MEMS

Based on Wikipedia: MEMS

Your smartphone knows which way is up. It tracks your steps, adjusts its screen brightness, and can even detect when you've fallen. Your car's airbag deploys in milliseconds. The projector in your local cinema produces stunningly sharp images. Behind all of these capabilities lies a technology so small you could fit thousands of its components on the head of a pin: Microelectromechanical Systems, or MEMS.

These are machines smaller than a grain of sand—and they're everywhere.

The World at a Hundred Micrometers

To understand MEMS, you first need to grasp just how small we're talking about. A human hair is roughly 70 micrometers wide. MEMS components range from 1 to 100 micrometers—meaning some are thinner than a single strand of your hair. The devices themselves, assembled from these microscopic parts, typically span 20 micrometers to about a millimeter. That's small enough to dance on the tip of a pencil, yet complex enough to sense motion, measure pressure, or redirect beams of light.

At this scale, the world behaves differently. The physics that govern your everyday experience—gravity pulling things down, objects moving when you push them—take a back seat to forces you rarely notice. Static electricity, which at human scale gives you an annoying shock when you touch a doorknob, becomes a primary engineering concern. Surface tension, the force that lets water bugs walk on ponds, suddenly matters enormously. Viscosity—how "thick" air feels—becomes surprisingly significant when your entire machine is smaller than a speck of dust.

This is what makes MEMS engineering so fascinating and so challenging. You can't simply shrink down a regular machine and expect it to work. You have to design for a world where electromagnetic forces and fluid dynamics dominate, where the ratio of surface area to volume changes everything about how things behave.

A Dream Before Its Time

The famous physicist Richard Feynman saw this coming. In 1959, he delivered a legendary lecture called "There's Plenty of Room at the Bottom," imagining a future where we could build machines at the molecular level. He described inscribing the entire Encyclopaedia Britannica on the head of a pin. He speculated about tiny surgical robots that could swim through your bloodstream.

The audience was intrigued but skeptical. The technology to build such things simply didn't exist.

It would take decades before Feynman's vision became reality, and the key breakthrough came from an unexpected source: the semiconductor industry. The same techniques used to etch transistors onto silicon chips—the technology powering computers—could be adapted to carve out tiny mechanical structures. By the mid-1960s, researchers were creating the first primitive MEMS devices.

Harvey C. Nathanson and Robert A. Wickstrom built something called a resonant-gate transistor in 1965. It was an adaptation of the MOSFET, or Metal-Oxide-Semiconductor Field-Effect Transistor, the fundamental building block of modern electronics. But unlike a regular transistor, this one had a tiny moving part—a mechanical resonator that could vibrate at specific frequencies. Around the same time, Raymond J. Wilfinger patented the "resonistor," another early electromechanical device.

These were proof-of-concept experiments, curiosities in the world of electronics research. Nobody quite knew what to do with machines this small.

A Name for the Revolution

For two decades, researchers worked on these microscopic devices without a unifying term. They were called microsensors, microactuators, micromachines—whatever seemed appropriate for the particular application. The field lacked an identity.

That changed in 1986 when S.C. Jacobsen and J.E. Wood at the University of Utah submitted a proposal to DARPA, the Defense Advanced Research Projects Agency—the Pentagon's technology incubator that had previously funded the creation of the internet. Their proposal introduced a new term: Micro Electro-Mechanical Systems, or MEMS. The acronym stuck. By the late 1980s, researchers worldwide had adopted the terminology, and MEMS emerged as a recognized field of engineering.

The DARPA connection wasn't coincidental. Military applications drove early MEMS development. Tiny sensors could make weapons smarter, navigation more precise, surveillance more discreet. But the technology quickly spread beyond defense applications into consumer electronics, automotive systems, and medical devices.

How to Build Impossible Machines

Building something you can barely see with the naked eye requires manufacturing techniques borrowed from the semiconductor industry but pushed in new directions. The fundamental process involves three steps: depositing material, patterning it, and etching away what you don't want. Simple in concept, extraordinarily complex in execution.

Deposition comes first. Engineers need to lay down thin films of material—anywhere from a few nanometers to about 100 micrometers thick. They use two main approaches.

Physical vapor deposition works almost like spray painting at the atomic level. Material is blasted off a target surface, usually by bombarding it with ions. These liberated atoms float through a vacuum chamber and settle on the surface where you want them. Alternatively, you can simply heat the target material until it evaporates, letting the vapor condense on your workpiece like dew forming on a cold window.

Chemical vapor deposition takes a different approach. You flow gases over your surface, and chemical reactions at the surface build up the material you want. It's like growing a crystal from solution, except you're growing a precisely controlled film one atomic layer at a time. Variations of this technique use different pressures and energy sources—low-pressure versions for uniform coatings, plasma-enhanced versions for lower-temperature processing.

Writing with Light

Once you've deposited your material, you need to shape it. This is where lithography enters—literally "writing with stone," though we've moved far beyond stone tablets.

The surface is coated with a photosensitive material called a resist. When light hits the resist, it changes chemically. Some resists harden when exposed; others become soluble. By shining light through a mask—essentially a stencil at the microscopic level—you transfer a pattern onto the resist. Where the light hit, the resist changes. Where the mask blocked it, the resist stays the same. Wash away one type or the other, and you've created a precise pattern for the etching step that follows.

But there's a fundamental problem with light: diffraction. Light waves bend around obstacles, blurring edges at extremely small scales. For features smaller than about 100 nanometers, optical lithography starts to struggle.

Electron beam lithography sidesteps this limit entirely. Instead of light waves, it uses streams of electrons, which have much shorter wavelengths and can define features just a few nanometers across. The downside? Speed. A light beam can expose an entire wafer in seconds through a mask. An electron beam must trace out every feature individually, like drawing with a very precise but very slow pen. Exposing a single wafer can take hours. If the beam drifts during that time—even by nanometers—your pattern is ruined.

More exotic techniques push the boundaries further. Ion beam lithography can write lines less than 50 nanometers wide. X-ray lithography uses radiation with even shorter wavelengths than ultraviolet light. Each technique trades off precision against speed, cost, and complexity.

The Art of Controlled Destruction

With your pattern defined in the resist layer, etching removes the material you don't want. This is where things get interesting—and where a lot of MEMS magic happens.

Wet etching is conceptually simple: you dunk your workpiece in a chemical bath that dissolves the exposed material. Hydrofluoric acid, for instance, eats through silicon dioxide but leaves pure silicon largely alone. This selectivity is crucial—you want to remove specific layers while preserving others.

But wet etching has a peculiar property in crystalline materials like silicon: the acid attacks different crystal planes at different rates. In potassium hydroxide solution, one particular plane of the silicon crystal lattice etches about 100 times slower than others. This means if you etch a square hole in a silicon wafer, you don't get a square hole. You get a pyramid with walls angled at precisely 54.7 degrees—the angle between crystal planes.

This sounds like a limitation, but clever engineers turned it into a feature. By understanding the crystal structure of their material, they could predict exactly what three-dimensional shapes the etching would produce. V-shaped grooves, precisely angled cavities, intricate stepped structures—all became possible through controlled chemical attack.

An interesting historical note: hydrofluoric acid etching of glass dates back to medieval times. Long before anyone dreamed of microchips, artisans used this dangerous acid to decorate glassware. Today, that same chemistry enables technology Feynman couldn't have imagined.

Etching Without Liquids

Dry etching takes a completely different approach. Instead of chemical solutions, it uses plasma—an energized gas where electrons have been stripped from atoms, creating a reactive soup of ions and radicals.

In a plasma etcher, gases containing fluorine or chlorine are energized until they become plasma. The reactive species attack the exposed surface, converting solid silicon into gaseous compounds that float away. Because the reactive particles come from all directions in the plasma chamber, this etching is typically isotropic—it cuts sideways as fast as it cuts down, undercutting the mask above.

But adjust the plasma conditions and you can change this behavior. In reactive-ion etching, you accelerate ions toward the surface with an electric field. They arrive traveling mostly straight down, attacking the horizontal surfaces far more aggressively than the vertical walls. The result is anisotropic etching—you can cut deep, narrow trenches with nearly vertical walls.

Deep reactive-ion etching, or DRIE, pushes this further still. By alternating between etching and depositing protective polymer layers, engineers can cut features with aspect ratios—the ratio of depth to width—that seem almost impossible. Trenches a hundred times deeper than they are wide. Holes punched cleanly through an entire silicon wafer.

One particularly elegant dry etchant deserves mention: xenon difluoride. This is a crystalline solid at room temperature that releases fluorine atoms when it encounters silicon. Unlike liquid etchants, it doesn't create surface tension forces that can pull delicate structures together and stick them in place—a phenomenon called stiction that has ruined many a MEMS device. Xenon difluoride was first applied to MEMS at UCLA in 1995, and it's become a standard tool for releasing fragile moving parts.

Switches That Bend

What do you do with all this microscopic manufacturing capability? One of the most straightforward MEMS devices is a switch—something that opens and closes an electrical connection. But at the MEMS scale, switches work differently than the toggle on your wall.

Capacitive MEMS switches use a moving plate that changes the capacitance of a circuit—its ability to store electrical charge. Move the plate closer to a fixed electrode, and the capacitance increases. Pull it away, and capacitance drops. The circuit doesn't actually open or close; it's the change in capacitance that signals the switch state.

Ohmic switches are more like traditional switches, making or breaking actual electrical contact. They typically use tiny cantilevers—microscopic diving boards—that bend when you apply a voltage. The electrostatic force between charged surfaces pulls the cantilever down until it touches a contact point, completing the circuit. Release the voltage and the cantilever springs back up.

These switches face failure modes that don't exist at larger scales. Metal fatigue in the cantilevers causes them to deform after millions of cycles. The contact points wear down, eventually failing to conduct. Engineers must design for billions of operations in devices that cost pennies to manufacture.

Where MEMS Meet the Everyday

The accelerometer in your smartphone is a MEMS device. A tiny proof mass hangs suspended on microscopic springs etched from silicon. When the phone accelerates, the proof mass lags behind, deflecting the springs. Capacitive sensors measure this deflection with exquisite precision, determining not just that you've moved but how fast you're accelerating in each of three dimensions.

Gyroscopes in navigation systems use vibrating MEMS structures. When the device rotates, the Coriolis effect—the same force that makes hurricanes spin—deflects the vibrating element in a measurable way. These devices can detect rotation rates accurate enough to navigate aircraft or guide missiles.

The digital micromirror device in projectors contains millions of tiny mirrors, each one capable of tilting between two positions thousands of times per second. These mirrors, smaller than the diameter of a human hair, switch light on or off for individual pixels. A single chip can create cinema-quality images.

Pressure sensors in car engines use silicon diaphragms thinner than paper. As pressure changes, the diaphragm flexes, stretching resistors embedded in its surface. The change in resistance translates directly to a pressure reading. These sensors help optimize fuel injection, monitor tire pressure, and trigger airbag deployment.

The Frontier Beyond MEMS

As fabrication techniques improve, MEMS are shrinking further. At the nanometer scale—billionths of a meter rather than millionths—they merge into a new field: Nanoelectromechanical Systems, or NEMS. Here, the boundary between mechanical engineering and chemistry blurs. Surface effects dominate so completely that you must consider how individual molecules interact with your device.

This is the realm Feynman dreamed about, where machines operate at the scale of molecules themselves. We're not there yet—not fully—but every year brings new capabilities, smaller features, more precise control over matter at scales that were unimaginable when the first MEMS devices flickered to life in the 1960s.

The next time you flip your phone to landscape mode and the screen rotates to match, you're watching physics that behaves nothing like the world you see around you. In a device smaller than a dust mote, tiny proof masses swing on springs finer than spider silk, capacitors measure displacements smaller than the wavelength of light, and all of it happens inside a chip that costs less than a cup of coffee to manufacture.

That's the miracle of MEMS: the world at a hundred micrometers, working millions of times a day in your pocket, utterly invisible and absolutely essential.