Researchers have developed a novel MoS₂ phototransistor with ultra-high gain, capable of detecting extremely weak light signals and attomolar concentrations of disease biomarkers at room temperature. This innovation paves the way for ultra-sensitive, rapid, and reliable point-of-care diagnostics.

To address the Ohmic losses and limited integration of conventional topological metasurfaces, Chinese scientists developed a novel approach by harnessing on-chip all-dielectric metasurfaces to precisely extract optical guided waves, replacing the loss-inducing role of metals. This strategy enables the creation and control of topological exceptional points in an all-dielectric environment, overcoming traditional limitations while leveraging the low-loss and high-integration advantages of on-chip dielectric platforms.

For the first time, light itself has been sculpted into a toron—a 3D pinwheel-like chiral topology that marries skyrmion textures with paired synthetic magnetic monopoles. Using vector structured beams, the scientists create, image and continuously tune these “pinwheels of light,” and watch them morph into typical topologies such as hopfions and monopole pairs on demand. The platform promises topologically protected information channels in free space and new routes for light–matter interaction.

Label-free detection of biological events at single-cell resolution in the brain can non-invasively capture brain status for medical diagnosis and basic neuroscience research. We have developed a new label-free, multiphoton photoacoustic microscope (LF-MP-PAM) with a near-infrared femtosecond laser to observe endogenous NAD(P)H in living cells. We demonstrated the detection of endogenous NAD(P)H photoacoustic signals in brain slices to 700 μm depth and in cerebral organoids to 1100 μm depth.

Researchers from UNamur and Stanford have developed a compact, energy-efficient photonic device that steers light using twisted crystal layers. This innovation enables precise beam control, potentially revolutionizing satellite tracking, VR headsets, lasers, and quantum computing. The breakthrough uses fast simulations and machine learning to optimize design, offering a powerful new tool in light manipulation.

To address growing demands for efficient AI computing, scientists in China developed a reconfigurable integrated photonic chip that supports diverse neural network models within a unified hardware architecture. By co-designing algorithms and photonic hardware, the photonic chip enables in situ switching among fully connected, convolutional, and recurrent models powered by a soliton microcomb. It demonstrates the ability to process image, speech, and text information, marking a key step toward multifunctional photonic AI processors.

We’re often told it is “unscientific” or “meaningless” to ask what happened before the big bang. But a new paper by FQxI cosmologist Eugene Lim, of King's College London, UK, and astrophysicists Katy Clough, of Queen Mary University of London, UK, and Josu Aurrekoetxea, at Oxford University, UK, published in Living Reviews in Relativity, in June 2025, proposes a way forward: using complex computer simulations to numerically (rather than exactly) solve Einstein’s equations for gravity in extreme situations. The team argues that numerical relativity should be applied increasingly in cosmology to probe some of the universe’s biggest questions–including what happened before the big bang, whether we live in a multiverse, if our universe has collided with a neighboring cosmos, or whether our universe cycled through a series of bangs and crunches.