The impact of volatile organic compound (VOC) exposure on the electrical breakdown of GaN in the inter channel region of dual channel microcantilever heaters has been studied. Exposure to three different VOCs with different latent heats of evaporation resulted in changes in breakdown voltage of varying magnitudes that can be correlated with their latent heats. A physical model has been proposed to explain the observed shift in breakdown voltage upon VOC exposure based on changes in thermal and electrical profiles at the microcantilever apex, which is caused by the molecular interaction and amplified by its unique tapered geometry. The critical breakdown field of the inter channel GaN has been observed to reduce dramatically by almost 50 times compared to that of bulk GaN at room temperature. The inter-channel current rises dramatically at the onset of breakdown induced by VOC exposure, at specific bias voltages corresponding to VOCs, which can be utilized for detecting them with high sensitivity as well as selectivity.

III-Nitride semiconductors are well known for their application in high power microwave and power electronic devices owing to their large bandgap and unique polarization properties which results in the formation of high sheet charge density and carrier mobility at heterostructure interfaces.1–3 In recent years, with the availability of high quality III-Nitride epitaxial layers on the Si substrate, there has been a renewed focus on developing III-Nitride based microelectromechanical systems (MEMS) taking advantage of their unique material properties.4,5 Polarized III-Nitride surfaces have a propensity to interact with polar molecules, a property that has been exploited to develop gateless III-Nitride heterostructure field effect transistors (HFETs) with exposed gate areas, which allow polar molecules to alter the 2DEG upon adsorption. These FET type sensors have shown success in sensing aqueous salts, nitrous oxide, and volatile organic compounds (VOCs).6,7 In recent years, piezoresistive and piezotransistive microcantilevers fabricated from this material system have been used for highly sensitive detection of displacement, trace explosives, surface potential, acoustic waves, and volatile organic compounds (VOCs).8–14 

Silicon based microcantilevers with heater elements were originally devised in the late 1990s which were often used in conjunction with an atomic force microscope (AFM) for thermomechanical data writing, nanoscale topography imaging, and deposition processes.15–18 Due to their highly polar surface and presence of a sub-surface electronic sheet charge of high density and mobility, AlGaN/GaN heterojunction based triangular microcantilevers are inherently suitable for integrating heater channels and developing low power sensors for detecting VOCs.12–14 Detection of VOCs, which are ubiquitous in residential and workplace environments, is highly significant as they have been linked to cancers, asthma, and various neurological problems.19,20 In contrast to hot bead pellistors, where the high temperature causes the VOCs to undergo an exothermic reaction with oxygen causing the bead resistance to increase,21 the TMHs (triangular microcantilever heaters) have been observed to undergo a reduction in resistance in the heated tip region when exposed to VOCs. This has been explained by considering VOC condensation in the surrounding region of the tip followed by rapid transport (induced by convection) and re-evaporation closer to the tip, removing latent heat and lowering the tip temperature.13 Two types of TMH based VOC sensors have been investigated so far: one with a single combined heating and sensing channel12 and the other with an outer sensing channel and a separate inner heating channel, which also demonstrated the capability to identify individual components in a VOC mixture.14 However, the change in the device signal in the presence of VOCs was generally <5%, and the limit of detection was found to be in the ppm range.14 A sensitivity down to the ppb range is typically required to ensure the widespread applicability of VOC sensors.22,23

Monolithic tip dual channel microcantilever heaters (MDC-MHs; scanning electron micrographs shown in Fig. 1 not only offer a platform to perform VOC sensing utilizing separate functionalities of the two channels but also present a unique geometry suitable for exploring the electrical breakdown characteristics of the GaN layer between the AlGaN/GaN channels. In this work, we have investigated the influence of polar VOC vapor exposure, in the presence of an elevated temperature profile, on the critical breakdown voltage of the inter-channel GaN layer in the MDC-MHs. A simple physical model has been proposed to explain the electrical breakdown of the inter-channel GaN layer, observed to be 50-fold lower than that of bulk GaN at room temperature. The inter-channel breakdown phenomenon, occurring at moderate bias voltages, presents a unique opportunity to devise a sensing scheme based on the inter-channel breakdown current (through GaN layer), which can increase by several orders of magnitude.

FIG. 1.

(a) SEM image of a monolithic tip dual channel microcantilever heater (MDC-MH), Ti/Au contacts, and substrate. (b) SEM image showing the two parallel AlGaN channels and the gap distance at the tip.

FIG. 1.

(a) SEM image of a monolithic tip dual channel microcantilever heater (MDC-MH), Ti/Au contacts, and substrate. (b) SEM image showing the two parallel AlGaN channels and the gap distance at the tip.

Close modal

The MDC-MHs used in this study were fabricated using AlGaN/GaN epitaxial layers grown on a 625 μm thick Si (111) substrate, purchased from NTT Advanced Technology Corporation, Japan. The wafers had a 2 nm i-GaN cap layer and 15 nm Al0.25Ga0.75N on top of 1 μm i-GaN, with a 300 nm buffer layer separating the GaN layer from the thick Si substrate. At the beginning of the fabrication process, the top 100 nm of the AlGaN/GaN layer was etched using BCl3/Cl2 plasma to isolate the cantilever mesa, followed by another deep etch to define the cantilever geometry. Then, a Ti(20 nm)/Al(100 nm)/Ti(45 nm)/Au(55 nm) metal stack was deposited followed by a rapid thermal annealing process at 825 °C for 1 min, to form the ohmic contacts on the mesa region. The Ti(20 nm)/Au(225 nm) stack was deposited afterwards to form the probe contacts. Finally, the Bosch process was used, with a plasma-enhanced chemical vapor deposited SiO2 masking layer at the backside, to perform through wafer etching of the silicon layer to release the cantilevers. The MDC-MHs feature separate heating and sensing channels. The lower resistance inner channel served as the heating channel for the entire device, which enabled desired temperature to be reached at a lower bias voltage. The outer channel typically served as the sensing channel and is biased at 1 V to generate a high enough current to be monitored easily for VOC detection without causing any significant heating effects.14 

Current-voltage (I-V) characteristics for the MDC-MHs were determined under dark, closed chamber conditions using both terminals of an Agilent B2900A Series Source/Measure Unit (SMU). The I-V characteristics recorded with the inner channel bias varied from 0 to 50 V are shown in Fig. 2(a). The outer channel was maintained at 1 V bias to monitor heating effects and record I-V characteristics. A schematic representation of the current flow directions in various channel sections of the MDC-MH before and after breakdown is shown in Fig. 2(b). Upon breakdown, the two channels become electrically connected at some location with the potential VB [see Fig. 2(b)] allowing current to flow from the inner channel to the outer channel. The source current Ii1 splits into three directions at the breakdown point with the relation Ii1 = Ii2 + Io1 + Io2, where Ii2, Io1, and Io2 are the different branch-off currents. Since the breakdown current Io1 in the left arm of the outer channel now opposes the current flowing there due to the outer channel bias of 1 V, the net outer channel current will be zero at some inner channel voltage. This inner channel voltage is defined as the “zero crossing” voltage (VZC) and has been used to differentiate the effect of VOCs (see discussion below).

FIG. 2.

(a) Breakdown I-V characteristic of a monolithic tip dual channel microcantilever heater (MDC-MH) when 1 V is applied to the outer channel. (b) Schematic of the currents in a MDC-MH at pre-breakdown and post-breakdown voltage biases.

FIG. 2.

(a) Breakdown I-V characteristic of a monolithic tip dual channel microcantilever heater (MDC-MH) when 1 V is applied to the outer channel. (b) Schematic of the currents in a MDC-MH at pre-breakdown and post-breakdown voltage biases.

Close modal

For accurate estimation of the critical breakdown field (EB), as influenced by the geometrical and thermal effects, the outer channel terminals should be connected to ground, shown in the inset of Fig. 3. Before breakdown, the measured outer channel current is typically in tens of nA, and we define the breakdown voltage as one where this pre-breakdown current increases by 100 times (reaching 1 μA). The average inner channel breakdown voltage over the tested devices was ∼39 V. Figure 3 shows a breakdown characteristic for a single device. Assuming that the breakdown occurred where the GaN thickness is minimum at 3 μm and VB is 39/2 or 19.5 V, EB for the inter-channel GaN layer is (19.5V/3 μm) = 6.5 kV/cm, this is a factor of 50 less than bulk GaN, 3.3 MV/cm.4,5 It should be noted that due to high tip resistance and gradual change in the gap between channels near the tip, the breakdown voltage is likely higher than half the applied bias and the breakdown field is underestimated to some extent.

FIG. 3.

Breakdown current in the outer channel with 0 V applied to the outer channel with line indicating the crossing of the 1 μA threshold and the breakdown voltage indicated by VB.

FIG. 3.

Breakdown current in the outer channel with 0 V applied to the outer channel with line indicating the crossing of the 1 μA threshold and the breakdown voltage indicated by VB.

Close modal

Although EB of bulk GaN is reported as 3.3 MV/cm,4,5 this number can change significantly for stressed thin films, as is the case for GaN cantilevers after removal from Si substrates.24 Indeed, Wang showed that for thin film GaN (2 μm) removed from the substrate, EB can be reduced by orders of magnitude, down to a few kV/cm. In addition, an elevated temperature profile can also reduce it significantly. Wang discovered that the breakdown voltage was reduced by more than 40% at 200 °C, as compared to 30 °C.25 Elevated temperatures can cause multiple physical changes that may lead to early breakdown including hot electron effects, lowering of bandgap, and defect/trap assisted tunneling.26–28 In previous reports on the AlGaN/GaN TMHs and MDC-MHs, the temperature of the tip was shown to exceed 300 °C,12,13 which can contribute to significant lowering of EB.

To investigate the effect of VOCs on the breakdown characteristics of the MDC-MHs, three different VOCs (isopropanol, methanol, and acetone) with characteristically distinct latent heats of evaporation were used in our studies. Saturated vapor (at 25 °C) of each VOC was diluted with dry N2 to a concentration of 25 000ppm using two different MFCs (mass flow controllers) at appropriate flow settings.12 The inner channel voltage of the device was swept from 0 to 40 V during the vapor flow, while the outer channel voltage was maintained at 1 V (see earlier discussion). Figure 4 shows the variation in outer channel current with inner channel voltage for all three VOCs (nominally at 25 000 ppm) in comparison to the control characteristics, for a typical cantilever. At voltages below the breakdown voltage, the control and VOC curves (in Fig. 4) are very similar, indicating that the reduction in current is due to heating and not due to the presence of VOCs. Once the swept voltage reaches ∼25 V, the VOC curves begin to differentiate from the control, but not enough from each other. Therefore, to have a proper definition to distinguish VOCs, we utilize VZC (see earlier discussion). Isopropanol exhibits the lowest VZC of 33.3 V, while acetone shows the highest VZC of 37 V, with methanol in between at 33.7 V (see Fig. 4 inset). As reported earlier, the latent heat of evaporation of a VOC can play a significant role in altering the thermal profile of a TMH,12–14 which can result in a change in the breakdown electric field of GaN. We have plotted the outer channel currents, showing the Vzc values for isopropanol for six different samples, which is shown in Fig. S1 in the supplementary material. We find that Vzc varies from 27.6 to 34.6 V for the set of samples tested, which can be attributed to potential variation in process parameters as well as AlGaN thickness, as well as charge density and mobility variations due to non-uniformity in the CVD growth process. We would also like to note there that the microcantilevers used in this work went through several hundred cycles of testing with the VOCs, over a period of several months without any significant degradation in their characteristics.

FIG. 4.

Current in the outer channel while in the inner channel is swept past breakdown in the presence of either neutral gas, acetone, methanol, or isopropanol. Inset: Close-up emphasizing the different zero-crossing point voltages of the different VOCs. Inset Table: Latent heat of evaporation (ΔHvap) of relevant VOCs.

FIG. 4.

Current in the outer channel while in the inner channel is swept past breakdown in the presence of either neutral gas, acetone, methanol, or isopropanol. Inset: Close-up emphasizing the different zero-crossing point voltages of the different VOCs. Inset Table: Latent heat of evaporation (ΔHvap) of relevant VOCs.

Close modal

We expect breakdown to happen on the high voltage side shifted from the apex, where a combination of a high electric field and temperature makes it the easiest for the GaN layer to break down. With the introduction of VOCs, we expect the breakdown location to move further away from the apex as the presence of VOCs has been shown to lower the apex temperature and increase the temperature in the adjacent region.12 To check this hypothesis, we biased the inner channel at an appropriate voltage (slightly below VZC of the control from Fig. 4), such that the introduction of a VOC would result in breakdown.

Isopropanol was chosen as the test VOC to initiate breakdown as it has the largest effect on shifting the breakdown voltage among the vapors tested (due to it having the largest latent heat of evaporation). With an inner channel biased at 38 V and an outer channel at 1 V, the magnitude of the currents through all the terminals was recorded as the VOC (25 000 ppm isopropanol diluted in pure N2) flow was started (∼15 s) and stopped (∼40 s), as shown in Fig. 5(a). Prior to breakdown (t < 18 s), the input and output currents (Ii1 and Ii2, respectively) at the inner channel are very close in magnitude, indicating that the inter-channel leakage current is very low, as expected. As breakdown occurs (t > 20 s), the magnitude of Ii1 and Ii2 starts to differ very significantly, with the former increasing in magnitude and the latter reducing due to the introduction of the two outer channel paths to ground. It is interesting to note that the current through the 1 V terminal became approximately 5 times that of the grounded terminal [shown by Io1 and Io2 in Fig. 5(a)]. Therefore, breakdown must occur at the left shoulder of the cantilever and not at the apex, which consequently attracted a lot more current to the 1 V terminal compared to the 0 V terminal, as the 1 V terminal offered a much lower resistance path for current flow. This clearly confirms our earlier postulation on the location of the breakdown point. Further verification of the location was obtained from the infrared microscopic image (taken using an FLIR SC1000 microscope) of the TMH before and after introduction of VOC. The hottest point shifted from the apex of the cantilever prior to breakdown, to significantly down the arm biased at high voltage (away from the apex) at breakdown caused by VOC exposure. The results are shown in supplementary material Fig. S2.

FIG. 5.

(a) Current at all four terminals of a monolithic dual channel microcantilever heater (MDC-MH) when isopropanol is introduced at 15 s to induce breakdown until the system is purged with clean air at 40 s. (b) Amplification factor of outer channel current through the 1 V terminal (Io1 where Io1 − 0 is the initial current pre-breakdown). The inset shows the measurement schematic of four terminal constant bias experiment.

FIG. 5.

(a) Current at all four terminals of a monolithic dual channel microcantilever heater (MDC-MH) when isopropanol is introduced at 15 s to induce breakdown until the system is purged with clean air at 40 s. (b) Amplification factor of outer channel current through the 1 V terminal (Io1 where Io1 − 0 is the initial current pre-breakdown). The inset shows the measurement schematic of four terminal constant bias experiment.

Close modal

The occurrence of inter-channel breakdown, with its consequent dramatic increase in breakdown current, offers interesting sensing possibilities. In previous sensing schemes using dual channel cantilevers, where breakdown did not occur, the sensor current merely changed by <5% in the presence of VOCs.14 This made it very challenging to detect a low VOC concentration, especially in the sub-ppm range. In this work, the current in the outer channel changes dramatically at breakdown offering the possibility to detect much lower concentration of VOCs. Figure 5(b) shows the variation in inter-channel current as the VOC is introduced into the sensing chamber, as well as when the VOC flow is replaced with purging N2 flow. The pre-breakdown current (Ibr,0 = Ii1 – Ii2) is ∼6.5 μA, while the post-breakdown current increases dramatically to ∼500 μA, which is a factor of 80 or 8000% higher. The inset of Fig. 5(b) shows the amplification factor calculated as the ratio of the current at a given time to the pre-breakdown current of 6.5 μA. Using the current change as the sensing parameter, a much lower level of VOC may be detected (potentially in the low ppb level) than the 2000 ppm formerly demonstrated with <5% change in resistance.12 

In conclusion, we have investigated the reversible breakdown behavior of dual channel III-Nitride microcantilevers as influenced by the presence of multiple VOCs. The critical breakdown field of the inter channel GaN was found to be dramatically lower than bulk GaN due to dimensional and thermal effects, which got further lowered in the presence of VOCs, due to the change in the thermal and electric profile near the cantilever apex. Different latent heats of evaporation of the VOCs resulted in varying magnitudes of breakdown voltage reduction, which can be used to preform selective detection of VOCs. Using the inter-channel current as a sensing parameter, which increased dramatically at breakdown triggered by VOC exposure, can lead to their highly sensitive detection, to much lower levels than was previously demonstrated using these types of dual channel cantilevers.

See supplementary material for VZC variation among tested samples and for IR microscopy images.

This work was financially supported by the National Science Foundation through Grant Nos. # CBET-1606882, IIP-1602006, and ECCS-1809891. The fabrication of the microcantilevers was performed at the Georgia Tech Institute for Electronics and Nanotechnology, a member of the National Nanotechnology Coordinated Infrastructure, which was supported by the National Science Foundation (Grant No. ECCS-1542174). We also acknowledge Mr. Robert Irvine and Mr. Harrison Eggers for help with the experimental set up.

1.
T. J.
Flack
,
B. N.
Pushpakaran
, and
S. B.
Bayne
, “
GaN technology for power electronic applications: A review
,”
J. Electron. Mater.
45
(
6
),
2673
(
2016
).
2.
G.
Simin
,
F.
Jahan
,
J.
Yang
,
M.
Gaevski
,
X.
Hu
,
J.
Deng
, and
M.
Shur
, “
III-nitride microwave control devices and ICs
,”
Semicond. Sci. Technol.
28
(
7
),
74008
(
2013
).
3.
M.
Stutzmann
,
G.
Steinhoff
,
M.
Eickhoff
,
O.
Ambacher
,
C. E.
Nebel
,
J.
Schalwig
, and
G.
Müller
, “
GaN-based heterostructures for sensor applications
,”
Diamond Related Mater.
11
(
3
),
886
(
2002
).
4.
M.
Rais-Zadeh
,
V. J.
Gokhale
,
A.
Ansari
,
M.
Faucher
,
D.
Theron
,
Y.
Cordier
, and
L.
Buchaillot
, “
Gallium nitride as an electromechanical material
,”
J. Microelectromech. Syst.
23
(
6
),
1252
(
2014
).
5.
A.
Dadgar
, “
Sixteen years GaN on Si
,”
Phys. Status Solidi B
252
(
5
),
1063
(
2015
).
6.
Y.
Alifragis
,
A.
Georgakilas
,
G.
Konstantinidis
,
E.
Iliopoulos
,
A.
Kostopoulos
, and
N. A.
Chaniotakis
, “
Response to anions of AlGaN∕GaN high-electron-mobility transistors
,”
Appl. Phys. Lett.
87
(
25
),
253507
(
2005
).
7.
P.
Offermans
and
R.
Vitushinsky
, “
Detection with AlGaN/GaN 2DEG channels for air quality monitoring
,”
IEEE Sens. J.
13
(
8
),
2823
(
2013
).
8.
A.
Talukdar
,
M.
Faheem Khan
,
D.
Lee
,
S.
Kim
,
T.
Thundat
, and
G.
Koley
, “
Femtoscale displacement transduction using piezotransistive GaN microcantilever
,”
Nat. Commun.
6
,
7885
(
2015
).
9.
F.
Bayram
,
D.
Khan
,
H.
Li
,
M. M.
Hossain
, and
G.
Koley
, “
Piezotransistive GaN microcantilevers based surface work function measurements
,”
Jpn J. Appl. Phys. Part 1
57
(
4
),
40301
(
2018
).
10.
D.
Khan
,
F.
Bayram
,
D.
Gajula
,
A.
Talukdar
,
H.
Li
, and
G.
Koley
, “
Plasmonic amplification of photoacoustic waves detected using piezotransistive GaN microcantilevers
,”
Appl. Phys. Lett.
111
(
6
),
062102
(
2017
).
11.
P.
Ray
,
S.
Pandey
, S.
, and
R.
Ramgopal
, “
V. Development of graphene nanoplatelet embedded polymer microcantilever for vapour phase explosive detection applications
,”
J. Appl. Phys.
116
(
12
),
124902
(
2014
).
12.
I.
Jahangir
,
E. B.
Quddus
, and
G.
Koley
, “
Unique detection of organic vapors below their auto-ignition temperature using III–V nitride based triangular microcantilever heater
,”
Sens. Actuators B: Chem.
222
,
459
(
2016
).
13.
I.
Jahangir
and
G.
Koley
,“
Modeling the performance limits of novel microcantilever heaters for volatile organic compound detection
,”
J. Micromech. Microeng.
27
(
1
),
015024
(
2017
).
14.
I.
Jahangir
and
G.
Koley
, “
Dual-channel microcantilever heaters for volatile organic compound detection and mixture analysis
,”
Sci. Rep.
6
(
1
),
28735
(
2016
).
15.
W. P.
King
,
T. W.
Kenny
,
K. E.
Goodson
,
G. L. W.
Cross
,
M.
Despont
,
U. T.
Durig
, and
P.
Vettiger
, “
Design of atomic force microscope cantilevers for combined thermomechanical writing and thermal reading in array operation
,”
J. Microelectromech. Syst.
11
(
6
),
765
(
2002
).
16.
J.
Lee
,
T.
Beechem
,
T. L.
Wright
,
B. A.
Nelson
,
S.
Graham
, and
W. P.
King
, “
Electrical, thermal, and mechanical characterization of silicon microcantilever heaters
,”
J. Microelectromech. Syst.
15
(
6
),
1644
(
2006
).
17.
Z.
Dai
,
E. A.
Corbin
, and
W. P.
King
, “
A microcantilever heater–thermometer with a thermal isolation layer for making thermal nanotopography measurements
,”
Nanotechnology
21
(
5
),
055503
(
2010
).
18.
J.
Lee
and
W. P.
King
,
J. Microelectromech. Syst.
17
(
2
),
432
(
2008
).
19.
J. E. C.
Lerner
,
T.
Kohajda
,
M. E.
Aguilar
,
L. A.
Massolo
,
E. Y.
Sánchez
,
A. A.
Porta
,
P.
Opitz
,
G.
Wichmann
,
O.
Herbarth
, and
A.
Mueller
, “
Improvement of health risk factors after reduction of VOC concentrations in industrial and urban areas
,”
Environ. Sci. Pollut. Res.
21
,
9676
(
2014
).
20.
H.
Guo
,
S.
Lee
,
L.
Chan
, and
W.
Li
, “
Risk assessment of exposure to volatile organic compounds in different indoor environments
,”
Environ. Res.
94
(
1
),
57
(
2004
).
21.
B.
Szulczyński
and
J.
Gębicki
, “
Currently commercially available chemical sensors employed for detection of volatile organic compounds in outdoor and indoor air
,”
Environments
4
(
1
),
21
(
2017
).
22.
Y.
Zhou
,
J.
Yu
,
Z.
Yan
,
C.
Zhang
,
Y.
Xie
,
L.
Ma
, and
F.
Li
, “
Application of portable gas chromatography-photo ionization detector combined with headspace sampling for field analysis of benzene, toluene, ethylbenzene, and xylene in soils
,”
Environ. Monit. Assess.
185
(
4
),
3037
(
2013
).
23.
P.
Boeker
, “
On “electronic nose” methodology
,”
Sens. Actuators B: Chem.
204
,
2
(
2014
).
24.
T.
Zimmermann
,
M.
Neuburger
,
P.
Benkart
,
F. J.
Hernandez-Guillen
,
C.
Pietzka
,
M.
Kunze
,
I.
Daumiller
,
A.
Dadgar
,
A.
Krost
, and
E.
Kohn
, “
Piezoelectric GaN sensor structures
,”
IEEE Electron Device Lett.
27
,
309
(
2006
).
25.
T.
Wang
,
B.
Wang
,
A.
Haque
,
M.
Snure
,
E.
Heller
, and
N.
Glavin
, “
Mechanical stress effects on electrical breakdown of freestanding GaN thin films
,”
Microelectron. Reliab.
81
,
181
(
2018
).
26.
A.
Hu
,
X.
Yang
,
J.
Cheng
,
L.
Guo
,
J.
Zhang
,
Y.
Feng
, and
B.
Shen
, “
Hot electron assisted vertical leakage/breakdown in AlGaN/GaN heterostructures on Si substrates
,”
Superlattices Microstruct.
107
,
240
(
2017
).
27.
K. B.
Nam
,
J.
Li
,
J. Y.
Lin
, and
H. X.
Jiang
, “
Optical properties of AlN and GaN in elevated temperatures
,”
Appl. Phys. Lett.
85
(
16
),
3489
(
2004
).
28.
J.-I.
Chyi
,
C.-M.
Lee
,
C.-C.
Chuo
,
X.
Cao
,
G.
Dang
,
A.
Zhang
, and
R.
Wilson
, “
Temperature dependence of GaN high breakdown voltage diode rectifiers
,”
Solid-State Electron.
44
(
4
),
613
(
2000
).

Supplementary Material