Comparison of annealed I-E and SiN x passivated samples

A comparison in lifetime (with iron in the FeB state) and bulk interstitial iron concentration for the silicon nitride passivated and iodine-ethanol passivated samples annealed at all three temperatures is presented in Figure 6.5 and Figure 6.6, respectively. The samples from the both the passivations were subjected to the same thermal treatment for the same cumulative period. The lifetimes and iron concentration values are normalised to the as-received values for both SiNxand I-E passivated samples so that the

data can be compared directly.

6.4.4.1

Bottomand top wafers

In the bottom and top samples with both the SiNx and I-E passivation, the behaviour

is quantitatively similar. Bottom samples have a relatively low starting lifetime and relatively high starting interstitial iron concentration experience lifetime improvements and longer-term interstitial iron reductions with both passivation types. The kinetics of the decay in interstitial iron concentration at a given temperature are very similar in samples with both the passivations (Figure 6.5) which suggest the main sink for interstitial iron is the same for both passivation types. This is perhaps surprising that the samples for SiNxand I-E passivation for a given annealing temperature were not sister, but adjacent,

samples so have similar properties by virtue of their ingot location but not near-identical microstructures. Lifetime improvement in the bottom samples annealed at 300 °C and 400 °C is higher with I-E passivated samples than the SiNxpassivated samples. This result

indicates that defects other than iron are involved in silicon nitride passivated samples. In the top wafers, lifetime is improved and interstitial iron concentration decayed in samples with both the passivations at 300 °C and 400 °C. The decay rate in iron concentration is slightly higher in SiNxpassivated top samples than the I-E passivated samples. Annealing

at 500 °C exhibits a lot of scatter in the data in top samples with both the passivations. The spatial lifetime distribution images for the bottom and top samples for both the passivations are presented in Figures 6.7(a) and 6.7(b) and 6.8 (a) and 6.8(b), respectively. The corresponding interstitial iron concentration maps are presented in Figures 6.9(a) and 6.9(b) and 6.10(a) and 6.10(b), respectively. Note that a different scale is maintained for SiNxand I-E passivated samples as they have different as-received values. It can be seen

both passivation types, the lifetimes increase with annealing and the interstitial iron concentrations decrease. The changes which occur are slow and much slower than if diffusion-limited internal gettering of interstitial iron is the only process involved. The gettering process requiring the longest impurity diffusion length would be gettering to the SiNxfilm (or surface in the case of I-E passivation), yet if known diffusion coefficients

of interstitial iron [18] are used with a two-sided surface diffusion calculation [128] the decay in interstitial iron concentration in the bottom samples would be faster than observed experimentally by a factor of ~ 6 at 300 °C, ~ 15 at 400 °C and ~ 20 at 500 °C, as shown in Figure 6.17. At 500 °C after an initial increase, the decay is slower than at 400 °C, which is indicative of multiple mechanisms of iron gettering occurring.

Figure 6.17: Comparison of measured bulk interstitial iron concentration as a function of cumulative annealing time and the kinetics using a double-sided diffusion model for the bottom samples according to Equation 6.3. Samples were passivated with silicon nitride passivation and annealed at 300 °C, 400 °C and 500 °C temperatures.

There are good reasons not to abandon the surface or SiNxfilm gettering mechanism

in bottom samples entirely, however. Factors in favour include the SIMS data for SiNx

samples, the same interstitial iron decay in non-sister samples, and previous studies which have found iron gettering to free surfaces in single crystal silicon [211, 212]. Trapping of interstitial iron at crystal defects in mc-Si could slow down the effective diffusion process of interstitial iron, and iron release from elsewhere in the material at 500 °C could complicate the kinetics further.

6.4.4.2

Middle wafers

The effect of low-temperature annealing in samples from the middle part of the ingot

(MB and MT) show a significant difference between SiNx and I-E passivated samples.

Annealing SiNxpassivated samples at 300 °C or 400 °C has a positive effect on lifetime

and decay in interstitial iron concentration, whereas in I-E case the effect is more mixed and can be highly detrimental.

The time dependence of bulk lifetime and interstitial iron concentration for samples from the middle of the ingot is strongly passivation-dependent as shown in Figure 6.5 and Figure 6.6, respectively. The abrupt reduction in lifetime observed in samples with I-E passivation upon annealing at 400 °C and 500 °C [53], was rarely observed in samples with SiNxpassivation. For example, the lifetime in bottom middle sample reduced from

46.5µs to just 7.6µs after 6 h of annealing at 400 °C with the I-E passivation, whereas the same treatment with SiNxpassivation increases lifetime from 112µs to 163 µs. The

key difference is in bulk interstitial iron concentration decreases substantially with SiNx

passivated samples whereas it increases with I-E upon annealing at 300 °C and 400 °C. It is therefore concluded that passivation strongly affects theinternalgettering behaviour. The spatially resolved lifetime images for the bottom middle and top middle samples annealed at 400 °C with both the SiNx and I-E passivation are shown in Figure 6.7(c),

6.7(d), 6.8(c) and 6.8(d), respectively. The corresponding interstitial iron concentration maps are shown in Figures 6.9(c) and 6.9(d) and 6.10(c) and 6.10(d), respectively. A decrease in lifetime in I-E passivated samples is clearly visible with I-E but not with SiNx

passivation. Although lifetime recovers with I-E passivation upon further annealing but it is lower in most of the bulk grain regions compared to the as-received values. Furthermore, the lifetime reduction in the sample with I-E passivation is accompanied by

an increase in interstitial iron concentration but this increase is far too small to account for the dramatic lifetime reduction. The samples with SiNxpassivation lifetime improves

in both the MB and MT samples at an intermediate stage and remains constant in further

annealing. Why does lifetime reduce substantially with I-E but not with SiNx? One

possibility is that bulk passivation from the SiNxfilm somehow prevents the formation of

the recombination centres which form with I-E passivation. Other possibilities include that fast diffusing impurities are released from locations within the bulk of the material upon initial annealing and are gettered by the SiNxfilm but not by a free surface, or that

contamination from outside the sample can enter in the I-E case but not in the SiNxcase

as the film might act as a diffusion barrier. The complexity of low-temperature gettering is discussed in Section 6.4.6. It is noted that grain structure and densities are slightly

different in the I-E and SiNx passivated samples which might cause a difference in

interstitial iron kinetics. The difference in lifetime improvement and interstitial iron

concentration reduction upon annealing between SiNxand I-E passivated samples could

be a result of bulk hydrogenation as recent studies by Liuet al. [138], Karzelet al. [137] and Leonardet al. [206] provided evidence for hydrogen interacting with bulk iron. In this study, there is no direct evidence of hydrogen present in the bulk material but SiNx

film has an effect on iron concentration which discussed in details in the following section.