SN 2025mkn: Lensed Type II Supernova

• SN 2025mkn is a gravitationally lensed Type II supernova at z=1.371, characterized by multiple images produced by an elliptical galaxy acting as a natural >100× telescope.
• The resolved fold-pair and fainter counterimage enable spectral phase comparisons that provide alternative avenues for measuring lensing time delays and probing explosion physics.
• Lens modeling combined with multi-instrument data confirms magnification estimates of 100–300× while highlighting potential systematic effects like microlensing on flux ratios.

Searching arXiv for the specified paper to ground the article in the cited source. SN 2025mkn is a gravitationally lensed Type II supernova at source redshift $z_{\mathrm{s}} = 1.371$, lensed by an elliptical galaxy at lens redshift $z_{\ell} = 0.4203 \pm 0.0001$, and presented as a case in which galaxy-scale strong lensing functions as a “natural $\gtrsim 100\times$ telescope” (Lemon et al., 9 Apr 2026). It was first detected in the Zwicky Transient Facility as a blue transient located $0.83^{\prime\prime}$ from the center of the lens galaxy, and subsequent spectroscopic and imaging follow-up with SNIFS on the University of Hawai‘i 2.2 m telescope, Keck/LRIS, and JWST established both its supernova classification and its multiply imaged lensing configuration (Lemon et al., 9 Apr 2026). The system combines unusually high magnification, resolved image multiplicity, and Type II spectral diagnostics, making it relevant both to high-redshift core-collapse supernova studies and to prospective time-delay cosmography.

1. Discovery and source identification

SN 2025mkn was first flagged in ZTF as a very blue transient near an early-type galaxy, and ZTF cross-matching to photometric redshifts immediately indicated anomalously high apparent luminosity (Lemon et al., 9 Apr 2026). The transient lies $0.83^{\prime\prime}$ from the center of the lens galaxy. A DESI spectrum of the galaxy fixed the lens redshift at $z_{\ell} = 0.4203 \pm 0.0001$.

Early SNIFS classification on 2025 May 30 showed a nearly featureless blue continuum, with a blackbody temperature around $27{,}000\ \mathrm{K}$, and narrow absorption lines at two distinct redshifts, $z = 1.256$ and $z = 1.371$. Keck/LRIS optical spectra, obtained as $600\ \mathrm{s} \times 2$ per epoch on 2025 June 24, July 25, and July 28, revealed Mg II $z_{\ell} = 0.4203 \pm 0.0001$0 and Fe II absorption systems consistent with $z_{\ell} = 0.4203 \pm 0.0001$1, while JWST/NIRSpec confirmed strong Balmer emission at the same redshift. Together, these observations established the source redshift at $z_{\ell} = 0.4203 \pm 0.0001$2 and the classification as a Type II supernova (Lemon et al., 9 Apr 2026).

The spectroscopic identification rests on both absorption and emission diagnostics. The NIRSpec spectra show broad Balmer emission in image A at a rest-frame phase of $z_{\ell} = 0.4203 \pm 0.0001$3 days, with H$z_{\ell} = 0.4203 \pm 0.0001$4, H$z_{\ell} = 0.4203 \pm 0.0001$5, and H$z_{\ell} = 0.4203 \pm 0.0001$6, and strong H$z_{\ell} = 0.4203 \pm 0.0001$7 in image B at the same redshift. The observed wavelengths obey the standard redshift relation

$z_{\ell} = 0.4203 \pm 0.0001$8

Using $z_{\ell} = 0.4203 \pm 0.0001$9 and $\gtrsim 100\times$0 gives

$\gtrsim 100\times$1

which lies within the NIRSpec bandpass (Lemon et al., 9 Apr 2026). The presence of broad Balmer lines and the continuum shape match expectations for Type II supernovae.

2. Resolved lensing configuration

At JWST angular resolution, the system was resolved into multiple images. NIRCam imaging revealed a bright image A that, upon PSF modeling, is itself a very close pair of point sources, A1 and A2, separated by $\gtrsim 100\times$2 and straddling the tangential critical curve, together with a much fainter counterimage B approximately $\gtrsim 100\times$3 to the opposite side of the lens and about $\gtrsim 100\times$4 fainter than A in flux (Lemon et al., 9 Apr 2026). Residuals also show an arc-like structure southwest of the lens, interpreted as the lensed host galaxy, and a hint of a fourth image C in the NIRSpec white-light slice.

Synthetic photometry and astrometry quantify the configuration in the F200W band. Relative to the lens center, A1 is at $\gtrsim 100\times$5, A2 at $\gtrsim 100\times$6, and B at $\gtrsim 100\times$7 (Lemon et al., 9 Apr 2026). The A1/A2 flux ratio is very close to unity, as expected for a fold-pair straddling a critical curve, while A/B flux ratios of $\gtrsim 100\times$8–$\gtrsim 100\times$9 are measured across the JWST bands, including $0.83^{\prime\prime}$0 in F277W and $0.83^{\prime\prime}$1 in NIRSpec F150W synthetic photometry.

The morphology is characteristic of a source located near a fold or cusp caustic. In the paper’s lensing notation, the magnification is

$0.83^{\prime\prime}$2

with $0.83^{\prime\prime}$3 the Jacobian matrix $0.83^{\prime\prime}$4 of the lens mapping (Lemon et al., 9 Apr 2026). Near critical curves, $0.83^{\prime\prime}$5 and $0.83^{\prime\prime}$6 can be very large. The near-equality of the absolute A1 and A2 fluxes and their placement across the critical curve are consistent with the fold-pair interpretation.

A plausible implication is that SN 2025mkn occupies an observationally valuable regime in which the local image geometry directly exposes the caustic structure of the lens potential while simultaneously delivering extreme magnification.

3. Spectroscopy, temporal ordering, and phase information

NIRSpec integral-field spectroscopy with G140M and F100LP, covering $0.83^{\prime\prime}$7–$0.83^{\prime\prime}$8 at $0.83^{\prime\prime}$9, cleanly resolved image B and extracted the spectrum of A with a PSF model anchored to the NIRCam positions (Lemon et al., 9 Apr 2026). The NIRSpec setup used NRSRAPID readout, $0.83^{\prime\prime}$0, $0.83^{\prime\prime}$1, total $0.83^{\prime\prime}$2, and a spaxel scale of $0.83^{\prime\prime}$3.

An important empirical result is that the spectral phase of B appears more evolved than A at the same epoch, indicating that B arrived first and had already faded by the time of discovery, whereas A arrived after a significant lensing time delay (Lemon et al., 9 Apr 2026). This ordering is described as a robust prediction of lens models for sources close to fold or cusp caustics, and it is borne out by the data. The leading image is therefore not the brightest image, but the faint counterimage.

This arrival sequence has direct methodological consequences. A search for a leading image in pre-discovery ZTF data yielded $0.83^{\prime\prime}$4 non-detections at all epochs, consistent with the faint first-arriving image being below ZTF thresholds. As a result, light-curve-based time-delay measurements are not possible, because the first image was the faintest and was undetected prior to the rise of A (Lemon et al., 9 Apr 2026). However, the resolved spectra offer a different route: time-delay estimation through spectral phase retrieval, using the evolution of Balmer line profiles and continuum temperature at matched rest-frame phases.

The lensing time-delay surface is given as

$0.83^{\prime\prime}$5

where $0.83^{\prime\prime}$6 is the lensing potential and $0.83^{\prime\prime}$7, $0.83^{\prime\prime}$8, and $0.83^{\prime\prime}$9 are the angular diameter distances to the lens, to the source, and between lens and source (Lemon et al., 9 Apr 2026). The distance combination often used is $z_{\ell} = 0.4203 \pm 0.0001$0, which sets the scale of time delays for a given mass model. Matching spectral phases between A and B should yield a delay of order weeks, whereas the model-predicted A1–A2 delay is only $z_{\ell} = 0.4203 \pm 0.0001$1 minute and is effectively unmeasurable.

4. Photometry, luminosity analogs, and magnification

The light curve of image A was tracked with ZTF $z_{\ell} = 0.4203 \pm 0.0001$2 and follow-up observations from SEDM/Palomar 60″ $z_{\ell} = 0.4203 \pm 0.0001$3, NOT/ALFOSC $z_{\ell} = 0.4203 \pm 0.0001$4, Liverpool $z_{\ell} = 0.4203 \pm 0.0001$5, and Wendelstein/FTW 3KK $z_{\ell} = 0.4203 \pm 0.0001$6, processed with scene modeling and image subtraction and calibrated to Pan-STARRS and 2MASS/UKIRT templates (Lemon et al., 9 Apr 2026). ZTF coverage had typical cadence $z_{\ell} = 0.4203 \pm 0.0001$7–$z_{\ell} = 0.4203 \pm 0.0001$8 days and only short seasonal gaps.

JWST photometry, compared to the nearby Type II SN 2023ixf scaled to $z_{\ell} = 0.4203 \pm 0.0001$9 in matched rest-frame filters and phases, shows remarkably similar light-curve shapes and colors (Lemon et al., 9 Apr 2026). Matching the observed brightness of image A to that of SN 2023ixf yields $27{,}000\ \mathrm{K}$0 mag, or $27{,}000\ \mathrm{K}$1, while spectral scaling at similar phases likewise suggests $27{,}000\ \mathrm{K}$2 and $27{,}000\ \mathrm{K}$3. The paper also notes that Type II supernovae exhibit $27{,}000\ \mathrm{K}$4 mag scatter at peak, so magnification estimates using luminosity analogs carry systematic uncertainty.

Independent lens modeling confirms that $27{,}000\ \mathrm{K}$5 is inevitable given the geometry and positions, and total magnifications of order $27{,}000\ \mathrm{K}$6–$27{,}000\ \mathrm{K}$7 are consistent with the resolved image arrangement and the source location near the astroid caustic (Lemon et al., 9 Apr 2026). In this sense, the “natural telescope” characterization is not purely descriptive but a quantitative statement about the effective flux amplification delivered by the lens.

A common misconception would be to treat the $27{,}000\ \mathrm{K}$8 estimate as a direct measurement independent of supernova population variance. The paper does not do that. Instead, it presents the similarity to SN 2023ixf as an empirical analog-based argument and explicitly notes the intrinsic scatter of SNe II. This suggests that the strongest conclusion is the lower bound $27{,}000\ \mathrm{K}$9, while values near $z = 1.256$0–$z = 1.256$1 are supported when photometric and geometric information are considered together.

5. Lens modeling and the fourth image

A parametric mass model was constructed with lenstronomy, using a singular isothermal ellipsoid aligned with the lens light plus external shear, constrained by the JWST astrometry and assuming that the lens mass centroid, ellipticity, and orientation follow the light, with axis ratio $z = 1.256$2 and position angle $z = 1.256$3 North of East (Lemon et al., 9 Apr 2026). Fitting five free parameters—Einstein radius $z = 1.256$4, shear $z = 1.256$5 and angle, and source position—to the three image positions yields $z = 1.256$6 and $z = 1.256$7 at $z = 1.256$8, with reduced $z = 1.256$9.

The model recovers A1 and A2 straddling the critical curve, with $z = 1.371$0 and $z = 1.371$1, respectively, B as a demagnified minimum at $z = 1.371$2, and predicts a fourth image C at $z = 1.371$3 with $z = 1.371$4 (Lemon et al., 9 Apr 2026). The small shear supports the assumption that lens mass follows the light.

Independent lens-mass estimates from broadband SED fitting of the lens galaxy, using Pan-STARRS, 2MASS, and WISE photometry with Prospector, FSPS, and a Chabrier IMF, give $z = 1.371$5 and $z = 1.371$6 (Lemon et al., 9 Apr 2026). The DESI-measured velocity dispersion of $z = 1.371$7, assuming an isothermal profile, implies $z = 1.371$8. Both estimates are stated to be consistent with the image separations and the presence of a distant counterimage.

The predicted A/B flux ratio in mass-light aligned models is $z = 1.371$9, larger than the observed $600\ \mathrm{s} \times 2$0–$600\ \mathrm{s} \times 2$1 after accounting for time delays (Lemon et al., 9 Apr 2026). The proposed resolution is microlensing of the saddle-point images: demagnifying C by $600\ \mathrm{s} \times 2$2 mag, found in $600\ \mathrm{s} \times 2$3 of microlensing realizations for typical smooth matter fraction and stellar mass assumptions, and/or boosting B by $600\ \mathrm{s} \times 2$4 mag, occurring in $600\ \mathrm{s} \times 2$5 of cases, can reconcile the A/B ratio and the apparent non-detection of C in NIRCam imaging. The NIRSpec white-light residual hints at C at a level $600\ \mathrm{s} \times 2$6 of B, consistent with substantial microlensing of the saddle image.

A plausible implication is that fuller incorporation of the extended host arc, which the paper identifies as future work, will sharpen the inference on both macro-model parameters and microlensing contributions.

6. Observational infrastructure and scientific significance

The observational campaign combined wide-field transient discovery, optical spectroscopy, high-resolution space imaging, infrared integral-field spectroscopy, and radio follow-up (Lemon et al., 9 Apr 2026). ZTF provided early discovery and cadenced $600\ \mathrm{s} \times 2$7 light curves, with scene modeling and subtraction against Pan-STARRS templates used to isolate the transient from the lens and a nearby star. SNIFS/UH2.2m integral-field spectroscopy covered the blue channel $600\ \mathrm{s} \times 2$8–$600\ \mathrm{s} \times 2$9 and red $z_{\ell} = 0.4203 \pm 0.0001$00–$z_{\ell} = 0.4203 \pm 0.0001$01 in a $z_{\ell} = 0.4203 \pm 0.0001$02 exposure obtained in $z_{\ell} = 0.4203 \pm 0.0001$03 seeing. Keck/LRIS used the B400/3400 grism and R400/8500 grating with a $z_{\ell} = 0.4203 \pm 0.0001$04 slit at parallactic angle and LPipe reductions. JWST/NIRCam employed program 3468, module B, FULL subarray, SW F150W and F200W paired with LW F277W, RAPID readout, SMALL-GRID-DITHER, and four integrations per filter; short exposures of $z_{\ell} = 0.4203 \pm 0.0001$05 per SW filter and $z_{\ell} = 0.4203 \pm 0.0001$06 in F277W were sufficient to resolve the close A1–A2 pair and the faint B image. JWST/NIRSpec G140M used a four-point dither and delivered synthetic photometry consistent with NIRCam within $z_{\ell} = 0.4203 \pm 0.0001$07 mag systematics.

VLA follow-up found the lens galaxy to be radio bright, with $z_{\ell} = 0.4203 \pm 0.0001$08 flux density $z_{\ell} = 0.4203 \pm 0.0001$09 and $z_{\ell} = 0.4203 \pm 0.0001$10, while the supernova images were undetected at rms $z_{\ell} = 0.4203 \pm 0.0001$11–$z_{\ell} = 0.4203 \pm 0.0001$12 (Lemon et al., 9 Apr 2026). The paper states that this disfavors an LFBOT interpretation and supports a normal core-collapse supernova.

With $z_{\ell} = 0.4203 \pm 0.0001$13 and likely $z_{\ell} = 0.4203 \pm 0.0001$14–$z_{\ell} = 0.4203 \pm 0.0001$15, SN 2025mkn extends high-redshift core-collapse supernova studies into a regime where rest-frame far-UV to near-IR spectroscopy and photometry at $z_{\ell} = 0.4203 \pm 0.0001$16 reach signal-to-noise comparable to local events (Lemon et al., 9 Apr 2026). The paper identifies stringent tests of explosion physics, hydrogen recombination and line formation, circumstellar interaction, dust formation, and metallicity effects after $z_{\ell} = 0.4203 \pm 0.0001$17 Gyr of cosmic evolution as enabled scientific directions. It also states that strongly lensed supernovae are rare but increasingly found in modern time-domain surveys, and that the large flux ratio and leading-image faintness in this system may fall outside some mock-catalog expectations that impose brightness thresholds on all images. This suggests that spectroscopic follow-up and high-resolution imaging are especially important for revealing asymmetric configurations of this kind.

7. Limitations, caveats, and future use in cosmography

Several uncertainties are explicit. The magnification inference relies partly on SN 2023ixf as a luminosity analog, and the paper notes intrinsic scatter in SNe II (Lemon et al., 9 Apr 2026). Possible microlensing of saddle images affects flux ratios, and the current lens model assumes that mass follows light without yet using all host-arc constraints. These are not peripheral issues: they directly affect the translation from image geometry and flux ratios into precise magnification and time-delay inferences.

Nevertheless, the paper identifies a specific cosmographic pathway. Because light-curve-based delay measurements are precluded by the first-arriving image being the faintest, the proposed route is phase-based delay estimation from the resolved NIRSpec spectra (Lemon et al., 9 Apr 2026). The broad hydrogen features are described as reliable phase diagnostics that are relatively insensitive to chromatic microlensing, and these can be combined with parametric lens models and host-arc constraints to infer $z_{\ell} = 0.4203 \pm 0.0001$18. The intended procedure is to match spectral phases between A and B while remaining blinded to exact model predictions, with an expected delay of order weeks.

In summary, SN 2025mkn is a multiply imaged Type II supernova at $z_{\ell} = 0.4203 \pm 0.0001$19 whose observed configuration consists of a close fold-pair A1/A2 separated by $z_{\ell} = 0.4203 \pm 0.0001$20, a counterimage B roughly $z_{\ell} = 0.4203 \pm 0.0001$21 fainter and $z_{\ell} = 0.4203 \pm 0.0001$22 from the lens on the opposite side, and a possible fourth image C (Lemon et al., 9 Apr 2026). Its luminosity and geometry imply extreme magnification, with $z_{\ell} = 0.4203 \pm 0.0001$23 suggested by comparison to SN 2023ixf and $z_{\ell} = 0.4203 \pm 0.0001$24 required by the lens configuration. The later spectral phase of B indicates that the faint counterimage arrived first, rendering classical light-curve delay measurement impractical but opening the prospect of cosmography through spectral phase retrieval. As presented, the system serves simultaneously as a highly magnified view of a high-redshift core-collapse supernova and as a promising laboratory for future time-delay measurements.