Printable Version

MEMS — A Small World with Big Opportunities

SwRI engineers take on the big challenge of working on the very small scale

By Heather Hanson


Heather Hanson, shown in front of a scanning electron microscope, is a senior research engineer in the Medical Systems Department in the Automation and Data Systems Research Division. Hanson has designed mechanical components, assemblies, packaging and fixtures for single unit, low volume and high volume production. Her work in microelectromechanical systems includes actuator and sensor design and development.


In modern technology applications, sometimes smaller is better - and microscopic is better yet. Microelectromechanical systems (MEMS) have moved out of the basic research arena to become a useable technology.

What are MEMS? The short answer is microscopic machines; however, they are much more diverse than that. The current driver in MEMS applications is the ability to make an existing device microscopically small, or to create a new device that would not work if it were inches in size but that works well at the micron scale. MEMS devices also potentially can be manufactured for low cost at high volumes, mirroring the semiconductor industry upon which they are based.

MEMS are an enabling technology, a building block for solving problems in nearly every technical field. They often are used to make sensors, including the air bag sensor in most modern automobiles. In other applications they interact with their environment to change it in some way, such as an ion propulsion device that can move tiny satellites in space or an optical system that diverts light beams. Sometimes, they interact with themselves, such as the timed-locking mechanism on a nuclear warhead. Such a mechanism comprises gears and linkages that can open a switch with the correct electrical input. Because they are micro-scale, they can be installed in small spaces. They also can interact with molecules, opening up a new realm of possible chemical, biological and medical applications.

MEMS researchers at Southwest Research Institute (SwRI) outsource the production of MEMS, enabling them to select the process steps best suited to the design of a particular device. The selection of a fabrication facility is an early part of the design process. This process determines such requirements as layer thicknesses, number of layers, materials, minimum feature sizes, residual stress and final package size. As a multidisciplinary research and development organization, SwRI provides the wide range of technical expertise needed to design, develop and package MEMS devices for a variety of applications.


An SwRI-developed MEMS device undergoes testing in a probe station. The one-centimeter-square chip contains more than 60 different mechanical devices. The microprobe tips provide drive signals and sensing connections to the chip.


MEMS methodologies

MEMS devices are fabricated using methods originally created to process the small features necessary for integrated circuits. These processes require a layer-based design - meaning that three-dimensional structures are created by stacking layers of material(s). Initially the materials were polysilicon and metals, but the technology has progressed to include numerous other materials, including polymers. Typically, the layers range in thickness from 1 micron to 1,000 microns, depending on the fabrication process. The diameter of a human hair is around 90 microns, fitting easily in the realm of MEMS.

The most common processes in use today involve photolithography, or etching a design into photosensitive materials. In this process, a structural material of constant thickness is deposited onto a substrate chip. A photoresistant material is applied onto the structural material in a particular pattern of interest. The structural layer is then etched according to the photoresist pattern, and the photoresist is removed. This process is repeated until the desired layers have been placed. Some of the materials deposited throughout this process are "sacrificial." When the deposition and photolithography processes are complete, the entire chip is exposed to an etchant, which removes the sacrificial layer(s), thus making room for the other layers to move. There are many variations of this basic process, which can be categorized into bulk micromachining, LIGA (a process that makes parts with significant depths from metals, metal alloys, plastics or ceramics) and surface micromachining, to name a few.

SwRI has extensive experience in designing MEMS devices. Outsourcing allows Institute staff members to custom design the devices and select the MEMS methodology best-suited for a particular project.


Although this MEMS device has more than 100 functioning mechanical devices, the entire chip can rest on a fingertip. MEMS devices are incredibly small yet highly functional.


Methods of actuation

Researchers can make small devices: but how do they work?

MEMS devices employ several basic methods of actuation. The most common is the electrostatic method, which requires high voltage (more than 100 volts) but little current. A receptor plate is charged, which causes it to be attracted to a grounded plate. One or both of these plates can move toward the other. Sometimes these plates are bonded on one end, forming a cantilever beam or a series of interdigited comb fingers. In other cases, they can be spokes on a wheel that the electrostatic attraction causes to turn. This attraction produces the motion and is repeated to obtain a desired effect. This effect may be raising and lowering mirrors in an optical device at a particular frequency, pumping fluid through a microchannel or operating a micromotor.

Joule resistive heating is another common actuation method that requires a more moderate voltage and current (about 5 volts, about 5 milliAmperes). This method can be used to produce devices containing either a bimetallic construct or a single material of varying widths. The bimetallic version requires two materials to be bonded together that have differing thermal expansion coefficients. When the current is applied, the two materials expand at different rates, causing a straight beam to bend in the same manner as its large-scale counterpart, the thermostat strip. A single-material version must use varying thicknesses or widths for the two beams that are attached to each other on one end. The difference in cross-sectional area causes the arms to heat at different rates, producing the same beam-bending effect. Unlike the electrostatic method, joule resistive heating cannot be used to resonate a structure because the structure must have time to cool between cycles. It can be used for discrete motions, similar to a mechanical relay switch.

MEMS devices also can be activated in other ways. A magnetic field can be applied to attract a magnetic material. A voltage can be applied to a device made from a piezoelectric material to cause motion. In the area of microfluidics, a stationary actuator may cause the motion of the fluid. For example, an ultrasonic pump uses piezoelectric material to create a surface acoustic wave. This wave, also known as acoustic streaming, causes motion in the contacting fluid. Electrohydrodynamic pumps use a traveling electric field to pull a dielectric fluid along a desired channel path. This pump and similar electrokinetic pumps are used in biomedical applications for such things as separating DNA molecules.


The illustration depicts the steps used in the surface micromachining process. A layering process is used to create MEMS components and obtain air gaps between layers that allow the components to move.


Small size, big challenges

The small size of MEMS devices creates design challenges. One of these is "stiction," the result of two surfaces not having great enough stiffness to overcome surface adhesion. During processing, liquids can stay between layers rather than drying out, pulling the layers together via capillary forces. Once the surfaces are held in contact in this manner, the devices are useless because the layers are not strong enough to overcome the capillary forces holding them together. Even room humidity can be the source of the liquid agent that causes this phenomenon. This can happen in varying degrees, and is most often seen as a type of friction that requires more force to overcome than had been planned.

As described earlier, MEMS devices are designed and fabricated using layers. The 3-D structures are made from these layers. Therefore, devices that need to stand up off the chip, such as mirrors, must be made flat and be able to lift up on their own or by an actuator on the chip, because they are too small to be effectively handled manually. MEMS devices can be manually activated using a microscope and special probes, but it is only practical to do this in a research laboratory setting, such as at SwRI.

The manufacturing process causes another difficult challenge for MEMS. When a layer of material is deposited, all of the atoms in the layer are not in the same energy state. This results in a compressive or tensile stress inside the layer, known as residual stress. After the sacrificial layers are released from the chip and the structural layers can move, structures made with high residual stresses will bend upward, off the chip (tensile) or downward toward the chip (compressive). Devices can be designed to make use of residual stress, but researchers usually design around it.

Once devices have been fabricated and assembled, they must be protected from their environment, yet in many cases they must be able to interact with their environment. Thus, the final packaging is typically individualized for the given device. For example, a chemical sensor must be in contact with the chemical it is sensing. If this chemical is in liquid form, it must be injected into the MEMS chip in some manner. If the chemical is in gaseous form, the MEMS device must be in contact with air but not clogged by dust particles because, to a MEMS device, a dust particle is a large object. Of course, all devices must be protected from vibration and handling.


Institute engineers developed vertical thermal actuators operated by a differential expansion between two layers of dissimilar materials in each arm. To obtain the initial upward curvature, the device takes advantage of residual stresses in the film layers. An electrostatic version with integrated flip-up mirrors was used to produce an optical switch. SwRI holds U.S. patent No. 6,608,714.


Solving industry problems with MEMS

The SwRI MEMS team seeks ways to apply MEMS technology to industry problems. An example of this is a stress corrosion-cracking sensor (see sidebar) being developed using internal research funds. Current methods available for in situ monitoring of stress corrosion cracking are too cumbersome to install in the minimal spaces on an airplane, for example. Because of this, airplane maintenance workers often replace parts according to a given schedule rather than first showing that stress corrosion cracking has occurred or may occur soon. MEMS fabrication techniques allow for smaller fabrication of standard test structures that could be employed in small locations.

SwRI project teams can consist of a variety of experts across several technical areas. For example, the SwRI team that designed an in vivo drug delivery system included experts in microencapsulation, chemistry, mechanical engineering, fluid dynamics, electronics, bioengineering and MEMS.

Conclusion

MEMS have the ability to impact almost every technical field. Their small size, high volume and low cost enable the creation of a suite of disposable sensors and devices. They can interact with the environment at the molecular level to achieve new goals.

Comments about this article? Contact Hanson.

Acknowledgments
The author gratefully acknowledges the contributions of the following SwRI staff members to the preparation of this article, as follow:

  • "Relieving Stress," Manager Dr. Sean Brossia and Research Engineer Andy Veit.
     
  • "In Motion," Research Engineer Joseph N. Mitchell.
     
  • "New Methods," Institute Scientist Dr. Steve Hudak, Principal Engineer Dr. Dan Nicolella, Research Engineer Joseph N. Mitchell and former SwRI staff member Rick Fess.

Published in the Winter 2004 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Joe Fohn.

Winter 2004 Technology Today
SwRI Publications SwRI Home