Display and Camera Together: The Fourier Pixel is Born in Zurich
On June 24, a group of researchers from the Swiss Federal Institute of Technology Zurich (ETH) presented in Nature the first bidirectional pixel, named Fourier pixel: a single optical element that generates light and, at the same moment, can collect it. It is the theoretical building block of a screen that can also function as a camera, without separating the element that displays the image from the one that captures it, two functions that today exist in distinct devices.
The work, authored by Yannik Glauser, Sander Vonk, and colleagues under the guidance of David Norris from the Optical Materials Engineering Laboratory at ETH, is based on two principles: interference of light waves and Fourier analysis. But the key to bidirectionality lies in a reversible conversion: the pixel does not act directly on ordinary light; it transforms it into waves that travel along its metallic surface (surface plasmon polaritons) and can perform this transformation in both directions. It is precisely the symmetry of the process that allows a single element to show an image and, with the same apparatus, to capture it.
In the "outgoing" direction, that of the display, as described by ETH, the pixel starts from surface waves and re-irradiates them as ordinary light. The area of the chip, sculpted with nanometer precision, determines how these re-irradiated waves overlap: where they arrive in phase they reinforce each other, and where they arrive out of phase they cancel each other out. This interplay of reinforcements and cancellations, which is simply wave interference, composes the image. Fourier analysis comes into play earlier in the process: given the desired outcome, it calculates what profile the surface must have to produce it.
In the "incoming" direction, that of the camera, the same mechanism operates in reverse. Light arriving from the outside generates surface waves, and from the way these waves recombine, the pixel reconstructs what the incoming light was carrying: not only its intensity but also its phase and polarization, that is, the direction in which the electric field of the wave oscillates. "In addition to light intensity, that is, the light and dark areas from which images arise, our Fourier pixels can also control other properties of light waves, for example, polarization," Glauser explains.
Doughnut Beams and Color Images
The pixel can produce beams of light shaped like doughnuts, with a dark hole in the center. As you move around the axis of the beam, the phase rotates continuously, and along the axis itself, opposite phases cancel each other out, nullifying the intensity. This is called phase singularity, and the beams that embody it are vortex beams. The tighter the rotation of the phase, the higher the number that measures it, the topological charge, that is, how many complete turns the phase makes in a rotation around the axis: the researchers generated it equal to +1, +3, and +5.
All of this works at different wavelengths, a necessary condition for obtaining color images.
The paper reports two quality indicators. The first is a speckle contrast of 17%: speckle is the random granularity that spoils images formed with coherent light, and such a low value indicates a uniform image, thus a phase reconstructed with great fidelity. The second is a signal-to-background ratio of 68: the useful signal is 68 times more intense than the background, meaning the residual contamination is almost nonexistent. The authors describe the result as a "scalable and universal architecture for vectorially programmable pixels": programmable, that is, treating light as a vector quantity that encompasses amplitude, phase, and polarization.
A significant point is that surface waves can be used for mathematical calculations directly on the pixel material. In the future, a pixel could react to a captured image and produce the corresponding light pattern without passing through a computer. The indicated applications range from adaptive optics to holographic displays, from optical communication to quantum information processing.
How Far is a Real Screen?
The limitations, at the current state, remain significant. Each pixel serves a fixed function, determined at the time of manufacturing, and is calibrated to a single wavelength; a conventional screen, on the other hand, shows arbitrary content across millions of elements. The demonstrations use coherent illumination, although the authors point out that incoherent excitation is possible and that coherent plasmons could be generated directly on the chip thermally, electrically, or through quantum emitters without the need for an external laser. The group has already created small arrays (one of 2x3 pixels): Norris's short-term goal is precisely to extend the method to arrays of many Fourier pixels, as is done in today’s displays and cameras. The patent request related to the research has been submitted for the ETH Spark Award.
In conclusion, it is worth remembering that the "picture element," later shortened to pixel, first appeared in print in 1927, in the magazine Wireless World: in 2027 it will celebrate its centenary, arriving at the threshold of the century with the premises of a profound revolution.