Magneto-Optical Properties of the Dominant Bound Excitons in Undoped6HSiC

Undoped single crystals of $6H$ SiC often exhibit two series of luminescence lines, sharp at low temperature, in good quality strain-free crystals. A high-energy series, in the violet close to the energy gap, contains three prominent no-phonon lines $P_0$, $R_0$, $S_0$. We report the Zeeman splittings of these lines as functions of the magnetic field strength and its orientation relative to the crystalline $c$ axis. These data indicate that the final state of the luminescence transitions contains a single electron bound to three distinct centers (donors) with lattice-site symmetry for the electron-attractive C sublattice, while the initial state contains an extra electron and a hole. The isotropic electron $g$ value is close to 2, while the hole $g$ value is $3.2 \overrightarrow{H}\parallel \overrightarrow{\mathrm{c}}$ and 0 for $\overrightarrow{H}\perp \overrightarrow{\mathrm{c}}$. We thus confirm a model proposed by Choyke and Patrick for these lines, although we can provide no additional evidence for the reasonable supposition that the donor on the three crystallographically distinct C lattice sites in $6H$ SiC is N. A lower-energy series, in the blue, contains three no-phonon lines labeled $A_0$, $B_0$, $C_0$ by Hamilton, Choyke, and Patrick (HCP). The distinctive Zeeman splittings of these lines can be understood only if an $L-S$-coupling model is assumed for an exciton bound to a center containing no additional electronic particles. This is in contrast to the $J-J$ coupling model used for the shallow $P_0$, $R_0$, $S_0$ lines, but such a difference is plausible for a semiconductor with an unusually small spin-orbit coupling like SiC. The detailed magneto-optical properties indicate that $A_0$ and particularly $B_0$ involve exciton decay at sites which do not possess the full lattice symmetry, most probably at axial (impurity-pair) defects. This finding, together with independent evidence of the properties of N donors in $6H$ SiC, particularly from electron-spin resonance, is inconsistent with the model of HCP that the $A_0$, $B_0$, $C_0$ lines involve the decay of excitons at ionized N donors. This conclusion supports a general theory for the stability of bound-exciton complexes, since the model of HCP was in strong violation of the predictions of this theory. In addition, we show that recent findings from the electronic Raman scattering of N donors in $6H$ SiC, together with a plausible upper-limit estimate of the effective-mass binding energy of donors, yield absolute values of donor binding energies $E_D$ in much better agreement with values obtained from...