Emitter sheet resistance

The evolution of the emitter sheet resistance was studied as a function of plateau time and plateau temperature of the standard RTP diffusion process (see Fig. 3.1) for the SODs P509 and P507. In an earlier work and had been identified as the most significant process parameters [128]. Due to the high heating and cooling rates of 100 K/s used throughout this work, and are almost equal to the actual diffusion time and temperature, respectively.

For the determination of , 5 5 cm² CP133-etched Cz wafers with a specific resistivity of 1 cm were used. After diffusion and removal of the PSG in HF, the mean was calculated from four point probe measurements at five spots on each wafer.

Spin-on source P509

In Fig. 3.2 is plotted for the spin-on source P509. The investigations were restricted to a maximum plateau duration of 120 s and a minimum duration of 5 s. Longer diffusion times contradict the idea of rapid thermal processing and shorter times make the actual diffusion time become improperly defined. As expected, decreases with increasing temperature and prolonged time since the amount of diffused and electrically active P increases.

The figure also shows that if is too short for a given , the amount of diffused P is not sufficient to overcompensate the boron background doping and to form a pn-junction. This explains why for = 850 C increases as increases from 5 to 30 s.

0 20 40 60 80 100 120 20

40 60 80 100 120 140 160 180 200 300400 500

T_P [°C]

850 875 900 925 950

R

[

/s q]

t

[s]

Fig. 3.2: Evolution of the emitter sheet resistance as a function of plateau time and plateau temperature using the standard RTD process and the highly concentrated SOD P509.

Tab. 3.3:Thermal budgets yielding a 100 Ohm/sq emitter using the P509 SOD.

Thermal budget I II III IV V

Plateau temperature [ C] 860 875 900 925 950

Plateautime[s] 120 60 30 15 5

A desired can be obtained by different combinations of diffusion temperature and diffusion time, that is by different thermal budgets. In particular, for a 100 /sq emitter which is favorable in the case of evaporated front contacts commonly used in this work, five thermal budgets can be extracted from Fig. 3.2 (see Table 3.3). Having in mind that there are various silicon materials with bulk properties reacting differently to thermal treatments, a priori it is not obvious to know which thermal budget is best with respect to solar cell efficiency.

Spin-on source P508

The spin-on source P508 was mainly used for the fabrication of 40 to 50 /sq emitter which, as shown later in this work, have proven to be suitable for screen-printed silver front contacts.

Table 3.4 shows three temperature-time combinations, i.e. thermal budgets, each resulting in of 45 /sq. In section 4.3.2 these budgets are used to check whether the bulk lifetime of Cz-Si is preserved during RTD. Additionally, in section 4.6.6 these budgets are applied and assessed with respect to the fabrication of EFG solar cells.

Tab. 3.4:Thermal budgets yielding a 45 Ohm/sq emitter using the P508 SOD.

Thermal budget I II III

Plateau temperature [ C] 860 925 1025 Plateau time [s] 300 60 5 Spin-on source P507

In Fig. 3.3 is plotted for the SOD P507. As the P507 features a lower P concen-tration than the P509 and the P508, a higher is necessary to obtain the same for the same . For example, in order to diffuse a 100 /sq emitter in 30 s, the respective is 975 C in the case of the P507 whereas for the P509, is only 900 C. It has to be mentioned that the PSG of the P507 is much less hydrophile than that of the P509 which is a good indication that the P content within the PSG is lower for the P507.

0 20 40 60 80 100 120 140 160

20 40 60 80 100 120 140 160 180

T_P[°C]

950 975 1000 1025

R

[

/s q]

t

[s]

Fig. 3.3: Evolution of the emitter sheet resistance as a function of plateau time and plateau temperature using the standard RTD process and the medium concentrated SOD P507.

Activation of inactive P during rapid thermal oxidation

is determined by the electrically active part of the diffused P profile. However, depending on the P source, a large amount of inactive P is incorporated in the high concentration region near the surface. A simple way to test on inactive P, is to subject the sample to a second high-temperature step after removal of the PSG. If inactive P is present, it serves as an internal

0 20 40 60 80 100 120 140 160 180 200 0

20 40 60 80 100 120 140 160 180 200

y=x Spin-on source

P509 P507

R

a fte r R TO [

/s q]

R

after RTD [

/sq]

Fig. 3.4: Change in caused by a post-diffusion RTO step at 950 C for 30-40 s. The PSG was removed in HF prior to RTO. The RTD conditions for the P509 SOD were =875 C and was varied from 15 to 120 s. The P507 was diffused at 975 C and was varied from 20 to 160 s.

source and hence decreases upon this drive-in. The amount of inactive P is important if a second high-temperature step is applied after the P diffusion, such as the growth of an oxide for surface passivation. In this work, rapid thermal oxidation (RTO) was used for emitter surface passivation. Horzel [63] observed an RTO oxide growth rate depending on the electrically inactive P concentration. The more inactive P present, the higher the growth rate. Similar observations have been made by many authors for RTO as well as for the conventional growth of oxides on heavily P-doped surfaces, e.g. P concentrations exceeding 10²⁰ cm^-3 [170]. It is widely accepted that this effect is caused by a Si-vacancy contribution mechanism in the silicon substrate which may provide reaction sites for converting Si to SiO2 and consequently increase the rate at which this reaction occurs [62].

In Fig. 3.4, the change in caused during RTO (950 C, 30-40 s) is plotted. The PSG was removed in HF prior to RTO. The RTD conditions for the P509 were = 875 C and was varied from 15 to 120 s. The P507 was diffused at 975 C and was varied from 20 to 160 s. For the P509 the initial decreases drastically during RTO. This hints towards a large amount of inactive P present after RTD which diffuses deeper into the bulk and becomes activated. Because of the decrease in during RTO, the RTD parameters have to be adjusted carefully in order to obtain the desired after RTO. On the contrary, for the P507 almost no change in is observed upon RTO. Only for higher sheet resistances a slight increase is observed as one expects if only active P is present. In this case, P diffuses deeper into the bulk at the expenses of the surface concentration of already active P.

0 50 100 150 200 250 10¹⁶

10¹⁷ 10¹⁸ 10¹⁹ 10²⁰ 10²¹

Base doping

Before RTO SIMS SHP After RTO

SIMS SHP

C on ce nt ra tio n [c m

]

Depth [nm]

Fig. 3.5: Comparison of atomic, measured by SIMS and electrically active P concentration, measured by SHP for the P509 SOD. After RTD at 900 C for 30 s, the PSG was etched off and 30 s of RTO at 950 C was carried out.