Surface Science 539 (2003) 37–48

www.elsevier.com/locate/susc

Electronic properties and interface characterization

of phthalocyanine and Ru-polypyridine dyes on TiO2 surface
Guangming Liu *,1, A. Klein, A. Thissen, W. Jaegermann
Department of Material Science, Darmstadt University of Technology, Petersenstrasse 23, D-64287 Darmstadt, Germany
Received 11 October 2002; accepted for publication 12 May 2003

Abstract
Electronic structures, energy levels alignment and electronic interactions of a series of phthalocyanine and Ru-
polypyridine dyes at the interfaces of dyes/TiO2 films have been studied by photoelectron spectroscopy. The shifts of
carboxylic C 1s core level binding energies of carboxylated dye complexes at different adsorption states indicate that the
dyes are mainly adsorbed on the nanoporous TiO2 surface via carboxylic groups. The relative positions of the highest
occupied molecular orbitals/the lowest unoccupied molecular orbitals (HOMO/LUMO) of these phthalocyanine and
Ru-polypyridine dyes with respective to the conduction band (CB) of TiO2 were determined from a combination of
X-ray and ultraviolet photoemission (XPS and UPS) measurements and further compared with their corresponding
electrochemical oxidation potentials. Both measurements support the general picture of Ru-polypyridine complexes
that their HOMO has predominately core metal Ru 4d(t2g ) character while for phthalocyanine dyes the HOMO energy
levels are more related to their chromophores.
Ó 2003 Elsevier B.V. All rights reserved.

Keywords: X-ray photoelectron spectroscopy; Visible and ultraviolet photoelectron spectroscopy; Interface states; Titanium oxide;
Surface energy

1. Introduction mechanism in a dye-sensitized solar cell involves

the transfer of photoexcited electrons from the
Dye-sensitized nanostructured TiO2 semicon- excited states of dye molecules to the conduction
ductor materials are of considerable interest and band of TiO2 [5], and the solar cells performance
intensively investigated nowadays due to their (short circuit photocurrent and open circuit photo-
demonstrated high conversion efficiency [1], low voltage) is affected by the driving forces derived
cost [2] and potential application in photovoltaic from the relative energies at the semiconductor/
devices [3,4]. It is well known that the basic dye/electrolyte interfaces [3,4]. However, much still
remains to be understood of the interface proper-
ties of dye/TiO2 at a molecular scale, such as ad-
*
Corresponding author. Present address: Department of sorption geometry, adsorption sites and electronic
Chemistry, University of California, Riverside, CA 92521, interactions. Chemical bonding formed between
USA. Tel.: +1-909-787-5736; fax: +1-909-787-4713.
E-mail addresses: gmliu@mail.ucr.edu, gmliu@surface. the dye and TiO2 not only anchor the sensitizer in
tu-darmstadt.de (G. Liu). place but may also alter conceivably surface state
1
Tel.: +49-6151-16-6354; fax: +49-6151-16-6308. energetics, and especially the determining factors

0039-6028/03/$ - see front matter Ó 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0039-6028(03)00721-0
38 G. Liu et al. / Surface Science 539 (2003) 37–48

in charge transfer dynamics at the semiconductor investigated some phthalocyanine dyes of zinc(II)
interface. The most important and fundamental phthalocyanine (ZnPc), 2,9,16,23-tetra (carbonyl-
property that should be considered is at which methylaminocarboxyl) phthalocyaninato zinc(II)
energetic levels the LUMO and HOMO of the dye (ZnPcGly), 2,9,16,23-tetra[1-carbonyl-2-(4-hydroxy-
molecule should be maintained to ensure efficient phenyl) ethylaminocarboxyl] phthalocyaninato
photoinduced electron transfer into the TiO2 zinc(II)––(ZnPcTyr), bis(4-carboxypyridine)-phtha-
conduction band upon illumination followed by locyaninato ruthenium(II)––(RuPc) and the
hole transfer to the electrolyte solution (regenera- Ru-polypyridine dye complex of cis-bis (4,40 -
tion of the dye by iodide) [6]. dicarboxy-2,2 0 -bipyridine)-bis-isothiocyanato)-
Photoelectron spectroscopy (PES) provides a ruthenium(II) [Ru(dcbpy)2 (NCS)2 ] (see molecular
method in order to better understand the electronic structures in Table 1), one of the most efficient
structures of interfaces and molecular composi- sensitizers in dye-sensitized thin film solar cells so
tion, adsorption geometries, distribution of energy far, by means of X-ray and UV excited PES (XPS,
levels and orbital symmetries of adsorbates. There UPS).
have been a number of reports on photoelectron
spectroscopy studies for thin films of various
polymers [7], zinctetraphenylporphyrin on semi- 2. Experimental section
conductor surfaces [8,9], modification and char-
acterization of phthalocyanines [10–12]. However, The preparation of nanoporous TiO2 films was
only few attempts have been made to investigate adapted from the literature [1,3]. Transparent
the exact positions of the relative energetics, elec- conductive oxide (TCO) glass plates (F doped
tronic structures and electronic interaction at the SnO2 ; sheet resistance 10–15 X=) were used as
interfaces between the dye sensitizers and semi- the substrate for deposition of TiO2 films. Colloi-
conductor TiO2 , which have important implication dal TiO2 solution (Ti-Nanoxide T) with solid
for the understanding of the charge injection pro- particle size ca. 13 nm was purchased from So-
cess in photovoltaic devices and also for the opti- laronix S.A., Switzerland. The colloidal TiO2 so-
mization of the dye-sensitized solar cells. So it is lution was spread uniformly on the surface of the
becoming one focus of the surface science study. TCO glass by raking off the excess suspension with
The results expected from such UHV studies will a glass rod. To control the thickness of the TiO2
provide a basic understanding on systematic vari- films, the TCO glass plate was covered with ad-
ation in interface properties. As the dye-sensitized hesive tape (Scotch) as a spacer along the length of
solar cells operate in electrolyte environment, the each edge. Nanocrystalline TiO2 (anatase) films
influence of the changed environment also must be were prepared by following heating at 450 °C for
considered when interpreting the results. For this 30 min in air. The thickness of the TiO2 films is 2.6
reason we will compare the results from the UHV lm, measured by a Dektak 3 surface profile mea-
study with measurements of oxidation potentials suring system.
in electrolyte solution. The TiO2 surfaces were coated with the dye
Phthalocyanine dyes have been extensively in- complexes by dipping them for at least 3 h into the
vestigated as sensitizers because of their chemical respective dye dissolved absolute ethanol solutions
stability and high absorption coefficient in the far at certain concentrations, for some phthalocyanine
red/near-IR region. They have shown sufficient dyes 3% of dimethylsulfoxide (DMSO, v/v) was
improvement in energy conversion efficiencies by added. These organic dye complexes with high
adequate synthetic modification [13–16]. Some pht- purity, such as ZnPc and Ru(dcbpy)2 (NCS)2 , were
halocyanines have been widely applied in photo- purchased from Aldrich and Solaronix without
voltaic device, as well as in chemical sensor and further purification. The synthesis and photo- and
electrophotography, also benefiting from such electrochemical properties of the other dye com-
increased light to electrical energy conversion pounds are described in the literature [15,16].
efficiencies [17]. In this work we comparatively Adsorption of the dyes was performed at a
 G. Liu et al. / Surface Science 539 (2003) 37–48 39

Table 1
HOMO/LUMO binding energies and electrochemical oxidation potentials of the dye complexes (vs. vacuum level)

Dyes Molecular structures Eox (eV) Eox (eV) HOMO (eV) LUMO (eV)
ZnPcGly
2,9,16,23-tetra COOH 5.21 [15] 3.39 [15] 5.72 3.90
(carbonylmethyl- HN O
C
aminocarboxyl)
phthalocyaninato
zinc(II) O
N N
N HN
C N Zn N C COOH
HN
N O
HOOC N N

C
O NH
COOH

ZnPcTyr
2,9,16,23-tetra 5.29 [15] 3.47 [15] 5.78 3.96
HO COOH
[1-carbonyl-
HN O
2-(4-hydroxyphenyl) C
ethylaminocarboxyl] OH

phthalocyaninato
zinc(II) N N
O N HN
C N Zn N C COOH
HN
N O
HOOC N N

OH C
O
NH
HO COOH

ZnPc
zinc(II) phthalocyanine 5.41–5.52 3.41–3.52 5.78 3.78
[27,28] [27,28]
N N N

N Zn N
N
N N

RuPc
bis(4-carboxypyridine)- O 5.69 3.75 5.91 3.97
phthalocyaninato OH
ruthenium(II) N N
N
N
N Ru N
N
N N
N
HO

(continued on next page)

40 G. Liu et al. / Surface Science 539 (2003) 37–48

Table 1 (continued)

Dyes Molecular structures Eox (eV) Eox (eV) HOMO (eV) LUMO (eV)
Ru(dcbpy)2 (NCS)2
cis-bis(4,40 -dicarboxy-2,20 - COOH COOH 5.7 [3] 3.4 [3] 5.91 3.61
bipyridine)-bis-
(isothiocyanato)-
ruthenium(II) N N

HOOC N Ru N COOH

NCS NCS

substrate temperature of ca. 80 °C immediately is available for in situ cleaning of the substrate
after the high temperature annealing of the TiO2 surface. The XP spectra of the samples were
film in order to avoid hydration of the nanoporous measured using monochromatized AlKa radia-
TiO2 films from ambient air. After completion of tion. The binding energy scale was calibrated
dye adsorption, the sample was withdrawn from against Cu 2p3=2 (932.67 eV), Ag 3d5=2 (368.26 eV)
the solution, rinsed with dry ethanol several times, and Au 4f7=2 (84.00 eV). The uncertainty in deter-
and dried with a stream of dry N2 of high purity. mining these binding energies was 0.05 eV. He(I)
This procedure is expected to yield monolayer (hm ¼ 21:2 eV) and He(II) (hm ¼ 40:8 eV) discharge
coverage of the dyes on the TiO2 film surface [18]. lamp light are used as excitation sources for UPS
In order to measure compounds without any ad- measurements. For UP spectra the sample was
sorption, these pure dye powders were pressed negatively biased at a voltage of 1.55 V to shift the
directly onto the soft surfaces of metal indium (In) spectra from the spectrometer threshold. The
substrates. This procedure was found to give high spectra were calibrated vs. the Fermi edge of poly-
quality spectra without surface charging effects. crystalline Au. All the given XPS core level spectra
Then the sample was immediately transferred into and HOMO maximum positions were determined
the UHV chamber for analysis. by a fitting procedure using Igor software.
The electrochemical redox potential of the dye
RuPc was measured in a three electrode cell using
Pt wire as a working electrode and a counter 3. Results and discussion
electrode, separately, the reference electrode was
Ag/AgCl (0.2368 V vs. NHE). The electrolyte was Fig. 1 shows XP spectra of the uncovered
0.1 M LiClO4 in water free acetonitrile/dimethyl nanostructured TiO2 substrate film deposited on
sulfoxide (DMSO) (7/1, v/v) media. Cyclic vol- transparent conductive oxide (TCO) glass (spec-
tammograms were obtained on the stationary Pt trum a) and the TiO2 film covered with various dye
electrode in the electrolyte solution containing dye sensitizers of ZnPcTyr, RuPc and Ru(dcbpy)2 -
RuPc (ca. 0.1 mM) using a potentiostat of Thales (NCS)2 (spectra b–d). Spectrum a indicates that
controlled by a personal computer for data ac- the TiO2 film is mainly composed of Ti and O,
quisition. with only very little amount of C 1s from surface
The PES experiments were carried out in an contaminations. The sensitivity corrected intensity
integrated ultra high vacuum (UHV) system ratio of Ti and O is 1:2 indicating the stoichio-
equipped with multi-technique surface analysis metry and oxidation state of the prepared TiO2
system (Physical Electronics 5700) photoelectron film. The electronic structures of such TiO2 films
spectrometer (XPS, UPS) which allows in situ were discussed in detail in our very recent work
characterization of the prepared films and surfaces [19]. After adsorption of different type of dyes, the
at a base pressure of 1  1010 Torr. An Ar ion gun substrate TiO2 signals decreased in intensity, with
 G. Liu et al. / Surface Science 539 (2003) 37–48 41

(spectra d). Beside some additional emission in-

tensity due to the O containing dyes in the O 1s
species there is no change in the spectral shape
(FWHM (Ti 2p3=2 ) ¼ 0.9 eV, FWHM (O 1s) ¼ 1.1
eV). It is very interesting that all the substrate
binding energies of Ti 2p and O 1s shifted to
smaller values following the adsorption of different
dye complexes although for some dyes the shifts
are rather small. Moreover the binding energy
shifts of the substrate varied with the different
molecular structures of the adsorbed dyes. For the
adsorption of the dye Ru(dcbpy)2 (NCS)2 on TiO2
surface, the binding energies of Ti 2p (459.37 eV)
and O 1s (530.58 eV) are only very little shifted ca.
0.05 eV to lower binding energies (from spectra a
to b). Similar results can also be found for the
Fig. 1. X-ray photoelectron spectra of (a) nanostructured TiO2 adsorption of the other dyes as RuPc (spectra c).
film on TCO glass; (b) TiO2 /ZnPcTyr; (c) TiO2 /RuPc; (d) TiO2 /
Ru(dcbpy)2 (NCS)2 .
Moreover the binding energies of TiO2 further
shifted to 459.30 eV (Ti 2p) and 530.48 eV (O 1s)
with the adsorption of dye ZnPcTyr (spectra d).
some new peaks appeared from the different dye The largest shift ca. 0.1 eV was found for the ad-
overlayers (C 1s, N 1s, Zn 2p and Ru 3d). sorption of the dye ZnPcGly on the nanoporous
Fig. 2 shows Ti 2p and O 1s core level spectra of TiO2 surface, as reported in our earlier work [19].
the substrate TiO2 before (spectra a) and after ad- These small shifts may be ascribed to charge re-
sorption of various dye complexes of Ru(dcbpy)2 - distribution developed in the interface of semi-
(NCS)2 (spectra b), RuPc (spectra c), ZnPcTyr conductor TiO2 /dye complexes arising from the

Fig. 2. X-ray photoelectron spectra in the Ti 2p region (a) and O 1s region (b). Spectra a of uncovered TiO2 film; spectra b of TiO2 /
Ru(dcbpy)2 (NCS)2 ; spectra c of TiO2 /RuPc; spectra d of TiO2 /ZnPcTyr.
42 G. Liu et al. / Surface Science 539 (2003) 37–48

charge transfer or chemical interaction between structures see also Table 1) with different substi-
the substrate and the dyes adsorbed on the surface. tuents, respectively. The observed differences in the
These chemical bonding formed have also effects spectra for the various dyes may be interpreted in
on the anchoring groups––carboxylic C 1s core terms of the different molecular complex struc-
level binding energies of the adsorbed dye com- tures. The C 1s core levels of the dyes (Fig. 3a)
plexes (see discussion below). reflect the different chemical states of carbon
Fig. 3 shows the C 1s, N 1s, Zn 2p and Ru 3d atoms in the dye molecules. The lower binding
core level spectra of four phthalocyanine dyes of energy emission of C 1s (I) originating from aro-
ZnPc, ZnPcGly, ZnPcTyr and RuPc (molecular matic carbons appears at a value of 285.18 eV for

Fig. 3. XPS core level spectra of the phthalocyanine dyes of ZnPc, ZnPcGly, ZnPcTyr and RuPc, respectively. (a) C 1s emission;
(b) N 1s emission; (c) Zn 2p emission; (d) Ru 3d emission. All the dyes were adsorbed on the TiO2 surface except the dye ZnPc on
In surface.
 G. Liu et al. / Surface Science 539 (2003) 37–48 43

the dyes ZnPcGly and ZnPcTyr and at a higher the aromatic N 1s (V) is shifted to higher binding
value of 285.40 eV for the dyes ZnPc and RuPc, as energy of 399.43 eV, which is caused by the
the former dyes have more substituents connected changing of core metals from Zn2þ to Ru2þ . The
with the aromatic rings in the dye molecules than binding energy of the Ru 3d5=2 line in RuPc is
the latter ones. The C 1s associated with carbon 281.45 eV as shown in Fig. 3(d). The N 1s of
atoms bonded to nitrogen are not resolved from substituents in the dyes of ZnPcGly and ZnPcTyr
the main C 1s emission line for the dyes of ZnPc (VI) show the same value of binding energy at
and ZnPcTyr tailing towards higher binding en- 400.58 eV due to the similar surroundings. As also
ergies. While such emission (II) can be obviously expected, the N 1s emission from the isonicotinic
distinguished at 286.43 eV for the dye ZnPcGly ligands of the dye RuPc (VII) at higher binding
and at 287.16 eV for the dye RuPc. The binding energy has a value of 400.53 eV, which is very close
energies of the carboxylic C 1s (III) are ca. 289.03 to the value (400.48 eV) of bi-isonicotinic N 1s of
eV for ZnPcTyr and RuPc and 288.61 eV for the dye Ru(dcbpy)2 (NCS)2 as discussed below.
ZnPcGly. This difference in the carboxylic C 1s The adsorption of dyes on nanoporous TiO2
binding energies for the different dye complexes semiconductor film may involve a variety of dif-
adsorbed on the TiO2 surface can be explained by ferent adsorption geometry and bonding struc-
the different amount of carboxylic groups in the tures due to the strong electronic interaction
different dye molecules bonding with TiO2 , because between the dye molecules and TiO2 , which is re-
the bonding carboxylic C 1s core level spectra shift flected in the core level spectra of ligands and an-
to lower binding energy than the non-bonding choring groups of the dyes. Fig. 4(a–c) shows the
ones (detailed discussion can be found below). The C 1s, N 1s and Zn 2p core level spectra of the dye
carboxylic bonding number is smaller for the dyes ZnPcGly pressed onto the In surfaces and ad-
ZnPcTyr and RuPc than for ZnPcGly, probably sorbed on the nanoporous TiO2 films, respectively.
arising from the steric effect related to their mo- The carboxylic C 1s emission of the dye shifted
lecular structures. The RuPc molecule can only be from 289.22 eV at the powder state on In surface
anchored to the nanocrystalline TiO2 films through to 288.61 eV at the adsorption state on TiO2 sur-
one of the axial pyridine 4-carboxylic acid ligands, face in Fig. 4(a). On the contrary, there is nearly
as evidenced by another ruthenium phthalocya- no shift for aromatic C 1s at 285.18 eV in Fig. 4(a)
nine type complex with a similar molecular struc- and for the Zn 2p emission at binding energy of
ture [20]. It is also conceivable that the presence of 1022.23 eV in Fig. 4(c). A very small shift (ca. 0.08
tyrosine groups in ZnPcTyr reduces the formation eV) to higher binding energy of 399.13 eV is found
of TiO2 –COOH bonding by comparison with the for the N 1s emission at the adsorption state in Fig.
adsorption mode of ZnPcGly on the TiO2 surface 4(b). Similar trend can also be found in the C 1s,
because of such steric effects. Ru 3d, N 1s and S 2p core level spectra of the dye
The binding energy of the core metal Zn 2p Ru(dcbpy)2 (NCS)2 in Fig. 4(d–f). These binding
emission of ZnPc, ZnPcGly and ZnPcTyr shown energy shifts of the core levels for the same dye
in Fig. 3(c) is 1022.23 eV for all of them. The ab- complex at different adsorption conditions obvi-
sence of shift in this binding energy indicates that ously indicate the bonding patterns of the complex
the same oxidation state (+2) and similar ligands to the TiO2 surface and support the general picture
coordinated to the metal center for all complexes. of adsorption involving primarily the carboxyl
Correspondingly, the aromatic N 1s coordinated groups although some influence from other groups,
to the metal center (Zn) would be expected to show such as the NCS ligand, can not be completely
similar trends for the three dyes of ZnPc, ZnPcGly excluded, while they do not contribute significantly
and ZnPcTyr. This is indeed the case as shown in to the chemical bonding to TiO2 [21,22]. In Fig.
Fig. 3(b), where spectrum of ZnPc, ZnPcGly and 4(d) the spectrum of Ru(dcbpy)2 (NCS)2 on In
ZnPcTyr exhibit the same low binding energies of shows two C 1s emission lines originating from
N 1s (IV) at 399.15 eV arising from aromatic ni- carboxylic carbons at 289.57 eV and pyridine car-
trogen atoms. While for spectrum of the dye RuPc bons at 285.58 eV, including a Ru 3d5=2 emission at
44 G. Liu et al. / Surface Science 539 (2003) 37–48

Fig. 4. XPS core level spectra of the dye ZnPcGly (a–c) and the dye Ru(dcbpy)2 (NCS)2 (d–f), pressed on metal In surface and
adsorbed on the TiO2 surface, respectively.

281.48 eV. The major C 1s emission from pyridine tensity of the complex adsorbed on TiO2 surface is
is obviously asymmetric due to some contributions too low to be recorded.
from C 1s of thiocyanate carbons and from the The position of the HOMO energy level of dye
superposed Ru 3d3=2 line (spin orbit splitting for sensitizers plays a very important role in photo-
Ru 3d: 4.2 eV). In Fig. 4(e) the spectrum on In induced charge transfer between the dye and the
surface shows two N 1s peaks, where the N 1s peak semiconductor TiO2 . The injection rate of the
associated with the NCS ligand has significantly electron into the conduction band will depend on
lower binding energy (398.48 eV) with respect to the energy level matching of the excited state
the pyridine N 1s peak (400.48 eV). The S 2p peak (LUMO) of the dye molecule with the conduction
associated with the NCS ligand of the complex band state of the semiconductor TiO2 . Fig. 5(a)
powder on In surface is shown in Fig. 4(f) with a shows the UP spectra (He I) of the dye RuPc ad-
binding energy of 162.63 eV, while the S 2p in- sorbed on the TiO2 surface (spectrum d). From the
 G. Liu et al. / Surface Science 539 (2003) 37–48 45

spectra (He II) measured for the same samples in

Fig. 5b, 1.60 eV for ZnPcGly (a), 1.65 eV for
ZnPcTyr and ZnPc (b,c), 1.78 eV for RuPc and
Ru(dcbpy)2 (NCS)2 (d,e), as shown in the insert.
The HOMO binding energy of Ru(dcbpy)2 (NCS)2
is also in accordance with the reported value ob-
tained using synchrotron radiation [12]. Spectra d
and e show that the dye complexes containing the
same core metal Ru(II) have the similar HOMO
emission although they have different ligands co-
ordinated with Ru(II) in their molecular struc-
tures. Moreover, this can be further verified by
their rather similar electrochemical redox potentials
as shown in Table 1. This result is also consis-
tent with the general picture of Ru(II) complexes
that the HOMO has predominately Ru4d(t2g )
character [23].
It is thus to be expected that the HOMO and
Ru 3d binding energies should show similar trends.
As discussed above this is indeed the case. The
binding energy of Ru 3d (281.45 eV) of RuPc in
Fig. 3(d) is very close to the value of 281.48 eV
of Ru(dcbpy)2 (NCS)2 in Fig. 4(d). However we
should note here that the negative ligands (thio-
cyanate) coordinated to the metal Ru center in
Ru(dcbpy)2 (NCS)2 did not shift its binding energy
toward lower values in comparison with that of
RuPc as expected. On the other hand, for the dyes
of ZnPc, ZnPcGly and ZnPcTyr with the same
core metal Zn, the HOMO positions slightly
change with variation of chromophores even when
the binding energy of Zn 2p is the same value of
Fig. 5. (a) Ultraviolet photoelectron spectra (He I) of the dye 1022.23 eV for the three phthalocyanine dyes as
RuPc on TiO2 surface. Insert: spectra a–e corresponding to
UPS of ZnPcGly/TiO2 , ZnPcTyr/TiO2 , ZnPc/In, RuPc/TiO2 ,
shown in Fig. 3(c). These results indicate that their
Ru(dcbpy)2 (NCS)2 /In in the valence band region. (b) Ultra- redox center is delocalized over the whole mole-
violet photoelectron spectra (He II) of the dye RuPc on TiO2 cular chromophores rather than being localized on
surface. Insert: spectra a–e corresponding to UPS of ZnPcGly/ a specific atom as Ru(II) in the Ru-polypyridine
TiO2 , ZnPcTyr/TiO2 , ZnPc/In, RuPc/TiO2 , Ru(dcbpy)2 (NCS)2 / complexes.
In in the valence band region.
Based on all the results obtained here and our
previous results of TiO2 films (the TiO2 films are
highly n-doped locating the Fermi level very near
insert the energetic positions of the HOMO emis- to the conduction band edge of the semiconductor
sion of the dyes ZnPcGly (spectrum a), ZnPcTyr [19]), we can estimate the relative LUMO positions
(spectrum b), ZnPc (spectrum c), RuPc (spectrum of the dyes with respect to the conduction band of
d), and Ru(dcbpy)2 (NCS)2 (spectrum e) were de- TiO2 by subtracting the band gap values Eopt (es-
termined to be 1.62, 1.68, 1.68, 1.81 and 1.81 eV, timated from the absorption spectra) and charge
respectively. Moreover these values are also con- redistribution values from the HOMO values
sistent with the HOMO values obtained from UP of the dyes.
46 G. Liu et al. / Surface Science 539 (2003) 37–48

Eoffset ¼ EHOMO  Eopt  DðTiO2 Þ values as determined in the photoelectron spectra

(referred to EF ), because the TiO2 work function
Eoffset ¼ 1:81  2:30  ð0:05Þ
was determined to be 4.1 eV, which corresponds to
¼ 0:44 eV for RuðdcbpyÞ2 ðNCSÞ2 the energy distance of EF to the vacuum level [19].
Eoffset ¼ 1:68  1:82  ð0:07Þ To convert to the solid state scale, 4.5 eV is added
to the electrochemical data (0 V NHE
4.5 eV on
¼ 0:07 eV for ZnPcTyr the solid state scale). We should note that the
Eoffset ¼ 1:81  1:94  ð0:05Þ absolute values of the converted electrochemical
data given here are subject to uncertainties due to
¼ 0:08 eV for RuPc
the use of such energy scale although the relative
values are not affected by those uncertainties. The
A detailed discussion on the HOMO/LUMO en- results presented in this paper show the oxida-
ergy levels alignment of the dye ZnPcGly on TiO2 tion potentials are significantly smaller than the
is given elsewhere [19]. In all cases the positions of HOMO binding energies. The variation is between
the LUMO states fit well with the energy require- 0.21–0.51 eV and depends on the dye molecular
ments for efficient electron injection into the con- structures. Two processes that may cause this ap-
duction band after light absorption. However the parent inequality in the dye potentials under dif-
dye Ru(dcbpy)2 (NCS)2 on the surface of semi- ferent working conditions: (1) Contributions of
conductor TiO2 has the largest offset between the solvation effects, such as from dry acetonitrile/
LUMO and the conduction band based on such dimethylsulfoxide (DMSO) (7/1, v/v) containing
energy levels alignment, which makes it fit better 0.1 M LiClO4 or 0.1 M tetrabutylammonium
with the energy requirements (more driving force) hexafluorophosphate (TBAPF6 ) that are normally
than the other dye sensitizers for efficient electron used as electrolyte additives in the CV electro-
injection into conduction band from the excited chemical measurements of dye oxidation poten-
state of the dye. This may be one of the reasons for tials. Such solvation effect on the energy levels
the dye Ru(dcbpy)2 (NCS)2 being one of the most distribution was also manifested by Cahen et al. in
efficient sensitizers in dye-sensitized solar cells up other system [24], such as: the TiO2 anatase energy
to date. However we should keep in mind that the levels ECB will shift 0.55 eV toward the vacuum
reasonable energy alignment at the interfaces of level upon contact with the solvent. (2) Adsorption
the TiO2 /dye/electrolyte is necessary and especially of the dye on the TiO2 surface may also give some
important in the designing and optimization of the contributions to the potentials shift. As reported in
solar cells while it does not control the overall the literature [6,25], the oxidation potentials of
conversion efficiency, which also depends on the Ru-polypyridine types of complexes bound to the
charge recombination and transport, surface ag- nanostructured surface maybe become sensitive to
gregation, etc. In addition, we should also point either the solution pH or the external bias. The
out that these results as LUMO estimated from adsorption induced potential changes were also
band gap values are only first step, further com- found to depend on both the dye structure and the
prehensive investigation on the LUMO positions electrolyte composition. Similar measurements
by inverse PES or NEXAFS is necessary to give indicate that there are only minor changes in the
more accurate energy levels alignment. redox potentials of the similar Ru-polypyridine
Table 1 shows a comparison of the HOMO complexes attached to the TiO2 surface comparing
maximum energy levels (ionization potentials) with that dissolved in free solution [26]. Moreover,
from the dye complexes obtained by PES under some other factors should also be considered, such
UHV conditions with the corresponding electro- as the distinctly different working conditions,
chemical oxidation potentials measured by cyclic HOMO values are obtained from strict UHV while
voltammetry (CV) in electrolyte. To refer the oxidation potentials are from complex solutions
HOMO energy position to the vacuum level a and more related to the external environment, etc.
value of 4.1 eV is added to these binding energy It is evident from the data that the relative differ-
 G. Liu et al. / Surface Science 539 (2003) 37–48 47

ences between the electrochemical and PES data electrochemical chamber connected to a photo-
are larger for the zinc phthalocyanine dyes than electron spectrometer.
for the Ru-polypyridine dyes. We tentatively ex-
plain this effect by the character and extension of
the HOMO state, whereas for Ru-polypyridine
Acknowledgements
dyes the HOMO state is centered on the transition
metal the HOMO state is delocalized on the whole
We acknowledge financial support of this work
chromophore for zinc phthalocyanine dyes. In the
by the Alexander von Humboldt Foundation in
former case any solvation induced stabilization
Germany. We gratefully thank the provision of
energy of the hole in the HOMO state should be
some of TiO2 electrodes and phthalocyanine
less effective than for the latter case and as a
compounds from Dr. J. He in Lund University
consequence smaller energy shifts are expected.
kermark, Dr. L. Sun in
and from Prof. B. A
Stockholm University of Sweden.

4. Conclusions
References
In conclusion, the electronic structures and
molecular interactions of a series of phthalocya- [1] B. OÕRegan, M. Gr€atzel, Nature 353 (1991) 737.
nine and Ru-polypyridine dye complexes at the [2] G. Smestad, C. Bignozzi, R. Argazzi, Sol. Energy Mater.
Sol. Cells 32 (1994) 259.
interface of nanoporous TiO2 films were compar- [3] A. Hagfeldt, M. Gr€atzel, Chem. Rev. 95 (1995) 49.
atively investigated by photoelectron spectro- [4] S.Y. Huang, G. Schlichth€ orl, A.J. Nozik, M. Gr€atzel, A.J.
scopy. The relative energy levels of HOMO/LUMO Frank, J. Phys. Chem. B 101 (1997) 2576.
and the offsets between the LUMO and the con- [5] G. Liu, X. Li, J. Zhao, H. Hidaka, N. Serpone, Environ.
Sci. Technol. 34 (2000) 3982.
duction band at the semiconductor/organic dyes
[6] A. Zaban, S. Ferrere, B.A. Gregg, J. Phys. Chem. 102
interfaces (ZnPcTyr/TiO2 : 0.07 eV; RuPc/TiO2 : (1998) 452.
0.08 eV; Ru(dcbpy)2 (NCS)2 /TiO2 : 0.44 eV) were [7] P. Dannetun, M. Fahlman, C. Fauquet, K. Kaerijima, Y.
estimated from a combination of XPS and UPS Sonoda, R. Lazzaroni, J.L. Bredas, W.R. Salaneck, Synth.
measurements. The adsorption induced C 1s core Met. 67 (1994) 133.
level binding energy shifts reveal that the carb- [8] S. Narioka, H. Ishii, D. Yoshimura, M. Sei, Y. Ouchi, K.
Seki, S. Hasegawa, T. Miyazaki, Y. Harima, K. Yamash-
oxylated dye complexes mainly attach to the TiO2 ita, Appl. Phys. Lett. 67 (1995) 1899.
surface via carboxyl groups. The measurements [9] H. Ishii, S. Narioka, D. Yoshimura, M. Sei, Y. Ouchi,
support the general picture that for Ru-polypyri- S. Hasegawa, T. Miyazaki, Y. Harima, K. Yamashita, K.
dine complexes the HOMO has predominately Ru Seki, J. Electron. Spectrosc. Relat. Phenom. 76 (1995) 559.
[10] D. Schmeisser, J. Pohmer, M. Hanack, W. Goepel, Synth.
4d(t2g ) character due to the redox center being
Met. 61 (1993) 115.
localized on core metal Ru(II) in the molecular [11] M. Suzuki, O. Bartels, R. Gerdes, G. Schneider, D.
structures rather than delocalized over the whole W€ ohrle, G. Schulz-Ekloff, M. Kimura, K. Hanabusa,
molecular chromophores as for phthalocyanine H. Shirai, Chem. Lett. (1999) 579.
Zn(II) dyes. An apparent inequality between the [12] H. Rensmo, K. Westermark, S. S€ odergren, O. Kohle, P.
HOMO energies (ionization energies) and the first Persson, S. Lunell, H. Siegbahn, J. Chem. Phys. 111 (1999)
2744.
oxidation potentials of the dye complexes is as- [13] J. He, A. Hagfeldt, S.E. Lindquist, H. Grennberg, F.
cribed to solvation effects. Extension of these Korodi, L. Sun, B. A
kermark, Langmuir 17 (2001) 2743.
studies is underway to a broad range of promising [14] Y. Shen, L. Wang, Z. Lu, Y. Wei, Q. Zhou, H. Mao,
dye sensitizers with strong energy matching H. Xu, Thin Solid Films 257 (1995) 144.
(HOMO/LUMO) and electronic coupling with the [15] J. He, G. Benk€ o, F. Korodi, T. Polıvka, R. Lomoth,
B. A
kermark, L. Sun, A. Hagfeldt, V. Sundstr€ om, J. Am.
conduction band of the semiconductor and also to Chem. Soc. 124 (2002) 4922.
the interfaces of semiconductor/dye/electrolyte by [16] M. Yanagisawa, F. Korodi, B. A
kermark, L. Sun, J. He,
in situ UHV experiments in our specially designed V. Sundstr€ om, J. Pophyr. Phtalocya. 6 (2002) 217.
48 G. Liu et al. / Surface Science 539 (2003) 37–48

[17] R.F. Ziolo, C.H. Griffiths, J.M. Troup, J. Chem. Soc. [23] H. Rensmo, S. S€ odergren, L. Patthey, K. Westermark,
Dalton Trans. (1980) 2300. L. Vayssieres, O. Kohle, P.A. Brhwiler, A. Hagfeld, H.
[18] V. Shklover, M.K. Nazeeruddin, S.M. Zakeeruddin, C. Siegbahn, Chem. Phys. Lett. 274 (1997) 51.
Barbe, A. Kay, T. Haibach, W. Steuer, R. Hermann, H.U. [24] D. Cahen, G. Hodes, M. Gr€atzel, J.F. Guillemoles, I.
Nissen, M. Gr€atzel, Chem. Mater. 9 (1997) 430. Riess, J. Phys. Chem. B 104 (2000) 2053.
[19] G. Liu, W. Jaegermann, J. He, V. Sundstr€ om, L. Sun, [25] A. Zaban, S. Ferrere, J. Sprague, B.A. Gregg, J. Phys.
J. Phys. Chem. B 106 (2002) 5814. Chem. 101 (1997) 55.
[20] M.K. Nazeeruddin, R. Humphry-Baker, M. Gr€atzel, B.A. [26] T.A. Heimer, S.T. DÕArcangelis, F. Farzad, J.M. Stipkala,
Murrer, Chem. Commun. (1998) 719. G.J. Meyer, Inorg. Chem. 35 (1996) 5319.
[21] H. Onishi, Y. Iwasawa, Chem. Phys. Lett. 226 (1994) 111. [27] R. Schlaf, B.A. Parkinson, P.A. Lee, K.W. Nebesny, N.R.
[22] L. Patthey, H. Rensmo, K. Westermark, L. Vayssieres, Armstrong, J. Phys. Chem. B 103 (1999) 2984.
P.A. Br€uhwiler, H. Siegbahn, N. M
artensson, P. Persson, [28] A.P.B. Lever, E.R. Milaeva, G. Speier, in: A.P.B. Lever,

. Petersson, S. Lunell, J. Chem. Phys. 110
A. Stashans, A C.C. Leznoff (Eds.), Phthalocyanines: Properties and
(1999) 5913. Applications, VCH Publications, New York, 1991, p. 9.