Fundamental Metallicity Relation (FMR)

• Fundamental Metallicity Relation (FMR) is an empirical three-parameter relation that links gas-phase metallicity, stellar mass, and star-formation rate in star-forming galaxies.
• It refines the traditional mass–metallicity relation by incorporating SFR to reduce scatter and provide a more precise depiction of galaxy chemical evolution.
• FMR is used to interpret diverse galaxy populations and test theoretical models of inflows, outflows, and star formation regulation in evolving systems.

The Fundamental Metallicity Relation (FMR) is the empirical three-parameter relation linking gas-phase metallicity, stellar mass, and star-formation rate in star-forming galaxies, usually expressed through $12+\log(\mathrm{O/H})$, $M_\star$, and SFR. It generalizes the classical mass–metallicity relation by incorporating the systematic decrease of metallicity with increasing SFR at fixed stellar mass, and was originally identified as a thin surface in $(M_\star,\mathrm{SFR},Z)$ space for local SDSS galaxies (Mannucci et al., 2010). Subsequent work extended the relation to lower masses, tested it in resolved and non-BPT-star-forming regimes, examined its dependence on metallicity diagnostics, and used it to interpret special populations such as long-GRB hosts, supernova hosts, and interacting galaxies (Mannucci et al., 2010).

1. Definition and mathematical parameterization

The classical mass–metallicity relation (MZR) correlates stellar mass with gas-phase oxygen abundance, conventionally parameterized as

$12+\log(\mathrm{O/H}).$

The FMR replaces the 2D MZR by a 3D dependence

$Z = f(M_\star,\mathrm{SFR}),$

with the defining empirical feature that, at fixed $M_\star$, metallicity decreases with increasing SFR (Mannucci et al., 2010).

A convenient projection introduced for the FMR is

$\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$

with $M_\star$ and SFR in solar units. For local SDSS galaxies, the value minimizing the metallicity scatter is

$$\alpha=0.32, \qquad \mu_{0.32}=\log(M_\star)-0.32\log(\mathrm{SFR}).$$

In this representation, metallicity depends primarily on $\mu_{0.32}$, although the original authors stressed that no part of the full 3D surface is exactly planar (Mannucci et al., 2010, Mannucci et al., 2010).

Using

$M_\star$0

the local FMR was parameterized as

$M_\star$1

for the main SDSS regime, while a later low-mass extension adopted

$M_\star$2

thereby producing a continuous description from $M_\star$3 upward (Mannucci et al., 2010). On the metallicity scale used in these formulations, solar abundance is

$M_\star$4

(Mannucci et al., 2010).

2. Empirical establishment in SDSS and claimed redshift behaviour

The original FMR was constructed from approximately $M_\star$5 local SDSS DR7 star-forming galaxies by binning the $M_\star$6 plane in 0.15 dex intervals and measuring the median metallicity in each bin (Mannucci et al., 2010). In this dataset, galaxies populate a smooth 2D surface in $M_\star$7 space. The overall dispersion around the FMR is $M_\star$8 dex, and in the central, well-populated region it is $M_\star$9 dex in metallicity, compared with $(M_\star,\mathrm{SFR},Z)$0 dex for the 2D MZR (Mannucci et al., 2010). This reduction in scatter was the empirical basis for describing the relation as “fundamental.”

The phenomenology is strongly mass-dependent. At low stellar mass, metallicity decreases sharply with increasing SFR at fixed $(M_\star,\mathrm{SFR},Z)$1; at high stellar mass, metallicity becomes nearly independent of SFR and approaches a saturation regime (Mannucci et al., 2010). In terms of specific SFR, the original work also identified a threshold near $(M_\star,\mathrm{SFR},Z)$2, above which metallicity drops steeply with increasing SSFR for all masses, while below that threshold the dependence weakens, especially for massive galaxies (Mannucci et al., 2010).

The original SDSS-based formulation further reported that galaxies up to $(M_\star,\mathrm{SFR},Z)$3 follow the same FMR with little or no evolution, and argued that much of the observed redshift evolution of the MZR results from selecting progressively higher-SFR, hence lower-metallicity, galaxies at fixed mass (Mannucci et al., 2010). The same studies also reported that galaxies at $(M_\star,\mathrm{SFR},Z)$4 lie about $(M_\star,\mathrm{SFR},Z)$5 dex below the local FMR, indicating evolution beyond $(M_\star,\mathrm{SFR},Z)$6 (Mannucci et al., 2010, Mannucci, 2011).

3. Extensions in mass, spatial scale, and dimensionality

Because the original SDSS calibration was limited to $(M_\star,\mathrm{SFR},Z)$7, a low-mass extension was required to study dwarf galaxies and many transient hosts. Using SDSS-DR7 galaxies down to $(M_\star,\mathrm{SFR},Z)$8, the low-mass extension showed that the FMR continues smoothly to smaller $(M_\star,\mathrm{SFR},Z)$9, but with substantially larger intrinsic scatter than at high mass (Mannucci et al., 2010). This increased scatter was interpreted as physically real and linked to the more stochastic star-formation histories of dwarf systems (Mannucci et al., 2010).

Integral-field studies reframed the FMR on resolved scales. An analytic derivation based on MaNGA showed that a local anti-correlation between residual gas metallicity and residual SFR surface density,

$12+\log(\mathrm{O/H}).$0

leads, after spatial integration, to the global FMR with the same slope $12+\log(\mathrm{O/H}).$1 (Almeida et al., 2019). This formulation makes the global FMR a consequence of a local disk-level anti-correlation between star formation and metallicity (Almeida et al., 2019). MUSE observations of nine nearby dwarf galaxies pushed this perspective to $12+\log(\mathrm{O/H}).$2 pc scales and found that the $12+\log(\mathrm{O/H}).$3–metallicity anti-correlation is tighter in the low-mass galaxies of the sample, while different star-forming regions within the same galaxy can exhibit different local behaviours (Bulichi et al., 2023).

A separate extension concerned dimensionality rather than mass. A principal component analysis of 29 physical parameters for 41,338 SDSS star-forming galaxies identified the surface density of stellar mass $12+\log(\mathrm{O/H}).$4 as the fourth most important parameter after stellar mass, metallicity, and molecular-gas mass (Hashimoto et al., 2017). Incorporating $12+\log(\mathrm{O/H}).$5 into a 4D molecular-gas FMR reduced the metallicity dispersion to 50% of that of the 3D molecular-gas FMR, indicating that compactness carries residual information not captured by $12+\log(\mathrm{O/H}).$6, metallicity, and gas content alone (Hashimoto et al., 2017).

4. Physical interpretations and theoretical frameworks

The standard physical interpretation of the FMR is that metal-poor gas inflow both fuels star formation and dilutes the interstellar medium, producing the observed anti-correlation between SFR and metallicity at fixed mass (Mannucci et al., 2010). In this picture, outflows and the depth of the potential well set the mass dependence, while the Schmidt–Kennicutt law links gas supply to star formation (Mannucci et al., 2010).

A simple analytic model formulated the FMR in terms of gas mass $12+\log(\mathrm{O/H}).$7, star formation $12+\log(\mathrm{O/H}).$8, inflow $12+\log(\mathrm{O/H}).$9, and outflow $Z = f(M_\star,\mathrm{SFR}),$0. In that model, the gas-phase oxygen abundance satisfies

$Z = f(M_\star,\mathrm{SFR}),$1

with $Z = f(M_\star,\mathrm{SFR}),$2 and $Z = f(M_\star,\mathrm{SFR}),$3 (Dayal et al., 2012). Calibrated against the local FMR, the model found that inflow efficiency is almost mass-independent, whereas outflow efficiency scales strongly with mass and is larger in low-mass systems, consistent with momentum-driven winds (Dayal et al., 2012). Massive galaxies then approach a regime in which enrichment by star formation is compensated by dilution from metal-poor inflow, making metallicity nearly independent of SFR; lower-mass galaxies, with shallower potential wells and stronger outflows, show the observed SFR dependence (Dayal et al., 2012).

Hydrodynamical simulations supplied a complementary time-domain interpretation. In Illustris and IllustrisTNG, offsets from the star-forming main sequence and offsets from the MZR evolve on similar timescales, are often anti-correlated, and track the halo dynamical time (Torrey et al., 2017). The simulations showed that a pronounced FMR requires metallicity and SFR to evolve in an anti-correlated sense with similar variability timescales; if SFR becomes globally bursty on much shorter timescales than metallicity, the residual correlation weakens (Torrey et al., 2017). This suggests that the strength and persistence of the FMR can discriminate between non-bursty and globally bursty feedback models (Torrey et al., 2017).

5. Diagnostics, calibration dependence, and contested universality

A central methodological issue is that the empirical visibility of the FMR depends on the metallicity diagnostic. Using $Z = f(M_\star,\mathrm{SFR}),$4 SDSS star-forming galaxies, one study compared the classical Maiolino et al. calibration to the Dopita et al. (2016) metallicity indicator

$Z = f(M_\star,\mathrm{SFR}),$5

where the dominant term is [N II]/[S II], effectively tracing N/S or N/O rather than instantaneous oxygen abundance (Kashino et al., 2016). With the classical O/H-sensitive indicators, the usual FMR anti-correlation reappears; with the D16 calibration, the anti-correlation largely disappears, and at high mass a slight positive correlation emerges (Kashino et al., 2016). This was interpreted not as evidence against the FMR, but as a consequence of using a diagnostic largely insensitive to short-timescale dilution by pristine inflow (Kashino et al., 2016).

Sample selection also affects the apparent tightness and slope of the relation. In the H$Z = f(M_\star,\mathrm{SFR}),$6-selected KISS sample of over 1,450 star-forming galaxies, the optimal compressed variable was

$Z = f(M_\star,\mathrm{SFR}),$7

with the fitted relation

$Z = f(M_\star,\mathrm{SFR}),$8

and a scatter of $Z = f(M_\star,\mathrm{SFR}),$9 dex, markedly larger than in canonical SDSS FMR analyses (Hirschauer et al., 2018). The authors attributed the broader relation to the activity-biased nature of the KISS sample and emphasized that $M_\star$0 and the residual scatter depend strongly on selection and metallicity calibration (Hirschauer et al., 2018).

High-redshift metal-poor starbursts provide a further challenge to strict invariance. For 35 eBOSS star-forming galaxies at $M_\star$1 with $M_\star$2-based metallicities, the relevant comparison used $M_\star$3, following the Andrews–Martini direct-method FMR (Gao et al., 2018). These galaxies lie below the local FMR, with a fitted high-redshift relation offset by about $M_\star$4 dex and a dispersion of about $M_\star$5 dex; many individual galaxies are lower by $M_\star$6 dex (Gao et al., 2018). In that study, the offset was linked to unusually high SFRs, younger stellar ages, and the direct-method metallicity scale, and was presented as evidence for both the existence of an FMR and its cosmic evolution (Gao et al., 2018).

6. Outliers, disequilibrium phases, and extension beyond the BPT star-forming sequence

The FMR is not a universal equilibrium manifold for all dynamical states. In a sample of $M_\star$7 SDSS galaxies, the residual distribution around the FMR showed an excess low-metallicity tail relative to a Gaussian core (Grønnow et al., 2015). A merger-driven model, in which interactions dilute central metallicity by an amount proportional to stellar mass ratio for a fixed time, fitted the tail with a dilution timescale

$M_\star$8

a $M_\star$9 merger depression

$\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$0

and a minimum discernible stellar mass ratio

$\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$1

The mean metallicity depression for mergers with mass ratio between $\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$2 and $\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$3 was $\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$4 dex (Grønnow et al., 2015).

A direct pair and post-merger analysis confirmed that interacting systems are genuine off-FMR populations. In SDSS galaxy pairs, metallicity dilution at small projected separations is stronger than predicted by the FMR, and post-mergers are about $\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$5 dex too metal-poor for their mass and SFR; pairs are consistent with the FMR only when their separation exceeds $\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$6 kpc (Bustamante et al., 2020). This established mergers as a transient, non-equilibrium population rather than merely an extreme part of the ordinary FMR (Bustamante et al., 2020).

The FMR has also been extended beyond BPT-star-forming galaxies by using empirical DIG/LIER metallicity calibrations for BPT-non-SF systems. In that extended framework, galaxies above the main sequence are more metal-poor than main-sequence counterparts, consistent with gas accretion; low-mass galaxies below the main sequence are more metal-rich than main-sequence counterparts, consistent with starvation; and massive nearly quiescent galaxies with LI(N)ER-like emission have gas metallicities much closer to main-sequence values than expected from their stellar metallicities, suggesting recent accretion of circum/intergalactic gas (Kumari et al., 2021).

7. Applications to special populations and transients

The FMR has proved useful in interpreting populations that appear anomalous in the 2D MZR. Long-GRB host galaxies, for example, lie systematically below the local mass–metallicity relation at fixed mass, but are fully consistent with the extended FMR once their elevated SFRs are taken into account (Mannucci et al., 2010). For optically selected long-GRB hosts at $\mu_\alpha=\log(M_\star)-\alpha\log(\mathrm{SFR}),$7, this result was used to argue that the apparent low metallicities arise from their preference for low-mass, actively star-forming galaxies, rather than from an additional global host-metallicity bias relative to the star-forming population (Mannucci, 2011).

The FMR has also been used as an indirect metallicity estimator for Type Ia supernova host galaxies. Recalibrating the Mannucci relation for photometric stellar masses and SFRs, one study found that SN Ia Hubble residuals correlate more strongly with FMR-based host metallicity than with host mass alone, and that the improvement over mass-only metallicity is significant for the Z-PEG host analysis (Hayden et al., 2012). That result was interpreted as evidence that metallicity underlies the host-mass dependence of SN Ia distances and that an FMR-based correction can reduce systematic errors in cosmological applications (Hayden et al., 2012).

Across these applications, the FMR functions less as a single immutable equation than as a diagnostic framework for locating galaxies on, above, or below the metallicity expected from their mass and star-formation state. The cumulative literature shows that it is exceptionally informative for ordinary star-forming systems, extensible to lower masses and alternative excitation classes, and physically revealing precisely where it fails—during mergers, in strongly selected metal-poor starbursts, and in other non-equilibrium phases (Mannucci et al., 2010, Kumari et al., 2021).