Optically pumped NMR (OPNMR) of direct gap and indirect gap semiconductors has been an area of active research interest, motivated by both basic science and technological perspectives. Proposals to enhance and to spatially localize nuclear polarization have stimulated interest in this area. Recent progress in OPNMR has focused on exploring the experimental parameter space in order to elucidate details of the underlying photophysics of optical pumping phenomena. The focus of this review is on recent studies of bulk samples of GaAs and InP, namely, the photon energy dependence, the magnetic field dependence, and the phase dependence of OPNMR resonances. Models for the development of nuclear polarization are discussed.

The intensity and the hyperfine shift of optically polarized NMR ͑OPNMR͒ signals of 71 Ga in semiinsulating GaAs have been found to depend on the photon energy and the helicity of light used for optical pumping. Single-crystal GaAs wafers of two different thicknesses, 400 and 175 m, were examined. The maximum intensity of the OPNMR signals was observed well below the band gap energy, and this maximum OPNMR signal shifted to higher photon energies for the thinner sample. In the range of photon energies for which the maximum OPNMR signals are obtained, there is little or no hyperfine shift of the 71 Ga OPNMR resonance. Hyperfine shifts with the largest magnitude are recorded for photon energies at or above the band gap. At a given photon energy, asymmetric OPNMR signals were observed, with + light producing a more intense emissive signal than the corresponding absorptive signal coming from − light. We developed a model that accounts for optical absorptivity of GaAs in order to simulate the observed OPNMR intensity, the change in the OPNMR maximum for the thinner sample, and the hyperfine shift dependence on photon energy. This model also accounts for the asymmetric OPNMR intensity for the two + and − helicities of light. The intensity dependence of the OPNMR signals arises as a consequence of two competing factors: the optical absorptivity of GaAs which directly impacts nuclear polarization ͗I Z ͘, and the number of accessible nuclear spins. The magnitude of the hyperfine shift of the OPNMR signals reflects the probability of occupation of optically relevant defects in the semiconductor, also related to optical absorptivity. Finally, the asymmetry in the OPNMR signals arises from the sign of electron spin polarization produced in the optical pumping process.

No abstract

A model of optically pumped NMR (OPNMR) behavior in GaAs that connects the photon energy dependence of the OPNMR signal intensity for (69)Ga with different polarizations of light has been developed. Inputs to this model include experimental conditions--external magnetic field (B(0)), temperature (T), and optical pumping parameters (tau(L), laser helicity)--as well as parameters that arise from sample-specific characteristics--electron spin lifetime (T(1e)), electron lifetime (tau(e)), electron-nuclear correlation time (tau(c)), and sample thickness (z). These various inputs affect the profile of the OPNMR signal intensity as a function of photon energy (E) in a predictable manner. Therefore, the profile can serve as a composite fingerprint by which individual parameters can be inferred when not known. Characteristics of the profile include the photon energy for maximum OPNMR signal intensity and the intensity ratio between sigma(+) and sigma(-) light.

On page 2 of this paper, we depicted in Fig. 2 a fit of the 69 Ga OPNMR shift offset with respect to L irradiation time, through numerical analysis of Eq. ͑3͒. We found an error in the code, and with the parameters presented in the paper, there no longer is a good match to the experimental data.We depict below the corrected plot. What is shown are two values for isotropic nuclear spin diffusion D of 2300 Å 2 sec ͑as reported in the original paper͒ and 0 Å 2 sec . The observed data fall between these two extreme values. The shape of the shift dependence on L is still maintained, suggesting that other value͑s͒ of D may be appropriate. Our conclusions for the paper must be modified slightly. In the absence of a good fit to the experimental data, the values of the electron spin polarization ͗S Z ͘ cannot be determined. These values were reported on page 3 as −0.10 for + and 0.14 for − light. In our concluding paragraph we stated: "We have presented a quantitative model of OPNMR that utilizes the hyperfine interaction, between localized electrons and surrounding nuclei, and nuclear spin diffusion." This statement is no longer correct. However, the other conclusions regarding the origin and the helicity dependence of the hyperfine shift still hold.The authors would like to thank Lee Sobotka for helpful discussions. FIG. 2. Experimental and simulated 69 Ga OPNMR shift offset. Experimental points obtained with + light ͑᭺͒, with − light ͑b͒. The lines are the simulated shift offsets using Eq. ͑4͒ with the diffusion coefficients, D, as indicated in the legend of the figure.