AURORA Survey: Ultradeep JWST/NIRSpec Study

• AURORA Survey is a JWST/NIRSpec Cycle 1 program that delivers continuous 1–5 µm spectroscopy of high-redshift galaxies to detect faint auroral lines for direct electron-temperature-based abundance measurements.
• It combines deep exposures, medium spectral resolution (R∼1000), and wide wavelength coverage to robustly analyze nebular excitation, dust attenuation, density structures, and galactic outflows.
• The program establishes a comprehensive framework for direct-metallicity, ionizing-photon production, and multi-phase ISM diagnostics, providing key insights into early galaxy evolution from cosmic noon to reionization.

The AURORA Survey—Assembly of Ultradeep Rest-optical Observations Revealing Astrophysics—is a JWST/NIRSpec Cycle 1 multi-object spectroscopic program designed to obtain ultradeep, continuous $1$–$5\,\mu\mathrm{m}$ spectroscopy of high-redshift galaxies, primarily to detect faint auroral lines and thereby enable direct, electron-temperature-based abundance measurements. In practice, its combination of depth, continuous wavelength coverage, and medium spectral resolution $R\sim1000$ has made it a broader survey of nebular excitation, attenuation, density structure, outflows, ionizing-photon production, and rare massive systems in the early universe, with secure spectroscopic redshifts spanning $z=1.4$–10.4 (Shapley et al., 2024, Topping et al., 12 Feb 2025).

1. Program definition and observational architecture

AURORA is a JWST/NIRSpec MSA program with Program ID 1914, described as a Cycle 1 survey and identified in the survey overview as having Co-PIs Shapley and Sanders (Shapley et al., 2024). The survey was built around two deep pointings in GOODS-N and COSMOS, using the three medium-resolution settings G140M/F100LP, G235M/F170LP, and G395M/F290LP to deliver continuous rest-optical and near-infrared spectroscopy across a wide redshift baseline (Pahl et al., 14 Oct 2025).

Its instrumental design is central to its scientific scope. The survey was explicitly optimized for faint auroral features such as $[\mathrm{OIII}]\,\lambda4363$, $[\mathrm{OII}]\,\lambda\lambda7320,7330$, $[\mathrm{SIII}]\,\lambda6312$, $[\mathrm{NII}]\,\lambda5755$, and $[\mathrm{SII}]\,\lambda\lambda4068,4076$, but the same data also provide strong leverage on Balmer and Paschen recombination lines, classical strong-line diagnostics, rest-frame near-IR line ratios, and near-UV absorption features (Shapley et al., 2024).

Property	Value	Source
Program type	JWST/NIRSpec Cycle 1 MSA, PID 1914	(Shapley et al., 2024)
Fields	GOODS-N and COSMOS	(Pahl et al., 14 Oct 2025)
Spectral setup	G140M/F100LP, G235M/F170LP, G395M/F290LP	(Pahl et al., 14 Oct 2025)
Coverage and resolution	Continuous $1$–$5\,\mu\mathrm{m}$0, $5\,\mu\mathrm{m}$1	(Shapley et al., 2024)
Total targeted galaxies	97	(Pahl et al., 14 Oct 2025)
Secure spectroscopic redshifts	95/97	(Topping et al., 12 Feb 2025)

Later AURORA papers report exposure times of 12.3 hr, 8.0 hr, and 4.2 hr in the three NIRSpec settings, with a typical $5\,\mu\mathrm{m}$2 line-flux limit of about $5\,\mu\mathrm{m}$3 (Pahl et al., 14 Oct 2025). A separate survey analysis emphasizes that the long integrations yield a typical sensitivity of about $5\,\mu\mathrm{m}$4 at roughly $5\,\mu\mathrm{m}$5 (Shapley et al., 2024). Taken together, these descriptions show that AURORA was engineered not merely for line detection, but for line-ratio precision in individual galaxies.

2. Sample construction and direct-method framework

The survey targeted 97 galaxies at $5\,\mu\mathrm{m}$6, with a tiered selection strategy. The primary targets were star-forming galaxies at $5\,\mu\mathrm{m}$7–4.4 chosen for expected detection of faint auroral lines, while additional slits were assigned to very high-redshift galaxies, quiescent galaxies at $5\,\mu\mathrm{m}$8, strong-line emitters, and photometric-$5\,\mu\mathrm{m}$9 sources (Topping et al., 12 Feb 2025). In the metallicity-calibration analysis, 89 AURORA targets are treated as star-formation-dominated, and 41 of those show at least one auroral line at $R\sim1000$0 (Sanders et al., 13 Aug 2025).

The direct-abundance methodology relies on temperature-sensitive auroral-to-nebular ratios. The AURORA metallicity framework uses $R\sim1000$1 for the high-ionization zone, $R\sim1000$2 for the low-ionization zone, $R\sim1000$3 for intermediate ionization, and $R\sim1000$4 as an additional low-ionization tracer. The analysis uses PyNeb to infer $R\sim1000$5 and $R\sim1000$6, adopts the relation

$R\sim1000$7

when only one oxygen temperature is directly measured, and computes total oxygen abundance through

$R\sim1000$8

This is the survey’s core “direct” or $R\sim1000$9 method (Sanders et al., 13 Aug 2025).

AURORA also made unusually strong use of hydrogen recombination lines. In the direct-$z=1.4$0 mass–metallicity analysis, electron density, electron temperature, and nebular reddening are solved iteratively, with dust correction based on all Balmer and Paschen lines detected at $z=1.4$1 (Khostovan et al., 18 Dec 2025). In later dust-focused work, this multi-line strategy becomes a survey-defining capability rather than a secondary correction.

3. Emission-line physics and the high-redshift metallicity framework

AURORA’s first major survey synthesis used emission-line measurements for 95 out of 97 targeted galaxies, focusing on 87 star-forming galaxies after excluding AGN and quiescent systems (Shapley et al., 2024). Across the classical BPT planes, the $z=1.4$2–$z=1.4$3 plane, the bluer $z=1.4$4 versus $z=1.4$5 diagram, and new rest-frame near-IR diagnostics, the survey found a coherent pattern: distant star-forming galaxies are chemically young, $z=1.4$6-enhanced, and photoionized by harder stellar ionizing spectra at fixed nebular metallicity than their $z=1.4$7 counterparts (Shapley et al., 2024).

The same analysis reported previously unseen evolution in the $z=1.4$8 versus $z=1.4$9 diagram prior to $[\mathrm{OIII}]\,\lambda4363$0: the $[\mathrm{OIII}]\,\lambda4363$1–4.0 sample is offset by roughly $[\mathrm{OIII}]\,\lambda4363$2 dex higher [OIII]/H$[\mathrm{OIII}]\,\lambda4363$3 at fixed [NII]/H$[\mathrm{OIII}]\,\lambda4363$4 relative to the $[\mathrm{OIII}]\,\lambda4363$5–2.7 sample (Shapley et al., 2024). The survey also produced the first statistical sample of rest-frame near-IR emission-line diagnostics at high redshift, including 55 galaxies with $[\mathrm{OIII}]\,\lambda4363$6 detections and 23 star-forming galaxies with the full line set needed for the He I, [Fe II], and Paschen-based diagrams (Shapley et al., 2024).

AURORA’s direct-$[\mathrm{OIII}]\,\lambda4363$7 abundance work then expanded this diagnostic program into an empirical calibration framework. One survey paper combined 41 AURORA auroral-line galaxies with 98 literature objects to form a 139-galaxy direct-method sample at $[\mathrm{OIII}]\,\lambda4363$8–10.6, covering $[\mathrm{OIII}]\,\lambda4363$9–8.6, or about $[\mathrm{OII}]\,\lambda\lambda7320,7330$0–$[\mathrm{OII}]\,\lambda\lambda7320,7330$1 (Sanders et al., 13 Aug 2025). It calibrated 19 emission-line ratios against oxygen abundance and found that calibrations based on $[\mathrm{OII}]\,\lambda\lambda7320,7330$2-element lines—O, Ne, S, and Ar—are broadly reliable, while N-based calibrations are substantially less reliable because of the large dispersion in N/O at fixed O/H. The same paper emphasizes that applying typical $[\mathrm{OII}]\,\lambda\lambda7320,7330$3 calibrations to high-redshift galaxies can bias metallicities by more than 0.1 dex in O/H (Sanders et al., 13 Aug 2025).

The survey also extended direct-$[\mathrm{OII}]\,\lambda\lambda7320,7330$4 metallicity work into galaxy scaling relations. Using 34 galaxies at $[\mathrm{OII}]\,\lambda\lambda7320,7330$5 with auroral oxygen-line detections, AURORA measured a $[\mathrm{OII}]\,\lambda\lambda7320,7330$6 mass–metallicity relation with slope $[\mathrm{OII}]\,\lambda\lambda7320,7330$7, normalization $[\mathrm{OII}]\,\lambda\lambda7320,7330$8 at $[\mathrm{OII}]\,\lambda\lambda7320,7330$9, and intrinsic scatter $[\mathrm{SIII}]\,\lambda6312$0 dex (Khostovan et al., 18 Dec 2025). The same analysis found that the sample is consistent with the $[\mathrm{SIII}]\,\lambda6312$1 fundamental metallicity relation within 0.1 dex in O/H, while also concluding that none of six simulations—EAGLE, SIMBA, Illustris, IllustrisTNG, FIRE, and NewHorizon—reproduce the observed normalization evolution of the MZR from $[\mathrm{SIII}]\,\lambda6312$2 to $[\mathrm{SIII}]\,\lambda6312$3 (Khostovan et al., 18 Dec 2025).

4. Dust attenuation, ionizing output, and enrichment tracers

One of AURORA’s most distinctive contributions is its treatment of nebular attenuation. In the $[\mathrm{SIII}]\,\lambda6312$4 galaxy GOODSN-17940, the survey used 11 unblended H I recombination lines to derive a nebular attenuation curve spanning $[\mathrm{SIII}]\,\lambda6312$5–$[\mathrm{SIII}]\,\lambda6312$6, then extended it with rest-UV spectroscopy and photometry to a combined $[\mathrm{SIII}]\,\lambda6312$7–$[\mathrm{SIII}]\,\lambda6312$8 curve (Sanders et al., 2024). The resulting curve is steeper than the Milky Way, SMC, and Calzetti curves at long wavelengths, has a similar slope in the blue optical, is shallower than the SMC and Calzetti curves in the ultraviolet, and shows no significant $[\mathrm{SIII}]\,\lambda6312$9 bump (Sanders et al., 2024). The survey paper presents this as direct evidence that commonly assumed dust curves are not appropriate for all high-redshift galaxies.

This object-specific work scales into a survey-level attenuation program in the $[\mathrm{NII}]\,\lambda5755$0 study. There, AURORA analyzes 63 star-forming galaxies at $[\mathrm{NII}]\,\lambda5755$1–6.9, of which 23 have individual nebular dust attenuation curves and 40 use the survey-average AURORA nebular curve (Pahl et al., 14 Oct 2025). The average curve has $[\mathrm{NII}]\,\lambda5755$2, compared to 3.1 for the Galactic curve, 2.74 for the SMC, and 4.05 for Calzetti (Pahl et al., 14 Oct 2025). Using these nebular curves, the survey defines

$[\mathrm{NII}]\,\lambda5755$3

finds a median

$[\mathrm{NII}]\,\lambda5755$4

and reports

$[\mathrm{NII}]\,\lambda5755$5

for the $[\mathrm{NII}]\,\lambda5755$6 subset (Pahl et al., 14 Oct 2025). Positive correlations are found with redshift, $[\mathrm{NII}]\,\lambda5755$7 equivalent width, and O32, while negative correlations are found with stellar attenuation, UV luminosity, stellar mass, and direct-method metallicity (Pahl et al., 14 Oct 2025). The same paper shows that adopting a Galactic nebular curve or assuming $[\mathrm{NII}]\,\lambda5755$8 yields systematically lower $[\mathrm{NII}]\,\lambda5755$9 and can flatten the $[\mathrm{SII}]\,\lambda\lambda4068,4076$0–$[\mathrm{SII}]\,\lambda\lambda4068,4076$1 relation (Pahl et al., 14 Oct 2025).

AURORA has also used less conventional elements to probe enrichment pathways. In a sample of 46 star-forming galaxies at $[\mathrm{SII}]\,\lambda\lambda4068,4076$2–3.5, split into stacks at $[\mathrm{SII}]\,\lambda\lambda4068,4076$3 and $[\mathrm{SII}]\,\lambda\lambda4068,4076$4, the survey measured

$[\mathrm{SII}]\,\lambda\lambda4068,4076$5

respectively, and argued that both measurements are $[\mathrm{SII}]\,\lambda\lambda4068,4076$6 below solar, indicating enrichment dominated by core-collapse supernovae with minimal Type Ia supernova contribution (Foley et al., 10 Dec 2025). Comparison with Galactic chemical-evolution models was found to be more consistent with the Milky Way Bulge than the Solar Neighborhood, implying a rapid star-formation timescale (Foley et al., 10 Dec 2025).

The helium-abundance program extends AURORA’s chemical ambitions still further. Using 20 galaxies at $[\mathrm{SII}]\,\lambda\lambda4068,4076$7 with multiple $[\mathrm{SII}]\,\lambda\lambda4068,4076$8 He I detections, including the critical He I $[\mathrm{SII}]\,\lambda\lambda4068,4076$9 line, the survey produced the first robust high-redshift helium abundances in normal star-forming galaxies (Berg et al., 22 Jul 2025). Most objects follow the extrapolated local He/H–O/H trend, but four galaxies show elevated helium mass fractions with $1$0 and no comparable enhancement in N/O or the $1$1-elements (Berg et al., 22 Jul 2025). The paper argues that this pattern is inconsistent with asymptotic giant branch enrichment and instead favors early helium enrichment from very massive stars with $1$2 (Berg et al., 22 Jul 2025).

5. Multi-phase ISM structure and galactic outflows

AURORA’s density measurements provide a systematic view of H II region structure across redshift. Using 51 galaxies with density-sensitive [S II] measurements and 8 with resolvable C III] doublets, the survey inferred median low-ionization electron densities of

$1$3

at $1$4, $1$5, and $1$6, following an evolutionary scaling of $1$7 (Topping et al., 12 Feb 2025). High-ionization gas traced by C III] yields a median density of

$1$8

about $1$9 times higher than the [S II]-based densities (Topping et al., 12 Feb 2025). The survey interprets this as evidence for a persistent multi-phase H II region structure in which dense, high-ionization interiors are surrounded by less dense, low-ionization gas.

The density analysis also reports weak positive correlations with SFR and SFR surface density, a significant correlation with Ne3O2, and a stronger correlation with distance from the local BPT sequence than can be reproduced by simple photoionization models (Topping et al., 12 Feb 2025). A comparison with the SPHINX simulations is used to argue that density is shaped jointly by residual molecular-cloud pressure, stellar age, metallicity, and feedback (Topping et al., 12 Feb 2025). A plausible implication is that AURORA’s deep rest-optical spectra are not only measuring integrated galaxy properties, but also constraining the internal stratification of their ionized gas.

The outflow program uses the same spectra in a different regime. In 41 and 43 galaxies at $5\,\mu\mathrm{m}$00, AURORA measures ISM kinematics from Fe II and Mg II absorption, respectively (Kehoe et al., 20 Jun 2025). The mean centroid velocities are

$5\,\mu\mathrm{m}$01

and

$5\,\mu\mathrm{m}$02

indicating outflows on average (Kehoe et al., 20 Jun 2025). Galaxies with outflow detections tend to have higher stellar masses, while composite spectra show that maximum outflow velocity increases with stellar mass, SFR, $5\,\mu\mathrm{m}$03, $5\,\mu\mathrm{m}$04, and $5\,\mu\mathrm{m}$05 (Kehoe et al., 20 Jun 2025). The paper also identifies 5 Mg II emitters, more common in lower-mass, higher-sSFR, and less dusty systems, and 10 Na D absorbers associated with higher stellar mass, SFR, and dust attenuation (Kehoe et al., 20 Jun 2025). These trends are described as consistent with lower-redshift work using the same tracers.

6. Rare systems, boundary cases, and survey significance

Although AURORA was designed primarily around auroral-line targets at cosmic noon, its depth and spectral coverage also make it sensitive to rare systems at much higher redshift. A notable example is GOODSN-100182, observed serendipitously within the program and confirmed at $5\,\mu\mathrm{m}$06 (Shapley et al., 2024). This galaxy has

$5\,\mu\mathrm{m}$07

a red UV slope of $5\,\mu\mathrm{m}$08, nebular reddening

$5\,\mu\mathrm{m}$09

and a dust-corrected

$5\,\mu\mathrm{m}$10

(Shapley et al., 2024). NIRCam imaging shows an extended disk with effective radius $5\,\mu\mathrm{m}$11 kpc in a multi-component fit or $5\,\mu\mathrm{m}$12 kpc in a single Sérsic fit, and the NIRSpec spectrum reveals a rich rest-optical line set extending from [O II] to [N II] (Shapley et al., 2024). The line ratios imply roughly $5\,\mu\mathrm{m}$13 solar metallicity from N2 and O3N2, or about $5\,\mu\mathrm{m}$14 solar from $5\,\mu\mathrm{m}$15, and the system resides in a $5\,\mu\mathrm{m}$16 overdensity with a spectroscopically confirmed companion (Shapley et al., 2024). The paper presents it as a mature, dusty, chemically enriched disk-like galaxy within the first billion years of cosmic time.

The broader significance of AURORA lies in the way these results interlock. Survey papers repeatedly show that high-redshift galaxies are not well described by a simple transplantation of local empirical tools: ordinary $5\,\mu\mathrm{m}$17 metallicity calibrations can be biased, a universal dust law is not supported, and one-zone gas models are incomplete (Sanders et al., 13 Aug 2025, Sanders et al., 2024, Topping et al., 12 Feb 2025). At the same time, AURORA’s direct-$5\,\mu\mathrm{m}$18 MZR and FMR results show that some large-scale baryon-cycle regularities were already in place at cosmic noon, even though current simulations fail to reproduce the observed normalization evolution of the MZR from $5\,\mu\mathrm{m}$19 to $5\,\mu\mathrm{m}$20 (Khostovan et al., 18 Dec 2025).

A common misconception is to treat AURORA as a narrowly defined auroral-line program. The published record shows a more expansive role. It is indeed a direct-metallicity survey by design, but its ultradeep, continuous NIRSpec spectroscopy has also turned it into a laboratory for attenuation curves, ionizing-photon production efficiencies, multi-phase densities, rest-frame near-IR diagnostics, chemical clocks based on Ar/O and He/H, and rare high-redshift systems whose properties are more reminiscent of $5\,\mu\mathrm{m}$21–3 galaxies than of the canonical young, blue, low-mass population expected at $5\,\mu\mathrm{m}$22 (Shapley et al., 2024, Shapley et al., 2024). In that sense, AURORA functions both as a targeted program and as a quantitative reference dataset for early galaxy evolution from cosmic noon into the epoch of reionization.