AZoSensors speaks with John Labram, from the Labram Research Group, Oregon State University on the team's latest development. The team has developed a bio-inspired perovskite sensor that operates like the mammalian eye.
Can you give our readers a summary of your recent research?
Our work was an experimental proof-of-principle demonstration of a new type of optical detector. Most conventional detectors operate by outputting a signal (e.g., a voltage), which is roughly proportional to the intensity of the light that falls upon it. E.g., twice as much incident light gives twice the voltage.
Our device operates on a different principle. It will only output a high voltage if the intensity of light is changing, regardless of intensity level. For example, if a light is switched on and left on over the detector, it will output a brief voltage spike followed by a quick decay back to zero volts. If the sensor is under a bright, but constant, light it will output zero volts regardless of the light intensity.
What led the team to take a bio-inspired focus when developing this sensor?
I am a firm believer in the “know a little about a lot” philosophy to science, and that it only takes a tiny aspect of another field to get inspired. Sometimes when I am doing something that does not require my full attention (e.g., formatting figures) in the background I will watch a lecture series from a field outside my own.
Having immediate and free access to lectures from some of the world's best educators is something that my generation of scientists are incredibly lucky to have. I watched a series by Nancy Kanwisher at MIT called, “The Human Brian.” It was a fantastic series, which I would recommend to anyone interested in neuroscience or neuromorphic computation, and it was from there that I got a rough idea of how the brain handles visual stimuli.
What were the experimental techniques involved in the recent research?
To fabricate the devices, we used well-known semiconductor growth protocols; primarily taken from the solar cell community. Our measurements were also very simple. We made single devices and tested them one at a time. We applied a constant voltage to each sensor and tracked the output voltage as a function of time using an oscilloscope. We then placed a light source (green LED) above the sensor and turned it on. We observed a short spike in the output voltage, followed by a decay back to zero volts. We carried out the same procedure on multiple devices, from this we were able to parameterize them and quantify device-to-device variability.
In what ways could this perovskite sensor change the way that computer processors operate?
You could achieve the desired effect in software or via complex hardware. The novelty of our work is that the sensor operates this way as part of its fundamental design. You could summarize it as a single pixel doing something that would currently require a microprocessor.
The difference will come down to speed and power consumption, ultimately determining how “life-like” artificial intelligence systems can be. Traditional processors/sensors were not designed to solve AI problems, but with modern advancements in speed they can now do certain AI tasks pretty well (voice recognition, etc.).
However, researchers are extremely ambitious about what could be achieved with AI in the future. The “Holy Grail” would be to enable real-time training of AI (e.g., having a robot learn about the physics of moving objects while watching their motion). This would require different hardware, and for this reason researchers are designing processors that operate the way the brain does (e.g., see Intel Loihi chips). Our sensor stems from the same motivation and, as far as we are aware, is the first sensor that works fundamentally like photoreceptors in the eye.
Image Credit: ra2 studio/Shutterstock.com
In what other applications can this type of perovskite be used?
The class of perovskites used in this application is called metal halide perovskites, often shortened to “perovskites” for brevity. They are materials which been widely studied for applications in solar energy generation for about 10 years, and that was our original motivation to study them. When I designed the sensor, I knew we needed a material that changes its electrical conductivity under illumination. Perovskites were therefore, an obvious choice.
In which industries is this sensor likely to be applied in the future?
The ability to easily differentiate moving objects from the background in a visual field is valuable in a range of applications. The most apparent immediate applications include things where real-time processing of visual information is crucial — for example, autonomous vehicles and robotics, or projectile tracking.
What are some of the other research areas of the Labram Research Group?
Our research sits at the intersection of physics, chemistry, materials science, and electrical engineering. Broadly, we look at the charge transport and device physics of disordered, generally solution-processed, semiconductors. The applications are mechanically flexible electronics and low-cost solar energy.
What’s next for the Labram Research Group?
With regards to this work, there are a few things we would like to do next. Below are a few examples:
- A high priority would be to better understand how signals such as this would be handled by a neural network. The knowledge gap between our team working on the sensor and those who write code is likely to be large and is going to require some collaboration between us and with computer scientists.
- Demonstrate these devices using organic semiconductors rather than perovskites. I have a strong suspicion that the material we used in the first demonstration (perovskite) is giving us some instabilities in the response.
- Improve some of the figures of merit (e.g., peak voltage height/input optical power) through device engineering. The design we demonstrated in the paper was the simplest design we could conceive of that would demonstrate the effect. So there are plenty of easy changes we can make that should give us better performance.
- There are quite a few fundamental questions on operation which are very important to us. In particular: performance limits, stability, and device-to-device variability.
- We also need to develop a robust mathematical framework to predict behavior reliably.
Where can readers find more information?
Readers can access our technical paper here.
About John Labram
John Labram received his undergraduate degree in Physics from the University of Warwick in 2008. In 2011 John received his Ph.D. in Physics from Imperial College London. Between 2011 and 2013 John took a break from academia to work as a currency options trader in the City of London.
In 2013 John returned to science, joining Imperial College as a postdoctoral research associate. In 2014 John was awarded the Elings Prize Fellowship to join the California NanoSystems Institute at the University of California, Santa Barbara. In 2017 John joined the School of Electrical Engineering and Computer Science at Oregon State University as an Assistant Professor.
Disclaimer: The views expressed here are those of the interviewee and do not necessarily represent the views of AZoM.com Limited (T/A) AZoNetwork, the owner and operator of this website. This disclaimer forms part of the Terms and Conditions of use of this website.