Perovskite-kesterite monolithic tandem solar cells with high open-circuit voltageTeodor Todorov, Talia Gershon, Oki Gunawan, Charles Sturdevant, and Supratik Guha Citation: Applied Physics Letters 105, 173902 (2014); doi: 10.1063/1.4899275 View online: http://dx.doi.org/10.1063/1.4899275 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/105/17?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Suns-VOC characteristics of high performance kesterite solar cells J. Appl. Phys. 116, 084504 (2014); 10.1063/1.4893315 Band tailing and efficiency limitation in kesterite solar cells Appl. Phys. Lett. 103, 103506 (2013); 10.1063/1.4820250 Indications of short minority-carrier lifetime in kesterite solar cells J. Appl. Phys. 114, 084507 (2013); 10.1063/1.4819849 Admittance spectroscopy in kesterite solar cells: Defect signal or circuit response Appl. Phys. Lett. 102, 202105 (2013); 10.1063/1.4807585 Copper-phthalocyanine-based organic solar cells with high open-circuit voltage Appl. Phys. Lett. 86, 082106 (2005); 10.1063/1.1871347
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.76.102.52 On: Fri, 14 Nov 2014 10:07:54
Perovskite-kesterite monolithic tandem solar cells with high open-circuitvoltage
Teodor Todorov, Talia Gershon, Oki Gunawan, Charles Sturdevant, and Supratik Guhaa)
IBM T. J. Watson Research Center, Yorktown Heights, New York 10598, USA
(Received 5 September 2014; accepted 3 October 2014; published online 28 October 2014)
We report a monolithic tandem photovoltaic device with earth-abundant solution processed absorb-
ers. Kesterite Cu2ZnSn(S,Se)4 and perovskite CH3NH3PbI3 solar cells were fabricated monolithi-
cally on a single substrate without layer transfer. The resulting devices exhibited a high open
circuit voltage (Voc) of 1350 mV, close to the sum of single-absorber reference cells voltages and
outperforms any monolithic tandem chalcogenide device (including Cu(In,Ga)Se2) reported to
date. Ongoing optimization of several device elements including the severely limiting top contact
electrode is expected to yield superior currents and efficiency. Importantly, our device architecture
demonstrates the compatibility and synergistic potential of two of the most promising emerging
photovoltaic materials and provides a path for optimization towards >20% efficiency. VC 2014AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4899275]
Chalcogenide solar cells such as Cu(In,Ga)Se2 (CIGS)
and Cu2ZnSn(S,Se)4 (CZTSSe) can benefit from a partner
material suitable for a tandem photovoltaic device that would
allow broader spectrum light harvesting and enhanced effi-
ciency.1 Researchers have attempted different approaches in
the past including mechanical stacking of separate devices
with different band gap absorbers via external wiring of
CIGS-CdTe,2 CIGS-dye sensitized solar cell,3 CIGS-CGS4
and CIGS-CIGS5 cells with different ratios of In:Ga.
However, the use of multiple transparent conducting layers
with related losses due to transmission and series resistance
as well as the additional interconnections needed for external
wiring of the two separate solar cells limits the practicality
of these concepts. A preferred monolithic design on a single
substrate would theoretically be capable of delivering supe-
rior performance and lower fabrication cost. However, the
processing and materials constraints associated with the
preparation of highest-efficiency chalcogenide devices such
as elevated absorber crystal growth temperatures (above
400 �C), the rapid deterioration of the p-n junction above
220 �C and non-transparent Mo bottom contact have been a
significant barrier to the fabrication of efficient CIGS or
CZTSSe based monolithic tandem devices. One attempt
towards a CIGS-organic solar cell tandem via a sophisticated
layer transfer reached an efficiency of 3.8%.6 With earth-
abundant CZTSSe solar cells, neither monolithic nor mechani-
cally stacked tandem devices have been reported to date.
Recently, perovskite organo-lead halide solar cells have
made rapid progress with the demonstration of solar cells
with over 16% efficiency.7 The low processing temperatures
of metal-organic (MO) perovskite devices (<150 �C) and
tunable band gap from 1.4 to over 2 eV,8 pair ideally with
the lower band gap, and therefore higher short-circuit current
density (Jsc) of the CIGS and CZTSSe families of chalcoge-
nide solar cells for forming a tandem photovoltaic device.
The highest-efficiency CIGS and CZTSSe are achieved at
relatively low band gaps from about 1.1 to 1.25 eV, where
the sulfur content is low and/or the In:Ga ratio is high
(for CIGS). Concomitantly, the open circuit voltage (Voc)
deficit9 in CZTSSe is also significantly reduced when shift-
ing to lower sulfur contents or band gaps closer to 1 eV. At
the same time, perovskites have demonstrated their highest
efficiencies at band gaps of 1.55 eV with high Voc values of
1.1 V (Ref. 10) for NH3CH3PbIxCl3�x. The band gaps can
be further increased to �1.7–1.9 eV (Ref. 8) by Br or Cl
addition for improved current matching with the bottom cell
in tandem structures. The two materials systems (CZTSSe
and MO-perovskites) are, therefore, well matched both in
terms of opto-electronic properties as well as processing
compatibilities for integration in a monolithic tandem solar
cell architecture.
In this paper, we demonstrate the fabrication and opera-
tion of a monolithic CZTSSe-perovskite tandem device
structure consisting of Glass/Mo/Cu2ZnSn(S,Se)4/CdS/ITO/
PEDOT:PSS/NH3CH3PbI3/PCBM/Al, where PEDOT:PSS
(poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) and
PCBM ([6,6]-phenyl C61 butyric acid methyl ester) are organic
hole- and electron-conducting species, respectively, and are
commonly employed in organic photovoltaics.
The scope of this report is a proof of concept of the
chalcogenide-perovskite monolithic architecture and does
not focus on individual optimization of the different ele-
ments, e.g., band gap tuning of the absorbers or transparent
top contact engineering. It does however demonstrate a via-
ble fabrication sequence for producing a working monolithic
tandem device with cumulative Voc. Preliminary results and
modeling indicate that improved transparent conductive con-
tact on our current CZTSSe and perovskite devices (power
conversion efficiency g� 12% each) could produce tandem
devices with efficiency up to 16.0%. Further band gap opti-
mization and integration of state-of-the-art top perovskite
cell with >15% efficiency is projected to reach tandems with
over 20% efficiency.
The device (SEM cross-section shown in Fig. 1) was
built on a molybdenum coated soda-lime glass substrate and
a)Author to whom correspondence should be addressed. Electronic mail:
0003-6951/2014/105(17)/173902/4/$30.00 VC 2014 AIP Publishing LLC105, 173902-1
APPLIED PHYSICS LETTERS 105, 173902 (2014)
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.76.102.52 On: Fri, 14 Nov 2014 10:07:54
comprises a bottom CZTSSe-based solar cell, prepared by a
solution process described in detail previously.11,12 The
CZTSSe absorber used in the study was prepared via a pure
solution deposition approach, where the precursors dissolved
in hydrazine are spin-coated onto a Mo-coated soda lime
glass substrate and subsequently annealed to form a large-
grained kesterite layer with a band gap of approximately
1.1 eV (details in Ref. 13). CdS was then deposited by chem-
ical bath deposition, followed by a sputtered ITO top elec-
trode. On top of this completed device, we prepared a top
cell based on a perovskite absorber in a manner similar to
that described previously.14 A PEDOT:PSS layer was spun
directly on the top ITO contact of the completed CZTSSe
cell, and was then annealed in air at �140 �C to remove
excess water. The samples were then moved to a nitrogen-
filled glove box for subsequent processing. The
NH3CH3PbI3 perovskite absorber was prepared by adapting
the vapor-assisted solution process described by Chen15 con-
sisting of spin-coating a layer of PbI2 and then annealing it
on a hot plate in a vapor of NH3CH3I until conversion to the
perovskite phase was complete. A PCBM layer was then pre-
pared by spin-coating from solution in chlorobenzene on top
of the perovskite layer. A thin Al contact was then deposited
by thermal evaporation on top of the PCBM layer. The Al
thickness (�20–25 nm) was tuned to allow 30%–40% optical
transmission enabling device characterization (Fig. 2). As
we will discuss below, the poorly transparent and highly
resistive (sheet resistance> 200 X/sq) Al layer is the main
performance limiting factor, requiring further optimization
of the transparent top contact. Device area of 0.45 cm2 was
defined by mechanical scribing. We also fabricated separate
stand-alone CZTSSe (glass\Mo\CZTSSe\CdS\ZnO\ITO)
with e-beam evaporated Ni-Al grid contact and superstrate
perovskite devices (Glass/ITO/PEDOT:PSS/NH3CH3PbI3/
PCBM/Al) by identical procedures, with the exception of a
thick (>50 nm) Al contact, for comparison as shown in
Table I. We did not observe any evidence of optical or elec-
trical deterioration of the ITO layer in the superstrate cell as
is most likely the case with the ITO layer in the tandem
device.
A current density-voltage (J-V) curve of a typical stand-
alone CZTSSe device (single-cell) is shown in Figure 3
(blue), and the parameters are tabulated in Table I. The Voc
values of the reference CZTSSe devices prepared identically
as the ones used for the tandem structures had range between
460–490 mV. Overlaid with the CZTSSe J-V curve is a simi-
lar measurement performed on a stand-alone perovskite ref-
erence device fabricated on ITO-coated glass substrate (red)
with parameters tabulated in Table I. Due to the larger band
gap of this material, the Jsc value is smaller and the Voc is
larger (typically 850–960 mV for our devices) than the val-
ues corresponding to the CZTSSe device. From the Tauc
plot shown in Fig. 2(b), we estimate the band gap of the per-
ovskite layer as �1.58 eV. The band gap of our bottom
FIG. 1. Cross-sectional SEM image of a monolythic tandem CZTSSe/
perovskite solar cell structure identifying individual layers.
FIG. 2. (a) Relationship between the average transmission (over the range of
400–1200 nm) and sheet resistance for a given aluminum film thickness. The
Al layer used in our device is expected to be similar to the middle data point.
(b) Tauc plot generated from the transmission curve of the stand-alone per-
ovskite layer assuming the relationship I=I0 ¼ e�ax and a perovskite thick-
ness of �200 nm.
TABLE I. Device parameters of various devices at simulated 1 sun illumina-
tion. RS is the series resistance and n is the ideality factor extracted using
Lambert W function fitting method.16
g FF Voc Jsc RS nDevice % % V mA/cm2 X.cm2
CZTSSe 11.6 68.2 0.477 35.7 0.63 1.66
Perovskite 12.3 76.6 0.953 16.8 2.53 2.76
CZTSSe-shadoweda 4.5 64.4 0.452 15.5 0.66 1.85
Tandem-projectedb 16.0 73.6 1.401 15.5 3.20 4.41
Tandem 4.6 60.4 1.353 5.6 15.70 6.40
a“CZTSSe-shadowed” is a CZTSSe cell shadowed by a perovskite absorber.bThe “tandem-projected” cell is a theoretical estimate based on the CZTSSe-
shadowed cell and the perovskite cell.
FIG. 3. J-V measurements of stand-alone devices (CZTSSe with Ni-Al grid
contacts and a superstrate perovskite on ITO-coated glass), and a tandem
CZTSSe/perovskite device.
173902-2 Todorov et al. Appl. Phys. Lett. 105, 173902 (2014)
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.76.102.52 On: Fri, 14 Nov 2014 10:07:54
CZTSSe layer is estimated to be �1.1 eV from quantum effi-
ciency measurement.
The J-V characteristics of the tandem CZTSSe/perovskite
device at 1 sun (red) is shown in Fig. 2 and tabulated in Table
I. We note that the Voc value of 1353 mV is nearly equal to
the sum of the individual Voc values of the CZTSSe and per-
ovskite devices, thus confirming that we have effectively
formed a series-connected, monolithic, two-terminal tandem
solar cell, and also confirming the integrated processing com-
patibility of these two materials systems (chalcogenide and
the perovskite) for tandem photovoltaic device fabrication.
The efficiency of our tandem device is still limited by
low current and fill factor with two main contributing ele-
ments. First is the significant optical transmission and reflec-
tion losses in the top Al layer17 (Fig. 2(a)) resulting in
extremely low Jsc. There is a tradeoff between optical trans-
parency and sheet resistance in thin metallic films such as
Al, where films with 30%–40% optical transmission
still have sheet resistances that are too high for optimal per-
formance (>200 X/sq, Fig. 2); this is in contrast to >90%
transmission for indium tin oxide (ITO) with sheet resistance
of <10 X/sq which can be achieved by sputtering. We have
used only Al as the top electrode in these feasibility studies
due to the risk of sputter damage associated with standard
ITO deposition. Future strategies for less aggressive top con-
tact fabrication could be evaporated ITO or solution-
processed alternatives such as silver nanowires.18
The tradeoff between transmittance and conductivity
dictates the additional limitation due to high series resistance
in the tandem cell (15.7 Xcm2) which is significantly higher
than the sum of the individual CZTSSe and perovskite cell.
While contribution of other device elements to the high se-
ries resistance cannot be ruled out at this point, the combina-
tion of drastically reduced illumination in all layers and high
sheet resistance of the Al contact are major contributors to
the low current and fill factor. Optimized top contact fabrica-
tion is currently under development and significant device
performance improvement is expected by enhancing the
transmission and conductivity of this element alone.
In order to assess the effect of the sub-optimal Al trans-
parent electrode and the related performance enhancement
potential, we estimate the upper limit of the efficiency of
our tandem cell based on the individual characteristics of
the stand-alone CZTSSe (g¼ 11.6%) and perovskite
(g¼ 12.3%) cells. We assume that we can have a high
quality transparent conductor on our perovsite top cell (e.g.,
ITO with 90% transmission and sheet resistance of �10 X/
sq) as in our standalone superstrate perovskite devices built
on ITO-coated glass.
To obtain a realistic estimate of the limit in current, we
measure the performance of the stand-alone CZTSSe cell
shadowed by the stand-alone perovskite, labeled as “CZTSSe-
shadowed” (see Table I). We observe a lower Jsc of
15.5 mA/cm2 (and g¼ 4.5%) as expected due to shadowing
(transmission, reflection, and absorption) losses of the top per-
ovskite device. A rough estimate of this shadowing test using
the unshadowed Jsc multiplied by the fraction of the photon
flux in the AM1.5 G spectrum in between the top and bottom
bandgap yields a consistent result of 14.6 mA/cm2. We then
combine the characteristics of the perovskite and the
“CZTSSe-shadowed” device in series to obtain the “tandem-
projected” device characteristics as shown in Table I and plot-
ted in Fig. 3 (dashed-red). In tandem device, we assume that
each individual device maintains the same diode characteris-
tics (ideality factor n, reverse saturation current J0, series
resistance RS, and shunt conductance GS) and the current is
limited by the cell with lower current. The “tandem-
projected” device shows a high efficiency of 16.0%, signifi-
cantly higher than both individual CZTSSe and perovskite
cell pointing to the high potential of this CZTSSe-perovskite
tandem device.
Optimization of the CZTSSe and perovskite bandgaps
for enhanced light harvesting and better current matching
provides another route for efficiency improvement. Figure
4(a) shows a projected calculation of a tandem CZTSSe and
perovskite devices with bandgap ranges of 1.0–1.5 eV and
1.5–2.2 eV, respectively. Each of the devices is assumed to
have a fixed efficiency (e.g., 12% for the CZTSSe and 12%
or 15% for the perovskite)—with device characteristics
(Voc, Jsc, FF, n, J0, RS, GS) based on parameter values that
our devices currently exhibit. We assume the Jsc and Voc of
each device scales proportionally to the bandgap and main-
tains the same efficiency, variable transmission to the bot-
tom cell due to shadowing loss of the top cell and current
matching condition for a series-connected tandem cell. The
transmission to the bottom cell is estimated based on the
fraction of the photon flux in the AM1.5 G spectrum with
energy in between the top and bottom cell bandgap, which
is consistent with the experimental shadowing test as shown
in Table I.
FIG. 4. The efficiency projection of
the CZTSSe/perovskite tandem solar
cells for 1 sun AM1.5 G spectrum with
varying bandgaps based on 12%
CZTSSe and (a) 12% perovskites (b)
15% perovskites devices.
173902-3 Todorov et al. Appl. Phys. Lett. 105, 173902 (2014)
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.76.102.52 On: Fri, 14 Nov 2014 10:07:54
For 12% CZTSSe and perovskites (Fig. 4(a)), we
observe that an optimum tandem efficiencies (�17%) are
obtained with bandgap combinations stretching almost line-
arly from Eg,BOTTOM and Eg, TOP¼ 1 and 1.65 eV to 1.5 and
2 eV, respectively. For our existing device configuration of
Eg¼ 1.1 eV CZTSSe and Eg¼ 1.58 eV perovskite, a large
efficiency gain can be obtained by improving the perovskite
bandgap to 1.70 eV while keeping the same CZTSSe
bandgap. Note that the current CZTSSe bandgap of 1.1 eV is
close to the empirical optimum bandgap in CZTSSe
(�1.13 eV).12 An example of similar calculation based on
12% CZTSSe and 15% perovskite is shown in Fig. 4(b),
which suggests optimum bottom and top bandgap combina-
tion stretching from 1 and 1.75 eV to 1.5 and 2.1 eV with a
maximum attainable efficiency of over 20%.
We have demonstrated a monolithic tandem photovoltaic
device architecture with solution processed kesterite and per-
ovskite absorbers. To date, this is the most efficient tandem
structure grown on a chalcogenide bottom device of any type,
despite the severely limiting Al top contact with low transmis-
sion and high resistivity. Optimization of the transparent
conductive electrode alone is expected to allow efficiency
enhancement of up to 16.0% with our current devices. Further
improvement potential with band gap engineering is projected.
We thank Professor Franky So and his group,
University of Florida, for helpful advice regarding
perovskite solar cell fabrication methods, Harry Hovel,
Ning Li, Ali Afzali, George Tulevski, and Jeehwan Kim
for useful discussion, Jay S. Chey, Ravin Mankad, and
Liang-yi Chang for their help with sample fabrication and
equipment operations.
1R. Klenk, J. Klaer, C. Koble, R. Mainz, S. Merdes, H. Rodriguez-Alvarez,
R. Scheer, and H. W. Schock, Sol. Energy Mater. Sol. Cells 95, 1441
(2011).2M. Symko-Davies and R. Noufi, Proceedings of 20th EuropeanPhotovoltaic Solar Energy Conference (WIP-Renewable Energies,
Munich, Germany, 2005), p. 2001.3P. Liska, K. R. Thampi, M. Gr€atzel, D. Bremaud, D. Rudmann, H. M.
Upadhyaya, and A. N. Tiwari, Appl. Phys. Lett. 88, 203103 (2006).4S. Nishiwaki, S. Siebentritt, and P. Walk, Prog. Photovoltaics 11, 243
(2003).5R. Kaigawa, K. Funahashi, R. Fujie, T. Wada, S. Merdes, R.
Caballero, and R. Klenk, Sol. Energy Mater. Sol. Cells 94, 1880
(2010); G. Cheek, F. Yang, and H. Lee, IEEE Photovoltaic SpecialistConference (IEEE, 2013).
6M. Reinhard, P. Sonntag, R. Eckstein, L. Burkert, A. Bauer, B. Dimmler,
U. Lemmer, and A. Colsmann, Appl. Phys. Lett. 103, 143904 (2013).7N. J. Jeon, J. H. Noh, Y. C. Kim, W. S. Yang, S. Ryu, and S. I. Seok, Nat.
Mater. 13, 897 (2014).8E. Edri, S. Kirmayer, M. Kulbak, G. Hodes, and D. Cahen, J. Phys. Chem.
Lett. 5, 429 (2014).9O. Gunawan, T. Gokmen, and D. B. Mitzi, J. Appl. Phys. 116, 084504
(2014).10M. Liu, M. B. Johnston, and H. J. Snaith, Nature 501, 395 (2013).11T. K. Todorov, K. B. Reuter, and D. B. Mitzi, Adv. Mater. 22, E156
(2010); T. Todorov, H. Sugimoto, O. Gunawan, T. Gokmen, and D. B.
Mitzi, IEEE J. Photovoltaics 4, 483 (2014).12W. Wang, M. T. Winkler, O. Gunawan, T. Gokmen, T. K. Todorov, Y.
Zhu, and D. B. Mitzi, Adv. Energy Mater. 4, 1301465 (2014).13T. Todorov and D. B. Mitzi, Eur. J. Inorg. Chem. 2010, 17; T. K. Todorov,
J. Tang, S. Bag, O. Gunawan, T. Gokmen, Y. Zhu, and D. B. Mitzi, Adv.
Ener. Mater. 3, 34 (2013).14J. You, Z. Hong, Y. Yang, Q. Chen, M. Cai, T. Song, C. Chen, S. Lu, Y.
Liu, H. Zhou, and Y. Yang, ACS Nano 8, 1674–1680 (2014).15Q. Chen, H. Zhou, T.-B. Song, S. Luo, Z. Hong, H.-S. Duan, L. Dou, Y.
Liu, and Y. Yang, Nano Lett. 14, 4158–4163 (2014).16C. Zhang, J. Zhang, Y. Hao, Z. Lin, and C. Zhu, J. Appl. Phys. 110,
064504 (2011).17H. Ehrenreich, H. R. Philipp, and B. Segall, Phys. Rev. 132, 1918
(1963).18L. Yang, T. Zhang, H. Zhou, S. Price, B. Wiley, and W. You, ACS Appl.
Mater. Interfaces 3, 4075–4084 (2011).
173902-4 Todorov et al. Appl. Phys. Lett. 105, 173902 (2014)
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.76.102.52 On: Fri, 14 Nov 2014 10:07:54