TL;DR
Today we explore the fascinating world of microelectromechanical systems (MEMS), which produce microscopic machines using the same manufacturing techniques as for semiconductors. MEMS serve a wide array of different use cases and are broadly used in modern phones and in a host of industries. We also comment briefly on the promising changes underway at Intel.
Surprising News from Intel
Pat Gelsinger, Intel’s new CEO, is taking some surprising steps to shake up Intel. He’s introduced what he’s calling the Integrated Device Manager 2.0 (IDM 2.0) model, in which Intel will both open out its foundry to other companies and scale up its usage of external foundries for its internal chip development projects. Intel has previously tried and failed to start an external foundry business, but the newly announced effort seems more serious, with the new Intel Foundry Services set up as a separate business line with its own profit and loss line. Intel is also committing to investing $20 billion into two new foundries in Arizona that will come online in 2024.
In the broader global setting, Intel’s news is very welcome. It sets up a plausible alternative to TSMC that will be based in the US. I hope that major players in the US (such as Nvidia and Google) will use the opportunity to de-risk their manufacturing by keeping some of their core manufacturing efforts within the US rather than in Taiwan or South Korea. Intel’s steps also make it more likely that the company will be able to compete effectively with TSMC. If Intel succeeds in capturing some of TSMC’s market share, it will gain additional revenue to fuel the R&D efforts needed to build a world class foundry. The success or failure of Intel’s ambitious new efforts will ultimately come down to Mr. Gelsinger’s leadership, but the steps taken seem highly promising for Intel’s future.
Stratechery and Anandtech have excellent write-ups of Intel’s new directions worth checking out as well.
Microelectromechanical Systems (MEMS)
The nanoscale manufacturing techniques that have primarily been used to construct new transistors and circuits can also be used to construct microscopic machines called MEMS. MEMS bring several advantages to the table compared with more traditional manufacturing techniques. For one, they allow for the miniaturization of complicated systems including microphones, accelerometers, gyroscopes, oscillators, microfluidics, cantilevers, magnetometers, bolometers, barometers, pressure sensors, ultrasound transducers, and much more. MEMS have a rapidly growing market, with spending growing steadily over the last few years across multiple different industries (as shown below).
MEMS systems have to satisfy different physical constraints than macroscale devices since their form factors are so radically different. In some cases, MEMS can be more robust than their macroscopic alternatives, but MEMS can also face unique design challenges arising from microscale physics. MEMS are visually compelling and strange devices, so it’s worth looking at a few pictures to understand the range of MEMS that can be fabricated today. Each picture below shows an example of a MEMS system used for a different application.
Discussion
MEMS miniaturize familiar machines down to the microscopic scale, allowing for powerful sensory modalities to be integrated directly onto chips. Accelerometers and gyrometers in phones already use MEMS, but future smartphones could perhaps have integrated gene sequencing, mass spectrometry, ultrasound, or radar! Progress in smartphones may have slowed in recent years, but the most radical sensory innovations (powered by MEMS) are yet to arrive. MEMS are manufactured at foundries, similar to chips, but typically on older process nodes than the latest generation logic devices. As a result, multiple companies still run MEMS Foundries. Modern MEMS devices have sizes in the micrometer scale. One frontier of MEMS research is to miniaturize devices even further to the nanoscale, but the physics become more challenging, necessitating more progress in simulation tools (source). I anticipate that genetic algorithms and reinforcement learning based tools could potentially design classes of MEMS that we might have trouble designing due to the unintuitive physics of the micro and nanoscales. If you work on reinforcement learning, consider exploring the MEMS design space!
Next week’s subscriber-only issue will investigate some companies in the MEMS space, while next week’s free issue will dive into photonic computing.
Highlights for the Week
https://www.wsj.com/articles/h-m-battered-with-criticism-in-china-over-xinjiang-forced-labor-stance-11616598679: Chinese netizens are attacking retailer H&M for its stance against using products from Xinjiang. Similar efforts are also gearing up on the Chinese internet to attack Nike which has made a similar statement about Xinjiang labor. The CCP looks to be forcing companies to commit to using Xinjiang forced labor products, becoming complicit in the on-going genocide, or be punished by boycotts. It’s a dark and unsettling trend that speaks to the power of propaganda bubbles.
https://lexi-lambda.github.io/blog/2021/03/25/an-introduction-to-typeclass-metaprogramming/: Alexis King has a new blog post out on advanced programming techniques in Haskell.
https://www.cnn.com/2021/03/25/politics/georgia-state-house-voting-bill-passage/index.html: Bad news from Georgia, which has passed a new voter suppression bill.
Subscription, Feedback and Comments
If you liked this post, please consider subscribing! We have weekly subscriber-only posts on Tuesdays.
Please feel email me directly (bharath@deepforestsci.com) with your feedback and comments! In particular, if you’re currently working in the semiconductor or battery industry, please get in touch! I’d love your input for future iterations in our semiconductor series.
About
Deep Into the Forest is a newsletter by Deep Forest Sciences, Inc. We’re a deep tech R&D company specializing in the use of AI for deep tech development. We do technical consulting and joint development partnerships with deep tech firms. Get in touch with us at partnerships@deepforestsci.com! We’re always welcome to new ideas!
Credits
Author: Bharath Ramsundar, Ph.D.
Editor: Sandya Subramanian
The chart on the growth of industry MEMS is interesting to an academic who sits in / started in a MEMS group, where it feels like net academic investment is way down. Much of the research may have moved internal (many alumni from my group end up at Apple).
Happy to chat on that if it'll help your post, DM me https://twitter.com/natolambert.
This is a fantastic post. I participated in the last DARPA ERI Summit and attended a seminar on small MEMS fabrication that doesn't require a large fab facility. There was a startup looking for funding to do it - but it seems they solved a lot of intricate material science problems to make it actually happen. Imagine creating MEMS chips in your garage! I'll dig up the reference...