Bond-strength engineering beyond mass fluctuation: unraveling the dominant mechanism of lattice thermal conductivity suppression in neighboring-site doped β-FeSi2 thermoelectrics

Abstract

Doping is the main strategy to optimize thermoelectric (TE) properties while reducing lattice thermal conductivity (κ_L), for example, in the optimization of β-FeSi₂. Experiments show that Ir doping in β-FeSi₂ reduces κ_L by 82%, in line with the point defect (PD) model, due to the large mass and size fluctuations between Ir and Fe. In contrast, Co doping still yields a 76% reduction despite negligible fluctuations, and neutron scattering experiments confirm that the size effect induced by Co doping is weaker than that induced by Ir doping. This work fully studies the effect of dopants (Co and Ir) on the phonon transport properties of β-FeSi₂ through first-principles calculations, aiming to address this abnormal κ_L reduction by the Co_Fe dopant with minor fluctuations. After considering multiple scattering mechanisms, the calculated κ_L values for Ir and Co doping are 4.40 and 5.95 W m⁻¹ K⁻¹ at 300 K, respectively, showing good agreement with the experimental values. Next, chemical bonding analysis reveals that Co doping replaces strong Fe–Si bonds with weaker Co–Si bonds. This imbalance in chemical bonding caused by the dopants induces significant softening of phonon group velocities and avoided-crossing behavior in phonon dispersion. Following this idea of bond-strength modification, the neighboring-site P-doped FeSi_1.75P_0.25 also shows a great reduction in κ_L. Our work pioneers chemical bond engineering as a universal strategy to bypass PD model limitations, offering a roadmap for designing high-efficiency TE materials through targeted bond-strength modulation.