Periodic Radio Emission from the T8 Dwarf WISE J062309.94–045624.6

We performed a search for highly circularly polarized objects in the RACS midband (1.36 GHz) data (RACS-mid; Duchesne et al. 2023). The 15 minute RACS-mid observations cover the whole sky south of decl. = +49° (covering 36,449 deg²) with a median angular resolution of ∼10'' and a median sensitivity of ∼0.15–0.40 mJy beam⁻¹. We used the circular polarization method presented by Pritchard et al. (2021) to identify interesting sources. We identified ASKAP J062309.2−0456227 for further investigation because we did not find a nearby (≲5'') positional crossmatch to any known astronomical objects.

ASKAP J062309.2−0456227 was detected in RACS-mid on MJD59216 with a time and frequency averaged Stokes I flux density of 4.17 ± 0.41 mJy beam⁻¹, with a beam size of 95 × 74, and an absolute fractional circular polarization of f_cp = 66.3% ± 9.0%. There was no source detected within 120'' of these coordinates in the RACS-low survey (0.88 GHz) with a 5σ flux density limit of 1.81 mJy beam⁻¹; see Table 1 for details of these radio observations.

We observed ASKAP J062309.2−0456227 with ATCA in the L band (1.1–3.1 GHz) on MJD 59805 for 6 hr using the hybrid H168 array configuration and an additional 11 hr on MJD 59929 in the extended 6C configuration (C3363). For both observations we used the ATCA calibrator source PKS 1934-638 as the primary flux calibrator and the calibrator source [HB89] 0607-157 for phase calibration scans.

We reduced the data from these observations using the Miriad software (Sault et al. 1995) and imaged using the Common Astronomy Software Application software (CASA; CASA Team et al. 2022). We used tclean with briggs weighting and a robust parameter of 0.5, with the multiscale multiterm multifrequency synthesis (mtmfs) deconvolver and clean scales of 0, 4, and 8 pixels. The MJD 59805 observation had limited uv coverage due to the hybrid array configuration and the source's proximity to the celestial equatorial, resulting in an extended PSF that confused ASKAP J062309.2−0456227 with nearby sources. This observation was dominated by artifacts from bright off-axis emission. These artifacts are the result of bright sources in the field, as well as a resolved source ∼135'' from WISE J062309.94−045624.6. The MJD 59929 observation produced more extensive coverage of uv space and did not suffer from the same PSF and source confusion issues as the MJD 59805 observation.

In the MJD 59805 ATCA observation we detected a time and frequency averaged Stokes I emission 0.39 ± 0.06 mJy beam⁻¹, with a beam size of 108 × 26, and marginal evidence of periodicity identified in the dynamic spectrum of the observation. We did not measure a clear Stokes V flux for the MJD 59805 ATCA observation. Our MJD 59929 observation produced a source detection with a time and frequency averaged Stokes I flux density of 0.30 ± 0.03 mJy beam⁻¹, with a beam size of 531 × 33, and an absolute fractional circular polarization of f_cp = 66.0% ± 11.9%. Figure 1 shows the Stokes V dynamic spectrum as well as Stokes I and V lightcurves from the MJD 59929 ATCA observation. We see clear evidence of highly circularly polarized periodic pulsed emission.

Zoom In Zoom Out Reset image size

We observed ASKAP J062309.2−0456227 with MeerKAT (Jonas & MeerKAT Team 2016) in the L band (0.86–1.71 GHz) on MJD 60030 in the c856M4k configuration with the standard 8 s integration time. We used the source PKS J0408-6544 as the primary flux calibrator, PKS J0521+1638 as the polarization calibrator, and the source PKS J0609-1542 for phase calibration scans. We reduced and imaged the data from this observation with the oxkat pipeline (Heywood 2020) and used the IDIA (Inter-University Institute for Data Intensive Astronomy) processMeerKAT pipeline (Collier et al. 2021) for cross polarization calibration. We used tclean with briggs weighting and a robust parameter of 0.0, with the multiscale multiterm multifrequency synthesis (mtmfs) deconvolver and clean scales of 0, 5, 10, and 15 pixels. The MeerKAT observation produced a source detection with a time and frequency averaged Stokes I flux density of 1.65 ± 0.17 mJy beam⁻¹, with a beam size of 74 × 64, and an absolute fractional circular polarization of f_cp = 73.8% ± 10.6%. Figure 2 shows the Stokes V dynamic spectrum and lightcurve from the MJD 60030 MeerKAT observation. We see strong evidence of repeating, highly circularly polarized pulses with multiple intrapulse peaks.

Zoom In Zoom Out Reset image size

In RACS-mid, ASKAP J062309.2−0456227 has coordinates R. A._J2000 = 06^h23^m0928, $\mathrm{decl}{.}_{{\rm{J}}2000}=-04^\circ 56^{\prime} 22\buildrel{\prime\prime}\over{.} 8$ (l = 21413, b = − 853) on MJD 59216, with uncertainties of ±2'' in both R.A. and decl. (McConnell et al. 2020; Duchesne et al. 2023). We extended our crossmatch radius and found WISE J062309.94−045624.6, with an on-sky separation of 10'' from ASKAP J062309.2−0456227 in the SIMBAD astronomical database (Wenger et al. 2000). WISE J062309.94−045624.6 has a proper motion of ${\mu }_{\alpha }=-0.93\pm 0.01\,\mathrm{arcsec}\,{\mathrm{yr}}^{-1}$ and ${\mu }_{\delta }=0.17\pm 0.02\,\mathrm{arcsec}\,{\mathrm{yr}}^{-1}$ in R.A. and decl., respectively, in the CatWISE2020 catalog (Marocco et al. 2021). Propagating the WISE object coordinates to the RACS-mid epoch resulted in an offset of 03 from ASKAP J062309.2−0456227. We therefore identified WISE J062309.94−045624.6 as the source of the radio emission.

WISE J062309.94−045624.6 is a high-proper-motion T8 dwarf discovered and spectroscopically confirmed (with IRTF/SpeX) by Kirkpatrick et al. (2011). This UCD has an effective temperature of T_eff = 699 K and is located 11.44 ± 0.38 pc away (Kirkpatrick et al. 2019). Zhang et al. (2021) derived constraints on the age, radius, and mass of WISE J062309.94−045624.6 with Sonora Bobcat models (Marley et al. 2021): $\mathrm{age}={738}_{-592}^{+2701}\,\mathrm{Myr}$, $M={13.18}_{-9.44}^{+31.26}\,{M}_{\mathrm{Jup}}$, $R={0.78}_{-0.13}^{+0.17}\,{R}_{\mathrm{Jup}}$.

WISE J062309.94−045624.6 was not detected in the VLA Sky Survey (VLASS; Lacy et al. 2020) epochs 1.1 and 2.1, with 5σ limits of 0.57 and 0.82 mJy respectively. Looking at the lower 700 MHz of the ATCA observing band (see Figure 1), the pulsed emission stops at a cutoff frequency ∼1.9–2.0 GHz. This is likely the reason for the 3 GHz VLA nondetections listed in Table 1, though the lack of detection could also be due to the short-duration observations in the VLASS on-the-fly observing mode. This also explains the lower time and frequency averaged flux density measured with ATCA compared to ASKAP and MeerKAT, which have narrower bandwidths that are below the cutoff frequency.

We did not identify archival radio observations of WISE J062309.94−045624.6 from ATCA, ASKAP, the Atacama Large Millimeter/submillimeter Array (Wootten & Thompson 2009), or the National Radio Astronomy Observatory (NRAO) VLA Sky Survey (NVSS; Condon et al. 1998). Nor were there detections with the ROentgen SATellite (ROSAT; Truemper 1982) or other X-ray telescopes.

We used a Lomb–Scargle periodogram to identify the dominant period in the Stokes V emission from the ATCA MJD 59929 observation. The pulsed emission has a double-peaked structure (as seen in the lower left panel of Figure 1), which repeats with a period of P = 1.889 ± 0.018 hr at a false alarm probability of <1% (VanderPlas 2018). Peaks within the pulses are separated by 0.971 hr, approximately half the pulse period, which appears as a less prominent peak in the Lomb–Scargle power spectrum. We used similar Lomb–Scargle analysis to find a pulse periodicity of P = 1.912 ± 0.005 hr, at a false alarm probability of <1% (VanderPlas 2018), in the MeerKAT data. The pulsed emission has a complex pulse profile with multiple peaks (as seen in the lower left panel of Figure 2). The leading and trailing peaks of the pulse are separated by 0.942 hr, approximately half the pulse period. False alarm probabilities were calculated using bootstrap resampling (VanderPlas 2018) of the 270 s (ATCA) and 64 s (MeerKAT) time-averaged data. We calculated the period uncertainties using the method presented in Equation (52) of VanderPlas (2018), which assumes that the periods are physical and not aliases. The pulse periodicities observed with ATCA and MeerKAT (see lower panel of Figure 2) are consistent with a P ∼ 1.9 hr period. Figure 3 shows the Lomb–Scargle periodograms and phase-folded lightcurves from the ATCA MJD 59929 Stokes I and Stokes V data and from the MeerKAT Stokes V data. Two phase cycles are shown for clarity and the folded lightcurves are cumulatively binned. Both plots in Figure 3 show that P ∼ 1.9 hr is the dominant period in both the Stokes I and Stokes V emission. A less prominent peak is visible in both periodograms at 0.48 hr. This peak appears to be a harmonic corresponding to a quarter of the period.

Zoom In Zoom Out Reset image size

Based on modeling of WISE J062309.94−045624.6 by Zhang et al. (2021) we assume an upper limit of 0.95 R_Jup for the emission region. Using the RACS-mid flux density, we obtain a lower limit for the brightness temperature of T_b > 2.6 × 10¹² K, which requires a coherent emission mechanism.

Using a distance of 11.44 ± 0.37 pc (Kirkpatrick et al. 2019), the measured RACS-mid flux density corresponds to an isotropic radio luminosity of L_ν ∼ 10^14.8 erg s⁻¹ Hz⁻¹. The peak radio luminosities of UCDs presented by Pineda et al. (2017) range from 10¹³ to 10^15.5 erg s⁻¹ Hz⁻¹. The isotropic radio luminosity we calculate for WISE J062309.94−045624.6 is comparable to the most luminous UCDs of similar spectral types (see Table 2).

Table 2. Known Late-type UCDs with Published Radio Detections

Name	SpT	Teff	${\mathrm{log}}_{10}({L}_{\nu ,\mathrm{avg}})$	${\mathrm{log}}_{10}({L}_{\nu ,\mathrm{peak}})$	Band	References
		(K)	(erg s−1 Hz−1)	(erg s−1 Hz−1)	(GHz)
2MASS J04234858-0414035	L6	1483 ± 113	12.8	13.7	4.0–12.0	Kao et al. (2016, 2018)
2MASS 10430758+2225236	L8	1336 ± 113	12.7	13.4	4.0–12.0	Kao et al. (2016, 2018)
SIMP J013656.5+093347.3	T2.5	1089 ± 62	12.2	13.0	4.0–12.0	Kao et al. (2016, 2018)
WISEP J112254.73+255021.5 a	T6	943 ± 113	...	14.9	4.0–7.0	Route & Wolszczan (2016b)
2MASS J10475385+2124234 b	T6.5	880 ± 76	12.1	13.2	4.0–18.0	Williams & Berger (2015)
2MASS J12373919+6526148	T6.5	851 ± 74	12.7	13.1	4.0–12.0	Kao et al. (2016, 2018)
BDR J1750+3809	T6.5	...	15.0	...	0.12–0.17	Vedantham et al. (2020)
WISEP J101905.63+652954.2 c	T7+T5.5	...	...	14.0	0.12–0.17	Vedantham et al. (2023)
WISE J062309.94−045624.6	T8	699	...	14.8	0.9–2.0	This Work

Notes. We include average and peak radio luminosities, arranged by spectral type (SpT), for dwarfs at the L/T transition (L4–T4) and cooler from Pineda et al. (2017), Vedantham et al. (2020, 2023), and references therein. We list the observing bands the UCDs have been detected.

^a WISEP J112254.73+255021.5 was first detected in radio by Route & Wolszczan (2016b) in the 4.0–6.0 GHz band and then by Williams et al. (2017) in the 5.0–7.0 GHz band. ^b 2MASS J10475385+2124234 was detected in the 5.0–7.0 and 9.0–11.0 GHz bands by Williams & Berger (2015), 4.0–6.0 GHz by Route & Wolszczan (2016a), and 12.0–18.0 GHz by Kao et al. (2018). ^c The latest-type radio-emitting UCD prior to this work is the T7+T5.5 binary detected by Vedantham et al. (2023) at 144 MHz.

Download table as:  ASCIITypeset image

The detected emission from WISE J062309.94−045624.6 has a brightness temperature beyond the 10¹² K coherent limit, implying that it is generated either by plasma radiation or by ECMI. ECMI is the primary source of coherent emission in UCDs, which is both highly circularly polarized and rotationally modulated (Hallinan et al. 2006, 2008). Because the WISE J062309.94−045624.6 emission is strongly circularly polarized and periodic, we favor the ECMI interpretation. This implies that the observed periodicity corresponds to the dwarf's rotation period.

Using a conservative emission upper cutoff of 2.0 GHz from the MJD 59929 ATCA observation, the minimum magnetic field strength in the emission region is B = 2.0/2.8 ≃ 0.71 kG (Dulk 1985). However, the ECMI mechanism could be operating at the first harmonic of the cyclotron frequency rather than the fundamental, in which case the magnetic field lower limit would instead be B ≃ 0.35 kG. The upper cutoff for ECMI emission is typically interpreted to represent the magnetic field strength at the base of the emission region, where the plasma density becomes sufficient to drive the local plasma frequency ν_p above the cyclotron frequency ν_c , and in principle provides a lower limit to the maximum stellar magnetic field strength at the photosphere. Because we detect radio emission up to 2.0 GHz in the MJD 59929 ATCA observation, we used the electron plasma frequency relation: ${\nu }_{p}\approx 9\times {10}^{3}{n}_{e}^{1/2}$ Hz (Dulk 1985) to calculate an upper limit of n_e < 5 × 10¹⁰ cm⁻³ on the local electron density at the base of the emission region. The spectral cutoff may also be due to the plasma frequency, which is dependent on the electron density, reaching the order of the cyclotron frequency at a higher-altitude void in the dwarf's magnetosphere. This would cut off the emission at a lower magnetic field strength and the true maximum magnetic field strength would be larger.

Using the upper limit radius r = 0.95 R_Jup (Zhang et al. 2021), we can use the P = 1.9 hr period to constrain the rotational velocity to v $\sin i\gt 63$ km s⁻¹. This projected rotational velocity is relatively high for a UCD but is still slower than the fastest known rotating T dwarfs found by Tannock et al. (2021) and Vos et al. (2022). The rotational period is reasonable as it is above the 1 hr empirical lower limit for UCD rotation periods (Tannock et al. 2021) and well above the breakup period lower limit ${P}_{\mathrm{break}-\mathrm{up}}={0.58}_{-0.22}^{+0.38}\,\,\mathrm{hr}$. We calculated this breakup limit by taking the parallax ϖ = 86.5 ± 1.7 mas and the Starfish-based values for the surface gravity $\mathrm{log}g={4.70}_{-0.42}^{+0.47}$ dex and solid angle $\mathrm{log}{\rm{\Omega }}=-{19.610}_{-0.156}^{+0.167}$ dex from Zhang et al. (2021). We then computed the breakup period as ${P}_{\mathrm{break}-\mathrm{up}}=\sqrt{\tfrac{4{\pi }^{2}{{\rm{\Omega }}}^{1/2}}{g\varpi G}}$ by using a Monte Carlo estimate for the uncertainty, drawing the values for $\mathrm{log}{\rm{\Omega }}$, $\mathrm{log}g$, and ϖ from independent Gaussians and computing the distribution in values of P_break-up that resulted.

The pulse profile detected in both ATCA and MeerKAT observations of WISE J062309.94−045624.6 has a dominant period of 1.9 hr, which we interpret as the stellar rotation period, and a duty cycle of ∼70%. Each pulse profile features a multipeaked structure with strong leading and trailing pulses separated by a phase of ∼0.5, multiple weaker pulses in between, and an apparent "off" period for the remaining ∼0.5 phase. ECMI radio emission is beamed into a hollow cone aligned with the local magnetic field and with half-angle that is dependent upon the nature of the instability in the electron velocity distribution, typically near perpendicular to the magnetic field (Treumann 2006). Emission from an extended region of the stellar magnetosphere such as an auroral oval is a superposition of individual ECMI sources with complex, overlapping beaming geometry.

The ∼0.5 phase separation of the strongest pulse peaks in our data can then be explained by the leading and trailing edges of an extended ECMI source analogous to Jupiter's main auroral oval (e.g., Cowley & Bunce 2001). This is the expected ECMI source distribution for aurorae that are powered by corotation breakdown between the stellar magnetosphere and ionosphere, which is a plausible scenario given the rapid 1.9 hr rotation period. Alternatively, a localized ECMI source with perpendicular beaming angle may produce strong leading and trailing pulses separated by ∼0.5 phase (e.g., Hallinan et al. 2007), and extension of the source over a small range of stellar longitudes can then explain the weaker intermediate pulses as separate ECMI cones are swept into the line of sight (e.g., Bastian et al. 2022). The pulse duty cycle of ∼70% is much larger than most previously detected UCD auroral pulses, with only a few similar examples such as the L3.5 dwarf 2M J0036+1821 (Hallinan et al. 2008) and the T6 dwarf WISE J1122+2550 (Williams et al. 2017) with duty cycles of ∼30% and ∼50% respectively, which may be explained by unresolved subpulse structure as shown in Figure 2. Modeling of the time-frequency structure of pulses in the dynamic spectra will inform the interpretation of the ECMI source configuration and electrodynamic engine powering the aurorae, and can provide constraints on the rotation and magnetic axis inclinations, which will be the subject of a future publication.

WISE J062309.94−045624.6 is the only T dwarf detected in our untargeted RACS-mid circular polarization search, which will be presented in an upcoming publication. While the radio luminosity distribution of T dwarf auroral pulses is still poorly constrained, to determine an order-of-magnitude estimate of the expected number of T dwarf detections we assume the observed value of L_ν ∼ 10^14.8 erg s⁻¹ Hz⁻¹ is representative of the population. With a 5σ limiting flux density of 1.25 mJy beam⁻¹ in RACS-mid, this corresponds to a detection horizon of 20.54 pc. Best et al. (2021) used a volume-limited 25 pc sample of 190 T dwarfs to calculate a space density of 5.45 ± 0.24 × 10⁻³ T dwarfs (T0–T8) per cubic parsec. We therefore expect a total of 180 ± 10 T dwarfs within the sampled 20.54 pc RACS-mid sensitivity horizon, accounting for the ∼88% RACS-mid sky coverage.

However, not all T dwarfs in a given volume are necessarily radio active. Williams (2018) suggested that the ≤10% detection fraction for UCDs in targeted radio surveys (Route & Wolszczan 2016a) may be due to a similarly small fraction of UCDs having the necessary magnetospheric conditions to produce detectable radio emission. Another factor contributing to the ≤10% detection fraction is the beaming geometry (Pineda et al. 2017), where some fraction of T dwarfs will have rotation and magnetic axes that are unfavorably aligned for the rotationally modulated ECMI beam to cross our line of sight. The overall ∼10% fraction of radio-pulsing UCDs implies that ∼18 T dwarfs within the RACS-mid sensitivity horizon are producing detectable auroral radio pulses.

Finally, radio-pulsing T dwarfs whose emission is favorably beamed may have been missed due to the short duration of RACS-mid observations, which are not capable of sampling the full rotational phase of these systems. Measured T dwarf rotational periods typically range from ∼1 to ∼15 hr (Tannock et al. 2021, and references therein), while the RACS-mid observations cover each candidate radio-loud T dwarf for 15 minutes, implying a rotation phase coverage of as little as ∼2% for slowly rotating T dwarfs to ∼25% for rapidly rotating T dwarfs. As discussed in Section 4.2, the duty cycle of auroral ECMI emission is dependent on multiple factors including magnetic and rotation axis orientation and the distribution of auroral sources in altitude and longitude. While the distribution of these factors in the T dwarf population are still poorly constrained, most auroral pulses detected to date have duty cycles of less than ∼10% (e.g., Kao et al. 2016; Pineda et al. 2017; Williams 2018). Using the average measured T dwarf rotation period of ∼5 hr and assuming a typical duty cycle of ∼10%, we expect a significant number of the estimated ∼18 detectable aurorally active T dwarfs in RACS-mid would be undetected due to limited rotation phase coverage. While a quantitative estimate of the expected detection rates would require stronger constraints on the population parameters of auroral T dwarf activity, we find that the expected number of detections in RACS-mid is of order unity. This is consistent with WISE J062309.94−045624.6 being the only T dwarf we found in our circular polarization search of RACS-mid.