Barred Galaxies: Structure and Evolution

• Barred galaxies are disc galaxies featuring an elongated stellar bar that redistributes angular momentum and fuels central star formation.
• Observational analyses—including Fourier decomposition, ellipse fitting, and IFU spectroscopy—robustly identify bars from the local universe to high redshifts with tools like JWST.
• Dynamical studies and simulations reveal that bars drive gas inflow, modulate star formation rates, and influence long-term structural evolution.

A barred galaxy is a disc galaxy whose stellar distribution contains a large-scale, non-axisymmetric bar: an elongated central structure that is dynamically coupled to the disc and often to spiral arms. In contemporary galaxy-evolution research, bars are treated as major drivers of secular evolution because they redistribute angular momentum and drive gas from the outer disk into the circumnuclear region, enhancing central star formation and fueling long-term structural transformation. Observationally, barred systems are studied from the local universe to the young universe with a heterogeneous toolkit that includes visual morphology, isophotal analysis, Fourier decomposition, IFU spectroscopy, cosmological simulations, and orbit-based dynamical modeling; JWST has now pushed direct bar identification to $z\sim2$ and to a spectroscopically anchored unlensed barred spiral at $z_{\rm spec}=3.159$ (Jr. et al., 2024, Ivanov et al., 22 Jun 2026).

1. Definition and structural diagnostics

In dynamical terms, a barred galaxy is a disc galaxy hosting a dominant $m=2$ non-axisymmetric mode in the stellar distribution. In photometric analyses, this is commonly quantified from the face-on stellar surface density through the normalized Fourier amplitude

$$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$

with the corresponding phase

$$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$

A galaxy is classified as “barred” in the TNG100 analysis if $A_{2,\max}\geq0.2$, $r_{\rm bar}>1$ kpc, and $\Phi(R)$ is approximately constant within $r_{\rm bar}$; “strongly barred” requires $A_{2,\max}\geq0.3$ (Rosas-Guevara et al., 2019).

A second diagnostic family is based on isophotal ellipse fitting. The canonical bar signature is that the ellipticity $z_{\rm spec}=3.159$0 rises steadily to a maximum, typically with $z_{\rm spec}=3.159$1, while the position angle remains approximately constant across the bar; at the bar end, $z_{\rm spec}=3.159$2 drops by at least $z_{\rm spec}=3.159$3 and the position angle twists as the fit transitions into the outer disc. These criteria are used in both local-cluster work and JWST high-redshift case studies, and remain the standard bridge between visual and quantitative classification (Jr. et al., 2024).

At the orbit level, the bar is the collection of stellar orbits whose apsides librate about the bar position angle in the bar’s rotating frame. The x1 family constitutes the backbone of the bar, while bifurcated and vertically extended families contribute to boxy/peanut and X-shaped morphologies. This orbital viewpoint is central to torque-based evolution studies and to Schwarzschild modeling, because it links the observable bar to trapped-orbit families, resonance structure, and angular-momentum exchange with the outer disc and halo (Petersen et al., 2019).

2. Identification across wavelength, resolution, and redshift

Bar detectability depends strongly on the rest-frame wavelength being observed. In the local universe, near-IR imaging is favored because it traces the old stellar population that structures bars and is minimally affected by dust extinction and patchy star-forming regions; at $z_{\rm spec}=3.159$4, H-band and optical $z_{\rm spec}=3.159$5 classifications were reported to be “essentially identical” for overlapping samples, but near-IR was adopted as the primary basis for uniformity and to mitigate dust bias (Giordano et al., 2011).

At high redshift, the wavelength dependence becomes explicit through

$z_{\rm spec}=3.159$6

For CEERS-30155 at $z_{\rm spec}=3.159$7, JWST imaging shows that bar visibility changes systematically with filter because each band probes a different rest-frame regime. The bar is not evident in F115W, which traces rest-UV/blue light; it is visible in F200W, which traces rest-optical light; and it is most prominent in F444W, which traces rest-NIR light and therefore the low-mass, long-lived stars that dominate the stellar mass. Quantitative detectability is additionally limited by the PSF: in the same study, ellipse fits of F444W only robustly detect bars with semimajor axis at least one PSF, corresponding at $z_{\rm spec}=3.159$8 to $z_{\rm spec}=3.159$9 or $m=2$0 kpc, whereas the sharper F200W PSF of $m=2$1 or $m=2$2 kpc is better suited to smaller bars if dust does not erase the rest-optical signature (Jr. et al., 2024).

Filter	Rest-frame at $m=2$3	Bar appearance in CEERS-30155
F115W	$m=2$4 Å	Not evident
F200W	$m=2$5 Å	Visible; size $m=2$6 kpc
F444W	$m=2$7	Most prominent; $m=2$8 kpc

The same logic extends to earlier epochs. COSMOS-74706 at $m=2$9 is identified as a barred spiral by three independent methods: Sérsic-residual morphology, ellipse-fitting, and Fourier decomposition. The inner elongated feature is most conspicuous in F200W, F277W, and F356W; the measured bar semimajor axis is $$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$0 kpc; and the Fourier “bar modulus” reaches $$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$1 in F200W, well above the threshold calibrated on $$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$2–3 barred spirals. This establishes an unlensed barred spiral within 2.049 Gyr of the Big Bang and sharpens the timeline for the assembly of rotationally supported discs (Ivanov et al., 22 Jun 2026).

3. Incidence across morphology, mass, environment, and epoch

Bar incidence is not uniform across the Hubble sequence. In a $$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$3 comparison of isolated AMIGA galaxies and Virgo-cluster members, the global barred fraction of morphologically selected discs was $$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$4 for AMIGA and $$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$5 for Virgo, with the total barred fraction remaining in the $$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$6–35% range and showing no significant dependence on environment. Within that same framework, early-type spirals ($$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$7–$$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$8) have $$A_2(R)=\frac{\left|\sum_j m_j e^{2i\theta_j}\right|}{\sum_j m_j},$$9–50%, late-type spirals ($$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$0–$$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$1) have $$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$2, and S0s show a lower overall barred fraction that becomes strongly mass dependent, dropping below $$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$3 in the least massive S0s (Giordano et al., 2011).

Dense environments introduce a more nuanced picture rather than a single consensus result. In the Coma cluster core, an ACS study found an optical S0 bar fraction of $$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$4 and reported that none of the identified bars had $$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$5 kpc, suggesting an elevated S0 bar fraction but limited bar growth in the densest local environment (Marinova et al., 2010). A later matched-sample ACS analysis of bright S0s in Coma, A901/902, and Virgo found Coma-core bar fractions of $$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$6, $$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$7, and $$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$8 for strict ellipse fitting, relaxed ellipse fitting, and visual classification, respectively, but concluded that the S0 bar fraction does not show a statistically significant variation across environments spanning $$\Phi(R)=\frac{1}{2}\arctan\!\left[\frac{\sum_j m_j \sin(2\theta_j)}{\sum_j m_j \cos(2\theta_j)}\right].$$9–10,000 gal/Mpc$A_{2,\max}\geq0.2$0 (Marinova et al., 2012). A Virgo–field structural comparison further showed that field bars are absolutely longer at fixed luminosity, yet the difference largely disappears when bar size is normalized by host-galaxy size, implying that “bars adapt to the discs in which they live” (Aguerri et al., 2023).

Cosmological simulations reproduce both the ubiquity of bars and their redshift evolution, though with method-dependent trends. In the EAGLE CNN analysis of 3,964 galaxies with $A_{2,\max}\geq0.2$1, the global bar fraction is roughly constant from $A_{2,\max}\geq0.2$2 to $A_{2,\max}\geq0.2$3 at $A_{2,\max}\geq0.2$4 to $A_{2,\max}\geq0.2$5, then declines to $A_{2,\max}\geq0.2$6 by $A_{2,\max}\geq0.2$7; the fraction is highest in spirals, from $A_{2,\max}\geq0.2$8 at $A_{2,\max}\geq0.2$9 to $r_{\rm bar}>1$0 at $r_{\rm bar}>1$1, while ellipticals remain below $r_{\rm bar}>1$2 (Cavanagh et al., 2021). In TNG100 at $r_{\rm bar}>1$3, 107 of 270 disc-dominated galaxies are barred, corresponding to $r_{\rm bar}>1$4, with $r_{\rm bar}>1$5 strongly barred and $r_{\rm bar}>1$6 weakly barred (Rosas-Guevara et al., 2019). JWST-based high-redshift demographics remain limited, but lower limits of $r_{\rm bar}>1$7 at $r_{\rm bar}>1$8 and $r_{\rm bar}>1$9 at $\Phi(R)$0 have been reported for mass-limited samples, and the spectroscopic discovery of COSMOS-74706 confirms that mature barred spirals existed by $\Phi(R)$1 (Ivanov et al., 22 Jun 2026).

4. Gas inflow, star formation, and quenching

Bars are central engines of secular evolution because they torque the ISM, drive inflows, and reconfigure where star formation occurs. In TNG100, strong bars form mainly between $\Phi(R)$2, with more massive galaxies hosting older bars, and the bar population at $\Phi(R)$3 differs sharply from the unbarred one: strongly barred discs are predominantly quenched with $\Phi(R)$4, whereas unbarred discs remain on the star-forming sequence at $\Phi(R)$5. Nuclear quenching occurs shortly after bar formation, with rapid drops in central SFR and star formation efficiency inside $\Phi(R)$6, while the outer disc remains comparatively unaffected. Gas fractions are also systematically lower in barred systems, with differences reaching $\Phi(R)$7 dex at the high-mass end (Rosas-Guevara et al., 2019).

Spatially resolved IFU work shows that bar-driven star formation is strongly mass dependent. In the MaNGA census of 684 barred galaxies, only low-mass galaxies host H$\Phi(R)$8 along their bars, and 94% of the H$\Phi(R)$9-along-bar systems have $r_{\rm bar}$0. By contrast, ring-like H$r_{\rm bar}$1 morphologies are predominantly a high-mass phenomenon, with 91% of ringed systems at $r_{\rm bar}$2, while H$r_{\rm bar}$3 at bar ends is likewise biased high, with 74% above $r_{\rm bar}$4. For the subsample with H$r_{\rm bar}$5 along the bar, 109 of 115 systems are star-forming on BPT diagnostics, and 67 $r_{\rm bar}$6 7% show H$r_{\rm bar}$7 bars leading the stellar bar by $r_{\rm bar}$8–20$r_{\rm bar}$9 in the direction of rotation (Fraser-McKelvie et al., 2020).

The bar effect changes again in strongly perturbed cluster galaxies. In the GASP jellyfish sample, barred galaxies affected by ram pressure show the strongest central $A_{2,\max}\geq0.3$0 enhancements and central rejuvenation, and barred jellyfish exhibit inner sSFR values more than an order of magnitude above those of unbarred jellyfish. Within the same RPS subsample, every case with centrally suppressed star formation also shows AGN activity, consistent with a sequence in which bar-driven inflow and ICM compression first boost central activity and can later help ignite nuclear activity (Sanchez-Garcia et al., 2023). On GMC scales, the M83 hydrodynamical model reaches the same qualitative conclusion from the ISM side: clumps in the bar have larger virial parameters than those in the spiral arms, and the SFE in the bar region is $A_{2,\max}\geq0.3$1 of the spiral arm region, despite a mean $A_{2,\max}\geq0.3$2–$A_{2,\max}\geq0.3$3 relation consistent with the Kennicutt–Schmidt law (Nimori et al., 2012).

5. Formation, growth, slowdown, and destruction

Simulations resolve barred galaxies not as static morphologies but as systems that assemble, grow, buckle, slow down, and sometimes dissolve. In EAGLE, bars in present-day massive discs develop after $A_{2,\max}\geq0.3$4; strong bars grow relatively quickly, in a few Gyr or $A_{2,\max}\geq0.3$5 disc rotation periods, whereas weak bars grow more slowly and are still increasing in strength at $A_{2,\max}\geq0.3$6. As strong bars grow, their pattern speeds drop by $A_{2,\max}\geq0.3$7 over the last $A_{2,\max}\geq0.3$8 Gyr, and by $A_{2,\max}\geq0.3$9 their corotation radii are roughly ten times the bar length. That makes them “slow” in a sense that is inconsistent with the observationally inferred fast-bar regime, where $z_{\rm spec}=3.159$00, and the discrepancy is explicitly identified as a potential challenge for disc galaxy formation in $z_{\rm spec}=3.159$01CDM if confirmed (Algorry et al., 2016).

Torque-based analyses resolve this evolution into three phases: assembly, secular growth, and steady-state equilibrium. During assembly, a local dynamical instability rapidly traps disc orbits into the nascent bar; during secular growth, the bar loses angular momentum to sinks in the halo or outer disc and lengthens by trapping new orbits at larger radii; and in steady-state equilibrium, net $z_{\rm spec}=3.159$02 exchange becomes small, the pattern speed stabilizes, and the bar neither lengthens nor slows appreciably. Which channel dominates depends on the halo: in a cusped halo, the halo mediates assembly and growth, whereas in a cored halo the outer disc can be the main torque sink (Petersen et al., 2019).

Bars also have finite lifetimes and can recur. In the EAGLE morphology-tracking study, some galaxies undergo episodes of bar creation, destruction, and regeneration, with a mean bar lifetime of 2.24 Gyr. Major mergers are more commonly linked to bar destruction, while minor merging and accretion are linked to both creation and destruction (Cavanagh et al., 2021). Zoom-in simulations further show that subgrid physics changes the entire bar pathway: in the Eris suite, enhanced effective feedback in Eris2k delays early stellar mass growth but ultimately produces an earlier, stronger, and longer bar, with strong-bar formation at $z_{\rm spec}=3.159$03 and bar size reaching $z_{\rm spec}=3.159$04 kpc, whereas ErisBH forms a later, shorter bar at $z_{\rm spec}=3.159$05 that slows secularly and eventually buckles into a boxy/peanut bulge (Zana et al., 2018).

6. Dynamical modeling, magnetic structure, and observational outlook

Bars are sufficiently non-axisymmetric that they cannot be treated adequately by standard axisymmetric dynamical models. Orbit-superposition work has therefore extended Schwarzschild modeling to barred galaxies by explicitly including barred structures, deprojecting the stellar light into an axisymmetric disk plus a triaxial barred bulge, and allowing figure rotation with the bar pattern speed as a free parameter. In a mock barred galaxy with IFU data created from an N-body simulation, this approach fit the observed 2D surface density and all kinematic features well, recovered the bar pattern speed with relative uncertainty smaller than 10%, and separated the system into an X-shaped bar, a boxy bulge, a vertically extended structure, and a disk (Tahmasebzadeh et al., 2022). A related implementation based on Fourier coefficients in the meridional plane showed that strongly barred analytic Ferrers-like density profiles could not be turned into stable, dynamically self-consistent 3D equilibria, whereas models reconstructed from N-body barred snapshots were near-stationary, underscoring that realistic barred structure is constrained as much by orbital support as by density fitting (Vasiliev et al., 2015).

Bar-driven structure is not limited to stars and gas. In global MHD simulations, a cosmic-ray-driven dynamo operating in barred galaxies amplifies weak seed magnetic fields up to a few $z_{\rm spec}=3.159$06G within a few Gyr, and the modeled magnetic configuration resembles observed high-frequency polarized radio maps. The resulting fields develop ridges along bar dust lanes, magnetic arms that drift into inter-arm regions, and X-shaped halo structures in edge-on projections, linking the bar not only to secular stellar evolution but also to the organization of galactic magnetism (Kulpa-Dybeł et al., 2011).

The current observational frontier is the young universe. The CEERS case study shows that a combined F200W+F444W strategy improves bar detection at $z_{\rm spec}=3.159$07–4 by balancing spatial resolution against dust-insensitive stellar-mass tracing, while COSMOS-74706 demonstrates that a short, strong stellar bar could already exist at $z_{\rm spec}=3.159$08. At even higher redshifts ($z_{\rm spec}=3.159$09), the Giant Magellan Telescope is identified as a cornerstone facility for extending bar studies to even younger epochs, where the dominant limitation is no longer merely sensitivity but the simultaneous requirement of infrared access and finer angular resolution (Jr. et al., 2024).