Photoelectron spectroscopy investigations of recombination contacts for tandem organic solar cells

Abstract

Recombination contacts play an important role in highly efficient organic tandem solar cells. We present a photoelectron spectroscopy study on contact systems that have previously been shown to work efficiently as recombination contacts. Here, the conversion of an electron current into a hole current is realized either by insertion of gold clusters or by a highly doped pn-junction. From the measured energy level alignments, we show that the working principles of these two approaches are significantly different. For gold clusters, the recombination current is promoted by an accumulation of charge carriers, while for doped pn-junctions, it is achieved by tunneling through a depletion layer.

Full text

Page 1

Photoelectron spectroscopy investigations of recombination contacts forPhotoelectron spectroscopy investigations of recombination contacts for
tandem organic solar cells tandem organic solar cells
Selina Olthof, Ronny Timmreck, Moritz Riede, and Karl Leo

Citation: Appl. Phys. Lett. 100100, 113302 (2012); doi: 10.1063/1.3693385
View online: http://dx.doi.org/10.1063/1.3693385
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v100/i11
Published by the American Institute of Physics.

Related ArticlesRelated Articles
Positron annihilation studies of the AlOx/SiO2/Si interface in solar cell structures
J. Appl. Phys. 111, 053515 (2012)
Overcoming the bandgap limitation on solar cell materials
Appl. Phys. Lett. 100, 083901 (2012)
Energy transfer in CaYAlO4: Ce3+, Pr3+ for sensitization of quantum-cutting with the Pr3+-Yb3+ couple
J. Appl. Phys. 111, 043104 (2012)
GaAs/GaInNAs quantum well and superlattice solar cell
Appl. Phys. Lett. 100, 073508 (2012)
Understanding the operation of quantum dot intermediate band solar cells
J. Appl. Phys. 111, 044502 (2012)

Additional information on Appl. Phys. Lett.Additional information on Appl. Phys. Lett.
Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors
Downloaded 14 Mar 2012 to 128.112.49.229. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

Page 2

Photoelectron spectroscopy investigations of recombination contacts
for tandem organic solar cells
Selina Olthof,a)Ronny Timmreck, Moritz Riede, and Karl Leo
Institut fu ¨r Angewandte Photophysik, Technische Universita ¨t Dresden, Dresden 01062, Germany
(Received 15 December 2011; accepted 20 February 2012; published online 14 March 2012)
Recombination contacts play an important role in highly efficient organic tandem solar cells. We
present a photoelectron spectroscopy study on contact systems that have previously been shown to
work efficiently as recombination contacts. Here, the conversion of an electron current into a hole
current is realized either by insertion of gold clusters or by a highly doped pn-junction. From the
measured energy level alignments, we show that the working principles of these two approaches
are significantly different. For gold clusters, the recombination current is promoted by an
accumulation of charge carriers, while for doped pn-junctions, it is achieved by tunneling through a
depletion layer. V
C 2012 American Institute of Physics. [http://dx.doi.org/10.1063/1.3693385]
Organic solar cells are promising candidates for flexible,
low cost, and large area photovoltaic applications. Even
though organic molecules have large absorption coefficients,
they often only absorb in a narrow energy band which limits
the solar cell performance. To circumvent this disadvantage,
it is possible to combine two or more cells consisting of dif-
ferent absorbers into a single device forming a tandem
cell.1,2If an ideal connection recombination contact between
the cells is realized, the open circuit voltage (Voc) of the tan-
dem cell equals the sum of the Vocof the single cells. Such a
recombination contact has to convert the electron current
from the first solar cell into a hole current for the second so-
lar cell and should meet two requirements: (1) under illumi-
nation, the splitting of the Fermi level at this interface has to
be avoided, so no reverse voltage is produced3,4and (2) an
efficient tunneling between electron and hole states at the
interface has to be provided for. A common way to realize a
recombination contact is the insertion of metal clusters or
thin metal layers at the interface.5,6In this case, the metal
islands quench the excitons, remove the Fermi level splitting,
and introduce a density of states (DOS) that reaches up to
the Fermi energy providing states for an electron-hole
recombination. Another concept is the use of highly doped
layers to form a pn junction between the two sub-cells.3,7
Such an interface is a suitable recombination contact as the
high concentration of free charge carriers moves the lowest
occupied molecular orbital (LUMO) of the acceptor and
highest unoccupied molecular orbital (HOMO) of the donor
close to the Fermi level and thereby reduces the energetic
barrier, while the thin depletion regions can easily be tun-
neled through.8
In this letter, we investigate the energy level alignment
of a commonly used donor acceptor interface consisting of
N,N,N0,N0-tetrakis(4-methoxyphenyl)-benzidine (MeO-TPD)
and C60by UV photoelectron spectroscopy (UPS). We previ-
ously studied and published prototype tandem devices
employing the same recombination contacts as under investi-
gation here.3A summary of the characteristic values of these
single and tandem cells is reproduced in Table I. This current
work is supposed to complement these device studies and
provide an understanding of the working principle and
design requirements of recombination contacts.
The organic layers are prepared by thermal evaporation
under ultra high vacuum. Details about the experimental
setup can be found elsewhere.9The materials MeO-TPD,
2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethan
TCNQ), C60, and acridine orange base (AOB) are deposited
from heated crucibles. Doping is achieved by co-evaporating
host and dopant, while the evaporation rates are controlled
independently by two quartz crystals. As substrate, a sputter
cleaned silver foil (purity 99.995%) is used, covered by
30nm of freshly evaporated gold.
First, the interface between the intrinsic layers of C60
and MeO-TPD is characterized in order to compare the
alignment to the cases when the metal clusters or dopants are
introduced. In Table I, this interface equals the one used in
tandem cell 1. It can be seen from the Vocthat such an inter-
face does not act as a recombination contact when imple-
mented into a device, but rather as a backward solar cell,
preventing an increase of Voccompared to the single cell
device. For the UPS measurements, a gold substrate is used
instead of the device relevant indium tin oxide (ITO), as ITO
often shows degradation under UV illumination.10The sub-
strate is covered by 10nm of p-doped MeO-TPD; the doping
(F4-
TABLE I. Characteristic values of a single solar cell (stack: ITO/p-MeO-
TPD/MeO-TPD/C60/BPhen/Al) and the tandem solar cells that exploit the
recombination contacts investigated in this paper. The structure of the
tandem cells is ITO/p-MeO-TPD/C60/x/MeO-TPD/C60/BPhen/Al, where x
stands for the different recombination contacts. The loss of Vocis the per-
centage missing for a full doubling of Voc. Values are taken from our previ-
ous publication of Timmreck et al.3
Interlayer xVoc(eV)Loss of Voc(%)FF (%)
Single cell
Tandem cell 1
Tandem cell 2
Tandem cell 3
—
none
0.53
0.57
0.96
1.04
—
46.2
9.4
1.9
50
39
53
48
1nm gold
pn junction
a)Author to whom correspondence should be addressed. Electronic mail:
solthof@princeton.edu.Presentaddress:
Engineering, Princeton University, New Jersey 08544, USA.
DepartmentofElectrical
0003-6951/2012/100(11)/113302/4/$30.00
V
C 2012 American Institute of Physics100, 113302-1
APPLIED PHYSICS LETTERS 100, 113302 (2012)
Downloaded 14 Mar 2012 to 128.112.49.229. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

Page 3

ensures an independence of the energetic alignment on the
substrate work function11and leads to an equal energetic
alignment of the following layers compared to the previously
reported tandem devices given in Table I. Additional intrin-
sic layers can be omitted during the investigation as they
have no or very little effect on the alignment.9On top of this
p-doped layer, 8nm of C60are evaporated and measured by
UPS shown as the bottom curve in Fig. 1(a). The ionization
potential (IP) is 6.46eV and the distance from the HOMO
onset to the Fermi energy is 1.45eV as can be seen in the
resulting energy level alignment in Fig. 2(a). Assuming an
electron affinity of 4eV (Ref. 12) and, therefore, a transport
gap of 2.46eV, the distance between the Fermi energy and
the LUMO is about 1eV. The measurement is carried on by
incrementally depositing the MeO-TPD interface in sub nm
steps on top while for every thickness an UPS spectrum is
recorded as shown in Fig. 1(a). Since the work function of
the C60layer is close to the IP of the MeO-TPD, a gradual
change of 300meV is observed in the vacuum level that
moves the MeO-TPD HOMO away from the Fermi energy.
However, beginning from 1nm thickness, the sample shows
charging, which expresses itself in a shift of several tenths of
meV of the whole spectra during repeated measurements.
This is expected, as the deeper lying HOMO of C60forms a
barrier for the holes generated by the removal of the photo-
electrons during measurement. The fact that this already hap-
pens at very low coverage gives first insight on the blocking
behavior of this contact and suggests that this is indeed not a
recombination contact. The offset between the LUMO of C60
and the HOMO of MeO-TPD at this interface is 1.36eV,
while the open circuit voltage measured for this interface in
a solar cell is Voc¼0.53eV (see Table I) and, therefore,
much smaller. This discrepancy has been observed often and
is due to contributions from the exciton binding energy13as
well as intermolecular charge-transfer states14and disorder
induced tail states.11,15,16
By repeating the measurement with a similar stack, but
this time inserting a thin gold layer at the interface between
C60and MeO-TPD, the alignment is significantly modified.
This interface used in tandem cell 2 in Table I was found to
be a well working recombination contact with a loss of Voc
of about 10% compared to a full doubling of the voltage. For
the device, a 1nm thick gold layer was used; however in the
UPS investigations, this led to a suppression of the C60fea-
tures, therefore the thickness was reduced to 0.5nm. Such an
interlayer leads to similar device performance (Voc¼0.91V
and FF¼53%). For the measurement of the energy level
alignment, C60is again evaporated on a gold substrate cov-
ered by p-doped MeO-TPD. The UPS measurement of this
layer, shown as the bottom curve of Fig. 1(b), gives within
FIG. 1. (Color online) UPS measurements of the interfaces between layers
of (a) intrinsic C60 and subsequently deposited MeO-TPD. (b) C60 and
MeO-TPD with 0.5nm of gold deposited in between and (c) C60doped by
AOB (20mol%) and MeO-TPD doped by F4-TCNQ (40mol %). The dotted
lines mark the change in high binding energy cutoff, C60peak position, and
MeO-TPD HOMO cutoff.
FIG. 2. (Color online) Schematic energy
level alignment deduced from the meas-
urements shown in Fig. 1: (a) alignment
between intrinsic MeO-TPD and C60, (b)
the same interface when 0.5nm of gold
are inserted, and (c) the highly doped pn
junction.
113302-2Olthof et al.Appl. Phys. Lett. 100, 113302 (2012)
Downloaded 14 Mar 2012 to 128.112.49.229. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

Page 4

the experimental error the same alignment as in the previous
sample. After the evaporation of 0.5nm gold, the HOMO of
C60exhibits a significant shift of 230meV away from the
Fermi energy, and the Fermi edge of the gold becomes visi-
ble. The shift is either due to the formation of an interface
dipole upon contact with the gold or the penetrating gold
clusters lead to an accumulation of electrons at the interface.
The change in the HOMO position is accompanied by a
downward shift of the vacuum level of 310meV. As before,
the MeO-TPD layer is incrementally built on top. Appa-
rently, this layer is affected by the gold interlayer as well
and shows a shifting of the HOMO position by 320meV
with increasing layer thickness, indicating a transfer of elec-
trons from the MeO-TPD to the gold at the interface. During
the deposition, the C60shows no further change in HOMO
position, so the charge transfer is solely between the MeO-
TPD and Au. The measurement can only be continued until
5nm MeO-TPD coverage, when again a charging of the sam-
ple is observed. We can conclude that the charge carrier con-
version at the interface is much better than in case of the
intrinsic interface, but still the process is not perfect, leading
to an accumulation of photogenerated holes. The resulting
energetic alignment is shown in Fig. 2(b). The total built-in
potential of this interface, taken from the change in vacuum
level position, is VB¼0.75eV. The offset between C60,LUMO
and MeO-TPDHOMOis reduced by about 270meV compared
to the intrinsic interface and in addition, there is a DOS
reaching up to the Fermi energy due to the gold.
For the interface employing a pn junction, we use F4-
TCNQ as the p-dopant in MeO-TPD at a high doping con-
centration of 40mol % and AOB as the n-dopant in C60at
20mol %, which was found to produce a very efficient
recombination contact that nearly doubles the Voc(see tan-
dem cell 3 in Table I). For this sample, the n-doped C60layer
is directly put on the gold foil. The p-doped MeO-TPD can
be omitted as the high doping concentration in C60leads to
Fermi level alignment of the film independent of the layers
underneath.11The UPS measurements are shown in Fig.
1(c). As expected, the energetic position of the n-doped C60
differs from those in Figs. 1(a) and 1(b) as this time, the n-
doping leads to an alignment, where the LUMO is only
170meV away from the Fermi energy. Upon the deposition
of doped MeO-TPD, the C60level shows a strong upward
bending, typical for the formation of a depletion region due
to charge transfer across the pn-junction. The HOMO of C60
can only be observed up to 0.5nm MeO-TPD deposition. By
then, it shows a level shifting of 920meV. The HOMO of
MeO-TPD shows no shift at all and remains 0.35eV below
EF. Apparently, the 40mol % doping by F4-TCNQ in MeO-
TPD produces considerably more free charge carriers than
the 20mol % doping of C60by AOB; therefore, the entire
voltage drop takes place in n-C60. The change in vacuum
level position is VB¼0.91eV. The spectra do not show
charging during the measurement done up to 10nm MeO-
TPD thickness. However, from these measurements, we can
only determine the amount of shifting and not the width of
the depletion region in n-C60; therefore, the measurement
was repeated with reversed deposition sequence (data not
shown). Just as before, the deposition of n-C60onto p-MeO-
TPD does not lead to a change in the MeO-TPD HOMO
position. All the level bending takes place in the C60which
is found to take 10nm to saturate. As can be seen in Fig.
2(c), the offset between C60,LUMOand MeO-TPDHOMOis
decreased to merely 0.52eV. Furthermore, it has been
reported that upon doping, the tailing states of the transport
levels are pinned at the Fermi energy,11,15so we can assume
the tail of the DOS distribution reaches up to the EFon both
sides of the 10nm thick depletion layer.
Comparing the alignment of the two recombination con-
tacts in Figs. 2(b) and 2(c) reveals that the effects causing an
efficient recombination contact are rather different. In both
cases, the distance between C60,LUMOand MeO-TPDHOMOis
reduced, and we observe states at or close to EFthat prevent
any Fermi level splitting under illumination which would
create an unwanted backward Voc. However, the sign of the
built in voltage at the interface shows in opposite directions.
In the case of the metal clusters, the 750meV field is created
by a dipole formation and accumulation of charge carriers at
the interface; in the case of the highly doped pn junction, the
built in voltage of 910meV is created by the formation of a
10nm wide depletion layer that the charge carriers tunnel
through. Here, the advantage of using pn junctions over the
metal recombination layers becomes obvious. Any suffi-
ciently high doped material combination will be able to form
a recombination contact by putting the transport levels in
close proximity to the Fermi energy. However, in the case of
the metals, it depends sensitively on the individual combina-
tion of organic material and metal whether the levels are
shifted in the right direction and the built in field is of suffi-
cient strength. This might explain why only few material
combinations have so far been found that produce a doubling
of the Vocwhen using the metal cluster approach,6,17while
most combinations suffer in some loss with regard to 2 Voc.
In summary, we have presented photoelectron spectros-
copy measurements of two different recombination contacts
used in organic tandem solar cells employing either gold
clusters or a highly doped pn junction. Comparing them to
an interface without modification, we showed that both
approaches reduce the tunneling distance for the recombina-
tion current between donorHOMOand acceptorLUMO. Further-
more, a density of states is appearing close to the Fermi
energy promoting the tunneling current and preventing a
Fermi level splitting under illumination. In both cases, a
large built in field is created at the interface, but the direction
of the field is opposed leading to a charge accumulation in
case of the metal clusters and to the formation of a depletion
region for the pn junction.
This work was supported by the BMBF within the
projects OPEG (Grant No. 13N9720) and R2flex (Grant
No.13N8855).
1J. Y. Kim, K. Lee, N. E. Coates, D. Moses, T.-Q. Nguyen, M. Dante, and
A. J. Heeger, Science 317, 222 (2007).
2T. Ameri, G. Dennler, C. Lungenschmied, and C. J. Brabec, Energy Envi-
ron. Sci. 2, 347 (2009).
3R. Timmreck, S. Olthof, M. K. Riede, and K. Leo, J. Appl. Phys. 108,
033108 (2010).
4P. Wuerfel, Physics of Solar Cells, From Principles to New Concepts
(Wiley-VCH, Weinheim, 2005).
5M. Hiramoto, M. Suezaki, and M. Yokoyama, Chem. Lett. 19, 327 (1990).
113302-3 Olthof et al. Appl. Phys. Lett. 100, 113302 (2012)
Downloaded 14 Mar 2012 to 128.112.49.229. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

Page 5

6A. Yakimov and S. R. Forrest, Appl. Phys. Lett. 80, 1667 (2002).
7J. Drechsel, B. Maennig, F. Kozlowski, M. Pfeiffer, K. Leo, and H. Hoppe,
Appl. Phys. Lett. 86, 244102 (2005).
8K. Harada, M. Riede, K. Leo, O. Hild, and C. M. Elliott, Phys. Rev. B 77,
195212 (2008).
9S. Olthof, R. Meerheim, M. Schober, and K. Leo, Phys. Rev. B 79, 245308
(2009).
10M. M. Beerbom, B. Laegel, A. J. Cascio, B. V. Doran, and R. Schlaf, J.
Electron Spectrosc. Relat. Phenom. 152, 12 (2006).
11S. Olthof, W. Tress, R. Meerheim, B. Luessem, and K. Leo, J. Appl. Phys.
106, 103711 (2009).
12W. Zhao and A. Kahn, J. Appl. Phys. 105, 123711 (2009).
13P. I. Djurovich, E. I. Mayo, S. R. Forrest, and M. E. Thompson, Org. Elec-
tron. 10, 515 (2009).
14D. Veldman, S. Meskers, and R. Janssen, Adv. Funct. Mater. 19, 1939
(2009).
15M. Kroeger, S. Hamwi, J. Meyer, T. Riedl, W. Kowalsky, and A. Kahn,
Org. Electron. 10, 932 (2009).
16W. L. Kalb, S. Haas, C. Krellner, T. Mathis, and B. Batlogg, Phys. Rev. B
81, 155315 (2010).
17G. Dennler, H. J. Prall, R. Koeppe, M. Egginger, R. Autengruber, and N.
S. Sariciftci, Appl. Phys. Lett. 89, 073502 (2006).
113302-4 Olthof et al.Appl. Phys. Lett. 100, 113302 (2012)
Downloaded 14 Mar 2012 to 128.112.49.229. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions