Spin-on dopants

SOD P508

Table 2.2 reports the mean sheet resistance and the respective standard deviation (SD) after diffusion using the highly concentrated phosphorous SOD P508 and the standard open-loop process. Comparison of the mean of the single-sided diffused samples 3 and 8P-4 gives no indication of UV-enhanced diffusion. The same holds for samples 8P-5 and 8P-6 which were subjected to a higher diffusion temperature since the THL power was increased from 66 to 73 %.

Tab. 2.2: Measured sheet resistance after RTD from the spin-on source P508 using the standard open-loop process, once with the excimer UV lamps switched off and once switched on.

Sample Wafer SOD P508 THL UV Frontside [ /sq] Backside [ /sq]

ID type 1^stdeposition on [%] lamps mean SD mean SD

8P-3 A - 66 off 47.0 1.08 -

-8P-4 A - 66 on 45.3 1.06 -

-8P-5 A - 73 off 23.4 0.50 -

-8P-6 A - 73 on 21.6 0.49 -

-8P-7 B frontside 66 off 25.4 0.61 27.3 0.56

8P-8 B frontside 66 on 24.9 0.95 25.8 1.50

For both sets the difference in between the case without UV and the case with UV is extremely small. The temperature difference which would explain the difference in can be estimated as follows. It was shown before that increasing the THL power from 66 to 73 % leads to an increase in the wafer temperature of approximately 50 C. For this reason, of sample 8P-3 is 23.6 /sq lower than of sample 8P-5. For the sake of a simple dependence, we assume that, in first approximation, sheet resistance varies linearly with temperature in the relevant range. Here, this means that 1 /sq difference in translates into a difference in temperature of approximately 2 C. Consequently, the differences in between samples 8P-3 and 8P-4 and between samples 8P-5 and 8P-6 can be explained by just 3 to 5 C, respectively. It was shown earlier that the comparison of open-loop diffusion without UV with open-loop diffusion with UV leads to an uncertainty in the wafer temperature of less than 3 C as long as optically identical wafers are used. However, taking into account that the spin-on procedure yields comparable but not identical optical properties one can conclude that the observed variations in are within the limits of experimental error and do not point

0 100 200 300 10¹⁶

10¹⁷ 10¹⁸ 10¹⁹ 10²⁰

10²¹ Spin-on dopant P508

RTG Berlin 8P-3

8P-4 with UV CEA Grenoble

8P-3

8P-4 with UV

C on ce nt ra tio n [c m

]

Depth [nm]

Fig. 2.5: Atomic concentration profile of P after RTD using the SOD P508 and the standard open-lopp process featuring a 5 min plateau with 66 % of THL power from the back. For sample 8P-4, the SOD film on the front was illuminated by UV light from the excimer lamps during RTD. SIMS measurements were carried out at RTG Mikroanalyse, Berlin and at CEA Grenoble on adjacent spots.

to UV-enhanced diffusion. In any case, if there is any difference at all it is of no relevance for industrial solar cell processes.

The same conclusion can be drawn from the atomic P profiles as measured by SIMS which are shown in Fig. 2.5 for the samples 8P-3 and 8P-4. The SIMS measurements were carried out at the CEA in Grenoble and for comparison at RTG Mikroanalyse GmbH in Berlin. Apart from differences in the near-surface concentration, the measurements agree well. Clearly, both measurements show that the profile diffused without UV coincides perfectly with the profile diffused with UV. There is no noticeable difference, neither in the lowly doped tail of the samples, nor in the junction depth and not in the highly doped surface near region.

In the case of the double-sided diffused wafer 8P-7 both sides exhibit similar . This is remarkable because it means that THL illumination alone does not in any case yield photon-enhanced diffusion which is in contradiction to what has been claimed by Singh and co-workers [165] and Ishikawa and co-workers [72], respectively.

SOD P507

Table 2.3 comprises the sheet resistance and the respective standard deviation after diffusion using the standard open-loop process with 66 % of THL power and the medium concentrated spin-on dopant source P507.

Comparing the single-sided diffused samples 7P-3 and 7P-4, no influence of the additional UV illumination can be seen. is identical within the limits of experimental error. The corresponding atomic P profiles as measured by SIMS are shown in Fig. 2.6. The two profiles lie above each other perfectly, and there is neither a difference in the plateau nor in the tail region. The near-surface P concentration matches well the solid solubility of P in Si at the diffusion temperature of approximately 900 C [122].

As shown in section 3.1.4, for the P507 SOD the profiles of the electron and the atomic P concentration always coincide. Hence, in contrast to the SOD P508, the P507 does not yield near-surface electrically inactive SiP precipitates. This indicates substantial differences in the chemical composition of the two SODs. Furthermore, there is no difference in of the front of sample 7P-5, diffused without UV and the front of sample 7P-6, diffused with UV which would point to some UV-enhanced diffusion.

Tab. 2.3: Measured sheet resistance after RTD from the spin-on source P507 using the standard open-loop process once with the excimer UV lamps switched off and once switched on.

Sample Wafer SOD P507 THL UV Frontside [ /sq] Backside [ /sq]

ID type 1^st depositon on [%] lamps mean SD mean SD

7P-3 A - 73 off 54.7 1.70 -

-7P-4 A - 73 on 55.5 ^1.09 -

-7P-5 B frontside 73 off 39.7 1.22 34.6 1.37

7P-6 B frontside 73 on 41.9 0.75 34.7 0.89

7P-7 B backside 73 off 35.9 0.72 42.6 1.23

7P-8 B backside 73 on 36.6 1.17 41.2 2.36

Remarkably, the backside of both wafers exhibits a significantly lower than the front.

One might be tempted to interpret this observation as photon-enhanced diffusion due to the illumination by the THLs. As mentioned earlier, several authors have claimed this effect.

However, in case of wafers with double-sided spin-on, care has to be taken whether the front or the backside was coated first, because the sheet resistance does not solely depend on the diffusion temperature but also on the baking conditions of the SOD film. Of course, the film deposited first is baked twice whereas the film deposited secondly is baked only once. Indeed, for samples 7P-7 and 7P-8 where the backside was coated first, the situation is reversed and the side facing the THLs shows higher . This demonstrates that careful and reproducible preparation of the spin-on dopant film has to be ensured. One might speculate that some of the formerly published experiments claiming the photon-enhanced diffusion from spin-on sources lacked this careful preparation.

In this work, no hint of UV or photon-enhanced diffusion can be found for the Filmtronics SOD P507. Measurable differences in are likely stemming from small deviations in the deposition and baking of the SOD film.

0 100 200 300 400 10¹⁶

10¹⁷ 10¹⁸ 10¹⁹ 10²⁰ 10²¹

Spin-on dopant P507

7P-3

7P-4 with UV

C on ce nt ra tio n [c m

]

Depth [nm]

Fig. 2.6:Atomic concentration profile of phosphorus after RTD using the spin-on dopant P507 and the standard open-lopp process featuring a 5 min plateau with 73 % of THL power from the back. The SOD film on the front of sample 7P-4 was illuminated by UV light from the excimer lamps during RTD.