Alfred J. van Roosmalen, Philips Semiconductors

Alfred J. van Roosmalen (Source: Philips Semiconductors)

Alfred J. van Roosmalen is the director of international partnerships and government affairs at Philips Semiconductors. In this position, he leads the company's worldwide network for leveraging public/private partnerships in cooperative R&D, interaction with academia, and new business creation. He chairs the ENIAC support group and is a member of the MEDEA+ board support group. Previously, he chaired the Sematech executive steering council and the European International Technology Roadmap for Semiconductors (ITRS) committee, and participated in the IMEC scientific advisory board and EC advisory group on human resources and mobility. Van Roosmalen has a M.Sc. in chemical technology from the Technical University Eindhoven, and received a Ph.D. from the University of Amsterdam. Philips Semiconductors (Eindhoven, Netherlands) is a leading supplier of solutions for mobile communications, consumer electronics, identification, and automobile entertainment and networking technologies. With ~37,000 employees worldwide, it had sales in 2005 of 4.6 billion euros.

SI: What is the status of the semiconductor operation now that Philips has sold a majority stake of it to an investment consortium?

Van Roosmalen: Following the recent announcement of the purchase of Philips Semiconductors by a consortium of private equity investors, the company is becoming a standalone semiconductor company (eighth worldwide and ready to move to second in Europe). The new name and brand introduction will be officially unveiled on Sept. 1. As a standalone semiconductor company with a strong financial foundation, we'll continue our existing strategy and business renewal program, but we now have the flexibility to pursue other strategic options, such as the acquisition of companies that fit into our strategy.

SI: You define the roadmaps for your company's technology and R&D programs, as well as whom you partner with. What's your current direction?

Van Roosmalen: We have a long history of partnering with other companies that dates back to the early 1990s. Now, it's a substantial part of our research and advanced development, as well as early manufacturing. In the Grenoble area, we have the well-known partnership in the Crolles2 Alliance with STMicroelectronics and Freescale. The Alliance has excellent relationships with a number of institutes and universities in Grenoble and elsewhere.

SI: In general terms, what's your R&D's current focus?

Van Roosmalen: Our growth area right now is in the CMOS arena. It's for us a major effort and major investment, which is why we have such an extensive partnering program. We have been doing CMOS research with IMEC for some time. Some five years ago, we transferred to the IMEC site our research on advanced CMOS as well as process options, and we have participated in their program ever since with a team of about 100 people who work there. They cooperate with IMEC as well as do our proprietary early work on process options in RF, power, high voltage, and embedded memory.

SI: Sounds like a practical model.

Van Roosmalen: It is. Besides the Moore's Law domain, we've been extending it to the More-than-Moore's domain, identified by the ITRS as a new dimension to create value as Moore's Law decelerates.

SI: You don't see it continuing?

Van Roosmalen: I'd be the last one to say that Moore's Law won't continue; it will, just not at the same rate. There's not much room left for things like power dissipation any more. We must create value elsewhere.

SI: MEMS is an option.

Van Roosmalen: Yes. We're looking at MEMS. Until relatively recently, MEMS seemed more like something that universities liked to play with — little practical or wide-scale application. But now microsystems are rapidly becoming mainstream. This is because there are many well-established older CMOS facilities that can do just about anything and are therefore well-suited to do microsystems, which simultaneously allows for easier logic integration. When you have a single MEMS device, you must have a way of driving it, packaging it. If you produce it in a factory that does CMOS, you can make it a part of the same design, as well as package it in some special way if you wish; for example, multi-chip modules have experienced great growth. We're actively using many of these modules as systems-in-a-package in our current product lines.

SI: Can you describe an application?

Van Roosmalen: Think of RF MEMS for multi-band switching in cell phones. This is an important emerging market. Just consider the multitude of bands that must be crammed into a single phone. Naturally, you don't want a transmitter for each, but rather prefer to reuse and combine them. Another application is tire-pressure monitoring. This is a requirement in the U.S., and I expect it to soon be in Europe also. For it you need a pressure sensor, an inertia sensor, a radio device, all in some small package that goes inside the tire. This opens another market for devices that have been around for a long time, although not in substantial volumes. Another is solid-state microphones. Look, for example, at the very thin cell phones you see nowadays. It requires a very thin microphone, and they've already gone about as far as is possible with a conventional microphone; and if one mike, why not several? Your phone could then better adapt its directionality in difficult environments, such as a noisy car.

SI: Unquestionably, the semiconductor industry is firmly rooted to the consumer market. There are no longer “killer apps.”

Van Roosmalen: It's practically all consumer, and I also include the automotive sector, which has been growing steadily. Semiconductor content in the car is not just control; it is sensing and networking as well. Cars are now like computer motherboards, only bigger.

SI: What do you see as some of the major drivers then?

Van Roosmalen: I don't think so much in terms of drivers as I do about things needing improvement or change. Everyone talks about hardware and advanced CMOS. Now we're going into 65 nm and we'll move to 45, and are confidently predicting we'll find ways of getting there, including the software that will run on that logic. I don't think it'll be that simple.

SI: Why is that?

Van Roosmalen: People think software is something that results when system designers give device parameters — somehow a way is found to develop it. The time when it was possible to improve speed and power dissipation from one generation to the next passed quite some time ago at the 0.18 µm node in the late 1990s. Since then, limitations were imposed by interconnect which, in turn, is also reaching an end. Solutions must be found at a system level, not at the design, device or circuit level. Digital devices are becoming increasingly analog-like. And system solutions will require software to be considered in terms of the hardware underneath. It'll become increasingly difficult to think about hardware-independent design. This was a very useful model for years, but it's doubtful how much longer this can continue.

SI: What's the solution?

Van Roosmalen: Hardware designers must think about the system that will run on top — the software. Software designers will have to consider what runs their programs. It is at this interface that effort must be increasingly exerted, and there are very few people left at this interface who can communicate with both hardware and software designers.

SI: Neither software or hardware can be an afterthought then.

Van Roosmalen: Absolutely! Both must now be created with a system-level mindset. A paradigm shift is needed in the concept of handing over responsibilities between the hardware and software engineering environments.

SI: How do you see the direction of the development of nanotech?

Van Roosmalen: I'm connected with ENIAC, the European platform for nanoelectronics. It's an effective organization, because this is the first time that not only industry is working cooperatively and thinking in term of long-range roadmaps, but also academia is being brought together at the same high level — institutes and universities — as well as the European countries and the European Commission. The latter was a key player in getting everything started. In this sort of environment, you're forced to think beyond just a decade. On the ENIAC platform, the scope of nanoelectronics goes beyond CMOS. It considers not only Moore's Law but also what lies beyond — non-charge-based devices, for instance. Packaging is also a long-range consideration; after all, it's rapidly becoming an integral part of the circuit itself. We're also closely looking at the connection with embedded systems and software. We must all work together, because there are too many challenges that are becoming common to everyone.

SI: With the increasing introduction of new materials, sometimes it appears as if the whole periodic table is being thrown at us. How do you view this development, particularly in the area of material characterization?

Van Roosmalen: Things reached a point, quite some time ago, where it became impossible for individual companies or universities to cope with it. In the European context, it's something on which even a large institute such as IMEC has admitted it cannot do fully on its own, regardless of its considerable resources and expertise.

SI: Surely it is not just a matter of cost?

Van Roosmalen: No. It's just that if we're trawling the periodic table in search of a gate dielectric material and you're running out of options, it isn't enough to be large — you need a network of people, companies and universities that cooperate with you, and will focus on one part while you focus on another.

SI: What do you view as the device designer's biggest challenge over the next five years?

Van Roosmalen: In the advanced CMOS environment, lithography's limitations are making things uncomfortable for the circuit designer. We're reaching the point at which the minimum dimension being patterned is about a quarter wavelength. At this level, there are several restrictions even if all you want is a few parallel lines. Some of these lines are beginning to vanish. This will make DFM the main production challenge. Another challenge is RF components. CMOS brings parasitics. If you're looking at high-speed digital logic, parasitics are increasingly becoming a problem. This is what I mean when I say that digital is becoming more analog. Digital design is rapidly turning into RF design, with all the complications that this brings with it.

SI: Sounds like we're in for interesting times.

Van Roosmalen: (Laughing) We are indeed!