Changing Doping Paradigm in Polymer Thermoelectrics

How does doping affect the thermoelectric properties of conjugated polymers? We discuss how we discovered the role of clustering of dopants on thermoelectric properties and their implications for designing the next generation of organic thermoelectric materials. The results appear in Nat. Commun.

2
0

Thermoelectric (TE) materials directly convert thermal energy (heat) to electric energy and have been the subject of intense study for over six decades. Thermoelectric devices have no moving parts and are robust enough for NASA to deploy in radio-isotope thermoelectric generators for deep-space probes [1]. However, they remain plagued by relatively low efficiency and high manufacturing cost. In the last two decades, nanostructuring pushed the needle on both of these fronts and led to several avenues of improvement, while materials informatics now point us to new, more efficient materials [2-3]. Although the focus has been on inorganic materials, polymers have recently emerged [4] as a low-cost alternative due to their scalable, environmentally-friendly solution processing.Thermoelectric (TE) materials directly convert thermal energy (heat) to electric energy and have been the subject of intense study for over six decades. Thermoelectric devices have no moving parts and are robust enough for NASA to deploy in radio-isotope thermoelectric generators for deep-space probes [1]. However, they remain plagued by relatively low efficiency and high manufacturing cost. In the last two decades, nanostructuring pushed the needle on both of these fronts and led to several avenues of improvement, while materials informatics now point us to new, more efficient materials [2-3]. Although the focus has been on inorganic materials, polymers have recently emerged [4] as a low-cost alternative due to their scalable, environmentally-friendly solution processing.

Earlier on in our collaboration [5], we noticed that there is a trade-off in every material between its thermopower (α), also called Seebeck coefficient, and electrical conductivity (σ). This trade-off is enacted by doping: more doping means higher carrier concentration, which boosts conductivity as there are more charge carriers to move a current but at the same time reduces the Seebeck coefficient, which is a measure of the average energy transported by each carrier. By filling the states available for conduction, doping brings the Fermi level EF closer to conducting states and reduces the average energy (E-Ef) they carry. At some point, the balance between conductivity and Seebeck will be optimal, maximizing the power factor α2 σ, which, in turn, controls the thermoelectric conversion efficiency via the dimensionless figure-of-merit zT=α2 σT/κ, with κ being the thermal conductivity and T the temperature. 

The amorphous structure of most polymers means a very low κ to begin with.  To further maximize the power factor and increase zT requires a fundamental change in the distribution of available states. This is known as the density of states (DOS) and is typically a fixed material property that arises from its physical and electronic structure in inorganics—doping simply fills the available states, lowest first. Our recent work [6] found that there is profound interaction between doping and the DOS in polymers. Adding dopants alters the DOS, broadening it, and the effect is particularly pronounced when dopants cluster together. The change is a consequence of long-range Coulomb interactions, poorly screened by the polymer, and can even result in a heavy tail at the end of the otherwise Gaussian DOS (see Figure). 

A direct consequence of the change in DOS is a shift downward of the entire α vs. σ trade-off curve and a dramatic change of its shape. By doping two polymers, P3HT and PDPP4T, with iodine and letting them de-dope gradually over time, we measured the entire α vs. σ curve and observed that samples doped at an elevated temperature, which aids in dispersing dopants uniformly, had a steeper curve than those doped at room temperature (see Figure). Numerical simulation allowed us to interpret the measurements and explore the direct link between the shape of the α vs. σ curve and the DOS, which was then confirmed by Kelvin probe force microscopy leading to the conclusion that how we dope is as important as how much we dope. Thus, further advances in organic thermoelectrics demand that we control not only the amount of doping but also its spatial homogeneity. 

REFERENCES:
[1] F. Ritz and C. E. Peterson, IEEE Aerospace Conference Proceedings, vol. 5, pp. 2950-2957 (2004).
[2] M. W. Gaultois et al., Chemistry of Materials, vol. 25, pp. 2911−2920 (2013).
[3] J. E. Gubernatis and T. Lookman, Physical Review Materials, vol. 2, 120301 (2018)
[4] O. Bubnova and X. Crispin, Energy and Environmental Science, vol. 5, 9345-9362 (2012)
[5] M. Upadhyaya et al., Scientific Reports, vol. 9, 5820 (2019).[Link]
[6] C. Boyle et al., Nature Communications, vol 10, 2827(2019).[Link]

This article was written by Zlatan Aksamija, D. Venkataraman, Meenakshi Upadhyaya and Michael Lu-Díaz.

Go to the profile of Dhandapani Venkataraman

Dhandapani Venkataraman

Professor, University of Massachusetts Amherst

No comments yet.