 |
Aug. 22, 2003 – The MEMS industry has seen mixed success over the last decade. Today, there are only a few high-volume MEMS products available, notably air bag accelerometers, desktop inkjet print heads and pressure sensors for automotive and biomedical applications.
 |
 |
Despite a major infusion of funds in other applications, such as optotelecommunications, RF (radio frequency, or wireless), and biomedical/microfluidics, these efforts have not yet produced working MEMS in high volumes.
The customer-foundry relationship is a significant factor. The “foundry” must be more than just a foundry; it must be a true manufacturing partner.
There is clearly a continuum of customer needs when they engage with a foundry. Some customers are more expert in MEMS than others. In some cases, the foundry acts as a true “foundry,” receiving wafers from the customer, performing specifically predefined process steps and returning the wafers to the customer. Other customers bring a design to the foundry with the expectation that the foundry will design the process flow for them. Regardless of design sophistication, this scenario requires some level of feedback on the design itself to ensure manufacturability. In many cases, this feedback is substantial and critical to the customer’s — and foundry’s — success.
Still other customers provide only specifications, such as speed, size, power consumption or, in the case of microfluidics, metered volume delivery. Here, the “foundry” must provide complete design services, working closely with the customers to ensure that their designs meet specifications. In all cases, the foundry must also provide prototyping, process development and manufacturing services.
Simply put, you can’t make a car in a toaster factory. While both are made primarily of metal, the car requires much more sophisticated manufacturing capability.
Varied materials, processes and tolerances are required to make very complex MEMS work. Pure CMOS fabs are extremely limited in the materials that can be used due to contamination issues. For complex MEMS, flexibility is needed to incorporate non-CMOS materials.
Novel processes are also required for many complex MEMS. The key to success is adaptability to the customer’s requirements and specifications. And the foundry must be able to invent outside fixed design rules, for example, to produce smooth mirrors on vertical etched surfaces or to actuate devices during wafer processing to avoid subsequent damage to the device.
Finally, successful manufacturing of complex MEMS requires superior tools and metrology; for example, submicron photolithography and metrology. Most MEMS work is done on contact lithography tools, but many devices require or can benefit from a reduction stepper’s finer lateral dimensions and superior dimensional control.
Even where the features do not require submicron lithography, the tighter dimensional control of the stepper can provide devices that work better and more reliably. Other required tools provide precision deposition and metrology to ensure tight tolerances for thickness, composition and morphology.
The MEMS adage, “You can make one of anything” rings true. Products designed with manufacturability in mind have a vastly superior opportunity for success. The design must be conceived with the requirement that it meet well-thought-out tolerances that are achievable in quantity production.
Extensive 1-D modeling is essential to designing for manufacturability. If the foundry does not offer these services, the design may often not meet the specific requirements and capabilities of the fab.
The importance of manufacturing controls to high-volume production cannot be over-emphasized. These include metrology and test for process optimization and implementation of statistical quality and process control.
Once the product is being prototyped and/or manufactured in a given foundry, moving that product to a different facility is painful and costly. Reasons for breaking up this marriage can range from the fact that the selected foundry partner has limited manufacturing capacity or expertise, or that a redesign is required after several iterations have not reached the required yield.
This problem can only be solved by selecting a full-service foundry/manufacturing partner. The foundry must offer full services, with the required capabilities, facilities, tools, experience and high-volume manufacturing capacity in one place, specifically:
• Flexibility for difficult customer requirements;
• High-volume manufacturing experience;
• Submicron photolithography for tight dimensional and overlay control;
• Large fab capacity with an automated and complete tool set;
• Non-CMOS materials flexibility;
• Extensive metrology and test capability, including the tools and the experience to optimize and control the production process.