Efficient spectral control of network lasers is possible and could lead to fast-switching, multifunctional light sources for on-chip photonic applications. Using experiments and mathematical models, researchers at Imperial College London demonstrated the ability to control the emission of network lasers, so that the laser emits only one desired color or color combination at a time. Network lasers are made from a mesh of nanoscale optical fibers that are fused together to form a web-like network. Light travels along the fibers, and the lightwaves interfere to create hundreds of colors simultaneously. However, the different colors, or wavelengths, are mixed in a complex fashion and emitted randomly in all directions. The researchers showed experimentally and theoretically that the nonlinearity associated with random lasing can be harnessed to achieve a high degree of design control over the lasing emissions. They demonstrated that the emission spectrum of network lasers can be precisely controlled through the optimization of spatially nonuniform patterns. Illumination patterns were created using a digital micromirror device (DMD), a computer-controlled device with thousands of mirrors. The DMD was optimized with an algorithm that enabled it to select the most suitable illumination pattern for each laser color. When the researchers shined illumination patterns on a network laser, they found that each pattern induced a different color or different combination of colors. The researchers created tools, based on physical modeling and theory, to optimize the illumination patterns and observed the network laser spectrum’s sensitivity to the patterns. The team showed that network lasers can display spectral control for over 90% of the top 50 laser modes, based on the choice of the pump profile. The degree of spectral control stems from the complexity of the network and can be increased for further flexibility or decreased for improved resiliency. The researchers believe that the complexity of the network modes is the key behind efficient spectral control and could open the way for the development of optical devices for on-chip communication, sensing, and computation. “This is an example where we saw math and physics coming together, showing how the properties of a network can affect and help control the lasing process,” professor Mauricio Barahona said. The researchers believe that network lasers could be made out of semiconductor materials to power next-generation programmable light sources and optical sensors. For example, they could be used as highly secure hardware keys, where the illumination patterns become the secure keys that generate the password in the form of the laser spectrum. Due to their extreme sensitivity, network lasers could be used as sensors for tracking changes on surrounding surfaces. “We have combined the mathematics of network theory with laser science to tame these complex lasers,” professor Riccardo Sapienza said. “We believe this will be at the heart of light processing on chips and we are testing it now as a machine learning hardware.” Barahona said that the next challenge for the team will be to design networks and illumination patterns to control the temporal profile of the laser light and encode information in it. The team is collaborating with research and industrial partners across Europe to explore applications in machine learning. The researchers are part of the EU Horizon 2020 project CORAL (Controlling Network Random Lasers on Chip), together with IBM Zurich, to drive the emerging field of network lasers toward commercialization. The research was published in Nature Communications (www.doi.org/10.1038/s41467-022-34073-3).