Santa Clara, CA - Near-infrared Nd:YAG lasers have been widely used for drilling alumina and for soft and hard marking and cutting of silicon. The Synova laser/waterjet process has been cutting silicon for some years. So why do so many in the industry believe that only doubled, tripled, or even quadrupled wavelengths are suitable for cutting silicon?
There are three well-known problems with laser cutting silicon; debris, thermal damage, and throughput. Multi-pass cutting has been developed for thin wafers using diode-pumped solid-state (DPSS) lasers but cutting thicker than 0.2 mm causes problems. Various schemes are available to improve the situation; active gases or various water-assist schemes are being used. A novel process known as ‘stealth dicing’ is also being developed but poor flexibility similar to that in the glass cutting or cleaving process using CO2 lasers is expected.
All laser machining processes are to some degree thermal, even the strongest exponents of femtosecond lasers will agree with that, if conditions are right. When this basic premise has been accepted, then the approach to reducing thermal effects when laser cutting is clear-reduce heat input per unit length.
Reflectivity decreases rapidly with increasing temperature. Combine this with the very high brightness of CW fiber lasers, small spot sizes, stability, and pulse energy control and we are part way to understanding recent results from the lab at SPI Lasers.
Cutting of 0.2mm-thick silicon using a 50W fiber laser may initially seem contradictory when compared to using a more costly frequency tripled 10W DPSS laser. But it has been shown that it is possible to single-pass cut five times faster with the fiber lasers than is possible with a multi-pass cut using DPSS, and this suggests that the heat input per unit length is approximately the same. If the amount of heat ejected through the kerf with the super-heated dross or molten ejecta is considered, then the heat input per unit length is likely to be much less.
SPI Lasers now has results that confirm this; 0.2mm-thick silicon can be cut crack free with a 50W air-cooled fiber laser in a single pass at up to 6 m/min. Using a 200W laser, up to 1.4 mm thickness has been cut crack free at 0.7 m/min. Cut surfaces compare favorably with the laser/water-assist process. The fiber laser process is robust, simple, and highly repeatable with a large operating envelope and takes advantage of a number of the intrinsic properties of semiconductor materials. All surface conditions can be cut at similar speeds. The process uses no liquids and off-the-shelf optics. Capital and running costs will be only a small percentage of DPSS or waterjet based systems.
If you want to debate this, contact Dr. Tony Hoult (email@example.com), applications manager, SPI Lasers.