RESUMEN
Cathodoluminescence (CL) spectroscopy provides a powerful way to characterize optical properties of materials with deep-subwavelength spatial resolution. While CL imaging to obtain optical spectra is a well-developed technology, imaging CL lifetimes with nanoscale resolution has only been explored in a few studies. In this paper we compare three different time-resolved CL techniques and compare their characteristics. Two configurations are based on the acquisition of CL decay traces using a pulsed electron beam that is generated either with an ultra-fast beam blanker, which is placed in the electron column, or by photoemission from a laser-driven electron cathode. The third configuration uses measurements of the autocorrelation function g(2) of the CL signal using either a continuous or a pulsed electron beam. The three techniques are compared in terms of complexity of implementation, spatial and temporal resolution, and measurement accuracy as a function of electron dose. A single sample of InGaN/GaN quantum wells is investigated to enable a direct comparison of lifetime measurement characteristics of the three techniques. The g(2)-based method provides decay measurements at the best spatial resolution, as it leaves the electron column configuration unaffected. The pulsed-beam methods provide better detail on the temporal excitation and decay dynamics. The ultra-fast blanker configuration delivers electron pulses as short as 30â¯ps at 5â¯keV and 250â¯ps at 30â¯keV. The repetition rate can be chosen arbitrarily up to 80â¯MHz and requires a conjugate plane geometry in the electron column that reduces the spatial resolution in our microscope. The photoemission configuration, pumped with 250â¯fs 257â¯nm pulses at a repetition rate from 10â¯kHz to 25â¯MHz, allows creation of electron pulses down to a few ps, with some loss in spatial resolution.