Skutterudite thermoelectrics: What can we learn from theory?

Presented at the 2021 Midlands Computational Chemistry Meeting.

Transcript

1. Dr Jonathan Skelton Department of Chemistry, University of Manchester (jonathan.skelton@manchester.ac.uk) Skutterudite thermoelectrics: What can we learn from theory?
2. The group Dr Jonathan Skelton MCCM Jan 2021 | Slide 2 Jonathan (me) Ioanna (PhD) Ben (MChem) Jianqin (Former MSc) Jiayi (Former MChem) Matt (PhD) Sophie (MChem)
3. Thermoelectrics: motivation Dr Jonathan Skelton MCCM Jan 2021 | Slide 3 34 % 26 % 19 % 18 % 3 % 1000 MW nuclear power plant: o 650 MW waste heat o 3 % ≈ 20 MW ≈ 50,000 homes 300-500 W from exhaust gases: o 2 % lower fuel consumption o 2.4 Mt reduction in CO2 Thermoelectric generators allow waste heat to be recovered as electricity TEGs with ~3 % energy recovery ( = 1) are considered industrially viable 1. Provisional UK greenhouse gas emissions national statistics (published June 2020) 2. EPSRC Thermoelectric Network Roadmap (2018)
4. Thermoelectrics: Dr Jonathan Skelton = 2 ele + lat - Seebeck coefficient - electrical conductivity lat - lattice thermal conductivity ele - electronic thermal conductivity G. Tan et al., Chem. Rev. 116 (19), 12123 (2016) MCCM Jan 2021 | Slide 4
5. Thermoelectrics: Dr Jonathan Skelton Phonons are generated at the hot side of the material and transport energy to the cold side MCCM Jan 2021 | Slide 5
6. Phonon glass electron crystal Dr Jonathan Skelton = 2 ele + lat - Seebeck coefficient - electrical conductivity lat - lattice thermal conductivity ele - electronic thermal conductivity Phonon scattering by “rattler” filler atoms Electron transport through crystalline host framework MCCM Jan 2021 | Slide 6 G. A. Slack in CRC Handbook of Thermoelectrics (1995)
7. Filled Skutterudites Dr Jonathan Skelton Composition CoSb3 0.05 (773 K) Ni0.3 Co3.7 Sb12 0.52 (773 K) Na0.48 Co3 Sb12 1.25 (800 K) Sr0.16 Tb0.03 Co4 Sb11.82 1.32 (850 K) Ba0.08 La0.05 Yb0.04 Co4 Sb12 1.7 (850 K) Yb0.2 Ba0.1 Al0.1 Ga0.1 In0.1 La0.05 Eu0.05 Co4 Sb12 1.2 (800 K) Ce0.12 Fe0.71 Co3.29 Sb12 0.8 (750 K) D. T. Morelli et al., Phys. Rev. B 51, 9622 (1995) Y. Lei et al., J. Mater. Sci. Mater. Electron. 30, 5929 (2019) Y. Z. Pei et al., Appl. Phys. Lett. 95, 042101 (2009) S. Q. Bai et al., Appl. Phys. A 100, 1109 (2010) MCCM Jan 2021 | Slide 7 X. Shi et al., J. Am. Chem. Soc. 133, 7837 (2011) S. Zhang et al., J. Alloys Compd. 814, 152272 (2020) X. F. Tang et al., J. Mater. Sci. 36, 5435 (2001)
8. “Toy model” systems Dr Jonathan Skelton Filler [amu] [pm] He 4.0026 31 Ne 20.180 38 Ar 39.948 71 Kr 83.798 88 Xe 131.29 108 Noble gases are chemically inert (closed shell, unlikely to reduce/oxidise host framework) and are likely closest it is possible to get to a “hard sphere” filler MCCM Jan 2021 | Slide 8 J. Tang and J. M. Skelton, J. Phys.: Condens. Matter (accepted), DOI: 10.1088/1361-648X/abd8b8
9. Modelling Dr Jonathan Skelton A. Togo et al., Phys. Rev. B 91, 094306 (2015) latt () = 1 0 ෍ () ⊗ () The simplest model for latt is the relaxation time approximation (RTA) - a closed solution to the phonon Boltzmann transport equations Modal heat capacity Mode group velocity λ Average over phonon modes λ Phonon MFP Mode lifetime λ = 1 2Γλ = MCCM Jan 2021 | Slide 9
10. Pristine CoSb3 : Dr Jonathan Skelton MCCM Jan 2021 | Slide 10
11. Filled XCo8 Sb24 : Dr Jonathan Skelton MCCM Jan 2021 | Slide 11 Filler ( = 300 K) [W m-1 K-1] - 9.98 He 9.11 (-9 %) Ne 8.86 (-11 %) Ar 9.17 (-8 %) Kr 8.77 (-12 %) Xe 8.49 (-15 %)
12. How do the fillers suppress ? Dr Jonathan Skelton Consider again the RTA model for latt : latt = 1 0 ෍ ⊗ Two mechanisms through which rattlers can affect latt : 1. Reduction of - avoided crossings 2. Reduction of - resonant scattering These are not necessarily mutually exclusive - both can be active in the same material MCCM Jan 2021 | Slide 12 E. S. Toberer et al., J. Mater. Chem. 21 (40), 15843 (2011)
13. Avoided crossings Dr Jonathan Skelton MCCM Jan 2021 | Slide 13
14. Avoided crossings Dr Jonathan Skelton YbFe4 Sb12 MCCM Jan 2021 | Slide 14 Ba8 Ga16 Ge30 (inorganic clathrate) M. Christensen et al., Nat. Mater. 7, 811 (2008) W. Li and N. Mingo, Phys. Rev. B 91, 144304 (2015)
15. Resonant scattering Dr Jonathan Skelton Resonant scattering is usually defined as a linewidth (inverse lifetime) derived from a model for “one-phonon scattering due to force-constant changes” of the form: −1 = ෍ 22 2 − 2 2 + 22 MCCM Jan 2021 | Slide 15 Schwartz and Walker, Phys. Rev. B 155, 959 (1967)
16. The rattling frequency Dr Jonathan Skelton XX, = Γ = 1 X ෍ ′ X0, X′ The phonon frequencies are obtained by constructing and diagonalising the dynamical matrix () for a given phonon wavevector A good conceptual definition of “rattling” is to consider the filler atom moving inside a rigid host framework at = Γ = 0 0 0 The frequency can be obtained by diagonalising a 3 × 3 () from the “self” force constants Filler [amu] [eV Å-2] ෨ [THz] He 4.0026 1.005 5.960 Ne 20.180 2.316 4.022 Ar 39.948 6.410 4.745 Kr 83.798 8.643 3.798 Xe 131.29 12.35 3.613 MCCM Jan 2021 | Slide 16
17. The rattling frequency Dr Jonathan Skelton MCCM Jan 2021 | Slide 17
18. vs. : the CRTA model Dr Jonathan Skelton Consider again the RTA model: latt = 1 0 ෍ = 1 0 ෍ ⊗ Replace the with a constant lifetime (relaxation time) CRTA defined as follows: latt CRTA = 1 0 ෍ = 1 0 ෍ ⊗ latt ≈ 1 0 ෍ ⊗ × CRTA HA AH HA AH MCCM Jan 2021 | Slide 18
19. vs. : the CRTA model Dr Jonathan Skelton Replace the with a constant lifetime (relaxation time) CRTA defined as follows: latt ≈ 1 0 ෍ ⊗ × CRTA MCCM Jan 2021 | Slide 19
20. vs. : the CRTA model Dr Jonathan Skelton The calculated rattling frequencies ሚ X suggest that most of the fillers introduce states among the CoSb3 optic modes These account for ~20 % of the overall latt in CoSb3 , so the impact of these fillers is somewhat limited - max. reduction of 15 % in XeCo8 Sb24 MCCM Jan 2021 | Slide 20
21. A “thought experiment” Dr Jonathan Skelton We defined a rattling frequency for the noble gas fillers X based on the XX, = Γ : XX, = Γ = 1 X ෍ ′ X0, X′ What happens to latt if we artificially change the X while keeping the fixed? MCCM Jan 2021 | Slide 21
22. A “thought experiment” Dr Jonathan Skelton Lowering the ሚ X into the acoustic region reduces the λ and the group velocities → considerably larger reduction of latt MCCM Jan 2021 | Slide 22
23. Current research: molecular fillers Dr Jonathan Skelton MCCM Jan 2021 | Slide 23
24. Why molecular fillers...? Dr Jonathan Skelton A. Gold-Parker et al., PNAS 115, 11905 (2018) MCCM Jan 2021 | Slide 24
25. Why molecular fillers...? Dr Jonathan Skelton MCCM Jan 2021 | Slide 25
26. Three-phonon scattering Dr Jonathan Skelton In the RTA model, the are calculated by considering three-phonon scattering processes: λ′ λ′′ λ λ λ′ λ′′ λ λ′′ λ′ Collision/Absorption Decay/Emission MCCM Jan 2021 | Slide 26
27. Why molecular fillers...? Dr Jonathan Skelton MCCM Jan 2021 | Slide 27
28. Why molecular fillers...? Dr Jonathan Skelton MCCM Jan 2021 | Slide 28
29. Why molecular fillers...? Dr Jonathan Skelton Trans 0 Rot MCCM Jan 2021 | Slide 29
30. Summary Dr Jonathan Skelton Filled CoSb3 Skutterudites are archetypal “phonon glass electron crystal” materials Noble-gas filled XCo8 Sb24 systems (X = He-Xe) are a good “toy model” to investigate how fillers suppress the thermal transport Possible to calculate a rattling frequency ሚ X from the harmonic force constants - shows a competition between the mass of the filler atoms and how strongly they interact with the framework A CRTA decomposition of the latt shows that the main impact of the fillers is on the group velocity λ of the optic modes, leading to a maximum reduction of 15 % Further “thought experiments” show that reducing the ሚ X to the acoustic mode frequencies leads to a sharp reduction in latt by reducing the λ ሚ X appears to be a good predictor of the effect of the filler on the latt - possibly useful for screening Currently looking at whether the alternative phonon scattering mechanism in MAPbI3 could potentially work for Skutterudites with molecular fillers MCCM Jan 2021 | Slide 30
31. Acknowledgements Dr Jonathan Skelton MCCM Jan 2021 | Slide 31
32. These slides are available on Speaker Deck: http://bit.ly/3hTtpJY