Page 59 - Mirjam-Theelen-Degradation-of-CIGS-solar-cells
P. 59
Stability of Cu(In,Ga)Se 2 Solar Cells
The formation of the undoped ZnO is here linked to the dissociative chemisorption of
water, which leads to the removal of free electrons from donor sites, which are more
populated at the grain boundaries. It should be noted that the measured decrease of
zinc to oxygen ratio can also be caused by the formation of other species like Zn(OH) .
2
Owen et al. [76] also proposed the formation of stoichiometric ZnO in the grain
boundaries due to the adsorption of water. This would lead to the loss of the effect
of aluminium doping and thus to a decrease of the local carrier concentration. This
can then lead to an increase of depletion regions and potential barriers at the grain
boundaries. This effect is not visible in the global carrier concentration, since this re-
duction only happens locally.
2.3.4.1.4 Influence of substrate temperature and crystallinity
Pern et al. [25] observed that ZnO:Al and (Zn,Mg)O:Al samples sputtered at a higher
substrate temperature degrade slower than those deposited at ambient tempera-
ture. Theelen et al. [71] confirmed this observation: a higher sputtering temperature
o
(200 C compared to RT) of ZnO:Al led to the formation of larger grains, which were
also more stable under damp heat treatments.
This relationship was not found for all deposition techniques: for Pulsed Laser Depo-
sition (PLD) samples, it was observed that a varying substrate temperature between
o
68 C and 200 C did not influence the degradation greatly [68]. However, it was also
o
found that films with a better crystallinity had a higher mobility, due to bigger crys-
tallite size and more crystals with the same orientation, reducing the effect of grain
boundary scattering.
2.3.4.1.5 Influence of sample thickness
It was observed that thick samples degrade slower than thin samples [25,67,68,75].
This behaviour can be expected if diffusion is the degradation rate determining
step in the degradation process. Pern et al. [25] described that thicker ZnO:Al or
(Zn,Mg)O:Al samples degrade slower than thinner ones. This was explained by either
a degradation process that started from the surface and gradually deepened into the
bulk [25] or a difference in grain sizes or structures for different thicknesses [75]. Since
the penetration of the atmospheric species is a slow process, the thick samples can
have a non-degraded bottom layer even after long exposure to a damp heat test,
while thin samples are then completely degraded.
In several studies, an increase in TCO layer thickness due to water exposure was ob-
served. Very strong effects were observed by Feist et al. [29], who degraded conduc-
57
The formation of the undoped ZnO is here linked to the dissociative chemisorption of
water, which leads to the removal of free electrons from donor sites, which are more
populated at the grain boundaries. It should be noted that the measured decrease of
zinc to oxygen ratio can also be caused by the formation of other species like Zn(OH) .
2
Owen et al. [76] also proposed the formation of stoichiometric ZnO in the grain
boundaries due to the adsorption of water. This would lead to the loss of the effect
of aluminium doping and thus to a decrease of the local carrier concentration. This
can then lead to an increase of depletion regions and potential barriers at the grain
boundaries. This effect is not visible in the global carrier concentration, since this re-
duction only happens locally.
2.3.4.1.4 Influence of substrate temperature and crystallinity
Pern et al. [25] observed that ZnO:Al and (Zn,Mg)O:Al samples sputtered at a higher
substrate temperature degrade slower than those deposited at ambient tempera-
ture. Theelen et al. [71] confirmed this observation: a higher sputtering temperature
o
(200 C compared to RT) of ZnO:Al led to the formation of larger grains, which were
also more stable under damp heat treatments.
This relationship was not found for all deposition techniques: for Pulsed Laser Depo-
sition (PLD) samples, it was observed that a varying substrate temperature between
o
68 C and 200 C did not influence the degradation greatly [68]. However, it was also
o
found that films with a better crystallinity had a higher mobility, due to bigger crys-
tallite size and more crystals with the same orientation, reducing the effect of grain
boundary scattering.
2.3.4.1.5 Influence of sample thickness
It was observed that thick samples degrade slower than thin samples [25,67,68,75].
This behaviour can be expected if diffusion is the degradation rate determining
step in the degradation process. Pern et al. [25] described that thicker ZnO:Al or
(Zn,Mg)O:Al samples degrade slower than thinner ones. This was explained by either
a degradation process that started from the surface and gradually deepened into the
bulk [25] or a difference in grain sizes or structures for different thicknesses [75]. Since
the penetration of the atmospheric species is a slow process, the thick samples can
have a non-degraded bottom layer even after long exposure to a damp heat test,
while thin samples are then completely degraded.
In several studies, an increase in TCO layer thickness due to water exposure was ob-
served. Very strong effects were observed by Feist et al. [29], who degraded conduc-
57
