Transcript No Slide Title

Defect-related trapping and recombination
in metamorphic GaAs0.72P0.28 grown on GaAs
Tim Gfroerer, Peter Simov, and Brant West, Davidson College, Davidson, NC
Mark Wanlass, National Renewable Energy Lab, Golden, CO
Abstract
Since the bandgap of metamorphic GaAsP can be tuned through a range of energies in a high-radiance region of the solar spectrum, this alloy may be
useful in multi-junction solar cells. For this application, achieving high conversion efficiency will require a good understanding of the defects that
accompany lattice-mismatch in this system. We use photoluminescence and deep-level transient spectroscopy (DLTS) on GaAs0.72P0.28 to identify defectrelated energy levels and assess their impact on photovoltaic device performance. Low temperature photoluminescence spectra reveal a broad defectrelated band approximately 0.10 eV below the band-to-band emission. At 165 K and above, thermal activation out of the defect-related band appears to
enhance the band-to-band radiative mechanism. DLTS measurements on p-type GaAs0.72P0.28 suggest the presence of a hole trap with a comparable
depth (~0.09 eV) and thermal escape temperature, but the collective results are difficult to interpret. An alternative approach assuming a reciprocal
rather than logarithmic dependence on time is compelling, but this hopping-transport model will require further testing.
DLTS: Trapping During a Bias Pulse
-2
-1
Solar Spectral Irradiance (Wm nm )
Motivation: Multi-junction solar cells
Depletion Layer
2.1eV GaInP
1.50
- - - - - - - N+ - - - -
1.75eV GaAsP
1.25
GaAs bandgap
1.00
0.75
0.50
Visible
0.25
1200
1600
2000
2400
GaAs0.72P0.28 (S)
N+
+
P
P
GaAs0.72P0.28 (Zn)
+
+-
Ga0.65In0.35P (Zn)
Barrier
+
Lattice-matched GaInP barrier
GaAsP buffer
GaAsP Lattice-mismatch step grade
GaAs Substrate
{
GaAs0.72P0.28
GaAs0.78P0.22
GaAs0.83P0.17
GaAs0.89P0.11
GaAs0.94P0.06
Metal-organic vapor phase epitaxy structures incorporate a GaAsP stepgraded buffer to accommodate the ~ 1% lattice mismatch between the
double heterostructure and the substrate.
40ms RF
40ms RS
Ea = 91meV
1
155K
160K
165K
170K
175K
180K
0.5
0.1
Ea = 81meV
1
10
0.2
0.3
Ea = 91meV
60
0.4
65
70
1.88
Energy (eV)
Photoluminescence spectra show a broad defect-related band (DRB)
approximately 0.1eV below the 77K and 120K band-to-band (BB) emission.
W
Hopping Sequence
2
1
3
+
-
3
10
155K
160K
165K
170K
175K
180K
Depletion Layer
Ea = 0.38eV
We thank Jeff Carapella for growing and processing the test
structures and the Donors of the American Chemical Society –
Petroleum Research Fund for supporting this work.
40 ms response
400 ms response
Time (s)
(with bias)
Acknowledgement
1
0.4
+
-
Reciprocal transient behavior may be due to anomalous transport,
where trapped carriers hop from defect site to defect site rather
than being thermally activated and propagating in a band state.
2
10
10
0.3
+
-
+ + +
+
+
+ +P
+ +
+ +
+
+
+
+
+
+ +
+
+ + +
+ +
5
0.2
• Deduced rates should not depend on the observation
time window
• Fast and slow rates should correspond to shallow and
deep levels with very different activation energies
• If the non-exponential behavior is due to a broad
defect-related band, we may need to account for the
continuous distribution of levels contributing to the
DLTS signal
• Alternatively, the functional form of the transients
themselves could be reconsidered (see compelling
reciprocal analysis below)
Mechanism for Reciprocal Transients?
Steady-state Bias = -2V
Pulse height = +1V
0.1
Discussion
1/kT (eV )
10
Capacitance Change (pF) on a reciprocal scale
1.84
Position
75
Transient capacitance following a brief pulse towards zero bias, plotted on a logarithmic scale.
The Arrhenius plot shows the fast and slow (RF and RS) escape rates deduced from double
exponential fits for short (40 ms) and long (400ms) observation windows.
1
0.0
1.80
Depletion
region
-1
Time (s)
BB
0
1.76
Ea = 93meV
DLTS Results: Reciprocal Analysis
DRB
+ + +
Valence band
----
400ms RS
2
Excitation = 532nm
2
Intensity ~ 20 W/cm
0.1eV
Quasi EF,n
The N+/P doping profile across the junction implies that capacitance transients
are dominated by hole trapping at the depletion edge in the p-type GaAsP.
10
3x10
1x10
Quasi EF,p
400ms RF
 Escape 

  e  Ea / kT
 Rate 
5
Photoluminescence Spectra
4
Energy
GaAs Substrate
Steady-state Bias = -2V
Pulse height = +1V
0.0
300K
250K
165K
120K
77K
Conduction band
GaAsP step-grading
-1
GaAs0.72P0.28 active layer
10
Capacitance Change (pF) on a log scale
GaInP window
{
N+ P Junction
DLTS Results: Conventional Exponential Analysis
PL Test Structure (not to scale)
PL Intensity (a.u.)
-
-
+-
Deep level transient spectroscopy (DLTS) employs transient capacitance
measurements on diodes during and after the application of a bias pulse to
monitor the capture/emission of carriers into/out of defect-related traps.
Higher energy photons are absorbed in higher bandgap alloys, reducing
the heat loss caused by excess photon energy relative to the gap.
1.72
-
-
+
Rate (s )
800
Wavelength (nm)
4
-
Slope of 1/C(t) (a.u.)
400
2x10
+
+
+
+
+
Ga0.65In0.35P (S)
Barrier
Depletion with bias
0.00
4
DLTS Test Structure (not to scale)
60
65
70
75
-1
1/kT (eV )
The same capacitance transients as those plotted logarithmically above, but here plotted on a
reciprocal scale. Here, the slope of the reciprocal transient is used on the Arrhenius plot for
the same observation time windows.
80