Ceramic Industry

The Laser Age

March 1, 2008
Despite a long history, new applications for the laser processing of microelectronic ceramics are still under development.

Lasers have been used for the cutting, scribing and drilling of the ceramic substrates used in microelectronics packaging for nearly three decades. Yet, despite this long history, new applications for the laser processing of microelectronic ceramics are still under development. Some laser manufacturers are even supporting these new techniques with products specifically optimized for emerging ceramic applications.

Figure 1. The nearly square wave pulse of a slab discharge laser delivers the same processing power as a much higher-power flowing gas laser.

Laser Processing of Thick Ceramics

The benefits of laser processing over traditional mechanical methods are fairly obvious and include speed, accuracy, consistency (no tool wear), reduced consumables costs, and the lack of downtime usually required for tool replacement. As a result of these advantages, lasers have become well established for certain ceramic processing tasks. In particular, they are often used with fired ceramics in the 0.010 to 0.125 in. thickness range, with alumina being by far the most common material.

Initially, the majority of ceramic processing applications used flowing gas CO2 lasers in the 300 to 750 watt (W) power range. These CO2 lasers emit light at 10.6 µm, a wavelength that is far into the infrared and easily absorbed by virtually all ceramics. However, the use of these types of lasers represents something of a “brute force” approach because of their output characteristics.

In particular, the temporal pulse shape of this type of laser is triangular and typically has a relatively slow rise and fall time. This means that, for much of the pulse duration, the laser power is not sufficient to process the material. The result is that a significant portion of the incident laser power goes into simply heating the work- piece, rather than cutting or drilling it. This produces a large heat-affected zone (HAZ) and also requires a relatively high-power laser to achieve a given result.

The sealed, slab discharge CO2 laser was introduced in the early 1990s to address the limitations of older flowing gas technology for precision processing applications, such as ceramic scribing. One important characteristic of this laser type is that it produces a nearly square wave pulse with a very fast rise and fall time. As seen in Figure 1, this pulse shape delivers much more of its energy above the material processing threshold. The result is that a sealed CO2 laser of a given power can provide the same processing power as a much higher-power (and therefore more expensive and physically larger) flowing gas laser.

The sealed CO2 laser also delivers several other distinct advantages for ceramic processing. For example, both pulse shape and power can be precisely controlled, allowing for the optimization of process parameters. In fact, the fast rise and fall times allow custom pulse shaping on a pulse-by-pulse basis. Just as importantly, the pulse energy can be independent of repetition rate for enhanced process flexibility.

The slab discharge design also yields very good beam characteristics (i.e., a beam that can be readily focused). This characteristic is typically quantified by the beam quality parameter M2. A perfect laser has a M2 of 1.0 and can be focused to the so-called diffraction limit, where the spot size has a diameter that is about the same as the wavelength of the laser light itself. In reality, the very best lasers are typically in the 1.1 to 1.2 range, and any M2 value of less than 1.5 is considered very good for materials processing.

A typical M2 for a sealed CO2 laser falls in the 1.2 to 1.5 range.* In contrast, most flowing gas lasers are in the 2 to 3 M2 range. The major benefit of high beam quality is sharper focus on the workpiece, which results in minimal HAZ. Concentrating the laser power through tight focusing also contributes to increasing the effective processing power of a laser of a given power.

In addition, these lasers deliver excellent reliability and reduced operating costs because they are completely sealed off. In particular, they do not consume gas and their sealed optics do not need to be replaced. Slab discharge lasers are also remarkably small, as a result of their high efficiency. For example, one model** can produce 400 W of average power, with peak power of over 1 kW, and yet the laser is only 42 in. long and weighs less than 75 lbs (including the integrated power supply).

The small size and low weight of sealed CO2 lasers also offer a number of practical benefits. The lasers can be mounted directly on a machine or work station, eliminating complex and expensive beam delivery systems. Moreover, the sealed-off configuration eliminates the cost and logistical problems of gas bottles and pumps. Finally, virtually no maintenance costs are associated with this type of laser, and there is no scheduled maintenance required within the laser’s expected lifetime.

*Values are for the Coherent Diamond product line.
**The Coherent Diamond E-400.

Figure 2. Representative construction of a multilayer LTCC surface mount substrate. LTCC image courtesy of American Technical Ceramics. © 2008 American Technical Ceramics.

LTCC Processing

Low-temperature, co-fired ceramics (LTCCs) are becoming an increasingly popular microelectronics substrate, especially for use in the construction of thin, multi-layered devices like flash RAM. LTCCs are produced in a tape format by applying a ceramic slurry to a silicone-coated polyethylene terephthalate (PET) carrier (see Figure 2).

The slurry contains fine particles of alumina, quartz and glass. The materials can be fired at temperatures below 900°C because of the glass content, allowing co-firing with low-melting-point conductors such as gold and silver. The thickness of the green ceramic typically ranges from 50 to 250 µm, while the PET carrier thickness typically falls between 40 and 60 µm.

To produce circuit traces, the conductive material is screenprinted onto the tape. Holes, called “vias,” must then be produced to make electrical connections between layers or to conduct heat through the layers. After via production, the carrier film is removed to allow lamination and firing.

Vias in first-generation products based on LTCCs were 250 µm in diameter. Holes of this size can be reliably produced by mechanical means. However, as with all microelectronics products, the market has been pressured to shrink product dimensions, reducing via diameters down into the 100 µm range. While holes of this size can be mechanically punched, tool wear becomes a significant issue. As a result, virtually all via drilling in LTCCs is now accomplished with CO2 lasers.

Unfortunately, traditional CO2 lasers present one serious drawback for drilling green ceramic tape. Specifically, drilling is usually done through the PET carrier, which then melts over a substantial area surrounding the drilled via. The melted plastic in this recast ring adheres to the ceramic, which makes it difficult to separate the backing from the substrate without damaging the substrate. Ultimately, this limits the spacing and density of holes that can be produced, compromising the utility of the laser drilling process.

The absorption characteristics of many plastics vary substantially with wavelength in the infrared. As a result, it is sometimes possible to achieve better results by working at wavelengths other than the 10.6 µm that is most common with CO2 lasers. This fact has already been exploited in the packaging industry, where it has been found that, for certain materials, process speed and quality can be increased by using other CO2 laser wavelengths where absorption is higher than at 10.6 µm.

Figure 3. The absorbance spectrum of PET shows that the material absorbs 9.4 µm light much more strongly than 10.6 µm, resulting in a much smaller HAZ at the shorter wavelength.

In the case of PET, the transmittance of a 50 µm thick film is 26% at 10.6 µm; this drops to 4.4% at a wavelength of 9.4 µm (see Figure 3). In contrast, the ceramic itself has a negligible difference in transmittance between these two wavelengths; it absorbs very strongly at both wavelengths. These properties can be exploited to significant advantage in the application of LTCC drilling. Specifically, it has been found that processing with a 9.4 µm CO2 laser essentially eliminates the sizeable HAZ produced by 10.6 µm laser processing, thus avoiding the problems experienced in separating the backing.

One hypothesis to explain these results is that infrared light is scattered in the PET and propagates laterally through the film. The propagation distance has a strong dependence on the absorptivity. For example, light at 10.6 µm will spread laterally for about 40 µm in a 50-µm-thick PET film, while 9.4 µm light will only spread about 16 µm. These distances correspond almost exactly with the observed HAZ profiles produced in testing under these conditions. Most importantly, the smaller melt zone around the holes drilled using a 9.4 µm laser produces less adhesion, allowing for the clean removal of the carrier from the ceramic.

The 10.6 µm wavelength is generally favored for commercial CO2 lasers because it produces the highest output power for a given size laser. However, it is quite possible to build high-reliability, high-performance CO2 lasers at other wavelengths through the use of wavelength-selective optics inside the laser resonator, together with the use of different isotopes of CO2. For example, one model can deliver up to 225 W of output power at 9.4 µm, yet offers the same beam characteristics and lifetime specifications as other sealed CO2 lasers at this power level.

†The Coherent Diamond K-225i laser.

Figure 4. Photos of 400 µm square holes laser drilled in PET clearly show the larger HAZ when working with 10.6 µm light.

Future Trends

For some applications, ceramic substrates must be scribed after other materials have been deposited onto them. For example, substrates for the surface mounting of passive devices typically consist of 250-µm-thick fired alumina with a 50-µm-thick layer of copper conductor on one side and a 50-µm-thick layer of stainless steel on the second side. Individual parts of about 2 x 3 mm in size are singulated from a larger substrate.

The current technology for performing this singulation is mechanical sawing. However, there are two significant problems with this process. The first is that the saw cuts can only be placed with an accuracy of about 25 µm. At worst, this can produce a 5% part-to-part variation in final dimensions (in the 2 mm direction). The second problem derives from the fact that metals, especially copper, are much softer than ceramic. As a result, the copper tends to stick in the teeth of the saw blade, lowering its cutting efficiency. Also, copper dust gets spread over the entire cut surface and can even provide an electrical pathway between front and back surfaces, thus creating a short circuit.

Preliminary investigations have shown that the laser is a viable solution for this application. In this case, the laser of choice is a diode-pumped, solid-state (DPSS) laser with output in the ultraviolet (355 nm). This short wavelength is necessary because it is absorbed very strongly by both ceramics and metals. In contrast, metals are very highly reflective at the 10.6 µm CO2 laser wavelength, as well as in the near infrared and visible. High absorption translates into efficient material removal of both metal and ceramic layers, as well as minimal HAZ. DPSS lasers are preferred over other UV laser sources because they offer the combination of performance, reliability and cost of ownership required to make this process competitive with mechanical sawing.

In terms of performance, this application needs relatively high power to achieve the desired feed rate, together with good beam quality. The latter is necessary because it determines how well the laser energy is coupled into the substrate, and it also affects the width of the scribe. Reliability and cost of ownership (specifically maintenance downtime), mean time between failures (MTBF), and cost of consumables ultimately determine the economic viability of laser processing.

Commercial DPSS lasers are now available with the output characteristics necessary for sawing multi-layer ceramics. For example, one DPSS laser can deliver 23 W of average power (at a 90 kHz repetition rate) at 355 nm and possesses excellent beam quality. This laser can scribe a metal/ceramic sandwich of the dimensions described previously at feed rates in the 50 mm/sec range. Recast metal in this process is limited to a distance of about 10 µm from the scribe, thus eliminating any potential electrical shorting problem.

The DPSS architecture employed in the series also provides excellent reliability characteristics. Specifically, the units employ a hermetically sealed laser head design with permanently solder-bonded and aligned optics so that no service within the laser head is ever required. The only consumables are the pump diodes; these are located in the power supply and their light is coupled by fiber optic cables into the laser head. This allows diodes to be easily replaced in the field by end users without the need for any optical realignment. The typical lifetime for these diodes exceeds 20,000 hours.

‡The Coherent AVIA-355-23.

Increasing Popularity

Lasers have long coexisted with mechanical methods for the scribing, sawing and drilling of ceramics. However, as ceramic substrates for microelectronics become more sophisticated, and as their physical dimensions and mechanical tolerances decrease, the advantages of laser processing should make this technique even more attractive, thus increasing its utilization.

For more information, contact Coherent Inc. at 5100 Patrick Henry Dr., Santa Clara, CA 95054; (860) 769-3130; fax (860) 243-9577; e-mail Stephen.Lee@coherent.com or Frank.Gaebler@coherent.com; or visit www.coherent.com.

Links