FEATURE FOCUS: Working Together—MEs and EEs

mixing it up

It takes a team of experts and a great deal of cross- fertilization to produce successful MEMS.

The name says it all: Microelectromechanical. The devices are produced by a team effort, in which electrical and mechanical engineers bring their own brands of expertise to bear on a number of tasks that often run concurrently. Different companies, from giant electronics firms to small start-ups, have their own unique ways of doing this. Yet many try, through frequent meetings, to encourage the exchange of ideas between engineering disciplines that fosters collaboration.

"The best design for MEMS is done by a team that is expert in both mechanical and electrical engineering. It's a multidisciplinary program," said Roger Grace, a sensor and semiconductor consultant based in Naples, Fla. Grace observed that the biggest applications for micro devices tend to be mechanical ones: pressure sensors, accelerometers, and inkjet printer heads. But new applications are reaching into different markets—bio, chemical, optical, radio frequency— that are adding new engineering disciplines to the MEMS mix. That trend is demanding flexibility, communication, and better design tools.

Michael Huff, founder and director of the MEMS Exchange in Reston, Va., a clearinghouse for MEMS design and fabrication centers, said the area is a specialty that typically imposes a learning curve on both electrical engineers and mechanical engineers. Electrical engineers familiar with integrated circuits typically want to rely on standard integrated circuit process technology, while mechanical engineers often lack fabrication experience and may not appreciate the degree of difficulty and cost involved.

THE BIG PICTURE

Companies that have successfully produced micro-scale devices talk of an environment in which teams of specialists work together toward a common goal.

Mike Judy is the MEMS design and computer-aided design manager of Analog Devices' Micromachine Product Division, located in Norwood, Mass. It produces accelerometers and gyroscopes. "The more effectively you can understand the big picture, the better you will be able to do your more specialized job," he said. Despite talk of generalists, team members' primary job functions in designing micro devices remain focused. For people to do their jobs effectively, Judy sees a need to bolster general knowledge while contributing expertise.

At Analog Devices, teams work together from early in the design process and continue to evolve as the project progresses. Small teams identify next-generation products. Typically, a marketing team will identify a niche for a specific product. Design teams start small—perhaps four or five people—and may grow to around 20 as the product develops, said Tim Brosnihan, a process development engineer.

Motorola capacitive pressure sensor's left-side diaphragm senses pressure changes; the thicker right-side diaphragm is the reference cell.

Early team members may include an electrical engineer who handles circuit design and a mechanical engineer. As a project progresses, teams discuss mechanical and electrical specifications to find a design that meets those specifications. Formal meetings take place weekly, yet team members meet informally more frequently. Collaboration is high, because much of the work runs concurrently, said Brosnihan.

Dave Monk is the systems development engineering manager for the Sensors Product Division of Motorola in Tempe, Ariz. He has responsibility for analog and mixed-signal design for transducers, and for MEMS design, package development, test development, and system engineering. Motorola's culture is strongly rooted in electrical engineering, yet a look at its MEMS operation brings home the multidisciplinary nature of the work.

Monk speaks of skill sets. Process engineers, who run fabrication processes such as etching and photolithography, typically have expertise in chemical engineering or materials science. The engineers responsible for the process flow of a given device are often electricals, who interact frequently with chemical engineers, materials scientists, and physicists. Generally, there are electrical or mechanical engineers who link product engineering and assembly. Project leaders, who may come from various backgrounds, tie it all together, and are responsible for project management, as well as for intra-project communication.

Recently, Motorola introduced two MEMS products, a tire pressure sensor and a low-g accelerometer. Products start out as a potential market opportunity; proposals are taken to a program steering group that often consists of general management and marketing.

The project enters a definition phase. A project leader is assigned who identifies its scope, required resources, and a general schedule. A planning stage follows when the lead people of different functional groups are brought in.

The sponsor of the project, the business manager, and the engineers get together to discuss what they think they can really do, taking into consideration the wishes of the customer. The tire pressure sensor, for example, required a new transducer, circuit, package, and test. Various engineers were involved in IC design, fabrication process development, packaging, and other aspects of the product.

The group also resolves timing conflicts and goes through a risk mitigation process.

The project then goes into an execution phase. In the case of the tire pressure sensor, this consisted of a team that included members with responsibility for transducer and analog and mixed-signal designs, process development in the manufacturing lab, and package and test development.

Product engineers provide the link to the assembly area. When the first prototypes are produced, the product engineer is responsible for providing input back to the designers and providing samples to customers.

Wafers at Motorola (top) are untouched by human hands. Microphoto (lower left) shows a high-aspect-ratio structure for inertial sensing. Making a finished package (lower right), a micro accelerometer is bonded to an application-specific IC inside a compact housing. The axis of sensitivity is determined by the chip, so package doesn't have to change with orientation.

Once the design appears practical, the working device undergoes qualification testing, such as shake, rattle, and roll tests, temperature cycling, and shock tests. The tire pressure sensor required a probe test, a testing of the wafer after completion of fabrication, before assembly. Any devices that don't pass this probe measurement are eliminated.

When the tests are complete, the device enters a ramp-up phase in which the manufacturing area is prepared for high-volume production. Documentation for qualification is completed and marketing materials are finished. Once the yields are in line and the group is meeting its cost estimate targets, the project goes back to the program steering group and the project is closed.

Monk said that this flow is fairly standard and includes once- or twice-a-month meetings with the project steering group, in which progress is discussed in detail. The tire pressure sensor project, which was rather broad in scope because it included new transducer, process, integrated circuit test, and package, took about four years from start to finish. Monk said the company tries to base new generations of products on the same platforms, so variations can be brought out much more quickly.

MEMS development is including ever more diverse disciplines as it reaches into new applications. Kurt Petersen, president of Cepheid, a Sunnyvale, Calif., supplier of MEMS-based test systems for DNA analysis, calls developing bio-MEMS products a very complex, interdisciplinary task. "An ideal situation is to have people who are not just electrical engineers or mechanical engineers, but who have experience on either side," he said.

It's the same throughout, he said. In the company's development area, mechanical engineers, electrical engineers, molecular biologists, chemists, quality assurance specialists, manufacturing engineers, and software developers are all lumped together intentionally.

Because of the relative complexity of Cepheid's instruments, mechanical engineers account for one of the highest portions of its technical staff, while electrical engineers are a relatively small group, Petersen said. There is a sizable software group, which is involved in building test systems. The largest group is bio, whose members develop assays and the various steps to process biological samples. A chemical group develops the reagents.

Communication is key, and mixing up various disciplines promotes exchanges of ideas. Core team meetings occur once a week, but informal meetings occur every day, Petersen said. He added that the facility is designed with plenty of small conference rooms that are constantly used. Core teams typically consist of 10 to 12 people.

"We make a really big effort to get quality and manufacturing people involved from day one," said Petersen. Most jobs are done concurrently. The one exception is fluidics. Biological chemicals react to various materials in different ways, so modifying designs or materials could change the biology, said Petersen. This hampers the biologist from doing a really thorough testing and assay development until the design is nearly complete, he said.

Michael Huff of the MEMS Exchange noted that MEMS is a difficult design environment because of the various fields —mechanical, electrical, and thermal—that interact with each other and must be modeled accurately during device design. "Simulation of that becomes an incredibly tricky business," he said. Another problem is the gap in the design tools used by electrical engineers and mechanical engineers. Linking the two often leads to errors.

At Corning IntelliSense, MEMS development is multidisciplinary. Wafers get careful handling (left) at the furnace in the fabrication facility. Team members in isolation suits (below) work in the lithography clean room. The results of their cooperative efforts could be something like these dual-axis actuated micro mirrors (lower left).

Mike Judy of Analog Devices notes that CAD software used in circuit design and finite element tools are separate, and there are very few examples of where the tools could be merged together. "It doesn't really overlap very well today," he said. Judy observed that MEMS software is continuing to bridge some shortcomings as it matures. Traditional finite element analysis software has begun to offer multiphysics coupled analysis, while specialized MEMS tools have become easier to use as they mature, he said. But he added, "Neither is there yet, as far as what MEMS companies really need."

Dave Monk of Motorola remarked that electrical engineers and mechanical engineers traditionally use different software. Electrical engineers are versed in two-dimensional layout tools, while mechanical engineers are more likely to be trained in finite element analysis or in solid modeling. Motorola's engineers use both: Spice or Cadence to design electronic circuits, and Ansys and solid modeling tools such as Pro/Engineer for mechanical design, he said.

At Motorola, resistors, capacitors, and transistors are built within an electronic engineering group, while transducers are built by a transducer design group. Transducer engineers must develop the mechanical model and translate that into an electrical representation that could be used in the electrical engineers' toolset.

"The real challenge is that we have got to get those systems to talk," said Monk. "We are still struggling."

Motorola has been working with various software houses to come up with a seamless design toolkit that would allow design files to be passed back and forth. That ability would help electronic engineers determine how a transducer affects the rest of the design, Monk said. He said a closer linking of the two could help manufacturers more clearly define product specifications, as well as manufacturing yields.

Paul Lethbridge, product manager for the electronics sector and the MEMS initiative manager at Ansys in Canonsburg, Pa., believes that electronics engineers and mechanical engineers tend to work more closely in the development of micro devices than their counterparts do in the macro world. He said that MEMS design and analysis software suppliers have made strides in applying multidisciplinary skills to design products of increasing sophistication.

PACKAGING AND TESTING

Tim Brosnihan of Analog Devices noted that engineers who handle packaging and in-package testing during manufacturing are involved early in the process. Those people have a lot to say about how the die is handled. If it were just a circuit die, there would be a variety of options. But mechanical elements require very specific ways to dice the wafer, handle the chips, and put them in packages, he said.

Packaging is largely the domain of the mechanical engineer, according to Monk. In MEMS, the mechanical engineers are working with silicon, where the system could be affected by the stress of the package, he said. The mechanical engineer must design the package and keep in mind the package's effect on the device itself.

A machine at Michigan Tech mills parts as small as 20 µm thick (top left). UV curing of bonded microtubes (right) is used to form a cochlear implant insertion tool. The tool changes shape when actuated (lower left).

In the bio-MEMS area, packaging involves an extra layer of complexity, because devices may combine fluidics with optical and electronic detection, said Peterson of Cepheid. There are also issues that are associated with human interaction and ergonomics.

Monk said that electrical engineers and mechanical engineers work closely together on testing. The company has a test development group that works with outside vendors to integrate test equipment into a system. He describes test systems at Motorola as semi-custom. Ideally these are standard on the electrical engineering side. Electrical engineers might build the test boards where the parts are mounted for testing and write the software for the tester. But testing also requires inputs and outputs for a mechanical component such as a tube or pressure chamber. Mechanical engineers would handle designing the physical stimulus module into the test.

STARTING EARLY

The flowering of MEMS design teams at many companies today sprouted from seeds planted much earlier at engineering schools. Roger T. Howe, professor and associate chair of electrical engineering at the University of California at Berkeley and director of the Berkeley Sensor and Actuator Center, said that the school had a goal of "blurring the identities" between electrical and mechanical engineering disciplines. His hope is that electrical engineering and mechanical engineering graduates could cross disciplines with-
out the concern of losing those identities. That is a trend that many in the industry credit for fostering how various teams work together to design MEMS.

Robert O. Warrington, the dean of engineering at Michigan Technological University in Houghton, said that prominent MEMS-focused doctoral-level programs, such as those at the University of California at Berkeley, are beginning to trickle down to the master's level at various universities.

Michigan Tech is a participant in the Wireless Integrated Microsystems program in cooperation with the University of Michigan and Michigan State University. An enterprise program at Michigan Tech, tied in with the Wireless Integrated Microsystems Center, allows engineering undergraduates in their sophomore year to participate in 30-person interdisciplinary teams.

Wafers at Analog Devices undergo inspection for defects (top). The die for an integrated angular rate sensor combines mechanical sensing and signal conditioning electronics on a single chip.

Michigan Tech has also initiated team teaching and course development by electrical engineering and mechanical engineering faculty, as well as by materials science and physics departments. The school emphasizes micromachining—milling, drilling, and electrical discharge machining at the micro level.

Jonathan Bernstein, vice president of technology at Corning IntelliSense in Wilmington, Mass., noted that historically micro fabrication labs have been within the electrical engineering departments of universities, yet students trained in MEMS are multidisciplinary, a phenomenon that is reinforced when students are hired by industry. "A lot of MEMS engineers started out doing electrical engineering, but they really don't do what is considered classical electrical engineering any more," he said.

Good communication between disparate groups is a critical component to designing and producing MEMS. This can only become more important as micro technology reaches new applications. Companies are trying to provide an environment that allows engineers with very specialized skills to collaborate on aspects of a project.

After all, they don't want to lose sight of the big picture. They are out to market a product that is more than the sum of its parts.

home | features | news update | marketplace | departments | about ME | back issues | ASME | site search

© 2003 by The American Society of Mechanical Engineers