Copper and BufferLayer Effects

1
Copper and Buffer-Layer Effects Dave Albin
(NREL) Samuel Demtsu (CSU) 
2
Introduction

• Reach-through diode model prediction
• - Low Voc independent of light intensity
• No roll-over , steep J-V
• Saturation Current dependent on Intensity
• Backcontact removal leaves high Vspv

Glass
Tech 15
TCO
1000 Å sputtered Buffer
CdS
1200 Å CSS
CdTe
3.5 ?m CSS
Back contact
Y.Roussillion, V.G. Karpov, Diana Shvydka,
J.Drayton, and A.D.Compaan, Back contact and
reach-through diode effects in thin-film
photovoltaics, J.Appl.Phys., 96(12) 2004.
3
JV Curve (UT Cell)
Y.Roussillion, V.G. Karpov, Diana Shvydka,
J.Drayton, and A.D.Compaan, Back contact and
reach-through diode effects in thin-film
photovoltaics, J.Appl.Phys., 96(12) 2004.
4
JV Curves (NREL cells)
CVD FSnO2
Glass
Glass
TCO
1000 Å CVD i-SnO2
TCO
CdS
CdS
800 Å CBD CdS
CdTe
CdTe
8-10 um CdTe
Graphite
Graphite
CuxTeHgxTe
Ag - paste
Ag - paste
5
Low-Light Voc (UT Cell)
Y.Roussillion, V.G. Karpov, Diana Shvydka,
J.Drayton, and A.D.Compaan, Back contact and
reach-through diode effects in thin-film
photovoltaics, J.Appl.Phys., 96(12) 2004.
6
Low-Light Voc (NREL Cells)
7
Low-Light Voc (NREL Cells)
Jsc and Voc varied with illumination as one might
expect from the diode equation (linear, ln)
8
Light-Soak Stress Testing (100C Voc bias)
Cu containing devices start at higher performance
(?) but degrade to a level comparable to their
non-Cu containing equivalents Devices with Buffer
degraded differently depending upon presence of
Cu. (With Cu, larger degradation in Voc, without
Cu, larger degradation in FF. Devices without
Buffer degraded in similar fashion
9
NREL buffer not Reactive
Early cell development (1994-1997) used a
buffer layer similar to one used by UT (room
temperature sputtered) buffer layer (also called
i-SnO2) susceptible to H2 anneals SnO2 reduced
to Sn
Albin, et al., NREL/SNL Photovoltaics Program
Review, AIP Press, 665 (1997)
10
NREL buffer not Reactive
If CBD CdS were present (e.g., making a device),
Sn was observed to react and form SnS Formation
of such defects correlated with reduced
Voc Hardening of room-temperature sputtered
buffer layers achieved using subsequent high T O2
anneals Today, we use high T, CVD-grown i-SnO2 as
our buffer which is NOT reactive
11
Conclusions

• In this experiment, we could not duplicate
  effects predicted by the reach-true diode model
• The presence (or absence) of buffer (i-SnO2) had
  only a minor effect on Voc.
• The presence (or absence) of Cu had the major
  impact on performance
• no roll-over in Cu-containing devices
• significant cross-over (light dark JV) in
  Cu-containing devices
• higher Voc with Cu
• Jsc and Voc varied with illumination in normal
  fashion (linear, ln respectively)
• Devices made on buffer layers showed strong
  differences in degradation between Cu-containing
  and Cu-free (intentional) devices
• Reasons exist for why results did not replicate
  previous UT results
• devices were made using old samples
• strong differences in buffer layer structure,
  chemistry, reactivity
• differences in CdTe thickness (UT devices
  thinner) and CdS type (CBD vs CSS)