betaFeedback

WorksPhysics and Astronomy

ZTF J2229+1928: A 23 Minute Eclipsing Double White Dwarf Binary

Abstract

We report the discovery of ZTF J2229+1928, a detached and eclipsing DA double white dwarf binary in a 23-minute orbit. The system was identified in a GPU-accelerated search of ZTF photometry and confirmed through follow-up observations with the high speed imager proto-Lightspeed on the 6.5-meter Magellan Clay telescope. Combining the proto-Lightspeed eclipse model with mass-radius relations gives component radii and masses of R1=0.0146−0.0008+0.0009 R⊙R_1 = 0.0146^{+0.0009}_{-0.0008} \, R_\odot, R2=0.0186−0.0008+0.0010 R⊙R_2 = 0.0186^{+0.0010}_{-0.0008} \, R_\odot, and M1=0.49−0.07+0.08 M⊙M_1 = 0.49^{+0.08}_{-0.07} \, M_\odot, M2=0.37−0.05+0.06 M⊙M_2 = 0.37^{+0.06}_{-0.05} \, M_\odot. Fitting the Keck/LRIS spectra gives component temperatures of approximately 11,10011,100 K and 8,2008,200 K, and scaling this model to the broadband spectral energy distribution gives a distance of ∼965\sim965 pc. Its short orbital period places it firmly in the millihertz gravitational-wave regime, and we estimate an expected signal-to-noise ratio SNR≈5.3\mathrm{SNR} \approx 5.3 with the upcoming LISA mission over its nominal four-year duration. ZTF J2229+1928 thus adds to the small but growing sample of eclipsing ultracompact binaries that will serve as important sources for future space-based gravitational-wave observatories.

Introduction

The growing sample of compact double white dwarf (DWD) binaries has transformed our ability to study both the outcomes of stellar binary evolution and the progenitors of gravitational-wave sources detectable by the upcoming Laser Interferometer Space Antenna (LISA). Over the past decade, large spectroscopic efforts such as the ELM Survey and SPY have identified more than a hundred close double white dwarf binaries (e.g., [5] (Napipowtzki et al. 2020)). More recently, wide-field time-domain surveys have expanded the sample to include several eclipsing systems, which provide valuable additional constraints on the properties of both components in the binary. In particular, the Zwicky Transient Facility (ZTF; [4, 20, 32]) has discovered several short-period detached eclipsing DWDs (Burdge et al. 2019, 2020a,b; Coughlin et al. 2020; van Roestel et al. 2025), as well as eclipsing AM CVn systems and other accreting ultracompact binaries whose donors may be white dwarfs ([45, 10]).

In this paper, we report the discovery of ZTF J2229+1928, a detached and eclipsing DWD with an orbital period of just 23.4 minutes. We describe the discovery of ZTF J2229+1928 from ZTF photometry (§ 2), present high-speed follow-up with proto-Lightspeed on Magellan (§ 2.2), and model the

high-speed imaging light curves, Keck/LRIS spectra, and spectral energy distribution to measure the parameters of the system (§ 3). We then compute the expected gravitational-wave signal with LISA and discuss the orbital decay and fate of the system (§ 4). Finally, in § 5, we summarize our findings and outline prospects for future characterization of ZTF J2229+1928.

Discovery of ZTF J2229+1928

Detection in ZTF

ZTF J2229+1928 was identified as part of a systematic search for short-period eclipsing systems in photometric data from the Zwicky Transient Facility (ZTF), which surveys the northern sky with 30 s exposures. We analyzed light curves for 1.3 billion Pan-STARRS sources using a GPU-accelerated implementation of the Box-Least Squares (BLS) algorithm ([26]). Our search flagged ZTF J2229+1928 as one of the most significant short-period candidates, with a strong BLS peak near P≈23.41P \approx23.41 minutes. The period reported in Table 1, P=23.413090±0.000013P = 23.413090 \pm0.000013 minutes, is the location of this BLS peak in the ZTF light curve, with the uncertainty given by the half width at half maximum of the periodogram peak. Unlike many known DWDs, ZTF J2229+1928 is so faint that Gaia does not even detect it as a point source, so it is not listed in published white dwarf catalogs. Nevertheless, because our search was performed over 1.3 billion Pan-STARRS sources rather than a catalog-selected subsample, the system was successfully recovered.

ParameterValue
R.A. (J2000)22:29:16.21 ± 0.01″
Decl. (J2000)+19:28:21.9 ± 0.01″
Pan-STARRS ID131363373176327976
Pan-STARRS gg (mag)21.21 ± 0.05
Pan-STARRS rr (mag)21.14 ± 0.04
Period (minutes)23.413090(13)
R1/aR_1/a0.057−0.003+0.0040.057^{+0.004}_{-0.003}
R2/aR_2/a0.072−0.003+0.0020.072^{+0.002}_{-0.003}
ii (deg)88.64−0.14+0.1488.64^{+0.14}_{-0.14}
T0T_0 (BJD − 2400000.5)60933.132815(3)
J2/J1J_2/J_10.351−0.015+0.0150.351^{+0.015}_{-0.015}
M1M_1 (M⊙M_\odot)0.49−0.07+0.080.49^{+0.08}_{-0.07}
M2M_2 (M⊙M_\odot)0.37−0.05+0.060.37^{+0.06}_{-0.05}
R1R_1 (R⊙R_\odot)0.0146−0.0008+0.00090.0146^{+0.0009}_{-0.0008}
R2R_2 (R⊙R_\odot)0.0186−0.0008+0.00100.0186^{+0.0010}_{-0.0008}
aa (R⊙R_\odot)0.257−0.007+0.0070.257^{+0.007}_{-0.007}
R1/RL,1R_1/R_{\mathrm{L},1}0.140−0.013+0.0160.140^{+0.016}_{-0.013}
R2/RL,2R_2/R_{\mathrm{L},2}0.204−0.017+0.0160.204^{+0.016}_{-0.017}

Table 1. Parameters for ZTF J2229+1928. WD 1 is the hotter white dwarf and WD 2 is the cooler one. The fitted ellc parameters are the fractional radii R1/aR_1/a and R2/aR_2/a in units of the orbital separation, inclination ii, mid-eclipse time T0T_0, and surface brightness ratio J2/J1J_2/J_1. The Roche lobe filling factors R1/RL,1R_1/R_{\mathrm{L},1} and R2/RL,2R_2/R_{\mathrm{L},2} give the ratio of each white dwarf radius to its Roche lobe radius, calculated using Eggleton [16]. The masses, radii, and orbital separation are derived using mass-radius relations.

Confirmation with proto-Lightspeed

On UT 2025 September 15, we obtained high-speed unfiltered photometric follow-up of ZTF J2229+1928 with proto-Lightspeed during a commissioning run on the Magellan Clay telescope ([29]). The observations ran continuously for approximately 4.7 hours, yielding 3350 consecutive 5-second exposures that captured multiple consecutive eclipses. proto-Lightspeed records each observation as a merged data cube, in which thousands of short-exposure frames are stacked with GPS-synchronized timestamps. We extracted the individual frames and performed aperture photometry on ZTF J2229+1928 and nearby comparison stars using a custom photometry pipeline that has also been used in recent work (e.g., [12]). This pipeline includes the nonlinearity calibration presented in [29], which corrects the flux measured in each pixel to account for photogenerated charge that is trapped during readout. A differential light curve was constructed by dividing the target flux by the combined flux of stable comparison stars to correct for transparency variations. We also rejected outliers by removing data points with flux uncertainties larger than the median uncertainty plus 3σ3\sigma, which removed 73 of the 3350 data points. The final differential light curve clearly shows repeated narrow eclipses at the period recovered in ZTF (Figure 1).

Figure 1. Top) High-speed follow-up light curve of ZTF J2229+1928 (green) obtained with proto-Lightspeed on Magellan [29 b], phase-folded to the period recovered with the ZTF search (P≈23.41P \approx23.41 minutes). The best fit ellc model is plotted in black. Bottom) ZTF g-band (top green) and r-band (bottom red) light curves of ZTF J2229+1928 used to discover the object, phase-folded to the same period.

Modeling

Light Curve Modeling

We modeled the proto-Lightspeed light curve using ellc ([33]). The orbital period was fixed to the value measured from ZTF, while both stars were assumed to be spherical since neither star is close to filling its Roche lobe. We did not include Doppler beaming, irradiation, or gravity darkening, and the spherical model does not include ellipsoidal modulation, because the expected signals are too small to detect in these data. For instance, irradiation is expected to be the largest of these effects for the properties of our white dwarfs, but it is still too small to detect: assuming the temperatures of 11,100 K and 8,200 K measured in § 3.2 and that all of the light from the hotter white dwarf incident on its companion is reprocessed, the irradiation amplitude is only ∼\sim1,100 ppm. We adopted a linear limb-darkening law with coefficients from the DA white dwarf tables of [13], averaging the Sloan g′g' and r′r' values at each star’s LRIS temperature and surface gravity because the proto-Lightspeed data are unfiltered ([29]). We fixed the coefficients to u1=0.362u_1 = 0.362 for the hotter white dwarf (WD 1) and u2=0.479u_2 = 0.479 for the cooler white dwarf (WD 2).

The fitted parameters of our ellc fit were R1/aR_1/a, R2/aR_2/a, the orbital inclination ii, the primary eclipse time T0T_0, and the surface brightness ratio J2/J1J_2/J_1. The posterior distributions were sampled using the emcee Markov chain Monte Carlo sampler ([18]) with 32 walkers and 5000 steps. We started the walkers from a tight distribution around a best-fitting preliminary parameter estimate, discarded the first 1000 steps as burn-in, and assessed convergence with the Gelman-Rubin statistic ([19]). Because the light curve constrains only the fractional radii and inclination, a mass-radius relation for each star is needed to determine the absolute radii and individual masses. For the hotter and higher mass white dwarf, we used the carbon-oxygen core, thick hydrogen layer cooling sequences of [3], whose grid covers 0.2 to 1.3 M⊙M_\odot. For the cooler white dwarf, we used the helium-core cooling tracks of [1], because white dwarfs below about 0.45 M⊙M_\odot are expected to have helium cores. These tracks cover 0.155 to 0.435 M⊙M_\odot, so above 0.435 M⊙0.435\,M_\odot we extrapolated linearly from the two heaviest tracks. For each posterior sample, we drew each LRIS temperature with 300 K Gaussian scatter and used the cooling models to relate radius to mass and temper-

ature. We also applied a 10 percent fractional radius model error draw, following the value adopted for white dwarf mass-radius relations by Burdge et al. [8]. The masses were then solved so that the model radii matched (R1/a)a(R_1/a)a and (R2/a)a(R_2/a)a, with aa set by Kepler’s third law for the total mass and fixed period. Roche lobe filling factors were calculated using the Eggleton [16] approximation. The posteriors are shown in Figure 2 and Table 1. Importantly, we note that the quoted uncertainties in Table 1 are the statistical posteriors conditional on the adopted mass-radius relations and do not include systematic uncertainty from the choice of cooling model.

Posterior distributions from ellc light curve modeling

Figure 2. Posterior distributions from our ellc light curve modeling of ZTF J2229+1928. The five sampled parameters are the fractional radii of the two components R1/aR_1/a and R2/aR_2/a (each in units of the orbital separation aa), the orbital inclination ii in degrees, the mid-eclipse time T0T_0, and the surface brightness ratio s=J2/J1s = J_2/J_1. The linear limb-darkening coefficients were held fixed and are not shown. The remaining parameters are derived from each posterior sample using the carbon-oxygen core, thick-hydrogen cooling sequences of Bédard et al. [3] for WD 1 and the helium-core cooling tracks of Althaus et al. [1] for WD 2, evaluated at temperatures drawn with 300 K scatter and with a 10 percent radius model error (Burdge et al. [8]), together with Kepler’s third law and the Eggleton [16] Roche lobe approximation (i.e., the component masses M1M_1 and M2M_2, the component radii R1R_1 and R2R_2 and the orbital separation aa in solar radii, and the Roche lobe filling factors R1/RL,1R_1/R_{L,1} and R2/RL,2R_2/R_{L,2}).

Keck/LRIS Spectroscopy

We obtained phase-resolved spectroscopy of ZTF J2229+1928 with the Low Resolution Imaging Spectrometer (LRIS; Oke et al. [37]) on the 10-m Keck I telescope on UT 2026 July 7. We acquired 22 blue-arm spectra with the 600/4000 grism and 22 red-arm spectra with the 600/7500 grating (120 s exposures, 1.0 arcsec longslit, D560 dichroic), covering approximately 2.4 cycles of the P≈23.41P \approx23.41 minute orbit. The spectra were reduced and flux calibrated with the automated LRIS pipeline lpipe (Perley [38]).

The blue arm covers the Balmer series from Hβ\beta through H8, with DA Balmer absorption and no emission lines, but the components are not resolved into separate line sets and yield no radial velocities.

We built a deep stack of 14 blue-arm spectra away from the two conjunctions, interpolating each frame onto a common wavelength grid and coadding them with inverse-variance weighting. To measure the component temperatures, we fit this stack with a composite of two DA atmosphere models from Koester [25], the two-component modeling of Burdge et al. [8] and the grid-based chi-squared matching of atmosphere models to an LRIS spectrum in Householder et al. [22]. The model grid is spaced by 250 K in effective temperature and 0.25 dex in surface gravity, so we interpolated linearly in both to make sure that the fit was not restricted to the grid points. For each pair of trial temperatures we built the composite model from the two interpolated components. We convolved each component to the instrumental resolution of R≈1000R \approx1000, broadened it by the rotational velocity of a tidally locked star of that radius, and weighted it by the square of its radius, which sets the relative contribution of each star to the combined light. We then Doppler shifted each component to the mid-exposure orbital phase of every contributing frame, using radial-velocity semi-amplitudes and surface gravities fixed at the median values implied by the light curve modeling. Next, we averaged the shifted models over the same 14 frames that were included in the stack, so the model carried the same orbital smearing as the data. We also normalized the local continuum in each Balmer line window with an iterative linear fit that rejects the absorption cores, following a similar approach to Householder et al. [23]. We then applied the identical normalization to the model and evaluated the fit on the normalized spectra, so the temperatures were constrained by the line shapes rather than the continuum calibration. Finally, we fit the spectra with a Gaussian prior from the light curve surface brightness ratio, which helps separate the temperatures of the two stars, and measured T1≈11,100T_1 \approx11{,}100 K and T2≈8,200T_2 \approx8{,}200 K (Figure 3). The uncertainties on these temperatures are dominated by choices made in the fit rather than by photon noise, and are difficult to quantify, so rather than report uncertainties we simply regard both temperatures as accurate to within a few hundred kelvin. We also attempted to measure radial velocities from the individual spectra, but no coherent velocity variations were recovered. In particular, no S-wave from either component is apparent in the trailed spectrogram (Appendix § 7). Measuring the radial velocities of this source will thus likely require more telescope time on a 10-m class telescope, or extremely large telescopes.

Figure 3. LRIS spectroscopy and spectral energy distribution (SED) of ZTF J2229+1928 with the best-fitting two-component model. Left) a deep stack of the 14 out-of-conjunction LRIS spectra (black) with the composite model (orange), white dwarf 1 (WD 1, blue dashed), and white dwarf 2 (WD 2, pink dashed), with the plotted models averaged over the same 14 orbital phases. Right) archival SED photometry (black points, each band labeled) with the composite LRIS fit scaled to the photometry (gray) and the implied distance. Residuals are shown beneath each panel.

Estimating the distance to ZTF J2229+1928

ZTF J2229+1928 does not have a measured Gaia parallax, so its distance cannot currently be determined from astrometry. We therefore estimated the distance by fitting its SED using photometry retrieved through VizieR [36], dereddened with E(B−V)=0.043E(B - V) = 0.043 from the Schlegel et al. (1998) dust map with the Schlafly & Finkbeiner (2011) recalibration and the Cardelli et al. (1989) extinction law with RV=3.1R_V = 3.1. We adopted the GALEX NUV, SDSS uu, and Pan-STARRS grigri measurements, and excluded points with only upper limits or large fractional uncertainties (Figure 3). Because this is a binary system, the observed SED was fit by the sum of two white dwarf spectra:

Fν(λ)=(R1d)2Fν,1(λ)+(R2d)2Fν,2(λ),(1)F_\nu(\lambda) = \left(\frac{R_1}{d}\right)^2 F_{\nu,1}(\lambda) + \left(\frac{R_2}{d}\right)^2 F_{\nu,2}(\lambda), \tag*{(1)}

where Fν,1(λ)F_{\nu,1}(\lambda) and Fν,2(λ)F_{\nu,2}(\lambda) are the surface fluxes from hydrogen-atmosphere (DA) white dwarf models from Koester (2010), R1R_1 and R2R_2 are the component radii from the light curve modeling, and dd is the distance to the system. The surface gravities for each component were computed from the MCMC posteriors, and the models were interpolated to those values. With the radii and the temperatures held at their fitted values, the distance is the only free parameter. We fit the dereddened GALEX NUV, SDSS uu, and Pan-STARRS gg, rr, and ii fluxes to the two-component model by minimizing the chi-squared (χ2\chi^2) statistic, with each data point weighted by one over the square of its photometric uncertainty. From this fit, we obtained d≈965d \approx965 pc.

Discussion

Gravitational-Wave Prospects

Compact double white dwarfs like ZTF J2229+1928 are among the strongest persistent gravitational-wave sources in the Galaxy. Their small orbital separations and high orbital frequencies place them squarely in the millihertz regime where the upcoming LISA mission is most sensitive. We estimated the expected LISA signal-to-noise ratio (SNR) for ZTF J2229+1928 using Monte Carlo sampling over the posterior distributions of the system parameters derived in § 3 (M1M_1, M2M_2, ii) at a distance of d=965d = 965 pc from the SED fit. For each posterior draw, we computed the chirp mass

Mc=(M1M2)3/5(M1+M2)1/5(2)\mathcal{M}_c = \frac{(M_1 M_2)^{3/5}}{(M_1 + M_2)^{1/5}} \tag*{(2)}

and the frequency-domain gravitational-wave amplitude

A=5241π2/3(GMc)5/6c3/2dfGW−7/6,(3)\mathcal{A} = \sqrt{\frac{5}{24}} \frac{1}{\pi^{2/3}} \frac{(G\mathcal{M}_c)^{5/6}}{c^{3/2}d} f_{\mathrm{GW}}^{-7/6}, \tag*{(3)}

where GG is the gravitational constant, cc is the speed of light, dd is the distance, and fGWf_{\mathrm{GW}} is the gravitational-wave frequency. We then averaged over polarization angle and the LISA antenna pattern to obtain the effective signal power, and compared it against the analytic LISA noise model of Robson et al. [40], assuming a four-year observation baseline. At a gravitational-wave frequency of fGW=2/P=1.42×10−3 Hzf_{\mathrm{GW}} = 2/P = 1.42 \times10^{-3}\,\mathrm{Hz}, we find a characteristic strain of hc=2.5×10−20h_c = 2.5 \times10^{-20}, corresponding to a four-year SNR=5.3−0.9+1.0\mathrm{SNR} = 5.3^{+1.0}_{-0.9} (Figure 4). The predicted four-year SNR falls within the SNR≈5\mathrm{SNR} \approx5–66 range commonly adopted for individually detectable LISA sources [28, 17]. Because the SNR grows as the square root of the observing time, a longer mission would increase the expected SNR.

Predicted characteristic strain of ZTF J2229+1928 and known ultracompact binaries

Figure 4. Predicted characteristic strain of ZTF J2229+1928 (pink). Other known ultracompact binaries compiled by Chickles et al. [11] are plotted in gray. The two gray dashed curves mark the characteristic strain at which a monochromatic source reaches a signal-to-noise ratio of 10 and 100 over a four-year LISA mission. We computed these curves from the Robson et al. [40] noise model, including Galactic confusion noise, using hc=SNRfSn(f)h_c = \mathrm{SNR}\sqrt{fS_n(f)}.

Orbital Decay and Evolutionary Fate

For a circular orbit, gravitational radiation alone predicts P˙≈−4.0×10−12 s s−1\dot{P} \approx-4.0 \times10^{-12}\,\mathrm{s\,s^{-1}} Peters [39] for the properties of ZTF J2229+1928. The corresponding decay timescale, (3/8)P/∣P˙∣(3/8)P/|\dot{P}|, is roughly 4.2 Myr, so ZTF J2229+1928 reaches contact well within a Hubble time. What remains uncertain is whether the system survives the onset of mass transfer as a stable accreting binary or instead merges. At Roche-lobe contact, material leaving the donor through the inner Lagrangian point follows a ballistic path whose closest approach to the accretor’s center is about 0.0029 R⊙0.0029\,R_\odot, assuming the component masses and radii remain fixed [35]. Because this distance is smaller than the 0.0146 R⊙0.0146\,R_\odot radius of the 0.49 M⊙0.49\,M_\odot white dwarf, the stream strikes the accretor instead of orbiting it to form a disk, so mass transfer is expected to begin through direct impact [31]. If mass transfer through this direct-impact phase remains stable, the orbit may widen as mass moves to the accretor, and once the stream no longer strikes the accretor a disk may form [31]. If it does not remain stable, the donor is disrupted and the binary merges. In the models of Marsh et al. [31] with a 10 yr synchronization timescale, binaries with accretors near 0.49 M⊙0.49\,M_\odot are stable only for donors below about 0.24 M⊙0.24\,M_\odot. In the models of Kremer et al. [27] with a 10 yr synchronization timescale for both stars at contact, binaries with accretors near 0.49 M⊙0.49\,M_\odot remain stable for donors up to about 0.32 M⊙0.32\,M_\odot. The median masses of 0.49+0.37 M⊙0.49+0.37\,M_\odot (q≈0.76q \approx0.76) therefore give unstable mass transfer in both models, while the lower 1σ1\sigma donor mass of 0.32 M⊙0.32\,M_\odot reaches the stable region of Kremer et al. [27]. Furthermore, even a binary that is considered stable in these models may still merge. For the 0.2+0.50.2+0.5 and 0.2+1.0 M⊙0.2+1.0\,M_\odot binaries modeled by Shen [44], drag during the first nova drives mergers within hours, leading Shen [44] to argue that most, if not all, interacting double white dwarfs likely merge. In the merger simulations of Dan et al. [15], a 0.49+0.37 M⊙0.49+0.37\,M_\odot binary lies outside the region that produces a helium detonation, so a Type Ia supernova is unlikely. Such mergers may produce R Coronae Borealis stars or hot subdwarfs [43, 48], with the subsequent evolution onto the white dwarf cooling track described by Schwab [43].

Conclusion

In this paper, we have presented the discovery and characterization of ZTF J2229+1928, a detached and eclipsing double white dwarf in a 23.4 minute orbit. Our main results are summarized below:

  • ZTF J2229+1928 was identified in a GPU-accelerated search of ZTF photometry and confirmed with high-speed follow-up observations using proto-Lightspeed on Magellan.

  • The high-speed light curve constrains the fractional radii and orbital inclination. Combining these constraints with the finite-temperature cooling models of Bédard et al. [3] for WD 1 and Althaus et al. [1] for WD 2 yields the component masses and absolute radii in Table 1.

  • Fitting the Keck/LRIS spectra with a Gaussian prior on the light curve surface brightness ratio gives effective temperatures of T1≈11,100 KT_1 \approx11{,}100\,\mathrm{K}

and T2≈8,200T_2 \approx8{,}200 K, and scaling the two-component model to the broadband SED gives a distance of ≈965\approx965 pc.

  • At a gravitational-wave frequency of fGW=1.42×10−3f_{\mathrm{GW}} = 1.42 \times10^{-3} Hz, we predict a characteristic strain of hc=2.5×10−20h_c = 2.5 \times10^{-20} and a four-year SNR =5.3−0.9+1.0= 5.3^{+1.0}_{-0.9}, making this system a promising target for the upcoming LISA mission.

Future follow-up will be crucial for characterizing this system. We are continuing high-speed photometric monitoring to measure the orbital period derivative, which traces the rate of orbital decay from gravitational-wave emission. Additional spectroscopic observations with large-aperture facilities could also help further characterize the properties of the DWDs, though the faintness of this system will likely make such efforts time-intensive.

Facilities: PO:1.2m (ZTF), Magellan:Clay (proto-Lightspeed), Keck:I (LRIS)

Software: astropy ([2]), numpy ([21]), scipy ([47]), matplotlib ([24]), ellc ([33]), emcee ([18]), LPipe ([38])

Acknowledgments

This work was supported by the MIT Undergraduate Research Opportunities Program (UROP). This material is also based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. 2141064 and the MIT Dean of Science Fellowship. We also acknowledge the use of computational resources provided by the MIT Engaging cluster at the Massachusetts Green High Performance Computing Center (MGHPCC).

This paper includes data gathered with the 6.5 meter Magellan Telescopes located at Las Campanas Observatory, Chile. This work was supported by a NASA Keck PI Data Award, administered by the NASA Exoplanet Science Institute. Data presented herein were obtained at the W. M. Keck Observatory from telescope time allocated to the National Aeronautics and Space Administration through the agency’s scientific partnership with the California Institute of Technology and the University of California. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation.

The authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of Maunakea has always had within the Native Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain.

This work is based on observations obtained with the Samuel Oschin Telescope 48-inch and the 60-inch Telescope at the Palomar Observatory as part of the Zwicky Transient Facility project. ZTF is supported by the National Science Foundation under Grants No. AST-1440341 and AST-2034437 and a collaboration including current partners Caltech, IPAC, the Weizmann Institute of Science, the Oskar Klein Center at Stockholm University, the University of Maryland, Deutsches Elektronen-Synchrotron and Humboldt University, the TANGO Consortium of Taiwan, the University of Wisconsin at Milwaukee, Trinity College Dublin, Lawrence Livermore National Laboratories, IN2P3, University of Warwick, Ruhr University Bochum, Northwestern University and former partners the University of Washington, Los Alamos National Laboratories, and Lawrence Berkeley National Laboratories. Operations are conducted by COO, IPAC, and UW.

Trailed Spectrogram

Figure 5 shows a trailed spectrogram of the 22 LRIS blue-arm spectra, built by interpolating the five blue Balmer lines (Hβ\mathrm{H}\beta through H8) of each frame onto a common velocity grid and coadding them in 12 orbital phase bins. The radial-velocity curves of both components predicted from the light curve masses are overlaid. No coherent S-wave from either component is apparent at the signal-to-noise ratio of these data.

Figure 5. Trailed spectrogram of the 22 LRIS blue-arm spectra, coadding the Hβ\mathrm{H}\beta through H8 lines in velocity space and repeated over two orbital cycles. Darker shades are deeper absorption. The dotted curves show the radial-velocity curves of WD 1 (blue, K1=341 km s−1K_1 = 341\ \mathrm{km\ s^{-1}}) and WD 2 (green, K2=459 km s−1K_2 = 459\ \mathrm{km\ s^{-1}}) predicted from the light curve masses. No coherent S-wave from either component is detected, likely because of the low signal-to-noise ratio of the individual spectra.

References

  1. [1]Althaus, L. G., Miller Bertolami, M. M., & Córsico, A. H. 2013, A&A, 557, A19, doi: 10.1051/0004-6361/201321868DOI
  2. [2]Astropy Collaboration, Price-Whelan, A. M., Lim, P. L., et al. 2022, ApJ, 935, 167, doi: 10.3847/1538-4357/ac7c74
  3. [3]Bédard, A., Bergeron, P., Brassard, P., & Fontaine, G. 2020, ApJ, 901, 93, doi: 10.3847/1538-4357/abafbeDOI
  4. [4]Bellm, E. C., Kulkarni, S. R., Graham, M. J., et al. 2019, PASP, 131, 018002, doi: 10.1088/1538-3873/aaecbeDOI
  5. [5]Brown, W. R., Kilic, M., Kosakowski, A., et al. 2020, ApJ, 889, 49, doi: 10.3847/1538-4357/ab63cdDOI
  6. [8]Burdge, K. B., Prince, T. A., Fuller, J., et al. 2020b, ApJ, 905, 32, doi: 10.3847/1538-4357/abc261DOI
  7. [10]Chakraborty, J., Burdge, K. B., Rappaport, S. A., et al. 2024, ApJ, 977, 262, doi: 10.3847/1538-4357/ad9563DOI
  8. [11]Chickles, E. T., Burdge, K. B., Chakraborty, J., et al. 2025, ApJ, 987, 206, doi: 10.3847/1538-4357/add34cDOI
  9. [12]Chickles, E. T., Chakraborty, J., Burdge, K. B., et al. 2026, ApJ, 1000, 237, doi: 10.3847/1538-4357/ae4871DOI
  10. [13]Claret, A., Cukanovaite, E., Burdge, K., et al. 2020, A&A, 634, A93, doi: 10.1051/0004-6361/201937326DOI
  11. [15]Dan, M., Rosswog, S., Guillochon, J., & Ramirez-Ruiz, E. 2012, MNRAS, 422, 2417, doi: 10.1111/j.1365-2966.2012.20794.xDOI
  12. [16]Eggleton, P. P. 1983, ApJ, 268, 368, doi: 10.1086/160960
  13. [17]Finch, E., Bartolucci, G., Chucherkо, D., et al. 2023, MNRAS, 522, 5358, doi: 10.1093/mnras/stad1288DOI
  14. [18]Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306, doi: 10.1086/670067
  15. [19]Gelman, A., & Rubin, D. B. 1992, Statistical Science, 7, 457, doi: 10.1214/ss/1177011136DOI
  16. [20]Graham, M. J., Kulkarni, S. R., Bellm, E. C., et al. 2019, PASP, 131, 078001, doi: 10.1088/1538-3873/ab006cDOI
  17. [21]Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357, doi: 10.1038/s41586-020-2649-2DOI
  18. [22]Householder, A., Shin, K., Burdge, K. B., et al. 2026a, Nature Astronomy, doi: 10.1038/s41550-026-02992-6DOI
  19. [23]Householder, A., Dai, F., Kesseli, A., et al. 2026b, AJ, 172, 48, doi: 10.3847/1538-3881/ae71b9DOI
  20. [24]Hunter, J. D. 2007, Computing in Science and Engineering, 9, 90, doi: 10.1109/MCSE.2007.55DOI
  21. [25]Koester, D. 2010, Mem. Soc. Astron. Italiana, 81, 921
  22. [26]Kovács, G., Zucker, S., & Mazeh, T. 2002, A&A, 391, 369, doi: 10.1051/0004-6361:20020802DOI
  23. [27]Kremer, K., Breivik, K., Larson, S. L., & Kalogera, V. 2017, ApJ, 846, 95, doi: 10.3847/1538-4357/aaa8557
  24. [28]Kupfer, T., Korol, V., Shah, S., et al. 2018, MNRAS, 480, 302, doi: 10.1093/mnras/sty1545DOI
  25. [29]Layden, C., Burdge, K., Furesz, G., et al. 2026a, PASP, 138, 065001, doi: 10.1088/1538-3873/ae6dfcDOI
  26. [31]Marsh, T. R., Nelemans, G., & Steeghs, D. 2004, MNRAS, 350, 113, doi: 10.1111/j.1365-2966.2004.07564.xDOI
  27. [32]Masci, F. J., Laher, R. R., Rusholme, B., et al. 2019, PASP, 131, 018003, doi: 10.1088/1538-3873/aae8acDOI
  28. [33]Maxted, P. F. L. 2016, A&A, 591, A111, doi: 10.1051/0004-6361/201628579DOI
  29. [35]Nelemans, G., Portegies Zwart, S. F., Verbunt, F., & Yungelson, L. R. 2001, A&A, 368, 939, doi: 10.1051/0004-6361:20010049DOI
  30. [36]Ochsenbein, F., Bauer, P., & Marcout, J. 2000, A&AS, 143, 23, doi: 10.1051/aas:2000169DOI
  31. [37]Oke, J. B., Cohen, J. G., Carr, M., et al. 1995, PASP, 107, 375, doi: 10.1086/133562DOI
  32. [38]Perley, D. A. 2019, PASP, 131, 084503, doi: 10.1088/1538-3873/ab215dDOI
  33. [39]Peters, P. C. 1964, Physical Review, 136, B1224, doi: 10.1103/PhysRev.136.B1224DOI
  34. [40]Robson, T., Cornish, N. J., & Liu, C. 2019, Classical and Quantum Gravity, 36, 105011, doi: 10.1088/1361-6382/ab1101DOI
  35. [43]Schwab, J. 2019, ApJ, 885, 27, doi: 10.3847/1538-4357/ab425dDOI
  36. [44]Shen, K. J. 2015, ApJL, 805, L6, doi: 10.1088/2041-8205/805/1/L6DOI
  37. [45]van Roestel, J., Kupfer, T., Green, M. J., et al. 2022, MNRAS, 512, 5440, doi: 10.1093/mnras/stab2421DOI
  38. [47]Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature Methods, 17, 261, doi: 10.1038/s41592-019-0686-2DOI
  39. [48]Zhang, X., & Jeffery, C. S. 2012, MNRAS, 419, 452, doi: 10.1111/j.1365-2966.2011.19711.xDOI

Paper details

vlnc:W_5pmad2vf
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents