Doping-driven electronic and lattice dynamics in the phase-change material vanadium dioxide

Doping is generally understood as a strategy for including additional positive or negative charge carriers in a semiconductor, thereby tuning the Fermi level and changing its electronic properties in the equilibrium limit. However, because dopants also couple to all of the microscopic degrees of freedom in the host, they may also alter the nonequilibrium dynamical properties of the parent material, especially at large dopant concentrations. Here, we show how substitutional doping by tungsten at the 1 at. % level modifies the complex electronic and lattice dynamics of the phase-change material vanadium dioxide. Using femtosecond broadband spectroscopy, we compare dynamics in epitaxial thin films of pristine and tungsten-doped $\mathrm{V}{\mathrm{O}}_2$ over the broadest wavelength and temporal ranges yet reported. We demonstrate that coupling of tungsten atoms to the host lattice modifies the early electron-phonon dynamics on a femtosecond timescale, altering in a counterintuitive way the ps-to-ns optical signatures of the phase transition. Density functional theory correctly captures the enthalpy difference between pristine and W-doped $\mathrm{V}{\mathrm{O}}_2$ and shows how the dopant softens critical V-V phonon modes while introducing new phononic modes due to W-V bonds. While substitutional doping provides a powerful method to control the switching threshold and contrast of phase-change materials, determining how the dopant dynamically changes the broadband optical response is equally important for optoelectronics.

Figure 1

Equilibrium electronic and structural properties of pristine and W-doped vanadium dioxide. (a), (c) White-light transmission measurements as a function of temperature and (b), (d) in situ x-ray diffraction rocking curves of the thermally induced insulator-to-metal transitions. “HC” and “CC” stand for heating and cooling cycle, respectively. The suppressed $z$ axis in the projected 3D isoline contours is the diffraction intensity. (e) Atomic-number scanning-transmission electron micrograph (z-STEM) of the W-doped $\mathrm{V}{\mathrm{O}}_2$ sample, combined with electron energy loss spectroscopy (EELS). The inset is an electron beam intensity profile acquired by Fourier analysis of mass peaks along the row of atoms (measured in $\AA$) marked by the red-colored dotted line. Evidence of substitutional doping allows for monoclinic structures of pristine and W-doped $\mathrm{V}{\mathrm{O}}_2$ to be calculated (f) using the density functional scheme described in the text. The W atom is represented in red here, with corresponding changes in the bond distances.

Figure 2

Broadband Ultrafast Response of Pristine versus W-doped $\mathrm{V}{\mathrm{O}}_2$. (a) Linear extinction spectra of $\mathrm{V}{\mathrm{O}}_2$ in the insulating (blue) and metallic (red) state (left scale). The change in optical contrast (extinction) between the two states is also plotted in black dashed lines ($\mathrm{\Delta }{\mathrm{Ext}}_{(M-I)}$), right scale). Surface plots of visible-to-IR broadband transient response of (b) pristine and (c) W-doped $\mathrm{V}{\mathrm{O}}_2$ films excited at $\sim 3.5\mathrm{mJ}/{\mathrm{cm}}^2$. The W-doped film shows completed switching behavior while the pristine ${\mathrm{VO}}_2$ shows only a transient semiconductor response. For comparison purposes, the 2D color bar have been set to the same range for both samples. (d) Fluence-dependent measurements taken at 200 ps and averaged at $\sim 600$ nm for the pristine (grey) and W-doped $\mathrm{V}{\mathrm{O}}_2$ (red). Inset plots the same data on a linear scale to show that the slope ($\left|\mathrm{\Delta }\mathrm{mOD}/\mathrm{\Delta }\mathrm{Fluence}\right|$) is smaller for the W:$\mathrm{V}{\mathrm{O}}_2$ sample. Red and black stars highlight salient features of the ultrafast dynamics that are discussed in extensively in the main text.

Figure 3

Dynamics of pristine versus W-doped $\mathrm{V}{\mathrm{O}}_2$ at their respective switching thresholds. Transient absorption (TA) dynamic traces at various wavelengths ($\lambda =480$, 580, and 680 nm) for (a) $\mathrm{V}{\mathrm{O}}_2$ at $3.5\mathrm{mJ}/\mathrm{c}{\mathrm{m}}^2$ and (b) W:$\mathrm{V}{\mathrm{O}}_2$ at $4.0\mathrm{mJ}/\mathrm{c}{\mathrm{m}}^2$ displaying the effect of changes in the coherent phonons on the optical signatures. The red star denotes the delay at which the samples have reached maximum change in optical contrast, corresponding to maximum R1 metallicities for the stated pump-laser fluence. Note the change in scale for the ordinate axis so as to display the coherent vibrations in the W:$\mathrm{V}{\mathrm{O}}_2$ sample.

Figure 4

Effect of W dopants on $\mathrm{V}{\mathrm{O}}_2$ coherent phonon dynamics. Normalized transient absorption (TA) data for only the first picosecond comparing $\mathrm{V}{\mathrm{O}}_2$ (gray) and W:$\mathrm{V}{\mathrm{O}}_2$ (red) at (a) below, (b) near, and (c) just above their respective switching thresholds. Note the systematic faster damping of the oscillation for the W-doped sample. (d) Experimental phonon spectra comparing pristine (gray) and tungsten-doped (red) $\mathrm{V}{\mathrm{O}}_2$ samples with lattice vibrations driven by below-threshold exciting pump pulse. Data were acquired by Fourier-transforming the kinetics data of Figs. 2 and 2 with spectral slices centered at 475 nm and with a 20 nm bandwidth. (e) Calculated phonon spectra at the $\mathrm{\Gamma }$ point using a finite displacement method comparing the $\mathrm{V}{\mathrm{O}}_2$ (gray) and W:$\mathrm{V}{\mathrm{O}}_2$ (red). The principal phonon vibrations for the V-V at 6 THz (for the $\mathrm{V}{\mathrm{O}}_2$ unit cell) and a representative 3.7 THz W-V mode in the 3.6–4.8 THz range (for the W:$\mathrm{V}{\mathrm{O}}_2$ unit cell) are shown in (f) and (g), respectively. The red atom represents the tungsten dopant.