FULL PAPER

Origin of the Open Circuit Voltage of Plastic Solar Cells

By Christoph J. Brabec,* Antonio Cravino, Dieter Meissner, N. Serdar Sariciftci, Thomas Fromherz,
Minze T. Rispens, Luis Sanchez, and Jan C. Hummelen

A series of highly soluble fullerene derivatives with varying acceptor strengths (i.e., first reduction potentials) was synthesized
and used as electron acceptors in plastic solar cells. These fullerene derivatives, methanofullerene [6,6]-phenyl C61-butyric acid
methyl ester (PCBM), a new azafulleroid, and a ketolactam quasifullerene, show a variation of almost 200 mV in their first
reduction potential. The open circuit voltage of the corresponding devices was found to correlate directly with the acceptor
strength of the fullerenes, whereas it was rather insensitive to variations of the work function of the negative electrode. These
observations are discussed within the concept of Fermi level pinning between fullerenes and metals via surface charges.

1. Introduction can be improved up to 2.5 % under AM1.5 irradiation, by

manipulating the solid state morphology (nanostructure) of the
Photovoltaic elements based on thin polymer films of solid components, as well as that of the blend. The results clearly
state composites of conjugated polymer/fullerene compounds show that the several-fold enhancement in the short circuit cur-
seem to be promising candidates for solar energy conver- rent of photovoltaic devices most likely originates from the
sion.[1±3] The mechanical flexibility of plastic materials is wel- improved mobility of the charge carriers and from enhanced
come for all photovoltaic applications on curved surfaces. By interfacial contacts between the donor and the acceptor.[4,5]
casting thin, semitransparent polymer photovoltaic films be- The excellent photosensitivity and relatively high energy
tween the glass panes of insulated windows, large, unused areas conversion efficiencies obtained from these interpenetrating
can be accessed, in addition to the limited roof areas of crowded network bulk-heterojunction materials are promising. One
cities. Even the bandgap of the polymer, i.e., the color of such route to enhance the device performance is by understanding
photovoltaic elements, can be varied due to the flexibility of or- the device physics and then optimizing the device parameters.
ganic synthesis, and the chemical tailoring of desired properties. An essential parameter of thin film photovoltaic devices, which
An important step towards efficient organic solar cells was influences charge dissociation, charge transport, and charge
the development of the bulk-heterojunction concept.[1] In con- collection, is the built-in potential. A generally accepted esti-
trast to bilayer devices, which are comprised of donor and mate for the built-in potential is given by the open circuit volt-
acceptor layers spatially separated, in analogy to the classical age, Voc, which underestimates the built-in potential at room
p±n junction, the bulk-heterojunction overcomes the limitation temperature and converges to the correct value at low tem-
of the charge generation at a two-dimensional interface by dis- peratures. Therefore, the question of the built-in potential is
tributing the acceptor more or less homogeneously into the directly related to an extensively discussed phenomenon, the
donor matrix, thereby generating a three-dimensional network origin of Voc. Mixing fullerenes with conjugated polymers into
of photoinduced charge generating interfaces. Recently, we a composite active layer completely modifies the nature of the
have demonstrated that the power conversion efficiency of thin film devices, compared with those made from conjugated
bulk-heterojunction plastic solar cells (PSCs) produced from a polymers alone,[6,7] and hence the Voc of the corresponding
conjugated polymer, poly(2-methoxy-5-(3¢,7¢-dimethyloctyl- solar cells.[1] Therefore, it is not surprising that models success-
oxy)-1,4-phenylenevinylene) (MDMO-PPV), and a methano- fully describing the situation in pristine conjugated polymer
fullerene [6,6]-phenyl C61-butyric acid methyl ester (PCBM) photodiodes, such as the metal-intrinsic-metal (MIM) or the
Schottky junction[9,10] picture, cannot satisfactorily explain the
± observed Voc in conjugated polymer/fullerene based (bulk-
[*] Dr. C. J. Brabec,[+] A. Cravino, Prof. D. Meissner, Prof. N. S. Sariciftci heterojunction) solar cells.
Linz Institute for Organic Solar Cells (LIOS), Physical Chemistry In order to investigate the critical parameters influencing the
Johannes Kepler University of Linz
Altenbergerstr. 69, A-4040 Linz (Austria) built-in potential in conjugated polymer/fullerene bulk-hetero-
E-mail: christoph.brabec@e&LS.Siemens.de junction solar cells systematically, a series of highly soluble ful-
Dr. T. Fromherz lerene derivatives with varying acceptor strengths (i.e., first
Institut für Halbleiter- und Festkörperphysik
Johannes Kepler University of Linz reduction potentials) was synthesized, and the open circuit volt-
Altenbergerstr. 69, A-4040 Linz (Austria) age of the corresponding devices as a function of acceptor
Dr. M. T. Rispens, Dr. L. Sanchez, Dr. J. C. Hummelen strength was analyzed. These fullerene derivatives, methanoful-
Stratingh Institute and Materials Science Centre, University of Groningen
Nijenborgh 4, NL-9747 AG Groningen (The Netherlands)
lerene PCBM,[11] a new azafulleroid and a ketolactam quasiful-
[+] Present address: Siemens AG, CT MM1, Innovative Electronics, Paul Gos- lerene (Fig. 1), show a variation of almost 200 mV in their first
senstr. 100, D-91052 Erlangen, Germany. reduction potentials. Additionally, cells made with [60]fullerene

374 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2001 1616-301X/01/0510-0374 $ 17.50+.50/0 Adv. Funct. Mater. 2001, 11, No. 5, October
C. J. Brabec et al./Open Circuit Voltage of Plastic Solar Cells

FULL PAPER
Scheme 1. Synthesis of N-3-(2-ethylhexyloxy)benzyl azafulleriod 5 and of N-3-
(2-ethylhexyloxy)benzyl ketolactam 6. EHO = 2-ethyl hexyloxy. a) NaOMe,
2-ethyl-1-bromohexane, MeOH, D, 16 h, 29 %. b) PBr3, Et2O, 3 h, 59 %. c) NaN3,
DMSO, 1.5 h, 95 %. d) C60, ODCB, 180 C, 2 h, 18 %. e) O2, halogen lamp, Kap-
ton filter, ODCB, 3 h, room temperature, 18 % (based on [60]fullerene).

synthesized from commercially available 3-hydroxymethylphe-

nol (1) by alkylation with 2-ethylhexylbromide,[13] conversion
of the benzyl alcohol 2 to a benzyl bromide 3,[14] followed by
reaction with sodium azide in dimethyl sulfoxide (DMSO),[15]
all analogously to known procedures. Subsequently, [5,6]-aza-
fulleroid 5 was prepared by the addition of azide 4 to a solution
of C60 in boiling o-dichlorobenzene (ODCB) and subsequent
reaction for 2 h at 180 C, yielding 5 in an optimized yield of
18 %, after column chromatography.
Fig. 1. Chemical structures of the investigated compounds, and device layout of Ketolactam 6 was prepared by self-sensitized photo-oxyge-
the solar cells. nation of azafulleroid 5 (Scheme 1). Since the separation of 5
from C60 was difficult, it was more practical to use a partially
(C60) were compared. It is important to emphasize that, apart purified mixture (consisting of 5 and C60) in the photo-oxyge-
from C60, these acceptors have comparably sized solubilizing nation reaction.[16] The cage-opening reaction, which most like-
groups. Effects due to different donor±acceptor distances, and/ ly proceeds via a dioxetane-intermediate, was complete in 3 h,
or different morphologies, should, in this way, be minimized, as as indicated by high performance liquid chromatography
is required for a comparative study. Nevertheless, we realize (HPLC). Column chromatography gave 6 in 18 % yield, indi-
that it is highly unlikely that the morphologies of the various cating the conversion of 5 to 6 to be quantitative (see above).
active layers are identical. (This is unavoidable, however, since Azafulleroid 5 showed 31 signals for the fullerene-sp2 car-
it is obviously impossible to alter the electron affinity of an bons in the 13C NMR spectrum, indicative of CS-symmetry.
acceptor component without altering its structure!) UV-vis spectroscopy featured the characteristic azafulleroid
MDMO-PPV (Fig. 1) was used as the polymeric donor, in all absorptions [341 nm (e = 426 000 M±1 cm±1), 437 nm (sh),
experiments. This polymer has been used routinely in plastic 551 nm (e = 13 900 M±1 cm±1)] are present.[17] In the Fourier
photovoltaic (PV) research over the last few years.[4,5,12] In the transform infrared (FTIR) spectrum, a strong peak at 525 cm±1,
second part of this research, we investigated the possibility of typical for a monosubstituted [60]fullerene, was observed.
influencing the built-in potential of the photodiodes by varying Matrix-assisted laser desorption ionization time-of-flight mass
the work function of the top electrode (i.e., the negative elec- spectrometry (MALDI-TOF-MS; S8) gave a M± at 953.7, in
trode, collecting electrons from the active layer). Four different accordance with a calculated value of m/z = 953.2.
metal electrodes were selected for this purpose: calcium, silver, Ketolactam 6 showed 57 signals in its 13C NMR spectrum
aluminum, and gold, thus varying the work function by more for the fullerene-sp2 carbons, showing a lack of symmetry
than 2 eV, while keeping the transparent positive front elec- compared to 5. The ketone carbon resonance was found at
trode material constant (i.e., a poly(3,4-ethylenedioxythio- 194.8 ppm, whereas for the lactam carbon the most likely
phene):poly(styrene sulfonate) (PEDOT:PSS) layer on an resonance is either at 160.7 or 159.6 ppm. The 1H NMR
indium tin oxide (ITO) coated support material). spectrum showed a double doublet for the diasterotopic
benzylic hydrogens at d 6.38 and 5.39 ppm (J = 15 Hz), as a
result of the chirality of the cage-opened fullerene. The
2. Results and Discussion UV-vis spectrum showed the characteristic absorption pat-
tern for a ketolactam fullerene moiety [333 (e = 488 000), 433
2.1. Synthesis of Two New Fullerene Acceptor Materials (44 400), 619 (4 560) and 698 nm (4 410 M±1 cm±1)], whereas
the FTIR spectrum showed absorption features at 1727 and
The synthesis of the target molecules [5,6]-N-3-(2-ethylhex- 1689 cm±1, indicative of the ketone and lactam moieties,
yloxy)benzyl azafulleroid (5) and N-3-(2-ethylhexyloxy)benzyl respectively. Moreover, a peak at 522 cm±1, assigned to the
ketolactam (6), started with the preparation of 3-(2-ethylhex- fullerene core, was observed. Both 5 and 6 gave satisfactory
yloxy)benzyl azide (4), as depicted in Scheme 1. Azide 4 was elemental analyses.

Adv. Funct. Mater. 2001, 11, No. 5, October Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2001 1616-301X/01/0510-0375 $ 17.50+.50/0 375
 C. J. Brabec et al./Open Circuit Voltage of Plastic Solar Cells
FULL PAPER

2.2. Cyclic Voltammetry pound is solvent dependent, it is the relative differences

between the acceptor strengths of the fullerene derivatives that
The redox behavior of both new fullerene derivatives was are important, and not so much their absolute values. In the
determined by cyclic voltammetry (CV), together with that of discussion below, we will assume that the trend found working
the parent C60 and PCBM, all were measured under identical with solutions is representative also for the fullerene deriva-
conditions. The voltammograms are shown in Figure 2, and the tives in the solid state.
data are tabulated in Table 1, for numerical comparison. All The tuning of the reduction potential of fullerene-based ma-
four CVs showed four reversible reduction waves correspond- terials has also been accomplished by other research groups.
Recently, qualitatively comparable results with respect to
increasing the electron affinity of fullerenes were obtained with
isoxazolo-fullerenes,[19] fulleropyrrolidium salts,[20] and certain
metal containing fullerene-crown ether conjugates.[21] The pos-
sible application of fullerene derivatives in an organic ªplasticº
solar cell, however, demands processability and a total reversi-
bility of the one-electron reduction. These criteria are met by
both azafulleroid 5 and ketolactam 6. Irradiation of 5 in the
presence of oxygen yields 6. This ketolactam is inert to further
photo-oxidation, which in itself is an interesting property for a
component in a ªplasticº solar cell.

2.3. Photovoltaic Devices

2.3.1. Variation of the Acceptor Materials

Fig. 2. Cyclic voltammograms of PCBM, azafulleroid 5 and ketolactam 6. The Thin film photovoltaic devices with an active layer thickness
experimental conditions were the following: reference electrode was a Ag wire;
GCE was the working electrode; Bu4NPF6 (0.1 M) was the supporting electro- of ~100 nm were produced by doctor blading the active layer
lyte; 100 mV/s was the scan rate; ODCB/MeCN (4:1) was the solvent. donor±acceptor composite from toluene solutions on top of a
PEDOT:PSS (~100 nm)/ITO (80 nm)/polyester (175 lm) sub-
Table 1. Redox potentials (V vs NHE) of C60 and fullerene derivatives [a]. strate. Subsequently, a top aluminum layer was deposited by
evaporation. Photovoltaic parameters were determined under
inert gas conditions under illumination with 60 mW/cm2 white
light from a halogen lamp. More than 80 devices were pro-
duced from each acceptor type to allow a statistical evaluation
of the observed open circuit voltage. A box plot diagram was
chosen to present the results from current/voltage (I/V) mea-
surements for the Voc (Fig. 3). The horizontal lines in the box
[a] Experimental conditions: Quasi Ag/AgCl wire (calibrated with ferrocene) as
reference electrode, Pt foils as working and counter electrodes; Bu4NPF6 (0.1 M)
denote the 25th, 50th, and 75th percentile values. The error
as supporting electrolyte; ODCB/MeCN (4:1) as solvent. Scan rate 100 mV/s; bars denote the 5th and 95th percentile values. The two sym-
room temperature. bols below and above the 5th/95th percentile error bars denote

ing to the reduction of the fullerene cage. However, the first

reduction wavesÐindicative of the electron acceptor strength
of the compoundsÐshow distinctive differences. Ketolactam 6
(±0.53 V) appeared to be a substantially better electron accep-
tor than C60 (±0.60 V),[18] whereas azafulleroid 5 (±0.67 V) is
close to C60, and PCBM (±0.69 V) showed a clearly diminished
electron affinity. In the case of azafulleroid 5, an additional
shoulder to the wave at ±2.02 V was observed. We assigned this
feature to the reduction of the addend, after comparison to
N,N-dimethyl (3-methoxy)benzyl amine (±1.88 V). The CV of
ketolactam 6 showed an additional wave (±1.93 V), which most
likely corresponds to the reduction of the ketone moiety. This
value compares nicely to that of benzophenone (±1.85 V),
observed under the same conditions. Hence, a difference of
Fig. 3. The Voc for the solar cells utilizing PCBM, azafulleroid 5, and ketolactam
160 mV is observed between the strongest and the weakest 6 as the acceptor components, in bulk-heterojunction solar cells comprising
acceptor. Since, in general, the reduction potential of a com- MDMO-PPV as the electron donor.

376 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2001 1616-301X/01/0510-0376 $ 17.50+.50/0 Adv. Funct. Mater. 2001, 11, No. 5, October
C. J. Brabec et al./Open Circuit Voltage of Plastic Solar Cells

FULL PAPER
the highest and the lowest observed values, respectively. For all devices utilizing Ca (uMe = 2.87 eV), Al (uMe = 4.28 eV), Ag
three acceptors presented, a relatively narrow distribution of (uMe = 4.26 eV), and Au (uMe = 5.1 eV)[22] as the negative
the open circuit voltage was observed, indicating excellent electrodes: the scale is logarithmic. A total variation of less
reproducibility: 75 % of the devices from each acceptor are dis- than 200 mV of the Voc was observed for a variation of the neg-
tributed less than 40 mV around their average value. The high- ative electrode work function by more than 2.2 eV. For the de-
est and the lowest average open circuit voltages were observed vices with a Au electrode, Voc was found to be slightly lower
for the PCBM containing cells and for the ketolactam contain- than the average value, but still as high as 650 mV. The Ca de-
ing cells, with 760 and 560 mV, respectively. vices exhibit a Voc of 814 mV. It is interesting to note that the
direction of flow of the short circuit current (i.e., the polarity of
2.3.2. Variation of Top Electrode Materials the device) was not reversed in the case of the Au electrode, as
would normally be expected for a MIM device due to the nom-
PCBM was chosen as the reference electron acceptor for inally slightly higher work function of Au compared to the
investigating the influence of the top (negative) electrodes with ITO/PEDOT:PSS electrode on the other side. In this device,
different work functions on the built-in potential of conjugated holes still flow towards the ITO/PEDOT:PSS electrode (posi-
polymer/fullerene bulk-heterojunction plastic solar cells. These tive electrode), while electrons are still collected at the (nega-
devices were produced by spin-casting the photoactive donor/ tive) Au electrode. The observation that the short circuit cur-
acceptor mixture on top of PEDOT:PSS (80 nm)/ITO rent Isc for the Au devices is clearly lower than for comparable
(100 nm)/glass substrates from toluene solutions. The top elec- devices with other electrodes will be discussed later.
trode materials were evaporated thermally at a vacuum lower
than 10±6 mbar. Figure 4 shows the I/V curves of four typical
2.4. Fermi Level Pinning

The experimental results on the variation of the acceptor

strength and on the variation of the top electrode work func-
tion are summarized in Figures 5a and 5b. Figure 5a shows the
highest observed Voc (from Fig. 3) versus the acceptor strength
for devices utilizing PCBM, C60, azafulleroid 5 and ketolactam
6 as electron acceptors. The data points were fitted to a linear
model, and a slope of S1 ~ 1 was derived from the fitting proce-

Fig. 4. a) I/V curves for MDMO-PPV/PCBM photovoltaic devices with different

metal electrodes. Al electrode (solid line), Ca electrode (dashed line), Au elec- Fig. 5. a) Voc versus acceptor strength and b) Voc versus negative electrode work
trode (dotted line), Ag electrode (dash-dotted line). b) I/V curves with a linear function. The slopes S1 and S2 of the linear fits to the data are given inside the
scale. figures.

Adv. Funct. Mater. 2001, 11, No. 5, October Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2001 1616-301X/01/0510-0377 $ 17.50+.50/0 377
 C. J. Brabec et al./Open Circuit Voltage of Plastic Solar Cells
FULL PAPER

dure. The fact that a slope of unity was obtained emphasizes ing at the semiconductor±metal interface. For n-type semicon-
the strong coupling of the Voc to the reduction strength of the ductors, these states are acceptor-like, and the semiconductor
acceptors, as expected for the ideal case of ohmic contacts at equilibrium may exhibit an upward (negative) band bending
between the different fullerenes and the aluminum contact. as the surface Fermi level moves toward the charged states. For
Deviations from this ªideal behaviorº might occur, since the a p-type semiconductor with donor-like surface states, the
reduction strengths of the acceptors plotted in Figure 5a on the semiconductor at equilibrium would exhibit downward band
x-axis were taken from electrochemical data in solution and bending as the surface Fermi level moves toward the charged
not from thin solid films. states. For the photodiodes presented in this study, we propose
The influence of the work function of the negative electrode that the Fermi level of the negative electrode metal is pinned
on the Voc of the MDMO-PPV/PCBM solar cells is shown in to the reduction potential of the fullerene. First, the correlation
Figure 5b. It is important to note that the x-axis now covers of the open circuit voltage with the reduction potential of the
more than 2 eV. Again, a linear model was fitted to the experi- acceptor (Fig. 5a) proves that the energy alignment of the met-
mental data and a slope S2 ~ 0.1 was calculated as the best fit. al is related to the lowest unoccupied molecular orbital
This result shows that the work function of the metal has a con- (LUMO) energy states of the acceptor and not of the polymer.
siderably weaker effect on the Voc values than does the reduc- Second, recent results from XPS[24,25] (X-ray-photoemission)
tion potential of the fullerene derivative. on mono- and multi-layers of C60 on different metals with var-
In order to understand this behavior, we use a concept that ious work functions showed clearly that there is considerable
has been developed to explain the built-in potential for inor- charge transfer to C60 adsorbed on metal surfaces. The align-
ganic semiconductor/metal interfaces, and which is based on ment of the ground state energies was determined by the inter-
the index of interface behavior S.[23] The parameter S is defined face dipole, i.e., the bond formation in the first layer, which en-
as the slope of a plot of the blockade potential of a semicon- sures the charged state's induced Fermi level alignment to the
ductor/metal interface versus the work function of the metal: substrate. Surface states in organic semiconductors are unlikely
due to the weak bonding forces between the molecular units.
qVB = S(uM ± uSC) + C (1) However, fullerenes, with their cage-like p-electron system and
their high tendency to crystallize, may be different, exhibiting
where qVB is the blockade potential of the interface barrier, strong charge transfer (up to 1.8 electrons per fullerene) at ful-
uM and uSC are the work functions of the metal and the semi- lerene/metal interfaces.[26] It is beyond the aim of this paper to
conductor, respectively, and C is a constant that describes the speculate on the nature of the charged states for thin spin-cast
interface potential for the ideal ohmic contact. From the films of fullerenes: this has been discussed for ultra high vacu-
results presented above (Figs. 5a and 5b), we propose the um (UHV) grown films.[27±29] The presence of large interface
following equation for the open circuit voltage in the PSC: dipoles between metals and organic semiconductors has also
been shown for a large number of other small molecules.[30]
VOC = (AOX ± S1Ered(A)) ± S2(uM ± Ered(A)) + C (2) Assuming that the mechanism of Fermi level pinning is
dominant for the contact formation at the negative electrode
where, Ered(A) is the reduction potential of the acceptor (fuller- of the photodiodes, the qualitative difference between the Au
ene derivative) and S1 and S2 are the slopes calculated from electrode and the Ca, Ag, and Al electrodes can be discussed.
Figures 5a and 5b. Aox is a constant representing all the contri- Whereas the low work function of Ca, Ag, and Al (uME
butions from the positive electrode to the Voc, and is expected < 4.3 eV) will favor ohmic contacts to fullerenes (EFermi (C60)
to be properly described by the oxidation potential of the con- ~ 4.7 eV), Au is the only one to form ohmic contacts to a hole-
jugated polymer. The negative sign in front of the second term transporting conjugated polymer such as MDMO-PPV, due to
of Equation 2 takes into account that this contribution works its high work function (uAu ~ 5.1 eV). The other three metals
as a counter diode. Since the parameter S2 was found to be suf- are known to form rectifying (blocking) contacts to the holes in
ficiently small (S2 < 0.1), it can be also neglected for the open a conjugated polymer.[31,32] For the devices presented in this
circuit voltage, leading to Equation 3, which yields a good esti- study, it was shown above that the pinning of the metal work
mation of our results: function to the fullerene reduction potential promotes a quasi-
ohmic contact between the metal and the fullerene, even for a
VOC = (AOX ± S1Ered(A)) + C (3) Au electrode (preceded by charge transfer between the metal
and the first fullerene monolayer, yielding an interfacial dipole
Comparing Equations 1 and 2, the slope S2 of the linear fit layer). For the proper function of a photodiode, asymmetric
calculated from Figure 5b resembles the index of interface contact conditions are essential. At the negative electrode, an
behavior S. Following this concept, the observation of a slope ohmic contact to the electron transporting phase of the donor±
as low as 0.1 in Figure 5b suggests a mechanism called ªFermi acceptor composite is favorable, while the holes should be
level pinningº.[23] In such a case, the work function of the metal blocked (rectifying contact). This condition seems to be ful-
is pinned to the work function of the semiconductor (typically filled for the photodiodes with a Ca, Ag, or Al negative elec-
via surface states), independent of whether the work function trode, due the low value of the work function compared to the
of the metal is higher or lower than the Fermi level of the semi- Fermi level of the fullerenes. In the case of the negative Au
conductor. The presence of surface charges leads to band bend- electrode, ohmic contacts may be formed to both the fullerene

378 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2001 1616-301X/01/0510-0378 $ 17.50+.50/0 Adv. Funct. Mater. 2001, 11, No. 5, October
C. J. Brabec et al./Open Circuit Voltage of Plastic Solar Cells

FULL PAPER
phase and the conjugated polymer phase. Such contacts will VXR-300 (300 MHz) instrument at 298 K, as indicated. Spectra recorded in CS2
employed a D2O insert as the external lock and 1H reference (d = 4.67 ppm rel-
enhance surface recombination, thereby reducing the short cir- ative to TMS) for the 1H NMR, and CS2 was used as the internal 13C reference (d
cuit current and the open circuit voltage. An interesting conse- = 192.3 ppm relative to TMS) in the 13C NMR. Coupling constants (J) are given
quence of the fact that the Au contact to the conjugated poly- in Hz. Multiplicities are denoted as follows: s = singlet, d = doublet, t = triplet,
sep = septet, dd = double doublet, m = multiplet, br = broad. FTIR spectra were
mer phase is a rectifying one for the Au-PSC is the occurrence recorded on a Mattson Galaxy 4020 instrument. UV-vis spectra were recorded
of a dipole layer at this interface with a reversed sign compared on a Hewlett Packard HP 8452 UV-vis spectrophotometer. MALDI-TOF-MS
measurements were performed on a Micromass TofSpec E apparatus (negative-
to Al, Ca, and Ag electrodes. This results in positively charged
ion reflectron mode) by using elemental sulfur as the matrix [34]. Measurements
states on the fullerene for the Au contact and in negatively were performed on 1:1 (v/v) aliquots of the analyte (~ 1 mg/mL in CS2) and the
charged states on the fullerene for the other contacts with high- matrix (elemental sulfur: 20 mg/mL). HPLC analyses were performed on a Hew-
lett Packard HP LC-Chemstation 3D (HP 1100 Series) using an analytical Cos-
er work functions. mosil Buckyprep column (4.6 ” 250 mm). Cyclic voltammetry was performed
using an Autolab PGStat 100 equipped with a BAS C3 cell stand. Elemental ana-
lyses were performed by the Microanalytical Department of this laboratory.
3-(2-Ethylhexyloxy)benzyl alcohol (2) [13]: Sodium methoxide (4.70 g,
3. Conclusion 87 mmol) was dissolved in methanol (40 mL), and a solution of 3-hydroxymethyl
phenol (9.92 g, 80 mmol) in methanol (50 mL) was added. The clear solution
In conclusion, we have found that the open circuit voltage in was heated under reflux for 30 min, and subsequently cooled to room tempera-
ture. A solution of 2-ethyl-1-bromohexane (17.0 g, 88 mmol) in methanol
plastic solar cells is directly related to the acceptor strength of (50 mL) was added dropwise. The mixture was heated under reflux overnight
the fullerenes. This result fully supports the view that the open and subsequently cooled to room temperature. The solvent was removed in
vacuo. Final purification was done by column chromatography (silica gel; tolu-
circuit voltage of this type of donor±acceptor bulk-heterojunc- ene/methanol = 7:1) giving pure product (5.55 g, 24 mmol, 29 %) as a colorless
tion cell is related directly to the energy difference between oil. 1H NMR (CDCl3, 300 MHz), d [ppm]: 7.27 (t, J = 8.0, 1H), 6.94±6.91 (m,
the HOMO level of the donor and the LUMO level of the 2H), 6.85 (dd, J = 8.1, J = 2.1, 1H), 4.66 (br s, 2H), 3.86 d, J = 6.0, 2H), 1.74 (sep, J
= 6.1, 1H), 1.55±1.27 (m, 8H), 0.96±0.89 (m, 6H). 13C NMR (CDCl3, 75 MHz), d
acceptor components. Furthermore, and also in full agreement [ppm]: 159.62, 142.40, 129.45, 118.76, 113.74, 112.88, 70.40, 65.22, 39.34, 30.49,
with this view, it was found that a variation of the negative elec- 29.03, 23.82, 23.00, 14.03, 11.05. FTIR (KBr), m [cm±1]: approx. 3331 (m, OH),
trode work function influences the open circuit voltage in only 1264 (s, ArOR), 1037 (m, ArOR), 782 (m, Ar). HRMS: Calcd for 12C15H24O2
(M+): m/z = 236.178. Found m/z = 236.177. Anal. calcd. for C15H24O2: C,
a minor way. In accord with results from XPS studies proving 76.23 %; H, 10.23 %. Found: C, 76.32 %; H, 10.28 %.
surface charges on C60 at metal interfaces, this electrode-insen- 3-(2-Ethylhexyloxy)benzyl bromide (3) [14]: 3-(2-Ethylhexyloxy)benzyl alco-
hol (5.50 g, 23 mmol) was dissolved in ether (235 mL), and phosphorous tribro-
sitive voltage behavior is identified as a result of Fermi level mide (7.33 g, 27 mmol) dissolved in ether (60 mL) was added. After stirring for
pinning between the negative metal electrode and the fullerene 3 h at room temperature, the mixture was poured into water (1.0 L) and
reduction potential via charged interfacial states. These obser- extracted with dichloromethane (4 ” 500 mL). The combined organic layers were
dried over sodium sulfate and evaporated in vacuo. Final purification was done
vations lead the way to molecular engineering of the open cir- by column chromatography (silica gel; diethyl ether/cyclohexane = 1:1) giving
cuit potentials in plastic solar cells by tailoring the electronic pure 3 (4.10 g; 13.7 mmol, 59 %) as a colorless oil. 1H NMR (CDCl3, 300 MHz),
structure of the acceptor. d [ppm]: 7.24±7.19 (t, J = 7.8, 1H), 6.95±6.91 (m, 2H), 6.82 (dd, J = 8.3, J = 1.1,
1H), 4.45 (s, 2H), 3.82 (d, J = 6.0, 2H), 1.70 (sep, J = 6.0, 1H), 1.54±1.25 (m, 8H),
0.94±0.87 (m, 6H). 13C NMR (CDCl3, 75 MHz), d [ppm]: 159.57, 139.00, 129.70,
120.97, 115.05, 114.66, 70.46, 39.37, 33.58, 30.51, 29.07, 23.84, 23.03, 14.06, 11.10.
FTIR (KBr), m [cm±1]: 1266 (s, ArOMe), 693 (m, RBr). HRMS: Calcd for
4. Experimental 12
C15H23BrO (M+): m/z = 298.093. Found m/z = 298.094. Anal. calcd for
C15H23BrO: C, 60.21 %; H, 7.75 %; Br, 26.70 %. Found: C, 59.84 %; H, 7.84 %;
Device Fabrication: ITO/glass substrates (MDT Darmstadt) and ITO/poly- Br, 26.41 %.
ester substrates (Cadillac Plastic, Germany) were cleaned in ultrasonic baths of 3-(2-Ethylhexyloxy)benzyl azide (4) [15]: Sodium azide (990 mg, 15.2 mmol)
acetone and isopropanol. Devices for acceptor strength variation were produced was dissolved in DMSO (22 mL) by stirring overnight at room temperature.
by doctor blading. Poly(3,4-ethylenedioxythiophene) doped with polystyrene sul- Then 3-(2-ethylhexyloxy)benzyl bromide (4.10 g, 13.7 mmol) was added and the
fonic acid (PEDOT:PSS, Bayer AG) was doctor bladed to a thickness of 100 nm resulting mixture was stirred for 1.5 h. Addition of water (75 mL) and subse-
on top of the ITO, from a water solution. The active layer was cast from a toluene quent cooling to room temperature was followed by extraction with ether (3 ”
solution with equal molar ratios of MDMO-PPV and the fullerene (1:1). Deposi- 50 mL). The combined organic layers were washed with water (2 ” 75 mL) and
tion of the top electrode was done by thermal evaporation of Al through a shad- pre-dried over brine (1 ” 75 mL). Drying over sodium sulfate and subsequent
ow mask to define a device area of 5 mm2. Devices for variation of the negative evaporation of the solvent in vacuo yielded the azide (3.39 g, 13.0 mmol, 95 %)
electrode work function were produced by spin-casting a thin film from a as a slightly yellow oil, pure enough to allow for the subsequent reaction. 1H
MDMO-PPV/PCBM (1:1 molar ratio) solution from toluene. After drying, the NMR (CDCl3, 300 MHz), d [ppm]: 7.27±7.21 (m, 1H), 6.88±6.81 (m, 3H), 4.27 (s,
top electrodeÐCa, Al, Ag, or AuÐwas thermally deposited through a shadow 2H), 3.82 (d, J = 5.4, 2H), 1.70 (sep, J = 6.0, 1H), 1.56±1.24 (m, 8H), 0.93±0.86 (m,
mask to define a device area of 7.5 mm2. All electrodes were non-transparent 6H). 13C NMR (CDCl3, 75 MHz), d [ppm]: 159.73, 136.74, 129.71, 120.03, 114.32,
with a thickness of ~ 80 nm. Spin-casting of the devices was performed under 114.22, 70.45, 54.73, 39.36, 30.49, 29.05, 23.82, 23.00, 14.03, 11.07. FTIR (KBr), m
ambient conditions, avoiding exposure to light resonant with the photoactive [cm±1]: 2099 (s, N3), 1266 (s, ArOR).
components, thereby preventing photooxidation [33]. After evaporation of the N-3-(2-ethylhexyloxy)benzyl azafulleroid (5): C60 (1.08 g, 1.5 mmol) was dis-
metal, electrode devices were characterized in a dry-box under an argon atmo- solved in ODCB (200 mL), degassed (3 ”), and heated under reflux under a ni-
sphere. I±V characteristics were taken with a source-meter (Keithley 2400) while trogen atmosphere. 3-(2-Ethylhexyloxy)benzyl azide (408 mg, 1.56 mmol) was
illuminating the devices through the ITO/PEDOT side with 60 mW/cm2 white added and the reaction mixture was heated under reflux for 2 h. After cooling to
light from a halogen lamp. room temperature, all ODCB was removed in vacuo. After repetitive column
Synthesis of Azafulleroid and Ketolactam: All reagents and solvents were used chromatography (silica gel; CS2/cyclohexane = 1:4) (CAUTION! The column
as received, or purified using standard procedures. [60]Fullerene (99.5 %) was should be poured in cyclohexane since CS2 is extremely flammable; silica gel grade
purchased from Bucky USA and used without purification. All reactions were I is sufficient to ignite CS2) pure 5 was obtained as a brown solid. This was dis-
performed under a nitrogen atmosphere, unless indicated otherwise. Nitrogen solved and precipitated using CS2/pentane, washed with pentane (2 ”), and dried
was deoxygenated using a copper column. Flash chromatography was performed overnight in a vacuum oven (T = 50 C) yielding the pure azafulleroid (255 mg,
using Kieselgel Merck Type 9385 (230±400 mesh). Analytical thin layer chroma- 268 lmol, 18 %). m.p.: 201.7±202.8 C. 1H NMR (CS2, 500 MHz), d [ppm]: 7.42
tography (TLC) was performed using aluminum coated Merck Kieselgel 60 F254 (t, J = 8.0, 1H), 7.38±7.37 (m, 2H), 6.97±6.95 (m, 1H), 5.03 (s, 2H), 4.07±4.03 (m,
plates. Melting points were determined with a Mettler FP1 melting point appara- 2H), 1.91 (sep, J = 6.1, 1H), 1.76±1.58 (m, 4H), 1.57±1.49 (m, 4H), 1.16±1.10 (m,
tus equipped with a Mettler FP2 microscope. 1H NMR and 13C NMR spectrosco- 6H). 13C NMR (CS2, 125 MHz), d [ppm]: 159.53, 147.58, 146.09, 144.88, 144.51,
py were performed on a Varian Unity Plus (500 MHz) instrument, or on a Varian 144.37, 144.26, 144.11, 143.98, 143.90, 143.64, 143.47, 143.37, 143.23, 143.05,

Adv. Funct. Mater. 2001, 11, No. 5, October Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2001 1616-301X/01/0510-0379 $ 17.50+.50/0 379
 C. J. Brabec et al./Open Circuit Voltage of Plastic Solar Cells
FULL PAPER

142.94, 142.75, 142.65, 142.64, 142.52, 141.23, 140.66, 140.63, 139.14, 138.33, [5] C. J. Brabec, S. E. Shaheen. T. Fromherz, F. Padinger, J. C. Hummelen,
138.09, 138.01, 137.67, 137.35, 137.02, 136.07, 135.47, 133.50, 129.65, 121.04, A. Dhanabalan, R. A. J. Janssen, N. S. Sariciftci, Synth. Met., in press.
115.10, 113.91, 70.08, 54.78, 39.70, 30.96, 29.55, 24.48, 23.76, 14.69, 11.70. FTIR [6] L. H. Campbell, T. W. Hagler, D. L. Smith, J. P Ferraris, Phys. Rev. Lett.
(KBr), m [cm±1]: 1264 (s, ArOR), 525 (s, [60]fullerene). UV-vis (toluene), m [cm±1]: 1996, 76, 1900.
341 (e = 426 000 M±1 cm±1), 437 (sh), 551 (e = 13 900 M±1 cm±1). MALDI-TOF [7] C. M. Heller, I. H. Campbell, D. L. Smith, N. N. Barashkov, J. P. Ferraris,
MS: Calcd. for 12C75H23NO (M-): m/z = 953.2. Found: m/z = 953.7. Anal. calcd. J. Appl. Phys. 1997, 81, 3217.
for C75H23NO: C, 94.42 %; H, 2.43 %; N, 1.47 %. Found: C, 94.48 %; H, 2.39 %; [8] I. D. Parker, J. Appl. Phys. 1994, 75, 1656.
N, 1.55 %. [9] H. Antoniadis, B. R. Hsieh, M. A. Abkowitz, M. Stolka, S. A. Jenekhe,
N-3-(2-ethylhexyloxy)benzyl ketolactam (6): C60 (1.08 g, 1.5 mmol) was dis- Polym. Prepr. 1993, 34, 490.
solved in ODCB (200 mL), degassed (3 ”), and heated under reflux under an [10] S. Karg, W. Riess, V. Dyakonov, M. Schwoerer, Synth. Met. 1993, 54, 427.
atmosphere of nitrogen. 3-(2-Ethylhexyloxy)benzyl azide (408 mg, 1.58 mmol) [11] J. C. Hummelen, B. W. Knight, F. Lepec, F. Wudl, J. Yao, C. L. Wilkins,
was added and the reaction was allowed to proceed for 2 h under reflux. After J. Org. Chem. 1995, 60, 532.
cooling to room temperature, all ODCB was removed in vacuo. Column chroma- [12] C. J. Brabec, F. Padinger, N. S. Sariciftci, J. C. Hummelen, J. Appl. Phys.
tography (silica gel; CS2/cyclohexane = 1:4) (CAUTION! The column should be 1999, 85, 6866.
poured in cyclohexane since CS2 is extremely flammable; silica gel grade I is suffi- [13] P. L. Burn, A. Kraft, D. R. Baigent, D. D. C. Bradley, A. R. Brown, R. H.
cient to ignite CS2) gave a mixture consisting of C60 and [5,6]-3-(2-ethylhexyloxy)- Friend, R. W. Gymer, A. B. Holmes, R. W. Jackson, J. Am. Chem. Soc.
benzyl azafulleroid. This mixture was redissolved in ODCB (200 mL) and irra- 1993, 115, 10 117.
diated with a 400 W flood lamp using a sheet of Kapton 500HN as UV filter, [14] A. van Oeveren, J. F. G. A. Jansen, B. L. Feringa, J. Org. Chem. 1994, 59,
while a slow stream of oxygen was bubbled through the reaction mixture. HPLC 5999.
(Cosmosil; toluene/cyclohexane = 1:3) indicated complete reaction in 2 h. [15] S. G Alvarez, M. T. Alvarez, Synthesis 1997, 413.
Removal of the solvent in vacuo and column chromatography (silica gel; CS2 to [16] J. C. Hummelen, M. Prato, F. Wudl, J. Am. Chem. Soc. 1995, 117, 7003.
remove the starting fullerene, followed by silica gel; toluene/cyclohexane = 1:1) [17] T. Grösser, M. Prato, V. Lucchini, A. Hirsch, F. Wudl, Angew. Chem. Int.
afforded the pure ketolactam fraction. After removal of the solvent, the product Ed. Engl. 1995, 34, 1343.
was precipitated from CS2/pentane, washed with pentane (2 ”) and dried over- [18] This has also been determined for the MEM-ketolactam of reference [16]
night in a vacuum oven (T = 50 C) yielding the pure product (269 mg, 18 % (J. C. Hummelen, B. Knight, F. Wudl, personal communication 1995).
based on [60]fullerene) as a brown solid. m.p.: 129.7±131.2 C. 1H NMR (CS2, [19] a) H. Irngartinger, P. W. Fettel, T. Escher, P. Tinnefeld, S. Nord, M. Sauer,
500 MHz), d [ppm]: 7.31 (t, J = 8.0, 1H), 7.13±7.12 (m, 2H), 6.87 (dd, J = 7.5, J = Eur. J. Org. Chem. 2000, 455. b) P. de la Cruz, E. Espíldora, J. J. García,
2.0, 1H), 6.38 (d, J = 15.0, 1H), 5.38 (d, J = 15.0, 1H), 3.95±3.94 (m, 2H), 1.84 A. de la Hoz, F. Langa, N. Martín, L. Sµnchez, Tetrahedron Lett. 1999, 40,
(hep, J = 6.0, 1H), 1.67±1.52 (m, 4H), 1.50±1.44 (m, 4H), 1.10±1.05 (m, 6H). 13C 4889.
NMR (CS2, 125 MHz), d [ppm]: 194.78, 160.71, 159.56, 149.52, 148.95, 147.62, [20] T. Da Ros, M. Prato, M. Carano, P. Ceroni, F. Paolucci, S. Roffia, J. Am.
147.17, 147.13, 147.08, 146.49, 146.31, 146.14, 146.03, 145.89, 145.82, 145.79, Chem. Soc. 1998, 120, 11 645.
145.73, 145.55, 145.48, 145.14, 145.10, 145.00, 144.94, 144.74, 144.62, 144.27, [21] J.-P. Bourgeois, P. Seiler, M. Fibbioli, E. Pretsch, F. Diederich, L. Eche-
144.18, 144.08, 143.98, 143.88, 143.85, 143.73, 143.56, 143.53, 143.40, 143.38, goyen, Helv. Chim. Acta 1999, 82, 1572.
143.27, 143.18, 142.56, 141.49, 141.03, 140.95, 140.61, 140.44, 140.17, 139.61, [22] Handbook of Chemistry and Physics, 75th ed. (Ed: D. R. Lide), CRC Press,
139.25, 139.02, 138.61, 137.94, 137.51, 136.35, 135.77, 135.72, 135.42, 134.85, Boca Raton, FL 1995, pp. 12±113.
133.79, 133.54, 132.92, 132.06, 131.72, 129.89, 128.69, 120.49, 114.67, 114.29, 70.10, [23] L. J. Brillson, Surf. Sci. Rep. 1982, 2, 145.
53.99, 39.62, 30.91, 29.51, 24.43, 23.75, 14.69, 11.67. FTIR (KBr), m [cm±1]: 1727 (s, [24] B. W. Hoogenboom, R. Hesper, L. H. Tjeng, G. A. Sawatzky, Phys. Rev. B
CO), 1689 (s, CO (amide)), 1260 (m, ArOR), 522 (m, [60]fullerene); UV-vis (tol- 1998, 57, 11 939.
uene), m [cm±1]: 333 (e = 488 000 M±1 cm±1), 433 (e = 44 400 M±1 cm±1), 619 (e = [25] T. R. Ohno, Y. Cehn, S. E. Harvey, G. H. Kroll, J. H. Weaver, R. E. Hau-
4 560 M±1 cm±1), 698 nm (e = 4 410 M±1 cm±1). MALDI-TOF MS: Calcd for fler, R. E. Smalley, Phys. Rev. B 1991, 44, 13 747.
12
C75H23NO3 (M-): m/z = 985.2. Found: m/z = 985.8. Anal. calcd. for C75H23NO3: [26] R. Hespers, Ph.D. Thesis, Rijksuniversiteit Groningen, Groningen, The
C, 91.36 %; H, 2.35 %; N, 1.42 %. Found: C, 91.42 %; H, 2.35 %; N, 1.41 %. Netherlands 2000.
[27] L. H. Tjeng, R. Hesper, A. C. L. Heessels, A. Heeres, H. T. Jonkman,
Received: February 14, 2001 G. A. Sawatzky, Solid State Commun. 1997, 103, 31.
Final version: June 8, 2001
[28] S. J. Chase, W. S. Basca, M. G. Mitch, L. J. Pilione, J. S. Lannin, Phys. Rev.
B 1992, 46, 7873.
± [29] M. R. C. Hunt, S. Modesti, P. Rudolf, R. E. Palmer, Phys. Rev. B 1995, 51,
[1] G. Yu, J. Gao, J. C. Hummelen, F. Wudl, A. J. Heeger, Science 1995, 270, 1789. 10 039.
[2] N. S. Sariciftci, A. J. Heeger, in Handbook of Organic Conductive Mole- [30] I. G. Hill, D. Milliron, J. Schwartz, A. Kahn Appl. Surf. Sci. 2000, 166, 354.
cules and Polymers (Ed: H. S. Nalwa), Vol. 1, John Wiley & Sons, New York [31] P. W. M. Blom, M. J. M. de Jong, M. G. van Munster, Phys. Rev. B 1997,
1997. 55, 656.
[3] a). C. J. Brabec, N. S. Sariciftci, J. C. Hummelen, Adv. Funct. Mater. 2001, [32] P. W. M. Blom, M. J. M. de Jong, J. J. M. Vleggaar, Appl. Phys. Lett. 1996,
11, 15. b). C J. Brabec, N. S Sariciftci, in Conjugated Polymers (Eds: 68, 3308.
G. Hadziannou, P. van Hutten), Wiley-VCH, Weinheim 1999. [33] H. Neugebauer, C. J. Brabec, J. C. Hummelen, N. S. Sariciftci, Sol. Energy
[4] S. E. Shaheen, C. J. Brabec, F. Padinger, T. Fromherz, J. C. Hummelen, Mater. Sol. Cells 2000, 61, 35.
N. S. Sariciftci, Appl. Phys. Lett. 2001, 78, 841. [34] D. C. Brune, Rapid Commun. Mass Spectrom. 1999, 13, 384.

______________________

380 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2001 1616-301X/01/0510-0380 $ 17.50+.50/0 Adv. Funct. Mater. 2001, 11, No. 5, October