Report 253 · Lab Science
What a single-molecule spectrum detected
In August 2026 a team from Innsbruck and Warsaw published, in Nature, an infrared absorption spectrum of one molecule: a single CaOH+ ion held in a trap next to a single calcium ion. Coverage called it spectroscopy extended to single molecules, which is fair. What it actually detected is stranger and more interesting than "a very sensitive spectrometer," and the paper is candid about what its own model still can't explain.
Spectroscopy is my research field (microwave and rotational work, not this group's infrared setup), and I have no connection to this paper or its authors. The paper is open access, so you can check every number below against it.
Why one molecule is hard
Ordinary absorption spectroscopy compares how much light goes into a sample with how much comes out. With trillions of molecules in the beam, the missing light is easy to see. With one molecule, the paper explains, fluctuations and the quantum noise of the light itself swamp the tiny amount that gets absorbed. Earlier single-atom and single-molecule absorption experiments worked with visible light, needed many photons to be absorbed, and relied on efficient low-noise detectors that don't exist for much of the spectrum.
Molecular ions add another problem: they repel each other, so you can't pack them into a dense sample. The usual workaround is to look for a side effect of absorption, such as the molecule breaking apart (photodissociation) or reacting. Those methods often disturb or destroy the molecule you were trying to measure.
What they measured instead: a kick
When a molecule absorbs a photon it also absorbs the photon's momentum, so it recoils very slightly. Zhenlin Wu, Philipp Schindler and colleagues trapped one CaOH+ ion together with one 40Ca+ ion in a Paul trap. The two ions push on each other electrically, so they move together, and a kick to the molecule shows up in the motion of the pair. The calcium ion can be controlled and read out with lasers in the same way trapped-ion quantum computers read their qubits. The molecule itself never has to emit or scatter anything you detect.
The catch is size. In the paper's units, one mid-infrared photon displaces the pair's motion by about 0.02 to 0.03, far too small to see directly against the experiment's noise. So the team first put the two-ion motion into a "Schrödinger cat" state, a superposition of the crystal moving two opposite ways at once, entangled with the calcium ion's internal state. A recoil during that window turns into a phase shift between the two halves, which is amplified by the size of the cat state. Undoing the cat state then maps that phase onto the calcium ion, and the calcium ion's fluorescence says whether a kick happened.
They calibrated this before touching the molecule, using tiny electric pulses sized to mimic a mid-infrared photon's kick. With the cat amplitude they settled on, decoherence (from magnetic-field and trap-voltage noise) cut the achievable signal to about half of the ideal value. Bigger cat states should amplify more, but in practice they also lost coherence faster, so the signal stopped growing.
The spectrum
The target was the O–H stretching vibration of CaOH+, predicted by the Warsaw group's coupled-cluster calculations at 3,783 cm−1 (about 2.64 µm). The light came from a femtosecond laser at 100 kHz, timed so every pulse kicked the ions in step with their motion. At 3,703 cm−1, a train of 34 pulses produced a signal of 0.12 on the calcium ion. Sweeping the laser from 3,600 to 4,000 cm−1 traced an absorption peak centered close to the predicted value, with a width that matched their simulation.
That is a genuine first. The paper describes it as the first use of this recoil method to detect single-photon absorption on a molecular ion, and the abstract's phrase is "single-photon sensitivity." Earlier versions of the technique had been shown on atoms, where many absorption and emission events added up.
Four things the coverage smoothed over
1. "Agreement with theory" is half true. One write-up said the spectrum "shows good agreement with theoretical calculations." The peak position and width did agree. But the paper also says the model "fails to explain the absolute magnitude" of the measured absorption probability, and that it captures the qualitative features "but does not fully explain the observed dynamics and spectral structure." The authors say they plan more experimental and theoretical work to fix that. A peak in the right place with the wrong height is real progress, not a closed case.
2. It is not single-shot yet. Each point on the spectrum comes from repeating the sequence many times and fitting the calcium ion's response. Detecting one absorption event in one shot is something the authors say they could do "with large-enough amplification," which makes it a goal, not a result. That matters for the quantum-technology uses they describe, like non-destructive readout of a molecule's state, which depend on single-shot detection.
3. The resolution is coarse on purpose. The laser's linewidth was about 127 cm−1, much wider than the molecule's rotational fine structure. The team chose that deliberately so the result wouldn't depend on which rotational state the molecule happened to be in. The cost is that this is a low-resolution band shape, not a high-precision line position. The paper sketches how a narrow-linewidth pulse train could get to high resolution, but that is future work.
4. "Single molecule" means one ion in ultrahigh vacuum. Making the sample meant loading two calcium ions, leaking in water vapor until one of them reacted and went dark on the camera (about five minutes), then pumping back down to around 10−10 mbar. Background-gas collisions make the two ions swap places every few seconds, and the team has to detect and reshuffle the order because the measurement only works in one configuration. This is a quantum-physics platform for charged molecules that can be trapped next to a suitable atom. It is not a way to take the spectrum of one molecule in a sample, a cell or a drug tablet.
Where it reaches, and where it doesn't
The kick scales with the photon's momentum, so the method favors high-frequency light. The authors estimate that with their parameters it should detect absorption on transitions down to about 2,000 cm−1 (roughly 60 THz). They also note it works best when the excited state lives much longer than it takes to build the cat state, which fits vibrational transitions better than the bright visible and ultraviolet ones.
From my side of spectroscopy, that floor is the interesting part. Rotational transitions, the ones I work with, sit in the microwave and millimeter-wave range. A 10 GHz microwave photon is about 0.33 cm−1, roughly 6,000 times less momentum than a 2,000 cm−1 infrared photon (my arithmetic). So recoil detection like this is an infrared and higher-frequency tool. It complements, rather than replaces, the quantum-logic methods that already read the internal states of simpler, mostly diatomic, trapped molecular ions, which the paper cites as its starting point. If you want the gas-phase rotational side of the story, I covered it in how a molecule is found in space and how to tell mirror-image molecules apart.
How I'd read it
This is careful, honest work. The authors calibrated their detector with a known kick before trusting it, stated the decoherence penalty, published their data openly, and wrote their model's shortfall into the main text instead of burying it. The fair one-line summary: a team detected infrared absorption in one trapped polyatomic ion by measuring the momentum kick of the absorbed photons, and the band showed up where theory put it, though the theory doesn't yet get its strength right. It's an important step toward controlling and reading out complex molecules one at a time. It is not a single-molecule spectrometer for everyday chemistry, and nobody involved claims it is.
What would change my read: a follow-up that gets the absolute absorption right, single-shot detection, or a high-resolution line measured this way. If one lands, I'll update this post in place.
Sources
- Zhenlin Wu, Tim Duka, Mariano Isaza-Monsalve, Miriam Kautzky, Vojtěch Švarc, Andrea Turci, René Nardi, Marcin Gronowski, Michał Tomza, Brandon J. Furey and Philipp Schindler, "Infrared absorption spectroscopy of a single polyatomic molecular ion," Nature 657, 364–368 (published online 26 August 2026; issue of 10 September 2026), DOI 10.1038/s41586-026-10915-8. (Primary source, open access, CC BY 4.0. Full main text and Methods read on nature.com; metadata checked via Crossref. Source of the method, the calibration with electric kicks, the 0.52 decoherence factor, the 3,783 cm−1 prediction, the 34-pulse signal of 0.12 at 3,703 cm−1, the 3,600–4,000 cm−1 scan, the 126.7 cm−1 laser linewidth, the molecule-loading and ion-swapping details, the 2,000 cm−1 estimate and every quoted phrase. Data on Zenodo, DOI 10.5281/zenodo.19727569, not downloaded.)
- Zhenlin Wu et al., "Infrared absorption spectroscopy of a single polyatomic molecular ion," arXiv:2511.19687 (preprint, November 2025). (The preprint version of the paper above. Downloaded; this post relies on the published Nature text.)
- "Quantum Boost Extends Spectroscopy to Single Molecules," Optics & Photonics News (Optica), 3 September 2026. (Read. Source of the "shows good agreement with theoretical calculations" phrasing and the 60 THz figure.)
- University of Warsaw, "Measuring a single molecule without disturbing its state" (news release, 2026). (Read. Source of the split of work: experiment in Schindler's Innsbruck group, theory by Tomza and Gronowski in Warsaw.)
The 2.64 µm wavelength, the 0.33 cm−1 value for 10 GHz and the roughly 6,000-fold momentum ratio are my arithmetic (wavenumber = frequency ÷ speed of light). The view that recoil detection complements rather than replaces quantum-logic methods for rotational states is my reading, not a claim the paper makes.