The ultraviolet absorption, photoluminescence, and electron-spin resonance (ESR) of pure CsI which had not been exposed to ionizing radiation and which was relatively strain and defect free was studied between room and liquid-helium temperatures. Bulk single crystals were used in the photoluminescence and ESR experiments and both bulk single crystals and thin films 500-2000 Å thick were employed for taking the ultraviolet absorption data. At room temperature before cooling no emissions or ESR signals were observed for any of the samples investigated. As the temperature was lowered to that of liquid nitrogen two emissions with ${\lambda }_{max}$ at 420 and 350 nm appear for excitation in the 205-235-nm range, i.e., from the excitonic region to the long-wavelength tail of the fundamental absorption. The 350-nm emission is excited most efficiently in the excitonic region 205-225 nm, whereas the 420-nm emission is more readily produced by excitation in the long-wavelength tail of the fundamental absorption at 235 nm. This general behavior continues down to liquid-helium temperatures. At the temperature of liquid helium (4.2°K) an emission at 2900 Å appears under excitation in the excitonic region and disappears at ∼25°K. This emission has been observed by other workers and will not be dealt with here. An interesting effect has been observed when the samples are temperature cycled between room and liquid-nitrogen and lower temperatures. Emissions which were not present at room temperature before cooling appear in the 350-550-nm range depending on the excitation wavelength. It has been found that the intensities of these room-temperature-after-cooling luminescences are approximately 30-40% stronger when the samples are exposed to radiation in the 205-235-nm range when at the lower temperatures before warming back to room temperature, relative to samples which were not similarly exposed. ESR data were taken on pure single crystals of CsI and single crystals of CsI doped with thallium (0.1 mole%) and sodium (0.01 mole%). Experiments were performed on the doped samples, which are known from previous studies to produce stable hole centers at liquid-nitrogen temperatures, to compare with the pure samples which under conditions of high gain hint at the presence of an anisotropic holelike paramagnetic center. At room temperature the pure samples showed no ESR signals. Cooling slowly to about 80°K produced an ESR signal described by magnetic parameters indicating the presence of an $F$ center. The signal is characterized by a $g$ value of 2.003±0.001 with a peak-to-peak linewidth of 100 G and is isotropic in nature. The signal intensity increased as the temperature was lowered to 80°K and changed very little between 80 and 20°K, the maximum amplitude occurring between 60 and 70°K, with the signal intensity remaining isotropic. The signal persists upon warming back to room temperature which correlates quite nicely with the optical data. Effects to correlate the holelike center observed in pure CsI with hole centers observed in CsI(Tl) and CsI(Na) were only moderately successful. Using the theoretical model developed by Bassani and Inchauspe for determining the positions of the $\alpha$ and $\beta$ bands in CsI we obtain values of 235 ± 5 and 224 ± 2 nm, respectively, for these bands, which coincide with the peaks observed in the excitation spectra for the 350- and 420-nm emissions, further suggesting that the observed emissions are due to the annihilation of excitons bound to negative-ion vacancies ($\alpha$ bands) and excitons bound to $F$ centers ($\beta$ bands). These observations seem to imply that in pure CsI cooling to liquid-nitrogen and lower temperatures produces traps for both electrons and holes which persist upon warming back to room temperature and that these centers are responsible for the luminescence observed at room temperature after cooling. At low temperatures these traps are probably produced by the thermal-expansion-coefficient mismatch between the crystal (or film), the vacuum grease, and the copper block which constitute the sample holder of the experimental Dewar.

• Received 28 May 1976

DOI:https://doi.org/10.1103/PhysRevB.15.5963