A HaloSat Analysis of the Cygnus Superbubble
Jesse Bluem, Philip Kaaret, William Fuelberth, Anna Zajczyk, Daniel M. LaRocca, R. Ringuette, Keith M. Jahoda, K. D. Kuntz
aa r X i v : . [ a s t r o - ph . H E ] J a n Draft version January 13, 2021
Typeset using L A TEX twocolumn style in AASTeX63
A HaloSat Analysis of the Cygnus Superbubble
Jesse Bluem, Philip Kaaret, William Fuelberth, Anna Zajczyk,
1, 2,3
Daniel M. LaRocca, R. Ringuette, Keith M. Jahoda, and K. D. Kuntz
2, 4 University of Iowa Department of Physics and Astronomy, Van Allen Hall, 30 N. Dubuque St., Iowa City, IA 52242, USA NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA Center for Space Sciences and Technology, University of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore, MD 21250, USA The Henry A. Rowland Department of Physics and Astronomy, Johns Hopkins University, 3701 San Martin Dr., Baltimore, MD 21218,USA (Received X N, 2019; Revised Y N, 2019; Accepted January 13, 2021)
Submitted to ApJABSTRACTThe Cygnus Superbubble (CSB) is a region of soft X-ray emission approximately 13 degrees wide inthe direction of the local spiral arm. Such a large region might be the result of strong stellar windsand supernovae from nearby stellar nurseries, or it could be the result of a single event - a hypernova.HaloSat observed 4 non-overlapping 10 degree diameter fields in the CSB region over the 0.4-7 keVband. The CSB absorption and temperature was found to be consistent over all 4 fields, with aweighted average of 6 . × cm − and 0 .
190 keV, respectively. These observations suggest that theCSB is a cohesive object with a singular origin. The total thermal energy for the CSB is estimatedat 4 × erg, based upon a shell-like physical model of the CSB. Absorption and distance estimatesto Cyg OB associations are examined. The CSB absorption is found to be most consistent with theabsorption seen in Cyg OB1, implying that the CSB lies at a similar distance of 1.1-1.4 kpc. Keywords: INTRODUCTIONIn 1980, an extended soft X-ray structure was discov-ered by HEAO 1 near the plane of the galaxy, in thedirection the constellation Cygnus (Cash et al. 1980).This region of soft X-rays was found to be related to pre-vious IR, optical, and radio structures observed in thesame region, and is now commonly referred to as theCygnus Superbubble (CSB). Cash et al. (1980) foundthe X-ray emission was spread over 13 degrees of thesky, corresponding to a diameter of 450 pc at roughly 2kpc distance, as estimated by absorption measurements.The CSB appears to resemble a horseshoe at first glance,but this is an artifact induced by the intervening CygnusRift (or, if the reader prefers, the Northern Coalsack orGreat Rift of Cygnus), a large dust cloud that obscuresthe central region of the CSB. Near the CSB are 9 OBassociations, including the notable Cygnus OB2 associ- [email protected] ation. Cygnus OB2 contains more than 100 O spectralclass stars, making it the largest such assemblage of Ostars and the highest mass of any young stellar associa-tion yet detected in our galaxy (Kn¨odlseder 2000).When we look towards the CSB, we are looking downthe length of the local spiral arm. Separate objects alongthis line of sight are superposed on this region of thesky, making it difficult to determine if observed struc-tures are discrete objects or multiple superposed objects.This effect, combined with conflicting measurements ofthe distance to regions of the bubble, has made under-standing the precise nature of the CSB difficult.One way to study distance is to look at absorption,typically parameterized by the total hydrogen columndensity. The farther away an object is, the more inter-vening Galactic material absorbs light radiated by theobject. In the case of the CSB, conflicting measure-ments of N H have resulted in evidence supporting bothcomposite and discrete origins for the observed struc-ture. Uyaniker et al. (2001) found differing N H valuesfor different regions of the CSB, implying that the CSB Bluem et al. is a composite object dependent upon our particular lineof sight along the spiral arm. However, more recentlyKimura et al. (2013) has found similar N H values acrossthe CSB - inferring that the CSB is a unified structureand not a line of sight composite.If the CSB is a unified structure, then explaining thelarge size of the region becomes problematic. Cash et al.(1980) estimated the total thermal energy of the CSB asexceeding 6 × erg, for a distance of 2 kpc, and favoran interpretation of a series of 30-100 supernovae ratherthan a single event. If the CSB did originate from asingular event, it would have to be a very rare sort ofsupernova, referred to as a hypernova (Paczy´nski 1998).Some observational evidence of these hypernovae exist.SN1998bw is a supernova with an estimated 2 − × erg initial kinetic energy, putting it an order of mag-nitude above other supernova, and may have stemmedfrom a progenitor star with a mass of 40 solar masses(Iwamoto et al. 1998). This observed energy is similarto the energy observed in the CSB.Another possibility is a combination of multiple super-novae and/or stellar winds from the OB associations inthe area. Cygnus OB2 in particular could be involved,being one of the most impressive star formation regionsin the galaxy. An analysis of the Cyg OB2 associa-tion found significant mass loss rates for some of themost notable members of the association, and that sucha strong wind could conceivably generate a bubble onthe scale of the CSB in 2 million years (Abbott et al.1981). Cash et al. (1980) suggests that the appearanceof a wind-blown bubble could be achieved by 30-100 su-pernova over the preceding 3-10 million years, meaningthat the current most massive Cyg OB2 association starsmust be from a younger generation than the progenitorsof these novae. However, given that Cygnus OB2 is off-set from the center of the fairly circular CSB, it is lesslikely to be the direct source of the bubble, be it fromwinds or supernovae. This is not an issue for a hyper-nova - the offset could be explained by a runaway starejected from Cyg OB2 (Kimura et al. 2013).Observations of the CSB were taken by HaloSat in Oc-tober 2018, September 2019, and October 2019. Halosatis specifically suited for studying diffuse emission, withan observing program that emphasizes reducing fore-ground contamination. The HaloSat energy range goesdown to 0.4 keV, allowing for good measurements of ab-sorption. Section 2 describes the details of these obser-vations. Section 3 covers the spectral analysis. Section 4details estimating a distance to the CSB by using fittedabsorption and previous literature on the Cyg OB as-sociations. Section 5 describes the spectral results and estimates global parameters of the CSB using the esti-mated distance. OBSERVATIONSHaloSat is a NASA funded CubeSat with instrumen-tation developed at the University of Iowa and was de-ployed from the International Space Station on 2018July 13 (Kaaret et al. 2019; LaRocca et al. 2020a). Us-ing three silicon drift detectors, HaloSat specializes inobserving diffuse X-ray emission, including extendedsources. The average energy resolution of these detec-tors is 85 eV at 676 eV (fluorine K α ) and 137 eV at5.9 keV (manganese K α ) (Kaaret et al. 2019). HaloSathas full response over a 10 degree diameter field of view,which then falls off linearly to zero at 14 degrees.The CSB was covered with four initial fields (HS0013,HS0014, HS0015, HS0016), selected with the intent ofcovering as much of the CSB as possible. The point-ing coordinates for the fields were chosen to avoid thestrong X-ray sources Cygnus X-1 and X-2, as well asto avoid the nearby Cygnus Loop, which was itself aseparate target observed by HaloSat. Early observa-tions were taken with a misalignment between the space-craft’s boresight and the intended observation target,resulting in a small offset of 1 degree. These early off-set fields (HS0355, HS0357, HS0358) were given theirown HaloSat IDs and analyzed separately from the laterobservations. HS0014’s pointing was moved before anydata was taken, becoming HS0356 and resulting in noobservations being taken under the HS0014 ID. TheHaloSat fields can be seen overlaid on a ROSAT 3/4keV (R4+R5) map (Snowden et al. 1997) in Figure 1.Cuts were performed based on a 0.16 ct s − countrate in the HaloSat “hard” band (3-7 keV) and a 0.75 cts − count rate in the HaloSat “very large event” band(7+ keV). Some observations with high particle-inducedbackgrounds (HS0359 and HS0360) were removed due tosignificant data loss from the cuts used. Table 1 includesthe observation parameters for each field, including rightascension and declination for the field pointing, nadirangle, and anti-Sun angles at the time or times of theobservations, and the exposure time and hard count ratefor each detector after cuts were applied. ANALYSISThe CSB spectra were analyzed with XSPEC version12.10.1f (Arnaud 1996). For each field, the CSB is mod-eled as collisionally-ionized diffuse gas using an absorbed apec model (Smith et al. 2001). Also included are mul-tiple background/foreground components. These in-clude detector specific power-laws for the particle back-grounds, fixed models for the absorbed cosmic X-ray
HaloSat Analysis of the Cygnus Superbubble Figure 1. ROSAT 3/4 keV (R4+R5) map (Snowden et al. 1997) showing the HaloSat CSB fields. Primary targets are markedwith an x and offset targets are marked with a cross. Cygnus X-1, X-2, and X-3 are marked with boxes. The plane of the MilkyWay is the line through the middle of the figure. For each field, the solid circle is the HaloSat 10 degree diameter full-responsefield of view. Only the fields of view for the primary targets are shown. The dashed circle is an example of where the HaloSatresponse drops to zero (14 degree diameter). The color bar is in units of 10 − ct s − arcmin − . background (CXB) and unabsorbed local hot bubble(LHB), and absorbed non-equilibrium ionization (NEI)models for Galactic ridge emission. Field HS0016 in-cludes absorbed model components for Cygnus X-3 (CygX-3) found from fitting MAXI data over the same timeinterval. Due to the CSB being in the Galactic plane,and the high column densities therein, the Galactic halois not included in the modeled background. All absorp-tions used the tbabs model and the associated Wilmsabundances (Wilms et al. 2000). All model componentsare folded through the HaloSat response matrix, exceptfor the instrumental particle backgrounds (Kaaret et al.2019; Zajczyk et al. 2020).The apec model used to represent the CSB is fit withfreed absorption, temperature, and normalization. Theabundances are fixed to 0.26, as found for the CSB in Kimura et al. (2013). The spectra for each HaloSat fieldcan be seen in Figures 2 and 3 and the fitted modelparameters can be found in Table 2.The particle backgrounds are power-laws (XSPECmodel powerlaw ) specific to each detector, with fixedphoton indices, unique to each detector, calculated us-ing an empirical relation for HaloSat’s detectors basedon an analysis of the southern halo (Galactic latitude < -30) (Kaaret et al. 2020). The power-law normaliza-tions are fit between 3 and 6.9 keV (some fields do nothave enough data for the bins used to extend to 7 keV,so a limit of 6.9 keV is used for all fields) and then fixedbefore any other model parameters are fit. All othermodel parameters are linked together between the threedetectors. The exceptions to this fitting procedure arefields HS0357 and HS0016. The calculated power-law Bluem et al. no r m a li ze d c oun t s s − k e V − HS001310.5 2−202 ( d a t a − m od e l ) / e rr o r Energy (keV) 0.010.10.020.050.2 no r m a li ze d c oun t s s − k e V − HS035510.5 2−202 ( d a t a − m od e l ) / e rr o r Energy (keV)0.010.10.020.050.2 no r m a li ze d c oun t s s − k e V − HS035610.5 2−202 ( d a t a − m od e l ) / e rr o r Energy (keV) 0.010.10.020.050.2 no r m a li ze d c oun t s s − k e V − HS035710.5 2−4−202 ( d a t a − m od e l ) / e rr o r Energy (keV)0.10.020.050.2 no r m a li ze d c oun t s s − k e V − HS001510.5 2−2−1012 ( d a t a − m od e l ) / e rr o r Energy (keV) 0.010.10.020.050.2 no r m a li ze d c oun t s s − k e V − HS035810.5 2−202 ( d a t a − m od e l ) / e rr o r Energy (keV)
Figure 2. Spectra for the HaloSat CSB fields. Overlapping fields are adjacent to each other. Each field is labeled with its HaloSatID at the top of the spectrum. Each detector is represented with a different color. Detector 14 is black, 38 is orange, and 54 isteal. The apec is the dominant component in each field, marked with a thick solid line. The NEI background components aremarked with dashed lines. All other background components are marked with thin solid lines.
HaloSat Analysis of the Cygnus Superbubble Table 1. Observation ParametersHS ID RA dec nadir solar angle detector exposure hard rate(deg) (deg) (deg) (deg) (s) (c s − )HS0013 318 .
49 +43 .
88 145 124 14 12224 0.07638 12480 0.06654 12928 0.070HS0355 319 .
43 +43 .
18 116 116 14 16320 0.05338 15936 0.04854 15680 0.050HS0356 307 .
97 +48 .
99 144, 174 116, 112 14 13888 0.09738 10310 0.09854 17216 0.090HS0357 308 .
48 +50 .
29 118 108 14 19264 0.05838 19136 0.05254 19200 0.054HS0015 294 .
96 +42 .
04 131 114 14 6656 0.08238 8256 0.08554 8128 0.088HS0358 295 .
67 +42 .
87 144 105 14 11584 0.06538 11328 0.06754 11456 0.064HS0016 308 .
43 +40 .
57 151 117 14 30144 0.08638 31616 0.08354 32384 0.084
Note —Earlier offset fields are paired with their later counterpart fields. Column 1 is theHaloSat field ID, Column 2 and 3 are right ascension and declination. Column 4 and5 are average nadir angle and average solar angle for the duration of the observation.Column 6, 7, and 8 are specific to each detector for each observation. Column 6 is thedetector unit ID, column 7 is exposure time for each detector after cuts, and column8 is the hard count rate in the detector.
Bluem et al. no r m a li ze d c oun t s s − k e V − HS001610.5 2−202 ( d a t a − m od e l ) / e rr o r Energy (keV)
Figure 3. Spectrum for the HaloSat CSB field HS0016. Each detector is represented with a different color. The apec is thedominant component in the field, marked with a thick solid line. The NEI background components are marked with dashedlines. All other background components are marked with thin solid lines. photon indices for HS0357 do not match the slope ofthe data, so the photon indices are left free and arefit over the full data range of 0.4 to 6.9 keV while si-multaneously fitting the other model parameters. TheHS0016 power-law normalizations are also fit over thefull range, simultaneously with the other model param-eters. Fitting the power-law at the high energies alone isproblematic because the Cyg X-3 components affect thefull energy range. For all fields, all particle backgroundpower-law components are fixed, and the energy rangeis narrowed to 0.4-3 keV for the final fit of the othermodel components.The full Galactic absorption for each HaloSat fieldis calculated using Planck dust opacity ( τ ) maps, asopacity serves as a better tracer of the total absorptioncolumn density at the high levels of absorption seen to-wards the CSB (Planck Collaboration et al. 2014). Theopacity maps are first converted to E(B-V) maps follow-ing Planck Collaboration et al. (2014). Then the E(B-V) map is converted to N H following Zhu et al. (2017).Next, the absorption is weighted by the HaloSat re-sponse and the best-fit equivalent N H value is found bycombining the shapes of the weighted absorption curvesfound over the entire field (LaRocca et al. 2020b). Thisproduces a single-valued N H that more appropriately re-flects the range of absorptions over the extended fieldwhen compared to a simple average absorption. As theobservations are close to the plane of the galaxy, the col- umn densities vary significantly over the entire CSB re-gion, and as such, vary significantly within each HaloSatfield. This absorption is applied to both the CXB andNEI components.The LHB is modeled as a collisionally-ionized diffusegas apec model. The parameters for the LHB apec are fixed in XSPEC according to values from Liu et al.(2017) ( kT = 0.097 keV), with the normalization calcu-lated from the emission measure. Each emission mea-sure is the average value from the Liu et al. (2017) mapfor a circular region corresponding to a 5 degree radius(the HaloSat full-response field of view). The CSB re-gion is extrapolated from surrounding areas in Liu et al.(2017) due to infringing X-ray sources making the LHBcomponent difficult to extract in that area. Overall, theLHB is not expected to vary much in any particular di-rection, so the CSB regions being based on extrapolatedemission measures should not be a concern.The CXB is modeled by an absorbed power-law(XSPEC model powerlaw ), using the photon index andnormalization from Cappelluti et al. (2017). The nor-malization is adjusted for the HaloSat field of view. Thefull Galactic absorption is sufficiently high that the ther-mal components of the CXB are insignificant.Another background component stems from the planeof the Galaxy cutting through the middle of the CSBregion. It has been documented in multiple articles(e.g. Iwan et al. (1982), Valinia & Marshall (1998), HaloSat Analysis of the Cygnus Superbubble nei ) componentmodel for fields 30 degrees from Galactic center. Theirhard component has model parameters of temperature kT = 7 keV, ionization timescale log n e t cm − s = 10 . Z = 0 .
8. The soft NEI component hasmodel parameters kT = 0 . n e t cm − s = 9 . Z = 0 .
6. Both components are included with all pa-rameters fixed to the values from Kaneda et al. (1997),except for normalization, which is left free for fitting.In most fields, one of the components is significantlyobserved while the second is only observed as an upperlimit. Additionally, different components are detected indifferent regions of the CSB (although all significantlyoverlapping fields are consistent). This is not unusualgiven the large field of view of HaloSat, the differing ori-entation of each field relative to the plane of the galaxy,and the relatively low scale heights of the NEI compo-nents. As an approximation of the true column density,the NEI components were also subjected to full galacticabsorption using the same value of N H as applied to theCXB.Field HS0016 includes the highly variable X-ray bi-nary Cyg X-3 within its field of view. The MAXI(Matsuoka et al. 2009) light curve for Cyg X-3 over thetime interval of the HS0016 observations was found tobe non-variable. We fit the MAXI data to find the spec-tral shape of the Cyg X-3 emission at the time of theHaloSat observation. Hjalmarsdotter et al. (2009) men-tions a blackbody spectrum combined with a power-lawto be a sufficient fit for Cyg X-3, although it is a non-physical simplification. This was found to be a goodstarting point for a Cyg X-3 fit, with an additional Gaus-sian component needed to handle the strong iron com-plex often observed in Cyg X-3 (Kallman et al. 2019).The central value for this Gaussian was found to besimilar to the value of a single Gaussian fit for the ironcomplex in Hjalmarsdotter et al. (2009). Because CygX-3 is a background component rather than a scientifictarget of interest in this analysis, this model was quiteacceptable with a reduced χ of 1.05. The fit parame-ters found for the MAXI fit (XSPEC models powerlaw+ bbody + gauss ) were as follows: a power-law photonindex of 0.0606, a blackbody kT of 1.81 keV, and a Gaus-sian line energy of 6.46 keV and a Gaussian line width of 0.221 keV. These parameters were then fixed for theHalosat spectra, while the normalizations of the com-ponents were allowed to vary while linked to each otherin the same ratios as were found in the MAXI fit. Thisconserves the overall spectral shape of Cyg X-3 in theHaloSat fits. Due to Cyg X-3 being quite a bit furtheraway than the CSB (approximately 3-9 kpc, Ling et al.(2009)), it was also subjected to the same full N H valueas the CXB and NEI components. DISTANCE ESTIMATESA distance must be assumed for further analysis ofthe properties of the CSB. Previous CSB analyses havebeen performed with the CSB distance being equivalentto the distance to the Cygnus OB2 association (CygOB2). This assumption results in the estimated dis-tance to the CSB changing over time as distance esti-mates to Cyg OB2 are refined. Cash et al. (1980) esti-mated a distance of 2 kpc based on their fitted absorp-tion of 0 . − . × cm − . Kimura et al. (2013)used a distance estimate of 1.7 kpc, predicated uponthe CSB being associated with Cyg OB2. Kimura et al.(2013) further justifies linking the distance to OB2 bycomparing their fitted absorption of 0 . − . × cm − to a set of stars within Cyg OB2 studied inYoshida et al. (2011), stating that that paper found afitted absorption of 0 . − . × cm − . Unfor-tunately, the Yoshida et al. (2011) absorptions are fitonly for the local circumstellar medium of those OB2stars, which is combined with a fixed ISM absorption of0 . − . × cm − from Waldron et al. (1998). Sim-ilar high values of extinction of A V ∼ . − .
07 ( N H ∼ . − . × cm − ) are found by Kiminki et al.(2015). As such, the absorption found by Kimura et al.(2013) is not consistent with the Cyg OB2 stellar asso-ciation. Neither is the absorption found in this paper.Since our measured absorption is inconsistent with thelocation of the CSB at the distance of Cyg OB2, lookingat other OB associations in Cygnus becomes a point ofinterest.Cygnus has 9 OB associations within the greater re-gion, but any related OB association is likely to be in-side the CSB rather than outside. This limits the as-sociations to Cyg OB1, OB2, OB3, OB4, OB6, OB8,and OB9. Uyaniker et al. (2001) has an excellent sum-mary of estimated distances as of 2001, but much morehas been learned in the last 19 years. Many newerestimates place Cyg OB2 closer, for example, ∼ . . − .
44 kpc (Kiminki et al. 2015). More recently,the distance to Cyg OB2 has also been studied usingGaia parallax measurements (Gaia Collaboration et al.
Bluem et al.
Table 2. Model Parameters (0.4-3.0 keV)HS ID l b hard NEI norm soft NEI norm apec N H apec kT apec norm chi / DoF(deg) (deg) (10 cm − ) (keV)HS0013 87.0 − . . +0 . − . < . . +0 . − . . +0 . − . +37 − − . . ± . < . . +0 . − . . +0 . − . +47 − . . ± . < . . +0 . − . . +0 . − . +36 − . . ± . < . . +0 . − . . +0 . − . +25 − . < . . +0 . − . . +0 . − . . ± .
03 26 +46 − . < .
03 0 . ± . . +0 . − . . +0 . − . +28 − . . ± . < . . ± .
04 0 . +0 . − . +21 − Note —Column 1 is the HaloSat target ID. Columns 2 and 3 are galactic coordinates of the field. Column 4 is the normalizationof the 7 KeV NEI component and Column 5 is the normalization of the 0.8 KeV NEI component. Columns 6 through 8 arethe CSB apec parameters. Column 9 is the χ over degrees of freedom for the fit over the 0.4-3 KeV interval. . ± . . +0 . − . kpc and 1 . +0 . − . kpc. Most of the uncer-tainty in the Berlanas et al. (2019) distance estimatesstems from systematic uncertainty in the Gaia paral-laxes (see Stassun & Torres (2018)). The OB associ-ation summary from Uyaniker et al. (2001) points to-wards Cyg OB3 being further away than Cyg OB2, andCyg OB4 being much closer. Mel’nik & Efremov (1995)performed a survey of stellar associations and split CygOB3 into two associations with distances of 1.82 kpcand 2.31 kpc. Cyg OB3, OB4, and OB6 are also moreon the outer edge of the CSB than the other associa-tions. Additional literature on Cyg OB3, OB4, and OB6is sparse, but the CSB being related to any of them ap-pears unlikely. This leaves Cyg OB1, OB8, and OB9 asthe remaining possibilities.Mel’nik & Efremov (1995) found Cyg OB1, OB8, andOB9 to be the same association at a distance of 1.37kpc. Kharchenko et al. (2005) found a distance for CygOB1 of 1.148 kpc and a distance for Cyg OB9 of 1.139kpc. These distance measurements support these as-sociations being interconnected. Straiˇzys et al. (2014)studied stars within the M29 (NGC 6913) cluster, itselffound within Cyg OB1, and found a distance for M29of 1 . ± .
15 kpc and a range of extinctions between A V = 2 .
45 and A V = 3 .
83 ( N H ∼ . − . × cm − ). This level of absorption found for stars in Cyg OB1 is more consistent with our measurements of theCSB absorption than those found for Cyg OB2.As a result of these new distances, Cyg OB1 and OB2would seem to be much closer to each other than previ-ously assumed. There are other observations that pro-vide additional evidence of these associations being closetogether. S106 is a molecular cloud generally thoughtto be at ∼ . ∼ . − . × cm − for potential embedded ob-jects in S106. While those objects are heavily obscureddue to their host cloud, the lower end of the absorptionrange would be consistent with a distance similar to CygOB1. The higher absorption seen in Cyg OB2 may becaused by these clouds, with the less absorbed Cyg OB1positioned more in front of the clouds. This would beconsistent with the estimated distances for Cyg OB2 be-ing slightly further away than Cyg OB1. We concludethat the CSB may then be positioned on the near sideof Cyg OB1, possibly near the foreground population ofCyg OB2 stars from Berlanas et al. (2019), somewherebetween 1.1 and 1.4 kpc. RESULTS AND DISCUSSION
HaloSat Analysis of the Cygnus Superbubble Figure 4. Plots of absorption and temperature for the HaloSat CSB fields. Top panel compares the fitted absorption with errorfor each field to the weighted average absorption (dashed line). Bottom panel compares the fitted temperature with error foreach field to the weighted average temperature (dashed line). Overlapping CSB fields are closer to each other than they are toother fields in the plots.
The fitted model parameters for absorption and tem-perature in the different HaloSat fields (Table 2) areconsistent with each other. Taking the weighted aver-age for absorption returns a value of 0 . × cm − for the data having a χ /dof deviation from the aver-age of 5.6/6. The weighted average for temperature is0.190 keV with a χ /dof deviation for the data from theaverage of 3.9/6. Figure 4 is a plot of the fitted absorp-tions and temperatures, and includes a comparison tothe weighted averages.While the best-fit column densities for HaloSat donot match those from Kimura et al. (2013), the valuesremaining consistent across the CSB regions still rein-forces the conclusion that the CSB is a single structure.The Kimura et al. (2013) model fits are simultaneousfits of ROSAT and MAXI spectra, and the resultingmodel fits for the MAXI data exhibit excess emissionat the lowest energies (0.7-0.8 kev). The HaloSat datagoes down to 0.4 keV, exhibits no such issue with ex-cess emission (see Figures 2 and 3), and avoids poten- tial sources of error stemming from simultaneously fit-ting two separate observatory’s data. Absorption hasa significant effect below 0.5 keV, so the energy rangeof HaloSat is sufficient in this regard. Also note thatwhile the calculated column densities from HaloSat ex-ceed Kimura et al. (2013) by a fair margin, they arewithin the range of column densities (0 . − . × N H ) found by Uyaniker et al. (2001), and are consistentwith the absorption of 0 . − . × cm − found byCash et al. (1980).Using the best-fit spectral parameters in Table 2, sev-eral additional physical characteristics of the CSB canbe calculated. First some initial assumptions about aphysical model of the CSB must be made. As seen inFigure 5, the CO contours (Dame et al. 2001) in thisregion trace gaps in emission (the CO contours alsotrace the dust opacity from Planck Collaboration et al.(2014)) and provide evidence that the southern arc ofthe CSB is not bound by absorption, rather it is an ex-pression of the actual structure of the CSB. This also0 Bluem et al.
Figure 5. ROSAT 3/4 keV (R4+R5) map (Snowden et al. 1997) showing the CSB physical model. The CSB emission contouris red. The blue CO contours from Dame et al. (2001) stretch across the middle of the image. The CSB is primarily modeledwith a shell, which is the thicker black lines. Two secondary regions to the upper right and lower left are modeled as spherespartially embedded in the shell and are marked with black lines as well. The color bar is in units of 10 − ct s − arcmin − . appears to be true of the upper edge of the northern arcas well. Due to the strength of the emission droppingoff in the CSB interior, the central volume of the CSBis assumed to form a shell and the thickness of the shellis based on the thickness of this southern arc, matchedwith the outer boundary of the northern arc. Gaps inthe shell are assumed to be absorbed by the interven-ing dust, which the CO contours trace. Two signifi-cant regions of emission remain outside the shell, andthese have been assumed to be spherical as they exhibitno obvious sign of being additional shell-like remnants.These two regions may be separate supernova remnantsin the region, or might be low-density voids where theCSB expansion has propagated faster. The CSB shellmodel is centered at galactic coordinates l = 82 . ◦ and b = − . ◦ , with the two spherical regions at l = 75 . ◦ , b = +6 . ◦ and l = 89 . ◦ , b = − . ◦ . The outer radius of the shell is 8.5 degrees and the inner radius is 4.0 de-grees. The secondary spheres each have a radius of 3.5degrees.In order to calculate the physical parameters of in-terest, an estimate of path length through the observedsections of the CSB is needed. This is done by gener-ating a grid of path lengths across the CSB model andremoving values that lie outside the CSB contour underthe aforementioned assumption that the emission fromthose parts are heavily absorbed. The two additionalspherical components are treated as being embedded inthe shell, so for any given point the greater path lengthof the two shapes is used. An average path length isfound for each observed CSB field. This average is foundover the entire field, with the path lengths reduced pro-portionally for any portions of the model that lie in theregion of the field with a reduced response. This pro- HaloSat Analysis of the Cygnus Superbubble Table 3. Derived ParametersHS ID Path length Emission measure Density Pressure Thermal Energy Luminosity(pc) (10 cm − ) (cm − ) (10 − dyne cm − ) (10 erg) (10 erg s − )HS0013 207 − . − . . . +0 . − . . +0 . − . − . − . . HS0355 207 − . − . . . +0 . − . . +0 . − . − . − . . HS0356 199 − . − . . . +0 . − . . +0 . − . − . − . . HS0357 189 − . − . . . +0 . − . . +0 . − . − . − . . HS0015 94 − . − . . . +0 . − . . +0 . − . − . − . . HS0358 95 − . − . . . +0 . − . . +0 . − . − . − . . HS0016 175 − . − . . . +0 . − . . +0 . − . − . − . . Full CSB ... ... 0 . +0 . − . . +0 . − . − ... Note —This table includes parameters derived from the best fit model parameters in XSPEC for each field, as wellas estimated global parameters for the entire CSB, assumed here as a cohesive object, using average values foundfor the HaloSat observations. The values are calculated assuming a distance of 1.4 kpc, while the listed error iscalculated by using different distance estimates. The upper error value is the difference between 1.15 kpc and 1.4kpc distances, while the bottom error bar is the difference between 1.4 kpc and 1.5 kpc. The difference between afilled spherical versus shell like model is roughly an additional 10% error. cedure is repeated for 3 separate possible distances of1.15 kpc, 1.4 kpc, and 1.5 kpc based upon the distanceestimates for Cyg OB1 and OB2 discussed in section 4.Table 3 includes the following parameters for eachfield: path length, emission measure, density, pressure,thermal energy, and luminosity. The error ranges ofthese values are based upon the aforementioned rangeof distances. The luminosity for each field is calculatedby finding the flux in XSPEC by using a dummy re-sponse and the apec model for the field, unabsorbed,over an energy range of 0.1-20 keV. This is repeated forthe three previously mentioned distances. The table val-ues would increase by up to ∼
10% if the emitting regionwere a filled sphere rather than a shell, depending onhow much of the shell is in the field of view.Despite the significant differences in what part of theCSB model is captured by each field, the derived param-eters are all rather similar, lending support to both theCSB being a hypernova remnant as well as support forthe shell remnant model. The similar values for densitycan be averaged and combined with the weighted aver-age temperature to calculate global values for the entireCSB. These values are also included in Table 3. Tak-ing the thermal energy for each field and dividing it bythe luminosity of the field returns an estimate of coolingtime (calculated using the 1.4 kpc distance estimate). Averaging all the fields, the CSB has an estimated cool-ing time of 18 million years. Note, however, that thisis functionally an upper bound to the cooling time thatdoes not take into account energy losses due to expan-sion and radiation at other wavelengths. This estimatecorresponds to an energy loss of ∼ × erg overthat time, based on an estimated total luminosity of theCSB of ∼ × erg s − . This is an initial energy of ∼ × erg, similar to the 2 − × erg found forhypernova SN1998bw by Iwamoto et al. (1998).If the CSB is a hypernova remnant, then the ques-tion remains of where the progenitor star came from.The relative positions of the CSB model and the CygOB associations can be seen in Figure 6. BD+43 3654,a 70 solar mass runaway star ejected from Cyg OB2,has proper motion of 5.7 mas yr − and an estimatedage of 1.6 Myr (Comer´on & Pasquali 2007). This cor-responds to a total distance traveled of ∼ . Bluem et al.
Figure 6. ROSAT 3/4 keV (R4+R5) map (Snowden et al. 1997) showing the CSB physical model in black and the Cyg OBassociations in red dashed lines. The coordinates and sizes of the OB associations are taken from Table 1 in Uyaniker et al.(2001), with Cyg OB4’s latitude corrected to the original value from Humphreys (1978). The diameter used for Cyg OB2 isfrom Kn¨odlseder (2000). The color bar is in units of 10 − ct s − arcmin − . proper motion measurements show that stars in the as-sociations are moving away from the center of the CSB,interpreting this as a sign that the expansion of the CSBtriggered the formation of the Cygnus association stars.This relationship is supported by this paper’s absorp-tion measurements, meaning that the relation betweenthe CSB and Cyg OB1 may not be that Cyg OB1 gaverise to the CSB progenitor star, rather that the CSBtriggered the formation of Cyg OB1. Cyg OB2 also ex-hibits proper motion away from the center of the CSB(Lim et al. 2019), so triggered star formation caused bythe CSB expansion may also be happening in Cyg OB2.Comer´on et al. (2016) finds that Cyg OB2 exhibits aolder secondary population of stars, with an age of 20Myr, based on observations of red supergiants in the re-gion. This would have interesting implications for CygOB2 as a potential source of a progenitor star, especially in combination with the results from Berlanas et al.(2019). However, the majority of the Gaia DR2 paral-laxes for the Comer´on et al. (2016) red supergiant sam-ple would be considered foreground/background con-taminants when added to the Berlanas et al. (2019)sample. Only RAFGL 2600 and IRAS 20341+4047 haveparallaxes that are consistent with Cyg OB2, with thosestars respectively falling in the 1.35 kpc and 1.76 kpcpopulations.The total thermal energy of the CSB at a distance of1.4 kpc is estimated to be 3 . × erg. This is sim-ilar to the total thermal energy found by Kimura et al.(2013) of 9 × erg, although they assumed a largerdistance for the CSB. The reason that the total ther-mal energy ends up being similar is due to a combi-nation of a higher calculated density (0.08 cm − ver-sus the Kimura et al. (2013) value of 0.02 cm − ) and HaloSat Analysis of the Cygnus Superbubble Table 4. Parallaxes for Candidate CygOB2 StarsStar Parallax(mas)IRAS 20315+4026 − . ± . . ± . . ± . . ± . . ± . . ± . . ± . Note —Gaia DR2 parallaxes(Gaia Collaboration et al. 2018) forthe 7 stars studied in Comer´on et al.(2016). Gaia systematic error is notincluded in these measurements. a significantly different geometry of emitting region forthe CSB resulting in a larger volume ( ∼ × cm or ∼ , ,
000 pc ). Kimura et al. (2013) uses a C-shaped toroidal model in the plane of the observation,with a volume of 4 × cm . The Kimura et al.(2013) model is motivated only to cover the observedemission, rather than attempting to describe the fullhighly-absorbed structure. The consistent parameters found for the CSB emis-sion, both fit and derived, imply that it stems from a sin-gular origin. If the interior volume of the CSB was blownout by varying degrees of stellar winds and/or extendedsupernova activity, then there is no particular expecta-tion that these different sides of the CSB would featuresimilar parameter values, given their differing points oforigin physically and chronologically. This makes thehypernova interpretation of the CSB origin more favor-able, as the edges of the CSB would thus be expectedto have similar history in terms of starting energy andcooling times. ACKNOWLEDGEMENTSThis research was supported by NASA grantNo. NNX15AU57G. This research has made useof MAXI data provided by RIKEN, JAXA and theMAXI team. This work has made use of data fromthe European Space Agency (ESA) mission Gaia
Gaia
Gaia
Multilateral Agreement.
Software:
DS9 (Joye & Mandel 2003), matplotlib(Hunter 2007), NumPy (Harris et al. 2020), XSPEC(v12.10.1f; Arnaud 1996)REFERENCES