I am obliged to admit that I’ve not spent as much time as perhaps I should, cogitating and ruminating over what I could do with an array of 250,000 ultrasonic transducers. Now, however, after chatting with Xavier Rottenberg, Fellow Sensors and Actuators at IMEC, I find I can think of little else.
Are you familiar with the Interuniversity Microelectronics Centre (IMEC)? (The folks there tend to write this in lowercase as imec, but this is the sort of usage “up with which I will not put,” as Winston Churchill may or may not have said.)
Founded in 1984, IMEC is an international research and development organization, active in the fields of nanoelectronics and digital technologies, with headquarters in Belgium. In addition to employing around 5,000 researchers and scientists from over 95 countries, plus having an ecosystem of more than 600 world-leading industry partners and a global academic network, IMEC boasts a unique infrastructure that includes a 2.5-billion-euro 300mm semiconductor pilot line.
Xavier told me that his team is focused on manipulating wave-based processes, both acoustical and optical, and that they can use things from one domain in the other, and vice versa. For example, they can do some things easily in acoustics that are much more difficult in optics, and they can do some things in the optical realm that would be much more difficult in the acoustic domain.
The topic of this particular conversation was the fact that IMEC has recently announced a piezoelectric Micromachined Ultrasound Transducer (pMUT) array that is compatible with flat-panel-display (FPD) process technologies (for “micromachined” read “microelectromechanical systems (MEMS)”).
Piezoelectricity was discovered by French physicists Jacques and Pierre Curie in 1880. The piezoelectric effect results from the linear electromechanical interaction between the mechanical and electrical states in certain crystalline materials. Deforming a piezoelectric crystal by about 0.1% of its static dimension causes an electrical potential to be generated, thereby allowing the crystal to be used as a sensor that can detect ultrasonic sound waves, for example. Contrariwise, applying an electric potential to a piezoelectric crystal causes it to deform by about 0.1% of its static dimension, thereby allowing the crystal to act as an actuator that can generate ultrasonic sound waves, for example.
Consider typical ultrasonic probes that are used in sonography or echography applications, for example. These typically involve around 128 transducers mounted in a line or as a 2D array. Each transducer has its own analog signal wire conveying data to a honking big workstation that performs the analog-to-digital conversion (ADC) followed by the digital signal processing (DSP) that generates images for the medical bods to “Ooh” and “Aah” over. The fact that there are 128 of these analog signals explains why the cables coming out of the probe are so thick and unwieldy. Furthermore, the continual flexing of this cable as the system is used results in the quality of the signals being degraded over time.
The term “beamforming” (a.k.a. “spatial filtering”) refers to a signal processing technique used in sensor arrays for directional signal transmission or reception. Unfortunately, there’s only so much beamforming you can do with 128 transducers. This is a case of “the more, the merrier,” which is why the announcement from IMEC is so exciting.
The guys and gals at IMEC have spent years developing different types of sensors and actuators on silicon wafers, but this limits the size of the transducer array to between 1mm² and 1cm², give or take. Now, the ability to realize these transducers on top of FPD substrates allows for the creation of transducer arrays that are 10cm x 10cm in size (actually, they could be 1m x 1m or larger, if required).
Furthermore, they can use the thin-film transistors (TFTs) that are part-and-parcel with FPDs to perform first-pass signal processing locally, and also to act as a switch network to funnel signals to a small CMOS device that can perform more sophisticated digitization, after which the digitized data can be transported over one or more high-speed serial communications channels to the host computer.
Xavier tells me that they can also delaminate the thin-film-plus-transducer from the FPD glass substrate and re-laminate it onto non-flat surfaces, which flings open the door to a whole new range of possibilities.
One thing I worried about is that TFTs are much larger than their CMOS counterparts. Xavier responded by noting that an ultrasonic transducer 200um x 200um in size is huge compared to a TFT. Point taken, “nuff said.”
Let’s take this further. Suppose each ultrasonic transducer is 200um x 200um (they can be constructed smaller if required). That means we can build 25 transducers per mm² or 2,500 transducers per cm². Good Golly Miss Molly! That’s 250,000 ultrasonic transducers in a 10cm x 10cm panel, which is the stuff beamforming aficionados drool over (it’s not a pretty sight).
So, what could we do with such a panel, especially if it were flexible? Well, first, there are medical applications. For example, you could slap it on someone’s chest over their heart. The system could quickly detect the location of the ribs and focus all its attention on the gaps between the ribs to image the heart. Have you seen how bad the images are that are generated by even a state-of-the-art ultrasound system based on traditional technology? As I previously noted, this is largely due to the fact that there’s only so much beamforming you can do with 128 transducers. Now imagine the quality of the images you can create using 100,000 transducers or more!
Another application for such an array is to deliver power. Using beamforming, the array can deliver large amounts of power to a precisely located spot. This ability can be used for a wide variety of non-invasive medical applications, like breaking up kidney stones and gall stones, manipulation of the vagus nerve to treat pain, ablating tissue and burning cancers, and… the list goes on.
Xavier also talked about the possibility of applying a rectangular ultrasonic array patch to a patient’s forehead and using it to mitigate epileptic secures. “Can you really do that?” I asked. In reply, he pointed me to the Focused Ultrasound Foundation website, which documents so many medical applications that can benefit from ultrasonic technology that it left my head spinning (I wonder if they can cure that?).
I was already reeling from what Xavier had told me, but it turned out we’d only scratched the surface of the myriad potential uses for this incarnation of ultrasound technology. For example, it would be possible to use one of his ultrasonic arrays to generate pulses and detect reflections to perform gesture recognition, like someone pressing virtual buttons on a virtual control panel, for example. The same array could also be used to focus the ultrasound and modulate it with a lower frequency to provide haptic feedback (e.g., when the user “presses” a virtual button, they actually feel some resistance).
Xavier said something else that really made me think. He pointed out that people are currently getting excited about all of the things we can do with millimeter wave (mmWave) radio (think 5G and 6G phones) and radar. However, mmWave in electromagnetics means ~100GHz (the official mmWave spectrum is defined as being between 30GHz and 300GHz) because radio waves travel at the speed of light, which is ~300,000,000 meters per second.
By comparison, the speed of sound is ~300 meters per second (if we round furiously), which means that if we are talking about mmWave in sound, we are only talking about 100 kHz. This means we can get the same resolution using sound at 1,000,000 times lower frequency, which equates to much less power and much simpler electronics.
All I can say is “Wow!” This has really opened my eyes to some of the possibilities we might expect to see in our ultrasonic future. What say you? Do you have any thoughts you’d care to share on any of this?