High-Pressure Optical Properties and Chemical Stability of Picene

Article pubs.acs.org/JPCC High-Pressure Optical Properties and Chemical Stability of Picene Samuele Fanetti,† Margherita Citroni,*,†,‡ Lorenzo Malavasi,§ Gianluca A. Artioli,§ Paolo Postorino,¶ and Roberto Bini†,‡ † LENSEuropean Laboratory for Nonlinear Spectroscopy, via N. Carrara 1, 50019 Sesto Fiorentino (FI), Italy Dipartimento di Chimica, Università di Firenze, via della Lastruccia 3, 50019 Sesto Fiorentino (FI), Italy § Dipartimento di Chimica, Università di Pavia, v. le Taramelli 12, 27100 Pavia, Italy ¶ CNR-IOM and Dipartimento di Fisica, Università di Roma Sapienza, p. le A. Moro 2, 00185 Roma, Italy ‡ ABSTRACT: Picene is a polycyclic aromatic hydrocarbon belonging to the class of phenacenes which have been recently found to behave as hightemperature superconductors upon alkali metal doping. The electronic properties of organic crystals can be finely and largely modified by the density changes obtained by the application of an external pressure. In this work, the role of pressure in tuning the optical properties of crystalline picene has been investigated from room conditions up to 15 GPa through the measurement of UV−visible absorption spectra, two-photon excitation profiles, and one- and two-photon excited fluorescence spectra in a diamond anvil cell. The pressure dependence of the optical band gap was determined, and the frequencies of several vibronic bands belonging to electronic transitions from the ground state (S0) to the four lowest-energy excited singlet states (S1 to S4) were determined as a function of pressure. We evidence a very different density dependence of the transition energy of S0 → S1, which undergoes a remarkable red shift of ∼400 cm−1/GPa, and of the transitions from S0 to the higher excited states, which remain constant in the whole investigated range. This is consistent with a S1 state of 1La character in solid picene. The high-pressure chemical stability of solid picene was investigated through visible absorption and Fourier transform infrared spectroscopy (FTIR). A chemical transformation involving the bulk picene crystal occurs above ∼23 GPa, giving rise to a disordered material similar to the amorphous hydrogenated carbon obtained in the pressure-induced reactivity of benzene. The combination of electronic and vibrational data allows us to identify the presence of reaction intermediates at ∼10 GPa, preferentially forming at crystal defects. It is known that the volume decrease induced by an applied static pressure in the kilobar to megabar range may dramatically alter the electronic structure of molecular systems,10−14 with also remarkable implications in their chemical reactivity.15,16 Depending on the system under examination, a range of phenomena related to modifications of the electronic structure may be observed due to the increase of intermolecular interactions, as for instance a different energy shift of the energy levels,17,18 a change in the equilibrium geometry of the ground and excited electronic states,19 and a change in the binding character of a given electronic state.20 Pressure may also be the decisive parameter in achieving superconductivity in carbon-based materials through the enhancement of the involved coupling mechanisms.21,22 The superconducting doped solid K3picene, with Tc = 18 K at room conditions, has been recently found to have a large positive pressure dependence of Tc (12.5 K/GPa up to 1.2 GPa).8 INTRODUCTION Organic molecular crystals have appealing electronic and optical properties that can be achieved or tuned by doping with guest moieties, as is the case for the superconductivity recently discovered in some alkali-metal-doped aromatic hydrocarbons,1−3 including picene.4 The picene molecule is a condensed aromatic hydrocarbon made of five benzene rings fused in an armchair configuration. This molecular geometry gives rise to a large resonance stabilization, reflected for instance in the higher chemical stability and larger band gap with respect to the linearly arranged pentacene molecule.5 The crystal structure of picene6 is monoclinic, P21, Z = 2, made by layers parallel to the ab planes of molecules arranged in a herringbone structure, with their long molecular axis roughly parallel to the c-axis. Besides the changes in the electronic structure due to the addition of the alkali atoms,7 doping induces structural modifications on the picene crystal including a 6% volume reduction, mainly involving a contraction along the b- and c-axes4 (different synthetic methods, however, lead to materials with different structural parameters8). Comparable volume changes are obtained with compressions to ∼1 GPa in anthracene, tetracene, and pentacene.9 ■ © 2013 American Chemical Society Received: January 21, 2013 Revised: February 13, 2013 Published: February 20, 2013 5343 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351 The Journal of Physical Chemistry C Article Figure 1. Microphotograph of the sample used in UV−vis absorption measurements, consisting of a picene single crystal in Ar as a pressure medium. The initial gasket diameter is 135 μm. Left: at 0.56 GPa after loading. Right: at 14.0 GPa. as a pressure medium. The measurements were performed using an unfocused halogen lamp as the source and measuring transmitted light, collected through a ultrashort Mitutoyo 20× objective and a spatial filter, with a CCD detector coupled to a single-stage monochromator (300 grooves/mm). The detection setup allows a spatial resolution better than 3 μm, and a spectral resolution better than 0.1 nm. At each pressure step, a reference spectrum was measured detecting the light transmitted by Ar, and the pressure was measured by the ruby fluorescence technique, using a few microwatts of the 647 nm line of a Kr+ laser. FTIR absorption measurements were performed on polycrystalline samples using a Bruker-IFS 120 HR spectrometer modified for high-pressure measurements,27 with an instrumental resolution of 1 cm−1. The pressure was either measured by the ruby fluorescence technique using the second harmonic (532 nm) of a continuous-wave Nd:YAG laser, or using a thin film of a 2% (in weight) mixture of NaNO2 in NaBr as the pressure gauge, following the frequency versus pressure calibration of the nitrite antisymmetric stretching mode as reported in ref 28. The use of an infrared pressure gauge was necessary to measure pressures above ∼15 GPa, due to the strongly increasing absorptivity of the sample with pressure, and to separate possible photochemical effects due to the laser radiation. The fluorescence measurements were performed on polycrystalline samples using a picosecond tunable source, and detecting the fluorescence in a backscattering geometry, with a setup described in detail in ref 19. In the present work, the beam was focused with a 100 mm focal length achromatic doublet obtaining a beam-waist diameter of comparable dimensions with the gasket aperture and a depth of focus longer than the sample thickness. A parabolic Al mirror was used to collect fluorescence, which was focused on a monochromator and measured by a cooled Hamamatsu R943-02 photomultiplier. For the fluorescence measurements, the sample was loaded with no pressure gauge and the pressure was determined by measuring the FTIR spectrum and using the frequency shift of the infrared absorption bands of picene calibrated by the ruby fluorescence method. Fluorescence spectra were measured with a resolution of ∼1 nm. The sampling of the excitation profiles was set to 1 nm. In this work, we investigate the optical properties of pristine picene up to 14 GPa in order to separately study the effect of density in tuning the electronic states, and to gain a deeper insight into the pressure dependence of the electronic properties of aromatic crystals. Since the discovery of superconductivity in K-doped picene, several theoretical and experimental studies have been published concerning the electronic structure of pure and doped solids (see, for example, refs 23 and 24 and references therein). Picene is shown to be a typical molecular crystal, with a band structure dominated by the molecular electronic states, but electronic correlations and anisotropy must be taken into account for an accurate description. A possible energy superposition or inversion of the two lowest-energy singlet excited states has been proposed to explain the enhanced optical density of the lowest-frequency observed absorption system relative to the higher-frequency systems.25 The different pressure dependence of the energies and cross sections of different transitions may be useful to qualitatively determine the relative extent of the changes in dipole moments and polarizabilities of the involved electronic states, thus being of help in their assignment. In the present work we have measured the UV−visible absorption edge, oneand two-photon excited fluorescence spectra, and two-photon excitation profiles of picene up to 14 GPa in a diamond anvil cell, thus obtaining the frequency versus pressure evolution of the optical band gap, and the pressure evolution of higherenergy electronic excited states deriving from HOMO−LUMO excitations. Moreover, we have investigated the high-pressure chemical stability upon compression up to 45 GPa through visible absorption and Fourier transform infrared spectroscopy (FTIR), identifying a reaction product and the conditions that trigger the room temperature reactivity. EXPERIMENTAL SECTION Picene was prepared by an optimized synthesis route as described in ref 26. All the measurements were performed in membrane diamond anvil cells (DAC) equipped with ultra-lowfluorescence IIa type diamonds. Rhenium gaskets were used, with initial dimensions of 50 μm in thickness and 150 μm in diameter. For the UV−vis absorption measurement, a single crystal of picene of dimensions 10 × 10 × 100 μm3 (prepared by crystallization from a chloroform solution) was loaded using Ar ■ 5344 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351 The Journal of Physical Chemistry C Article the sample that was in contact with the gasket and at the central part that was damaged upon loading. The sample at 14.0 GPa is shown in Figure 1. The spectra measured in these regions have a red-shifted absorption edge and a larger pressure dependence than those measured on picene (Figure 2) and were not reversible with the pressure download. These data indicate that a transformation of the pristine material occurs, preferentially at crystal defects and not reversible with pressure. The optical band gap in a molecular crystal is the wavelength of the maximum of the 0−0 band of the lowest-energy electronic transition (exciton band). Due to saturation, it is impossible to determine its value in the absorption spectra of our sample. This is usually the case for pure solid samples in the DAC, with typical thickness >10 μm. In order to describe the pressure dependence of the optical gap one may choose a fixed absorbance value and follow the pressure evolution of the wavelength at which this value is reached. Since the pressure may also affect the intensity and bandwidth of the electronic transitions, in addition to their frequency, the pressure shift of the optical gap may be overestimated in the presence of absorption saturation,29 while the measurement of excitation profiles gives in general more reliable results.19 In Figure 3 we In the polycrystalline samples, the crystal dimensions were not resolved under an optical microscope with resolution of about 1 μm, and scattering of light was observed as a background in the FTIR spectra at wavelengths shorter than 1 μm, strongly increasing with the radiation frequency. We can thus confidently assume that the crystallites were of submicrometric size. RESULTS Optical absorption. The UV−visible absorption spectrum in the region 370−500 nm was measured as a function of pressure up to 14 GPa on a single crystal using Ar as a pressure medium (Figure 1). At each pressure, several spectra were measured at different points of the crystal and the absorbance was calculated using as reference the transmission spectrum measured on Ar. The spectra (Figure 2) become saturated from just above the absorption edge, due to the high absorptivity of solid picene25 ■ Figure 3. Pressure evolution of the bandgap of picene, taken as the wavelength at which the absorbance of the sample reached a value of 0.36 (see text). Full symbols: compression. Open symbols: decompression . report the pressure dependence of the wavelength at which the absorbance of the transparent edge of the single crystal is 0.36. This is the absorbance value at which the lowest-pressure spectrum is saturated. The pressure dependence is approximately linear, with a slope of ∼8 nm/GPa (∼400 cm−1/GPa). By choosing different absorbance values to define the optical gap, we obtain the same slope of the wavelength as a function of pressure, which apparently indicates that the bandwidth and intensity of the exciton band are not appreciably modified by pressure with respect to how much the frequency is affected. This result is also confirmed by the fluorescence measurements, as described in the next paragraph, where the wavelength of the 0−0 fluorescence band is found to have the same pressure dependence. The spectrum of the transparent, nontransformed part of the sample is perfectly reversible with pressure (Figure 3). On the other hand, as shown in the lower panel of Figure 2, the absorption spectrum of the transformed part undergoes a blue shift down to 7.8 GPa and more important modifications at lower pressures, where it becomes progressively weak in the Figure 2. Absorption spectra of picene single crystal at different pressures, indicated in the figure in GPa units. (a) Spectra measured during compression on the transparent end of the sample (black lines) and on the red part of the sample (red lines). (b) Spectra measured during decompression on the red part of the sample. The spectra measured at P ≤ 3.59 GPa are reported in different colors to emphasize the remarkable spectral changes occurring with pressure. and to the sample thickness. The absorbance value at which saturation occurs depends on the source intensity and detector sensitivity, both of them rapidly increasing going to longer wavelengths. The spectrum of picene could be measured up to 14 GPa and was found to be perfectly reversible during the pressure download. However, at 11 GPa some regions of the sample acquired a reddish coloration, and their extent and absorbance increased with increasing the pressure. These regions can be identified as strained or defected parts of the crystal. In fact, the reddish coloration started at the extremity of 5345 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351 The Journal of Physical Chemistry C Article blue region and more intense at λ > 480 nm, indicating a possible chemical transformation. Fluorescence Spectra and Excitation Profiles. Fluorescence measurements were performed on polycrystalline samples loaded with no pressure medium, at several pressures from 0.1 to 12 GPa (Figure 4). After compression to 6.5 GPa, Figure 5. Deconvolution of the fluorescence spectrum at 0.1 GPa, two-photon-excited at 588 nm. Figure 4. Fluorescence spectra (left panel) and two-photon-excitation profiles (right panel) of polycrystalline picene measured at different pressures. The pressure values, in GPa units, are indicated above each spectrum. In the left panel, solid lines indicate TP excited spectra (excitation in the range 620−670 nm) whereas dotted lines indicate OP excited spectra (excitation at 440 nm). In the right panel, color lines indicate the different excited states reached by TP excitation as a function of excitation wavelength (see text). Red, S1; green, S2; blue, S3; and magenta, S4. the pressure was decreased to 4.5 GPa and then increased again. This cycle was performed to check the reversibility of the fluorescence and excitation profiles in a pressure range below the onset of the relevant changes observed in the absorption spectrum, and they were found to be perfectly reversible. The dependence of the fluorescence signal intensity upon the laser power is linear or quadratic, respectively, when fluorescence follows a one-photon (OP) or a two-photon (TP) absorption process. Thus, at each pressure step, the fluorescence intensity dependence upon the laser power was checked at different excitation wavelengths to determine whether excitation followed TP or OP process. At each pressure, the TP excitation profile was measured detecting the fluorescence around its maximum, whereas several fluorescence spectra were measured using different excitation wavelengths. The fluorescence spectrum was always found to be independent of the excitation wavelength, indicating a relaxed fluorescence. The only difference is found between TP and OP excited fluorescence spectra, the latter having a higher intensity in the blue region. As already discussed in ref 25, this OP excited blue emission is probably due to the formation of excited states at surface defects, particularly relevant in nanocrystalline samples. The TP excited fluorescence spectrum measured at the lowest pressure is superimposable to that measured on a similar nanocrystalline sample at ambient conditions.25 A fit of the spectrum with Voigt profiles gives five bands, approximately equally spaced by ∼1250 cm−1, as shown in Figure 5 and reported as a function of pressure in Figure 6. In fitting the fluorescence spectra, the bands were fixed to have an equal line shape, and constant with pressure. The full-width- Figure 6. Peak frequencies as a function of pressure in the twophoton-excitation profiles (black; different symbols indicate vibronic transitions likely belonging to different excited electronic states) and in the fluorescence spectra (red), along with the frequencies representing the optical band gap (blue) obtained in the absorption measurements and also reported in Figure 3. half-maximum (fwhm) resulted to be ∼1350 cm−1 for all of the bands excepted for the 0−0 band, that was ∼450 cm−1. This is probably ascribable to a partial reabsorption of fluorescence by the sample. With increasing the pressure, we observe a red-shift of the whole spectrum. As shown in Figure 6, this frequency shift is in perfect agreement with that observed in the absorption measurements, reflecting the energy shift of the S0 → S1 transition. At pressures above 9 GPa, OP excited fluorescence spectra were used for the analysis, because, as discussed below, the TP excitation wavelengths in the accessible range increasingly overlapped the emission band. 5346 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351 The Journal of Physical Chemistry C Article dependence as for the optical band gap, only red-shifted due to the much higher sensitivity of the measurements, detecting a large fluorescence signal even for very low absorbance values. At 10.8 GPa the vibronic structure of the TP excitation profile of picene is lost, and at 12 GPa a different TP excitation profile is observed. The excitation profile of picene is not recovered with pressure download. In fact, as was observed in the UV/vis absorption measurements, a transformation of the pristine material occurs at crystal defects, which are particularly numerous in this highly polycrystalline sample. FTIR Spectra and Reactivity. FTIR experiments have been primarily performed for the purpose of unveiling if the changes in the UV−vis absorption properties occurring above 11 GPa correspond to a chemical reaction, and, in that case, of identifying the reaction product. In addition, the IR experiments were used to calibrate the frequencies of picene vibrational modes as a function of pressure, in order to use them as an internal pressure gauge in the fluorescence measurements. Four polycrystalline samples were investigated. Two of them were compressed up to 16 GPa and decompressed, and the pressure was measured by the ruby fluorescence technique. The strong visible absorption of picene prevented an accurate measurement of pressure by the ruby fluorescence technique above this pressure because of strong absorption of the laser beam and emission by the sample, and reabsorption of the ruby fluorescence by the transformed sample. For this reason, and also to separate the effect of irradiation with visible light from the effect of pressure on chemical reactivity, one sample was loaded with an infrared pressure gauge (see Experimental Section), compressed to 23.4 GPa, and decompressed. To further check if the presence of the salt has an effect on reactivity, a fresh sample was loaded with no pressure gauge, compressed to about 45 GPa, and then decompressed. In this case, the pressure was estimated using the infrared bands of picene as calibrated in the other experiments, and the highest pressure values were extrapolated. The room-pressure IR spectrum is in perfect agreement with the previously reported data.26,30 All the IR bands have a blue shift with pressure, with different slopes ranging from 1 to 2.5 cm−1/GPa. No discontinuities of the frequencies versus pressure evolution and no spectral changes indicating a phase transition were observed, suggesting that the same crystal structure is maintained in the whole investigated pressure range. All the samples acquired an orange coloration (in transmission with a microscope, as shown in Figure 8) at about 9 GPa, and became increasingly red on further compression, appearing black with reflected light. The coloration was only partially reversible with pressure, in agreement with the results of the UV−vis absorption measurements showing an additional contribution to the sample visible absorption above 11 GPa, not reversible with pressure download. The slightly lower pressure value (9 GPa) for the appearance of the red color with respect to the single crystal used for the absorption measurements (11 GPa) is probably due to the larger amount of crystal defects in the polycrystalline samples. However, when the sample is compressed to 16 GPa and decompressed, the IR spectrum is reversible with pressure, with only a slight broadening of the absorption bands due to the residual strain, and no evidence of a chemical reaction is gained. Instead, as shown in Figure 9, the presence of a reaction product is revealed when the pressure exceeds ∼23.5 GPa by a broad absorption centered at 3000 cm−1. The intensity of this band The TP excitation profile measured at the lowest pressure is the same as measured on a powder at room conditions25 (Figures 4 and 7). The longest available wavelength from our Figure 7. Deconvolution of the two-photon-excitation profiles measured at 0.1, 3.1, and 9.5 GPa. source is 680 nm, which corresponds to excitation at 340 nm by a TP process. At room pressure, the transitions belonging to the S0 → S1 system fall in the region 382−330 nm.25 However, the lowest-energy band in the TP excitation profile is at 638 nm, and it was assigned to the electronic origin of the S0 → S2 transition,25 while S0 → S1 has a too low TP absorption cross section to be observed in our experiments. A well-defined vibronic structure is observed, as shown in Figures 4 and 7, due to transitions from the ground state to to S2, S3, and S4. Likely, the two bands at 622 and 609 nm belong to a vibronic series starting at the 0−0 band of the S0 → S2 system and built on a A1 vibrational mode with frequency ∼700 cm−1. The peak at 590 nm can be assigned to the origin of the S0 → S3 transition,25 and the bands at 564 and 546 nm probably belong to a vibronic series starting at the 0−0 band and built on a A1 vibrational mode with frequency ∼1300 cm−1. In fact, the ν29 mode of A1 symmetry with frequency 730 cm−1 and several A1 modes with frequencies between 1200 and 1350 cm−1 in S026,30 could be responsible for the vibronic progressions in the two cases. The two peaks at 504 and 519 nm can be likely assigned to TP excitation to S4 due to the large frequency separation (∼2000 cm−1) between the peak at 504 nm and the highest-frequency peak assigned to the S0 → S3 transition. This structure does not undergo any substantial change up to 9.5 GPa, remaining well-defined and not appreciably shifted in frequency. As can be seen in Figure 4, a steep increase of the signal is detected on the high-energy side of the excitation profiles, with a rapidly red-shifting edge with increasing pressure. The signal intensity measured on this edge has a linear dependence on the laser beam intensity, indicating that it is due to OP excitation, so that it can be assigned to the onephoton absorption edge to S1. In fact, if we plot the frequency at which the signal dependence upon laser intensity deviates from quadraticity as a function of pressure, we obtain the same 5347 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351 The Journal of Physical Chemistry C Article increased with increasing pressure and further increased upon decompression. The same evolution of the IR spectrum with compression was observed in both the samples compressed above 23 GPa, indicating that the presence of the IR pressure gauge did not appreciably affect the kinetics or the energetics of the reaction. Also, the coloration of the sample occurred at the same pressure in all the samples, indicating that even laser irradiation did not appreciably affect the chemical properties of picene. The final amount of reaction product after decompression depends on the maximum pressure reached in the experiment. The sample recovered after compression to 23.5 GPa contains a much larger quantity of unreacted picene than the sample that was compressed to 45 GPa. To quantify the extent of reaction, we can calculate the ratio of the integrated area before and after the compression/decompression cycle of an infrared absorption band of picene in the two experiments, which gives the fraction of unreacted sample. The narrow band at 680 cm−1 (room pressure value) is a good choice for this calculation because it is in scale in the pristine material but it is clearly detectable even in the presence of a low concentration of picene. In the sample that was compressed up to 23.5 GPa, after the complete pressure download the area of this band is 35% of the initial value, whereas in the sample compressed to 45 GPa it is only 8% of the initial value. The spectrum of the recovered sample at room pressure, obtained after compression to 45 GPa, can be compared to the spectrum of the amorphous hydrogenated carbon (a:CH) obtained from the high-pressure reaction of benzene.31 The two spectra are very similar, the main difference being the smaller relative intensity of the C−H stretching in our spectra, mainly ascribable to the smaller hydrogen content in picene with respect to benzene. Differently from benzene, which reacted completely upon decompression once the reaction was triggered by pressure (P ≥ 40 GPa in annealed samples32), the reaction is not complete in picene, as already noted. The absorptions at 766, 826, 870, and 960 cm−1 are due to the residual of the very strong absorptions (multiplets) of picene centered at 750, 820, 860, and 945 cm−1. Higher frequency absorptions of picene, in the range 1000−1800 cm−1, are not as intense as those mentioned but still contribute to the spectrum of recovered sample. The underlying spectrum is very similar to that of a:CH, where the 700−1000 cm−1 region can be assigned to CH or CH2 rocking and wagging modes, the 1130−1530 cm−1 region to C−C stretching and other CH or CH2 bending modes, and the absorption at 1615 cm−1 to CC stretching.32 The integrated area of the latter is 53% of that of the picene multiplet before compression, centered at 1600 cm−1, indicating that it is mainly due to a CC stretching of the product. In the C−H stretching region, two bands are observed at 2920 and 3030 cm−1, respectively, assigned to vibrations involving C sp3 and C sp2 atoms, so that the ratio of the integrated areas of the two bands, corrected by the different transition moments of the two types of vibration, can be used to determine the ratio of saturated to unsaturated carbon atoms.31 In our spectrum we have to consider that the residual picene contributes to the intensity of the C sp2 band. So we subtracted the 8% of the integrated area of the multiplet at ∼3000 cm−1 in the spectrum of the starting picene to the integrated area of the 3030 cm−1 peak in the spectrum of the recovered product. After this subtraction, the intensity ratio is 2.0, corresponding to a C sp3/C sp2 ratio of 1.3, whereas in the high-pressure reaction of benzene it was 3.2.31 This is in Figure 8. Microphotographs of polycrystalline picene in the DAC during compression at 3.4 GPa (a), 9.3 GPa (b), 12.3 GPa (c), and 15.2 GPa (d), and during decompression at 10 GPa (e) and after complete pressure download (f). Figure 9. Solid lines: Selected FTIR spectra of polycrystalline picene measured at different pressures during compression (black lines) and decompression (red lines). Pressure values are indicated in GPa units. The star indicates the most intense band of the product spectrum, used to identify the reaction threshold pressure. Blue line: FTIR spectrum of the amorphous hydrogenated carbon obtained from the pressure-induced reaction of benzene (ref 31). 5348 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351 The Journal of Physical Chemistry C Article distributions, which are differently stabilized by intermolecular interactions. For instance, an electronic state with a nonzero dipole moment is stabilized by the crystalline environment in a similar way as it would be in a polar solvent, and its energy is expected to lower linearly with increasing the pressure, within the dipole approximation. Even if the involved electronic states have a zero dipole moment, as is the case of linear acenes in the ground-state geometry, their different polarizabilities determine a different extent of stabilization in the solid through dispersion forces, and thus a shift of the vertical transition frequency (observed in absorption spectra and excitation profiles) upon crystallization and with increasing density. The frequency shift of the transition depends on the relative stabilization of the involved electronic states, and thus it is not a pure information on the excited state. However, it has been shown by TD-DFT calculations35 that the S0 → 1La vertical transition in polycyclic aromatic hydrocarbons involves a larger degree of local charge reorganization than the S0 → 1Lb vertical transition, leading to a more polarizable electronic distribution having a charge-transfer or charge-separation character. This finding is in accordance with the larger density dependence of the absorption spectra of aromatic systems with a La than those with a Lb lowest-energy excited state. In fact, the pressure evolution of the lowestenergy singlet electronic transition has been studied through fluorescence or optical absorption measurements for benzene and for several fused-ring aromatic hydrocarbons. The band gap of solid benzene (where S1 is 1Lb) has a red shift of 60 cm−1/ GPa in the pressure range 0−14 GPa.19 In linear acenes, from anthracene to hexacene (where S1 is 1La), a more pronounced pressure dependence of the transition energy is observed.38−40 In pentacene, for instance, the absorption edge undergoes a red shift of ∼650 cm−1/GPa in the range from ambient pressure to 7 GPa,39 and a chemical reaction, involving the formation of saturated bonds, occurs above 10 GPa. The band gap of picene undergoes a red-shift of 400 cm−1/GPa, entering the visible region already at 2 GPa. This red shift is similar to that observed for the 1La states in the cited examples, giving support to the hypothesis that in solid picene S1 derives from the 1La molecular state. This cannot be a definitive assignment, because the extent of the density stabilization of a polar or polarizable electronic state also depends on the particular equation of state of the molecular system, which is at present not available for picene. However, it is useful to compare the behavior of the S0 → S1 transition with the transitions from S0 to the higherenergy states, observed in the TP excitation profiles, whose frequencies remain constant with pressure. It is therefore evident that S1 is characterized by a much larger stabilization by intermolecular forces than the higher-energy states, which would be consistent with a 1La electronic distribution. Electron energy loss spectroscopy (EELS) spectra measured on solid picene at room pressure as a function of momentum transfer41 showed that the band observed at 3.61 eV (29 000 cm−1) is characterized by a charge-transfer character, differently from the adjacent bands that were assigned to exciton bands. This band was observed in optical absorption measurements on picene at room pressure25 and assigned to a vibronic component of the S0 → S1 transition. The change in the intensity pattern and the frequency shift with respect to the solution spectra suggested that the whole electronic transition, and not just a vibronic component, is strongly affected by intermolecular interactions in the solid state, and this is confirmed by the present work. agreement with the higher intensity of the CC stretching band of our sample. DISCUSSION The absorption spectrum of solid picene has been recently related to that of the isolated picene molecule, measured in dilute solutions.25 In the isolated molecule, the four lowestenergy electronic excited singlet states, S1 to S4, can be labeled as 1Lb, 1La, 1Ba, and 1Bb, respectively, according to a nomenclature based on the Platt’s perimeter model.25,33 This nomenclature is widely used because it is independent of symmetry and allows to rationalize and relate the electronic spectra of very different aromatic molecules.34,35 The 1La state is mainly a HOMO−LUMO excitation and has electronic densities mainly localized on the atoms of the aromatic system, and its energy is strongly dependent on the chemical environment, on the molecular geometry, and on the presence of substituents. The 1Lb state derives from the mixing of HOMO−1−LUMO and HOMO−LUMO+1 excitations and has nodal planes on the atoms of the aromatic system. Transitions from S0 to 1Lb are 10−100 times weaker than those to 1La (in the perimeter model; they are forbidden by the selection rule Δk = ±1, where k is a quantum number related to the z-component of the angular momentum of the electron involved in the transition). Due to the different electronic distributions and polarizabilities of the two states, their relative energies depend on the molecular structure and on the environment. In the picene isolated molecule S1 is 1Lb.35,36 In the C2v molecular symmetry 1Lb and 1Bb belong to the A1 irreducible representation, while 1La and 1Ba belong to B2. In the crystal (site group C1, factor group C2), under the assumption that the crystal field acts as a perturbation on the molecular electronic properties, each molecular state is split into two states of A and B symmetry. However, the coupling of the molecular states in the solid may result in additional, even more important, effects, such as different frequency shifts for the different states with respect to the single-molecule level scheme. For instance, in pyridine an energy inversion between the nπ* and the first ππ* state (1Lb) is observed in going from the liquid to the solid phase due to intermolecular hydrogen bonding.17 In picene, upon crystallization, the lowest-energy electronic transition (S 0 → S 1 ) undergoes a relative intensification by 2 orders of magnitude with respect to the higher-energy transitions.25 A possible explanation is the superposition or energy inversion of the 1La and 1Lb states, due to a strong intermolecular coupling, so that the observed lowest-energy transition in the crystal involves an excited state with 1La character. Indeed, in naphthalene the molecular S1 state is Lb, but in the crystal a mixing of components having the same symmetry and deriving from the molecular La and Lb states was assumed to explain the large Davydov splitting and pressure dependence of the transition frequency.37 In naphthalene crystal, the A and B components of the S0 → S1 transition undergo a red shift of 190 and 110 cm−1/GPa respectively, in the pressure range 0−4.5 GPa. Due to the different character of 1La and 1Lb that is responsible for their different solvatochromic shift and energy dependence upon molecular structure, we can expect that the crystal states deriving from 1La or 1Lb may be differently affected by pressure. As a first approximation, we can assume that in a molecular crystal the electronic transition energies are affected by a change in density inasmuch as the involved electronic states are characterized by different electronic ■ 5349 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351 The Journal of Physical Chemistry C Article A further interesting observation is the perfect superposition, within the accuracy of the measurement, of the exciton energy and the 0−0 fluorescence band in the whole pressure range examined. In general, the pressure shifts observed in fluorescence spectra can be more pronounced than those observed in absorption spectra or excitation profiles (see, for example, refs 18 and 19) due to the structural relaxation of the excited molecule and its surrounding following the vertical excitation. This is particularly relevant when the relaxed geometry of the excited molecule substantially differs from that of the ground state, so that the surrounding is locally perturbed and the effect is enhanced by the density (it can be said that the ground and excited states have different shapes as a function of density, or compressibilities). In the case of picene, the frequencies of the vertical excitation and of the relaxed fluorescence have the same pressure dependence, likely indicating that the ground- and excited-state equilibrium geometries are similar so that excitation does not distort the crystal structure, and are similarly affected by the pressure. In addition, a similarity between the equilibrium geometries of S0 and S1 is also suggested by the fact that reactivity in picene seems to be not much affected by the laser irradiation. In fact, in most molecular systems the high-pressure reactivity is enhanced by photochemical effects. In some cases, the activation is due to a photodissociation42−44 or to the achievement of a different, more reactive, molecular geometry45−47 that is not usually obtained in aromatic systems, where the excitation is delocalized on the π electron system. In the case of benzene,19 the reaction threshold pressure is strongly lowered by HOMO−LUMO excitations through the formation of excimers, that become reaction initiators. The crystal structure of picene prevents excimer formation, because molecules in the required geometry (a sandwich or slippedparallel relative position) are too far from each other along the a- or b-axes, while inequivalent molecules inside a unit cell are in T configuration.6 Accordingly, we do not observe excimer emission in the fluorescence spectra even at the highest pressures reached. Extrapolating the linear fit of the S0 → S1 transition energy as a function of pressure, band gap closure in solid picene would occur at 64 GPa, a much higher pressure than the reaction threshold pressure. Comparison of the IR and electronic absorption data suggests that the reaction detected above 23 GPa in IR spectra likely starts at about 11 GPa through the formation of reactive nuclei at crystal defects. For this mild compression, there is no bulk reactivity and the reaction does not propagate during the pressure download, as evidenced thanks to the high spatial resolution in the measurement of the visible absorption spectra. The species formed at this stage are reaction intermediates from which a further transformation takes place toward a product or back to picene. The red coloration of the polycrystalline samples reproducibly starts at 9 GPa, but was found to be reversible in the samples that were not compressed above 15 GPa. The applied pressure can therefore be the tool to trigger the formation of activated complexes, at least in the presence of crystal defects, which revert back to the pristine molecules if the density does not overcome a specific threshold. Also, the product obtained by this reactivity at defects may be different from that obtained for compressions above 23 GPa. The threshold pressure of 23.5 GPa detected for the reaction by IR spectra is quite reproducible in different samples, and after reaching this pressure value the reaction mostly propagates upon decom- pression. The existence of a reproducible threshold pressure could indicate a reactivity occurring in the bulk crystal and driven by the intermolecular interactions in the crystal structure. The FTIR spectra show that the recovered product is a disordered branched material, similar to the amorphous hydrogenated carbon obtained from the high-pressure reaction of benzene but containing a larger amount of unsaturated carbon atoms, due to the lower hydrogen content of the starting system. CONCLUSIONS The pressure evolution of the four lowest-energy singlet− singlet transition up to 14 GPa, and the pressure-induced reactivity of solid picene have been characterized in this work, giving an extensive insight into the optical properties and chemical stability of picene under high-pressure conditions. We find that the S0 → S1 transition frequency decreases linearly with pressure, with a slope similar to that found in aromatic systems where S1 is of 1La type, indicating that S1 excited molecules are more stabilized by the crystalline environment than ground state molecules, due to their larger polarizability or possibly a nonzero dipole moment. By contrast, the higherenergy transitions do not undergo a frequency shift with pressure. This may be an appealing property to be considered in designing materials whose electronic properties can be tuned by pressure. At room temperature the system is found to be chemically stable up to 11 GPa. Above this pressure a limited reactivity occurs at crystal defects while above 23.5 GPa an amorphization reaction involves the bulk crystal. Reactivity also prevents the achievement of the pressure conditions where the band gap possibly closes. Low-temperature studies on hydrostatically compressed single crystals could elucidate the nature of the reactivity and the role of defects. In conclusion, there are attractive possibilities of pressure-tuning the electronic properties of picene that from a chemical point of view has a relatively high stability with respect to other aromatic hydrocarbons. ■ ■ AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]fi.it. Notes The authors declare no competing financial interest. ACKNOWLEDGMENTS Supported by the European Union under the Contract LASERLAB EUROPE 228334 and by the Italian Ministero dell’Istruzione, dell’Università e della Ricerca (MIUR) under the Contract FIRBFuturo in Ricerca 2010 no. RBFR109ZHQ. ■ ■ REFERENCES (1) Xue, M.; Cao, T.; Wang, D.; Wu, Y.; Yang, H.; Dong, X.; He, J.; Li, F.; Chen, G. F. Superconductivity above 30 K in Alkali-MetalDoped Hydrocarbon. Sci. Rep. 2012, 2 (389), 1−4. (2) Wang, X. F.; Liu, R. H.; Gui, Z.; Xie, Y. L.; Yan, Y. J.; Ying, J. J.; Luo, X. G.; Chen, X. H. Superconductivity at 5 K in Alkali-MetalDoped Phenanthrene. Nat. Commun. 2011, 2 (507), 1−7. (3) Kubozono, Y.; Mitamura, H.; Lee, X.; He, X.; Yamanari, Y.; Takahashi, Y.; Suzuki, Y.; Kaji, Y.; Eguchi, R.; Akaike, K.; et al. H. Metal-Intercalated Aromatic Hydrocarbons: a New Class of CarbonBased Superconductors. Phys. Chem. Chem. Phys. 2011, 13, 16476− 16493. 5350 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351 The Journal of Physical Chemistry C Article (4) Mitsuhashi, R.; Suzuki, Y.; Yamanari, Y.; Mitamura, H.; Kambe, T.; Ikeda, N.; Okamoto, H.; Fujiwara, A.; Yamaji, M.; Kawasaki, N.; et al. Superconductivity in Alkali-Metal-Doped Picene. Nature 2010, 464, 76−79. (5) Portella, G.; Poater, J.; Bofill, J. M.; Alemany, P.; Solà, M. Local Aromaticity of [n]Acenes, [n]Phenacenes, and [n]Helicenes (n = 1− 9). J. Org. Chem. 2005, 70, 2509−2521. (6) De, A.; Ghosh, R.; Roychowdhury, S.; Roychowdhury, P. Structural Analysis of Picene, C22H14. Acta Crystallogr. C 1985, 41, 907−909. (7) Kosugi, T.; Miyake, T.; Ishibashi, S.; Arita, R.; Aoki, H. FirstPrinciples Electronic Structure of Solid Picene. J. Phys. Soc. Jpn. 2009, 78 (113704), 1−4. (8) Kambe, T.; He, X.; Takahashi, Y.; Yamanari, Y.; Teranishi, K.; Mitamura, H.; Shibasaki, S.; Tomita, K.; Eguchi, R.; Goto, H.; et al. Investigation on Synthesis and Physical Properties of Metal Doped Picene Solids. arXiv:1210.1632 [cond-mat.supr-con] 2012. (9) Oehzelt, M.; Aichholzer, A.; Resel, R.; Heimel, G.; Venuti, E.; Della Valle, R. G. Crystal Structure of Oligoacenes under High Pressure. Phys. Rev. B 2006, 74 (104103), 1−7. (10) Drickamer, H. G. Forty Years of Pressure Tuning Spectroscopy. Annu. Rev. Mater. Sci. 1990, 20, 1−18. (11) Loubeyre, P.; Occelli, F.; LeToullec, R. Optical Sstudies of Solid Hydrogen to 320 GPa and Evidence for Black Hydrogen. Nature 2002, 416, 613−617. (12) Goncharov, A. F.; Gregoryanz, E.; Mao, H. K.; Liu, Z.; Hemley, R. J. Optical Evidence for a Nonmolecular Phase of Nitrogen above 150 GPa. Phys. Rev. Lett. 2000, 85, 1262−1265. (13) Luo, H.; Desgreniers, S.; Vohra, Y. K.; Ruoff, A. L. HighPressure Optical Studies on Sulfur to 121 GPa: Optical Evidence for Metallization. Phys. Rev. Lett. 1991, 67, 2998−3001. (14) Goettel, K. A.; Eggert, J. H.; Silvera, I. F.; Moss, W. C. Optical Evidence for the Metallization of Xenon at 132(5) GPa. Phys. Rev. Lett. 1989, 62, 665−668. (15) Drickamer, H.; Frank, C. Electronic Transitions and the High Pressure Chemistry and Physics of Solids; Studies in Chemical Physics; Chapman and Hall: New York, 1973. (16) Schettino, V.; Bini, R.; Ceppatelli, M.; Ciabini, L.; Citroni, M. Chemical Reactions at Very High Pressure. Adv. Chem. Phys. 2005, 131, 105−242. (17) Fanetti, S.; Citroni, M.; Bini, R. Pressure-Induced Fluorescence of Pyridine. J. Phys. Chem. B 2011, 115, 12051−12058. (18) Citroni, M.; Costantini, B.; Bini, R.; Schettino, V. Crystalline Indole at High Pressure: Chemical Stability, Electronic, and Vibrational Properties. J. Phys. Chem. B 2009, 113, 13526−13535. (19) Citroni, M.; Bini, R.; Foggi, P.; Schettino, V. Role of Excited Electronic States in the High-Pressure Amorphization of Benzene. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 7658−7663. (20) Fanetti, S.; Ceppatelli, M.; Citroni, M.; Bini, R. Changing the Dissociative Character of the Lowest Excited State of Ethanol by Pressure. J. Phys. Chem. B 2011, 115, 15236−15240. (21) Kim, J. S.; Boeri, L.; Kremer, R. K.; Razavi, F. S. Effect of Pressure on Superconducting Ca-Intercalated Graphite CaC6. Phys. Rev. B 2006, 74 (214513), 1−6. (22) Takabayashi, Y.; Ganin, A. Y.; Jeglic, P.; Arcon, D.; Takano, T.; Iwasa, Y.; Ohishi, Y.; Takata, M.; Takeshita, N.; Prassides, K.; et al. The Disorder-Free Non-BCS Superconductor Cs3C60 Emerges from an Antiferromagnetic Insulator Parent State. Science 2009, 323, 1585− 1590. (23) Sato, T.; Iwahara, N.; Tanaka, K. Critical Reinvestigation of Vibronic Couplings in Picene from View of Vibronic Coupling Density Analysis. Phys. Rev. B 2012, 85, 2−6. (24) Xin, Q.; Duhm, S.; Bussolotti, F.; Akaike, K.; Kubozono, Y.; Aoki, H.; Kosugi, T.; Kera, S.; Ueno, N. Accessing Surface Brillouin Zone and Band Structure of Picene Single Crystals. Phys. Rev. Lett. 2012, 108, 1−5. (25) Fanetti, S.; Citroni, M.; Bini, R.; Malavasi, L.; Artioli, G. A.; Postorino, P. HOMO-LUMO Transitions in Solvated and Crystalline Picene. J. Chem. Phys. 2012, 137 (224506), 1−7. (26) Joseph, B.; Boeri, L.; Malavasi, L.; Capitani, F.; Artioli, G. A.; Protti, S.; Fagnoni, M.; Albini, A.; Marini, C.; Baldassarre, L.; et al. Vibrational spectrum of solid picene C22H14. J. Phys.: Condens. Matter 2012, 24 (252203), 1−6. (27) Bini, R.; Ballerini, R.; Pratesi, G.; Jodl, H. Experimental Setup for Fourier Transform Infrared Spectroscopy Studies in Condensed Matter at High Pressure and Low Temperatures. Rev. Sci. Instrum. 1997, 68, 3154−3160. (28) Ceppatelli, M.; Serdyukov, A.; Bini, R.; Jodl, H. J. Pressure Induced Reactivity of Solid CO by FTIR Studies. J. Phys. Chem. B 2009, 113, 6652−6660. (29) Ciabini, L.; Santoro, M.; Bini, R.; Schettino, V. High Pressure Photoinduced Ring Opening of Benzene. Phys. Rev. Lett. 2002, 88 (085505), 1−4. (30) Girlando, A.; Masino, M.; Bilotti, I.; Brillante, A.; Della Valle, R. G.; Venuti, E. Phonon Dynamics and Electron-Phonon Coupling in Pristine Picene. Phys. Chem. Chem. Phys. 2012, 14, 1694−1699. (31) Ciabini, L.; Santoro, M.; Bini, R.; Schettino, V. High Pressure Reactivity of Solid Benzene Probed by Infrared Spectroscopy. J. Chem. Phys. 2002, 116, 2928−2935. (32) Ciabini, L.; Santoro, M.; Gorelli, F. A.; Bini, R.; Schettino, V.; Raugei, S. Triggering Dynamics of the High-Pressure Benzene Amorphization. Nat. Mater. 2007, 6, 39−43. (33) Platt, J. R. Classification of Spectra of Cata-Condensed Hydrocarbons. J. Chem. Phys. 1949, 17, 484−495. (34) Klessinger, M.; Michl, J. Excited States and Photochemistry of Organic Molecules; VCH Publishers, Inc.: New York, 1995; Chapter 2. (35) Richard, R. M.; Herbert, J. M. Time-Dependent DensityFunctional Description of the L-1(a) State in Polycyclic Aromatic Hydrocarbons: Charge-Transfer Character in Disguise? J. Chem. Theory Comput. 2011, 7, 1296−1306. (36) Gallivan, J. B. Polarization of Electronic Transitions of Aromatic Hydrocarbons. J. Chem. Phys. 1969, 50, 1590−1595. (37) Meletov, K. High-Pressure Study of Intermolecular Resonance Interaction in a Naphthalene Crystal. Chem. Phys. 1991, 154, 469− 475. (38) Aust, R. B.; Bentley, W. H.; Drickamer, H. G. Behavior of Fused-Ring Aromatic Hydrocarbons at Very High Pressure. J. Chem. Phys. 1964, 41, 1856−1864. (39) Farina, L.; Syassen, K.; Brillante, A.; Della Valle, R.; Venuti, E.; Karl, N. Pentacene at High Pressure. High Pressure Res. 2003, 23, 349− 354. (40) Dreger, Z.; Lucas, H.; Gupta, Y. High-Pressure Effects on Fluorescence of Anthracene Crystals. J. Phys. Chem. B 2003, 107, 9268−9274. (41) Roth, F.; Mahns, B.; Büchner, B.; Knupfer, M. Exciton Character in Picene Molecular Solids. Phys. Rev. B 2011, 83, 83−86. (42) Fanetti, S.; Ceppatelli, M.; Citroni, M.; Bini, R. High-Pressure Photoinduced Reactivity of CH3OH and CD3OH. J. Phys. Chem. C 2012, 116, 2108−2115. (43) Ceppatelli, M.; Fanetti, S.; Citroni, M.; Bini, R. Photoinduced Reactivity of Liquid Ethanol at High Pressure. J. Phys. Chem. B 2010, 114, 15437−15444. (44) Ceppatelli, M.; Bini, R.; Schettino, V. High-Pressure Photodissociation of Water as a Tool for Hydrogen Synthesis and Fundamental Chemistry. Proc. Natl. Acad. Sci.U.S.A. 2009, 106, 11454−11459. (45) Citroni, M.; Bini, R.; Pagliai, M.; Cardini, G.; Schettino, V. Nitromethane Decomposition under High Static Pressure. J. Phys. Chem. B 2010, 114, 9420−9428. (46) Chelazzi, D.; Ceppatelli, M.; Santoro, M.; Bini, R.; Schettino, V. High-Pressure Synthesis of Crystalline Polyethylene Using Optical Catalysis. Nat. Mater. 2004, 3, 470−475. (47) Citroni, M.; Ceppatelli, M.; Bini, R.; Schettino, V. Laser-Induced Selectivity for Dimerization versus Polymerization of Butadiene under Pressure. Science 2002, 295, 2058−2060. 5351 dx.doi.org/10.1021/jp4006789 | J. Phys. Chem. C 2013, 117, 5343−5351