Sunrise Galaxy: Lensed Arc & Young Clusters

• Sunrise Galaxy is a strongly lensed system at z≈5.93, displaying a 17″ tangential arc that resolves into compact, parsec-scale star-forming knots.
• Advanced lens-modeling and JWST spectroscopy provide precise redshift determination and characterization of young massive star clusters’ sizes, masses, and ages.
• The study reinterprets Earendel as a proto-globular cluster candidate, offering key insights into star formation and cluster evolution during the reionization era.

Searching arXiv for the specified Sunrise Galaxy papers and closely related work. arxiv_search(query="(Pascale et al., 7 Jul 2025) OR \"Is Earendel a Star Cluster?\" OR \"Sunrise Arc\" OR (Vanzella et al., 2022)", max_results=10) Searching for exact titles individually to ensure precise citation targets. arxiv_search(query="ti:\"Is Earendel a Star Cluster?: Metal Poor Globular Cluster Progenitors at z~6\"", max_results=5) arxiv_search(query="ti:\"JWST/NIRCam Probes Young Star Clusters in the Reionization Era Sunrise Arc\"", max_results=5) The Sunrise galaxy, commonly termed the Sunrise arc, is a strongly lensed galaxy at $z \approx 5.93$ observed behind the massive foreground cluster WHL0137 at $z_{\rm cluster}=0.566$. In imaging it appears as a highly stretched, $\sim 17''$-long tangential arc with four multiple images (“a, b, c, d”), while source-plane reconstructions resolve it into compact knots down to $\sim 1$ pc scales. JWST/NIRCam imaging and JWST/NIRSpec PRISM spectroscopy have made Sunrise a reference system for parsec-scale star formation in the first Gyr of cosmic history, for the study of young massive star clusters (YMCs), and for the reassessment of the highly magnified feature Earendel, previously identified as a candidate star or binary but now also analyzed as a possible star cluster (Vanzella et al., 2022, Pascale et al., 7 Jul 2025).

1. Lensed morphology and source-plane geometry

Sunrise is lensed by WHL0137 and reconstructed with lens-modeling frameworks including WSLAP+, Lenstool, and GLAFIC. The full arc spans $\sim 17''$ in the image plane, corresponding to $\sim 95$ kpc tangentially, and to $\sim 1.6$ kpc length in the source plane. The principal star-forming complex (SFC) has size $\sim 200$ pc, and the entire analyzed host region (“TOTd”) extends $\sim 1000$ pc (Vanzella et al., 2022).

The global magnification is reported as $\mu_{\rm tot} \simeq 300$, while individual “b” knots have $z_{\rm cluster}=0.566$0 and knot 1b has $z_{\rm cluster}=0.566$1. Lens-model reconstructions place critical curves through the arc and yield tangential stretches $z_{\rm cluster}=0.566$2–$z_{\rm cluster}=0.566$3, with $z_{\rm cluster}=0.566$4 for the “b” knots. This geometry resolves the galaxy into compact components on source-plane scales that are otherwise inaccessible at $z_{\rm cluster}=0.566$5 (Vanzella et al., 2022).

For Earendel, the magnification is model-dependent. Macro models discussed by Welch et al. (2022b) predict tangential magnifications of order $z_{\rm cluster}=0.566$6–$z_{\rm cluster}=0.566$7, implying an intrinsic FUV half-light radius $z_{\rm cluster}=0.566$8 pc. Ji & Dai (2024) showed that including dark subhalos can relax the limit to $z_{\rm cluster}=0.566$9 pc without violating observed flux ratios or astrometric constraints. Scofield et al. (2025) instead find $\sim 17''$0–$\sim 17''$1 from a joint strong+weak lens model, still sufficient for Earendel to remain unresolved in the F090W, F115W, and F200W NIRCam images, whose native pixel scale is $\sim 17''$2 pc at $\sim 17''$3 (Pascale et al., 7 Jul 2025).

2. Redshift determination and spectroscopic basis

Deep archival JWST/NIRSpec PRISM spectroscopy established a spectroscopic redshift for Sunrise by extracting the NIRSpec MSA slit 2282_12001, which covers the SFC, and simultaneously fitting Gaussian profiles to the nebular lines [O II] $\sim 17''$4, [Ne III] $\sim 17''$5, H$\sim 17''$6 $\sim 17''$7, [O III] $\sim 17''$8, and H$\sim 17''$9 $\sim 1$0. The formal fit yields

$\sim 1$1

in excellent agreement with the Lyman-continuum break seen in the continua of Earendel and 1b (Pascale et al., 7 Jul 2025).

This spectroscopic redshift supersedes earlier photometric estimates of $\sim 1$2 from Welch et al. (2022b). The same analysis places Sunrise at a lookback time of $\sim 1$3 Gyr. The earlier JWST/NIRCam study had already inferred $\sim 1$4 at 95% confidence from SED fitting, consistent with HST estimates; the PRISM result converts that approximate placement into a line-anchored systemic redshift suitable for continuum modeling and lensing interpretation (Vanzella et al., 2022, Pascale et al., 7 Jul 2025).

The spectroscopic confirmation is methodologically important because the stellar-population analysis of Earendel and 1b uses observed-frame $\sim 1$5–$\sim 1$6m spectroscopy, corresponding to $\sim 1$7–$\sim 1$8 Å in the rest frame. At this redshift, the Balmer-break region and the rest-UV continuum both fall within the PRISM coverage, which is central to the subsequent age–metallicity analysis (Pascale et al., 7 Jul 2025).

3. Young massive star clusters in the Sunrise arc

JWST/NIRCam imaging identified six young massive star clusters in Sunrise with measured radii spanning $\sim 1$9 pc down to $\sim 17''$0 pc, estimated stellar masses of $\sim 17''$1, and ages $\sim 17''$2–$\sim 17''$3 Myr based on SED fitting to photometry measured in 8 filters extending to rest-frame $\sim 17''$4 Å (Vanzella et al., 2022).

The size-measurement procedure uses F150W for UV-continuum knots and F277W for nebular-line knots. Each candidate is modeled as a 2D Gaussian with $\sim 17''$5, ellipticity $\sim 17''$6, and position angle $\sim 17''$7, convolved with the filter PSF and fit in a $\sim 17''$8 pixel box. The Gaussian is extrapolated to infinite radius to obtain total flux, and the intrinsic source-plane half-light radius is computed as

$\sim 17''$9

with $\sim 95$0 converted from image-plane $\sim 95$1 by dividing angular size by $\sim 95$2 and multiplying by $\sim 95$3 kpc/$\sim 95$4 at $\sim 95$5 (Vanzella et al., 2022).

The six clusters span the following measured values: 1b has $\sim 95$6 pc, $\sim 95$7, and age $\sim 95$8 Myr; 2b has $\sim 95$9 pc, $\sim 1.6$0, and age $\sim 1.6$1 Myr; 3b has $\sim 1.6$2 pc, $\sim 1.6$3, and age $\sim 1.6$4 Myr; 4b has $\sim 1.6$5 pc, $\sim 1.6$6, and age $\sim 1.6$7 Myr; 5b has $\sim 1.6$8 pc, $\sim 1.6$9, and age $\sim 200$0 Myr; 6b has $\sim 200$1 pc, $\sim 200$2, and age $\sim 200$3 Myr (Vanzella et al., 2022).

The stellar surface density is defined as

$\sim 200$4

and exceeds $\sim 200$5 for the sample, reaching up to a few $\sim 200$6. For 1b specifically, the tabulated value is $\sim 200$7 (Vanzella et al., 2022).

A dynamical classification was carried out through the crossing time and the dynamical age,

$\sim 200$8

with $\sim 200$9. The resulting values indicate that five of the six candidates, all except 4b, satisfy $\sim 1000$0 and therefore qualify as gravitationally bound; for 1b, $\sim 1000$1 (Vanzella et al., 2022).

4. Continuum fitting of Earendel and 1b

The 2025 spectroscopic analysis tested whether Earendel could be explained by a compact stellar population rather than an individual star or binary. Over the observed-frame range $\sim 1000$2–$\sim 1000$3m, the rest-UV through optical continua of Earendel and 1b were fit with instantaneous-burst simple stellar population (SSP) models from three libraries: BPASS v2.3 with binaries and a broken power-law IMF ($\sim 1000$4 below $\sim 1000$5, $\sim 1000$6 above), BC03 (2016) with a Kroupa IMF ($\sim 1000$7–$\sim 1000$8), and FSPS/MIST with MILES empirical stellar spectra and a Kroupa IMF ($\sim 1000$9–$\mu_{\rm tot} \simeq 300$0) (Pascale et al., 7 Jul 2025).

The spectral-energy distributions were generated on a fine grid of $\mu_{\rm tot} \simeq 300$1 from $\mu_{\rm tot} \simeq 300$2 Myr to $\mu_{\rm tot} \simeq 300$3 Gyr and $\mu_{\rm tot} \simeq 300$4, guided by the native grid points of each library and linearly interpolated in log–log space. A nebular component computed with CLOUDY 17 tied $\mu_{\rm tot} \simeq 300$5 and allowed $\mu_{\rm tot} \simeq 300$6 and covering factor $\mu_{\rm tot} \simeq 300$7 to vary. Dust attenuation followed the Salim et al. (2018) prescription with slope $\mu_{\rm tot} \simeq 300$8 and 2175 Å bump strength $\mu_{\rm tot} \simeq 300$9. The fit adopted a Gaussian prior on $z_{\rm cluster}=0.566$00 centered at $z_{\rm cluster}=0.566$01, convolved each model to the NIRSpec PRISM resolution, and minimized

$z_{\rm cluster}=0.566$02

including an extra free white-noise term $z_{\rm cluster}=0.566$03 to scale the pipeline uncertainties upward as needed. Parameter inference used Nautilus nested sampling with 3,000 live points across a 10–12-parameter space (Pascale et al., 7 Jul 2025).

All three SSP libraries produce formally acceptable fits, with reduced $z_{\rm cluster}=0.566$04 before noise scaling, and the BC03 and FSPS results agree with BPASS within $z_{\rm cluster}=0.566$05–$z_{\rm cluster}=0.566$06. The paper reports that the continuum of Earendel is well described by an SSP nearly equivalently to 1b, which is confidently a star cluster (Pascale et al., 7 Jul 2025).

Property	Earendel	1b
Age	$z_{\rm cluster}=0.566$07	$z_{\rm cluster}=0.566$08
Interpreted age	$z_{\rm cluster}=0.566$09 Myr	$z_{\rm cluster}=0.566$10 Myr
Metallicity	$z_{\rm cluster}=0.566$11	$z_{\rm cluster}=0.566$12
Metallicity limit	$z_{\rm cluster}=0.566$13 (95% C.L.)	$z_{\rm cluster}=0.566$14 (99% C.L.)
Stellar mass	$z_{\rm cluster}=0.566$15	$z_{\rm cluster}=0.566$16
Interpreted lensed mass	$z_{\rm cluster}=0.566$17	$z_{\rm cluster}=0.566$18
Dust	$z_{\rm cluster}=0.566$19 mag, $z_{\rm cluster}=0.566$20, $z_{\rm cluster}=0.566$21	$z_{\rm cluster}=0.566$22 mag

A central result is that the pronounced Balmer break at $z_{\rm cluster}=0.566$23 Å rest, together with the UV and optical continuum slopes, provides leverage to disentangle age, metallicity, and $z_{\rm cluster}=0.566$24 even without strong absorption features at PRISM resolution. In the authors’ formulation, this is what enables spectroscopic characterization of intermediate-age clusters that are seldom probed at high redshift (Pascale et al., 7 Jul 2025).

5. Earendel, 1b, and the proto-globular-cluster interpretation

The status of Earendel is the main interpretive controversy associated with Sunrise. It was previously identified as a candidate star or binary because extreme lensing magnification appeared to require a sub-parsec source size. The subsequent relaxation of the size constraint to $z_{\rm cluster}=0.566$25 pc opened the possibility that Earendel could instead be a compact stellar cluster. The 2025 analysis explicitly explored that hypothesis and found that Earendel’s continuum is compatible with an SSP solution similar to that of 1b (Pascale et al., 7 Jul 2025).

The neighboring knot 1b provides an anchor for that interpretation. In the NIRCam study, 1b was already one of the most compact and dense YMCs, with $z_{\rm cluster}=0.566$26 pc, $z_{\rm cluster}=0.566$27, age $z_{\rm cluster}=0.566$28 Myr, and $z_{\rm cluster}=0.566$29, which qualified it as gravitationally bound. In the later spectroscopic analysis, adopted magnifications $z_{\rm cluster}=0.566$30–$z_{\rm cluster}=0.566$31 together with photometrically inferred sizes $z_{\rm cluster}=0.566$32–$z_{\rm cluster}=0.566$33 pc imply stellar surface densities $z_{\rm cluster}=0.566$34 (Vanzella et al., 2022, Pascale et al., 7 Jul 2025).

Both Earendel and 1b fall in the intermediate-age ($z_{\rm cluster}=0.566$35–$z_{\rm cluster}=0.566$36 Myr), metal-poor ($z_{\rm cluster}=0.566$37) regime and are reported to be consistent with the formation age–metallicity trend seen in local globular clusters. In the comparison presented in Fig. 4 of the 2025 study, their positions on the age–metallicity plane align most closely with the simulated relation for SMC/LMC-mass halos in the E-MOSAICS cosmological runs, although the total stellar mass of Sunrise, $z_{\rm cluster}=0.566$38–$z_{\rm cluster}=0.566$39, also leaves a Milky Way-like host as a plausible alternative. The same paper states that their inferred $z_{\rm cluster}=0.566$40 and $z_{\rm cluster}=0.566$41 align with in-situ or ex-situ formation epochs envisaged for metal-poor GCs (Pascale et al., 7 Jul 2025).

The earlier NIRCam analysis had already advanced a related argument for 1b and 2b: with $z_{\rm cluster}=0.566$42, $z_{\rm cluster}=0.566$43 pc, $z_{\rm cluster}=0.566$44, $z_{\rm cluster}=0.566$45, ages $z_{\rm cluster}=0.566$46–$z_{\rm cluster}=0.566$47 Myr, and inferred $z_{\rm cluster}=0.566$48, they were described as direct analogues of proto-globular clusters. Even with $z_{\rm cluster}=0.566$49 mass loss over a Hubble time, they would retain $z_{\rm cluster}=0.566$50, comparable to present-day metal-poor GCs (Vanzella et al., 2022).

6. Reionization-era star formation and observational significance

The Sunrise arc is also used to investigate the relation between clustered star formation and ionizing output during the reionization era. The ages of the six YMCs map a progression of star formation along the arc, with evolved systems older than $z_{\rm cluster}=0.566$51 Myr followed by very young clusters. The youngest stellar clusters, younger than $z_{\rm cluster}=0.566$52 Myr, show photometrically inferred rest-frame equivalent widths $z_{\rm cluster}=0.566$53 Å, and they are hosted in a $z_{\rm cluster}=0.566$54 pc-sized SFC (Vanzella et al., 2022).

For the SFC region, the reported photometric excesses correspond to $z_{\rm cluster}=0.566$55 Å in F356W and $z_{\rm cluster}=0.566$56 Å in F444W. The ionizing photon production efficiency is defined as

$z_{\rm cluster}=0.566$57

and the inferred value is $z_{\rm cluster}=0.566$58, assuming at least 50% conversion to nebular lines. The SFC is described as dominating the ionizing photon production (Vanzella et al., 2022).

The integrated star-formation accounting is likewise cluster-centric. SED fits with Prospector non-parametric and BAGPIPES/BPASS constant-SFH models yield a host stellar mass $z_{\rm cluster}=0.566$59–$z_{\rm cluster}=0.566$60 formed over $z_{\rm cluster}=0.566$61 Myr. The six YMCs together contain $z_{\rm cluster}=0.566$62 formed over $z_{\rm cluster}=0.566$63 Myr, implying $z_{\rm cluster}=0.566$64, while the host has $z_{\rm cluster}=0.566$65–$z_{\rm cluster}=0.566$66. The resulting cluster formation efficiency is

$z_{\rm cluster}=0.566$67

The 2022 study speculated that YMC-driven feedback may carve low-density channels and that progressively older clusters could facilitate Lyman-continuum escape during bursty episodes of star formation (Vanzella et al., 2022).

The 2025 work extends the significance of Sunrise from photometric identification of very young clusters to continuum-only spectroscopy of intermediate-age systems. It argues that the ability to measure ages and metallicities of $z_{\rm cluster}=0.566$68–$z_{\rm cluster}=0.566$69 star clusters at $z_{\rm cluster}=0.566$70 from continuum-only spectroscopy constitutes a critical new frontier. At NIRSpec PRISM resolution and for magnitudes $z_{\rm cluster}=0.566$71–$z_{\rm cluster}=0.566$72 AB, broad continuum features can break the classical age–dust–metallicity degeneracies when the signal-to-noise per resolution element is $z_{\rm cluster}=0.566$73. Strong-lensing caustics with $z_{\rm cluster}=0.566$74 boosted by factors of tens to hundreds are therefore essential, and the same paper anticipates that similar continuum-fitting techniques will become applicable to statistical samples of proto-globular clusters as lensed arcs accumulate in the JWST archives (Pascale et al., 7 Jul 2025).