Professional Documents
Culture Documents
Abstract
presence of the collector-tub, which is realized by etching the buried oxide (BOX)
peak electric field at the silicon film-BOX interface and secondly, facilitates the
collector potential to be absorbed both by the collector drift and substrate regions. It is
shown that the BVCEO of CTLBT is enhanced by 2.7 times when compared with a
conventional lateral bipolar transistor (LBT) with identical drift region dopings.
silicon-on-insulator (SOI)
I. Introduction
have drawn wide attention in the recent past due to their compatibility with BiCMOS
technology and possibility of integration in smart power ICs. This is particularly true
ballast circuits where high voltages LBTs are often used [6-8]. Several studies have
been made to improve the breakdown voltage of these LBTs, either by increasing the
critical electric field for breakdown [9-12] or using REdistribution of SURface Field
problem associated with such thin silicon film (≤ 1μm) on SOI, is the fact that the
breakdown voltage is limited by the high electric field at the silicon film-BOX
interface due to difference in dielectric constant between silicon film and BOX. The
breakdown voltages, therefore, do not increase with the increase in drift region length
and saturate at a lower value. This is because the device is operated with the substrate
grounded and hence, the substrate-BOX interface acts as an equi-potential line and the
entire collector voltage is dropped at the BOX and the silicon film on the collector
high-low (NN+) junction side. The present paper highlights the fact that an
reduce the peak electric field at the silicon film-BOX interface, which we propose to
LBT structure on SOI. Our two-dimensional simulation studies show that the presence
breakdown voltage (BVCEO) by a factor of about 2.7. The paper also presents an
analysis of the effect of different device parameters on the breakdown characteristics
have first created the lateral bipolar transistor structure with and without the collector-
tub using standard experimental parameters available in the literature [24-27] in the 2-
dimensional process simulator ATHENA [28] so that the simulated device structures
dimensional device simulator ATLAS [29] to evaluate the device characteristics using
(NS=5.0x1013 cm-3) and an N-type silicon film (ND=1.0x1015 cm-3) on SiO2 as the
starting material in the two-dimensional process simulator ATHENA. The SOI film
and BOX thickness are 1.0μm and 1.2μm respectively. As illustrated in Fig. 1, the
process flow begins with the formation of the collector-tub by etching the silicon film
and the BOX (Fig. 1(a)) and followed by phosphorus implantation. The energy, dose,
time and anneal temperature are varied to obtain different diffusion junction depths
(2.8μm-20μm). The etched region is then refilled with in-situ doped N+-polysilicon
having a concentration of 5.0×1019 cm-3 (Fig. 1(b)). Next the emitter and collector
regions are implanted with phosphorus at a dose of 5.7×1015 cm-2 and an energy of
130 keV and annealed at a temperature of 950OC for 20 min (Fig. 1(c)). The base
region is then opened and implanted with boron at an energy of 62 keV and a dose of
5.0×1012 cm-2 and a drive-in for 120 min at a temperature of 1100OC (Fig. 1(d)). This
is followed by the opening of the base contact window and deposition of 0.3μm layer
Finally the emitter, base, and collector metalizations are made with Al deposition
(Fig. 1(f)). The Al deposition is extended 5μm long [30-31] both at the base-collector
and NN+-junction side for realizing metal plate junction termination with field oxide
thickness of 0.5μm. The LBT structure is obtained following the same process
sequence as described for the CTLBT but without the collector-tub. The overall
process sequence gives the emitter/collector and base doping concentration values of
5.0×1019 cm-3 and 5.0×1016 cm-3 respectively, and base-emitter junction depth of 7
μm. Fig. 2 shows the generated CTLBT and LBT structures and Fig. 3 gives their
B. Device Simulation
The structures obtained in ATHENA are imported for simulation in the two-
are Fermi-Dirac distribution for carrier statistics, Klaassen’s unified mobility model
for dopant-dependent low-field mobility, analytical field dependent mobility for high
electric field, Slotboom model for bandgap narrowing, Selberherr’s ionization rate
model for impact ionization and Shockley-Read-Hall (SRH) and Klaassen Auger
calculated at a collector current of 1.0μA. Figs. 4 and 5 show the Gummel plots and
current gain curves for both LBT and CTLBT for VCB=0 V. The simulated peak
Fig. 6 shows the output characteristics of CTLBT and LBT. Clearly, due to
the presence of the collector-tub, the BVCEO of CTLBT is significantly enhanced when
compared with that of the LBT. We notice that the breakdown voltage has increased
The reason for this significant improvement in breakdown voltage can be understood
from Fig. 7, which shows the electric field profile along the drift region length for
expected, the collector-tub allows the electric field build-up to shift from NN+-
junction side to base-collector junction. Also, the applied reverse voltage is now
supported both by the drift and substrate regions. This is illustrated by the potential
contours in Fig. 8 and electric field vector diagram in Fig. 9, both of which show
however, is limited by the peak electric field at the collector-base junction. On the
other hand, as shown in the potential contour lines and electric field vector diagram of
Figs.10 and 11, the LBT structure shows crowding of electric field lines in the BOX
at the collector NN+-junction and breakdown of silicon film at the silicon-BOX
interface.
junction depth, drift region doping affect the break down voltage of CTLBT, we have
varied the above parameters and estimated the collector breakdown voltage. Fig. 12
illustrates the effect of substrate doping (NS) on breakdown voltage. We notice that
the breakdown voltage is maximum for an optimum substrate doping. This can be
understood from Fig. 13, which gives the electric field profile at various substrate
dopings. At low substrate dopings, the electric field build-up takes place at the
collector-base junction. The depletion volume in the substrate becomes large and the
substrate leakage current limits the device breakdown. At high substrate dopings,
however, the electric field builds up at the NN+-junction and breakdown takes place at
the substrate and collector-tub junction interface. A low substrate doping and deep
collector-tub junction depth makes the electric field build-up at the drift region and
the device breakdown is then limited both by the drift region length and its doping.
Fig. 14 shows the effect of collector-tub junction depth (Xj) on the breakdown
voltage and Fig. 15 gives the electric field profile at various Xj. The breakdown
voltage increases with increasing Xj, reaches a maximum and then decreases. This is
because a deep Xj makes the device to behave as a bulk device and the electric field
build-up begins at the collector-base junction. On the other hand, a shallow Xj makes
the electric field to build-up at the NN+-junction and the device acts as a conventional
and at different drift dopings (ND). As seen from the figure, both CTLBT and LBT
breakdown is different in both the structures. The CTLBT structure breaks down at
low drift dopings as the drift region is entirely depleted and because the electric field
at the NN+-junction is high. For LBT, breakdown occurs because the electric field at
the silicon-box interface is high. In this case, the electric field distribution is two-
dimensional in nature. The component of the electric field at the NN+ junction side
depletes the lowly doped thin silicon film. When the depletion front reaches the
The breakdown characteristics at high drift dopings are, however, same for both the
structures, which is due to high electric field at the collector-base junction end.
Fig. 17 shows the effect of drift region doping on the breakdown voltage of
CTLBT and LBT. Clearly, both the structures show maximum breakdown voltage for
an optimum doping. However, the optimum drift doping for maximum breakdown
voltage in CTLBT is smaller than that of the LBT structure. This can be explained as
follows. In CTLBT, due to the presence of the collector-tub, the critical electric field
limited by the drift doping. In the case of LBT, a high drift doping makes the vertical
component of the electric field large, and, therefore, the lateral electrical field
component becomes responsible to cause depletion in the drift region beginning from
the collector-base junction end. However, once the drift region is entirely depleted,
the vertical component becomes dominant and the electric field at the silicon-BOX
interface determines the maximum breakdown voltage. The critical electric field for
breakdown at the silicon-BOX interface is increased with the increase in drift region
doping and hence the LBT shows optimum breakdown voltage at higher doping when
Fig. 18 shows the influence of the drift region length (LD) on the breakdown
voltage both for the LBT and CTLBT. For LBT, the breakdown voltage saturates at
lower values of LD, however, for CTLBT, the saturation is observed at higher values
of LD. This is because, in CTLBT, the collector-tub makes the lateral electric field
dominant, and therefore, the drift region length can accommodate the spread of lateral
electric field and the breakdown voltage increases with increasing LD. The breakdown
voltage, however, saturates when both the vertical and lateral components of the
electric field become high as evidenced by the two peaks in the electric field
distribution curve (Fig. 7). For LBT structure, both the vertical and lateral component
contributes to the breakdown process and a longer collector drift region length is of no
V. Conclusions
Two-dimensional numerical simulation studies of a collector-tub lateral bipolar
has an N-diffusion region at the collector high-low (NN+) junction. The collector-
emitter breakdown voltage (BVCEO) of CTLBT is about 2.7 times higher than that of
the conventional LBT on SOI with identical doping profile. The increased breakdown
voltage in CTLBT is explained as due to the shifting of the electric field from the
collector high-low junction side to the base-collector junction side and also due to the
distribution of the applied reverse potential in the substrate and drift regions. This
potential redistribution, for a given BOX thickness and SOI film thickness, is found to
be dependent on substrate doping (NS), drift doping (ND) and collector-tub junction
depth (Xj). To realize the CTLBT structure in a standard CMOS process, a process
[08] R. Gomez, R. Basir, and G. W. Neudeck, “On the design and fabrication of
novel lateral bipolar transistor in a deep-submicron technology,”
Microelectronics Journal, Vol. 31, pp. 199-205, 2000.
[09] W. Wondark, R. Held, E. Stein, and J. Korec, “A new concept for high-
voltage SOI Devices,” Proceedings of the 4th International Symposium on
Power Semiconductor Devices and ICs, pp. 278-281, May 1992.
[11] S. R. Banna, P. C. H. Chan, and J. Lau, “On buried oxide effects in SOI lateral
bipolar transistors,” IEEE Transactions on Electron Devices, Vol. 44, No. 1,
pp. 139-143, Jan. 1997.
[12] G. Amaratunga and G. Udrea, “Power devices for high voltage integrated
circuits: new device and technology concepts,” Proceedings of the
Semiconductor Conference, Vol. 2, pp.441 – 448, 9-13 Oct. 2001.
[21] R. Sunkavalli, A. Tamba, and B. J. Baliga, “Step drift doping profile for high
voltage DI lateral power devices” Proceedings of the IEEE International SOI
Conference, pp. 139-140, Oct. 1995.
[28] Athena User’s Manual. Santa Clara, CA: Silvaco International, 2000.
[29] Atlas User’s Manual. Santa Clara, CA: Silvaco International, 2000.
[30] V. P. O’Neil and P. G. Alonas, “Relation between oxide thickness and the
breakdown voltage of a planer junction with field relief electrode,” IEEE
Transactions on Electron Devices, Vol. 26, No. 7, pp. 1098-1100, July 1979.
Fig. 2(a) Lateral bipolar transistor (LBT) and (b) Collector-tub lateral bipolar
transistor (CTLBT) on SOI.
Fig. 3 Doping profile for CTLBT and LBT as obtained from ATHENA.
Fig. 5 Collector current versus Current gain of CTLBT compared with that of the
LBT.
Fig. 7 Electric field at the silicon film/BOX interface of CTLBT compared with that
of the LBT.
Fig. 9 Electric field vector diagram for CTLBT showing electric field spreading at the
Silicon film/BOX interface and breakdown at the collector-base interface or at the
collector junction/substrate interface.
Fig. 10 Potential contours at the Silicon film of LBT at breakdown voltage (93V),
potential contours=10V/step, LD=40μm, NS=5.0x1013 cm-3, ND=2.5x1015 cm-3,
tOX=1.2μm, and Xj=3.0μm.
Fig. 11 Electric field vector diagram for LBT showing field crowding at the BOX and
breakdown at the silicon-BOX interface.
Fig. 15 Electric field profile at the silicon film/field oxide interface of CTLBT at
various junction depths.
Fig. 16 Electric field profile along the silicon/BOX and silicon-field oxide interface of
CTLBT for different drift region dopings compared with that of the LBT.
Phosphorus
N N+ P N N+
BOX BOX
(a) N (d)
P- substrate P- substrate
P+ polysilicon contact
N N+
BOX N+ P N N+
N (b)
BOX
P- substrate N (e)
P- substrate
Phosphorus Phosphorus E B C
N N+ N+ P N N+
BOX BOX
N (c) (f)
N
P- substrate P- substrate
Fig. 1
E B C
N+ P N N+
BOX
(a)
P- substrate
E B C
N+ P N N+
Buried oxide
BOX
(b)
N-collector
tub
P- substrate
Fig. 2
20
10
+ Emitter Collector +
19 N N
10
Net doping [ cm ]
-3
18
10
17
10
Base
16
10 P
Drift region
15 N
10
14
10
0 10 20 30 40 50 60
Lateral distance [ μm ]
-3
10
CTLBT, Xj=3.0μm
-5
10 - - - - LBT
VCB=0 V
-7 IC
10
IC , IB [ A ]
15 -3
ND=2.5x10 cm
-9 13 -3
10 NS=5.0x10 cm
LD=40μm IB
-11
10 tOX=1.2μm
-13
10
-15
10
-17
10 0.2 0.4 0.6 0.8 1.0
Base-emitter voltage, VBE [ V ]
Fig. 4
35
CTLBT, Xj=3.0μm
30 - - - - LBT
VCB=0 V
Current gain, β
25
15 -3
ND=2.5x10 cm
20 13 -3
NS=5.0x10 cm
15 LD=40μm
tOX=1.2μm
10
5
0 -11 -9 -7 -5
-13
10 10 10 10 10
Collector current, IC [ A ]
7
N =2.5x10 cm
15 -3
CTLBT, Xj=3.0μm
Collector current [ μA ]
6 ND=5.0x1013cm-3 - - - - LBT
S
IB=0 to 0.25μA @ 0.05μA
5 LD=40μm
tOX=1.2μm
4
3
2
1
IB=0
0
0 50 100 150 200 250 300
Collector-emitter voltage, VCE [ V ]
Fig. 6
2.5 CTLBT, Xj=3.0μm
-1
- - - - LBT
2.0 BV =255V
5
CEO VCEO increment @ 25 V
15 -3
1.5 ND=2.5x10 cm
13 -3
NS=5.0x10 cm
1.0 LD=40μm
tOX=1.2μm
0.5
0.0 BVCEO=25V
10 20 30 40 50 60
Lateral distance [ μm ]
Fig. 7
E B C
-3.2
-2.2
N+ P N N+
-1.2
Vertical distance [ µm ]
BOX N
0
10
20
30 Depletion layer
40
50
VCEO=0 V VCEO=255 V
60
P-substrate
70
0 10 20 30 40 50 60
Lateral distance [ µm ]
Fig. 8
E B C
-3.2
-2.2
N+ P N N+
-1.2
Vertical distance [ µm ]
BOX N
0
10
Depletion layer
20
Breakdown
30
40
50
60 P-substrate
70
0 10 20 30 40 50 60
Lateral distance [ µm ]
Fig. 9
E B C
-3.2
-2.2
N+ P N N+
-1.2
Vertical distance [ µm ]
BOX
0
10
20 VCEO=0 V VCEO=93 V
30 Depletion layer
40
50
60 P-substrate
70
0 10 20 30 40 50 60
Lateral distance [ µm ]
Fig. 10
E B C
-3.2
-2.2
N+ P N N+
-1.2
Vertical distance [ µm ]
BOX
0
10
20 Breakdown
30 Depletion layer
40
50
60 P-substrate
70
0 10 20 30 40 50 60
Lateral distance [ µm ]
Fig. 11
200 15 -3
Breakdown voltage [ V ]
ND=1.0x10 cm
LD=40μm
180
Xj=3.0μm
160 tOX=1.2μm
140
120
100
80 13 14 15
10 10 -3 10
Substrate doping [ cm ]
Fig. 12
13 -3
NS=1.0x10 cm , BVCEO=110V
1.5 ND=1.0x10 cm
15 -3
LD=40μm
1.0 Xj=3.0μm
tOX=1.2μm
0.5
0.0
10 20 30 40 50 60
Lateral distance [ μm ]
Fig. 13
280
Breakdown Voltage [ V ]
15 -3
ND=1.0x10 cm
260 13
NS=5.0x10 cm
-3
LD=40μm
240 tOX=1.2μm
220
200
180
3 6 9 12 15 18 21
Collector-tub junction depth [ μm ]
Fig. 14
2.0 Xj=2.6μm, BVCEO=192V
Electric field [ x10 Vcm ]
-1
Xj=13.2μm, BVCEO=270V
Xj=19.5μm, BVCEO=226V
1.5 15 -3
ND=1.0x10 cm
5
13 -3
NS=5.0x10 cm
1.0 LD=40μm
tOX=1.2μm
0.5
0.0
10 20 30 40 50 60
Lateral distance [ μm ]
Fig. 15
4.0
CTLBT
Si-field oxide interface
3.5 Si-BOX interface
Electric field [ x10 Vcm ]
-1
LBT
3.0 Si-field oxide interface
14 -3
Si-BOX interface ND=1.0x10 cm
2.5 16
ND=1.6x10 cm
-3
5
2.0 13 -3
NS=5.0x10 cm
1.5 LD=40μm
1.0 Xj=3.0μm
tOX=1.2μm
0.5
0.0
10 20 30 40 50 60
Lateral distance [ μm ]
Fig. 16
CTLBT, Xj=3.0μm
Breakdown voltage [ V ]
250 LBT
13 -3
NS=5.0x10 cm
LD=10μm
200 tOX=1.2μm
150
100
14 15 16
10 10 10
-3
Drift doping [ cm ]
Fig. 17
300
CLBT, Xj=3.0μm
Breakdown voltage [ V ]
250 LBT
15 -3
ND=2.5x10 cm
13 -3
200 NS=5.0x10 cm
tOX=1.2μm
150
100
50
10 20 30 40 50
Drift region Length, LD [ μm ]
Fig. 18