Interconnect formation typically requires three scribing operations known in the industry as P1, P2 and P3. A fourth operation, often referred to as P4, removes shunts at the edges and prepares the module for electrical interconnect and encapsulation. Laser scribe processes based on nanosecond (ns) pulsed lasers are preferred for both a-silicon and CdTe thin film PV manufacturing. CIGS (copper indium gallium selenide) is a different story. In this case, the shunt and series resistance of laser processed modules suffers unacceptably due to thermal degradation when traditional ns pulse lasers are used.
Success has been claimed utilizing picosecond (ps) lasers for CIGS P2 and P3, however, negative thermal effects still cannot be avoided as is evident in Fig. 2a. Furthermore, such lasers are much more expensive, much larger and considered far less industrially robust than ns lasers. Consequently, they have yet to gain broad acceptance across the solar industry. Because of these concerns, mechanical stylus scribing continues to be the industry norm for CIGS P2 and P3 scribing.
The tailored-pulse laser
The invention of a new type of fiber laser has enabled discovery and exploitation of a previously unknown process window for the CIGS P2 and P3 process. The laser is a 1064nm pulse programmable fiber laser [1]. Unlike more traditional lasers, this laser technology allows the pulse duration to be varied from approximately two to several 100s of nanoseconds, independent of the laser repetition rate, which can be varied up to 500kHz. In addition, each pulse can be arbitrarily programmed to generate a specific desired temporal profile of instantaneous laser power. Pulse trains comprised of these shaped pulses can be applied to the scribing process at high repetition rates. It is precisely these attributes that are not available in Q-switched or ps lasers that enable this novel and effective process.
|
||||||||||||
|
Table 1. Transmission of several, various CIGS films. |
Laser-CIGS material interaction
Understanding the laser-material interaction itself is key to uncovering why this novel process window has previously proven inaccessible using existing laser sources. For a typical CIGS film, the band gap corresponds to optical wavelengths around 1030nm. Because its photon energy is just below the band edge, when an individual 1064nm laser pulse first interacts with the CIGS film, the film is significantly transparent to it. We have confirmed this transparency for several CIGS films from various sources as shown in Table 1.
 |
|
Figure 1. SEM showing MoSe2 between the CIGS and Mo layer. Used with permission. [The company asked not to be identified] |
Because of this transparency, a large fraction of the laser light passes thorough the CIGS layer to the underlying molybdenum (Mo) layer. As evidenced by the resulting morphology of the Mo surface, substantial heating of the Mo layer occurs with local temperatures that can approach the melting point of Mo (Fig. 2b). These very high local temperatures set two competing processes in motion.
Explosive evolution of vapor
First of all, between the Mo layer and the CIGS, there is typically a thin layer (~100Å) of MoSe2 (Fig. 1) that serves as an interface layer between the CIGS and Mo layer [2]. MoSe2 has been shown to decompose without melting at 1170°C [3], well below the melting point of Mo at 2617°C. Furthermore, other work has shown that under intense laser irradiation, CIGS begins to decompose and that the first component to leave is selenium [4]. We postulate that this decomposition of the interface layer (and potentially of the CIGS material itself), and the resultant rapid evolution of selenium vapor, explosively drives the CIGS layer off the substrate surface before the CIGS layer becomes hot enough to melt.
 |
|
Figure 2. Morphology of various CIGS scribes under different scribe conditions: a) Typical P3 scribe with picosecond laser at 515nm wavelength (used with permission); b) SEM image of a tailored-pulse scribe in CIGS at higher than optimal pulse energy. Dark regions in center indicate areas where the Mo has reached melting point; c) Optical image of scribe when peak fluence is too high; d) SEM image of an optimized, tailored-pulse scribe; e) SEM image of tailored-pulse scribe when the pulse duration exceeds the critical time, τ; f) SEM image of tailored-pulse scribe with less than optimal energy. |
While we have no direct evidence that it is vaporization of the MoSe2 layer that drives the process, a simple calculation based on the ideal gas equation that assumes complete vaporization of the selenium suggests that decomposition of the MoSe2 layer alone would generate adequate pressure to explosively remove the overlying layers. Additionally, EDX analysis also indicates nearly a complete absence of Cu and In from the groove, which suggests that the motive force originates below the CIGS layer.
The end result is the explosive removal of the overlying material in the solid phase in a brittle fracture manner [5]. Experiments conducted near the process energy threshold show that indeed, the CIGS layer is removed in the solid state as an intact layer. Higher energies cause the material to fragment and to be completely ejected as smaller flakes. Fig. 2f shows the lifting of the film at low energy.
Parasitic thermal conduction
The second process put in motion by the laser pulse is thermal conduction into the CIGS layer from the hot Mo layer. Modeling of this conduction indicates that it occurs on the time scale of a few ns [5]. This conductive heating of the CIGS layer in turn has two effects. First, the band gap of the CIGS material rapidly red-shifts, thus increasing the absorption coefficient of the CIGS at the 1064nm laser wavelength [6]. This means that the absorption of the laser light transitions from being primarily in the Mo to being primarily in the CIGS. Secondly, the mechanical properties of the film change as it approaches its melting point. At these high temperatures, the film transitions from being brittle to being ductile. For long pulse durations, the film will actually melt.
 |
|
Figure 3. Modeled dependence of CIGS film transmission on pulse duration for a tailored pulse (red) and a generic Q switched pulse (blue). The dotted lines show the actual temporal pulse shapes, and the solid lines show how much of the pulse penetrates to the CIGS/Mo interface. |
Therefore, there is a critical time, τ, within which enough energy must be delivered to the Mo layer to initiate the explosive release of selenium, without allowing sufficient thermal conduction back into the CIGS to raise its temperature sufficiently to initiate the undesirable mechanisms introduced by significant heating of the CIGS. Our experiments show that this critical time τ is approximately 5ns. Fig. 2e shows the morphology that results when the pulse width extends significantly beyond t, and stands in contrast with Fig. 2d, which shows the result obtained using a similar pulse energy but with a pulse width that does not exceed τ.
Picosecond pulses: excess peak power
Finally, we observe that if the peak fluence is too high, the mechanism of Mo heating and brittle fracture is completely absent. The laser energy is deposited entirely at the CIGS surface as shown in Fig. 2c. We hypothesize that this is caused by the non-linear threshold for absorption in CIGS. This behavior occurs at a peak fluence about five times higher than that where the brittle fracture process dominates. By extension, we believe that this suggests a lower limit in pulse duration of about 1ns below which the peak power is so high that non-linear absorption prevents the laser pulse from reaching the Mo layer. This is likely the reason this brittle fracture process has not been reported with ps lasers.
The brittle fracture process window
The behavior described above therefore defines both temporal and fluence process windows within which the laser pulse must fit for successful brittle-fracture ablation. Ideally, to meet these requirements, all of the required energy must be delivered within a pulse not exceeding 5ns duration, but also not less than 1ns, and with a temporal pulse shape that would be close to rectangular.
With this new understanding of how this “brittle fracture” ablation process unfolds, we can now understand why ns and ps lasers have not successfully addressed this application to date. Pulse durations have been either too short, too long, or parasitic energy in the pulse tail has spoiled the process. To illustrate this point, Fig. 3 shows a simulation of two pulses of equal energy and full-width at half maximum (FWHM), one tailored to a square temporal shape, and one typical of a Q-switched laser (dashed curves). Heating of the CIGS induced by thermal conduction shuts down the transmission of light to the Mo interface after about 5ns resulting in less energy delivered to the desired process and more energy dumped into parasitic heating of the film (solid curves).
 |
|
Figure 4. Sample of P3 (right) scribe as applied to the NREL mini-modules. Note the P1 scribe (left) and the P2 scribe (middle). The P1 scribe appears dark grey due to the CIGS layer deposited over it. The P2 scribe has a bluish appearance due to the presence of the AZO layer that was deposited between the P2 and the P3 steps. |
Application
The tailored-pulse process was applied in the production and test of four monolithically integrated CIGS mini modules for which we also developed a tailored-pulse P1 process. For direct comparison, two of the mini modules were produced with the P2 and P3 scribes being mechanically scribed by a manufacturer of widely-used scribing equipment. The performance of modules produced entirely with tailored-pulse laser scribes was found to be as good as, or better, than equivalent modules produced with the industry standard mechanical scribing process. A thorough review of the laser process and the production and test of the mini modules can be found in reference [5]. Fig. 4 shows the morphology of the tailored-pulse P1, P2 and P3 scribes on a mini module, and Table 2 shows the performance of the mini-modules compared to their mechanically-scribed counterparts.
|
||||||||||||||||||||||||||||||
|
Table 2. I-V parameters as measured at NREL under standard test conditions (STC). |
Conclusion
CIGS material has, until now, resisted the development of a viable laser scribe process for the P2 and P3 steps. Utilizing time domain tailored pulses, we have achieved excellent process results with performance at least comparable to that of the existing mechanical scribing process. Utilizing the properties of the tailored-pulse laser we have been able to study and determine the limits of the temporal process window for ablation of CIGS. Our work indicates that this temporal process window extends from about 5ns down to about 1ns. Finally, we have demonstrated the viability of the tailored-pulse ablation process by producing mini-solar modules and demonstrating similar performance compared to equivalent modules that are mechanically scribed.
Acknowledgments
The authors would like to acknowledge the contribution of JENOPTIK Automatisierungstechnik GmbH to this paper, and in particular to thank Dr. Gabriele Eberhardt and Dr. Torsten Reichl for their support. All process development work and metrology was performed at National Research Council (NRC) in London, Ontario, and process development samples were provided by the National Renewable Energy Laboratory in Golden, Colorado [5].
Mathew Rekow received a BS in physics and electrical engineering from the University of Idaho and is a Principal Applications Engineer at PyroPhotonics Lasers, a subsidiary of ESI, Inc., 48660 Kato Road, Fremont, CA 94538; ph.: 831-332-4380; email: [email protected].
Richard Murison received his BSc in physics from Loughborough University of Technology, UK, and is the Chief Technology Officer and Director of Product Marketing at PyroPhotonics Lasers, a subsidiary of ESI, Inc., 275 Kesmark , Dollard-des-Ormeaux, QC Canada H9B 3J1; ph.: 514.904.9000; email: [email protected].
References
1. R. Murison, US Patent 7742511 2008.
2. D. Abou-Ras, “Formation and Characterisation of MoSe2 for Cu(In,Ga)Se2-based Solar Cells,” Thin Solid Films, 1 June 2005, pp. 433-438.
3. Y.C. Lee, “Temperature Dependence Anisotropic Photoconductivity in 2H-MoSe2 Single Crystals,” Journal of Alloys and Compounds, January 2008, Vol. 448, 1&2.
4. P.O. Westin, “Laser Patterning of P2 Interconnect Via in Thin-film CIGS PV Modules,” Solar Energy and Solar Cells, Oct. 2008, Vol. 92
5. M. Rekow, R. Murison, C. Dunsky, C. Dinkel, J. Pern, L. Mansfield, T. Panarello, S. Nikumb, “CIGS P1, P2, P3 Scribing Processes Using a Pulse Programmable Industrial Fiber Laser.” EU PVSEC Proceedings, Sept. 2010
6. Band Gap. wikipedia.org. [Online] http://en.wikipedia.org/wiki/Band_gap, Wikimedia.org, Jan 9, 2011. [Cited: Jan 19, 2011] http://en.wikipedia.org/wiki/Band_gap
