Ronald D. Schaeffer and John O'Connell
Beam quality, focus, depth of focus and pulse repetition rate are important factors to consider when processing on a micro scale
It is fair to say that the CO2 laser is the fastest laser for materials processing applications, while UV lasers give better edge quality and higher resolution features, especially in plastics. From a cost-of-ownership perspective, it is almost always a more economical route to use a CO2 laser if the desired end result can be obtained. Compact sealed CO2 lasers eliminate gas handling and they operate in excess of 10,000 hours without maintenance or refurbishment. In general, however, the photon material interaction with plastic materials is sometimes not desirable because the first order interaction is via a thermal process, which sometimes causes charring, melting or burning. Therefore, UV lasers are used to make parts for the best edge quality at the sacrifice of speed and cost to manufacture.
Among the important factors in materials processing applications using lasers are edge quality, aspect ratio and taper of features, achievable spot size (or kerf width for cutting applications), processing speed and cost. Let's examine each in detail.
Edge quality-Are the edges smooth and straight? Is there a heat-affected zone (HAZ), and, if so, how far does it go into the material? These important factors affect processing speed and therefore cost. Most laser beams are 'round' or Gaussian, so straight cuts are made by overlapping lots of shot positions to form a continuous line. A benefit of using a round beam is that making different shapes and patterns is a matter of programming and not of changing part orientation.
Smooth edges can be affected by a number of things. The smoothest lines will generally be achieved using maximum pulse spot overlap. However, smoothness must be balanced with processing speed, because a 20-percent overlap will be much faster than say a 50-percent overlap. Ideally the edges should be crisp and clean, but this is rarely achieved, especially when cutting plastics with infrared (IR) lasers.
Normally, one sees some sort of heat effect-even when using the supposedly 'heat-free' UV lasers. The basic fact is that, even if the first order photon/material interaction is purely non-thermal, there are still thermal side effects that can take place because of heat deposition in the plume or because all of the photons incident on the target are not involved in the work of material removal and their energy is therefore dissipated as heat. This is more evident as the pulse lengths get longer and/or the pulse repetition rates get higher. The effect is greatly minimized in femtosecond machining because the pulse length is so short that most photons are used to create material removal and the tail is not as susceptible to interacting with the rising plume.
Aspect ratio and taper-The ILS definition of micromachining is that feature sizes are below 1 mm, usually on the order of a couple hundred microns or less. The same is true for material thickness. Most lasers used in micromachining are below 100 W, with the frequency-shifted solid-state lasers being less than 10 W. With almost any laser process, there will be a taper associated with the features. The entrance side is larger than the exit or bottom side. The taper angle is a function of the material, spot size, focal length of the lens and divergence of the beam and, finally and most importantly, the fluence of the laser light on target. All things considered, minimal taper is achieved with higher fluences.
The aspect ratio is the difference in material thickness with respect to hole size. As an example, a 1mil exit hole in 5 mils of material will give a 5:1 aspect ratio. One of the benefits of laser processing over, for instance chemical etching, is the ability to produce aspect ratios greater than 1:1. We usually feel comfortable with aspect ratios of 10:1 under almost any conditions, and we have done up to 100:1 in some cases (with the caveat that we are describing the exit diameter, while the entrance diameter is larger than the exit by some amount). Generally, but not always, customers are interested in low taper features.
Achievable spot size (kerf width)-Most important, the achievable spot size on target is a function of the laser wavelength of the laser light. IR lasers have larger minimum achievable spot sizes than ultraviolet (UV) lasers because of the longer wavelength. Optical set-ups also come into play. Expanding the incident light and filling the optical aperture of a fast, fixed lens gives spot sizes on the smaller side, while under filling an aperture gives larger spot sizes. Also, if galvo fields are used, then the minimum achievable spot size is related to the field size. Having said this, the question is whether a minimum spot size is always desired. Smaller spot size may minimize the heat-affected zone, but a smaller spot has to be scanned for a longer period of time with more overlap to make a straight-line cut. So for the purpose of speeding up the process, compromises are usually made.
Processing speed/cost-Speed and capacity directly correlate to cost. IR lasers are faster than UV lasers, galvo beam deliveries are faster than simple table motion and higher-energy and/or higher-pulse-repetition-rate lasers are faster than smaller lasers. In terms of operating costs, sealed CO2 and diode-pumped UV lasers cost less than $10 per hour to operate, not including capitalization, labor or overhead. Very high galvo speeds are possible using small input apertures and mirrors, but there is a limitation on spot size with the smaller apertures. Also, higher speeds do not give the best positioning accuracy. Therefore, sometimes compromises in design must be made so that the best combination of speed, accuracy and spot size is achieved.
Solid-state lasers at about 1 µm are widely used in industry except for plastic processing. Therefore, we can concentrate on the frequency-shifted wavelengths. The doubled wavelength (532 nm) is useful for thin film removal, especially from visibly transparent materials like glass and Mylar. Some work has been done with this wavelength on plastics, especially in cases where small spot sizes are desired, coupled with fast processing speeds. For example, we produced great holes with a CO2 laser, but the job ran on a web that was 18 in. wide and, with a galvo field of this size, the 500µm spot size on target was not sufficient. The best holes were made with the 355nm wavelength, but the process was too slow. So, we compromised and used the 532nm laser in order to get about 100µm spot size on target, reasonably good hole quality and fast processing speed.
At present 355 nm is the wavelength of choice for processing of plastics (and other materials like Cu, which cannot be directly processed using the CO2 laser). The UV wavelength couples nicely with most materials. The achievable spot size-even for 12-in. galvo fields-is about 25 µm (less for smaller fields), and the laser can be used at high repetition rates. Commercially available 7W and 10W lasers are now common. These lasers are small, easily integrated and maintained. Lasers around 1.5W output have long lifetimes with infrequent changing of the crystal site. Lasers greater than 3Ωrequire changing crystal sites every thousand hours, but there are typically 25 crystal sites to use. Interestingly, doubling the laser power increases processing speed by a factor of 3 to 5 instead of 2. In cases where the 355nm laser does not couple well, the 266nm frequency-quadrupled laser can also be considered. This laser has less output power and requires more care and maintenance than the 355nm laser, yet it can provide interesting complementary capabilities.
Compact, sealed RF CO2 lasers never require alignment or optics cleaning and have good M2 values-meaning good focusability. Standard broadband output gives emission lines from about 9 to 11 µm wavelength, so it is not a strictly monochromatic laser. However, many plastics absorb quite highly around 9.3 to 9.4 µm, and it is possible to filter the laser output such that other emission lines are suppressed and output is mostly confined to this narrower region. The total output of the laser goes down by about 20 percent, but the usable power is probably greater. Processing speeds for the line-narrowed lasers are essentially the same as for broadband lasers, but the quality is better.
Another type of CO2 laser is the TEA laser, a pulsed laser with a large, non-circular output beam. This laser is used in an imaging mode and has a pulse length much shorter than the RF lasers can achieve. Therefore, the quality of the cut is very good, especially in the case of using a line-narrowed 'T' laser. The lasers are only pulsed at a few hundred Hertz, so processing speed may be limited.
A final type that is new to the market is the Q-switched CO2 laser that has a very short pulse length of about 120 ns, which makes it quite interesting for materials processing applications where HAZ is an issue.
To demonstrate the speed and quality of laser cuts, we used seven different lasers-four CO2 lasers and three solid-state UV lasers-on two different materials-2mil-thick Mylar and Kapton. Figures 1 through 7 show both materials for each laser type used. The part that we make on a daily basis in high volume is a circle about 1 in. in diameter and contains some cutout features in the central portion of the circle. In each case 'a' is the Mylar and 'b' is the Kapton, and it is fair to point out that in general the polyimide results are better than the Mylar for any particular laser because the polyimide absorbs better at all laser wavelengths used. All samples are shown as from the laser with no post cleaning. All parts were made using galvanometer-based beam delivery, and processing speeds represent the speed at which the best part quality was produced, not in general the maximum speed possible.
Figure 1 shows the 1.5W, 355nm laser results. Note that the cuts in both materials are relatively clean with a small HAZ. Using this laser it took about one minute to make a complete part. Figure 2 shows the 3.0W, 355nm laser results. There is a slightly larger HAZ, but still the cut quality is pretty good. Using this laser, the part took 18 seconds to make, more than three times faster with a two-times increase in laser power. Figure 3 shows the 1.5W, 266nm laser results. Note the exceptional cut quality. Surprising, this part took only 12 seconds to make.
Next let's look at the CO2 lasers. Figure 4 shows the broadband RF CO2 laser results. As expected, there is a large HAZ, especially evident in the Mylar. The part took about 400 ms to cut-very fast but not good quality. Figure 5 shows the 9µm, RF CO2 laser. The cut quality is better than with the broadband laser, but still not as good as the UV lasers. The cut time was also about 400 ms. Figure 6 shows the results using the line-narrowed CO2-T laser.
Figure 7 shows the Q-switched CO2 results. These results are very interesting, as the cut quality is nearly as good as, or perhaps better than, the UV lasers. The processing speed was about 700 ms. We are interested in continuing to evaluate this laser as we see a lot of potential. Note that we have not used the excimer laser because, even though the cut quality would be pretty good, the time to make these parts would be on the order of minutes, not seconds, and that is not cost effective.
In conclusion, there are many lasers used for material processing and in particular for processing plastics. We have compared several different lasers with respect to processing speed and feature quality on two different materials. These results can be used as a guide to determine which laser of those available might be the best choice for new applications.
Ronald D. Schaeffer and John O'Connell are partners in PhotoMachining Inc. Pelham, NH. They can be contacted at Tel. (603) 882-9944.