Metal-Insulator Phase Transition in thin VO2 films: A Look from the Far Infrared Side

Vanadium dioxide (VO2) displays a well-known metal-insulator (MI) transition at atemperature of 68oC. The MI transition in VO2 has been studied extensively by a widerange of optical, electrical, structural, and magnetic measurements. In spite of this there isstill some controversy about the nature of the phase transition.

Mid-IR studies1 indicatethat a model based on growth of metallic domains accounts for the phase transitionwhereas other studies2 argue that a homogeneous increase of the carrier concentration isresponsible for the change in properties across the phase transition.In this study we use Terahertz Time-Domain Spectroscopy (THz-TDS) to investigate theoptical properties of laser-ablated VO2 films in the vicinity of the MI transition in the 0.1-2 THz frequency range.

At frequencies below the phonon resonances of the material wecan obtain a clean spectroscopic signature of the carrier dynamics of the material.Radiation in the THz region interacts with an electron gas in a characteristic way, oftenwell described by the Drude model. It turns out that the capability of THz-TDS tomeasure both amplitude and phase of the transmission is crucial for the interpretation ofthe results.We observe the typical temperature hysteresis of the far-infrared transmission through thethin film with temperature.

Interestingly the temperature-dependent transmissionamplitude shows a markedly different switching temperature than the transmission phase.This effect has not been observed before, and is very important for the interpretation ofthe results.We have performed simulations of the temperature-dependent transmission amplitudeand -phase based on two different models of the phase transition, namely the domaingrowth model and a model based on homogenous increase of the carrier concentration asthe temperature is increased across the transition point.

Our data shows perfect agreementwith the domain growth model, whereas the homogeneous medium model is inconsistentwith our experimental results. 1 H. S. Choi et al, Phys. Rev. B 54, 4621 (1996) 2 A. Zylbersztejn and N. F. Mott, Phys. Rev. B 11, 4383 (1975)