Effects of Thermal Fluctuations on the Structure, Level Alignment, and

Article pubs.acs.org/JPCC

Effects of Thermal Fluctuations on the Structure, Level Alignment, and Absorption Spectrum of Dye-Sensitized TiO2: A Comparative Study of Catechol and Isonicotinic Acid on the Anatase (101) and Rutile (110) Surfaces He Lin,*,† Guido Fratesi,‡ Sencer Selçuk,§ Gian Paolo Brivio,† and Annabella Selloni§ †

Dipartimento di Scienza dei Materiali, Università di Milano-Bicocca, Via Cozzi 55, I-20125 Milano, Italy Dipartimento di Fisica, Università degli Studi di Milano, Via Celoria 16, I-20133 Milano, Italy § Department of Chemistry, Princeton University, Princeton, New Jersey 08544, United States ‡

S Supporting Information *

ABSTRACT: The adsorption of catechol and isonicotinic acid on the TiO2 anatase (101) and rutile (110) surfaces has been studied by means of first-principles molecular dynamics simulations and time-dependent density functional calculations. Our results show that thermal fluctuations induce changes in the position of the molecular levels around the TiO2 valence band edge. For the anatase (101) surface, the alignment of the molecular levels with the TiO2 valence band edge has a significant effect on the absorption spectrum. For rutile (110), instead, the adsorption of catechol and isonicotinic acid induces only a minor sensitization. The sensitization of anatase (101) by catechol and isonicotinic acid can be enhanced by increasing the hybridization between the adsorbed dye and TiO2 states.

1. INTRODUCTION Dye-sensitized solar cells (DSSCs) continue to attract tremendous interest due to their ability to convert sunlight into electrical energy at low cost and with high efficiency.1−3 DSSCs are constituted by organic or transition-metal-based organometallic dye molecules adsorbed on a semiconductor, generally a metal oxide such as TiO2.4 In particular, catechol adsorbed on TiO2 has received considerable attention in recent years as a prototypical model for DSSCs.5−14 Catechol has two acidic hydroxy functionalities which are used to anchor it on a TiO2 surface. The highest occupied molecular orbital (HOMO) energy level of anchored catechol lies in the band gap of TiO2,5−8 hence reducing the optical gap and allowing for an efficient dye−substrate electron injection. Another prototypical dye is isonicotinic acid. Similarly to catechol, the HOMO energy level of adsorbed isonicotinic acid is also within the TiO2 gap.15−17 However, isonicotinic acid is an electron-poor aromatic molecule due to both the substitution of a carbon with the more electronegative nitrogen atom in the aromatic ring and the presence of an electron withdrawing carboxylic substituent, whereas catechol is an electron-rich aromatic molecule thanks to the electron donating capabilities of hydroxyl groups. Because of these different electronic properties, the energy level of the lowest unoccupied molecular orbital (LUMO) of the isolated isonicotinic acid is 1.0 eV lower than that of free catechol when the HOMO is taken as a reference (see Figure S1, Supporting Information). © 2016 American Chemical Society

Motivated by these properties, several theoretical studies have utilized density functional theory (DFT) and timedependent DFT (TDDFT)18 methods to examine the adsorption of catechol5−8,19−23 and isonicotinic acid15−17 on TiO2 surfaces. Risplendi et al. 7 recently performed a comparative investigation of catechol and isonicotinic acid on TiO2 rutile (110) using hybrid B3LYP functional calculations. They analyzed geometries, adsorption energies, and electronic structures but did not investigate the optical spectra. Experimentally, catechol-sensitized TiO2 nanoparticles24 show an absorption band at ∼430 nm (∼2.88 eV) that is attributed to the excitation from the catechol HOMO orbital in the TiO2 band gap to the conduction band of the nanoparticle.25 Since TiO2 nanoparticles generally prefer the metastable anatase phase rather than the thermodynamically stable rutile form,26−30 it is desirable to model the absorption spectrum not only on rutile but also on anatase TiO2, especially on the most frequently exposed anatase (101) surface.31 In another study, Sánchez-de-Armas et al. analyzed the electronic structures of several dyes including catechol, either free or adsorbed on TiO2, and provided optical absorption spectra calculated using TDDFT.23 After verifying that the calculated spectra for the free dyes agree well with experiment,32 they Received: December 4, 2015 Revised: February 1, 2016 Published: February 3, 2016 3899

DOI: 10.1021/acs.jpcc.5b11885 J. Phys. Chem. C 2016, 120, 3899−3905

Article

The Journal of Physical Chemistry C discussed different electron injection mechanisms for the various sensitizers. However, they used a small (TiO2)9 cluster to simulate the absorption spectra of dye-sensitized TiO2.23 To provide a more realistic description of the electronic structure and of the optical absorption spectrum, an extended model such as a surface slab is desirable. Furthermore, since experimental absorption spectra are generally recorded at room temperature,33 it is convenient to take into account the effects of thermal fluctuations, usually not included in theoretical studies.7,23 In this paper, we compare the finite temperature optical spectra of catechol and isonicotinic acid on the TiO2 anatase (101) and rutile (110) surfaces. To describe finite temperature effects, we perform first-principles molecular dynamics (MD) simulations, select several snapshots along the MD trajectory, and calculate the optical absorption spectrum of each snapshot using TDDFT. The main questions we want to address with our study are the following: (1) How does the atomic motion affect the electronic structure and the optical absorption spectrum of dye-sensitized TiO2? (2) What is the relationship between the electronic structure and the absorption spectrum? (3) What is the difference in the electronic structure and the absorption spectrum between TiO2 sensitized with catechol and isonicotinic acid? (4) What is the difference between dyesensitized anatase (101) and rutile (110)?

Figure 1. Optimized geometries of (a) catechol/anatase TiO2(101), (b) isonicotinic acid/anatase TiO2 (101), (c) catechol/rutile TiO2(110), and (d) isonicotinic acid/rutile TiO2(110). Relevant bond distances are indicated (Å). O atoms are red, Ti atoms are light gray, C atoms are yellow, and H atoms are blue. This notation is used throughout this paper.

Two protons resulting from the dissociation are transferred to two neighboring surface O2c sites on the left and right, and the molecular benzene-like ring is slightly tilted toward the surface. This configuration agrees well with previous experimental and theoretical reports.24,33,42,43,19 To check the stability of our system, we computed the dye adsorption energy (Eads) as follows

2. COMPUTATIONAL METHODS All calculations were performed within DFT using the Perdew− Burke−Ernzerholf (PBE) functional34 and the plane wave pseudopotential scheme as implemented in the QuantumESPRESSO package.35 Electron−ion interactions were described by ultrasoft pseudopotentials, with electrons from C and O 2s, 2p and Ti 3s, 3p, 3d and 4s shells explicitly included in the calculations. Plane-wave basis set cutoffs for the smooth part of the wave functions and the augmented density were 30 and 300 Ry, respectively. K-point sampling was restricted to the Γ point. Optical absorption spectra were computed by TDDFT using the Liouville−Lanczos approach.36,37 First-principles molecular dynamics (MD) simulations were performed within the Car−Parrinello (CP) approach.38 In MD simulations, a time step of 0.194 fs and a fictitious electronic mass of 400 atomic units (amu) were used. The ionic temperatures were controlled by means of a Nosé thermostat.39 To model the surface, we used the periodically repeated slab geometry.40,41 The anatase (101) surface was modeled with three layers of oxide (∼9.5 Å thick). For the 1 × 1 surface, a p(3 × 1) surface supercell was used, with 108 atoms and a surface area of 11.34 × 10.31 Å2. For the rutile (110) surface, we adopted a four-layer slab (∼12.5 Å thick) and a p(3 × 2) surface cell with 144 atoms and a surface area of 8.86 × 13.03 Å2. For both surfaces, the vacuum region between facing slabs was 10 Å wide. To check the adequacy of such a choice, the adsorption of dissociated catechol was calculated using vacuum sizes of 10 and 20 Å. A difference in adsorption energy of only 0.01 eV was found between the two cases.

Eads = E(dye/TiO2 ) − E(TiO2 ) − E(dye)

(1)

where E(dye/TiO2), E(TiO2), and E(dye) are the total energies of the chemisorbed system, the clean TiO2 slab, and the isolated molecular dye, respectively. The adsorption energy for catechol/anatase is Eads = −0.72 eV, which is in good agreement with previous DFT calculations (−0.70 eV).19 As shown in Figure 2, during the MD simulation at 300 K, the bond distances between the two oxygen (O*) atoms of the

Figure 2. Time evolution of the bond distances between the O* atoms of the adsorbed catechol and the surface Ti5c atoms of anatase (101) during a MD simulation at 300 K. Selected snapshots during the simulation are also shown.

3. RESULTS AND DISCUSSION A. Catechol on the Anatase (101) Surface. Our adsorption geometry studies of catechol on anatase (101) show that the most stable structure is a completely dissociated molecule in a bidentate configuration (Figure 1a), bonded to two surface Ti5c atoms via its two deprotonated OH groups.

adsorbed catechol and the surface Ti5c atoms oscillate around ∼1.85 Å, the average of the two Ti5c−O* bond lengths, which are 1.87 and 1.83 Å in the optimized structure (Figure 1a). In addition, the adsorbed catechol swings back and forth around an axis passing along the two O* atoms, thus showing alternate molecular orientations during the simulation. 3900


Article

The Journal of Physical Chemistry C We analyze the electronic states at the catechol/anatase interface, focusing on the states with energies close to the TiO2 valence and conduction band edges (VBE and CBE, respectively). The projected DOS (PDOS) of the optimized structure and selected snapshots along the MD trajectory are plotted in Figure 3; the snapshots were chosen every 2 ps after

Figure 4. Absorption spectrum for (a1, a2) the ground state and (b−f) selected snapshots of catechol/anatase along a 300 K MD simulation. The dotted blue (solid red) curve is the absorption spectra for clean anatase TiO2 (catechol/anatase), and the dashed black curve in (a1) is the absorption spectrum of the gas phase catechol molecule.

Figure 3. Projected density of states (PDOS) for (a) the ground state structure and (b−f) selected snapshots of catechol/anatase during a 300 K MD simulation. The solid black curve (red shaded area) is the PDOS for TiO2 (catechol). For better clarity, the PDOS of catechol has been multiplied by 3.

catechol/anatase in the optimized structure is red-shifted compared to clean TiO2 (see Figure 4a1,a2), which is in agreement with experiment,24,33 but with a much smaller shift than the experimental one.24,33 For comparison, the computed absorption spectrum of free catechol is also reported in Figure 4a1; this shows a peak at ∼4.5 eV, in good agreement with experiment32 as well as with previous calculations.23 From Figure 4b−f, we can also see that the catechol-induced shift of the absorption threshold changes from one snapshot to the other, indicating that thermal fluctuations have a significant effect on the optical properties. Note that for the ground state (Figure 3a) and the snapshot at 10 ps (Figure 3f) the HOMO− 2 lies in the band gap of TiO2 and has predominant catechol character. In these cases, the hybridization between catechol and TiO2 is relatively weak, and a very small shift in the absorption edge position is obtained (Figure 4a,f). By contrast, for the snapshot at 2 ps (Figure 3b), 4 ps (Figure 3c), 6 ps (Figure 3d), and 8 ps (Figure 3e), the HOMO−2 is below the TiO2 VBE. When the snapshot displays a HOMO−2 state that lies very close to the VBE of TiO2 (e.g., at 4, 6, and 8 ps), the hybridization between catechol and TiO2 is strong, and correspondingly we obtain a more pronounced red-shift in the absorption edge position. On the other hand, if the HOMO−2 is well below the TiO2 VBE, as for the snapshot at 2 ps, the energy of the electronic transition to the conduction band is larger, and correspondingly the red-shift of the absorption is smaller. B. Isonicotinic Acid on the Anatase (101) Surface. Unlike catechol, which prefers to adsorb in dissociated form, for

the initial thermalization, a time chosen so as to separate them enough to represent uncorrelated configurations of the dye/ TiO2 system. For each structure, we show the PDOS for the TiO2 slab and the adsorbed catechol. In the following we use the notation HOMO, HOMO−1, HOMO−2, etc., to indicate the states of the adsorbed dye. In the optimized (T = 0 K) structure (Figure 3a), the HOMO, HOMO−1, and HOMO−2 energies lie in the band gap of TiO2, and the ground state wave functions of HOMO and HOMO−1 are plotted in Figure S2. As shown in Figure 3b−f, the energy levels of the HOMO and HOMO−1 fluctuate between the TiO2 VBE and CBE during the MD simulation, whereas the energy level of the HOMO−2 is sometimes above and other times below the TiO2 VBE. Thus, a significant effect of thermal fluctuations is to change the position of the HOMO−2 level around the TiO2 VBE. In addition to the gap states, which have clear catechol character, it is also interesting to study the distribution of the other states contributed by catechol, in particular the empty ones. The unoccupied electronic structure of catechol/anatase shows TiO2-related states at lower energy, i.e., close to the CBE, and states that are closely related to free catechol at higher energy. The two sets of unoccupied states occur at about the same energy for the various snapshots, suggesting that thermal fluctuation have less effect on the unoccupied electronic structure of catechol/anatase than on the occupied one. The calculated optical absorption spectra of catechol/TiO2 are shown in Figure 4, where they are compared to those of the clean surface. We can see that the onset of absorption of 3901


Article

The Journal of Physical Chemistry C

compared to the case of catechol (see Figure 3), where the position of the LUMO-induced features are pinned at an energy of ∼3 eV above the CBE, suggesting that thermal fluctuations have a more pronounced effect on the unoccupied electronic structure of isonicotinic acid/anatase than on that of catechol/anatase. The calculated absorption spectrum of isonicotinic acid/ TiO2 is also shown in Figure 5. Similar to catechol, the absorption threshold of isonicotinic acid/TiO2 is red-shifted compared to that of the clean TiO2 surface, and the red-shift changes during the MD simulation. Since for the ground state (Figure 5a1), the HOMO−1 lies in the gap of TiO2 and has predominant isonicotinic acid character, the hybridization between the adsorbate and TiO2 is relatively weak, and the shift in the absorption edge position is very small (Figure 5a2). However, for the snapshot at 6 ps, as the HOMO−1 already overlaps with the TiO2 valence band (Figure 5b1), the hybridization between isonicotinic acid and TiO2 is enhanced, leading to a more significant red-shift of the absorption spectrum (Figure 5b2). However, when the HOMO−1 is deeper below the TiO2 VBE (Figure 5c1), the energy for the optical electronic transitions increases; hence a less effective sensitization results (Figure 5c2). C. Rutile (110) and Comparison to Anatase (101). On the rutile (110) surface, dissociative adsorption is highly favored for both catechol and isonicotinic acid, resulting in the bidentate configurations shown in Figure 1c,d. As on anatase (101), the dissociated catechol (isonicotinic acid) is attached to rutile (110) through the bond between the oxygen (O*) of its hydroxyl groups (carbonyl group) and two Ti5c atoms, but unlike on anatase (101) in this case the molecule is almost perpendicular to the surface. The Ti5c−O* bond lengths are similar to those of catechol (isonicotinic acid)/anatase. However, among the four geometries in Figure 1, we obtain the smallest magnitude of the adsorption energy (−0.45 eV) for catechol/rutile and the largest one (−0.82 eV) for isonicotinic acid/rutile. These configurations are stable during MD simulations (300 K). Along the MD trajectory, the adsorbed dye also sways with respect to an axis along the two O* atoms and shows alternate molecular orientations. The calculated PDOS and absorption spectra for the ground state structures of catechol and isonicotinic acid on rutile are shown in Figures 6 and 7, respectively, while analogous results for various snapshots along the corresponding MD simulations are shown in Figures S3−S6. In comparison to anatase (101) (Figures 6b,d and 7b,d), a shoulder in the PDOS appears near the TiO2 VBE (see Figure 6a,c), and the absorption threshold occurs at lower energy than on anatase (see Figure 7a,c). For catechol/rutile, the onset of the absorption shows a very small red-shift compared to that of the clean rutile surface (Figure 7a), while for isonicotinic acid/rutile the absorption is even slightly blue-shifted compared to clean rutile TiO2 (Figure 7c), which is opposite to what found for isonicotinic acid/anatase (Figure 7d). To make closer contact with experiments, in Figure 7a−d we also show the dye/TiO2 absorption spectra obtained from the average over the different snapshots of each simulation. It appears that thermal fluctuations tend to enhance the absorption in the low-energy tail with a more pronounced effect on anatase than on rutile (see the insets of Figure 7a−d). To summarize our results and compare the alignments of the molecular levels with the TiO2 band edges, we report in Table 1 the energy differences between HOMO and HOMO−1 energies in the gap and the VBE of the clean TiO2 surface.

isonicotinic acid/anatase the nondissociative monodentate configuration is energetically more favorable than the dissociative bidendate one (Figure 1b). However, despite the more favorable adsorption energy, the molecule in the monodentate configuration desorbed from the surface during the MD simulation. Therefore, we only discuss the bidentate configuration in the following. In this configuration, the dissociated isonicotinic acid is attached to the surface through the oxygens (O*) of the carbonyl group which are bound to two surface T5c atoms. The bond lengths between the surface Ti5c atoms and O* are 2.10 Å (see Figure 1b), thus longer than those of catechol/anatase (1.83 and 1.87 Å), while the adsorption energy is −0.60 eV, whose magnitude is 0.12 eV smaller than that of catechol/anatase. Similarly to catechol, the adsorbed isonicotinic acid oscillates back and forth around an axis along the two O* atoms during MD simulation. The PDOS of selected snapshots along the MD trajectory at T = 300 K are shown in Figure 5. Differently from catechol, in

Figure 5. PDOS (left) and absorption spectrum (right) for (a) the ground state and (b, c) selected snapshots during a MD simulation of bidentate isonicotinic acid on anatase (101). In the left panels, the solid black curves and orange shaded areas are the PDOS of TiO2 and isonicotinic acid, respectively. For better clarity, the PDOS of isonicotinic acid has been multiplied by 3. In the right panels, the dotted blue and solid red curves are the absorption spectrum of clean anatase TiO2 and isonicotinic acid/anatase, respectively.

the optimized structure of isonicotinic acid/anatase, only the HOMO and HOMO−1 energies lie in the gap of anatase TiO2, and the main effect of thermal fluctuations is to change the position of the HOMO−1 level around the TiO2 VBE. The ground state wave function of HOMO and HOMO−1 are plotted in Figure S2. In addition, the empty states look quite different from those of catechol. States of adsorbed isonicotinic acid coexist with TiO2-related states at lower energy relative to the CBE, and the LUMO shows a stronger coupling with substrate states, forming a wide structure in the DOS in the energy range from 2 to 3.5 eV in Figure 5a1. As shown in Figure 5b1 and especially in Figure 5c1, during the MD simulation the DOS contribution by the LUMO of adsorbed isonicotinic acid is transferred closer to the CBE. This can be 3902


Article


Note that for both molecules the molecular states of dye/rutile are closer to the VBE than those of dye/anatase. This is consistent with the relative alignments of the rutile and anatase band edges proposed by Scanlon et al.44 On the other hand, comparing different dyes, the HOMO and HOMO−1 states of adsorbed isonicotinic acid are closer to the VBE than those of catechol. This is possibly related to the pinning of catechol’s HOMO−2 to the VBE, which is observed on both the anatase (101) and rutile (110) surfaces. We further remark that there is no clear correlation between these energy level alignments and the adsorption energies in Table 1.

4. CONCLUSIONS In this work, we have presented a comprehensive periodic DFT-TDDFT study of prototypical dye/oxide semiconductor models of DSSCs. To make closer contact with experiments and applications, which are most often performed at room temperature, we accounted for thermal fluctuations on the adsorbate structures by first-principles MD and studied their effects on the absorption spectra by calculating the spectra of several uncorrelated snapshots along the MD trajectories and determining their averages. Considering the significant amount of research reported in recent years, it is interesting to compare our results to those of previous studies.7,8,23 Risplendi et al.7 examined different adsorption modes of catechol and isonicotinic acid onto the rutile TiO2 (110) surface within a hybrid DFT framework. For both molecules, they found that a dissociated bidentate configuration is favored, that in the most stable configuration, the HOMO of both molecules lies in the energy gap of TiO2, and the HOMO of adsorbed catechol is a delocalized π orbital, while the HOMO of isonicotinic acid corresponds to a σ orbital. All these findings are in agreement with our results. Sánchez-de-Armas et al. reported a DFTTDDFT study of the electronic structure and the optical response of various dyes, free and bound to a (TiO2)9 cluster.23 The calculated spectrum of free catechol in ref 23 agrees well with ours. For catechol adsorbed on the (TiO2)9 cluster, however, Sánchez-de-Armas et al. obtained a wide absorption band at lower energy, which is not present in the experimental spectrum. Unfortunately, these authors did not report the spectrum of the bare (TiO2)9 cluster, so that it is not possible to infer the effect of the sensitization due to the dye adsorption. A limitation of our approach is the use of PBE Kohn−Sham (KS) eigenvalues to evaluate the alignment of the molecular levels relative to the TiO2 band edges. The use of PBE may also explain why the effect of TiO2 sensitization (notably the catechol induced red-shift of the anatase absorption) appears to be underestimated with respect to the experiment.24,33 However, the limitations of the Kohn−Sham description of the DOS do not alter the conclusion that the sensitization is improved by enhancing the hybridization between the dye and TiO2 states, as observed during our molecular dynamics simulations. In our case, the alignment of the hybridizing orbital of catechol (HOMO−2) with the TiO2 VBE depends on the geometry, and the highest orbitals (HOMO−1 and HOMO) experience the same dependence. In this respect, we note that a recent G0W045,46 calculation for catechol on rutile (110)8 shows that the HOMO−1 energy of the adsorbed catechol overlaps the valence band, and thus its energy is lower than in our results (the HOMO is still found to be at higher energy than the VBE in ref 8). Our analysis of the molecular dynamics induced hybridization and its implication for the

Figure 6. PDOS of (a) catechol/rutile, (b) catechol/anatase, (c) isonicotinic acid/rutile, and (d) isonicotinic acid/anatase. The solid black curves are the PDOS of TiO2, and the red (orange) shaded areas are the PDOS for catechol (isonicotinic acid). The PDOS of catechol and isonicotinic acid have been multiplied by 3.

Figure 7. Absorption spectra of (a) catechol/rutile, (b) catechol/ anatase, (c) isonicotinic acid/rutile, and (d) isonicotinic acid/anatase. The dotted blue (solid red) curves are the absorption spectra for the ground state (T = 0 K) of clean TiO2 (dye/TiO2), while the thick solid black curves are the absorption spectra at T = 300 K, obtained from the average over the various snapshots of each dye/TiO2 MD simulation. The dashed black curves are the absorption spectra of the free catechol (a, b) and isonicotinic acid (c, d) molecules. The insets of (a), (b), (c), and (d) show the corresponding absorption spectra in the low-energy tail.

Table 1. Energy Differences between Mid-Gap States (HOMO and HOMO−1) and Valence Band Edge (VBE) (EHOMO − EVBE, EHOMO−1 − EVBE, in eV) and Adsorption Energy Eads (eV) for the Configurations Given in Figure 1a configurations

EHOMO − EVBE

EHOMO−1 − EVBE

Eads

catechol/anatase isonicotinic acid/anatase catechol/rutile isonicotinic acid/rutile

1.48 1.04 1.10 0.78

0.85 0.21 0.34 0.03

−0.72 −0.60 −0.45 −0.82

a

We determined the energy of the VBE (EVBE) on the clean surfaces, using the O 2s level as a common reference for the clean and sensitized surfaces.

3903


Article


(8) Mowbray, D. J.; Migani, A. Using G0W0 Level Alignment to Identify Catechol’s Structure on TiO2 (110). J. Phys. Chem. C 2015, 119, 19634−19641. (9) Li, S.-C.; Losovyj, Y.; Diebold, U. Adsorption-Site-Dependent Electronic Structure of Catechol on the Anatase TiO2 (101) Surface. Langmuir 2011, 27, 8600−8604. (10) Li, S.-C.; Chu, L.-N.; Gong, X.-Q.; Diebold, U. Hydrogen Bonding Controls the Dynamics of Catechol Adsorbed on a TiO2 (110). Science 2010, 328, 882−884. (11) Rangan, S.; Theisen, J.-P.; Bersch, E.; Bartynski, R. A. Energy Level Alignment of Catechol Molecular Orbitals on ZnO(112¯0) and TiO2 (110) Surfaces. Appl. Surf. Sci. 2010, 256, 4829−4833. (12) Calzolari, A.; Ruini, A.; Catellani, A. Surface Effects on Catechol/Semiconductor Interfaces. J. Phys. Chem. C 2012, 116, 17158−17163. (13) Duncan, W. R.; Prezhdo, O. V. Electronic Structure and Spectra of Catechol and Alizarin in the Gas Phase and Attached to Titanium. J. Phys. Chem. B 2005, 109, 365−373. (14) Duncan, W. R.; Prezhdo, O. V. Theoretical Studies of Photoinduced Electron Transfer in Dye-Sensitized TiO2. Annu. Rev. Phys. Chem. 2007, 58, 143−84. (15) Persson, P.; Lunell, S.; Ojamäe, L. Electronic Interactions between Aromatic Adsorbates and Metal Oxide Substrates Calculated from First Principles. Chem. Phys. Lett. 2002, 364, 469−474. (16) Nilsing, M.; Persson, P.; Ojamäe, L. Anchor Group Influence on Molecule-Metal Oxide Interfaces: Periodic Hybrid DFT Study of Pyridine Bound to TiO2 via Carboxylic and Phosphonic Acid. Chem. Phys. Lett. 2005, 415, 375−380. (17) Fratesi, G.; Motta, C.; Trioni, M. I.; Brivio, G. P.; SánchezPortal, D. Resonant Lifetime of Core-Excited Organic Adsorbates from First Principles. J. Phys. Chem. C 2014, 118, 8775−8782. (18) Onida, G.; Reining, L.; Rubio, A. Electronic Excitation: DensityFunctional versus Many-Body Green’s-Function Approaches. Rev. Mod. Phys. 2002, 74, 601−659. (19) Liu, L.-M.; Li, S.-C.; Cheng, H.; Diebold, U.; Selloni, A. Growth and Organization of an Organic Molecular Monolayer on TiO2: Catechol on Anatase (101). J. Am. Chem. Soc. 2011, 133, 7816−7823. (20) Syres, K. L.; Thomas, A. G.; Flavell, W. R.; Spencer, B. F.; Bondino, F.; Malvestuto, M.; Preobrajenski, A.; Grätzel, M. AdsorbateInduced Modification of Surface Electronic Structure: Pyrocatechol Adsorption on the Anatase TiO2 (101) and Rutile TiO2 (110) Surfaces. J. Phys. Chem. C 2012, 116, 23515−23525. (21) Giorgi, G.; Fujisawa, J.-i.; Segawa, H.; Yamashita, K. Unraveling the Adsorption Mechanism of Aromatic and Aliphatic Diols on the TiO2 Surface: A Density Functional Theory Analysis. Phys. Chem. Chem. Phys. 2013, 15, 9761−9767. (22) Sánchez-de-Armas, R.; San-Miguel, M. A.; Oviedo, J.; Márquez, A.; Sanz, J. F. Electronic Structure and Optical Spectra of Catechol on TiO2 Nanoparticles from Real Time TD-DFT Simulations. Phys. Chem. Chem. Phys. 2011, 13, 1506−1514. (23) Sánchez-de-Armas, R.; Oviedo, J.; San-Miguel, M. A.; Sanz, J. F. Direct vs Indirect Mechanisms for Electron Injection in Dye-Sensitized Solar Cells. J. Phys. Chem. C 2011, 115, 11293−11301. (24) Moser, J.; Punchihewa, S.; Infelta, P. P.; Grätzel, M. Surface Complexation of Colloidal Semiconductors Strongly Enhances Interfacial Electron-Transfer Rates. Langmuir 1991, 7, 3012−3018. (25) Persson, P.; Bergstrom, R.; Lunell, S. Quantum Chemical Study of Photoinjection Processes in Dye-Sensitized TiO2 Nanoparticles. J. Phys. Chem. B 2000, 104, 10348−10351. (26) Zhang, H. Z.; Banfield, J. F. Thermodynamic Analysis of Phase Stability of Nanocrystalline Titania. J. Mater. Chem. 1998, 8, 2073− 2076. (27) Ranade, M. R.; Navrotsky, A.; Zhang, H. Z.; Banfield, J. F.; Elder, S. H.; Zaban, A.; Borse, P. H.; Kulkarni, S. K.; Doran, G. S.; Whitfield, H. J. Energetics of Nanocrystalline TiO2. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 6476−6481. (28) Li, G.; Li, L.; Boerio-Goates, J.; Woodfield, B. F. High Purity Anatase TiO2 Nanocrystals: Near Room-Temperature Synthesis,

sensitization could be applied to such state, given its similar energy dependence on thermal fluctuations. In conclusion, our results show that finite temperature effects are small but not negligible for catechol- and isonicotinic acidsensitized TiO2 system. In particular, the changes of atomic geometry induced by thermal fluctuations modify the position of the molecular levels around the TiO2 valence band edge. For the anatase (101) surface, these fluctuations enhance significantly the absorption in the low-energy tail of the spectrum. Sensitization effects are less relevant for the rutile (110) surface.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.5b11885. Kohn−Sham energies of the gas phase catechol and isonicotinic acid molecules; contour plots of HOMO and HOMO−1 for the optimized dye/anatase geometries; PDOS for the ground state structure and selected snapshots of dye/rutile; absorption spectrum for the ground state structure and selected snapshots of dye/ rutile (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]; Tel +39-02-64485183; Fax +39-02-64485400 (H.L.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS H. Lin is grateful to Pirelli Co. for funding his PhD scholarship within the PCAM network. A.S. and S.S were supported by DoE-BES, Division of Chemical Sciences, Geosciences and Biosciences under Award DE-FG02-12ER16286. We used resources of the TIGRESS high performance computer center at Princeton University. We also thank A. Papagni and X. Shi for fruitful discussions.

■

REFERENCES

(1) O’Regan, B.; Grätzel, M. A low-cost, High-Efficiency Solar Cell Based on Dye-Sensitized Colloidal TiO2 Films. Nature 1991, 353, 737−740. (2) Grätzel, M. Recent Advances in Sensitized Mesoscopic Solar Cells. Acc. Chem. Res. 2009, 42, 1788−1798. (3) Hagfeldt, A.; Boschloo, G.; Sun, L.; Kloo, L.; Pettersson, H. DyeSensitized Solar Cells. Chem. Rev. 2010, 110, 6595−6663. (4) Zhang, L.; Cole, J. M. Anchoring Groups for Dye-Sensitized Solar Cells. ACS Appl. Mater. Interfaces 2015, 7, 3427−3455. (5) Abuabara, S. G.; Rego, L. G. C.; Batista, V. S. Influence of Thermal Fluctuations on Interfacial Electron Transfer in Functionalized TiO2 Semiconductors. J. Am. Chem. Soc. 2005, 127, 18234− 18242. (6) Li, S.-C.; Wang, J.-g.; Jacobson, P.; Gong, X.-Q.; Selloni, A.; Diebold, U. Correlation between Bonding Geometry and Band Gap States at Organic-Inorganic Interfaces: Catechol on Rutile TiO2 (110). J. Am. Chem. Soc. 2009, 131, 980−984. (7) Risplendi, F.; Cicero, G.; Mallia, G.; Harrison, N. M. A QuantumMechanical Study of the Adsorption of Prototype Dye Molecules on Rutile-TiO2(110): A Comparison between Catechol and Isonicotinic Acid. Phys. Chem. Chem. Phys. 2013, 15, 235−243. 3904


Article

The Journal of Physical Chemistry C Grain Growth Kinetics, and Surface Hydration Chemistry. J. Am. Chem. Soc. 2005, 127, 8659−8666. (29) Barnard, A. S.; Zapol, P.; Curtiss, L. A. Modeling the Morphology and Phase Stability of TiO2 Nanocrystals in Water. J. Chem. Theory Comput. 2005, 1, 107−116. (30) Barnard, A. S.; Curtiss, L. A. Prediction of TiO2 Nanoparticle Phase and Shape Transitions Controlled by Surface Chemistry. Nano Lett. 2005, 5, 1261−1266. (31) Burnside, S. D.; Shklover, V.; Barbé, C.; Comte, P.; Arendse, F.; Brooks, K.; Grätzel, M. Self-Organization of TiO2 Nanoparticles in Thin Films. Chem. Mater. 1998, 10, 2419−2425. (32) Wang, Y.; Hang, K.; Anderson, N. A.; Lian, T. Comparison of Electron Transfer Dynamics in Molecule-to-Nanoparticle and Intramolecular Charge Transfer Complexes. J. Phys. Chem. B 2003, 107, 9434−9440. (33) Janković, I. A.; Šaponjić, Z. V.; Č omor, M. I.; Nedeljković, J. M. Surface Modification of Colloidal TiO2 Nanoparticles with Bidentate Benzene Derivatives. J. Phys. Chem. C 2009, 113, 12645−12652. (34) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (35) Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Chiarotti, G. L.; Cococcioni, M.; Dabo, I.; et al. Quantum Espresso: A Modular and Open-Source Software Project for Quantum Simulations of Materials. J. Phys.: Condens. Matter 2009, 21, 395502. (36) Rocca, D.; Gebauer, R.; Saad, Y.; Baroni, S. Turbo Charging Time-Dependent Density-Functional Theory with Lanczos Chains. J. Chem. Phys. 2008, 128, 154105. (37) Malcıoğlu, O. B.; Gebauer, R.; Rocca, D.; Baroni, S. TurboTDDFT-A Code for the Simulation of Molecular Spectra Using the Liouville-Lanczos Approach to Time-Dependent DensityFunctional Perturbation Theory. Comput. Phys. Commun. 2011, 182, 1744−1754. (38) Car, R.; Parrinello, M. Unified Approach for Molecular Dynamics and Density-Functional Theory. Phys. Rev. Lett. 1985, 55, 2471−2474. (39) Nosé, S. A Molecular Dynamics Method for Simulations in the Canonical Ensemble. Mol. Phys. 1984, 52, 255−268. (40) Brivio, G. P.; Trioni, M. I. The Adiabatic Molecule-Metal Surface Interaction Theoretical Approaches. Rev. Mod. Phys. 1999, 71, 231−265. (41) Brivio, G. P.; Butti, G.; Caravati, S.; Fratesi, G.; Trioni, M. I. Theoretical Approaches in Adsorption: Alkali Adatom Investigations. J. Phys.: Condens. Matter 2007, 19, 305005. (42) Connor, P. A.; Dobson, K. D.; McQuillan, A. J. New Sol-Gel Attenuated Total Reflection Infrared Spectroscopic Method for Analysis of Adsorption at Metal Oxide Surfaces in Aqueous Solutions. Chelation of TiO2, ZrO2, and Al2O3 Surface by Catechol, 8Quinolinol, and Acetylacetone. Langmuir 1995, 11, 4193−4195. (43) Araujo, P. Z.; Mendive, C. B.; Rodenas, L. A. G.; Morando, P. J.; Regazzoni, A. E.; Blesa, M. A.; Bahnemann, D. FT-IR−ATR as a Tool to Probe Photocatalytic Interfaces. Colloids Surf., A 2005, 265, 73−80. (44) Scanlon, D. O.; Dunnill, C. W.; Buckeridge, J.; Shevlin, S. A.; Logsdail, A. J.; Woodley, S. M.; Catlow, C. R. A.; Powell, M. J.; Palgrave, R. G.; Parkin, I. P.; et al. Band Alignment of Rutile and Anatase TiO2. Nat. Mater. 2013, 12, 798−801. (45) Aryasetiawan, F.; Gunnarsson, O. The GW Method. Rep. Prog. Phys. 1998, 61, 237−312. (46) Fratesi, G.; Brivio, G. P.; Molinari, L. G. Many-Body Method for Infinite Nonperiodic Systems. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 69, 245113.

3905


Effects of Thermal Fluctuations on the Structure, Level Alignment, and

Recommend Documents