UV laser micromachining: Q-switched or mode-locked?
a recent comparison study shows that the answer depends on the particular material being processed
Q-switched, solid-state, ultraviolet (UV) lasers with pulse durations in the nanosecond regime are now firmly established in precision materials processing. Recently, a new type of mode-locked, UV laser has become available for industrial micromachining applications with pulse durations of around 10 picoseconds and a pulse repetition frequency (PRF) of around 80 MHz. In a recent study, we directly compared the performance of the two laser types for processing three materials commonly used in the electronics industry: polyimide film, FR4 and copper. Our study showed that the optimum choice of laser is very much a function of the material and the desired process speed.
New mode-locked lasers
Historically, mode-locked lasers with picosecond and even femtosecond pulse durations have been too complex, expensive and sensitive to consider for industrial applications. However, a new type of Nd:YVO4 laser has now changed this situation completely, thanks to a special type of mirror called Saturable Brag Reflector (SBR). When used as one of the mirrors in a continuous-wave laser cavity, the SBR forces the laser to operate mode-locked. In mode-locked operation, the cavity modes are locked in phase, resulting in a single, intense pulse at a very high PRF. One example of this is the new Vanguard from Spectra-Physics, which has typical pulse duration of 12 picoseconds and a PRF of 80 MHz. This laser delivers more than 20 watts of output at a wavelength of 1.06 microns that is efficiently converted to wavelengths of 532 nm, 355 nm or 266 nm using pre-aligned frequency conversion modules in the laser head.
Practical laser comparison
We conducted a preliminary study on several materials to evaluate the relative merits of Q-switched and mode-locked lasers for materials processing. The Q-switched laser was a Spectra-Physics Navigator. Both lasers generated 4 watts at 355 nm. The Q-switched laser PRF was set at 30 kHz for the purposes of this study and the mode-locked laser at 80 MHz. In each case, the laser beam was passed through a 10X beam expander and a galvanometer scan head, followed by a 100mm telecentric objective to focus the beam. This delivered about 3.5 W to the work surface with a spot diameter of around 15 microns.
Processing polyimide films
Figure 1. Micrographs of thin (51 microns) polyimide film cut by (a) a mode-locked laser at 600 mm/sec, (b) a Q-switched laser at 80 mm/sec.
The first tests involved cutting polyimide (Kapton) film of two different thicknesses: 50.8 microns and 127 microns. Both lasers cut these films efficiently with only a small heat-affected zone (HAZ). We steadily increased the scanning speed to find the maximum cutting speed that would allow complete cut through of the films. In the case of the Q-switched laser, we were able to cut the 51-micron Kapton at speeds up to a maximum of 80 mm/.sec. With the mode-locked laser, the maximum speed was 600 mm/sec, a factor of seven times greater. Moreover, cutting at 600 mm/sec with the mode-locked laser produced a smaller heat affected zone than when cutting at 80 mm/sec with the Q-switched laser (see Figure 1). While the shorter pulses of the mode-locked laser might be expected to produce a smaller HAZ, much of the difference may just be due to the scan speed-how long the laser dwells in one particular position. In the case of the thicker Kapton film, a higher maximum cutting speed was again obtained with the mode-locked laser, although the difference was less dramatic. The actual maximum speed was 45 mm/.sec versus 25 mm/sec for the Q-switched laser.
Scribing of FR4
FR4 is the glass/epoxy composite used as the substrate for many printed circuit boards (PCBs). In our experiments, we scribed a series of straight-line grooves with a nominal length of 20 mm. We studied the effect of scan speed on groove depth, width and quality for both lasers. The quantitative results are summarized in Table 1.
Figure 2. Scribing FR4 with (a) a mode-locked laser at 1000 mm/sec, and (b) a Q-switched laser at 100 mm/sec. Note the smaller HAZ in (a).
As with the polyimide samples, the mode-locked laser appears to deliver better results. For example, in one set of tests, we set at an arbitrary target groove depth of 20 microns and then increased the scanning speed to determine the maximum speed at which each laser could deliver this depth of scribing. The maximum scribing speed was around 1000 mm/sec for the mode-locked laser and 100–200 mm/sec for the Q-switched laser. (The slight uncertainty arises from limitations in precisely measuring the groove depth). However, the grooves produced by the mode-locked laser were much narrower than with the Q-switched laser. Also, the mode-locked laser at 1000 mm/sec produced less thermal damage than the Q-switched laser at 100 mm/sec (see Figures 2a and 2b).
Figure 3. FR4 samples scribed at 1000 mm/.sec by (a) a mode-locked laser and (b) a Q-switched laser. The slower PRF (30 kHz) of the Q-switched laser means that it cannot produce a continuous line at such a high feed rate.
We also compared the characteristics of scribes produced by these lasers at a fixed scan speed of 500 mm/.sec. Here the mode-locked laser produced a groove depth of 20 microns, whereas the Q-switched could only produce a groove depth of 5 microns. Interestingly, the mode-locked laser also produced a wider scribe at these speeds, clearly indicating that it removes more material. We originally intended to also compare scribe characteristics at 1000 mm/sec. However, at 30 kHz, the Q-switched laser could not produce a continuous scribe: the pulses are no longer overlapped on the work surface, resulting in a "dotted" line (see Figure 3).
Hole drilling and scribing in copper
In addition, we compared the two lasers' performance for both scribing and drilling copper, with the results summarized in Table 2. In this case, these data clearly show that now it is the Q-switched laser that offers the superior processing power, by a fairly wide margin. In particular, the Q-switched laser produced a 10-micron (depth) × 15-micron (width) scribe at 200 mm/second. Even when the scan speed was slowed to 100 mm/sec, the mode-locked laser could not come close to replicating this performance, with a maximum depth of only 2 microns and a width of only 9 microns.
For percussion drilling holes in copper the Q-switched laser is quite clearly a superior source as summarized in Table 3. For example, with a drill time of 10 milliseconds/hole, the Q-switched laser delivered a hole diameter of 20 microns, whereas the mode-locked laser produced holes with a diameter of only 8 microns. The Q-switched laser also appeared to produce much deeper holes than the mode-locked laser at the same drill times. Unfortunately, our optical microscope could not deliver accurate absolute information on the depth of these narrow holes.
These results indicate that the mode-locked laser is a superior source for processing polyimide and FR4, removing more material in less time than a Q-switched laser of the same average power: an advantage that can be used to produce deeper, wider or faster cuts. Moreover, for given cut dimensions at the maximum processing speed for both lasers, the mode-locked laser produces less peripheral thermal damage, due in part to the faster processing speed and lower dwell time with the mode-locked laser. This may also be due to the fact that ultrashort pulses tend to produce less thermal damage, as demonstrated with other types of mode-locked lasers. In the case of the polyimide, it is interesting to note that the advantage of the mode-locked laser decreases with increasing sample thickness, that is, cut depth. This could indicate that the major speed benefit is in piercing, rather than the actual cutting processing. Both lasers reach similar peak powers during their duty cycle, the difference is likely due to pulse duration, rather than peak power. But more detailed studies by materials scientists would be necessary to confirm or negate this idea.
Copper is different from the other materials tested in having a higher ablation threshold and much higher thermal conductivity. It would seem that in micromachining copper, there are two possible reasons for the superiority of the Q-switched laser: the higher energy per pulse and/or the longer pulse duration. When processing metals with infrared lasers, it is a well-known phenomenon that as the metal surface melts during a laser pulse (or when processed by a high-power CW laser) the reflectivity goes down allowing greater absorption. This may be what is happening here. Or, it could be an effect due to the higher ablation threshold of copper. In addition, copper is a very good conductor of heat, requiring sustained irradiation to cause significant local heating, compared to non-metallic samples. Clearly, however, this is a complex issue that merits further investigation by materials researchers.
In summary, it is obvious, there is an upper limit on feed rate of easily machined materials with a Q-switched laser due to the laser's PRF. This is not the case with a mode-locked laser. And, the mode-locked laser appears to be the superior choice for non-metals, whereas the Q-switched laser is better for metals such as copper.
Mingwei Li, Andy Held and Kathleen Hartnett, are with Spectra-Physics Inc., Mountain View, CA. Email email@example.com, www.splasers..com.