PE2-5.2 Materials for MEMS

What this is and why it exists

Material choice here is not the same question as in chip design. The material has to work mechanically as well as electrically.

That reframes the obvious answer. Silicon is used in this field for a mechanical reason that most people find surprising, quite separate from its electronic properties.

The vocabulary

• Substrate — the wafer the device is built on.
• Active substrate — a substrate that takes part in the device's function rather than only supporting it.
• Single crystal — material whose atomic arrangement is unbroken throughout.
• Elastic — returning exactly to its original shape when a load is removed.
• Fatigue — weakening under repeated loading, leading eventually to failure.
• Hysteresis — not returning by the same path it went out by.
• Piezoelectric material — one that produces charge when stressed, and deforms when a voltage is applied.
• Biocompatible — safe to put in contact with living tissue.

The mental model

Start with the surprising fact. Single-crystal silicon is an excellent spring. It is stronger than steel by weight, and it is elastic right up to the point where it fractures. It also shows almost no fatigue and almost no hysteresis.

That combination is unusual and it is exactly what a mechanical structure wants. A metal spring flexed millions of times gradually weakens and eventually fails. It also does not return by quite the same path it went out by. A sensor made from one has a small error depending on which way it last moved. Silicon has neither problem, because it has no grain boundaries for a crack to grow along. It either behaves perfectly or it snaps, and there is very little in between.

So the material was chosen twice, for different reasons, and that is why the field could adopt chip fabrication wholesale.

The substrate is the starting point, and the choice constrains every process step afterwards. What temperature it can survive, what etches it, what can be grown on it: all of that is fixed before anything else is decided.

The active substrate is a distinction specific to this field. In ordinary chip making the wafer is a support. Here it may be part of the device itself, etched into a diaphragm or a cantilever that does the mechanical work. The substrate is then not a platform but a component.

Silicon compounds serve as insulators, as masks and as structural layers in their own right. Their residual stress is a design parameter here rather than an afterthought. A film that will be freed carries its stress into the finished structure.

Piezoelectric materials earn their place for a reason worth noticing. They convert in both directions. Stress produces charge, and an applied voltage produces strain. So the same film can sense and can actuate, and a device can be made to drive itself and read itself with one material.

Polymers matter at the end of the field furthest from electronics. They are cheap, they can be moulded, and some are biocompatible. Where a device is disposable, or must touch a patient, or needs to be flexible, a polymer beats silicon regardless of its inferior mechanical properties. That is why they dominate the microfluidic end of the field.

Packaging materials matter more here than in ordinary chips, and for the reason from the previous topic. An ordinary chip wants a sealed package. A pressure sensor must feel the pressure, and a chemical sensor must be reached by the chemical. The package has to protect what must be protected and expose what must be exposed, and that is a harder problem than sealing everything.

What you should now be able to explain or do

Say why silicon is chosen for mechanical reasons as well as electronic ones. Explain what fatigue and hysteresis are and why their absence matters in a sensor. Distinguish a supporting substrate from an active one. Say why a piezoelectric material can both sense and actuate. Explain when a polymer beats silicon, and why packaging is harder here.

Check yourself