Multi-Project Wafer (MPW) services were first introduced in the 1980s (by MOSIS) to share space on silicon wafers and counteract very high Integrated Circuit (IC) development costs. For ICs, this was relatively straightforward, given the highly standardised fabrication processes and excellent function libraries, which allow developers to efficiently transfer sub-units between designs. Following on from MPW prototyping, scale-up to higher volumes was relatively easy.

Now MEMS foundries also offer MPW services, even though microfabrication covers a much wider range of processes and materials, including bulk or surface micromachining, silicon-on-insulator (SOI), wafer bonding integrated with electronics/photonics/piezoelectrics/bio-elements.

MPWs can be a low-cost and accessible way of making your first MEMS chip, but what do you need to know before choosing this route? In this blog, we discuss the pros and cons, alongside examples of three TTP-internal MPW projects, ranging from more conventional bulk microfabrication to lesser-known polymer and plastic-related processes.

Key points in this blog:

  • Examples of MPW MEMS projects
  • Design rules and their importance
  • Scale-up considerations
  • Scheduling and costs involved

Multi-Project Wafer project examples

Bulk microfabrication using SOI wafers

In this example, we used SOI wafers to manufacture four different “micro” devices: micro-mirrors, micro-antennas, micro-nozzles and micro-valves. The project used the same mask-set, process and materials for all the devices, thereby saving approximately 25 percent on the overall project cost and timescale. A key advantage was the inclusion of multiple designs on the same wafer, with 10mm x 10mm sized chips made using low-risk processes available at most MEMS foundries.

Shared SOI wafer micro-devices: mirrors, valves, antennas and nozzles.

Silicon-polymer process

This project involved the making of polymer layers containing thermal elements for two applications, heated pixel and thermal sensor arrays. Devices consisted of arrays of resistors at 0.5mm pitch, horizontal and vertical metal tracks separated by SU8 polymer layers and via holes to connect each layer electrically. There were nine designs, each with a die size of 22mm x 22mm. The range of devices and layout allowed us to address the needs of multiple stakeholders on a single chip, dramatically reducing development cost.

Shared polymer-based thermal micro-devices.

Hot embossing and sealing using low-cost polymers

Shared hot-embossed plastic micro-devices.

Design rules

An inflexible and complex set of design rules can create higher barriers to MPW entry. Process uniformity is another potential issue to deal with. For example, in the D-RIE process, variations in the mask pattern density can produce etch non-uniformities and such knowledge, as well as the adding of dummy structures to improve the pattern density, can really help. Top MEMS foundries and their designers are aware of all these issues and have established, documented (SOPs) and qualified (repeatable and reliable) processes to minimise them. Alternatively, the specification window is widened to account for such process variations and loss of yield.

Scale-up

Wafer testing can represent 25–50 percent of the overall manufacturing cost due to the multi-physics nature of MEMS chips and so may be dropped by some MPW providers. So, some foundries deliver unpackaged and untested prototypes, but at additional cost some test along with dicing, encapsulation and wire bonding can be included. Packaging requirements are generally more bespoke and depend on the specific chip and application, so they are either dealt with at wafer scale (anodic bonding) or individually, say through 3D printing for example.

Scheduling and costs

MPW costs are proportional to the die number, total area used, complexity and masking steps, typically in £/sq-cm or £/chip. As an example, a European university-linked foundry can charge 1,000–1,500 Euros for approximately 100 chips to students, and 4,000–6,000 Euros for industrial clients.

Summary

However, lack of MEMS standardisation, wide range of processes and materials, lack of testing due to the multi-physics nature, and design-related process variations can all affect the yields. This in-turn can create differences between the different MPW providers. Europractice is a good place to find foundries with core strengths that they then offer as MPWs, thereby reducing any overlap or competition, whilst maintaining certain quality standards. So, aligning your process steps and materials with one of these foundries, through MPWs could enable your MEMS idea to soon become a scaled-up product.

Healthcare | Industrial | Healthcare Blogs | Industrial Blogs | Healthcare Case Studies | Industrial Case Studies

TTP is an independent technology company where scientists and engineers collaborate to invent, design and develop new products and technologies.