Klaus Stolberg, Susanna Friedel, Jenoptik Laser GmbH, Jena, Germany
Observing the PV market today, we are faced with two major technologies: traditional multicrystalline Si-based cells and thin-film PV cells. Multicrystalline Si-based cells have high efficiencies on the order of 16%, but the cost level is nearly twice as much as thin-film PV cells. The latest studies  show that crystalline Si cells clearly dominate world production of solar cells with a market share of ~77%. So there is much pressure for cost reduction in this market because production volume is very attractive and growing.
There are innovative cell concepts for silicon cells that can help to meet this challenge, however, there are some problems to be overcome. For example, a significant loss factor occurs when using front contacts that reduce the effective photosensitive area available on the front side of the solar cell. But so-called backcontact solar cells “wrap” the contacts to the backside. To achieve back contact cells, micro-vias are drilled into the silicon wafer, which are used to guide the contacts (metal filling) or the emitter to the backside (doping). Thus, the back-contact can be accessed from the backside as well as wrapped front-side contacts.
Metal-wrap-through (MWT) (Fig.1) technology , which is closer to production, has different factors that enhance the cell efficiency: 1) bus bars can be located at the backside to reduce shadowing, 2) bus bars can be designed broader to reduce ohmic losses, and finally 3) all connection steps can be processed from the backside. Because a front contact pattern can take up to 10% of the complete area, a total efficiency gain of about 0.5% is expected.
|Figure 1. The new MWT cell concept “wraps” front contacts to the cell backside. Complete backside wiring is a technological advantage because contacts can take larger area at backside to reduce ohmic losses.|
In modern PV cell production, a throughput of 1800-2400 wafers per minute is state-of-the-art; this creates hard requirements on the drilling time to create the through-holes. Laser technologies are favored for this task.
Because of the required tact time (or cycle time), the processing time for a default 156 x 156mm² wafer is only 1-2s. Within this time period for MWT, about 100 via holes with a diameter of 50-100µm through the silicon wafer, must be drilled. Fundamental mode lasers with a laser pulse length in the range of nanoseconds (ns) and pulse energy in the range of millijoules (mJ) are well suited for this. Usually the pulse duration and pulse repetition rate of the laser source are linked to each other, and thus both parameters cannot be tuned independently. In this paper, we present application results processed with a new kind of IR laser source, for which the tuning is not limited in that manner.
Compared to laser drilling, traditional wet etching technology is not fast enough, i.e., the typical etching rate is ~3.8µm/min (30% KOH, 100°C)  and additional process steps (masking) are required. Therefore, the estimated process time for a 200µm-thick wafer will be in the range of 1 hour. Furthermore, the disposal of the by-products of wet etching technology are an environmental problem and prevention of them is highly desirable.
It is estimated that if one replaces 1.6mm-wide front side bus bars with about 100 point contacts of 200µm in diameter, the result would be a reduction of front side area consumption from 5.3% to 2.9%. If this directly contributes to the cell efficiency, it would increase cell efficiency by 0.3-0.4 %. Indeed, an enhancement of c-Si solar cell efficiency from 0.5% to 16.5% was reported recently using MWT technology .
Using a laser to drill the holes is favored because it introduces energy selectively at the drilling position without any mechanical contact of the drilling tool. Because there is no tool wear-out and laser emission can be controlled precisely, the laser drilling process is very reproducible—a demand of factory automation. It is known that when ablating silicon, the ablation rate can be optimized if the laser pulse duration is in the range of hundreds of nanoseconds .
For a wavelength of 1030nm, the absorption coefficient in silicon is 4.5 x 10-3µm-1 which gives an optical penetration depth of ~200µm . Within this dimension, 63.2% of the incoming laser intensity is absorbed (reflection losses at the surface are neglected). The typical wafer thickness for PV applications now is 180-300µm, therefore, energy from an IR laser will be absorbed efficiently inside the complete wafer thickness, resulting in efficient thermal ablation.
We also know from further trials  that the ablation rate will increase for a pulse duration of more than 500ns because optical absorption is temperature-dependent. For “large” pulse lengths of hundreds of nanoseconds, there will be a significant temperature rise in the leading edge of the laser pulse and energy absorption in the trailing edge will therefore be enhanced.
To conduct the work described in this paper, we used a JenLas disk IR70 laser, which is a power-enhanced version of another laser already introduced elsewhere . The laser works with a proprietary active control mechanism for stable pulse generation and independent tuning of pulse duration and pulse repetition rate. Usually, both parameters are linked to each other. This laser also allows tuning of the laser pulse duration over a large range between 0.2 and 2 µs and offers the opportunity to find optimized drilling parameters.
Further characterization  shows symmetric beam profile, fundamental-mode beam quality and absence of astigmatism. This laser is equipped with a fast beam shutter that allows the gating of the pulse train down to a single pulse and enables the emission of burst-like pulse trains with free control of timing.
For the application trials we used optical setup depicted in Fig. 2. We aligned the laser (1) through a beam expander (3) into a Galvoscanner (4) for fast beam deflection. The Galvoscanner was equipped with 254mm focussing optics (5) to cover a working field of 160 x 160mm². Finally, the sample (6) was fixed to a support structure (7). A Galvo controller board (2) was used to synchronize the Galvoscanner with the laser.
|Figure 2. Application setup for high-speed silicon drilling using a Galvoscanner.|
The traditional way of drilling holes into the silicon wafer is as follows: the Galvoscanner’s mirrors are accelerated to the desired target position at the wafer surface, the mirrors stop their motion and the position control loop is activated. Then the laser is triggered to fire a pulse burst, which contains as many laser shots as required for drilling through the wafer. The disadvantage of this “point-and-shoot” method (Fig. 3) is the large number of time consuming start-stop operations for the Galvo. As an example: for 3 rows with 33 holes, there are 99 start-stop procedures required.
|Figure 3. Comparison of “point-and-shoot” (left); and “on-the-fly” drilling mode (right). The number of start-stop procedures is significantly reduced for the “on the fly” method.|
A smarter method is “on the fly” drilling (Fig. 3). In this case, the scanner mirrors are moving continuously from one side of the sample to the other. The laser is triggered by the motion controller at the “correct” position, i.e., when the mirror points to the desired wafer location. Thus, only three start-stop procedures are necessary for one complete scan of the wafer. We have measured a minimum number of two shots for achieving a through-hole in a 200µm-thick silicon wafer; this means that a total of six start-stop operations are sufficient, which is a significant reduction in effective drilling time.
Because there are multiple laser shots required for a through hole (a minimum of two), the scanning operation in the attractive “on the fly” mode must be repeated multiple times. The firing position for all scans must be exactly the same to match the drilling pattern of the earlier scans. This has to be ensured not only by the Galvoscanner, but also by the laser, which means the internal delay time between the trigger pulse of the Galvo controller and emission of the laser pulse must be constant; additionally, pulse energy for each laser shot must be equal/constant.
In a former publication  we have demonstrated a minimum number of 15 shots to drill through a 200µm thick silicon wafer, requiring a total of 1.9ms. With the modified laser we have been able to reduce this value to a minimum number of five shots (254mm focal length). The laser can emit the required pulse energy with a repetition rate of 50kHz, giving a theoretical drilling rate of up to 10000 holes per second (entrance diameter 65µm, exit 28µm). Whereas for the “point-and-shot” method, the effective drilling time per hole is ~10ms, so we can reach 100µs “on the fly.” In reality, these values are slightly higher, because there is an additional amount of motion turnaround time consumed by the Galvo.
After changing the focal length to 100mm, finally two pulses are sufficient to drill through 200µm of silicon (entrance diameter 43µm, exit 26µm) with a theoretical drilling rate of 20000 holes per second. In this case, one has to take into account that the working field of the Galvo is drastically reduced to 60 x 60mm².
In Fig. 4, the shape of laser drilled holes is shown as well as the shape of the hole after post-processing. In the PV industry, laser drilling is usually followed by an etch step to remove lattice damage caused by the thermal character of laser drilling, as well as to remove some melting residuals at the walls. Thus, the hole diameter is extended by another 10-20µm. Additionally, hole roundness – especially for monocrystalline wafers – may be influenced because of anisotropic etching velocity relating to the crystal orientation. Thus, hole quality must be evaluated after performing all relevant process steps.
Figure 4. Comparison of a pure hole just after laser drilling (left, entrance side) with another hole after the next process step (right, after etching). Etching widens the hole diameter by 10-20µm and affects “roundness.” Note the different scaling: i.e., left, the hole diameter is 63µm; right, the hole diameter is 66µm.
JenLas is a registered trademark of Jenoptik Laser GmbH.
1. www.nanomarkets.net, based on “Materials for Thin-Film Silicon Photovoltaic,” NanoMarkets, 3/25/2009.
2. A. Grohe, G. Kray, “Strukturieren und Dotieren von kristallinen Siliziumsolarzellen mit dem Laser” (in German), Proc. Laser + Photonics Conference, Fellbach/Stuttgart, pp.155-168, 2008.
3. H. Seidel et al., “Anisotropic Etching of Crystalline Silicon in Alkaline Solutions,” J. Electrochem. Soc. Vol 137, 11, pp. 3612-3632, Nov. 1990.
4. F. Clement et al., “Processing and Comprehensive Characterisation of Screen-printed mc-Si Metal Wrap Through (MWT) Solar Cells,” Proc. 22nd European Photovoltaic Solar Energy Conference, Milano, pp.1399-1402, 2007.
5. M. Leitner, E.Körner, ”Vereinzeln dünner Silizium-Wafer mittels innovativer Lasertechnologien” (in German), Proc. Stuttgarter Lasertage, Stuttgart, pp.119-122, 2003.
6. A. Grohe et al., “Laser Processes for the Industrial Production of High-efficiency Silicon Solar Cells,” Proc. 22nd European Photovoltaic Solar Energy Conference, Milano, 2007.
8. M. Leitner, K. Stolberg, “Innovative gepulste Scheibenlaserkonzepte – vom Nanosekunden- bis zum Femtosekundenbereich” (in German), Proc. 6. Jenaer Lasertagung, Jena, pp. 20-30, 2008.
9. K. Stolberg et al., “Systematic Optimization of Process Parameters in Laser Drilling of 200µm Photovoltaic Silicon Wafers Using New Kind of Nanosecond IR Lasers,” JLMN Vol. 4, No. 3, p. 231-233, 2009.
Klaus Stolberg received his degree in physics at Friedrich Schiller U. Jena/Germany, and is Head of Laser Applications Lab at Jenoptik Laser GmbH, Goeschwitzer Str. 29, 07745 Jena, Germany; ph.: +49-3641-654334; email [email protected].
Susanna Friedelreceived her degree in engineering at U. of Applied Sciences in Jena/Germany and is laser applications engineer at Jenoptik Laser GmbH.