Thermoelectric performance of Ni-Au metallic alloys determined by resonant scattering

TL;DR

Ni-Au metallic alloys exhibit exceptionally high thermoelectric power factors due to resonant scattering from Ni impurities and a distinct flat band below the Fermi level. The authors combine full-potential KKR-CPA with the Kubo-Greenwood transport formalism to show that energy-dependent resonant scattering shapes the transport function $\sigma(E)$ and Seebeck coefficient $S$, outperforming Ni-Cu constantan. They demonstrate that Ni-Au's higher lifetimes near $E_F$ and the asymmetric scattering across $E_F$ drive large $|S|$, and they show lattice-parameter variation as a viable route to further enhancement. The work provides design principles for metallic thermoelectrics and highlights the importance of scattering mechanisms in addition to the electronic DOS. The results agree with experiments and offer pathways for optimization via chemical pressure and layered structures.

Abstract

This work presents a theoretical study of the electronic structure and transport properties of Ni-Au alloys, recently identified as excellent thermoelectric metals with a power factor significantly exceeding that of conventional semiconductor thermoelectrics. Using first-principles calculations based on the Korringa-Kohn-Rostoker method combined with the coherent-potential approximation (KKR-CPA) and the Kubo-Greenwood formalism, we demonstrate the key role of resonant scattering in determining the thermoelectric properties of these alloys. This is supported by calculated densities of states, Bloch spectral functions, electrical conductivity, and thermopower. Alloying Ni with Au not only induces resonant scattering but also leads to the formation of a flat band below the Fermi level. The combination of these two features results in high thermopower, arising from a transition between resonant and weak scattering regimes near the Fermi level. Our findings are further compared with analogous calculations for constantan, a Ni-Cu alloy long regarded as a reference thermoelectric metal. We show that differences between the Ni-Au and Ni-Cu systems explain why Ni-Au exhibits nearly twice the thermopower of Ni-Cu. Finally, we simulate the effect of lattice parameter variation on the thermoelectric performance of Ni-Au and suggest that this is a promising pathway for further enhancement, for example through additional alloying or layer deposition.

Figures (17)

• Figure 1: (a) Calculated density of states (DOS) of $\mathrm{Ni_{0.1}Au_{0.9}}$ with partial density of states in panel (b) and (c); (d) evolution of total DOS of Ni$_x$Au$_{1-x}$ with increasing Ni content.
• Figure 2: Plots of Bloch spectral functions at $\Gamma$ point (with light violet line as a guide to the eye) as well as 2D projections of BSFs for (a,b) $\mathrm{Ni_{0.1}Au_{0.9}}$, (c,d) $\mathrm{Ni_{0.2}Au_{0.8}}$, (e,f) $\mathrm{Ni_{0.3}Au_{0.7}}$, (g,h) $\mathrm{Ni_{0.4}Au_{0.6}}$, (i,j) $\mathrm{Ni_{0.6}Au_{0.4}}$ and (k,l) $\mathrm{Ni_{0.8}Au_{0.2}}$. In case of 2D projections (panels b,d,f,h,j,l) color represents BSF value in logarithmic scale with black color corresponding to values over 300 atomic units.
• Figure 3: Calculated BSFs of Ni$_x$Au$_{1-x}$ in the vicinity of the Fermi level for the X high symmetry point for several Ni concentrations. We see a gradual formation of virtual band as sharpening of the spectral function. Above 40% a two-peak structure is observed as two bands at X can be distinguished in Fig. \ref{['fig:bsf_overview']} in this energy range.
• Figure 4: Transport properties of Ni$_x$Au$_{1-x}$ for $x=0.1-0.9$. (a) Energy-dependent conductivity $\sigma(E)$; (b) calculated Seebeck coefficient as a function of temperature $S(T)$; (c) thermopower at 300 K $S^{300\ \mathrm{K}}$ and conductivity at Fermi level $\sigma(E_F)$ as a function of Ni content. Panels (d-f) show comparison between experimental Garmroudi2023 and calculated temperature-dependent Seebeck coefficient for chosen concentrations. In case of the Ni$_{0.3}$Au$_{0.7}$ shown in panel (f), experimental points obtained during heating are shown, as due to phase segregation occurring at high temperatures results upon cooling did not correspond to single-phase material Garmroudi2023.
• Figure 5: (a) Comparison of DOS of Ni$_{11}$Au$_{16}$ supercell ($\mathrm{Ni_{0.407}Au_{0.593}}$) (APW+lo method) with the KKR-CPA. (b) Comparison of calculated temperature-dependent Seebeck coefficients from the supercell method with the constant relaxation time approximation (CRTA), which neglect electron scattering, with KKR-CPA and Kubo formalism, where disorder-induced electron scattering is taken into account. CRTA strongly underestimates the absolute value of thermopower. In addition, experimental results for Ni$_{0.37}$Au$_{0.63}$ and Ni$_{0.4}$Au$_{0.6}$ from Ref. Garmroudi2023 are shown and agree rather well with the KKR-CPA + Kubo calculations confirming that the resonant scattering is the dominating feature leading to a high thermopower. Experimental data are shown for measurement obtained only during heating process due to phase segregation occurring at high temperatures.