Wanted: New Model for MEMS Manufacturing

MEMS is booming. MEMS sales grow over 10 percent a year in dollars[1], and almost double that in units—so what’s the problem?

The MEMS manufacturing model of today is fragmented and—when compared to the IC foundry model – still not ready for prime time. The industry’s shipments are still dominated by a small handful of players, and yet the experiences of the IC industry show us that when an industry expands, it does so because all of the elements of the ecosystem are in place to encourage new players and products to emerge.

For the MEMS industry to realize the growth potential of the market, it needs a fundamental shift in how hardware is designed, developed, built and ramped. It also needs a model where an order of magnitude more players can see a path to entering this market, and where the investors and backers of these players can see a path where the technology and markets are attractive, and not blindly risky.

Figure 1: The Shanghai Industrial and Technology Research Institute (SITRI) aims to help the MEMS market make the most of growth opportunities says the author.

Figure 1: The Shanghai Industrial and Technology Research Institute (SITRI) aims to help the MEMS market make the most of growth opportunities says the author.

There is a new concept being put in place in Shanghai that aims to build this new model—not just for China, but also for the industry as a whole. By combining investment, incubation, design, integration, supply chain, customers and market entry under a single organizational umbrella, this new approach strives to create a “global innovation network” for the industry and become the entity that makes MEMS work, through focus on not only the individual steps of company development, but more importantly the boundaries between these steps.

The Ramp Fab Issue

The model that we have for MEMS manufacturing today offers a history lesson in innovation and adaptation. It is said that innovation is taking existing inventions and applying them to new applications, and nowhere is this as clear as in the history of MEMS.

Despite the unique challenges that MEMS brings to manufacturing, fabs have built tremendous knowledge about creating 3D structures out of silicon wafers using IC-derived tools. But look across the foundry landscape and you see no common foundation: each foundry promotes its “proprietary” process flows and know-how based on its in-house tools and the projects it’s done in the past. Though most of these processes contain virtually identical process steps, we have no common standard or methodology for describing these process steps to facilitate fab-to-fab transfer or calibration.

Rather than being the catalyst for invention, the challenges of getting MEMS to market have killed VC interest in this area and stifled the ability of entrepreneurs to see if their ideas can become reality. For those who do line up funding, early development will be with a ramp fab that will work for NRE. This will likely be on 6-inch wafers or even 4-inch wafers for cost-saving purposes. And even though the foundry is set up as a commercial foundry, its old equipment set and limited clean room space, coupled with single installs of several tools, mean that the processes, even when completed, will be highly specific to that fab and subject to multiple points of failure in production support

Ideas targeting the largest volume applications can’t afford to take the path just described. Yet the chances of these new ideas attracting the attention and resources of a major foundry are increasingly small. The result is that only “ramp fab-sized” opportunities are pursued, which turns the maxim “90 percent of MEMS opportunities don’t need 8-inch wafers” into a self-fulfilling prophecy.

Hardware has Lost its Sexy

Investors want their entrepreneurs to be great successes, but success can be a curse as well. As soon as you outgrow the ramp fab’s limitations, you immediately need to get the attention, interest and resources of a high volume fab (or two) to support your growth needs. By and large, the ramp fab has no economic interest in supporting this effort, so the typical approach is that the company keeps its first generation product with the ramp fab and in parallel engages with the high-volume fab for Gen2 Keeping its relationship with the high-volume fab a secret from the ramp fab is done to ensure the company still gets an unfair share of attention from its foundry ‘partner.’ This drives duplication, inefficiency and costs that harm the company and industry.

While tech-based venture capital investment has exploded in the past five years, interest in semiconductor-based ideas has dwindled. Overall tech investments have amassed to $53.9 billion over the past five years while semiconductor investments have shrunk to their lowest level in decades, a mere $1.1 billion over the last five years, with almost all in later stage rounds, not in new ventures[2].

Not that all hardware is out of favor: Maker Faires and hardware accelerators populate the valley, and already have strong ties to Asia through the Electronics Manufacturing Services (EMS) companies that sponsor or back these efforts. Providing ways for startups to get their idea prototyped and their app built so that they can start their Kickstarter campaign to prove the market pull for their idea is a great way of taking your company from stage to stage. And the transition from prototype to full-scale production is generally well understood, straightforward and doesn’t require complete re-engineering of the product. It is what gives us the Nests and Fitbits and Cardboards of the world today.

But this hardware is based on today’s components, which are based on yesterday’s ideas. And while we might not know what the new sensors are going to be, history consistently shows that integration and innovation go hand in hand in this area, and that sustained, long term innovation requires sustained efforts at the integration scale. So where is this progress in integration-level innovation going to come from?

Free Range Momentum

Great innovation comes when an industry sees a Pre-Cambrian explosion of ideas being tried out, testing themselves against the market and against each other to see which ideas will survive. Just like what we see today in the Hardware Accelerator case, where out of a thousand different ideas for wearable bands and bracelets, a small handful of really useful ideas will emerge—so too we need this revolution to happen in new sensors, components and materials. You need the open space of unleashed creativity to see in which direction and how far these different ideas will run, and which ones look like winners in the long run.

If you look at the center of gravity for emerging market potential, it is quite clearly in the Far East. From now until 2040, the global population of middle class households will nearly triple from 1.8 billion to almost 5 billion, with essentially all of that growth coming from Asia[3]. Global middle class consumption will increase by thirty trillion dollars by 2030 alone.

And while China has been almost entirely dependent on imported MEMS sensor components and systems up to now, this will shift as China establishes and builds its “More than Moore” capabilities and ecosystem, and takes this opportunity to build the companies that will feed its own growth. New opportunities will emerge for new sensors and products: the growth in Asia is a growth market for the entire industry. Even with China’s moves to establish and build its domestic industry, the demand will outstrip this capacity for many years, and there will always be a preference among certain buyers for Western technology, regardless.

Integrating Development and Commercialization

What is needed is an approach that spans the entire industry: financing, processing, prototyping, production and market entry. Ideally, such an approach would work across the industry and not just within certain regions. Ultimately, this approach would become fab-agnostic and yet recognize and protect the fab’s unique value.  It’s an approach whose objective is to make hardware sexy again.

The Shanghai Industrial and Technology Research Institute (SITRI) is building this model, in Shanghai and around the world. Working with major supply chain partners and research institutions, SITRI provides an integrated approach toward development and commercialization of new “More than Moore” solutions for the new economies. It does this by partnering across the industry, by building world class R&D and integration teams in the major electronics hubs, by working with strategic investor funds to seed and sustain these early ideas until they are proven enough to attract the VCs to them, and through the strong support from China and local industries, to establish and launch new companies into the exploding China market.

SITRI is built on five pillars to enable rapid innovation in the “More than Moore” space:

1.     State of the art 8-/12-inch “More than Moore” R&D fab: a 5000 square meter fab near SITRI’s Shanghai headquarters to support the development and integration of new processes, materials and structures. Using this in-house facility and SITRI’s foundry partner network, we can support full integration needs directly to commercial volume requirements,
2.     Early stage funding and support: working with SITRI’s own early stage funds as well as other development, investment and industry sources, we provide the critical early stage support until the new venture can attract institutional investment interest.
3.     Engineering and analytical services: design, integration, analysis, test and characterization services as needed to support complete solutions development.
4.     Supply chain engagements: leveraging SITRI’s focal role in the electronics industrial base, we assist in introductions and negotiations for both manufacturing and logistics support.
5.     Market engagement and entry: SITRI’s direct relationships with the largest Chinese brands, as well as the industry associations with strong China and international participation, bring broad engagement and deep understanding of market trends and requirements to SITRI partners, as well as critical customer engagements including China state-owned enterprises where government policies require a strategic partner like SITRI to reach.

Sanity in the Face of Chaos

SITRI works with companies from inception to market launch and offers supply chain support, process development and integration support, foundry and back-end support, and access to market intelligence services. SITRI works with foundries around the world in matching each opportunity with the best of the available houses. It works with the major high-volume fabs for direct process transfer after in-house integration. It works with leading for Outsourced Semiconductor Assembly and Test (OSATs) and IC houses to create advanced and integrated packaged solutions. It works with leading IP houses to scan, find, secure and integrate needed IP, doing so in a growing “global innovation network” with centers being established in Shanghai, Silicon Valley, Taiwan and, in the future, Boston and Europe.

And SITRI is not just for the entrepreneur: the challenge for finding a viable, cost effective and efficient path toward process development, integration and commercialization, coupled with growing imperatives to establish China manufacturing for the China market, make SITRI an ideal strategic partner for IDMs, OEMs and established fabless companies worldwide.

PeterHimesheadshotPeter Himes is GM of SITRI’s Silicon Valley office and VP for International Marketing for SITRI.


Contact Information


235 Chengbei Rd

toll-free: 86 21-69923266

Share and Enjoy:
  • Digg
  • Sphinn
  • del.icio.us
  • Facebook
  • Mixx
  • Google
  • TwitThis


Extension Media websites place cookies on your device to give you the best user experience. By using our websites, you agree to placement of these cookies and to our Privacy Policy. Please click here to accept.