Page 28 - EE Times Europe Magazine | June2020

28 EE|Times EUROPE

Wanted: Process Engineers Versed in Packaging

setup and hold time. I did not understand how Raychowdhury: I think one of those areas EETE: So when you showed up, what
complex that was before I started working in would be that process engineers now need to things did you wish you had studied
the industry. And if you really want to under- understand packaging. The way I look at that before joining the company?
stand what hold time is, and if you design a trend now, if you look at what AMD is doing Raychowdhury: I had all the background
flip flop and how you clock the two stages of a with the Zen processor or what Intel is doing knowledge, but what I was missing was, I
flip flop so that you can get the best hold time with Lakefield, you see it’s all about hetero- didn’t know how to connect the dots. How is
… these are the things that you only learn by geneous integration of the package. A lot of this particular subject material connected to
building it and seeing how it works and then back-end engineers in these companies are that subject material?
deploying it in millions and billions. We need now going to use their skills to build dense I think that’s something that you don’t
to get someone to write about it. I think you packages. I can see that happening. learn in college, because you’re learning
need someone who, first, [knows] the theory Many of our research programs [at Georgia all these courses in separate semesters. At
but also understands what’s going on in prac- Tech] are industry-funded. So I can see that some point, all of these things need to click
tice. This is exactly the kind of thing missing in those in the industry are asking us to look together. My first six or eight months at
books. It’s missing in most articles as well. at these kinds of things — work from a design TI were essentially spent trying to under-
perspective as well as from a processing stand how these different components are
2D, 2.5D, AND 3D INTEGRATION perspective. connected.
EETE: In the academic world, it takes There is very little understanding, at least Universities now are trying to do more
time for a book to be published. It takes in the U.S., about how this transition is going hands-on design-based courses in the senior
time for a course to be developed to to happen. TSMC is probably doing a better year and even for graduate students, which I
cover a technology. Are there things job of making sure that people can under- think will help a lot.
you see that you wish you had an stand that this is a smooth transition between Some people have the natural ability to
opportunity to create a course for? what is on-die versus what is on-package. And hack things and write software and break
Raychowdhury: For example, look at I think this would be one area where I would things and put them back together. I think
technology today, and people are saying that suspect people need to be ready very soon. they are better engineers. They will be able
transistor scaling probably is coming to an end, to connect better when they enter the work-
whatever it means. Technologically speaking, MOST LIKELY TO SUCCEED force. So I think more vertically integrated
maybe scaling isn’t going as fast as we would EETE: As a professor, when you see a projects and courses are important.
want. But in terms of memory technologies, for student start work at TI, for example,
example, or back-end-of-line transistor tech- what general knowledge do you think he NEW COURSES
nologies, I think there are lots of new things or she should have? What are the basics EETE: As you develop new courses for the
happening. The industry is moving very fast in — other than a personal specialty — that next year, what are the things that you
that particular domain. And there are no good he or she should absolutely know or think will be must-teach topics?
books, because these are all the black magic seek to know? Any advice? Raychowdhury: I teach circuit courses.
that the industry talks about. Raychowdhury: I feel like people who do From a circuits perspective, I think a lot of
Another trend that I see, which is not well in the industry eventually have a broad students have gone down to the deep end.
covered in books as well as it should be, is understanding. Today, I tell my students to They’re just interested in, you know, under-
integration — 2D, 2.5D, and 3D. We all won- take courses all across the board, not just standing how EDA tools work. Like, how do
der: What does it mean, then, and what are courses in your area of research. A good you know the different new options that
the different implications? example would be people who are doing pro- Synopsys has come up with or new back-end
But [discussions about it in the industry cess technologies. They need to understand tools Cadence is producing? And they want
are] optimistic. They talk about layers and physics and chemistry really well. to just do projects and get the skill set.
layers of transistors, which is never going to Device people really need to understand I’m actually taking the fall semester
happen. Even if it’s technologically feasible, materials. If they don’t, it’s a problem. It’s off and trying to build a new course I can
it’s just economically not feasible. Even for harder for materials people to understand use in the spring when I go back. I want
the short-term means of getting there — like devices and technology. I think understanding to teach the fundamentals. When grad
the chiplet technologies that Intel has been how basic chemistry works helps. Because as students — and I mostly teach graduate
working on. I don’t think there are good books we change the paradigm, we are looking at students — graduate without having a good
on [that]. new materials. understanding of signal and noise, that’s a
Similarly, I feel like a lot of circuit design- problem. And these are circuit designers.
TRANSITION FROM PROCESS TO ers, particularly analog designers, who go In the VLSI space, we don’t have a lot of
PACKAGING into the industry do not have or have forgot- good fundamental courses. That’s mostly
EETE: Are there emerging technologies ten basic math. So I think they have to because of demand from the students.
that we ought to be highlighting or we relearn those things once they go to TI or Students mostly want things that can help
should be bringing to people’s attention whatever company. their skill set. They don’t really understand
or getting them prepared to be able to Sometimes they have the skill sets, but they that it doesn’t take long to pick up those
use? There’s a lot of excitement about do not necessarily have a broader, general skills after you graduate. They need more of
things like quantum computing, but in understanding of engineering — including a fundamental understanding.
reality, quantum computing is a decade math, physics, and chemistry, which are often If you look at the VLSI curriculum — and
away. [Are there] things that are a little required. this is common across the entire country
closer so that we might say, “Hey, you — none of the schools are teaching funda-
know, in the next year or two, you might EETE: When you showed up at the mental VLSI circuit design. ■
be called upon to work in this area, and doorstep at TI, you were the young guy.
here’s some background information to Did you already have a Ph.D. at that time? Junko Yoshida is global co-editor-in-chief at
get you started”? Raychowdhury: No. AspenCore.

JUNE 2020 | www.eetimes.eu