Measurement of Dose Rate Dependence of Radiation Induced Damage to the Current
Gain in Bipolar Transistors1
D. Dorfan, T. Dubbs, A. A. Grillo, W. Rowe, H. F.-W. Sadrozinski,
A. Seiden, E. Spencer, S. Stromberg, R. Wichmann
SCIPP, University of California, Santa Cruz, CA 95064 USA
N. E. Ipe, S. Mao
Stanford Linear Accelerator Center, Stanford, CA 94309 USA
Abstract
We report the study of radiation induced change in the
current gain of bipolar transistors for three different gamma
dose rates. The dose rates differed by a factor of 60 with the
lowest close to that anticipated for the LHC, and the highest at
a rate we have been routinely using for radiation damage tests.
The maximum dose attained was 200kRad, which is high
enough to compare with other measurements. The importance
of annealing high dose rate data is demonstrated.
I. INTRODUCTION
One of the major challenges for High Energy Physics
experimentation at the future Large Hadron Collider (LHC) at
CERN is the expected radiation damage to electronic devices.
For example, the front-end electronics (FEE) of the silicon
tracking device SCT in the ATLAS experiment is expected to
accumulate a total dose of approximately 10 MRad over an
anticipated lifetime of 10 years [1]. Previous studies have
examined the effect of total irradiation dose on bipolar
transistors and electronics, and found lowered current gain for
increased doses [2,3]. Recently, new results have shown that
for a given dose, more damage is induced if the irradiation is
performed at a lower dose rate [4-10]. This effect may have a
significant impact on electronics being developed for the
readout of the silicon tracking detectors for use at the LHC.
Radiation damage to bipolar transistors is characterized by a
rapid decrease of the current gain in the first few hundred
kRad, with a leveling off at much larger total doses. While it
has been demonstrated that the transistors will still operate
adequately after 10 MRad when irradiated at a dose rate of
3Rad/s, (yielding a current gain of about 50 in typical npn
transistors [3]), there is concern that the low irradiation rate at
the LHC of about 0.05Rad/s may lead to increased damage
such that the electronics are no longer functional for their
designed lifetime. The existence of dose rate effects has been
observed in radiation damage to CMOS transistors [11] and
silicon detectors [12], and in both cases they are explained by
annealing which occurs during and after the irradiation.
Recent measurements [4-10] show large dose rate
dependence beginning at a relatively low total dose of about
100kRad. We thus decided to test dose rate dependence
____________________________
1This work was supported in part by the US. Department
of Energy.
systematically up to 200kRad over a period of a few weeks,
which yields a dose rate similar to that expected at the LHC.
Our previous measurements [2,3] indicated that a large fraction
of the damage occurs early in the irradiation. It should be
pointed out that in Refs. [4-10], the radiation damage was
investigated at extremely low currents and correspondingly
small VBE. In this investigation, we will concentrate on
damage at constant collector current IC, which tends to be kept
stabile in bipolar circuits, and determine the change in current
gain at realistic collector currents IC.
In this study we have irradiated 32 transistors from the
Maxim CB2 process identical to those which are currently
being used in the prototype CAFE-M bipolar chip designed for
the silicon tracker of the ATLAS experiment [13]. We have
exposed these transistors to three different rates of gamma's
from a 60Co source at the Santa Cruz Institute for Particle
Physics (SCIPP), including the expected dose rate for the
LHC of 0.05Rad/s, and at the highest dose rate of the source
(3Rad/s), employed in previous measurements [2,3].
Measurements of induced damage to the current gain of the
transistors were performed at regular intervals throughout the
irradiation to a total dose of 200 kRad. In addition, careful
studies of the effect of annealing under controlled conditions
were obtained.
II. TEST STRUCTURES AND IRRADIATION
The Maxim CB2 was developed by Maxim as an
enhancement to the Tektronix SHPi process by adding vertical
pnp structures. The process is a fully complementary bipolar
process which uses oxide isolation. The typical p+ isolation
is replaced with a recessed oxide isolation to reduce the
collector area.
This process is designed for superior
performance and flexibility. The CB2 process uses polysilicon
emitters in both the pnp and npn transistors and typically
shows dc gains (hFE) of about 90.
The various size
transistors are arranged in standard test keys included on the
wafers of CAFE-M VLSI chips. Table 1 summarizes the
various transistors used in this study. The reference current
listed in the table is the nominal current at which the transistor
is run in the CAFE-M chip.
To characterize the transistors we performed forward
Gummel plots following the specifications of the CB2 process
published by Maxim in their design documentation to extract
the dc parameters and characteristics of the various BJT's under
test [14].
Table 1
Summery of MAXIM CB2 Transistors.
Device
Emitter size
Emitter #
Ref. current
NPN1
NPN2
NPN8
NPN32
PNP1
PNP2
PNP8
PNP16
1.0x3.2 µm2
1.0x7.2 µm2
1.0x7.2 µm2
1.0x31.2 µm2
1.0x3.2 µm2
1.0x7.2 µm2
1.0x31.2 µm2
1.0x31.2 µm2
1
1
4
4
1
1
1
2
5 µA
10µA
40µA
160µA
5µA
10µA
40µA
80µA
The photon irradiation took place at the 60Co source at UC
Santa Cruz. The devices were irradiated to a total dose of
200 kRad at three different rates: 3, 0.1 and 0.05Rad/s. Two
opti-chromic dosimeters [15] were used to calibrate the dose
rate at each of the three different device and exposure positions.
The measured dose rate at the different positions agreed to
better than 5%. We estimate the relative error on the dose
between different dose rates to be 5%.
In all cases the transistors were biased with an emitter
current of approximately 100 µA during the irradiation at a
temperature of 20C. During the annealing period, the devices
were stored in the same location to eliminate temperature
effects to skew the data, but without applied bias. Forward
Gummel plots were obtained periodically throughout the
irradiation for the lower dose rates. Table 2 summarizes the
irradiation and annealing history of the devices under test.
Table 2
Irradiation history for CB2 Test Structures.
Device
Rate (Rad/s) Dose
(kRad)
1
0.05
2
0.1
3
3.0
III. RESULTS
A. Annealing
During the measurements taken throughout the irradiation,
as indicated in Table 2, we looked carefully for room
temperature annealing of 1 week, 2 weeks and 3 weeks
respectively. The data shows appreciable detrimental post-rad.
annealing, with the radiation damage increasing by a large
factor within the first week after the exposure for structures
irradiated with high dose rate, but with very little change
afterwards. However, for the test structures irradiated with the
lower dose rates, little change appears between measurements
taken directly after the irradiation, and those performed a week
later.
Thus we can conclude that for exposures lasting much
longer than one week, the radiation damage in low rate
exposures will have annealed out during the irradiation time
and the measurements taken immediately after the final
irradiation will reflect the stable radiation damage. For the
10Mrad irradiation of Ref. 3 at 3Rad/s, the exposure time was
5.5 weeks, so the error was less than 20%, assuming an anneal
time of less than one week. The data taken during the
irradiation, on the other hand, would have systematically
underestimated the damage. In the following, we will compare
data of low and high dose rate exposures after annealing times
of up to 13 weeks.
4
3.0
0
14.3
35.6
64.0
64.0
98.9
157.1
200.5
200.5
200.5
200.5
0
28.5
71.3
128.0
128.0
197.9
197.9
197.9
0
220.4
220.4
220.4
220.4
220.4
220.4
0
53.6
53.6
53.6
53.6
53.6
219.6
219.6
219.6
219.6
Anneal Time
(hrs)
0
0
0
170
0
0
0
170
1020
2200
0
0
0
0
170
340
1020
2500
340
510
680
850
2040
3500
0
170
340
510
680
0
170
510
2200
B. Change of Current gain as a function of Dose
During irradiation, we measured the current gain hFE as a
function of collector current for all eight of the transistors at
the dose and anneal points listed in Table 2. They correspond
to values of VBE >= 0.7V. They were fit with a second order
polynomial in log (IC), which describes the data very well:
hFE = A + B*ln(Ic) + C*ln(Ic)2 .
(1)
For comparison of the radiation induced damage between
different transistors and differing rates of irradiation, the data
for different transistors could not be compared directly because
of the difference in pre-rad current gains. Instead we compared
the percent change form the pre-irradiation current gain at the
nominal collector currents specified in Table1.
0.05
0.00
0
-0.05
-0.05
-0.10
-0.1
-0.15
-0.15
-0.25
-0.2
-0.25
-0.30
0
50
100
150
200
-0.3
250
Relative Change in Current Gain h
MAXIM npn2 @ I
0
npn2-0.05Rad/s
npn2-0.1Rad/s
npn2-3Rad/s
-0.05
npn2-3Rad/s
-0.05
-0.10
-0.1
-0.15
-0.15
-0.20
-0.2
-0.25
-0.25
-0.30
0
50
100
150
200
-0.3
250
0.05
0.00
0
-0.05
-0.05
-0.10
-0.1
-0.15
-0.15
-0.20
-0.2
-0.25
-0.25
-0.30
0
MAXIM npn8 @ I
= 40microA
c
0.00
0
-0.05
-0.05
-0.10
-0.1
-0.15
-0.15
-0.20
-0.2
npn8-0.05Rad/s
npn8-0.1Rad/s
npn8-3Rad/s
npn8-3Rad/s
-0.25
-0.25
-0.30
0
50
100
150
200
-0.3
250
= 160microA
c
0.00
0
-0.05
-0.05
-0.10
-0.1
-0.15
-0.15
-0.20
-0.2
npn32-0.05Rad/s
npn32-0.1Rad/s
npn32-3Rad/s
npn32-3Rad/s
-0.25
-0.25
-0.30
0
50
100
150
200
-0.3
250
Dose [kRad]
Figure 1: Relative change in transistor current gain vs. dose for
different dose rates for the four npn transistor at scaled collector
currents Ic.
Relative Change in Current Gain h
MAXIM npn32 @ I
100
150
200
-0.3
250
MAXIM pnp2 @ I
= 10microA
c
0.10
0.1
0.05
0.05
0.00
0
-0.05
-0.05
-0.10
-0.1
-0.15
-0.15
pnp2-0.05Rad/s
pnp2-0.1Rad/s
pnp2-3Rad/s
pnp2-3Rad/s
-0.20
-0.25
-0.2
-0.25
-0.30
0
50
100
150
200
-0.3
250
Dose [kRad]
MAXIM pnp8 @ I
= 40microA
c
0.10
0.1
0.05
0.05
0.00
0
-0.05
-0.05
-0.10
-0.1
-0.15
-0.15
0.05 Rad/s
0.1 Rad/s
3 Rad/s
3 Rad/s
-0.20
-0.25
-0.2
-0.25
-0.30
0
50
100
150
200
-0.3
250
Dose [kRad]
FE
Dose [kRad]
50
Dose [kRad]
FE
Dose [kRad]
FE
Relative Change in Current Gain h
= 10microA
c
0.00
= 5microA
0.1
pnp1-0.05Rad/s
pnp1-0.1Rad/s
0.05
pnp1-3Rad/s
pnp1-3Rad/s
FE
FE
Dose [kRad]
c
0.10
Relative Change in Current Gain h
npn1-0.05Rad/s
npn1-0.1Rad/s
npn1-3Rad/s
npn1-3Rad/s
Relative Change in Current Gain h
0.05
-0.20
MAXIM pnp1 @ I
FE
= 5microA
0.1
FE
Relative Change in Current Gain h
c
Relative Change in Current Gain h
Relative Change in Current Gain h
FE
MAXIM npn1 @ I
0.10
MAXIM pnp16 @ I
0.10
c
= 80microA
0.1
0.05
0.05
0.00
0
-0.05
-0.05
-0.10
-0.1
-0.15
-0.15
0.05 Rad/s
0.1 Rad/s
3 Rad/s
3 Rad/s
-0.20
-0.25
-0.2
-0.25
-0.30
0
50
100
150
200
-0.3
250
Dose [kRad]
Figure 2: Relative change in transistor current gain vs. dose for
different dose rates for the four pnp transistor at scaled collector
currents Ic.
shown as a function of the collector current. The curves are
calculated using the fits to the forward Gummel plots
mentioned above [eq. (1)]. The values for the two lower dose
rates are scaled linearly by 10% to account for the different
total dose. The conclusions from these plots are the same as
from the figures showing the dose dependence (Figs. 1 and 2):
the radiation damage in the pnp is about half of that in the
npn, and the scatter of the two high dose rate curves are as
large as between the high and the low dose rate sets, of about
30%.
FE
FE
Relative Change in Current Gain dh / h
0
npn2 @220kRad
-0.1
-0.2
-0.3
0.05Rad/s
0.1Rad/s
3Rad/s
3Rad/s
-0.4
-5
-4
10
10
Collector Current IC [A]
10
-3
FE
Relative Change in Current Gain dh / h
FE
-0.5
-6
10
0
npn8 @220kRad
-0.1
-0.2
-0.3
0.05Rad/s
0.1Rad/s
3Rad/s
3Rad/s
-0.4
-0.5
1 0- 6
1 0- 5
1 0- 4
Collector Current IC [A]
1 0- 3
FE
In Figures 1 and 2, a line connecting the origin with the
200kRad dose point for the 3.0 Rad/s dose rate is displayed.
A deviation from this line for the data with lower dose rates
will indicate a dose rate dependence of the radiation damage.
We can first compare the points of the two test structures
irradiated with the high dose rate of 3Rad/s, which had different
histories of irradiation and annealing, as listed in table 2. At a
total dose of 220kRad, they agree with each to 30% for the
npn (Fig. 1) and much better for the pnp transistors (Fig. 2).
This level of agreement is a measure of the sensitivity of our
test. Since the low dose rate data deviate from the high dose
rate data at 200kRad by about the same amount, we can state
that the maximal dose rate effect is less or of the order of 30%.
In general, the initial low dose points deviate more from
the expectation than the final point. The npn and the pnp
show different trends, with the pnp's generally lying above the
expectation. We attribute this to the fact that the data taken at
early times did not have time to anneal out completely. For a
given radiation dose, it is apparent that the induced damage for
a pnp transistor is roughly half that of a npn transistor, as
mentioned before. This makes the use of pnp's more attractive
for applications where the current gain is affected by radiation.
A comparison of the dose rate dependence at the highest
total dose point and more than 10 weeks of annealing is shown
in Figs. 5 and 6. Here the relative change in current gain is
∆hFE/hFE = [hFE(220kRad)-hFE(0kRad)]/hFE(0kRad) (2)
We have investigated the dose rate dependence of radiation
induced damage to the current gain of bipolar Maxim CB2
transistors at realistic collector currents. Test structures were
irradiated with gamma’s from a 60Co source at dose rates
differing by a factor 60 up to a total dose of 200kRad.
FE
C. Dose Rate Dependence
IV. CONCLUSIONS
Relative Change in Current Gain dh / h
Figures 1 and 2 summarize the results for 8 transistor
sizes. Each figure contains data from all three dose rates for a
given size transistor. For data corresponding to the same total
dose, the one with the longest anneal time is displayed. Error
bars representing our estimate of the relative dose uncertainties
(5%) are shown. The error bars in the vertical scale are of the
order 1%. The smallest transistors (npn1, pnp1) show similar
trends as the other 6 transistor sizes, but have special features
which we will investigate further in the future.
The npn transistors show a larger initial gain than the pnp
transistor, however, the relative damage in the npn transistors
appears larger, especially at the lower collector currents. These
trends are common across the 32 transistors tested. The
increased radiation hardness of the pnp transistors has not been
seen in previous Maxim C-Pi or SHPi processes. For the pnp
transistors, we observe for low dose an initial increase in the
current gain, followed by the expected decrease for larger doses.
-0.1
npn32 @220kRad
-0.2
-0.3
-0.4
-0.5
-0.6
1 0- 6
0.05Rad/s
0.1Rad/s
3Rad/s
3Rad/s
1 0- 5
1 0- 4
Collector Current IC [A]
1 0- 3
Figure 3: Relative change in transistor current gain for npn
transistors vs. collector current Ic at the maximum dose of
220kRad for different dose rates. Note the vertical offset in the
npn32 data.
FE
Relative Change in Current Gain dh / h
FE
weeks [2,3] are thus only marginally affected. To a level of
30%, the total induced damage appears to be dependent on dose
received, at least for the dose rates employed in this study.
A direct comparison with previous measurements
indicating dose rate dependence of the radiation damage in
bipolar circuits [4-10] is difficult because they were determined
in very different current regimes, i.e. at very much lower
currents where the radiation damage is much larger and surface
effects are large and possibly technology dependent.
0
pnp2 @220kRad
-0.1
-0.2
-0.3
0.05Rad/s
0.1Rad/s
3Rad/s
3Rad/s
-0.4
-5
-4
10
10
Collector Current IC [A]
10
-3
FE
Relative Change in Current Gain dh / h
FE
-0.5
-6
10
V. REFERENCES
0
pnp8 @220kRad
-0.1
-0.2
-0.3
0.05Rad/s
0.1Rad/s
3Rad/s
3Rad/s
-0.4
1 0- 5
1 0- 4
Collector Current IC [A]
1 0- 3
FE
Relative Change in Current Gain dh / h
FE
-0.5
1 0- 6
0
pnp16 @220kRad
-0.1
-0.2
-0.3
-0.4
-0.5
1 0- 6
0.05Rad/s
0.1Rad/s
3Rad/s
3Rad/s
1 0- 5
1 0- 4
Collector Current IC [A]
1 0- 3
Figure 4: Relative change in transistor current gain for pnp
transistors vs. collector current Ic at the 220kRad dose.
No evidence for rate dependence was found in the region of
0.05Rad/s to 3Rad/s. The significance of the former is that it
is at the anticipated dose rate at the LHC, and the one of the
latter that it is the dose rate of our Co source, used before
frequently for our radiation damage studies. We find that for
higher dose rates the total damage to the transistor current gain
is not apparent until an annealing period of about a week at
room temperature has elapsed. Measurement taken directly
after irradiation would yield reduced damage effects for these
higher rate exposures [12]. Earlier long-term studies of several
[1] ATLAS Inner Detector TDR, CERN/LHCC/97-16/17.
[2] E. Barberis et al., “Radiation Hardness Measurements on
Components of a full Custom Bipolar Process,” SCIPP
92/38, Aug. 1992.
[3] A. A. Grillo et al., “Radiation Hardness Measurements
on MAXIM full Custom Bipolar Processes C-Pi and
CB2,” SCIPP 96/34, Aug. 1996.
[4] E. W. Enlow et al., “Response of Advanced Bipolar
Processes to Ionizing Radiation,” IEEE Trans. Nucl. Sci.,
vol. 38, pp. 1242, 1991.
[5] R. N. Nowlin et al., “Hardness-Assurance and Testing
Issues for Bipolar/BiCMOS Devices,” IEEE Trans. Nucl.
Sci., vol. 40, pp. 1686, 1993.
[6] A. H. Johnston et al., “Enhanced Damage in Linear
Bipolar Integreated Circuits at Low Dose Rate,” IEEE
Trans. Nucl. Sci., vol. 42, pp. 1650, 1995.
[7] D. M. Schmidt et al., “Modeling Ionizing Radiation
Induced Gain Degradation of the Lateral PNP Bipolar
Junction Transistor,” IEEE Trans. Nucl. Sci., vol. 43, pp.
3032, 1996.
[8] H. Barneby et al., “Analysis of Bipolar Linear Circuit
Response Mechanisms for High and Low Dose Rate Total
Dose Irradiations,” IEEE Trans. Nucl. Sci., vol. 43, pp.
3040, 1996.
[9] A. H. Johnston et al., “Enhanced Damage in Bipolar
Devices at Low Dose Effects at Very Low Dose Rates,”
IEEE Trans. Nucl. Sci., vol. 43, pp. 3049, 1996.
[10] S. McClure et al., “Dependence of Total Dose Response
of Bipolar Linear Microcircuits on Applied Dose Rate,”
IEEE Trans. Nucl. Sci., vol. 41, pp. 2544, 1994.
[11] D. W. Fleetwood et al., “A Reevaluation of worst-case
Postirradiation Response for hardened MOS Transistors,”
IEEE Trans. Nucl. Sci., vol. 34, pp. 1178, 1987.
[12] H. Ziock et al., Nucl. Instr. Methods, vol. A342, pp. 96,
1994.
[13] Ref. 1 and ATLAS-SCT Project Specification, Project
Name: CAFE-M, Version 3.01.
[14] MAXIM Quickchip8 IC Design Documentation, 1994,
Beaverton, Oregon 97003.
[15] Opti-chromic dosimeters of type FWT 70-40M, Batch 0-1
from Far West Technology Inc., 330-D S. Kellogg,
Goleta, CA 93017, USA.
[16] N. E. Ipe et al., “High Level Dosimetry at SLAC”,
SLAC-Pub 5686, Nov 1991.