A deep semiconductor defect with continuously variable activation energy
and capture cross section
M. A. Lourenc¸o,
a)
Wai Lek Ng, and K. P. Homewood
The School of Electronic Engineering, Information Technology and Mathematics, University of Surrey,
Guildford, Surrey, GU2 5XH, United Kingdom
K. Durose
Department of Physics, Durham University, Durham DH1 3LE, United Kingdom
͑Received 21 December 1998; accepted for publication 18 May 1999͒
A deep level with a continuously varying activation energy and capture cross section has been
observed in CdS/CdTe thin-film solar cells. Given that the activation energy and capture cross
section of a level are usually considered to be a unique identifier or signature for a particular deep
level, this has important implications for the application of deep-level transient spectroscopy and
related techniques for the characterization of deep levels in this and similar systems. We believe this
phenomenon explains the well-known but poorly understood efficacy of CdCl
2
treatment for CdS/
CdTe thin solar cells. © 1999 American Institute of Physics. ͓S0003-6951͑99͒03328-8͔
Polycrystalline thin-film CdS/CdTe heterojunctions are
one of the leading contenders to produce low-cost, large-
area, stable, and efficient solar cells. The calculated theoret-
ical efficiency for CdTe cells is close to 29%,
1
but practically
achieved results are around 16%.
2,3
The cell’s efficiency has
been shown to be improved dramatically by a postdeposition
of a CdCl
2
layer followed by annealing, typically, at
400°C.
4,5
The structural effect of the CdCl
2
treatment is to
modify grain growth and to promote recrystallization, hence,
reducing strain at grain boundaries and lowering the defect
density.
6
This treatment has been found to produce a number
of other changes to the system; for example, it has been
reported to change the dominant transport mechanism ͑from
interface recombination/tunneling to depletion region recom-
bination͒, increase the barrier height of the CdS/CdTe junc-
tion, improve the cell’s optical response as well as its elec-
trical characteristics, and also promote interdiffusion
between CdS and CdTe at the interface producing a thin
layer of CdS
x
Te
1Ϫx
.
2,5
Despite these observations, the basic
mechanism for the efficiency gains has not been understood.
Deep levels, electronically active defects, in semicon-
ductors play the major role in determining ultimate device
performance. The most commonly used technique for evalu-
ating and identifying deep levels in semiconductor materials
and devices is deep-level transient spectroscopy ͑DLTS͒.
The power of this technique relies on its ability to produce a
unique fingerprint of the deep-level content in the sample ͑at
a constant rate window͒ and to provide values of the trap
parameters. In particular, trap activation energy and trap cap-
ture cross section that can be directly associated with a par-
ticular defect in a particular semiconductor. In the results
presented here, we give an example where we have observed
monotonic changes in all these features as a function of pro-
cessing. This feature is attributed to the influence of grain-
boundary modification in the polycrystalline material. Given
that such granular materials are being increasingly used and
mooted for large-area semiconductor device applications we
suggest that the use of DLTS to investigate deep levels in
polycrystalline materials should be treated with caution.
Cells with the CdTe/CdS/ITO/glass structures were pro-
vided by ANTEC GmbH. Tin–oxide:indium–tin–oxide
͑TO/ITO͒ was deposited by sputtering to a total thickness of
50 nm. An n
ϩ
-CdS layer ͑ϳ100 nm͒, doped to around
10
14
cm
Ϫ3
, followed by a thicker layer of CdTe ͑15
m͒ was
deposited by close-space sublimation. The samples were
then evaporated with CdCl
2
prior to being annealed in air at
400°C for 25 min and were distinguished by the thickness of
the CdCl
2
layer: 15, 30, 60, and 120 nm. Au was used as a
contact to the CdTe.
Deep-level transient spectroscopy experiments were per-
formed in the dark using a Bio-Rad DL4600 system.
Samples were mounted on a stage in a liquid-nitrogen cry-
ostat. The temperature was monitored by a platinum resis-
tance thermometer attached directly to the stage, giving an
uncertainty of Ϯ0.5 K on the measured value. All DLTS
spectra were taken twice ͑ramping the temperature up and
then down͒ to account for the temperature lag which might
have occurred due to the glass substrate. Arrhenius plots
were obtained from the average values of the up and down
peak positions. Prior to the DLTS measurements,
capacitance–voltage and current–voltage measurements
were performed at various temperatures, and standard p–n
junction characteristics were obtained for all samples.
A typical DLTS spectrum indicated the presence of a
dominant majority-carrier trapping center ͑associated with a
single exponential positive peak present in the temperature
range of 150–300 K͒ and one or more minority-carrier trap-
ping centers ͑associated with broad, negative peaks͒.Itis
noted that all the results discussed subsequently were ob-
tained under the same bias and pulse conditions ͑reverse
voltage, V
R
ϭϪ0.4V; forward voltage, V
F
ϭ2.0 V; fill
pulse, tϭ8ms͒to enable direct comparison between the dif-
ferent treated samples. Under these excitation conditions,
around 2–3
m of bulk CdTe layer is sampled. However, we
have also carried out experiments under a wider range of
a͒
Electronic mail: m.lourenco@ee.surrey.ac.uk
APPLIED PHYSICS LETTERS VOLUME 75, NUMBER 2 12 JULY 1999
2770003-6951/99/75(2)/277/3/$15.00 © 1999 American Institute of Physics