Supplementary Material: A Theoretical Investigation of Adsorption ...
Supplementary Material: A Theoretical Investigation of Adsorption, Dynamics, Self-Aggregation and Spectroscopic Properties of the D102 Indoline Dye on an Anatase (101) Substrate. Susanna Monti,∗,†,‡ Mariachiara Pastore,∗,¶,§ Cui Li,‡,k Filippo De Angelis,¶ and Vincenzo Carravettak CNR-ICCOM, Institute of Chemistry of Organometallic Compounds, via G. Moruzzi 1, I-56124 Pisa, Italy, Theoretical Chemistry and Biology, School of Biotechnology, KTH Royal Institute of Technology, SE-10044 Stockholm, Sweden, Computational Laboratory for Hybrid Organic Photovoltaics (CLHYO), Institute of Molecular Science and Technologies (ISTM-CNR), Via Elce di Sotto, 8, I-06123, Perugia, Italy, CNRS, Théorie−Modélisation−Simulation, SRSMC, Boulevard des Aiguillettes, 54506 Vandoeuvre−lés-Nancy, Frnace, and CNR-IPCF, Institute of Chemical and Physical Processes, via G. Moruzzi 1, I-56124 Pisa, Italy E-mail: [email protected]; [email protected],[email protected]
∗ To
whom correspondence should be addressed Institute of Chemistry of Organometallic Compounds, via G. Moruzzi 1, I-56124 Pisa, Italy ‡ Theoretical Chemistry and Biology, School of Biotechnology, KTH Royal Institute of Technology, SE-10044 Stockholm, Sweden ¶ Computational Laboratory for Hybrid Organic Photovoltaics (CLHYO), Institute of Molecular Science and Technologies (ISTM-CNR), Via Elce di Sotto, 8, I-06123, Perugia, Italy § CNRS, Théorie−Modélisation−Simulation, SRSMC, Boulevard des Aiguillettes, 54506 Vandoeuvre−lés-Nancy, Frnace k CNR-IPCF, Institute of Chemical and Physical Processes, via G. Moruzzi 1, I-56124 Pisa, Italy † CNR-ICCOM,
1
Validation of the Reactive Force Field: comparison with DFT optimized structures (Cluster Models). The force field selected for this work is a relatively recent development of the original glycine/TiO2 parametrization, which was extended to simulate, on the one hand, all the amino acids, by fitting the parameters to large quantum chemistry-based, training sets, including, minimum energy conformations of all residues, dipeptides, tripeptides, and longer peptide chains, and, on the other hand, titanium oxide structures, by fitting equations of state for bulk anatase, rutile, brookite, high energy titanium dioxide crystals, surface energies, etc. 1–7 This reactive parametrization was validated extensively by comparison to quantum mechanics and dynamics data but also classical MD simulations based on highly specialized force fields tuned to describe biomolecular systems. However, even though it was employed, subsequently, for other types of molecules successfully, a check on the conformation of the dye was mandatory. The first step of the whole procedure consisted of a plain geometry optimization of a complex made of the dye and a small nanoparticle of anatase, 8 by using the molecular mechanics ReaxFF approach. The starting configuration was obtained through energy minimization by means of density functional theory calculations carried out at the B3LYP/6-31G* level. From the comparison of the quantum and classical optimized structures it was apparent that the MM methodology was suitable and reliable for these kinds of systems because the two geometries, namely quantum and classical structures, were almost identical (Figure 1), having a root mean square deviation (RMSD) of all the atomic coordinates of at most 0.55 Å, which was reduced to 0.07 Å when only the heavy atoms of the dye were taken into account in the superimposition process. The largest value is essentially due to a slight readjustment of the atoms of the slab and to a shift of the molecule to a farther distance from the interface. Indeed, the new distances between the titanium atoms of the interface and the oxygen atoms of the dye changed by about 0.2 Å (passing from about 2.1 to about 2.3 Å), in agreement with the earlier parametrization of this type of interactions. 2,3,9 These results can be considered very satisfactory because, on the one hand, the structure of the dye is preserved, not distorted and 2
Figure 1: Geometry optimization of a single molecule adsorbed on a small anatase (101) cluster.
appropriately represented as a considerably rigid system and, on the other hand, the whole complex has a three dimensional configuration more compatible with the anatase crystal obtained in previous studies. 7 As a consequence, it was neither necessary to develop new parameters for the dye nor to re-parametrize the interactions between the dye and the atoms of the support.
Single Dye Molecular Dynamics Simulations: Validation of the Reactive Force Field. The second check comprised the evaluation of the dynamics of the molecule when adsorbed on top of a periodic anatase (101) interface, which was more realistic than a relatively small nanoparticle. Indeed, the size and morphology of nanoparticles should be determined in agreement with experimental data, or otherwise considering the length of dye in order to explore different ways of
3
adsorption. The choice of an infinite surface is justified further by the fact that no border effects are present. Indeed, in the case reported in ref. 8 the dimensions of the substrate are too small and part of the molecule, which is very long (about 20 Å), could bend and stretch out of the edges of the anatase cluster, thus being subjected to different forces caused by the frontier atoms. These spurious perturbations could play important roles in the adsorption and dynamic of the molecule and be misleading in the comparison of the computational data with the experimental findings, especially when the samples, even though multi-faced, are larger.
Preliminary Check: Dimers on a Anatase Nanoparticle. Considering that self-interactions are important in the description of adsorption, conformational organization and distribution of the dye on the surface, before starting the production phase, it was verified if these interactions were well accounted for by the force field. Indeed, the data obtained in the case of dimers optimized at the DFT level 8 were compared with the force field-based energy minimizations. The six most stable DFT conformations found in the earlier investigation 8 were considered. The tendency of the dye to be inclined relatively to the surface was apparent in all the models. Indeed, the molecules were not aligned vertically, but they were quite bent to self-assemble effectively through stacking and T-shaped interactions of the donor portion of the adsorbates (phenyl rings). Their main contacts with the support were established through carboxyl oxygen and sulfur coordination to the titanium sites. The ReaxFF optimized structures were very similar to the initial configurations with a maximum RMSD of all heavy atoms of about 1 Å (the minimum value was about 0.65 Å). This was mainly due to the readjustment of the atoms of the cluster (maximum RMSD = 0.9 Å) which became more compactly arranged. As already observed, Ti-O(dye) distances were longer (about 0.2 Å) and a slightly reorientation of the phenyl rings took place (Figure 2), probably because, differently from the B3LYP functional, dispersion interactions are taken into account in the force field picture.
4
Figure 2: Two different views of the D102(40) 8 optimized geometry at the DFT (left hand side) and ReaxFF (right hand side) levels of theory. This is the most stable arrangement in the ReaxFF description.
5
Surface Deformation To elucidate the role played by the dye layer in influencing the surface relaxation and thus the morphology of the interface, benchmark MD simulations of the anatase bulk and clean surface (small and large supercell) models were performed and the results were compared with the functionalized substrates trajectories choosing as descriptors all possible atom-atom radial distribution functions, namely the Ti-Ti, O-O and Ti-O pairs (see plotted data in Figure 3). The results indicate that the slab has the tendency to conserve the internal crystal structure, which is very close to its bulk (light orange areas). However, as depicted by the Ti-Ti radial distribution functions, the substrates seem more crystalline in the presence of the adsorbate than in vacuum. This partial loss of crystallinity is suggested by the slight decrease in the height of the peak at about 3 Å and the parallel increase in the height of the minimum located between the first and the second peak of all the curves. However, the effect is not negligible and affects mainly the interfaces. In fact, MD simulation results evidence that the dyes and the top layer of the slab are attracted towards each other and just the portion of the surface facing the molecule becomes distorted.
6
Figure 3: Radial distribution functions of the titanium and oxygen atoms of the slab in the different simulations (see legend). The behavior of the clean and functionalized surfaces is show for both the small and large supercell models. Comparison with the bulk anatase (light orange area under the curves) is also displayed. 7
Dye Aggregates Dynamics Snapshots
Figure 4: Final snapshot of the adsorption of D102 on a small anatase (101) supercell (40.8397×15.1040 Å2 ) - low density (8 molecules) and slow equilibration. The simulation supercell has been replicated in order to visualize complete molecular structures. Hydrogens have been undisplayed for clarity.
8
Figure 5: Final snapshot of the adsorption of D102 on a small anatase (101) supercell (40.8397×15.1040 Å2 ) - high density (12 molecules). The simulation supercell has been replicated to visualize complete molecular structures. Hydrogens have been undisplayed for clarity.
References (1) Monti, S.; Corozzi, A.; Fristrup, P.; Joshi, K. L.; Shin, Y. K.; Oelschlaeger, P.; van Duin, A. C. T.; Barone, V. Exploring the Conformational and Reactive Dynamics of Biomolecules in Solution using an Extended Version of the Glycine Reactive Force Field. Phys. Chem. Chem. Phys. 2013, 15, 15062–15077. (2) Monti, S.; van Duin, A. C. T.; Kim, S.-Y.; Barone, V. Exploration of the Conformational and Reactive Dynamics of Glycine and Diglycine on TiO2 : Computational Investigations in the Gas Phase and in Solution. J. Phys. Chem. C 2012, 116, 5141–5150.
9
(3) Rahaman, O.; van Duin, A. C. T.; III, W. A. G.; Doren, D. J. Development of a ReaxFF Reactive Force Field for Glycine and Application to Solvent Effect and Tautomerization. J. Phys. Chem. B 2011, 115, 249–261. (4) Kim, S.-Y.; van Duin, A. C. T. Simulation of Titanium Metal/ Titanium Dioxide Etching with Chlorine and Hydrogen Chloride Gases Using the ReaxFF Reactive Force Field. J. Phys. Chem. A 2013, 117, 5655–5663. (5) Kim, S.-Y.; Kumar, N.; Persson, P.; Sofo, J.; van Duin, A. C. T.; Kubicki, J. D. Development of a ReaxFF reactive force field for Titanium dioxide/water systems. Langmuir 2013, 29, 7838– 7846. (6) Kim, S.-Y.; van Duin, A. C. T.; Kubicki, J. D. Simulations of the Interactions between TiO2 Nanoparticles and Water with Na+ and Cl− , Methanol and Formic Acid Using a Reactive Force Field. J. Mater. Res. 2013, 28, 513–520. (7) Raju, M.; Kim, S.-Y.; A. C. T. van Duin, K. A. F. ReaxFF Reactive Force Field Study of the ˘ S10572. Dissociation of Water on Titania Surfaces. J. Phys. Chem. C 2013, 117, 10558âA ¸ (8) Pastore, M.; Angelis, F. D. Aggregation of Organic Dyes on TiO2 in Dye-Sensitized Solar Cells Models: An Ab Initio Investigation. ACS NANO 2010, 4, 556–562. (9) Li, C.; Monti, S.; Carravetta, V. Journey Toward the Surface: How Glycine Adsorbs on Titania in Water Solution. J. Phys. Chem. C 2012, 116, 18318–18326.
10
∗ To
whom correspondence should be addressed Institute of Chemistry of Organometallic Compounds, via G. Moruzzi 1, I-56124 Pisa, Italy ‡ Theoretical Chemistry and Biology, School of Biotechnology, KTH Royal Institute of Technology, SE-10044 Stockholm, Sweden ¶ Computational Laboratory for Hybrid Organic Photovoltaics (CLHYO), Institute of Molecular Science and Technologies (ISTM-CNR), Via Elce di Sotto, 8, I-06123, Perugia, Italy § CNRS, Théorie−Modélisation−Simulation, SRSMC, Boulevard des Aiguillettes, 54506 Vandoeuvre−lés-Nancy, Frnace k CNR-IPCF, Institute of Chemical and Physical Processes, via G. Moruzzi 1, I-56124 Pisa, Italy † CNR-ICCOM,
1
Validation of the Reactive Force Field: comparison with DFT optimized structures (Cluster Models). The force field selected for this work is a relatively recent development of the original glycine/TiO2 parametrization, which was extended to simulate, on the one hand, all the amino acids, by fitting the parameters to large quantum chemistry-based, training sets, including, minimum energy conformations of all residues, dipeptides, tripeptides, and longer peptide chains, and, on the other hand, titanium oxide structures, by fitting equations of state for bulk anatase, rutile, brookite, high energy titanium dioxide crystals, surface energies, etc. 1–7 This reactive parametrization was validated extensively by comparison to quantum mechanics and dynamics data but also classical MD simulations based on highly specialized force fields tuned to describe biomolecular systems. However, even though it was employed, subsequently, for other types of molecules successfully, a check on the conformation of the dye was mandatory. The first step of the whole procedure consisted of a plain geometry optimization of a complex made of the dye and a small nanoparticle of anatase, 8 by using the molecular mechanics ReaxFF approach. The starting configuration was obtained through energy minimization by means of density functional theory calculations carried out at the B3LYP/6-31G* level. From the comparison of the quantum and classical optimized structures it was apparent that the MM methodology was suitable and reliable for these kinds of systems because the two geometries, namely quantum and classical structures, were almost identical (Figure 1), having a root mean square deviation (RMSD) of all the atomic coordinates of at most 0.55 Å, which was reduced to 0.07 Å when only the heavy atoms of the dye were taken into account in the superimposition process. The largest value is essentially due to a slight readjustment of the atoms of the slab and to a shift of the molecule to a farther distance from the interface. Indeed, the new distances between the titanium atoms of the interface and the oxygen atoms of the dye changed by about 0.2 Å (passing from about 2.1 to about 2.3 Å), in agreement with the earlier parametrization of this type of interactions. 2,3,9 These results can be considered very satisfactory because, on the one hand, the structure of the dye is preserved, not distorted and 2
Figure 1: Geometry optimization of a single molecule adsorbed on a small anatase (101) cluster.
appropriately represented as a considerably rigid system and, on the other hand, the whole complex has a three dimensional configuration more compatible with the anatase crystal obtained in previous studies. 7 As a consequence, it was neither necessary to develop new parameters for the dye nor to re-parametrize the interactions between the dye and the atoms of the support.
Single Dye Molecular Dynamics Simulations: Validation of the Reactive Force Field. The second check comprised the evaluation of the dynamics of the molecule when adsorbed on top of a periodic anatase (101) interface, which was more realistic than a relatively small nanoparticle. Indeed, the size and morphology of nanoparticles should be determined in agreement with experimental data, or otherwise considering the length of dye in order to explore different ways of
3
adsorption. The choice of an infinite surface is justified further by the fact that no border effects are present. Indeed, in the case reported in ref. 8 the dimensions of the substrate are too small and part of the molecule, which is very long (about 20 Å), could bend and stretch out of the edges of the anatase cluster, thus being subjected to different forces caused by the frontier atoms. These spurious perturbations could play important roles in the adsorption and dynamic of the molecule and be misleading in the comparison of the computational data with the experimental findings, especially when the samples, even though multi-faced, are larger.
Preliminary Check: Dimers on a Anatase Nanoparticle. Considering that self-interactions are important in the description of adsorption, conformational organization and distribution of the dye on the surface, before starting the production phase, it was verified if these interactions were well accounted for by the force field. Indeed, the data obtained in the case of dimers optimized at the DFT level 8 were compared with the force field-based energy minimizations. The six most stable DFT conformations found in the earlier investigation 8 were considered. The tendency of the dye to be inclined relatively to the surface was apparent in all the models. Indeed, the molecules were not aligned vertically, but they were quite bent to self-assemble effectively through stacking and T-shaped interactions of the donor portion of the adsorbates (phenyl rings). Their main contacts with the support were established through carboxyl oxygen and sulfur coordination to the titanium sites. The ReaxFF optimized structures were very similar to the initial configurations with a maximum RMSD of all heavy atoms of about 1 Å (the minimum value was about 0.65 Å). This was mainly due to the readjustment of the atoms of the cluster (maximum RMSD = 0.9 Å) which became more compactly arranged. As already observed, Ti-O(dye) distances were longer (about 0.2 Å) and a slightly reorientation of the phenyl rings took place (Figure 2), probably because, differently from the B3LYP functional, dispersion interactions are taken into account in the force field picture.
4
Figure 2: Two different views of the D102(40) 8 optimized geometry at the DFT (left hand side) and ReaxFF (right hand side) levels of theory. This is the most stable arrangement in the ReaxFF description.
5
Surface Deformation To elucidate the role played by the dye layer in influencing the surface relaxation and thus the morphology of the interface, benchmark MD simulations of the anatase bulk and clean surface (small and large supercell) models were performed and the results were compared with the functionalized substrates trajectories choosing as descriptors all possible atom-atom radial distribution functions, namely the Ti-Ti, O-O and Ti-O pairs (see plotted data in Figure 3). The results indicate that the slab has the tendency to conserve the internal crystal structure, which is very close to its bulk (light orange areas). However, as depicted by the Ti-Ti radial distribution functions, the substrates seem more crystalline in the presence of the adsorbate than in vacuum. This partial loss of crystallinity is suggested by the slight decrease in the height of the peak at about 3 Å and the parallel increase in the height of the minimum located between the first and the second peak of all the curves. However, the effect is not negligible and affects mainly the interfaces. In fact, MD simulation results evidence that the dyes and the top layer of the slab are attracted towards each other and just the portion of the surface facing the molecule becomes distorted.
6
Figure 3: Radial distribution functions of the titanium and oxygen atoms of the slab in the different simulations (see legend). The behavior of the clean and functionalized surfaces is show for both the small and large supercell models. Comparison with the bulk anatase (light orange area under the curves) is also displayed. 7
Dye Aggregates Dynamics Snapshots
Figure 4: Final snapshot of the adsorption of D102 on a small anatase (101) supercell (40.8397×15.1040 Å2 ) - low density (8 molecules) and slow equilibration. The simulation supercell has been replicated in order to visualize complete molecular structures. Hydrogens have been undisplayed for clarity.
8
Figure 5: Final snapshot of the adsorption of D102 on a small anatase (101) supercell (40.8397×15.1040 Å2 ) - high density (12 molecules). The simulation supercell has been replicated to visualize complete molecular structures. Hydrogens have been undisplayed for clarity.
References (1) Monti, S.; Corozzi, A.; Fristrup, P.; Joshi, K. L.; Shin, Y. K.; Oelschlaeger, P.; van Duin, A. C. T.; Barone, V. Exploring the Conformational and Reactive Dynamics of Biomolecules in Solution using an Extended Version of the Glycine Reactive Force Field. Phys. Chem. Chem. Phys. 2013, 15, 15062–15077. (2) Monti, S.; van Duin, A. C. T.; Kim, S.-Y.; Barone, V. Exploration of the Conformational and Reactive Dynamics of Glycine and Diglycine on TiO2 : Computational Investigations in the Gas Phase and in Solution. J. Phys. Chem. C 2012, 116, 5141–5150.
9
(3) Rahaman, O.; van Duin, A. C. T.; III, W. A. G.; Doren, D. J. Development of a ReaxFF Reactive Force Field for Glycine and Application to Solvent Effect and Tautomerization. J. Phys. Chem. B 2011, 115, 249–261. (4) Kim, S.-Y.; van Duin, A. C. T. Simulation of Titanium Metal/ Titanium Dioxide Etching with Chlorine and Hydrogen Chloride Gases Using the ReaxFF Reactive Force Field. J. Phys. Chem. A 2013, 117, 5655–5663. (5) Kim, S.-Y.; Kumar, N.; Persson, P.; Sofo, J.; van Duin, A. C. T.; Kubicki, J. D. Development of a ReaxFF reactive force field for Titanium dioxide/water systems. Langmuir 2013, 29, 7838– 7846. (6) Kim, S.-Y.; van Duin, A. C. T.; Kubicki, J. D. Simulations of the Interactions between TiO2 Nanoparticles and Water with Na+ and Cl− , Methanol and Formic Acid Using a Reactive Force Field. J. Mater. Res. 2013, 28, 513–520. (7) Raju, M.; Kim, S.-Y.; A. C. T. van Duin, K. A. F. ReaxFF Reactive Force Field Study of the ˘ S10572. Dissociation of Water on Titania Surfaces. J. Phys. Chem. C 2013, 117, 10558âA ¸ (8) Pastore, M.; Angelis, F. D. Aggregation of Organic Dyes on TiO2 in Dye-Sensitized Solar Cells Models: An Ab Initio Investigation. ACS NANO 2010, 4, 556–562. (9) Li, C.; Monti, S.; Carravetta, V. Journey Toward the Surface: How Glycine Adsorbs on Titania in Water Solution. J. Phys. Chem. C 2012, 116, 18318–18326.
10
