Full text of DOI:10.1007/BF02665838

Journal of ELECTRONIC MATERIALS, Vol. 30, No. 6, 2001
Current Mechanisms in VLWIR Hg1-xCdxTe Photodiodes
Special Issue P5a8pe5r
Current Mechanisms in VLWIR
Hg1–xCdxTe Photodiodes
A.I. D’SOUZA,1 R.E. DEWAMES,2 P.S. WIJEWARNASURIYA,2
G. HILDEBRANDT,2 and J.M. ARIAS2
1.—Boeing Sensor & Electronic Products, 3370 Miraloma Avenue, Anaheim, CA 92803;
e-mail: arvind.i.d’souza@boeing.com. 2.—Rockwell Science Center, 1049 Camino Dos Rios,
Thousand Oaks, CA 91360
VLWIR (lc ~ 15 mm to 17 mm at 78 K) detectors have been characterized as a
functionoftemperaturetodeterminethedominantcurrentmechanismsimpact-
ing detector performance. Id – Vd curves indicate that VLWIR detectors are
diffusion limited in reverse and near zero bias voltages down to temperatures in
the40Krange. At30Kthedetectorsarelimitedbytunnelingcurrentsinreverse
bias. Since the detectors are diffusion limited near zero bias down to 40 K, the
RoAimp versus temperature data represents the diffusion current performance of
the detector as a function of temperature. The detector spectral response
measurement and active layer thickness are utilized to calculate the HgCdTe
layer x value and the optical activation energy Ea optical. The activation energy,
Ea electrical, obtained from the measured diffusion limited RoAimp versus temperature
data is not equal to the activation energy, Ea optical, obtained from the spectral
response measurement for all x values measured. Ea
= b* Ea optical, where
electrical
b ranges between 0.64 and 1.0 For cutoff wavelengths in the £ 9 mm at 78 K,
Ea electrical = Ea optical. Ea electrical = 0.65* Ea optical have been measured for lc = 17 mm at
78 K detectors. As the band gap energy decreases to values in the range of 70 meV
and lower, it is reasonable to expect a more dominant role of band tailing effects
on the transport properties of the material system. In such a picture, one would
expect the optical band gap to be unmodified, whereas the intrinsic concentra-
tion could be enhanced from its value for the ideal semiconductor. Such a picture
couldexplaintheobservedbehavior. Furtherprobingexperimentsandmodeling
efforts will help clarify the physics of this behavior.
Key words: Molecular beam epitaxy (MBE), VLWIR HgCdTe
detectors. Advances in the molecular beam epitaxy
(MBE) growth of HgCdTe and p-on-n double layer
planar heterostructure (DLPH) photovoltaic (PV)
detector architecture have resulted in producing high
performance detectors for infrared focal plane arrays
(IRFPAs). MBE is used to grow in-situ double layers;
VLWIR Hg1–xCdxTe base layer followed by wider
bandgapcaplayer. Layersdiscussedhereweregrown
on (211)B lattice-matched CZT substrates. Detectors
with p-type implants ranging in size from 8 mm in
diameter to 250 mm ¥ 250 mm square were fabricated.1,2
INTRODUCTION
Hg1–xCdxTe material lends itself to the fabrication
of high quantum efficiency detectors. The bandgap of
Hg1–xCdxTematerialcanbeeasilytunedfromthevery
shortwaveinfrared(VSWIR—lc =1–2mm)tothevery
long wave infrared (VLWIR—lc ~ 14–17 mm). This
bandgap tunability coupled with the high quantum
efficiency obtained in Hg1–xCdxTe detectors results in
Hg1–xCdxTe being extensively utilized for a range of
infrared imaging and sounding applications.
Infrared imaging radiometers and sounders re-
quire VLWIR (lc > 15 mm) detectors. It is, therefore,
important to understand the current mechanisms
that impact detector performance in these VLWIR
VLWIR DETECTOR RESULTS
Optical Measurements
Hg1–xCdxTe material transmission measurements
at room temperature are combined with processed
material detector spectral measurements at cryo-
(Received November 21, 2000; accepted January 13, 2001)
585

D’Souza, DeWames, Wijewarnasuriya,
Hildebrandt, and Arias
586
the model. The xopt value is a fitting parameter that
best replicates the shape of the measured spectral
responseorquantumefficiencyversuswavelengthchar-
acteristicsofthedetector.Anxopt =0.1997wasextracted
from the quantum efficiency versus wavelength dis-
played in Fig. 1. The xopt values for all the layers were
extracted in this manner. The xopt value thus ex-
tracted determines the bandgap at any temperature
via the Hansen, Schmit, Casselman equation,4
Eg (x,T) = -0.302 + 1.93x - 0.81x3
(1)
+0.832x3 + 5.35 ¥ 10-4 T(1- 2x)
I
det – Vdet versus Temperature
Fig. 1. Quantum efficiency versus wavelength for a lc = 15 mm at
78 K detector.
Idet – Vdet versus temperature curves for an 8 mm
circle (lc = 15 mm at 78 K) detector is displayed in
Fig. 2. For small bias values, the detector is diffusion
limited when T > 40 K. Analysis of the T = 50 K Idet
–
Vdet curve for low bias values in the forward and
reversedirection,resultsinanidealityfactorn=1.01,
provingthatthedetectorisdiffusionlimitednearzero
bias. The dark current values at Vdet = –50 mV,
–100 mV, and –150 mV are plotted in Fig. 3 as a
function of inverse temperature. At all three biases
the detector dark current is limited by diffusion cur-
rents for T > 60 K. At Vdet = –50 mV, the detector dark
current is limited by diffusion currents down to T ª
50 K. No generation-recombination currents are ob-
served in these measurements and is consistent with
previously reported data.5,6 At Vdet = –150 mV, T < 60 K
and Vdet = –100 mV, T < 50 K, the dark current has a
negativetemperaturecoefficient(dIdet/dT<0)andthe
detector is limited by band-to-band tunneling.
Fig. 2. Id – Vd curves versus temperature for an 8 mm diameter, lc =
15 mm detector.
RoAimp versus Temperature
RoAimp (Aimp is the detector p-n junction area) mea-
surements were made as a function of temperature
for a series of Hg1–xCdxTe detectors with x values
ranging from x = 0.1925 to x = 0.3127. Modeled RoA
and measured RoAimp versus inverse temperature
dataisplottedinFig. 4foraxopt =0.2981detector. The
theoretical model3 is a one-dimensional model that
assumes diffusion currents are dominated by the n-
side active layer and minority carrier recombination
is via Auger and radiative processes. Input param-
eters for the model are the Hg1–xCdxTe layer xopt value
as determined from spectral response and thickness
measurements, and carrier concentration from Hall
data. The experimental RoAimp versus inverse tem-
perature data displayed in Fig. 4 is parallel to the
modeled RoA versus 1/T calculation. Therefore, for
this xopt = 0.2981 case, Ea electrical = Ea optical The large
detector RoA data being equal to the modeled RoA
versus 1/T calculation, indicates that the detector is
limited by Auger and radiative processes.
Fig. 3. Dark current versus temperature at –50 mV, –100 mV, and
–150 mV detector bias.
genic temperatures to extract the material x value. A
Fourier transform infrared spectrometer is used to
measuretransmissionversuswavelengthforthelayer,
yielding material x value and thickness information.
Following Hg1–xCdxTe detector processing, the spec-
tral response and quantum efficiency of the detector
is measured. A modeled3 and measured quantum
efficiency versus wavelength curve is displayed in
Fig. 1 for a lc = 14.7 mm at 85 K (lc = 15.0 mm at 78 K)
detector.ThemeasuredHgCdTeactivelayerthickness,
detector cutoff wavelength, lc, are input parameters to
The data displayed in Fig. 5 is similar to the data
displayed in Fig. 4 for an xopt = 0.1997 detector. A
combination of Idet – Vdet curves and RoAimp versus
inverse temperature data was used to calculate the
temperature range within which the detectors are

Current Mechanisms in VLWIR Hg1-xCdxTe Photodiodes
587
(2)
diffusion limited. At temperatures greater than 100 K,
the modeled RoA line has a kink reflecting the inad-
equacy of the model in the intrinsic region. The
measured data deviates such that the RoAimp versus
inverse temperature slope at T is shallower than at
low temperatures. This shallower slope is also caused
by the detector now operating in the intrinsic region
where the detector exhibits properties of a p-i device7
with a dependence of qV/2kT. At temperatures be-
tween 100 K and 40 K, the RoAimp data is diffusion
limited as explained in the previous section, there is
nochangeintheslopeatlowtemperaturestoindicate
any shift to generation-recombination current as the
dominant current mechanism. However, as can be
observedinFig. 5, theslopeofthemeasuredelectrical
RoAimp versus inverse temperature data, is not equal
to the slope of the modeled diffusion limited RoA data.
The activation energy Ea electrical obtained from the
measured RoAimp versus inverse temperature data
displayed in Fig. 5 is,
Ea electrical = b*Ea optical
where b = Ea electrical/Ea optical = 0.67 in this case. Ea optical
istheactivationenergycalculatedat4Kusingthexopt
Fig. 5. RoAimp versus temperature for an 8 mm diameter detector from
the same wafer as the detector in Fig. 1.
Fig. 4. RoAimp versus temperature for a 250 mm diameter, lc = 5.12 mm
at 78 K detector .
Fig. 6. Calculated and measured majority carrier concentration versus
T, xelectrical = 0.1840.
Table I. MWIR to VLWIR Wavelength Measured and Calculated Data
lc at 78 K
in mm
Ea optical (T=4K)
in meV
Ea electrical
in meV
xopt
b
xelectrical
Ea opt – Ea
elect
17.4
17.0
17.0
16.2
16.2
15.0
15.0
14.2
12.8
12.1
11.8
10.8
10.8
10.1
9.7
0.1925
0.1942
0.1942
0.1956
0.1960
0.1997
0.2003
0.2025
0.2084
0.2121
0.2137
0.2195
0.2196
0.2245
0.2280
0.2616
0.2981
0.3081
0.3127
46.8
49.7
49.8
52.1
52.7
59.0
60.1
63.8
73.8
80.1
82.8
92.6
92.8
101.1
107.0
163.4
224.3
241.1
248.7
32.7
32.3
32.4
38.5
40.1
39.6
42.0
40.8
62.7
73.7
76.2
85.2
81.6
95.0
105.9
163.4
224.3
241.1
248.7
0.70
0.65
0.65
0.74
0.76
0.67
0.70
0.64
0.85
0.92
0.92
0.92
0.88
0.94
0.99
1.00
1.00
1.00
1.00
0.1843
0.1840
0.1841
0.1877
0.1888
0.1883
0.1897
0.1890
0.2019
0.2083
0.2098
0.2151
0.2130
0.2209
0.2274
0.2616
0.2981
0.3081
0.3127
14.03
17.38
17.40
13.53
12.66
19.48
18.02
22.96
11.07
6.41
6.62
7.41
11.13
6.07
1.07
6.8
0.00
5.1
0.00
4.8
0.00
4.7
0.00

D’Souza, DeWames, Wijewarnasuriya,
Hildebrandt, and Arias
588
Fig. 8. DE in meV versus Ea optical in meV.
Fig. 7. Ea eletrical /Ea optical versus lc.
ments described previously and Ea electrical the activa-
tion energy extracted from the measured RoAimp ver-
sus inverse temperature data.
The ratio Ea/Eg is plotted versus the detector cutoff
wavelength at 78K in Fig. 7. The data seems to fit to
an equation
value extracted from the spectral response curve and
the expression listed in Eq. 1. Equation 1 is also used
to extract xelect values from Ea electrical
.
Majority Carrier Concentration
Majority carrier concentration doped n-type with
Nd = at 7.8 ¥ 1014 cm–3 was measured as a function of
temperature for a HgCdTe layer with xoptical = 0.1942
asdeterminedfromspectralresponsemeasurements.
Usingthemeasuredmajoritycarrierconcentrationat
T = 300 K, the Hansen and Schmit8 expression for
intrinsic carrier concentration, Hansen, Schmit,
Casselman equation4 for bandgap and the Williams3
model, an x = 0.1820 value is extracted from the
models to match the measured majority carrier con-
centration at T = 300 K. The x = 0.1820 value is then
used to calculate3,8 the majority carrier concentration
as a function of temperature and with the measured
majority carrier concentration is plotted in Fig. 6 as
a function of temperature. Deviations exist in the
intermediate temperature range. The x = 0.1820
value extracted from the T = 300 K measurement is
close to the xelectrical = 0.1840 value obtained from the
Ea electrical measured value.
Ea
Ea
(lc - 10)
electrical
optical
b =
= 1-
(3)
25
However, this is just an empirical fit to the data, no
underlying physics information is obtained. What is
clear is that the ratio b is not a constant as the cutoff
wavelength(bandgap)changes, itrangesfrom0.64to
1.0. This implies that the electrical measurements
are not dominated by a midgap state or a state that
exists at a fixed percentage of the gap as has been
observed in SWIR detectors.9 Figure 8 is a plot of
bandgap difference DEg versus Ea optical. There does not
exist a constant DE for all the measured layers.
Therefore there does not exist a defect level in the gap
that is a fixed level of energy below the conduction
bandedgeorabovethevalencebandedge. Theconjec-
ture is that there exists band tailing of the density of
states10,11 into the gap that affects the electrical prop-
erties, giving rise to activation energies Ea electrical that
are smaller than the gap values Ea optical obtained from
optical measurements, the optical absorption coeffi-
cient between states in the band tails being low, and
thus not contributing to the optical response of the
detector. Therefore the observed behavior is related
to the intrinsic properties of the material and not to
discrete defect levels.
DISCUSSION
Table I lists parameters extracted from measured
data, the layers ranged from the MWIR to VLWIR
wavelengths. Column 1 is the detector cutoff wave-
length at 78 K. The xopt values in Column 2 are
obtained from the spectral response curves that in-
clude the effects of the active layer thickness. There-
fore layers with the same cutoff wavelength at 78 K
can have slightly different x values. Column 3 is the
calculated cutoff wavelength and the bandgap Ea optical
at 4 K using the Hansen, Schmidt, Casselman Eq. 1.
The activation energy Ea electrical obtained from the
RoAimp versus inverse temperature data is in Column 4.
Column 5 is a list of the b values for all the layers. The
values range from 0.64 to 1.0. A b value of 1.0 implies
that the measured RoAimp versus inverse temperature
data is parallel to the modeled diffusion limited data.
Column 6 is the xelect extracted from Ea electrical using
Eq. 1. Column 7 in Table I is a calculation of the
differenceDE=Eaoptical –Eaelectrical betweenthebandgap
energy Ea optical as measured from optical measure-
CONCLUSION
MWIR to VLWIR detectors have been character-
ized as a function of temperature to determine the
dominant current mechanisms impacting detector
performance. Thedetectorsarediffusionlimitednear
zero bias down to 40 K, the RoAimp versus temperature
data thus representing the diffusion current perfor-
mance of the detector as a function of temperature. In
small (Ea optical ~ 70 meV and lower) bandgap material,
the activation energy Ea electrical, obtained from the
measured diffusion limited RoAimp versus tempera-
ture data is not equal to the activation energy Eaoptical
obtained from the spectral response measurement.
,

Current Mechanisms in VLWIR Hg1-xCdxTe Photodiodes
589
What is obtained is, Ea electrical = b* Ea optical, where b
ranges between 0.64 to 0.75 for Ea optical < 70 meV. For
large (Ea optical > 200 meV) bandgap material, Ea electrical
= Ea optical. As the band gap energy decreases to values
in the range of 70 meV and lower, it is reasonable to
expect a more dominant role of band tailing effects on
the transport properties of the material system. It is
expected that the optical band gap be unmodified,
whereastheintrinsicconcentrationcouldbeenhanced
from its value for the ideal semiconductor. Such a
picture could explain the observed behavior.
Additional work is needed to validate this hypoth-
esis. Models that include the effects of band tailing
and overlaying bands should be used to calculate
intrinsic carrier concentration ni and the optical/
electrical properties of Hg1–xCdxTe material as a func-
tion of x. Measurement of Hg1–xCdxTe material with x
values ranging from 0.15 to 0.22, are needed to corre-
late measured optical absorption coefficient, electri-
cal conductivity, and intrinsic carrier concentration
with modeled values.
REFERENCES
1. J.M. Arias, J.G. Pasko, M. Zandian, S.H. Shin, G.M. Will-
iams, L.O. Bubulac, R.E. De Wames, and W.E. Tennant, J.
Electron. Mater. 22, 1049 (1993).
2. J. Bajaj, J.M. Arias, M. Zandian, D.D. Edwall, J.G. Pasko,
L.O. Bubulac, and L.J. Kozlowski, J. Electron. Mater. 25,
1394 (1996).
3. G.M. Williams and R.E. DeWames, J. Electron. Mater. 24,
1239 (1995).
4. G.L. Hansen, J.L. Schmit, and T.N. Casselman, J. Appl.
Phys. 53, 7099 (1982).
5. R.E. DeWames, G.M. Williams, J.G. Pasko, and A.H.B.
Vanderwyck, J. Cryst. Growth, 86, 849 (1988).
6. R.E. DeWames, J.G. Pasko, E.S. Yao, A.H. B. Vanderwyck,
and G.M. Williams, J. Vac. Sci. Technol. A6, 2655 (1988).
7. J. Lindmayer and C.Y. Wrigley, Fundamentals of Semicon-
ductor Devices (New York: Van Nostrand Reinhold Co.,
1965).
8. G.L. Hansen and J.L. Schmit, J. Appl. Phys. 54, 1639 (1983).
9. R.E. De Wames, D.D. Edwall, M. Zandian, L.O. Bubulac,
J.G. Pasko, W.E. Tennant, J.M. Arias, and A. D’Souza, J.
Electron. Mater. 27, 722 (1998).
10. M.H. Cohen, H. Fritzsche, and S.R. Ovshinsky, Phys. Rev.
Lett. 22, 1065 (1969).
11. N.F. Mott and E.A. Davis, Electronic Processes in Non-
Crystalline Materials, Second Edition (Oxford, U.K.:
Clarendon Press, 1979).