Page 73 - Mirjam-Theelen-Degradation-of-CIGS-solar-cells
P. 73
Stability of Cu(In,Ga)Se 2 Solar Cells
If Na O was present in the glass, additional sodium migrated in the PID samples due
2
to the voltage from the glass to the solar cell. Already after 25 hours exposure to heat
and voltage, all samples with a high Na O content in the glass had degraded severely.
2
On the other hand, solar cells on glass without or with small quantities of NaO did
2
retain most of their efficiency. It should be noted that the sample with a relatively high
amount of K O and a low amount of Na O did have a constant efficiency, but it did
2
2
show a roll-over in the IV curve. It was observed that the degraded samples contained
a large amount of sodium in the CdS layer and maybe also in the top region of the
CIGS layer. It was proposed that this sodium actually migrated from the glass and that
the sodium flux is the product of the concentration of sodium in the glass and the mo -
bility of the sodium ions. High ion mobilities occur within glass with low electrical re-
sistivity, which is more susceptible to sodium ion drift and subsequent out-diffusion.
Based on the results of Fjallström et al., it can be expected that the diffusion of alkali
elements, and likely mainly sodium, plays a dominant role in the degradation of CIGS
solar cells and modules under combined damp heat and electrical bias exposure. The
presence of an electrical bias can be obtained by both illumination and an externally
applied electrical bias. Due to the limited studies on this topic, the differences and
similarities between the influences of electrical bias and illumination exposures was
not yet distinguished.
Several studies based on dry heat tests combined with the application of electrical
biases have been executed by Mack and Ott [98-100].
Mack et al. [98] compared the impact of 200 hours simultaneous exposure to various
biases (electrical biases of +100, +200, +400 or -100 mV or illumination) and 165 o C. They
observed that illumination and positive electrical biases had a positive impact on the
net doping content and the non-radiative recombination and thus on the V, while
oc
the negative electrical bias led to faster performance decrease. A positive electrical
bias also led to stabilisation of the fill factor, which otherwise changed due to the ac-
cumulation of negative charges at the hetero-interface. Since the changes can be in-
duced by light soaking or by a voltage, it was concluded that they are caused by the
bias across the junction and not the photogeneration itself. It was proposed that the
change in net doping concentration, which impacted the V , influenced the position
oc
of the Fermi level at the hetero-interface. This leads to the formation of a p + layer, which
influences the fill factor as well, as is described extensively by Ott et al. [99]:
Mack et al. exposed CIGS solar cells to elevated temperatures in the dark (24 hours
o
at 165 C). When the current voltage curves of these solar cells were measured under
spectral edge filters, it was observed that the fill factor decreased when measured by
light with a wavelength over 550 nm, while shorter wavelength measurements still
71
If Na O was present in the glass, additional sodium migrated in the PID samples due
2
to the voltage from the glass to the solar cell. Already after 25 hours exposure to heat
and voltage, all samples with a high Na O content in the glass had degraded severely.
2
On the other hand, solar cells on glass without or with small quantities of NaO did
2
retain most of their efficiency. It should be noted that the sample with a relatively high
amount of K O and a low amount of Na O did have a constant efficiency, but it did
2
2
show a roll-over in the IV curve. It was observed that the degraded samples contained
a large amount of sodium in the CdS layer and maybe also in the top region of the
CIGS layer. It was proposed that this sodium actually migrated from the glass and that
the sodium flux is the product of the concentration of sodium in the glass and the mo -
bility of the sodium ions. High ion mobilities occur within glass with low electrical re-
sistivity, which is more susceptible to sodium ion drift and subsequent out-diffusion.
Based on the results of Fjallström et al., it can be expected that the diffusion of alkali
elements, and likely mainly sodium, plays a dominant role in the degradation of CIGS
solar cells and modules under combined damp heat and electrical bias exposure. The
presence of an electrical bias can be obtained by both illumination and an externally
applied electrical bias. Due to the limited studies on this topic, the differences and
similarities between the influences of electrical bias and illumination exposures was
not yet distinguished.
Several studies based on dry heat tests combined with the application of electrical
biases have been executed by Mack and Ott [98-100].
Mack et al. [98] compared the impact of 200 hours simultaneous exposure to various
biases (electrical biases of +100, +200, +400 or -100 mV or illumination) and 165 o C. They
observed that illumination and positive electrical biases had a positive impact on the
net doping content and the non-radiative recombination and thus on the V, while
oc
the negative electrical bias led to faster performance decrease. A positive electrical
bias also led to stabilisation of the fill factor, which otherwise changed due to the ac-
cumulation of negative charges at the hetero-interface. Since the changes can be in-
duced by light soaking or by a voltage, it was concluded that they are caused by the
bias across the junction and not the photogeneration itself. It was proposed that the
change in net doping concentration, which impacted the V , influenced the position
oc
of the Fermi level at the hetero-interface. This leads to the formation of a p + layer, which
influences the fill factor as well, as is described extensively by Ott et al. [99]:
Mack et al. exposed CIGS solar cells to elevated temperatures in the dark (24 hours
o
at 165 C). When the current voltage curves of these solar cells were measured under
spectral edge filters, it was observed that the fill factor decreased when measured by
light with a wavelength over 550 nm, while shorter wavelength measurements still
71
