Stability and degradation in CdTe/CdS and
CIGS photovoltaic cells
Gary Hodes and David Cahen
Dept. of Materials and Interfaces
Weizmann Institute of Science, Rehovot
76100, Israel
Degradation rates of different types of modules/systems
<2000 >2000 <2000 >2000 <2000 >2000 <2000 >2000 <2000 >2000
Jordan & Kurtz, 2011 (August), National Renewable Energy Laboratory (NREL)
Photovoltaic degradation rates – An analytical review
CdTe/CdS
CdS
CdTe
CdSxTe1-x
TCO front contact
CdS
CdTe
back contact
Cu/Cu2-xTe/CdTe:Cu
Te-rich CdTe surface
Actual CdTe/CdS module stabilities
From: Photovoltaics International
Aug. 2012. N. Strevel, L. Trippel &
M. Gloeckler, First Solar, Perrysburg,
Ohio, USA
8.4% drop over 16.6 yr
0.5%/yr
Beginning 2011
0.6%/yr
From: Francesco Biccari
Master’s thesis 2012
480 kW thin film CdTe solar field installed in 2003
Tucson Electric/First Solar
First Solar’s long-term degradation modelling recommendation:
-0.5%/yr for moderate climates and –0.7%/yr for hot climates (-0.5% with ZnTe)
Diffusion of Cu from the back contact through the cell
From: Dobson et al., Sol. En. Mater. Sol. Cells, 62, 295 (2000)
Diffusion of Cu from back contact to CdS
Cu 200ºC/N2/15 h
Cu
From: Visoly-Fisher et al., Adv. Funct. Mater. 13, 289 (2003)
Accumulation of Cu in CdS due to stronger Cu-S bond (than Cu-Te) and relatively
Larger grain surface (smaller grains) of CdS.
How impurities move in the cell from different back contacts
Au
Au/Cu
Mo/Sb
Mo/Sb2Te3
(n.b. Cu contacts gave somewhat more efficient cells)
Au
Back contact
Au
Bätzner et al., Thin Solid Films, 451-2, 536 (2004)
What is the effect of Cu on the cell performance?
In the short term – positive
In the long term – if too much, causes degradation
Why?
Sensitivity of CdTe (CdCl2-treated) resistivity to Cu
CuCd
Cui
2.1018/cc
From: Perrenoud et al., J. Appl. Phys. 114, 174505 (2013)
There is always some Cu (and other impurities) in CdTe
Some of the causes of temporal changes in cell
Cu into CdTe and via grain boundaries to CdS
CuCd and [VCd-ClTe] acceptors in CdTe
Cui donor in CdTe
Optimum doping – space charge layer width
Cu deep acceptor in CdS
Cui donor in CdS
A little Cu acceptor compensates.
A lot forms recombination centres and
Increases RS. Also forms shunt path via
grain boundaries between CdTe and CdS
Cu+ to CdS enhanced under illumination since barrier is decreased
CuCd has a charge of -1 (relative to lattice) and therefore will drift in opposite direction to Cu+
Recovery of slightly/medium degraded cells by
Storage or anneal in dark could be due to
dissociation of acceptors (CuCd) or VCd-Cui to
VCd + Cui and back diffusion of Cu+ in field
CdS
CdTe
Predicted radiation damage to different cell types by protons or electrons
Bätzner et al., Thin Solid Films, 451-2, 536 (2004)
Decrease in efficiency for CdTe at high DS
Such high Ds would not normally occur in space for many years.
Performance recovery occurs over days/weeks in absence of radiation.
Not to forget – mechanical stability
Module thermal cycling (TC) is a standard test exposing modules in a controlled
environment between -40 ºC to + 85 ºC to test mechanical stability of the modules
(connections and differences in thermal expansion).
First Solar modules are not affected by these (or even more cycles) tests.
Long-term degradation – probable cause:
Drop in fill factor due to increase in RS believed due to changes in back contact
CIGS/CdS
<2000 >2000 <2000 >2000 <2000 >2000 <2000 >2000 <2000 >2000
??? Fundamental diff between the 2 cells due to thermodynamic
Preference of Cu for CdS in CdTe cell and for CIGS in CIGS cell?
ZSW
S. Niki et al., Prog. Photovolt: Res. Appl., 18, 453, (2010)
Thermodynamic reaction possibilities for CIGS cells
∆G KJ/mol
Mo
If CIS not Cu-rich (no Cu2Se )
No reaction
Increasing Ga, increased
Possibility of reaction
CIGS
CdS ZnO
Cu2Se + CdS Cu2S + CdSe
-5.6
In2Se3 + 3CdS 3CdSe + In2S3
-26
CdS + In2Se3 CdIn2(Se3)S
High temperature stability of cells
(Kijima and Nakada, Appl. Phys. Xpress, 1 075002 (2008))
not
heated
280ºC
360ºC
400ºC
At higher temperature, Cd diffuses to CIGS
Space charge layer moves towards Mo.
Stable up to 320 ºC
(30 min in vacuum)
Shows high chemical stability
Even if chemically stable, there could be other
electrical degradation mechanisms
Formation of defects, e.g. through Cu+ diffusion/migration
e.g., low energy (of formation) defects:
[Vcu] [2VCu InCu]
complex neutral defect
or
changes in ZnO stoichiometry
Large variation in degradation rates from module to module,
but main effect is drop in fill factor (increase in RS)
Bias-induced Cu migration in CIGS
CIGS
Cu+
CIGS
Cu+
CdS
CdS
Operating conditions (PMAX)
Field at the interface of CIS/CdS pulls the Cu+ into the CIS makes the CIS ‘surface’ Cu poor
Reverse bias may occur for single cells within a series-connected module
By partial shading of the module.
Thus, the cell should be stable against:
Forward bias, found under normal operation
Reverse bias
We thus conclude that the electrical stability of CI(G)S is not static
but is rather dynamic. In other words, stability is not obtained
because the material is “strong”, but because it is flexible.
Guillemoles et al. J. Phys. Chem. B, 104, 4849 (2000)
Maybe also true to some extent for CdTe/CdS?
Overall conclusions
There are many possible degradation pathways for CdTe and CIGS cells
Results show module/system stabilities have improved
(along with efficiencies) over the years
Present mean panel/system efficiencies are 0.5% for CdTe and
1% for CIGS, but can expect this to fall considerably for CIGS
<2000 >2000
>2000
<2000 >2000
<2000 >2000
<2000 >2000
<2000
Thanks to Kevin Dobson (Brian McCandles and Steve Hegedus)
(U. Delaware) for helpful input