One-Shot Microscope Captures Space and Energy in Femtoseconds

A conventional image answers one question beautifully: where? A spectrum answers another: what is it, and what electronic or chemical state is it in? Modern instruments can join those answers, but often by scanning a specimen point by point and assembling the result later. That bargain breaks down when the subject changes faster than the scan—or when the first flash of X-rays permanently changes the thing being observed.

A team from the University of Tokyo, RIKEN and collaborating institutions has now demonstrated a different bargain. Its soft-X-ray hyperspectral microscope collects spectra from many sample positions during one pulse shorter than 100 femtoseconds. A femtosecond is one quadrillionth of a second.

The information a normal image leaves behind

Hyperspectral imaging gives every sampled location a spectrum rather than a single brightness value or three broad color channels. In the soft-X-ray region, absorption changes around the characteristic energies of elements and chemical bonds. The result can become a map of composition or chemical state, not merely shape.

The research was led by Assistant Professor Yoko Takeo and Associate Professor Takashi Kimura of the University of Tokyo’s Institute for Solid State Physics, Assistant Professor Satoru Egawa of the university’s Research Center for Advanced Science and Technology, and Group Director Makina Yabashi of the RIKEN SPring-8 Center. Their paper appeared in Optica on August 20, 2026.

“Single shot” requires care. The instrument did not record the entire evolution of a reaction as a self-contained movie. It recorded spatially distributed spectral information for the instant sampled by one ultrashort pulse. Following motion over time would still require a time-resolved experimental design, often with a pump pulse that starts a process and a probe pulse whose delay is controlled.

An astronomical mirror turned inward

Soft X-rays cannot be handled like visible light in an ordinary glass microscope. The new system begins with Wolter mirrors, nested curved surfaces that reflect X-rays at grazing angles. Because the microscope is reflective, it can enlarge images across a range of photon energies without the wavelength-dependent focusing error known as chromatic aberration.

The Wolter geometry has an unusual history. German physicist Hans Wolter described a two-reflection X-ray imaging design in 1952. It became a foundation of focusing X-ray telescopes. Here, the same broad optical idea that helps observatories form images of distant high-energy sources is used to enlarge a microscopic specimen.

The Tokyo team had already developed high-precision Wolter-mirror microscopy. In 2024, related work used a roughly 30-femtosecond soft-X-ray pulse to image carbon-rich structures in a living mammalian cell before radiation damage altered the specimen. The 2026 work adds a different capability: it retains energy information at many positions.

The chip with 417 tiny spectrometers

After the Wolter microscope enlarges the image, light reaches a newly designed multi-aperture grating. The component contains 417 apertures made with semiconductor microfabrication in the University of Tokyo’s Takeda Super Clean Room. Each aperture selects a spatial channel; a variable-line-spacing transmission grating within it diffracts photon energies in different directions.

A two-dimensional detector receives the separated spectra. The challenge is geometrical: each spectrum must preserve its connection to a sample position without colliding with its neighbors. In effect, the device runs hundreds of small spectroscopic measurements in parallel rather than moving a single slit across the field.

This is not a picture with artificial color painted on afterward. It is an image in which each sampled location carries its own soft-X-ray absorption curve. — Japan.co.jp analysis

What happened at SACLA

The proof-of-principle experiment took place at BL1, the soft-X-ray beamline of SACLA in Hyogo Prefecture’s Harima Science Garden City. SACLA—the SPring-8 Angstrom Compact free electron LAser—began research operation in 2012; its soft-X-ray beamline opened to users in 2016.

The test specimen was a silicon-nitride film. Near silicon’s absorption edge at about 104 electron volts, the system distinguished changes in transmitted spectra associated with the presence of the film. The published field of view was 39 micrometers square. Aperture pitch produced spatial sampling of 2.4 micrometers horizontally and 1.6 micrometers vertically.

Those sampling figures should not be relabeled as the microscope’s spatial resolution. They describe the spacing of sampled channels. The paper also gives a theoretically estimated grating-limited resolving power of about E/ΔE = 1,000 around 100 eV. That is a theoretical limit set by the number of grating lines in each aperture, not a blanket measurement of the full instrument’s performance under every condition.

The system also revealed that the XFEL beam’s spectrum varied from shot to shot and across the beam profile. This makes the microscope useful as a potential diagnostic of the illumination itself. When a source changes within a single pulse, knowing only an average spectrum can blur the distinction between a feature of the sample and a feature of the beam.

Why eliminating the scan matters

Scanning instruments remain powerful because they can accumulate signal and build fine maps from stable specimens. But many scientifically interesting systems are not stable. A catalytic intermediate may be fleeting. A phase transition can propagate unevenly. A battery particle may change locally. A biological specimen can move or suffer radiation damage.

If each point is measured at a different time, the finished map may combine states that never existed simultaneously. Parallel acquisition reduces that temporal mismatch. It also increases the information recovered from an event that cannot be repeated precisely.

One pulse does not remove every compromise. Dividing light among many channels affects signal. Complex specimens produce overlapping spectral features. Detectors, source bandwidth, radiation dose and data reconstruction all impose limits. The present apparatus also extends roughly 6.5 meters from specimen to detector and depends on a national-scale XFEL facility. This is not a compact laboratory product.

From Röntgen to an X-ray laser

X-rays were discovered in 1895. For decades they were powerful probes but difficult to focus, because they pass through or are absorbed by materials that bend visible light. Grazing-incidence mirrors solved part of the imaging problem; synchrotrons later provided intense, tunable beams for spectroscopy and microscopy.

X-ray free-electron lasers changed the time scale. They drive relativistic electrons through magnetic structures to create extraordinarily bright, coherent pulses measured in femtoseconds. SACLA was the second facility in the world to offer an X-ray free-electron laser for research when it opened in March 2012, according to RIKEN.

Brightness brings a paradox. An intense pulse can reveal a specimen and damage it. Ultrafast experiments exploit the interval before structural damage develops—a strategy sometimes summarized as recording before destruction. Single-shot hyperspectral microscopy adds another objective: gather more kinds of information before the moment is gone.

Proof today, possible chemical movies tomorrow

The study does not establish chemical-state movies of a functioning device or cell. Turning the method into that instrument will require pump-probe integration, higher sensitivity, careful dose control and validation on chemically complex samples. The group describes those outcomes as future directions.

The achievement is narrower and more fundamental: it weakens the old trade-off between knowing where an X-ray was absorbed and knowing its energy. In ultrafast science, an experiment is often limited not by how many events can be produced, but by how much truth can be recovered from the one event that actually occurred.