To convert waste heat into electricity efficiently, a material must possess a seemingly contradictory combination of properties: it needs to conduct electricity like a perfect metallic crystal, yet block heat like an amorphous glass. This ideal is quantified by the dimensionless thermoelectric figure of merit:
$$ZT = \frac{S^2 \sigma}{\kappa} T$$
where $S$ is the Seebeck coefficient (voltage generated per unit temperature difference), $\sigma$ is electrical conductivity, and $\kappa$ is thermal conductivity.
The fundamental challenge in thermoelectric design is that these parameters are intrinsically coupled. As Han et al. note, optimizing one property typically degrades the others. For instance, increasing the charge carrier concentration to boost electrical conductivity ($\sigma$) inherently reduces the Seebeck coefficient ($S$) because the average energy per carrier decreases (Su et al.; Chen et al.). Furthermore, because electrons carry both charge and heat, a higher $\sigma$ naturally drives up the electronic component of thermal conductivity.
To bypass this paradox, researchers focus on suppressing the lattice thermal conductivity ($\kappa_L$)—the heat carried by atomic vibrations, or phonons—without disrupting the flow of electrons. Achieving this "phonon-glass electron-crystal" ideal requires mechanisms that selectively target and scatter phonons while leaving charge carriers unhindered (Zhang et al.; Johnston et al.).
Geometrical Filtering and Interface Engineering
How do you stop a phonon without stopping an electron? The answer lies in exploiting their different characteristic length scales. Phonons responsible for carrying heat often have a much longer mean free path (MFP) and different wavelengths than charge-carrying electrons. By introducing structural features at the nanoscale, we can create a geometrical filter.
[Phonon (Long MFP)] ----> || (Nanoparticle / Grain Boundary) ----> [Scattered / Blocked]
[Electron (Short MFP)] ----> || (Nanoparticle / Grain Boundary) ----> [Passes Through]
One effective strategy is embedding ultra-small secondary-phase nanoparticles within a host matrix. Lv et al. demonstrated that embedding metallic nanoparticles within a graphitized carbon matrix creates dense interfacial boundaries. These interfaces heavily scatter heat-carrying phonons but preserve continuous pathways for electrical conduction. To target a wider spectrum of phonons, Xie et al. utilized a hierarchical scattering approach, combining a layered microstructure with dispersed nanoparticles to filter out intermediate-frequency phonons while stabilizing the Seebeck coefficient.
Grain boundaries offer a similar filtering mechanism. Li et al. showed that nanocrystalline thin films with grain sizes in the tens of nanometers disrupt lattice thermal conductivity far more effectively than they impede charge carriers. At the extreme limit of this geometric confinement, surface roughness in ultra-scaled nanostructures can collapse the phonon mean free path down to mere nanometers. As Luisier demonstrates through atomistic simulations, this drastic reduction creates a severe mode mismatch between adjacent regions, causing thermal transport to collapse while electronic transport remains viable.
Harnessing Lattice Anharmonicity
Beyond physical boundaries, transport decoupling can be engineered directly into the chemical and electronic bonds of a material by exploiting lattice anharmonicity. In a perfectly harmonic crystal, atomic vibrations propagate smoothly as independent phonon waves. In an anharmonic crystal, the potential energy well of the atoms is distorted; as atoms vibrate, they interact strongly with one another, causing phonons to spontaneously scatter off each other and lose energy.
This intrinsic scattering can be driven by higher-order atomic interactions. For example, Lan et al. demonstrated that four-phonon scattering—an effect often ignored in simpler models—can suppress lattice thermal conductivity by up to 80% in certain two-dimensional monolayers.
Harmonic Lattice: (Atom) <--- smooth wave ---> (Atom) [High Heat Flow]
Anharmonic Lattice: (Atom) <### distorted ###> (Atom) [Low Heat Flow]
Researchers can also use external forces or specific chemical structures to trigger this anharmonicity. Wan et al. showed that applying tensile strain can manipulate the spatial "expression" of lone pair electrons in a crystal. This strain triggers intense, localized atomic vibrations that slash lattice thermal conductivity by over 50% without negatively impacting the mobility of valence electrons.
An even more radical form of decoupling occurs when the electronic charge centers themselves do not align with atomic positions. Gao et al. found that localizing electrons in the interstitial spaces of a crystal lattice allows these "floating" electrons to facilitate charge movement while completely decoupling electrical conductivity from bulk thermal transport.
Accelerating Discovery via Computation
Because navigating the multi-dimensional trade-offs of thermoelectric parameters is incredibly complex, traditional trial-and-error synthesis is increasingly being replaced by high-throughput computational workflows.
Machine learning regression models and first-principles calculations allow researchers to screen vast chemical spaces for materials that naturally exhibit low thermal conductivity and high carrier mobility (Han et al.; Gharavi et al.). By identifying structures with intrinsic bonding mismatches, heavy element mass differences, or complex multi-valley electronic bands, these computational models provide a direct blueprint for realizing the next generation of decoupled electro-thermal materials.
Key Foundational Work
Phonon Transport and Thermoelectricity in Silicon Nanostructures (2008) — contributed 4 concept elements to this direction.