Partida Arancelaria, preferencias y restricciones

Frequency tripled solid state lasers require less capital to buy and are less expensive to maintain in comparison with excimer lasers. A more detailed cost of ownership comparison of the two laser technologies is presented in Section 6.1.3. Solid state lasers typically have a much higher beam quality than excimer lasers, and their close to diffraction limited performance is exploited in laser direct write systems by focusing the laser to a small spot. Given the cost advantage, laser direct write systems with applications in microelectronics have been the subject of much research, for example [6, 50-55]. Such systems usually pass the laser beam through a galvanometer scan head which deflects the beam across a substrate. An f-theta lens mounted underneath the scan head keeps the beam in focus across a flat substrate regardless of deflection angle. The laser is used to vector mark grooves in a substrate sequentially such that the process time is dependent on the circuit pattern complexity.

As a vector marking system with CAD file input, LDW processes are ideal for low volume prototyping since there is no need for a mask or the longer cycle times associated with lithographic processes [56]. They do, however, suffer from process times dependent on pattern complexity, design rules to keep beam velocity constant across the substrate and some limitations in the circuit features they can process. Complicated control systems have been researched to try to overcome some of these limitations [26, 53], but, for example, it remains challenging and time inefficient to ablate large ground plane areas to a uniform depth by overlapping a comparatively very small focused spot.

Feature resolutions down to around 10μm are currently achievable across approximately 20×20 mm scan fields using commercially available f-theta lenses with direct write laser structuring methods using frequency tripled solid state lasers [57]. SEM micrographs of 9 µm channels machined in an ABF type substrate can be seen in Figure 2.16 [26]. As with the excimer laser, the silica filler particles cannot be machined by the laser and are either ejected from the substrate remain in the channels.

27

Figure 2.16 SEM micrographs of 9 µm channels machined with a frequency tripled solid state laser and a vector direct write system with galvanometer scanhead and acousto-optic deflector. The ABF type substrate is a glass filled epoxy resin, and the silica particles are visible in the trenches – from [26].

It is possible to extend the area an f-theta lens and scanner can address without stitching scan fields by moving the substrate under the scanner. Bow Tie Scanning (BTS) is a technique where the substrate stage is moved in an axis orthogonal to the scan direction. In addition to scanning the beam across the sample, the galvo-scanner also compensates for the motion of the substrate in the axis orthogonal to the scan direction as depicted in Figure 2.17 [58]. As the laser is scanned across the moving substrate to form parallel lines in the reference frame of the substrate, the commanded moves sent to the scanner outline a bow tie in the static reference frame.

Figure 2.17 Schematic diagram illustrating the bow tie scanning methodology. As the substrate is translated in an axis orthogonal to the scan direction, the scanner compensates for the stage motion such that parallel lines are produced at the substrate – from [58].

When used in conjunction with a silicon mask, this technique can be used to ablate repeating patterns across large substrates. Figure 2.18 shows T bar electrodes ablated into an ITO on glass substrate using a solid state laser in a BTS regime. The image shows a 4×4 array of laser pulses, where each laser pulse contains the image of the

28

electrodes. This technique could be used to efficiently pattern arrays of repeating features in a chip package across panel sized substrates, for example vias.

Figure 2.18 Microscope image of electrodes ablated into ITO on glass using the BTS technique. Each element in the 4×4 array contains the image silicon mask projected onto the substrate using the f-theta lens – from [47].

Solid state lasers have long been the industry standard in via drilling below 100μm, where the thermal effects induced by the cheaper CO2 lasers become too great, and as

such vias down to 10μm are the current state of the art [59]. The smallest hole sizes achievable with commercial UV solid state laser systems at throughputs of thousands of holes per second, such as the Orbotech Emerald 150 and ESI 5330 series, is 30 µm and 25 µm diameter respectively. Laser drilling of hole diameters less than 5 µm in diameter in production are only currently achievable with excimer laser systems.

Reference [60] is a simplistic comparison of a frequency tripled solid state laser and a KrF 248nm excimer laser source for the production of fine lines in printed circuits. The excimer laser source is used in a direct write type application. The analogue to the focused spot in solid state systems is achieved by imaging mask apertures of different sizes, since the beam quality of an excimer laser is too low to focus to a small spot. They report a smallest spot size of 25μm achieved by imaging a 100μm aperture to the workpiece where apertures smaller than this result in too little energy reaching the substrate. However this could be easily rectified by changing the beam size at the mask with a telescope. The focused spot of the solid state laser was also 25μm, so the features ablated by the two lasers were of similar dimensions. The excimer laser trench was squarer in profile due to the uniform illumination of the aperture by the laser. In part due to the inefficient use of the beam, since most of it was blocked at the aperture,

29

the solid state laser process was found to be 120 times faster than the excimer process despite the solid state laser only having 0.12W average power and the excimer laser 7W. They also report larger a resolidification height, the height of the melt relative to the rest of the substrate at the edge of the trench, with the frequency tripled Nd:YAG laser. This could be in part due to the shorter excimer wavelength leading to a more photochemical ablation mechanism, but the repetition rate of the excimer laser being two orders of magnitude smaller than that of the Nd:YAG laser likely contributed significantly as well.

Solid state laser systems are less expensive and more convenient than excimer laser systems. They offer a good solution for marking simple patterns, or low volume prototyping but are less suitable for marking the complex patterns with large areas of uniform ablation required for advanced chip packaging.