The nature of dark matter remains one of the main unsolved problems in astrophysics, and primordial black holes are an attractive hypothesis because they require no new particles. Microlensing, the temporary brightening of a star's light when a compact object passes near the line of sight, directly tests this idea. For masses below $10^{-10}$ solar masses, events last minutes or less, and their amplitude is suppressed because of the finite size of the star. Ground-based surveys are limited by the atmosphere and cannot reliably track such short flares in bright fields. This work stands somewhat apart from neighboring studies: most either search for more massive black holes or use other methods, whereas here JWST is used for the first time, with its stability and high resolution, for nanolensing in M31.
The authors built the FLASH pipeline, which starts with raw NIRCam level-1 data and preserves time information within the readout ramp, rather than averaging it into standard calibrated images. First, they selected 464,204 stars from F150W mosaics using a correlation with an empirical point-spread function. Then, for each frame of the ramp, they performed PSF photometry with simultaneous background modeling, which is robust to cosmic rays. Additionally, they introduced empirical corrections for detector nonlinearity and for field-dependent sensitivity variations, derived from the data itself using a ubercalibration method. Stellar radii were estimated by fitting multi-color photometry from JWST and the archival PHAT survey to BOSZ synthetic spectra. To search for events, they fit a grid of model light curves that account for the finite source size, and constraints on the primordial black hole fraction were obtained by combining an analytic calculation of the event rate with Monte Carlo simulations of the detection efficiency.
On two sets of observations of the M31 disk with a cadence of 21.5 seconds (total duration about 1 hour in each filter), the pipeline achieved photometric accuracy of about 1% for bright stars, with the empirical nonlinearity calibration noticeably improving accuracy over the standard one. After selecting 480 candidates and visual inspection, no convincing microlensing event was found. Monte Carlo simulations showed that even if all dark matter consisted of primordial black holes, the expected number of events would be about 0.03 for a mass of $10^{-9}$ solar masses. This yields a 95% upper limit of $f_{PBH}$ < 100 at a mass of about $10^{-8}$.5 solar masses. The sensitivity is set by stars with radii roughly from 5 to 100 solar radii, and the contribution of lenses in the M31 halo is negligible because of the short observing time.
This work demonstrates for the first time that JWST can act as a high-precision photometer with sub-minute time resolution in crowded fields, opening a new window for the search for light primordial black holes. Unlike neighboring works, which either use ground-based telescopes with poorer angular resolution or consider more massive objects, this shows the applicability of a space telescope precisely to the finite-source-size regime, where events are fainter but more numerous. The obtained limit is by itself weaker than ground-based constraints, but it is derived from data not optimized for this purpose, and thus serves as a conservative test of the method. This changes the assessment of prospects: a dedicated monitoring program on JWST could improve the constraints by several orders of magnitude and, for the first time, probe the mass range below $10^{-10}$ solar masses.
The topic could develop toward dedicated JWST observing programs with continuous monitoring of selected fields in M31 or other targets. Optimizing filter choices (for example, F150W2 and F322W2), avoiding frequent dithering, and increasing observation durations could boost sensitivity by tens of thousands of times. In addition, improved bad-pixel flagging and a move to pixel-lensing would allow the use of even more stars. In the longer term, the method could be extended to searches for rapid stellar variability, flares, and occultations by Solar System objects.
The results will influence the planning of future dark matter surveys, especially in connection with the Roman Space Telescope mission, as well as JWST data processing techniques for high-precision photometry.
The next step will be to conduct a dedicated JWST observing program with optimized parameters, as well as to implement full end-to-end event simulations to verify the selection efficiency.
This work is directly connected to the unsolved problem of the nature of dark matter: it tests the hypothesis that a significant fraction of it could consist of primordial black holes in a mass range that is currently poorly constrained. It also touches on the problem of calibrating JWST detectors for photometry with better than percent-level accuracy, which is important for many astrophysical applications, including measurements of the Hubble constant.
At a mass of $10^{-9}$ solar masses, a primordial black hole has a Schwarzschild radius of about 3 nanometers, comparable to the size of a virus, yet its gravitational field can brighten a star in the Andromeda galaxy by a few percent.
The idea of invisible compact objects distorting the light of distant stars echoes the novel 'Roadside Picnic' by the Strugatsky brothers, where anomalous zones are left by mysterious aliens, although here the focus is on natural dark matter candidates.