The laser drilling horse race: UV lasers, C[O.sub.2] lasers and hybrids all offer advantages and disadvantages. Will one technology make the others obsolete?

The development of PCB technology during the last few years was driven by the industry's technical needs. The main driver for this speedy development was mobile device technology, in particular the requirements set by mobile phones and PDAs. These applications were and are under constant pressure to drive miniaturization of components and interconnection solutions.

Parallel to the interconnection technology development, the functionality of these products was enhanced. A lot of new features can be found today in multimedia platforms: high-resolution cameras in the megapixel range with video capability, high-resolution screens for enabling digital video broadcast on handhelds (DVB-H) available in the near future and MP3 players with hard disc micro drives with a storage capacity comparable to a mid-nineties desktop PC. The interconnection technology has had a tremendous push since 1998, the year when HDI started in Europe. Looking into the high-end handhelds, you will find flex boards and rigid HDI multilayer boards. A move towards merging the two technologies is seen, in particular for addressing the cost benefits of each technology. New rigid-flex technologies are in the pipeline using low-cost materials and a much better automation of their production processes.

All these efforts for improving the interconnection density have one big driver--new packaging technologies. The CSP package with a pitch size of 0.8 mm was the reason HDI technology was introduced. Blind microvia interconnection brought much higher flexibility in routing these high-density components. Two HDI layers on each side were recommended when the pitch dropped to 0.5 mm. The small diameters of the laser via, together with the optical registration of the drilling pattern to the production panel, have the potential to meet the requirements of the new design rules. Throughput and cost are today's drivers in production. During the whole time since laser drilling was introduced for HDI, different methods have been competing to drill holes. Every technology has strengths and weaknesses that must be taken into account when considering its advantages for existing products and for future demand. Laser drilling is one process step that has strong interaction with the interfaces of a multilayer board and the base materials. Hole geometries--top diameter, bottom diameter and aspect ratio--are the most important parameters to influence the results of the plating process. Galvanic copper-filled vias to be used for stacked via construction are the next challenge; their hole shapes have to be in tight tolerances to yield the desired filling behaviour.

Process Options

There was once a competition between ultraviolet (UV) lasers and the C[O.sub.2] laser sources. The UV YAG laser at the third harmonic produces laser light at 355 nm. At these wavelengths most of the materials have a high absorption rate and therefore the ablation process is very efficient. Metals, glass and organic materials like epoxies can be ablated at low pulse energy. The typical energy is at a level of 200 [micro]J when using a beam diameter of 20 [micro]m. Therefore it is understandable that the first industrial UV laser drilling machines had a laser power of only 1 W. Today the sources are running at 5-7 W at repetition rates of 30-50 kHz. The newest diode-pumped UV YAG lasers have 20 W power, running up to 150 kHz. This gives an indication of the fast development in this area.

C[O.sub.2] laser sources running on longer wavelengths--10.6 [micro]m--can be used for ablating organic materials like epoxies and polyimide. These lasers are used for routing of flex boards or drilling of holes. Currently, C[O.sub.2] lasers of 200 W and more are used in state-of-the-art PCB drilling machines for hole drilling. And C[O.sub.2] lasers are faster in bare laminate.

So, where has laser drilling technology development been in the last few years? C[O.sub.2] laser drilling technology came from Japan, UV drillers from the U.S. In the beginning, UV technology was accepted in the U.S. and Europe, but very soon it became obvious that C[O.sub.2] machines had throughput and cost advantages. A new type of hybrid machine was developed, combining the advantages of both laser sources, the UV and C[O.sub.2]. These hybrid drillers combined the UV laser's capability of opening the copper and making holes down to 20-30 lam with the throughput of the C[O.sub.2] laser. The combined process is very stable. At 100 [micro]m hole size, hybrid throughput is between 120 and 180 holes per second depending upon the design and material.

Conformal mask and large window technology came from the Far East, where the hole definition was made by a photo and etching process. Early on, the beam diameters of lasers were 300 lam or more, and the mask for the laser process was etched for these tiny holes, but not with the best yield. Hole diameter variations and missing holes were issues. With improved Optics for the 10.6 [micro]m wavelength, a much better beam-shaping was realized and large windows technology started its triumph. Up to 1,000 holes per second can be drilled on the newest generation of machines. The holes are slightly tapered with low wall roughness even in FR-4 materials, which is ideal for the plating process.

This additional mask-forming photo process for the C[O.sub.2] and the higher cost for electroless plating when drilling in bare laminate raise cost concerns.

With the optimized laser optics, which can generate very high power on a small diameter, a new possibility is available: direct copper drilling. Together with a treatment on the copper surface (typically black oxide) that improves the coupling of the laser light into the copper, ablation can be done with a C[O.sub.2] laser. This treatment process has to run in tight tolerances to get constant absorption of the laser pulses. It was several years until this process was used in production, and today it is used for 100 [micro]m holes. There is a trend toward this technology, and some companies in the Far East are running direct copper drilling in high volume.

There are some other issues like desmear, plating and registration in general, which must be mentioned. One of the real advantages of UV laser drilling is the cleanliness of the inner layer copper after drilling. A light desmear process is enough to get an excellent interface of the microvia barrel to the inner layer copper. With small holes the cleaning process tends to be more challenging, but the UV laser-drilled holes show fewer problems than C[O.sub.2] laser-drilled holes. This non-aggressive process prevents wedge voids that could lead to reliability problems. C[O.sub.2] laser-drilled holes will never be clean on the bottom of the holes due to physical limitations. The laser light will be reflected at the copper surface of the inner layer out of the low absorption of this surface and therefore the energy level is not high enough for complete ablation of the resin.

Registration Capability

The automatic registration of the drilling pattern to optical targets (drilled or etched) is this drilling technology's main advantage. With the individual registration and the adaptation of the drilling pattern to the dimension of each board, pad sizes could be reduced dramatically. Microvia target lands have a diameter between 300 [micro]m and 250 [micro]m and they will continue to drop.

We carried out an analysis of the registration capabilities of existing laser drilling equipment. Most of the existing machines drill the pattern on a working area of 2 x 2". Within this field, the galvo motors move the laser beam to the specified location. When all holes are drilled in a section, the machine steps to the next section by moving the xy table. FIGURE 1 shows the registration analysis randomly clone over one complete production panel. This is typical for a hybrid machine, which opens the copper with UV and ablates the resin with a C[O.sub.2] laser.

[FIGURE 1 OMITTED]

The registration tolerance was weil below 30 [micro]m over the production area of 12 x 24". The measurement was performed against mechanically drilled holes, which have been the registration targets for the laser drilling process and the position of the drilled laser holes. Figure 1 shows the offset between the measured position and the theoretical position of the CAD data.

As a comparison, an analysis on a C[O.sub.2] laser machine for copper direct drilling was performed (see FIGURE 2). This machine had some slight performance problems--an offset in x and y of about 15 [micro]m, which can be compensated for, and a larger standard deviation of measured position values. An in-depth analysis of the misregistration was carried out. It could be shown that the comers of the fields exhibit the maximum positioning failure. In addition, a stepping error of the xy table can be seen.

[FIGURE 2 OMITTED]

The vector graphic in FIGURE 3 shows much more detailed information. The field in the middle shows position failures caused by the galvo motors. At the left side of the board, the xy offset is a maximum. The marked area shows the combination of the galvos and the table error. For the next generation of board designs these tolerances have to be reduced--towards less than 15 lain.

[FIGURE 3 OMITTED]

Another analysis was done on the diameter variation of holes that were drilled with the direct copper process. This parameter is critical for the following hole cleaning and plating processes. The surface of the copper was treated with black oxide and the optics of the C[O.sub.2] laser machine were optimized for this process. FIGURE 4 gives an impression of the process stability of 90 [micro]m holes.

[FIGURE 4 OMITTED]

The +/- 9.5 [micro]m variation of the hole diameter is nearly the same as with the UV- C[O.sub.2] process. It is acceptable for a hole diameter of 90 [micro]m.

As a next step, holes with a nominal diameter of 70 [micro]m and 50 [micro]m were drilled on the same machine. About 5% of the holes were missing at 70 [micro]m diameter and 30% at 50 [micro]m. Small changes in the morphology of the oxide treatment, together with the technical limits of the optics, could be the reason for this result. Nevertheless, the pressure to improve the direct copper drilling process is on the developer of the C[O.sub.2] machines.

What's Next

The pitches of CSP components are setting the design rules, and we'll see 0.4 mm CSP on telecom products in volume in 2006. With the advent of these components, several changes in design will take place. Stacked microvias and 2-3 HDI layers on each side will be needed. Lines/spaces will drop to below 60 [micro]m and target lands will have a diameter of 220 [micro]m. The number of laser vias will increase by a factor of 3, and their diameter will drop from 90 to 70 [micro]m. Providing capacity for laser drilling even when the hole diameter is reduced will be a major challenge for the future. The large window process will be the only one that can provide high throughput but it cannot compensate for the dramatic growth of the microvia.

Looking two years into the future, the next dramatic change in design rules will take place when 300 [micro]m pitch CSPs will roll out. Dielectrics will drop in thickness and microvia will be at 50 [micro]m diameter. This may be when laser-drilling technology could switch from C[O.sub.2] to UV to produce accurate holes with high throughput. PCD&M

HANNES STAHR is the manager of R&D at AT&S. He is currently responsible for technology implementation. He can be reached at [e-mail].

FIGURE 4. The process stability of 90 [micro]m holes drilled with
C[O.sub.2] lasers.

Process Data

LSL                    0.04000
Target                 0.09000
USL                    0.14000
Sample Mean            0.10915
Sample N                   839
StDev (within)         0.00314
StDev (overall)        0.00385