AR Aur: HgMn and Am Binary System

• AR Aur is a detached eclipsing binary with nearly twin late-B stars that exhibit markedly different chemical abundances despite similar masses.
• TESS-based light curve analysis provides highly precise orbital and stellar parameters, resolving previous discrepancies in radius and light ratio estimates.
• Spectroscopic disentangling reveals a classical HgMn pattern in the primary and a milder weak Am profile in the secondary, highlighting sensitivity to small temperature differences.

Searching arXiv for papers on AR Aur / AR Aurigae to support the encyclopedia entry. AR Aur, or AR Aurigae, is a detached eclipsing, double-lined spectroscopic binary comprising two late-B stars on a circular orbit of period $4.135$ d. The primary, AR Aur A, is a classical HgMn chemically peculiar star, whereas AR Aur B is chemically peculiar in a much milder and more regular way, commonly described as a weak Am star and, in later abundance work, as a possible distinct late-B subtype. Because the two components are nearly twins in mass, age, rotation, and original composition yet show markedly different photospheric abundance patterns, the system has become a benchmark for studies of atomic diffusion, radiative levitation, chemical spotting, and the interpretation of eclipsing-binary observables in chemically peculiar atmospheres (Takeda, 14 Jan 2025, Southworth, 29 Jul 2025, Folsom et al., 2010).

1. System configuration and multiplicity

AR Aur is a close eclipsing, double-line spectroscopic binary with components of spectral type B9 V and B9.5 V. The system is nearly edge-on, and the inner orbit is circular. The primary is the star eclipsed at the deeper minimum. Both stars rotate slowly to moderately, with $v \sin i \simeq 23\ {\rm km\,s^{-1}}$, and the short orbital period makes synchronous rotation a standard working assumption; the later TESS-based rediscussion found synchronous velocities of $22.555 \pm 0.027~\mathrm{km\,s^{-1}}$ and $21.604 \pm 0.032~\mathrm{km\,s^{-1}}$, in excellent agreement with measured $v\sin i$ values (Southworth, 29 Jul 2025).

The eclipsing pair is part of a wider multiple system. A third body produces a light-time effect with $P_3 = 23.452 \pm 0.096$ yr and $e_3 = 0.262 \pm 0.023$, with a minimum third-body mass of $\sim 0.51\,M_\odot$. Its luminosity is negligible compared to the B-type pair, so it is detectable primarily through timing variations rather than direct photometric or spectroscopic light (Southworth, 29 Jul 2025).

The system’s astrophysical importance follows from a strong internal contrast. AR Aur A is a strong HgMn star, while AR Aur B is only mildly peculiar. Since both stars almost certainly formed from the same material, their present-day atmospheric differences are attributed to internal processes rather than differing birth composition. Earlier work emphasized that AR Aur is the only eclipsing binary known to contain a HgMn star, which makes the system unusually valuable because masses and radii can be derived from eclipsing-binary geometry rather than inferred indirectly (Folsom et al., 2010).

2. Photometric rediscussion and absolute dimensions

A longstanding issue in the literature concerned the radius ratio. Earlier eclipse solutions constrained by spectroscopic light ratios from metal lines, especially Mg II $4481$ Å, yielded $R_{\rm B}/R_{\rm A} > 1$, so the less massive secondary appeared slightly larger than the primary. This was interpreted cautiously as evidence that B might still be contracting toward the zero-age main sequence. The TESS reanalysis removed the need for such spectroscopic priors and showed that the spectroscopic light ratios were unreliable because chemical peculiarity alters the diagnostic lines from which they were derived (Southworth, 29 Jul 2025).

The modern solution is based on eight TESS sectors, all at 120-s cadence, analyzed sector by sector with JKTEBOP. For a well-detached system with nearly spherical components, the fitted parameters were $v \sin i \simeq 23\ {\rm km\,s^{-1}}$0, $v \sin i \simeq 23\ {\rm km\,s^{-1}}$1, the central surface brightness ratio $v \sin i \simeq 23\ {\rm km\,s^{-1}}$2, the inclination $v \sin i \simeq 23\ {\rm km\,s^{-1}}$3, third light $v \sin i \simeq 23\ {\rm km\,s^{-1}}$4, the period $v \sin i \simeq 23\ {\rm km\,s^{-1}}$5, a reference eclipse time $v \sin i \simeq 23\ {\rm km\,s^{-1}}$6, and the linear limb-darkening coefficient in the power-2 law

$v \sin i \simeq 23\ {\rm km\,s^{-1}}$7

The adopted mean solution gave $v \sin i \simeq 23\ {\rm km\,s^{-1}}$8, $v \sin i \simeq 23\ {\rm km\,s^{-1}}$9, $22.555 \pm 0.027~\mathrm{km\,s^{-1}}$0, $22.555 \pm 0.027~\mathrm{km\,s^{-1}}$1, $22.555 \pm 0.027~\mathrm{km\,s^{-1}}$2, $22.555 \pm 0.027~\mathrm{km\,s^{-1}}$3, $22.555 \pm 0.027~\mathrm{km\,s^{-1}}$4, and a TESS-band light ratio $22.555 \pm 0.027~\mathrm{km\,s^{-1}}$5 (Southworth, 29 Jul 2025).

Combining this light-curve solution with literature radial velocities yielded highly precise absolute dimensions:

Quantity	AR Aur A	AR Aur B
Mass	$22.555 \pm 0.027~\mathrm{km\,s^{-1}}$6	$22.555 \pm 0.027~\mathrm{km\,s^{-1}}$7
Radius	$22.555 \pm 0.027~\mathrm{km\,s^{-1}}$8	$22.555 \pm 0.027~\mathrm{km\,s^{-1}}$9
Surface gravity	$21.604 \pm 0.032~\mathrm{km\,s^{-1}}$0	$21.604 \pm 0.032~\mathrm{km\,s^{-1}}$1
Mean density	$21.604 \pm 0.032~\mathrm{km\,s^{-1}}$2	$21.604 \pm 0.032~\mathrm{km\,s^{-1}}$3

These values show that B is definitively smaller than A, not larger. The resulting stellar properties are matched by PARSEC 1.2 evolutionary tracks for a slightly super-solar metallicity, $21.604 \pm 0.032~\mathrm{km\,s^{-1}}$4, and an age of $21.604 \pm 0.032~\mathrm{km\,s^{-1}}$5 Myr after the zero-age main sequence. In this framework, both stars are young main-sequence objects, and the previously favored pre-main-sequence interpretation for B is no longer supported (Southworth, 29 Jul 2025).

3. Spectroscopy, disentangling, and atmospheric analysis

The most detailed abundance study used 11 BOES spectra obtained on 2010 December 14, 15, 16, 18, and 20, with wavelength coverage of approximately $21.604 \pm 0.032~\mathrm{km\,s^{-1}}$6–$21.604 \pm 0.032~\mathrm{km\,s^{-1}}$7\,Å. Because AR Aur is double-lined and the component spectra overlap, the spectra were numerically separated with the public disentangling code CRES. The calculation was performed for 69 partially overlapping spectral segments between $21.604 \pm 0.032~\mathrm{km\,s^{-1}}$8 and $21.604 \pm 0.032~\mathrm{km\,s^{-1}}$9 Å. The resulting disentangled spectra of A and B achieved typical signal-to-noise ratios from about 200 to 600, highest around $v\sin i$0 Å (Takeda, 14 Jan 2025).

Equivalent widths were measured for lines judged suitable after comparison with synthetic spectra. Profiles were fitted with a convolution of a rotational broadening function and a Gaussian, appropriate because rotational broadening dominates for $v\sin i$1. The analysis used 606 lines for A and 538 for B. Atmospheric parameters were then determined spectroscopically from Fe lines using ATLAS9 models at fixed $v\sin i$2, with $v\sin i$3 sampled from 9500 to 12000 K and microturbulence $v\sin i$4 from $v\sin i$5 to $v\sin i$6. The adopted solutions were Fe II based, because in late-B atmospheres Fe is almost entirely ionized and Fe II lines are more robust than Fe I against temperature sensitivity and modeling uncertainties (Takeda, 14 Jan 2025).

The adopted atmospheric parameters were

$v\sin i$7

and

$v\sin i$8

These temperatures are higher by $v\sin i$9–300 K than earlier SED-based values, but the temperature difference between the components remains close to $P_3 = 23.452 \pm 0.096$0–$P_3 = 23.452 \pm 0.096$1 K. The microturbulence result is especially notable: both stars have small $P_3 = 23.452 \pm 0.096$2, but the hotter primary has the larger value (Takeda, 14 Jan 2025).

Abundances were derived for 34 ionic species of 28 elements using modified WIDTH9. Non-LTE calculations were performed for elements with $P_3 = 23.452 \pm 0.096$3: He, C, N, O, Ne, Na, Mg, Al, Si, and P. Abundances were expressed relative to the Sun as

$P_3 = 23.452 \pm 0.096$4

For both stars, the global abundance pattern shows an increase of $P_3 = 23.452 \pm 0.096$5 with atomic number $P_3 = 23.452 \pm 0.096$6, but the slope is steeper and the scatter far larger for A than for B (Takeda, 14 Jan 2025).

4. AR Aur A as a HgMn star

AR Aur A displays the classical HgMn pattern: a global rise of $P_3 = 23.452 \pm 0.096$7 with $P_3 = 23.452 \pm 0.096$8 combined with conspicuous element-to-element departures from any smooth trend. The star is strongly He deficient, with $P_3 = 23.452 \pm 0.096$9. Among the light elements, $e_3 = 0.262 \pm 0.023$0, $e_3 = 0.262 \pm 0.023$1, and $e_3 = 0.262 \pm 0.023$2. Na is nearly solar at $e_3 = 0.262 \pm 0.023$3, and Mg is effectively solar within the quoted uncertainties. By contrast, Al is strongly underabundant, with upper limits of $e_3 = 0.262 \pm 0.023$4 for Al I and $e_3 = 0.262 \pm 0.023$5 for Al II; Sc is also very underabundant with $e_3 = 0.262 \pm 0.023$6, and Ni is strongly deficient at $e_3 = 0.262 \pm 0.023$7 (Takeda, 14 Jan 2025).

The heavy and iron-peak elements show the complementary behavior expected of a HgMn star. Si is mildly enhanced at $e_3 = 0.262 \pm 0.023$8, P is highly overabundant at $e_3 = 0.262 \pm 0.023$9, Ti is at $\sim 0.51\,M_\odot$0, Cr II at $\sim 0.51\,M_\odot$1, Mn II at $\sim 0.51\,M_\odot$2, and Fe II at $\sim 0.51\,M_\odot$3. Among the heavy elements, the enhancements become extreme: $\sim 0.51\,M_\odot$4, $\sim 0.51\,M_\odot$5, $\sim 0.51\,M_\odot$6, $\sim 0.51\,M_\odot$7, $\sim 0.51\,M_\odot$8, $\sim 0.51\,M_\odot$9, and $4481$0 (Takeda, 14 Jan 2025).

Earlier spectropolarimetric abundance work had already established the same qualitative picture and quantified the species from which the class takes its name. In that analysis, Hg was found at $4481$1 versus solar $4481$2, corresponding to $4481$3 dex, and Pt at $4481$4 versus solar $4481$5, corresponding to $4481$6 dex; Xe was also extraordinarily enhanced, and the primary’s abundance pattern was explicitly identified as characteristic of HgMn stars (Folsom et al., 2010).

The defining feature of A is therefore not merely enhancement of high-$4481$7 material, but strong selectivity. Very deficient N, Al, Sc, and Ni coexist with markedly overabundant P, Mn, Sr, Y, Zr, Xe, and rare-earth elements. This large dispersion around the global $4481$8-versus-$4481$9 trend is the central spectroscopic signature that distinguishes AR Aur A from its companion (Takeda, 14 Jan 2025).

5. AR Aur B as a weak Am star and a possible distinct mild peculiarity

AR Aur B is chemically peculiar, but its pattern is smoother and much less extreme than that of the primary. The later abundance analysis gave $R_{\rm B}/R_{\rm A} > 1$0, $R_{\rm B}/R_{\rm A} > 1$1, $R_{\rm B}/R_{\rm A} > 1$2, $R_{\rm B}/R_{\rm A} > 1$3, $R_{\rm B}/R_{\rm A} > 1$4, $R_{\rm B}/R_{\rm A} > 1$5, and [Mg/H] $R_{\rm B}/R_{\rm A} > 1$6. Al is slightly overabundant, with $R_{\rm B}/R_{\rm A} > 1$7 from Al I and $R_{\rm B}/R_{\rm A} > 1$8 from Al II; Si is at $R_{\rm B}/R_{\rm A} > 1$9, P at $v \sin i \simeq 23\ {\rm km\,s^{-1}}$00, and S near solar at $v \sin i \simeq 23\ {\rm km\,s^{-1}}$01 (Takeda, 14 Jan 2025).

The traditional Am diagnostics are present but subdued. Ca is slightly subsolar at $v \sin i \simeq 23\ {\rm km\,s^{-1}}$02, and Sc is significantly deficient at $v \sin i \simeq 23\ {\rm km\,s^{-1}}$03. The iron-peak elements are mildly enhanced, with $v \sin i \simeq 23\ {\rm km\,s^{-1}}$04, $v \sin i \simeq 23\ {\rm km\,s^{-1}}$05, $v \sin i \simeq 23\ {\rm km\,s^{-1}}$06, $v \sin i \simeq 23\ {\rm km\,s^{-1}}$07, and $v \sin i \simeq 23\ {\rm km\,s^{-1}}$08. The heavy elements continue the same progression: $v \sin i \simeq 23\ {\rm km\,s^{-1}}$09, $v \sin i \simeq 23\ {\rm km\,s^{-1}}$10, $v \sin i \simeq 23\ {\rm km\,s^{-1}}$11, and $v \sin i \simeq 23\ {\rm km\,s^{-1}}$12. Only upper limits were obtained for Xe, Ce, and Nd, but these limits still suggest some enhancement (Takeda, 14 Jan 2025).

The crucial difference from A is the regularity of the pattern. For B, the abundance sequence is described by the approximate relation

$v \sin i \simeq 23\ {\rm km\,s^{-1}}$13

with only small dispersion except for a few outliers, especially Sc and some very high-$v \sin i \simeq 23\ {\rm km\,s^{-1}}$14 species. Folsom and collaborators interpreted the star as a weak Am object, emphasizing the underabundance of Ca and Sc together with enhanced Ba and Nd (Folsom et al., 2010). Takeda argued that the almost straight-line dependence on $v \sin i \simeq 23\ {\rm km\,s^{-1}}$15 is unusual even among Am stars and suggested that B might represent a distinct late-B peculiarity subtype, perhaps “Bm”-type; this remains a suggestion rather than an established classification (Takeda, 14 Jan 2025).

6. Line-profile variability, magnetic constraints, and physical interpretation

High-resolution spectropolarimetry showed that AR Aur A exhibits line-profile variability in Cr, Mn, Sr, Y, Ba, Pt, and Hg, whereas the secondary showed no convincing variability in that dataset. The variability is rotationally modulated, with spectra obtained at the same orbital phase weeks apart agreeing closely, consistent with tidal synchronization of the $v \sin i \simeq 23\ {\rm km\,s^{-1}}$16-d orbital and rotational periods. The element-to-element differences in variability pattern indicate inhomogeneous surface abundance distributions, usually described as chemical spots (Folsom et al., 2010).

The same spectropolarimetric study found no statistically significant circular-polarization detection in either star. Using Least Squares Deconvolution of 1168 lines, the longitudinal field measurements had typical uncertainties of $v \sin i \simeq 23\ {\rm km\,s^{-1}}$17–$v \sin i \simeq 23\ {\rm km\,s^{-1}}$18 G and remained consistent with zero. The conservative upper limit on the longitudinal magnetic field in each star was

$v \sin i \simeq 23\ {\rm km\,s^{-1}}$19

and modeled $v \sin i \simeq 23\ {\rm km\,s^{-1}}$20 limits on a dipolar surface field were about $v \sin i \simeq 23\ {\rm km\,s^{-1}}$21–$v \sin i \simeq 23\ {\rm km\,s^{-1}}$22 G for A and $v \sin i \simeq 23\ {\rm km\,s^{-1}}$23–$v \sin i \simeq 23\ {\rm km\,s^{-1}}$24 G for B, depending on assumed obliquity. This places AR Aur in a regime very different from classical Ap/Bp stars with kilogauss organized fields and supports the conclusion that the primary’s chemical spots do not require a strong global magnetic field (Folsom et al., 2010).

The physical interpretation of the A–B contrast remains centered on diffusion in stable radiative envelopes. AR Aur A and B share similar masses, the same age and original composition, similar gravities, similar rotation, and the same binary environment, yet their abundance patterns differ radically. The later abundance analysis emphasized that A is hotter by only $v \sin i \simeq 23\ {\rm km\,s^{-1}}$25 K, with $v \sin i \simeq 23\ {\rm km\,s^{-1}}$26 K and $v \sin i \simeq 23\ {\rm km\,s^{-1}}$27 K, and noted that HgMn stars are empirically confined to about $v \sin i \simeq 23\ {\rm km\,s^{-1}}$28–$v \sin i \simeq 23\ {\rm km\,s^{-1}}$29 K. This suggests that the onset of the HgMn phenomenon is extremely sensitive to $v \sin i \simeq 23\ {\rm km\,s^{-1}}$30, with a threshold near $v \sin i \simeq 23\ {\rm km\,s^{-1}}$31 K; above it, diffusion may produce large element-specific anomalies, whereas below it the result may be the smoother abundance–$v \sin i \simeq 23\ {\rm km\,s^{-1}}$32 gradient seen in B (Takeda, 14 Jan 2025).

The revised dynamical and evolutionary solution strengthens this interpretation. With $v \sin i \simeq 23\ {\rm km\,s^{-1}}$33 and age $v \sin i \simeq 23\ {\rm km\,s^{-1}}$34 Myr, both stars are young main-sequence objects, so the system no longer requires a pre-main-sequence secondary to explain the observed chemistry. A plausible implication is that AR Aur constrains both the timescale and the parameter sensitivity of chemical separation processes: strong HgMn peculiarities, evolving spots, and a weak-Am-like or quasi-linear late-B peculiarity can all arise early in main-sequence evolution under nearly identical external conditions (Southworth, 29 Jul 2025).