Christopher Ryder, of MKS Instruments’ ESI products group, describes how CO2 lasers can be used to address new material challenges being faced in the manufacture of 5G technologies
5G is more than a buzzword for anyone directly involved with the materials and components needed to make it a reality. While not necessarily requiring a paradigm shift in terms of manufacturing, this higher frequency bandwidth brings with it some significant new challenges, and with 5G set to ‘go live’ within the next few months around the world, finding robust, impactful solutions to these challenges has become the focus of various tiers of the micro-electronics supply chain. In this article I will explore how CO2 lasers can be used to perform via drilling on rigid PCB panels featuring complex base materials – which are at the core of emerging 5G technologies.
Typically, rigid PCB panels consist of copper-cladded FR4 dielectrics comprising of woven glass fibres and a resin compound, which ensures the required panel flatness and stiffness, as well as the desired inter-layer electrical/signal insulation. Given the CO2 laser’s ability to process these materials effectively and quickly, it remains a popular solution for high-volume manufacturing. The power to the work surface is an order of magnitude larger than what one would typically see with a UV laser used for PCB via drilling, making it highly versatile for the target HDI (high density interconnect) materials. Nonetheless, as base materials become more complex, new ways of approaching these laser processing applications may become necessary.
For our purposes, we should understand the main material challenges stemming from the need for high frequency applications for electrical interconnect. These applications, as such, are not particularly new, and neither are the various material solutions and processes used in manufacturing. But with the rising applications and production volumes associated with applications such as 5G, IoT (Internet of Things), and autonomous driving, a whole new emphasis is needed to characterise expected base material performance in terms of its processability. In particular, we see a convergence of lower Df/Dk (dissipation factor and dielectric constant) and lower profile Cu (copper) foils, which help to meet functional requirements, with generally reduced material thicknesses (both Cu and dielectric) driven by smaller form factors. These smaller, thinner form factors potentially complicate high frequency demands for multilayer PCBs.
ESI’s Geode system is capable of performing CO2 laser-based via drilling on the rigid PCB panels at the core of emerging 5G technologies.
To address the need for lower Df/Dk and the concurrent need for thinner base material stack-ups, base material suppliers can offer products that are modified from existing recipes, which is typically an FR4 with high and specialised resin filler content designed to better insulate signals between Cu layers. Issues may arise in processing these due to the differing resin melt properties onset from the filler combinations and the effects this may have on material ablation.
Materials such as PTFE (Teflon) offer excellent Df/Dk properties with very low thicknesses, but also tend to have a higher CTE (coefficient of thermal expansion) and typically lack the structural integrity (rigidity) of an FR4. Therefore, the high thermal loading of a CO2 pulse train can negatively affect via quality (barreling, residues, etc) and panel scaling. Some base material manufacturers offer composite dielectric materials, which offer great performance potential, but can present process challenges given the variations in constituent glass-transition temperature (Tg) – i.e. each material is melting/ablating at a different rate.
When we add to these complications to the demand for thinner inner layer low-profile Cu that can be easily damaged or penetrated with excessive power, the CO2 process faces some setbacks, largely driven by overall thermal load, the temporal displacement of that load and the spatial distribution of that energy. To understand the difficulty, we need to take a look at the pulse and how it is delivered to the work surface. In essence, a round imaged beam is formed through various apertures, which provide the desired spot size for the via diameter in question. However, the larger the spot size, the lower the fluence of that beam. This is sometimes called a ‘punch’ process, as we’re using the full-imaged spot to punch through the Cu and dielectric down to the bottom Cu.
Figure 1: A typical CO2 laser pulse used to ‘punch’ through the copper material in rigid PCB panels.
Power and duration of the pulse can be attenuated, and under normal circumstances we see delivery of the pulse with a desired ‘peak power’, followed by a usually less-desired residual ‘pulse tail’. The pulse tail (see Figure 1) is the energy that exits the laser after RF (radio frequency) excitation in the laser source has ceased. This is the water pouring out of the hose after we have shut off the valve. This can result in an over-delivery of energy versus what one is targeting. Of course, there are ways to address this potential excessive energy being delivered to the material, but these typically involve employing multiple lower power pulses, which can adversely affect drilling throughput (more pulses = more time per via).
It would appear then, if presented with materials potentially sensitive to high thermal loading over time (more time = more melting/ablation = higher Cu undercut and other effects impacting quality negatively), the desire would be to more finely contain/control energy amplitude and duration, as well as spatial distribution. In other words, there could be an advantage found through applying the desired peak power when and where needed, without the disadvantages of a large residual pulse tail or the throughput-inhibiting lower fluence spot size needed to drill.
Figure 2: Pulse amplitude and temporal modification with acousto-optic deflector technology.
In UV-based via drilling, devices known as AOD’s (acousto-optic deflectors) have been used for some time to essentially slice and dice the pulse train in terms of amplitude and duration, and simultaneously distribute the energy spatially. Hence with a small, higher fluence spot size that is rastered over the material, vias are created. Different to the punch process described above, this rastering of vias takes advantage of the high speed of AOD deflection to accommodate complex multi-diameter toolpaths without spot recalibration. However, this has largely been limited to the UV wavelengths and the accompanying material sets thus far.
With its recent acquisition of ESI, MKS leverages substantial history with AODs in UV flex PCB drilling to benefit the HDI CO2 via drilling space with a new system, named Geode. With the system’s Hypersonix AOD-based technology, pulses can be tuned and edited for amplitude, duration and spatial distribution with a high-power laser engine (see Figure 2).
Figure 3: Example of a thin bottom copper application.
The capability to finely tune energy and duration of the pulse creates a new toolset for managing the above-mentioned challenging base materials. In addition, the elimination of the pulse tail supports a more predictable process quality window with a more precisely dosed drilling recipe. The use of delays in the pulse execution also allows for additional leverage in managing material melt/ablation behavior and potential benefits for reducing via undercut and bottom Cu damage on thin inner Cu layers (see Figure 3).
The use of AODs, however, also brings with it some spatial distribution benefits. In the punch process, for a given material, there are general limitations which result in ‘one spot size = one via size’ operations. But if a single spot and trepan that pulse in a given pattern is maintained, multiple via sizes may be achieved without any changes to the optics (i.e. no recalibration of new apertures, etc). This helps to maintain, and indeed often increase throughput (see Figure 4), determined by how many vias per second can be drilled.
Figure 4: Drilling multiple via diameters in a single pass with acousto-optic deflector technology.
The use of the technologies described above are helping manufacturers meet new 5G material challenges with increased photonic flexibility, without any fundamental changes to existing process landscapes. Whichever material compositions will set new baselines for the HDI PCB manufacturing, it is clear that having more physics in our system’s toolkit will help to meet existing and emerging technical and commercial demands.
Christopher Ryder is the director of product marketing for ESI products, MKS Instruments