PHOTONICS APPLIED: MICROELECTRONICS PROCESSING: Ultrafast lasers improve efficiency of CIS thin-film solar cells

GERHARD HEISE, HELMUT VOGT, ANDREAS HEISS, and HEINZ P. HUBER

The market for photovoltaic (PV) modules is growing at a rate greater than 40% per year.¹Thin-film solar cells continue to increase their market share in the growing solar PV industry. Copper-indium-diselenide (CIS)—or more specifically Cu(In, Ga)(S, Se)₂—thin-film solar cells have the potential to achieve optical-to-electrical energy conversion efficiencies comparable to bulk silicon wafers, with demonstrated production module efficiencies of up to about 13%.²

Many laser processes are enabling the reliable and efficient production of crystalline and thin-film solar cells, including junction isolation, laser doping, and laser cutting or welding. An important technique to enhance the overall conversion efficiency of thin-film solar cell modules is through an integrated, monolithic serial interconnection that reduces ohmic losses in the module by dividing the large modules into smaller cells connected in series to achieve a high-voltage/low-current output from the module. And while nanosecond laser ablation and mechanical scribing can be used in this process, shorter-pulse picosecond lasers are proving far more successful in the operation.

Serial interconnection processes

A typical thin-film solar module (about 0.5 m × 1 m) is divided into individual cells with a strip width of about 5 mm. Thin-film solar cells consist of the absorber layer sandwiched between a metallic contact and a transparent conducting contact through which the solar light irradiates the absorber (see Fig. 1). In the substrate configuration applied to CIS, the metallic contact is deposited on the glass substrate.

For the monolithic serial interconnection, three functional line patterns have to be created in the thin-film layers. The first so-called P1 (pattern 1) has to achieve a galvanic separation of the molybdenum (Mo) p-contact. In P2, the CIS layer on top of the Mo has to be removed to enable a good contact of the Mo to the zinc oxide (ZnO) layer. This ZnO layer is deposited in a later process step and acts as the transparent n-contact. The patterning process P3 separates this n-contact, completing the monolithic serial interconnection.

Mechanically scribed P3 trenches extend down to the Mo layer; however, for its functionality, it is sufficient to separate only the conductive ZnO layer. The patterns repeat typically every 5 mm, producing a considerable amount of dead area where no electric power can be harvested.

Commonly applied structuring processes for the integrated interconnects are either based on nanosecond laser ablation for P1 or on mechanical scribing for P2 and P3. Both methods introduce damage to the thin films by thermal effects and mechanical forces. Nanosecond-laser-structured P1 is mainly performed with repetition rates of about 100 kHz and at process speeds of several meters per second. Unfortunately due to its lack of selectivity this process typically introduces some damage.^4,5

For P2 and P3, the task is more complex, because one layer has to be structured on top of another thin film. Nanosecond laser structuring is always connected with a thermal diffusion length in the range of microns leading to a lack of selectivity for P2 and P3.⁶ Therefore, production commonly employs mechanical scribing for P2 and P3, which creates 50–70-μm-wide trenches by chipping off of the CIS or the CIS/ZnO on top of the Mo. The structured lines are irregular and force a large safety distance of 150 to 250 μm. Furthermore, the achievable processing speed is limited to about 1 m/s and the scribing needles have a limited lifetime.

Much effort in the scientific and industrial community is going into research to find suitable laser processes to replace mechanical scribing. Using the picoREGEN ultrafast laser from High Q Laser (Rankweil, Austria; a Newport Spectra-Physics company) at a wavelength of 1064 nm with a pulse duration of about 10 ps, a repetition rate of up to 950 kHz, and a maximum power of 30 W, the Laser Centre of the Munich University of Applied Sciences successfully realized all three patterning steps as demonstrated on 300 × 300 mm² samples from the R&D pilot line at AVANCIS (see Fig. 2).

FIGURE 2. Patterned structures are compared for the monolithic serial interconnections in CIS thin films, showing P1, P2, and P3 lines from top to bottom. Confocal microscope images (left side) show commonly applied lines structured by a nanosecond laser (P1) and by a mechanical tip (P2 and P3). Confocal image profiles (right side) show P1, P2, and P3 lines structured with picosecond laser pulses. The P1 line has been structured from the glass side. The laser lines have typical widths of 30 μm and less visible damage, chipping, and burr compared to the mechanical lines with a width of 50 to 70 μm, requiring a safety distance of 150 to 250 μm between the lines [7].

P1 processing

The galvanic separation of Mo is achieved by irradiating the metal layer from the glass side (see Fig. 3). This liftoff process is based on directly induced laser ablation and utilizes the laser's energy very efficiently, allowing a high-speed structuring process.⁹ Trenches have been scribed with repetition rates up to 950 kHz. We achieved processing speeds up to 15 m/s, limited by the speed of our scanner system.

FIGURE 3. A P1 pattern in molybdenum scribed from the glass side (schematic sketch to the right) is shown with increasing fluence at a scribing speed of 15 m/s (microscopic images with front- and backside illumination). The light blue region corresponds to the exposed buffer layer. The length of the red bars is 30 μm [8].

The process window is determined by the requirement to separate every cell galvanically from its neighbors without damaging the silicon nitride barrier layer. Optical microscope images show trenches written with increasing fluence; in some cases, the fluence is too low to punch out the Mo. But if the fluence is too high, damage of the buffer layer below the Mo electrode is visible with backlit illumination.

P2 scribing

The P2 scribe must enable a highly conductive contact between the n-conducting ZnO and the Mo p-contact. While the P1 process is based on single-pulse ablation, P2 trenches are scribed by multipulse ablation at higher pulse overlaps that accordingly reduce the achievable process speed. A single 10 ps laser pulse at 1064 nm typically ablates about 100 nm of the CIS layer. Here, the challenge is to find a suitable combination of the pulse energy and the number of laser pulses applied at a position to selectively ablate the CIS down to the Mo layer, avoiding Mo damage or CIS residues in the trench.

The threshold value measured for CIS ablation with 10 ps laser pulses is in the range of 0.1 J/cm², but this value depends strongly on the detailed composition of the absorber layer, which will vary with the manufacturing process in use. We found that a repetition rate of 950 kHz at a process speed of 4 m/s achieved a scribe without causing thermal damage. Furthermore, the electrical contact of the Mo to the ZnO was improved significantly compared to mechanically scribed P2 trenches.¹⁰

P3 patterning

For P3 patterns scribed with ultrashort laser pulses, it is advantageous to separate only the conductive ZnO layer. Like P1 from the glass side, this P3 patterning is also an induced laser ablation process with low pulse overlap and no interaction between the single pulses. A scribing speed of 15 m/s has been achieved for this process. Contrary to P1, the underlying CIS is the absorbing partner and the ZnO is the transparent partner. The laser ablation of the ZnO layer is indirectly induced by the underlying CIS.

Compared to single cells, the efficiency of thin-film solar modules is partially reduced by the area lost for the serial interconnection. All laser patterning is a means to reduce the so-called dead area of solar modules. Considering conventional mechanical patterning processes for P2 and P3, the active area of monolithically integrated solar modules is reduced by about 100 μm per cell as a result of the scribe width of the three individual pattern trenches and the required in-between spacing. As laser patterning avoids chipping adjacent to the trench, the spacing between P1, P2, and P3 can be significantly decreased resulting in more active area and increased short-circuit current (see Fig. 4).

FIGURE 4. Optical micrographs show patterning trenches as applied for the 14.7% efficient module (a) in comparison to the previously used nanosecond laser P1 and mechanical P2, P3 processes (b) [11].

With the gain of active area and the optimization of the contact from Mo to ZnO in the P2 process, significant efficiency improvements of 15–20% over conventional structuring processes have been achieved in the successful cooperation between AVANCIS and the Laser Centre in Munich. In September 2011 a record efficiency of 15.8% for CIS modules was achieved using picosecond lasers for P1 and P2 patterns.

ACKNOWLEDGMENTS

This work was funded by the German Federal Ministry for the Environment, Nature Conservation and Nuclear Safety within the project "SECIS" under grant No. 0325043A/B. We thank all members of the Laser Centre in Munich who contributed to the success of this work.

REFERENCES

1. "EPIA Global Market Outlook for Photovoltaics until 2015"; www.epia.org/publications/epiapublications.html.
2. M.A. Green et al., Progress in Photovoltaics: Research and Applications, 19, 5, 565–72 (August 2011).
3. G. Heise et al., Appl. Phys. A: Mat. Sci. and Processing, 104,1, 387–393 (2011).
4. V. Probst et al., Solar Energy Mat. and Solar Cells, 90, 18–19, 3115–3123 (2006).
5. B. Dimmler et al., Conf. Record of the IEEE Photovoltaic Specialists Conf., Lake Buena Vista, FL, 189–194 (2005).
6. A.D. Compaan et al., Opt. and Lasers in Eng., 34, 1, 15-45 (2000).
7. G. Heise et al., Progress in Photovoltaics: Research and Applications online, (2012); doi:10.1002/pip.1261, or http://onlinelibrary.wiley.com/doi/10.1002/pip.1261/abstract.
8. G. Heise et al., Applied Phys. A, 102, 1, 173–178 (2011).
9. G. Heise et al., Physics Procedia, 12, B, 149–155 (2011).
10. T. Dalibor et al., 26th European Photovoltaic Solar Energy Conf. and Exhibition, paper 3CO.2.1, Hamburg, Germany, 2407–2411 (2011).
11. H. Vogt et al., 26th EUPVSEC, Hamburg, Germany, paper 3DV.2.9 (2011).

Gerhard Heise is senior scientist and Heinz P. Huber is professor for photonics and director at the Laser Centre of the Munich University of Applied Sciences (MUAS), Department of Engineering Physics, Lothstraße 34, D-80335 Munich, Germany; e-mail: [email protected]; www.hm.edu. Andreas Heiss is development engineer and Helmut Vogt is project manager at AVANCIS GmbH & Co. KG, Otto-Hahn-Ring 6, D-81739 Munich, Germany; www.avancis.de.