高效太阳能电池的晶圆清洗程序比较

A number of wafer cleaning technologies compete for the use in high efficiency solar cell processing. Some are  borrowed from IC manufacturing, e.g. the well-known RCA cleaning, the H2SO4 /H2O2 based SPM clean, and the  ozone-based HF/O3 clean, others were especially developed for solar cell processing such as the BASF Seluris® C  cleaning approach. In this work the mentioned technologies are compared qualitatively, and the latter three are  compared experimentally in terms of cleaning efficiency and suitability as pre-diffusion cleans for industrial  application in the field of processing bifacial and IBC n-type solar cells with efficiencies exceeding 21 %. It is  found that all three cleaning recipes reduce metal contamination from values up to 1E15 atoms/cm2 down to the  level of 2-3E11 atoms/cm2 . Both, the Seluris® C clean and HF/O3 clean perform similar as the reference in use (SPM clean). The suitability for mass production of the two procedures was tested by artificial bath aging (by  spiking with Fe and Cu according to an enrichment simulation), and testing these baths in terms of outplating and  cleaning efficiency. Threshold contamination levels were determined by measuring implied Voc values after  boron and phosphorus diffusion of intentionally contaminated wafers. Cell precursor lifetime samples were  manufactured and good results with the tested, industrially applicable cleaning methods were obtained.  Keywords: wafer cleaning, n-type, ICP-MS, metal surface contamination.

It is generally assumed that high efficiency solar cell  devices  such as PERC cells, bifacial n-type solar  cells such as the ISC’s BiSoN concept and interdigitated  back contact (IBC) solar cells such as the ISC’s ZEBRA  concept require more advanced wafer cleaning than  standard aluminium back surface field solar cells. Cell  efficiencies around 20% (BiSoN)  and above 21%  (ZEBRA) can only be achieved with sufficiently  clean surfaces before high temperature steps. For  industrial implementation of such solar cell concepts  straight forward and cost-efficient high throughput  cleaning techniques are required.

In a first experiment four different cleaning  procedures were compared: the current ISC Konstanz  standard clean for high efficiency solar cells, our process  of record (POR), the HF/O3 clean,the BASF Seluris® C clean, and an industrial clean (HCl+HF). The cleaning  baths were freshly made up and tested on alkaline  textured wafers and cleaning efficiencies were  determined by measuring the metal surface  contamination using the Sandwich-Etch-ICP-MS method.

Fig1

To determine the impact of bath aging on the wafer  surface concentration, freshly made up cleaning baths  were spiked using Cu and Fe standard solutions  (1000g/L, AAS standard solution in 2% HNO3 , Carl  Roth). For the first measuring point an equivalent of 5  ppb Fe/Cu were added, 50 for the second, and 250 for the  third (resulting in 5, 55, and 305 ppb in the bath). SPM  cleaned wafers were immersed into the spiked cleaning  baths for 5 min (Seluris® C) and 10 min (HF/O3 ),  respectively. Wafers were rinsed with water before the  metal surface contamination was extracted for ICP-MS  analysis. The resulting Cu and Fe concentrations on the  wafer surfaces can be found in Fig. 5.

No significant contamination of the clean wafer  surfaces after immersion in spiked cleaning baths could be measured for the Seluris® cleaning bath, whereas the  HF/O3 cleaning bath had an increasing tendency for Fe contamination from the bath (“outplating”) and a strong  tendency for Cu outplating at the highest contamination  level. This also has an impact on the cleaning efficiency on alkaline non-cleaned textured wafers. In Fig. 6 we  show that the efficiency for Cu removal of the HF/O3 bath deteriorates with bath aging, probably due to the  outplating effect, whereas no impact on the Seluris® C  can be detected. These results show that the assumed  feed+bleed conditions for the Seluris® C bath are well  suited for mass production. The HF/O3 bath, however, might need different feed+bleed settings (replacement of  larger volumes of cleaning liquid) to maintain low metal  concentrations in the bath and hence good cleaning  efficiency.

High efficiency cell processes require more advanced  cleaning. However, cleaning needs to be simple and cost  effective. Two alternatives were shown to be suitable for  mass production. Threshold Cu surface concentrations  before the diffusion steps of ISC Konstanz’ standard  process for high efficiency bifacial n-type solar cells  (BiSoN) were deduced from experimental results and  found to be in the range of 1E12 atoms/cm2.