Laser applications in photovoltaics

In challenging economic times, the industry readily accepts processes offering high throughput at low cost, and this has led to the adoption of lasers in a number of important PV manufacturing processes.

In challenging economic times, the industry readily accepts processes offering high throughput at low cost, and this has led to the adoption of lasers in a number of important PV manufacturing processes.

Peter G. Borden, Applied Materials, Santa Clara, CA USA

Photovoltaics (PV) – the direct conversion of sunlight to electricity – is undergoing a dramatic expansion as a number of drivers have fallen into place. These include increased energy prices, concern for energy security, and the threat of global climate change. In the United States, for example, a number of states have adopted aggressive renewable portfolio standards that compel utilities to generate a significant fraction of their electricity from renewable sources. For example, PG&E, the utility for Northern California, has committed to 1.65GW of PV [1], 30% of the 5.5GW world production in 2008 [2].

Figure 1. Cross-section of the laser grooved buried contact cell [8].

This growth has driven scaling of production, which in turn drives down manufacturing cost in a predictable manner. PV has followed an 18% learning curve (manufacturing cost drops 18% for each doubling of cumulative production) for 30 years [3], and First Solar has reported costs of $0.98/watt [4]. It is possible to buy silicon modules as low as $2.06/watt today [4], prices that will further fuel expansion.

In this climate, the industry seeks all efficiency improvements consistent with high throughput at low cost, and this has led to the adoption of lasers in a number of important PV manufacturing processes. The effect of even small efficiency improvements cascades through the entire system: higher efficiency translates to fewer modules, lower installation costs, less land use, and the ability to afford improvements that further increase generation, such as tracking. Here we survey both the applications where lasers are used to make both wafer-based silicon solar cells and thin film (TF) panels, and some of the new applications that may emerge in the near future.

Wafer-based silicon applications

Wafer-based silicon panels represent ~85% of today’s PV market [5], with a wide base of manufacturers located largely in Europe and the Pacific Rim. Most of the production uses a well-established phosphorous diffusion process with screen-printed contacts, yielding typical efficiencies of 15-17% [7]. These manufacturers seek incremental improvements, many of which require patterning, opening the door for a number of laser applications. We first review the processes currently in manufacturing, followed by opportunities that have potential to provide significant efficiency gains.

The primary application today is edge isolation. A YAG or vanadate laser is used to cut a several-micron deep groove around the periphery of a silicon solar cell as the last step in the process before final test. This disconnects the diffused front junction from the edge of the cell to reduce shunting.

Other applications have found their way into manufacturing, although only as relatively small niches. One is the laser grooved buried contact cell (Fig. 1), developed at the University of New South Wales (UNSW) [8] and put into production by BP Solar. A laser cuts deep narrow grooves through a silicon nitride coating after diffusion of the front junction. A second diffusion provides a heavily doped contact surface. The grooves are then filled with plated nickel and copper. This provides a high efficiency front junction and narrow, high aspect ratio conductors. Laboratory efficiencies exceeded 20%, although somewhat lower efficiencies were obtained in volume manufacturing.

Figure 2. Cells using through holes: a, top) emitter wrap through [9] and b, bottom) metal wrap through [10].

Some cell designs use laser drilled holes to form emitter wrap-through (EWT, Advent Solar) [9] or metal wrap-through (MWT, ECN/Solland Solar) [10] cells, shown in Figs. 2a and b. These cells use the holes to bring the front contact to the back side, enabling the use of surface-mount methods to provide a high packing density and low resistance losses to increase efficiency at the module level. The EWT junction is on both the front and back, so carriers travel a shorter distance to be collected, and has no front metal. This improves efficiency, especially on lower quality material. The holes are drilled using multiple passes of vanadate lasers. The EWT cell needs ~10,000 vias; the MWT needs a few dozen. Throughputs are on the order of two seconds per wafer.

Figure 3. Comparison of silicon nitride coated textured solar cell front surfaces after a) nanosecond and b) picosecond laser ablation [16].

More advanced designs require patterning. Indeed, the highest efficiency cell ever made – the 25% PERL cell from UNSW [11] – and the highest efficiency production cell at 23.5% from SunPower [12] – both use multiple patterning steps. The patterning is necessary to provide small area contacts with heavily doped regions localized under the contact metal. This minimizes carrier recombination at the contacts. The SunPower design additionally interdigitates p- and n-type contacts on the back to eliminate light blocking on the front and provide higher packing density at the module level.

Laser patterning in on the verge of commercialization to enable use of similar methods for incremental efficiency improvements. On the front side, the structure is called a selective emitter (SE). In conventional cells, the emitter (front diffused region) embodies a compromise between heavy doping to both enable low contact resistance and minimize resistance losses as collected current flows to the grid lines, and light doping to improve the efficiency of collecting photons absorbed in the emitter. The SE provides heavy doping only under the contacts, and lighter doping in the field. Efficiency gains from this process alone are on the order of 0.8% absolute.

Using laser patterning, it is possible to implement a selective emitter with as little as one laser exposure. Rödel et al. [13] of IPE Stuttgart demonstrated a process in which the phosphorus doped glass formed during a light field diffusion is a dopant source that can be driven in using a single laser exposure, resulting in 0.4% absolute efficiency gain. The Wenham group at UNSW [14] has shown a laser doping process that forms deeper doped regions combined with narrow plated grid lines to obtain additional gains.

On the back side, the preferred structure creates point contacts and a back reflector tuned to infrared light that is not absorbed in a single pass through the cell. The best known structure is the laser fired contact (LFC) developed by Preu and Grohe at Fraunhofer ISE [15]. A field dielectric is formed and coated with aluminum. Firing at contact points drives the aluminum into the silicon. Aluminum, a p-type dopant, makes contact to the p-type bulk at the firing sites.

Laser ablation of dielectrics is being investigated as a method to pattern dielectric etch or diffusion masks. Two concerns with this method are effectiveness on the highly textured front surface and induced damage. These have both motivated consideration of pico-second UV lasers [16]. Figure 3 shows a comparison of textured surfaces after ablation of a silicon nitride layer with a nano-second and pico-second laser. The nano-second laser melts the surface; the shorter pulse laser shows no visible damage. Englehart and Hermann [17] have compared diode saturation currents in SE structures made with ablation using the two types of laser and, for reference, using HF etching. The the HF reference is ~3X better than the pico-second laser and 30X better than the nano-second.

Figure 4. Views of a P2 laser scribe: a) side, b) top low magnification, and c) top high magnification showing TCO texture. Inset: TF interconnect scheme.

In one of the most ambitions demonstrations of laser patterning, Englehart et al. of ISFH Hameln [18] demonstrated the RISE cell, which is formed entirely using laser patterning. This device uses inter-digitated back contacts formed using etching of a laser patterned masking layer; 22% efficiency has been reported.

Thin film panel applications

Thin film (TF) panels provide low cost by coating a low cost substrate such as glass and creating a set of series connected cells using a combination of laser patterning and conductor deposition. The interconnection is necessary to convert the high current, low voltage output to a low current, high voltage output. This minimizes ohmic power losses, which scale as the square of the current. In a sense, the wafer-based silicon approach uses a set of discrete devices packaged into a module; the TF approach is an integrated circuit packaged into a module.

The inset in Fig. 4 shows the most common interconnect scheme, which uses three laser scribes per cell along the length of the panel. These are often called P1, P2 and P3 (for pattern 1, 2 and 3). The P1 scribe cuts the transparent conducting oxide (TCO) coating on the glass into stripes. The P2 scribe cuts the absorber layer (e.g., amorphous silicon, CdTe, CIGS) into parallel stripes, stopping at the TCO. The P3 scribe breaks the back conductor into cells, also stopping at the TCO. This process divides the panel into a set of cells ~1cm wide. A typical panel is ~1m wide, providing ~100 series-connected cells. The scribes can be as narrow as 50μm wide, but some allowance is required for small angular errors that prevent the lines from being perfectly parallel. As a result, the scribe structure can occupy 300-500μm, ~3-5% of the active area.

A typical process uses a 1064nm pulsed vanadate laser for the P1 scribe and 532nm pulsed vanadate lasers for the P2 and P3 scribes. There are a number of critical parameters, including pulse repeatability, layer thickness control, spot size control, registration of parallel scribe lines, throughput, and control of ejecta, especially after the P1 scribe. Figure 4 shows cross-section and top views of the P2 scribe through the absorber to the TCO. The top views show the scalloped shape of the scribe line due to the laser pulses and the rough texture of the TCO under the absorber. This roughness creates diffuse light scattering to enhance light absorption.

An additional application is edge deletion, which is the TF equivalent of the edge isolation process. The TF absorber and TCO layers run to the edge of the glass. A few millimeters of these layers are removed around the periphery. This provides isolation of the active layers from the environment and creates a good bonding surface for laminating the glass sheet containing the PV layers to a second glass sheet that provides protection against the environment.

Conclusion

The largest PV laser applications today are edge isolation of wafer-based cells and scribing and edge deletion of TF panels. It is likely that a number of additional applications will emerge, especially in the wafer-based arena, driven by the need to pattern cells to obtain incremental efficiency gains. Some of these applications, such as laser doping to create SE structures, are nearing commercialization. Others that provide more elaborate patterning, create damage, or have narrow process windows, will take longer to emerge. However, it is clear that the low cost, high-throughput and non-contact nature of laser processing ensures that the number of PV applications will grow.

References

1. 1. Photon International, May 2009, p. 18.
2. 2. P. Mints, “Photovoltaic industry 2009: a Journey into Uncertainty,” Photovoltaics International, second quarter, 2009.
3. 3. R. Swanson, SPIE Photonics Innovation Summit, 25 Nov. 2008, Burlingame, CA.
4. 4. First Solar Q4 financial report, 2/24/09.
5. 5. Ref 1 ibid, p. 76.
6. 6. Ref 2 ibid.
7. 7. Handbook of Photovoltaic Science and Engineering, A. Luque and S. Hegedus Ed., 2003 (Wiley), Ch. 7.
8. 8. M.A. Green, “Silicon Solar Cells: Advanced Principles & Practice,” UNSW Press, 1995, Ch. 11.
9. 9. P. Hacke, et. Al., “Busbarless Emitter Wrap-Through Solar Cells and Modules,” 33rd IEEE PVSC, San Diego, CA (2008).
10. 10. C.J.J. Tool, et. Al, “17% mc-Si Solar Cell Efficiency Using Full In-Line Processing with Improved Texturing and Screen-Printed Contacts on High-Ohmic Emitters, 20th European PVSEC, Barcelona, Spain, 6-10 June 2005.
11. 11. Ref 8 ibid, Ch. 10.
12. 12. R. Swanson, 33rd IEEE PVSC, San Diego, CA (2008).
13. 13. T. Röder, et al., “0.4% Absolute Efficiency Gain of Industrial Solar Cells by Laser Doped Selective Emitter,” 34th IEEE PVSC, Philadelphia, PA (2009).
14. 14. S. Wenham and M. A. Green, US patent 6,429,037.
15. 15. A. Grohe, et al., “Laser Processes for the Industrial Production of High Efficiency Silicon Solar Cells,” Proceedings of the 22nd European PVSEC, Milan, Italy, Sept. 2007.
16. 16. V. Rana, Photonics West Technical Symposium, San Jose CA, January 2009.
17. 17. P. Englehart, S. Hermann, et. Al, “Laser Ablation of SiO₂ for Locally Contacted Si Solar Cells With Ultra-Short Pulses,” Prog. Photovolt. Res. Appl, 15, 6, p. 521-527, 2007.
18. 18. P. Englehart, et. Al, Prog. Photovolt. Res. Appl, 15, p. 237-243, 2007.

Peter Borden received his PhD in applied physics from Stanford U. This paper evolved from work he did while at Applied Materials. He is currently executive chairman of Wakonda and teaches classes on PV at Santa Clara U; [email protected].