A better approach to SiN removal is to select the laser source in such a way that the laser energy is absorbed in the coating directly.
A better approach to SiN removal is to select the laser source in such a way that the laser energy is absorbed in the coating directly.
Klaus Stolberg, JENOPTIK Laser, Optik, Systeme GmbH, Jena, Germany
SiN is a compound material that is widely used in the photovoltaics (PV) industry as a top and backside coating layer for silicon wafers [1]. By changing the thickness of the layer, it can be used as an efficient anti-reflection (AR) coating to reduce reflection losses at the front side of the solar cell. Because it is non-conducting, it also works as a passivation and isolation layer.
Figure 1. Commercial fs laser JenLas D2.fs by Jenoptik delivers 40µJ pulse energy with 100kHz at 1025nm. |
To establish a conducting link between the doped substrate (emitter) to the current contacts at the front side, or to the metallic contact layer at the backside of a solar cell, it is necessary to again open this AR coating after a full-area coating step that covers both the front side and the back side [2, 3].
Lasers are well-established tools in the PV industry to set up a thin film ablation process that is high-speed, non-contact, and easy to control. Because the intended AR coating is optically transparent to the solar spectrum, it is not easy to remove the SiN layer using conventional lasers. There is one way of indirect removal of SiN: if the laser is transmitted through the coating, but absorbed by the silicon, one can ablate the silicon substrate by increasing the laser intensity. In this case, silicon will be “ablated” from the wafer surface together with the SiN layer. However, the ablation quality using this method is very poor.
A better approach is to select the laser source in such a way that the laser energy is absorbed in the coating directly. This can be done by choosing the correct wavelength or by using non-linear effects. The correct wavelength is UV, where SiN shows strong absorption. Nonlinear absorption would imply that laser pulses are used that are short enough to reach a peak power density of >108W/cm⊃2;. Some free electrons in the SiN coating can serve as initial absorbing centers—the number rapidly growing by avalanche ionization. This can be effectively done by using a ps or fs pulse length.
It is known from other investigations that laser processing with ns pulses will cause very strong heat effects in the samples. By using ns laser pulses, the ablation process is “classical,” i.e., the progression is: temperature rise—melting—evaporation. Within this ns timescale, there is strong heat conduction into the bulk. This generates cracks, tensions, and crystalline disorders in the silicon bulk. Such changes in microstructure will act as recombination centers for the photoelectrons, and cell efficiency will therefore decrease.
It is also known that there is a change-over of the ablation mechanism in the timescale below 10ps. Here we have very rapid energy transfer from the laser photons to the electrons, followed by energy transfer from the electrons to the lattice. From this point, heat conduction – which is a lattice effect – will start. It can be assumed that there will be a change in the ablation mechanism and thus quality, if the pulse duration comes into the range when heat conduction is just starting.
The value (of pulse duration) depends on the bulk material (silicon) and its purity and lattice defects; it is normally between 1ps and 10ps. For this reason, we have investigated SiN ablation at different silicon substrates with both a fs and a ps laser. Because frequency-doubling and multiplying is expensive and power-consuming, we have favored an IR wavelength laser. The laser was focused by a Galvoscanner system to maximise processing speed; the focus was 254mm, which allows the process of a complete standard wafer size of 156×156 mm⊃2;.
Femtosecond ablation
As a laser source we used the Jenoptik JenLas D2.fs (Fig. 1), a regenerative laser amplifier with a fs fiber seed oscillator. This gives a maximum pulse energy of 40µJ with a repetition rate of 100kHz and a fundamental mode beam quality at an emission wavelength of 1025nm. This parameter set is a very good fit relative to the ablation threshold of SiN, as well as to the scanning speed of the Galvo, which is >5m/s in this case. By using this scanning speed, single shots can be resolved as separate ablation spots.
Figure 2. Ablation curve for fs and ps ablation. 50–60nm SiN coated at c-Si can be ablated with a single laser pulse, resulting ablation diameter is 60µm. |
Starting with a 50–60nm SiN layer coated on crystalline silicon (SiN:c-Si, polished), we have measured ablation diameters up to 62µm (Fig. 2). Melt-free ablation works up to a laser fluence of 0.35J/cm⊃2;; at larger values, melting can be observed with a microscope. We are therefore limited to a melt-free diameter of 60µm. There is a very sharp boundary of the ablation spot, showing strong threshold behavior of the ablation.
Figure 3. Microscope image of ablation of SiN layer at etched c-Si substrate. Etched microstructures (lattice planes) are also preserved under fs processing; no detectable melt. |
An etched crystalline substrate coated with SiN was also investigated. Figure 3 shows the preservation of microstructures (pyramids formed by silicon lattice planes): microstructures can be observed in the coated as well as in the ablated area. If melting had occurred, the microstructures would have disintegrated.
In a second series of trials, we also investigated an ablation of a SiN layer on a multicrystalline silicon substrate (SiN:m-Si), which is also an important substrate material for a PV cell. The measured ablation curve is similar to the crystalline curve of Fig. 2. The maximum diameter for a 40µJ laser pulse is a 65µm ablation spot diameter and the limit for a melt-free condition is 0.18J/cm⊃2; at a 52µm diameter. Preservation of surface microstructure is demonstrated, which is a strong marker for very low damage.
Picosecond ablation
For our evaluation of ps ablation, we investigated SiN ablation with 10ps pulses and up to 100µJ at 1064nm. We have found that it is necessary to use a higher laser fluence for ps ablation of SiN:c-Si to reach the same ablation diameter as with fs pulses, though melting limits the ablation diameter to 45µm. Even for multicrystalline SiN:m-Si, a larger laser fluence is required to get the same ablation spot diameter than if a fs laser is used (Fig. 2).
Comparing ablation results at 0.25 J/cm⊃2; fluence, we see a “washed-out” boundary of the ablation spots with the ps laser (Fig. 4). Our explanation for this is anisotropic heat conduction. Anisotropy is caused by lattice defects (grain boundaries of the polycrystalline silicon). Together with other reports of melting effects of ps processing [4], this gives the first hint of the advantages of fs processing. A more detailed investigation of the influence of lattice damage on cell efficiency is already underway.
Robustness of the ablation process is a challenge to set up a scribing machine to do laser ablation. To check this, we moved the sample out of focus and then measured the ablation spot diameter. The result was that the depth-of-field is 1200µm, i.e., a change of the ablation spot diameter of ±5% (Fig. 5). This will help to prevent a dynamic focusing system because this depth-of-field is far above the specified flatness of the wafers and of the topology of etching structures, which can reach a budget of 50µm in total.
| Figure 4. Microscope images of SiN layer at multicrystalline silicon; ablation performed with 400fs (lower part) and 10ps IR lasers (upper part) at 0.25J/cm⊃2;. |
Conclusion
We have demonstrated reliable SiN layer ablation with a commercial IR fs laser by Jenoptik. Laser fluence can be tuned to about 50µm ablation spots for crystalline silicon without detectable melting. We have also demonstrated this process on other substrates: etched crystalline and non-etched multicrystalline silicon. Separate ablation spots can be generated with a 100kHz repetition rate at a process speed of >5m/s. Use of an IR ps laser causes higher laser fluence to generate an identical ablation spot size with lesser quality (melting, boundary sharpness).
Figure 5. Depth-of-field measurement of 50–60nm SiN coated at c-Si. The process is robust (DOF is 1.2mm), which reduces efforts needed for focus distance control. |
Further investigation of the influence of both fs and ps ablation on cell efficiency is required, as well as investigation of additional ps laser wavelengths.
Acknowledgment
JenLas is a registered trademark of Jenoptik.
References
- 1. B.Sopori, “Silicon Nitride Processing for Control of Optical and Electronic Properties of Silicon Solar Cells,” Jour. of Electronic Materials, Vol. 32, Nr. 10, 2003, pp. 1034-42.
- 2. F. Book et al., “Two Diffusion Step Selective Emitter: Comparison of Mask Opening by Laser or Etching Paste,” 23rd EC PVSEC, Sept. 1-5, 2008, Valencia.
- 3. K. Neckermann et al., “Local Structuring of Dielectric Layers on Silicon for Improved Solar Cell Metallisation,” 22nd EPSEC, September 3-7, 2007, Milano.
- 4. V. Rana, “Laser Processes for Advanced c-Silicon Solar Cells and Large Area Thin Film Solar Panels,” laser + photonics conference, April 22-23, 2008, Fellbach/Stutttgart.
Klaus Stolberg received his degree in physics at the U. of Jena, German,y and is applications manager, industrial lasers, at JENOPTIK Laser, Optik, Systeme GmbH; [email protected].
