Fast Nuclear Spin Hyperpolarization of Phosphorus in Silicon

We experimentally demonstrate a method for obtaining nuclear spin hyperpolarization, that is, polarization significantly in excess of that expected at thermal equilibrium. By exploiting a nonequilibrium Overhauser process, driven by white light irradiation, we obtain more than 68% negative nuclear polarization of phosphorus donors in silicon. This polarization is reached with a time constant of $\sim 150\sec$, at a temperature of 1.37 K and a magnetic field of 8.5 T. The ability to obtain such large polarizations is discussed with regards to its significance for quantum information processing and magnetic resonance imaging.

Figure 1

(a) Sketch of the energy levels of the four spin eigenstates of a phosphorus donor atom in silicon in presence of very high magnetic fields. The dashed arrows indicate allowed transitions with their respective rate coefficients. ${\Gamma }_1$ is for longitudinal relaxation processes, ${\Gamma }_{\mathrm{CE}}$ for relaxation driven by capture-emission of conduction electrons and ${\Gamma }_X$ for the Overhauser flip-flop process. The two different nuclear orientations are offset horizontally. (b) Simplified sketch of the change from a thermally polarized spin ensemble to a hyperpolarized spin ensemble using $T_p>T_e$, to illustrate qualitatively the polarization process. Note that the spin relaxation processes act continuously (not sequentially as illustrated).

Figure 2

(a) ESR spectra, measured at $T=3\mathrm{K}$ with $f_{\mathrm{res}}=240\mathrm{GHz}$, with (top) and without (bottom) illumination by a mercury discharge lamp. The polarization is determined by comparing the areas of the two resonances, obtained by fitting the data with two Gaussian line shapes separated by the phosphorus hyperfine splitting, $\Delta B=4.17\mathrm{mT}$ (solid line). (b) The ${}^{31}\mathrm{P}$ nuclear polarization obtained from EPR spectra, measured as a function of illumination time, at $T=3\mathrm{K}$. The solid line is a single exponential fit to the data.

Figure 3

(a) Electrically detected magnetic resonance spectrum of Si:P at $T=1.37\mathrm{K}$ under illumination. The spectrum was fit with two Gaussian line shapes, and the nuclear polarization determined by comparing the area of the hyperfine split resonances. Here, $P=-68\pm 1\%$. (b) The polarization determined from EDMR measurements for a number of different temperatures. The line is a numerical simulation of the expected polarization using the rate model presented in this Letter. The shaded region indicates where our model does not hold. See text for details. The measurements in (a) and (b) were obtained using a xenon discharge lamp. (c) The polarization measured simultaneously with ESR and EDMR, as a function of the illumination intensity of the mercury discharge lamp used to generate the photocarriers, measured at $T=3\mathrm{K}$. The lines are guides to the eye, and are scaled by a factor of approximately 2.5 between the ESR and the EDMR.