pubs.acs.org/Langmuir © 2010 American Chemical Society
Effect of Different Anchoring Groups on the Adsorption of Photoactive Compounds on the Anatase (101) Surface Pipsa Hirva* and Matti Haukka Department of Chemistry, University of Eastern Finland, P.O. Box 111, FI-80101 Joensuu, Finland Received June 17, 2010. Revised Manuscript Received August 18, 2010 The effect of replacing the anchoring carboxylate groups in the Ru(H2dcbpy)2(NCS)2 (H2dcbpy = 4,40 -dicarboxylic acid-2,20 - bipyridine) photoactive dye was studied by computational density functional theory (DFT) and timedependent DFT (TD-DFT) methods. The main emphasis in the study was to compare a series of attaching groups, including COOH, B(OH)2, PO(OH)2, SO2(OH), OH, NO2, and SiCl3, by the relative adsorption strength and geometry of the sensitizer molecules on the anatase (101) surface. Additionally, the substituent effect on the absorption signals in simulated UV-vis spectra was calculated with isolated dye molecules. Most of the selected substituents produced only small changes in the absorption characteristics of the dyes. However, OH groups were found to show a quite large blueshift compared to traditional COOH anchor groups in the simulated UV-vis spectra, while NO2 groups had an opposite effect of red-shifting the signals. On the other hand, although the NO2 substituents on the bipyridine ligands led to favorable absorption characteristics, the calculated adsorption strength of the NO2-substituted bipyridine models on the surface of anatase (101) was much smaller than that of the COOH-substituted one, indicating that larger modifications are necessary for both attaching the dye molecules on the surface and for tuning the absorption properties of photoactive compounds in the DSSC applications. The computational methods utilized here proved to be an efficient tool to study the effect of subtle structural changes on the properties of the dye molecules.
1. Introduction Dye-sensitized solar cell (DSSC) devices have attracted wide interest in the search for environmentally friendly alternatives for energy production. In order to obtain maximum efficiency in the DSSCs, it is especially important to find sensitizers, which not only attach firmly on the semiconductor surface, but also have optimal photochemical and charge transfer properties to enhance light-harvesting of the cells. Typically, the photoactive dyes in dye-sensitized solar cells consist of a transition metal complex attached to a semiconductor surface by different linking groups. A commonly used and one of the most effective compounds is Ru(H2dcbpy)2(NCS)2 (N3; H2dcbpy = 4,40 -dicarboxylic acid-2,20 - bipyridine),1,2 which has two carboxylic acid anchor groups in each bipyridine ligand (Figure 1). There have been numerous attempts to outperform the efficiency of the N3 dye. Different substituents have been added to replace the carboxylate groups, in order to either enhance the light-harvesting capacity, for example, by introducing ancillary ligands with thiophene containing long-chain hydrocarbons,3 or modify solubility of the compounds as well as the spectroscopic properties.4 Organic bridges or spacers of different lengths have been introduced between the anchor group and the acceptor *To whom correspondence should be addressed. E-mail: pipsa.hirva@ uef.fi. (1) Gr€atzel, M. Inorg. Chem. 2005, 44, 6841–6851. (2) Shklover, V.; Ovchinnikov, Yu. E.; Braginsky, L. S.; Zakeeruddin, S. M.; Gr€atzel, M. Chem. Mater. 1998, 10, 2553. (3) Cao, Y.; Bai, Y.; Yu, Q.; Cheng, Y.; Liu, S.; Shi, D.; Gao, F.; Wang, P. J. Phys. Chem. C 2009, 113, 6290–6297. (4) Schwalbe, M.; Sch€afer, B.; G€orls, H.; Rau, S.; Tschierlei, S.; Schmitt, M.; Popp, J.; Vaughan, G.; Henry, W.; Vos, J. G. Eur. J. Inorg. Chem. 2008, 3310– 3319. (5) Galoppini, E. Coord. Chem. Rev. 2004, 248, 1283–1297. (6) Persson, P.; Lundqvist, M. J.; Ernstorfer, R.; Goddard, W. A., III; Willig, F. J. Chem. Theory Comput. 2006, 2, 441–451.
Langmuir 2010, 26(22), 17075–17081
complex to prevent charge recombination and aggregation and to tune the properties of the dyes.5,6 Additionally, the effect of replacing the auxiliary NCS ligands on the binding of ruthenium dyes has been investigated.7-9 Replacing the metal complex with Zn-porphyrin-, Zn-phthalocyanine-, or organic perylene-based sensitizers has led to DSSCs with similar efficiency as the N3-derived complexes.10,11 Also the importance of the solvent at the TiO2/solvent interface has been addressed by molecular dynamics (MD) calculations.12 A high-level computational study of indoline dyes on TiO2 surface showed a large impact of intermolecular aggregation on the optical properties of the system.13 An important aspect of the performance of the DSSCs is how the dyes are attached to the semiconductor surface. The role of the anchor group is not only to bind the molecules on the surface, but also to inject the electrons from the excited dye molecule to the conducting band of the semiconductor. Even though other types of anchor groups have been suggested, such as chlorosilanes14 and ethoxysilanes,15 they have been much less studied than the traditional carboxylic acid and phosphonic acid groups. A recent review introduces the effect of adsorbing organic or ruthenium dyes on different single crystal semiconductor electrodes on the efficiency of DSSCs.16 Experimental studies have shown that the (7) Kilsa˚, K.; Mayo, E. I.; Brunschwig, B. S.; Gray, H. B.; Lewis, N. S.; Winkler, J. R. J. Phys. Chem. B 2004, 108, 15640–15651. (8) Aiga, F.; Tada, T. Sol. Energy Mater. Sol. Cells 2005, 85, 437–446. (9) De Angelis, F.; Fantacci, S.; Selloni, A.; Nazeeruddin, M. K. Chem. Phys. Lett. 2005, 415, 115–120. (10) Rochford, J.; Chu, D.; Hagfeldt, A.; Galoppini, E. J. Am. Chem. Soc. 2007, 129, 4655–4665. (11) Imahori, H.; Umeyama, T.; Ito, S. Acc. Chem. Res. 2009, 42, 1809–1818. (12) Schiffmann, F.; Hutter, J.; VandeVondele, J. J. Phys.: Condens. Matter 2008, 20, 064206–1-064206-8. (13) Pastore, M.; De Angelis, F. ACS Nano 2010, 4, 556–562. (14) Christ, C. S.; Yu, J.; Zhao, X.; Palmore, G. T. R.; Wrighton, M. S. Inorg. Chem. 1992, 31, 4439–4440. (15) Ford, W. E.; Rodgers, M. A. J. J. Phys. Chem. 1994, 98, 3822–3831. (16) Spitler, M.; Parkinson, B. A. Acc. Chem. Res. 2009, 42, 2017–2029.
Published on Web 10/05/2010
DOI: 10.1021/la102468s
17075
Article
Figure 1. Structure of the Ru(H 2 dcbpy)2 (NCS)2 (N3) dye. 2 The acidic protons have been placed at idealized positions.
phosphonic acid groups bind more strongly to the TiO2 surface than the carboxylic acid anchor groups.17,18 This is in accordance with a computational study of the adsorption of H3PO3 and COOH on TiO2 surface models, which estimated much larger adsorption energy for the phosphonate anchor group than the carboxylate one.19 Nevertheless, the detailed influence of the anchor groups on the efficiency of the cells is still relatively poorly known. The importance of the sensitizer adsorption modes of the N3-sensitized DSSCs has been addressed in several experimental and computational studies. Different binding modes of the dicarboxylate anchor groups have been suggested, depending on the degree of substitution and/or protonation of the bipyridine ligands. Periodic density functional theory (DFT) calculations on different adsorption modes for bi-isonicotinic acid on anatase (101) surface have led to conclusions that the dissociative bidentate bridging mode of both of the carboxylates is the most stable among the different options.20 Similar bidentate bridging binding of both dicarboxylate anchor groups belonging to the same bipyridine ligand has been found also on the rutile TiO2 surface,21 as well as for the N3 dye on TiO2 nanoparticles.22 Heteroleptic N3-based sensitizers have been found to adsorb onto TiO2 via a single bipyridine, which leads to considerable alteration in the electronic properties of the system.23 The N719 dye, which is a doubly deprotonated analogue of N3, has also been suggested to adsorb via two carboxylic groups residing on different bipyridines.24 In a very recent report, De Angelis et al. have shown by extensive Car-Parrinello molecular dynamics optimizations that doubly deprotonated N719 dye can adsorb onto TiO2 surface (17) Wang, P.; Klein, C.; Moser, J.-E.; Humphry-Baker, R.; Cevey-Ha, N.-L.; Charvet, R.; Comte, P.; Zakeeruddin, S. M.; Gr€atzel, M. J. Phys. Chem. B 2004, 108, 17553–17559. (18) Park, H.; Bae, E.; Lee, J.-J.; Park, J.; Choi, W. J. Phys. Chem. B 2006, 110, 8740–8749. (19) Nilsing, M.; Lunell, S.; Persson, P.; Ojamae, L. Surf. Sci. 2005, 582, 49–60. (20) Labat, F.; Adamo, C. J. Phys. Chem. C 2007, 111, 15034–15042. (21) Schnadt, J.; Br€uhwiler, P. A.; Patthey, L.; O’Shea, J. N.; S€odergren, S.; Odelius, M.; Ahuja, R.; Karis, O.; B€assler, M.; Persson, P.; Siegbahn, H.; Lunell, S.; Ma˚rtenson, N. Nature 2002, 418, 620–623. (22) Persson, P.; Lundqvist, M. J. J. Phys. Chem. B 2005, 109, 11918–11924. (23) De Angelis, F.; Fantacci, S.; Selloni, A.; Gr€atzel, M.; Nazeeruddin, M. K. Nano Lett. 2007, 7, 3189–3195. (24) De Angelis, F.; Fantacci, S.; Selloni, A.; Nazeeruddin, M. K.; Gr€atzel, M. J. Am. Chem. Soc. 2007, 129, 13156–14157. (25) De Angelis, F.; Fantacci, S.; Selloni, A.; Nazeeruddin, M. K.; Gr€atzel, M. J. Phys. Chem. C 2010, 114, 6054–6061.
17076 DOI: 10.1021/la102468s
Hirva and Haukka
via three carboxylate groups.25 Persson and Lundqvist found also additional surface interactions between the terminal sulfur of the auxiliary NCS ligands and the surface titanium atoms, which may further enhance the adsorption strength.22 This adds to the complexity of the binding modes and might alter the favorability of the anchoring groups. On the other hand, experimental evidence indicates that a considerable fraction of the N3 dye molecules is attached to the surface via only one of the dcbpy ligands.26 In the current work, we used computational density functional theory (DFT) and time-dependent DFT (TD-DFT) methods to study the effect of modifying the anchor groups at the bipyridine rings in the N3-based ruthenium dye. We were especially interested in the relative adsorption energies of the complexes on the anatase (101) surface. The motivation was to find groups with optimal attaching properties and which could also be used to tune the absorption characteristics of the dye molecules on the titanium oxide surfaces. Surface Models. On the basis of our earlier studies on the adsorption of the N3 dye on the anatase (101) surface, we selected a two-layer surface model Ti14O42H28 (S14) for the preliminary studies.27 The results were compared to a larger model Ti26O75H46 (S26) which would be more suitable for larger adsorbates such as substituted bipyridine. Both models, shown in Figure 2, were terminated and neutralized with hydrogen atoms. In the optimizations, the effect of relaxation was investigated by allowing the active titanium sites and the neighboring oxygens to relax, while the surrounding atoms were kept fixed at their bulk structure. Surface Adsorbates. The photochemical properties were studied with isolated dye complexes. Initial geometries were taken from the crystal structure of the N3 dye (Figure 1),2 which was used as a basic structural motif where the anchoring groups were varied. Each dye molecule was allowed to fully optimize. TD-DFT calculations on the photochemical properties were performed in acetonitrile solvent using the gas phase optimized geometries. Since the goal was to model the effect of different attaching groups on the surface, we chose a series of pyridine derivatives substituted by anionic CO2-, BO22-, PO32-, SO3- and O-, groups (Figure 3) as model compounds. In all cases, only the dissociative adsorption was considered, where the protons were attached to adjacent oxygen sites at the surface. Furthermore, neutral attaching groups NO2 and SiCl3 were also tested. Finally, bipyridine adsorption with two attaching groups was also calculated to see the effect of the larger ligand on the preference of the adsorbates. In the bipyridine models, the central N-C-C-N torsion was constrained to zero in order to mimic the cis-conformation found in the dye complexes. Figure 3 shows also a schematic representation of the adsorption modes described in the text.
2. Computational Details All computations were done with the Gaussian03 program package.28 We used a DFT method with hybrid GGA functional PBE1PBE (PBE0),29 which has been found effective in simulating spectral properties and geometries in organic and inorganic sensitizers.30,31 The basis set for the surface Ti and O atoms was a standard LANL2DZ basis set,32 which includes an effective core potential for Ti, and Dunning’s double-ζ valence basis for O. The basis set for the adsorbates was a standard 6-31G(d) for all atoms except for binding oxygens, where a slightly more flexible 6-311þ G(d) basis was used. Basis set superposition (BSSE) correction to the interaction energies was done by full counterpoise method. (26) Benk€o, G.; Kallioinen, J.; Myllyperki€o, P.; Trif, F.; Korppi-Tommola, J. E. I.; Yartsev, A. P.; Sundstr€om, V. J. Phys. Chem. B 2004, 108, 2862–2867. (27) Haukka, M.; Hirva, P. Surf. Sci. 2002, 511, 373–378.
Langmuir 2010, 26(22), 17075–17081
Hirva and Haukka
Article
Figure 2. Surface cluster models for anatase (101) surface. Left, Ti14O42H28 (S14); right, Ti26O75H46 (S26). scaling was performed for the isolated dye molecules to ensure optimization to true minima.
3. Results and Discussion
Figure 3. Schematic presentation of the model adsorbates. (a) Pyridine model, (b) bipyridine model, and (c) presentation of the adsorption modes. Including BSSE corrections in the adsorption energies was important, since owing to a rather limited basis set description of the surface models the basis set superposition error was found to have a large impact on the adsorption energies, up to 20-30%. However, although the absolute adsorption energies were substantially affected, the relative trends in the energies remained the same as they did without the BSSE correction. In the separate Ru(X2bpy)2(NCS)2 complexes, the ruthenium atom was described with the quasirelativistic Stuttgart&Dresden small core effective core potential.33 For other atoms, the same basis sets were used as with the small surface adsorbates. Solvent effects were included in the TD-DFT calculations by the conductor-like CPCM method.34Frequency analysis with no (28) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; M. A. Al-Laham, Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; , Pople, J. A. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (29) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865– 3868. (30) Jacquemin, D.; Perpte, E. A.; Scuseria, G. E.; Ciofini, I.; Adamo, C. J. Chem. Theory Comput. 2008, 4, 123–135. (31) Vlcek, A., Jr.; Zalis, S. Coord. Chem. Rev. 2007, 251, 258–287. (32) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310. (33) Andrae, D.; Haeussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim. Acta 1990, 77, 123–141. (34) Cossi, M.; Rega, N.; Scalmani, G.; Barone, V. J. Comput. Chem. 2003, 24, 669–681.
Langmuir 2010, 26(22), 17075–17081
TD-DFT Studies on Ruthenium Dyes. At the first stage, we calculated the effect of different anchoring groups on the photoactive properties of the ruthenium dyes, which are based on modified N3 (Figure 1). Table 1 lists the spectral data and the energies of the HOMO and LUMO orbitals of the complexes. In all cases, the absorption spectrum consists of three main signals: one in the visible region, one in the near-UV, and a large signal in the UV region. The two experimental spectra found in the literature (for Ru[(COOH)2bpy]2(NCS)2 and Ru[(PO(OH)2)2bpy]2(NCS)2, Table 1) were very well reproduced by the TD-DFT calculations. The tendency of the PO(OH)2 substituents to blue-shift the spectral signals can also be seen in the computational results; therefore,we were confident that the computational method would give reliable information on the effect of different anchor groups on the photochemical properties of the dye complexes. When the different substituents at the bipyridine rings are compared, the most drastic spectral changes can be found with the strongly electron-withdrawing NO2 group, which produced a large red-shift for all three signals. Complexes with OH groups showed an opposite shift of the signals, and hence, a blue-shift is observed compared to the N3 dye. Furthermore, the strongest effect can be observed at the second signal λ(II), which corresponds to transitions between the orbitals with mixed thiocyanate metal-to-ligand character and the bipyridine π* orbitals, as already interpreted in the literature.31,35,36 However, the mixed nature of the transitions makes the signal very broad, thus complicating the determination of the actual wavelength. Interestingly, the signal at the UV region, λ(III), which involves the intraligand π-π* transitions of the bipyridine ligand was almost unaffected by the substitutions, except with X = OH or NO2. The change in the signal wavelengths according to the modification of the anchor group is shown in Figure 4. The change in the energy of the spectral signals can be correlated to the energies of the frontier molecular orbitals, as shown in Figure 5. There is a direct correlation of the signal energy E(I) with the HOMO-LUMO gap of the molecules, although the gap energy is somewhat larger than the corresponding signal maxima. Therefore, calculating the change in the HOMO-LUMO gap gives information on the direct vertical absorption characteristics of the dye molecules. The Effect of Relaxation of the Surface Models. The effect of relaxation on the geometries and energies of the active surface sites was first studied with the separate surface models S14 and S26. Figure 6 presents the changes in the relaxed surface atoms for the smaller S14 model. The effect of relaxation for both models is listed in Table 2. It should be noted that the same surface atoms, DOI: 10.1021/la102468s
17077
Article
Hirva and Haukka
Table 1. Spectral Data for Modified Ruthenium Dyes, Ru(X2bpy)2(NCS)2 in Acetonitrile Solvent According to the TD-DFT Calculationsa X= COOH
B(OH)2
PO(OH)2
SO2(OH)
OH
NO2
SiCl3
λ(I) 528 478 512 524 459 617 525 λ(II) 410 361 354 363 324 558 385 λ(III) 282 278 275 274 252 326 279 E(I), eV 2.35 2.59 2.42 2.37 2.70 2.01 2.36 E(II), eV 3.02 3.43 3.50 3.42 3.83 2.22 3.22 E(III), eV 4.40 4.46 4.51 4.53 4.92 3.80 4.44 ε(HOMO) -5.75 -5.47 -5.70 -6.02 -5.14 -5.94 -5.74 ε(LUMO) -2.92 -2.33 -2.74 -3.08 -1.88 -3.61 -2.86 gap 2.83 3.14 2.96 2.94 3.27 2.33 2.88 a Wavelengths are given in nm, and energies in eV. b Water solvent (pH = 1).35 c Solvent 0.1 M H2SO4.36
exp COOHb
exp PO(OH)2c
520 390 312 2.30 3.12 3.95
494 360 303 2.51 3.44 4.09
Figure 4. Effect of different anchor groups in the N3-derived dyes on the simulated spectra relative to X = COOH according to TD-DFT calculations.
Figure 5. Correlation between the spectral signal Ε(I) and the HOMO-LUMO gap (in eV) of the N3-derived dye molecule.
representing the adsorption site, were relaxed also in the larger model and the resulting changes were consequently the same. 17078 DOI: 10.1021/la102468s
When the active surface atoms were allowed to relax, the oxygens at the step site (Ostep) moved inward, and the oxygens at the plane site (OS) moved upward. At the same time, the surface titanium atoms (TiS) moved slightly down toward the second layer. As a result, the Ti-O distances became either smaller (Ostep-TiS; TiS-O2 L) or larger (Ostep-Ti2 L ; OS-TiS) than those in the nonrelaxed surface model. Similar tightening of the Ti-O bonds upon relaxation has been observed with periodic DFT studies of the anatase surface.20 Because of the similarity of the modifications, the relative energies compared to the nonrelaxed models were very similar for both surface models. Adsorption Energies of the Pyridine Models. The TD-DFT results indicated that modification of the anchor groups in the N3-derived dyes creates a strong effect on the absorption characteristics of the separate molecules. The next step in the current work was to study the influence of the different attaching groups on the adsorption strength of them at the anatase (101) semiconductor surface. The corresponding BSSE corrected adsorption energies of the pyridine models are shown in Table 3. In the case of small pyridine models (Figure 1a), we only considered bidentate bridging Langmuir 2010, 26(22), 17075–17081
Hirva and Haukka
Article
Figure 6. Optimized structure of the relaxed surface model S14. The arrows show the direction of the relaxation. Table 2. Effect of Relaxation on the Selected Distances (A˚) and Energies (kJ/mol) in the Surface Models S14 and S26 parameter
S14 (relax)
S26 (relax)
nonrelaxed
Ostep-TiS Ostep-Ti2 L TiS-O2 L OS-TiS ΔErelax-nonrelax
1.723 2.003 1.841 1.931 -180
1.745 1.976 1.834 1.940 -194
1.973 1.930 1.930 1.930
Table 3. BSSE Corrected Adsorption Energies (kJ/mol) for Adsorbates with Different Anchor Groups and Optimized Distances from the Surface Sites (S = Surface, A = Adsorbate)a X= COOH B(OH)2 PO(OH)2 SO2(OH) ΔΕ(S14) ΔΕ(S26) ΔΕ(S14)fixed d(S-A)1 d(S-A)2
-154 -149 -205 2.025 2.047
-123 -116 -191 1.840 1.887
-223 -229 -324 1.902 1.918
-179 -182 -253 2.045 2.077
OH
NO2 SiCl3
-146 -44 -15 -148 -49 -19 -208 -78 -43 2.002 2.380 2.782 2.511 2.876
a Geometrical parameters are reported for the larger (relaxed) surface model (S26) only.
(1BB) adsorption mode between two neighboring Ti atoms at the surface, since for single attaching group it was found to lead to optimal adsorption in our previous study.27 Both S14 and S26 models gave the same adsorption energies for all substituted pyridine models, which indicates that the smaller surface model is sufficient for studying adsorption on the anatase (101) surface as long as the adsorbates are not very large. The smaller model, which can be calculated about 10 times faster than the larger one, can therefore be used to predict the relative adsorption strength of different anchor groups. According to the adsorption energies, the strongest interaction was found with the tridentate PO(OH)2 and SO2(OH) groups, even though their adsorption mode was effectively bidentate and similar to the other 1BB adsorptions. The interaction was further strengthened by the fact that the third oxygen was able to form hydrogen-bonding interactions to the dissociated hydrogen atoms at the neighboring Ostep sites. The OH group showed similar adsorption energies to those of the COOH group, and can therefore be estimated to make a potential anchor group. The effect of relaxation on the adsorption energies of the substituted pyridine models was also studied with the smaller S14 surface model. Although the relaxed surface models were (35) Fantacci, S.; De Angelis, F.; Selloni, A. J. Am. Chem. Soc. 2003, 125, 4381– 4387. (36) Zabri, H.; Gillaizeau, I.; Bignozzi, C. A.; Caramori, S.; Charlot, M.-F.; Cano-Boquera, J.; Odobel, F. Inorg. Chem. 2003, 42, 6655–6666.
Langmuir 2010, 26(22), 17075–17081
much more stable than the nonrelaxed ones, the stability of the surface-adsorbate systems was also enhanced when relaxation was taken into account and the overall effect on the adsorption energetics was smaller, as can be seen from the adsorption energies for the fixed surface in Table 2. Furthermore, the relative adsorption energies of the pyridines with different anchor groups were not essentially changed, indicating that the stability order of the adsorbates can also be estimated on the nonrelaxed surface models. The overall effect of the adsorbates on the relaxed surface sites was to cancel the relaxation, so that the surface atoms moved toward their idealized positions. This effect was more complete with the adsorbates which adsorbed strongly on the surface, such as PO(OH)2 and SO2(OH), and therefore, the difference between the adsorption energies on the relaxed and fixed surfaces was larger with these anchor groups. Bipyridine Adsorption. The adsorption studies with the small pyridine model gave information on the relative adsorption strength of the different anchoring groups in the case where the dye molecule binds via only one group. In previous studies,20-23 it has been suggested that at least two of the anchor groups, either in the same bipyridine ligand or in different bipyridines, could bind to the surface sites of anatase, although even more complicated adsorption modes involving up to three adsorption sites have been suggested.25 In most cases, the results indicate that the most stable adsorption mode would be the 2BB mode, where both anchor groups have a bridging bidentate interaction with two surface titanium atoms. However, this adsorption mode led to substantial twisting of the two pyridine rings as well as off-plane twist of the substituents, and/or large modifications in the geometry of the surface sites, in order to enable adsorption of all four oxygens on the surface. In the actual N3 dye, the twist between the py rings is limited because of the binding to the central ruthenium atom, leading to effectively zero twist. The reorganization is not necessary, if the two carboxylate groups are adsorbed via monodentate modes (2M), which makes it easier to compare different anchoring groups. It should be stressed, however, that the adsorption of a single bipyridine model does not necessarily give accurate adsorption characteristics of the actual N3-derived dyes, where more complicated adsorption modes are possible. We tested the relative stability of substituted bipyridine models (Figure 3b) to account for the different geometrical properties of the anchor groups. To facilitate a comparison between several different anchor groups, only one likely binding mode was selected, the 2M mode, where both of the anchor groups attach monodentately to the surface Ti atoms. Depending on the geometries of the bpy ligands, different surface sites are possible. In this study, we considered two different binding sites, A and B, which have slightly different Ti-Ti distances. The 2MA and 2MB binding modes are shown in Figure 7. As in the case of pyridine models, only dissociative adsorption was considered with the hydrogens attached to neighboring oxygen sites. The BSSE corrected adsorption energies together with the key geometrical parameters for the bipyridine models in binding sites A and B are shown in Table 4. It should be noted that, because of the larger size of the bipyridine model, the larger surface model S26 was required in order to minimize the edge effects even though the smaller surface model was found to reproduce the adsorption energies very well in the 1BB binding of pyridine. Furthermore, the inherently larger number of optimized parameters prevented the relaxation of the surface sites; therefore, we optimized the adsorbates on a fixed surface model. However, we tested the effect of surface relaxation compared to fixed surface in the case of pyridine models and, as already described in the previous section, the effect on the relative adsorption energies was DOI: 10.1021/la102468s
17079
Article
Hirva and Haukka
Figure 8. Partial 1BBþ1M binding mode of the SO2(OH) substituted bipyridine model.
Figure 7. (a) Schematic presentation for the double monodentate binding modes for bipyridine models with different anchoring groups. Distance between the active surface Ti is shown. (b) Optimized structure of the bipyridine model with X = COOH (dissociative adsorption) in the binding mode 2MA. Table 4. BSSE Corrected Adsorption Energies (kJ/mol) for Bipyridine Adsorbates with Different Anchor Groups According to Double Monodentate Binding Modes 2MA and 2MB, and Optimized Distances from the Surface Sites (S = Surface, A = Adsorbate)a X= COOH B(OH)2 PO(OH)2 SO2(OH) OH
NO2 SiCl3
ΔΕ(2MA) -383 -130 -462 -424 -301 -121 -32 ΔΕ(2MB) -342 -24 -296 -331 -334 -143 5 1.908 1.860 1.989 2.002 2.085 2.651 d(S-A)1A 1.932 1.888 1.878 1.979 2.172 2.123 2.704 d(S-A)2A 1.954 1.831 1.872 1.944 2.000 2.162 2.723 d(S-A)1B 1.915 1.933 1.889 2.009 2.003 2.156 2.947 d(S-A)2B 1.920 a Adsorbates were fully optimized on the fixed surface model (S26).
small compared to the differences between various anchoring groups. The energetics in Table 4 show similarity in the relative trends of the adsorption strength of different anchor groups between the bipyridine and the pyridine models (Table 2). Even though the relative trends are mainly the same, the matching geometry with the surface sites becomes more important with the larger bipyridine adsorbates. Generally, the adsorbates should fit better in the 2MB mode. However, the flexibility in the bipyridine ligand enabled it to bend slightly so that for most anchor groups the A site became more favorable. This flexibility was already observed in our previous study with the N3 dye.27 Furthermore, the torsional freedom of the anchor groups can lead to partial bidentate bridging adsorption mode, which further strengthens the interaction. Hence, the tridentate PO3 and SO3 groups optimized in 1BBþ1M interaction mode, where one of the anchor 17080 DOI: 10.1021/la102468s
groups interacted with two titanium sites, as shown in Figure 8. Consequently, this additional interaction increased the adsorption energies compared to pure 2M adsorption modes. The only options, where the B site was clearly more favorable than the A site, were (OH)bpy and (NO2)bpy models. The adsorption strength more or less correlates with the distance of the adsorbates from the active surface sites: the shorter the distance, the stronger the adsorption. Naturally, the whole geometry of the adsorbate plays an important role, which can be seen, for example, in the adsorption of B(OH)2 substituted bipyridine. The longer B-O distances in the B(OH)2 group compared to C-O distances in COOH led to less symmetric adsorption and to decreased match to the distance between the surface titanium atoms, and consequently to decreased adsorption energies, even though with the smaller pyridine models the difference between the B(OH)2 and COOH groups was much smaller. Therefore, with the larger adsorbates, such as the N3-derived dyes, a sufficiently large surface model will give more reliable information on the actual energetics and geometries of the adsorbed species. Furthermore, our model adsorbates give no information on the effect of the central ruthenium atoms. It can be predicted, however, that the metal atoms would have a large effect on other sensitizing processes, such as electron injection, but only a minor effect on the relative adsorption energies. These types of calculations are very demanding at the DFT level of theory, and limitations on the models/optimized parameters enable faster screening of the potential anchoring groups.
4. Conclusions Ruthenium dyes based on Ru(H2dcbpy)2(NCS)2 complexes (N3) were modified by replacing the anchoring carboxylate substituents with different acidic or neutral groups, and their effect on the absorption characteristics and interaction with anatase (101) surface sites was studied by computational DFT methods. The results for the frequently applied carboxylate and phosphonate groups were in excellent agreement with the experimental observations, verifying that the utilized computational methods are an effective tool, when the effect of subtle geometrical modifications is studied. OH groups, which until now have not been considered suitable anchor groups, were in this work found to be a potential option for attaching the N3-derived dye molecules with bipyridine ligands on the TiO2 semiconductor surface. The (OH)bpy ligands exhibit suitable geometries to match the surface sites on the anatase (101) surface, and therefore, rather strong adsorption can be formed. However, TD-DFT studies on the isolated dye molecules showed that OH groups lead to a large blue-shift in the UV-vis spectrum compared to typically used COOH groups, Langmuir 2010, 26(22), 17075–17081
Hirva and Haukka
Article
which might limit the light-harvesting properties of the dye. Another strongly affecting group would be NO2, which was found to produce a large red-shift compared to carboxylic acid substituents, but its adsorption strength was found much weaker than that of most of the other anchor groups (although in the case of bipyridine model the suitable geometry of the attaching groups was found to enhance the adsorption strength). Therefore, another more strongly attaching group would be required in combination of the NO2 group. In summary, it seems that it will be very difficult to outperform the N3 dye, at least by introducing modifications on its bipyridine
substituents. Of course, many other aspects should be accounted for when designing more efficient dye molecules for DSSCs. For example, although the thiocyanate auxiliary ligands have shown the best performance, probably because of their active role in the electronic excitations, the effect of other, more neutral ligands should be investigated more extensively. Also, the replacement of the original bipyridine ligands with other nitrogen ligands would make a difference in the optical properties of the complex, as was shown in our recent work with 1D ruthenium chains.37 A computational study of such combinatorial effects is one of our future goals.
(37) Niskanen, M.; Hirva, P.; Haukka, M. Phys. Chem. Chem. Phys. 2010, 12, 9777–9782.
Acknowledgment. Financial support from the Academy of Finland is gratefully acknowledged (Grant 129772).
Langmuir 2010, 26(22), 17075–17081
DOI: 10.1021/la102468s
17081
