Firms Forge Alliance to Bring Piezoelectric MEMS to the Middle Market

Today, A. M. Fitzgerald & Associates (AMFitzgerald) and Sumitomo Precision Products (SPP) announced an alliance to combine AMFitzgerald’s microelectromechanical systems (MEMS) product development expertise with SPP’s full-service PZT (epitaxial lead zirconate titanate) MEMS wafer foundry, MEMS Infinity. The alliance will expedite the commercialization of thin-film PZT MEMS chip technologies in small to medium size markets. 

PZT is an up and coming high-performance piezoelectric MEMS material, but developers face challenges due to the limited availability of production quality prototype substrates. One of the challenges facing developers of PZT MEMs products is that the raw materials are difficult to come by in adequate quality.

Small quantities of thin-film piezoelectric PZT materials, the base for many MEMs devices, are only available in lesser quality R&D wafer batches, which differ in performance than production materials. That limitation increases the cost and difficulty of designing new PZT MEMs devices. The new alliance aims to make high quality PZT piezoelectric wafers available in small quantities with a path to higher volumes without the need for materials-dictated redesigns.

AMFitzgerald is a 20 year MEMS design firm that specializes in aerospace, medical and industrial devices. SPP is a full-service MEMS fab with a complete prototype design facility and PZT wafer fab operation. PZT has solid dielectric, ferroelectric piezoelectric and pyroelectric properties which make the material well suited for electromechanical and thermoelectric devices.

 

Why MEMS Development Matters

MEMS don’t get as much attention as systems like batteries and wireless protocols, but they are everywhere. The MEMS installation base is growing from two directions. First, MEMS are replacing legacy technology sensors with smaller, lower cost, higher performing MEMS equivalents. Second, the small size and low cost of MEMS mean that it is now economically feasible to add sensors in a whole host of areas and applications that could not economically take larger traditional sensors.

A Yole Group study put the MEMS sensor market in 2021, at $13.6 billion, growing to $22.3 billion by 2027 due to this “sensorization” trend.

 

Growth in the MEMS market. Image courtesy of Yole Group
 

Passing Silicon in Performance

MEMS (micro-electromechanical systems) combine both electrical and mechanical components in the same substrate. Components making up a MEMS device are typically between 0.001 mm and 0.1 mm in size with the entire part being as large as 0.02 mm to 1.0 mm. They operate based on mechanical stress that results in changing electrical properties. 

Most current MEMS devices rely on etched silicon (Si) architecture from the 1990s. Over the last 15 to 20 years, piezoelectric base material has been developed which allows a far greater variety of MEMS devices than does Si.

Si MEMS devices rely on capacitive or resistive changes brought about by mechanical deformation of small silicon struts. Piezoelectric devices rely on the piezoelectric effect, which is the creation of a small electric charge based on mechanical deformation, or the reverse: mechanical deformation based on a small added electric charge.

 

Piezoelectric actuator and sensor in action. Image courtesy AMFitzgerald

 

Piezoelectric can deliver more force as a micro actuator than can silicon-based deep Si etched MEMS actuators. Piezo sensors are also more sensitive and more accurate than Si MEMS sensors.

 

Silicon vs. Piezoelectric 

A silicon MEMS accelerometer has a structure etched in the semiconductor substrate that is affixed to the base and free on the end—somewhat like a stick protruding from a wall on one end and hanging free on the other end. This part is insulated from the substrate and is referred to as the moveable electrode. The substrate is called the fixed electrode. These features are etched using techniques similar to the way integrated circuits are etched.

When the device is moved, the moveable electrode bends slightly at the free end due to inertia. When it bends, the distance between it and the fixed electrode changes. With an electrical charge applied, the bending causes a change in the capacitance between the fixed and moveable electrodes. Nearby electronics, often etched into the same Si substrate, convert that changing capacitance into a software readable signal that indicates the strength of the acceleration.

A piezoelectric accelerometer has a similar movable structure etched in a piezoelectric substrate. When moved, the piezoelectric material generates a small electric current under the bending action. Again, nearby electronics turn the current into a usable signal.

 

AlN vs. PZT

Piezoelectric MEMS come in two primary flavors: AlN (aluminum nitride) and PZT. AlN was matured into a manufacturable technology faster than thin-film PZT. However, according to Dr. Alissa M. Fitzgerald, founder and managing member of AMFitzgerald, AlN is more limited in applications.

Dr Fitzgerald said that AlN is well suited to devices that require smaller motion than PZT, or sensor applications such as RF filters, oscillators, microphones and touch sensors, but PZT has much wider application potential and is best for wafer-level or chip-level electronics integration. PZT piezoelectric applications include Inkjets, autofocus, ultrasound, microphones/speakers, micromirrors, pumps, fluidics, gyroscopes, energy harvesters and more.

PZT is not new. Medical ultrasound wands have used a thick-film PZT transducer for many years now. However, thick film devices are very expensive compared to thin-film. It is only recently that thin-film PZT has become commercially viable. Making the small quantity prototype wafers at full manufacturing quality levels, which is the goal of this alliance, is one of the remaining steps to wide-spread adoption of thin-film PZT for MEMS construction.

 

Shortening the Design Process

The PZT MEMS development process today, essentially has two prototyping phases. First, the initial design is made with R&D quality wafers. Second, after foundry selection, design adjustments and optimization are required to adapt to the production fab volume quality materials.

 

The alliance will enable a reduction in design steps. Image courtesy AMFitzgerald

 

The “foundry selection” stage is problematic as well. Fitzgerald explained that the big Si fab houses are selective with their customer base. They only want customers that will sell billions of devices per year, but a lot of innovation is coming from markets that will support millions of devices, which translates to 500 to 1,000 wafers per year. That makes innovation in the middle markets very difficult now.

By making production quality wafers available for low-volume prototype builds the post-fab redesign phase is eliminated. The initial design can be created under final manufacturing conditions. With this alliance, developers in specialized non-mass production markets will have much better access to the raw materials they need for successful product design and manufacturing programs.