of Lucent's Microelectronics Group, and one of the original authors of PISCES. Here is an interview on the future of TCAD:
Mark Law recently commented that SUPREM4 and PISCES were originally created to solve particular problems (OED and latch-up), and not to write simulators. Yet much of TCAD development since then has been on tool development, rather than problem solving. I think this has impacted the "relevance" of TCAD. Do you think TCAD has lived up to its potential?
I agree to a large extent with Mark. I do think the best TCAD developments are done with a specific problem in mind as opposed to blindly trying to solve the full boltzmann equation or whatever and then subsequently looking for an application or example. Ideally I think one should start with the driving example, pick a few possible new simulation directions (algorithmic and/or physics) to look at that might have impact, validate the new capability and explain something we didn't understand before, and finally predict something new based on the new capability. Now one might also start with the capability that would seem to be of general interest and find an example to drive it. In the end I believe the work is immensely more valuable - both for the code developed and the results on the specific problem. I do think that TCAD development more recently has been driven more by tool development than solving problems and has contributed to disappointment. I myself remain very strongly supportive of the role TCAD can (and does here) play.
Cadence recently bought the Bell Labs EDA department. But there was no mention of TCAD. Does this mean that Cadence wasn't interested in adding TCAD to its portfolio (as Avant! did last year), or that Lucent feels TCAD is too valuable, or perhaps you are negotiating with a 3rd party for distribution. Can you comment on this?
The spin-out of the Lucent EDA business did not include anything that was intimately connected to technology - e.g. compact models and TCAD. We strongly believe these need to continue to be done within Lucent and never even discussed them with any of the bidders for the business that was sold. That said, we are interested in ways the industry can work together to improve the TCAD infrastructure yet allow us maintain proprietary advantages where this makes sense. We believe the best approach to this is common platforms, APIs, etc.. I'd ask Conor Rafferty for his comments on this - I am not up to date on how the community has taken to these ideas.
The TCAD market is much smaller than the ECAD market, perhaps only a few percent of the ECAD total. Yet some of us in the TCAD field like to think TCAD is nearly as important as ECAD to the development of new products. How can we, as a TCAD community, ensure continued development of TCAD in lieu of the small market size?
Again I would say common platforms where we don't really need to be differentiated. Thereby we can concentrate on key models, etc. and even share and collaborate on these where appropriate. My assessment of the difficulty here is that the current embedded base of vendors doesn't see a business model for them in this approach.
Lucent has always had a strong simulation activity, and presumably, with you in your current position, that will continue. Yet in many companies, TCAD still struggles for respect. How do you think this dichotomy will resolve? Will TCAD become widely accepted, or will it fall out of use except in a few companies?
Without the acceptance of the platform type model, I think the dichotomy is likely to continue because I don't see the tools vendors doing the things necessary to meet our company's needs. They simply aren't equipped. I don't know enough about experiences with the commercial tools to judge if TCAD will fall out of use if the current model continues.
This page last updated Sept 16, 1998 by