Under the UV irradiation, a small but repeatedly steady decrease of the EPR signal is observed in comparison with the sample in dark. The decrease can be clearly seen at the peak position as zoomed in Fig. 6 (curves a and b). The difference of subtracting curve a from curve b is shown in curve c with 30 magnification. Curve d shows that difference becomes smaller at the second cycle after 100 s recovering in dark from first cycle. The decreases of Mn2þ EPR signal
and the photoluminescence in the trapping state after UV excitation suggest a population decrease of the Mn2þ ions that may be ionized to Mn3þ (non-Kramer’s system). Once the valence of Mn2þ ions is changed in the current host lattice, they lose the role as the luminescent centers, leading to the decrease of photoluminescence that is consistent with the results in Fig. 5b. In addition, the ionization may just occur for the Mn2þ ions in the surface, which has been observed for Mn2þ in other hosts [16], depleting the limited population of Mn2þ at surface even upon weak excitation. The initial maximum emission peaks gradually decrease after each on–off cycle since the recovering process of Mn2þ from Mn3þ is slow, as illustrated in curve d (Fig. 6), where just part of Mn3þ ions returns to Mn2þ after 100 s in dark. The Mn3þ ions may also serve as hole traps (Mn3þ ¼holeþMn2þ), yielding phosphorescence. The recovering time is in the same scale of phosphorescence in Mn2þ doped samples. This also explains the results for undoped sample shown in Fig. 5a, where the initial peaks are absent or no ionization occurs when UV light is on. Finally, the EPR signal maintains the same if irradiating the doped sample with UV at longer wavelengths (4300 nm), which is consistent with the observation of photoluminescence, i.e., no initial peaks when UV is on for all the samples, as discussed in previous section.