 |
August 24, 2001
Valluri Rao is an Intel Fellow in Intel's Technology and Manufacturing Group. He was formerly responsible for directing the development of advanced analytical tools and methods for microprocessor performance characterization, silicon debug and yield enhancement. Rao is currently directing Intel's MEMS research and development activities.
Q1: Valluri, you've been with Intel longer than most - tell us what you've been up to.
A1: I joined Intel in 1983, and my first project involved developing new silicon debug tools for the Intel® 286 microprocessor. Subsequently I enhanced this technology to adapt to higher clock speeds, pin-count and power dissipation for the Intel® 386, 486 processors, and the Pentium® generation.
As clock frequencies rapidly climbed to over 500MHz, Intel moved to flip chip packaging and I was responsible for developing new chip diagnostic tools that combined Physical and Design For Test (DFT) techniques for product debug with speed improvements. These technologies are now the workhorse physical silicon debug methods used throughout Intel for microprocessors.
My job over the years has really involved taking new capabilities from the lab in the research phase and bringing them into mainstream use, in a very short timeframe. Over the years I was fortunate to have a team of very talented engineers work for me - the success of these projects was a direct outcome of this teamwork.
Q2: Your recent emphasis has shifted to Micro Electro Mechanical Systems (MEMS) research and development. Can you tell us more about this?
A2: In 1999, Intel made the decision to play a more active role in developing MEMS as an emerging technology. Because MEMS is based on semiconductor fabrication and micro-machining - which overlapped with some of my areas of expertise - I was asked to explore the extent to which MEMS might support Intel's business plans. After an initial six-month exploratory phase, I realized there was a great deal of potential and drafted a MEMS research and development proposal. This was eventually supported by Intel's Technology and Manufacturing Group and funded by Intel Research, in Intel Capital.
Q3:What exactly are MEMS?
A3:MEMS is essentially a technology that is used to create micro-miniature mechanical devices out of silicon. These devices can respond to external stimuli - for example, in sensing applications, and on-chip voltages can be used to move (or actuate) these mechanical structures. MEMS technology is used to build accelerometers in automobile airbags, pressure sensors, flow rate sensors, and so forth. Micro-mechanical mirror arrays have also been developed for projection display applications. MEMS is based on fabrication technologies similar to CMOS, with the added ability to incorporate moving and mechanical structures.
For example, let's look at potential methods of cooling some of the new, high-power processors. This could be done most efficiently with micro-machined channels that are attached to the chip. This means making miniature pumps and condensers - essentially micro refrigerators that cool the processors. We can also use MEMS to build micro-radiators, which is similar to an automobile engine. We are funding some universities that are spearheading research in this area.
Q4: Can you define "small" in the MEMS technology?
A4:Compared to today's leading edge silicon process technology, which is at 0.13 to 0.18 microns, MEMS devices are relatively large - typically ranging from one micron to 100 microns.
Q5: You referred to MEMS as an "emerging technology." What exactly does that mean?
A5:Just as silicon was an emerging technology 30 years ago, MEMS is at an early stage today. The difference is that MEMS has the potential to develop much more quickly, as MEMS is based on many of the same processing techniques as conventional silicon. And since silicon has long been one of Intel's core competencies, we expect to be able to advance this technology very quickly.
Q6: What is Intel doing with MEMS research?
A6:We have a small group of engineers working in MEMS research and development, complemented by Intel's investment in companies with parallel interests. We haven't moved into volume manufacturing, but prototype fabrication is happening in both the United States and at our fab in Israel. By engaging with selected companies, as well as with universities, Intel is using its leadership position in silicon technology and putting the benefits of MEMS to work in some pretty exciting areas.
Q7: Like what?
A7: We're focusing on some specific key applications - MEMS for wireless communications and bio-medical technologies. For example, there are passive components in the wireless area, such as filters, switches and tuning elements used in cell phones, cable modems, and LANs. These passive components take up a lot of real estate on printed circuit boards. MEMS allows some of these high-value passive components to be fabricated out of silicon and integrate with each other. This could lead to reductions in size and cost, plus improvements in reliability. We are also exploring applications where MEMS can enhance our core products. Chip cooling is an example of this.
Q8: How can MEMS advance biotechnology?
A8:Earlier we discussed the use of MEMS to cool high-power chips; there's a related application in the bio-medical field. If we think back to the chip, with channels and pumps, we see an alignment with fluid analysis for medical diagnostics. Biochips essentially pump fluids through long channels, leading to a separation of molecules that can then be analyzed with sensors. The biochips enable rapid detection of the chemical composition of the fluid, which is potentially a valuable tool for physicians.
Q9:So Intel is entering the bio-medical field?
A9:No. An internal research effort has been started by Intel Research, to explore and exploit future opportunities in this space. This is technology that, in high volume, can be advanced with Intel's silicon experience. The bio-medical field is expected to grow in the next decade, and Intel's research and development in the MEMS area can help support this important technology.
Q10:When do you think we'll see MEMS in our everyday lives?
A9:While MEMS is considered an emerging technology, it's already part of things we take for granted. The sensor that triggers the airbag in your car is MEMS-based, as are the pressure, temperature, and other engine controls. MEMS micro mirror chips can now be found in projection display. MEMS is also beginning to appear in wireless communications, optical networks, and bio-medical technologies.
Q11:If we're in the early stages of MEMS research now, where do you see us four or five years from now?
A9:The areas we've discussed - wireless, optical, and bio-medical - will continue to be key areas of focus. MEMS devices serve the input (sensing) and output (actuation) function, taking analog information into and out of the digital computation world. As such, they will need to scale in performance and size with the computational engines with which they co-exist.
The term "MEMS" has actually evolved over the years to embrace not only chips that have moving structures, but a broader range of devices that are fabricated with micro-machining. I personally feel this is what MEMS means today.
The technical processes will continue to mature, and solutions for some of today's challenges will begin to surface. There will have to be breakthroughs in how MEMS devices are packaged, and in the integration of MEMS devices with silicon CMOS chips. Right now every MEMS chip requires a custom package and integration scheme - industry standards must evolve to address this.
As a technologist, this is the excitement in what I do at Intel. I have the unique opportunity to be part of something with great potential. It will be very exciting to take this technology from the lab and into real-life applications.
|
|
