Chapter 1
Charge Transfer in Thermoelectric Nanocomposites: Power Factor Enhancements and Model Systems
Matt Beekman1, Sage R. Bauers2, Danielle M. Hamann3 and David C. Johnson3,*
1Physics Department, California Polytechnic State University, San Luis Obispo, California, United States
2National Renewable Energy Laboratory, Golden, Colorado, United States
3Department of Chemistry and Materials Science Institute, University of Oregon, Eugene, Oregon, United States
Abstract
Nanocomposites provide unique opportunities to capitalize on nanoscale effects not present in conventional composite materials. Recent research on thermoelectric nanocomposites has demonstrated significant performance enhancements in the power factor, related to nanoscale features as well as charge transfer effects between the constituents. The latter can be used in an approach known as modulation doping, whereby dopant atoms are spatially separated from the region through which charge carriers are transported, resulting in reduced ionized impurity scattering and increased carrier mobility. In this chapter, we review recent work that has demonstrated thermoelectric enhancements using charge transfer effects in nanocomposites, including both bulk nanocomposite materials and thin film heterostructures. Challenges in synthesis, characterization, and theoretical modeling are discussed, as well as model systems for understanding charge transfer and nanoscale effects in heterogeneous solids. The opportunities to engineer charge transfer by chemical design of the constituents in layered nanocomposites is also highlighted.
Keywords: Thermoelectric, power factor, nanocomposite, mobility, charge transfer
1.1 Introduction
The formidable challenges in finding thermoelectric materials with higher performance are apparent from the primary metric used to evaluate thermoelectric performance, the dimensionless material figure of merit ZT = (S2σ/κ)T, where S is the Seebeck coefficient, σ is the electrical conductivity, κ is the thermal conductivity, and T is the absolute temperature. Few materials have the requisite combination of large magnitude S, high σ, and low κ needed to achieve high ZT, corresponding to high coefficient of performance for thermoelectric refrigeration, or energy conversion efficiency for power generation [1]. σ = neµ, where n is the charge carrier concentration, e is the fundamental charge, and µ is the charge carrier mobility. The total thermal conductivity is the sum of lattice and electronic contributions, but is typically dominated by the lattice in most thermoelectric materials, κ = κL + κe ≈ κL. Replacing σ and κ into the expression for ZT yields ZT ≈ (S2n)(µ/κL)eT. This further reveals the contradictory interplay of material properties. In semiconducting materials, typically used for thermoelectric applications, increasing n is typically accompanied by a lowering of S, and the introduction of the types of defects that might strongly reduce κL through strong phonon scattering also reduces µ [1]. Furthermore, ionized impurity atoms introduced as dopants to optimize S2n also typically degrades µ [2].
In the nearly four decades following the pioneering work of Ioffe [3] and Goldsmid [4] in the 1950s, the prevailing approach to thermoelectric materials research was to optimize the carrier concentration for maximum power factor while making isovalent substitutions to reduce the lattice contribution to the thermal conductivity. While these approaches brought thermoelectrics into the realm of practical and useful applications in cooling and power generation, progress stagnated [5, 6]. In the 1990s, however, two new ideas revolutionized TE materials research. In the first of these ideas, Dresselhaus and coworkers predicted that nanoscale materials could have significantly higher thermoelectric efficiency than bulk materials due to lower lattice thermal conductivity and sharp features in their density of states induced by reduced dimensionality [7, 8], predictions which were supported by subsequent experiments [9]. Nanostructuring of materials, particularly using nanoscale inclusions within a bulk matrix of an already good thermoelectric material, have further resulted in increased thermoelectric performance [10–13]. Interestingly, despite experiments that have supported the basic premise of the original predictions of power factor enhancements by nanostructuring, the majority of the overall performance enhancements observed in nanoscale materials can be attributed to a significant decrease in lattice thermal conductivity rather than an increase in the power factor [14]. About the same time, Slack suggested that compounds containing a weakly bound cation or molecule in an otherwise tightly bound lattice might simultaneously behave as a crystal with respect to carrier transport, but as a glass with respect to phonons [15]. The aim was to obtain reasonable electronic mobility characteristic of a high quality, but heavily doped crystal while reducing lattice thermal conductivities to exceptionally small values characteristic of a structural glass. This “phonon glass-electron-crystal” concept prompted the discovery of several new families of high performance thermoelectric materials, for example materials with the skutterudite and clathrate crystal structures, noted for their low lattice thermal conductivities and relatively large power factors. Furthermore, Slack’s ideas formed the impetus for the current focus on using unusual features of structure and chemical bonding to design or discover new crystalline materials with very low lattice thermal conductivities [16, 17].
Since thermal conductivity values are approaching theoretical minimums [17, 18] in many thermoelectric materials, efforts to achieve even better thermoelectric performance are now increasingly focused on enhancing the power factor. While high average ZT only suggests high-energy conversion efficiency or coefficient of performance, large power factors are simultaneously needed to achieve either meaningful power generation or cooling power in most practical applications [19, 20]. Since the power factor S2σ is proportional to (S2n)(µ), the task is to find ways to achieve enhanced S and/or µ for a given carrier density. Strategies to enhance the Seebeck coefficient have included quantum confinement [7, 21, 22], energy filtering of charge carriers [23–25], and the introduction of resonance levels via selective doping [26–28...
