IRAS 00450+7401 and the Mid-infrared Fade/Burst Cycle of R Coronae Borealis-type Stars

Stratospheric Observatory for Infrared Astronomy (SOFIA) imaging observations were obtained for IRAS 00450+7401 on multiple flights in 2022 with the FORCAST (Herter et al. 2013) and HAWC+ (Harper et al. 2018) instruments. All observations at wavelengths longer than 11 μm were performed at an aircraft altitude of ∼43,000 feet. Table 1 reports all observation dates and measured flux values.

While FORCAST hosts short- and long-wavelength channels (SW and LW, respectively) that are capable of simultaneously taking images, most observations of IRAS 00450+7401 utilized only a single channel at a time to avoid sensitivity loss from using the dichroic. Dual-mode observations that simultaneously used both the SW and LW channels were done for the 19.7 and 31.36 μm pair, and for the 25.24 and 33.57 μm pair. Observations were obtained in C2N (2-point, symmetrical) chopnod mode with a chop throw of ±30'' (resulting in one positive beam and two negative beams with half the intensity of the positive beam). The telescope was nodded in an ABBA pattern with nod offsets of 60''. The chopping and nodding position angles on the sky varied depending on the time of observation, but always aligned such that chopping and nodding angles were matched. Total on-source integration times ranged from roughly 1 to 8 minutes.

Images are reduced and calibrated into physical flux units by the SOFIA Data Cycle System (DCS) pipeline; pipeline-reduced images have pixel scales of 0768 pixel⁻¹. These Level 3 data products are retrieved from the SOFIA archive service. IRAS 00450+7401 is detected as a point source in all FORCAST imaging data with varying levels of signal-to-noise. As such, we attempted to follow the guidance of the FORCAST Photometry data analysis cookbook ⁵ for measuring photometry. In short, this entails using a 12 pixel (922) radius aperture (and larger sky annulus) to extract flux and estimate an uncertainty.

As discussed in detail within the cookbook and in Su et al. (2017), using aperture radii other than 12 pixels is not recommended for extracting flux as the point-spread function (PSF) is highly variable (even within a flight) and aperture corrections cannot be reliably established. IRAS 00450+7401 is detected with high signal-to-noise (≳20 at the aperture, meaning we take the signal in the aperture divided by the relative error in the photometry measurement as described in the FORCAST Photometry cookbook) for filters between 6 and 20 μm, but is only detected at moderate or low signal-to-noise (<15 at the aperture) for filters longward of 20 μm. Unfortunately, we found that standard 12 pixel radius aperture extractions for these low- to moderate-signal-to-noise images resulted in photometry dominated by noise; the resulting behavior of the source's flux between 30–40 μm appeared quite erratic and unphysical.

The FORCAST Photometry cookbook suggests, following the work of Su et al. (2017), that PSF fitting can be used to measure photometry when using high signal-to-noise observations of standard stars. Although the cookbook discusses this within the context of performing aperture corrections, we decided to explore the possibility of using PSF fitting to directly obtain photometric measurements. We use bright standard star images as an empirical model of the PSF, multiply them by a constant factor to match the image of interest for IRAS 00450+7401, and then perform the regular cookbook aperture extraction on the scaled standard star image. Such aperture extractions are not dominated by noise and should more faithfully measure the flux from the science target.

All standard star observations obtained on the same night and with the same filter as observations for IRAS 00450+7401 were downloaded. Standard star observations for the F315 filter were not obtained on the same night as F315 observations for IRAS 00450+7401, so we used F315 standard star data from the subsequent night. All standard star images are Level 3 data products and calibrated into physical flux units.

With in-house IDL routines we modeled each image of IRAS 00450+7401 with the same filter image of the standard star effectively obtaining a scaling factor for the standard star image. Aperture photometry with a 12 pixel radius aperture was then performed on the scaled standard star image while the uncertainty was calculated from the IRAS 00450+7401 image as specified in the cookbook. This approach yielded consistent results for filters between 6–20 μm and what appeared to be more accurate (at least more physically reasonable) results for filters longward of 20 μm.

For moderate and especially low signal-to-noise detections in FORCAST data it seems standard aperture photometry with a 12 pixel radius aperture can produce questionable results. We have found PSF fitting using standard star observations seems to produce more robust photometry values and strongly encourage such an approach when analyzing FORCAST data.

Far-infrared imaging was obtained toward IRAS 00450+7401 with HAWC+. Images were obtained at an aircraft altitude of 43,000 feet in total intensity mode with the Band A filter and used Lissajous scan imaging. Data are reduced within DCS and Level 3 calibrated data products are obtained from the SOFIA science archive. HAWC+ observations of IRAS 00450+7401 were obtained in one of the final SOFIA flights. During these last runs, HAWC+ experienced heightened consumption rates for its coolant that sometimes impacted end-of-flight targets. Unfortunately, IRAS 00450+7401 was one such target and HAWC+ ran out of coolant after ∼5 minutes of on-source observation. IRAS 00450+7401 was not detected in this short HAWC+ sequence and we report an upper limit in Table 1.

FORCAST spectroscopic observations were conducted with the G063 and G111 grisms in the SW channel using the 47 slit resulting in a spectral resolving power of ∼140.

G063 spectra were obtained on 2022 February 17 UT (MJD 59627.25208) and covered roughly 5–8 μm. During these observations the aircraft started at an altitude of 37,000 feet and ended at 39,000 feet. Chopping and nodding were performed in C2N mode. A total on-source integration time of 941 s was accrued with the G063 grism. G111 spectra were obtained on 2022 February 19 UT (MJD 59629.24306) and covered roughly 8.75–14 μm. The aircraft altitude for these observations was ∼39,000 feet and C2N chopping and nodding were performed. G111 observations amassed a total on-source integration time of 469 s.

Spectroscopic data are reduced, extracted, and calibrated within DCS and Level 3 data products retrieved from the SOFIA science archive. G063 data had manual telluric corrections performed with an ATRAN model computed for a precipitable water vapor content of 3 μm. The final spectral signal-to-noise ratio is ∼60 at 7 μm and ∼20 at 11.5 μm. However, this does not take into account systematic noise from telluric correction (especially apparent near the 9.5 μm ozone feature).

We incorporate NEOWISE 3.35 and 4.60 μm data into our analysis (Wright et al. 2010; Mainzer et al. 2014).

IRAS 00450+7401 is saturated in NEOWISE single-frame images. While the NEOWISE single-frame detection lists attempt to use the PSF wings of saturated sources to recover valid fluxes, the Explanatory Supplement ⁶ shows that there are still biases for the brightest sources (like IRAS 00450+7401).

We perform an additional correction to the NEOWISE-reported magnitudes retrieved for IRAS 00450+7401 as follows: W1/2_corr = W1/2mpro + W1/2mcorr for W1/2mpro < W1/2_satlimit and W1/2mcorr and W1/2_satlimit replaced by appropriate values for either W1 or W2. Uncertainties are the quadrature sum of the W1/2mpro catalog-quoted uncertainties and the uncertainty on W1/2mcorr.

Palomar Hale 200 inch (P200) near-infrared imaging was obtained for IRAS 00450+7401 on 2022 February 15 UT using the Wide-field Infrared Camera (WIRC; Wilson et al. 2003) and the 2048 × 2048 Hawaii-II detector that was installed in 2016. Observations were obtained in the J and H bands, although only one H-band image was obtained as the target source was found to saturate with the minimum exposure time of 0.92 s. Five un-dithered exposures of 0.92 s each were obtained in the J band. Two of the J-band images have peak counts for IRAS 00450+7401 that are well above the 1% nonlinearity level (but still below full-well); we do not find results from these images to be discrepant with those from the other images in the set and thus include them in our analysis (excluding them does not have a significant impact on the final measured values).

WIRC images were reduced by subtracting a 0.92 s dark exposure and then dividing by a twilight flat for each respective band. Photometric extractions were performed for IRAS 00450+7401 and several in-field reference stars with 2MASS magnitudes brighter than ∼12 in the J and H bands. An aperture that encloses ∼97% of the flux was used for all sources, resulting in ≲1% variance within the set of five J-band images.

To obtain an H-band magnitude for IRAS 00450+7401, we performed PSF fitting with bright, unsaturated point sources serving as the PSF reference. PSF fits were anchored on the unsaturated (and linear) wings of the source profile; saturated and strongly nonlinear pixels were masked. As a test case, we first attempted a PSF fit for a field source also saturated in the H-band image and found excellent agreement with the 2MASS H-band magnitude quoted for the source. Despite this encouraging result, we adopt a generous uncertainty of 0.1 mag for the IRAS 00450+7401 H-band measurement.

Final P200 infrared photometry measurements for IRAS 00450+7401 are reported in Table 1.

Optical photometric measurements were taken with the San Diego State 40 inch Mount Laguna Observatory (MLO) telescope with the goal of characterizing the optical brightness of the star contemporaneous with the infrared measurements. B-, V-, R-, and I-band images were taken on four separate epochs in February–March of 2022, along with another epoch in September of 2022 (see Table 2).

Data were reduced within IRAF, including performing bias subtraction and flat-fielding for each frame. Differential photometry for each frame was obtained for the science target with the nearby star TYC 4308 226 1 serving as a reference source (adopted magnitudes for TYC 4308 226 1 can be found in Table A1). Typically three frames for each filter were obtained and the final magnitude measurement is taken to be the average of the three frames with the uncertainty given by the standard deviation.

Final magnitudes are converted into flux density values in Jy following information provided in the NASA/IPAC Infrared Science Archive. ⁷ The complete set of optical photometry can be found in Table 2.

IRAS 00450+7401 was observed through clear skies on 2022 January 28 UT at Lick Observatory with the Kast Double Spectrograph mounted on the Shane 3 m telescope. Before placing IRAS 00450+7401 on the slit, we obtained guider camera images of the science target and the nearby reference star TYC 4308 226 1 through a Spinrad-R filter. From these data we estimate an R-band Vega magnitude for IRAS 00450+7401 on 2022 January 28 UT (MJD 59607.12210) of 14.4 ± 0.1.

Kast observations simultaneously employed the blue and red arms with light split by the d57 dichroic around 5700 Å. After splitting, blue light was passed through the 600/4310 grism while red light was passed through the 600/7500 grating. A slit width of 15 was used resulting in resolving powers of about 1100 in the blue spectra and 1750 in the red.

Kast data are reduced using standard IRAF long-slit tasks including bias subtraction, flat-fielding, wavelength calibration with arc lamps, and instrumental response calibration via observations of flux calibration standard stars. Arc lamp frames were not obtained close in time to science frames and as such the zero-point of the wavelength scale is not accurate. The final signal-to-noise ratio is ∼30 at 4900 Å in the blue spectrum and >50 across the entire red spectrum. We do not attempt to correct the Kast spectra for reddening.

We measured the brightness of IRAS 00450+7401 in optical bands to assess what level of dimming the star was experiencing as the infrared data were being collected. During the four initial epochs when visual data were taken, the star got consistently brighter and marginally bluer. In nearly a month, from 2022 February 05 UT to 2022 March 02 UT, the star brightened by ∼1 mag in the B, V, and R bands and brightened by ∼0.6 mag in the I-band (Table 2). The R−I color of IRAS 00450+7401 decreased from 1.56 ± 0.08 to 1.31 ± 0.05 over the same time span. A fifth epoch of optical photometry on 2022 September 15 UT revealed that the star had dipped again in brightness, fading by nearly 4 mag in the R band compared to the 2022 March epoch.

To better put MLO optical monitoring into context we utilized the dreams ⁸ monitoring webserver that contains brightness records for many R Cor Bor stars. ATLAS o-band monitoring (Tonry et al. 2018) provides the most complete modern optical photometry data set for IRAS 00450+7401 covering the times when the new data discussed above were being obtained and several years leading up to their acquisition. The ATLAS data also covered what appears to be the brightest magnitudes and hence what we tentatively interpret as maximum light for IRAS 00450+7401. Around MJD 57966 (UT 2017 August 1), IRAS 00450+7401 had an o-band magnitude of 12.17 ± 0.03. Following that maximum, IRAS 00450+7401 fell into a deep dimming state (o-band magnitudes consistently fainter than 15 and frequently >18) before beginning a resurgence around MJD 59490 (roughly 2021 October). At the time of Kast observations (where we measured a rough R-band magnitude of 14.4, see Section 2.6) ATLAS measured an o-band magnitude of 14.70 ± 0.03. The star continued to brighten from that epoch consistent with MLO measurements, but unfortunately ATLAS did not visit the star between 2022 March 02 UT and 2022 June 11 UT (presumably because it was at poor elevation). By the time ATLAS measurements for IRAS 00450+7401 resumed, it had begun a new decline confirmed with the faint state measured at MLO in 2022 September.

In a recent work Crawford et al. (2023) define a spectral classification system for hydrogen-deficient carbon stars, including R Cor Bor stars. While mostly based on observed spectral appearance, the new classification system allows one to make a rough estimate of the temperature for a star of known spectral type.

Figure 1 shows a comparison between optical spectra for IRAS 00450+7401 and the earliest-type matching standard from Crawford et al. (2023). Earlier-type standards show a clear mismatch in the position of the steep molecular shelves (e.g., near 4700 Å for IRAS 00450+7401) or a lack of such features altogether, ruling out the HdC4 and earlier classifications (see Figure 4 of Crawford et al. 2023).

Zoom In Zoom Out Reset image size

Confounding a more exact spectral type estimate for IRAS 00450+7401 is the fact that the Kast spectra were obtained when the source was still recovering from a prolonged and deep dimming event (see Section 3.1). The Kast spectra appear to have been obtained when the source was ∼2.4 mag below maximum light while all spectral standards from Crawford et al. (2023) were obtained near maximum light or at most 0.5 mag below it.

Applying an A_V = 2.23 magnitude (assumed R_V = 3.1; Hecht et al. 1984 and references therein report R_V values of ∼3 for R Cor Bor stars recovering from dimming events) reddening correction to the Kast spectra for IRAS 00450+7401 results in excellent agreement with the HdC5 standard shown in Figure 1. We cannot verify such a V-band extinction value with the available photometric monitoring data for IRAS 00450+7401. The Shane-measured R-band magnitude obtained on the same night as the Kast spectra (Section 2.6) and the ATLAS photometric monitoring around the same time (Section 3.1) suggest that A_R∼2.4 would be more appropriate (this would correspond to A_V = 3.06 for R_V = 3.1).

R Cor Bor stars appear to have hot gas surrounding them that becomes more apparent during declines (e.g., Clayton et al. 1992; Section 3.2.2 of Clayton 1996; Kameswara Rao et al. 2004). While reddening by dust can impact the overall slope of the Kast spectra, veiling by such emission can act to weaken spectral features giving an R Cor Bor star the appearance of an earlier spectral type (e.g., Lambert et al. 1990 and references therein). What is not known is at what dimming state such veiling begins to become important (for the case of IRAS 00450+7401, such emission appears to be present in near-infrared spectra taken during a dimming event observed by Karambelkar et al. 2021).

Given the uncertainties around veiling for IRAS 00450+7401 at the time of the Kast spectroscopic observations, it is possible that it could be an HdC6 or even HdC7 type star. The best way to resolve this issue would be to obtain a spectrum of IRAS 00450+7401 near maximum light, although it is not clear when that may happen again. Until that time, we conservatively assign it a spectral type of HdC5 or later.

Based on Table 5 of Crawford et al. (2023), a spectral type of HdC5 or later corresponds to a stellar effective temperature of ≲4500 K. We thus conclude that IRAS 00450+7401 is a cool R Cor Bor star.

Mid-infrared spectroscopic data have been collected for ∼35 R Cor Bor stars with ISO and Spitzer (Lambert et al. 2001; Kraemer et al. 2005; García-Hernández et al. 2011a; García-Hernández et al. 2011b; García-Hernández et al. 2013). These mid-infrared spectra reveal a (likely carbon dust) "hump" feature between 6 and 8 μm for most R Cor Bor stars, emission from polycyclic aromatic hydrocarbons for ∼10% of R Cor Bor stars, and a smaller fraction still that may be host to C₆₀ fullerenes.

The FORCAST spectra for IRAS 00450+7401 demonstrate that this R Cor Bor star falls in line with the majority of other R Cor Bor systems, displaying only a mild 6–8 μm hump (Figure 2). In an effort to better categorize IRAS 00450+7401 among other R Cor Bor stars, we attempted to match Spitzer IRS spectra presented by García-Hernández et al. (2011b) with the FORCAST spectra for IRAS 00450+7401. IRS spectra are obtained from the CASSIS database (Lebouteiller et al. 2011) and are presented as retrieved. The best match overall is V739 Sgr, which is shown in Figure 2. However, we note that V348 Sgr, Z Umi, RZ Nor, and WX Cra also provide acceptable matches to the IRAS 00450+7401 FORCAST spectra. Table 4 of Crawford et al. (2023) reports spectral types for four out of five of these R Cor Bor stars, with all but RZ Nor probably belonging to cool subtypes (types of HdC5 or later). This small sample possibly indicates that R Cor Bor dust does not depend on the temperature class of the host star, a finding that should be more exhaustively investigated before being considered conclusive.

Zoom In Zoom Out Reset image size

Figure 3 presents a 2022 epoch spectral energy distribution for IRAS 00450+7401 with an accompanying blackbody fit for data points >10 μm to roughly characterize the dust temperature as measured in the mid-infrared (T_MIR). The data cover 0.44–53.56 μm with 53.56 μm being an upper limit and the blackbody fit yields a temperature of T_MIR = 900 ± 90 K. When doing this blackbody fit we do not include data at wavelengths less than 10 μm; as seen in Figure 3 the fit overpredicts the flux at 3–6 μm wavelengths. Fitting a 3-blackbody decomposition to all measurements from Table 1—representing contributions from the star, an inner dust component, and an outer dust component—results in temperatures of 2700, 1350, and 650 K, respectively. Such a fit reasonably reproduces all Table 1 measurements and is not unique, but outer dust component temperatures <600 K are ruled out.

Zoom In Zoom Out Reset image size

While the three-blackbody fitting procedure produces a better fit to the data, we will employ throughout this paper the mid-infrared-only fitting approach. As will be seen in the next section, this approach is necessary to maintain a consistent and homogeneous dust temperature measurement methodology when working with IRAS and AKARI data that only have data near 10 and 20 μm. Temperatures estimated following the T_MIR approach are better viewed as mid-infrared dust color temperatures since they miss contributions of dust components closer to the star.

Table 3 of García-Hernández et al. (2011b) presents multiblackbody-fit dust temperatures for ∼30 R Cor Bor stars observed with Spitzer IRS. They find inner dust components ranging in temperature from 272–1600 K and outer dust components (when present) ranging from 100–900 K, thus placing the mid-infrared temperatures we report for IRAS 00450+7401 somewhere near the high-end of the mean.