Terahertz Radiation from Gallium Phosphide …
51
Table 1 Optimized design parameters
S. No. f d (THz) W n (nm) W p (nm) N D (×10 23
m −3 )
N A (×10 23
m −3 )
J 0 (×10 8 A
m −2 )
D j (µm)
1
0.10
1125.0
1125.0
0.300
0.400
2.50
35.0
2
0.15
800.0
800.0
0.480
0.680
5.10
25.0
3
0.22
510.0
510.0
0.800
1.150
9.40
20.0
4
0.30
366.0
366.0
1.250
1.510
15.00
15.0
5
0.50
224.0
224.0
1.950
2.270
22.00
10.0
6
1.00
109.0
109.0
3.150
3.850
32.50
5.0
2 Structure and Fabrication Steps
Schematic view of the double-drift GaP IMPATT diode is illustrated in Fig. 1. The
thicknesses of different semiconductor layers are defined in that figure. Those thicknesses, corresponding doping densities (N D , N A , N n+ and N p+) , and bias current
density (J 0 ) of the diodes designed to operate at 0.1, 0.15, 0.20, 0.30, 0.50, and
1.00 THz frequencies are optimized [14]. The effective diameters (D j ) of the circular
cross-sectional diodes are optimized subject to obtain stable thermal operation by
following the thermal design method reported elsewhere [15]. Table 1 provides the
optimized design parameters. The doping density of n
+ -GaP substrate layer is kept
around 5.0 × 10
25 m
−3 and the doping density of p
+ -GaP contact layer is kept between
2.0 × 10
25
− 3.0 × 10
25 m
−3 in all the diodes under consideration.
The n, p and p
+ -layers may be successively developed on n
+ -GaP substrate/wafer
(<100>) by using either molecular beam epitaxy (MBE) technique [13]. The n-type
and p-type doping may be done by using Si and Zn dopants, respectively. Appropriate
etching and polishing techniques must be used to thin down the n
+ -GaP substrate
in the order of 2–3 µm in order to reduce the parasitic series resistance. Sputter
deposition of Be/Au and Au/Ge metal systems on p
+ and n
+ -surfaces can be used to
realize the electrodes [16].
3 Simulation
Well-established non-sinusoidal voltage-excited large-signal (LS) simulation
technique-based drift-diffusion model is used to evaluate the THz performance of the
diodes under consideration [14, 15]. The one-dimensional diode model is shown in
Fig. 2; it is used for microscopic simulation of the device. The justification behind the
use of 1D model instead of 2D or 3D models has already been discussed elsewhere
[14, 15]. The external circuit model containing the frequency dependent equivalent
circuit of the diode, inductive load (waveguide or antenna), and biasing current has
been shown in Fig. 3. From 3, it can be understood that the equivalent circuit of
diode contains parallel connected negative conductance (G) and susceptance (B),
51
Table 1 Optimized design parameters
S. No. f d (THz) W n (nm) W p (nm) N D (×10 23
m −3 )
N A (×10 23
m −3 )
J 0 (×10 8 A
m −2 )
D j (µm)
1
0.10
1125.0
1125.0
0.300
0.400
2.50
35.0
2
0.15
800.0
800.0
0.480
0.680
5.10
25.0
3
0.22
510.0
510.0
0.800
1.150
9.40
20.0
4
0.30
366.0
366.0
1.250
1.510
15.00
15.0
5
0.50
224.0
224.0
1.950
2.270
22.00
10.0
6
1.00
109.0
109.0
3.150
3.850
32.50
5.0
2 Structure and Fabrication Steps
Schematic view of the double-drift GaP IMPATT diode is illustrated in Fig. 1. The
thicknesses of different semiconductor layers are defined in that figure. Those thicknesses, corresponding doping densities (N D , N A , N n+ and N p+) , and bias current
density (J 0 ) of the diodes designed to operate at 0.1, 0.15, 0.20, 0.30, 0.50, and
1.00 THz frequencies are optimized [14]. The effective diameters (D j ) of the circular
cross-sectional diodes are optimized subject to obtain stable thermal operation by
following the thermal design method reported elsewhere [15]. Table 1 provides the
optimized design parameters. The doping density of n
+ -GaP substrate layer is kept
around 5.0 × 10
25 m
−3 and the doping density of p
+ -GaP contact layer is kept between
2.0 × 10
25
− 3.0 × 10
25 m
−3 in all the diodes under consideration.
The n, p and p
+ -layers may be successively developed on n
+ -GaP substrate/wafer
(<100>) by using either molecular beam epitaxy (MBE) technique [13]. The n-type
and p-type doping may be done by using Si and Zn dopants, respectively. Appropriate
etching and polishing techniques must be used to thin down the n
+ -GaP substrate
in the order of 2–3 µm in order to reduce the parasitic series resistance. Sputter
deposition of Be/Au and Au/Ge metal systems on p
+ and n
+ -surfaces can be used to
realize the electrodes [16].
3 Simulation
Well-established non-sinusoidal voltage-excited large-signal (LS) simulation
technique-based drift-diffusion model is used to evaluate the THz performance of the
diodes under consideration [14, 15]. The one-dimensional diode model is shown in
Fig. 2; it is used for microscopic simulation of the device. The justification behind the
use of 1D model instead of 2D or 3D models has already been discussed elsewhere
[14, 15]. The external circuit model containing the frequency dependent equivalent
circuit of the diode, inductive load (waveguide or antenna), and biasing current has
been shown in Fig. 3. From 3, it can be understood that the equivalent circuit of
diode contains parallel connected negative conductance (G) and susceptance (B),