4th August 2016
Solar cell dye further investigations report
Following on from the work of the schools day and previous intern I continued investigations, working with the computational chemistry department and Careers Academy at Daresbury Laboratory into how the orientations of the molecules with respect to the TiO2 slab affected their final energy values when run through the simulation on the Blue wonder supercomputer. While working with the molecules I used the input files from the schools day and the previous intern to create new input files based on key features of the molecules, for example with the ethyl group closest to the slab (shortened to EGC in the graphs). I used the visualisation software ‘Material Studio’ when working with the molecules to create the new orientations and input files for simulations on the supercomputer.
We ran two different simulations on each molecule; the first simulation we ran involved allowing the electrons in each molecule to relax to their ground state while keeping the positions of the atoms fixed, this simulation is known as a single point energy calculation. These simulations were run at two different energy convergence criteria in order to determine whether reducing the calculation time (as the second set of simulations run had a looser tolerance on the criteria, meaning they would complete faster) would affect the results.
Figure 1: This is the graph of the energies of molecule 1 at different orientations to the TiO2 surface and at different energy convergence criteria (tolerances). There are small differences between the two tolerances but no trend is shown, additional information can be found in the final energies excel document in supplementary information. For these dyes the most negative result is the best.
Figure 2: Molecule 2 has some more visible differences between the tolerances, especially between the BRC results, jumping from the second most to the most negative value
Figures 3&4: Both molecules show relatively consistent results with a few large peaks. Molecule 4 shows quite a large variation between tolerances for the more negative results with a difference of -3 eV between the results for the NGC orientation
After collecting the results from the energy calculations I took the most negative and 2 least negative orientations from each tolerance of each molecule and carried out a second simulation in which the atomic positions were allowed to relax as well as the electrons, known as a geometry optimisation, which yielded some interesting results. There were no variations between the results from the different tolerances, however there were much more pronounced changes for the orientations of molecule 4 which were submitted than the changes to molecule 1 orientations submitted, which were minimal.
Figure 5: 5A: Molecule 1 HGC orientation before geometry optimisation, 5B: Molecule 1 after geometry optimisation. Little change is visible but what this simulation has done is reduced the length of the bonds in the molecule and slab slightly. Note: green lines on the molecule show the position of the centre of the molecule
Figure 6: 6A: Molecule 2 F_SS orientation before geometry optimisation, 6B: Molecule 2 F_SS orientation after geometry optimisation. The change is similar to that of molecule 1 with one exception: the bonds to the sulphur atom in the molecule, as shown above are longer, distorting the shape of the ring it is in slightly.
Figure 7: Molecule 3 45degCW orientation 7A: before, 7B: after. This optimisation showed quite a considerable number of differences; the lower C≡N group bonded to an oxygen atom in the slab and the triple bond has become a double bond, in addition, the two cyanide groups have moved slightly farther apart and a hydrogen bond has formed between the hydrogen atom in the OH group and an oxygen atom in the slab (shown by dashed blue line).
Figure 8: Molecule 4 SRC orientations, the sulphur atom acts in the same way as with molecule 2 and the C≡N groups move farther apart
From this data it is possible to conclude that the best orientations for each molecule are hydroxide group closest for molecules 1 and 2, rotated 45 degrees anticlockwise with the oxygen atom of the squarine ring closest for molecule 3 and sulphur closest for molecule 4. Although these results are not consistent across the tolerances for the single point energy calculations, they were for the geometry optimisations. This has implications for future investigations into this area. The next stage of this research is to discuss these results with our experimental colleagues in Bangor University.
For more information, please look at the following links:
Report Author: Aedan Baker