2012-11 Boston MRS - Computational design of materials for organic photovoltaics

Boston Fall MRS, contributed talk, mainly Tight Binding in Organics with a Python code: https://github.com/jarvist/LongSnakeMoan.

Transcript

1. Computational Design of Materials for Organic Photovoltaics Jarvist Moore Frost, Sam Foster, James Kirkpatrick, Thomas Kirchartz, Jenny Nelson Imperial College London jarvist.frost@imperial.ac.uk MRS Boston Fall 2012 O8.08 – Wednesday 28th November
2. Designing New Copolymer Donors Generate Structures via SMILES (now easy!) → HOMO 'for free' (well defined in KS DFT) → LUMO Needs care (don't use KS virtual orb.) → Absorption Spectra TD-DFT (comp. Intensive) → Estimate Mobility: (Need packing structure) Inner Sphere Reorg. Energy (easy) Outer Sphere Reorg. Energy (???) Transfer Integrals (researcher time intensive) Kinetic Monte Carlo (ToFeT, other codes) Create model of Density of States: → continuum device modelling, transport behaviour
3. Designing New Copolymer Donors Generate Structures via SMILES (now easy!) → HOMO 'for free' (well defined in KS DFT) → LUMO Needs care (don't use KS virtual orb.) → Absorption Spectra TD-DFT (comp. Intensive) → Estimate Mobility: (Need packing structure) Inner Sphere Reorg. Energy (easy) Outer Sphere Reorg. Energy (???) Transfer Integrals (researcher time intensive) Kinetic Monte Carlo (ToFeT, other codes) Create model of Density of States: → continuum device modelling, transport behaviour
4. Density of States • • Difficult to extract experimentally • • Would be nice to calculate ab-initio • • Make predictions of charge transport behaviour / device operation •
5. Tight Binding - The Hückel Method [[ S J 0. 0. ] [ J S J 0. ] [ 0. J S J ] [ 0. 0. J S ]] H = Hamiltonian very easy to solve Parameters semi-empirical (can derive from DFT) Can treat extremely large systems
6. Tight Binding on 200 P3HT unit chain [no disorder] - predicts optical gap and structure in DoS but not exponential tail
7. Molecular Modeling of Crystalline Alkylthiophene Oligomers and Polymers Margherita Moreno, Mose Casalegno, Guido Raos, Stefano V. Meille, and Riccardo Po ̀ The Journal of Physical Chemistry B 2010 114 (4), 1591-1602 40 independent copies of a 16-mer P3HT Packed Loosely in a (10nm)^3 box Equilibrated for 1ns under 1.atm with a thermostat
8. Density 1.108 → g/cm^3 9ns production run Representative snapshots produced at 200,300,400,500 K Frames extracted from every 20ps
9. Molecular Orbital Overlap calculated to give transfer integral between oligomers used as input for Tight Binding calculation MD QC / MOO Tight Binding → → Simulated XRD (Inter Monomer) RDF
10. Interchain contributions only: Urbach tail ~3.5meV
11. T
12. Variation of the intrachain transfer integral due to overlap of Pi-like orbitals, which results in a cosine variation of J: We calculate J0 with a projective method (B3LYP/6-31g*) J = 0.8 eV
13. Interchain + Intrachain contributions: Urbach tail ~38meV
14. P3HT DoS Conclusions Experimentally: E ~ 35 meV (subgap quantum efficiency, transient photocurrent at short times) E ~ 60 meV (further inside the bandgap, at longer time scales, which would explain charge extraction measurements and transient photocurrent at longer timescales) Modelling directly produces a realistic DoS Suggests that inter-monomer torsional disorder dominates density of states ! Note that the extended regions of conjugation are traps !
15. (Boon Kar, 2008) High Mobility Polyfluorine & the -phase β
17. Transition Time for Backbone Movements (extracted from MD oligomer simulation, PF8)
19. MD Representation Stiff QM Representation → (via Votca) Polyfluorene Tetramer Simulation 250 tetramers – 3.5ns
20. Transfer Integrals: PF8 vs. PF5:8 F8-F8-F8-F8 F8-F5-F8-F5
21. Predicted Mobility in PF8 vs. PF5:8 cm2 / Vs Simulated Hole Mobilities Simulated Tetramer Densities
22. Processing magic di-octyl Polyfluorene can be persuaded to have high mobilities → Samples are glassy Induction of the Beta-phase strongly reduces mobility
23. β-Phase α-Phase HOMO Spin Density B3LYP/6-31g* "Formation of the -phase effectively β corresponds to crystallization in one dimension, a remarkably uncommon phenomenon in nature." http://pubs.acs.org/doi/pdfplus/10.1021/nl071207u
24. 0 1 2 3 4 5 6 7 eV In Vacuum, Beta Phase structures are strained But at accessible energies
25. Need larger systems for representative DoS → Python TightBinding http://www.physics.rutgers.edu/pythtb/ Hybrid DFT on Octamers (B3LYP/6-31g* & tuned BNL/6-31g*) - get some trap formation (up to 140 meV)
27. 50meV Gaussian Site Energy Disorder
28. Beta Phase Energetic Trap (5 sites)
29. Beta Phase Energetic Trap (10 sites) For Wide Enough trap ~10 sites (no confinement), Trap depth = site Energy change
30. P3HT–like polymer (normally flat) Approximate torsional distribution as normal distribution Consider beta phase to be small pockets of flat regions of polymer – (i.e. J is maximised is these regions to J0)
31. Disordered (torsionally) But FLAT (minima) polymer 70meV
32. Polyfluorene–like polymer (Helix / twisted rod) Approximate torsional distribution as normal distribution (around 45 degree minima) Consider beta phase to be small pockets of flat regions of polymer – (i.e. J is maximised is these regions to J0)
33. Disordered (torsionally) TWISTED (~PFO) polymer 596meV
34. Conclusions • In polymers, Torsion is important – In P3HT, torsion sufficient to describe DoS – In PFO, natural background twist makes extended regions (Beta phase) deep traps – Traps are not always what one imagines (and be careful what you wish for) Even (or especially?) 1930s electronic-structure techniques can offer useful insights
35. Computational Design of Materials for Organic Photovoltaics Jarvist Moore Frost, Sam Foster, James Kirkpatrick, Thomas Kirchartz, Jenny Nelson Imperial College London jarvist.frost@imperial.ac.uk Acknowledgements: Sumitomo, Donal Bradley, Boon Kar Yap, EPSRC (Scallops Project)
36. Designing New Copolymer Donors Generate Structures via SMILES (now easy!) → HOMO 'for free' (well defined in KS DFT) → LUMO Needs care (don't use KS virtual orb.) → Absorption Spectra TD-DFT (comp. Intensive) → Estimate Mobility: (Need packing structure) Inner Sphere Reorg. Energy (easy) Outer Sphere Reorg. Energy (???) Transfer Integrals (researcher time intensive) Kinetic Monte Carlo (ToFeT, other codes)
37. Which materials? BT PFO (+Si) T iPFO (+Si) TT TPT (+Si) T-TT-T DBT TT-T-TT TTB DPP DPP Carbazole + Se / Ge substitions Fluorination ⊗ ~800+ copolymers
38. P1, Bronstein et al. Experimental Calculated (Dimer) HOMO -5.05 eV -4.90 eV Bg 1.37 eV 1.70 eV LUMO -3.68 eV -3.20 (via Bg) eV DPP – T – TT – T DFT/TD-DFT B3LYP/6-31g*
39. HOMO LUMO (via TD-DFT 1 → st Singlet & NTOs) Totally distributed... Push-Pull ? Beautifully flat structure
42. Triplet vs. Donor Moiety eV If present, DPP dominates & pins Triplet
43. HOMO vs. Donor Moiety Again, DPP pins...
44. LUMO vs. Donor Moiety Low (ish) LUMO as a result of pinned HOMO + low
46. AutoJ Generate non-overlapping (vdW radii) pairs [PACKMOL] Calculate Transfer Integral (and/or inner sphere reorganisation energy) → MOO (ms) or Projective DFT (hrs) ← Repeat! Much less researcher time than MD
47. via PACKMOL