Water soluble nano-scale transient material germanium oxide for zero toxic waste based environmentally benign nano-manufacturing

Abstract

In the recent past, with the advent of transient electronics for mostly implantable and secured electronic applications, the whole field effect transistor structure has been dissolved in a variety of chemicals. Here, we show simple water soluble nano-scale (sub-10 nm) germanium oxide (GeO2) as the dissolvable component to remove the functional structures of metal oxide semiconductor devices and then reuse the expensive germanium substrate again for functional device fabrication. This way, in addition to transiency, we also show an environmentally friendly manufacturing process for a complementary metal oxide semiconductor(CMOS) technology. Every year, trillions of complementary metal oxide semiconductor(CMOS)electronics are manufactured and billions are disposed, which extend the harmful impact to our environment. Therefore, this is a key study to show a pragmatic approach for water soluble high performance electronics for environmentally friendly manufacturing and bioresorbable electronic applications.

Full text

Water soluble nano-scale transient material germanium oxide for zero toxic waste
based environmentally benign nano-manufacturing
A. S. Almuslem, A. N. Hanna, T. Yapici, N. Wehbe, E. M. Diallo, A. T. Kutbee, R. R. Bahabry, and M. M. Hussain
Citation: Appl. Phys. Lett. 110, 074103 (2017); doi: 10.1063/1.4976311
View online: http://dx.doi.org/10.1063/1.4976311
View Table of Contents: http://aip.scitation.org/toc/apl/110/7
Published by the American Institute of Physics
Articles you may be interested in
Electrically driven and electrically tunable quantum light sources
Appl. Phys. Lett. 110, 071102071102 (2017); 10.1063/1.4976197
Surface acoustic wave unidirectional transducers for quantum applications
Appl. Phys. Lett. 110, 073105073105 (2017); 10.1063/1.4975803
A complete laboratory for transport studies of electron-hole interactions in GaAs/AlGaAs ambipolar bilayers
Appl. Phys. Lett. 110, 072105072105 (2017); 10.1063/1.4976505
Distributed angular double-slit interference with pseudo-thermal light
Appl. Phys. Lett. 110, 071107071107 (2017); 10.1063/1.4976575

Water soluble nano-scale transient material germanium oxide for zero toxic
waste based environmentally benign nano-manufacturing
A. S. Almuslem,
1
A. N. Hanna,
1
T. Yapici,
2
N. Wehbe,
2
E. M. Diallo,
2
A. T. Kutbee,
3
R. R. Bahabry,
3
and M. M. Hussain
1,a)
1
Integrated Nanotechnology Lab and Integrated Disruptive Electronic Applications (IDEA) Lab,
Electrical Engineering, Computer Electrical Mathematical Science and Engineering Division,
King Abdullah University of Science and Technology (KAUST), Thuwal 23955-6900, Saudi Arabia
2
Analytical Chemistry Lab and Imaging and Characterization Lab, KAUST Core Facilities, KAUST, Thuwal,
Saudi Arabia
3
Integrated Nanotechnology Lab and Integrated Disruptive Electronic Applications (IDEA) Lab,
Material Science and Engineering, Physical Science and Engineering Division, King Abdullah University of
Science and Technology (KAUST), Thuwal 23955-6900, Saudi Arabia
(Received 4 December 2016; accepted 25 January 2017; published online 14 February 2017)
In the recent past, with the advent of transient electronics for mostly implantable and secured
electronic applications, the whole field effect transistor structure has been dissolved in a variety of
chemicals. Here, we show simple water soluble nano-scale (sub-10 nm) germanium oxide (GeO
2
)
as the dissolvable component to remove the functional structures of metal oxide semiconductor
devices and then reuse the expensive germanium substrate again for functional device fabrication.
This way, in addition to transiency, we also show an environmentally friendly manufacturing pro-
cess for a complementary metal oxide semiconductor (CMOS) technology. Every year, trillions of
complementary metal oxide semiconductor (CMOS) electronics are manufactured and billions are
disposed, which extend the harmful impact to our environment. Therefore, this is a key study to
show a pragmatic approach for water soluble high performance electronics for environmentally
friendly manufacturing and bioresorbable electronic applications. Published by AIP Publishing.
[http://dx.doi.org/10.1063/1.4976311]
Transient electronics have received growing attention
recently owing to its dissolvability for implantable and
secured electronic applications.
1,2
In these examples, the
whole structure is dissolved over the time in a variety of
chemicals. Here, we show that water soluble germanium
dioxide (GeO
2
) can be employed as a transient material
besides its function as a dielectric layer. The usage of GeO
2
as the dissolvable material will still remove the key structure
to serve the purpose of dissolution. At the same time, since it
is simply water soluble, we can reuse the expensive germa-
nium (Ge) substrate for further device fabrication.
GeO
2
can have three forms, two of them are crystalline
and the third one is vitreous. Crystalline forms are repre-
sented by hexagonal and tetragonal crystalline structures.
3
The growth conditions determine the resultant form of GeO
2
,
and the difference between them was reported by Johnson.
4
Only hexagonal and vitreous form is water soluble. The
hydrolysis mechanism of GeO
2
undergoes the following
chemical reaction:
5
GeO2ðsÞþH2OðlÞ$H2GeO3ðaq:Þ:(1)
Surface oxidation of Ge has been considered as an unde-
sirable feature, which prevents Ge to be incorporated in the
Complementary Metal Oxide Semiconductor (CMOS) indus-
try for a long time. However, many studies have addressed a
surface passivation challenge for the purpose of making use
of superior material properties of Ge. Ge possesses favorable
properties such as high carrier mobility (up to 3900 cm
2
V
1
s
1
for electrons and 1900 cm
2
V
1
s
1
for holes),
6
direct
and small energy band gap (0.66 eV), small optical band gap
(wide absorption wavelength spectrum), and low dopant
thermal activation energies in comparison with the universal
CMOS material Silicon (Si).
7
As a result of unique proper-
ties, Ge becomes the material of choice for manufacturing
high speed, low power devices,
8
as well as optoelectronic
devices.
9
In this work, water soluble GeO
2
is a hallmark since it
serves two different purposes as it allows for both recycling
the substrate and reducing the pollution originated from dis-
carded consumer electronics. We show that it would be of
great benefit to meet the rapid growth demands for consumer
electronics and keeping the electronic waste at bay simulta-
neously. Electronic waste is reported to be up to 42 10
6
tons yearly.
10
Moreover, water as an etchant of GeO
2
is
highly preferable to other wet reactive chemicals as water is
used frequently in the IC fabrication process for cleaning
and etching purposes because: (i) it does not have an envi-
ronmental impact, (ii) it is safe to handle easily, and (iii) it is
relatively of low-cost.
11
On the other hand, the Ge/GeO
2
interface has a variety of satisfactory properties such as
acceptable dielectric constant (j¼2.8),
12
moderate refrac-
tive index, and adequate thermal stability.
13
Besides that,
GeO
2
has the advantage of suppressing the Ge dangling
bond at the surface, which in turn minimizes the interface
trap density.
14
In this study, Metal Oxide Semiconductor Capacitors
(MOSCAPs) are built on p-type Ge substrates using GeO
2
as
a)
Author to whom correspondence should be addressed. Electronic mail:
muhammadmustafa.hussain@kaust.edu.sa
0003-6951/2017/110(7)/074103/5/$30.00 Published by AIP Publishing.110, 074103-1
APPLIED PHYSICS LETTERS 110, 074103 (2017)

a dielectric layer. After characterizing the capacitors, the sub-
strate is rinsed with tap water in order to dissolve GeO
2
,which
in turn takes away all the top layers. The substrate is then
effectively reused to build the similar MOSCAPs again. GeO
2
is thermally grown on a p-type Ge (Ga-doped) 175lmthick
substrate with resistivity ranging from 0.01 to 0.05 Xcm.
We start with 4-in. Ge wafer, which is then diced up to
22cm
2
experimental samples. Then, the pieces are cleaned
with HF (49%), rinsed with Deionized (DI) water, blown
driedinNitrogen(N
2
), followed by acetone and isopropanol
based cleaning, and blown dried in N
2
. After that, the proc-
essed experimental samples are transferred into an Ultra High
Vacuum Chemical Vapor Deposition (UHVCVD) chamber in
order to grow GeO
2
. The chamber’s temperature is stabilized
at 400 Cfor4hinOxygen(O
2
) ambient with a flow rate of
100 sccm.
X-ray Photoemission Spectroscopy (XPS) was used to
verify the substrate surface properties for a clean Ge surface,
grown GeO
2
layer, GeO
2
removal, and regrown GeO
2
layer
as it is capable to provide the chemical composition.
15
XPS
experiments were performed on a KRATOS Analytical
AMICUS instrument equipped with an achromatic Al Ka
X-ray source (1468.6 eV). Typically, the source was operated
at a voltage of 10kV and a current of 10mA. The pressure in
the analysis chamber was below 4 10
6
Pa. In XPS, it is
well known that 2p
3/2
and 3d
5/2
photoemission lines of the
elemental Ge are detected at 1217.3 eV and 29.3 eV, respec-
tively, and any peak shift in the binding energy is related to
altering the oxidation state of Ge. XPS data recorded for a
clean germanium sample and for the processed samples are
compared in Fig. 1(a). The three processed samples refer
to post GeO
2
growth (GeO
2
grown on pristine Ge after
cleaning), post GeO
2
removal, and post GeO
2
re-growth
(GeO
2
grown after the first GeO
2
has been removed). XPS
peaks of GeO
2
grown and GeO
2
regrown samples are identi-
cal and reveal a clear shift with respect to the reference sam-
ple. Indeed, the 2p
3/2
and 3d
5/2
photoemission lines are
shifted by 2.8 eV and 3.5 eV, respectively, toward higher
binding energy, which is specified to the fourth oxidation
state (Ge
þ4
) of Ge and confirms the existence of the GeO
2
layer.
7,13,16,17
However, the spectrum acquired for the refer-
ence germanium substrate is very similar to the one belong-
ing to the rinsed sample, suggesting that both layers have the
same composition. The quantitative data indicate that the ref-
erence and rinsed samples are composed of almost 50% of
germanium and 50% of oxygen, whereas the grown and
regrown layers contain 67% of oxygen and 33% of germa-
nium. Precisely speaking, the major components of Ge 2p
3/2
and Ge 3d
5/2
of the reference and rinsed samples are
detected, respectively, at 1218 eV and 29.6 eV. These values
can be assigned to metallic germanium Ge.
7,13,16,17
The
smaller component detected mainly for Ge 2p
3/2
at 1219.8
can be assigned to germanium monoxide (GeO). Hence, the
most probable chemical composition of both reference and
rinsed layers can be described by a thin over layer of native
germanium monoxide (GeO) covering the metallic germa-
nium. In contrast, the major components of Ge 2p
3/2
and Ge
3d
5/2
of the grown and regrown samples are detected, respec-
tively, at 1221.2 eV and 32.8 eV. These values can be clearly
assigned to germanium dioxide GeO
2
.
7,13,16,17
The intensity
of the Ge 3d component measured at 29.3 eV for the regrown
layer is slightly higher than that measured for the grown one,
suggesting that the regrown GeO
2
layer is thinner.
Knowledge of the Ge/GeO
2
interface is essential for
evaluating the quality of the growth method. A valuable
insight into the Ge/GeO
2
interface is obtained from TEM
FIG. 1. (a) High resolution XPS spectra of Ge 2p and Ge 3d peaks acquired for the reference Ge substrate, for the grown GeO
2
layer, for the same layer after
submerging into tap water for 3 days, and for the re-grown GeO
2
layer after water based removal. (b) and (c) TEM images of GeO
2
for post-growth and post-
re-growth samples, respectively. (d) and (e) AFM images of grown and regrown GeO
2
layers, respectively. (f) Thickness variation of GeO
2
across 10.76 mm
2
by spectroscopic ellipsometry. (g) Refraction index and extinction coefficient of GeO
2
.
074103-2 Almuslem et al. Appl. Phys. Lett. 110, 074103 (2017)

images (Figs. 1(b) and 1(c)), which show a discontinuity free
GeO
2
layer with a clear and slightly zig zag Ge/GeO
2
inter-
face in both cases of post GeO
2
growth and post-regrowth
GeO
2
samples. However, the surface of the Ge substrate
becomes rougher after regrown GeO
2
and that introduces pro-
nounced variation in the GeO
2
thickness. Atomic force
microscopy (AFM) is employed to obtain the topographic
imaging of 400 lm
2
surface area of GeO
2
(Figs. 1(d) and
1(e)) and the RMS roughness found to be 1.89 nm and 5 nm
for post GeO
2
growth and post-regrowth GeO
2
, respectively.
The thickness uniformity of the thermally grown GeO
2
layer
is examined using spectroscopic ellipsometry. The map of the
scanned area of 10.76 mm
2
reveals uniform growth of germa-
nium dioxide with an average thickness of 7.260.3 nm varia-
tion across the scanned area (Fig. 1(f)). The optical constants
representing the refraction index (n) and extinction coefficient
(K) of the GeO
2
layer have been measured and depicted in
Fig. 1(g). A couple of features of this plot are worth pointing
out. First, the measured refractive index at 550 nm is equal to
1.6, which is in total agreement with the reported value in pre-
vious studies.
13,18
Furthermore, the maximum value of the
extinction coefficient (K) of GeO
2
is observed at 246 nm,
which means that GeO
2
has a tendency to be highly absorbing
at 246 nm.
In order to get accurate information about the dissolu-
tion rate of GeO
2
, Emission-spectrometric detection of the
elements at the ultra-trace level using inductively coupled
plasma optical emission spectrometry (ICP-OES) is hired.
The dissolution rate is systematically studied at different
temperatures within the range of T ¼23 C and T ¼70 C for
tap water, DI water, and Phosphate Buffered Solutions
(PBS). The GeO
2
is thermally grown on p-type Ge pieces
that have the same size (1 1cm
2
). Then, every piece is
submerged into 100 ml of a solution in a sealed Teflon
bottle, which is placed in a thermal water bath at steady
temperature. Later, 5 ml of the solution is drawn every cou-
ple of hours and used to measure the concentration of Ge
ions. In more detail, the solution is fed into a 50-MHz radio
frequency (RF) induced argon plasma chamber from sample
introduction nebulization systems, which is exposed to an
elevated temperature up to approximately 7000–8000 K. At
such atmosphere, all the analyte species are thermally
excited through collisional excitation within the plasma. A
photon at quantized energy is released when the atomic and
ionic excited state species relax to the ground state, which in
turn is used to identify the analyte species from which they
originated. The concentration of the analyte species is
directly proportional to the total number of photons.
19
The
detected wavelengths are equal to 219.871 nm, 204.377 nm,
259.253 nm, 209.426 nm, 265.117 nm, and 206.866 nm,
which is certainly a fingerprint of Ge ions.
20
This demon-
strates the removal of GeO
2
by dissolution with well-defined
kinetics. The concentration of Ge ions is measured for all
stipulated wavelengths. Then, the average of the concentra-
tion is plotted as a function of time (Figs. 2(a)–2(c)). The
slope of solubility curves in linear regimes represents the dis-
solution rate of GeO
2
. The dissolution rate of GeO
2
is plotted
at different temperatures for tap water, DI water, and PBS
solution as depicted in Fig. 2(d). Fig. 2(d) sheds some light
on the dissolution rate of GeO
2
from different aspects. First
of all, there is a highly significant distinction between the
dissolution rate of GeO
2
in tap water on one side and DI
water and PBS solutions on the other side. Further, the disso-
lution rate of GeO
2
increases remarkably with temperature in
tap water, whereas it increases slightly in DI water and PBS
solutions, as well. As a result, the dissolution rate of GeO
2
could be controlled based on the temperature and pH values
of the solution. From the chemistry point of view, the three
solutions used in this study differ from each other in the pH
value. The tap water has the highest pH value (8.9) among
FIG. 2. (a)–(c) Concentration of Ge ions
measured by (ICP-OES). (b) Dissolution
rate of GeO
2
.
074103-3 Almuslem et al. Appl. Phys. Lett. 110, 074103 (2017)

other employed solutions and has a clear impact on speeding
up the dissolution rate of GeO
2
(Fig. 2(d)). The pH value
indicates that tap water has a higher relative number of
hydroxyl ions (OH

) in comparison to DI water (pH ¼6.6)
and PBS (pH ¼7.4). OH

ions are of great importance to ini-
tiate the hydrolysis reaction of GeO
2
with water. Strictly
speaking, the physical properties, in particular, the ionic
charge z and the ionic radius of the material, will determine
under which pH value will the material hydrolyze.
21
Additionally, the presence of OH

ion in the solution will
speed up the hydrolysis reaction because the reactivity of
OH

ions is four orders of magnitude greater than H
2
O.
22
In
future, the scope of the influence of pH needs to be studied
in more detail. On the other hand, the solution’s temperature
can accelerate the dissolution rate, which can be explained
based on the kinetic molecular theory and the temperature
dependence of reaction rate constant, which in turn quantifies
the rate of a chemical reaction.
23
MOSCAPs (with area 2.19 10
4
mm
2
) are fabricated
in order to study the utility of GeO
2
as a dielectric layer in
the case of post GeO
2
growth and post-regrowth samples.
The process flow starts with the Ge substrate that is first
diced to 2 2cm
2
pieces and cleaned with acetone and iso-
propanol. Then, Ti (40 nm)/Pt (60 nm) is sputtered in the
backside to serve as a back electrode and to prevent GeO
2
from growing on the backside of the pieces. Then, GeO
2
is
grown. Later on, sputtered Al (250 nm) is patterned through
a shadow mask to serve as a gate electrode Fig. 3(a). The
data of CV measurement were collected by an LCR meter at
1 MHz. After characterizing the MOS capacitors, the piece is
immersed into a tap water at RT and sonicated for 60 min
every 24 h for three days with replacing the water daily. The
dissolution of the GeO
2
layer starts at the exposed area and
at the edge of the Al patterned layer, which in turn facilitates
removing the Al layer (Fig. 3(b)). Then, the pieces were
examined by XPS to verify the absence of GeO
2
.Later,the
piece is treated exactly as the first time in order to clean it and
rebuild the MOSCAPs. The C-V characteristics of capacitors
in both cases (post GeO
2
growth and post-regrowth samples)
are shown in Fig. 3(c). For the first-time growth sample, we
notice a capacitance value of 495 pF for devices. For the pur-
pose of extracting the value of dielectric constant of the GeO
2
layer, the universal equation is employed
C¼je0A
d;(2)
where jis the dielectric constant, e
0
is the vacuum permittiv-
ity, A is the device area, and d is the dielectric thickness.
According to the thickness scan map shown in Fig. 1(f), the
d value of 7.2 nm has been chosen for the calculation of j.
The calculated value came to be 1.84, which is consistent
with the previous literature,
18
showing similar dielectric
properties as shown from ellipsometry data, Figs. 1(g) and
3(d), where the increase in the dielectric constant for photon
energies larger than 3 eV is attributed to sub-bandgap photo-
absorption due to significant density of gap states. The exis-
tence of mid-gap states could also explain the bump shown
in the CV curve in Fig. 3(c).
24
After the regrowth of the
GeO
2
layer, the oxide capacitance becomes 43% higher than
the original value. Spectroscopic ellipsometry data (Fig.
3(d)) have shown the dielectric constant of the regrown sam-
ple to be identical to that of the first-time growth. Therefore,
using the extracted dielectric constant from the first growth
capacitance measurement, 1.84, we extracted a dielectric
thickness of 5 nm and this was confirmed by the spectro-
scopic ellipsometry measurement. This proves that the thick-
ness reduction of GeO
2
is behind the 43% higher capacitance
after regrowth. We have independently calculated the dielec-
tric constant using the C-V measurements and measured film
thicknesses for post-growth and post-regrowth samples,
which yielded a similar dielectric constant (Figure S1 in the
supplementary material).
We have shown how water soluble sub-10 nm GeO
2
can
be effectively used for environmentally benign zero toxic
waste CMOS manufacturing for scaled high performance
CMOS electronics, transient electronics for water solvable
low-cost transient sensory systems.
FIG. 3. (a) Fabrication flow of MOS capacitors. (b) Chronological optical images gathered at different stages of dissolution of GeO
2
(0–3 days). (c) CV mea-
surement of MOS capacitors at 1 MHz for post-growth and post regrowth GeO
2
samples. (d) Dielectric constant of GeO
2
post-growth and post-regrowth.
074103-4 Almuslem et al. Appl. Phys. Lett. 110, 074103 (2017)

See supplementary material for extracting the relative
dielectric constant jfrom CV measurement and atomic force
microscopy of GeO
2
after immersed in tap water. Two fig-
ures (Figs. S1 and S2) are also included.
This publication is based on the work supported by the
King Abdullah University of Science and Technology
(KAUST).
1
X. Huang, Y. Liu, S.-W. Hwang, S.-K. Kang, D. Patnaik, J. F. Cortes, and
J. A. Rogers, Adv. Mater. 26, 7371 (2014).
2
S. K. Kang, R. K. J. Murphy, S. W. Hwang, S. M. Lee, D. V. Harburg,
N. A. Krueger, J. H. Shin, P. Gamble, H. Y. Cheng, S. Yu, Z. J. Liu, J. G.
McCall, M. Stephen, H. Z. Ying, J. Kim, G. Park, R. C. Webb, C. H. Lee,
S. J. Chung, D. S. Wie, A. D. Gujar, B. Vemulapalli, A. H. Kim, K. M.
Lee, J. J. Cheng, Y. G. Huang, S. H. Lee, P. V. Braun, W. Z. Ray, and
J. A. Rogers, Nature 530, 71 (2016).
3
M. K. Murthy and H. Hill, J. Am. Ceram. Soc. 48, 109 (1965).
4
O. H. Johnson, Chem. Rev. 51, 431 (1952).
5
M. F. Ehman, K. Vedam, W. B. White, and J. W. Faust, J. Mater. Sci. 6,
969 (1971).
6
Q. Xie, S. Deng, M. Schaekers, D. Lin, M. Caymax, A. Delabie, X.-P. Qu,
Y.-L. Jiang, D. Deduytsche, and C. Detavernier, Semicond. Sci. Technol.
27, 074012 (2012).
7
J. Oh and J. C. Campbell, J. Electron. Mater. 33, 364 (2004).
8
C. O. Chui, S. Ramanathan, B. B. Triplett, P. C. McIntyre, and K. C.
Saraswat, IEEE Electron Device Lett. 23, 473 (2002).
9
E. S. Goh, T. Chen, C. Sun, and Y. Liu, J. Appl. Phys. 107, 024305
(2010).
10
X. Zeng, Q. Song, J. Li, W. Yuan, H. Duan, and L. Liu, J. Cleaner Prod.
90, 55 (2015).
11
J.-G. Park, Ph.D. thesis, The University of Arizona, Tucson, 1993.
12
J. F. Binder, Ph.D. thesis, Ecole Polytechnique Federale De Lausanne,
Lausanne, 2012.
13
N. R. Murphy, J. T. Grant, L. Sun, J. G. Jones, R. Jakubiak, V.
Shutthanandan, and C. V. Ramana, Opt. Mater. 36, 1177 (2014).
14
N. Taoka, K. Ikeda, Y. Yamashita, N. Sugiyama, and S.-I. Takagi,
Semicond. Sci. Technol. 22, S114 (2006).
15
D. Briggs and J. T. Grant, Surface Analysis by Auger and X-ray
Photoelectron Spectroscopy (IM Publications, Chichester, 2003), p. 31.
16
A. Molle, M. N. K. Bhuiyan, G. Tallarida, and M. Fanciulli, Appl. Phys.
Lett. 89, 083504 (2006).
17
V. Grossi, L. Ottaviano, S. Santucci, and M. Passacantando, J. Non-Cryst.
Solids 356, 1988 (2010).
18
K. Kita, C. H. Lee, T. Nishimura, K. Nagashio, and A. Toriumi, ECS
Trans. 19, 101 (2009).
19
See http://onlinelibrary.wiley.com/ for Inductively Coupled Plasma
Optical Emission Spectrometry. Encyclopedia of Analytical Chemistry:
Instrumentation and Applications; accessed 9 October 2016.
20
J. E. Sansonetti, W. C. Martin, and S. L. Young, J. Phys. Chem. Ref. Data
34, 1734 (2005).
21
C. Ekberg and P. L. Brown, Hydrolysis of Metal Ions (Wiley-VCH Verlag
GmbH & Co. KGaA, Weinheim, 2016), p. 39.
22
R. A. Larson, Reaction Mechanisms in Environmental Organic Chemistry
(Lewis Publishers, Boca Raton, 1994), p. 105.
23
S. K. Upadhyay, Chemical Kinetics and Reaction Dynamics (Springer
Science & Business Media, New York, 2006), p. 46.
24
K. Martens, C. O. Chui, G. Brammertz, B. D. Jaeger, D. Kuzum, M.
Meuris, M. Heyns, T. Krishnamohan, K. Saraswat, and H. E. Maes, IEEE
Trans. Electron Devices 55, 547 (2008).
074103-5 Almuslem et al. Appl. Phys. Lett. 110, 074103 (2017)