Critical Slowing Down of the Charge Carrier Dynamics at the Mott Metal-Insulator Transition

We report on the dramatic slowing down of the charge carrier dynamics in a quasi-two-dimensional organic conductor, which can be reversibly tuned through the Mott metal-insulator transition (MIT). At the finite-temperature critical end point, we observe a divergent increase of the resistance fluctuations accompanied by a drastic shift of spectral weight to low frequencies, demonstrating the critical slowing down of the order parameter (doublon density) fluctuations. The slow dynamics is accompanied by non-Gaussian fluctuations, indicative of correlated charge carrier dynamics. A possible explanation is a glassy freezing of the electronic system as a precursor of the Mott MIT.

Figure 1

(a) Resistance measurements of $\kappa \text{−}{\mathrm{D}}_8/{\mathrm{H}}_8\text{−}\mathrm{Br}$ for various cooling rates $q$. (b) Arrow indicates the position of the slowly cooled, pristine sample in the generic phase diagram. Open circles indicate values taken from the literature of the second-order critical end point of the first-order Mott MIT (thick solid curve), see [17] and references therein. The change of the sample position with $q$ is schematically indicated by the colored lines. Blue represents fully deuterated $\kappa \text{−}{\mathrm{D}}_8\text{−}\mathrm{Br}$ [18].

Figure 2

(a)–(c) Resistance noise PSD taken at $f=1\mathrm{Hz}$ of $\kappa \text{−}{\mathrm{D}}_8/{\mathrm{H}}_8\text{−}\mathrm{Br}$ compared to $\kappa \text{−}{\mathrm{D}}_8\text{−}\mathrm{Br}$ (slowly cooled, same data as in [18]) normalized to the value of the local maximum around 100 K for increasing cooling rates (a) $q=0.5$, (b) 2, (c) 5 and $10\mathrm{K}/\min$. Note the rescaling of the ordinate in (c). (d) Frequency exponent $\alpha$ vs $T$ of the $1/f^{\alpha }$-type noise spectra for $q=5\mathrm{K}/\min$. Line is a description with a model of nonexponetial kinetics (DDH model) as described in the text. (e) Relative noise level $a_R\equiv f\times S_R/R^2$ vs $f$ vs $T$ for the same cooling rate.

Figure 3

Second spectra $S^{(2)}(f_2,f)$ vs $f_2/f$ of $\kappa \text{−}{\mathrm{D}}_8/{\mathrm{H}}_8\text{−}\mathrm{Br}$ for $q=10\mathrm{K}/\min$ at three different temperatures. Horizontal lines are guides for the eyes. At $T=34\mathrm{K}$, where the first noise spectra $S_R/R^2(f)$ peaks at $f=1\mathrm{Hz}$, see Fig. 2, the data at low frequencies can be fitted to $S^{(2)}\propto 1/f_2^{\beta }$ yielding $\beta =1.35$ (black line). At all temperatures the spectra measured at different frequency octaves are scale invariant, i.e., do not depend on the scale of $f$.