In microelectronic production lines, reclaimed silicon wafers are often used for process monitoring out of cost-saving reasons. To make a re-utilization possible, the reclaim process must accomplish a complete removal of all layers (dielectric, metallic or other) deposited on the original silicon wafer and ensure a contamination-free surface. Furthermore, the recycling process aims to remove only little material from the wafer to make a multiple re-use possible.
Reclaimed wafers from microelectronic industry represent an interesting option as potential low-cost substrate material for epitaxial silicon thin-film solar cells. The preparation of a thin-film solar cell on a reclaimed silicon wafer is less sensible to substrate properties compared to microelectronic devices. A less sophisticated reclaim procedure might be sufficient thus further contributing to the aspect of cost-saving.
Single-crystal silicon reclaim wafers from microelectronics industry were supplied by the company AstroPower (Delaware, USA) for a re-use as substrates in epitaxial thin-film solar cells. The reclaim procedure included a mechanical removal of all devices by sand-blasting and a wet-chemical surface cleaning in a NaOH solution. The specific resistivity and the thickness of the electronic grade wafers was determined to 0.01 Ωcm and 600 µm respectively. Prior to further processing, the 6” wafers were cut into 50x50 mm² samples by laser scribing.
The wafers were treated by different cleaning methods, resulting in different surface morphologies: a. No additional treatment. Mean roughness: 3.7 µm.
b. Removal of 100 µm from the surface by wet-chemical CP133-damage etch. Mean roughness: 1.6 µm. The strong etching led to an increased fragility of the edges and a significant reduction in wafer area. The latter aspect presented a severe problem to the formation of the front contact grid which is defined to a specific area.
c. Mechanical grinding of one surface with a final wafer thickness of 400 µm. The surface damage introduced by the grinding was removed by KOH etching. Mean roughness: 0.6 µm. The epitaxial layer is grown on the ground surface.
d. Mechanical grinding of both surfaces with a final wafer thickness of 400 µm. Further treatment and mean roughness are identical to c.
The main characteristic for pre-treatment b to d is that a large amount of material is removed from the surface prior to epitaxy. All samples were RCA-cleaned prior to epitaxy. Solar cells were prepared using process 3 with industrial screen printing technology and for reference, a selection of samples was introduced into the standard cleanroom process 1.
6.4 Solar cells on reclaimed silicon wafers 117
For an evaluation of the pre-treatment methods, solar cells from process 3 were analyzed. Within each pre-treatment group, samples with similar or equal epilayer thickness were used for the calculation of the mean values which are depicted in Table 6.9. For an interpretation of the different pre-treatment methods the difference in epilayer thickness between each group has to be accounted for.
Best efficiencies were obtained for epitaxial thin-film solar cells on ground reclaim substrates. Solar cells of type c and d feature similar characteristics, indicating that the condition of the rear surface does not affect the solar cell performance. The difference in short-circuit current density can be attributed to the different epilayer thickness. The high reproducibility of the grinding and the epitaxy process respectively are reflected in the comparatively low standard deviations.
Comparing solar cells of type b and d a decrease in VOC, JSC and FF can be observed for the wet-
chemically treated samples. The mean epilayer thickness is larger for type b and therefore the decrease in short-circuit current cannot be traced back to a thinner base layer. The same feature can be found for solar cells with type a pre-treatment. This characteristic (reduced JSC and VOC with increasing
epilayer thickness) can be explained if low minority carrier lifetimes in the base are assumed. The comparatively low fill factor for solar cells with pre-treatment b is attributed to technological problems in contact formation and edge isolation due to the fragility of the samples.
Pre-Treatment dbase VOC JSC FF Efficiency
[µm] [mV] [mA/cm²] [%] [%]
a none 37 597 ± 5 22.3 ± 0.4 77.5 ± 0.6 10.3 ± 0.4
b CP-133 35 610 ± 3 23.0 ± 0.2 74.6 ± 3.8 10.5 ± 0.5
c 1 side ground 33 614 ± 1 24.4 ± 0.1 77.1 ± 0.1 11.6 ± 0.1
d 2 sides ground 30 613 ± 2 23.6 ± 0.4 78.1 ± 0.1 11.3 ± 0.2
Table 6.9: Mean values for epitaxial solar cells on reclaimed Cz-Si wafers prepared by solar cell process 3. The different substrate pre-treatments are compared.
Figure 6.20 shows internal quantum efficiency characteristics measured for solar cells of type a, b and d. The epilayer thickness of 25 µm is the same for all cells. For comparison, the IQE curve for a solar cell on epitaxial reference material (30 µm epilayer grown on highly doped Cz-Si in a commercial system) prepared in the same solar cell process is included.
For the entire spectral range, the solar cell without additional pre-treatment (type a) reveals the lowest response. The characteristics for type b and d are similar with type d being slightly superior. The spectral response is clearly correlated to the pre-treatment of the samples. Wet-chemical CP-133 damage-etch and a grinding of the surface result in a comparable epilayer quality. The mechanical removal of the devices by sand-blasting and the subsequent NaOH treatment possibly results in a highly defected wafer surface, inadequate for high-quality epitaxy. Another explanation for the low response of type a solar cells might be that surface near regions are still contaminated by impurities after the reclaim procedure. During high-temperature CVD these impurities could easily diffuse into the epilayer leading to reduced carrier lifetime. Future work on reclaim material will have to deal with a characterization of impurities in the substrate material and in the epilayers e.g. by means of GDMS (Glow-Discharge Mass Spectrometry).
Compared to the internal quantum efficiency measured for the epitaxial reference cell, the CP-etched or ground samples show a similar blue response and are only slightly inferior in the mid-wavelength range. The superiority of the epitaxial reference cell in red response is attributed to a combination of better minority carrier diffusion length, thicker base layer and probably a reduced interface recombination velocity.
The best efficiencies were reached on ground substrates for both solar cell process types. In Table 6.10 the illuminated I/V solar cell parameters of the best solar cells are summed up.
400 600 800 1000 1200 0.0 0.2 0.4 0.6 0.8 1.0 Epitaxial Ref. a - as received b - CP133
d - ground on both sides
IQ
E
λ
[nm]
Figure 6.20: Internal quantum efficiency for epitaxial thin-film solar cells on highly doped reclaim wafers with different epitaxial pre-treatment.
Area Sample dbase VOC JSC FF Efficiency
Process [cm²] [µm] [mV] [mA/cm²] [%] [%] 1 21.2 Reclaim (d)* 22 630 28.6 74.5 13.4 FZ-Si Ref. 634 35.9 78.5 17.8 3 23.0 Reclaim (d) 35 615 23.9 78.3 11.5 FZ-Si Ref. 621 30.7 76.7 14.6
* Confirmed measurement by ISE Calibration Laboratory
Table 6.10: Solar cell parameters of best solar cells achieved on epilayers grown on ground, highly doped reclaim wafers.
Comparing the solar cell results of the epitaxial cells to the FZ-Si reference cells shows the potential of the realized concept. Apart from small reductions in open-circuit voltage, the main loss for the epitaxial cells can be observed in short-circuit current.
Future activities will focus on the optimization of the reclaim procedure for an application of reclaim wafers as substrate material for epitaxial thin-film solar cells.