We present a detailed analysis concerning the inhuence of the defect density XD on the lowtemperature (T=-10 K) geminate recombination kinetics. It is shown that the lifetime distribution of a-Si:H, measured by frequency-resolved photoluminescence spectroscopy, can be explained quantitatively if it is assumed that radiative recombination is directly competing with nonradiative tunneling into defect states. This allows us to determine the radiative lifetime distribution with high accuracy, which in turn provides the unique opportunity to describe the recombination kinetics entirely, i.e. , including its nonradiative contribution. Interesting and nonintuitive consequences that are related to the competition model are discussed.
I. INTROBUCTIQNIt is well known that hydrogenated amorphous silicon (a-Si:H) gives rise to photoluminescence (PL) of high efficiency at low temperatures. PL experiments thus provide an efFective tool for studying the recombination kinetics in this material, detecting a substantial part of the excess carrier population. The lifetime distribution of the photogenerated carriers has been shown to be of particular interest, ' ' and most of the recombination models were derived on the basis of this quantity. There are two experimental techniques commonly used for lifetime measurements:time-resolved spectroscopy (TRS) and quadrature frequency-resolved spectroscopy (QFRS). While TRS records the PL decay, following a short laser pulse, QFRS measures the 90' phase-shifted frequency response under stationary excitation conditions. Though equivalent in theory, it seemed as if the two methods were producing confhcting results. Based on TRS data Tsang and Street' derived the geminate-pair model, according to which the recombining electron-hole pairs are correlated, being created in the same absorption process.The QFRS data, ' ' however, seemed to indicate that there is no such correlation, which led to the distant-pair model, where it is proposed that electrons and holes are randomly distributed in space. This problem, which arises due to the accumulating background carrier concentration in the band tails was finally resolved, and it was shown by Bort et al. that geminate recombination is also observable with QFRS, once the generation rate is decreased below the critical value Go -5 X 10' cm s There is, however, another important issue where a discrepancy between TRS and QFRS data has remained. This concerns the question of whether the nonradiative channel introduced by the defects is actually directly competing with luminescence, afFecting both quantum efficiency and radiative recombination kinetics. At first glance it appears as if this question should be easy to answer, since, in the case of direct competition, the de-crease of the PL quantum efficiency is usually expected to correspond to a proportional decrease of the observed lifetime [see Eqs. (4) and (5) in Sec. IIIB]. However, while Tsang and Street' find that the fast contribution to the PL decay becomes much more significant with increasing def...
