Prof. Abdennaceur Karoui

Goal: Study Oxygen Precipitation driven by Nitrogen Segregation and N-O Coupling in silicon.

We investigated O precipitation in very lightly N doped CZ silicon (NCZ), as a part of the global effort for introducing 300mm (and larger) silicon wafers in electronic device and integrated circuit production lines. To that end, the microelectronics industry identified the need for lowering O concentration in large Si wafers, and increasing wafer toughness by N doping (< 10¹⁴ cm^-3). However, N changed the mechanisms of O precipitation in silicon. This was manifested by a drastic increase in precipitate concentration  in thermal processed wafers, while there was a positive reduction of the precipitate mean size.

     Defect size distribution in annealed wafers were obtained, as a function of depth, using high-resolution Oxygen Precipitate Profiler (OPP, i.e., high resolution IR scattering tomography). Annealed wafers for denuded zone (DZ) formation, unexpectedly produced defected denuded zones. In addition, an unusual high O precipitate concentrations in the near-surface region (device zone) appeared. These new phenomena were problematic for the microelectronics industry.

Fig. 1: Bevel polished specimen of Cz Si HiLoHi treated and Wright etched to delineate the distribution of O-precipitates, and the frontal denuded zone.

     We could explain the origin of these new defects by combining defect etching, SIMS depth profiling and imaging, HRTEM, and Fourier Transform Infrared Spectroscopy (FTIR) along the wafer depth. SIMS images in Fig. 2 show that O and N coexist within precipitates and in the silicon crystal (dissolved). Their depth dependent concentrations were calculated and found highly correlated.

Fig. 2: Ultra-high resolution nitrogen and oxygen SIMS images as a function of depth, from the surface down to 42um depth. [O] and [N] distributions are given respectively in the top row and bottom. The N and O SIMS images were taken simultaneously at each depth. Due to SIMS beam divergence, the dots do not give the actual shape of the precipitates.

     We have developed theories to explain the origin of these new phenomena: N segregation from the bulk to the surface, high concentration of N at the surface (even though the N concentration was extremely low), N-O coupling, the role of silicon vacancies (V), and the diffusion of the N-O-V species. The theories were backed by computational modeling of the clustering of O, N, V and interstitials species and their diffusion.

Fig. 3: Model for nitrogen pair diffusion and segregation to the surface.

     These discoveries and the developed detailed understanding allowed microelectronic industry to improve the DZ processes and to employ NCZ silicon in their IC production lines. We also found that the local N induced stresses play a great role in these phenomena. We could also explain, using these results, the N effects on O and C precipitations in solar grade multicrystalline silicon sheets grown by Astropower. The mechanisms are similar though the boundary conditions and the stress driving forces are slightly different.

     Another related discovery was based on the explained role of N in the segregation of dopant impurities; we could create a nanoscale diode under electron beam bombardment. In-situ analysis under HR-TEM showed the dopant diffusion and the formation of a nanoscale electronic p-n junction. We could write the junction by steering the electron beam motion.