15 February 2001
Sleeping technogiants are the MEMS
Just as transistors replaced vacuum tubes and changed electronics, these tiny silicon devices will establish a new technology base.
Not yet knee-high to an ant in the technological scheme of things, microelectromechanical systems (MEMS) are young technology. They've existed for fewer than 25 years.
These tiny mechanical devices, which are produced on silicon wafers using semiconductor production techniques, have long enjoyed a worldwide reputation as an advanced technology but have seen modest commercial success only in the past five to seven years.
Now, particularly with the advent of MEMS applications in medical and communications applications, the floodgates have opened, and it seems the whole world is turned on to MEMS.
Engineers are looking to MEMS to solve design challenges that otherwise could not easily be met. In such fields as telecommunications, medical devices, and instrumentation, MEMS components are making inroads because they are smaller, more integration friendly, and less expensive than traditional technologies.
MEMS' grand entrance into these areas follows commercial success in the 1990s in the automotive industry (airbag accelerometers), computer peripherals (ink-jet printers, disk drive heads), medical applications (disposable sensors), and the process control industry (pressure sensors).
MEMS now account for $3 billion to $5 billion in annual sales. MEMS will move into the mainstream of component technology. Application designers and engineers in a variety of fields must take note.
Etching in the sand
Like their cousin the semiconductor, MEMS devices are tiny devices fabricated using silicon and films deposited onto silicon. The biggest difference is that MEMS components are mechanical devices, often with moving parts, whereas integrated circuits (ICs) move and control only electrical signals.
A MEMS device can be smaller than the width of a human hair, containing one or more of a long list of moving parts: actuators, motors, toothed gears, wheels, inclined planes, hinges, and other parts.
To create these moving mechanical structures, several layers of different materials are stacked on one another. By selectively etching a part of a layer, one can construct patterns in the silicon material.
The resulting built-up device may move in response to heat, electrical signals, or magnetic flux. MEMS devices can also integrate with electronics at the chip level, expanding usage beyond strictly mechanical applications.
Following development of the first MEMS devices a quarter-century ago, researchers viewed MEMS as an exotic technology but not necessarily one that was designed to solve engineering or design problems.
Airbags inflate hope
For years, MEMS didn't venture much beyond the university campus. That changed with the commercial introduction of MEMS-based accelerometers in automobile airbags in 1993, which roughly coincided with the development and introduction of MEMS micronozzles on a chip for ink-jet printer cartridges.
The industry developed expertise around these and other specific applications where industry could design and manufacture MEMS in quantities that were profitable.
However, these applications were geared toward vertically oriented market segments (such as printers and airbag accelerometers), and the MEMS industry was not mature enough to provide for easy transfer of the technology to other applications.
More recently, MEMS have come to the forefront of technological advancement in telecommunications, biotechnology, and instrumentation. In the optical communications segment of telecom, MEMS hold great promise as components such as attenuators, switches, and cross-connects are being shown to offer enabling performance in optical networks.
Expertise seeking work
MEMS-based biotech devices, which will take longer to develop because of required regulatory approvals in the medical field, are starting to appear on the market.
In instrumentation, MEMS sensors and devices will be useful to dramatically reduce size and increase functionality in industrial and military applications.
In much the same way that the transistor replaced the vacuum tube and ushered in the era of electronics, MEMS are displacing traditional electromechanical devices to establish a new technology base for mechanics.
This new technology base, standing on the shoulders of 30 years of semiconductor manufacturing experience, provides two fundamental enablers. It significantly enhances performance and reliability for mechanical devices, and it provides a low-cost, high-volume manufacturing platform that did not exist previously.
From this base, new industries will grow, and existing fields will see a number of improvements.
One, two, three ... MEMS
Over the years, the MEMS industry has developed and refined its methods for making the devices. There are three basic steps: design, fabrication, and packaging. Through the work of researchers and companies dedicated to the development of MEMS, design and fabrication methods are standardized.
When it comes to packaging the devices so their mechanical properties are not compromised and then integrating them into systems and products, industry standards do not yet exist.
The manufacturing process typically begins with computer-aided design (CAD), in which MEMS CAD software companies have focused efforts to make MEMS design as simple and uniform as possible over a range of applications.
There are three fabrication methods, all of which have their roots in the semiconductor industry.
Build device up
Surface micromachining, developed directly from IC fabrication methods, uses the IC process of deposit, pattern, and etch. This method systematically builds the device up layer by layer.
Thicker films, however, are required for the mechanical behavior of MEMS devices. Because the final processing step is to free the moving parts by selectively etching a sacrificial layer, the device materials must be resistant to this final etchant (such as hydrofluoric acid).
The bulk micromachining process grew out of methods used in developing pressure sensors. The process involves etching the silicon wafer to form through holes and cavities for trenches, chambers, beams, and other structures.
A more advanced form of bulk micromachining known as deep reactive ion etching (DRIE) allows for more geometric freedom and control over the etching. The DRIE process is driving a wider acceptance of MEMS.
Tools for DRIE are readily available in the market, the process is standardized and well established, and the devices made from the process easily transition into manufacturing.
Throw in some metal
Lithographie, galvanoformung und abformung (LIGA, German for lithography, electroplating, and molding), while not used as commonly as surface and bulk micromachining, remains a good methodology for MEMS devices with high aspect ratios or tall structures with vertical sidewalls.
LIGA structures differ from other MEMS structures in that the moving mechanical parts of the device are made of plated metal rather than silicon.
Taken together, these three fabrication methods create a sophisticated actuation technology, which is the driving force for active MEMS components. Actuation triggers movement by thermal, electrostatic, electromagnetic, or piezoelectric impulses.
Perfection of these actuation technologies is leading to the development of MEMS actuators with robust and reliable performance, providing added displacement and force compared with traditional approaches. These device improvements have recently converged with the process standardization (see DRIE discussed above), driving MEMS technology to a new level of maturity.
As a result, the newest class of MEMS devices-microrelays, optical attenuators, photonic switch components, and such-is suitable for the demanding applications in optical communications, biotech, and other emerging areas.
Don't brown-bag chips
Packaging issues represent the biggest hurdle to widespread deployment of MEMS. Unlike ICs, MEMS devices must have protection for the moving parts. Epoxies used to underfill ICs are not suitable for MEMS devices because they can interfere with device movement.
Each MEMS component carries unique packaging requirements, and the MEMS industry has not put as many research and development dollars into uniform standards for packaging.
On the plus side, MEMS devices can integrate at the system level and are thus suitable for hybrid, monolithic, and flip-chip technologies. At the system level, packaging costs are lower because the relative value of the package technology rapidly diminishes as more devices integrate into that package.
In spite of the inevitable limitations that still exist, it is clear a more robust and widely applicable infrastructure for MEMS design and manufacture is emerging. With rapidly maturing building blocks of CAD, fabrication processes, standardized components, systems integration, and packaging, the MEMS industry has the tools in place to lead to an application-specific approach to MEMS, much like the application-specific nature of the IC industry.
As a consequence, the previous approach of vertical integration of unique MEMS technology into market niches is disappearing, and we see a much broader acceptance of more standardized approaches across many industries.
Switch to instrumentation
Aside from a maturing infrastructure, MEMS are muscling their way in on the strength of their advantages over more traditional component technologies-namely, size, form factor, performance, enabling functionality, reliability, and cost.
Just like economies of scale developed in the electronics industry through the use of semiconductor manufacturing techniques, MEMS stand to benefit from scale as processes become more standardized and widespread.
While many in MEMS believe the bulk of resources during the next few years will be devoted to the hot fields of optical communications and biotech, many other industries stand to gain from the advantages MEMS technology has to offer.
Instrumentation and process control industries, for example, are considering MEMS on several levels. In controller applications, MEMS devices replace some existing switches and relays at the board level.
In instrumentation, various forms of MEMS-based sensors are under consideration. In wireless communications, various forms of switches, inductors, and capacitors are emerging. IT
Figures and Graphics
Jesko von Windheim is vice president and general manager of Cronos Integrated Microsystems/JDS Uniphase, a MEMS designer and producer in Research Triangle Park, N.C. He holds degrees in chemistry, physics, physical chemistry, and business.