Pore comparison from four different stacks
In this post, pores from four different structures are compared. Those four different structures are nitride/silicon/nitride (NSN), nitride/silicon/oxide (NSO) , oxide/silicon/nitride (OSN) and oxide/silicon/oxide (OSO) . The thicknesses of nitride, oxide and silicon are 30nm, 30nm and 25nm, respectively. Wafers with four different structures were etched first to remove the substrate. The four different free stand three-stacks were then annealed at different temperatures arrange from 800C to up to 1100C with 100C interval for 1 minute with a 50C/s ramp rate.
The following TEM images are four different samples annealed at 1100C. It can be clear seen that NSN structure yields the a lot of big pores in the four structures. Lots of silicon crystals as well as small pores are found in OSN sample. All samples show very high density of pores except the OSO one.
The next figure is the porosity plot of four structures annealed at different temperatures. This plot shows that the NSN sample creates the highest porosity all the time at different annealing temperature and the OSO sample is just the opposite one. The porosities from OSN sample and NSO sample are comparable and they are in between of the sample NSN and sample OSO.
The following figure is the pore size plot from four different structures. This plot shows a different situation compared with the porosity plot. The pore size of sample OSO increases rapidly with the annealing temperature and reaches to the biggest one when annealed at 1100C. Sample NSN yields the largest pore size until the annealing temperature increases to 1100C. The sample NSO shows a slight larger pore size than that of the sample OSN. The pore size plot looks different from the porosity plot and the main reason is due to the pore density, which can be seen from the pore density figure.
From the pore density plot, it can be seen that the pore density of OSO is very low compared with those of the rest three, especially when the annealing temperature is high. This tells us that in the OSO sample, pores keep growing bigger with increasing annealing temperature however only a limited amount of new pores are formed at higher annealing temperature. It is not true for NSO and OSN sample because the pore densities increase rapidly when annealing temperature increases from 800C to 1000C and then reach to a plateau. For NSN sample, it seems that the pore density reaches to stable when the annealing temperature is above 900C.
In sum, the nitride experiments show that the pore formation is promoted when the oxide protective layers are fully replaced by nitride layers. Highly dense pores are formed with the existence of nitride layers. When the protective layers are mixed by nitride and oxide layer, porosities in between of the nitride/nitride and the oxide/oxide structures are achieved and they are very comparable no matter which protective layer is on the top or the bottom.