See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/224445725

CuPc:C60 blend film: A photoemission

investigation

Article in Journal of Vacuum Science & Technology A Vacuum Surfaces and Films · August 2006
DOI: 10.1116/1.2194923 · Source: IEEE Xplore

CITATIONS READS

13 96

7 authors, including:

Luca Lozzi Silvia Picozzi

Università degli Studi dell'Aquila Italian National Research Council
216 PUBLICATIONS 4,346 CITATIONS 222 PUBLICATIONS 5,573 CITATIONS

SEE PROFILE SEE PROFILE

Mirko Simeoni Sandro Santucci

Consorzio C.R.E.O. Università degli Studi dell'Aquila
9 PUBLICATIONS 164 CITATIONS 398 PUBLICATIONS 7,558 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

organic spintronic devices View project

Nanotools for brain delivery View project

All content following this page was uploaded by Mirko Simeoni on 22 September 2017.

The user has requested enhancement of the downloaded file.

CuPc: C60 blend film: A photoemission investigation
L. Lozzia兲 and V. Granato
CNISM and Department of Physics, University of L’Aquila, Coppito, L’Aquila 67010, Italy
S. Picozzi
INFM-CNR and Department of Physics, University of L’Aquila, Coppito, L’Aquila 67010, Italy
M. Simeoni
CNISM and Department of Physics, University of L’Aquila, Coppito, L’Aquila 67010, Italy
S. La Rosa
Sincrotrone Trieste SCpa, Basovizza, I-34012 Trieste, Italy
B. Delly
Paul Scherrer Institut, WHGA/123, CH-5232 Villigen PSI, Switzerland
S. Santucci
CNISM and Department of Physics, University of L’Aquila, Coppito, L’Aquila 67010, Italy
共Received 15 May 2005; accepted 13 March 2006; published 23 June 2006兲
The electronic structure of thin blend films composed of copper phthalocyanine 共CuPc兲 and
fullerene 共C60兲 have been studied by x ray, soft-x-ray photoemission spectroscopy, and by
theoretical simulations. Samples with different C60 / CuPc concentration ratios were prepared by
thermal evaporation in ultrahigh vacuum conditions. A strong shift of the highest occupied
molecular orbital 共HOMO兲 of CuPc towards the Fermi level and of the HOMO C60 states towards
higher binding energy was observed. The comparison between valence band structures and core
level line shape indicates that the central part of the CuPc molecule is lightly modified by the
presence of the C60 molecule. The theoretical analysis, performed using density functional theory,
confirms this small interaction between the two molecules, with a small charge density localized
between the two molecules and a structural modification of the CuPc molecule. © 2006 American
Vacuum Society. 关DOI: 10.1116/1.2194923兴

I. INTRODUCTION area is achieved within a bulk material. If the donor-acceptor

In the last years a strong research effort was devoted in
the realization of solar cells whose active element is made by
an organic layer.1–3 The use of organic compounds 共as con-
jugated polymers or phthalocyanines兲 is driven by the possi-
bility to reduce the cost of the organic cell, to tailor better the
cell characteristics due to the enormous variety of available
compounds and to prepare solar cells on flexible substrates.
One of the most important problems is the very low effi-
ciency of these organic cells 共about 3%–4%兲, because of the
recombination of the electron-hole pair 共exciton兲 before its
separation at the metallic electrodes.4 In order to reduce the
exciton recombination an electron acceptor 共generally
fullerene, C60兲 was introduced. These devices are formed as
a bilayer 共metal 1/active layer/fullerene/metal 2兲 or by mix-
ing the active layer with fullerene forming a blend film. This
last configuration should allow a better charge separation be-
cause of both the increased organic-fullerene interface area,
forming a “bulk heterojunction,” and the strong reduction of
the exciton path 共the exciton diffusion length is quite short,
about 10– 20 nm兲.5,6 Through careful control of the morphol-
ogy of the blend film and of the transformation from a phase
separation into an interpenetrating network, a high interfacial
a兲
Author to whom correspondence should be addressed; electronic mail:
luca.lozzi@aquila.infn.it
network is continuous also the collection efficiency can be
optimized because the efficiency of the charge separation
process is nearly 100% as a result of the fast electron transfer
from the donor compound towards the fullerene 共⬍50 fs兲.7
However, the blend configuration did not improve the device
characteristics.6,8 An important parameter influencing the
overall cell efficiency is the position of the molecular orbitals
involved in the photon absorption, that is the highest occu-
pied molecular orbital 共HOMO兲 and lowest unoccupied mo-
lecular orbital 共LUMO兲 levels. Their positions, in particular,
with respect to the Fermi level of the metallic contacts, affect
the charge transfer process because Schottky barriers at the
interfaces can be formed. Moreover the alignment of the do-
nor and acceptor energy levels, makes the exciton dissocia-
tion energetically favorable and any variation of these states
may influence the dissociation probability.5
Generally, in simplified energy level diagrams used for
a priori evaluation of the electrical behavior of an organic/
C60 blend film, the HOMO-LUMO levels of the organic
compound and C60 are taken from the literature as those of
the pure elements8 and any variation induced by the chemical
interaction among the two compounds is not taken into ac-
count. Actually, in the case of a bilayer composed as a thin
film of C60 deposited onto copper phthalocyanine 共CuPc兲
共one of the most used active layers兲 a shift of the CuPc
HOMO level towards the Fermi level was detected, and it

1668 J. Vac. Sci. Technol. A 24„4…, Jul/Aug 2006 0734-2101/2006/24„4…/1668/8/$23.00 ©2006 American Vacuum Society 1668
1669 Lozzi et al.: CuPc: C60 blend film: A photoemission investigation 1669

was attributed to the formation of a surface band bending in

the CuPc film, while no effects were observed for the C60
electronic structure.9 Therefore we think that the electronic
structure of any blend layer needs understanding. With this
aim, in this work we present the photoemission results on the
valence band of CuPc: C60 blend films, prepared with differ-
ent concentrations. These experimental results will be com-
pared with calculated data which were used to determine the
structural and electronic properties of the CuPc: C60 system.

II. EXPERIMENTAL AND THEORETICAL

DETAILS
The low energy photon experiment was performed at the
spectromicroscopy beam line in Elettra, Trieste 共Italy兲. The
photon energy was 95 eV and the beam was strongly defo-
cused 共beam diameter of about 100 ␮m兲 in order to avoid
any sample damage. The blend films were prepared by ther-
mal evaporation in ultrahigh vacuum conditions onto a thick
gold film. CuPc and C60 powders 共Aldrich兲 were evaporated
from quartz crucibles. Both the gold and blend films deposi-
tions were performed in two UHV chambers connected with
the experimental apparatus, therefore the samples were not
exposed to air before the analysis. Then the samples were
analyzed using a nonmonochromatic x-ray photoemission
spectroscopy 共XPS兲 apparatus. From the analysis of the core
level intensities the C60 / CuPc concentration ratios were de-
termined. The blend film thickness is greater than 10 nm, as
judged by the lack of any Au 4f signal in the XPS spectra.
The zero energy scale was fixed at the gold substrate Fermi
level.
All the calculations were performed by means of the all-
electron DMOL3 共Ref. 10兲 ab initio simulation code within the
framework of density functional theory11 共DFT兲 with gener-
alized gradient approximations.12 The relativistic effects 共ex-
cept for spin-orbital coupling兲 are considered using a
pseudopotential scheme.13 The electronic structures were in-
vestigated by an analysis of the Kohn-Sham14 共KS兲 energy
levels, of the density of states 共in particular, focusing on
states around the HOMO-LUMO gap兲 and of charge density
contours. A Gaussian broadening of 0.1 eV was used for the
density of states 共DOS兲 plots in order to simulate that experi-
mental resolution.
III. RESULTS AND DISCUSSION
A. Experimental results
In Fig. 1 we show the valence band spectra acquired on
the CuPc bulk 共bottom curve兲, the C60 film 共top curve兲 and
the different CuPc: C60 blend films, with the C60 concentra-
tion increasing from bottom to top. The spectra were ac-
quired using 95 eV photon energy. The CuPc bulk film
shows the typical spectrum acquired using this photon
energy.15 The spectrum shows the HOMO peak at about
1.8 eV considering the maximum 共and 1.4 for the leading
edge of the HOMO peak, but in this article all the reported
energies will indicate the peak position兲. The photoemission
HOMO peak is mainly due to Cu 3d states, while the con-
FIG. 1. Valence band spectra acquired on the CuPc film 共bottom curve兲, C60
film 共upper curve兲, and for different CuPc: C60 concentrations: 共a兲
C60 / CuPc= 0.3, 共b兲 C60 / CuPc= 1.7, 共c兲 C60 / CuPc= 3, 共d兲 C60 / CuPc= 4, and
共e兲 C60 / CuPc= 13.
tribution of carbon states 共mainly pyrrol carbon 2p, C1, see
Fig. 7兲, whose energy is about 0.15– 0.2 eV higher than the
Cu states, is negligible because of the cross section.16 A more
detailed discussion is reported in the theoretical section be-
low. Then there are all the features related to different mo-
lecular orbitals, whose intensity depends on the involved
atomic states and cross section. The C60 spectrum 共top curve兲
shows the HOMO peak at about 2.4 eV and the other mo-
lecular orbitals in agreement with previously published
data.17 Starting from the pure CuPc film and increasing the
fullerene concentration, the C60 states start to influence the
spectrum. The most interesting part of the spectra is the up-
per part, between 7 eV binding energy and the Fermi level,
and, in particular, the behavior of the CuPc and C60 HOMO
peaks. The CuPc HOMO peak, suddenly moves from 1.8 eV
共CuPc bulk兲 to about 1.4 eV 共C60 / CuPc= 0.3兲 then, increas-
ing the C60 concentration, there is a further small shift up to
about 1.2 eV for the highest C60 percentage sample. The sec-
ond peak of the CuPc bulk spectrum at about 2.5 eV is
quickly hidden by the strong HOMO peak of the C60, that
already appears for the C60 / CuPc= 1.7 film. The C60 HOMO
peak is still missing for the C60 / CuPc= 0.3 film. Then its
intensity grows and it is located at about 2.7 eV 共for the
C60 / CuPc= 1.7 film兲. Finally, it moves up to 2.4 eV for the
C60 bulk film. We have to underline that the shift of the
HOMO states of the two molecules is not equal 共about
0.6 eV for CuPc HOMO states and 0.3 eV for the C60
HOMO one兲, ruling out any simple electrostatic effects. All
the other valence structures do not follow a simple shift,
because of the superposition and probably an interaction be-

JVST A - Vacuum, Surfaces, and Films

1670 Lozzi et al.: CuPc: C60 blend film: A photoemission investigation
FIG. 2. Comparison between the upper part of the valence band spectra of
CuPc 共continuous line兲 and of C60 / CuPc= 0.3 sample. These spectra were
aligned at the HOMO peak.
tween CuPc and C60 molecular levels. For example, the
CuPc peak at about 6 eV does not shift as the CuPc HOMO
one, on going from CuPc bulk to the C60 / CuPc= 0.3 film.
This is evident in Fig. 2 where we compare the valence band
spectrum of the CuPc bulk sample 共continuous line兲 with that
of the C60 / CuPc= 0.3 sample 共dashed line兲. This latter spec-
trum was shifted by +0.4 eV in order to line up the HOMO
peak of the two spectra. The shift of the peak at 6 eV is
0.2 eV lower than that of the HOMO peak. This different
shifts cannot be simply attributed to the growth of the C60
contribution under the CuPc signal. In fact, if we try to fit the
C60 / CuPc= 0.3 experimental spectrum by a linear combina-
tion using the bulk CuPc and C60 spectra 共using as fitting
parameters both the intensities and the shifts with respect to
the position of the pure film spectra兲 we cannot obtain the
measured energy position for all the main peaks, in particu-
lar, the three low energy peaks 共1.4, 3.9, and 6 eV兲. More-
over the contribution of the C60 states is also depressed by
the low cross section of the carbon states for the used photon
energy.16 These data suggest that the coevaporation of CuPc
and C60 molecules determine an interaction between them,
with the variation of the molecular levels.
A further indication of the interaction between the mol-
ecules comes from the core level photoemission. In Fig. 3 we
report the carbon 1s core level spectra for the same samples
reported in Figs. 1 and 2. The bottom curve is the typical C1s
peak for the CuPc molecule, showing the characteristic three
peak shape, due to carbon-carbon peak 共284.4 eV兲, carbon-
nitrogen bond 共285.8 eV兲, and to the shake-up 共␲-␲* transi-
tion兲 effect 共287.8 eV兲.18 This shake-up feature is generally
attributed to a satellite of the carbon-nitrogen peak, while the
satellite of the carbon-carbon peak is hidden under the
J. Vac. Sci. Technol. A, Vol. 24, No. 4, Jul/Aug 2006
1670
FIG. 3. Carbon 1s peak for the CuPc film 共bottom curve兲, C60 film 共upper
curve兲, and for CuPc: C60 with different concentrations: 共a兲 C60 / CuPc= 0.3,
共b兲 C60 / CuPc= 1.7, 共c兲 C60 / CuPc= 3, 共d兲 C60 / CuPc= 4, and 共e兲 C60 / CuPc
= 13. In the insert a magnification of the shake-up region is reported.
carbon-nitrogen main peak and can be evidenced by a careful
analysis of the spectrum. The C60 film 共top curve兲 presents a
single peak located at about 284.5 eV and a quite wide
shake-up region 共see insert兲. These features are related to the
excitation from different occupied molecular levels to sev-
eral unoccupied ones.19 The HOMO-LUMO transition deter-
mines the first peak at about 1.8 eV from the main C1s peak
共284.5 eV兲. Because of the energy resolution, we have only a
shoulder at about 2 eV. Then there are two clear peaks at
about 4 and 6 eV, attributed to higher energy molecular tran-
sitions. These three structures are in good agreement with
high resolution XPS spectra.19
All the CuPc: C60 blend films show a shift of the main
peak towards higher binding energy 共about +0.3 eV兲 and its
position is independent of the film composition. While for
low CuPc concentration samples 共curves c–e兲 the line shape
is very similar to that of C60, that is quite symmetric, for high
CuPc concentrations the line shape becomes asymmetric,
showing new structures on both sides and the main line
width broadens, on going from about 1 eV 共C60兲 to 1.5 eV
共C60 / CuPc= 3兲. This is evident from the C60 / CuPc= 1.7
sample 共curve b兲, where a shoulder at about 284 eV and a
peak at about 287 eV appear. Then, for the C60 / CuPc= 0.3
sample 共curve a兲 the line shape becomes very similar to the
pure CuPc and the C60 contribution is hidden under the
shoulder at about 285.5 eV. Similar effects are shown in the
shake-up region. For the C60 / CuPc= 13 sample 共the C60 rich-
est one兲 the shake-up structure is very similar to the C60 pure
film, apart from a shift to higher binding energy 共+0.3 eV兲,
as the shift of the main C1s peak. Increasing the CuPc con-
1671 Lozzi et al.: CuPc: C60 blend film: A photoemission investigation
FIG. 4. Comparison between the carbon 1s peak of the CuPc 共continuous
line兲, C60 共dashed line兲, and C60 / CuPc= 0.3 sample 共dotted line兲. The spectra
were aligned in order to compare the line shape.
centration the shape of the shake-up region strongly changes
because of the growing of the tail of the shake-up signal of
the CuPc C1s contribution. In Fig. 4 we compare the C1s
spectra of pure CuPc 共continuous line兲, pure C60 共dashed
line兲, and C60 / CuPc= 0.3 samples 共dotted line兲. The three
spectra were aligned in order to allow a better comparison
between the line shapes. It is evident that the C60 / CuPc
= 0.3 sample spectrum is not simply a combination of the
other two spectra. In fact, the C–N peak seems to move
towards lower binding energy and its intensity increases, al-
though in the blend film the concentration of C–N bonds is
lower than in pure CuPc film. This comparison suggests that
the interaction between CuPc and C60 molecules influences
the measured binding energies of the carbon spectra and that
they have to be carefully analyzed in order to better under-
stand this interaction. For this reason a careful fitting analy-
sis of the C1s core level was performed on all the samples.
In this fitting we followed the evolution of the CuPc and C60
components for the different samples. In Fig. 5 we report the
fitting analysis on three samples, CuPc bulk 共bottom spec-
trum兲, C60 / CuPc= 0.3 共middle spectrum兲, and C60 / CuPc
= 1.7 共upper spectrum兲. For the CuPc contribution we intro-
duced, following the results by Schwieger et al.,18 four com-
ponents as already indicated, while for the C60 contribution
we included two components, the main peak and one
shake-up peak. In the fitting procedure, we fixed the intensity
ratio between the different contributions 共that is for the CuPc
contribution the C – C / C – N intensity ratio was fixed to 3.0
as in the bulk sample兲 and the energy distance between the
main peak and its shake-up was allowed to vary within
±0.3 eV with respect to the bulk value. Moving from the
CuPc bulk to the C60 richest samples, apart from the decrease
JVST A - Vacuum, Surfaces, and Films
1671
FIG. 5. Carbon 1s peak for CuPc pure film 共bottom curve兲, C60 / CuPc= 0.3
共middle curve兲, and C60 / CuPc= 1.7 共upper curve兲. The result of the fitting
analysis is also reported.
of the intensity of the contribution due to the carbon atoms of
the CuPc molecule, a shift of the different peaks is notice-
able. In particular, as the C60 concentration increases, the
C–C peak of C60 moves continuously towards lower binding
energy values 共from 285.4 to 284.5 eV兲, while the binding
energies of the peaks due to the C–C and C–N bonds in the
CuPc molecule first increase then decrease. This effect is
clearly shown in Fig. 6 where the position of the main peaks
is reported as a function of the C60 / CuPc concentration ratio.
This figure also reported the behavior for the N1s main peak
关this core level spectrum presents in the CuPc bulk film a
main perk at about 398.8 eV and a shake-up peak at about
400.8 eV 共Ref. 20兲, which is very similar to that followed by
the CuPc C1s contributions 共C–N and C–C兲. These effects
are more evident and important when the C60 / CuPc ratio is
quite small 共0.3 or 1.7兲. In fact, passing from 0.3 to 1.7 the
number of CuPc and C60 molecules interacting each other is
quite high, that is, almost all CuPc molecules are probably
surrounded by one or more C60 molecules and vice versa. A
quite similar behavior is also observed for the Cu2p core
level 共here not reported兲. The fitting results indicate that the
interaction between CuPc and C60 molecules determines the
variation of the core level binding energies, which could be
attributed to some charge redistribution inside the molecule
or between them. Another possible effect influencing the
core level binding energy can be the electron acceptor char-
acteristic of the C60 molecule. Any “free” electron originated
1672 Lozzi et al.: CuPc: C60 blend film: A photoemission investigation
FIG. 6. Binding energy position, as a function of the C60 / CuPc ratio, for the
components of the C1s spectrum, obtained using the fitting procedure shown
in Fig. 5, and for the main N1s peak.
on the CuPc molecule, for example, by the deexcitation pro-
cess following the photoionization effects 共the same effect
that determines the presence of shake-up peaks兲, could be
very quickly removed from the CuPc molecules resulting in
an increase of the measured binding energy. Note that the
shifts reported in the core level spectra cannot be attributed
to any charging effects due to variation of the sample con-
ductivity 共for example, increasing the C60 concentration兲. In
fact, since all the samples showed a small oxygen contami-
nation 共due to the transfer from the soft-x-ray photoemission
system to the x-ray apparatus for the core level analysis兲, we
have used this signal as surface binding energy reference.
These spectra show, within 0.1 eV, the same binding energy
ruling out any electrostatic contribution on the core level
shifts.
B. Theoretical results
The correspondence between the results observed on the
different core levels and the evolution of the HOMO levels
suggests that in a CuPc: C60 blend film some interaction be-
tween the two molecules is likely to occur. In order to con-
firm our experimental data a simulation of the CuPc: C60 film
共with a C60 / CuPc= 1 ratio兲 was performed. For this calcula-
tion we firstly determined the electronic properties of the two
separated molecules, C60 and CuPc, respectively. Then the
electronic structure of the joint system was determined.
C60, which is essentially a carbonic “soccer ball,” has a
total of 240 valence electrons, but up to 180 are used to link
each carbon atom to its neighbors with three ␴ bonds. These
␴ bonds do not form a plane and the ␲ electrons tend to be
localized mainly outside the C60 carbon cage, whereas in a
planar graphene sheet they have the same probability of be-
ing on one side or the other. The remaining 60 valence elec-
trons are distributed on carbon-carbon ␲ orbitals; however,
in C60 the C–C bond lengths are not uniform, therefore the ␲
electrons are not homogeneously distributed all around the
J. Vac. Sci. Technol. A, Vol. 24, No. 4, Jul/Aug 2006
1672
FIG. 7. Three-dimensional 共3D兲 view of the isolated CuPc molecule 共upper兲
and of the joint system 共bottom兲.
six-membered ring, at variance with benzene. The geometry
of C60 共icosahedral point group Ih兲 is fully determined by two
bond lengths, the pentagon edge, and the hexagon-hexagon
bond. Our calculation gives bond lengths of 1.452 and
1.398 Å, respectively, and internal pentagon angles of 108.0°
in good agreement with previous computational and experi-
mental works.21 The basic calculated electronic structure of
fullerene is consistent with previous works.22 The main elec-
tronic levels, that is the HOMO, show hu symmetry, and the
LUMO with a t1u symmetry. The calculated energy gap be-
tween these states is 1.67 eV in good agreement with previ-
ous computational results,23 but quite lower than the experi-
mental value of ⬃3.3 eV,17 as expected from a DFT
approach.
The single CuPc molecule is a planar macrocycle 共D4h
point group兲 constituted by four units presenting a p-electron
aromatic cloud delocalized over an arrangement of alternated
carbon and nitrogen atoms 共see Fig. 7, upper part兲. The cal-
culated HOMO-LUMO gap is 1.30 eV, again smaller than
experimental values 关the optical gap is ⬃1.7 eV 共Ref. 24兲
and the transport gap is ⬃2.3 eV 共Ref. 18兲兴. In Fig. 8 we
show the comparison between the experimental spectra
共dashed line兲 and the theoretical spectra 共continuous line兲 for
C60 molecule 共lower panel兲 and CuPc 共upper panel兲. In order
to compare the experimental and theoretical spectra we
aligned them setting the HOMO peaks at zero binding en-
ergy. The theoretical spectra were calculated multiplying the
DOS by the cross section for the used photon energy
共95 eV兲.16 For the C60 both HOMO and LUMO states essen-
tially originate from the l = 1 共p states兲 component, while the
1673 Lozzi et al.: CuPc: C60 blend film: A photoemission investigation
FIG. 8. Comparison between the experimental 共dashed line兲 and the theoret-
ical 共continuous line兲 spectra for the CuPc 共upper panel兲 and C60 共lower
panel兲. The spectra were aligned at the HOMO peak. The binding energy
reference was positioned at the leading edge of the theoretical HOMO peak.
In the insert the contributions to the HOMO peak 共continuous line兲 due to
Cu states 共dashed line兲 and C1 states 共dotted line兲 are shown. On the left are
reported the DOS and PDOS intensities, on the right the DOS and PDOS
times the cross section. The arrows indicate the same states before and after
the cross section use.
contribution of all the other states is negligible. For this mol-
ecule there is a very good correspondence between the the-
oretical and experimental data.
CuPc is a much more complex system, due to the pres-
ence of different atomic species, in particular, the Cu central
atom, and thereby its spectral analysis is less straightforward.
By taking into account the number of equivalent atoms, the
density of states shows that HOMO and LUMO essentially
originate from the C1 共pyrrolic兲 carbon atomic levels while
all the other atoms give smaller 共and very similar兲 contribu-
tions. However, the introduction of the cross section strongly
modified the weight of the different atoms in the CuPc
HOMO spectrum and, as already reported,15 the photoemis-
sion spectrum is dominated by the copper contribution while
the other states are strongly depressed. In the insert of Fig. 8
we report on the left the HOMO total DOS 共continuous line兲,
the Cu partial density of states 共PDOS兲 共dashed line兲, and the
C1 PDOS 共dotted line兲, while on the right the same contri-
bution after the introduction of the cross section 共the arrows
indicate the same contribution before and after the applica-
tion of the cross section兲. All the other contributions are very
similar to that of Cu, except for N2 and C2 which do not
significantly contribute to the HOMO. It is evident that, al-
though the HOMO states are mainly located on the carbon
TABLE I. Bond lengths 共in Å兲 in CuPc molecule and the corresponding in the joint system.
Bonds 共A兲 Cu–N1 N1–C1 N2–C1 C1–C2
CuPc 1.968 1.379 1.328 1.457
Joint 1.960 1.367 1.317 1.446
JVST A - Vacuum, Surfaces, and Films
1673
FIG. 9. Side view of the charge density contour of the CuPc: C60 joint
system.
atoms and, in particular, on the C1 sites, the photoemission
spectrum gives us the effect of the copper atom only.
Move now to the CuPc: C60 system. The “joint” system
consists of a C60 molecule and a CuPc molecule ⬃3 Å apart.
Figure 7 共lower image兲 shows the optimized structure. The
main result is the bending of CuPc molecule that loses its
planarity 共upper image兲. However, the bond lengths 共cf Table
I兲 do not change significantly 共all variations are below tenths
of angstroms兲 with respect to the isolated molecule.
Then, we calculated the electronic properties for the opti-
mized geometrical configuration, paying particular attention
to the HOMO-LUMO region.
First of all, we note that the charge density contours 共Fig.
9兲 show a small charge density in the interstitial region be-
tween the two molecules, suggesting a very weak interaction
between them. This hypothesis is confirmed by the analysis
of Mulliken populations 共see Table II兲 which points to an
intramolecular electronic charge transfer towards the nitro-
gen atoms from hydrogen and copper atoms in the CuPc,
whereas there is no charge transfer from C60 to CuPc or vice
versa. This is also in agreement with the core level spectra,
in particular, for the CuPc core levels, which do not show a
shift of all the peak towards the same direction 共a fingerprint
of a charge transfer兲, but a variety of effects which can be
partially interpreted as due to a partial charge transfer be-
tween atoms on the same CuPc molecule.
In Fig. 10 we present the density of states of the joint
system compared with that of the two subsystems. These
theoretical spectra were aligned fixing the Fermi level at the
midgap. Using this common reference line we observe a shift
of the states of the CuPc towards the Fermi level, in particu-
lar, the CuPc HOMO states, while the C60 ones moves to-
wards high binding energy values. This effect does not indi-
C2–C3 C3–C4 C4–C5 C2–C6 C–H
1.398 1.398 1.410 1.410 1.093
1.386 1.388 1.401 1.399 1.098
1674 Lozzi et al.: CuPc: C60 blend film: A photoemission investigation 1674

TABLE II. Mulliken analysis of the density of charge for the isolated CuPc IV. CONCLUSIONS
molecule and for the joint system. In the last column 共␦ charge兲 we list the
Mulliken charge into the joint system minus the charge into the isolated In conclusion we have shown that in the formation of a
CuPc molecule. CuPc: C60 blend film the upper molecular levels of both
components are influenced by a weak intermolecular interac-
CuPc: C60
tion. In particular, the CuPc HOMO level is shifted towards
CuPc Joint
the Fermi level, while the C60 HOMO peak moves towards
Atom Charge Charge ␦ charge higher binding energies. Core level spectra show differences
N1 −0.46 −0.47 −0.01 in the binding energy of the different components, confirm-
N2 −0.27 −0.29 −0.02 ing a small interaction. These data are confirmed by the cal-
C1 +0.28 +0.28 +0.00 culation of the electronic structure of the blend film, that also
C2 +0.03 +0.03 +0.00
show a bending of the CuPc molecule when it is close to a
C3 −0.27 −0.27 +0.00
C60 molecule. Of course the determination of the effects of
C4 −0.25 −0.25 +0.00
Cu +0.47 +0.51 +0.04
the interaction between CuPc and C60 molecules on the elec-
H +0.25 +0.27 +0.02 tronic properties of a blend film cannot fully explain the
missing of the improvement in the solar cell device charac-
teristics. However, since the electronic state energies, in par-
ticular, the HOMO binding energies, can severely affect the
charge transfer between the different compounds, our results
cate a strong interaction between the two molecules, because may help for a deeper understanding of the device perfor-
if we look at the PDOS on the single atoms they are not very mances. These results have to be carefully considered when
different with respect to those of the separated molecules. CuPc: C60 blend films are used in the realization of organic
Moreover if we simply sum the DOS of the CuPc and C60 solar cells.
with no interaction at all this “ideal” system presents a DOS
very similar to the joint ones. However, the system shows,
ACKNOWLEDGMENT
with respect to the separate molecules, a shift of the upper
levels, in particular, those indicated in the figure by the ver- This work has been partially supported by Ministero
tical bars, which correspond to the C60 and CuPc upper lev- dell’Istruzione, dell’Università e della Ricerca Scientifica
els. These shifts are in agreement with those reported in the 共PRIN No. 2003093440兲.
experimental photoemission spectra 共Fig. 1兲. The HOMO
1
state of the joint system depends on the CuPc molecule 共and, G. Yu, J. Gao, J. C. Hummelen, F. Wudi, and A. J. Heeger, Science 270,
1789 共1995兲.
as previously reported, mainly by the Cu states for the pho- 2
P. Peumans, S. Uchida, and S. R. Forrest, Nature 共London兲 425, 158
toemission spectrum兲, while the LUMO state is mainly lo- 共2003兲.
3
calized on the C60 molecule. Note that the shift of the LUMO H. Spanggaard and F. C. Krebs, Sol. Energy Mater. Sol. Cells 83, 125
is slightly higher 共0.5 eV兲 than that of the C60 HOMO state 共2004兲.
4
J. Xue, S. Uchida, B. P. Rand, and S. R. Forrest, Appl. Phys. Lett. 84,
共0.4 eV兲, determining a reduction of the band gap in the 3013 共2004兲.
fullerene molecule. 5
P. Peumans, A. Yakimov, and S. R. Forrest, J. Appl. Phys. 93, 3693
共2003兲.
6
P. Sullivan, S. Heutz, S. M. Schultes, and T. S. Jones, Appl. Phys. Lett.
84, 1210 共2004兲.
7
C. J. Brabec, G. Zerza, G. Cetrullo, S. De Silvestri, S. Luzzati, J. C.
Hummelen, and S. Sariciftci, Chem. Phys. Lett. 340, 232 共2001兲.
8
S. Uchida, J. Xue, B. P. Rand, and S. R. Forrest, Appl. Phys. Lett. 84,
4218 共2004兲.
9
C. Schlebusch, B. Kessler, S. Cramm, and W. Eberhardt, Synth. Met. 77,
151 共1996兲.
10
B. Delley, J. Chem. Phys. 92, 508 共1990兲; 105, 192 共1996兲; 113, 7756
共2000兲.
11
M. Springborg, Methods of Electronic Structure Calculations: From
Molecules to Solids 共Wiley, New York, 2000兲 and references therein.
12
P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 共1996兲.
13
B. Delley, Phys. Rev. B 66, 155125 共2002兲; Int. J. Quantum Chem. 69,
423 共1998兲.
14
W. Kohn and L. J. Sham, Phys. Rev. 140, A1133 共1965兲.
15
L. Lozzi, S. Santucci, S. La Rosa, D. Delly, and S. Picozzi, J. Chem.
Phys. 121, 1883 共2004兲.
16
J. J. Yeh and I. Lindau, At. Data Nucl. Data Tables 32, 1 共1985兲.
17
R. Schwedhelm, L. Kipp, A. Dallmeyer, and M. Skibowski, Phys. Rev. B
58, 13176 共1998兲.
18
T. Schwieger, H. Peisert, M. S. Golden, M. Knupfer, and J. Fink, Phys.
Rev. B 66, 155207 共2002兲.
19
FIG. 10. Theoretical DOS for the C60 molecule 共lower curve兲, CuPc mol- C. Enkvist, S. Lunell, B. Sjögren, S. Svensson, P. A. Brühwiler, A.
ecule 共middle curve兲, and CuPc: C60 system 共upper curve兲. The vertical bars Nilsson, A. J. Maxwell, and N. Mårtensson, Phys. Rev. B 48, 14629
show the shift of the HOMO peaks. 共1993兲.

J. Vac. Sci. Technol. A, Vol. 24, No. 4, Jul/Aug 2006

 1675 Lozzi et al.: CuPc: C60 blend film: A photoemission investigation 1675

20 22
L. Lozzi, S. Picozzi, S. Santucci, C. Canalini, and B. Delly, J. Electron L. Forrò and L. Mihaly, Rep. Prog. Phys. 64, 649 共2001兲.
Spectrosc. Relat. Phenom. 137–140, 101 共2004兲. 23
I. M. L. Billas et al., J. Chem. Phys. 111, 6787 共1999兲.
21
V. de Coulon, J. L. Mrtins, and F. Reuse, Phys. Rev. B 45, 13671 共1992兲 24
I. G. Hill, A. Kahn, Z. G. Soos, and R. A. Pascal, Jr., Chem. Phys. Lett.
and reference therein. 327, 181 共2000兲.

JVST A - Vacuum, Surfaces, and Films

View publication stats