With recent memories of Halloween, I’ve pulled together a few short lessons from skeletons that have jumped out at me over the years to disrupt the growth journey in companies that I’ve been involved with together with reflections on what I would do differently next time around.
This one resulted in a lot of wasted metal…
Shrinking geometries in the semiconductor industry result in device features being squashed closer and closer together in the x-y plane. Layer thicknesses in the z-direction don’t reduce to the same extent, as certain thicknesses are required to provide sufficient electrical isolation between conductors. This results in higher and higher aspect ratios in the interconnects between layers, making them more difficult to fill with metal to form the electrical contacts. We had a highly innovative process that used a high pressure and temperature to force the metal into these narrow (sub-micrometer) interconnects. This worked well providing you could hold the temperature and pressure within a narrow process window: too low and the metal didn’t flow into the hole, too high and you would damage the other layers in the structure.
The extremely narrow process window led to enormous engineering effort being applied to control the process conditions throughout the process chamber. Months and months of engineering time were applied to new heater designs, new control systems, and modelling of thermal profiles. Several customers adopted the technology on a trial basis and continued the engineering effort by running huge matrices of experiments at different pressures and temperatures to identify the optimum process conditions to get the metal to flow into their structures. After several years of effort, it was finally concluded that the process window was impossibly small so the process could not be taken into production. Meanwhile, our competitors had developed a new process using a different metal and a different deposition technology which, despite having a lower electrical conductivity and higher cost, was controllable enough to go into production.
The engineering developments achieved in this project were amazing; we gave some very clever people a really tough challenge and they delivered some incredible hardware. With the benefit of hindsight, we probably should have applied some more effort to the fundamental science… were there some other metals or alloys that would have allowed a wider process window? Were there other ways of getting the metal into the interconnects? Were we keeping a close enough eye on competitive processes? We probably should have also investigated and understood other variabilities in the production process more fully – even if we could control our process perfectly, would other variations in the incoming materials and customer product have meant that our process couldn’t have worked every time?
In the next instalments of this blog, I’ll consider the importance of keeping the right people in the business. Telegraph Materials offers advisory services and practical support to fast-growth businesses bringing new materials-science based technologies and processes to international B2B markets. For more information, see www.telegraphmaterials.com, where you can also find the previous editions of this blog series.