Comparison of RTP and CFP P-Al co-diffused solar cells

Experimental

Solar cells were processed from RGS Si ribbons that were rapidly cooled after crystallization.

The solar cell process flow is shown in Fig. 4.16. The foils 2, 3 and 9 of the crystallization runs 6520 and 6521 were used. First, the foils´ surface was levelled by polishing, setting the cell thickness to approximately 300 m. Then each foil was cut into four 2.5 2.5 cm² samples which were damage-etched afterwards. After a modified RCA clean, two samples of each foil were subjected to a heavy CFP P diffusion for 2 h at 920 C using POCl3. During this pre-diffusion step the initially rapid RGS is assumed to be transformed into the annealed state as described for approach I. This means, large O-precipitates have formed at the expenses of [Oi].

The PSG and the diffused n layers were removed afterwards. Together with the remaining two non pre-diffused samples per foil, they were then coated with a 1 m thin Al film on the

back. P-Al co-diffusion for simultaneous emitter and BSF formation was performed either by CFP for 1 h at 820 C or by RTP for 30 s at 920 C., both resulting in an emitter sheet resistance of approximately 80 /sq. The CFP diffusion utilized POCl3 as a P source whereas prior to the RTP diffusion the phosphorus spin-on dopant P509 was deposited on the front side. The profile of the RTD emitter is essentially the same as the one shown in Fig. 3.5 and the profile of the CFP emitter is shown in the right graph of Fig. 3.8. Eventually, the four samples stemming from one foil were subjected to each of the four thermal sequences A to D shown in Fig. 4.16. Please note that 820 C lies well in the temperature range of the formation of New Donors whereas the 920 C is above. In the case of RTP, the heating and cooling rates were 100 K/s. This ensures a quick ramping through the critical temperature range and hence the suppression of new donor formation. During the CFP pre-diffusion the formation of large oxygen precipitates from interstitial oxygen is expected so that the subsequent ramping down should not be critical any more. After P-glass removal in HF the front contact was prepared by photolithography and evaporation of Ti/Pd/Ag with subsequent Ag plating. The edges were isolated by laser cutting and cleaving. A 30 min forming gas anneal step at 350 C was followed by a standard remote plasma hydrogen passivation step (RPHP) at 350 C for 45 min⁴. Finally, a double layer TiO2/MgF2 antireflection coating was deposited.

Solar cell results

The distribution of the single solar cell IV parameters under illumination is shown in Fig. 4.17.

Table 4.7 contains the average values and the respective standard deviations. A look at the efficiencies of the non pre-diffused solar cells reveals that the all-RTP processed solar cells (Seq. D) exhibit the best performance. Their mean efficiency is 8.6 % with a remarkable top value of 9 %. In contrast, the non pre-diffused CFP solar cells (Seq. C) show the poorest efficiencies of all groups. The average efficiency is only 5.2 % and the maximum value is 6.1 %. For comparison, after CFP processing of many solar cells from different RGS growth runs, Hahn [51] reported best efficiencies of 6.0 before and 8.8 % after an optimized hydrogen passivation step, respectively. Of course, the efficiencies they obtained are superior to our CFP results which we attribute to their optimized hydrogen passivation because the remaining process steps were similar. However, they are still lower than our RTP results. We think that this shows the superiority of RTP to CFP in the case of rapidly cooled RGS. In the case of the pre-diffused solar cells we have obtained mean efficiencies of 7.5 after RTP and 7.9 % after CFP P-Al co-diffusion, respectively.

The differently processed solar cells show characteristic differences in their IV parameters

, and" ". The CFP P pre-diffused solar cells exhibit by far the highest average

of about 25 mA/cm^-2 regardless whether the subsequent P-Al co-diffusion was carried out by CFP (Seq. A) or RTP (Seq. B). As will be shown, this is due to the formation of conducting

4Hahn [51] observed an increase in of solar cells from rapid RGS even after 2 h of remote hydrogen passivation at 350 C. Full passivation was observed only after 8 h of passivation, which is attributed to the high [Oi]. Retrospectively, we can say that the 45 min of H passivation applied in this work were too short to describe the cells´ state as after hydrogen passivation.

2 2 2 2 3 3 3 3 9 9 9 9 2 2 2 2 3 3 3 3 9 9 9 9

Fig. 4.17: Overview of the IV characteristics of solar cells fabricated from the rapidly cooled RGS foils 2, 3 and 9 of the growth runs 6520 and 6521, respectively. The cells were processed according to the thermal sequences A to D presented in Fig. 4.16.

Tab. 4.7:Average IV parameters of the solar cells fabricated from rapidly cooled RGS foils.

Process sequence " " (

Seq. P pre-diffused P-Al co-diffusion [mV] [mAcm^-2] [%] [%]

A yes CFP 500 5 25.1 3.0 62.6 2.9 7.9 0.7

B yes RTP 487 9 24.8 4.5 61.8 5.8 7.5 0.8

C no CFP 481 11 16.1 0.9 67.6 4.6 5.2 0.7

D no RTP 525 7 22.0 0.8 74.5 1.4 8.6 0.4

(inversion) channels during the pre-diffusion step which enhances the carrier collection in deep the bulk and hence increases the apparent carrier diffusion length. It does not seem to make a significant difference whether CFP or RTP is carried out afterwards. It has to be mentioned, that for the solar cells made from foil number 9 the pre-diffused solar cells are not superior to the all RTP processed counterparts with respect to as shown in Fig. 4.17. Apparently, foil number 9 does not allow the formation of inversion channels. The all RTP processed solar cells show a very homogeneous distribution of across all foils. The mean value is 22 mA/cm^-2 with a small standard deviation of 3-4 %. The solar cells processed according to Seq. C show a minimum of only 16 mA/cm^-2on average. Compared to Seq. D, the low after Seq. C indicates a degradation of the diffusion length. Considering the situation is different. The RTP solar cells show the highest with a mean value of 525 mV. This is significantly higher than the 500 and 487 mV of the pre-diffused CFP and RTP solar cells, respectively. In case of foils from the growth run 6521, Seq. B apparently yields lower than Seq A. The lowest has been obtained after Seq. C with an average value of 481 mV. With respect to" ", the following ranking becomes clear. Sequence D yields by far the highest fill factors. A mean value of 74.4 % has been achieved with top values close to 77 %. Very much in contrast, the pre-diffused solar cells exhibit only about 62 % for both sequences. It seems that for these cells the fill factor increases with the foil number and that it behaves detrimental to : the higher the lower " ". Remarkably, the non pre-diffused CFP cells (Seq. C) are superior to the pre-diffused ones, showing a mean" " of almost 68 %.