Stability of CdTe Cells with Different

Device Structures
Jim Sites, Colorado State University
NREL Photovoltaic Module Reliability Workshop
February 26, 2014

with former students: Jason Hiltner, Alex Pudov,

Sam Demtsu, and Caroline Corwine
and current students John Raguse and Russell Geisthardt,
and research associates Jennifer Drayton and Tyler McGoffin
Funding from DOE F-PACE program

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 1

 Cell-Level Stability Testing

Motivation for cell-level CdTe stability testing:

● Possible intrinsic issues with CdTe cells
● Additional issues can arise with different device structures
● Device analysis generally assumes good stability; flawed
otherwise

Multiple stress parameters important:

● Temperature ● Illumination level
● Voltage bias ● (Humidity)

Multiple tracking measurements important:

● Average cell properties: J-V, C-V, (QE, J-V-T)
● Cell uniformity: LBIC, EL

Focus on plausible physical mechanisms

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 2

 Accelerated Life Testing

Uses compact
lab oven
(60-100°C)
originally
stadium lights

John Raguse,
Jennifer Drayton, and
Russell Geisthardt

LED arrays for

illumination
(3 cm diam;
0-2 suns)
Shelf to hold cells

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 3

 Changes in CdTe Current-Voltage
Change seen earlier in first quadrant, later in power quadrant.
Increasingly greater at higher temperatures.

From Samuel Demtsu

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 4

 Voltage and Illumination Dependence
Bias voltage during stress
seemed to be the larger factor,
but contacting also played a role

1.0
Light Stress
0.9
V=0
0.8 Dark Stress

 0.7 V<0 V>0


0.6
Hiltner and Sites, AIP Conf.
Series 462, 170 (1998) 0.5
J >> 0
0.4
Cells from NREL and
Solar Cells, Inc 0.3
V<0 SC Rmp/2 MP 2Rmp OC J>0

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 5

 Long-Term Expectations
,

Define Acceleration Factor a = exp(Ea/kTeff – Ea/kTcell)

where an activation energy Ea (typically ~1 eV) can be
deduced from rate of change at different temperatures
and Teff from a histogram of daylight module
temperatures (and an estimate of Ea)
Effective Temperature Assuming Activation Energy E a

350

Ea
300 Teff 
  Ni 
k B ln   N exp(i Ea ) 
Number of Hours per Year

250
i i kBTi

200
Teff[Ea = 0.5]
Teff[Ea = 1.0]
150
Teff[Ea = 1.5]

100

50
NREL Data

0
-20 0 20 40 60 80
o
Module Field Temperature [ C]

For CdTe with Tcell = 85°C and typical field

temperatures, a appears to be in the range of 100-1000

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 6

 Copper Model
CSU Cells
30 Copper reduces the CdTe back-
Initially contact barrier, but can diffuse away.
Current Density [mA/cm2]

20

10
Positive ions diffuse faster when field
is reduced at VOC and above.
0
Small amount of copper may not be
-10 no Cu
sufficient, and may diffuse from back.
increasing
-20 Cu

-30
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
Voltage [V]

30
Small amount of copper
Current Density [mA/cm 2]

20

10

-10 increasing stress time

no stress
-20
Caroline Corwine et al, SOLMAT 82, 481 (2004)
-30
-0.2 0.0 0.2 0.4 0.6 0.8 1.0
Voltage [V]

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 7

 Effect on Capacitance

Copper diffusion appears to also increase carrier density

CSU CdTe cells Hole density is derived from C-V

Factor
of two

Thickness
Jennifer Drayton and John Raguse of CdTe

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 8

 Alternative CdTe Structures

Transparent
Window and Transparent
Standard
Electron Back Contact
Baseline
Reflector
Glass

TCO
CdS

CdTe

CMT
MoOx, MoON
Al/Au
Ni

 = 13%  ≈ 19%
Target Studies in progress
16.2% achieved

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 9

 Varying the Back Contact

with
graphite

Demtsu, Albin, Pankov, and Davies,

SOLMAT 90, 2934 (2006)

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 10

 Light-Beam-Induced Current
Steps through 10,000 points in 10 min

Resolution
down to 1 μm at
1 sun intensity

Built by Jason Hiltner

Updated by Tim Nagle
and Russell Geisthardt

638
nm 830
Initial 8 hrs 8 days Three Primary nm
wavelengths
405 nm for CdTe

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 11

 Different Amounts of Copper
100°C, illuminated, short circuit (CSU cells)
30
Current Density [mA/cm ]
2

20 4 min Cu
10 0 - 50 days
0
Initial Stressed
-10

-20
30
Current Density [mA/cm ]
2

20 0.5 min Cu
10
0 - 32 days

-10

-20
30
Current Density [mA/cm ]
2

20 No Cu
0 - 16 hours
10

0 Cu
-10

-20

-0.2 0.0 0.2 0.4 0.6 0.8 1.0

Voltage [V]

A. Pudov et al, PVSC-29 (2002)

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 12

 Electroluminescence (EL)

John Raguse and Tyler McGoffin

IEL ~ exp(VOC/kT)

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 13

 EL and Voltage of Stressed CdTe Cells
EL tracks voltage; gives confidence to both
Pre-stress 1 hr stress 24 hr stress 36 hr stress 78 hr stress 144 hr stress

VOC =
802 mV 787 mV 795 mV 801 mV 801 mV 809 mV

803 mV 790 mV 790 mV 807 mV 805 mV 812 mV

808 mV 792 mV 801 mV 804 mV 802 mV 806 mV

John Raguse and Jennifer Drayton

• Devices are standard CSU cell recipe

• Devices stressed at 65°C, at VOC under nominal 1 sun illumination
• Decline and recovery appears real. Two effects from copper diffusion?
 Summary

(1) Cell-level CdTe stability is generally

good, but needs to be tested with new
device structures.
(2) Copper used with back contact is
responsible for at least some of the
change.
(3) A mix of tracking measurements,
including uniformity, is highly desirable.
(4) There can be small stability issues that
have little effect on performance, but
can compromise analysis and could be
precursor to later trouble.

February 26, 2014 Reliability Workshop – Jim Sites, Colorado State 15