CEO interview: InVisage’s Lee discusses an image sensor revolution
InVisage Technologies Inc. (Menlo Park, Calif.) is an image sensor startup that is nearly ten years old and that has taken in more than $100 million in venture capital. The company has been developing image sensors based on a quantum-dot material called QuantumFilm that replaces conventional silicon photodiodes but that is integrated on supporting CMOS technology.
It now has two products out in the market and is starting to gain traction, according to Lee (see Quantum-dot image sensor launch threatens silicon and Quantum-dot sensor brings benefits to near-IR.)
EETE Q1: What is QuantumFilm made of and how does it differ from other quantum dot materials?
Jess Lee: “We have not said too much about the material structure of QuantumFilm. There are concerns over cadmium in quantum dots so the first thing to say is there is no cadmium. What we have said is that it is a metal-chalcogenide material, similar to a II-VI material surrounded by ligands in a matrix.”
Lee added that it is the combination of the quantum dots and the carrier material that allows charge transfer. “It’s because of the packing regime we are in.” The dots are of a diameter of between about 3 and 5 nanometers and it is these dimensions that affect the electron band structure and govern the sensitivity to light.
In addition, InVisage has developed an automated process to control the diameter of quantum dots produced as part of film production, Lee said.
Q2: Quantum dots are known for having multiple applications besides sensing, such as displays and energy storage. Is InVisage only interested in image sensor applications?
JL: “Image sensors are our focus for the short- and medium-term future. It is where our expertise is. But semiconductor development is getting harder and More-than-Moore opportunities are increasing, so there is the possibility it [QuantumFilm] can be applied elsewhere in the future.
Q3: InVisage has shown that QuantumFilm has certain advantages over silicon photodiodes in a number of parameters, such as sensistivity, viewing angle, but how does it compare in terms of speed?
JL: Lee answered by saying that for machine vision applications it may well be speed of response, rather than resolution, and other features such as global shutter that determine design wins. Lee pointed out that QuantumFilm is not limited in terms of its speed of response.
“We are limited by the state of silicon I/O and analog-to-digital converters. As that rises we can go at hundreds, even thousands of frames per second. There is no difference between us and comparable CMOS image sensors.
However, Lee said that QuantumFilm has an additional advantage: that it is able to capture scenes very quickly, in periods of the order of a millisecond, and apply a “global shutter” electronically. This is not done easily in CMOS image sensors, which conventionally use a scheme of continuously rolling exposure and line-by-line read out. Although this is adequate for still images it can produce visual artifacts in images of fast moving objects.
Next: Ready for primetime in smartphones?
Q4: CMOS image sensor manufacturing is mature and optimized for volume production and low unit costs. Surely InVisage cannot compete on price, as it is yet to build up volume sales and so must compete on performance? So can InVisage go after the smartphone mass market immediately?
JL: “The QuantumFilm does cost something but we have shown that this is viable in high volume. And we get the great optimization that TSMC can bring (see InVisage, TSMC, join forces for image sensor attack). So in the material we have something exclusive that we can make on a very high volume platform.”
Lee points out that QuantumFilm sensors are not the same as CMOS image sensors and that the smartphone market is itself dynamic and is looking for superior performance. “The space we are in is performance and value add, but the primary motivation is performance.”
And as with most engineering, the devil is in the detail. Most CMOS image sensors used in smartphones use thinned wafers and back-side illumination (BSI) “We don’t require a BSI process which is also a fundamental cost advantage,” said Lee.
So as to whether InVisage goes after the smartphone market from its market debut, Lee said: “Yes. We have and we are. But we are also going after the machine vision market.”
Q5: The CMOS image sensor market is reckoned to have been worth about $10 billion in 2015 and on a compound annual growth rate of about 11 percent (see Pause forecast for CMOS image sensor market). What penetration pattern over time do you envisage for QuantumFilm?
JL: “We are aggressive in how we attack the market and integrate in product form but we also realise we can’t bite off the whole market by ourselves. We’re still a venture-backed startup. We can’t build 20 fabs to produced QuantumFilm sensors. But we are looking to raise money with partners as investors,” said Lee.
Lee added: “We see QuantumFilm as a platform used by us as the first and second customer. To have third and fourth customers is better for us. We are prepared to work with partners to enable them, with QuantumFilm. We wouldn’t license the technology out but there are a number of other ways to enable partners.” Lee said it is too early to name any potential customers for QuantumFilm or potential partners and investors.
Next: Pay-back time?
Q6: Is there more engineering that can be done with QuantumFilm besides replacing silicon photodiodes in an array?
JL: “Absolutely there is. Near-infrared is something we are doing. Deploying multiple sensors with computational processing is possible although it is orthogonal to us. But we have a lot of latitude because we are on CMOS.” Lee added that, in addition, InVisage customers are telling him that the global shutter capability makes it easier to do multi-sensor imaging because it aids synchronization of images.
“The product roadmap has two arcs; pushing QuantumFilm and its performance and spectral range. The second arc is to build in the silicon. At the moment that is the readout circuit but there is no reason to stay there. We can also approach from the opposite direction. We can put QuantumFilm on other circuits. And we could also do optical sensors for ambient light, infrared for simple sensing,” said Lee.
Q7: InVisage has raised more than $100 million in venture capital. How can you pay those investors back?
JL: “It’s true the conventional semiconductor sector is a jungle with companies fighting at the leading edge. This is one of the reasons venture capital has turned away from that sector. But we are on a mature silicon platform – 110nm. It’s a very different capital investment process there. We have our own fab in Taiwan but it is only focused on a couple of processes – a spin-on process to add the film and the definition of the pixels.”
But Lee is confident that it is possible for startups to be successful and reward their investors. Lee cites Ambarella Inc. (Santa Clara, Calif.) a developer of video compression and image processing products, and InvenSense Inc. (San Jose, Calif.), a provider of MEMS-based inertial sensor components and sub-systems. These companies were founded in 2004 and 2003 respectively and are both now publicly owned.
“Ultimately it [paying investors back] is only achieved by making product and making it competitive, but to this we have added that we don’t just want to be a product company. We will provide our technology platform as well,” said Lee.
Related links and articles:
White paper on QuantumFilm applied to near-infrared
News articles:
Quantum-dot image sensor launch threatens silicon
Quantum-dot sensor brings benefits to near-IR
InVisage, TSMC, join forces for image sensor attack
China VC lifts sensor startup to record funding
16 analog, MEMS and sensor startups to follow in 2016
EE Times Silicon 60: 2015’s Startups to Watch
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
