(oral talk) Compatibility of Al-neal in processing of Si devices with Al₂O₃ layer

Surface and bulk defects are a well-known reason for limited operation of silicon devices. These defects cause increased recombination and leakage current, and hence, their avoidance is an important factor in device processing. The traditional choice for surface passivation of silicon devices has been thermally grown silicon dioxide (SiO2). The development of device designs has led to increasing utilization of other thin films as well, including aluminum oxide (Al2O3). However, SiO2 is still often needed alongside Al2O3 in devices, such as induced junction detectors [1]. For effective elimination of defects, SiO2 typically needs to be annealed with a thin Al layer deposited on top of the thin film (so-called Al-neal process) [2]. If both SiO2 and Al2O3 are simultaneously present on the wafer, also the Al2O3 film will experience the Al-neal, which is usually performed as the last processing step in device fabrication. However, it has so far remained unresolved whether Al-nealing would have impact on the passivation performance of Al2O3, and hence, this needs to be studied.

Table 1 demonstrates the well-known result that Al-neal is highly beneficial in the case of SiO2 to reduce the interface defect density. Although Al-neal improves the carrier lifetime in SiO2-passivated Si by a factor of >20, Al2O3 does not need Al-film for superior passivation, as it is provided by regular annealing already. This is mainly due to efficient field-effect passivation induced by the high density of negative fixed charges (see Table 1). However, Figure 1 reveals that the Al-neal process required for SiO2-passivated regions is detrimental for the passivation performance of Al2O3. Especially, if the Al2O3 film has already been annealed once before Al-neal, the lifetime in the Al2O3-passivated regions is reduced by an order of magnitude. Al-nealing Al2O3 directly after its deposition without a separate post-deposition anneal provides good surface passivation but cannot reach the lifetimes achieved without Al-nealing.

The root causes for such behavior are further investigated by separately examining the substeps of Al-nealing and their impact on the passivation performance of Al2O3 film. The two details that set the Al-neal apart from a regular annealing treatment are sputtering of Al on top of Al2O3 and the presence of this Al during annealing. To study the impact of these steps individually, the above experiment is repeated following the same process with the exception that Al is etched from top of Al2O3 before the final annealing. The obtained lifetimes are slightly higher than after Al-nealing but still far from the case without any sputtering (see Table 2), which proves that sputtering damage does indeed have an impact on passivation. Furthermore, the discrepancy still exists between cases with and without separate post-deposition annealing performed prior Al-nealing. This result indicates that sputtering damage can only partly explain the degraded passivation performance during the Al-neal of Al2O3. Another affecting factor could be depletion of hydrogen from the Al2O3 film during multiple annealings. This could result in annealed film not having enough hydrogen left to re-passivate the Si/Al2O3 interface after being damaged by sputtering. The presented findings can be considered in process design to achieve higher performance in silicon devices involving both Al2O3 and SiO2. Subsequently, we have already utilized the optimized Al-neal parameters in fabrication of Si detectors resulting in lower leakage current in the devices.
References:
[1] M. A. Juntunen, J. Heinonen, V. Vähänissi, P. Repo, D. Valluru, H. Savin, Nat. Photonics 10(12), 777-781 (2016). 
[2] P. L. Castro, B. E. Deal, J. Electrochem. Soc. 118(2), 280 (1971).