Photoelectron spectroscopy on thin films of extended zinc porphyrazines

Article abstract of DOI:10.1021/jp046576i

Photoemission measurements were performed on a series of stepwise benzoannelated zinc porphyrazine molecules in thin films. The electronic structure of tert-butyl-substituted zinc tetraazaporphyrin, phthalocyanine, and naphthalocyanine is investigated using mainly EUV synchrotron radiation. A detailed analysis of the zinc satellites in the spectra of the valence region is performed in an attempt to infer the effect of ligand size extension on the metal-ligand interactions. No differences in the character of the bond between zinc and ligand were detected as a function of ligand size. The results are compared with those for the respective metal-free and copper-containing molecules.

Full text of DOI:10.1021/jp046576i

7826
J. Phys. Chem. B 2005, 109, 7826-7833
Photoelectron Spectroscopy on Thin Films of Extended Zinc Porphyrazines
D. Pop,* B. Winter, W. Freyer, W. Widdra,† and I. V. Hertel‡
Max-Born-Institut fu¨r Nichtlineare Optik und Kurzzeitspektroskopie, Max-Born-Strasse 2A,
D-12489 Berlin, Germany
ReceiVed: July 30, 2004; In Final Form: February 25, 2005
Photoemission measurements were performed on a series of stepwise benzoannelated zinc porphyrazine
molecules in thin films. The electronic structure of tert-butyl-substituted zinc tetraazaporphyrin, phthalocyanine,
and naphthalocyanine is investigated using mainly EUV synchrotron radiation. A detailed analysis of the
zinc satellites in the spectra of the valence region is performed in an attempt to infer the effect of ligand size
extension on the metal-ligand interactions. No differences in the character of the bond between zinc and
ligand were detected as a function of ligand size. The results are compared with those for the respective
metal-free and copper-containing molecules.
1. Introduction
studies11,12,18,21-29 were carried out for clarifying the electronic
structure of zinc porphyrazines, photoemission investigations
on these compounds are scarce. Moreover, to the best of our
knowledge, no photoemission spectra of the valence region have
been previously reported for zinc tetraazaporphyrin or for the
respective tert-butyl-substituted compound.
Due to the completely filled Zn 3d orbitals, the photoemission
spectra of zinc porphyrazines exhibit satellite features that are
of different nature (shakeup) as compared to those of similar
copper complexes (charge transfer). In addition to changes in
the electronic structure, in this paper we study the intensity of
the zinc shakeup satellites as a function of ligand size, in an
attempt to infer the effect of benzoannelation on the Zn-ligand
interactions.
Porphyrazine molecules, including tetraazaporphyrins (TAPs),
phthalocyanines (Pc’s), and naphthalocyanines (Nc’s), have long
been known as dyes, and more recently they have been
investigated for use in gas sensors, computer read/write disks,
electrochromic devices, liquid crystals, photovoltaic and solar
cells, sensitizers for photodynamic therapy, photocopying ap-
plications, nonlinear optics, and molecular electronics.1-8 These
molecules are also attractive due to their similarity to the natural
porphyrins. Porphyrazines contain a central C8N8 ring sur-
rounded by four pyrrole groups; hence they differ from
porphyrins only by the presence of four meso-nitrogen atoms
instead of four methine groups in the central ring. The high
chemical versatility of the porphyrazine macrocycle allows for
tailoring the electronic structure/properties of the molecule
through changes of the ligand and/or the central atom(s).
Because these molecules have great potential for industrial
applications and in fundamental research, a detailed understand-
ing of their electronic structures and intramolecular charge
transfer processes is essential. We have recently used photo-
electron spectroscopy to investigate the electronic structures of
a series of metal-free9 and copper10 porphyrazines with stepwise
increased ligand size: tert-butyl-substituted tetraazaporphyrin,
phthalocyanine, and naphthalocyanine. The system of copper-
containing porphyrazines10 allowed us to also study the interac-
tions between the ligand and a metal with a 3d9 configuration
in the molecular ground state. Changes in the character of the
Cu-ligand bond as a function of the degree of benzoannelation
were thus inferred.10 In the present work we contrast the
previous results with data for Zn tert-butyl-substituted tetraaza-
porphyrin, phthalocyanine, and naphthalocyanine in thin films.
The Zn atom has a full 3d shell in the ground state of these
molecules. Therefore, the absorption spectra of zinc porphyr-
azines show no ligand-to-metal and metal-to-ligand charge
transfer bands, and these compounds are ideal for studying the
electronic states of the reduced symmetry (D4h) ring.11,12 Even
though a large amount of experimental11,13-20 and theoretical
2. Experimental Section
2.1. Materials. The zinc porphyrazine molecules studied in
this work are sketched in Figure 1. The chemical names of the
molecules, from left to right in the figure, are zinc tetra(tert-
butyl)porphyrazine (or zinc tetra(tert-butyl)tetraazaporphyrin),
zinc tetra(tert-butylbenzo)porphyrazine (or zinc tetra(tert-butyl)-
phthalocyanine), and zinc tetra(tert-butylnaphtho)porphyrazine
(or zinc tetra(tert-butyl)naphthalocyanine), respectively. For
convenience, we denote these molecules as ZnP0, ZnP1, and
ZnP2. “P” symbolizes the porphyrazine unit, and the numbers
“0”, “1”, and “2” represent the number of benzo units fused to
each pyrrole group. Accordingly, the metal-free compounds are
denoted H2P0, H2P1, and H2P2, respectively. Notice that for
no substituent (symbolized by R), the molecules in Figure 1
correspond to zinc tetraazaporphyrin (or porphyrazine) (ZnTAP),
zinc phthalocyanine (ZnPc), and zinc naphthalocyanine (ZnNc).
ZnP0, ZnP1, and ZnP2 were synthesized from their corre-
sponding metal-free compounds,30 in the presence of zinc acetate
in ethanol or pentanol. ZnPc and H2Pc were prepared and
purified according to the procedure given in ref 31. tert-Butyl
substitution eases the synthesis of materials and improves their
solubility; its effect on the optical absorption spectra is, however,
insignificant. On the other hand, the increase in the number of
benzo units causes a pronounced shift of the Q-band toward
longer wavelengths, which is ascribed to the stronger delocal-
ization of the molecular π-electron system.32 The center of the
* Corresponding author.
† Present address: Martin-Luther-Universita¨t Halle-Wittenberg, D-06099
Halle, Germany.
‡ Also at Physics Department, Freie Universita¨t Berlin.
10.1021/jp046576i CCC: $30.25 © 2005 American Chemical Society
Published on Web 04/05/2005

Thin Films of Extended Zinc Porphyrazines
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7827
Figure 1. Illustration of the zinc porphyrazine molecules investigated in the present work. R denotes a substituent tert-butyl group. The different
characters of the nitrogen atoms in the molecules are marked for the ZnP0 molecule: Np denotes a pyrrole nitrogen atom and Nm denotes a
meso-nitrogen atom. The center of the Q-band for each compound dissolved in dioxane is given by λmax ((2 nm uncertainty).
Q-band for each compound dissolved in dioxane is indicated
in Figure 1. For ZnTAP (ZnPz),25,27 ZnPc,11,12,25,26 and ZnNc11
the Q-band is primarily attributed to the transition from the a1u
HOMO orbital to the eg LUMO. The assignment of these
absorption bands is similar for the tert-butyl-substituted mol-
ecules.33
more detail elsewhere.34 It provided photon energies between
20 and 180 eV. For 64 eV photons the energy resolution was
about E/dE ) 6000, with a photon flux of ca. 4 × 1012 s-1 at
100 mA ring current. The analysis chamber (base pressure better
than 2 × 10-10 mbar) is equipped with a hemispherical electron
energy analyzer, Omicron EA 125. For the present experiments,
photoelectrons were detected in normal emission geometry with
grazing light incidence (83° with respect to the surface normal)
and polarization of the synchrotron light at 80° with respect to
the surface normal.
The binding energies reported in the paper were determined
with respect to the vacuum level for thin films that also showed
the substrate Au 4f levels. For this kind of films the charging
was considered negligible. From spectra measured with a
negative bias applied to the sample, the low energy cutoff was
determined. Based on this, the binding energies were determined
via BE ) hν - (Ekin - Ecutoff), where hν, Ekin, and Ecutoff
represent the excitation energy, the kinetic energy, and the low
energy cutoff of the spectrum. For the thicker films, to
compensate for charging, the binding energy scales were aligned
to those obtained for the corresponding thin films.
2.2. Thin Film Preparation and Photoemission Measure-
ments. Thin films of ZnP0, ZnP1, ZnP2, and ZnPc, as well as
those of H2P0, H2P1, and H2P2, were deposited by sublimation
on an Au(111) single crystal in ultrahigh vacuum (2 × 10-10
mbar base pressure). The materials were evaporated from open
quartz crucibles mounted at ca. 5 cm distance from the 10 mm
diameter gold substrate. During deposition the crystal was kept
at room temperature. Prior to deposition the gold substrate was
cleaned by repeated cycles of Ar+ sputtering and annealing to
750 °C. The gold Au(111) single crystal was mounted on a
manipulator which allowed for sample rotation (azimuth, tilt,
polar) and translation in front of the energy analyzer used for
photoelectron detection. The sublimation temperatures for film
preparation, as measured by thermocouples connected to the
quartz crucibles, were 195, 400, 500, and 390 °C for ZnP0,
ZnP1, ZnP2, and ZnPc, respectively. For H2P0, H2P1, and H2P2
the sublimation temperatures were 180, 350, and 500 °C,
respectively. The existence of only intact molecules in the films
was confirmed by optical absorption measurements (spectra
recorded between 220 and 1100 nm). The absorption spectrum
of a given native material dissolved in dioxane and that obtained
by dissolving the corresponding film, grown under the actual
preparation conditions on a quartz substrate, exhibited no
differences. We have also measured the absorption spectra of
the remaining material (left in the quartz oven after film
preparation) dissolved in dioxane, again observing no effect of
decomposition.
3. Results and Discussion
3.1. Effect of tert-Butyl Substituent on Photoemission
Spectra. Figure 2 compares the photoemission spectra of ZnP1
and unsubstituted ZnPc, both obtained for 98.7 eV excitation
energy. The spectra are normalized to the background. For
clarity they are vertically displaced with respect to each other,
and in addition, to facilitate the comparison, on top of the ZnP1
spectrum we also present the spectrum of ZnPc (dotted). At
first glance the intensities of bands B, E, F, and F′ seem
surprisingly high for ZnPc; they appear larger in the ZnPc
spectrum than in the ZnP1 one, even though the latter compound
contains additional tert-butyl groups. As no change of the
background shape occurs for the two spectra, the relative
intensity of these peaks is attributed to the normalization
procedure. More precisely, this is because the normalization of
the spectra was done at the background, but the background
intensity increases with the tert-butyl addition.
The ZnPc films were typically prepared to be about 400 (
40 Å thick, as determined from the deposition rate monitored
by a quartz crystal microbalance. The ZnP0, ZnP1, and ZnP2
films were made to contain about an identical number of
molecules by scaling the monitored thickness to the respective
molecular mass. Following the film preparation the sample was
transferred under ultrahigh vacuum conditions into the analysis
chamber for measurements.
The direct comparison of ZnPc and ZnP1 photoemission
curves reveals that the tert-butyl addition to the ZnPc macrocycle
leads to a signal increase at the position of band G and in the
9.5-13.5 eV binding energy region. However, considering the
comparison between H2Pc and H2P1 spectra9 and that between
The photoemission experiments were performed at the MBI-
BESSY beamline (undulator U125) at the synchrotron radiation
facility BESSY in Berlin. This beamline has been described in

7828 J. Phys. Chem. B, Vol. 109, No. 16, 2005
Pop et al.
Figure 3. Comparison of photoemission spectra of ZnP1 and H2P1
thin films.
in the region of band G. In the H2P1 spectrum the features C
and D appear with higher intensities than in the ZnP1 spectrum
as a consequence of the normalization procedure. This is because
the normalization is done in a region where the photoemission
signal is strongly overlapped with the background signal, the
latter being larger in the spectra of ZnP1.
Figure 2. Photoemission spectra of ZnP1 and ZnPc thin films at 98.7
eV photon energy. The dotted line represents the ZnPc spectrum plotted
on top of the ZnP1 spectrum. Electron binding energies are given with
respect to the vacuum level. Labels and their assignment are discussed
in the text.
In view of the previously mentioned theoretical results, and
accounting for the fact that ZnP0, ZnP1, and ZnP2 are
chemically very similar to ZnTAP, ZnPc, and ZnNc, respec-
tively, it is concluded that band A in the ZnP1 spectrum in
Figure 3 originates from the HOMO of the molecule. To make
the assignment of features B, C, and E, we use the calculations
for the energies of ZnPc molecular orbitals presented in ref 12.
We compared the difference in energy between the various
molecular orbitals and the HOMO with the difference in energy
between the photoemission peaks and the center of the peak
derived from the HOMO (Figure 3). Thus, we deduce that band
B includes contributions from several in-plane and π molecular
orbitals. As comes out by comparing our results to those from
ref 12, the in-plane molecular orbitals that give a photoemission
signal at the position of band B are 11b1g, which is the HOMO
- 1; 9b2g, which is the HOMO - 2; 19eu; and 13a1g. While the
HOMO - 1 has strong contributions from the pyrrole nitrogen
CuPc and CuP1,10 it results that tert-butyl contributions are also
found in the 13.5-16 eV binding energy range. The latter
contributions are not observed in the ZnP1-ZnPc spectra
comparison due to the significant Zn signal in the respective
region and to the normalization procedure.
Generally, the peak positions are roughly the same in ZnPc
and ZnP1 spectra. In the ZnP1 spectrum, bands C and G are
shifted by about 0.2 eV toward lower binding energies as
compared to the ZnPc case. Peaks F and F′, as well as D and
C, are no longer well resolved for tert-butyl substitution. Good
agreement exists between the above-mentioned changes and the
modifications appearing in the H2Pc9 or CuPc10 spectra upon
tert-butyl substitution. Overall, the differences between the
photoemission spectra of substituted and unsubstituted com-
pounds are mainly the additional contributions from the tert-
butyl groups. No significant changes in the electronic valence
structure of the macrocycle are detected.
2
2
,
lone pairs mixed with 12% contributions from the Zn dx -y
each of the other three molecular orbitals is largely derived from
the meso-nitrogen lone pairs.12 The π molecular orbitals that
have a photoemission signal at the position of band B are 3b2u,
6eg, 5a2u, 5eg, 2b1u, 4a2u, and 1a1u. Among them, the 4a2u
molecular orbital has strong contributions from the pyrrole
nitrogen and meso-nitrogen atoms; the other π orbitals are
derived mainly from benzene and pyrrole carbon. Thus, band
B contains a photoemission signal originating from nitrogen,
benzene, pyrrole carbon, and Zn, though the latter is in a small
amount. We attribute the much higher intensity of band B
relative to band A to the fact that band B contains signals from
several π and localized molecular orbitals, while band A
includes only the emission originating from the HOMO of the
molecule. Compared to the H2P1 case, in the ZnP1 spectrum
band B is found at lower binding energy (by ∼0.3 eV). It is
likely that this shift results from the fact that in ZnP1 the orbital
11b1g is pushed up by antibonding with the Zn dx -y (see ref
12). Concerning the signal from the 4eg orbital mentioned in
ref 12, we interpret that this is found in the border region
between peaks B and C. However, band C includes the signal
from the 18eu and 17eu in-plane orbitals, which each have 26%
pyrrole nitrogen lone pair contributions.12 The emission from
the 10b1g orbital, which is derived from benzene, is also found
at the position of band C.
3.2. Peak Assignment and Metal Features in the Spectra.
Calculations using D4h symmetry for ZnPc (molecule placed in
an xy-plane with the axis passing through the pyrrole nitrogen
atoms) showed that the HOMO is an a1u orbital having exclusive
ligand contributions and π character.25,26,35,36 It has in fact the
same composition (C 2p atomic orbitals) as the HOMO of H2Pc.
In another study, 2% contribution from pyrrole nitrogens,
additional to the C 2p character, has been claimed.12 The HOMO
2
2
- 1 is calculated to be a b1g orbital with relatively low Zn dx -y
contribution (determined as 12% in ref 12 and 20% in ref 26),
but with significant pyrrole nitrogen lone pair (∼66%) and
carbon character.12 A large energy separation (∼1.1 eV12) is
found between the HOMO - 1 and the HOMO. As determined
in ref 12, the HOMO - 2 mainly arises from the meso-bridging
nitrogen atoms with some contributions from pyrrole nitrogen
and carbon. The very next occupied orbitals are calculated to
have mainly mixed contributions from carbon and nitrogen
atomic orbitals. The Zn 3d shell is considered to be unperturbed
for the low energies of the electronic spectrum. Except for the
HOMO - 1, occupied molecular orbitals that include Zn 3d
contributions have energies that are separated by 7.96-8.94 eV
from the HOMO energy.12
2
2
In the case of ZnTAP25 and ZnNc,29 the HOMO is calculated
to derive also from the carbon atomic orbitals. It is an a1u orbital,
well separated from the HOMO - 1.
The photoemission curves of ZnP1 and H2P1 thin films,
measured at 89.6 eV excitation energy, are displayed in Figure
3. The spectra have been normalized to achieve the best overlap
Reference 12 does not present the states found between the
above-mentioned 17eu orbital and the orbitals that have signifi-
cant Zn 3d contributions. However, based on the comparison
between ZnP1 and ZnPc spectra, it is concluded that also tert-
butyl signal contributes to band C. Feature D in the spectrum

Thin Films of Extended Zinc Porphyrazines
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7829
of ZnP1 appears to have a one-to-one correspondence to that
of the analogous metal-free complex, H2P1.9 Thus, it contains
emission from benzene and tert-butyl. As concluded from the
comparison of ZnP1 and H2P1 photoemission spectra and as
also results from ref 12, zinc signal is found at the position of
band E. This signal was assigned to the direct emission from
the Zn 3d level.37 More exactly, molecular orbitals contributing
to band E are 7b1g, 5b2g, 8a1g, 1eg, 6b1g, and 7a1g. These orbitals
include mainly Zn 3d contributions but also some contributions
from benzene and from the carbon and nitrogen atoms found
in the central ring.12 As deduced in section 3.1, this band
includes also tert-butyl signal. Since we find one single band
for the Zn 3d photoemission signal, our results support the
calculations from ref 12 according to which the Zn 3d level
remains almost unperturbed in the low energy part of the
electronic spectrum; the only 3d orbital that interacts to a certain
2
2
extent with the macrocycle is the Zn dx -y . The intensity of
band E relative to that of the other bands in the photoemission
spectrum rises with increasing photon energy. This is a result
of the changes in the cross sections for Zn 3d, C 2s, C 2p, N
2s, and N 2p, with varying photon energy. The cross sections
for all these orbitals decrease with increasing excitation energy,
but for Zn 3d the decrease is slower.38 Thus, the ratio between
the cross sections for Zn 3d and C 2s, C 2p, respectively N 2s,
N 2p, becomes larger with increasing photon energy. Reference
12 does not present calculations for the molecular orbitals that
correspond to larger binding energies. The spectra comparison
in Figure 3 indicates that the Zn signal extends up to the position
of feature F. However, this feature includes also benzene
contributions, as in H2P1 (see ref 9). Because the comparison
of ZnP1 and H2P1 photoemission curves does not indicate the
presence of Zn signal at the position of bands F′, G, H, and I,
these features appearing to have a one-to-one correspondence
in the two spectra, for further peak assignment we use the
calculations done for H2Pc by Orti and Bre´das39 and the results
from our previous study on the analogous metal-free com-
pounds.9 We thus conclude that feature F′ includes signal
originating from the 2s orbitals of the C atoms found in the
central molecular ring, while peak G contains contributions
similar to those from the C 2s 1e1u molecular orbital of benzene39
and tert-butyl.9 Band H contains emission from the same
constituents as in the C 2s 1a1g molecular orbital of benzene,39
while band I includes signal from pyrrole nitrogen.9,39
Figure 4. Photoemission spectra (vertically displaced with respect to
each other) of zinc porphyrazines in the series. The ligand size increases
from bottom to top. Spectra were obtained for 91.7 eV photon energy.
naphthalocyanine in ref 40. There it is stated that the electronic
structure of the macrocycle is dominated by the electronic levels
of the polyacetic units.
On the basis of the results from ref 25, we deduce that in the
ZnP0 spectrum band A also originates from the HOMO of the
molecule. In the same study, calculations for several other
molecular orbitals of the unsubstituted compound ZnTAP are
performed. Comparing the energy of these orbitals relative to
the HOMO with the difference in energy between the ZnP0
photoemission peaks and the peak originating from the HOMO,
we conclude that the calculated orbitals 7b1g, 3a2u, 2b2u, 7b2g,
4eg, 2a2u, and 11eu would give signals at the position of band
B, the orbital 8a1g would give a signal in the border region
between peaks B and C, and the emission from the 3eg orbital
would contribute to feature C. Clearly, in making the assignment
of the ZnP0 photoemission features, the benzene contributions
in the spectra must be disregarded and only the contributions
from the central C-N ring, pyrrole units, zinc, and tert-butyl
should be considered.
However, we noted that apparently some disagreements still
exist in the literature concerning the calculations of the
molecular orbitals and of the atomic contributions to the
molecular orbitals for ZnTAP and ZnPc compounds.
The two features marked by arrows and found at binding
energies larger than 33 eV in the ZnP1 spectrum shown in
Figure 3 are absent in the spectrum of the metal-free compound,
and are barely distinguishable in the spectra of ZnPc and ZnP1
at 98.7 eV (Figure 2). These features are assigned to the Zn
satellite signal,37 and are strongly enhanced for the 3p-4s
3.3. Evolution of the Photoemission Features as a Function
of Ligand Size. In Figure 4 one notices that the peaks containing
signals from the aromatic units and also the background increase
relative to the Zn 3d band with enlarging ligand size.
The position of the Zn 3d main line is identical for all the
Zn compounds studied here, 15.2 ( 0.07 eV binding energy.
Also the Zn satellite peaks exhibit the same binding energy in
ZnP0, ZnP1, and ZnP2 spectra. The changes in the positions of
bands H and G with increasing ligand size are consistent with
those observed for copper10 and metal-free9 compounds. They
have constant binding energy for the benzene-containing
complexes (ZnP1 and ZnP2). For ZnP0, relative to the ZnP1
case, bands H and G shift toward lower binding energies by
∼0.2 and 0.4 eV, respectively. We interpret that the origin of
this shift is the absence of benzene signal in the case of the
ZnP0 molecule.
excitation of the central Zn2+
.
The spectra of ZnP0, ZnP1, and ZnP2 obtained for 91.7 eV
photon energy are plotted in Figure 4. All spectra were
normalized to unity at the maximum of peak E and were then
vertically translated for clarity.
In the case of ZnP2 we deduce from ref 29 that band A is
also derived from the HOMO of the molecule. To the best of
our knowledge, calculations similar to those done for ZnPc in
ref 12 showing the contributions of individual atoms to the
molecular orbitals do not presently exist for ZnNc. However,
comparing the photoemission spectra of ZnP1 and ZnP2, we
conclude that the ZnP2 peaks include contributions similar to
those found for ZnP1 with the difference that the benzene signal
is now replaced by the naphthalene one. This affirmation is
based also on the study done for metal-free phthalocyanine and
A blowup of the edge of the valence region is presented in
Figure 5. As for the case of the metal-free9 and copper10
compounds, a binding energy decrease of feature A (derived

7830 J. Phys. Chem. B, Vol. 109, No. 16, 2005
Pop et al.
Figure 6. Photoemission spectra of a ZnP0 thin film, measured on an
identical surface spot after different exposure times to synchrotron
radiation.
Figure 5. Photoemission spectra of ZnP0, ZnP1, and ZnP2 thin films
in the HOMO region. Labels mark the respective HOMO binding
energies.
as a result of photodegradation. It smears out and broadens,
and eventually for long exposures it appears only as a shoulder
in the spectra. Features B, C, D, F′, and G broaden and smear
out with a lower rate than peak A. The Zn satellite peaks are
not significantly affected by irradiation, as inferred from the
spectra of ZnP0 at 91.7 eV photon energy. The changes of the
ZnP1 spectrum occur for longer irradiation times as compared
to ZnP0 (data not presented here), and manifest themselves by
weaker similar modifications. ZnP2 is the most stable compound
within the series with respect to synchrotron radiation exposure.
Very similar photostability trends were observed for the
respective metal-free9 and copper10 compounds. Since ZnP0 is
the most sensitive compound with respect to synchrotron
radiation exposure, the central porphyrazine ring appears to be
most strongly affected by photodegradation. A possible explana-
tion for these changes is the breaking of bonds in the molecule
as a result of irradiation with EUV photons. Irrespective of the
different photosensitivities, measurements were, however, al-
ways performed by frequently changing the exposed sample
spot.
from the HOMO) is observed for increasing ligand size. The
HOMO binding energies are determined to be 6.63 ( 0.07, 5.97
( 0.07, and 5.69 ( 0.07 eV for ZnP0, ZnP1, and ZnP2,
respectively. Because for these zinc compounds the HOMO has
an atomic orbital composition similar to that of the HOMO for
the corresponding metal-free compounds (see refs 25, 26, and
40 on unsubstituted ZnPc and H2Pc), its destabilization upon
linear benzoannelation must have the same origin in the two
classes of molecules. As shown in ref 40, the HOMO is mainly
localized on the central porphyrazine ring, and the antibonding
interactions with the fused benzo units at linear benzoannelation
cause a decrease of its binding energy. As these interactions
are weaker for the outer benzene units than for the inner ones,
the difference in HOMO binding energy between ZnP2 and
ZnP1 is smaller than the one between ZnP1 and ZnP0. The
difference in HOMO binding energy between any two com-
pounds in the Zn series is nearly the same as for the corre-
sponding metal-free complexes.
3.5. Zinc Satellites. As the photon energy is increased from
86 to 101 eV, one notices in the photoemission spectra a growth
and subsequent decrease of the zinc satellite peaks that are found
at approximately 34.8 and 37.8 eV binding energy. For the
excitation energies used in the present experiments, the maxi-
mum intensity of these peaks occurs for 91.7 eV. The photo-
emission spectra of ZnP0, ZnP1, and ZnP2 films obtained for
91.7 and 87.1 eV excitation energy are displayed in Figure 7.
The inset presents photoemission spectra (smoothed by adjacent
average) of the ZnP0 thin film in the region of the satellite peaks,
for several photon energies. These spectra were normalized to
the signal for 30 eV binding energy and were then vertically
translated for clarity. The inset allows direct tracing of the
evolution of the satellite intensity as a function of photon energy.
For all of the present zinc compounds, the satellite features are
clearly observed only for excitation energies around 91.7 eV.
A blowup of the zinc satellite region for ZnPc and ZnP1
spectra measured at 86.6, 91.7, and 100.2 eV is displayed in
Figure 8. For ZnPc also the 85.6 eV spectrum is shown. One
can note that the ZnPc spectrum exhibits a weak zinc satellite
signal for 85.6, 86.6, and 100.2 eV excitation, while for the
same photon energies satellite features are not distinguishable
for ZnP1. A behavior similar to that of ZnP1 is also found for
ZnP0 and ZnP2. We conclude that in the photoemission spectra
of ZnP0, ZnP1, and ZnP2 for 86.6 and 100.2 eV excitation, the
satellites are masked by the larger background as compared to
the ZnPc case. A reduced background signal (as for ZnPc)
Additionally, band B shifts slightly with the extension of the
ligand, yielding binding energies of 8.2 ( 0.07, 8.1 ( 0.07,
and 8.00 ( 0.07 eV for ZnP0, ZnP1, and ZnP2, respectively.
This is likely to reflect the effect of benzene contributions, which
gradually increase with ligand size.
To further characterize the Zn series of molecules, in addition
to the investigations with extreme ultraviolet (EUV) light, X-ray
photoelectron spectroscopy (XPS) using nonmonochromatized
Al KR radiation (main excitation energy hν ) 1486.7 eV) was
performed. The main focus in the XPS studies was on the
analysis of the N 1s and Zn 2p core level regions, since these
are particularly sensitive to changes in the Zn-N bonding
character. However, within the experimental error bars (∼0.2
eV) the XPS measurements showed no changes in the separation
between C 1s and N 1s or C 1s and Zn 2p1/2,3/2 peaks with
increasing ligand size.
3.4. Photostability. To examine the photostability of these
zinc compounds, photoemission spectra were recorded for
different synchrotron radiation exposure times. Spectral changes
induced by the synchrotron radiation need to be identified to
allow for a meaningful analysis and interpretation of the
measured data.
Figure 6 presents the photoemission spectra of a ZnP0 thin
film obtained from an identical surface spot but for different
synchrotron radiation exposure times. Peak A is one of the
strongly changing features in the ZnP0 photoemission spectrum

Thin Films of Extended Zinc Porphyrazines
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7831
in the molecular ground state. Nevertheless, due to the fact that
the bonding between metal and ligand is not purely ionic, having
also a covalent character, the 4s and 4p orbitals get partially
occupied. Thus, in ref 26 the charge distribution is calculated
to be 3d104s0.554p0.81. However, the next discussion in this paper
will consider just the formal configuration for the central Zn
atom.
ZnP0, ZnP1, and ZnP2 are similar in chemical structure
and properties to ZnPc, which implies that the central zinc is
also divalent in these porphyrazine compounds. The above-
mentioned zinc satellite peaks were previously observed for
ZnPc37 and ZnO.41 They arise from a 3d ionization process with
shakeup, which leads to a formal 3d84s1 final-state configuration
of zinc (two holes in the Zn 3d shell and one electron in the 4s
level):42
3p63d104s0 + hν f 3p6(3d84s1) + e-
The enhancement of the satellite peaks observed for photon
energies around 91.7 eV is a result of the Zn 3p-4s excitation
accompanied by an Auger decay of a 3d electron, with the
emission of another electron from the 3d level.
Actually, for the particular case of ZnPc a similar interpreta-
tion was previously proposed, but treating the central zinc as
quasiatomic.37 As claimed by the authors, the s-p hybridization
would lead to empty 4s states which, in the presence of a core
hole, are localized. Then the existence of two satellite peaks
was attributed to the 3d8 multiplets.37 In contrast, we stress that
the enhancement of the Zn satellite peaks in ZnPc (and ZnP0,
ZnP1, ZnP2) spectra indeed arises from the 3p-4s excitation
accompanied by an Auger decay of a 3d electron, but the holes
in the Zn 4s level exist mainly because the central zinc has a
formal Zn2+ oxidation state.
Figure 7. Photoemission spectra of ZnP0, ZnP1, and ZnP2 thin films
for 91.7 eV (solid lines) and 87.1 eV (dotted lines) photon energy.
The inset shows photoemission spectra (smoothed by adjacent average)
of the ZnP0 thin film in the region of the satellite peaks, obtained for
several photon energies (from bottom to top: 86.6, 87.1, 87.6, 88.1,
88.6, 89.1, 89.6, 90.1, 90.6, 91.1, 91.7, 92.2, 92.7, 93.2, 93.7, 94.2,
94.7, 95.1, 95.7, 96.2, 96.8, 97.2, 97.8, 98.2, 98.7, 99.2, 99.8, 100.2,
101.2, 101.7 eV). Those spectra are normalized to the signal for 30 eV
binding energy and vertically translated.
In a simple model37 this process can be schematically
described as
3p63d104s0 + hν f 3p53d104s1 T
3p63d84s1 + Auger electron
so the obtained final state is again 3d84s1. Thus, an identical
final state of the system is reached by ionization with shakeup
and through the Zn 3p-4s excitation. Consequently, the coherent
superposition of the electron emerging from photoionization with
shakeup and of the Auger electron occurs, and leads to the
enhancement of the satellite peaks.
In addition to the satellite features, above the threshold energy,
M
2,3M4,5M4,5 Auger peaks appear.37 In the present series of
spectra they are observed on the high binding energy side of
the satellites for excitation energies equal to or exceeding 91.7
eV (see Figures 7 and 8). The Auger signal is not clearly
distinguished in the spectra, and therefore these features will
not be further examined here. We notice, though, that the Auger
signal in the vicinity of the satellite peaks complicates the
analysis of the data.
To get information about the satellite intensity dependence
on the excitation energy, we have fitted the respective region
of the photoemission spectra. The fit function was a sum of
two Gaussians of equal widths, 1.07 eV, and a constant
separation of 3 eV between their centers (accounting for the
satellite peaks) plus a linear function (accounting for the
background signal in that region). Both the widths of the satellite
peaks and their energy spacing were thus fixed parameters for
the fit. Their values have been determined from the analysis of
the data measured at excitation energies where the satellite peaks
are well observed but no Auger signal is detected. Figure 9
Figure 8. Photoemission spectra of ZnPc and ZnP1 thin films in the
region of zinc satellite peaks, obtained for 86.6, 91.7, and 100.2 eV
photon energy. For ZnPc also the spectrum measured at 85.6 eV is
shown. The dashed lines in the upper panel serve to guide the eye.
makes the satellites better observable over a larger range of
excitation energies.
It is generally accepted that in ZnPc the zinc atom is
divalent,12,26 which implies a Zn 3p63d104s0 formal configuration

7832 J. Phys. Chem. B, Vol. 109, No. 16, 2005
Pop et al.
Figure 10. Overlapped ZnP2 and H2P2 spectra, both after background
subtraction (bottom), and their difference curve, obtained by subtracting
the H2P2 spectrum from the ZnP2 one (top).
Figure 9. Area of the largest zinc satellite peak normalized to the
background intensity as a function of excitation energy for ZnP0.
shows for ZnP0 the area of the largest zinc satellite peak (Area-
SL) normalized to the background intensity as a function of
excitation energy. Similar dependencies were also obtained for
ZnP1 and ZnP2, and also for the sum of both satellite peak
areas. The curve in Figure 9 exhibits a double peak structure
and can be well fitted by a superposition of two Gaussians of
equal widths, separated by 2.8 eV. This indicates the excitation
of the spin-orbit split components 3p1/2 and 3p3/2 of the Zn 3p
level. Due to the inaccuracies in the fitting procedure, Figure 9
shows overestimated values of the satellite areas for off-resonant
excitations, when practically no satellite peak is observed by
direct examination of the ZnP0, ZnP1, or ZnP2 spectra.
The resonance curves for ZnP0 (Figure 9), ZnP1, and ZnP2
were obtained by normalization of the satellite areas with respect
to the background signal, which has different intensities for the
three compounds. Hence, these curves are not suited for
analyzing the variations in the satellite intensity as a function
of ligand size. This information can, however, be accessed by
comparing the ratios between the area of the satellite signal and
the area of the Zn 3d direct photoemission signal, for the
different molecules in the series.
To determine the area of the Zn 3d contributions, the ligand
signal was extracted from the measured Zn porphyrazine spectra.
First, a background that had a Shirley and a linear component
was subtracted from both the Zn compound and the correspond-
ing metal-free compound spectra. This combination of a linear
plus Shirley-type background was found to account best for the
experimental data. In a next step, for a given excitation energy,
each metal-free compound spectrum (after background subtrac-
tion) was overlapped at the ligand features with the correspond-
ing zinc compound spectrum (also background subtracted).
Subsequently, the metal-free compound spectrum was subtracted
from the one of the zinc compound, and the difference spectrum
was integrated at the position of the Zn 3d main line. Identical
integration ranges were assured for all compounds at all
excitation energies. The procedure is illustrated in Figure 10,
which shows the overlapped ZnP2 and H2P2 spectra (both after
background subtraction) and the difference spectrum obtained
for 94.2 eV photons.
Figure 11. Ratios between the area of the largest zinc satellite peak
and the area of Zn 3d main line for ZnP0, ZnP1, and ZnP2.
the intensity ratios are nearly identical for ZnP0, ZnP1, and
ZnP2, while for higher photon energies larger values are
obtained for ZnP2 than for the other two compounds. We
attribute the enhancement of the ZnP2 ratios to an artifact
generated by the following reason. For photon energies equal
to or exceeding 91.7 eV Auger peaks appear in the spectra. The
Auger signal in the close vicinity to the satellite peaks introduces
additional inaccuracies in the spectral analysis (not included in
the error bars from Figure 11). Consequently, the different values
of the satellite-to-main line intensity ratios for ZnP2 are likely
to result from a less precise estimation of the satellite areas
within the fitting procedure. We thus conclude that, within our
accuracy, the satellite-to-main line intensity ratios in the present
measurements are identical for ZnP0, ZnP1, and ZnP2. This
would suggest the existence of an equal number of free 4s states
for all three compounds.
As previously mentioned, in ZnPc, although the metal atom
has a formal charge of +2, the 4s and 4p orbitals are partially
populated (the charge distribution calculated in ref 26 is
3d104s0.554p0.81). Similar to the case of other metal phthalo-
cyanines, the fact that there is charge on 4s and 4p indicates
that the bonding between metal (Zn) and Pc ligand is not purely
ionic but significantly covalent.26 Therefore, our photoemission
studies provide no indication of any difference in the degree of
covalent (or ionic) character of the Zn-ligand bond when
comparing ZnP0, ZnP1, and ZnP2.
The ratios between the area of the larger Zn satellite peak
(Area-SL) and the area of the Zn 3d main line (Area-Zn3d) as
a function of excitation energy for ZnP0, ZnP1, and ZnP2 are
displayed in Figure 11. The same type of dependencies were
also obtained for the ratios between the summed areas of both
satellite peaks and the area of the Zn 3d main line. Error bars
are only shown for ZnP2 in several points, but they are
calculated to be similar for all three compounds. One can see
in Figure 11 that for a given excitation energy below 91.7 eV
This behavior is different from that observed for the analogous
set of copper porphyrazines. In the latter system, an increase in
the ionic character of the Cu-N bond with increasing ligand
size was inferred experimentally.10 Therefore, we expected a
similar effect also for the zinc series of porphyrazines, which
should have resulted in an increased ratio between the zinc
shakeup satellites and zinc main line with the enlargement of
the ligand (because more free Zn 4s states exist if the Zn-N

Thin Films of Extended Zinc Porphyrazines
J. Phys. Chem. B, Vol. 109, No. 16, 2005 7833
(9) Pop, D.; Winter, B.; Freyer, W.; Hertel, I. V.; Widdra, W. J. Phys.
Chem. B 2003, 107, 11643.
(10) Pop, D.; Winter, B.; Freyer, W.; Weber, R.; Widdra, W.; Hertel, I.
V. J. Phys. Chem. B 2004, 108, 9158.
(11) Kobayashi, N.; Mack, J.; Ishii, K.; Stillman, M. J. Inorg. Chem.
2002, 41, 5350.
bond becomes more ionic). Note, however, that in analyzing
the intensity variations of the zinc satellite peaks we are less
sensitive than in the case of the copper satellites. For the copper
porphyrazines, the satellites from the valence band were
sufficiently intense to reveal a dependence with the ligand size
even at off-resonant excitations. Off-resonance, the Zn shakeup
satellites have too small an intensity to be well observed in the
spectra, and therefore their analysis can be done only on-
resonance. However, within our accuracy the resonant increase
of the Zn shakeup satellites does not differ for ZnP0, ZnP1,
and ZnP2. Whether a change in the Zn-N bond character is
simply too small to be detected in the present photoemission
measurements or whether it is really absent remains unresolved
at the present moment.
(12) Ricciardi, G.; Rosa, A.; Baerends, E. J. J. Phys. Chem. A 2001,
105, 5242.
(13) Schechtman, B. H.; Spicer, W. E. J. Mol. Spectrosc. 1970, 33, 28.
(14) Edwards, L.; Gouterman, M. J. Mol. Spectrosc. 1970, 33, 292.
(15) Huang, T. H.; Rieckhoff, K. E.; Voigt, E. M. J. Phys. Chem. 1981,
85, 3322.
(16) Stillman, M. J.; Nyokong, H. Absorption and Magnetic Circular
Dichroism spectral Properties of Phthalocyanines. In Phthalocyanines.
Properties and Applications; Leznoff, C. C., Lever, A. B. P., Eds.; VCH:
New York, 1989; Vol. 1, p 133.
(17) Nyokong, T.; Gasyna, Z.; Stillman, M. J. Inorg. Chem. 1987, 26,
1087.
(18) Vancott, T. C.; Rose, J. L.; Misener, G. C.; Williamson, B. E.;
Schrimpf, A. E.; Boyle, M. E.; Schatz, P. N. J. Phys. Chem. 1989, 93,
2999.
(19) Mack, J.; Stillman, M. J. J. Phys. Chem. 1995, 99, 7935.
(20) Plows, F. L.; Jones, A. C. J. Mol. Spectrosc. 1999, 194, 163.
(21) Weiss, C.; Kobayashi, H.; Gouterman, M. J. Mol. Spectrosc. 1965,
16, 415.
(22) Schaffer, A. M.; Gouterman, M.; Davidson, E. R. Theor. Chim.
Acta 1973, 30, 9.
(23) Soares, L. D.; Trsic, M.; Berno, B.; Aroca, R. Spectrochim. Acta
A 1996, 52, 1245.
(24) Hashimoto, T.; Choe, E. K.; Nakano, H.; Hirao, K. J. Phys. Chem.
4. Conclusions
The analysis of the photoemission spectra of thin films of
benzoannelated zinc porphyrazine molecules deposited on a gold
single crystal has been performed. The focus was on the
evolution of the electronic structure and of the zinc-ligand
interactions with increasing ligand size. The measurements
revealed that the enlargement of the ligand system causes a shift
of the HOMO toward lower binding energies, which is similar
to that observed for the analogous metal-free compounds. Zinc
shakeup satellite signal is clearly observed in the spectra of the
valence region for photon energies near the 3p-4s resonant
excitation. However, for off-resonant excitations the satellite
peaks were observable only in the ZnPc spectrum, with an
intensity of at most 2% compared to that of the main line. For
a given excitation energy, the intensity ratios between the
satellite peaks and the Zn 3d main line are similar for ZnP0,
ZnP1, and ZnP2. Thus, differences in the covalent (or ionic)
character of the bond between zinc and ligand as a function of
ligand size were not detected in the present measurements.
A 1999, 103, 1894.
(25) Nguyen, K. A.; Pachter, R. J. Chem. Phys. 2001, 114, 10757.
(26) Liao, M. S.; Scheiner, S. J. Chem. Phys. 2001, 114, 9780.
(27) Baerends, E. J.; Ricciardi, G.; Rosa, A.; van Gisbergen, S. J. A.
Coord. Chem. ReV. 2002, 230, 5.
(28) Liao, M. S.; Scheiner, S. J. Comput. Chem. 2002, 23, 1391.
(29) Kobayashi, N.; Ogata, H. Eur. J. Inorg. Chem. 2004, 906.
(30) Freyer, W.; Mueller, S.; Teuchner, K. J. Photochem. Photobiol.
A: Chem. 2004, 163, 231.
(31) Barrett, P. A.; Dent, C. E.; Linstead, R. P. J. Chem. Soc. [London]
1936, 1719.
(32) Kobayashi, N.; Nakajima, S.; Osa, T. Inorg. Chim. Acta 1993, 210,
131.
(33) Konami, H.; Hatano, M. Chem. Lett. 1988, 1359.
(34) Gatzke, J.; Winter, B.; Quast, T.; Hertel, I. V. Proc. SPIE 1998,
3464, 14.
References and Notes
(35) Cheng, W. D.; Wu, D. S.; Zhang, H.; Chen, J. T. Phys. ReV. B
(1) Phthalocyanines. Properties and Applications; Leznoff, C. C.,
Lever, A. B. P., Eds.; VCH: New York, 1989; Vol. 1.
(2) Phthalocyanines. Properties and Applications; Leznoff, C. C.,
Lever, A. B. P., Eds.; VCH: New York, 1993; Vol. 3.
(3) Reimers, J. R.; Lu, T. X.; Crossley, M. J.; Hush, N. S. Chem. Phys.
Lett. 1996, 256, 353.
2001, 64, 125109.
(36) Schlettwein, D.; Hesse, K.; Gruhn, N. E.; Lee, P. A.; Nebesny, K.
W.; Armstrong, N. R. J. Phys. Chem. B 2001, 105, 4791.
(37) Iwan, M.; Koch, E. E.; Chiang, T. C.; Himpsel, F. J. Phys. Lett. A
1980, 76, 177.
(38) Yeh, J.-J. Atomic Calculations of Photoionization Cross Sections
and Asymmetry Parameters; Gordon and Breach: Langhorne, PA, 1993.
(39) Orti, E.; Bre´das, J.-L. J. Am. Chem. Soc. 1992, 114, 8669.
(40) Orti, E.; Crespo, R.; Piqueras, M. C.; Tomas, F. J. Mater. Chem.
1996, 6, 1751.
(4) Tomiyama, T.; Watanabe, I.; Kuwano, A.; Habiro, M.; Takane,
N.; Yamada, M. Appl. Opt. 1995, 34, 8201.
(5) Ali, H.; van Lier, J. E. Chem. ReV. 1999, 99, 2379.
(6) Hanack, M.; Durr, K.; Lange, A.; Barcina, J. O.; Pohmer, J.; Witke,
E. Synth. Met. 1995, 71, 2275.
(7) Nalwa, H. S.; Hanack, M.; Pawlowski, G.; Engel, M. K. Chem.
Phys. 1999, 245, 17.
(41) Didziulis, S. V.; Cohen, S. L.; Butcher, K. D.; Solomon, E. I. Inorg.
Chem. 1988, 27, 2238.
(8) Nalwa, H. S.; Kakuta, A.; Mukoh, A. J. Phys. Chem. 1993, 97,
1097.
(42) Ishii, T.; Taniguchi, M.; Kakizaki, A.; Naito, K.; Sugawara, H.;
Nagakura, I. Phys. ReV. B 1986, 33, 5664.