Control of Electron Injection Barrier by Electron Doping of Metal

Jun 24, 2010 - Progressive occupation of the former lowest unoccupied molecular ... the central metal atoms are involved in the electron donation proc...
8 downloads 0 Views 1MB Size
12258

J. Phys. Chem. C 2010, 114, 12258–12264

Control of Electron Injection Barrier by Electron Doping of Metal Phthalocyanines Pierluigi Gargiani, Antonio Calabrese, Carlo Mariani,* and Maria Grazia Betti Dipartimento di Fisica, UniVersita` di Roma “La Sapienza”, Piazzale A. Moro 5, I-00185 Roma, Italy ReceiVed: April 30, 2010; ReVised Manuscript ReceiVed: June 3, 2010

Metal phthalocyanine (MPc, with M ) Fe, Co, Ni, Cu, Zn) thin-films grown on Au(110) have been electrondoped by exposition to alkali metal (potassium). Progressive occupation of the former lowest unoccupied molecular orbital (LUMO) up to the filling of the 2-fold degenerate LUMO with 2eg symmetry has been observed in photoemission spectra for the NiPc, CuPc, and ZnPc films. The evolution of the electron filling induces a different process for FePc and CoPc, whose empty electronic states localized on the central metal atoms are involved in the electron donation process by the alkali metal. Electron filling is accompanied by a sudden decrease of the work function mainly due to a variation of the surface dipole. The molecular films remain semiconducting, with the hole injection barrier strongly reduced at saturation of the electron filling. The selective influence of the electron donation process on the molecular states (ligands, metal-related) is discussed. Introduction The appealing optical and transport properties of organometallic semiconducting materials and the processing flexibility of molecular films has led to exceptionally rapid progress in the development of organic devices over the past decade.1 Design of molecular devices requires an atomic level understanding of the parameters that control the transport properties of these molecular systems. Modification of the electronic properties of organic semiconductors via charge donation or acceptance from dopants plays a crucial role in determining the electron and hole injection barriers of these molecular systems.2–7 Metal phthalocyanines (MPcs, M-C32H16N8) are stable aromatic dyes constituted by four pyrrolic and four benzene rings symmetrically arranged around the central metal atom. Condensed MPcs are molecular semiconductors with energy gap in the visible region, giving rise to large injection barriers.8 MPcs upon alkali metal doping have been investigated by combined scanning tunneling spectroscopy (STM) and transport measurements:6,7 they are semiconductors in the condensed phase, they can be turned metallic through alkali doping, before turning again insulating when electrons fill up the empty states. These results seem independent of the MPc central metal atoms (Fe, Co, Ni, Cu).6,7 Despite this evidence, recent photoemission studies of CuPc4,9–12 and ZnPc13–15 films intercalated with alkali metals do not present spectral density of electronic states at the Fermi level, at any doping level until saturation. In this paper we present high-resolution photoemission data on the series of 3d-metal MPcs grown on Au(110). The reduction of the electron injection barrier, the absence of metal-insulator transition, and the symmetry of the filled electronic states are discussed comparing alkali doping induced electronic states in MPc thin films with M ) Fe, Co, Ni, Cu, Zn. We show that the electron injection processes depend on the degeneracy and on the symmetry of the actual molecular states available for electron allocation. While MPcs with almost filled d states (NiPc, CuPc, and ZnPc) behave identically upon alkali doping, with the injected electrons filling the LUMO states * To whom correspondence [email protected].

should

be

addressed.

E-mail:

localized on the macrocycles, FePc and CoPc present a different evolution of the electron filling process, involving empty states also localized on the central metal atom. Despite a different filling of the empty states, the alkali-doped organic MPc films remain semiconducting, with a strongly reduced hole injection barrier (HIB) for all measured MPcs. At saturation doping, the nature of the new highest occupied molecular orbital (HOMO) of FePc involves d metal states, suggesting a different nature of the new molecular gap. Thus, the different response to electron injection of the MPcs with different central d-metal occupancy, in particular for FePc, opens perspectives and a new scenario for controlling and tailoring the electronic and eventually magnetic response of organometallic thin films. Experimental Details Experiments have been carried out at the surface physics laboratory LOTUS at the Universita` “La Sapienza” in Roma, in an ultra-high-vacuum (UHV) chamber equipped with highresolution ultraviolet photoelectron spectroscopy (HR-UPS), low-energy electron-diffraction (LEED), and all the ancillary facilities for sample preparation. Photoelectron spectra have been excited with a He discharge source (He IR and He IIR photons, hν ) 21.218 and 40.814 eV, respectively). The photoemitted electrons were analyzed in the plane of incidence, with a highresolution Scienta SES-200 hemispherical analyzer, by using a two-dimensional multichannel-plate detector. Angular integration has been obtained over (8° around normal emission. Intrinsic energy resolution is 16 meV. A sharp long-range ordered (1×2) LEED reconstruction pattern of the clean Au(110) surface was obtained by means of a double-step sputtering-annealing treatment (Ar+ ions at 1000 eV, T ) 725 K, followed by 500 eV, 530 K). MPc thin films (∼20 nm thick) were prepared by evaporation on the Au(110) surface from quartz crucibles, after purification obtained by degassing cycles of several hours. The film thickness was estimated by means of a quartz microbalance, and a low deposition rate was used (about 2 Å/min), to maintain good vacuum conditions and smooth film growth. Potassium was sublimated through a SAES-Getters dispenser, keeping the pressure in the low 10-10 mbar range during deposition.

10.1021/jp103946v  2010 American Chemical Society Published on Web 06/24/2010

Control of Electron Injection Barrier

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12259

Deposition has been measured as a function of the exposure time, and the saturation coverage (Θsat) has been determined for each MPc thin film (TF). Θsat is assumed as the exposition dose for which the K-3p core level does not increase in intensity, the work-function saturates, and no significant spectral change is observed in valence band photoemission. The work function variation (∆Φ) has been estimated by measuring the shift of the low-energy photoemission cutoff, with -9 V biased sample with respect to ground, to get rid of the analyzer work function. Results and Discussion In the following, we present and discuss photoemission experiments on potassium-doped FePc, CoPc, NiPc, CuPc, and ZnPc thin films grown on Au(110). Although previous photoemission experiments on K-doped CuPc9 and ZnPc13,15 thin films are available, we report herewith the complete MPc series, for a straightforward comparison of the central metal atom influence on the electronic properties. To this purpose, all the present MPc thin films have been grown on the Au(110) surface with the same procedure, which has been shown to lead to a flat16 ordered monolayer,17 followed by thin-film formation with a planar molecular orientation.18 First, we will focus on the work function evolution, then on the valence-band spectra. Electron doping of the different MPc thin films with K produces consistent changes in the work function and in the spectral density of electronic states in the valence band energy region, depending on the central metal atoms. Doping with potassium results in a transfer of the outer K 4s electron to the MPc formerly empty molecular states, whose evolution as a function of K doping is discussed and correlated to the different central metal atoms. The energy level diagram resulting from the K-induced electron donation as a function of the occupancy of the d metal states is eventually discussed. Filling Empty Molecular States. The work function variation (∆Φ) of the MPc thin films grown on Au(110) as a function of K coverage is shown in Figure 1. For all MPcs, a decrease of the work function from the initial phase of potassium deposition until saturation is displayed. In particular, the work function drops by 0.49 (FePc), 1.06 (CoPc), 0.62 (NiPc), 0.65 (CuPc), and 0.67 eV (ZnPc) at saturation of potassium doping. Donation of one electron from the 4s K level to the molecular empty states of the MPc molecules upon K intercalation induces the decrease of the work function, as also observed in previous works on alkali-metal doped CuPc4,9–12 and ZnPc.13–15 We can attribute the work function reduction to the variation of the surface dipole, induced by electron charge donation from the alkali metal atoms to the empty states of the MPc molecules. While NiPc, CuPc, and ZnPc molecular films present a monotonic work-function decrease upon doping, with a comparable ∆Φ reduction at saturation, the FePc film shows a monotonic decrease as well, but with a lower saturation value, while CoPc displays a discontinuity, with a slope change at about 1/3 of Θsat, and a stronger reduction at saturation. The electron donation from the K atoms can find empty states with different symmetry and degeneracy for allocation of the transferred charge, depending on the specific MPc. In fact, while the molecular orbitals residing on the aromatic macrocycles are common to all MPcs, there is a different sequence of occupied and empty states in the d-symmetry levels localized on the central metal atom.17,19,20 This issue will be clarified by careful analysis of the electronic spectral density evolution in the valence band, as a function of K-induced electron filling. Normal emission photoelectron spectra of FePc, CoPc, NiPc, CuPc, and ZnPc thin films grown on Au(110) as a function of

Figure 1. Work function variation of the MPc thin films grown on Au(110), as a function of K coverage: (a) ∆Φ for FePc (open squares) and CoPc (filled upward and downward triangles, from different experimental runs); (b) NiPc (filled dots), CuPc (filled triangles), and ZnPc (open squares).

K coverage, in the energy region of the highest occupied molecular states, are displayed in Figure 2a-e. The undoped MPc thin films are semiconducting with the HOMO with a1u symmetry localized on the pyrrole ring at a binding energy (BE) of 1.35 (FePc), 1.33 (CoPc), 1.59 (NiPc), 1.70 (CuPc), and 1.57 eV (ZnPc), with corresponding full-width at half-maximum (fwhm) as reported in Table 1. These BE values are determined with respect to the Fermi level of the system, neglecting the downward band-bending in the semiconducting films (∼0.2 eV). The different MPc thin films present a hole-injection barrier (HIB), as estimated from the HOMO onset at low-binding energy, ranging from 0.9 to 1.2 eV, leading to an ionization energy (IE) of about 5.2-5.3 eV for all the present MPc thin films (see Table 1). These IEs are in agreement with previous observations, once their estimation calculated with respect to the HOMO peak maximum is taken into account.21,22 For CoPc, we notice that the actual highest occupied state is the partially occupied a1g level, only visible at high photon energy,17 due to excitation cross section reasons.23 Thus, if we evaluate the IE from the a1g slope edge for CoPc, it results in a value of 4.6 eV, which is in agreement with a previous determination.22 The values, lower than the corresponding IEs for the gas-phase molecules (6.3-6.4 eV for these MPcs,24,25 see Table 1), are due to the effect of intermolecular polarization by neighboring molecules, experienced by the photohole into the condensed film. The polarization reaction occurs on the time scale of the photoemission process, lowering the expected binding energies. The next molecular level at higher BE with respect to the HOMO is at 1.71 (FePc), 2.19 (CoPc), 2.89 (NiPc), and 2.66 eV (CuPc). The peak at 2.66 eV for CuPc is due to the b1g state, while the structures observed in FePc, CoPc, and NiPc can be attributed to the b2g state (Table 1), which becomes

12260

J. Phys. Chem. C, Vol. 114, No. 28, 2010

Gargiani et al.

Figure 2. High-resolution UPS data of the MPc thin films grown on Au(110) as a function of K coverage. Single and sum fitting curves are superimposed to the experimental data (open dots). For FePc and CoPc, fitting curves of the former-empty levels being filled upon doping are shadowed.

TABLE 1: Energy Levels of Undoped MPc Thin Films Grown on Au(110) FePc CoPc NiPc CuPc ZnPc

WF (eV)

HIB (eV)

IE (eV)

a1u-HOMO (eV)

fwhm (eV)

b2g (eV)

fwhm (eV)

4.31 4.28 4.03 3.95 4.11

0.98 0.92 1.19 1.28 1.20

5.29 5.20 5.22 5.23 5.31

1.35 1.33 1.59 1.70 1.57

0.37 0.49 0.42 0.48 0.40

1.71 2.19 2.89

0.55 0.54 0.49

hidden by higher BE molecular structures in CuPc and ZnPc. The b2g state is a hybridized molecular level associated with the overlapping between the central metal atom states and the π-orbital localized on nitrogen,20 and it presents an increasing binding energy upon increasing the number of 3d-metal electrons. On the other hand, the a1u HOMO molecular state, localized on the π-orbital molecular states of the pyrrole ring, is much less affected by the metal substitution (see Table 1) when considering the HOMO ionization energy (with respect to the vacuum level of each system). We do not observe any charging in the ∼20-nm-thick thin films, and all the bindingenergy values are in agreement with previously reported ones,17

once the small downward band-bending of the molecular layers is taken into account. Upon initial K doping, all the spectral features shift to higher binding energy (Figure 2a-e), thus the common energy variation results from a shift of the conduction band toward the Fermi level, as expected for an n-type doping process. In particular, for all MPcs the common energy shift of the HOMO/ex-HOMO peaks is comparable with the work function decreasing, attributed to a dipolar change and band-bending, except for FePc where the shift is consistently lower than the ∆Φ in absolute value (Table 2). This latter different energy shift indicates that more effects than a pure dipolar change take place in particular

Control of Electron Injection Barrier

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12261

TABLE 2: Energy Levels of K-doped MPc/Au(110) at Saturation FePc CoPc NiPc CuPc ZnPc

FePc CoPc NiPc CuPc ZnPc

ex-HOMO (eV)

fwhm (eV)

ex-2eg (eV)

fwhm (eV)

1.54 2.29 2.14 2.30 2.09

0.61 0.72 0.70 0.74 0.72

0.89 1.01 1.09 1.01

0.73 0.69 0.74 0.65

ex-a1g/1eg (eV)

fwhm (eV)

0.96 1.57

0.61 0.60

∆WF (eV)

HIB (eV)

IE (eV)

∆ex-HOMO (eV)

-0.46 -1.06 -0.62 -0.65 -0.67

0.36 0.24 0.36 0.36 0.36

4.21 3.46 3.77 3.66 3.80

-0.19 -0.96 -0.55 -0.60 -0.52

for FePc, suggesting a different charge transfer mechanism for K-doped FePc, as discussed later. The -∆Φ trend as a function of coverage in the very first deposition steps gives a linear slope in a semilogarithmic plot, whose fitting gives 0.22 (ZnPc), 0.22 (CuPc), 0.18 (NiPc), 0.11 (CoPc), and 0.11 eV/electron (FePc). These values are obtained estimating 4 injected electrons per molecule at saturation (doubly-degenerate LUMO level). However, even assuming an uncertainty of 50% in saturation occupancy, the error in the slope estimation is of the order of (0.01 eV/electron. As a consequence, even considering this large uncertainty in the number of transferred electrons per molecule, we observe that the slope is much higher than the thermal energy kBT (0.026 eV) expected in the classical theory for inorganic semiconductors, and it is in good agreement with the slope values previously determined at the Cs-doped4,5 and Na-doped CuPc.11 Potassium deposition leads to the appearance of induced peaks in the former band gap, which are caused by the filling of former-unoccupied molecular orbitals by the K 4s electrons. As the potassium concentration increases, the intensity of the new structures is enhanced, but none of the spectra show emission from the Fermi level, i.e., none of the MPc thin films intercalated with K is metallic at any coverage until saturation. The insulating nature of the alkali metal-intercalated MPcs is generally observed in photoemission experiments, with occupation of the former-LUMO.4,5,9–15 The absence of a metallic phase can be related to electron correlation effects26 associated to the occupation of the localized molecular bands. The evolution of the former-LUMO and of the other spectral features being occupied by the K-4s electron has been analyzed by fitting curves and shown in Figure 2a-e, superimposed to the experimental data. Fitting has been conducted with use of Gaussian curves, whose relevant fitting parameters (energy position and peak widths) are reported in Table 2. Although the K intercalation induces a similar behavior (absence of any metallic phase) for all MPc semiconducting films, we can notice evident differences in the electron filling process depending on the central metal atom. ZnPc, CuPc, NiPc, and FePc K-doped films clearly show the occupation of the former-LUMO, appearing as a peak in the former molecular energy gap. Upon increasing K exposure, the former-LUMO is progressively filled and its intensity increases up to a saturation coverage. The centroid of the former-LUMO states filled by the electrons coming from the ionized K adatoms shifts continuously to higher binding energy, as the other molecular levels. This energy shift is comparable (about 0.5-0.6 eV) for the NiPc, CuPc, and ZnPc systems, and it is lower for the alkalidoped FePc film (=0.2 eV). The CoPc K-doped film presents a different evolution: there is a first stage where a state appears at about 0.96 eV BE, it shifts to higher BE at subsequent K

deposition, while a further formerly empty molecular state is involved in the filling process. A quantitative description of the electron donation process can be achieved by evaluating the K 3p core-levels intensity, comparing the different MPc thin films at alkali saturation coverage, as reported in Figure 3. K 3p core levels are excited by HeII radiation, thus laying above the decreasing background due to the HeI photoemitted electrons, and they have been normalized to the spectral signal associated to the macrocycledependent σ-π molecular states at 22 eV binding energy (not shown in Figure 3), independent of the central metal atom. At saturation coverage, the K 3p core levels present comparable intensity (evaluated as the peak area, after a spline background subtraction) for the NiPc, CuPc, and ZnPc thin films, while a lower intensity is observed for the FePc film and a higher intensity for CoPc. The relative intensity of the K 3p core levels indicates that CoPc and FePc present about 1.5 ((20%) and 0.5 times ((20%) the number of K atoms, respectively, than NiPc, CuPc, and ZnPc. These data suggest that each CoPc molecule can host more electrons with respect to the almost d-filled central metal MPcs, while FePc has fewer states available for hosting the alkali metal electrons. It is useful to summarize the main experimental observations subsequent to potassium doping of the MPc/Au(110) thin films: (a) the work function decreases with different saturation values: higher for the CoPc doped thin film, lower for FePc, in agreement with a larger (lower) energy shift of the molecular levels for CoPc (FePc), due to the induced band bending in the semiconducting films; (b) all MPcs are semiconducting at any doping level, with a comparable reduction of the hole-injection barrier (HIB); (c) the K s electrons fill different unoccupied states, depending on the central metal atom; (d) the filled states

Figure 3. (Left) UPS data of the K-3p core-levels at the K-doped MPc/Au(110) thin films, at K saturation; (right) intensity of the K-3p core-levels at the K-doped MPc/Au(110) thin films, at K saturation.

12262

J. Phys. Chem. C, Vol. 114, No. 28, 2010

present different peak broadening; and (e) the number of potassium atoms per MPc molecule at saturation coverage is different, as deduced from the K 3p core levels. The comparison of the electron filling process among MPc/ Au(110) with different central atoms unravels some experimental evidence in contrast with previous works. In particular, an insulator-to-metal transition has been observed upon alkali metal electron filling half of the empty levels,6 regardless of the central metal atom, suggesting a similar filling process involving the LUMO states with the same degeneracy and symmetry. Considering the present experimental observations and previous photoemission data in literature on MPc thin films4,5,11,15 and at MPc single layers,10,12 the semiconducting response of electrondoped MPc has been established due to electron correlation effects.26 Furthermore, the differences in the spectral density of electronic states at saturation coverage for the doped MPc semiconducting films can be correlated to the specific MPc molecular orbitals involved in the filling process. We suggest that electrons can be added to orbitals that are centered either on the ligands or on the central metal atoms, depending on the specific metal phthalocyanine. Our results are consistent with the hypothesis of a common behavior in the molecules with almost filled d-states associated to the central metal atom (ZnPc, CuPc, NiPc), where the double degenerate LUMO empty state 2eg with π*-symmetry20 can be filled by four s electrons donated by the potassium intercalated atoms, while for FePc and CoPc the first state attracting electrons seems to be different from the higher lying 2eg state and to involve molecular levels located on the central metal atom.17,20 Theoretical predictions19,20,27–32 on the ground state of FePc and CoPc molecules are discordant on the energy distribution of the molecular levels mainly localized on the central metal atoms, i.e., involving d-symmetry partially occupied orbitals. The Fe and Co atoms, embedded in the molecular environment with a tetragonal symmetry, break the degenerate 3d level symmetry, giving rise to filled b2g states with dxy symmetry (clearly detected in photoemission spectra) and empty b1g state with dx2-y2 symmetry, both hybridized with the π-orbital localized on the nitrogen atoms, while the occupancy of the a1g (mostly dz2) and eg (mostly dzy, dxz) mainly localized on the magnetic atoms is more controversial, because they are not clearly detected in the photoemission spectra17 and the theoretical predictions are affected by the use of different exchange-correlation functionals, producing different energy sequence and symmetry mixing.27,30–32 In particular, the a1g state with spin-up configuration is expected to be occupied for FePc,31 while its spin-down counterpart is empty, as observed in recent X-ray absorption data at the FeL2, 3 edges.33 In CoPc, recent photoemission data show spectral evidence of the a1g occupation,17 while theoretical calculations predict the presence of partially empty levels in the degenerate eg state for both FePc and CoPc. Thus, for FePc we can attribute the orbital occupation to the empty a1g/eg states, and a more intriguing situation takes place for CoPc: the a1g/eg states become occupied first, while further electron injection causes filling of the 2eg level with π* character, at saturation. The orbital character of the filled formerempty states is confirmed by experimental data taken at two different photon energies, exploiting the different excitation cross sections,23 by comparing the relative valence band peak intensity, as shown in Figure 4. At saturation K-doping, the FePc/Au(110) thin film presents the new-HOMO feature with higher intensity when excited with 40.814 eV photons, in agreement with an enhanced excitation cross-section expected for d-like states,23 confirming its attribution to the electron filling of the a1g/eg

Gargiani et al.

Figure 4. Comparison between UPS spectra collected at 21.218 eV (lines) and 40.814 eV (dots) for FePc (left panel) and CoPc (right panel) at K saturation. Data are normalized to the ex-HOMO level.

states. In the case of CoPc/Au(110) at K saturation, the filled state at 1.57 eV BE has d-character, while the new-HOMO state at 0.89 eV BE has a lower intensity at higher photon energy, being associated to a π-like state like the macrocycle 2eg level. Previous electronic absorption data of electrochemically produced MPc reduction showed that CoPc hosts more than 4 electrons involving both 2eg and Co-related electronic states, while NiPc, CuPc, and ZnPc are saturated with the complete occupation of the 2eg molecular levels.34 In the same work, the participation of Fe-related states is suggested for FePc reduction, in close agreement with our interpretation. As a consequence of different level filling at saturation doping for the measured MPc/Au(110), we do have a different number of injected electrons from K in the different MPc/Au(110) thin films. While those with almost filled d-states associated to the central metal atom (ZnPc, CuPc, NiPc) host about 4 electrons at K saturation, FePc saturates with 2 electrons, while CoPc allocates 5-6 electrons, with complete potassium intercalation. Although these numbers present an uncertainty, they are in close agreement with the previously discussed (Figure 3) intensity ratio measured for the K 3p core-level intensity at saturation. The K intercalation influences the spectral densities of the molecular states, also introducing a different peak broadening, as reported in Table 2 (compared with data in Table 1). The broadening of formerly empty states has been observed already9,15 and attributed to competing effects like structural order, Jahn-Teller-like distortion, and electron-phonon interaction. Although it is not easy to discriminate these effects, we observe a different broadening or a different influence on the molecular states after electron filling. The peaks associated to the delocalized π-like orbitals, namely 2eg ex-LUMO and a1u ex-HOMO, present a wider line width after doping, with respect to the former-occupied states with the same symmetry. On the other hand, the peaks associated to states localized on the central metal atom maintain a comparable line width. The fact that solid MPc films are correlated systems also harbors the explanation of the nonmetallic nature of K-intercalated MPc phases, independently on the symmetry and degeneracy of the empty molecular levels hosting the K-4s electrons. In the case of MPcs with almost filled metal d-states (NiPc, CuPc, and ZnPc), the charge-induced structural relaxation might even be stronger than in many other related systems, because the LUMO is doubly degenerate while the HOMO is not degenerate, which renders the molecule in a negatively charged state susceptible to a Jahn-Teller-like distortion.26 The combination of correlation and structural relaxation effects makes the materials insulating.

Control of Electron Injection Barrier

Figure 5. Energy level diagram for the pristine MPc thin films grown on Au(110) and after K-doping at saturation.

Recent experiments on free-standing MPc thin films (previously prepared on KBr crystals) showed structural phase transitions as a function of K-doping,35,36 from the β-phase of the pristine film, to two further phases corresponding to K2MPc first, and to K4-MPc at saturation. Different structural phases as well have been suggested for K-doped ZnPc/Si.15 These phases are accompanied by modification of the electron energy loss signal, thus of the optical response, due to electron injection to different orbitals involving both ligand and metal d-states.37 In particular, the involvement of the eg and of a metal d-state is assumed for the first K2-FePc phase,37 in agreement with our attribution. However, in our data the K saturation coverage corresponds to K2-FePc, while the K4-FePc is not reached. We expect this difference in the observed saturation doping for the FePc thin films to be due to different pristine structural phase. In our all-UHV-prepared MPc thin films on Au(110), where a highly ordered and flat epitaxial monolayer is formed,16,17 the subsequent layer growth retains the interface flat-lying configuration, as observed in recent preliminary absorption experiments for FePc and CoPc thin films grown on Au(110).18 We remark that Au and Cu single-crystal surfaces generally drive the flat-lying configuration of π-conjugated oligomers,16,38–40 while decoupling the interaction with the metal through buffer layers41–44 or by using noninteracting substrates37,45,46 results in a standing-up molecular configuration. We observe that despite the different structural arrangements, these MPc thin films present the same sequence and electronic spectral density of the relevant molecular states in the valence band region,47 basically due to the low molecule-molecule interaction. However, the structural differences in the achieved thin films (planar vs. standing-up stacking) can allow a different number of intercalated K atoms to be allocated into the thinfilms. In particular, the standing-up stacking seems to favor the allocatin of more K atoms per molecule. Energy Level Diagram and Charge Injection Barrier. The energy shift of the main valence-band features, together with the evolution of the work function upon doping, are schematically summarized in the energy level diagram of Figure 5, synthesizing the previous observations. There is a common HIB (0.36 ( 0.05 eV) measured for all the K-doped MPc/Au(110) at saturation, but CoPc/Au(110) where a lower HIB of 0.24 ( 0.05 eV is estimated (Table 2). In the case of CoPc/Au(110), the filling of both d-like and π-like formerly empty states further reduces the HIB. In the pristine MPc semiconducting thin -films on Au(110), the Fermi level (aligned to that of the Au substrate)

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12263

Figure 6. Work function variation of the MPc thin films grown on Au(110), as a function of K coverage, with the coverage renormalized to the number of transferred electrons. (a) ∆Φ for FePc/Au (open squares) and CoPc/Au (filled upward and downward triangles, from different experimental runs); (b) NiPc/Au (filled dots), CuPc/Au (filled triangles), and ZnPc/Au (open squares).

lies in the energy gap, while upon increasing potassium doping moves it closer to the vacuum level, lowering the work function up to a value corresponding to four potassium atoms for NiPc, CuPc, and ZnPc, about five-six for CoPc, and two for FePc. This issue, also discussed previously while considering the K 3p core-level intensity and the specific empty levels occupied by the injected electrons, allows the coverage scale for the workfunction variation to be redefined. In fact, if we take into account the actual K occupancy at saturation, we can redraw the ∆Φ as a function of the number of doping K atoms, as shown in Figure 6. One can clearly observe that occupation of the a1g/eg states for FePc provides a -∆Φ value compatible with half occupancy with respect to the MPc with filled d-band central metals, while a further level of occupation (CoPc/Au) produces an even stronger reduction of the ionization energy when the d-related orbitals and the 2eg states are involved. The shift of all valence band features is lower than the shift (in absolute value) of the work function for FePc. Thus, its valence band moves closer to the vacuum level with increasing potassium doping, meaning that we do not observe a rigid shift of the electronic levels of the valence band. This fact supports the assumption of a relaxation of the molecular structure and variation of its electronic levels, due to the charge transfer to the molecules. Finally by comparing the measured ionization potential energy with respect to the ionization energy of gas-phase molecules, we estimate a polaronic relaxation of the valence levels by about 1-1.5 eV, in agreement with the expected IE difference between free molecules and condensed phase. Conclusions Potassium doping of MPc thin films grown on Au(110) gives rise to electron injection depending on the central metal atom. In particular, while for the MPcs with filled d-band metals the electrons are transferred to the former-LUMO with 2eg symmetry, in FePc and CoPc the charge injection process is driven by the actual empty states close to the Fermi level. In particular, FePc hosts electron charge on the a1g/eg states, presenting a wide d-like character from Fe, and CoPc first attracts electrons into the a1g/eg states, then it allocates further charge in the 2eg state. As a consequence of the different orbital sequence in the different MPc thin films, K-intercalation transfers about 2 electrons in FePc, 5-6 electrons in CoPc, and 4 electrons in

12264

J. Phys. Chem. C, Vol. 114, No. 28, 2010

NiPc, CuPc, and ZnPc. The electron charge donation at saturation drives the energy level diagram of all presently studied MPcs/Au(110) to a reduced hole-injection barrier, much lower than in the pristine MPc clean films. Finally, the main energy gap for these K-doped MPc thin films is constituted by different symmetry of the new-HOMO states, useful for possible tailoring of the transport/optical response of these systems. The different response to electron injection of the MPcs with different central d-metal occupancy, in particular for FePc, opens perspectives and a new scenario for controlling and tailoring the electronic and eventually magnetic response of organometallic thin films. References and Notes (1) Forrest, S. R. Nature 2004, 428, 911–918. (2) Walzer, K.; Maennig, B.; Pfeiffer, M.; Leo, K. Chem. ReV. 2007, 107, 1233–1271. (3) Kahn, A.; Koch, N.; Gao, W. J. Polym. Sci., Part B: Polym. Phys. 2003, 41, 2529–2548. (4) Gao, Y.; Yan, L. Chem. Phys. Lett. 2003, 380, 451–455. (5) Ding, H.; Park, K.; Gao, Y. J. Electron Spectrosc. Relat. Phenom. 2009, 174, 45–49. (6) Craciun, M.; Rogge, S.; den Boer, M.-J.; Margadonna, S.; Prassides, K.; Iwasa, Y.; Morpurgo, A. AdV. Mater. 2006, 18, 320–324. (7) Craciun, M. F.; Rogge, S.; Morpurgo, A. F. J. Am. Chem. Soc. 2005, 127, 12210–12211. (8) Phthalocyanines: properties and applications; Leznoff, C. C., Lever, A. B. P., Eds.; Wiley-VCH: Weinheim, Germany, 1989. (9) Schwieger, T.; Peisert, H.; Golden, M.; Knupfer, M.; Fink, J. Phys. ReV. B 2002, 66, 155207. (10) Betti, M.; Crispoldi, F.; Ruocco, A.; Mariani, C. Phys. ReV. B 2007, 76, 125407. (11) Ding, H.; Park, K.; Green, K.; Gao, Y. Chem. Phys. Lett. 2008, 454, 229–232. (12) Calabrese, A.; Floreano, L.; Verdini, A.; Mariani, C.; Betti, M. Phys. ReV. B 2009, 79, 115446. (13) Schwieger, T.; Knupfer, M.; Gao, W.; Kahn, A. Appl. Phys. Lett. 2003, 83, 500. (14) Gao, W.; Kahn, A. J. Phys.: Condens. Matter 2003, 15, S2757– S2770. (15) Giovanelli, L.; Vilmercati, P.; Castellarin-Cudia, C.; Themlin, J.M.; Porte, L.; Goldoni, A. J. Chem. Phys. 2007, 126, 044709. (16) Floreano, L.; Cossaro, A.; Gotter, R.; Verdini, A.; Bavdek, G.; Evangelista, F.; Ruocco, A.; Morgante, A.; Cvetko, D. J. Phys. Chem. C 2008, 112, 10794–10802. (17) Gargiani, P.; Angelucci, M.; Mariani, C.; Betti, M. G. Phys. ReV. B 2010, 81, 085412. (18) Betti, M. G.; Cossaro, A.; Frisenda, R.; Gargiani, P.; Mariani, C.; Verdini, A. To be submitted for publication. (19) Rosa, A.; Baerends, E. J. Inorg. Chem. 1994, 33, 584–595. (20) Liao, M.-S.; Scheiner, S. J. Chem. Phys. 2001, 114, 9780.

Gargiani et al. (21) Ellis, T. S.; Park, K. T.; Ulrich, M. D.; Hulbert, S. L.; Rowe, J. E. J. Appl. Phys. 2006, 100, 093515. (22) Grobosch, M.; Aristov, V. Y.; Molodtsova, O. V.; Schmidt, C.; Doyle, B. P.; Nannarone, S.; Knupfer, M. J. Phys. Chem. C 2009, 113, 13219–13222. (23) Yeh, J.; Lindau, I. At. Data Nucl. Data Tables 1985, 32, 1–155. (24) Evangelista, F.; Carravetta, V.; Stefani, G.; Jansik, B.; Alagia, M.; Stranges, S.; Ruocco, A. J. Chem. Phys. 2007, 126, 124709. (25) Berkowitz, J. J. Chem. Phys. 1979, 70, 2819. (26) Tosatti, E.; Fabrizio, M.; To´bik, J.; Santoro, G. E. Phys. ReV. Lett. 2004, 93, 117002. (27) Calzolari, A.; Ferretti, A.; Nardelli, M. B. Nanotechnology 2007, 18, 424013. (28) Reynolds, P. A.; Figgis, B. N. Inorg. Chem. 1991, 30, 2294–2300. (29) Kroll, T.; Aristov, V. Y.; Molodtsova, O. V.; Ossipyan, Y. A.; Vyalikh, D. V.; Bu¨chner, B.; Knupfer, M. J. Phys. Chem. A 2009, 113, 8917–8922. (30) Marom, N.; Kronik, L. Appl. Phys. A: Mater. Sci. Process. 2009, 95, 159–163. (31) Marom, N.; Kronik, L. Appl. Phys. A: Mater. Sci. Process. 2009, 95, 165–172. (32) Kuz’min, M. D.; Hayn, R.; Oison, V. Phys. ReV. B 2009, 79, 024413. (33) Bartolome´, J.; Bartolome´, F.; Garcia, L. M.; Filoti, G.; Gredig, T.; Colesniuc, C. N.; Schuller, I. K.; Cezar, J. C. Phys. ReV. B 2010, 81, 195405. (34) Clack, D. W.; Yandle, J. R. Inorg. Chem. 1972, 11, 1738–1742. (35) Roth, F.; Ko¨nig, A.; Kraus, R.; Knupfer, M. J. Chem. Phys. 2008, 128, 194711. (36) Flatz, K.; Grobosch, M.; Knupfer, M. J. Chem. Phys. 2007, 126, 214702. (37) Ko¨nig, A.; Roth, F.; Kraus, R.; Knupfer, M. J. Chem. Phys. 2009, 130, 214503. (38) Baldacchini, C.; Mariani, C.; Betti, M. G. J. Chem. Phys. 2006, 124, 154702. (39) Baldacchini, C.; Mariani, C.; Betti, M. G.; Gavioli, L.; Fanetti, M.; Sancrotti, M. Appl. Phys. Lett. 2006, 89, 152119. (40) Ferretti, A.; Baldacchini, C.; Calzolari, A.; Di Felice, R.; Ruini, A.; Molinari, E.; Betti, M. Phys. ReV. Lett. 2007, 99, 046802. (41) Kanjilal, A.; Ottaviano, L.; DiCastro, V.; Beccari, M.; Betti, M.; Mariani, C. J. Phys. Chem. C 2007, 111, 286–293. (42) Chiodi, M.; Gavioli, L.; Beccari, M.; Di Castro, V.; Cossaro, A.; Floreano, L.; Morgante, A.; Kanjilal, A.; Mariani, C.; Betti, M. Phys. ReV. B 2008, 77, 5321. (43) Betti, M. G.; Kanjilal, A.; Mariani, C. J. Phys. Chem. A 2007, 111, 12454–12457. (44) Betti, M.; Kanjilal, A.; Mariani, C.; Va´zquez, H.; Dappe, Y.; Ortega, J.; Flores, F. Phys. ReV. Lett. 2008, 100, 027601. (45) Halik, M.; Klauk, H.; Zschieschang, U.; Schmid, G.; Dehm, C.; Schu¨tz, M.; Maisch, S.; Effenberger, F.; Brunnbauer, M.; Stellacci, F. Nature 2004, 431, 963–966. (46) Thayer, G.; Sadowski, J.; Meyer Zu Heringdorf, F.; Sakurai, T.; Tromp, R. Phys. ReV. Lett. 2005, 95, 256106. (47) Lozzi, L.; Santucci, S.; La Rosa, S.; Delley, B.; Picozzi, S. J. Chem. Phys. 2004, 121, 1883–1889.

JP103946V