Theoretical Study of Near-Edge X-ray Absorption Fine Structure

Feb 23, 2012 - In the H2Pc spectrum (panel c), the excitations from the Npyr-H sites are reported in green. The main calculated features below the N 1...
0 downloads 7 Views 2MB Size
Article pubs.acs.org/JPCA

Theoretical Study of Near-Edge X-ray Absorption Fine Structure Spectra of Metal Phthalocyanines at C and N K-Edges R. De Francesco, M. Stener, and G. Fronzoni* Dipartimento di Scienze Chimiche e Farmaceutiche, Università di Trieste, Via L. Giorgieri 1, I-34127, Trieste, Italy ABSTRACT: The inner shell excitation of CuPc, NiPc, and H2Pc phthalocyanines at both C and N K-edges has been investigated theoretically by density functional theory calculations. The selected molecules allow one to study the effect on the spectra of the presence and the nature of the atom in the central cavity of the macrocycle. The individual characteristics of the spectra can be rationalized in terms of the position of the unequivalent C and N atomic sites, showing that sensible changes are present in the spectral features deriving from the N atoms directly bound to the atom at the center of the Pc macrocycle. The minor variations present in the spectral C 1s profiles of the phthalocyanines reflect the little perturbation experienced by the peripheral atomic sites.

1. INTRODUCTION During past decades, organic semiconductors have been the subject of an intense research activity,1−4 due to their interesting electronic and optical properties that make them suitable to form well-ordered thin films on various kind of substrates, showing a typical self-assembly behavior that strictly depends on the conditions of deposition and growth on the substrate. Ordering and orientation of such molecules are generally recognized as crucial for controlling the related device efficiency. One of the most promising classes of organic semiconductors is represented by phthalocyanines (Pcs),5−7 which form a fundamental group of stable organic compounds, representative of low molecular weight organic molecules with extended π-systems. More than 70 different ions are known to be included in the central cavity of the Pc molecule, and an enormous number of combinations can be obtained by varying the substituent groups on the macrocycles.8−11 The extremely high chemical and thermal stability of metal Pcs (MPcs)12 allows the formation of high-quality, well-ordered films;13 Pcs are ubiquitous in nature, where they can play several roles, such as in oxygen transport, electron transfer, oxidation catalysis, and photosynthesis. These considerations explain why Pcs show such a huge number of potential low-cost, large-scale, flexible electronic device applications, ranging from the original use as blue-green synthetic pigments5 to photodynamic therapy of cancer,14 from catalysis15 to nanotechnology,16 from nonlinear optics12,17,18 to magneto-optical disks and gas sensors,19 from organic field-effect transistors (OFETs)20 to organic light-emitting devices (OLEDs),21 from wide-range industrial applications22 to optoelectronics, and artificial solar energy conversion.23 An extensive comprehension of the electronic structure of such systems is essential to correctly interpret and exploit the mechanisms that underlie their potential device applications; this issue enhances the importance of a joint theoretical and experimental analysis to perform a reliable interpretation of the observed properties. This goal can be reached by the use of nearedge X-ray absorption fine structure (NEXAFS) spectroscopy, © 2012 American Chemical Society

which has proven to be a very powerful tool to investigate both the geometric and the electronic structure of a wide range of systems.24 The spectral features take origin from the electronic excitations from a core orbital that is strictly localized on a particular atom toward unoccupied electronic levels lying around the ionization limit of that specific core hole; the excitation process is governed by dipole selection rules, and the oscillator strengths (transition intensities) are directly connected with the atomic site component of the virtual orbitals, which is dipole allowed. Therefore, K-edge NEXAFS spectra are expected to map the np character of virtual orbitals involved in the electronic transitions from 1s core orbitals. The strongly localized nature of the core hole makes NEXAFS spectroscopy very sensitive to the local environment of the absorbing atom in a very narrow spatial range; thus it has proven to be particularly useful for the characterization of single free molecules and even of very complex systems, such as adsorbate structures on several kind of substrates, for which NEXAFS spectroscopy is the method of choice to determine the orientation of the adsorbate with respect to the substrate.24 This work is devoted to the theoretical simulation and interpretation of the N and C K-edge NEXAFS spectra of copper phthalocyanine (CuPc), nickel phthalocyanine (NiPc), and metal-free phthalocyanine (H2Pc). We have chosen these three molecules because their NEXAFS experimental spectra available in the literature are relative to the “free” Pcs,25−27 while most of the spectroscopic data are relative to Pcs adsorbed on a variety of substrates.28−33 NEXAFS spectroscopy of adsorbed molecules is in fact influenced by the interaction between the molecule and the substrate,24 so it is not completely convenient to compare the experimental spectrum of the adsorbate with the calculated one of the free molecule. Furthermore, we have Received: November 15, 2011 Revised: February 14, 2012 Published: February 23, 2012 2885

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

the TS KS one-electron eigenfunctions, φiTS for the core orbital and φfTS for the virtual orbital:

considered the present series of Pcs to assess the effect of the presence and the nature of the metal in the central cavity on the NEXAFS spectra. CuPc is probably the most studied among all MPcs: several works are dedicated to the study of its electronic and geometric structure, both from an experimental8,25,32−39 and from a theoretical8,25,34,39−41 point of view, mainly due to its openshell electronic structure that allows a very wide range of potential applications; in particular, Evangelista et al.25 have recently presented an experimental and theoretical study performed on CuPc that has allowed them to assign the main spectral C 1s X-ray photoemission spectroscopy (XPS) and NEXAFS features to the different nonequivalent C atoms by associating these contributions to the corresponding calculated electronic structure. NiPc is a closed-shell molecule whose electronic structure has been described in a number of experimental8,26,27,42 and computational8,12,40,42 works. In detail, Milev et al.26 present C 1s NEXAFS data performed on NiPc powder to characterize the Pc system before and after milling; even more interesting appear the results by Krasnikov et al.,27 which present N 1s NEXAFS spectra for a number of Ni macrocyclic complexes, including NiPc. Both works represent an important source of information to be compared to our theoretical results. Also, the H2Pc molecule has been the subject of intense experimental13,33,43−45 and theoretical41,43,45−47 research activity; it has been included in the present study to obtain a reliable Pc series and to assess the influence of the macrocycles on the NEXAFS spectrum of the Pc molecule. In this work, we have performed Density Functional Theory (DFT) calculations to assess the relationships between the electronic structure and the most salient NEXAFS spectral features. We have first addressed the description of the electronic structure of the systems under study at the ground-state level, to establish the nature of the virtual valence molecular orbitals (MOs) involved in the core excitations, with special regard to the relative contributions of ligand and metal atomic sites. In a second step, we have calculated the NEXAFS spectra, including the relaxation effects upon formation of the core hole by means of the transition state (TS) scheme, with the aim to compare the results with respect to the experiment as well as to discuss the effect of the presence and the nature of the metal on the spectra.

fi → f =

(2)

with ni being the ground-state occupation number of the core orbital. It is well established in the literature that for such systems the importance of the inclusion in the computational scheme of the relaxation effects following the core hole formation is essential for a correct reproduction of the spectral feature.45,48,49 The inclusion of relaxation effects is achieved, in the present KS DFT scheme, by means of the TS approach50,51 previously described.

3. COMPUTATIONAL DETAILS The calculations have been performed with the 2009.01 version of the ADF program;52,53 all of the basis sets employed in this work, consisting of Slater-Type-Orbital (STO) functions, have been taken directly from the ADF database. The first part of the work consists of the DFT geometry optimization of the molecular structure of the systems under analysis. The TZP basis set and the VWN54 exchange correlation functional have been adopted after preliminary tests on the NiPc molecule. The comparison with available experimental8,55−57 and calculated8,40−42 structural parameters shows a very strict agreement on the order of 10−2 Å in bond lengths and 1−2° in bond angles. All of the molecules have been assumed to be planar (D4h symmetry group in the yzplane, with the y and z axes along the M−N bonds), as it is largely confirmed in the literature;8,25,27,41 the geometry optimization tests starting with a nonplanar geometry as initial guess have clearly evidenced that the molecules tend to become planar, so the molecular planarity has been assumed in all calculations, with a significant reduction of computational effort due to the use of molecular symmetry. The H2Pc molecule has been considered only in its trans configuration, because a preliminary test has confirmed that the cis configuration is energetically less stable.46 Both N K-edge and C K-edge spectra have been calculated for the CuPc−NiPc−H2Pc series employing the optimized geometries; in each case, the presence of symmetry nonequivalent atoms has to be considered (see Figure 1). In particular, for metal Pcs, four “aza” or “bridging” N atoms (N aza ) and four “pyrrolic” N atoms (N pyr ) must be distinguished, as well as eight C1, C2, C3 atoms (all “benzene” C atoms) and eight C4 atoms (“pyrrolic” C atoms). We have performed a separate calculation of the excitation spectrum for each individual core hole localized on an atomic non equivalent site; the total NEXAFS spectrum at the C and N edge is then obtained by summing the single spectra weighted by their relative abundance of symmetry equivalent nitrogen or carbon atoms. For H2Pc, it has been taken into account that two kinds of pyrrolic N atoms exist, identified as Npyr or Npyr‑H depending on whether they are bound to a H atom or not, and that each set of C atoms (C1, C2, C3, C4) is split into two subgroups depending on whether they belong to the pyrrole units with a N−H bond or not. All of these considerations force the overall molecule symmetry (D4h for MPc and D2h for H2Pc) to be lowered to C2v or Cs as an effect of the core hole localization. It is well-known that the inclusion of diffuse functions in the basis set employed to describe the atom carrying the core hole is mandatory for a proper calculation of excitation to Rydberg

2. THEORETICAL METHOD The N and C K-edge excitation spectra of the series of Pcs considered for this work have been calculated adopting the TS approach in the framework of the Kohn−Sham (KS) DFT. The TS scheme here employed consists of assuming an electron configuration obtained removing one-half an electron from the (initial) core orbital and leaving all of the virtual orbitals unoccupied during the Self Consistent Field (SCF) procedure; then the one-electron TS-KS equations are to be solved: TSφTS = εTSφTS i = 1, ..., n HKS i i i

2 2 ·ni ·ΔEi → f ·|⟨ϕiTS| r |⃗ ϕTS f ⟩| 3

(1)

TS where HKS is the KS Hamiltonian built with the TS electron configuration. The excitation energies ΔEi→f are then obtained as TS eigenvalues differences between virtual and core orbitals, while the oscillator strengths (transition intensities, f i→f) are calculated directly from the dipole transition moments between

2886

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

(fwhm) values, to allow an easier comparison with respect to the experimental spectra (see the corresponding figure captions for further details). Increasing fwhm values have been used at relatively high excitation energies (few electronvolts beyond the lowest energy structures) to approximate, at least at a semiquantitative level, both the natural line bandwidth of the experimental profile due to the lifetime broadening and the experimental resolution.

4. RESULTS AND DISCUSSION In Figure 1 are reported the Partial Density Of States (PDOS) profiles for CuPc (a), NiPc (b), and H2Pc (c) to provide a direct pictorial representation of the calculated electronic structure of the ground state for these systems. The geometric structures of the MPc and H2Pc molecules and the labels assigned to the different, nonequivalent C and N atoms have been reported as insets. In these diagrams, the energy levels have been rescaled to the lowest unoccupied molecular orbital (LUMO) so that the zero energy is relative to the LUMO, while negative and positive energy values are associated, respectively, to occupied and virtual molecular orbitals. Only C 2p, N 2p, and metal 3d contributions are reported, as they are expected to be the almost exclusive contributors in the energy region around the highest occupied molecular orbital (HOMO)−LUMO gap. It is convenient to start the discussion with a comparison between the CuPc (a) and the NiPc (b): the main difference between the two diagrams is represented by the two metal 3d profiles. CuPc has an open-shell electronic structure, as evidenced by the presence of the so-called singly occupied molecular orbital (SOMO) at −0.8 eV with respect to LUMO (i.e., within the HOMO−LUMO gap, as the HOMO is located around −1.5 eV on this scale); the SOMO is contributed mainly by Cu 3d (in particular 3d x2−y 2) with N pyr 2p σ components. This description confirms what was presented in spatial orbital distribution plots by Evangelista et al.25 who assign the SOMO essentially to Cu 3dx2‑y2 with small in-plane N 2p contributions. NiPc is instead characterized by a closed-shell electronic structure; therefore, no energy levels are situated between the HOMO and the LUMO (notice that the HOMO−LUMO gap value differs only by 0.03 eV from CuPc to NiPc), and the Ni 3d profile shows a deep valley in correspondence to this gap. The composition of the HOMO is essentially the same in the two systems: the highest contribution comes from C4 2pπ, with lower 2pπ contributions from the other C atoms, while metal 3d and N 2p participation is negligible. This is in complete agreement with other theoretical analyses,34 and in particular with what was stated by Liao and Scheiner40 and Cheng et al.,12 who stress that the HOMO is purely associated with the macrocycle and its orbital energy is rather unaffected by metal substitution, and also with orbital plots by Evangelista et al.,25 which locate the HOMO on C and N atoms around the central metal ion. The LUMO is again very similar in both MPcs calculated electronic structures: it is much more delocalized than the HOMO, as it has a very mixed character taking origin mostly from C4 2pπ and Naza 2pπ; interestingly, the 2pπ contribution coming from Npyr is enhanced in NiPc with respect to CuPc. The metal 3d participation is very low (around 4%) in both CuPc and NiPc, in agreement with other theoretical results.12,25 It is worth noting that the LUMO+1 orbital of NiPc is located only 0.12 eV above the LUMO and has the same nature

Figure 1. PDOS profiles versus energy levels in terms of KS eigenvalues; energy is rescaled with respect to LUMO KS eigenvalue, taken as zero. PDOS are reported respectively for CuPc (panel a), NiPc (panel b), and H2Pc (panel c). PDOS profiles are obtained by a fixed Lorentzian broadening (fwhm = 0.4 eV). PDOS profiles have been divided by the corresponding number of nonequivalent atoms for each kind (i.e., by 4 for N atoms and by 8 for C atoms) to obtain “normalized” profiles. Insets: The molecular structures of (a) MPc (M = Cu, Ni) and (c) H2Pc, with labels relative to the different nonequivalent C and N atoms.

orbitals. Thus, the QZ3P-3DIF basis set has been used for the excited atom (respectively N or C), while a TZ2P basis set has been used for all of the remaining atoms; core orbitals of the latter atoms have been treated by the Frozen Core (FC) technique. Furthermore, the FC procedure ensures the localization of the half core hole on one of the equivalent N (or C) atomic sites. The spectra calculations have been performed using the GGA approximation for the exchange-correlation energy functional with the parametrization “PW86x Perdew”,58,59 as it has proven to be particularly accurate in similar calculation.60 Other functional choices have been tested during the NiPc N K-edge case study, for example, “PW91”61 and “BP86”;59,62 in all of these test calculations, only minor variations emerged in the spectrum. Calculated spectral profiles have been convoluted by Gaussian functions of appropriate full width at half maximum 2887

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

Table 1. Theoretical Excitation Energies E (eV) and Oscillator Strength f for N K-Edge Spectra of CuPc, NiPc, and H2Pca

(mainly 3dx2−y2) as the SOMO of CuPc: in fact, while LUMO is mainly contributed by the ligands and therefore does not change its energy position along the series, the metal 3dx2−y2 orbital energy depends strongly on the metal nuclear charge and therefore in NiPc is appreciably destabilized with respect to CuPc. Looking at the atomic character of the virtual orbitals in a wider energy range (some electronvolts above the LUMO), only minor differences are observed when moving from CuPc to NiPc: contributions belonging to the different C and N atoms are quite well-distinguished and recognizable, while as expected the metal 3d profiles break down very quickly. In particular, it is interesting to notice that benzene C atoms (C1−C3) are quite close in energy, while the pyrrolic C atom (C4) is separated from them; moreover, a certain degree of C 2p−N 2p mixing is observed almost exclusively for C4 atoms (see around 1.5 eV and around 3.7−4 eV above the LUMO), which are the only C atoms in the molecule to be bound to N atoms (both types). All of these considerations are in agreement with the behavior observed by XPS data,25 which display a definite separation between C1−C3 and C4 peaks (1.35 eV); this chemical shift appears to be particularly strong with respect to other organic molecules and can be ascribed to the presence of electronegative N atoms (both Naza and Npyr). The PDOS profiles calculated for the H2Pc molecule are reported in panel (c) of Figure 1: the above considerations relative to MPcs can be extended also to this system, in agreement with the analysis by Alfredsson et al.,43 with the exception of the absence of the metal 3d contribution. It is interesting to notice that for H2Pc the LUMO resembles closely the CuPc composition, and this suggests that the enhancement of the 2pπ contribution coming from Npyr observed in NiPc should be attributed to the mixing with Ni 3d atomic orbitals. 4.1. N K-Edge Spectra. The theoretical N 1s core excitations for the series CuPc−NiPc−H2Pc are collected in Table 1 and Figure 2. We remind that in the metal Pcs there are two different nitrogen sites: the N atoms of the pyrrole rings surrounding the metal (Npyr) and the N atoms in the meso positions of the macrocyclic ring (Naza). The red and blue lines of the calculated spectra in Figure 2 refer to the excitations from the Naza and Npyr nonequivalent sites, respectively, while the total convoluted profile is reported in black. In the H2Pc spectrum (panel c), the excitations from the Npyr‑H sites are reported in green. The main calculated features below the N 1s ionization thresholds have been classified with labels for an easier comparison with the theoretical results reported in Table 1. We underline that the virtual orbital set is modified by the creation of a core hole; therefore, the energies, intensities, and atomic contributions reported in Table 1 are specific for each excitation site. The spectra show some common aspects along the series, in particular, the group of low energy features labeled A and B, generally associated with π* transitions toward the LUMO and LUMO+1 orbitals, while the higher energy features, labeled C, show different shape, which reveals a more complex nature. In the lower energy region, the most significant difference is the presence of an extra peak, labeled S, at lower energy in the CuPc spectrum, so we start the discussion considering the CuPc results. To the best of our knowledge, this is the first calculation of the N 1s NEXAFS spectrum of the free CuPc; most experimental measurements are relative to CuPc films and are angle-resolved to discuss the molecular orientation.32−37,39 Therefore, a specific analysis of which contributions from the unequivalent N atoms give rise to

site

peak

E (eV)

f ×103b

Naza

A1 B1 C C* C* C* S A2 B2 C C

406.51 408.06 410.43 411.08 411.59 411.60 405.68 406.73 408.36 410.01 410.37

33.60 20.19 18.30 9.76 4.04 13.55 10.83 16.57 22.39 3.88 14.35

C* C*

411.76 411.92

16.11 7.21

A1 B1 C C* C* A2 A2 B2 C C

406.60 408.15 410.49 411.18 411.62 407.04 407.16 408.66 410.43 410.73

34.38 21.31 19.84 10.48 17.60 19.00 18.52 22.10 5.98 12.27

C* C*

412.10 412.32

18.89 4.55

A1 B C C* C* A2 B C′

406.43 408.01 410.43 411.34 411.54 406.03 407.69 409.70

31.97 20.83 16.62 12.34 11.24 8.76 12.04 7.08

C B C′ C* C* C* C*

411.17 407.94 409.48 411.31 411.61 412.56 412.92

13.53 7.32 9.30 4.12 5.48 8.82 7.22

Npyr

Naza

Npyr

Naza

Npyr

Npyr‑H

main atomic character of final MO CuPc C4 2pπ; Naza 2pπ C3 2pπ; Naza 2pπ; C4 2pπ C1 2pπ; C4 2pπ Naza 3p; Naza 2pπ Naza 3p; Naza 4p Naza 3d ; C4 2pπ Cu 3d ; Npyr 2pσ C4 2pπ; Naza 2pπ C2 2pπ; Npyr 2pπ C1 2pπ; C3 2pπ C4 2pπ; C2 2pπ; C3 2pπ; Npyr 2pπ C4 2pπ; Naza 2pπ C4 2pπ; Npyr 4p NiPc C4 2pπ; Naza 2pπ C3 2pπ; C1 2pπ; Naza 2pπ C1 2pπ; C4 2pπ; Naza 2pπ Naza 3d; Naza 3p C4 2pπ; Naza 3d; Naza 3p Cu 3d; Npyr 2pσ C4 2pπ; Naza 2pπ C2 2pπ; Npyr 2pπ C1 2pπ; C4 2pπ C4 2pπ; C2 2pπ; C3 2pπ; Npyr 2pπ C4 2pπ; Naza 2pπ C4 2pπ; Npyr 4p H2Pc C4 2pπ; Naza 2pπ C1 2pπ; C3 2pπ; Naza 2pπ C1 2pπ; C4 2pπ; Naza 2pπ Naza 3d; C4 2pπ; Naza 2pπ Naza 3d; C4 2pπ; Naza 2pπ C4 2pπ; Naza 2pπ C2 2pπ; C4 2pπ; Npyr 2pπ C4 2pπ; C2 2pπ; C3 2pπ; Npyr 2pπ C4 2pπ; Naza 2pπ; Npyr 2pπ C4 2pπ; Naza 2pπ C2 2pπ C1 2pπ; C3 2pπ C4 2pπ; C2 2pπ; C3 2pπ Npyr 5p; Npyr 5s C4 2pπ

LUMO

SOMO LUMO

LUMO

LUMO

LUMO

LUMO

LUMO

a Calculated DFT-KS N 1s ionization limits. CuPc: Naza = 411.89 eV, Npyr = 411.99 eV. NiPc: Naza = 411.96 eV, Npyr = 412.41 eV. H2Pc: Naza = 411.87 eV, Npyr = 411.36 eV, Npyr‑H = 413.29 eV. bOnly calculated transitions with f > 3.5 × 10−3 are reported.

all of the spectral features below edge is still lacking. The experimental spectrum reported in panel a of Figure 1 is relative to a sample prepared by vacuum sublimation of CuPc powders on tantalum substrates;63 it has been properly shifted along the theoretical energy axis (by about 7.2 eV) to reach the best match with the calculated profile. This is a usual procedure adopted to overcome the general overestimation of the 2888

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

from each site, an electron is promoted toward the LUMO, giving rise to the A1 and A2 transitions of peak A, and toward LUMO+1 orbitals, giving rise to the B1 and B2 transitions of peak B. The energy splitting between the two LUMO transitions and the two LUMO+1 transitions amounts to 0.2 and 0.3 eV, respectively, therefore higher than the splitting between the two Naza and Npyr core holes (0.1 eV); this accounts for the different relaxation of the LUMO and LUMO+1 following the core hole formation on the two unequivalent N 1s sites. The significant oscillator strength calculated for transitions A and B reflects the presence of Naza 2pπ and Npyr 2pπ components into the LUMO and LUMO+1 orbitals, which are mainly C 2pπ-based orbitals localized on the pyrrolic and benzene rings, respectively (see Table 1). Structure C is contributed by two main transitions of similar intensity starting from the two N 1s sites toward virtual MOs having a main C 2pπ character of outer benzene and pyrrolic rings; the mixing of 2pπ pyrrolic (C4) atoms with small Npyr 2pπ and Naza 2pπ components is responsible for the calculated oscillator strengths of these two lines. The higher energy structure C* corresponds to transitions from both Naza and Npyr sites toward empty orbitals having a quite significant Naza 2pπ character mixed with C4 (pyrrolic) 2pπ components. Consider now the NiPc spectrum reported in Figure 2 and in Table 1. The experimental spectrum by Kransikov et al.27 is also reported, with a shift along the energy axis by about 8.2 eV, to reach the best match with the calculated profile. A good agreement is obtained between experiment and theory, as concerns both the relative energy positions and the intensity distribution of the first two A and B features, as is apparent from the figure. At higher energy, in the region of the C bands, the experiment has a broader shape, so the comparison with the theory is more qualitative. In the NiPc calculated spectrum, the lack of a lower energy peak in correspondence to the S peak in CuPc spectrum is well apparent, thus reflecting the closed-shell electronic structure of NiPc molecule, as previously commented. The first calculated feature (A1−A2) falls in the same energy region as the corresponding one in CuPc and again derives from π* excitations at the Naza and Npyr sites. The first peak A1 is ascribed to the LUMO transition from the Naza site, while peak A2 is contributed by two transitions of comparable intensity from the Npyr site toward the LUMO and LUMO+1 orbitals; the double-peak shape of the A1-A2 feature mainly derives from the splitting between the two Naza and Npyr core holes (0.46 eV), significantly higher than the splitting in CuPc. The A1 transition has the same nature of the analogous A1 peak in CuPc; in fact, the LUMO optimized for the core hole localized on the Naza site is characterized by the mixing of the C4 2pπ and the Npyr 2pπ and Naza 2pπ components; when the core hole is localized on the Npyr site, the first A2 transition is toward a LUMO orbital with mixed Ni 3d and Npyr (2pπ and 2pσ) character, which can be classified as σ* due to the predominance of the N 2p in-plane orbitals of the ligand atoms, as in the SOMO orbital of CuPc. The second A2 transition involves the LUMO+1 orbital, which has a composition analogous to that of LUMO orbital relative to Naza core hole optimization. Therefore, the localization of the core hole on the Npyr site reverses the energetic order of the first two virtual orbitals of the ground-state description, while the energy splitting between the two A2 transitions is kept at 0.12 eV. Feature B also shows a two-peaked shape associated with the two transitions from the Naza and Npyr sites to empty MOs mainly localized on the benzene rings with 2pπ component

Figure 2. Calculated N K-edge excitation spectra for (a) CuPc, (b) NiPc, and (c) H2Pc. Convoluted profiles are obtained by a fixed Gaussian broadening (fwhm = 0.4 eV) up to 409 eV and with an increasing Gaussian broadening (fwhm = 0.4−1.0 eV) from 409 to 414 eV. Black lines, total profiles; red lines, Naza transitions; blue lines, Npyr transitions; green lines (only panel c), Npyr‑H transitions. Panel a, dashed line, experimental CuPc spectrum from ref 63; panel b, dashed line, experimental NiPc spectrum from ref 27; panel c, dashed line, experimental H2Pc spectrum from ref 63. The vertical lines show the ionization limits, calculated as opposite eigenvalues of the corresponding core orbitals, which are reported in Table 1.

calculated excitation energies with respect to the experimental ones, due to the electron self-interaction error typical of density functional methods where the exchange operator is approximated. For this reason, we should consider the relative energy shift among the calculated transitions, which actually represent the most significant observables. The comparison between experiment and theory is satisfactory, although the experimental profile does not resolve fine structures, in particular at the lower energy side of the first band. Here, the calculated spectrum shows a single well-defined peak (labeled S), which arises from the transition of the Npyr 1s electron to the SOMO orbital. According to the electronic structure description, the contribution from the pyrrolic N 2p components to a virtual orbital mainly localized on the central Cu atom is responsible for the intensity calculated for the SOMO transition, which can be classified as σ* due to the predominance of the N 2p inplane orbitals of the ligand atoms. The following two bands (A and B) arise from the π* excitations at the Naza and Npyr sites: 2889

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

Table 2. Theoretical Excitation Energies E (eV) and Oscillator Strength f for C K-Edge Spectra of CuPc, NiPc, and H2Pca site

peak

E (eV)

f ×103b

C3 C1 C2 C2 C4 C1 C3 C1 C1 C3 C4 C4 C1 C1 C2 C2 C4

A A A B B B B B D D D E E E E F F

291.35 291.60 291.67 292.40 292.61 292.61 292.64 293.03 294.12 294.16 294.24 294.42 294.66 294.97 295.11 295.70 295.74

33.87 31.31 13.44 147.22 72.55 88.91 58.58 15.52 65.71 44.46 26.38 14.49 22.23 24.61 14.86 13.73 86.64

C3 C1 C2 C2 C1 C3 C4 C3 C1 C1 C3 C4 C1

A A A B B B B B B D D D E

291.38 291.62 291.69 292.37 292.58 292.62 292.65 292.86 293.04 294.12 294.17 294.25 294.66

34.09 31.43 13.45 144.02 84.85 49.36 70.33 19.31 18.98 67.97 45.36 36.52 23.33

main atomic character of final MO CuPc C4 C4 C4 C2 C4 C2 C2 C3 C1 C1 C3 C2 C1 C1 C2 C2 C1 NiPc C4 C4 C4 C2 C2 C2 C4 C3 C3 C1 C1 C3 C1

2pπ; Naza 2pπ 2pπ; Naza 2pπ 2pπ; Naza 2pπ 2pπ; C1 2pπ; C3 2pπ 2pπ; Naza 2pπ 2pπ; C1 2pπ; C3 2pπ 2pπ; C1 2pπ; C3 2pπ 2pπ; C1 2pπ 2pπ; C2 2pπ; C3 2pπ 2pπ; C3 2pπ; C4 2pπ 2pπ; Naza 2pπ 2pπ 3s 3p, 4p; Cu 4s, 5s 3p, 4p; Cu 4s, 5s 3d 2pπ; C3 2pπ

LUMO LUMO LUMO

2pπ; 2pπ; 2pπ; 2pπ 2pπ 2pπ 2pπ; 2pπ; 2pπ; 2pπ; 2pπ; 2pπ; 3s

Naza 2pπ Naza 2pπ Naza 2pπ

LUMO LUMO LUMO

Naza 2pπ C1 2pπ C1 2pπ C2 2pπ; C3 2pπ C3 2pπ; C4 2pπ C1 2pπ; Naza 2pπ

LUMO

site

peak

E (eV)

f ×103b

C1 C2 C2 C4

E E F F

294.99 295.13 295.70 295.76

24.74 16.41 13.42 88.68

C3 C1 C3‑H C1‑H C2‑H C4 C2 C2‑H C3 C1 C1‑H C3‑H C4‑H C3 C1 C4 C1‑H C3‑H C4‑H C1 C1‑H C1 C1‑H C4 C4‑H

A A A A A B B B B B B B B D D D D D D E E E E F F

291.14 291.45 291.48 291.63 291.71 292.40 292.40 292.44 292.53 292.61 292.62 292.76 292.82 294.00 294.02 294.07 294.21 294.32 294.50 294.64 294.75 294.99 295.13 295.66 295.93

14.36 12.43 17.74 18.10 10.33 36.19 79.35 68.03 37.33 51.50 41.57 25.82 37.46 23.93 35.34 12.95 27.25 19.81 13.55 13.16 14.39 11.51 12.09 35.07 47.42

LUMO

main atomic character of final MO NiPc C2 3p, 4p; Cu 4s, 5s C2 3p, 4p; Cu 4s, 5s C2 3d C1 2pπ; C3 2pπ H2Pc C4 2pπ; Naza 2pπ C4 2pπ; Naza 2pπ C4‑H 2pπ; Naza 2pπ C4‑H 2pπ; Naza 2pπ C4‑H 2pπ; Naza 2pπ C4 2pπ; Naza 2pπ C2 2pπ C2‑H 2pπ C2 2pπ; C1 2pπ; C3 2pπ C2 2pπ; C1 2pπ; C3 2pπ C2‑H 2pπ C2‑H 2pπ C4‑H 2pπ; Naza 2pπ C1 2pπ; C4 2pπ C1 2pπ; C4 2pπ Naza 2pπ; C3‑H 2pπ C1‑H 2pπ; C2 2pπ C1‑H 2pπ; C4‑H 2pπ Naza 2pπ C1 3s C1‑H 3s C1 3p, 4p C1‑H 3p, 4p C1 2pπ; C3 2pπ C1‑H 2pπ; C3‑H 2pπ

LUMO LUMO LUMO LUMO LUMO LUMO

LUMO

a Calculated DFT-KS C 1s ionization limits. CuPc: C1 = 296.69 eV; C2 = 296.76 eV; C3 = 296.62 eV; C4 = 298.02 eV. NiPc: C1 = 296.68 eV; C2 = 296.74 eV; C3 = 296.62 eV; C4 = 298.03 eV. H2Pc: C1 = 296.60 eV; C1‑H = 296.81 eV; C2 = 296.66 eV; C2‑H = 296.88 eV; C3 = 296.47 eV; C3‑H = 296.82 eV; C4 = 297.90 eV; C4‑H = 298.33 eV. bOnly calculated transitions with f > 13.0 × 10−3 (for CuPc and NiPc) and with f > 10.0 × 10−3 (for H2Pc) are reported.

atoms bound to the two central H atoms, indicated as Npyr‑H, and the pyrrole Npyr atoms not bound to H. The lower energy peak is mainly due to the excitation to the LUMO orbital from the Naza site (A1 line), while the LUMO transition from the Npyr site (A2 line) appears as a lower energy shoulder. There is therefore an inversion with respect to the energetic order of the A1−A2 excitations in the MPc spectra; in particular, the energy of the Npyr transition is lowered by about 0.7 and 1.0 eV with respect to the CuPc and NiPc analogous transitions, respectively. This red-shift affects also the Npyr ionization threshold; therefore, it seems to be essentially an initial state effect, which characterizes also the other transitions from this site. The sensitivity of the Npyr 1s ionization threshold to the absence of the metal atom might be ascribed to a different conjugation scheme of the free macrocycle with respect to the metal-phthalocyanine. The first resonance in the H2Pc spectrum has a π* nature, as in MPc; in fact, the LUMO orbitals specific for the two N excitation sites are both characterized by the mixing between the dominant C4 2pπ component and the smaller Naza 2pπ component. The Npyr‑H LUMO transition is instead found at higher energy (at 407.94 eV), shifted by about 1.5 and 1.9 eV from the Naza and Npyr LUMO transitions, respectively; these

from Naza atoms; their energy separation is about 0.5 eV and reflects the splitting of the two core holes. These transitions are similar both in nature and in intensity to the corresponding ones in CuPc. The higher energy C and C* features are again contributed by several transitions from both N unequivalent sites; in particular, structure C is assigned to virtual MOs with predominant C 2pπ character of both benzene and pyrrolic rings, while structure C* derives from transitions to empty MOs mainly localized on the benzene rings with p contributions from the N atoms in meso positions. The nature of the transitions in this energy region corresponds to that in CuPc; in particular, in the case of C* band, this shows that the empty MOs with predominant Naza contributions are less sensitive to the change of the central metal ion with respect to the Npyr sites directly bound to the metal. Finally, we consider the N 1s excitations of H2Pc (panel c). Also, the experimental spectrum63 is reported for comparison, with an energy shift of about 7.9 eV on the theoretical energy scale. A good match between the two profiles is observed; in particular, the energy position of the first two bands is correctly reproduced by the theory. To analyze the origin of these bands, we have to consider that the transitions can start from three unequivalent N sites: the bridging Naza atoms, the pyrrole N 2890

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

values indicate that the LUMO energies have a trend similar to that of the corresponding site-dependent ionization thresholds, confirming their specificity for each excitation site as a consequence of different conjugation. The Npyr‑H−LUMO transition contributes to feature B, which takes most of its intensity from the transitions from the other two Naza and Npyr sites to the respective LUMO+1 levels. These orbitals are mainly localized on the benzene rings with contributions from the 2pπ component from Naza and Npyr atoms, respectively, as found in MPc molecules; the Npyr excitation has still a lower energy than the Naza one. Feature C′ has not a counterpart in the MPc spectra being contributed by the Npyr‑H−LUMO+1 transition, which falls at a higher energy than the corresponding transitions from the Naza and Npyr sites, and by a Npyr transition to a virtual MO with predominant C 2pπ component of both benzene and pyrrolic rings. This kind of transitions characterize peak C in the MPc spectra, while in H2Pc peak C is contributed by the analogous transition from the only Naza site, with a consequent reduction of its intensity with respect to the corresponding peak in MPc spectra. Structure C* is assigned to a series of transitions from all three N unequivalent atoms, with those from the Npyr‑H site shifted at higher energies. The final MOs are mainly localized on the benzene rings, as in the MPc spectra. Summarizing, our calculations assign all of the below-edge features to transitions toward virtual orbitals mainly contributed by out-of-plane C 2p and N 2p atomic orbitals apart from the SOMO transition in CuPc, while the σ* transitions (i.e., transitions toward orbitals with main in-plane C 2p and N 2p character) are found at higher energies just above the ionization threshold. The spectral variations observed along the series are mostly due to the Npyr 1s partial profiles, while Naza profiles show only minor variations both in energy position and in intensity distribution. This agrees with the general observation that changes in the nature of the central metal atom are strongly felt by its nearest neighbors Npyr, while the rest of the molecule remains less sensitive;8,40 this is also confirmed by the variations observed in the metal-free phthalocyanine. 4.2. C K-Edge Spectra. The theoretical C 1s excitations for the three Pc molecules are collected in Table 2 and Figure 3. This figure shows the total profile and the individual contributions of the four unequivalent C 1s sites (labeled C1, C2, C3, and C4 in Figure 1) in different colors. Also, the available experimental spectra are reported for comparison (ref 25 for CuPc, ref 26 for NiPc, and ref 63 for H2Pc), following the same procedure before commented; in this case, the experimental spectra have been shifted on the calculated energy scale by 6.8, 8.0, and 7.2 eV, respectively. The C 1s spectra appear more structured than the N 1s ones due to the higher number of unequivalent core sites, while less variations in the spectral shape are observed along the series. The experimental spectrum of the CuPc molecule25 is dominated by an intense double-peaked feature (labeled A and B), which is followed by an asymmetric structure (D) with a higher energy shoulder (E) and by a wide feature (F) just before the first ionization threshold. A good general agreement of our calculated spectrum with the experiment is observed, allowing us to propose an assignment of the below-edge features. The lower energy peak A is assigned to the LUMO transitions at carbon atoms in the benzene rings. The LUMO levels have a main C4 2pπ component, and, as a result, the relative transitions acquire low intensity as they come from the small C 2pπ contribution of the benzene ring to the LUMO

Figure 3. Calculated C K-edge excitation spectra for (a) CuPc, (b) NiPc, and (c) H2Pc. Convoluted profiles are obtained by a fixed Gaussian broadening (fwhm = 0.3 eV) up to 294 eV and with an increasing Gaussian broadening (fwhm = 0.3−1.0 eV) from 294 to 300 eV. Black lines, total profiles; pink lines, C1 transitions; green lines, C2 transitions; blue lines, C3 transitions; gray lines, C4 transitions. Panel a, dashed line, experimental CuPc spectrum from ref 25; panel b, dashed line, experimental NiPc spectrum from ref 26; panel c, dashed line, experimental H2Pc spectrum from ref 63. The vertical lines show the ionization limits, calculated as opposite eigenvalues of the corresponding core orbitals, which are reported in Table 2.

orbitals. This attribution agrees with the interpretation in ref 25 based on static exchange (STEX) calculations. It is worth mentioning the lack of a SOMO-related peak due to its main Cu 3d and Npyr 2p nature; the C 2p participation is very small, so therefore such excitation has a negligible intensity in the C K-edge spectrum. The intense peak B is contributed by excitations at all four unequivalent C sites: in particular, the pyrrole C4 to LUMO transition is to be noticed, which is shifted by about 1.3 and 1.0 eV from the lower energy benzene C3 and C1 LUMO transitions, respectively, thus reflecting the energy trend of the calculated ionization thresholds. The pyrrole C4 to LUMO transition is higher in intensity than the benzene LUMO ones, as expected on the basis of the LUMO nature. The benzene carbon sites participate to peak B with transitions to LUMO+2 orbitals, which have a strong C 2pπ benzene character. This interpretation does not completely agree with the previous one,25 which ascribes feature B to the excitation only at the pyrrolic carbon atoms, while the 2891

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

going from n = 1 to n = 4, which can be related to the progressive approaching of the Cn/Cn‑H couple of sites to the pyrrole N/N−H sites. The same trend is followed by the transitions contributing to peaks A and B, although many of them are not reported in Table 2 due to their very low oscillator strengths. In particular, peak A derives from transitions from the six phenyl carbon sites to the corresponding LUMO orbitals, which are mostly localized on the pyrrolic carbon sites, therefore acquiring low intensity. The calculated IPs of these six C 1s sites differ by only about 0.4 eV, while the pyrrole C4/C4‑H thresholds are shifted to higher energy by about 2 eV; therefore, C4 and C4‑H transitions to the LUMO orbitals contribute to peak B. This peak also contains transition contributions from the other six C 1s sites to LUMO+1 orbitals, which are mostly localized on the C2 carbon atom of the phenyl ring. Also, the transitions at higher energies substantially follow the energy trend of the C 1s sites ionization thresholds previously explained and involve virtual orbitals localized on the other carbon atoms of the phenyl ring; the very large number of lines with relative low intensity gives rise to overlapping features (D, E, F) whose nature is therefore similar to that of the corresponding features in MPc molecules. The present assignment of the H2Pc spectrum is in general agreement with that proposed by Kera et al.45 to interpret the experimental spectrum of metal-free Pc films deposited on Ag(111) surface using the improved virtual orbital approximation.65 As a general comment, we observe that the total C 1s calculated profiles maintain a similar shape along the series, with some differences only in the H2Pc molecule due to the presence of a higher number of C 1s unequivalent sites; this behavior confirms that the peripheral atoms are less influenced by the presence and the nature of the central metal atom than its nearest neighbor atoms. Furthermore, the energy positions of the C site transitions are affected by the core hole localization and therefore reflect the splitting observed for the relative ionization thresholds.

transitions to LUMO+2 are supposed to contribute to the wide energy feature F. We stress again that the virtual orbitals set is modified by the creation of a core hole on the unequivalent carbon atoms; in particular, the LUMO−LUMO+1 shift decreases significantly with respect to the ground state (from 1.4 to 0.1−0.2 eV for the excitations at C1−C4 sites). As a consequence, the LUMO+1 transitions from the benzene carbon sites C1−C3 fall in the energy range of peak A; the LUMO+1 orbitals are mostly localized on the pyrrolic C4 atom, and this causes the negligible intensity of the relative transitions, which are not reported in Table 2 except for C2. Also, structure C is associated with very weak transitions from the unequivalent C1 and C3 sites into virtual MOs mostly localized on C2, which accounts for their negligible intensity; we do not report these transitions in Table 2. Feature D is ascribed to two excitations from the C1 and C3 benzene sites toward a virtual MO delocalized on the benzene carbon atoms and to the LUMO+2 transition from the pyrrolic C4 site. It has to be noted that the present assignment of the main A, B, and D peaks stands in some contrast to that provided by recent calculations based on the restricted open-shell CPP approach.64 The main difference is associated with the energy position of the main transition from the pyrrolic C4 atom, which is calculated in the region of peak B in the present work, while it is associated with the higher energy peak D in ref 64. Also, peak A is differently assigned: we associate it with a collective absorption of the three benzene C atoms (C1, C2, and C3), while it is ascribed as due to carbon C3 by ref 64. The weak shoulder E corresponds to a series of weak transitions, again derived from the benzene carbon sites. Finally, structure F is mainly contributed by a C4 site transition to a virtual MO mainly delocalized on the benzene carbon atoms and corresponding to the final states of the C1 and C3 excitations of feature D, according to the blue-shift of C4 transitions. There are no relevant changes in the general shape of the NiPc calculated spectrum; furthermore, also the attribution of the main C 1s transitions of NiPc strictly follows those proposed for CuPc. The agreement with the experimental spectrum26 is good, although in this case the experimental profile is less resolved; in fact, the CuPc spectrum25 has been recorded in gasphase experimental conditions with a photon energy resolution of about 40 meV, while the NiPc spectrum26 has been acquired using NiPc commercial powders, thus resulting in a rather poorer energy resolution (not explicitly specified). The NiPc spectrum does not reveal the lower energy A shoulder, which is instead present in the calculated spectrum and assigned as in CuPc spectrum, nor the small C feature whose intensity is negligible also in the calculated spectrum. The similarity between the two MPc C 1s spectra confirms that they are predominantly due to transitions to virtual states of the Pc macrocyclic rings, which do not contain contributions from the central metal atom. The convoluted profile of the H2Pc spectrum appears similar to the previous MPc ones; however, Figure 3 reveals an increased number of lines contributing to each feature and over which the intensity redistributes. This complexity is due to the larger number of unequivalent carbon atoms (eight) deriving from the two kind of pyrrole units, with or without a N−H bond, to which the set (C1, C2, C3, and C4) can refer. Therefore, the transitions at each C1−C4 site, relative to the pyrrole unit without N−H bond, have a counterpart at the sites C1‑H−C4‑H, relative to the pyrrole unit with a N−H bond. The calculated IPs of these eight carbon sites show a definite trend of increasing energy shift between the Cn−Cn‑H thresholds on

5. CONCLUSIONS N K-edge and C K-edge NEXAFS spectra have been calculated for CuPc, NiPc, and H2Pc phthalocyanines by means of DFT calculations using the transition state approximation to include the relaxation effects following the core hole formation. The site-resolved excitation spectra obtained with separate calculations for each unequivalent N or C atomic site represent a deconvolution of the total spectral profile into components that allows a great flexibility in analysis of the transitions and facilitates the attributions of the spectral features to specific portions of the molecule. The calculations reproduce well the few experimental data available for the free Pc molecules, confirming the reliability of the computational approach. Both N K-edge and C K-edge features in the below-edge energy region are attributed to transitions toward virtual orbitals mainly contributed by out-of-plane C 2p and N 2p atomic orbitals. N K-edge spectra show significant variations along the series in particular on going from the open-shell CuPc to the closed-shell NiPc and H2Pc systems; these differences can be mostly associated with Npyr−origin transitions, while Naza associated transitions do not change significantly along the series. This behavior is related to the perturbation experienced by pyrrolic Npyr atoms, which are directly bound to the atom at the center of the macrocycle (Cu, Ni, or H), while the bridging Naza sites are much less influenced by the change of the central atom. 2892

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

(28) Salomon, E.; Papageorgiou, N.; Angot, T.; Verdini, A.; Cossaro, A.; Floreano, L.; Morgante, A.; Giovanelli, L.; Le Lay, G. J. Phys. Chem. C 2007, 111, 12467. (29) Molodtsova, O. V.; Knupfer, M.; Ossipyan, Yu. A.; Aristov, V. Yu. J. Appl. Phys. 2008, 104, 083704. (30) Schulte, K.; Swarbrick, J. C.; Smith, N. A.; Bondino, F.; Magnano, E.; Khlobystov, A. N. Adv. Mater. 2007, 19, 3312. (31) Peltekis, N.; Holland, B. N.; Piper, L. F. J.; DeMasi, A.; Smith, K. E.; Downes, J. E.; McGovern, I. T.; McGuinness, C. Appl. Surf. Sci. 2008, 255, 764. (32) Biswas, I.; Peisert, H.; Casu, M. B.; Schuster, B.-E.; Nagel, P.; Merz, M.; Schuppler, S.; Chassé, T. Phys. Status Solidi A 2009, 206, 2524. (33) Okajima, T.; Fujimoto, H.; Sumitomo, M.; Araki, T.; Ito, E.; Ishii, H.; Ouchi, Y.; Seki, K. Surf. Rev. Lett. 2002, 9, 441. (34) Aristov, V. Yu.; Molodtsova, O. V.; Maslyuk, V.; Vyalikh, D. V.; Zhilin, V. M.; Ossipyan, Yu. A.; Bredow, T.; Mertig, I.; Knupfer, M. Appl. Surf. Sci. 2007, 254, 20. (35) Floreano, L.; Cossaro, A.; Gotter, R.; Verdini, A.; Bavdek, G.; Evangelista, F.; Rocco, A.; Morgante, A.; Cvetko, D. J. Phys. Chem. C 2008, 112, 10794. (36) Holland, B. N.; Peltekis, N.; Farrelly, T.; Wilks, R. G.; Gavrila, G.; Zahn, D. R. T.; McGuinness, C.; McGovern, I. T. Phys. Status Solidi B 2009, 246, 1546. (37) Chen, W.; Qi, D. C.; Huang, Y. L.; Huang, H.; Wang, Y. Z.; Chen, S.; Gao, X. Y.; Wee, A. T. S. J. Phys. Chem. C 2009, 113, 12832. (38) Holland, B. N.; Cabailh, G.; Peltekis, N.; McGuinness, C.; Cafolla, A. A.; McGovern, I. T. Appl. Surf. Sci. 2008, 255, 775. (39) Qi, D.; Sun, J.; Gao, X.; Wang, L.; Chen, S.; Loh, K. P.; Wee, A. T. S. Langmuir 2010, 26, 165. (40) Liao, M.-S.; Scheiner, S. J. Chem. Phys. 2001, 114, 9780. (41) Day, P. N.; Wang, Z.; Pachter, R. J. Mol. Struct. (THEOCHEM) 1998, 455, 33. (42) Rosa, A.; Ricciardi, G.; Baerends, E. J.; van Gisbergen, S. J. A. J. Phys. Chem. A 2001, 105, 3311. (43) Alfredsson, Y.; Brena, B.; Nilson, K.; Åhlund, J.; Kjeldgaard, L.; Nyberg, M.; Luo, Y.; Mårtensson, N.; Sandell, A.; Puglia, C.; Siegbahn, H. J. Chem. Phys. 2005, 122, 214723. (44) Casu, M. B.; Zou, Y.; Kera, S.; Batchelor, D.; Schmidt, Th.; Umbach, E. Phys. Rev. B 2007, 76, 193311. (45) Kera, S.; Casu, M. B.; Schöll, A.; Schmidt, Th.; Batchelor, D.; Rühl, E.; Umbach, E. J. Chem. Phys. 2006, 125, 014705. (46) Cortina, H.; Senent, M. L.; Smeyers, Y. G. J. Phys. Chem. A 2003, 107, 8968. (47) Linares, M.; Stafström, S.; Norman, P. J. Chem. Phys. 2009, 130, 104305. (48) Wilks, R. G.; MacNaughton, J. B.; Kraatz, H.-B.; Regier, T.; Moewes, A. J. Phys. Chem. B 2006, 110, 5955. (49) Otero, E.; Wilks, R. G.; Regier, T.; Blyth, R. I. R.; Moewes, A.; Urquhart, S. G. J. Phys. Chem. A 2008, 112, 624. (50) Slater, J. C. The Self Consistent Field for Molecules and Solids: Quantum Theory of Molecules and Solids; McGraw-Hill: New York, 1974; Vol. 4. (51) Ziegler, T.; Rauk, A. Theor. Chim. Acta 1977, 46, 1. (52) Baerends, E. J.; Ellis, D. E.; Ros, P. Chem. Phys. 1973, 2, 41. (53) Fonseca Guerra, C.; Snijders, J. G.; te Velde, G.; Baerends, E. J. Theor. Chem. Acc. 1998, 99, 391. (54) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200. (55) Robertson, J. M.; Woodward, I. J. Chem. Soc. 1937, 219. (56) Brown, C. J. J. Chem. Soc. A 1968, 2488. (57) Hoskins, B. F.; Mason, S. A.; White, J. C. B. Chem. Commun. 1969, 554. (58) Perdew, J. P.; Wang, Y. Phys. Rev. B 1986, 33, 8822. (59) Perdew, J. P. Phys. Rev. B 1986, 34, 7406. (60) Bolognesi, P.; O’Keeffe, P.; Ovcharenko, Y.; Coreno, M.; Avaldi, L.; Feyer, V.; Plekan, O.; Prince, K. C.; Zhang, W.; Carravetta, V. J. Chem. Phys. 2010, 133, 034302. (61) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C. Phys. Rev. B 1992, 46, 6671.

The C K-edge spectra appear more structured than the N 1s spectra because of the increased number of unequivalent sites, while their dependence on the chemical nature of the central atom is less intense and only minor variations are observed in the total C 1s profiles along the series. A detailed analysis of the results shows the influence of the core hole localization on the transitions from the unequivalent C sites, which follows closely the energy trend of the relative ionization thresholds.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



REFERENCES

(1) Forrest, S. R. Nature 2004, 428, 911. (2) Crone, B.; Dodabalapur, A.; Lin, Y.-Y.; Filas, R. W.; Bao, Z.; LaDuca, A.; Sarpeshkar, R.; Katz, H. E.; Li, W. Nature 2000, 403, 521. (3) Chua, L.-L.; Zaumseil, J.; Chang, J.-F.; Ou, E. C.-W.; Ho, P. K.-H.; Sirringhaus, H.; Friend, R. H. Nature 2005, 434, 194. (4) Muccini, M. A. Nat. Mater. 2006, 5, 605. (5) Leznoff, C. C.; Lever, A. B. P. Phthalocyanines: Properties and Applications; VCH: New York, 1989; and refs therein. (6) Kadish, K. M.; Smith, K. M.; Guilard, R. The Porphyrin Handbook: Phthalocyanines, Properties and Materials; Academic Press: San Diego, CA, 2003. (7) McKeown, N. B. Phthalocyanine Materials; Cambridge University Press: New York, 1998. (8) Mastryukov, V.; Ruan, C.-Y.; Fink, M.; Wang, Z.; Pachter, R. J. Mol. Struct. 2000, 556, 225. (9) Miyake, K.; Hori, Y.; Ikeda, T.; Asakawa, M.; Shimizu, T.; Sasaki, S. Langmuir 2008, 24, 4708. (10) Oison, V.; Koudia, M.; Abel, M.; Porte, L. Phys. Rev. B 2007, 75, 035428. (11) Kong, X.-H.; Deng, K.; Yang, Y.-L.; Zeng, Q.-D.; Wang, C. J. Phys. Chem. C 2007, 111, 17382. (12) Cheng, W.-D.; Wu, D.-S.; Zhang, H.; Chen, J.-T. Phys. Rev. B 2001, 64, 125109. (13) Kera, S.; Casu, M. B.; Bauchspiess, K. R.; Batchelor, D.; Schmidt, Th.; Umbach, E. Surf. Sci. 2006, 600, 1077. (14) Juzenas, P. Trends Cancer Res. 2005, 1, 93. (15) McKeown, N. B.; Budd, P. M. Chem. Soc. Rev. 2006, 35, 675. (16) Elemans, J. A. A. W.; van Hameren, R.; Nolte, R. J. M.; Rowan, A. E. Adv. Mater. 2006, 18, 1251. (17) Chen, Y.; Hanack, M.; Blau, W. J.; Dini, D.; Liu, Y.; Lin, Y.; Bai, J. J. Mater. Sci. 2006, 41, 2169. (18) Wróbel, D.; Dudkowiak, A. Mol. Cryst. Liq. Cryst. 2006, 448, 617. (19) Trometer, M.; Even, R.; Simon, J.; Dubon, A.; Laval, J.-Y.; Germain, J. P.; Maleysson, C.; Pauly, A.; Robert, H. Sens. Actuators, B 1992, 8, 129. (20) Guillaud, G.; Al Sadoun, M.; Maitrot, M.; Simon, J.; Bouvet, M. Chem. Phys. Lett. 1990, 167, 503. (21) Huang, Y.-S.; Jou, J.-H.; Weng, W.-K.; Liu, J.-M. Appl. Phys. Lett. 2002, 80, 2782. (22) Gregory, P. J. Porphyrins Phthalocyanines 2000, 4, 432. (23) Walter, M. G.; Rudine, A. B.; Wamser, C. C. J. Porphyrins Phthalocyanines 2010, 14, 759. (24) Stöhr, J. NEXAFS Spectroscopy; Springer: Berlin, 1992. (25) Evangelista, F.; Carravetta, V.; Stefani, G.; Jansik, B.; Alagia, M.; Stranges, S.; Ruocco, A. J. Chem. Phys. 2007, 126, 124709. (26) Milev, A.; Kannangara, G. S. K.; Tran, N.; Wilson, M. Int. J. Nanotechnol. 2007, 4, 516. (27) Krasnikov, S. A.; Preobrajenski, A. B.; Sergeeva, N. N.; Brzhezinskaya, M. M.; Nesterov, M. A.; Cafolla, A. A.; Senge, M. O.; Vinogradov, A. S. Chem. Phys. 2007, 332, 318. 2893

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894

The Journal of Physical Chemistry A

Article

(62) Becke, A. D. Phys. Rev. A 1988, 38, 3098. (63) Koch, E. E.; Jugnet, Y.; Himpsel, F. J. Chem. Phys. Lett. 1985, 116, 7. (64) Linares, M.; Stafström, S.; Rinkevicius, Z.; Ågren, H.; Norman, P. J. Phys. Chem. B 2011, 115, 5096. (65) Kosugi, N. Theor. Chim. Acta 1987, 72, 149.

2894

dx.doi.org/10.1021/jp2109913 | J. Phys. Chem. A 2012, 116, 2885−2894