers;9 and
once an electron is captured by a defect state near midgap,
the lack of holes inhibits the recombination of the electrons.
However, conventional methods for the extraction of the
interface trap density (Dit ) rely upon the change of trap occupancy at different frequencies while the Fermi energy level
is swept through the band gap by the applied DC bias. Thus,
only trap levels close to the conduction band edge are accessible, because they are the ones able to respond to the AC
signal at a given DC bias. To enable the emission of captured
electrons throughout the band gap, photoionization of the
interface traps can be employed by ultraviolet (UV) illumination with a photon energy, ht, larger than the GaN band
gap.
The photoassisted C–V method, proposed by Swenson
and Mishra,4 is based upon the comparison of the C–V curve
measured before and after UV illumination, where the density of interface traps can be determined from the resultant
voltage shift at a constant capacitance (corresponding to a
constant band bending). Details of the photoassisted C–V
FIG. 1. Cross section (not to scale) of the fabricated metal–insulator–semiconductor capacitor with ALD-grown Al2O3.
01A106-2
method are described elsewhere.4,5 For completeness, however, a brief summary of the technique is given in the following (Fig. 2).
First, the system is biased into accumulation to force all
of the interface traps to capture electrons and to set the system into a defined state of trap occupancy, whereupon the
initial C–V curve is measured in the dark from depletion to
accumulation. Assuming large time constants at room temperature and no change of trap occupancy, the initial dark
C–V is considered to be the “ideal” C–V curve. For the
extraction of Dit across the band gap, the ideal C–V is used
to determine the doping concentration, Nd , and surface
potential, Ws , such that4
Nd ¼
2
qeGaN e0 A2
Ws ¼
dð1=C2meas Þ
dV
;
qeGaN e0 A2 Nd
2½COx Cmeas =ðCOx Cmeas Þ2
(1)
;
(2)
where eGaN is the relative permittivity of the GaN, e0 is the
permittivity in vacuum, q the electron charge, A is the capacitor area, COx is the insulator capacitance, and Cmeas is the
measured capacitance value. Also, the parameter Nd is the
mean value of the doping concentration Nd . The measured
capacitance corresponds directly to the depletion width and,
therefore, to the band bending.
In the next step, the system is biased into deep depletion
under UV light exposure, whereupon electron–hole pairs are
generated and lead to a depletion capacitance increase. As
the depletion capacitance saturates, the UV light is switched
off and the bias voltage is maintained in deep depletion until
the system is in equilibrium. Finally, the post-UV C–V curve
is measured from the depletion to accumulation, wherein the
voltage shift to the negative voltage direction results from
FIG. 2. (Color online) Representative photoassisted C–V measurement procedure of ALD-grown Al2O3 on n-GaN at 1 MHz. (1) and (2) The initial
C–V sweep is performed in the dark after biasing the Al2O3/GaN system in
accumulation. (3) Next, the system is held in depletion under UV illumination for 10–15 min. (4) Finally, a post-UV C–V profile is measured in the
dark. The interface trap density is determined from the voltage shift, DV, at
a constant capacitance, which corresponds to the GaN band bending and the
surface potential.
J. Vac. Sci. Technol. B, Vol. 33, No. 1, Jan/Feb 2015
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 87.172.47.218 On: Mon, 05 Jan 2015 13:13:33
01A106-3 Winzer et al.: Impact of postdeposition annealing upon film properties
the captured holes (emitted electrons) in the interface traps
during UV illumination. Determining the voltage shift, DV,
at a constant capacitance, the Dit is derived as4
dNit COx dDV
Dit ¼
¼
:
(3)
dWs
qA dWs
Because the voltage shift of the photoassisted C–V technique
is a consequence of all of the charges close to the interface,
the conductance method and the border trap analysis are
applied to separate the fast-response trap states at the GaN
interface from the border trap states in the Al2O3.
The conductance method by Nicollian and Goetzberger6
only allows the extraction of interface traps with short emission
times in Al2O3/GaN owing to the large time constants of deep
interface trap levels and the lack of minority charge carrier generation in GaN. The technique relies upon the measured parallel conductance as a function of frequency and bias voltage,
which can be related to the capture and emission of carriers in
the interface trap. From the peak conductance, GP =x, with the
measurement frequency x ¼ 2pf , the Dit is extracted by6
2:5 GP
Dit ¼
:
(4)
x max
q
Border traps were first introduced by Fleetwood10 for the
Si/SiO2 system. In terms of the Al2O3/GaN structure, border
traps are oxide traps located in the Al2O3 close to the GaN
interface and are therefore able to respond to the Fermi level
sweep across the GaN bandgap. The border trap density, Nbt ,
can be estimated from the difference of the C–V curves
measured in the forward (depletion to accumulation) and in
the reverse (accumulation to depletion) bias direction. As
seen from Fig. 3, the ðð1=qAÞjDCreverse–forward jÞ difference
curve can be related to the distribution of the border trap
density across the applied bias voltage range. The total effective border trap density, Nbt; total , is obtained from the integration of the absolute value of the C–V hysteresis over the
applied bias voltage, such that7
ð
1
jCreverse Cforward jdV:
Nbt; total ¼
(5)
qA
FIG. 3. (Color online) Forward and reverse C–V measurements of an unannealed 14 nm-thick ALD-grown Al2O3 on n-GaN. The C–V hysteresis, DC,
for a constant bias voltage correlates with the density of border traps, Nbt.
01A106-3
For the conductance method as well as the border trap
analysis, the measured parallel capacitance, Cm, and parallel
conductance, Gm, are compensated for the series resistance
using the relations11
2
Gm þ x2 C2m Cm
Cc ¼
;
(6)
a2 þ x2 C2m
2
Gm þ x2 C2m a
;
(7)
Gc ¼
a2 þ x2 C2m
where x ¼ 2pf is the angular measurement frequency and
a ¼ Gm ðG2m þ x2 C2m Þrs . The series resistance, rs , is calculated from the measured capacitance, Cm;a , and conductance,
Gm;a , in accumulation, such that
rs ¼ h
Gm;a =xCm;a
2
:
2 i
1 þ Gm;a =xCm;a Gm;a
(8)
IV. RESULTS AND DISCUSSION
Unannealed ALD-grown Al2O3 is deposited on n-type
GaN without the formation of an interfacial layer, where the
absence of a GaOx interface is confirmed by transmission
electron microscopy (not shown) and the 1=COx versus
Al2O3 thickness plot from the C–V curves. Additionally, the
best fits of the XRR measurements are achieved without an
interfacial GaOx layer. From the C–V measurements, we
extract a dielectric constant of 8.9 for the unannealed Al2O3
films, which is consistent with the literature.12,13 With
increasing Al2O3 film thickness, the flat-band voltage of the
C–V curve shifts toward negative voltages (Fig. 4 with
inset), indicating the presence of positive fixed charges at the
oxide/semiconductor interface with the approximate density
of these fixed charges Nfix ¼ 5 1012 cm2. For all of the
ALD-grown Al2O3 films, the absolute value of the leakage
current densities are compared at a negative electrical field
of 2 MV/cm, where trap-assisted tunneling can be considered
FIG. 4. Normalized C–V curves of 9, 14, and 20 nm-thick unannealed
ALD-grown Al2O3 (e ¼ 8.9) films on n-GaN. The bias voltage is swept
from accumulation to depletion. The linear increase (see inset) of the flatband voltage (VFB) with the Al2O3 thickness (tAl2 O3 ) indicates a homogeneous distribution of positive fixed charges at the Al2O3/n-GaN interface
with Nfix ¼ 5 1012 cm2.
JVST B - Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 87.172.47.218 On: Mon, 05 Jan 2015 13:13:33
01A106-4 Winzer et al.: Impact of postdeposition annealing upon film properties
as the dominant leakage current mechanism. The leakage
current density through the unannealed Al2O3 is very low
and is found to be J ¼ 1.4 109 A/cm2 at 2 MV/cm (Fig.
5). With respect to the annealing atmosphere, thermal treatments at 300 and 500 C significantly affect the leakage current level of the Al2O3 thin films on GaN. Annealing
processes performed in N2 or a forming gas atmosphere degrade the ALD Al2O3 detrimentally and result in high leakage currents (see Fig. 5), with the leakage current density
found to be increased by more than one order of magnitude
for the N2-annealed films (J300 C ¼ 5.3 108 A/cm2,
J500 C ¼ 8.7 108 A/cm2 at 2 MV/cm) and two orders of
magnitude for the forming gas-annealed films (J300 C ¼ 1.3
107 A/cm2, J500 C ¼ 1.2 107 A/cm2 at 2 MV/cm)
compared to the unannealed Al2O3. In contrast to the N2 and
forming gas thermal treatments, O2 postdeposition anneals at
300 and 500 C both have a negligible impact upon the leakage current density (J300 C ¼ 1.7 109 A/cm2, J500 C ¼ 5.2
109 A/cm2 at 2 MV/cm).
To investigate the influence of the 500 C postdeposition
anneal in different ambient atmospheres upon the Al2O3/
GaN interface, the density of fast response traps are
extracted from the photoassisted C–V measurements for the
14 nm-thick Al2O3 films (Fig. 6). In the unannealed Al2O3
film, a Dit peak of 2 1012 cm2eV1 is observed close to
the conduction band edge. After annealing, the maximum Dit
is reduced for all annealing atmospheres. After the N2 thermal treatment, however, the Dit peak is broadened and levels
at 5–10 1012 cm2 eV1 for 0.2–0.6 eV from the GaN conduction band edge. The O2 anneal process is found to eliminate the interface traps more efficiently, but the forming gas
anneal at 500 C shows the best passivation of the interface
traps, resulting in the lowest Dit peak value of 5 1010 cm2
eV1 at 0.3 eV from the conduction band edge. Long et al.14
correlate the passivation effect of the forming gas anneal
with an increased incorporation of hydrogen at the Al2O3/
GaN interface, whereas the nitrogen annealing process
FIG. 5. (Color online) Influence of the annealing temperature and atmosphere (O2, N2, and forming gas) upon the leakage current density of ALDgrown Al2O3 films on n-GaN, measured at 2 MV/cm. The electric field
across the oxide is extracted from the capacitive voltage divider of the semiconductor and oxide capacitance with respect to the flat-band voltage shift
after the thermal treatment.
01A106-4
FIG. 6. (Color online) Interface trap density vs energy from the GaN conduction band, extracted from the photoassisted C–V technique at 1 MHz for Pt/
14 nm Al2O3/n-GaN metal–insulator–semiconductor capacitors. The samples are annealed for 10 min at 500 C in O2, N2, or forming gas. The gray
shaded area indicates the methodical limit of the applied analysis.
reduces the near-interface hydrogen content in the ALDgrown Al2O3 film.
The photoassisted C–V technique detects all of the
charges at the interface that are capable of responding to the
Fermi level sweep across the GaN bandgap, including border
traps and interface trap states. To investigate the influence of
the postdeposition annealing upon the density of interface
traps without the effect of the border traps, the conduction
method is applied. Devoid of border traps, the results of the
Dit extraction by the conductance method confirm the trends
of the results obtained via the photoassisted C–V technique
(Fig. 7). The impact of the annealing ambient atmosphere at
500 C upon the interface traps is found to be significant,
with the O2 annealing process reducing the Dit and the N2
thermal treatment lowering the maximum Dit marginally.
Considering the interface traps located at deep energy levels
(E EC > 0.8 eV), the N2 anneal is found to result in an
increased Dit compared to both the unannealed and the O2annealed Al2O3 films, meaning that the N2 anneal creates
interface traps. For the forming gas-annealed Al2O3/GaN
FIG. 7. (Color online) Interface trap density as a function of energy in the
GaN band gap, as determined from the conductance method for Pt/14 nm
Al2O3/n-GaN metal–insulator–semiconductor capacitors after a 10 min postdeposition annealing at 500 C in O2 or N2 compared with the unannealed
reference sample.
J. Vac. Sci. Technol. B, Vol. 33, No. 1, Jan/Feb 2015
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 87.172.47.218 On: Mon, 05 Jan 2015 13:13:33
01A106-5 Winzer et al.: Impact of postdeposition annealing upon film properties
stack, the conductance method does not give reasonable
results because the Dit value level is beyond the detection
limit of 5 1010 cm2 eV1.
The border trap analysis reveals that the influence of the
thermal treatment upon the overall border trap density in
the investigated Al2O3/GaN structures is much less than its
impact upon the Dit (see Fig. 8). In terms of the total effective border trap density, however (inset in Fig. 8), the unannealed Al2O3 exhibits the lowest density of border traps
(Nbt; total ¼ 1.2 1012 cm2) followed by the N2 and the
forming gas anneal. The O2 treatment produces the maximum total border trap density of 1.7 1012 cm2, which is
a factor of 1.42 increase compared to the unannealed
Al2O3.
The results of the interface trap characterization and the
leakage current measurements show that the O2 annealing at
500 C is the most beneficial of the investigated postdeposition treatments for the ALD-grown Al2O3 on GaN. On the
contrary, an annealing process in a nitrogen-rich atmosphere
degrades the Al2O3/GaN stack, with the thermal treatment at
500 C in pure N2 found to create interface traps and to
impair the leakage current through the Al2O3 film. These
results suggest that the N2 annealing process produces a
reduction of the Al2O3 stoichiometry at the GaN interface,
wherein an oxygen deficiency in the Al2O3 might be formed
during the 500 C postdeposition thermal treatment. The oxygen vacancies act as defect levels that diminish the Al2O3
bandgap, as reported by Ealet et al.,15 and therefore, the substoichiometric Al2O3 at the GaN interface cannot be considered to be completely insulating. This promotes the
formation of conductive paths through the oxide and a rise in
the leakage current, which can be inferred from these N2annealed Al2O3 films.
In contrast to the N2 annealing, a forming gas anneal
reduces the Dit , but for the forming gas postdeposition
annealing, a clear increase in leakage current is seen. This
01A106-5
sort of detrimental effect upon the leakage current occurring
because of hydrogen incorporation in the ALD Al2O3 thin
films has been already observed in other works.16–18 Here,
the influence of the hydrogen-containing oxygen precursor
has been studied without additional postdeposition annealing. Kozen et al.18 have reported an increase of the leakage
current density with higher hydrogen impurity concentration
in Al2O3. Moreover, Jennison et al.19 have found that interstitial neutral hydrogen creates a defect level near midgap in
the Al2O3-mediating electronic conduction through the oxide. Although hydrogen incorporation during a forming gas
postdeposition anneal passivates interface traps most efficiently, it does produce hydrogen-related defects traps within
the Al2O3 and induces the observed increase in the leakage
current. Therefore, a forming gas postdeposition annealing
process is not favored for the processing of future HFETs
with Al2O3 MIS-gate contacts.
V. SUMMARY
We investigate the influence of the annealing temperature and the annealing atmosphere upon the electrical properties of ALD-grown Al2O3 thin films on GaN. With
respect to the annealing atmosphere, thermal treatments
both at 300 and 500 C have a significant impact upon the
leakage current density of the Al2O3 thin films compared to
the unannealed films. Postdeposition annealing in N2 or
forming gas results in a detrimental increase of the leakage
current. For all of the different annealing atmospheres,
thermal treatments at 500 C marginally affect the border
oxide trap density. Interface traps are passivated by hydrogen during the forming gas annealing, while the N2 thermal
treatment creates interface trap charges. In terms of producing low leakage currents and a reduced density of interface
traps, the O2 annealing at 500 C is the most beneficial postdeposition treatment for the investigated ALD-grown
Al2O3 on GaN.
ACKNOWLEDGMENT
The authors thank the German Federal Ministry for
Education and Research for funding this work within the
project TeleGaN (01BM1204).
1
J. Robertson and B. Falabretti, J. Appl. Phys. 100, 014111 (2006).
A. Freese, M. Grube, A. Wachowiak, M. Geidel, B. Adolphi, S. Schmult,
and T. Mikolajick, J. Vac. Sci. Technol., B 31, 03C111 (2013).
3
J. Robertson, Rep. Prog. Phys. 69, 327 (2006).
4
B. L. Swenson and U. K. Mishra, J. Appl. Phys. 106, 064902 (2009).
5
R. Yeluri, X. Liu, B. L. Swenson, J. Lu, S. Keller, and U. K. Mishra,
J. Appl. Phys. 114, 083718 (2013).
6
E. H. Nicollian and A. Goetzberger, Bell Syst. Tech. J. 46, 1055 (1967).
7
D. M. Fleetwood and N. S. Saks, J. Appl. Phys. 79, 1583 (1996).
8
K. K€
ohler, S. M€
uller, P. Waltereit, L. Kirste, H. P. Menner, W. Bronner,
and R. Quay, Phys. Status Solidi A 206, 2652 (2009).
9
M. Miczek, C. Mizue, T. Hashizume, and B. Adamowicz, J. Appl. Phys.
103, 104510 (2008).
10
D. M. Fleetwood, IEEE Trans. Nucl. Sci. 39, 269 (1992).
11
E. H. Nicollian and J. R. Brews, MOS Physics and Technology (Wiley,
New York, 1982).
12
Z. 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).
2
FIG. 8. (Color online) Estimation of the border oxide trap density as
derived from the C–V hysteresis ðð1=qAÞjDCjÞ as a function of the applied
bias voltage for Pt/14 nm Al2O3/n-GaN metal–insulator–semiconductor
capacitors after a 10 min postdeposition anneal at 500 C in O2 (upside
down triangle), N2 (triangle), or forming gas (circle), compared to the
unannealed reference sample (square). The inset shows the influence of the
annealing atmosphere in a 500 C thermal treatment upon the total effective border trap density.
JVST B - Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 87.172.47.218 On: Mon, 05 Jan 2015 13:13:33
01A106-6 Winzer et al.: Impact of postdeposition annealing upon film properties
13
Y. C. Chang, M. L. Huang, Y. H. Chang, Y. J. Lee, H. C. Chiu, J. Kwo,
and M. Hong, Microelectron. Eng. 88, 1207 (2011).
14
R. D. Long et al., Appl. Phys. Lett. 103, 201607 (2013).
15
B. Ealet, M. H. Elyakhloufi, E. Gillet, and M. Ricci, Thin Solid Films 250,
92 (1994).
16
S.-C. Ha, E. Choi, S.-H. Kim, and J. S. Roh, Thin Solid Films 476, 252
(2005).
01A106-6
17
L. Aarik, T. Arroval, R. Rammula, H. M€andar, V. Sammelselg, B.
Hudec, K. Husekova, K. Fr€
ohlich, and J. Aarik, Thin Solid Films 565, 19
(2014).
18
A. C. Kozen, M. A. Schroeder, K. D. Osborn, C. J. Lobb, and G. W.
Rubloff, Appl. Phys. Lett. 102, 173501 (2013).
19
D. R. Jennison, P. A. Schultz, and J. P. Sullivan, Phys. Rev. B 69, 041405
(2004).
J. Vac. Sci. Technol. B, Vol. 33, No. 1, Jan/Feb 2015
Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 87.172.47.218 On: Mon, 05 Jan 2015 13:13:33