An early detection of blue luminescence by neutral PAHs in the direction of the yellow hypergiant HR 5171A?
AAstronomy & Astrophysics manuscript no. y.26392˙am˙2 c (cid:13)
ESO 2018October 19, 2018
An early detection of blue luminescence by neutral PAHs in thedirection of the yellow hypergiant HR 5171A?
A.M. van Genderen , H. Nieuwenhuijzen , and A. Lobel Leiden Observatory, Leiden University, Postbus 9513, 2300RA Leiden, The Netherlands SRON Laboratory for Space Research, Sorbonnelaan 2, 3584 CA Utrecht, The Netherlands Royal Observatory of Belgium, Ringlaan 3, 1180 Brussels, BelgiumReceived... / Accepted....
ABSTRACT
Aims.
We re-examined photometry (
V BLUW , UBV , u v b y ) of the yellow hypergiant HR 5171A made a few decades ago. In that studyno proper explanation could be given for the enigmatic brightness excesses in the L band ( V BLUW system, λ e ff = L band.Our goals were to investigate other possible causes, and to derive the fluxes of the emission. Methods.
We used two-colour diagrams based on atmosphere models, spectral energy distributions, and di ff erent extinctions andextinction laws, depending on the location of the supposed BL source: either in Gum48d on the background or in the envelope ofHR 5171A. Results.
False L –excess sources, such as a hot companion, a nearby star, or some instrumental e ff ect, could be excluded. Also,emission features from a hot chromosphere are not plausible. The fluxes of the L excess, recorded in the data sets of 1971, 1973, and1977 varied (all in units of 10 − Wm − µ m − ) between 1.4 to 21, depending on the location of the source. A flux near the low side ofthis range is preferred. Small brightness excesses in u v ( u v b y system) were present in 1979, but its connection with BL is doubtful.For the L fluxes we consider the lowest values as more realistic. The uncertainties are 20–30 %. Similar to other yellow hypergiants,HR 5171A showed powerful brightness outbursts, particularly in the 1970s. A release of stored H-ionization energy by atmosphericinstabilities could create BL emitted by neutral PAHs. Key words.
Interstellar medium, nebulae: molecules – Interstellar medium, nebulae: HII regions – Stars: HR 5171A – Stars: hyper-giants – Technique: photometric
1. Introduction
The fluorescence of small neutral polycyclic aromatic hydro-carbon molecules (PAHs) in the Red Rectangle (RR) nebula in2003, and in reflection nebulae, was discovered by Vijh et al.(2004, 2005a; 2006, respectively). As this emission is containedin a broad band, somewhere between 3550 and 4400 Å, withmost of the peaks between 3700 Å, and 4050 Å, it was calledblue luminescence (BL). Vijh et al. (2005a) found a very closecorrelation between the BL distribution and the 3.3 µ m UIR-band emission. The latter is attributed to the C–H stretch bandradiated by neutral PAH molecules. The central source of theRR nebula is HD 44179, a binary, and presumably a post-AGBstar of spectral type A ( T e ff ∼ ff ect in the L band ( λ = V BLUW photometricsystem of the yellow hypergiant (YHG) HR 5171A = V766 Cen,reported by van Genderen (1992, Figs. 1 and 5, hereafter PaperI), made in 1971 and 1973 (and 1977, Sects. 3.1., 3.3–3.3.1.).Hereafter, these data sets are abbreviated to sets 71, 73, and 77,respectively. No proper explanation could be o ff ered at the time.Surprisingly, that excess appeared to be absent in the V BLUW time series 1980–1991. We also investigated
U BV photometryof HR 5171B made in 1971, and u v b y photometry of HR 5171A, made in 1979 and between 1991–1994. The rationale was thatthe blue and ultraviolet bands have some overlap with the BLspectrum considering its FWHM ∼
450 Å. Thus, if the excesswas really due to BL, the location could be either Gum48d in thebackground, or the dense CS matter of HR 5171A.HR 5171A is thought to be located in the old HII regionGum48d, known for its gas, dust, and molecular complexes.However, this would implicate that HR 5171A has a luminos-ity that is too high for a hypergiant, with the consequence thatits distance should be smaller (Nieuwenhuijzen et al. in prep.).Moreover, Karr et al. (2009) concluded that the present energybudget of Gum 48d only needed one ionizing O-type star, i.ethe present B-type star HR 5171B a few million years ago onthe main sequence, just like HR 5171A roughly at the sametime. Thus, this might be another argument that HR 5171A is nomember of Gum48d, unless the contribution of the secondary’sbrightness (unknown; Sect. 2.) would cancel this inconsistency.
2. Characteristics of HR 5171A and the Gum48dnebula
HR 5171A = V766 Cen = HD 119796, in 1971 of spectral typeG8Ia + (Humphreys et al. 1971), T ∼ Geneva seven-colour photometry exists as well, made in the 1970sand 1980s. The catalogue only lists two averages based on an unknownnumber of observations, and precise dates are lacking (Rufener 1988).1 a r X i v : . [ a s t r o - ph . S R ] S e p .M. van Genderen et al.: Transient blue luminescence in the direction of HR5171A? evolved massive star, showing various types of photometric in-stabilities. It is suspected to be a contact binary surroundedby a mid-IR nebula (Chesneau et al. 2014). It belongs to thevery small group of yellow hypergiants (YHG; de Jager 1980,1998; de Jager & van Genderen 1989). The distance according toHumphreys et al. (1971) is ∼ B − V ) J ( U BV system) between 1960 and 1980 (Paper I; Chesneau et al.2014).Gum48d, containing the photo-dissociation region RCW 80(Schuster et al. 2006; Schuster 2007), is located in the CentaurusArm. At present, its ionization is sustained by the near-by bluestar HR 5171B (Schuster 2007), a B0Ibp star (Humphreys etal. 1971), at a distance of ∼ (cid:48)(cid:48) . Most of the UIR bands ofGum48d contain emission features of PAHs (Schuster et al.2006; Schuster 2007; Karr et al. 2009). The 3.4 µ m C–H stretchband emission from the WISE archive, which is the best signa-ture for neutral PAHs (Bakes et al. 2004), turned out to be sat-urated at a flux of 95 Jy (Cutri et al. 2012, see also in Chesneauet al. 2014, the Appendix Table D.1). Thus, based on these ob-servations, the nebula is a serious candidate to be the source forthe BL. On the other hand, Chesneau et al. (2014), believe thatHR 5171A is responsible for the high far- and mid-IR excess(see their Table D.1.) as the instruments used have apertures notlarger than a few arcsec, centred on HR 5171A. However, withrespect to temperature, HR 5171A is too cool ( ∼ ffi cient to produce these kinds of photons(Sect. 4).HR 5171A shows a prominent 10 µ m silicate feature(Humphreys et al. 1971) and a strong Na I 2.2 µ m emission(Chesneau et al. 2014), also detected in other massive evolvedstars like the YHG (Oudmaijer & de Wit 2013). This could pointto the presence of a pseudo-photosphere in an optically thickwind, shielding the neutral sodium from ionization by directstarlight (Oudmaijer & de Wit 2013). These authors also claimthat the NaI emission in other YHG originates in a region insidethe dust condensation radius. Considering the controversy aboutthe precise origin of the L excess: the CS material of HR 5171Aand Gum48d, we compute the fluxes for these two alternatives.A possible instrumental e ff ect responsible for a false emissionin the L channel during the 1970s is also discussed (Sect. 3.1).We also discuss two emission features radiated by a hot chromo-sphere, often a feature exhibited by cool stars, which under ab-normal hot and active conditions also radiate in the wavelengtharea of the L channel (Sect. 3.2.).
3. The photometric observations
V BLUW photometry. Other possible causes toexplain the L –excess A total of seven observations, made between 1971 and1977, showing the strong brightness excess in the L band( λ e ff = Fig. 1.
The fluxes log f for the V BLUW data sets as a functionof the wavelength. The theoretical SED for a star with log g = = L flux as it includes the suspected BL) does not agree withany model. At the bottom, the filter coverage for the V BLUW bands, and the BL spectrum for only the highest peak. The L excesses are indicated with triangular peaks.which was then still located at the Leiden Southern Station inSouth Africa, Broederstroom (Hartebeestpoortdam). At the time,the telescope was equipped with the same simultaneous five-colour photometer of Walraven as between 1980–1991 (no L excess) after the telescope was moved to the ESO in 1979, LaSilla, Chile (Lub & Pel 1977; de Ruiter & Lub 1986; Pel & Lub2007). It is customary to express the brightness and colour in-dici of the V BLUW system in log intensity scale. The observingand calibration procedures were similar for all data sets (PaperI). However, the aperture was 21 (cid:48)(cid:48) / or 16 (cid:48)(cid:48) (cid:48)(cid:48) . m (cid:48)(cid:48) –5 (cid:48)(cid:48) .Two observations were obtained in 1971 (Pel 1976), two ob-servations in 1973 (Paper I) and three in 1977 (Pel 2013, priv.comm.). The time intervals between the individual observationsin a set were a few days up to a few weeks. All photometric dataare given in the 1980 system (Paper I). Those made in 1971 and1973, on both sides of maximum no. 3 (see Fig. 1 of Paper I) aremarked ’ L excess’. The 1977 observations are located in the ris-ing branch to maximum no. 7, very near the top (not plotted inFig. 1 of Paper I, as we obtained them from Pel only recently.The star was brighter and bluer than usual because of intrinsicchanges (Sect. 3.3.).The amount of L excess is based on the position in the the-oretical two-colour diagram V − B / B − L (empirical versionin Fig. 5 in Paper I) based on Kurucz-atmospheres (Castelli & Kurucz 2003; Castelli 2014). The excesses in B − L representthe colour index di ff erence between the observed and expectedposition (at roughly constant V − B ), if there was no excess in L .Table 1 lists the observed photometric parameters for thethree data sets of the 70s and the median values of the 80–91set. The parameters V J and ( B − V ) J were transformed from V and V − B by formulae of Pel 1987). The last column lists the L excesses, labelled L Ex.The L excesses in the early 1970s, which is very obvious inthe empirical two-colour diagram V − B / B − L , could not be dueto any stray light from the blue nearby star HR 5171B; whateverthe size of the diaphragm, as HR 5171A was always decentredin such a way that the blue star’s distance to the edge was en-larged. Moreover, HR 5171A would have shown progressivelyincreasing excesses in U and W , if this was done insu ffi ciently.The suggested presence of a mysterious blue star C in the di-aphragm, at a 0 (cid:48)(cid:48)
15 distance, nearly as bright as HR 5171A, isunlikely. Otherwise, the U and W fluxes should have shown evenlarger excesses. Moreover, the 71 / B − U and U − W indices(in 1977 the star brightened intrinsically, Sect. 3.3.) were verysimilar to those of 80 /
91 lacking any L excess!Another possible explanation of the false L excesses, withthe observed intensity, is if the ’field of view’ of the photomul-tiplier of the L channel was accidentally shifted by ∼ (cid:48)(cid:48) , (withrespect to the other channels) in such a manner that HR 5171Bwas shifted up to, or on the edge of the diaphragm. This scenariois considered almost impossible from a technical point of view(Pel 2014, priv.comm.). Moreover, if this had been the case, itwould mean that hundreds of program stars and tens of thou-sands of L measurements of comparison and standard stars dur-ing the seventies, were badly centred by about 5 (cid:48)(cid:48) . The muchless favourable seeing conditions at the location of the SouthernStation, compared to the seeing at the ESO, would have entaileda much larger scatter for B − L than for the three other colour in-dices. A quality check of, e.g. the 150 and 100 complete light andcolour curves of Population I Cepheids (Pel 1976) and RR Lyraestars (Lub 1977), respectively, proves that this was not the case. L band Two emission sources, which can give rise to radiation in thewavelength area of the L band, are related to a hot chromo-sphere. Cool stars like the twin sister of HR5171 A, ρ Cas, attimes showed the signature of a hot chromosphere. One pre-sumes that this is caused by the dissipation of mechanical en-ergy delivered by the di ff erent types of pulsations and occasionalshell ejections (Sargent 1961; Zsoldos & Percy 1991; Percy &Kolin 2000; Lobel et al. 2003). This object showed a variableBalmer continuum radiation starting at ∼ L band was ∼ . m L band. (Dommanget & Nys 1994; Dommanget 2002; Mason et al. 2001(WDS13472-6235); Mason & Hartkopf 2006; ESA 1997 (HIP 67261);van Leeuwen 2012, priv.comm.). However, its existence is still a mat-ter of debate: HST images do not show this component (Schuster &Humphreys 2014, priv.comm. On the other hand, the analysis of threeyears of Hipparcos photometry, revealed two 4 d . m
05 (Eyer 1998). Thus, a variable star should beinside the aperture used by the Hipparcos satellite.
Normally these resonance lines are in absorption, but in a veryheated chromosphere they appear in emission, as in flare stars.There are convincing reasons to reject a hot chromosphere as thesource of the L excesses in the 1970s (Sects. 3.3, 4.). V BLUW data sets
An important issue is the location of the supposed BL. There aretwo possibilities. Scenario (1) in Gum48d, or (2) in the extendedenvelope of HR 5171A. Depending on the location, di ff erent cor-rections for extinction (interstellar IS and / or circumstellar CS)should be applied to the observed L fluxes.Scenario (1) In this case the reddening of HR 5171B has beenused: E( B − V ) J = ± ff average of the red-denings based on the photometry by Humphreys et al. (1971),Dean (1980) and Paper I. Using the extinction law for the di ff useIS extinction R J = V j = ± V BLUW system are:E( V − B ) = ± W = V = ± L band: A L = ± L excesses and for the stellar SED. The total redden-ing for the four data sets was determined with the reddening linesin the two-colour diagrams defined by Pedicelli et al. (2008),and the position of the 80-91 set as reference. This is because itwas free of any L excess and if dereddened, its position in the V − B / B − L and B − U / U − W diagrams moves to T ∼ =
0. This is in agreement with spectral analyses(Humphreys et al. 1971; Warren 1973; Chesneau et al. 2014;Lobel 2014, priv.comm.; Oudmaijer 2014, priv.comm.).The reddening (IS and CS) in log scale was E( V − B ) = ± B − V ) J = ± J = U BV , Str¨omgren u v b y , and Walraven V BLUW , we used observations (almost)simultaneously made in the three systems. We found R J = S = W = Hence, the extinctions are A
V j = ± L = ± resulted from the addition of the calibrationconstants (de Ruiter & Lub 1986) and the extinction-free bright-nesses. They are listed in Table 2.Fig. 1 shows the SEDs of HR 5171A for scenario (2) only;the green and blue symbols denote the 71, 73, and 80–91 sets,respectively, and the red symbols denote the 77 set. It is obviousthat the brightness di ff erences between the sets 71, 73 and 80–91were small, therefore, they could be used as one homogeneousset. It appears that a satisfactory match, even with the small pos-itive Balmer jump, is obtained for V , B , ( L only for the 80-91set), and U with the model SED for 5000 K, adopting log g = turb = − , using Kurucz models (Castelli & Kurucz2003; Castelli 2014). The W flux is too high by about 0.2 log This matching eliminated small reddening di ff erences, which areevident from the V − B values between the 70s and 80s; see Table 1.We emphasize that these R values are only applicable to this star. Thereason is that the cause of the peculiar and long-lasting stellar reddeningis not well understood. 3.M. van Genderen et al.: Transient blue luminescence in the direction of HR5171A? Table 1.
The observed photometric parameters of the four data sets and the brightness excesses in the L -band. The L (the totalbrightness in the L -band) and L Ex. are means of the data set. Only V J and ( B − V ) J are in magnitudes, the remainder is given inlogarithmic scale JD–244 0000 set V J ( B − V ) J V V − B B − L B − U U − W L L
Ex.1164.5 71 6.80 2.30 -0.004 1.124 0.528 1.010 0.50 -1.660 0.251169.5 6.75 2.31 0.017 1.126 0.555 1.001 0.601758.6 73 6.88 2.34 -0.034 1.140 0.395 1.042 0.60 -1.572 0.371787.5 6.90 2.30 -0.042 1.122 0.410 0.987 0.543248.5 77 6.41 2.31 0.154 1.125 0.394 0.813 0.39 -1.343 0.183252.5 6.35 2.29 0.178 1.121 0.395 0.781 0.483269.5 6.32 2.27 0.191 1.109 0.407 0.796 0.484312–8315 80–91 6.60 2.55 0.060 1.23 0.83 0.98 0.54 -2.00 0
Table 2.
The extinction-free brightnesses log I and fluxes log f for the five channels of the four data sets according to scenario(2) Set band log I log f
71 V 2.086 -8.086B 1.592 -8.318L 1.300 -8.518U 0.986 -8.807W 0.716 -8.95773 V 2.042 -8.130B 1.541 -8.360L 1.388 -8.430U 0.926 -8.867W 0.636 -9.03777 V 2.254 -7.918B 1.766 -8.144L 1.617 -8.201U 1.369 -8.424W 1.199 -8.47480–91 V 2.140 -8.032B 1.593 -8.317L 0.960 -8.858U 0.960 -8.833W 0.710 -8.963 scale ( ∼ . m
5) for all three sets, thus, possibly caused by chro-mospheric radiation. Obviously, these W fluxes represented veryweak Balmer continuum excesses, otherwise U should show ex-cesses as well, which should then be greater than those in L .Therefore, these L excesses should be due to some other source.The L excesses are indicated with the two triangles.The SED of the 77 set is a di ff erent story: no match with anyhotter theoretical SED is possible. The slope of the Paschen con-tinuum (from V to B ), is slightly steeper than for the other sets.Thus, a possible photospheric temperature rise is small (by a fewhundred degrees), but the extreme high excess in the ultravioletis stunning, with on top the L excess. The pulsational activityand / or outbursts of ionization energy, represented by an excessof optical light (see Fig. 1 in Paper I), heated the chromospherepresumably to such a high level that it resulted in an excessiveBalmer continuum emission of up to more than 1 m in U and W (Sect. 4.). It should be emphasized that the observed L bright-nesses (71, 73, and 77) contain a stellar and a non-stellar con-tribution. The stellar contribution was likely subject to a muchhigher reddening than the non-stellar one (Sect. 3.3.1.). L excess fluxes In Table 3, we list the date (set); the brightness of HR 5171Ain the L band (taken from Table 1), including the excess L Ex;the corrected brightness of HR 5171A in L without the excess( L - L Ex); the brightness of the excess (E L ), which is the dif-ference of the anti-logs of L and L - L Ex., after transforming itinto a log scale; and the unreddened fluxes, f(E L ) , from sce-nario (1), in 10 − Wm − µ m − . The uncertainty is ± . m
10 uncertaintyin the reddening. As the reddening di ff erences between the threedata sets were small (Sect. 3.3.), the di ff erence in the fluxes of1971 and 1973 / (cid:48)(cid:48)
5, and / or 16 (cid:48)(cid:48) (cid:48)(cid:48) ff ect of the Gum48d’s radiationtravelling through HR 5171A’s envelope, with its much higherreddening (see scenario [2]), can be neglected because its diam-eter is about 3 (cid:48)(cid:48) L ), are listed in thesixth column. It appears that the fluxes are now larger than forscenario (1) by a factor of ten (due to the higher reddening)as well as the uncertainty. Thus, 14.7 in 1971 and 21.4 ± / L flux, is appropriate. However, it islikely that the supposed BL source did not su ff er from the samehigh extinction as the central star HR 5171A because its loca-tion could have been near the border of the envelope (e.g. fac-ing the Earth), where the extinction is much lower. Then, thelower reddening used in scenario (1) would be much more ap-propriate, resulting in fluxes about ten times smaller. The enve-lope of HR 5171A may be more or less similar to the envelopesurrounding the AGB star IRC + v ∼ . m ∼ m ), decreasing exponentially outward and beingalmost negligible halfway to the edge. Thus, by virtue of thismodel, it is much more realistic to suppose that the fluxes forthe L excesses should have been near the low side of the totalrange 1.4 to 21. A dust model of HR 5171A by Lobel (2015,priv. comm.) will be released soon.Thus, then the remarkable size of the L excesses, as evidentin Table 3, and in the empirical V − B / B − U diagram in Fig. 5in Paper I, being almost of the order of the stellar flux, can beunderstood (see our comment on the size of the 1977 flux below, Table 3.
The computation of the fluxes f(E L ) (in 10 − Wm − µ m − ) of the BL, based on the brightness excesses in the L band aftercorrection for interstellar extinction and assuming that: (1) the source lies in Gum48d ; column f(E L )(1), and (2) that the source liesin the CS envelope of HR 5171A; column f(E L )(2). See Sect. 3.3.1. set L L - L Ex. E L f(E L )(1) f(E L )(2)71 -1.660 -1.910 -2.019 1.37 14.773 -1.572 -1.942 -1.813 2.20 21.377 -1.343 -1.523 -1.812 2.21 21.4 and in Sect. 4). Moreover, the star was, and still is, intrinsicallyvery red.It is evident that the flux measured in the 77 set containsthe excess of the unknown source and those radiated by the hotchromosphere. Thus, the value listed in Table 3 represents thetotal excess flux of di ff erent sources. U J band (Johnson U BV system), and in the u v bands ( u v b y system) Any u v observation of HR 5171A, made after the 1971–1977time interval, and one U J observation of its close neighbourHR 5171B, made in 1971 (by Humphreys 2014, priv.comm.),has been carefully inspected and analysed. Despite their partialoverlap with the BL spectrum no excess in U J was found, and in u v only a small amount of excesses, which we do not considerconclusive. Therefore, they are only briefly discussed.As the U J magnitude of HR 5171B in 1971 was not a ff ectedby any measurable emission, we conclude that either the contri-bution to the L brightness of set 71 originated accidentally be-hind HR 5171A in a small area of Gum48d, or its origin was theCS material of HR 5171A.Olsen (1983) obtained seven observations in the Str¨omgren u v b y system in the period 1979, February 26 and October 6,with the Danish 50-cm reflector at the ESO in La Silla, Chile,equipped with the simultaneous u v b y photometer (Grønberg etal. 1976; Grønberg & Olsen 1976). The aperture was 30 (cid:48)(cid:48) andincluded HR 5171B, for which we had to correct before inspect-ing the v magnitude of HR 5171A for any brightness excess. Themean errors in the average photometric parameters appear to bevery small (Olsen 1983).As no u v b y photometry has ever been made of HR 5171B, wehad to use models (solar abundancy) for early B-type stars: thoseof Lester et al. (1986) and of Castelli & Kurucz (2003) / Castelli(2014). We adopted T e ff =
26 500 K, log g = y magnitude almost equal to the observed V J = V J ( ∼ y ) above was dereddened using E( B − V ) J = ± J = u v b y channels, we consulted the tables of Steenman & Th´e (1989).Table 4 lists Olsen’s u v b y observations of both stars together(first line), the mean of the two models for the blue star (sec-ond line), and then in the third line the ”observed” models ofHR 5171A only. Thus, the magnitudes on the second line sub-tracted from the magnitudes on the first line. The estimated un-certainties in the models are smaller than 0 . m
1, with the exceptionof the u magnitude, which is ± . m
2. The fourth line in Table 4lists the mean of three u v b y observations made in 1991 (see be-low).Str¨omgren u v b y observations were also made by the LTPVGroup of Sterken (1983) from 1991 until 1994 with the 50-cmESO telescope in La Silla, Chile (Manfroid et al. 1991; Sterkenet al. 1993). The very first three u v b y observations by the LTPVGroup, coincided with the last six observations of the 80–91 V BLUW data set in maximum no. 17 (Fig. 1 in Paper I; van
Fig. 2.
Log f , for observations made almost simultaneously intwo di ff erent photometric systems (green and red symbols) as afunction of λ with the indication of the λ coverage. U jB jV j forJohnson, u v b y for Str¨omgren, V BLUW for Walraven. The bluecurves are SEDs for 5000 K and 5250 K models. Vertical arrows:see the text in Sect. 3.5.Leeuwen et al. 1998). We used these three 1991 u v b y obser-vations to compare them with the 1979 observations of Olsen(listed in Table 4, third line). Although the aperture used by theLTPV group was 20 (cid:48)(cid:48) , star HR 5171B could generally be ex-cluded by decentring HR 5171A with ∼ (cid:48)(cid:48) . We constructed an SED (Fig. 2) simultaneously using obtained B J V J ( U BV system) photometry by Dean (1980); see Fig. 1 inPaper I, together with the simultaneously observed u v b y and V BLUW data from the 1991 sets. The band widths are indicatedat the bottom with horizontal line pieces. The central mark in-dicates λ e ff . For v and u , two widths are indicated: the smallestwas used by the LTPV group (Sterken 1983), and the largest wasused by Olsen (1983). The B J and V J bands are only representedby their λ e ff . The band width (FWHM) of the BL for the RRnebula is indicated (Vijh et al. 2004, 2005b, 2006). The full lineis the portion containing the highest emission peaks. The samereddening was used as in Sect. 3.3. (scenario 2), as well as the Table 4.
The comparison of the observed 1979 u v b y magnitudes of HR 5171A (corrected for the light of HR 5171B) on the thirdline with the 1991 observed u v b y magnitudes on the fourth line. The mean reddened model of HR 5171B is on the second line. set / type / object y b v u set 79, observed, HR 5171A + HR 5171B 6.78 8.45 10.16 11.45mean reddened model HR 5171B 10.03 10.67 11.20 11.86set 79, ”observed”, HR 5171A (corr. for star HR 5171B) 6.84 8.60 10.68 12.71set 91, observed, HR 5171A 6.74 8.51 10.89 13.24
Table 5.
The reddening free fluxes of HR 5171A (Wm − µ m − )measured in three di ff erent photometric systems. Each pair ofsets was simultaneously observed. The last column lists themean visual magnitude. In the last data set from 1991, the V J magnitude represent the computed equivalent value of V . set band log f V J set band log f V J y -8.093 6.84 91 y -8.053 6.74 b -8.200 8.60 b -8.166 8.51 v -8.526 10.68 v -8.609 10.89 u -8.813 12.71 u -9.025 13.2479 V J -8.126 6.83 91 V -8.058 0.044 B J -8.386 9.47 V J B -8.393 -1.186 L -8.867 -2.021 U -8.945 -2.232 W -9.092 -2.806 extinction laws for the three photometric systems, and the ex-tinctions for each pass band. Calibration constants are from Cox(1999), for U BV ; Helt et al. (1991); for ub vy , and de Ruiter &Lub (1986), for V BLUW .The computed fluxes log f are listed in Table 5. The obtainedSEDs of the simultaneous sets 79 u v b y and B J V J , match very sat-isfactorily. They also match well the 91 sets u v b y and V BLUW .Fig. 2 shows brightness excess by about 0.1-0.2 (see arrow) inthe u band in log f with respect to the combined SED. However,considering the errors, we are very hesitant to consider it as dueto BL. Errors in the models and physical di ff erences of the starbetween 1979 and 1991 are possible, but some contribution byBalmer continuum radiation is possible as well. The excess in W : ∼ . m
3, represents a weak Balmer continuum emission, butweaker than during the 1970s and 1980s (Fig. 1: ∼ . m
5, Sect. 3.3.
4. Discussion and conclusions
It is quite well possible that the brightness excesses in the L band( V BLUW system) detected in the photometry of the yellow hy-pergiant HR 5171A made in the 1970s were due to blue lumines-cence (BL) that was discovered in the RR nebula by Vijh et al.(2004; 2005a,b; 2006). Our suspicion is possible in view of thecomplete overlap of its spectrum, including the highest peaks,with the L band. We first considered various possibilities, whichmight have created false brightness excesses in the L band, butthey turned out to be implausible. The same can be said about thefollowing emission sources previously mentioned in Sect. 3.2.:the Balmer continuum emission due a hot chromosphere and theCaII H and K resonance lines (in the L band), which only appearin emission in extreme circumstances, when the chromosphereis extremely heated. However, in the first case one expects muchhigher brightness excesses in W and U than in L (see Fig. 1 inJoshi & Rautela 1978, showing an increasing Balmer continuumemission from λ ρ Cas). It is obvious in our Figs. 1 and 2 that apart from the 77 set observed during a large amplitude pulsation showing highBalmer continuum excesses in W and U , the 71 and 73 sets showno excesses in U and the excesses in W are smaller than in L .Note that in the 1980s there was no L excess at all other than a W excess, presumably due to the presence of a chromosphere. Weconclude that the L excesses in 1971 and 1973 were not causedby Balmer continuum emission. Obviously, the L excess of the77 set in Fig. 1 contains the contribution from the supposed BLsource, but also should include Balmer continuum emission, andperhaps also some emission from the CaII doublet. Therefore,the L flux listed in Table 3 is too high for the supposed BL alone,indicating that this emission was declining. After all, in 1979, noexcess in that wavelength area of the u v b y system was recorded,apart from some excess in the u band, presumably from uncer-tainties in the atmosphere models and / or from some Balmer con-tinuum radiation (Fig. 2).The unreddened L fluxes, depending on the location of theBL source, in units of 10 − Wm − µ m − , varied between 1.4 and2.2 (71 and 73 sets), when located in Gum48d (scenario 1). If theunreddened L fluxes were located in the dense centre of the CSenvelope of HR 5171A (scenario 2) then the values would be afactor of 10 higher, i.e. 15–21. However, these values were muchlower, approaching the lower values that we prefer, if they werelocated near the outskirts of the envelope.No significant trace of an excess due to BL was detected inother bands apart from the L band. The distribution of the highestBL peaks in the direction of HR 5171A occurred in the L band,which does not contradict the possibility that BL was the cause.As mentioned above, no L excess was present in the 1980–1991 V BLUW time series, while the colour indices B − U and U − W were almost the same as in the early seventies(1971 / B − L colour index.The short duration of the supposed BL emission could meanthat we were dealing with a small-scale source. The dynamicalcircumstances of the CS matter surrounding HR 5171A wouldthen be a good candidate, as shielded clouds containing neu-tral PAH molecules can quickly be exposed to highly energeticphotons. On the other hand, one may wonder whether this coolstar ( ∼ ∼ . m V , with a timescale of ∼ ρ Cas (Lobel et al. 2003) and HR 8752 (Nieuwenhuijzen et al.2012). These authors concluded that they showed at times strongatmospheric ionization driven instabilities. As a result, explosive’flash’ outbursts occurred by the release of stored ionization en-ergy due to H recombination as the temperature of the atmo- sphere declined. These processes might have delivered the nec-essary high energetic photons (3.5–5 eV, Witt 2013, priv.comm.).
Acknowledgements.
We are very grateful to a number of colleagues: Prof.Xander Tielens for his support, who drew our attention to the papers of Vijh et al.(2004, 2005a); Prof. Jan Willem Pel for o ff ering his 1977 VBLUW photometry;and the following colleagues for many fruitful discussions, help, and invaluableadvices: Dr. Floor van Leeuwen, Prof. Jan Willem Pel, Prof. Adolf Witt, Dr.Jan Lub, Prof. Roberta Humphreys, and Dr. Ren´e Oudmaijer, and especially toDr. Erik-Heyn Olsen, who spent a lot of time studying his log books, making iteasy for us to establish the date of the seven u v b y observations. We took muchadvantage of his and Dr. Chris Sterken’s expertise on the Str¨omgren photomet-ric system. A.L. acknowledges partial financial support by the Belgian FederalScience Policy O ffi ce under contract No. BR / / A2 / BRASS and in connectionwith the ESA PRODEX programmes ’Gaia-DPAC QSOs’ and ’Binaries, extremestars and solar system objects’ (contract C90290). We are grateful to the refereefor the invaluable comments and suggestions.
References