Figure 1

(a) $\sigma (t)$ for $V_g^i=11\mathrm{V}$ $[n_s({10}^{11}{\mathrm{cm}}^{-2})=20.26]$, $V_g^f=7.4\mathrm{V}$ $[n_s({10}^{11}{\mathrm{cm}}^{-2})=4.74]$, and $T=3.3\mathrm{K}$. (b) Experimental protocol: $V_g(t)$ and $T(t)$.Reuse & Permissions

Figure 2

$\sigma (t)$ for $V_g^i=11\mathrm{V}$, $V_g^f=7.4\mathrm{V}$, and several $T$, as shown. For $3.2\le T(\mathrm{K})\le 4.2$, $T$ was changed in steps of 0.1 K but some traces have been omitted for clarity. Similarly, the data obtained at $0.24<T(\mathrm{K})<1$ are shown in Fig. 3b.Reuse & Permissions

Figure 3

(a) Scaling of the Fig. 2 data for $t$ just below the minimum in $\sigma (t)$. The data obey a stretched exponential dependence on $t$; the dotted line is a fit. The prefactor $a(T)\propto ({\tau }_{\mathrm{low}}{)}^{-\alpha }$, $\alpha \approx 0.07$. (b) At short $t$, relaxations follow a power law $\sigma (t,T)/{\sigma }_0(T)\propto t^{-\alpha }$ [17]. The dashed lines are fits with slopes $\alpha =0.07\pm 0.01$. (c) While the lowest $T$ data clearly deviate from the stretched exponential dependence (dashed line), all the data are consistent with the Ogielski relaxation $\sigma (t,T)/{\sigma }_0(T)\propto ({\tau }_{\mathrm{low}}{)}^{-\alpha }(t/{\tau }_{\mathrm{low}}{)}^{-\alpha }\exp [-(t/{\tau }_{\mathrm{low}}{)}^{\beta }]$ (dotted line). The data have been collapsed with respect to the 2.4 K curve. (d) Scaling parameter ${\tau }_{\mathrm{low}}$ as a function of $T$. The dashed line is a fit with a slope equal to an activation energy $E_a\approx 19\mathrm{K}$; ${\tau }_{\mathrm{low}}(2.4\mathrm{K})\sim 500\mathrm{s}$.Reuse & Permissions

Figure 4

(a) The exponents $\alpha$ and $\beta$ vs $n_s$. Dashed lines guide the eye. (b) $\log {\tau }_{\mathrm{low}}$ plotted vs $n_s^{1/2}$. The values of ${\tau }_{\mathrm{low}}$ correspond to $T=3\mathrm{K}$. The dashed line is a fit with a slope $(\log e)\gamma \approx 4.5({10}^{-11}{\mathrm{cm}}^2{)}^{1/2}$.Reuse & Permissions

Figure 5

(a) Scaling of the Fig. 2 data at long $t$, i.e., above the minimum in $\sigma (t)$. The data have been collapsed with respect to the 5 K curve. (b) The scaling function, which describes the approach to equilibrium value ${\sigma }_0$ at long times. The dashed lines are fits with slopes $k_1\approx 2.89\times {10}^{-2}$ and $k_2\approx 4.37\times {10}^{-3}$ for the faster and slower exponential processes, respectively. (The corresponding relaxation times are ${\tau }_{\mathrm{high}}/k_1$ and ${\tau }_{\mathrm{high}}/k_2$.) Inset: $T$ dependence of ${\tau }_{\mathrm{high}}/k_1$ for different $V_g^i$ and $V_g^f$ as shown on the plot. The dashed line is a fit with the slope equal to an activation energy $E_A\approx 57\mathrm{K}$.Reuse & Permissions