STABLE RECORD EFFICIENCIES FOR EFG AND STRING RIBBON SOLAR CELLS Giso Hahn and Patric Geiger
University of Konstanz, Department of Physics, P.O.Box X916, 78457 Konstanz, Germany email: giso.hahn@uni-konstanz.de
ABSTRACT: Crystalline silicon ribbon materials have a high potential to significantly reduce wafer costs in PV due to their good use of the silicon feedstock without any losses related to block casting and subsequent sawing steps. This could lead to a significant reduction in Watt-peak (Wp) costs, if efficiencies are comparable to the ones obtained with standard cast multicrystalline wafers. Gettering and hydrogenation steps are necessary to improve the inhomogeneous as-grown material quality of the ribbon silicon materials due to the higher defect densities present. In this study we present results from three different solar cell processes for EFG and String Ribbon solar cells. Special attention is paid on the Al- gettering step with two different procedures tested (evaporation with subsequent alloying as well as screen-printing of Al-paste followed by a firing step). Efficiencies of the best cells benefit from the better back-surface-field (BSF) resulting from the screen-printed Al, and stable record efficiencies of 16.7% for EFG and 17.7% for String Ribbon could be achieved on 4 cm2 solar cells. The latter represents the highest value stable under illumination for ribbon silicon material.
1. SILICON RIBBONS
Ribbon silicon has a high potential to bring down Wp costs in photovoltaics (PV) because of its good use of the silicon feedstock. As no sawing steps have to be carried out and no time and energy consuming block casting is involved, modules made out of ribbon silicon solar cells should have a substantial cost advantage. But this is only true if efficiencies are in the same range as for standard ingot cast multicrystalline silicon, which has the highest share in PV module production (58%) [1]. To increase the inhomogeneous (and locally quite poor) as-grown material quality, gettering and hydrogenation steps have to be applied to the defected ribbon materials. Recent studies [2-5] revealed that P- as well as Al-gettering steps can significantly enhance lifetimes of minority carriers in RWE Schott Solar’s EFG (Edge-defined Film-fed Growth) and Evergreen Solar’s SR (String Ribbon) material. Even more dramatic is the effect of hydrogenation, with a synergetic effect observed for a gettering step preceding the hydrogenation. Both, hydrogenation via a hydrogen-rich PECVD (Plasma-Enhanced Chemical Vapour Deposition) SiNx with a subsequent firing step to release the hydrogen into the silicon bulk [2,3], as well as in-diffusion of hydrogen by a remote plasma [4,5] show large improvements in minority carrier lifetime. Spatially resolved measurements revealed that gettering is more effective in better areas of the as-grown wafer, whereas hydrogenation improves all areas significantly [4,5]. Lifetimes above 300 µs could be detected locally in wafers that underwent a gettering as well as a hydrogenation step. But locally in the wafer, areas have been identified which could not be improved by gettering and/or hydrogenation at all [4,5].
This inhomogeneity within typical wafer dimensions of 10x10 cm2 is still a problem to be addressed, and there is some research activity going on at the moment to clarify the nature of these defected regions which cannot be improved substantially within the solar cell process.
2. CELL PROCESSES
In the present study we were interested in the efficiency potential of EFG and SR material. Therefore we have chosen a solar cell process including evaporation of contacts to minimise grid shadowing losses and obtain a good blue response. In Fig. 1 the three applied processes are shown schematically. We used standard material quality with a base resistivity of 2-4 Ωcm. Wafers of 5x5 cm2 size have been acidic etched prior to processing. In process A (standard process) an open-tube POCl3 diffusion (80-100 Ω/sq.) is followed by a dry thermal oxidation (10-17 nm thick oxide) and the evaporation of 2 µm of Al at the back side. The Al is alloyed at 800 °C for 30 min and during this Al-gettering step a back-surface-field (BSF) is formed resulting in a back surface recombination velocity Sb of about 1500 cm/s. The front contact (evaporation of Ti/Pd/Ag) is defined by photolithography and the back contact is formed by evaporation of 2 µm Al. An Ag electroplating step was applied to reduce the series resistance. Cells of 2x2 cm2 size
Erschienen in: 13th Workshop on Crystalline Silicon Solar Cell Materials and Processes : Extended Abstracts and Papers / Sopori, B. L. et al. (Hrsg.). - S. 146-150
are separated by dicing and the solar cells are characterized by IV measurements. Afterwards, the incorporation of hydrogen is carried out on the processed cells by microwave-induced remote hydrogen plasma passivation (MIRHP) and cells are characterized again.
Al-gettering Photolithography Ti/Pd/Ag front contact
Thermal oxidation P-Diffusion (90Ω/sq)
Acidic etching Acidic etching
Ti/Pd/Ag front contact Ti/Pd/Ag front contact Al back contact Al back contact
Ag electroplating Al back contact Ag electroplating
P-Diffusion (90Ω/sq) P-Diffusion (90Ω/sq) Thermal oxidation Thermal oxidation
Photolithography Photolithography Al screen printing + firing Al screen printing + firing
Al-gettering Process A
standard process
Process B Al screen printing (SP)
Process C Al-gettering + SP
Acidic etching
Cell separation MIRHP passivation
IV measurement
IV measurement IV measurement IV measurement
MIRHP passivation MIRHP passivation IV measurement IV measurement
Cell separation Cell separation Ag electroplating
Fig. 1: Overview of the three processes used in this study for fabrication of 4 cm2 EFG and SR solar cells.
For process B the formation of the BSF is carried out by screen printing of Al paste on the cell’s back side followed by a firing step in a belt furnace. This results in Sb of about 200 cm/s (BSF thickness 8-10 µm).
The lower Sb as compared to process A should lead to higher internal quantum efficiencies (IQEs) in the areas of high minority carrier lifetime. On the other hand, the short firing step and the use of Al-paste instead of pure Al in process B might result in a reduced Al-gettering effect. Therefore, in process C we applied the same Al-gettering step used in process A as an Al-pregettering step prior to POCl3 diffusion.
After evaporation of 2 µm Al and subsequent gettering at 800 °C for 30 min we etched off the eutectic layer as well as the formed BSF. Afterwards, cells are processed in parallel to process B.
3. RESULTS
Parts of the results have already been published elsewhere [6,7]. Illuminated IV results of the best cells before and after hydrogenation for each process shown in Fig. 1 are listed in Table 1. It can be seen that all parameters of all cells can be significantly improved by hydrogenation. This effect is even more dramatic for cells with lower efficiencies. Furthermore, the best cells fabricated according to process B show significantly higher short-circuit current densities Jsc and open-circuit voltages Voc as compared to the standard process A. Comparing the results for processes B and C in Table 1 reveals the material inhomogeneity of the silicon ribbons. While no increase in efficiency could be observed for the best EFG cells, the Al-pregettering step led to a further increase in efficiency for the best SR cell.
To assess the effect of the different Al-gettering steps in processes A, B, and C a large statistic would therefore be necessary. As solar cell production capacity was limited (about 25 cells for each material and process have been fabricated) we have chosen another way to compare the effectiveness of the different gettering steps. Pairs of adjacent wafers with comparable grain structure have been processed according to the processes shown in Fig. 1. Results from pairs of such wafers are presented in the following and the effect of the different gettering steps can be observed on the same crystal grains of the different wafers.
In Fig. 2 the IQEs at 980 nm for pairs of adjacent wafers are shown. The lower Sb of process B leads to a significantly higher IQE as compared to process A in regions of high τ, despite of the shorter Al-gettering step in combination with Al screen-printing and firing. The Al-pregettering can further increase IQEs
Table 1: Cell parameters for the best EFG and SR cells processed according to processes A, B, and C respectively (no antireflective coating). Shown are the values before and after hydrogenation (MIRHP).
* Parameters for the best cells after ZnS/MgF2 DARC have been independently measured at JRC Ispra.
Process EFG SR
Jsc
[mA/cm2] Voc
[mV] FF
[%] η [%]
Jsc
[mA/cm2] Voc
[mV] FF
[%] η
[%]
A before MIRHP, no ARC 21.4 560 77.0 9.2 22.5 557 78.4 9.8 A after MIRHP, no ARC 22.8 577 80.2 10.5 23.1 568 79.8 10.5 B before MIRHP, no ARC 20.0 547 77.0 8.4 24.4 590 76.9 11.1 B after MIRHP, no ARC 23.3 590 79.6 10.9 24.7 600 77.9 11.5 C before MIRHP, no ARC 21.4 551 75.3 8.9 24.6 601 76.8 11.4 C after MIRHP, no ARC 23.7 587 78.1 10.9 25.1 609 77.9 11.9 Best cell after DARC* 35.1 601 79.0 16.7 37.0 615 77.7 17.7
Fig. 2: IQEs at 980 nm for adjacent EFG and SR solar cells processed according to the sequences shown in Fig. 1 after hydrogenation. The lower Sb of the screen-printed BSF for process B results in higher IQEs as compared to process A (top) in areas of high bulk lifetime. Some areas of mainly good quality benefit from the Al-pregettering step of process C as compared to process B (bottom).
especially in good quality areas of the wafer due to the more effective gettering as can be visualized by comparing the pairs of wafers processed according to processes B and C respectively.
Interestingly, IQE values before hydrogenation for cells processed according to process B can be lower as compared to the adjacent cells processed with process A (Fig. 3). This could be due to the better Al- gettering effect of process A. Without hydrogenation lifetimes are not high enough to get the full benefit of the lower Sb for the screen-printed BSF. The situation changes when lifetimes are increased after hydrogenation and the lower Sb for process B can be used. Now the IQE especially in the long wavelength part of the spectrum is higher for process B. Additionally, the hydrogenation step might passivate defects introduced during the screen-printing step.
400 600 800 1000 1200
0.0 0.2 0.4 0.6 0.8 1.0
adjacent EFG cells Process B before MIRHP
Process A before MIRHP
Process B after MIRHP
Process A after MIRHP
IQE
λ [nm]
Fig. 3: IQEs of two adjacent EFG cells before and after hydrogenation (MIRHP). After hydrogenation the lower Sb for process B increases the IQE for long wavelengths, whereas without hydrogenation the lower Al-gettering efficiency results in lower IQE values (same two cells as already shown in Fig. 2, top).
To investigate the Al-gettering effect further, we stripped off the metal contacts, the BSF, and the emitter for the four EFG cells shown in Fig. 2. After a surface passivation using iodine/ethanol solution we measured the spatially resolved bulk lifetime τ using microwave-detected photoconductance decay as described in [4]. Results can be seen in Fig. 4 and the following conclusions can be drawn.
Fig. 4: Bulk lifetimes of the four EFG cells already presented in Fig. 2. The influence of the different Al- gettering steps can be seen by comparing adjacent wafers with the same grain structure.
Comparing process A and B in Fig. 2 and 4: for process A (sample partly broken) no use of the high τ in the top part of the cell can be made due to the higher Sb. A higher IQE can be reached for process B despite of only average τ because of the lower Sb as compared to process A.
Comparing process B and C: a higher τ can be reached in the good quality region due to Al-pregettering, which can be used (higher IQE) due to the lower Sb for the Al screen-printing process.
4. BEST CELLS
IQE mappings at 980 nm for the two best cells can be seen in Fig. 5. Several electrically active grain boundaries are visible, whereas some grain boundaries present in the SR cell (invisible in Fig. 5) show no recombination activity at all.
The stability under an illumination of 1 sun was tested and results for the best SR cell are presented in Fig. 5 (right). The cell was measured immediately after a hotplate anneal (200 °C, 30 min). Within measurement uncertainty the cell seems to be totally stable under illumination. This is in contrast to other results published recently [8,9]. These cells had a PECVD SiN/MgF2 double layer ARC and it was found that the initial efficiency for cells with η >16% after annealing decreases 0.5-1% absolute within a short time. An explanation for this decrease could be a less effective gettering under the rapid thermal firing conditions used in [8] with some Fe remaining in the wafer. Or the observed instable efficiency in [8,9]
may be related to the used SiN, too, and could be caused by a change of the fixed charge within the SiN as in our process a thermal oxide for surface passivation and a ZnS/MgF2 double layer ARC is used.
1
0.5
0 0.25 0.75 IQE [980 nm]
1
0.5
0 0.25 0.75 IQE [980 nm]
0 5 10 565 570 575
99.0 99.5 100.0 100.5 101.0
best SR cell
η rel [%]
illumination [min]
Fig. 5: IQEs at 980 nm for the best EFG (left) and SR cell (middle). Although grain boundaries with remaining electrical activity are still visible, efficiencies of 16.7% (EFG) and 17.7% (SR) could be achieved. The stability under illumination of the best SR cell is shown on the right.
5. SUMMARY
Stable record efficiencies of 16.7% for EFG and 17.7% for String Ribbon solar cells have been reached with a screen-printed Al BSF which leads to a lower Sb and therefore to a higher IQE in areas of average or good τ. For cells with a thin BSF (evaporated Al) areas of high τ do not lead to better IQEs because of the higher Sb. An Al-pregettering step can lead to better τ in good quality areas and in combination with a low Sb to higher IQEs.
ACKNOWLEDGEMENTS
Part of this work was financed by the German BMWi under contract numbers 0329858J and 0329846J.
REFERENCES
[1] P. D. Maycock, PV News 2003; 22(3), 1-5
[2] J-W. Jeong, A. Rohatgi, M. D. Rosenblum, J. P. Kalejs, Proc. 28th IEEE PVSC, Anchorage 2001, 83-86 [3] V. Yelundur, A. Rohatgi, J-W. Jeong, A. M. Gabor, J. I. Hanoka, R. L. Wallace, Proc. 28th IEEE PVSC,
Anchorage 2001, 91-94
[4] P. Geiger, G. Hahn, P. Fath, E. Bucher, Proc. 17th EC PVSEC, Munich 2001, 1715-1718 [5] P. Geiger, G. Hahn, P. Fath, E. Bucher, Proc. 17th EC PVSEC, Munich 2001, 1754-1756 [6] P. Geiger, G. Hahn, P. Fath, Proc. 2nd WC PVSEC, Osaka 2003, in press
[7] G. Hahn, P. Geiger, Progr. Photovolt: Res. Appl. 11, 341-346 (2003) [8] D. S. Kim et al., Proc. 2nd WC PVSEC, Osaka 2003, in press
[9] B. Damiani et al., Proc. 2nd WC PVSEC, Osaka 2003, in press
