An Optical Illusion: High Electron Densities Create Extremely Metal-Poor Galaxy Impostors

Abstract: JWST has enabled the discovery of dozens of extremely metal-poor galaxies (EMPGs) with metallicities below $5\%\,Z_{\odot}$, representing a significant leap toward detecting the first galaxies without metals. However, accurate metallicity measurements require careful determination of physical conditions in the ionized gas. In this paper, we study four galaxies that appear to be EMPGs when analyzed using the direct $T_e$ method with the common low-density assumption ($n_e=10^3\,{\rm cm}<sup>{-3}$). To test whether their metal-poor status holds when accurately measuring the density, we apply the direct method with a self-consistent determination of electron temperature ($T_e$) and density ($n_e$) in the high-ionization zone using the [OIII]$λ$5008, [OIII]$λ$4364, and [OIII]$λ$1666 lines, using JWST/NIRSpec data from the SPURS program in the Abell 2744 lensed field. We find that three out of four galaxies in our sample have extremely high electron densities ($$n_e \sim 10^{{5}-10^{6}\,{\rm} cm<sup>{-3}}$$), which suppress the [OIII]$λ$5008 line and lead to underestimates of metallicity by up to $\sim1.1\,$dex when this density is not accounted for, and have true metallicities of 12+log(O/H) $\sim7.3-8.2$. Failure to account for high densities can therefore lead to systematic misclassification of metal-poor galaxies and biased conclusions about early chemical enrichment.

• The paper presents a method using three different O$^{++}$ emission line transitions to accurately determine electron densities ($n_e$) and temperatures ($T_e$).
• High ionization zones ($n_e$ up to $10^{6} cm^{-3}$) in metal-rich galaxies can mimic temperature signatures of extreme metal-poor galaxies under the traditional $n_e = 10^{3}cm^{-3}$ assumption, making up half to three-quarters of the galaxy sample.
• Contrast with $n_e$ in $C^{+3}$ indicates strong density gradients in early galaxies, meaning only self-consistent direct methods can correctly assign metallicity from gas composition.

Motivation and method

JWST has enabled the identification of dozens of extremely metal-poor galaxy (EMPG) candidates at $z\gtrsim5$, with metallicities below $5\%\,Z_{\odot}$, but most classifications rest on strong-line diagnostics such as the R3 ratio ($\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$), and the few direct-method measurements typically assume a low, uniform electron density ($n_e=10^3\,\mathrm{cm^{-3}}$). Hsiao et al. (2608.20339) test whether this assumption survives scrutiny in the high-ionization gas where the auroral-based temperature is actually measured. The key physical point is that $\lambda5008$ becomes collisionally de-excited above $n_e\sim10^5\,\mathrm{cm^{-3}}$, so the traditional $[\mathrm{O\,III}]\lambda5008/\lambda4364$ ratio no longer traces $T_e$ alone: a given ratio can be produced by hot, diffuse gas or by cooler, denser gas. The authors adopt the self-consistent direct method of Berg (2025) and Arellano-Córdova et al. (2026), which combines three O$^{++}$ transitions — $\lambda4364$, $5\%\,Z_{\odot}$0, and the UV O III]$5\%\,Z_{\odot}$1 line — each with different excitation energies and critical densities, to solve for $5\%\,Z_{\odot}$2 and $5\%\,Z_{\odot}$3 simultaneously within the same ionization zone.

Data and sample

The analysis uses ultra-deep JWST/NIRSpec medium-resolution spectroscopy from the SPURS survey (GO 9214) of the Abell 2744 lensing cluster, with custom flat-field extensions that push G140M and G235M coverage to 3.2 and 5.2 μm respectively; cross-grating comparisons indicate the extended-range flux calibration is accurate to ~10%. From 75 spectra at $5\%\,Z_{\odot}$4, the authors select galaxies with R3 < 5 (implying $5\%\,Z_{\odot}$5 under empirical calibrations) and >3σ detections of all three O$5\%\,Z_{\odot}$6 lines, yielding four candidates at $5\%\,Z_{\odot}$7. Dust corrections use an iteratively updated Case B Balmer decrement rather than a fixed intrinsic ratio.

Results: impostors revealed

Under the traditional direct method with $5\%\,Z_{\odot}$8, all four candidates show implausibly high temperatures ($5\%\,Z_{\odot}$9 K; capped at 30,000 K) and metallicities of 12+log(O/H) ≈ 7.0–7.2 — consistent with genuine EMPG status. The self-consistent solution tells a different story:

Galaxy	R3	log $\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$0(O III])	$\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$1 (K)	Traditional 12+log(O/H)	Self-consistent 12+log(O/H)
SPURS-A2744-415	3.0	5.74	15,000	7.00	7.79
SPURS-A2744-422	4.1	5.96	13,000	7.05	8.14
SPURS-A2744-437	4.3	<5.14	22,000	7.24	7.34
SPURS-A2744-544	2.8	5.34	17,000	7.08	7.48

Three of the four have $\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$2 in the high-ionization zone, suppressing $\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$3 and inflating the inferred $\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$4; their true metallicities are 12+log(O/H) ≈ 7.3–8.2, i.e., underestimated by up to ~1.1 dex under the low-density assumption. Only SPURS-A2744-437 remains a plausible genuine EMPG. Because these "impostors" retain low observed R3 ratios despite higher metallicities, neither strong-line diagnostics nor the traditional direct method with a low-$\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$5 prior can distinguish them from real EMPGs. The authors note that impostors could constitute 50%–75% of EMPG candidates selected this way, while explicitly flagging that a completeness analysis is needed before generalizing this fraction.

Selection effects and density stratification

A potential concern is whether requiring an O III]$\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$6 detection biases the sample toward high-density objects. The authors test this directly: for fixed $\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$7 and $\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$8 fluxes, lower O III] flux corresponds to higher inferred $\mathrm{[O\,III]}\lambda5008/\mathrm{H}\beta$9, meaning a genuine low-density EMPG would produce a brighter O III] line than an impostor with the same R3 and auroral ratio — well above the detection threshold. The O III] cut therefore does not manufacture the impostor population; high-density objects are simply common among low-R3 galaxies. A corollary is that O III]/λ4364 is a superior $n_e=10^3\,\mathrm{cm^{-3}}$0 diagnostic for metal-poor systems.

The paper also reports what it identifies as the first sample at these redshifts with both $n_e=10^3\,\mathrm{cm^{-3}}$1(C III]) and $n_e=10^3\,\mathrm{cm^{-3}}$2(O III]). In two impostors, $n_e=10^3\,\mathrm{cm^{-3}}$3(O III]) exceeds $n_e=10^3\,\mathrm{cm^{-3}}$4(C III]) by ~2 dex — about 1 dex more contrast than seen in literature samples using N IV] as a high-ionization proxy — indicating strong density stratification across ionization zones. However, the comparison is not fully like-for-like: literature galaxies lack direct $n_e=10^3\,\mathrm{cm^{-3}}$5(O III]), nitrogen lines are undetected in the impostors, and the expected stratification ordering ($n_e=10^3\,\mathrm{cm^{-3}}$6(O II) < $n_e=10^3\,\mathrm{cm^{-3}}$7(C III]) < $n_e=10^3\,\mathrm{cm^{-3}}$8(O III]) < $n_e=10^3\,\mathrm{cm^{-3}}$9(N IV)) remains observationally unverified, with some studies reporting the reverse trend.

Broader implications

If a substantial fraction of high-redshift EMPG candidates are impostors, then the low-mass end of the mass-metallicity relation and the apparent evolution of the fundamental metallicity relation at $\lambda5008$0 may be biased by incorrect density assumptions. Relatedly, Arellano-Córdova et al. (2026) showed that N/O can be overestimated by up to ~0.8 dex when high-ionization densities are ignored, so claimed nitrogen and carbon enhancements in early galaxies should be re-examined with multi-zone density measurements. The authors also argue that all direct-$\lambda5008$1 metallicities at high redshift — not only EMPG candidates — warrant re-analysis, since many strong-line diagnostics rely on transitions with critical densities even lower than $\lambda5008$2.

Limitations and open questions

Several caveats bound the conclusions. The sample contains only four galaxies, one of which has large uncertainties on $\lambda5008$3(O III]) ($\lambda5008$4) leaving a low-density solution viable, so the 50%–75% impostor fraction is indicative rather than definitive. The low-ionization-zone temperature is estimated via the Campbell et al. (1986) relation with an assumed $\lambda5008$5, and when [O II] is undetected the total abundance reduces to O$\lambda5008$6/H$\lambda5008$7 alone. The physical origin of the extreme high-ionization densities — whether dense natal cloud structure or something else — is not established, and the proposed stratification scenario awaits confirmation through matched tracers across ionization zones. The authors identify concrete follow-ups: archival re-analysis of all available high-redshift spectra with O III] coverage, empirical relations linking $\lambda5008$8 across ionization zones, independent checks with N IV] and ALMA far-infrared [O III] lines, and incorporation of high $\lambda5008$9 into SED-fitting stellar mass estimates.

Conclusion

Using self-consistent simultaneous constraints on $n_e\sim10^5\,\mathrm{cm^{-3}}$0 and $n_e\sim10^5\,\mathrm{cm^{-3}}$1 from three O$n_e\sim10^5\,\mathrm{cm^{-3}}$2 lines in SPURS spectroscopy of Abell 2744, Hsiao et al. demonstrate that three of four low-R3 EMPG candidates at $n_e\sim10^5\,\mathrm{cm^{-3}}$3 are high-density impostors whose metallicities were underestimated by up to ~1.1 dex. The result implies that both strong-line selection and traditional direct-method confirmation can systematically misclassify dense high-redshift galaxies as extremely metal-poor, and that accurate chemical abundances in the early universe require density measurements in the same high-ionization zones as the abundance diagnostics themselves.