In a conversation with Physics World’s Tami Freeman Krausz talks about his research into using ultrashort-pulsed laser technology to develop a diagnostic tool for early disease detection. He also discusses his collaboration with Semmelweis University to establish the John von Neumann Institute for Data Science, and describes the Science4People initiative, a charity that he and his colleagues founded to provide education for children who have been displaced by the war in Ukraine.
On 13–14 May, The Economist is hosting Commercialising Quantum Global 2025 in London. The event is supported by the Institute of Physics – which brings you Physics World. Participants will join global leaders from business, science and policy for two days of real-world insights into quantum’s future. In London you will explore breakthroughs in quantum computing, communications and sensing, and discover how these technologies are shaping industries, economies and global regulation. Register now and use code QUANTUM20 to receive 20% off. This offer ends on 4 May.
Until now, researchers have had to choose between thermal and visible imaging: One reveals heat signatures while the other provides structural detail. Recording both and trying to align them manually — or harder still, synchronizing them temporally — can be inconsistent and time-consuming. The result is data that is close but never quite complete. The new FLIR MIX is a game changer, capturing and synchronizing high-speed thermal and visible imagery at up to 1000 fps. Visible and high-performance infrared cameras with FLIR Research Studio software work together to deliver one data set with perfect spatial and temporal alignment — no missed details or second guessing, just a complete picture of fast-moving events.
Jerry Beeney
Jerry Beeney is a seasoned global business development leader with a proven track record of driving product growth and sales performance in the Teledyne FLIR Science and Automation verticals. With more than 20 years at Teledyne FLIR, he has played a pivotal role in launching new thermal imaging solutions, working closely with technical experts, product managers, and customers to align products with market demands and customer needs. Before assuming his current role, Beeney held a variety of technical and sales positions, including senior scientific segment engineer. In these roles, he managed strategic accounts and delivered training and product demonstrations for clients across diverse R&D and scientific research fields. Beeney’s dedication to achieving meaningful results and cultivating lasting client relationships remains a cornerstone of his professional approach.
A new retinal stimulation technique called Oz enabled volunteers to see colours that lie beyond the natural range of human vision. Developed by researchers at UC Berkeley, Oz works by stimulating individual cone cells in the retina with targeted microdoses of laser light, while compensating for the eye’s motion.
Colour vision is enabled by cone cells in the retina. Most humans have three types of cone cells, known as L, M and S (long, medium and short), which respond to different wavelengths of visible light. During natural human vision, the spectral distribution of light reaching these cone cells determines the colours that we see.
Spectral sensitivity curves The response function of M cone cells overlaps completely with those of L and S cones. (Courtesy: Ben Rudiak-Gould)
Some colours, however, simply cannot be seen. The spectral sensitivity curves of the three cone types overlap – in particular, there is no wavelength of light that stimulates only the M cone cells without stimulating nearby L (and sometimes also S) cones as well.
The Oz approach, however, is fundamentally different. Rather than being based on spectral distribution, colour perception is controlled by shaping the spatial distribution of light on the retina.
Describing the technique in Science Advances, Ren Ng and colleagues showed that targeting individual cone cells with a 543 nm laser enabled subjects to see a range of colours in both images and videos. Intriguingly, stimulating only the M cone cells sent a colour signal to the brain that never occurs in natural vision.
The Oz laser system uses a technique called adaptive optics scanning light ophthalmoscopy (AOSLO) to simultaneously image and stimulate the retina with a raster scan of laser light. The device images the retina with infrared light to track eye motion in real time and targets pulses of visible laser light at individual cone cells, at a rate of 105 per second.
In a proof-of-principle experiment, the researchers tested a prototype Oz system on five volunteers. In a preparatory step, they used adaptive optics-based optical coherence tomography (AO-OCT) to classify the LMS spectral type of 1000 to 2000 cone cells in a region of each subject’s retina.
When exclusively targeting M cone cells in these retinal regions, subjects reported seeing a new blue–green colour of unprecedented saturation – which the researchers named “olo”. They could also clearly perceive Oz hues in image and video form, reliably detecting the orientation of a red line and the motion direction of a rotating red dot on olo backgrounds. In colour matching experiments, subjects could only match olo with the closest monochromatic light by desaturating it with white light – demonstrating that olo lies beyond the range of natural vision.
The team also performed control experiments in which the Oz microdoses were intentionally “jittered” by a few microns. With the target locations no longer delivered accurately, the subjects instead perceived the natural colour of the stimulating laser. In the image and video recognition experiments, jittering the microdose target locations reduced the task accuracy to guessing rate.
Ng and colleagues conclude that “Oz represents a new class of experimental platform for vision science and neuroscience [that] will enable diverse new experiments”. They also suggest that the technique could one day help to elicit full colour vision in people with colour blindness.
Deterministic entanglement through holonomy: A system of four coupled optical waveguides (A, C, E, W), with three inter-waveguide coupling coefficients (k_A,k_E,k_W) vary in such a way to define a closed path γ. (Courtesy: Reprinted with permission from http://dx.doi.org/10.1103/PhysRevLett.134.080201)
Physicists at the Georgia Institute of Technology, US have introduced a novel way to generate entanglement between photons – an essential step in building scalable quantum computers that use photons as quantum bits (qubits). Their research, published in Physical Review Letters, leverages a mathematical concept called non-Abelian quantum holonomy to entangle photons in a deterministic way without relying on strong nonlinear interactions or irrevocably probabilistic quantum measurements.
Entanglement is fundamental to quantum information science, distinguishing quantum mechanics from classical theories and serving as a pivotal resource for quantum technologies. Existing methods for entangling photons often suffer from inefficiencies, however, requiring additional particles such as atoms or quantum dots and additional steps such as post-selection that eliminate all outcomes of a quantum measurement in which a desired event does not occur.
While post-selection is a common strategy for entangling non-interacting quantum particles, protocols for entangled state preparation that use post-selection are non-deterministic. This is because they rely upon making measurements, and the result of obtaining a certain state of the system after a measurement is associated with a probability, making it inevitably non-deterministic.
Non-Abelian holonomy
The new approach provides a direct and deterministic alternative. In it, the entangled photons occupy distinguishable spatial modes of optical waveguides, making entanglement more practical for real-world applications. To develop it, Georgia Tech’s Aniruddha Bhattacharya and Chandra Raman took inspiration from a 2023 experiment by physicists at Universität Rostock, Germany, that involved coupled photonic waveguides on a fused silica chip. Both works exploit a property known as non-Abelian holonomy, which is essentially a geometric effect that occurs when a quantum system evolves along a closed path in parameter space (more precisely, it is a matrix-valued generalization of a pure geometric phase).
In Bhattacharya and Raman’s approach, photons evolve in a waveguide system where their quantum states undergo a controlled transformation that leads to entanglement. The pair derive an analytical expression for the holonomic transformation matrix, showing that the entangling operation corresponds to a unitary rotation within an effective pseudo-angular momentum space. Because this process is fully unitary, it does not require measurement or external interventions, making it inherently robust.
Beyond the Hong-Ou-Mandel effect
A classic example of photon entanglement is the Hong–Ou–Mandel (HOM) effect, where two identical photons interfere at a beam splitter, leading to quantum correlations between them. The new method extends such interference effects beyond two photons, allowing deterministic entanglement of multiple photons and even higher-dimensional quantum states known as qudits (d-level systems) instead of qubits (two-level systems). This could significantly improve the efficiency of quantum information protocols.
Because state preparation and measurement are relatively straightforward in this approach, Bhattacharya and Raman say it is well-suited for quantum computing. Since the method relies on geometric principles, it naturally protects against certain types of noise, making it more robust than traditional approaches. They add that their technique could even be used to construct an almost universal set of near-deterministic entangling gates for quantum computation with light. “This innovative use of non-Abelian holonomy could shift the way we think about photonic quantum computing,” they say.
By providing a deterministic and scalable entanglement mechanism, Bhattacharya and Raman add that their method opens the door to more efficient and reliable photonic quantum technologies. The next steps will be to validate the approach experimentally and explore practical implementations in quantum communication and computation. Further in the future, it will be necessary to find ways of integrating this approach with other quantum systems, such as matter-based qubits, to enable large-scale quantum networks.
I recently met an old friend from the time when we both worked in the light-emitting diode (LED) industry. We started discussing how every technology has its day – a window of opportunity – but also how hard it is to know which companies will succeed long-term. As we got talking, I was reminded of a very painful product launch by an LED lighting firm that I was running at the time.
The incident occurred in 2014 when my company had a technology division that was developing wirelessly connected lighting. Our plan was to unveil our new Bluetooth control system at a trade show, but everything that could go wrong for us on the day pretty much did. However, the problems with our product weren’t down to any
particular flaws in our technology.
Instead, they were triggered by issues with a technology from a rival lighting firm that was also exhibiting at the event. We ended up in a rather ugly confrontation with our competitors, who seemed in denial about their difficulties. I realized there was a fundamental problem with their technology – even if they couldn’t see it – and predicted that, for them, the writing was on the wall.
Bother at the booth
Our plan was to put microprocessors and radios into lighting to make it smart, more energy efficient and with integrated sensors and Bluetooth controls. There were no fundamental physics barriers – it just needed some simple thermal management (the LEDs and particularly the radios had to be kept below 70 ºC), some electronics and lots of software.
Back then, the LED lighting industry was following a technology roadmap that envisaged these solid-state devices eventually generating 320 lumens per watt, compared to about 10 lumens per watt from conventional incandescent lamps. As the road map progressed, there’d be ever fewer thermal challenges.
With more and more countries phasing out conventional bulbs, LEDs are continuing their march along the roadmap. Almost every bulb sold these days is an LED and the overall global lighting market was worth $140bn in 2023, according to Fortune Business Insights. Lighting accounts for 15–18% of all electricity consumption in the European Union alone.
Back at that 2014 trade show, called LuxLive, all initially seemed to be going well as we set up our display. There was the odd software bug, but were able to work around that and happily control our LED lights with a smart phone connected via WiFi to a low-cost lighting server (a bit like a Raspberry Pi), with the smart LED fixtures and sensors connected via Bluetooth.
Opportunity knocks Almost all lights now sold are based on light-emitting diodes, but – as with all new technologies – it wasn’t initially clear which firms’ products would succeed. (Courtesy: iStock/KirVKV)
With final preparations over, the trade show opened and our first customer came up to the stand. We started giving them a demo but nothing seemed to be working – to our surprise, we simply could not get our lights to respond. A flurry of behind-the-scenes activity ensured (mostly us switching everything off and on again) but nothing made a difference.
Strangely, my phone call appeared to go dead just as I passed another booth from a rival firm
To try to get to the bottom of things, I stepped a decent distance away from our booth and rang one of our technical team up in London for support. I explained our problem but, strangely, as I walked back towards the booth, I got cut off. My call appeared to have gone dead just as I passed another booth from a rival firm called Ceravision.
It was developing high-efficiency plasma (HEP) lamps that could generate up to 100 lumens per watt – roughly where LEDs were at the time. Its lamps used radio-frequency waves to heat a plasma without needing any electrodes. Designed for sports stadia and warehouses, Ceravision’s bulbs were super bright. In fact, I was almost blinded by its products, which were pointing into the aisle, presumably to attract attention.
Back at base, our technical team frantically tried to figure out why our products weren’t working. But they were stuck and I spent the rest of the day unable to demonstrate our products to potential customers. Then, as if by magic, our system started working again – just as someone noticed we weren’t being blinded by Ceravision’s light any more.
I walked over to Ceravision’s booth as its team was packing up and had a chat with a sales guy, who I asked how the system worked. He told me it was a microwave waveguide light source but didn’t appear to know much more. So I asked him if he wouldn’t mind turning on the light again to demonstrate its output, which he did.
I glanced back at my team, who a few moments earlier had been all smiles that our system was now working, even if they were confused as to why. Suddenly, as the Ceravision light was turned on, our system broke down again. I requested to speak to one of Ceravision’s technical team but was told they wouldn’t be back until the following day.
Lighting the way
I left for the show’s awards dinner, where my firm won an innovation award for the wireless lighting system we’d been trying – unsuccessfully – to demonstrate all day. Later that night, I started looking into Ceravision in more detail. Based in Milton Keynes, UK, its idea of an electrodeless lamp wasn’t new – Nikola Tesla had filed a patent for such a device back in 1894.
Tesla realized that this type of lamp would benefit from a long life and little discolouration as there are no electrodes to degrade or break. In fact, Tesla knew how to get around the bulbs’ technical drawbacks, which involved constraining the radio waves and minimizing their power. Eventually, in the late 1990s, as radio-frequency sources became available, a US company called Fusion Lighting got this technology to market.
Its lamps consisted of a golf ball-sized fused-quartz bulb containing several milligrams of sulphur powder and argon gas at the end of a thin glass spindle. Enclosed in a microwave-resonant wire-mesh cage, the bulb was bombarded by 2.45 GHz microwaves from a magnetron of the kind you get in a microwave oven. The bulb had a design lifetime of about 60,000 hours and emitted 100 lumens per watt.
Unfortunately, its efficiency was poor as 80–85% of the light generated was trapped inside the opaque ceramic waveguide. Worse still, various satellite companies petitioned the US authorities to force Fusion Lighting to cut its electromagnetic emissions by 99.9%. They feared that otherwise its bulbs would interfere with WiFi, cordless phones and satellite radio services in North America, which also operate at 2.4 GHz.
Tasty stuff LED lights these days are used in all corners of modern life – including to grow plants for food. (Courtesy: iStock/Supersmario)
In 2001 Fusion Lighting agreed to install a perforated metal shield around its lamps to reduce electromagnetic emissions by 95%. However, this decision only reduced the light output, making the bulbs even more inefficient. Ceravision’s solution was an optically clear quartz waveguide and integrated lamp that yielded 100–5000 watts of power without any damage to the lamps.
The company claimed its technology was ideal for growing plants – delivering blue and ultraviolet light missing from other sources – along with everything from sterilizing water to illuminating TV studios. And whereas most magnetrons break down after about 2000 hours, Ceravision’s magnetrons lasted for more than 40,000 hours. It had even signed an agreement with Toshiba to build high-efficient magnetrons.
What could possibly go wrong?
The truth hurts
Back at the trade show, I arrived the following morning determined to get a resolution with the Ceravision technical team. Casually, I asked one of them, who had turned up early, to come over to our booth to see our system. It was working – until the rest of Ceravision arrived and switched their lights on. Once again, our award-winning system gave up the ghost.
Ceravision refused to accept there was a correlation between their lights going on and our system breaking down
Things then got a little ugly. Ceravision staff refused to accept there was a correlation between their lights going on and our system breaking down. Our product must be rubbish, they said. If so, I asked, how come my mobile phone had stopped working too? Silence. I went to talk to the show organizers – all they could do was tell Ceravision to point its lights down, rather than into the aisle.
This actually worked for us as it seemed the “blocking signal” was reasonably directional. It became obvious from Ceravision’s defensive response that this WiFi blocking problem must have come up before – its staff had some over-elaborate and clearly rehearsed responses. As for us, we simply couldn’t show our product in its best light to customers, who just felt it wasn’t ready for market.
Messaging matters For companies developing new products, it’s vital to listen and act on customer feedback as the market develops. (Courtesy: Shutterstock/magic pictures)
In the intervening years, I’ve talked to several ex-employees of Ceravision, who’ve all told me it trialled lots of different systems for different markets. But its products always proved to be too expensive and too complex – and eventually LEDs caught up with them. When I asked them about lights disrupting WiFi, most either didn’t seem aware of the issue or, if they did, knew it couldn’t be fixed without slashing the light output and efficiency of the bulbs.
That in turn forced Ceravision to look for ever more niche applications with ever-tinier markets. Many staff left, realizing the emperor had no clothes. Eventually, market forces took their toll on the company, which put its lighting business into receivership in 2020. Its parent company with all the patents and intellectual property suffered a similar fate in 2023.
I suspect (but don’t know for sure) that the window of opportunity closed due to high system costs, overly complex manufacturing and low volumes, coupled with LEDs becoming cheaper and more efficient. Looking back on 2014, the writing was really on the wall for the technology, even if no-one wanted to read the warning signs. A product that disrupts your mobile or WiFi signal was simply never going to succeed.
It was a classic case of a company having a small window of opportunity before better solutions came along and it missed the proverbial boat. Of course, hindsight is a wonderful thing. Clearly the staff could see what was wrong, but it took a long time for managers and investors to see or tackle the issues too. Perhaps when you are trying to deliver on promises you end up focusing on the wrong things.
The moral of the story is straightforward: constantly review your customers’ feedback and re-evaluate your products as the market develops. In business, it really is that simple if you want to succeed.
Connexion
