Nuclear spin relaxation in $n$-GaAs: From insulating to metallic regime

Nuclear spin relaxation is studied in $n$-GaAs thick layers and microcavity samples with different electron densities. We reveal that both in metallic samples where electrons are free and mobile, and in insulating samples where electrons are localized, nuclear spin relaxation is strongly enhanced at low magnetic fields. The origin of this effect could reside in the quadrupole interaction between nuclei and fluctuating electron charges, that has been proposed to govern nuclear spin dynamics at low magnetic fields in the insulating samples. The characteristic values of these magnetic fields are given by dipole-dipole interaction between nuclei in bulk samples, and are greatly enhanced in microcavities, presumably due to additional strain, inherent to microstructures and nanostructures.

Figure 1

Sketch of experimental setup used for PL (a), FR (b), and SN (c) measurements.

Figure 2

Typical examples of raw measurements using three different experimental techniques: PL (a), FR (b), and SN (c). Red lines indicate the pumping slots, blue arrows indicate when magnetic field is changed, black lines are the data.

Figure 3

Spin relaxation rate measured as a function of applied magnetic field for different samples at $T\sim 4–5$ K. Symbols are the relaxation rates extracted from the data, solid lines are Lorentzian functions corresponding to the best fit to the data. Bulk samples (a) and microstructures (b) are shown separately. Inset shows the details of the low-field behavior. (c) Half-width at half-maximum of the Lorentzians for all samples.

Figure 4

Temperature dependence. (a) Nuclear spin relaxation rate in the strong-field limit, the data (symbols) are obtained from the Lorentzian fits of the measured field dependence at each temperature. Solid lines (same color code as for the data) are temperature dependencies expected in metallic [Eq. (8)] and insulating samples [Eq. (9)]. (b) Half-width at half-maximum of the nuclear relaxation rate field dependence, extracted from the Lorentzian fits as those shown in in Fig. 3. Broadening (narrowing) of the field dependence in different samples. (c) Nuclear spin relaxation times in the strong-field limit at $T\sim 5$ K (symbols) for different samples, compared to the theoretical estimations within the diffusion-limited relaxation on isolated donors in insulating samples [Eq. (1), magenta line] and on donor pairs in metallic samples [Eq. (5), green line]. Calculation using Korringa formula [Eq. (7)] at $T=5$ K is shown by the blue line.