From Dust to Nanodust: Resolving Circumstellar Dust from the Colliding-wind Binary Wolf-Rayet 140

We observed WR 140 (J2000 R.A. 20:20:27.98, decl. +43:51:16.29; Lindegren et al. 2021) on 2020 September 2 UT with SCExAO coupled to the CHARIS integral-field spectrograph operating in low-resolution/broadband mode, covering the major JHK passbands simultaneously (λ_o = 1.16–2.39 μm) and using the ∼0113 radius Lyot coronagraphic mask. Seeing values measured from the Maunakea Weather Center ranged between 05 and 08 but with high (20–25 mph) winds. The observations covered −2.3 to −0.94 in hour angle and were conducted in "pupil tracking"/angular differential imaging mode (Marois et al. 2006) to allow the sky to rotate on the CHARIS detector with time (ΔPA = 30^o). Because WR 140 was slightly off center from the CHARIS detector, we have full 360^o coverage out to ∼096 instead of the more typical 105. For astrometric calibration, we took four exposures with the artificial satellite spots (Jovanovic et al. 2015b) turned on and obtained our other observations without satellite spots. We observed the K0-star HD 194479 afterwards as a PSF reference star and for spectrophotometric calibration.

We used the standard CHARIS pipeline (Brandt et al. 2017) to extract CHARIS data cubes from the raw data. For subsequent data reduction, we used the CHARIS Data Processing Pipeline (Currie et al. 2020, 2023). For spectrophotometric calibration drawn from the HD 194479 data, we adopted a Kurucz stellar atmosphere model appropriate for a K0V star. Simple inspection showed that the "bar" and C1 emission features are bright (Figure 1, top left-hand panel), especially at longer wavelengths, and detected with conservative PSF subtraction methods. Thus, to subtract the PSF, we used a full-frame implementation of reference star differential imaging (RDI) using the Karhunen–Loève Image Projection (KLIP; Soummer et al. 2012) algorithm, as in Currie et al. (2019), retaining only the first three KL modes. Separately, we performed a simple classical spectral differential imaging (SDI) reduction of the post-RDI/KLIP residuals. While classical SDI produced some attenuation of the bar at the shortest wavelengths, it fully preserved the overall shape and morphology of the emission feature with reduced residual speckles. The bright and linear horizontal emission feature at a vertical offset of +10 near the edge of the field of view is an artifact from the data reduction (Figure 1, top left-hand panel).

Zoom In Zoom Out Reset image size

To estimate flux loss in RDI/KLIP-reduced data due to over-subtraction (Pueyo 2016), we inserted the wavelength-collapsed final image produced using classical SDI on the post-RDI/KLIP residual images into our RDI/KLIP sequence. We then forward-modeled this image through our RDI/KLIP sequence to estimate signal loss. Forward-modeling showed that the flux from the bar is well preserved, with a throughput varying from 0.90 to 0.98 across the entire bar morphology and over the full range of CHARIS channels. We therefore conservatively adopt a 1σ photometric uncertainty of 10% across the CHARIS channels.

Flux densities of the C1 and the bar were extracted from the CHARIS data cubes using a circular aperture with a 250 mas radius and a rectangular aperture with a size of 1480 × 300 mas, respectively. The rectangular aperture for the bar was tilted by 17 south of west to match the angle of the bar feature.

Keck/NIRC2 L'-band (λ_c = 3.8 μm) imaging observations of WR 140 were obtained on 2020 August 18 UT in the program C258 (PI—M. Hankins). The vector vortex coronagraph (Vargas Catalán et al. 2016; Serabyn et al. 2017) was used to suppress the diffraction pattern of the star with an inner working angle of λ/D ≈ 80 mas at L'-band. The infrared PyWFS (Bond et al. 2020) was used to obtain good adaptive optics correction because the star is much brighter in the infrared than at visible wavelengths due to extinction. Observations were taken using the narrow camera with a 10'' × 10'' field of view and a pixel scale of 9.94 mas. QACITS, the standard focal plane wave front sensing technique for the vector vortex coronagraph, was used to obtain observations of the system while keeping the star aligned behind the coronagraph (Huby et al. 2017). In total, we obtained 58 coronagraphic images, each consisting of 100 coadds of 0.18 s exposures. As standard in the QACITS sequence, we also obtained sky background images and unocculted images of the star for calibration. In addition to WR 140, we used the same configuration to obtain 15 frames on two reference stars: the K5-star HD 181681 immediately before, and the K0-star HD 194479 immediately afterwards.

We used the vortex preprocessing pipeline (Xuan et al. 2018) to perform bad-pixel correction, flat fielding, thermal background subtraction, and co-registration of the images. The reference images from HD 181681 and HD 194479 were used to subtract off the stellar diffraction pattern of WR 140 using the KLIP-RDI implementation in the open-source pyKLIP package (Wang et al. 2015). Given the brightness of the dust in the data, we used 1 KL mode to model the stellar diffraction pattern. The fully reduced image is shown in Figure 1 (top right-hand panel). To accurately measure the flux of the dust, we forward-modeled the over-subtraction caused by KLIP-RDI by projecting the KL mode onto a model of the dust, and measured a small (1%–2%) flux loss due to KLIP-RDI that we compensate for. We conservatively adopt a 1σ $L^{\prime}$-band photometric uncertainty of 10%.

Flux densities of the C1 and the bar were extracted from the NIRC2 image using a circular aperture with a 250 mas radius and a rectangular aperture with a size of 1490 × 300 mas, respectively. The rectangular aperture for the bar was tilted by 17 south of west to match the angle of the bar feature. The flux density of the C1 "Fossil" dust feature (see Section 3.3) was extracted using an elliptical aperture with a semimajor axis and semiminor axis of 600 and 150 mas, respectively. The elliptical C1 Fossil aperture was rotated by 116 north of west to match the orientation of the dust feature.

Mid-IR imaging observations of WR 140 were obtained using the COMICS (Kataza et al. 2000; Okamoto et al. 2003) on the Cassegrain focus of the Subaru Telescope with the N11.7 filter (λ_c = 11.7 μm, Δλ = 1.0 μm) on 2020 June 27 UT (Figure 2). Individual 100 s exposures were performed using a perpendicular chopping and nodding pattern with a chop throw and nod amplitude of 10''. The total integration time on WR 140 was 400 s. The pixel scale of the detector is 013 pixel⁻¹.

Zoom In Zoom Out Reset image size

Observations of the K3II-star Gamma Aql were used as photometric and PSF references based on the list of mid-IR standards in Cohen et al. (1999). The measured full-width at half maximum (FWHM) in the N11.7 imaging observations of Gamma Aql was ∼035, which is consistent with near diffraction-limited performance at 11.7 μm (029).

The data reduction was carried out using the COMICS Reduction Software ²¹ and the Image Reduction and Analysis Facility (IRAF; Tody 1986, 1993). The chopping and nodding pair subtraction was employed to cancel out the background radiation and to reduce a residual pattern from the chop subtraction. Flat fielding was achieved using self-sky-flat made from each image. A conversion factor for flux density calibration was calculated by the reduced images of the standard star and its photometric data by Cohen et al. (1999). We adopt a 1σ N-band photometric calibration uncertainty of ∼10%.

The flux densities of the C1 and the bar were extracted from the COMICS image using a circular aperture with a 520 mas radius and a rectangular aperture with a size of 1820 × 650 mas, respectively. The rectangular aperture for the bar was tilted by 17 south of west to match the angle of the bar feature. The flux density of the C1 Fossil dust feature was extracted using an elliptical aperture with a semimajor axis and semiminor axis of 1170 and 520 mas, respectively. Background "sky" annuli were used to measure and subtract the mid-IR background emission. A circular annulus with inner and outer radii of 780 and 1560 mas were used for the background annulus of C1. A rectangular annulus with a 650 mas width and the same inner dimensions and orientation as the rectangular aperture for the bar was used for the bar. Lastly, the C1 Fossil background annulus had inner radii consistent with the dimensions of the elliptical C1 Fossil aperture and width of 650 mas.

The Subaru and Keck observations of WR 140 are summarized in Table 1.

The Subaru/CHARIS 1.1–2.4 μm, Keck/NIRC2 3.8 μm $L^{\prime}$-band, and 11.7 μm N-band Subaru/COMICS observations of WR 140 resolve and detect emission from dust formed during the periastron passage that occurred in 2016 December (Williams et al. 2009; Thomas et al. 2021). These features are consistent with the persistent dust features that were identified by IR imaging of WR 140 in Williams et al. (2009) after its 2001 periastron passage. The most prominent features are the two bright emission peaks located 086 and 115 southeast and southwest of the central binary, respectively (Figures 1 and 2). These southeast and southwest emission peaks were labeled as C0 and C1, respectively, by Williams et al. (2009). At wavelengths longer than ∼1.6 μm, we detect the aforementioned faint, linear bar of emission between C0 and C1. The near-IR CHARIS and 3.8 μm NIRC2 observations of the bar reveal a clumpy morphology that appear to bridge C0 and C1 (Figure 1).

To the east of the bar, the NIRC2 and 11.7 μm COMICS images show a faint arc of emission that curves to the north. Williams et al. (2009) refer to this feature as the arm, and, more recently, Han et al. (2022) label it as the Eastern Arm. In this work, we refer to this feature as the Eastern Arm in order to distinguish it from a fainter Western Arm that was detected in the NIRC2 image (Figure 1; Top Right). The Eastern Arm exhibits an integrated flux density ∼25% of the bar, and in the COMICS image the Eastern Arm is ∼35% the intensity of the bar. The high spatial resolution of the NIRC2 image reveals the clumpy morphology of the Eastern Arm, similar to that of the bar. Although the CHARIS observations did not detect significant emission from the Eastern Arm, this feature is located near/slightly beyond the edge of the CHARIS field of view.

Among the three observations, only the NIRC2 image reveals the Western Arm, which extends from C0 through the central source and continues north/northwest for ∼1'' (Figure 1; top right-hand panel). Northward of the central source, the curvature of the Western Arm resembles that of the Eastern arm. The integrated flux of the Western Arm in the NIRC2 image is a few percent of that of the bar. Non-detection of the Western Arm in the CHARIS and COMICS observations is likely to be due to its faintness, as well as the lower SNR toward the central source for the CHARIS observations.

With the exception of the Western Arm, all three observations capture the same set of persistent dust features given their similar morphology and positions relative to the central source. In the COMICS image, which was taken at φ = 0.444 in WR 140's orbital phase, C0 and C1 were located ∼2.5× and ∼3.0× the 11.7 μm FWHM from the central source which corresponds to ∼08 and ∼10, respectively. The extent of the bar in the COMICS image is ∼10. In both CHARIS and NIRC2 observations, which were taken at φ = 0.467 and φ = 0.462 in WR 140's orbital phase, respectively, C0 and C1 were located 09 and 11 from the central source, and the extent of the bar was 12. The slightly increasing size and displacement of the dust features measured over the ∼2 months spanning the COMICS, CHARIS, and NIRC2 observations are due to the high proper motion of the expanding dust (240–320 mas yr⁻¹; Williams et al. 2009).

A geometric model of dust formed in the wind collision interface between the WR 140 binary components around periastron passage can be derived from its well-defined orbital properties (Monnier et al. 2011; Thomas et al. 2021) and the previously characterized half-opening angle, θ ≈ 40°, of the colliding-wind shock from the central binary (Williams et al. 2009; Fahed et al. 2011). Similar geometric models have successfully reproduced the morphology of circumstellar dust formed in the colliding-wind WC binaries WR 104 (Soulain et al. 2018), WR 112 (Lau et al. 2020), and Apep (Callingham et al. 2019; Han et al. 2020). Figure 3 shows the orbital-plane and on-sky projections of the dust-forming wind–wind interface model of WR 140's dust shell assuming a constant dust expansion velocity of 2450 km s⁻¹, which is consistent with the observationally derived expansion velocity from Williams et al. (2009). The orbital parameters of WR 140 in the geometric models are adopted from the Monnier et al. (2011) results with the Fahed et al. (2011) prior: e = 0.9, ω = 468, Ω = 3536, i = 1196. The orbital configuration of the central binary at an orbital phase of φ = 0.47 and the orbital range of dust formation is also shown in Figure 3. The adopted true anomaly range where dust formation occurs is −135° to 135°; however, note that relatively little dust is observed around anomaly ranges consistent with periastron passage due to variable dust-formation rates (Han et al. 2022).

Zoom In Zoom Out Reset image size

Figure 3 (right-hand panel) indicates the position of the C0 and C1 dust features, as well as the bar and the Western and Eastern Arms. These features likely correspond to regions of the dust shell with larger integrated dust column densities along the line of sight. The morphology of the Eastern Arm and the Western Arm, the latter of which captured only in the NIRC2 image, appear consistent with the east and west edges of the near side of the geometric model. Based on the geometric model, the southern edge of the near side of the dust shell (i.e., the southern extent of the Arms) is coincident with C0 along the bar. Given the lower angular resolution of the 11.7 μm COMICS observations relative to those of NIRC2 and CHARIS, the blended dust emission of those two regions in the dust shell may contribute to C0 appearing brighter than C1 in the COMICS image.

A notable difference between the geometric wind–wind interface model of dust formation and the observations of WR 140 is the absence of bright dust emission along the northern edge of the dust shell, where the integrated column density along the line of sight toward the observer should be comparable to that of southern region associated with the bar. This is likely to be due to WR 140's highly variable dust-formation rates throughout periastron passage (Eatson et al. 2022b; Han et al. 2022). Therefore, further investigations comparing colliding-wind simulations (e.g., Eatson et al. 2022a, 2022b) with observations are important for testing our understanding of dust formation in colliding-wind binaries; however, this is beyond the scope of this paper.

The 3.8 μm NIRC2 and 11.7 μm COMICS images of WR 140 detect emission from the C1 feature that formed during the 2009 January periastron passage, one orbital period (7.93 yr) before the 2016 December periastron passage. Figure 4 shows the NIRC2 and COMICS images and the location of this C1 Fossil that is located along the same position angle as C1. The COMICS image also shows a faint detection of the C0 and Eastern Arm Fossils to the east of the C1 Fossil. The COMICS image resembles previous Gemini South imaging observations at 12.5 μm by Marchenko & Moffat (2007) and Williams et al. (2009), as well as JWST imaging where repeated dust features from multiple past dust-formation episodes were detected. The NIRC2 image presents the first 3.8 μm detection of a previous dust-formation episode.

Zoom In Zoom Out Reset image size

To verify that the C1 Fossil in the NIRC2 image is indeed from the 2009 January dust-formation episode, the distance between C1 and the C1 Fossil can be compared to the expected separation distance based on previously measured proper motions of C1. The proper motion of C1 is 326 mas yr⁻¹ (Williams et al. 2009); therefore, over a 7.93 yr orbital period the separation distance between C1 and the C1 Fossil should be 26, which is almost exactly the measured separation distance in the NIRC2 image.

The resolved mid-IR emission from circumstellar dust around WR 140 has previously been characterized and attributed to thermal emission heated radiatively by the central binary (Williams et al. 2009). Monnier et al. (2002) had first resolved near-IR emission around WR 140 with observations taken shortly after periastron passage (φ ≈ 0.04–0.06) and suggested that it originated from thermal emission from newly formed dust. Based on JHK light curves of WR 140 and the decreasing dust temperatures as dust drifts further from the central heating source, the near-IR emission from WR 140 was thought to fade away by around φ ∼ 0.4–0.6 (Williams et al. 2009). However, the 0.04 mag uncertainties in the JHK magnitudes from the Williams et al. (2009) light curve of WR 140 indicate that faint near-IR emission could still persist at late times. Indeed, the CHARIS observations reveal the presence of near-IR circumstellar emission around WR 140 at an orbital phase greater than φ ≳ 0.4, which highlights the importance in understanding the origin of the near-IR emission. The two most likely sources of circumstellar near-IR emission at φ ∼ 0.45, which is ∼3.6 yr after the 2016 December dust-forming episode, are thermal dust emission and scattered light from the central binary.

To investigate the properties of the spatially resolved near-IR emission, the SED of the bar and C1 were extracted from the CHARIS observations. The first two channels of the CHARIS data cube, which correspond to 1.16 and 1.20 μm, were heavily affected by data reduction artifacts and thus omitted from the SED analysis. The channels corresponding to 1.37, 1.42, 1.87, and 1.93 μm are likely telluric-dominated and were also omitted from the SED analysis. The extracted near-IR flux densities were corrected for interstellar extinction using the average observed extinction curve of the Milky Way "G21_MWAvg" from the dust_extinction Python package (Gordon et al. 2021). ²² A value of A_v = 2.21 ²³ was adopted for the interstellar extinction toward WR 140 (Rate & Crowther 2020).

Figure 5 shows the near-IR SEDs of the bar and C1 from the CHARIS data cube before and after extinction correction. The shape of the near-IR SEDs of both the bar and C1 clearly demonstrates that the dust emission increases with increasing wavelength. Note that because the bar and C1 are spatially resolved from the central stars, it is unlikely that there is any contamination by emission from stellar winds in the SEDs. To understand the source of the near-IR flux from the bar, the two most likely scenarios were investigated: scattered light from the central binary or thermal dust emission. For the scattered light model, ${F}_{\nu }^{\mathrm{Sca}}(a)\propto {Q}_{\mathrm{Sca}}(a){F}_{\nu }^{* }$, the scattering efficiencies Q_Sca(a) for aromatic-rich H-poor amorphous carbon (a-C) models with radii a = 0.01 μm and a = 1.0 μm were adopted (Jones et al. 2013). This a-C dust model is also used for SED modeling in Sections 3.5 and 3.6.

Zoom In Zoom Out Reset image size

The near-IR continuum emission from WR 140 is assumed to be dominated by the free–free emission from the ionized stellar winds of the WC star, which can be approximated as a power law ${F}_{\nu }^{\mathrm{ff}}\propto {\nu }^{0.96}$ (Morris et al. 1993). Both 0.01 and 1.0 μm grain size scattered light models, which were normalized to the H-band flux density of the SEDs, fail to reproduce the positive near-IR slope of SEDs. This indicates that there is little or no contribution from scattered light (Figure 5).

A modified blackbody with an emissivity power-law index of β = 1 was assumed for the thermal dust emission model,

$\begin{eqnarray}&&{F}_{\nu }^{\mathrm{MBB}}({T}_{d})\propto {B}_{\nu }({T}_{d}){\nu }^{\beta },\end{eqnarray} \tag{ 1 }$

where T_d is the temperature of the emitting dust. The dust temperature and a multiplicative scaling factor for ${F}_{\nu }^{\mathrm{MBB}}({T}_{d})$ were fit to the CHARIS SEDs of the bar and C1. The best-fit thermal dust emission models with temperatures of T_d ≈ 960 K and T_d ≈ 1020 K for the bar and C1, respectively, demonstrate a close agreement with the shape of the near-IR SEDs. Even at late times (φ ≈ 0.45), the near-IR circumstellar emission around WR 140 is therefore most likely to be dominated by hot thermal dust emission.

Table 1. WR 140 Observation Summary

	Subaru/CHARIS	Keck/NIRC2+PyWFS	Subaru/COMICS
Observing Mode	Broadband+Lyot 113 mas	$L^{\prime}$ Vortex Coronagraph	Imaging
Observation Date (UTC)	2020 Sep 2 (MJD 59094.2)	2020 Aug 18 (MJD 59079.4)	2020 Jun 27 (MJD 59027.5)
WR 140 Orbital Phase (φ)	0.467	0.462	0.444
Wavelength (μm)	1.15–2.39	3.8	11.7
Spec. Resolving Power (λ/Δλ)	∼19	∼5	∼12
Angular Res. (FWHM)	003–006	0074	035
Pixel Scale (mas)	16.4	9.94	130
Field of View	∼2'' × 2''	10'' × 10''	42'' × 32''

Note. Summary of WR 140 observations with Subaru/SCExAO/CHARIS, Keck/NIRC2, and Subaru/COMICS. The WR 140 orbital phase (φ) is derived from the ephemeris derived by Monnier et al. (2011) with the Fahed et al. (2011) prior.

Download table as:  ASCIITypeset image

Based on previous resolved mid-IR imaging studies of WR 140, the circumstellar 3.8 and 11.7 μm emission observed by NIRC2 and COMICS likely arises from thermal dust emission (Williams et al. 2009). The observed near-IR, 3.8 μm, and 11.7 μm flux densities extracted from the bar and C1 are provided in Table 2, along with the extinction correction factors derived from the Gordon et al. (2021) interstellar-extinction curve with A_v = 2.21. Table 2 also provides the 3.8 and 11.7 μm flux densities measured from the C1 Fossil.

Adopting the same modified blackbody thermal dust emission model fit to the near-IR emission (Equation (1)), the mid-IR 3.8 and 11.7 μm flux densities of the bar and C1 indicate temperatures of T_d ≈ 560 K and T_d ≈ 530 K, respectively. These temperatures are notably consistent with the circumstellar dust temperature derived from mid-IR observations by Williams et al. (2009) at a similar phase, where T_d ≈ 560 K. The dust temperature of the C1 Fossil can also be estimated from its measured mid-IR 3.8 and 11.7 μm flux densities assuming they originate from the same dust component. Equation (1) provides a dust temperature of T_d ≈ 490 K for the C1 Fossil.

There are two notable disparities in the dust temperature calculations:

• 1.  
  The near-IR thermal dust emission from the bar and C1 (T_d ∼ 1000 K) exhibits much hotter temperatures than those derived by the mid-IR 3.8 and 11.7 μm observations (T_d ∼ 550 K).
• 2.  
  The derived temperature of the C1 Fossil (T_d ≈ 490 K) from the 3.8 and 11.7 μm observations is a factor of 1.6 hotter than predicted from the decreasing radiative equilibrium temperature as a function of distance from the heating source (T_d ∝ r^−0.4).

Since the resolved circumstellar emission components at near- and mid-IR wavelengths are largely co-spatial, it is highly unlikely that the first disparity is due to spatial variations in the exposure to the central binary's radiation field that dominates the dust heating. This disparity between the near- and mid-IR dust temperatures of the bar and C1 instead suggests that the dust components probed by near- and mid-IR wavelengths possess different physical properties, such as grain size.

The second disparity arises because dust continuum emission in radiative thermal equilibrium with a central heating source should decrease with distance (e.g., Williams et al. 2009; Lau et al. 2020). The C1 Fossil is located over a factor 3.8 further than C1 from the central binary, which implies that the dust temperature of the C1 Fossil should be ∼310 K given the 530 K temperature measured for C1 and adopting T_d ∝ r^−0.4 for dust in equilibrium heating. This is significantly lower than the estimated 490 K dust temperature from the 3.8 and 11.7 μm flux measurements from Equation (1). The second disparity can also be resolved if there are distinct dust components probed by the 3.8 and 11.7 μm observations at the C1 Fossil. This scenario at the C1 Fossil would invalidate the dust temperature estimate from Equation (1), which assumed that the 3.8 and 11.7 μm emission originates from the same dust component. These disparities are investigated in a detailed dust SED analysis in the following sections.

To investigate the properties of the near- and mid-IR dust components of the circumstellar dust around WR 140, dust emission models were fit to the 1.2–11.7 μm SEDs of the bar and C1 using DustEM, which is a numerical tool that computes the emission from dust heated by an input radiation field in the optically thin limit with no radiative transfer (Compiègne et al. 2011). Using the formalism of Desert et al. (1986), DustEM also calculates grain temperature distributions, which is important for characterizing the emission from very small (a ≲ 10 Å) dust grains that are stochastically heated by single photon absorption (e.g., Draine & Li 2001). Unlike larger a ≳ 100 Å grains, these very small nanodust grains may not be in local thermal equilibrium (LTE) with the radiation field given their smaller absorption cross sections and lower heat capacities. Single photon absorption by nanodust grains that are not in LTE result in transient temperature spikes that exceed the temperatures of larger grains in LTE, and can therefore produce an excess in shorter wavelength IR emission relative to the larger grains (Sellgren et al. 1983; Desert et al. 1986; Guhathakurta & Draine 1989). Notably, due to their lower heat capacities, smaller grains exhibit higher temperatures than larger grains when exposed to the same radiation field. Emissions from such nanodust grains are likely to be important in explaining the hot temperatures of the near-IR component in WR 140.

The input parameters for our DustEM SED models were the radiation field of the central binary, the radial distance of the dust components from the central binary, the dust mass, and the grain size distribution properties. A log-normal grain size distribution with ${dn}/d\mathrm{log}(a)\propto {e}^{-{\left(\mathrm{log}(a/{a}_{0})/\sigma \right)}^{2}}$ was adopted for the DustEM models, where the free parameters were the centroid grain size a₀ and the distribution width σ. Minimum and maximum grain sizes in the models were fixed at 9 × 10⁻⁴ μm and 1.0 μm, respectively, where the minimum grain size limit was defined by where the maximum dust temperatures exceeded a sublimation temperature of ∼1800 K when exposed to the WR 140 radiation field.

To approximate the dust-heating radiation field from the WC7 and O5 stars in WR 140, we used synthetic spectra from model atmospheres computed with the PoWR code (Gräfener et al. 2002; Sander et al. 2012; Hainich et al. 2019). The models were calculated to reproduce the spectral appearance of both stars (including the light ratio) using a detailed binary analysis similar to those described in Shenar et al. (2019). A consistent treatment of the hydrostatic part is included in both the O and the WR model (Sander et al. 2015). A more dedicated paper about the spectral modeling of WR 140 is currently being worked on (T. Shenar et al. 2023, in preparation). The derived effective temperatures at an optical depth of τ = 2/3 are T_2/3,WR = 57,000 K and T_2/3,O = 36,000 K, and the luminosities are L_*,WR = 1 × 10⁶ L_⊙ and L_*,O = 5 × 10⁵ L_⊙.

Based on the H-poor environment around WR stars and the recent claim of C-rich aromatic compounds in the circumstellar dust around WR 140 (Lau et al. 2022), dust grains are assumed to be composed of aromatic-rich H-poor amorphous carbon for SED modeling. The properties of these grains are consistent with the aromatic-rich H-poor carbonaceous grains defined in Jones et al. (2013), which have a bulk density of ρ_b = 1.6 g cm⁻³. Notably, given its strong UV-radiation field, the central WR star is capable of rapidly photoprocessing (τ_UV,pd ≪ 1 yr; Equation (7) from Jones et al. 2014) newly formed carbonaceous dust as large a ≲ 1000 Å to a stable and fully evolved state consistent with the aromatic-rich H-poor amorphous carbon grains described by Jones et al. (2013). The distance to the circumstellar dust from the central binary system can be approximated by the spatially resolved imaging and the geometric model (Figures 1 and 3). By adopting the Gaia-derived distance to WR 140 of d = 1.64 kpc (Rate & Crowther 2020), the distance to the bar and C1 is r_dust ≈ 1650 au and the distance to the C1 Fossil is r_C1,Fossil ∼ 6200 au at the time of the observations.

DustEM SED model fits to the bar and C1 were initially attempted using a single population of a ≈ 100 Å dust grains, which was motivated by the 100 Å grain size estimates from previous studies of circumstellar dust around WR 140 (Williams et al. 2009; Lau et al. 2020). However, the single population fits failed to reproduce the emission at either near- and mid-IR wavelengths. A second population of ∼10 Å, nano-sized grains (i.e., nanodust) was then introduced into the SED fitting. The best-fit DustEM SED models were obtained from a reduced χ² analysis, where the amplitude of the dust emission of the two components is modulated by their dust mass, and a grid search through two parameters: the grain size centroid of the nanodust and larger grain distributions. Due to degeneracies with the grain size centroid, the nanodust and larger grain size distributions' widths were fixed at values of 0.1 and 1.0, respectively. The DustEM SED grid search parameters and the other DustEM model parameters are provided in Table 3.

Table 3.  DustEM SED Parameters

Property	Value
Grain Type	Aromatic-rich H-poor a-C
Bulk Density	1.6 g cm−3
Size Distribution	log-normal
${a}_{\min }-{a}_{\max }$	9 × 10−4 − 1.0 μm
rdust	1650 au
rC1,Fossil	6200 au
L*,WR	9.3 × 105 L⊙
T2/3,WR	57,000 K
R2/3,WR	9.79 R⊙
L*,O	5.0 × 105 L⊙
T2/3,O	36,000 K
R2/3,O	18.2 R⊙
dWR140 a	1.64 kpc
SED Grid Search Parameters
a0 (Nanodust)	9–13 Å (Δa = 0.5 Å)
σ (Nanodust)	0.1
a0 (Larger Grains)	50–1100 Å (Δa = 100 Å)
σ (Larger Grains)	1.0

Notes. The aromatic-rich H-poor amorphous carbon (a-C) models are defined in Jones et al. (2013). The parameters a_min, a_max, r_dust, and r_C1,Fossil correspond to the minimum and maximum grain sizes in the log-normal grain size distribution, the distance between the circumstellar dust and the central heating source, and the distance between the C1 Fossil and the central heating source, respectively. The heating source parameters L_*, T_2/3, and R_2/3 correspond to the luminosity, temperature (at τ = 2/3), and radius (at τ = 2/3) of the stellar heating sources, respectively. The SED grid search parameters show the range of the centroid grain sizes a₀ and step intervals Δa for the nanodust and larger grain dust components in the DustEM model fitting

^a Gaia-derived distance to WR 140 (Rate & Crowther 2020).

Download table as:  ASCIITypeset image

The best-fit DustEM SED models for the emission from the bar and C1 are shown in Figure 6. Both of the SED model fits demonstrate that the near-IR flux observed by CHARIS is dominated by thermal emission from ∼10 Å nanodust. Emission from the nanodust component between 3 and 14 μm also exhibits IR emission bands that are consistent with unidentified infrared (UIR) features found in various astrophysical environments (Leger & Puget 1984; Allamandola et al. 1985; Tielens 2008). Although the UIR features from dust around WR 140 were not covered by the ground-based observations presented in this work, they have been identified from mid-IR integral-field unit spectroscopy with JWST (Lau et al. 2022).

Zoom In Zoom Out Reset image size

The mid-IR flux observed by NIRC2 and COMICS is dominated by thermal emission from larger ∼300–500 Å grains in the best-fit SED models. Although consistent within uncertainties, it is interesting to note that the grain sizes of this larger population lie between the ∼100 Å grains inferred from the analysis in Williams et al. (2009), Lau et al. (2020), and the ∼700 Å grains determined by Marchenko et al. (2003) from optical photometric absorption around periastron passage. The dust SED properties of the best-fit models are provided in Table 4.

Table 4. WR 140 DustEM SED Properties of C1 and the Bar

	C1		bar
Property	Larger Grains	Nanodust Grains	Larger Grains	Nanodust Grains
Grain Size Dist. Centroid (a0)	${530}_{-210}^{+490}$ Å	${10.5}_{-0.5}^{+2.0}$ Å	${320}_{-200}^{+360}$ Å	${10.0}_{-1.0}^{+2.5}$ Å a
Grain Size Dist. Width (σ)	1.0	0.1	1.0	0.1
Dust Temp. (Td )	480 K	930 K	520 K	930 K
IR Luminosity (LIR)	${21.5}_{-3.3}^{+1.2}$ L⊙	${4.4}_{-0.3}^{+1.2}$ L⊙	${42.6}_{-7.0}^{+3.3}$ L⊙	${8.0}_{-2.3}^{+2.8}$ L⊙
Mass (Md )	$({2.4}_{-0.9}^{+1.6})\times {10}^{-9}$ M⊙	$({1.1}_{-0.1}^{+0.3})\times {10}^{-10}$ M⊙	$({3.2}_{-1.3}^{+2.0})\times {10}^{-9}$ M⊙	$({2.1}_{-0.6}^{+0.7})\times {10}^{-10}$ M⊙
Md,Tot	$({2.5}_{-0.9}^{+1.6})\times {10}^{-9}$ M⊙		$({3.4}_{-1.3}^{+2.0})\times {10}^{-9}$ M⊙
MNd/Md,Tot	${0.05}_{-0.02}^{+0.03}$		${0.07}_{-0.03}^{+0.05}$

Notes. Best-fit DustEM SED model parameters of WR 140's circumstellar C1 and bar dust features. The dust temperature T_d corresponds to the average dust temperature T_moy output by DustEM at the best-fit grain size centroid a₀. L_IR is the integrated IR luminosity of the larger grain or nanodust components. The individual dust masses (M_d ) of the larger and nanodust grains of the best-fit models are provided, as well as the total (M_d,Tot) dust masses of both components. M_Nd/M_d,Tot corresponds to the dust mass ratio of the nanodust component to the total dust mass.

^a The lower uncertainty of a₀ for the nanodust grain population of the bar hits the limit of the grid search.

Download table as:  ASCIITypeset image

The dust temperature, ²⁴ T_d , of the nanodust and larger grain size populations are 930 K and ∼500 K, respectively, and are consistent with the modified blackbody temperatures estimated in Section 3.4. The total integrated IR luminosity of C1 and the bar are a very small fraction of the luminosity of the central binary (L_IR/L_* < 0.01%).

The total dust mass derived for the bar is ∼3 × 10⁻⁹ M_⊙. As expected, this value is slightly lower than the total WR 140 circumstellar dust mass (${6.4}_{-3.3}^{+3.8}\times {10}^{-9}$ M_⊙) from similar SED modeling results from Lau et al. (2020), which was fitted to archival data taken at similar orbital phase (φ ≈ 0.43). In both C1 and the bar, the ratio of the nanodust mass to the total dust mass is ∼5%. The general agreement between the dust temperatures and nanodust mass ratios of C1 and the bar suggests that C1 does not possess significantly different dust properties from the other regions in the bar.

To estimate changes in mass from dust formed in the 2016 and 2009 dust-production episodes, the DustEM SED model fit to the C1 feature was applied to the 3.8 and 11.7 μm detections of the C1 Fossil (Figure 4). Several free parameters were therefore fixed due to the limited wavelength coverage of the C1 Fossil: the grain size distribution centroids (a₀) of the nanodust and larger grains was adopted from the derived C1 values (Table 4). The same heating source properties as the C1 and bar models were used. The separation distance between the central binary and the C1 Fossil was estimated from the NIRC2 observations to be ∼6200 au, which resulted in cooler dust temperatures for the larger and nanodust grains in the C1 Fossil. The only free parameters were therefore the dust masses of the nanodust and larger grain size populations, which were derived by scaling the nanodust and larger grain models to the 3.8 and 11.7 μm flux measurements.

The scaled SED model of the C1 Fossil shown in Figure 7 and the dust properties are provided in Table 5. The dust SED fit suggests that the 3.8 μm emission is dominated by nanodust, while the 11.7 μm emission is dominated by larger grains. Note that in the C1 and bar SED models (Figure 6), the emission at both 3.8 and 11.7 μm emission was dominated by the larger grains. The different dust population probed by the 3.8 μm emission in the C1 Fossil is likely to be due to the cooler dust temperatures, as well as an enhancement in the abundance of nanodust. The cooler dust temperatures of 300 K and 590 K exhibited by the larger grains and nanodust grains, respectively, in the C1 Fossil is expected due to r⁻² decrease in the radiative flux from the central binary heating source. A potentially intriguing result is the increase in the C1 Fossil nanodust mass relative to the C1 dust mass by a factor of ∼3 and the decrease of the C1 larger grain dust mass by a factor ∼0.4. The ratio of the nanodust mass to the total dust mass in the C1 Fossil, ∼40%, presents a substantial increase from the nanodust mass ratio in C1 of ∼5%.

Zoom In Zoom Out Reset image size

The total dust mass in the C1 Fossil, ∼1.3 × 10⁻⁹ M_⊙, is approximately half that of the total dust mass in C1, which suggests that circumstellar dust around WR 140 is destroyed as it propagates away from the central binary. This is consistent with the interpretation and observations of the decreasing circumstellar dust mass by Williams et al. (2009) from their multi-epoch mid-IR imaging of WR 140. However, the ∼4 μm imaging observations presented by Williams et al. (2009) did not capture the fossil dust emission, which was likely to be due to the limiting sensitivity of the observations. Therefore, the NIRC2 observations of WR 140 uniquely present evidence of an increasing abundance of nanodust. If the nanodust population is enhanced by the disruption of the larger grain population, then ∼20% of the total mass removed from the larger grain population between C1 and the C1 Fossil must have been transferred to new nanodust grains. It is important to note that these interpretations are derived from only two flux density measurements and should consequently be treated with caution until verified by observations at higher sensitivity and with broader wavelength coverage. However, UIR features that are typically attributed to very small carbonaceous grains and/or complex hydrocarbons were observed from the C1 Fossil with JWST (Lau et al. 2022), which supports the presence of nanodust grains derived from our SED models. The observation of such features, in addition to the detection of the C1 feature from the past ∼17 dust-formation episodes at 7.7 μm (Lau et al. 2022), suggest that nanodust grains are indeed present and survive in the extended circumstellar environment around WR 140.

In this section, we consider the mechanisms that may be responsible for the apparent enhancement of nanodust and destruction of larger ∼300–500 Å grains in the circumstellar environment of WR 140. This process must occur over the 7.93 yr timescale (i.e., the WR 140 orbital period) and the 1650–6200 au distance scale that characterize how dust propagates from C1 to the location of the C1 Fossil. Based on the increasing nanodust abundance, we can rule out several dust destruction mechanisms that would produce the opposite result and destroy nanodust grains more efficiently than larger dust grains. Thermal sublimation, which is caused by heating grains to, and beyond, their sublimation temperature, would preferentially tend to destroy small stochastically-heated grains because they are heated to higher temperatures than larger grains when exposed to an identical heating source, due to their lower heat capacities

Thermal and non-thermal sputtering of dust grains by collisions with gas in the circumstellar environment of WR 140 would destroy grains on timescales proportional to the grain size (e.g., Tielens et al. 1994), which implies that ∼10 Å nanodust would be destroyed via sputtering ∼50 times faster than ∼500 Å grains. Therefore, both thermal sublimation and (non-)thermal sputtering can be ruled out as dominant mechanisms for enhancing nanodust abundance.

We now consider the two most likely dust processing mechanisms that can produce smaller dust grains from larger grains: shattering via grain–grain collisions and rotational disruption from radiative torques (Hirashita 2010; Asano et al. 2013; Hoang et al. 2019). The shattering of larger grains via grain–grain collisions is believed to be an important mechanism to significantly alter grain size distributions and enhance the population of small grains (e.g., Jones et al. 1994, 1996; Hirashita 2010; Asano et al. 2013). Another mechanism that can increase the abundance of small grains relative to large grains is dust disruption from the centrifugal stress within fast-rotating grains spun up by radiative torques. Hoang et al. (2019) claim that this radiative torque disruption (RATD) process should be particularly relevant in environments with strong radiation fields from sources such as massive stars and supernovae. The viability of grain–grain collisions compared to disruption from radiative torques in the circumstellar environment of WR 140 can be investigated by comparing the relevant timescales of each mechanism. We specifically focus on the C1 dust concentration and its derived properties to estimate and compare the timescales.

Grain–grain collisions. For grain shattering, the mean timescale for collisions between two grains of size a can be estimated as follows:

$\begin{eqnarray}&&{\tau }_{\mathrm{gg}}=\displaystyle \frac{1}{\pi {a}^{2}{n}_{\mathrm{gr}}{v}_{\mathrm{gr}}},\end{eqnarray} \tag{ 2 }$

where n_gr is the number density of dust grains and v_gr is the relative velocity of the grains. Given the complexity of estimating the relative velocity of circumstellar dust formed via colliding winds, we adopt an upper limit of v_gr as the drift velocity of dust grains through gas, v_drift (i.e., v_gr < v_drift). The drift velocity can be estimated from the observationally derived radial expansion velocity of the circumstellar dust around WR 140, v_dust ∼ 2600 km s⁻¹ (Lau et al. 2022), and the terminal velocity of the WC star, v_∞ ≈ 2900 km s⁻¹ (Williams & Eenens 1989). Therefore, v_drift = v_∞ − v_dust ∼ 300 km s⁻¹.

$\begin{eqnarray}&&{v}_{\mathrm{drift}}={v}_{\infty }-{v}_{\mathrm{dust}}\sim 300\,\mathrm{km}\,{{\rm{s}}}^{-1}.\end{eqnarray} \tag{ 3 }$

The number density of dust grain in the circumstellar dust shell is difficult to measure given its complex 3D morphology (Figure 3). However, we can estimate an upper limit on the dust grain number density for the partially spatially resolved C1 dust feature assuming that its derived dust mass (Table 4) is concentrated in a sphere with a radius equivalent to half the FWHM of the angular resolution in the Keck/NIRC2 imaging observations at the 1.64 kpc distance to WR 140, Δr_C1 ≈ 60 au. An upper limit of the dust grain number density in the C1 feature can be expressed as

$\begin{eqnarray}&&{n}_{\mathrm{gr},{\rm{C}}1}\lesssim \displaystyle \frac{{M}_{d,{\rm{C}}1}}{{m}_{\mathrm{gr}}\tfrac{4}{3}\pi {{\rm{\Delta }}{r}_{{\rm{C}}1}}^{3}}\approx \displaystyle \frac{{M}_{d,{\rm{C}}1}}{(\tfrac{4}{3}\pi {a}^{3}{\rho }_{\mathrm{bulk}})\tfrac{4}{3}\pi {{\rm{\Delta }}{r}_{{\rm{C}}1}}^{3}},\end{eqnarray} \tag{ 4 }$

where M_d,C1 is the dust mass of the larger grains in C1, m_gr is the mass of an individual larger grain, and ρ_bulk is the bulk density of the dust grain. Based on the adopted and derived SED model dust properties of C1, we derive the following value as an upper limit

$\begin{eqnarray}&&\begin{array}{l}{n}_{\mathrm{gr},{\rm{C}}1}\lesssim 1.8\,\times \,{10}^{-6}\left(\displaystyle \frac{{M}_{d,{\rm{C}}1}}{2.4\times {10}^{-9}\,{M}_{\odot }}\right){\left(\displaystyle \frac{{\rm{\Delta }}{r}_{{\rm{C}}1}}{60\,\mathrm{au}}\right)}^{-3}\\ \qquad \times \ {\left(\displaystyle \frac{{\rho }_{\mathrm{bulk}}}{1.6\,{\rm{g}}\,{\mathrm{cm}}^{-3}}\right)}^{-1}{\left(\displaystyle \frac{a}{500\,\mathring{\rm A} }\right)}^{-3}\,{\mathrm{cm}}^{-3}.\end{array}\end{eqnarray} \tag{ 5 }$

A lower limit on the grain–grain collision timescale of τ_gg,C1 ≳ 7.5 yr can then be derived by combining Equations (3) and (5) with Equation (2):

$\begin{eqnarray}&&\begin{array}{l}{\tau }_{\mathrm{gg},{\rm{C}}1}\gtrsim 7.5{\left(\displaystyle \frac{{n}_{\mathrm{gr},{\rm{C}}1}}{1.8\times {10}^{-6}\,{\mathrm{cm}}^{-3}}\right)}^{-1}\\ \quad \times \,{\left(\displaystyle \frac{{v}_{\mathrm{drift}}}{300\,\mathrm{km}\,{{\rm{s}}}^{-1}}\right)}^{-1}{\left(\displaystyle \frac{a}{500\,\mathring{\rm A} }\right)}^{-2}\,\mathrm{yr}.\end{array}\end{eqnarray} \tag{ 6 }$

Given that grain–grain collision timescales are less than the 7.93 yr orbital period of WR 140, it is possible that grain–grain collisions may be causing the redistribution of mass from larger to smaller dust grains.

Radiative torque disruption (RATD). In strong radiation fields (U ≫ 1) with mean wavelength $\bar{\lambda }$, the RATD mechanism can disrupt dust grains with radii larger than a critical grain size a_disr (Hoang et al. 2019). By utilizing the derivation of a_disr from Equation (26) in Hoang et al. (2019) for environments where ${a}_{\mathrm{disr}}\lesssim \bar{\lambda }/1.8$, we can estimate a_disr for circumstellar dust around WR 140 as follows:

$\begin{eqnarray}&&\begin{array}{l}{a}_{\mathrm{disr}}\simeq 84\left[{\left(\displaystyle \frac{\gamma }{1.0}\right)}^{-1}{\left(\displaystyle \frac{{L}_{* ,\mathrm{WR}+{\rm{O}}}}{1.43\times {10}^{6}\,{L}_{\odot }}\right)}^{-1/3}\right.\\ \quad \times {\left.{\left(\displaystyle \frac{{r}_{\mathrm{dust}}}{1650\,\mathrm{au}}\right)}^{2/3}{\left(\displaystyle \frac{{\bar{\lambda }}_{\mathrm{WR}+{\rm{O}}}}{0.11\,\mu {\rm{m}}}\right)}^{1.7}{\left(\displaystyle \frac{{S}_{\max }}{{10}^{9}\,\mathrm{erg}\,{\mathrm{cm}}^{-3}}\right)}^{1/2}\right]}^{\tfrac{1}{2.7}}\,\mathring{\rm A} ,\end{array}\end{eqnarray} \tag{ 7 }$

where γ corresponds to the anisotropy of the radiation field (i.e., γ = 1 for unidirectional radiation sources), L_*,WR+O is the combined luminosity of the WR and O stars in WR 140, r_dust is the distance between the circumstellar dust and the central binary, ${\bar{\lambda }}_{\mathrm{WR}+{\rm{O}}}$ is the mean wavelength of the radiation field from the central WR+O binary, and S_max is the maximum tensile strength of the grain material.

In Equation (7), we have adopted ${S}_{\max }={10}^{9}$ erg cm⁻³, which is a typical value for compact grains and is also consistent with polycrystaline graphite (Hoang et al. 2019); however, we note that strong materials such as monocrystaline graphite and diamond can exhibit tensile strengths of ${S}_{\max }\gtrsim {10}^{11}$ erg cm⁻³ (Hoang et al. 2019). The calculation of a_disr in Equation (7) demonstrates that the larger grain size population in our SED models of the bar and C1 (a ∼ 300–500 Å; Table 4) satisfies the conditions for which RATD should occur.

The RATD disruption time for grains of size a can be defined as the time required to spin-up a grain to its rotational disruption velocity. Hoang et al. (2019) provide the expression for the RATD disruption time in their Equation (27), which we can apply to the circumstellar environment of WR 140 for dust located at r_dust = 1650 au (e.g., the C1 feature) from the central binary:

$\begin{eqnarray}&&\begin{array}{l}{\tau }_{\mathrm{RATD}}\simeq 1.6{\left(\displaystyle \frac{{\bar{\lambda }}_{\mathrm{WR}+{\rm{O}}}}{0.11\,\mu {\rm{m}}}\right)}^{1.7}{\left(\displaystyle \frac{{L}_{* ,\mathrm{WR}+{\rm{O}}}}{1.43\times {10}^{6}\,{L}_{\odot }}\right)}^{-1}\\ \quad \times \,{\left(\displaystyle \frac{a}{500\,\mathring{\rm A} }\right)}^{-0.7}{\left(\displaystyle \frac{{r}_{\mathrm{dust}}}{1650\,\mathrm{au}}\right)}^{2}{\left(\displaystyle \frac{{S}_{\max }}{{10}^{9}\,\mathrm{erg}\,{\mathrm{cm}}^{-3}}\right)}^{1/2}\,\mathrm{days}.\end{array}\end{eqnarray} \tag{ 8 }$

A comparison of the grain–grain collision and RATD timescales (Equations (6) and (8)) indicates that τ_RATD ≪ τ_gg for circumstellar dust around WR 140 at C1 located r_dust = 1650 au from the central binary. Therefore, dust disruption from radiative torques by the strong radiation field of the central binary may be the dominant grain disruption mechanism; however, such short RATD timescales suggest the larger grain population should be entirely disrupted by the time C1 reaches the position of the C1 Fossil, which disagrees with the presence of larger grains in the C1 Fossil (Table 5, Figure 7). It is possible that the maximum tensile strength of the dust grains (S_max) around WR 140 is underestimated by factors of ∼10–100, which would result in increased RATD timescales by factors of ∼3–10 (Equation (8)). In the RATD calculations, we have also disregarded the possible effects of grain rotational damping (e.g., by gas collisions), which may mitigate the grain disruption from radiative torques (Hoang et al. 2019). We also note that there are significant uncertainties in the grain density estimates for the grain–grain collision timescale calculation (Equation (4)). Ultimately, given the strong and hard radiation field around WR 140, we slightly favor RATD over grain–grain collisions as the dominant mechanism that increases the abundance of nanodust grains relative to the larger grains in WR 140.

We can investigate the impact of RATD in the circumstellar environment of WR 140 starting from the point of dust formation at r_dust ∼ 125 au (Williams et al. 2009). Figure 8 shows the critical grain size a_disr as a function of r_dust from WR 140 based on Equation (7). The ∼10 Å nanodust grains notably fall below a_disr, which implies that RATD should not destroy newly formed nanodust created from disrupted larger grains. Interestingly, Figure 8 indicates that larger grains (a ≳ 100 Å) around WR 140 should continue to be disrupted by RATD at separation distances as large as the position of the C1 Fossil (r_dust ≈ 6200 au). Therefore, the nanodust to total dust mass ratio should increase beyond the ∼40% measured from our SED analysis of the C1 Fossil observations (Table 5).

Zoom In Zoom Out Reset image size

Owing to their rapid evolutionary formation timescales (∼Myr) and high dust production rates ($\dot{{M}_{d}}\sim {10}^{-10}\mbox{--}{10}^{-6}$ M_⊙ yr⁻¹; Lau et al. 2020), dust-forming WC binaries have been revisited in recent studies as early and important sources of interstellar dust (Marchenko & Moffat 2017; Lau et al. 2020, 2021, 2022). Therefore, the confirmation of nanodust formation around a colliding-wind WC binary system like WR 140 presents potentially significant implications on the origin of very small carbonaceous grains in the ISM of galaxies in the local and early universe. Here, we consider whether or not all dust-forming WC binaries are capable of nanodust production like WR 140 based on the RATD mechanism.

Newly formed dust in the circumstellar environment around all WC binaries should be exposed to a strong radiation field powered by the high luminosities (∼10⁵–10⁶ L_⊙) and hot photospheric temperatures (≳40,000 K) of the central WC star, as well as that of the OB-star companion. Based on Equations (7) and (8), even the lowest luminosity WC9 stars (L_* ∼ 10⁵ L_⊙) should be capable of disrupting grains via RATD that are larger than a_disr ∼ 100 Å at a distance of r_dust = 1650 au on timescales as short as τ_RATD ∼ 20 days. Interestingly, in their dust SED model fitting of Galactic dust-forming WC systems, Lau et al. (2020) find that the episodic dust-forming WC systems such as WR 140, WR 125, WR 137, and WR 19 are better fit by smaller grain size models (a = 0.01–0.1 μm), whereas the persistent dust-forming systems such as WR 95, WR 98a, WR 104, WR 106, and WR 118 are better fit by larger grain size models (a = 0.1–1.0 μm). The dust production rates are higher in the persistent dust-forming systems compared to the episodic systems, which suggests that self-shielding and larger optical depths could play a role in enhancing grain growth and mitigating the RATD mechanism in persistent dust makers.

The Lau et al. (2020) dust SED model fits of the persistent dust-forming WC systems that favor larger grains do not, however, preclude the presence of nanodust in these systems given the challenges of detecting nanodust emission. Indeed, previous mid-IR spectroscopic studies of dust-forming WC systems that included persistent dust-formers like WR 104 show that they exhibit UIR features, and therefore indicate the presence of very small carbonaceous grains and/or complex hydrocarbons (Chiar et al. 2002; Marchenko & Moffat 2017; Endo et al. 2022). Our observations of WR 140 notably demonstrate that high-angular resolution multi-wavelength IR imaging observations are crucial for distinguishing nanodust from the longer wavelength IR emission by cooler and larger dust grains, as well as the near-IR free–free emission from the ionized winds of the central WR star. Therefore, we speculate that all dust-forming WC binaries should exhibit nanodust production due to the RATD mechanism on larger grains. The nanodust-production efficiency may differ based on the dust-formation conditions around the WC binary.

Given that the evolutionary timescales to form WR stars (∼Myr) is much shorter than that of AGB stars (≳100 Myr), which have been considered as leading sources of dust in the local universe (e.g., Gall et al. 2011), we argue that WC binaries present an early source of C-rich nanodust. We speculate that nanodust formed by WC binaries are the C-rich aromatic carriers of the UIR features detected from dusty WC binaries (Endo et al. 2022; Lau et al. 2022). Nanodust from WC binaries also presents a potential candidate as a carrier of the 2175 Å UV extinction bump, which is thought to arise from very small carbonaceous grains (Stecher & Donn 1965; Duley & Seahra 1998; Draine 2009). Recent observations with JWST of galaxies from the first billion years of cosmic time notably present evidence of the 2175 Å absorption feature, which highlights the possible presence of carbonaceous dust in the early universe (Witstok et al. 2023). Continued ground-based, high contrast, and high-angular resolution near/mid-IR observations in combination with sensitive spaced-based observations with JWST will be essential to investigate the impact of nanodust formation in WC binaries.