IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 43, NO.

12, DECEMBER 1996 2305

This feature has been predicted before in a non-Monte-Carlo model Reverse Leakage Current Calculations
by Plimmer et al. [9], though the feature here is less pronounced. The for SIC Schottky Contacts
reason for the overshoot is due to a population inversion in the carrier
energy distribution after traversing the dead space (panel 2c). This J. Crofton and S. Sriram
in turn causes a larger than normal loss of energy due to the higher
impact ionization rate which in turn reduces the impact ionization
Abstract-Reverse leakage current calculations as a function of Schot-
rate below its equilibrium value, causing the trough. tky barrier height and temperature have been performed for metal gates
In conclusion, we have shown that a simple analytical band stucture on n-type 6H-SiC. These ca~cu~ationswere performed using a WKB
may be used to describe accurately high field transport phenomena evaluation of the tunneling probability through a reverse biased Schottky
in a Monte Carlo model of GaAs. Such a model calculates accurate barrier and numerically integrating over all energies to find the reverse
drift velocities, ballistic overshoot, mean energies, impact ionization current density. This method is shown to yield much better agreement
with previously published reverse leakage currents on 6H-Sic than can
and quantities associated with impact ionization, i.e., dead space, and be obtained using thermionic emission theory.
spatial oscillations in ionization rate for both electrons and holes.
I. INTRODUCTION
REFERENCES Silicon Carbide devices are currently emerging as important can-
didates for RF power amplification [l], low-frequency power control
B. K. Ridley, “Soft-Threshold lucky drift theory of impact ionization in
semiconductors,” Semicond. Sci. Technol., vol. 2, p. 116, 1987. devices [2], and for high-temperature and radiation hard applications
D. C. Herbert, “Breakdown voltage in ultra thin PIN diodes,” Semicond. due to its unique combination of wide bandgap, high breakdown
Sci. Technol.. vol. 8. U. 1993. 1993. strength. high thermal conductivitv. and high saturated electron
P. G. Scrobohaci andLT.-W Tang, “Modeling of the hot-electron sub- velocity. A -variety of devices ha& been s;ccessfully fabricated
population and its application to impact ionization in submicron silicon
on both 6-H and 4-H polytypes of sic including Schottky diodes,
devices-Part I: Transport equations,” IEEE Trans. Electron Devices,
vol. 41. U. 1197. 1994. MESFET’s, MOSFET’s, and p-n junctions. The Schottky contact is
M. V. &hetti, ‘‘Monte Carlo simulation of transport in techno- the most important active element in MESFET’S and also in rectifiers
logically significant semiconductors of the diamond and zinc-blend for high-sDeed Dower control devices. The advantages of Schottkv

zinc-blend structures,” Phys. Rev., vol. 141, p. 789, 1966. junction devices. However, due to the high electric fields normally
C. Jacoboni and P. Lugli, The Monte Carlo Method For Semiconductor encountered in S i c devices, the reverse leakage current of Schottky
Device Simulation, S. Selberherr, Ed. New York Springer-Verlag, diodes can be significantly enhanced prior to junction breakdown due
1989, ch. 1-3.
F. Osaka, T. Mikawa, and 0. Wada, “Analysis of impact ionization to mechanisms such as thermionic field emission and field emission.
phenomena in InP by Monte Carlo simulation,” Jpn. J. Appl. Phys., vol. The leakage current is undesirable and is therefore important to
25, p. 394, 1986. quantitatively determine these effects in order to understand the
M. Ershov and V. Ryzhii, “Procedure for fitting Monte Carlo calculated present limitations of these devices and to obtain further improvement
impact ionization coefficients to experiment,” J. Appl. Phys. vol 76, p.
1672, 1994. in the future by proper device optimization.
S. A. Plimmer, J. P. R. David, D. C. Herbert, G. J. Rees, P. A. Houston, The purpose of this paper is to point out that for normal operating
P. N. Robson, R. Grey, T. W. Lee, A. W. Higgs, and D. R. Wight, conditions, the high electric fields encountered in S i c and other
“Investigation of impact ionization in thin GaAs diodes,” IEEE Trans. wide bandgap materials leads to significant tunneling through the
Electron Devices, vol. 43, pp. 1066-1072, July 1996. Schottky barrier thus increasing the leakage current by many orders
G. E. Bnlman, V. M. Robbins, and E. Stillman, “The determination
of impact ionization coefficients in (100) GaAs using avalanche noise of magnitude over that predicted by simple thermionic emission
and photocurrent multiplication measurements,” IEEE Trans. Electron calculations. It is also shown that the calculation of the tunneling
Devices, vol. ED-32, p. 2454, 1985. current can be easily accomplished by using a numerical technique
rather than relying on the complicated tunneling expressions derived
from field and thermionic field emission theories.

11. FIELDAND THERMIONIC FIELDEMISSION

The formalism that determines the tunneling current density be-
tween two dissimilar materials for high electric fields known as field
emission was first introduced by Stratton in 1962 [3].The theory is
Manuscript received October 11, 1995; revised August 7, 1996. The review
of this brief was arranged by Editor T. P. Chow. This work was supported
in part by the Department of the Air Force, Wright Patterson AFB, under
Contract F33615-92-C-5912 and Contract Monitors T. Kensky, B. Mitchell,
and J. King, and by Dr. R. Askew at the NASA Center for Commercial
Development (CCD) at the Auburn University Space Power Institute.
J. Crofton is with the Department of Physics and Astronomy, Murray State
University, Murray, KY 42071 USA.
S. Sriram is with the Northrop Grumman Science and Technology Center,
Pittsburgh, PA 15235 USA.
Publisher Item Identifier S 001 8-9383(96)08650-9.

0018-9383/96$05.00 0 1996 IEEE

Authorized licensed use limited to: NXP Semiconductors. Downloaded on December 20,2020 at 07:15:56 UTC from IEEE Xplore. Restrictions apply.
2306 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 43, NO 12, DECEMBER 1996

very useful in that it allows one to see which parameters have the most IO’ r 1 I I 1 I I I I
impact on the tunneling current, but since it was derived by expanding
the tunneling probability in a Taylor series about the Fermi energy of
the source material, it is sometimes difficult to use and has conditions
that must be met in order to ensure accuracy. The subsequent theory “E IO“
T = 250K
i
///-==j
by Padovani and Stratton that describes tunneling at energies greater $
v
10-3
/--- /--/

3 10-4 N, = 1x1017 cm-3 ,

;La-
than the Fermi energy of the source material, known as thermionic
field emission, also utilizes a Taylor expansion about some energy .$ 10-5
greater than E f and therefore also has restrictions that must be met to E 104
ensure accuracy [4]. Unfortunately, the field emission and thermionic E 10-7
.e
field emission theories are distinct and do not reduce to one another g IO-@
as the fields and energies approach the correct values. The transition
between field emission and thermionic field emission is given by a
2 10-9
10-10
v /’ I
constant cl defined by Padovani and Stratton. Field emission occurs L

for l / c l > IcT and thermionic field emission for l / c l < kT. Near the 3 10-1’ thermionic field emission
field emission
intermediate regime where l / c l S IcT, neither field nor thermionic
field emission accurately describes the conduction process.
5 10-13
o!
10-14
CALCULATIONS OF REVERSE LEAKAGE
111. NUMERICAL CURRENT
10-16
The approach taken here was to simply calculate the tunneling 40 60 80 100 120 140 160 180 200
probability through a parabolic Schottky barrier for a wide range of
reverse bias (Volts)
electric fields, temperatures, and barrier heights [ 5 ] . This approach
has successfully been used to calculate specific contact resistance Fig 1 A comparison of the calculated reverse current density for a 1 eV
as a function of doping [6], [7] and as a function of temperature barrier Schottky gate on 1 x 1017 cmP3 n-type 6H-SIC at 250 K using field
[8]. In both cases, relatively good agreement between theory and emission, thermionic field emission, and a numerical evaluation No image
experiment was obtained. The tunneling probability as a function of force effects are included.
energy through a reverse biased Schottky barrier can be approximated
using the WKB theory as the components of the effective mass tensor, m,, m,, and m, and

T ( E ) = exp ( - 2 1;’ k ( z ) (12)

the direction cosines I , m , and n of the unit normal to the interface
relative to the principal axes of the constant energy ellipsoid. The
transport effective mass m* can be written
for O < E < E b - q V + E f
m* = (Pm,m, + m2m,mx+ n2mxm,)1’2.
?-(E) = 1 for E > Eb - qV + E f
The effective masses used to calculate the tunneling mass are from
where 2 1 and x:, are the classical turning points, E f is the semi- reference [ 9 ] . For conduction along the c-axis of 6H-SiC, the calcu-
conductor fermi energy, and k ( z ) is the free electron wave vector lated tunneling mass using the above equation is 0.92mo. This mass
calculated using a parabolic potential obtained from the depletion must then be multiplied by 6 to account for the number of ellipsoidal
approximation. The first turning point 2 1 is 0 while the second is surfaces in the first Brillouin zone.
given by A comparison between thermionic field emission, field emission,
~2 =d m [ d E b - qv i-
Ef - a]. and this numerical calculation at 250 K for a 1-eV barrier on
1 x l o i 7 cm-3 n-type 6H-Sic is ,shown in Fig. 1. The doping
After integrating, the tunneling probability for 0 < E < Eb - qV+ E f and temperature were chosen so that the transition between field
can be written as
T ( E ) = exp [ t E { E h( E;+ c)
El -E
and thermionic field emission would be clear. These calculations do
not include any image force lowering of the barrier so that they
may be compared with the theory of Padovani and Stratton which
does not include the image force effect. The transition between field
and thermionic field emission occurs at approximately 130 V reverse
bias. At this bias, l / c l is approximately equal to kT. The numerical
calculation agrees with the appropriate expression, that is with field
where enlission results for l / c l > k T , (a reverse bias greater than 130
V) and with the thermionic field emission results for l / c l < IcT (a
reverse bias less than 130 V). The numerical approach is valid for
The tunneling probability is then inserted into the following equation all fields and temperatures considered including the transition region
for the current density and numerically integrated between field and thermionic field emission.
The calculations of reverse current density also require the flat band
barrier height which is the barrier height at the metal-semiconductor
1 + exp(Ef - E / k T ) interface with zero electric field. It is the flat band barrier height which
is measured using capacitance-voltage techniques [ 101. The effective
. (1 + exp(Ef - E
- qV/kT barrier height is the actual barrier that the reverse biased contact
Whether a numerical calculation is made using the method de- sees while biased with current flowing. Although there are several
scribed above or the field or thermionic field emission equations terms which cause the effective barrier height to differ from the flat
are used, the appropriate transport effective mass is required. This band barrier height, the most significant generally is the image force
mass can be calculated for conduction in a given direction using lowering term [lo]. The image force effect which was not included

Authorized licensed use limited to: NXP Semiconductors. Downloaded on December 20,2020 at 07:15:56 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 43, NO. 12, DECEMBER 1996 2307

the measured data, the present result clearly shows the importance of
tunneling currents during reverse bias operation of Schottky diodes in
high field materials such as Sic. In this context, we speculate that the
discrepancy between the theory and measured data at 300 K (Fig. 2)
arises partly due to the presence of a significant edge leakage current
caused by the higher electric fields at the edges. Comparison of the
theoretical calculations in Fig. 2 for the two temperatures shows that
when the voltage is increased from 50 V to 400 V, the leakage current
increase by a factor of 3 x lo5 at 300 K and shows a much smaller
increase of 3 x lo3 at 373 K. Therefore, any leakage currents that
arise due to high fields at the edges will have a greater significance at
lower temperatures thus possibly leading to the discrepancy between
0 Bhatnagar at T = 300K
Bhatnagar at T = 373K theory and measured data at 300 K in Fig. 2.
calculated for T = 300K
calculated for T = 373K V. CONCLUSIONS
-.
...... thermionic emission 300K
thermionic emission 373K When calculating reverse leakage currents of Schottky diodes on
Y I 1
materials such as Sic, the thermionic emission theory will be many
orders of magnitude in error since this theory does not take into
account any quantum mechanical tunneling through the Schottky
barrier. A simple WJSB approximation of the tunneling probability
50 100 150 200 250 300 350 400 through a reverse biased Schottky barrier and a subsequent numerical
reverse bias (Volts) evaluation of an integral can be used to calculate the reverse leakage
current density for a Schottky gate. This method has the advantage
Fig. 2. Comparison of calculated and experimental reverse leakage currents
over the formalism of Padovani and Stratton in that it is applicable
for Pt-Sic Schottky diodes on n-type 6H-Sic. Experimental results are from
Bhatnagar [ 111. All calculated results were calculated using the image force over a wider range of biases and temperature than either the field
lowering of the harrier height. This accounts for the mild bias dependence on or thermionic field emission equations for the current density and
the thermionic emission current. can be used to calculate leakage currents in the transition regions
between field and thermionic field emission. This method has been
in the calculations of Fig. 1, has been included in all subsequent shown to be in much better agreement with previously published
numerical calculations by simply subtracting the image force term, reverse leakage data on 6H-Sic than can be obtained using thermionic
AEb, from the flat band barrier height. The image force lowering of emission theory.
the barrier is important for S i c Schottky gates due to the high electric
fields at the metal-semiconductor interface. Note that the image force REFERENCES
lowering of the barrier increases with increasing reverse bias since
the reverse bias increases the semiconductor electric field. The image H. Morkoc, S . Strite, G. B. Gao, M. E. Lin, B. Sverdlov, and M.
force lowering of the barrier height can be written as Burns, “Large-band-gap Sic, 111-V nitride, and 11-VI ZnSe-based semi-
conductor device technologies,” J. Appl. Phys., vol. 76, no. 3, p. 1363,
1994.
M. Bhatnagar and B. J. Baliga, “Comparison of 6H-SiC, 3C-SiC, and Si
for power devices,” ZEEE Trans. Electron Devices, vol. 40, pp. 645-655,
where E,,, is the maximum electric field at the metal-semiconductor Apr. 1993.
R. Stratton, “Volt-Currentcharacteristics for tunneling through insulating
interface. Note that the image force term is not zero at zero bias since
films,” J. Phys. Chem. Solids, vol. 23, pp. 1177-1190, 1962.
the electric field at zero bias is not zero. F. A. Padovani and R. Stratton, “Field and thermionic-field emission in
Schottky harriers,” Solid State Electron., vol. 9, pp. 695-707, 1966.
Iv. COMPARISON OF CALCULATED J. Crofton and P. A. Barnes, “A comparison of one-, two-, and three-
hand calculations of contact resistance for a GaAs ohmic contact using
RESULTSWITH EXPERIMENTAL DATA the Wentzel-Kramers-Brillouin approximation and a numerical solution
Fig. 2 shows a comparison of calculated reverse leakage currents to the Schrodinger equation,” J. Appl. Phys., vol. 69, no. 11, pp.
over a 350 V range with experimental results obtained with Pt 7660-7663, 1991.
J. Crofton, P. A. Barnes, J. R. Williams, and J. A. Edmond, “Contact
Schottky diodes on n-type 6H-Sic previously published by Bhatnagar resistance measurements on p-type 6H-SiC,” Appl. Phys. Lett., vol. 62,
[ l l ] . The voltages less than 50 V were omitted since for low voltages, no. 4, pp. 384-386, 1993.
the currents are very small, difficult to measure, and of little interest. J. Crofton, E. D. Luckowski, J. R. Williams, T. Isaacs-Smith, M. J.
The flatband barrier height used in the calculations was 1.1 eV, since Bozack, and R. Siergiej, “Specific contact resistance as a function of
doping for n-type 4H and 6H-SiC,” in Silicon Carbide and Related
this was the barrier height Bhatnagar measured using capacitance- Materials, VI, no. 142, ch. 3, Inst. Phys. Con$ Ser. IOP Puhlishing
voltage techniques and thus was a flat band barrier height not subject Ltd., 1996.
to any image force lowering. J. Crofton, J. R. Williams, M. J. Bozack, and P. A. Barnes, “A TiW
It can be seen in Fig. 2 that the numerical technique of the high-temperature Ohmic contact to n-type 6H-SiC,” Silicon Carbide and
present work shows good agreement with the measured data at 373 Related Materials V. New York: Springer-Verlag, 1994, pp. 719-722.
N. T. Son, 0. Kordina, A. 0. Konstantinov, W. M. Chen, E. Sorman,
K, and also correctly predicts increased reverse leakage currents at B. Monemar, and E. Janzen, “Electron effective masses and mobilities
higher reverse bias voltages for both 300 K and 373 K, as observed in high-purity 6H-Sic chemical vapor deposition layers,” Appl. Phys.
experimentally. In contrast, it may be noted that thermionic emission Lett., vol. 65, no. 25, pp. 3209-3211, 1994.
theory predicts a much smaller voltage dependence and is in error E. H. Rhoderick and R. H. Williams, Metal-Semiconductor Contacts,
2nd ed. Oxford: Clarendon Press, 1988.
by more than three orders of magnitude at 400 V. Even though more M. Bhatnagar, P. K. McLarty, and B. J. Baliga, “Sic high-voltage (400
detailed experimental work (including guard rings to minimize edge V) Schottky barrier diodes,” ZEEE Electron Device Lett., vol. 13, no.
leakage currents) will be required to obtain better fit of the theory to 10, pp. 501-503, 1992.

Authorized licensed use limited to: NXP Semiconductors. Downloaded on December 20,2020 at 07:15:56 UTC from IEEE Xplore. Restrictions apply.