A team of researchers from Nanjing University and Tsinghua University has demonstrated a fundamentally new way to cool nanoscale devices using nothing more than a laser beam and a pair of atomically thin semiconductor layers. The approach, described in Nature on June 24, creates a temperature difference of more than 100 kelvin across a gap of less than a nanometer, a thermal gradient steep enough to open new possibilities for on-chip cooling in quantum devices and ultra-thin electronics.
The mechanism, which the researchers call interfacial charge transfer (ICT) cooling, represents a departure from how laser cooling has worked for decades.
A different kind of cooling
Conventional optical refrigeration, also known as upconversion photoluminescence cooling, works by shining a laser on a material that re-emits the absorbed energy as higher-energy photons. Because the emitted photons carry away more energy than the absorbed ones, the material cools. The catch is that this process requires near-perfect external quantum efficiency, the material must re-emit nearly every photon it absorbs, with virtually no parasitic heating. Only a handful of materials, such as cadmium sulfide nanoribbons and certain halide perovskites, can meet this standard, and even then only under carefully tuned resonant conditions.
ICT cooling works by an entirely different principle. The researchers built stacks of two different transition metal dichalcogenides (TMDs), specifically, heterobilayers of tungsten diselenide (WSe₂) paired with either molybdenum diselenide (MoSe₂) or tungsten disulfide (WS₂). Each layer is three atoms thick.
Here is what happens. A continuous-wave laser excites electron-hole pairs in the WSe₂ layer. Those photoexcited carriers then cross a type-II band-aligned junction into the adjacent MoSe₂ or WS₂ layer. The charge transfer is exceptionally fast, approximately 56 femtoseconds at room temperature in the WSe₂/WS₂ system. But crucially, the transfer is not momentum-matched. The electron must gain energy from the WSe₂ lattice vibrations, phonons, to overcome the energy barrier at the interface. That phonon absorption cools the WSe₂ lattice directly.
At the same time, the interface itself acts as a thermal barrier. Molecular dynamics simulations show that the interfacial thermal resistance is enormous, scaling exponentially with interlayer distance. This prevents heat from flowing back from the acceptor layer to the cooled donor layer, sustaining the gradient.
What was actually achieved
The authors report creating a temperature difference exceeding 100 kelvin between the WSe₂ and MoSe₂ layers under laser excitation. The exact magnitude depends on the measurement method: Raman thermometry of the WSe₂ layer shows cooling of several kelvin relative to ambient, while the temperature gradient across the sub-nanometer interlayer gap, extrapolated from simulations and supported by photoluminescence and Raman signatures, exceeds 100 K. The cooling effect was confirmed by three independent spectroscopic signatures: a drop in the anti-Stokes-to-Stokes Raman intensity ratio, a blue shift in photoluminescence, and a decrease in the electron temperature extracted from the emission spectrum.
The mechanism has several practical advantages over conventional optical refrigeration. It requires no resonant excitation, the cooling effect persists across a broad wavelength range of 1.7 to 2.0 electronvolts (approximately 620 to 730 nanometers). It tolerates a wide range of laser powers. And critically, it does not depend on photoluminescence quantum yield: cooling was observed even in chemical-vapor-deposited WSe₂ with a quantum yield of only 0.1%, far below the near-unity efficiency required for upconversion cooling.
The coupling sweet spot
The effect depends on achieving what the team calls an “intermediate coupling” state between the two layers. By precisely controlling the interlayer distance and twist angle through aligned dry-transfer stacking, they create a regime in which charge transfer remains efficient while the momentum mismatch necessary for phonon absorption is preserved. Too-strong coupling (small interlayer distance, strong hybridization) suppresses the momentum mismatch and kills the effect. Too-weak coupling (large separation, negligible interaction) prevents efficient charge transfer altogether.
The team identified three coupling regimes: weak (labeled T for tribological, negligible interlayer interaction), the sweet spot (H for heterobilayer, intermediate coupling), and strong (S for strongly hybridized). Only the intermediate H regime produces the cooling effect.
What it enables
The potential applications are in nanoscale thermal management. Current cryogenic cooling for quantum devices requires bulky cryostats that limit miniaturization and integration. An on-chip cooling method that works at the few-nanometer scale and requires only a laser source could change how thermal management is designed in quantum optoelectronic systems, ultra-thin electronics, and densely packed photonic circuits. The materials used, WSe₂, MoSe₂, WS₂, are among the most widely studied TMD semiconductors and can be fabricated by standard methods.
Caveats
The demonstration remains at the single-heterostructure flake level. Scaling to practical device arrays has not been addressed. The cooling effect is stronger when the heterobilayer is suspended and isolated from the substrate, a thermally conductive substrate acts as a heat sink and reduces efficiency, meaning practical devices will need careful thermal isolation engineering. The interfacial thermal resistance that enables the gradient also limits how fast heat can be extracted from the cooled layer, potentially capping steady-state cooling power.
The mechanism has been demonstrated only in two material combinations: WSe₂/MoSe₂ and WSe₂/WS₂. Whether it generalizes to other TMD pairs or other 2D material families, such as black phosphorus or III-VI compounds, remains to be shown.
First author Jiamin Lin of Nanjing University and co-first authors Baixu Xiang and Renguang Liu of Tsinghua University, along with corresponding authors Weigao Xu, Qihua Xiong, and Huajian Gao, have demonstrated a genuinely new physical mechanism for laser cooling. The temperature gradient itself may not yet match the headline numbers that traditional optical refrigeration can achieve in its best materials, but the mechanism, broad wavelength tolerance, impurity-tolerant, working in the most common family of 2D semiconductors, opens a thermal engineering toolbox that did not exist before.
Source: Lin J, Xiang B, Liu R, et al. Optical cooling by interfacial charge transfer in 2D heterostructures. Nature. Published online June 24, 2026. doi:10.1038/s41586-026-10662-w