Low temperature micro-photoluminescence spectroscopy on laser doped silicon with different surface condition

Full text: Although there have been many studies on laser doping techniques and their applications, the fundamental impacts of the laser process and the resultant characteristics of the processed region are still not completely characterised and understood. In this paper we utilise low temperature micro-photoluminescence spectroscopy (µ-PLS) to investigate shallow layers of laser doped silicon, exploring the impacts of laser doping processes on electronic quality, and thus providing insights into the impacts on the underlying silicon crystal structure. Micro-scale measurements (with resolution down to 3 µm) enable us to investigate effects of laser processing at specific positions such as boundary/edge of processed and unprocessed regions. In addition, low temperature PLS enables us to identify signals corresponding to laser induced defects or laser induced doping of silicon, which otherwise cannot be observed at room temperature. Samples were prepared with two different surface conditions, chemical mechanical polishing (CMP) and Tetramethyl Ammonium Hydroxide (TMAH) etching, to observe any effect of surface condition on laser doping process. Commercially available, boron-rich spin-on dopant was deposited on high-resistivity n-type float-zone crystalline silicon substrate as a dopant precursor. We employ a 248 nm nanosecond excimer laser, whose beam is homogeneous and rectangular, to melt a shallow layer, thus introducing dopants into the melt subsequent to recrystallisation. A variety of single or multiple repeat pulses is employed, using a range of fluences suitable for light to heavy doping. Neighbouring pulses are arranged such that clear 'overlap' boundaries are present as well as edges between doped and undoped regions. The resultant PLS signals, excited with 532nm laser at 79K, for three repeat pulses with fluence of about 4J/cm², measured at different positions on different wafer surface types. It was observed that for the undoped surfaces, indistinguishable PL signals are observed from CMP and TMAH samples. However, significantly different PL signals are observed after laser irradiation. Evidence of heavy doping on both samples is indicated by the local PLS peak at around 1160nm due to band gap narrowing effect of heavy doping, though with different peak heights observed due to different depth of doping. Longer wavelength PLS peaks correspond to the signature emissions from dislocation defects, with the well-known D3 dislocation peak observed around 1280 nm. The defect PLS signal on the CMP sample is barely observed at the middle of the doped region, but appears as a 'tail' for the measurements at the boundary regions. On the other hand, the TMAH sample gives dominantly high defects signal in the range of 1200~1500nm, albeit more pronounced again for boundary measurements. Moreover, silicon band-band peak (~1130nm) is buried under the defect signal at the edge due to significantly deep defects. These observations suggest that dislocation defects can be strongly associated with the recrystallization interface at the boundaries of laser-induced melt zones. By utilizing μPLS, we observed fundamental characteristic of laser doped regions, in particular the presence of both doping and dislocation defects. Notably, different PLS signals at the middle and boundaries of laser doped regions highlights the significant role that the edges of laser doped pulses play in terms of electronic quality and recombination properties. Moreover, different surface conditions or substrate properties appear to affect significantly the impact of laser doping processes on crystal quality and presence of dislocation defects, with a considerably greater level of induced defects observed on initially rougher surfaces than on polished surfaces. (author)