The surface trans effect: Influence of axial ligands on the surface chemical bonds of adsorbed metalloporphyrins

Article abstract of DOI:10.1021/ja1093502

The chemical bond between an adsorbed, laterally coordinated metal ion and a metal surface is affected by an additional axial ligand on the metal ion. This surface analogon of the trans effect was studied in detail using monolayers of various M(II)-tetrap

Full text of DOI:10.1021/ja1093502

ARTICLE
pubs.acs.org/JACS
The Surface Trans Effect: Influence of Axial Ligands on the Surface
Chemical Bonds of Adsorbed Metalloporphyrins
,
†
‡
‡
§
§
Wolfgang Hieringer,* Ken Flechtner, Andreas Kretschmann, Knud Seufert, Willi Auw €a rter,
§
†
‡
,‡
Johannes V. Barth, Andreas G €o rling, Hans-Peter Steinr €u ck, and J. Michael Gottfried*
†
Lehrstuhl f €u r Theoretische Chemie, Universit €a t Erlangen-N €u rnberg, Egerlandstrasse 3, 91058 Erlangen, Germany
‡
Lehrstuhl f €u r Physikalische Chemie II and Interdisciplinary Center for Molecular Materials (ICMM), Universit €a t Erlangen-N €u rnberg,
Egerlandstrasse 3, 91058 Erlangen, Germany
§
Physik-Department E20, Technische Universit €a t M €u nchen, James-Franck-Strasse, 85748 Garching, Germany
S Supporting Information
b
ABSTRACT: The chemical bond between an adsorbed, later-
ally coordinated metal ion and a metal surface is affected by an
additional axial ligand on the metal ion. This surface analogon of
the trans effect was studied in detail using monolayers of various
M(II)-tetraphenylporphyrins (MTTPs, M = Fe, Co, Zn) and
their nitrosyl complexes on a Ag(111) surface. X-ray photo-
electron spectroscopy (XPS) shows that the oxidation state of the Fe and Co (but not Zn) ions in the MTPP monolayers is reduced
because of the interaction with the substrate. This partial reduction is accompanied by the appearance of new valence states in the
UV photoelectron and scanning tunneling spectra (UPS and STS), revealing the covalent character of the ion-substrate bond.
Subsequent coordination of nitric oxide (NO) to the metal ions (Fe, Co) reverses these surface-induced effects, resulting in an
increase of the oxidation states and the disappearance of the new valence states. Removal of the NO ligands by thermal desorption
restores the original spectroscopic features, indicating that the described processes are fully reversible. The NO coordination also
changes the spin state and thus the magnetic properties of the metal ions. Density-functional theory (DFT) calculations on model
systems provide structural and energetic data on the adsorbed molecules and the surface chemical bond. The calculations reveal that
competition effects, similar to the trans effect, play a central role and lead to a mutual interference of the two axial ligands, NO and
Ag, and their bonds to the metal center. These findings have important implications for sensor technology and catalysis using
supported planar metal complexes, in which the activity of the metal center is sensitively influenced by the substrate.
1
. INTRODUCTION
additional ligand to the metal center and thus play a similar role
as the thiolate ligand for the Fe center in the cytochrome P450
enzymes. For example, cobalt tetraphenylporphyrin (CoTPP)
Adsorbed metal complexes are promising candidates for novel
30
heterogeneous catalysts with well-defined active sites. The uni-
form reactivity of these sites ensures a similarly high selectivity as
known from homogeneous catalysts, while the practical advan-
on TiO catalyzes the reduction of NO with H or CO, while
2
x
2
unsupported CoTPP or TiO alone are inactive. The activity
2
1
ꢀ7
enhancement is apparently related to electron transfer from the
oxide to the Co ion, as was deduced from investigations of
CoTPP/TiO 26 powder samples with UV and EPR
tages of heterogeneous catalysts are maintained.
Another
important application of adsorbed complexes is the sensing of
gases, in which the coordination of gas molecules to the metal
center causes measurable changes of electronic properties, color,
,7
spectroscopy. Other support materials have a similar influence
8
ꢀ12
on the activity of Co porphyrins; for example, thiol-functiona-
lized Co porphyrin anchored to Au(111) catalyzes the liquid-
etc.
phthalocyanines are particularly well suited for anchoring on
solid substrates, because they readily form ordered monolayers
Planar complexes such as metalloporphyrins and
-
phase oxidation of hydrochinone to benzochinone with O more
2
3
1
3ꢀ28
efficiently than the unsupported porphyrin. Likewise, Mn
by self-assembly
and possess two axial coordination sites
porphyrin on Au(111) catalyzes the epoxidation of stilbene, as
was observed with scanning tunneling microscopy (STM) on the
available as centers of catalytic activity or sensor functionality.
This coordinatively unsaturated character of the metal ions is
the key to the specific reactivity and the biological importance of
metalloporphyrins in a large number of enzymes. For example, in
heme-thiolate proteins, such as the cytochrome P450 family, the
active center consists of an iron porphyrin with an axial thiolate
ligand, which influences the electronic structure and facilitates
1
liquidꢀsolid interface.
These and further observations make a fundamental under-
standing of the chemical bond between metal complexes and
surfaces desirable. This aim requires a detailed investigation of
29
the OꢀO bond cleavage and other reactions. In the case of
Received: October 17, 2010
Published: April 04, 2011
adsorbed metal complexes, the substrate can operate as an
r 2011 American Chemical Society
6206
dx.doi.org/10.1021/ja1093502
|J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
the influence of the surface on the electronic structure of the
and the XP spectra (A) and (C) in Figure 1 were recorded in normal
emission; the other XP spectra were collected in grazing emission (80ꢀ
with respect to the surface normal) for increased surface sensitivity.
5,10,15,20-Tetraphenylporphyrin (2HTPP) and its metal complexes
zinc(II)-tetraphenylporphyrin (ZnTPP) and cobalt(II)-tetraphenylporphyr-
in (CoTPP) (purity >98%) were degassed in vacuo for 24 h at 420 K prior to
use and evaporated with a Knudsen cell at 638 K. The resulting flux of
typically 0.05 Å/s at the sample position was measured with a quartz crystal
microbalance. Monolayers of CoTPP and ZnTPP were prepared by deposi-
tion of CoTPP and ZnTPP multilayers, respectively, followed by thermal
30,31
metal center. Recently, we studied monolayers of CoTPP
32
and Co(II)-octaethylporphyrin (CoOEP) on a Ag(111) sur-
face with photoelectron spectroscopy and found a new electronic
valence state, which results from a covalent/coordinative interac-
27,30ꢀ33
tion between the complexed Co ion and the Ag substrate.
This interaction is accompanied by a transfer of electron density
from the Ag surface to the Co ion, as indicated by a substantial shift
of the Co(2p) level to lower binding energy.
Previously, it was shown that this CoꢀAg interaction is
31,44
desorption of the excessive multilayers at 550 K as explained in refs.
the preparation of iron(II)-tetraphenylporphyrin (FeTPP) monolayers,
HTPP monolayers (prepared as described above for CoTPP and ZnTPP,
For
suppressed if a nitric oxide (NO) ligand binds to the coordinated
3
4
Co ion. This observation was preliminarily interpreted as a
manifestation of the trans effect, which is an established concept in
classical, solution-based coordination chemistry and describes
the competition between two ligands in trans position binding to
2
31,45
cf. refs. ) were metalated directly on the surface by deposition of the
19,20,45ꢀ47
stoichiometric amount of Fe atoms as described in refs.
For the preparation of the monolayers of (NO)CoTPP and
3
5ꢀ39
the same metal center.
NO was chosen as a ligand, because
(
NO)FeTPP, NO gas (>99.5%) was dosed onto CoTPP and FeTPP
38
it exerts an exceptionally large structural trans effect and has
ꢀ8
monolayers, respectively, with a pressure of 3 ꢂ 10 mbar at a sample
found continued interest as a ligand to adsorbed tetrapyrrole
temperature of 140 K. The total NO dosage was 300 L (Langmuir, 1
4
0ꢀ43
complexes.
Due to the trans effect, a ligand can modify the
ꢀ6
L = 1 ꢂ 10 Torr ꢂ s). No direct adsorption of NO on the Ag(111)
metal-to-ligand (MꢀL) bond of other ligands within a complex.
In particular, it can weaken the MꢀL bond of the ligand in trans
position or, in other words, two trans ligands compete for the
stronger bond to the metal center. Typical manifestations of this
effect are the increased MꢀL bond length of the “weaker” trans
ligand (structural trans effect) and the increased kinetic lability of
48
surface was observed under these conditions. (NO)CoTPP multi-
ꢀ8
layers were prepared by dosing NO (3 ꢂ 10 mbar) during the vapor
deposition of CoTPP at 140 K (total NO dosage: 800 L). The XP and
UP measurements on (NO)CoTPP and (NO)FeTPP were performed
ꢀ
8
at 140 K at an NO pressure of 3 ꢂ 10 mbar.
The coverage θ is defined as the number of adsorbed molecules or
atoms divided by the number of substrate atoms, whereas the term
“monolayer” denotes a complete, saturated adlayer of molecules in direct
contact with the substrate surface at 300 K. Monolayer coverage
the weaker ligand toward substitution (kinetic trans effect).
ꢀ
“
N
Strong” ligands with a very high trans activity, such as NO ,
3ꢀ
2ꢀ
2ꢀ
, O and S , induce large structural trans effects in
octahedral and square-planar complexes. H , R , RCO and
RN2 also show substantial trans activity, while the ligands CO,
ꢀ
ꢀ
ꢀ
31
corresponds to θ = 0.037.
ꢀ
2.2. Scanning Tunneling Spectroscopy. The scanning tunnel-
ing spectroscopy (STS) and microscopy (STM) experiments were
performed in a custom-designed, stand-alone ultrahigh vacuum
ꢀ
ꢀ
ꢀ
38
CN , NO , PR and RS are relatively “weak”.
2
3
The observation of a trans effect on adsorbed complexes under
participation of the substrate acting as an axial ligand is very
important with respect to catalysis on supported metal com-
plexes, because it reveals how a coordinated reactant molecule
alters the properties of the active site. Therefore, a detailed
investigation of chemical bonding and electronic interactions of
reactive adsorbed metal complexes must take these ligand- and
surface-induced effects explicitly into account. This is the focus of
the present work, which reports a joint experimental and
theoretical investigation of the chemical bond between com-
plexed metal ions (Fe, Co) and a Ag(111) surface and the
influence of an axial NO ligand on this bond. These more reactive
systems will be compared to complexed Zn ions in a similar
environment as a relatively inert counterpart and data of the
4
9
(
UHV) apparatus
comprising
a
commercial low-temperature
5
0
ꢀ10
STM. The system base pressure is below 2 ꢂ 10
mbar. All ST
spectra were recorded under open feedback loop conditions via lock-in
technique using electrochemically etched tungsten tips. The STM
images were taken in constant current mode. For STM and STS
measurements, the sample was kept at 6 K. Submonolayer coverages
of CoTPP and in situ metalated FeTPP were applied in order to have
bare Ag(111) patches available to characterize the tip electronic
structure prior to any experiment. In the figure caption, V refers to the
bias voltage applied to the sample. While the general experimental
procedures were followed as described in Section 2.1, we applied lower
NO doses (around 10 L) to not fully saturate the metallotetraphenyl-
porphyrin (MTPP) layers. This allows for a direct spectroscopic
comparison of NO/MTPP and MTPP with an identical tip.
interaction of ZnTPP/Ag(111) with an NH ligand.
3
2
.3. Computational Methods. All quantum-chemical calcula-
51,52
tions presented here are based on density-functional theory (DFT).
Both periodic and nonperiodic calculations were performed to study two
different types of model systems that, in a complementary way, describe
different aspects of the experimental system. As the first type of model
systems, periodic slab calculations of metalloporphyrin (MP) molecules
adsorbed on a Ag(111) surface slab have been performed using the
2
. EXPERIMENTAL AND COMPUTATIONAL DETAILS
2
.1. Photoelectron Spectroscopy (PES). All PES investigations
were performed with a Scienta ESCA-200 photoelectron spectrometer
which was described in detail elsewhere. The sample was a discoidal Ag
single crystal (purity >99.999%) with a polished (111) surface, which
32
5
3
Vienna Ab Initio Simulation Package (VASP). For the second type of
model systems, nonperiodic calculations on isolated MTPP complexes
coordinated to a single Ag atom or a cluster of Ag atoms were carried out.
3
1
was aligned to <0.1ꢀ with respect to the nominal orientation (see ref.
for details about sample preparation). Two type K thermocouples were
mounted to the rim of the Ag crystal for temperature measurement ((5
K). The crystal was attached to a manipulator with facilities for cooling
and heating in the temperature range 95ꢀ900 K.
54
For the geometry optimizations of these systems, the Turbomole
program package was used in most cases. For the analysis of the
5
5
electronic structure, the ADF program was employed. The calculations
were performed in the spin-unrestricted framework, and care was taken
that the lowest-energy spin state was attained.
UP spectra were taken with HeꢀI radiation (21.22 eV) and a sample
bias of ꢀ10 V. The reported XPS and UPS binding energies are
referenced to the Fermi edge of the clean Ag surface (E
B
ꢁ 0). The
For the periodic slab calculations, the tetraphenylporphyrins inves-
tigated in the experiments have been simplified to the corresponding
following photoelectron detection geometries were used: All UP spectra
6
207
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
basic porphins, that is, the peripheral phenyl groups have been replaced
by hydrogen atoms. While these metalloporphins are not expected to
quantitatively reproduce all properties of the MTPP complexes used in
the experiments, they feature reduced steric repulsions of the ligand with
the metal surface and hence allow us to isolate the electronic trans effect
analogon which is central to this study. An orthorhombic unit cell of the
employed for better efficiency. For the Turbomole calculations, the
6
3
resolution-of-the-identity (RI) technique for the Coulomb integrals
6
4
and its multipole-accelerated version (MARIJ) was used through-
6
5
out. A similar density-fitting scheme is used by ADF. Full
(unconstrained) geometry optimizations were performed in general
(using Turbomole), and the minima was confirmed by frequency
calculations. For the larger clusters, constraints were used during the
geometry optimizations to ensure that the clusters can be regarded as
cutouts from the Ag fcc bulk structure. In these cases, relaxation was
allowed for those Ag atoms which are involved in the bonding to the
metalloporphyrins (see the SI for further details). Additional calcula-
tions, especially the fragment orbital based analysis of the electronic
structure, have been performed with the help of ADF.
3
size 14.4 ꢂ 15.0 ꢂ 35.4 Å has been used. The unit cell contained a single
metalloporphin (MP) molecule (M = Co, Fe) plus a five-layer Ag slab
with a total of 150 Ag atoms. This unit cell implies a vacuum layer of
approximately 18 Å, which serves to decouple the individual slabs in the
vertical dimension. The MP molecules have been placed on the Ag(111)
surface such that the central metal M was next to a top-site Ag atom
(Agtop) of the silver surface. Test calculations revealed that adsorption
over a bridge site or an fcc hollow site of the Ag(111) surface are
associated with similar binding energies (within about 0.1 eV) at the
present theoretical level. We chose top-site adsorption for our analysis in
order to allow for a direct comparison with model Ag complexes. Details
of the final geometries can be found in the Supporting Information (SI).
In all geometry optimizations involving the Ag slab the two topmost
layers of the slab were fully relaxed, while the bottom three layers were
kept fixed at the bulk positions of an fcc Ag crystal. All other atomic
positions have been optimized without constraints until the energy
The geometric and electronic structures were subsequently used for
the interpretation of the XPS, UPS and STS data using the same
exchange-correlation functionals and basis sets. For the nonperiodic
5
1,66,67
calculations, the so-called Slater transition state
method to calcu-
late shifts in core ionization potentials was used, and the results were
6
8,69
compare with experimental XPS peak shifts
in order to verify the
validity of the computational model systems. Partial charges according to
7
0
the Bader scheme were used in this study to estimate the direction of
charge transfer upon adsorption and ligand coordination (see the SI for
details). The densities for the Bader partitioning were obtained from the
periodic calculations as described above.
ꢀ
3
gradients dropped below 5 ꢂ 10 eV/Å. The PBE generalized gradient
5
6
approximation (GGA) functional due to Perdew, Burke and Ernzerhof
has been used unless stated otherwise, and the projector-augmented
57,58
wave method (PAW)
has been employed to treat the atomic cores.
3. EXPERIMENTAL RESULTS
The “normal” precision setting was used in the VASP calculations, which
corresponds to a global plane-wave energy cutoff of 400.0 eV (29.40 Ry,
3.1. X-Ray Photoelectron Spectroscopy (XPS). The central
5
.42 au). A single k-point has been used for the unit cell described above
aim of this work is to explore the influence of the Ag surface, of the NO
ligand, and of the combination of both on adsorbed, complexed metal
ions. For this purpose, we studied monolayers and multilayers of
metallo-tetraphenylporphyrins (MTPP, M = Co, Fe, Zn) and their
nitrosyl complexes ((NO)MTPP, M = Co, Fe), using the M(2p3/2) core
level signals as a monitor of the ligand-induced changes (Figures 1ꢀ3).
Figure 1a shows the Co(2p ) signal of CoTPP multilayers, that is, a
(
5 ꢂ 6 Ag atoms per layer). We have checked our results against a 3 ꢂ 3
ꢂ 1 k-point mesh for some cases, but found that the main results were
not affected. The convergence criterion for the self-consistent field
ꢀ
5
procedure for the electronic wave functions was set to 10 eV. All
calculations have been carried out in a spin-polarized fashion; the
minority spin channel is labeled β in this work. Core level shifts have
been estimated from the local average electrostatic potential at each ion
following a standard procedure for the periodic calculations.
3
/2
system in which the majority of the molecules does not interact with the
Ag substrate. The maximum is located at 780.0 eV, which is typical of
cobalt in a formal þ2 oxidation state and agrees with the nominal
oxidation state of the ion. The multiplet structure of the signal is caused
In general, most GGA density functionals (including those used here)
are too repulsive in the long-range dispersion regime, which makes bond
energies and bond distances beyond the covalent regime unreliable. In
some explicitly indicated cases, we have therefore included an empirical
long-range dispersion correction to account for the effect of van der
Waals forces (labeled PBEþvdW for the PBE functional with van der
7
31
by the unpaired electron of the Co(II) ion (d ).
The Co(2p3/2) signal of the CoTPP monolayer (Figure 1b) has its
maximum at 778.2 eV and also shows a multiplet structure. The chemical
shift of ꢀ1.8 eV relative to the multilayer signal exceeds by far the
respective surface-induced core-level shifts for the other elements,
carbon (ꢀ0.2 eV) and nitrogen (ꢀ0.3 eV), as was discussed
59
Waals dispersion correction). In most cases, however, we have not
included any dispersion correction to the standard GGA functional,
because we are predominantly interested in the electronic (i.e., covalent
or coordinative) interaction of the M(TP)P molecules with the surface,
which is usually well described by common GGA functionals. In general,
the inclusion of van der Waals dispersion reduces the molecule-surface
distances somewhat due to the attractive nature of the interaction.
For the nonperiodic calculations on isolated MTPP complexes using
Turbomole and ADF, the exchange-correlation functionals due to
3
1
previously. While these small shifts of the C(1s) and N(1s) signals
can be attributed to final state effects (relaxation shift), the much larger
shift of the Co(2p) signal suggests operation of initial state effects, such
as transfer of electron density from the Ag surface to the Co ion. This
assumption is supported by the theoretical analysis, which indicates a
charge transfer of 0.37e from the surface to cobalt porphin (cf. Section
2.3 and the SI for details; additional evidence (not shown) is given by the
calculated (initial-state) N 1s orbital energies or core-potential shifts
upon Ag or Ag(111) coordination, which are very small (<0.3 eV)). The
shoulder around 778 eV in the Co(2p3/2) multilayer spectrum
(Figure 1a) arises from molecules in the first layer, since its intensity
depends on the thickness of the multilayer and eventually vanishes for
thick multilayers.
60
61
Becke and Perdew (hereafter denoted BP) and due to Perdew,
56
Burke and Ernzerhof (PBE) were used. Note that we prefer the BP
functional in the study of the molecular model complexes rather than the
PBE functional which is used for the periodic density functional
calculations. The BP functional has been found suitable for the study
of molecular coordination compounds in many previous studies. For the
Turbomole and ADF calculations the TZVP and TZ2P basis set and
TZ2P have been used, respectively. The influence of metal coordination
to the MTPPs was studied by attaching a Ag atom or a small cluster of
varying size (Ag19, Ag72) to the metal center (Co, Fe, or Zn) of the
6
2
The Co(2p3/2) signal of (NO)CoTPP multilayers (Figure 1c) shows
how the NO ligand influences Co ions that do not interact with the Ag
surface. Comparison to the multilayer signal of CoTPP reveals two
major differences: First, the maximum is shifted to higher binding energy
by þ0.8 eV and second, the multiplet structure vanished. The positive
peak shift indicates a partial oxidation of the Co ion by the NO ligand,
metalloporphyrins. For complexes bearing larger Ag clusters (Ag19
Ag ), only a double-ζ basis set (Ahlrich’s SV(P) basis set) was
,
72
6
208
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
Figure 3. Coordination of NO to a FeTPP monolayer on Ag(111) and
thermal desorption of the NO ligand, Fe(2p3/2) XP spectra: (a) FeTPP
monolayer, (b) (NO)FeTPP monolayer, (c) after heating the
(
NO)FeTPP monolayer to 600 K.
Figure 1. Co(2p3/2) XP spectra of CoTPP and (NO)CoTPP on
Ag(111): (a) CoTPP multilayers (∼4 layers), (b) CoTPP monolayer,
N(1s) (ꢀ0.3 eV) signals and much less than the Co(2p3/2) shift of ꢀ1.8
eV between multilayer and monolayer CoTPP in the absence of the NO
ligand. Obviously, it makes little difference whether the (NO)CoTPP
complex is in contact to the surface or not, suggesting that the CoꢀAg
interaction is reduced by the NO ligand. This conclusion is confirmed by
the UPS, STS, and DFT results discussed below.
(
c) (NO)CoTPP multilayers (∼4 layers), and (d) (NO)CoTPP
monolayer. The small shoulder around 778 eV in (a) stems from the
CoTPP monolayer. The other curves in the signal deconvolution of (a)
and (c) are displayed for illustrative purposes and do not necessarily
represent the full complexity of the spectra.
Figure 2 illustrates that the suppression of the CoꢀAg interaction is
fully reversible: heating of the (NO)CoTPP monolayer to 500 K
removes the NO ligand and restores position and shape of the original
Co(2p3/2) signal prior to NO coordination on the Co ion.
Similar observations were made for FeTPP. Figure 3a shows the
Fe(2p3/2) signal of the FeTPP monolayer with a main peak at 706.7 eV
and a complex multiplet structure, which is attributed to the open-shell
7
2
character of the Fe(II) ion. (The present and earlier DFT calculations
predict indeed a triplet ground state for isolated FeTPP, see below).
Formation of (NO)FeTPP by coordination of NO to the FeTPP
monolayer (Figure 3b) shifts the maximum of the Fe(2p3/2) signal by
þ2.7 to 709.4 eV. In analogy to (NO)CoTPP, this large shift toward
higher binding energy is interpreted as an NO induced suppression of
the FeꢀAg interaction. Part of the shift may also be attributed to a partial
oxidation of the Fe ion by the NO ligand (see the discussion in the next
paragraph and the SI for calculated partial charges). In contrast to
(NO)CoTPP, the (NO)FeTPP monolayer signal still shows multiplet
structure, reflecting the complex electronic structure of the Fe ion in a
Figure 2. Reversibility of the changes in the electronic structure
induced by NO coordination: Co(2p3/2) XP spectra of CoTPP and
7
3
(NO)CoTPP on Ag(111). (a) CoTPP monolayer, (b) (NO)CoTPP
porphyrin complex. Due to the additional unpaired electron from the
monolayer, (c) after heating the (NO)CoTPP monolayer to 500 K for
thermal desorption of the NO ligand.
NO ligand, the Fe ion probably changes from the triplet state as the
73,74
lowest-energy state
to a doublet or quartet state. Which of these
states dominates cannot be decided on the basis of our XP spectra;
however, our DFT calculations predict a doublet ground state, see
below. In addition, Figure 3b shows good agreement with spectra of
other Fe(III) complexes. As was the case for the (NO)CoTPP
complex, thermal desorption of the NO ligand restores the original
spectrum within the noise limit, that is, the described effects are
reversible also for the FeTPP system (Figure 3c). The NO ligand is
apparently stronger bound than in the Co complex, since a temperature
of 600 K is necessary for its desorption.
which in turn acquires a partial anionic (nitroxylic) character. (The
theoretical analysis suggests a negative charge of 0.2e on the coordinated
NO ligand, see the SI for details.) The multiplet splitting disappears
presumably because, upon formation of the NOꢀCo bond, the spin of
the unpaired electron of the NO molecule compensates the spin of the
Co(II) ion. Indeed, our own DFT calculations and previous results
favor a low-spin (singlet) state for (NO)CoTPP, as will be discussed in
detail below.
7
5
7
1
In the (NO)CoTPP monolayer, the Co ion interacts with two axial
ligands”: the NO molecule and, in a trans position to it, the Ag surface.
The related Co(2p3/2) spectrum in Figure 1d shows a single peak shifted
by ꢀ0.4 eV compared to the (NO)CoTPP multilayer signal. This is only
slightly more than the relaxation shifts of the C(1s) (ꢀ0.2 eV) and
The FeTPP monolayers were prepared by direct metalation of
2HTPP monolayers on Ag(111) with the stoichiometric amount of
“
1
9,20,47
vapor-deposited Fe as described in Section 2.1.
This in situ
approach was chosen because of the high sensitivity of FeTPP toward
oxidation, which makes preparation and handling outside the vacuum
6
209
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
Figure 4. Zn(2p3/2) XP spectra of (a) multilayers and (b) monolayer of
ZnTPP on Ag(111).
Figure 6. Valence electronic structure of FeTPP and (NO)FeTPP on
Ag(111): HeꢀI UP spectra of (a) FeTPP monolayer, (b) (NO)FeTPP
monolayer, and (c) after heating the (NO)FeTPP monolayer to 600 K
for removal of the NO ligand.
by Figure 4, which shows the Zn(2p3/2) signal of multilayers and
monolayer of ZnTPP on Ag(111). The peak positions differ here by
only 0.25 eV, which is in the same range as the C(1s) and N(1s) core
level shifts and probably largely due to relaxation. Apparently, there is
little or no electronic interaction between the Zn(II) ion and the Ag
surface. This interpretation is supported by the UP spectra and the
computational results discussed below. Unsurprisingly, the coordination
of a ligand (here NH
ions of a ZnTPP monolayer causes only a small shift of the Zn(2p3/2
level, ꢀ0.2 eV, and the ZnꢀNH bond is relatively weak (desorption of
3
, since no nitrosyl ligation was observed) to the Zn
)
3
7
9
NH
is computed to 49 kJ/mol (BP/TZVP density-functional level), which is
in good agreement with experiment where a ZnꢀNH bond dissociation
3 3
around 150 K). The binding energy of NH to gas-phase ZnTPP
3
Figure 5. Manifestations of the surface coordinative bond in the valence
electronic structure and the influence of the NO ligand on the CoꢀAg
interaction in CoTPP monolayers on Ag(111): HeꢀI UP spectra (hν =
7
9
energy of 40 kJ/mol was estimated.
.2. UV Photoelectron Spectroscopy (UPS). In this section, we
3
21.22 eV) of (a) clean Ag(111) surface, (b) CoTPP monolayer, (c)
focus on the influence of the Ag surface and the NO ligand on the
valence electronic structure of the metal complexes. The HeꢀI UP
spectra (a) and (b) in Figure 5 refer to the clean Ag(111) surface and to a
monolayer of CoTPP on Ag(111), respectively. As previously shown,
the signal around 0.6ꢀ0.7 eV in Figure 5b results from the electronic
(NO)CoTPP monolayer, and (d) after heating the (NO)CoTPP
monolayer to 500 K for thermal desorption of the NO ligand.
76
31
difficult. While yielding high degrees of metalation in the monolayer
>90%), this method is less efficient for the metalation of multilayers,
because only a fraction of the Fe atoms deposited is coordinated by the
HTPP multilayers. Since the remaining part of the Fe retains the (0
interaction of the coordinated cobalt ion with the Ag surface. This
(
signal was tentatively attributed to the interaction of Co 3d orbitals with
31,33
electronic states of the porphyrin macrocycle and the Ag surface.
It
2
was proposed that this interaction leads to two new electronic states,
20
oxidation state, probably by formation of Fe clusters, comparison with
pure FeTPP and (NO)FeTPP multilayers is not possible here. However,
we can compare with multilayers and monolayers iron(II) phthalocya-
nine (FePc), which have been studied previously and in which Fe has a
which are located below the Fermi energy (E ) of Ag and thus are
F
populated by electrons from the substrate. Therefore, this model
suggests transfer of electron density from the surface to the Co ion
and is thus in agreement with the XPS results discussed above. (This
tentative model is analyzed in more detail using our DFT calculations, as
will be explained below.) Upon coordination of NO to the Co ion
(Figure 5c), the interaction-induced signal around 0.6ꢀ0.7 eV disap-
pears and the signal around 2 eV shifts by þ0.4 eV, reaching the position
77
very similar coordination environment. The Fe(2p3/2) signal of FePc
multilayers shows a broad multiplet structure around 708.8 eV. In the
FePc monolayer signal, the maximum is shifted by ꢀ1.7 to 707.2 eV,
while the signal shape is very similar to Figure 3a. This comparison
suggests that the larger part (þ1.7 eV) of the NO-induced shift of þ2.7
eV is caused by the suppression of the FeꢀAg interaction, while a smaller
part (þ1.0 eV) is due to electron transfer from the Fe ion to the NO
3
1
of the highest occupied state of a CoTPP multilayer. These changes are
reversed when the NO ligand is removed by heating to 500 K
(Figure 5d).
78
ligand. This implies that the NO ligand has predominantly anionic
Figure 6 shows the analogous results for FeTPP. The interaction-
induced peak (Figure 6a, marked in green) is here located slightly below
EF and is less intense than in the case of CoTPP. Its intensity is strongly
reduced by coordination of NO to the Fe ion (Figure 6b) and restored
upon thermal desorption of the NO ligand by heating to 600 K
(Figure 6c).
ꢀ
(
NO ) character (cf. the SI for calculated partial charges).
The close similarities in the interaction of Co and Fe ions with the Ag
surface and toward the coordination of NO may suggest that other
complexed transition metal ions in proximity to a metal surface may
behave similarly. This, however, is not generally the case, as is illustrated
6
210
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
For both CoTPP and FeTPP, the NO-induced changes in the valence
region are consistent with the conclusion drawn from the XPS data: Both
sets of UP spectra show that the central metal ions in (NO)MTPP
monolayers (M = Co, Fe) are far less influenced by the Ag surface than in
MTPP monolayers. In other words, coordination of the NO ligand
weakens the interaction of the metal ion with the substrate.
The interaction of ZnTPP with the Ag surface does not lead to
significant changes of the valence electronic structure, as can be seen in
Figure 7, which compares monolayer UP spectra of ZnTPP, CoTPP,
show the interaction-induced signals described above. The relative
inertness of ZnTPP is attributed to the completely occupied Zn(3d)
8
0
levels and their low energies (i.e., high binding energies). In addition,
monolayer and multilayers of ZnTPP show very similar UP spectra; no
additional signals appear in the monolayer spectrum. A table with all
valence peak positions of ZnTPP and CoTPP monolayers and multi-
layers and of the FeTPP monolayer is reproduced in the SI (Table S1).
7
9
3
Coordination of NH to the Zn ion does not lead to significant
changes in the UP spectra in the range between E and 3.5 eV (Figure S3
F
FeTPP, and 2HTPP. Between E
almost identical to the spectra of the 2HTPP monolayer and of the pure
Ag surface (cf. Figure 5a). In contrast, the spectra of CoTPP and FeTPP
F
and 1.5 eV, the ZnTPP spectrum is
of the SI). This is consistent with the low tendency of ammonia to bind
to ZnTPP as predicted by our DFT calculations (see below).
3.3. Scanning Tunneling Spectroscopy (STS). Up to now we
reported results on the valence electronic structure of MTPP molecules
on Ag(111) gained by spatially averaging photoemission experiments for
occupied states. In this section, we compare these findings with spectro-
scopic data acquired on the single molecule level by scanning tunneling
spectroscopy (STS). Besides being a local probe, STS has the advantage
F
to yield information also on unoccupied levels above E . Figure 8 shows
scanning tunneling microscopy (STM) images of CoTPP (Figure 8a)
and FeTPP (Figure 8b) species with and without coordination of an
additional NO ligand, together with the corresponding ST spectra
(Figure 8c and d, respectively). A detailed discussion of the STM images
is beyond the scope of this article, but it should be noticed that bare
2
7
MTPP molecules can be clearly discriminated from the (NO)MTPP
species. All the spectra presented in Figure 8c and d are recorded at an
identical off-center position above the porphyrin macrocycle (see
Figure 8a and b). This location was chosen to maximize the contribution
Figure 7. Comparison of the HeꢀI UP spectra of CoTPP (blue),
FeTPP (red), ZnTPP (green), and 2HTPP (black, dashed).
Figure 8. Scanning tunnelling microscopy (STM) images and ST spectra of MTPP and (NO)MTPP on Ag(111) showing the quenching of the
interaction-induced peak (marked in green) upon NO coordination. (a) STM image of a mixed CoTPP and (NO)CoTPP array (I = 55 pA, V = ꢀ0.7 V).
The NO related species are characterized by a round central protrusion. The colored dots represent the off-center position above the porphyrin
macrocycle where ST spectra were recorded (compare c and d), see text for discussion). (b) Mixed FeTPP and (NO)FeTPP array (I = 0.1 nA, V = ꢀ0.7
V). (c) ST spectra of CoTPP (blue) and (NO)CoTPP (light blue), compare to the UP spectra in Figure 5. (Stabilizing parameters for the STM tip before
opening the feedback loop: I = 55 pA, V = ꢀ0.6 V, modulation amplitude 25 mV, modulation frequency 967 Hz.) (d) ST spectra of FeTPP (red),
(
NO)FeTPP (light red) and 2HTPP (black) on Ag(111), compare to Figure 6 (I = 0.2 nA, V = ꢀ0.28 V, modulation amplitude 25 mV, modulation
frequency 398 Hz).
6
211
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
33
of the interaction-induced state. As these spectra are taken on
individual molecules, the observed peak width is reduced in comparison
to the laterally averaging UP spectra. Addressing first the occupied states
of the bare MTPPs, we observe a well-defined peak around 0.6ꢀ0.7 eV
binding energy (between ꢀ0.6 and ꢀ0.7 V sample bias) for CoTPP and
a similar feature at a binding energy of 0.1 eV for FeTPP. These spectral
fingerprints are in full agreement with the UPS data presented in the
previous section and represent the interaction-induced state (cf. Fig-
ures 5 and 6). Similar occupied resonances were previously reported in
STS studies on Co- and Fe-porphyrin molecules in contact with noble
13,33,81,82
metal surfaces.
Above the Fermi energy, the curves of the bare
MTPPs are featureless in the energy range investigated here. From the
theoretical calculations presented in Section 4.3 (Figure 12), we estimate
the gap between the bonding and the antibonding orbital to at least 1.7
eV, which means that this antibonding level is above the energy range
shown in Figure 8c and d.
Upon coordination of NO to either the Co or Fe center of the
MTPPs, the interaction-induced state is completely quenched. The
resulting spectra are featureless in the displayed binding energy range
Figure 9. Transformations considered in the theoretical analysis, M =
Fe, Co, Zn. In the periodic slab calculations, the MTPP molecules are
modeled as metalloporphins MP (i.e., by omitting the peripheral phenyl
rings) adsorbed to a five-layer Ag slab (see Section 2 for details). In the
molecular model systems, calculations with the full MTPPs have been
performed, but the Ag(111) surface has been replaced by a single Ag
atom. See Tables 1 and 3 for structural data and Tables 2 and 4 for
associated binding energies.
below E . Thus, the results from UPS are confirmed on a single molecule
F
level. In addition, NO coordination induces an enhanced spectral
interaction of the complexes with an ideal Ag(111) surface. The
second class of models provide additional insight into the changes
in the local bonding situation at the metal center upon coordination
of the NO ligand and has the advantage that a direct connection to
molecular coordination chemistry can be made. We show that the
core-level shifts calculated for both of these model systems
qualitatively reproduce those observed in the XPS experiments
described above (see Table 6 and the SI, Table S2 for details).
The transformations considered in the present computational
analysis of both model systems are summarized in Figure 9. For
each of the displayed species, geometry optimizations were
performed as detailed in Section 2. We start our analysis with a
brief discussion of the structural and electronic properties of gas-
phase MP and MTPP molecules (in the following summarized as
M(TP)P) and then proceed to the effect of Ag coordination.
density above E
exhibit an increase in intensity around 100 mV above E
F
. Both the (NO)CoTPP and the (NO)FeTPP spectra
, where the bare
F
MTPP spectra showed no features. On first sight, the additional intensity
upon coordination to NO might be explained as originating from the
antibonding state suggested in Figures 13 and 14 (see below). However,
a comparison of the (NO)MTPP spectra with a spectrum taken on free-
base 2HTPP (cf. Figure 8d) shows very close similarities. Accordingly,
we suggest that the NO coordination not only reduces the interaction
between the MTPP and the underlying Ag(111) substrate, but may also
modify the interaction between the central ion’s d states and the
porphyrin macrocycle. Thus, the electronic structure of NO/MTPP
F
around E seems to resemble its free-base 2HTPP counterpart. Based on
these STS data, it cannot be decided which part of the spectral density
above E is related antibonding states of the MꢀNO interaction and
F
which part is due to a partial decoupling of the metal ion from the surface
and from the porphyrin ligand.
4.1. Molecular Structures of Metalloporphyrins. The equi-
librium and electronic structure of metallotetraphenylporphyrins
7
2,83ꢀ85
(
MTPP) have been extensively discussed in the literature.
4
. THEORETICAL ANALYSIS
However, most of the early studies considered a highly symmetric
geometry (D4h point group), with the peripheral phenyl rings
perpendicular to the otherwise completely flat MTPP core
structure. It should be noted, however, that such highly symmetric
structures in general do not represent the global minimum on the
ground-state energy hypersurface, although they may closely
The experimental data presented in Section 3 suggest pro-
nounced electronic interactions of CoTPP and FeTPP with a
Ag(111) surface and indicate significant changes in the electronic
structure of the adsorbate system when an NO ligand is
coordinated to the metal center. In this section, we present
theoretical data on the geometric structures, binding energies,
and electronic structure of the adsorbate systems which yield a
more complete characterization and qualitative understanding of
the bonding mechanism and the surface trans effects in these
systems. To achieve this goal, we invoke two kinds of theoretical
model systems which emphasize different aspects of the system.
On the one hand, we present results from periodic density-
functional calculations on the simple metalloporphins (MP, M =
Co, Fe, Zn), which are adsorbed on an ideal Ag(111) surface.
Here, the metal center of the MP molecules has been placed over
a top-site of the Ag(111) surface. In the second class of model
systems, the coordination of a single Ag atom or small cluster to
the MTPP molecules is studied. Both approaches are not
expected to fully and quantitatively predict all the properties
72,86,87
approximate it, depending on the central metal M.
In the doublet ground state of CoTPP, symmetry breaking
87
from D4h symmetry occurs, resulting in a deformation of the
initially planar porphyrin core to a saddle shape or twisted saddle
structure (D2d or S point group symmetry, respectively). The
4
energy differences between the D2d or S structures (about 3 kJ/
4
mol at the BP/TZVP level) must be considered below the
presumed accuracy of the present computational methods,
however, and therefore it is best to regard the MTPPs as
somewhat flexible, not necessarily completely planar structures.
It should be noted that the saddle-shape deformation of the
CoTPP core is induced by the peripheral phenyl rings, and the
DFT calculations predict a planar minimum structure for the
simple, unsubstituted CoP. For the triplet ground state of
FeTPP, the lowest-energy conformation features a slightly
saddle-shaped TPP core as well. The peripheral phenyl groups
are twisted by 69ꢀ with respect to the porphyrin plane, resulting
(
such as absolute moleculeꢀsurface distances) of the experi-
mental adsorbate system. The periodic metalloporphin models
have been chosen such that steric effects are largely eliminated,
and hence these models allow us to study the electronic
in a D2 molecular symmetry. For ZnTPP, we find an almost
d
6
212
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
Figure 10. Optimized geometries of the CoP/Ag(111) and (NO)CoP/Ag(111) adsorbate systems from periodic DFT calculations at the PBEþvdW
level; geometries for the FeP system are similar; cf. Table 1 for details.
Table 1. Structural Parameters of the CoP, FeP and ZnP
Adsorbate Systems Optimized using Periodic Density-Func-
tional Calculations in Angstroms and Degrees
and the CoꢀAg distance is reduced by 0.05 to 2.83 Å
top
(Table 1).
a
Upon coordination of NO to the adsorbed CoP molecule in
trans position to the Ag surface, the CoꢀAg distance increases
significantly (cf. Table 1) beyond that of a covalent or coordi-
native interaction, indicating that the surface covalent bond has
been essentially broken. The computed geometry of the now
merely physisorbed ONꢀCoP complex is very similar to the
computed geometry of free ONꢀCoP without the presence of
structure
d(MꢀAgtop) d(MꢀNO) d(NꢀO) —(MꢀNꢀO)
b
CoPꢀAg(111)
2.88/2.83
b
ONꢀCoPꢀAg(111) >3.5/3.52
ONꢀCoP
1.81
1.80
1.19
1.19
123
123
b
FePꢀAg(111)
3.13/2.88
b
any Ag(111) surface. We estimate the CoꢀAg distance in
top
ONꢀFePꢀAg(111) >3.5/3.60
ONꢀFeP
1.71
1.70
1.19
1.19
145
146
ONꢀCoP Ag(111) to 3.52 Å at the PBEþvdW level, which is
3
3 3
0
.69 Å larger than the corresponding CoꢀAg of CoP adsorbed
b
top
ZnPꢀAg(111)
>3.5/3.30
to Ag(111). The total binding energy between NOꢀCoP and
the Ag(111) surface is reduced to 221 kJ/mol including disper-
sion (only 19 kJ/mol at the PBE level without dispersion
correction). In summary, we conclude from the structural and
energetic data that CoP is subject to a directed electronic
interaction with the Ag(111) surface (top-site), and that this
interaction is largely suppressed upon coordination of NO.
The FeP/Ag(111) system is different from CoP/Ag(111) in
that the chemical bond between the molecule and the surface is
considerably weaker according to our DFT calculations. The
binding energy is calculated to only 29 kJ/mol at the PBE
density-functional level (excluding van der Waals correction).
a
PBE functional; d(MꢀAgtop) refers to the distance of M to the top-site
b
Ag atom of the Ag(111) surface. Value including van der Waals
correction (PBEþvdW level).
planar TPP core with a very modest saddle shape deformation,
and the phenyl rings twisted to an angle of 67ꢀ with respect to the
porphyrin core (D symmetry). More details on the optimized
2
d
ground state geometries are provided in the SI.
A central point in this study is the nature of the coordinative
bond of the central metal ion in MP and MTPP to a Ag surface
and the competition with the coordination of an NO ligand in
trans position. We will first discuss the structural data obtained
from our periodic DFT calculations on the MP/Ag(111) ad-
sorbate systems. Illustrations of the optimized geometries are
shown in Figure 10, and salient structural data are summarized in
Table 1.
The optimized FeꢀAg distance is calculated to 3.13 Å. It is
top
thus considerably larger than the sum of covalent radii
(2.77...2.97 Å) and also significantly longer than the correspond-
ing CoꢀAg distance in CoP/Ag(111). Nevertheless, we find a
top
similar elevation of the top-site Ag atom above the Ag(111) plane
as was found for the more strongly interacting CoP/Ag(111)
system. As expected for such an apparently weak bond, inclusion
of the empirical van der Waals correction (PBEþvdW level)
4
.2. Periodic Systems. CoP adsorbs as a virtually flat mole-
cule, see Figure 10. The total magnetization of the adsorbate
system is zero, i.e., it proves to be a nonspinpolarized system. We
find a CoꢀAg distance of 2.88 Å at the PBE level (Table 1),
results in a smaller FeꢀAg distance of 2.88 Å, which is, perhaps
top
which is in the same range as the sum of the covalent radii of both
top
somewhat surprisingly, similar to the one found for the
88
atoms (CoꢀAg: 2.71...2.95) and thus suggests a covalent (or
coordinative) interaction between the cobalt center and the top-
site Ag atom (Agtop) of the surface. It is interesting to note that
this top-site Ag atom (i.e., the one closest to the Co center) is
slightly lifted above the ideal surface plane toward the Co center
by 0.3 Å with respect to its six neighbors in the same plane of the
relaxed Ag(111) surface. The binding energy is computed to 54
kJ/mol at the PBE level, that is, without inclusion of dispersion
interactions. If van der Waals dispersion is taken into account
CoꢀAg distance in the CoP/Ag(111) system. The mole-
top
cule-surface interaction in FeP/Ag(111) thus appears to be
effected by dispersion forces to an even larger extent than in
the case of CoP/Ag(111). The corresponding total binding
energy is calculated to 244 kJ/mol.
Upon coordination of NO to FeP in trans position to the Ag
surface, the FeꢀAg distance is significantly increased by 0.72 Å
top
to 3.60 Å (PBEþvdW level, i.e., including dispersion), and the
binding energy is reduced to 229 kJ/mol. The calculated binding
energy without taking into account van der Waals dispersion is
(PBEþvdW level), the binding energy increases to 275 kJ/mol,
6
213
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
Table 2. Calculated NO Binding Energies to Free and Ad-
sorbed MP (M = Co, Fe) in Axial Position
0.20e for M = Co and 0.35e for M = Fe. (For more details and
comparison with the other model systems, see the SI.) Charge
transfer from a metal surface to metal complexes has been
observed previously for other surface-confined coordination
a
system
kJ/mol (eV)
system
kJ/mol (eV)
91
CoP
FeP
159 (1.65)
180 (1.86)
CoPꢀAg(111)
FePꢀAg(111)
124 (1.29)
168 (1.75)
compounds.
In conclusion, the higher NO binding energy to surface-bound
Fe(TP)P compared to surface-bound Co(TP)P is thus due to
both the higher intrinsic binding energy of NO to Fe(TP)P on
one hand and due to the weaker surface coordinative bond on the
other hand.
a
Energies in kJ/mol and eV; PBE Density-Functional Level.
computed to 15 kJ/mol at the PBE level. These data indicate the
absence of any covalent or coordinative bond between the NO-
coordinated iron center and the Ag surface. Hence, just as for the
CoP/Ag(111) system, our DFT calculations indicate a strong
trans effect of NO, which essentially turns the chemisorbed FeP
molecule into a ON-FeP complex which is merely physisorbed.
Finally, the experimental data reported above suggest that
ZnTPP essentially does not interact electronically with the Ag
surface. This is confirmed by our density-functional calculations
for the ZnP/Ag(111) system. The computed ZnꢀAgtop distance
is calculated to 3.30 Å at the PBEþvdW level, and is thus
4.3. Interaction of MTPP and (ON)MTPP with a Ag atom.
The experimental data and periodic DFT calculations reported
so far support the view that coordinated NO exerts a pronounced
structural trans effect on adsorbed CoP and FeP. The phenom-
enon of a trans effect, however, has its historical roots in
molecular coordination chemistry. It is therefore interesting to
investigate the trans effect of an NO ligand on a single Ag ligand
in trans position to CoTPP and FeTPP. These coordination
compounds incorporating metalꢀmetal bonds represent our
second type of model systems. They allow us to better quantify
the trans effect in terms of the changes in MꢀAg distances (M =
Co, Fe) induced by the trans-coordination of NO in comparison
to conventional ligands, and provide a simple and transparent
picture of the MꢀAg bonding interaction from a molecular
orbital (MO) interaction perspective. Moreover, we will now
consider the full MTPP molecules instead of the MP models used
markedly greater than the corresponding MꢀAg distances
top
for CoP and FeP.
In summary, our periodic DFT calculations suggest that CoP
interacts more strongly with a Ag(111) surface than FeP, and that
the MꢀAg bonds in both systems are significantly weakened
top
upon NO coordination to MP. The experimental data reported
above in Sections 3.1 and 3.2 suggest that NO coordination is
reversible at elevated temperatures (500 or 600 K, see above),
and that a removal of the NO ligand requires higher temperature
for ONꢀFeTPP/Ag(111) than for ONꢀCoTPP/Ag(111). It is
therefore interesting to consider the energies associated with the
coordination (or dissociation) of the NO ligand to (from) these
systems. The calculated binding energies of NO to MP are
collected in Table 2 for both the gas-phase and the adsorbed MP
molecules.
2
7,48
in our periodic DFT studies. It has been shown before
that
the peripheral phenyl rings can be of relevance in cases where the
conformation of the porphyrins is important.
Nevertheless, one might wonder whether coordination of a
simple Ag atom bears any resemblance to the surface coordina-
tion problem discussed so far. The computed core-level shifts,
which agree qualitatively with experimental data, suggest that this
is indeed the case (see the SI, Table S2). Moreover, the
coordinated Ag atom might be considered, to some approxima-
tion, as a representation of the top-site Ag atom of the (111)
surface which is closest to the metal centers of the adsorbed
metalloporphyrins. Note that a related computational approach
was successfully employed to explain the unusual 3-fold coordi-
nation of carbonitrile ligands to Co centers identified at the same
The data suggest that NO binds stronger to FeP by 21 kJ/mol
than to CoP in the gas phase. For the adsorbed complexes, the
FeꢀNO bond is even more favored: The NO detachment from
FePꢀAg(111) is less favorable by 44 kJ/mol than from
CoPꢀAg(111). In general, NO is found to detach more readily
from the surface-bound complexes as compared to the free
metalloporphyrins. This is because the loss in total bond order,
caused by the release of NO, is in part compensated for by the
reestablished MPꢀAg(111) surface coordinative bond, which is
92
substrate.
Graphical representations of the optimized molecular geome-
tries (computed at the BP/TZVP level) of molecular AgꢀMTPP
model complexes with and without NO are shown in Figure 11.
Structural data of their optimized ground state geometries are
listed in Table 3. (See also refs 71 and 93ꢀ99 for previous results
on NO porphyrin complexes without a Ag atom in trans
position.) Table 4 collects the computed binding energies for
various ligand spheres according to the reactions given in
Figure 9.
Coordination of NO to MTPP. The binding energy of NO to
bare CoTPP is computed to 142 kJ/mol (cf. Table 4, BP/TZVP
level, see Section 2.3); this is reasonably close to but somewhat
lower than the binding energy to CoP obtained from our periodic
DFT calculations reported above (159 kJ/mol, PBE functional).
The lowest-energy structure found for (NO)CoTPP features a
twisted conformation of the porphyrin core which, leaving the
89
stronger for CoP than for FeP (bond order conservation ). This
result is in line with our experimental observations reported
above, which show that NO desorption from FeTPP monolayers
requires higher temperatures than for CoTPP monolayers. The
calculated CoꢀNO and FeꢀNO binding energies of 124 and
1
68 kJ/mol correspond to NO desorption temperatures of 460
and 615 K, respectively, according to an estimate based on the
9
0
13
Redhead equation with an assumed frequency factor of 10
ꢀ
1
s
and a heating rate of 1 K/s. This is in good agreement with
the experimental findings.
70
According to a partial charge analysis, the entire MP moiety
gains 0.37 (M = Co) or 0.22 (M = Fe) electron charges upon
interaction with Ag(111). The charge-transfer from the surface
to the complex is mitigated for the NO-coordinated complexes,
for example, (NO)CoP has a negative charge of only ꢀ0.03 e in
the presence of the surface. In the isolated complexes (NO)MP
NO ligand aside, can be assigned a S point group symmetry. The
4
ground state of the complex is computed to be a spin singlet, in
agreement with the XPS data for (NO)CoTPP multilayers. The
ground state of (NO)FeTPP is found to be a spin doublet (cf.
(M = Co, Fe), that is, in the absence of the surface, the
coordinated NO ligand adopts a negative partial charge of
71
also ref. ). The ONꢀFe bond in (NO)FeTPP (172 kJ/mol; cf.
6
214
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
Table 3. Salient Geometric Parameters (in angstroms and
degrees) from Optimized Geometries of Various MTPP
a
Complexes (M = Co, Fe, Zn)
complex
d(MꢀNO) d(MꢀAg) d(NꢀO) —(MꢀNꢀO)
(
(
(
(
(
(
(
Ag)CoTPP
2.48
ON)CoTPP
ON)(Ag)CoTPP
Ag)FeTPP
1.82
1.86
1.18
1.18
123ꢀ
124ꢀ
2.68
2.54
ON)FeTPP
ON)(Ag)FeTPP
Ag)ZnTPP
1.71
1.72
1.18
1.18
145ꢀ
148ꢀ
2.57
2.77
a
BP/TZVP Level of Theory.
Table 4. Computed NO and Ag Binding Energies (in kJ/mol)
a
in Various MTPP Complexes
Figure 11. Optimized geometries and ligand binding energies of
(
Ag)CoTPP (top, left), (NO)(Ag)CoTPP (top, right), (Ag)FeTPP
complex
NO
Ag
(left, bottom) and (NO)(Ag)FeTPP (right, bottom); BP/TZVP level of
theory.
(
(
(
(
(
(
(
NO)(Ag)CoTPP
NO)(Ag)FeTPP
NO)CoTPP
NO)FeTPP
Ag)CoTPP
81
149
142
172
30
49
1
80 kJ/mol for the periodic calculations on NOꢀFeP) is
stronger (by 29 kJ/mol) and shorter (by 0.11 Å) than the
corresponding bond in (NO)CoTPP. (NO)FeTPP shows a
twisted porphyrin core with the Fe center displaced toward the
NO ligand with respect to an imaginary plane defined by the four
porphyrin N atoms.
92
72
18
Ag)FeTPP
Ag)ZnTPP
a
BP/TZVP level of theory.
Both the (NO)CoTPP and (NO)FeTPP complexes feature a
bent NO ligand. This indicates an NO character of the ligand
ꢀ
38
and is thus in agreement with the experimental results, which show a
transfer of electron density from the metal center to the NO ligand.
occurs upon coordination of Ag. The FeꢀAg bond is slightly
longer (by 0.07 Å) and weaker (by 20 kJ/mol) than the CoꢀAg
bond (Table 4). In general, all AgꢀMTPP bonds are weaker than
the corresponding NOꢀMTPP bonds. In contrast, coordination
of a Ag atom to ZnTPP yields a fairly long ZnꢀAg distance of
2.77 Å and a very low binding energy of only 18 kJ/mol (see
Table 4). This underlines the weakness of the ZnꢀAg interaction
as suggested by the photoemission data reported above.
For the molecular complexes, the same trends in MꢀAg
binding energy and bond distances have been found as for the
periodic systems above. In particular, the binding energy of the
Ag atom or the Ag(111) surface to metalloporphyrins decreases
in the order Co(TP)P > Fe(TP)P > Zn(TP)P. However,
(leaving van der Waals dispersion corrections aside) the absolute
binding energies to a single Ag atom are higher by about 50 kJ/
mol than to a Ag(111) surface, and the MꢀAg bond lengths in
the (Ag)MTPP complexes are shorter than in the surface model
(shortest distance MꢀAg = 2.48 Å vs 2.88 Å for M = Co, and 2.54
Å vs 3.13 Å for M = Fe). These differences meet our expectations
because the single Ag atom is more reactive than a Ag atom with
nine nearest Ag neighbors in a Ag(111) surface.
(
The NO ligand bears a negative partial charge of 0.20e in
(NO)CoP or 0.35e in (NO)FeP, see the SI for details.) The angle
FeꢀNꢀO of 145ꢀ is less acute compared to that found in
NO)CoTPP (123ꢀ). This together with the shorter FeꢀN(O)
distance indicates an increased π-donation from NO to the more
(
95
electron-deficient Fe center. The FeꢀN(O) unit shows a peculiar
tilt of the NO ligand toward one side of the TPP core (angles
(
O)NꢀFeꢀN(por) 87ꢀ/94ꢀ/102ꢀ/103ꢀ), aneffect which recently
100
95
has been observed experimentally and explained theoretically.
From the experimental procedure discussed above, we cannot
exclude the coordination of two NO ligands to MTPP. Additional
DFT calculations for (NO) MTPP show, however, that the
2
binding energy of a second NO ligand in trans position to the
first NO is rather low (24 kJ/mol for M = Co, 47 kJ/mol for M =
90
Fe). According to the Redhead equation, bond energies of 24
and 47 kJ/mol correspond to desorption temperatures below
1
3
1
s
00 and 200 K, respectively (assuming a frequency factor of 10
ꢀ
1
), which means that the second NO ligand should not be
bound under the conditions of our experiment.
Coordination of Ag to MTPP. Next, the interaction of the
single Ag atom with MTPP (M = Co, Fe) is considered.
trans-Coordination of NO and Ag to MTPP. The structural
data obtained from the periodic DFT calculations suggest that
the adsorbed (ON)MP complexes are not covalently bound to
the Ag surface for both M = Co, Fe (physisorption), in contrast
the MP adsorbates without NO. To further clarify how this trans
(
Ag)CoTPP in its lowest-energy conformation (spin-singlet
ground state) shows a saddle-shaped porphyrin core (close to
D symmetry) with the usual twist of the peripheral phenyl rings
2
d
35ꢀ39
(
67ꢀ), see Figure 11. The Ag atom is bound to the Co center at a
effect
analogon affects the chemical bond to Ag, we now
distance of 2.48 Å (cf. Table 3) and with a binding energy of 92
consider the simultaneous coordination of NO and Ag to CoTPP
and FeTPP in the molecular model complexes (NO)(Ag)MTPP
(see Figure 11 and the SI).
The computed geometric parameters of the trans-(NO)-
(Ag)MTPP complexes are collected in Table 3 and Figure 11.
kJ/mol (Table 4, BP/TZVP level). In comparison, for
2
(
Ag)FeTPP (doublet ground state, spin contamination S =
1
.074), a modest symmetry breaking of the porphyrin core
toward a twisted conformation (approximate S symmetry)
4
6
215
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
The CoꢀAg distance in (NO)(Ag)CoTPP (spin-doublet
ground state) is increased from 2.48 to 2.68 Å upon coordination
of the NO ligand, indicating again a pronounced trans influence
of NO on the AgꢀCo bond. The binding energy of the Ag atom
is reduced from 92 kJ/mol in (Ag)CoTPP to only 30 kJ/mol.
Additional calculations included in the SI (Table S7) show that
the structural trans effect exerted by NO on Ag is in fact stronger
ꢀ
than the one exerted on other ligands such as NH , CO, Cl , and
3
ꢀ
NCS . On the other hand, the local coordination geometry of
the NO ligand does not change much upon Ag bonding
compared to (NO)CoTPP (CoꢀNꢀO angle: 124.2ꢀ). We
furthermore note that the porphyrin core shows a pronounced
twist (ruffling) of the pyrrole rings.
The situation isagain somewhatdifferentfor(NO)(Ag)FeTPP,
in which the FeꢀAg distance (2.57 Å) is only marginally longer
than in (Ag)FeTPP (2.54 Å). The dissociation energy of the Ag
atom (49 kJ/mol) is reduced by 23 kJ/mol compared to that of
(Ag)FeTPP (72 kJ/mol, Table 4). (NO)(Ag)FeTPP is predicted
to be a spin singlet, and, like the Co complex, shows a small
deviation of the porphyrin core from planarity (approximate S4
symmetry) with a slight displacement of the Fe center toward the
NO ligand. The FeꢀNꢀO angle is similar (148ꢀ) as that in
(
(
NO)FeTPP (145ꢀ), and is thus less acute than in (NO)CoTPP
123ꢀ). This agrees well with expectations from the Enemark-
Figure 12. Qualitative orbital energy interaction diagram for the
interaction of Ag and CoTPP in the complex (Ag)CoTPP and illustra-
tions of the bonding and antibonding MOs; orbital energies of the
CoTPP and Ag fragments have been shifted for illustration purposes; Co
and Ag d levels which are not involved in the bonding are shaded. For the
101
Feltham electron count. The AgꢀFeꢀN(O) unit is not linear
but shows an angle of 166ꢀ in the optimized geometry (cf. the
corresponding angle of 179ꢀ in (NO)(Ag)CoTPP). This again
reflects the peculiar tilt of the NO ligand toward one side of the
TPP core (angles (O)NꢀFeꢀN(por) 88ꢀ/89ꢀ/103ꢀ/103ꢀ)
(Ag)CoTPP complex, levels not directly involved in the interaction have
95,100
been omitted for clarity but can be found in the SI.
which is also observed in (NO)FeTPP.
All in all, our
calculations suggest that Ag and NO ligands do not significantly
affect their mutual coordination geometry in FeTPP, unlike the
situation for CoTPP.
for the binding of the NO or Ag ligands to MTPP. (Detailed
listings of computed orbital energies can be found in the SI).
Electronic Structure of (Ag)MTPP. Figure 12 shows a simpli-
fied orbital interaction diagram of the most important contribu-
tions to the MꢀAg bond in (Ag)MTPP (see Table S6 of the SI
for more details). The left-hand side of the Figure shows the
orbital energy levels of the CoTPP orbitals (open-shell spin
doublet, R is majority spin) with predominant d character, and
the right-hand side shows the s-type orbital energy levels of the
Ag atom (4d 5s electron configuration). The AgꢀCoTPP
interaction can thus be characterized as a simple two electron,
four spinꢀorbital interaction (or two-electron, two orbital
interaction in closed-shell terminology) built by the Ag(5s)
(spin) orbital(s) and the M(dz2) (spin) orbital(s). A similar
analysis for (Ag)FeTPP leads to an interaction diagram which
looks slightly more complicated due to the open-shell character
of (Ag)FeTPP but which leads to the same qualitative inter-
pretation, see the SI for details (Figure S5). In summary, the
analysis proves a direct, covalent orbital interaction between the
metal ion in CoTPP or FeTPP and the Ag ligand, thus giving
support for our interpretations of the calculated structural data
and the spectroscopic data reported above.
Electronic Structure. The relevant features of the electronic
structure of MTPP (M = Co, Fe) are briefly discussed as a
prequel to the analysis of the interaction with Ag and NO. For
convenience, the molecular orbitals will be labeled according to
the D4h (CoTPP) or D2h (FeTPP) point groups, even if the
underlying geometry does not have the full symmetry indicated
by these point groups. Furthermore, only those frontier orbitals
with significant coefficients for the d-type atomic orbitals of the
transition metal center M (i.e., Co or Fe) will be discussed,
because these are responsible for the binding of the axial ligands
10
1
(
Ag and NO) to MTPP.
According to the present calculations on the doublet ground
state of CoTPP, the degenerate set of Co d , d atomic orbitals
xz yz
(
AOs) labeled 12e represent the (global) HOMO in the
g
minority spin channel β (at ꢀ4.68 eV; majority (R) component
is HOMO-4 at ꢀ5.07 eV). The d AO 25a is singly occupied
z2
1 g
in CoTPP. This orbital is the LUMO in the β spin channel
(
ꢀ3.76 eV), while the corresponding R spin component is found
at lower energy (HOMO-10, ꢀ6.26 eV). The d
spin orbitals
x2-y2
are both unoccupied. The remaining d orbitals dominantly
participate in occupied MOs and are found within a similar
energy range (see the SI for details). In contrast to CoTPP, the
Electronic Structure of (NO)(Ag)MTPP. The electronic struc-
ture of (NO)(Ag)MTPP reveals that the covalent part of the
AgꢀMTPP bond is significantly weakened or virtually disap-
pears if the NO ligand coordinates to the metal center. From
dz AO (labeled 36a ) of the Fe center in the triplet ground state
2
g
of FeTPP is doubly occupied, the β component being the
HOMO (ꢀ3.89 eV) while the R component is the HOMO-2
7
1,97,101
previous work on nitrosyl metalloporphyrin complexes,
it
(
ꢀ4.48 eV). The d /d AOs (12b , 12b ) are half-occupied,
is known that the MꢀNO bond can be decomposed into π- and
σ-type interactions mediated by unoccupied π* orbitals of NO
and metal d-orbitals of appropriate local symmetry. On the other
hand, we have shown above that the MꢀAg interaction in
xz yz
2g
3g
the R spin component being occupied (ꢀ5.11 eV), while the β
components represent the LUMO and LUMOþ1 (ꢀ3.30 eV).
Further below, these sets of orbitals are shown to be responsible
6
216
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
Figure 13. Selected molecular orbitals for the complex
Figure 14. Selected molecular orbitals for the complex
(NO)(Ag)CoTPP; orbitals with minor or no Ag(5s) contribution are
(NO)(Ag)FeTPP.
shaded.
atoms take those positions at which they are found in the
optimized geometry of (NO)(Ag)CoTPP. The main differ-
ence to the relaxed structure of (Ag)CoTPP is thus the much
longer AgꢀCo distance in the unrelaxed structure (Δd = 0.21
Å, cf. Table 3). Inspection of the corresponding (relaxed) MOs
shows that, in spite of the long AgꢀCo distance, a similar
(
Ag)MTPP is built by M(3dz ) and Ag(5s) orbitals which have
2
local σ-symmetry with respect to the MꢀAg axis. To obtain
insight into the effect of NO coordination on this bond, we
therefore focus our analysis to those molecular orbitals which are
(
predominantly) composed of these M(3dz ) and Ag(5s) orbitals
2
and the π* orbital of NO which lies in the plane of the ONꢀM
Co(d
2
)ꢀAg(5s) orbital interaction is established (albeit with
z
fragment and therefore is capable of forming a σ-bond with the
less mixing, of course; data not shown). This proves that it is
indeed the presence of the NO ligand and the resulting
competition for the frontier orbitals of Co which weakens
M(3dz ) orbital (see Figure 13).
2
Inspection of the molecular orbitals in (NO)(Ag)CoTPP
some of which are illustrated in Figure 13) shows that almost
(
the Co(dz
2
)ꢀAg(5s) orbital interaction. If, in contrast, the Ag
no direct CoꢀAg orbital interaction is discernible in the presence
of a trans-coordinated NO ligand any more, in contrast to the
situation in (Ag)CoTPP. A closer analysis reveals that the lowest
ligand is deleted from (NO)(Ag)CoTPP without relaxing the
structure, the orbital interaction pattern describing the bond-
ing of NO to CoTPP is virtually unchanged compared to that
of the fully relaxed (NO)CoTPP. This is in line with the above
analysis of CoꢀNO binding in the optimized geometry of
(NO)(Ag)CoTPP.
(occupied) molecular orbital of the subset shown in Figure 13
represents the σ-type CoꢀNO interaction and bears only a small
contribution (about 7%) of the Ag(5s) orbital, which had been
crucial for the covalent CoꢀAg bond in (Ag)CoTPP. The main
contribution of the Ag(5s) orbital is found in two spin orbitals,
one of which is occupied while the other is unoccupied (see
Figure 13). However, neither of these orbitals contains a
significant contribution from an orbital located on the Co center.
In other words, the interaction of the Ag(5s) orbital with the
In the iron porphyrin complex (NO)(Ag)FeTPP, the NO
induced changes in FeꢀAg bond length are less pronounced, as
outlined above. This fact is also reflected by the corresponding
orbital interactions, which are sketched in Figure 14. In contrast
to the situation with M = Co, the Fe(3d)ꢀAg(5s) interaction
remains partly intact along with the NOꢀFe(3d) σ-type inter-
action, as shown by the doubly occupied molecular orbitals in
Co(3dz
Ag)CoTPP, has now essentially vanished, and the Co(3dz
2) orbital, which was responsible for the bonding in
103
(
2)
Figure 14.
orbital is largely involved in a σ-type interaction with one of the
NO π* orbitals. Please note that the CoꢀNO bonding mechan-
ism (not shown here) is largely unaffected by the presence of Ag.
This observation is in line with the fact that only a fairly small
change in the NOꢀCo bond distance occurs when the Ag binds
to (NO)CoTPP (Δd = 0.04 Å, cf. Table 3).
In conclusion, the suppression of the CoꢀAg orbital interac-
tion in the presence of an NO ligand can be held responsible for
the prolonged CoꢀAg distance in (NO)(Ag)CoTPP (Table 3).
However, one might argue that such a long distance, howsoever
caused, may by itself prevent a possible direct orbital interaction
between Co and Ag. To exclude this possibility, the NO ligand
was deleted from (NO)(Ag)CoTPP without relaxing the geo-
metry, that is, a complex (Ag)CoTPP was analyzed where all
4.4. Interaction of CoTPP and (NO)CoTPP with Ag19 and
Ag72 Clusters. The simple (NO)(Ag)MTPP model systems
provide an illustrative MO picture of the bonding mechanism
and explain the competition between the trans ligands, but a
single Ag atom cannot represent all properties of a Ag surface.
The AgꢀM bond in the model systems is likely to be shorter than
the actual distance MꢀAg(111), due to the increased reactivity
of the Ag atom and the neglect of steric repulsions between the
TPP framework and the surface. In addition, a single Ag atom is
expected to possess a different electron donor capacity as an
ensemble of Ag atoms.
To alleviate these limitations and to establish a link between
the periodic DFT calculations and the studies on
(NO)(Ag)MTPP model systems, another class of model systems
6
217
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
Figure 15. Optimized structures of (a) (Ag19)CoTPP, (b) (NO)(Ag19)CoTPP, (c) (Ag72)CoTPP, (d) (NO)(Ag72)CoTPP ; PBE/SV(P) level.
is introduced, in which the Ag(111) surface is approximated by
clusters with 19 and 72 Ag atoms. These more complex (and
computationally more demanding) model systems may provide a
more realistic description of the interaction with a Ag(111)
surface but have the drawback that an interpretation of the results
is less straightforward than for the (NO)(Ag)MTPP models. We
thus restrict ourselves here to a qualitative discussion of the
structural changes induced by NO coordination to the
Table 5. Salient Geometric Parameters (in angstroms and
degrees) from Optimized Geometries of (Ag )CoTPP and
(NO)(Ag )CoTPP Adsorbate Cluster Models, n = 1, 19, 72
n
a
n
model complex
d(MꢀN(O)) d(MꢀAg) d(NꢀO) —(MꢀNꢀO)
(Ag)CoTPP
2.48
2.80
(
(
(
(
(
Ag19)CoTPP
Ag72)CoTPP
3.6ꢀ3.7
2.68
3.37
(
Ag )CoTPP (n = 19, 72) model systems, in which the Co ion
n
ON)(Ag)CoTPP
ON)(Ag19)CoTPP
ON)(Ag72)CoTPP
1.86
1.81
1.81
1.18
1.17
1.17
124ꢀ
123ꢀ
123ꢀ
is located on top of the central Ag atom in the Ag19 cluster and
above a hollow site in the Ag72 cluster, respectively.
>4
Optimized structures of (Ag )CoTPP and (NO)(Ag )-
n
n
a
PBE/SV(P) level of theory (correction for van der Waals dispersion
CoTPP (n = 19, 72) are depicted in Figure 15, and salient
structural parameters are collected in Table 5 (see the SI for more
details). The computed distance between Co and the nearest Ag
not included).
1
04
5. DISCUSSION
atom(s)
increases with the cluster size, thus indicating a
weaker molecule-cluster interaction for the bigger clusters. For
the Ag72 cluster, the adsorbed CoTPP molecule undergoes a
saddle-shape distortion of the porphyrin core, accompanied by
rotation of the peripheral phenyl rings of TPP toward the
molecular plane (cf. the SI for details on the geometry.) This
can be explained by steric repulsions of these phenyl rings and the
Ag surface. Our computational results agree with a combined
STM and NEXAFS study of CoTPP on Cu(111), in which the
CoTPP molecule was found to acquire a saddle-shaped con-
This study focuses on the question how the chemical bond of a
coordinated metal center to a surface is influenced by additional
ligands in trans position to the surface. Previous theoretical
analysis of molecular complexes shows that it is energetically
unfavorable for trans ligands to share the same orbitals of the
38,106
central metal ion.
In general, it was found that σ- and π-
bonding effects must be considered for a full mechanistic under-
standing. In a complex with the trans-positioned ligands L and
S
LW, both ligands donate electron density to the metal center M
105
formation in the adsorbed state. Similar observations were
via a σ bond. If L is a stronger donor that L , then the σ-bond
S
W
27,32,33
made for CoTPP on Ag(111).
MꢀL will be weakened (L fM LW). The strong structural
3 3 3
ꢀ ꢀ
W
S
Upon NO coordination to the Co center, the CoꢀAg distance
trans effects of H and R , which are exclusively σ donor ligands,
are attributed to this mechanism (σ trans effect). In addition, M
can donate electron density from occupied d levels into empty
(usually non- or antibonding) ligand orbitals via a π-type
interaction. This back-donation increases the bond strength of
the ligand that is the stronger acceptor at the expense of the MꢀL
increases significantly. The relative increase in this distance is
larger for the bigger clusters than for the Ag model (þ8% for
1
Ag , þ20% for Ag , see Table 5, cf. also Table 1). This is
1
19
probably due to the fact that for the bigger clusters the interaction
between Co and Ag becomes weaker in general and is therefore
even less capable of competing with the trans-coordinated NO,
the binding of which to Co is largely unchanged by the Ag cluster.
For the biggest cluster model (NO)(Ag )CoTPP, just as for the
3
9
bond of the other (weaker) ligand (L rM L ). In nitrosyl
S
3 3 3
W
complexes (NO)ML , the NO ligand can bind in a linear or a
n
72
bent geometry. If it adopts a bent bonding mode, as is the case for
the systems investigated here (see Section 4), it is a very strong σ
donor, but also engages in π-backbonding and is known to exert a
particularly strong structural trans effect on a wide range
periodic calculations on Ag(111), the Co center is pulled away
from the Ag cluster by the coordinated NO (cf. the SI for details
on the computed geometry). These results on finite cluster
models confirm again the strong trans effect of NO on the
interaction of CoTPP with the Ag substrate.
38
of ligands in trans-position. An NO-related trans effect has
previously also been observed on metalloporphyrin-nitrosyl
6
218
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
Table 6. NO Induced Changes of Bond Length, Bond Energy
and XPS Co(2p3/2) Core Level Shifts for the CoTPP System
(M = Fe, Co) changes dramatically upon NO coordination.
(Some illustrative DFT results and experimental data for the
(
NO)(Ag)CoTPP system are summarized in Table 6.) The
(Ag)CoTPP (NO)(Ag)CoTPP
Δ
magnitude of this trans effect also depends on the nature of the
d
CoꢀAg in Å
CoꢀAg in kJ/mol
2.48
92
2.68
30
þ0.20
complexed metal center M in MTPP.
In line with the NO-induced reduction of the MꢀAg bond
energies, the respective bond lengths increase. The periodic calcula-
E
ꢀ62
a
Co(2p3/2) XPS shifts in eV exp. ꢀ1.8
ꢀ0.4
ꢀ0.2
b
tions of MP on Ag(111) show that the MꢀAg distances are
top
theor.
ꢀ0.6
a
increased upon NO coordination and indicate a physisorption
situation. The molecular model complexes, which allow for a more
precise quantification of the structural trans effect, show that the
effect of NO coordination is larger for the AgꢀCo (þ0.20 Å) than
for the AgꢀFe bond (þ0.03 Å). These values agree with data for
other Co and Fe nitrosyl complexes with bent NO ligands, where
thebondlengthto varioustrans ligands increases between 0.04and
0.29 Å (see ref 38 and Table S7 of the SI).
Relative to the CoTPP or (NO)CoTPP without the Ag ligand
multilayers of the complexes in the experiment and CoTPP or
NO)CoTPP molecule in the calculation). Slater transition state
calculations; cf. Table S2 in the SI for more detail.
(
(
b
complexes, in which the NO ligand influences amino acid ligands
coordinated to metalloporphyrins; this effect is very important
for biological signaling processes.
71,95
The calculations also illustrate the mutual nature of the trans
effect: While the NO ligand weakens the MꢀAg bond, the Ag
atom also weakens the MꢀNO bond, but to a lesser extent,
because the Ag atom is the inferior trans ligand. For the molecular
model complexes, the MꢀNO binding energy is reduced by
ꢀ43% for M = Co and ꢀ13% for M = Fe upon coordination of
the Ag atom.
On a phenomenological level, the concept of the trans
3
5ꢀ39
effect
chemistry.
has only recently been applied by us to surface
3
0,34
From the point of view of an adsorbed metal
complex, the surface which interacts with the metal center can
be considered as an unusually bulky ligand with a “cone
angle”
1
07
of 180ꢀ. As such, it can influence the electronic
properties of the complexed metal ion, as has been shown in
Section 3 for Fe and Co porphyrins: The interaction of the
coordinated Fe and Co ions with the Ag substrate results in
substantial Fe(2p) and Co(2p) core-level shifts and the occur-
rence of new valence states in the UP and ST spectra, which
reveal the covalent character of this interaction. In contrast, no
indications for bond formation were found for Zn porphyrin.
As shown above, coordination of NO to the Co and Fe centers
leads to an almost complete suppression of the electronic
interaction with the surface: the core-level shifts are reversed
and the new valence states disappear. These observations can
be interpreted as manifestations of the trans effect with NO as
the stronger and Ag as the weaker ligand. The process of
formation and cleavage of the substrate-metal bond is fully
reversible: if the NO ligands are removed by thermal desorp-
tion, the interaction between surface and metal center is
restored. Furthermore, in agreement with the expectation of
CO ligands showing a weak trans effect, recent findings
revealed that the surface bonding of CoTPP and FeTPP
adsorbed on Ag(111) or Cu(111) is only weakly affected by
The frontier orbitals of (Ag)MTPP (Figures 12 and S5) reveal
that the AgꢀM bond is a covalent two-electron, two-orbital
interaction which involves the Ag(5s) and the M(d ) orbitals.
2
z
Upon coordination of NO (Figures 13, 14), this bond is
suppressed or weakened, and the Ag(5s) orbitals become largely
nonbonding for the case of CoTPP. Instead of interacting with
the Ag(5s) orbital, the M(dz ) orbital engages in a σ-type bond to
2
the NO π * orbital, combined with a π-type interaction between
x
the NO π* and the Co(d /d ) orbitals. In (NO)(Ag)FeTPP,
xz yz
the Ag(5s) orbital is still involved in the FeꢀAg bond, but to a
lesser extent than in the absence of NO.
Additional calculations on finite cluster model systems (Ag19
and Ag ) further support our interpretation of the experimental
72
data. For (Ag )CoTPP, a CoꢀAg bond length of 2.8 Å is
19
calculated, which is between the lengths of AgꢀAg (2.90 Å) and
the CoꢀCo (2.52 Å) bond and thus in the range of typical
covalent or metallic bonds. Upon coordination of NO to the Co
ion, the CoꢀAg bond length increases by 20% to 3.37 Å, proving
substantial weakening of this bond.
The calculations predict that CoTPP interacts more strongly
with Ag than FeTPP, while the interaction of NO with FeTPP is
stronger than that with CoTPP. This agrees with experimental
results, which show that the decomposition temperature of
adsorbed (NO)FeTPP is higher than that of adsorbed
(NO)CoTPP and that NO coordination to a FeTPP monolayer
leads to an larger M(2p) XPS shift than for the CoTPP
monolayer (þ2.7 vs þ2.2 eV).
1
08
carbonyl ligation.
Detailed mechanistic insight into the surface trans effect is
provided by our DFT calculations of the coordination of NO to
Co and Fe porphyrins both with and without a Ag atom, a Ag
cluster or a Ag(111) surface in trans position. The different
model systems emphasize complementary aspects of the effect.
In particular, the seemingly simplistic model bearing just a single
Ag atom provides a direct link to molecular coordination
chemistry and allows for a conceptual, qualitative interpretation
of the local bonding situation. (Note that coordinative me-
talꢀmetal bonds between a central transition metal ion and
metal-containing ligands are not unusual and occur for example
The validity of our model systems is confirmed by the
computed core level shifts (see Table 6 and the SI, Table S2),
which are in qualitative agreement with the XPS data. All models
predict a reduction in Co(2p) binding energies when CoTPP
binds to the Ag atom or Ag(111) surface, in agreement with the
experimental XPS shift between multilayer and monolayer.
(Expectedly, the calculated shift is smaller, because a single Ag
atom has less capacity to influence the Co ion than the Ag(111)
surface. This holds both for initial state effects, for example,
electron donation, and final state effects, for example, core hole
screening.) Most importantly, the experimental finding that NO
coordination counteracts the influence of the Ag ligand and leads
109
in stannylene complexes, for which also trans effects have been
1
10
reported. ) Here, the Ag atom can be envisaged as the simplest
approximation of a top-site Ag atom in a real surface. The
calculations show that NO binds in the bent mode in all cases
and is thus likely to exert a very strong trans effect. In agree-
ment with this expectation, the strength of the bond between
the Ag atom, cluster, or surface and the transition metal ion
6
219
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
to an increase of the Fe(2p) and Co(2p) binding energies is
reproduced by the calculations. The computed core level shifts,
bond lengths and binding energies thus give support to the
interpretation deduced from the XPS experiments, namely,
that the interaction of the coordinated metal center with the
Ag substrate can be significantly reduced due to the co-
ordination of NO.
’ ASSOCIATED CONTENT
Supporting Information. Additional UP and N(1s) XP
spectra, UP signal positions, additional information on the
computed structures, electronic structure, and core-level shifts.
This material is available free of charge via the Internet at http://
pubs.acs.org.
S
b
’
AUTHOR INFORMATION
6
. SUMMARY AND CONCLUSIONS
Corresponding Author
Adsorption of iron(II) and cobalt(II) tetraphenylporphyrin
Dr. Michael Gottfried, Lehrstuhl f €u r Physikalische Chemie II,
Department Chemie and Pharmazie, Universit €a t Erlangen-
N €u rnberg, Egerlandstr. 3, 91058 Erlangen, Germany. Tel. þ49-
(
FeTPP, CoTPP) on Ag(111) results in the formation of a
covalent bond between the coordinated metal center and the Ag
surface. For zinc(II) tetraphenylporphyrin (ZnTPP), no bond
formation was observed, indicating that the bond strength
depends on the electronic structure of the metal center. Co-
ordination of a nitrosyl (NO) ligand to the metal centers of
FeTPP and CoTPP leads to a reversible suppression of the bond
to the substrate. This was concluded from XPS, UPS and STM/
STS experiments and confirmed by DFT calculations on molec-
9
131-8527320. Fax þ49-9131-8528867. E-mail: michael.gott-
fried@chemie.uni-erlangen.de. Dr. Wolfgang Hieringer, Theore-
tische Chemie, Department Chemie and Pharmazie, Universit €a t
Erlangen-N €u rnberg, Egerlandstr. 3, 91058 Erlangen, Germany.
Tel. þ49-9131-85-28646. Fax þ49-9131-85-27736. E-mail: hier-
inger@chemie.uni-erlangen.de
ular and periodic model systems of the type (NO)(Ag )M(TP)P
n
(
n = 1, 19, 72, ¥). Specifically, the adsorption of the Co and Fe
’
ACKNOWLEDGMENT
porphyrins on the Ag substrate leads to new valence states in
UPS and STS and to substantial core-level shifts in XPS. DFT
reveals that the M� Ag bond can locally be described as a
covalent two-orbital, two-electron bond between the Ag(5s)
Financial support by the Deutsche Forschungsgemeinschaft
through Sonderforschungsbereich 583 and the Cluster of Ex-
cellence “Engineering of Advanced Materials” and by the ERC
Advanced Grant MolArt (247299) and TUM-IAS is gratefully
acknowledged. We thank W. G. Schmidt for sharing with us his
van der Waals DFT module for the VASP program.
and M(3dz ) orbitals. Upon coordination of NO, which adopts
2
a bent geometry, this bonding interaction is suppressed and the
Ag(5s) orbitals become nonbonding in CoTPP and weakly
bonding in FeTPP. Structural data obtained from the periodic
and nonperiodic DFT calculations show that the M� Ag bond
length increases upon NO coordination.
For a conceptual understanding of these effects, we have
applied principles of chemical bonding in transition metal
complexes to the field of surface-confined coordination
’ REFERENCES
(1) Hulsken, B.; Van Hameren, R.; Gerritsen, J. W.; Khoury, T.;
Thordarson, P.; Crossley, M. J.; Rowan, A. E.; Nolte, R. J. M.; Elemans, J.
A. A. W.; Speller, S. Nat. Nanotechnol. 2007, 2, 285–289.
(2) Mochida, I.; Suetsugu, K.; Fujitsu, H.; Takeshita, K. J. Phys.
Chem. 1983, 87, 1524–1529.
3
0,111
chemistry.
In particular, the concept of the trans effect
proved suitable for the qualitative interpretation of the competi-
tion between the bonds of the NO ligand, and silver substrate, to
the transition metal centers. The similarity of the results obtained
from molecular and periodic model systems emphasizes the close
relationship of the classical trans effect and the surface trans effect,
which we have analyzed here in detail. This effect has important
implications for the potential catalytic activity or sensor func-
tionality of adsorbed metal complexes. Our results show that the
substrate surface can sensitively influence the electronic structure
and, thus, the reactivity of the metal centers. The strength of this
substrate influence, however, is also controlled by the additional
ligands on the metal center, that is, by potential reactant
molecules. Depending on the relative strengths of the trans
influence (the trans activity of the ligands), either the substrate
or, as is the case for the systems described here, the additional
ligand can dominate the electronic properties of the metal center.
On the basis of this systematic insight, the interaction between
the metal center and the surface can be adjusted with respect to a
desired reactivity. This approach mimics, to some extent, enzy-
matic systems such as heme-thiolate proteins, in which an axial
thiolate ligand controls the reactivity of the metal center. Our
findings illustrate how the electronic state (including the spin
state and thus the magnetic properties) and the reactivity of an
adsorbed, complexed metal center results from a delicate inter-
play between the influences of the surface and the coordinated
reactant on this metal center.
(
3) Berner, S.; Biela, S.; Ledung, G.; Gogoll, A.; Backvall, J. E.; Puglia,
C.; Oscarsson, S. J. Catal. 2006, 244, 86–91.
4) Hassan, S. A.; Hassan, H. A.; Hashem, K. M.; Dayem, H. M. A.
(
Appl. Catal., A 2006, 300, 14–23.
(5) Ferreira, A. D. Q.; Vinhado, F. S.; Iamamoto, Y. J. Mol. Catal. A
2006, 243, 111–119.
(6) Mochida, I.; Suetsugu, K.; Fujitsu, H.; Takeshita, K. J. Catal.
1
982, 77, 519–526.
7) Mochida, I.; Tsuji, K.; Suetsugu, K.; Fujitsui, H.; Takeshita, K.
J. Phys. Chem. 1980, 84, 3159–3162.
(
(
(
8) Rakow, N. A.; Suslick, K. S. Nature 2000, 406, 710–713.
9) Gulino, A.; Gupta, T.; Mineo, P. G.; van der Boom, M. E. Chem.
Commun. 2007, 4878–4880.
(
10) Takulapalli, B. R.; Laws, G. M.; Liddell, P. A.; Andreasson, J.;
Erno, Z.; Gust, D.; Thornton, T. J. J. Am. Chem. Soc. 2008,
130, 2226–2233.
(11) Andringa, A. M.; Spijkman, M. J.; Smits, E. C. P.; Mathijssen,
S. G. J.; van Hal, P. A.; Setayesh, S.; Willard, N. P.; Borshchev, O. V.;
Ponomarenko, S. A.; Blom, P. W. M.; de Leeuw, D. M. Org. Electron.
2
010, 11, 895–898.
12) Di Natale, C.; Monti, D.; Paolesse, R. Mater. Today 2010,
3, 37–43.
13) Scudiero, L.; Barlow, D. E.; Mazur, U.; Hipps, K. W. J. Am.
Chem. Soc. 2001, 123, 4073–4080.
14) Hipps, K. W.; Scudiero, L.; Barlow, D. E.; Cooke, M. P. J. Am.
(
1
(
(
Chem. Soc. 2002, 124, 2126.
(15) Barlow, D. E.; Scudiero, L.; Hipps, K. W. Langmuir 2004,
20, 4413–4421.
6
220
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
(
16) Katsonis, N.; Vicario, J.; Kudernac, T.; Visser, J.; Pollard, M. M.;
Feringa, B. L. J. Am. Chem. Soc. 2006, 128, 15537–15541.
17) Auw €a rter, W.; Weber-Bargioni, A.; Riemann, A.; Schiffrin, A.;
Gr €o ning, O.; Fasel, R.; Barth, J. V. J. Chem. Phys. 2006, 124, 194708.
18) Buchner, F.; Comanici, K.; Jux, N.; Steinr €u ck, H.-P.; Marbach,
H. J. Phys. Chem. C 2007, 111, 13531–13538.
19) Buchner, F.; Schwald, V.; Comanici, K.; Steinr €u ck, H.-P.;
Marbach, H. ChemPhysChem 2007, 8, 241–243.
20) Buchner, F.; Flechtner, K.; Bai, Y.; Zillner, E.; Kellner, I.;
Steinr €u ck, H.-P.; Marbach, H.; Gottfried, J. M. J. Phys. Chem. C 2008,
12, 15458–15465.
21) Sekiguchi, T.; Wakayama, Y.; Yokoyama, S.; Kamikado, T.;
(47) Auw €a rter, W.; Weber-Bargioni, A.; Brink, S.; Riemann, A.;
Schiffrin, A.; Ruben, M.; Barth, J. V. ChemPhysChem 2007, 8, 250–254.
(48) Buchner, F.; Seufert, K.; Auw €a rter, W.; Heim, D.; Barth, J. V.;
Flechtner, K.; Gottfried, J. M.; Steinr €u ck, H.-P.; Marbach, H. ACS Nano
2009, 3, 1789–1794.
(49) Auw €a rter, W.; Schiffrin, A.; Weber-Bargioni, A.; Pennec, Y.;
Riemann, A.; Barth, J. V. Int. J. Nanotechnol. 2008, 5, 1171–1193.
(50) www.createc.de.
(51) Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and
Molecules; Oxford University Press: New York, 1989.
(52) Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density
Functional Theory, 1st ed.; Wiley-VCH: Weinheim, 2000.
(53) Kresse, G; Hafner, J. Vienna Ab Initio Simulation Package
(VASP), version 4.6; Institut f €u r Materialphysik, Universit €a t Wien:
Vienna, since 1991; www.vasp.at.
(
(
(
(
1
(
Mashiko, S. Thin Solid Films 2004, 464ꢀ65, 393–397.
(
22) Nilson, K.; Åhlund, J.; Brena, B.; G €o thelid, E.; Schiessling, J.;
ꢁ
Martensson, N.; Puglia, C. J. Chem. Phys. 2007, 127, 114702.
(
23) Åhlund, J.; Schnadt, J.; Nilson, K.; G €o thelid, E.; Schiessling, J.;
(54) Ahlrichs, R.; et al. Turbomole, version 5.7; University of
Karlsruhe: Karlsruhe, since 1988; www.turbomole.com.
(55) Baerends, E. J.; et al. Amsterdam Density-Functional Package
(ADF), version 2007, Vrije Universiteit Amsterdam, Amsterdam, since
1978; www.scm.com.
ꢁ
Besenbacher, F.; Martensson, N.; Puglia, C. Surf. Sci. 2007,
6
01, 3661–3667.
24) Cheng, Z. H.; Gao, L.; Deng, Z. T.; Jiang, N.; Liu, Q.; Shi, D. X.;
Du, S. X.; Guo, H. M.; Gao, H.-J. J. Phys. Chem. C 2007, 111, 9240–9244.
(
(
(
25) Walzer, K.; Hietschold, M. Surf. Sci. 2001, 471, 1–10.
26) Gopakumar, T. G.; Lackinger, M.; Hackert, M.; M €u ller, F.;
(56) Perdew, J. P.; Burke, K.; Ernzerhoff, M. Phys. Rev. Lett. 1996,
77, 3865.
Hietschold, M. J. Phys. Chem. B 2004, 108, 7839–7843.
27) (a) Buchner, F.; Warnick, K.-G.; W €o lfle, T.; G €o rling, A.;
Steinr €u ck, H.-P.; Hieringer, W.; Marbach, H. J. Phys. Chem. C 2009,
13, 16450–16457. (b) Seufert, K.; Auw €a rter, W.; Barth, J. V. J. Am.
Chem. Soc. 2010, 132, 18141–18146.
28) Buchner, F.; Kellner, I.; Hieringer, W.; G €o rling, A.; Steinr €u ck,
H.-P.; Marbach, H. Phys. Chem. Chem. Phys. 2010, 12, 13082.
29) Auclair, K.; Moenne-Loccoz, P.; Ortiz de Montellano, P. R.
J. Am. Chem. Soc. 2001, 123, 4877–4885.
(57) Bl €o chl, P. E. Phys. Rev. B 1994, 50, 17953.
(
(58) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758.
(59) Ortmann, F.; Bechstedt, F.; Schmidt, W. G. Phys. Rev. B 2006,
73, 205101.
(60) Becke, A. D. Phys. Rev. A 1988, 38, 3098.
(61) Perdew, J. P. Phys. Rev. B 1986, 33, 8822.
1
(
(62) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005,
7, 3297–3305.
(
€
(63) Eichkorn, K.; Treutler, O.; Ohm, H.; H €a ser, M.; Ahlrichs, R.
(30) Gottfried, J. M.; Marbach, H. Z. Phys. Chemie 2009, 223, 53–74.
Chem. Phys. Lett. 1995, 240, 283–290.
(31) Lukasczyk, T.; Flechtner, K.; Merte, L. R.; Jux, N.; Maier, F.;
(64) Sierka, M.; Hogekamp, A.; Ahlrichs, R. J. Chem. Phys. 2003,
Gottfried, J. M.; Steinr €u ck, H. P. J. Phys. Chem. C 2007, 111, 3090–3098.
32) Bai, Y.; Buchner, F.; Kellner, I.; Schmid, M.; Vollnhals, F.;
Steinr €u ck, H.-P.; Marbach, H.; Gottfried, J. M. New J. Phys. 2009,
1, 125004.
33) Auw €a rter, W; Seufert, K; Klappenberger, F.; Reichert, J.;
Weber-Bargioni, A.; Verdini, A.; Cvetko, D.; Dell’Angela, M.; Floreano,
L.; Cossaro, A.; Bavdek, A.; Morgante, A.; Seitsonen, A. P.; Barth, J. V.
Phys. Rev. B 2010, 81, 245403.
118, 9136.
(
(65) te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca
Guerra, C.; van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. J. Comput.
Chem. 2001, 22, 931–967.
(66) Slater, J. C. Adv. Quantum Chem. 1972, 6, 1–92.
(67) Birgersson, M.; Almbladh, C.-O.; Borg, M.; Andersen, J. N.
Phys. Rev. B 2003, 67, 045402.
(68) Egelhoff, W. F. Surf. Sci. Rep 1987, 6, 253–415.
(69) Johansson, B.; Martensson, N. Phys. Rev. B 1980, 21, 4427.
(70) (a) Bader, R. F. W. Atoms in Molecules: A Quantum Theory;
Oxford University Press: New York, 1990. (b) Tang, W.; Sanville, E.;
Henkelman, G. J. Phys.: Condens. Matter 2009, 21, 084204.(c) The bader
program by Arnaldsson, A.; Tang, W.; , Henkelman, G.; , version 0.27a,
has been used to perform the partial charge analysis; http://theory.cm.
utexas.edu/vtsttools/bader.
1
(
(
34) Flechtner, K.; Kretschmann, A.; Steinr €u ck, H.-P.; Gottfried,
J. M. J. Am. Chem. Soc. 2007, 129, 12110–12111.
35) Appleton, T. G.; Clark, H. C.; Manzer, L. E. Coord. Chem. Rev.
973, 10, 335–422.
(
1
(
36) Hartley, F. R. Chem. Soc. Rev. 1973, 2, 163–179.
(37) Atwood, J. D. Inorganic and Organometallic Reaction Mechan-
isms, 2nd ed.; VCH: Weinheim, 1997.
(
(
38) Coe, B. J.; Glenwright, S. J. Coord. Chem. Rev. 2000, 203, 5–80.
39) Jordan, R. B. Reaction Mechanisms of Inorganic and Organome-
(71) Ghosh, A. Acc. Chem. Res. 2005, 38, 943–954.
(72) Ghosh, A. In The Porphyrin Handbook; Kadish, K. M., Smith,
K. M., Guilard, R., Eds.; Academic Press: San Diego, CA, 2000; Vol. 7, p
1ꢀ38.
tallic Systems, 3rd ed.; Oxford University Press Inc: New York, 2007.
40) Wackerlin, C.; Chylarecka, D.; Kleibert, A.; M €u ller, K.; Iacovita,
C.; Nolting, F.; Jung, T. A.; Ballav, N. Nat. Commun. 2010, 1, 61.
41) Suarez, S. A.; Fonticelli, M. H.; Rubert, A. A.; de la Llave, E.;
Scherlis, D.; Salvarezza, R. C.; Marti, M. A.; Doctorovich, F. Inorg. Chem.
(
(73) Collman, J. P.; Hoard, J. L.; Kim, N.; Lang, G.; Reed, C. A. J. Am.
Chem. Soc. 1975, 97, 2676–2681.
(
(74) Due to a spin equilibrium, fractions of the Fe ions may exist in a
singlet state (cf. Lecomte, C.; Rohmer, M.-M.; Benard, M. In The
Porphyrin Handbook; Kadish, K. M., Smith, K. M., Guilard, R., Eds.;
Academic Press: San Diego, 2000; Vol. 7, p 39; Scheidt, W. R. In The
Porphyrin Handbook; Kadish, K. M., Smith, K. M., Guilard, R., Eds.;
Academic Press: San Diego, 2000; Vol. 3, p 49; Collman, J. P.;
Hoard, J. L.; Kim, N.; Lang, G.; Reed, C. A. J. Am. Chem. Soc. 1975,
97, 2676–2681) or a quintet state, cf. ref 46.
(75) Grosvenor, A. P.; K., B. A; Biesinger, M. C.; McIntyre, N. S.
Surf. Interface Anal. 2004, 36, 1564–1574.
(76) Sanders, J. K. M.; Bampos, N.; Clyde-Watson, Z.; Darling, S. L.;
Hawley, J. C.; Kim, H.-J.; Mak, C. C.; Webb, S. J. In The Porphyrin
Handbook; Kadish, K. M., Smith, K. M., Guilard, R., Eds.; Academic
Press: San Diego, 2000; Vol. 3, pp 1ꢀ48.
2
010, 49, 6955–6966.
42) Domke, K. F.; Pettinger, B. Chemphyschem 2009,
0, 1794–1798.
43) Isvoranu, C.; Wang, B.; Schulte, K.; Ataman, E.; Knudsen, J.;
Andersen, J. N.; Bocquet, M. L.; Schnadt, J. J. Phys.-Condens. Matter
010, 22, 472002.
44) Kretschmann, A.; Walz, M.-M.; Flechtner, K.; Steinr €u ck, H.-P.;
Gottfried, J. M. Chem. Commun. 2007, 568–570.
45) Gottfried, J. M.; Flechtner, K.; Kretschmann, A.; Lukasczyk, T.;
Steinr €u ck, H. P. J. Am. Chem. Soc. 2006, 128, 5644–5645.
46) Shubina, T. E.; Marbach, H.; Flechtner, K.; Kretschmann, A.;
(
1
(
2
(
(
(
Jux, N.; Buchner, F.; Steinr €u ck, H.-P.; Clark, T.; Gottfried, J. M. J. Am.
Chem. Soc. 2007, 129, 9476–9483.
6
221
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222

Journal of the American Chemical Society
ARTICLE
(
77) Bai, Y.; Buchner, F.; Wendahl, M. T.; Kellner, I.; Bayer, A.;
Steinr €u ck, H.-P.; Marbach, H.; Gottfried, J. M. J. Phys. Chem. C 2008,
12, 6087–6092.
78) Note that the distance between the Fe ion and the Ag surface
(106) Burdett, J. K.; Albright, T. A. Inorg. Chem. 1979,
18, 2112–2120.
(107) Tolman, C. A. Chem. Rev. 1977, 77, 313–348.
1
(
(108) Seufert, K.; Bocquet, M.; Auw €a rter, W.; Weber-Bargioni, A.;
Reichert, J.; Lorente, N.; Barth, J. V. Nat. Chem. 2011, 3, 114–119.
(109) (a) Weidenbruch, M.; Stilter, A.; Peters, K.; von Schnering,
H. G. Chem. Ber. 1996, 129, 1565–1567. (b) Weidenbruch, M.; Stilter,
A.; Saak, W.; Peters, K.; von Schnering, H. G. J. Organomet. Chem. 1998,
560, 125–129. (c) Braunstein, P.; Veith, M.; Blin, J.; Huch, V. Organo-
metallics 2001, 20, 627–633. (d) Zabula, A. V.; Pape, T.; Hepp, A.; Hahn,
F. E. Dalton Trans. 2008, 5886–5890. (e) Cabon, Y.; Kleijn, H.; Siegler,
M. A.; Spek, A. L.; Gebbink, R.; Deelman, B. J. Dalton Trans. 2010,
39, 2423–2427.
may be different for FePc and FeTPP, because the peripheral phenyl
groups of FeTPP act as spacer groups, cf. ref 31. This would also
influence the magnitude of the ion-surface interaction.
(
79) Flechtner, K.; Kretschmann, A.; Bradshaw, L. R.; Walz, M. M.;
Steinr €u ck, H. P.; Gottfried, J. M. J. Phys. Chem. C 2007, 111, 5821–5824.
(80) Liao, M.-S.; Scheiner, S. J. Chem. Phys. 2002, 117, 205–219.
(
81) Comanici, K.; Buchner, F.; Flechtner, K.; Lukasczyk, T.;
Gottfried, J. M.; Steinr €u ck, H. P.; Marbach, H. Langmuir 2008,
4, 1897–1901.
82) Zotti, L. A.; Teobaldi, G.; Hofer, W. A.; Auw €a rter, W.; Weber-
Bargioni, A.; Barth, J. V. Surf. Sci. 2007, 601, 2409–2414.
83) Rovira, C.; Kunc, K.; Hutter, J.; Ballonelt, P.; Parrinello, M.
J. Phys. Chem. A 1997, 101, 8914–8925.
2
(
(110) Laurent, A. Low-valent group 14 metal containing ligands:
Versatile building blocks for the synthesis of transition metal complexes;
PhD Thesis: Saarbr €u cken, 2007.
(
(111) Barth, J. V. Surf. Sci. 2009, 603, 1533.
(
(
84) Liao, M.-S.; Scheiner, S. J. Chem. Phys. 2001, 114, 9780–9791.
85) Nguyen, K. A.; Pachter, R. J. Chem. Phys. 2001,
1
1
2
14, 10757–10767.
86) Vangberg, T.; Ghosh, A. J. Am. Chem. Soc. 1999,
21, 12154–12160.
87) W €o lfle, T.; G €o rling, A.; Hieringer, W. Phys. Chem. Chem. Phys.
008, 10, 5739.
88) Cordero, B.; G ꢀo mez, V.; Platero-Prats, A. E.; Rev ꢀe s, M.;
Echeverría, J.; Cremades, E.; Barrag ꢀa n, F.; Alvarez, S. Dalton Trans.
008, 2832–2838.
(
(
(
2
(
(
(
89) Shustorovich, E. Surf. Sci. Rep. 1986, 6, 1–63.
90) Redhead, P. A. Vacuum 1962, 12, 203–211.
91) Shi, X. Q.; Lin, C.; Minot, C.; Tseng, T.-C.; Tait, S. L.; Lin, N.;
Zhang, R. Q.; Kern, K.; Cerd ꢀa , J. I.; Van Hove, M. A. J. Phys. Chem. C
010, 114, 17197–17204.
92) Schlickum, U.; Decker, R.; Klappenberger, F.; Zoppellaro, G.;
2
(
Klyatskaya, S.; Ruben, M.; Silanes, I.; Arnau, A.; Kern, K.; Brune, H.;
Barth, J. V. Nano Lett. 2007, 7, 3813–3817.
(
93) Nappa, M.; Valentine, J. S. J. Am. Chem. Soc. 1978,
00, 5075–5080.
94) Groves, J. T.; Roman, J. S. J. Am. Chem. Soc. 1995,
17, 5594–95.
95) Ghosh, A.; Wondimagegn, T. J. Am. Chem. Soc. 2000,
22, 8101–8102.
96) Selcuki, C.; van Eldik, R.; Clark, T. Inorg. Chem. 2004,
3, 2828–2833.
97) Praneeth, V. K. K.; Neese, F.; Lehnert, N. Inorg. Chem. 2005,
4, 2570–2572.
98) Lehnert, N.; Praneeth, V. K. K.; Paulat, F. J. Comput. Chem.
006, 27, 1338–1351.
99) Nakashima, H.; Hasegawa, J.-Y.; Nakatsuji, H. J. Comput. Chem.
006, 27, 426–433.
100) Scheidt, W. R.; Ellison, M. K. Acc. Chem. Res. 1999,
2, 350–359.
101) Enemark, J. H.; Feltham, R. D. Coord. Chem. Rev. 1974,
3, 339–406.
1
1
1
4
4
2
2
3
1
(
(
(
(
(
(
(
(
(102) Paulat, F.; Lehnert, N. Inorg. Chem. 2007, 46, 1547–1549.
(103) It appears to the alert reader that we construct four molecular
orbitals from only three fragment orbitals here. Please note, however,
that other fragments orbitals (e.g., from the porphyrin ligand) also mix in
to these orbitals but are not shown for simplicity. The orbitals we show
here are those which bear a major contribution from the relevant
fragment orbitals which are included in the analysis.
(104) For the hollow-site coordination of CoTPP to the Ag72
cluster, the distance discussed corresponds to the distance between
the Co atom and the plane defined by the three adjacent Ag atoms
confining the hollow site.
(105) Weber-Bargioni, A.; Auw €a rter, W.; Klappenberger, F.; Reich-
ert, J.; Lefrancois, S.; Strunskus, T.; W €o ll, C.; Schiffrin, A.; Pennec, Y.;
Barth, J. V. ChemPhysChem 2008, 9, 89–94.
6
222
dx.doi.org/10.1021/ja1093502 |J. Am. Chem. Soc. 2011, 133, 6206–6222