Vous êtes sur la page 1sur 6

Nuclear Instruments and Methods in Physics Research A 477 (2002) 198203

3-D GaAs radiation detectors


A.R. Meiklea,*, R.L. Batesa, K. Ledinghama, J.H. Marshb, K. Mathiesona,
V. OSheaa, K.M. Smitha
b

a
Detector Development Group, Department of Physics and Astronomy, University of Glasgow, UK
Optoelectronics Group, Department of Electrical and Electronic Engineering, University of Glasgow, UK

Abstract
A novel type of GaAs radiation detector featuring a 3-D array of electrodes that penetrate through the detector bulk
is described. The development of the technology to fabricate such a detector is presented along with electrical and
radiation source tests. Simulations of the electrical characteristics are given for detectors of various dimensions.
Laser drilling, wet chemical etching and metal evaporation were used to create a cell array of nine electrodes, each
with a diameter of 60 mm and a pitch of 210 mm. Electrical measurements showed IV characteristics with low leakage
currents and high breakdown voltages. The forward and reverse IV measurements showed asymmetrical
characteristics, which are not seen in planar diodes.
Spectra were obtained using alpha particle illumination. A charge collection efciency of 50% and a S/N ratio of 3 : 1
were obtained.
Simulations using the MEDICI software package were performed on cells with various dimensions and were
comparable with experimental results. Simulations of a nine-electrode cell with 10 mm electrodes with a 25 mm pitch
were also performed. The IV characteristics again showed a high breakdown voltage with a low leakage current but
also showed a full depletion voltage of just 8 V. r 2002 Published by Elsevier Science B.V.

1. Introduction
Planarised GaAs radiation detectors have recently been used extensively in the research of
High Energy Physics (HEP) and X-ray imaging.
Normally, these detectors are hundreds of microns
thick, increasing their X-ray efciency and making
their fabrication and operation simple. HEP
radiation studies [1,2] have shown, however, that
high radiation levels introduce deep level defects to
the bulk which affect the main parameters of
planar GaAs detectors, namely the leakage current
*Corresponding author.
E-mail address: meikle@physics.gla.ac.uk (A.R. Meikle).

density, mean free drift length and charge collection efciency.


A novel architecture [3] that uses a 3-D array of
electrodes that penetrate all the way through the
thickness of the detector bulk offers the possibility
of overcoming these limitations. The electrode
pitch would be sufciently small to collect the
majority of the charge after irradiation and the
need for a detector bulk hundreds of microns thick
would be satised. These cylindrical electrodes
create an electric eld that sweeps the electron and
hole charge carriers laterally through the bulk for
collection at oppositely biased electrodes, and
charge collection will take place throughout the
thickness of the bulk.

0168-9002/02/$ - see front matter r 2002 Published by Elsevier Science B.V.


PII: S 0 1 6 8 - 9 0 0 2 ( 0 1 ) 0 1 8 9 6 - 4

A.R. Meikle et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 198203

2. Fabrication
To fabricate a 3-D detector cell array, holes
have to be created in a substrate and electrodes
created within those holes. Semi-insulating, undoped Liguid Encapsulated Czochralski (SI-U
LEC) was used as the substrate material for the
fabrication of the detector. Fig. 1 shows the
complete fabrication process as follows.
1. The substrate was thinned to a thickness of
200 mm using mechanical lapping and chemical
polishing.

Fig. 1. 3-D detector fabrication process ow.

199

2. Resist 5 mm thick was spun on both sides to


protect the sample surface from subsequent processing, namely ejected debris from the hole drilling process, wet chemical etchant used to treat
the internal walls of the holes and from the metallisation process used to create the electrodes.
3. Holes were drilled through the sample using a
10 ns pulse duration, 355 nm wavelength
Nd : YAG laser with a 5 cm focal length lens,
1 mm iris, and by ring the laser for an
additional 100 pulses immediately after the hole
is created. This process created repeatable,
round holes of 30 mm diameter, smooth internal
walls and minimal tapering. A heat-affected
zone of approximately 500 nm will be created in
the bulk region immediately surrounding the
holes. This drilling process is described in
greater detail in Ref. [4].
4. The internal walls were etched with a solution
of sulphuric acid, hydrogen peroxide and water
to remove the heat-affected zone and to extend
the holes to a diameter of 60 mm. This diameter
will give the electrode formation process a
greater chance of success.
5. Schottky barrier electrodes were formed within
the holes by evaporation of Ti/Pd/Au (33/30/
150 nm) from both sides of the sample. This
recipe has been used extensively in the past for
fabricating Schottky contacts on planar diodes
[5]. Before the metal evaporation was performed the hole internal walls were de-oxidised
for 30 s in a 1 : 1 solution of ammonia and
water. The evaporation was performed on both
sides of the sample to maximise the possibility
that the evaporated metal from both sides
would extend as far as the middle of the hole.
6. The photoresist on both surfaces was removed
using acetone under ultrasonic agitation. This
removed excess metal and debris from previous
processes. The sample was then cleaned with
methanol and water.
7. Al wire 25 mm thick, tipped with silver-loaded
epoxy adhesive, was inserted into the holes to
form electrical connections.
The completed cell consisted of a nine electrode
array of Schottky barrier electrodes. The electrodes had 60 mm diameters and 120 mm pitches.

200

A.R. Meikle et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 198203

3. Detector characterisation

limit mechanism on the Keithley was activated


immediately.

3.1. Currentvoltage characteristics


3.2. Charge collection
Electrical characterisation was carried out using
a Keithley-237 source measurement unit. The
central electrode was biased while the surrounding
eight electrodes were held at zero. All the
electrodes in the cell are of the Schottky type so
all measurements taken will be with a Schottky
Schottky conguration.
Fig. 2 shows the forward and reverse characteristics for the nine electrode cell. The forward and
reverse bias characteristics are asymmetrical,
which is not seen in planar diodes with this
conguration. This may be due to a difference in
active volume, depending on how it is biased,
caused by this particular electrode pattern i.e.
under reverse bias, only the central electrode will
have a reverse biased Schottky barrier, but under
forward bias the eight surrounding electrodes will
have reverse biased Schottky barriers. This inequality may be the cause of the asymmetric
characteristics. The breakdown voltage in both
regions is approximately 500 V and the leakage
current is very low, being only approximately 5 nA
at 500 V in the reverse region. The increase in
current at breakdown was so great that the current

The response of the cell to alpha particle


irradiation from a 241Am source under vacuum
was determined. Due to surface contamination on
the source, which reduced the energy of the alpha
particles, the average energy was measured as
4.1 MeV [6]. The cell was biased and read-out with
an EG&G Ortec-142 pre-amplier. The signal was
shaped with an Ortec-485 post-amplier with a
shaping time of 500 ns and sent to a PC-based
multichannel analyser.
Fig. 3 shows the spectrum obtained at an
applied bias of 300 V at room temperature. The
spectrum is clearly separated from the noise
pedestal and shows the familiar Gaussian shape
of an alpha spectrum obtained with a planar
diode. Both types of charge carrier will contribute
to the total charge. The charge collection efciency
(CCE) for this spectrum was 48%, although the
rate of counting was low in comparison to
standard planar diodes. A Gaussian curve was
tted to the above spectrum to obtain the FWHM.
From this the signal-to-noise ratio (S=N) was
found to be 3 : 1. This is a poor ratio and would
have been due, in part, to the contaminated alpha
source described earlier.
The CCE measurements as a function of applied
bias are shown in Fig. 4. The maximum CCE in
this plot was 50%. This incomplete charge
collection is a manifestation of charge trapping,
which affects both types of charge carrier and is
seen in planar diodes made from the same
substrate material.

3.3. Simulation
Electrical characteristics were calculated using
the MEDIC1 software package [7]. The inputs to
the model were:

Fig. 2. Nine-electrode cell IV characteristics.

1. the cell dimensions,


2. Schottky barrier height=0.79 eV,
3. ND 1:5  1013 ;

A.R. Meikle et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 198203

201

Fig. 3. 4.1 MeV alpha particle spectrum of nine-electrode cell at room temperature.

Fig. 4. CCE measurements of 4.1 MeV alpha particles at room temperature.

4. zero charge trapping,


5. impact ionisation turned on.
Comparing simulation with experiment is an
essential contribution to the validation of experimental results. The rst simulation used identical
cell dimensions to those of the fabricated cell i.e.
nine electrodes of 60 mm diameter and 210 mm

pitch. The bias was applied to the central electrode


while the eight surrounding electrodes were held at
zero.
The reverse biased IV characteristic behaviour
is similar to the reverse region shown experimentally in Fig. 3. The leakage current was very low
and breakdown commenced between 450 and
500 V. Such an agreement between simulation

202

A.R. Meikle et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 198203

and experiment inspires condence in the accuracy


of the experimental results, especially since the
experimental IV curve showed a vastly different
characteristic to that obtained with planar diodes
made from the same substrate and electrode
material.
The simulated forward biased IV characteristic
differs from that of the reverse biased simulation,
which again shows similarity with the experiment.
The leakage current had a slightly higher average
value than that seen in the reverse bias characteristic and began to increase rapidly at a lower value
of applied bias. Again, this is similar to what was
seen experimentally.
Fig. 5 shows the potential distribution across the
cell. It can be seen clearly that it does not extend
all the way to the other electrodes (at the corners

and edges of the cuboid) and therefore the device is


not fully depleted at 550 V. The undepleted area
covers a sizeable percentage of the total area and
this may go some way to explain the incomplete
charge collection described earlier. With reverse
breakdown occurring just below 500 V and incomplete depletion at 550 V, it seems it would not
be possible to fully deplete a 3-D detector with
these dimensions. It was therefore interesting to
see what the response from a cell with different
dimensions would be.
The second simulated cell consisted of nine
electrodes with 10 mm diameters and 25 mm pitches.
The biasing conditions were the same as those for
the previous cell. The potential distribution
characteristics up to 20 V showed clearly that this
device was fully depleted at this bias and, infact,

Fig. 5. Sixty micrometer diameter; 210 mm pitch electrode cell simulation of potential distribution.

A.R. Meikle et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 198203

the program showed the device was fully depleted


at just 8 V. This is a dramatically different result
from the rst cell simulation, where the device had
not fully depleted at 550 V.
The reverse bias IV characteristic for the
second cell found the leakage current to be low
and the breakdown region commenced between
200 and 250 V. Given that full depletion occurs at
just 8 V, it was obvious that this device could
function at full depletion without any dramatic
increase in current. It would seem that a device
with these dimensions could offer full depletion,
low voltage operation.

203

breakdown voltage and this may have contributed


to the incomplete charge collection observed.
Simulations of a cell with 10 mm diameter
electrodes with 25 mm pitches were also performed.
The leakage current again was much lower than
that of a planar diode and the breakdown voltage
was approximately 250 V. However, the most
dramatic observation was that this cell fully
depleted at just 8 V.
The simulations also showed that there were
undepleted regions in this type of cell due to the
layout of the electrodes. This problem could be
overcome simply by designing a cell with six
electrodes in a hexagonal array.

4. Conclusions
Acknowledgements
A 3-D detector has been fabricated and tested.
A nine-electrode test cell consisting of electrodes
with 60 mm diameters and 210 mm pitches was
fabricated on a 200 mm thick SI-U GaAs substrate.
Holes were created by drilling with a 10 ns pulse
duration Nd : YAG laser with a 5 cm focal length
objective lens. Electrodes were formed within these
holes by metal evaporation from both sides.
Currentvoltage characteristics showed asymmetrical forward and reverse characteristics. The
breakdown voltage in the reverse region was
approximately 500 V with a corresponding leakage
current of just 5 nA.
Charge collection experiments were carried out
using 4.1 MeV alpha particles. The CCE was seen
to increase with applied bias before reaching a
maximum value of 50%. A S=N ratio of 3 : 1 was
obtained, although a contaminated source may
have contributed to this low value.
Computer modelling was performed using the
MEDICI software package. A simulation of a cell
with 60 mm electrodes with 210 mm pitches was
performed to compare with the experimental
results. The forward and reverse IV characteristics were asymmetrical, the leakage current much
lower than that seen in planar diodes and the
reverse breakdown region commenced at approximately 500 V. These results were all similar to what
was seen experimentally. However, the simulations
showed that the cell was not fully depleted at the

The authors would like to thank Mr. T.


McCanny of the Laser Ionisation Group in the
Department of Physics and Astronomy at the
University of Glasgow for assistance in the laser
drilling of holes. One of usFK. MathiesonFgratefully acknowledges funding through a CASE
studentship award from BNFL. The authors
would also like to thank Mr. S. Passmore for
some useful discussions.

References
[1] R.L. Bates, C. DaVia, S. DAuria, V. OShea, C. Raine,
K.M. Smith, Nucl. Instr. and Meth. A 395 (1997) 54.
[2] M. Rogalla, M. Battke, N. Duda, R. Geppert, J. Ludwig, R.
Irsigler, T. Schmid, K. Runge, A. SoldnerRembold, Nucl.
Instr. and Meth. A 410 (1998) 41.
[3] S. Parker, 3-DFa new architecture for solid state radiation
detectors, University of Hawaii internal report, 1996, UH
511-839-96.
[4] A.R. Meikle, K. Ledingham, J.H. Marsh, B. More, V.
OShea, C. Raine, K.M. Smith, Nucl. Instr. and Meth. A
410 (1998) 115.
[5] R.L. Bates, Gallium arsenide radiation detectors for the
ATLAS experiment, Ph.D. Thesis, Department of Physics
and Astronomy, University of Glasgow, 1997, RAL-TH-97010.
[6] R.L. Bates, C. DaVia, V. OShea, C. Raine, K.M. Smith, R.
Adams, Nucl. Instr. and Meth. A 410 (1998) 46.
[7] MEDICI, Technology Modeling Associates Inc., 3950
Fabian Way, Palo Alto, CA, USA.

Vous aimerez peut-être aussi