Warm DQ White Dwarfs: A Transitional Class

• Warm DQ white dwarfs are a rare subclass with effective temperatures of 10,000–18,000 K, marking a transition between hot carbon-dominated and cool helium-rich DQs.
• They exhibit distinct spectral features with strong atomic carbon lines, coherent photometric variability, and measurable magnetism that challenge standard cooling models.
• Their high masses, minimal H/He envelopes, and peculiar kinematics support merger-remnant and crystallization delay scenarios in white dwarf evolution.

Warm DQ white dwarfs are a rare carbon-rich branch of the DQ spectral class characterized by optical spectra dominated by atomic carbon rather than the molecular C$_2$ Swan bands of classical cool DQs, and by effective temperatures usually placed between about $10{,}000$ and $18{,}000$ K. In the standard phenomenological sequence they occupy the interval between the hot, carbon-dominated DQs at $T_{\rm eff}\approx 18{,}000$–$24{,}000$ K and the cooler helium-dominated DQs below about $10{,}000$–$13{,}000$ K, although some recent survey analyses have applied a broader temperature domain to carbon-bearing objects. Their significance derives from a distinctive conjunction of properties: high masses, carbon-rich and often helium-poor atmospheres, coherent photometric variability in several prototypes, concentration on the Gaia Q branch, and kinematics that are difficult to reconcile with ordinary single-star cooling, making them central to merger-remnant and crystallization-delay scenarios (Macfarlane et al., 2017, Koester et al., 2019, Kilic et al., 16 Jul 2025).

1. Definition and taxonomic position

DQ white dwarfs are white dwarfs whose spectra show carbon in the atmosphere. Within this class, the empirical separation most often used in recent work is between classical or cool DQ stars, which show optical C$_2$ Swan bands and have $T_{\rm eff}\lesssim 10{,}000$ K, warm DQ stars, which show atomic C I lines and sometimes weak Swan bands above $T_{\rm eff}\gtrsim 10{,}000$ K, and hot DQ stars above about $10{,}000$0 K, where carbon-dominated atmospheres become common (Koester et al., 2019). In the formulation used for OW J1753-3107, hot DQs occupy $10{,}000$1 with $10{,}000$2, whereas warm DQs lie at $10{,}000$3–$10{,}000$4 K with $10{,}000$5 to $10{,}000$6, and are thought to bridge the hot-DQ stage and the more common cooler helium-dominated DQs (Macfarlane et al., 2017).

At medium optical resolution, warm DQs are recognized by neutral carbon absorption features such as C I $10{,}000$7, $10{,}000$8, and $10{,}000$9, with infrared C I multiplets such as $18{,}000$0 sometimes present. Unlike cool DQs, they do not display strong Swan bands, and unlike hot DQs the C II features, including $18{,}000$1, remain weak or absent in many survey objects. In the 500 pc Gaia-selected GTC/OSIRIS sample, He I $18{,}000$2 and $18{,}000$3 were never detected in the warm DQs, leading that study to describe them as helium-poor atmospheres (Zamora et al., 15 May 2026).

The class also contains spectroscopic subtypes. Warm DQAs show shallow Balmer lines; DQZAs add weak O I features; and DAQs are hydrogen-dominated but retain visible carbon lines. Kilic et al. argued that DAQ, DQA, and pure warm DQ stars form a single continuous population, so that the distinction depends on whether H, He, or C lines are strongest in the photosphere rather than on a fundamentally different interior or evolutionary channel (Kilic et al., 2024, Kilic et al., 6 Feb 2026). A plausible implication is that “warm DQ” is best understood as a temperature-composition regime rather than a sharply bounded spectroscopic species.

2. Prototypical objects and the emergence of the class

The first extensively characterized variable warm DQ was SDSS J103655.39+652252.2, usually abbreviated SDSS J1036+6522. Joint spectroscopic and photometric analysis yielded a self-consistent parameter set of $18{,}000$4 K, $18{,}000$5, $18{,}000$6, and a mean magnetic field $18{,}000$7 MG. The star was described as unique among DQ white dwarfs because it lies halfway between the hot carbon-dominated DQs and the cool helium-dominated DQs, and its temperature and abundances were interpreted as evidence for a transition object (Williams et al., 2013).

The second known variable warm DQ was OW J175358.85-310728.9, or OW J1753-3107. It was discovered in the OmegaWhite Survey and was reported as the brightest of the then known warm or hot DQs. Spectral modeling yielded $18{,}000$8 K, $18{,}000$9, $T_{\rm eff}\approx 18{,}000$0, and a mean field $T_{\rm eff}\approx 18{,}000$1 MG, making it closely similar to SDSS J1036+6522 in temperature and gravity, but different in carbon abundance, field strength, modulation period, and amplitude (Macfarlane et al., 2017).

Parameter	SDSS J1036+6522	OW J1753-3107
$T_{\rm eff}\approx 18{,}000$2	$T_{\rm eff}\approx 18{,}000$3 K	$T_{\rm eff}\approx 18{,}000$4 K
$T_{\rm eff}\approx 18{,}000$5	$T_{\rm eff}\approx 18{,}000$6	$T_{\rm eff}\approx 18{,}000$7
Carbon abundance	$T_{\rm eff}\approx 18{,}000$8	$T_{\rm eff}\approx 18{,}000$9
Magnetic field	$24{,}000$0 MG	$24{,}000$1 MG
Photometric period	$24{,}000$2 s	$24{,}000$3 min

These two stars established several defining themes of the subclass: intermediate effective temperature, high gravity, conspicuous carbon lines, measurable magnetism in at least some members, and coherent single-period variability. They also provided the first direct evidence that warm DQs are not merely a static spectroscopic bridge, but a dynamically interesting class with nontrivial links to the variable DQ white dwarfs.

3. Atmospheric composition, magnetic line formation, and parameter inference

Warm DQ atmospheric analysis is technically demanding because the relevant spectra are governed by mixed C-He or C-H compositions, strong carbon line blanketing, and in some cases magnetic splitting that leaves the simple Zeeman limit. For SDSS J1036+6522, the field strength was measured from the Zeeman splitting of the well-isolated C I triplet at $24{,}000$4 Å. In the weak-field regime, the first-order shift is

$24{,}000$5

but at $24{,}000$6 MG some carbon multiplets enter the Paschen-Back regime, where $24{,}000$7–$24{,}000$8 coupling breaks down and the full Hamiltonian

$24{,}000$9

must be diagonalized term by term (Williams et al., 2013).

That treatment was implemented for all major C I transitions in SDSS J1036+6522, while C II and He I were kept in the Zeeman limit. The LTE atmosphere grid for that object spanned $10{,}000$0–$10{,}000$1 K in steps of $10{,}000$2 K, $10{,}000$3–$10{,}000$4 in steps of $10{,}000$5 dex, and $10{,}000$6 to $10{,}000$7 in steps of $10{,}000$8 dex. Each model assumed a uniform 3 MG surface field, no magnetic modification of the hydrostatic temperature structure, and Stokes-$10{,}000$9 radiative transfer only. The best-match model reproduced the continuum slope and line profiles to within the uncertainties of the approximations (Williams et al., 2013).

OW J1753-3107 was modeled with LTE atmospheres including Paschen-Back splitting of C I and C II lines under a uniform surface field. Its field strength was likewise measured from the C I triplet around $13{,}000$0 Å, using the fact that in the Paschen-Back regime each fine-structure component splits into multiple subcomponents whose wavelength separations directly yield $13{,}000$1 when compared to zero-field laboratory wavelengths (Macfarlane et al., 2017).

For larger samples, atmospheric parameters are typically obtained by iterating between photometric and spectroscopic fits. In the SDSS-Gaia analysis of Koester and Kepler, ugriz and Gaia $13{,}000$2 photometry were combined with parallaxes to determine $13{,}000$3, $13{,}000$4, and a flux scaling, while spectroscopy was used to solve for $13{,}000$5; the procedure was iterated to convergence (Koester et al., 2019). In the far-UV-selected survey, magnitudes were converted to mean fluxes and fitted by minimizing

$13{,}000$6

after which the radius followed from the solid angle and parallax, and the mass from

$13{,}000$7

That analysis used 1D LTE model atmospheres with C+H, C+He, C+H+He, and pure-C grids over $13{,}000$8 K and $13{,}000$9 (Kilic et al., 16 Jul 2025).

An observational tension remains. The two prototype variable warm DQs are clearly magnetic, but no Zeeman splitting was detected in any of the 28 warm DQs in the GTC/OSIRIS 500 pc sample at $_2$0 and $_2$1 (Zamora et al., 15 May 2026). This suggests either that magnetism is heterogeneous within the class or that field detectability is strongly dependent on spectral resolution, $_2$2, field strength, and line selection.

4. Coherent variability, rotation, and the pulsation question

Warm DQ variability was first established through time-series photometry of SDSS J1036+6522. Eleven observing runs over a 16-month span yielded a single coherent modulation with period $_2$3 s and amplitude $_2$4. Least-squares fitting with a sinusoid,

$_2$5

showed no detectable change in period, amplitude, or phase over the full data set. No harmonics above a 99% false-alarm limit of about $_2$6 mma were present, and the folded light curve was indistinguishable from a pure sine wave to within $_2$7 mma RMS (Williams et al., 2013).

OW J1753-3107 displays the same phenomenology at a longer timescale. Its photometric modulation has $_2$8 min, remains stable over two years, and is well fitted by a single-period sinusoid with peak-to-peak amplitude $_2$9 in $T_{\rm eff}\lesssim 10{,}000$0 and up to $T_{\rm eff}\lesssim 10{,}000$1 in the blue. No additional frequencies were found up to $T_{\rm eff}\lesssim 10{,}000$2 cycles d$T_{\rm eff}\lesssim 10{,}000$3, and the phase and amplitude remained constant within observational errors (Macfarlane et al., 2017).

Several proposed mechanisms have been tested against these data. For SDSS J1036+6522, classical nonradial $T_{\rm eff}\lesssim 10{,}000$4-mode pulsation models developed for hot DQs predict instability only at $T_{\rm eff}\lesssim 10{,}000$5 K and require carbon-dominated atmospheres, so that star is both too cool and too He-rich for the standard hot-DQ pulsation framework. Accretion-driven AM CVn-like models can be ruled out by the absence of emission lines or radial-velocity shifts. The remaining possibilities discussed explicitly were an oblique magnetic, roAp-like pulsator or rapid rotation with a surface spot (Williams et al., 2013). For OW J1753-3107, the lack of emission lines or radial-velocity shifts at the 35 min period, with $T_{\rm eff}\lesssim 10{,}000$6 km s$T_{\rm eff}\lesssim 10{,}000$7, led its discoverers to favor rotational modulation of a spotted magnetic white dwarf (Macfarlane et al., 2017).

The shape of the pulse profile has been especially informative. SDSS J1036+6522 has a nearly perfect sinusoid despite its strong magnetic field, whereas some magnetic hot DQ variables show highly non-sinusoidal, boxy light curves with large harmonic content. The direct conclusion drawn from SDSS J1036+6522 is that pulse shape alone cannot diagnose the presence or absence of a strong magnetic field in DQ variables (Williams et al., 2013). This conclusion is reinforced by the broader warm DQ/DAQ family, in which at least two DAQs show coherent $T_{\rm eff}\lesssim 10{,}000$8 min photometric variations interpreted as spin periods rather than nonradial pulsations (Kilic et al., 2024).

5. Population properties, survey selection, and the Q-branch locus

The subclass has expanded rapidly from a handful of prototypes to statistically useful samples. In the SDSS-Gaia study of DQ white dwarfs, 254 clear DQs remained after quality cuts, of which 221 were classical DQ, 29 were warm DQ with $T_{\rm eff}\lesssim 10{,}000$9 K, and 7 were hot DQ. The warm DQs had $T_{\rm eff}\gtrsim 10{,}000$0–$T_{\rm eff}\gtrsim 10{,}000$1 K and $T_{\rm eff}\gtrsim 10{,}000$2–$T_{\rm eff}\gtrsim 10{,}000$3, and clustered around $T_{\rm eff}\gtrsim 10{,}000$4–$T_{\rm eff}\gtrsim 10{,}000$5, producing a bimodal mass distribution with the cool DQs near $T_{\rm eff}\gtrsim 10{,}000$6 and the warm/hot DQs near $T_{\rm eff}\gtrsim 10{,}000$7 (Koester et al., 2019).

Independent envelope modeling of 26 warm DQs with $T_{\rm eff}\gtrsim 10{,}000$8–$T_{\rm eff}\gtrsim 10{,}000$9 K gave Gaia-based masses from about $10{,}000$00 to $10{,}000$01, with $10{,}000$02 and $10{,}000$03. The same study inferred extremely small envelope reservoirs, with typical limits $10{,}000$04 and $10{,}000$05, and noted mean observed radial velocities of $10{,}000$06 km s$10{,}000$07, consistent with large gravitational redshifts for high-mass objects (Koester et al., 2020).

Survey methodology has also reshaped the field. In the Gaia machine-learning-selected 500 pc sample, many objects pre-classified as “massive DB” were shown by medium-resolution spectroscopy to be magnetic white dwarfs or warm DQs; of 112 such candidates, only 5 were genuine DBs, whereas 23 were warm DQs, 1 was a DAQ, and 4 were hot DQs (Zamora et al., 15 May 2026). In the DESI DR1 hot-white-dwarf sample, 68 warm DQs were isolated, including 9 DAQs and 36 DQAs, with a mean photometric mass of $10{,}000$08 and $10{,}000$09; no warm DQs were found below about $10{,}000$10, and the most massive member reached $10{,}000$11 (Kilic et al., 6 Feb 2026).

Far-UV selection has proved especially efficient because abundant carbon produces millions of FUV absorption lines, making warm DQs unusually red in FUV-optical colors. In the all-sky survey based on Gaia plus GALEX, the primary color cut was

$10{,}000$12

with $10{,}000$13. That strategy yielded 167 candidates, 75 of which proved to be warm DQs. Their mean properties were $10{,}000$14 K and $10{,}000$15 (Kilic et al., 16 Jul 2025).

Kinematically and photometrically, warm DQs occupy a distinctive locus. In the GTC sample they cluster along the Gaia Q branch at $10{,}000$16–$10{,}000$17 and $10{,}000$18 to $10{,}000$19, and 13 of the 33 warm+hot DQs had $10{,}000$20 km s$10{,}000$21, compared with 52 of 322 stars in the full sample (Zamora et al., 15 May 2026). In the all-sky far-UV sample, 41 of 75 warm DQs had $10{,}000$22 km s$10{,}000$23, and the class clustered on the Gaia Q branch with

$10{,}000$24

identified there with the onset of crystallization (Kilic et al., 16 Jul 2025).

6. Evolutionary origin, crystallization, and unresolved problems

Two broad evolutionary pictures coexist in the literature. The first treats warm DQs as morphological intermediates between hot carbon-dominated DQs and cooler helium-dominated DQs. In this view, a very late thermal pulse leaves a carbon-oxygen core, a thin helium veneer rises to the surface to create a DB phase, and later carbon dredge-up produces the hot DQ stage, with objects such as SDSS J1036+6522 occupying the transition to helium-dominated cooler DQs [(Williams et al., 2013); (Macfarlane et al., 2017)]. The second emphasizes the strong mass dichotomy: Koester and Kepler argued that warm/hot DQs and cool classical DQs are two distinct populations, because the former are concentrated near $10{,}000$25 while the latter cluster near $10{,}000$26 (Koester et al., 2019).

More recent work strongly favors a merger-remnant interpretation for the warm-DQ population. The envelope calculations of Koester et al. showed that the inferred hydrogen and helium masses are far smaller than expected from post-AGB single-star evolution, and proposed a double white dwarf merger, for example a $10{,}000$27 pair, as the favored origin because merger spiral-in and disk formation can strip nearly all of the outer H/He (Koester et al., 2020). The DESI and GTC studies likewise describe warm DQs as an ultramassive carbon-rich sequence probably produced by double-degenerate or WD+subgiant mergers, and the far-UV survey states explicitly that their masses, kinematics, and crystallization-driven cooling delays all point to a double-white-dwarf merger origin (Kilic et al., 6 Feb 2026, Kilic et al., 16 Jul 2025).

Within that merger framework, DAQ stars appear not as a separate species but as the hydrogen-rich end of the warm-DQ continuum. In Kilic et al., the larger warm DQ/DQA sample spans approximately $10{,}000$28, and all five bona fide DAQs lie in a narrow mass-temperature interval, $10{,}000$29–$10{,}000$30 and $10{,}000$31–$10{,}000$32 K. The paper concludes that the distinction between DAQ, DQA, and warm DQ is superficial and reflects the photospheric balance of H, He, and C rather than a different evolutionary route (Kilic et al., 2024).

Crystallization physics has become central to this interpretation. Warm DQs are found on or near the crystallization sequence, and several studies argue that their cooling ages, typically of order $10{,}000$33 Gyr from standard tracks, are inconsistent with their thick-disk or halo-like kinematics unless cooling has been delayed for much longer. The far-UV survey proposes that these stars may remain stuck on or near the crystallization sequence for $10{,}000$34 Gyr owing to substantial cooling delays from the distillation of neutron-rich impurities and from phase separation in the solidifying C/O core (Kilic et al., 16 Jul 2025). This is consistent with the Q-branch concentration and the kinematic mismatch emphasized in the GTC study (Zamora et al., 15 May 2026).

Several unresolved issues remain. Koester et al. reported that the apparent correlation of stellar mass with $10{,}000$35 in warm DQs remains unexplained, and noted that helium abundances are usually only upper limits because He I $10{,}000$36 Å is never seen, while neutral-carbon atomic data in the optical remain uncertain. Their diffusion results also depend on conductive opacities and overshoot at the convective-zone base; including one pressure-scale height of overshoot changes $10{,}000$37 by about $10{,}000$38 dex, and neglecting Coulomb terms changes it by about $10{,}000$39 dex (Koester et al., 2020). On the variability side, the root cause of the modulation in SDSS J1036+6522 and related DQ variables is still unclear, and any unified explanation must account for both nearly pure sinusoidal and strongly distorted pulse shapes across the DQ sequence (Williams et al., 2013).

Warm DQ white dwarfs therefore occupy a pivotal position in white-dwarf astrophysics. Spectroscopically they connect carbon-rich hot DQs, helium-poor or hydrogen-bearing mixed atmospheres, and the DAQ/DQA continuum; structurally they define an ultramassive, carbon-rich sequence; and evolutionarily they are among the most compelling observed candidates for long-lived white-dwarf merger remnants trapped near the onset of crystallization.