Diode-pumped, solid-state lasers with Q-switched output are now well-established tools in a wide range of processes for industries such as semiconductor electronics, printed circuit boards, ceramics and plastics. The diversity of these applications means that laser manufacturers must deliver a broad product range in terms of wavelength, power, repetition rate, pulse duration, and peak power. However, much of the laser development in this field is currently concentrated in two areas: achieving higher repetition rates (hundreds of kilohertz), and expanding the performance envelope at 266 nm.
Increased process throughput
Most of these lasers are used for some type of micromachining or marking. Micromachining refers to the creation of small parts, features, grooves, or cuts with micron precision. Compact, diode-pumped lasers are ideal for these applications, in part because they deliver excellent beam quality with a stable output mode and long-term reliability.
FIGURE 1. With a high pulse repetition rate the cut will be continuous and smooth—even at fast scanning rates—thus delivering high process throughput.
For many industrial applications, overall process throughput is a paramount consideration. Typically these applications involve fast scanning of a focused laser beam using galvanometer mirrors, sometimes in conjunction with rapid, automated motion and handling of the parts themselves. At high scanning speeds, the use of a laser with a modest (30 kHz, for example) repetition rate can result in poor overlap of features created by consecutive pulses. The end result could be a cut or mark with ragged edges, or worse, a series of holes, rather than the desired, smooth, continuous cut (see Fig. 1). Laser repetition rate cannot be increased at the expense of average power, however, because it would decrease process throughput. This is why many of today's industrial diode-pumped lasers are based on Nd:YVO4 (vanadate); the high stimulated emission cross section of this material inherently supports much higher pulse repetition rates and higher average output power than does Nd:YAG.
Until recently, most of these Nd:YVO4 lasers were limited to a maximum repetition rate between 30 and 100 kHz. Recently, novel designs have been developed to deliver repetition rates as high as 400 kHz, with average power up to 8 W at 1064 nm. Rather than being optimized for short pulse duration, the typical pulse duration is 120 ns. These new lasers are targeted specifically at high-speed materials-processing applications.
The drive to UV
The high peak power delivered by Q-switched lasers enables efficient harmonic conversion to visible (532 nm) and ultraviolet (355 and 266 nm) wavelengths. Much of the industry's recent development efforts have focused on producing lasers at 266 nm with higher reliability and longer lifetimes. The two main advantages of UV lasers for micromachining are now well documented. First, UV photons directly break the interatomic bonds holding a material together. This cold process minimizes peripheral thermal damage or the HAZ (heat-affected zone) compared to a strictly thermal process. Second, shorter wavelengths can produce smaller feature sizes because of diffraction. However, losses during harmonic generation mean that the shorter the wavelength, the lower the power for a given input. As a result, the infrared fundamental is generally used when power, and hence processing speed, is key, whereas the UV wavelengths provide better precision and edge quality.
These 355-nm all-solid-state lasers are now widely used in micromachining applications because of their simplicity, excellent beam quality, pulse-to-pulse stability and low operating costs. However, shorter UV-wavelength lasers are desirable for several micromachining applications, particularly with exotic and "difficult" materials such as sapphire. This is because the absorption of many of these materials is much stronger at shorter ultraviolet wavelengths. Consequently, the material is removed mainly through cold ablation due to direct absorption, speeding material removal while minimizing any thermal processes and hence peripheral damage.
For these reasons, laser manufacturers have vigorously worked to improve the performance and reliability of 266-nm all-solid-state lasers. This effort has involved the development of harmonic-generation schemes and materials with improved longevity, and the fabrication of optics and coatings that can withstand high peak powers at 266 nm for extended periods. As a result of these efforts, the latest 266-nm lasers now offer reliability similar to that of 355-nm lasers; the major consumable is the diode array, which is typically rated for more than 10,000 hours.
Solar-cell manufacture
There are now many micromachining and precision-marking applications for all-solid-state lasers. Examination of two different applications—scribing solar panels and scribing blue LEDs—illustrates the need for high repetition rate and short output wavelength.
For solar energy to successfully compete with traditional fuel sources, it is critically important to maximize efficiency, reduce manufacturing costs, and maximize useful lifetime. High-repetition-rate infrared lasers are now playing a key role in these efforts. Exitech (Foster City, CA) is a leading integrator of laser-processing workstations for several diverse industries, including solar-panel manufacturers. "Lasers are used to create front surface grooves in silicon panels; these contact grooves are subsequently metallized in order to act as electrodes,'' says Herbert Pummer, president of Exitech. "This buried-electrode structure is more electrically efficient than surface-deposited electrodes. Just as important, these laser-formed electrodes can withstand extreme thermal cycling without being damaged." Pummer explains that this is a critical advantage in device longevity, because solar cells cycle between ambient temperature (-20°C in some locations) and 60°C in bright sunlight.
FIGURE 2. This narrow surface groove was produced in a silicon nitride-coated solar-cell wafer at a processing speed of a few seconds per wafer using a 1.06-µm laser with a pulse repetition rate of 120 kHz.
Single-crystal silicon solar cells are generally produced on 5-in.-square wafers from 6-in.-diameter crystals. These usually have a thin top layer of very hard silicon nitride. The laser is used to create a front-side groove that is less than 20 µm wide and at least 40 µm deep (see Fig. 2).
Pummer notes that, "Manufacturers need to create about 14 m of total groove length per wafer at the highest possible speed. To avoid being limited by the laser pulse rate, we need to operate at 120 kHz or more, so as to achieve 80% pulse-to-pulse overlap, which is necessary for smooth, clean grooves. Otherwise, at our high scribe speeds, we would have ragged edges or even discontinuous cuts (dotted lines) due to poor overlap of laser pulses. Because we're primarily concerned about speed, and don't need the ultimate in cut quality, the Nd:YVO4 infrared solid-state laser is ideal for this application."
Scribing high-brightness LEDs
Demand for high-brightness, blue-emitting LEDs is growing at a healthy rate, for applications such as data storage and display. A blue LED consists of several epitaxially grown layers of gallium nitride (GaN) on either silicon carbide or sapphire substrates. Silicon carbide (SiC) is conventionally diced using high-precision saws, whereas sapphire is usually mechanically scribed with a diamond tip and then cleaved with a fracturing machine. Both of these substrate materials are extremely hard and the LED die size is relatively small, making it problematic to dice the individual LEDs out of a wafer by these traditional methods. Specific problems include low die yield, low throughput and high operating costs. Moreover, mechanical sawing cannot be used on wafers that are slightly warped, further reducing the overall process yield.
JPSA (Hollis, NH) is a laser integrator that provides turnkey laser workstations to the semiconductor industry, as well as offering contract laser manufacturing services. According to Jeff Sercel, president, the solid-state ultraviolet laser is now having a significant impact in blue LED production.
Silicon carbide strongly absorbs wavelengths shorter than 370 nm, but in sapphire, even 266 nm is absorbed only weakly. However, when Q-switched pulses are tightly focused into sapphire, the high fluence drives efficient multiphoton absorption and hence material ablation (see Fig. 3).
Sercel notes that his company has customers using lasers to pre-scribe sapphire substrates; the wafer is laser cut to a depth of 30% to 50% and then cleaved mechanically. Currently, the most cost-effective source for this application is a 355-nm Q-switched laser with a pulse energy of 0.1 mJ at 50 kHz and repetition rates of up to 300 kHz.
For typical 2-in. sapphire wafers with 350 × 350-µm die pitch, Sercel estimates the operating cost is less than $2 per wafer for a UV-laser-based system. Moreover, the UV laser boosts wafer throughput by 500% compared to traditional mechanical methods. Also, the laser's high reliability delivers uptimes exceeding 99%.
However, recent studies suggest the final LED light output is reduced by more than 10% by use of the 355-nm laser. For some LED applications, this power loss is unacceptable. In response, JPSA has pioneered a system based on our 266-nm laser. Sercel claims that initial results at customer sites indicate that the 266-nm laser eliminates the light-reduction problem. In addition, the higher absorption at 266 nm means that the throughput is similar to that with the 355-nm laser, even though the laser output power is lower (see Fig. 3).
MINGWEI LI is an applications sngineer and ANDREW HELD is director of marketing at Spectra-Physics, 1335 Terra Bella Ave., Mountain View, CA 94039; e-mail: [email protected].