next previous
Up: Dust features in the

Subsections

2 The observational investigation


 

 
Table 1: AGB stars observed
SourceIRAS namesOther namesVariabilitySpectralObservationCalibrator8.0 $\mu $m flux
   Type*TypeDate (10-12 Wm-2 $\mu $m-1)
Y Cas00007+5524IRC+60001, HD 225082MM7eOct. 5 '92$\epsilon$Cyg5.46
T Cet00192-2020IRC-20007, AFGL 53, HD 1760SRbM5/M6I/IINov. 1 '93$\beta$Peg10.0
T Cas00205+5530IRC+60009, AFGL 57, HD 1845MM7eOct. 5 '92$\epsilon$Cyg21.3
RW And00445+3224IRC+30015, AFGL 109, HD 4489, CIT 2MS6,2eOct. 5 '90$\alpha$Tau1.97
SAO 3767301556+4511IRC+50049, AFGL 278, HD 11979SR?M7Oct. 4 '92$\alpha$Cet12.8
o Cet02168-0312Mira, IRC+00030, AFGL 318, HD 14386MM7IIIeOct. 6 '92$\alpha$Lyr185
R Cet02234-0024IRC+00032, AFGL 4195, HD 15105MM4eOct. 4 '92$\alpha$CMa1.01
RR Per02251+5102IRC+50062, AFGL 335, HD 15186MM6e-M7eOct. 7 '92$\alpha$Cet3.20
RZ Ari02530+1807IRC+20051SRbM6IIIOct. 5 '90$\alpha$Tau7.94
RT Eri03318-1619IRC-20043, AFGL 500, HD 22228MM7eOct. 4 '92$\alpha$Tau5.42
IK Tau03507+1115IRC+10050, AFGL 529, NML TauMM6eOct. 7 '92$\alpha$Cet138
V Eri04020-1551IRC-20049, AFGL 542, HD 25725SRc?M5/M6IVOct. 4 '92$\alpha$Tau7.48
W Eri04094-2515IRC-30033, AFGL 552, HD 26601MM7eOct. 4 '92$\alpha$Cma2.13
BX Eri04382-1417IRC-10075, AFGL 615SRM3Oct. 4 '92$\alpha$CMa3.23
R Cae04387-3819AFGL 617, HD 29844MM6eOct. 4 '92$\alpha$CMa3.72
TX Cam04566+5606IRC+60150, AFGL 664MM9Oct. 4 '92$\alpha$Cma14.3
UX Aur05121+4929IRC+50138, HD 33877SRM4(II)Oct. 4 '92$\alpha$Cet0.988
R Aur05132+5331IRC+50141, AFGL 715MM7Oct. 4 '92$\alpha$Cet13.2
X Ori05351-0147IRC+00080, AFGL 786MM8Oct. 7 '92$\alpha$Aur3.26
RU Aur05367+3736IRC+40135, AFGL 794MM8eOct. 7 '92$\alpha$Aur3.99
SZ Aur05384+3854IRC+40136, AFGL 802, HD 37645MM8eOct. 7 '92$\alpha$Aur2.51
U Aur05388+3200IRC+30126, AFGL 822, HD 37724MM7eOct. 7 '92$\alpha$Aur6.33
RT Lep05404-2342IRC-20077, AFGL 8105MM9eOct. 4 '92$\alpha$CMa2.05
S Col05450-3142IRC-30049MM6eOct. 4 '92$\alpha$CMa1.30
CH Pup06434-3628AFGL 1008MMeOct. 7 '92$\alpha$Cet2.28
AZ Mon06551+0322IRC+00140,MM6eOct. 7 '92$\alpha$Cet1.26
GX Mon06500+0829AFGL 1028MM9Oct. 7 '92$\alpha$Cet9.32
Y Lyn07245+4605IRC+50180, AFGL 1120, HD 58521SRM6SMay 24 '91$\alpha$Boo5.40
Z Pup07304-2032IRC-20123, AFGL 1140, HD 60218MM4-M9eOct. 6 '92$\alpha$Cma4.97
DU Pup07329-2352IRC-20134, AFGL 1151MMOct. 6 '92$\alpha$CMa5.21
U Pup07585-1242IRC-10184, AFGL 1215, HD 65940MM5e-M8eOct. 6 '92$\alpha$Cma1.88
RS Cnc09076+3110IRC+30209, AFGL 1329, HD 78712SRM6SMay 25 '91$\alpha$Boo25.4
R Leo09448+1139IRC+10215, AFGL 1380, HD 84748MM8eMay 22 '91$\alpha$Boo67.4
R Hya13269-2301IRC-20254, AFGL 1627, HD 117287MM7eMay 25 '91$\alpha$Boo57.8
W Hya13462-2807IRC-30207, AFGL 1650, HD 120285SRaM8eMay 21 '91$\alpha$Boo152
RX Boo14219+2555IRC+30257, AFGL 1706, HD 126327SRbM7.5eMay 24 '91$\alpha$Boo25.8
S CrB15193+3139AFGL 4990S, HD 136753MM7eMay 24 '91$\alpha$Boo7.08
RU Her16081+2511IRC+30283, AFGL 1832, HD 145459, CIT 8MM7eMay 24 '91$\alpha$Boo8.74
U Her16235+1900IRC+20298, AFGL 1858, HD 148206MM7eMay 24 '91$\alpha$Boo12.6
BG Her17072+1844IRC+20314,MM3May 24 '91$\beta$Peg0.948
V1692 Sgr18320-1918 MM9Oct. 7 '92$\beta$Peg0.813
V1111 Oph18349+1023IRC+10365, AFGL 2206MM9Oct. 4 '92$\alpha$Lyr11.0
X Oph18359+0847IRC+10360, AFGL 2213, HD 172171MM5e-M9eOct. 4 '92$\alpha$Lyr13.2
V2059 Sgr18501-2132 MM8Oct. 4 '92$\alpha$Lyr2.92
R Aql19039+0809IRC+10406, AFGL 2324, HD 177940MM7eOct. 6 '92$\alpha$Lyr16.0
AG Sgr19044-2856IRC-30403, HD177868MM5-M6eOct. 7 '92$\beta$Peg0.619
V342 Sgr19093-3256IRC-30404, AFGL 5556MM9Oct. 5 '92$\beta$Peg4.29
W Aql19126-0708IRC-10497, AFGL 2349MS6,6eOct. 7 '92$\beta$Peg65.2
Z Sgr19167-2101IRC-20555, HD 181060MM5eOct. 5 '92$\beta$Peg2.02
V635 Aql19343+0912 M?Oct. 7 '92$\beta$And0.252
BG Cyg19369+2823IRC+30379, AFGL 2426MM7eOct. 4 '92$\alpha$Lyr2.45
V462 Cyg19384+4346IRC+40355, AFGL 2429MM7eOct. 4 '92$\alpha$Lyr1.15
RR Sgr19528-2919IRC-30419, AFGL 5569, HD 188378MM5eOct. 7 '92$\beta$Peg4.76
Z Cyg20000+4954IRC-50314, HD 190163MM5eOct. 5 '92$\beta$Peg1.88
SX Cyg20135+3055IRC+30423, HD 192788MM7eNov. 2 '93$\alpha$Lyr1.30
RU Cap20296-2151IRC-20590MM9eOct. 7 '92$\beta$And0.533
Y Del20392+1141IRC+10475MM8eOct. 7 '92$\beta$And1.87
Y Aqr20417-0500IRC-10546MM6.5eOct. 7 '92$\beta$Peg1.36
W Aqr20438-0415IRC+00489MM7Oct. 7 '92$\beta$And1.95
RZ Cyg20502+4709IRC+50347SRaM7Oct. 5 '92$\beta$Peg3.71
UW Cep20581+5841IRC+60301MM8Oct. 5 '92$\beta$Peg1.34
UU Peg21286+1055IRC-10498, AFGL 2775MM7eOct. 7 '92$\epsilon$Cyg6.76
RU Cyg21389+5405IRC+50390, AFGL 2790, HD 206483SRaM8eOct. 5 '92$\beta$Peg9.46
EP Aqr21439-0226IRC+00509, AFGL 2806, HD 207076, SAO 145652SRbM8Oct. 6 '92$\alpha$Lyr20.3
YY Cep22000+5643IRC+60337MM6Oct. 5 '92$\beta$Peg1.04
SV Peg22035+3506IRC+40501, AFGL 2845, HD 209872SRbM7Oct. 7 '92$\epsilon$Cyg9.47
CU Cep22097+5647IRC+60345, AFGL 2865SRbM5Oct. 5 '92$\beta$Peg4.99
R Peg23041+1016IRC+10527, AFGL 3023, HD 218292MM7eOct. 5 '92$\beta$Peg10.8
V Cas23095+5925IRC+60389, HD 218997MM5.5eOct. 5 '92$\beta$Peg3.67
BU And23212+3927IRC+40536, AFGL 3088MM7eOct. 6 '92$\beta$Peg4.41
R Aqr23412-1533IRC-20642, AFGL 3136, HD 222800MM7eOct. 6 '92$\alpha$Lyr32.2
Z Cas23420+5618IRC+60418, AFGL 3141, HD 222914MM7eOct. 7 '92$\epsilon$Cyg3.56
R Cas23558+5196IRC+50484, AFGL 3188, HD 224490MM7eMay 21 '91$\alpha$Boo41.5

*Variability Type: M = Mira; SR = semi-regular variable.
T Cet and V Eri are italicized because it is unclear whether each star is a giant or a supergiant - see Sect. 2.2.



 

 
Table 2: Supergiant stars in the sample
SourceIRAS namesOther namesVariabilitySpectralObservationCalibrator8 $\mu $m flux in
   Type*TypeDate 10-12 Wm-2 $\mu $m-1
MZ Cas00186+5940IRC+60008,LcM1.3IabOct. 5 '92$\epsilon$ Cyg1.25
BD+4748501400+4815IRC+50043, HD 10465Lc?M2IbAug. 16 '95$\beta$ And0.660
BD+5834201550+5901IRC+60070, HD 236915??M2.4IbAug. 16 '95$\beta$ And0.433
XX Per01597+5459IRC+50052, HD 12041SRcM3.6IbOct. 6 '92$\beta$ Peg2.58
KK Per02068+5619IRC+60074, HD 13136LcM1.9IbOct. 7 '92$\alpha$ Cet0.889
V550 Per02116+5754HD 13658??M5.4IabOct. 7 '92$\alpha$ Cet0.274
BU Per02153+5711IRC+60078SRcM3.7IbOct. 4 '92$\alpha$ Tau0.705
T Per02157+5843IRC+60079, HD 14142SRcM2.1IabOct. 4 '92$\alpha$ Tau0.400
V506 Cas02167+5926IRC+60081, HD 14242??M5.7IabOct. 7 '92$\alpha$ Cet0.591
AD Per02169+5645IRC+60082, HD 14270SRcM2.4IabOct. 4 '92$\alpha$ Tau0.771
FZ Per02174+5655IRC+60083, HD 14330LcM0.3IabOct. 7 '92$\alpha$ Cet0.521
PR Per02181+5738IRC+60085, HD 14404LcM0.7IabOct. 7 '92$\alpha$ Cet0.552
SU Per02185+5622IRC+60086, HD 14469SRcM3.3IbOct. 4 '92$\alpha$ Tau1.12
RS Per02188+5652IRC+60087, HD 14488SRcM4.4IbOct. 6 '92$\beta$ Peg1.33
S Per02192+5821IRC+60088, HD 14528, AFGL 323SRcM4.5IabOct. 4 '92$\alpha$ Tau6.81
V439 Per02196+5658BD+56595??M5.8IabOct. 7 '92$\alpha$ Cet0.368
V441 Per02217+5712IRC+60090, HD 14826??M3.1IabOct. 7 '92$\alpha$ Cet0.929
YZ Per02347+5649IRC+60093, HD 236979??M1.9IabOct. 7 '92$\alpha$ Cet0.976
GP Cas02360+5922IRC+60094, AFGL 359LcM2.8IabAug. 16 '95$\beta$ And0.983
V648 Cas02473+5738HD 237010, BD+57647??M2.9IabAug. 16 '95$\beta$ And1.01
IO Per03030+5532IRC+60110LcM3IOct. 4 '92$\alpha$ Tau4.88
AH Sco17080-3215IRC-30282, HD 155161SRcM5IAug. 16 '95$\eta$ Sgr14.0
IRC-3031217374-3156 ??M2.6IaAug. 16 '95$\eta$ Sgr1.68
KW Sgr17488-2800IRC-30326, HD 3167486, AFGL 2017SRcM2.4IaAug. 16 '95$\eta$ Sgr3.50
V540 Sgr17566-3555HD 163869LcM5IabAug. 16 '95$\eta$ Sgr1.10
VX Sgr18050-2213IRC-20431, AFGL 2071, HD 165674SRcM5-M6IOct. 4 '92$\beta$Peg73.3
IRC-1041918227-1347 ??M2.5IabAug. 17 '95$\gamma$ Aql0.450
UY Sct18248-1229IRC-10422, AFGL 2162SRcM3.4IabJun. 27 '98$\gamma$ Aql3.72
HD 17109418305-1408IRC-10435, AFGL 2186??M3IAug. 17 '95$\gamma$ Aql0.875
UW Aql18550+0023IRC+00398LcM2.2IabJun. 27 '98$\gamma$ Aql0.945
V1302 Aql19244+1115IRC+10420, AFGL 2390??F8IOct. 6 '92$\alpha$ Lyr18.2
IRC-2056519272-1929 ??M2IbAug. 17 '95$\gamma$ Aql0.285
NR Vul19480+2447IRC+20438, HD 339034LcM1.1IabAug. 16 '95$\gamma$ Aql2.46
BC Cyg20197+3722IRC+40409, AFGL 2560LcM3.2IabAug. 16 '95$\gamma$ Aql7.47
KY Cyg20241+3811IRC+40415, AFGL 2575LcM3.9IabAug. 16 '95$\gamma$ Aql6.95
AZ Cyg(20h56m)IRC+50351, AFGl 2683??M3.1IabAug. 16 '95$\gamma$ Aql2.26
AZ Cep22069+5918IRC+60343, AFGL 2857LbM1.6IAug. 16 '95$\beta$ Peg0.480
ST Cep22282+5644IRC+60357, AFGL 2916, HD 239978LcM2.6IaAug. 17 '95$\beta$ Peg1.01
U Lac22456+5453IRC +50446, HD 215924, AFGL 2957SRcM2.5IaAug. 17 '95$\beta$ Peg2.95
V355 Cep22471+5902 ??M1.1IabAug. 16 '95$\beta$ Peg0.442
V358 Cas23281+5742IRC+60410, AFGL 3110LcM2.8IaAug. 16 '95$\beta$ And1.42

*Variability Type: L = irregular variable; SR = semi-regular variable.



 

 
Table 3: 10 $\mu $m classifications of AGB stars
featureless AGB broad AGB broad+sil AGB silicate AGB A silicate AGB B silicate AGB C silicate AGB D
BU And AG Sgr BX Eri CU Cep DU Pup IK Tau CH Pup
R Hya BG Cyg RR Per RS Cnc GX Mon R Aqr R Cae
R Peg R Aql RW And RU Cyg R Cet RT Lep R Cas
RZ Ari R Aur RX Boo U Her RU Cap RU Aur W Eri
T Cas R Leo RZ Cyg UU Peg S CrB RU Her Y Aqr
UX Aur RR Sgr SAO 37673 X Ori U Aur EP Aqr  
V Cas RT Eri UW Cep Mira V1111 Oph TX Cam  
  S Col V462 Cyg   V342 Sgr U Pup  
  SZ Aur W Aqr   V635 Aql V2059 Sgr  
  V1692 Sgr Y Cas   Z Cyg Y Del  
  W Aql Z Cas   Z Pup Y Lyn  
  W Hya SV Peg     Z Sgr  
  YY Cep AZ Mon        
  V Eri          
  X Oph          



 

 
Table 4: 10 $\mu $m classifications of supergiant stars
broad Super A broad Super B silicate Super 1 silicate Super 2 silicate Super 3 silicate Super 4
AD Per BD+58342 AH Sco AZ Cyg AZ Cep V1302 Aql
FZ Per V439 Per KW Sgr BC Cyg V648 Cas IRC-30312
V506 Per V550 Per   BU Per GP Cas S Per
PR Per     HD 171094 IRC-10419 UY Sct
V441 Per     IRC-20565 KY Cyg VX Sgr
T Per     ST Cep MZ Cas  
KK Per     V358 Cas NR Vul  
BD+47485     V540 Sgr RS Per  
      XX Per SU Per  
      YZ Per UW Aql  
        V355 Cep  



  \begin{figure}\includegraphics[width=9cm]{h1934f2.eps}\end{figure} Figure 2: Continuum-subtracted AGB star spectra classed as showing broad features. x-axis is wavelength in $\mu $m. y-axis is flux in W m$^{-2}~\mu$m-1


  \begin{figure}\includegraphics[width=9cm]{h1934f3.eps}\end{figure} Figure 3: Continuum-subtracted AGB star spectra classed as showing broad+sil features. x-axis is wavelength in $\mu $m. y-axis is flux in W m$^{-2}~\mu$m-1


  \begin{figure}\includegraphics[width=9cm]{h1934f4.eps}\end{figure} Figure 4: Continuum-subtracted AGB star spectra classed as showing group A silicate features. x-axis is wavelength in $\mu $m. y-axis is flux in W m$^{-2}~\mu$m-1


  \begin{figure}\includegraphics[width=9cm]{h1934f5.eps}\end{figure} Figure 5: Continuum-subtracted AGB star spectra classed as showing group B silicate features. x-axis is wavelength in $\mu $m. y-axis is flux in W m$^{-2}~\mu$m-1


  \begin{figure}\includegraphics[width=9cm]{h1934f6.eps}\end{figure} Figure 6: Continuum-subtracted AGB star spectra with group C silicate features. x-axis is wavelength in $\mu $m. y-axis is flux in W m$^{-2}~\mu$m-1


  \begin{figure}\includegraphics[width=9cm]{h1934f7.eps}\end{figure} Figure 7: Continuum-subtracted AGB star spectra classed as showing group D silicate features. x-axis is wavelength in $\mu $m. y-axis is flux in W m$^{-2}~\mu$m-1

We present here a survey of dust features present in the mid-IR spectra of 80 oxygen-rich AGB stars and 62 M-supergiants, with a view to identifying similarities and differences between the features amongst these various types of stars. The selection of targets was largely dictated by their accessibility to the 3.8-m UKIRT on Mauna Kea at the dates of several different observing runs. The AGB stars were observed mainly during two runs, in May 1991 and October 1992, although RW And and RZ Ari were observed in October 1990 and SX Cyg was observed in November 1993. The supergiant stars were observed over a wider range of dates, from October 1992 to June 1998. The supergiant spectra from October 1992 and August 1995 have already been published by Sylvester et al. (1994, 1998). They are included here because they are used for a different purpose to that of Sylvester et al.

The nature of our survey means that these are snapshot spectra, obtained at a random phase of the stellar pulsation cycle. Work by Little-Marenin et al. (1996), Creech-Eakman et al. (1997) and Monnier et al. (1998) has indicated that AGB star silicate emission bands can vary systematically as a function of pulsation cycle, primarily in the strength of the feature but sometimes also in its profile. However, these observed profile variations have in general been significantly less pronounced than the differences between the emission feature types that are classified here.


  \begin{figure}\includegraphics[width=9cm]{H1934F8.ps}\end{figure} Figure 8: The continuum-subtracted spectrum of T Cet. x-axis is wavelength in $\mu $m. y-axis is flux in W m$^{-2}~\mu$m-1

2.1 Observations

Observations were taken using the 3.8-m United Kingdom Infrared Telescope (UKIRT) with the common-user spectrometer CGS3 (see Cohen & Davies 1995). We obtained 7.5 - 13.5 $\mu $m spectra with a 5.5-arcsec circular aperture, and a spectral resolution of 0.17 $\mu $m. Two grating settings gave a fully sampled 64-point spectrum. Wavelength calibration was with respect to observations of a Kr arc-lamp. The telescope secondary was chopped east-west at 5 Hz using a 30-arcsec throw. The error bars in the plots represent 1$\sigma$ standard errors on the fluxes. Due to imperfect cancelation of the time-varying atmospheric ozone band, the error bars are larger at approximately 9.6 $\mu $m. Details of the observed stars can be found in Table 1 for the AGB stars and Table 2 for the supergiants.

2.2 Classifying the features

For each source in our sample a 3000 K blackbody representing the stellar photosphere was normalized to the spectrum at 8.0 $\mu $m. This was then subtracted from the observed astronomical spectrum to yield the spectra that are plotted in Figs. 2-8 and 10-14. Since the mean level of the spectra between 7.5 and 8.0 $\mu $m is zero and the error bars are larger because of greater atmospheric absorption between these wavelengths, we have only plotted our continuum-subtracted spectra from 8.0 $\mu $m onwards. Our continuum subtraction procedure implicitly assumes that the 8.0 $\mu $m continuum is dominated by blackbody-like stellar photospheric emission, i.e. that neither continuum dust emission nor gaseous SiO fundamental-band emission or absorption is important. Higher resolution spectra of the SiO fundamental and first overtone bands (e.g. Tsuji et al. 1997; Waters et al. 1999; Aringer et al. 1999) indicate that the gaseous SiO band is not likely to have significantly perturbed our low resolution spectra. A much more detailed modeling of wide spectral coverage ISO-SWS spectra would be needed to investigate whether continuum dust emission in general makes a significant contribution at 8 $\mu $m. However, the ISO-SWS spectra of Tsuji et al. (1997) indicate that for high mass loss stars, such as the supergiant S Per, continuum dust emission is significant at 8 $\mu $m but that for lower mass loss rate stars (the majority of our sample) the contribution from continuum dust emission at this wavelength is much less pronounced.

In producing our continuum-subtracted spectra (CSS), we have assumed that the photospheric temperature for each star is 3000 K. However, we find that varying the adopted blackbody temperature by $\pm$1000 K about the adopted 3000 K has very little effect on the shape and strength of the derived continuum-subtracted dust features, since we are in the Rayleigh-Jeans domain of the photospheric Planck distributions. Therefore the exact effective temperatures of the stars, within the natural scatter that is expected for these stars, should not affect our data analysis.

In order to aid comparisons between different spectra, we also produced normalized "continuum-subtracted spectra'' (CSS) for the AGB stars and supergiants in our sample, by normalizing the CSS to a peak flux of unity. The normalized CSS of each AGB star was compared to those of the other AGB stars to find similarities and form groups. Leaving aside the seven AGB stars in our sample whose individual spectra show no pronounced emission features (classed as "featureless'' in Table 3), the dust features of the AGB stars have been classified into three groups: broad AGB, where the feature extends from 8 $\mu $m to about 12.5 $\mu $m with little structure; broad+sil AGB, which consists of a broad feature with an emerging 9.7 $\mu $m silicate bump; and silicate AGB, which is the "classic'' 9.7 $\mu $m silicate feature. In addition, the silicate AGB group can be classified further into four subgroups, A, B, C, D, which show slightly differently shaped silicate features, as shown by the mean spectra. These variations seem to show a sequence from broader to narrower silicate feature. The stars in each group are summarized in Table 3, and the mean feature profile for each group is shown in Fig. 1. The original CSS are shown in Figs. 2-7, grouped in the classes described above.

The broad+sil AGB groups can also be split into two groups according to whether the spectra show the 12.5 - 13.0 $\mu $m feature. This will be discussed later.

Two of our sources, T Cet and V Eri, have been classified both as AGB stars and as supergiants in various published works. For this reason, exactly whether these stars should be included in the AGB star groupings or the supergiant groupings is ambiguous. In fact T Cet has been classified as an AGB star (e.g. Bedding & Zijlstra citeB98), a supergiant (e.g. SP98), M-star (e.g. Bedding & Zijlstra 1998), MS-star (e.g. Groenewegen & de Jong 1998) and S-star (e.g. Skinner et al. 1990; Groenewegen 1993). This source also has a markedly different spectrum from any others in the sample (e.g. Speck 1998; Skinner et al. 1990). We can see in Fig. 8 that the spectral features for this source are unlike any others in our sample and it therefore needs to be treated separately. For this reason, T Cet was removed from the sample. V Eri has variously been classed as an AGB star and as a supergiant (see e.g. Hashimoto 1994; Habing 1996; Neri et al. 1998; Triglio et al. 1998; Cernicharo 1998 etc.). Kwok et al. (1997) give V Eri the spectral classification M6II, i.e. intermediate between a giant and a supergiant, while Houk & Smith-Moore (1988) assign V Eri a spectral class of M5/M6IV, which would make it a subgiant. Furthermore, V Eri exhibits the 12.5 - 13.0 $\mu $m feature, which is seen in many AGB star spectra but only in one other supergiant spectrum (that of S Per; see Sect. 3). We have therefore placed it in the AGB grouping, rather than in the supergiant grouping.


  \begin{figure}\includegraphics[width=9cm]{h1934f9.eps}\end{figure} Figure 9: Mean spectra for each of the supergiant star groups - progressing between the broadest feature (bottom right) and the narrowest silicate feature (top left)

In the same way as for the AGB stars, the normalized CSS of each supergiant was compared to those of the other supergiants to find similarities and form groups. The dust features of the supergiant stars can also be classified into three basic groups: featureless, whose emission above the continuum is weak and hard to classify; broad Super, where the feature extends from $\sim $9 $\mu $m to $\sim $13 $\mu $m; and silicate Super, which again is the "classic'' 9.7 $\mu $m silicate feature. Since the featureless spectra show little of interest regarding dust they will not be discussed further here. The broad feature can be classified into two further subgroups, one extending from 9 to 13 $\mu $m, and one with a short wavelength onset at 9.5 $\mu $m. The difference between these two broad groups may be entirely due to differences in the underlying dust continuum. Examination of the silicate features in these spectra is slightly compromised by the appearance of UIR bands (see Sylvester et al. 1994, 1998). The strong 11.3 $\mu $m UIR band has been edited out of the averaged spectra shown in Fig. 9. Again, the "classic'' silicate group (silicate Super) can be classified further into four subgroups, 1, 2, 3 & 4, where the exact shapes of the silicate features vary slightly and the averages show this. One supergiant with a silicate feature, U Lac, did not fit into any of these four groups, but does have the basic silicate feature and matches group B from the AGB star classifications. The stars in each group (excluding featureless) are summarized in Table 4, the mean feature from each group is shown in Fig. 9, and the original CSS are shown in Figs. 10-14, grouped in the classes described above.

Our classifications of AGB stars and supergiant can be compared to those of SP95 and SP98. These comparisons are shown in Tables 5 and 6. The classification system used by SP95 & SP98 gives each spectrum a number: SE1 to SE8, where SE1 refers to the broad feature; the SE number increases as the silicate feature emerges up to SE8 which refers to the "classic'' narrow silicate feature. We would therefore expect spectra in the our broad feature class to have low SE numbers. As we progress through the classes up to the narrow silicate feature class, the SE number should increase. In general we concur with the SP95 and SP98 progression, as can be seen in Tables 5 and 6. However, there are a couple of spectra for which the classifications are markedly different: W Hya, which we class as having a broad feature while SP95 give it a strong silicate feature; and R Cae, which we class as silicate D and they give a broad feature class. These differences may be due to their classification method, which only uses relative fluxes at 10, 11 and 12 $\mu $m rather than the overall shape of the features. However, in comparing the classifications used we found that none of the stars that we have classified as broad Super appear in their dataset and they do not have any examples of the broad feature. This is unfortunate, since it appears to be one of the few ways in which the observed dust features differ between AGB stars and supergiants.


  \begin{figure}\includegraphics[width=9cm]{h1934f10.eps}\end{figure} Figure 10: Continuum-subtracted supergiant spectra classed as showing broad features. x-axis is wavelength in $\mu $m. y-axis is flux in W m$^{-2}~\mu$m-1


 

 
Table 5: Comparing classifications - AGB star features
Source Variability Our SP95 Source Variability Our SP95
  Type Class Class   Type Class Class
BU And M featureless SE2 CH Pup M silicate D SE4
R Hya M featureless SE2t R Cae M silicate D SE3
R Peg M featureless SE4 R Cas M silicate D SE5
T Cas M featureless SE1 W Eri M silicate D SE6
V Cas M featureless SE1 Y Aqr M silicate D SE4
BG Cyg M broad SE2t IK Tau M silicate C SE5
R Aql M broad SE5 R Aqr M silicate C SE6
R Leo M broad SE2 RT Lep M silicate C SE6
RR Sgr M broad SE4 RU Aur M silicate C SE8
RT Eri M broad SE3 RU Her M silicate C SE5
S Col M broad SE2 EP Aqr SR silicate C SE4
SZ Aur M broad SE2t TX Cam M silicate C SE5
W Aql M broad SE3 U Pup M silicate C SE6
W Hya SR broad SE8 Y Lyn SR silicate C SE8
YY Cep M broad SE4 Z Sgr M silicate C SE6
V Eri SR broad SE3t DU Pup M silicate B SE5
X Oph M broad SE1 GX Mon M silicate B SE6
RR Per M broad+sil SE2 R Cet M silicate B SE6
RW And M broad+sil SE3 RU Cap M silicate B SE8
RX Boo SR broad+sil SE3t S CrB M silicate B SE5
RZ Cyg SR broad+sil SE3t U Aur M silicate B SE4
UW Cep M broad+sil SE3 V1111 Oph M silicate B SE5
V462 Cyg M broad+sil SE1 V342 Sgr M silicate B SE6
W Aqr M broad+sil SE3 Z Cyg M silicate B SE8
Y Cas M broad+sil SE3 Z Pup M silicate B SE6
Z Cas M broad+sil SE1 CU Cep SR silicate A SE6
SV Peg SR broad+sil SE3t RU Cyg SR silicate A SE5
AZ Mon M broad+sil SE3t U Her M silicate A SE4
        UU Peg M silicate A SE5
        X Ori M silicate A SE4
        o Cet M silicate A SE8


2.3 Comparing dust features from AGB stars and supergiants

Having classified the spectra into groups, we have compared the spectral features for AGB stars and supergiants.

2.3.1 The "classic'' silicate feature

Amongst both the AGB and supergiant classes we have distinguished sub-groupings of the "classic'' silicate feature. Comparing these sub-groups, we find that the various silicate features found in the spectra of AGB stars are very similar to those found in the spectra of supergiants. Figures 15a-d shows the comparisons of the various silicate features. In each case the mean AGB silicate feature for each sub-group is plotted along with the closest matching supergiant silicate features. Figure 15a compares AGB silicate group A with two supergiant silicate groups (1 & 2). The AGB feature is found to match the average of these two supergiant features. Figure 15b shows AGB silicate group B, together with the spectrum of the unclassified supergiant U Lac. Figure 15c shows AGB silicate group C with two supergiant silicate groups (3 & 4). As with AGB silicate group A, this AGB group appears to be intermediate between the two supergiant groups. Figure 15d shows AGB silicate group D with the supergiant group 4. It is clear from these figures that, although the silicate feature varies from star to star, similar shaped features are present in the spectra of both AGB stars and supergiants, and seem to form a progression between broader and "classic'' narrow 9.7 $\mu $m silicate features.

2.3.2 The broad feature

As we can see from Fig. 16, the broad AGB star feature is quite different from the two broad supergiant features. The broad features for supergiants are slightly narrower than for AGB stars, extending from 9.0 - 9.5 $\mu $m to $\sim $13 $\mu $m, rather than 8 - 12.5 $\mu $m as in the AGB star spectra. Furthermore, the flux continues to drop longward of 12.5 $\mu $m, rather than leveling off as in the case of the AGB stars. In addition, the broad feature for AGB stars seems to develop a silicate peak on its way to becoming a "classic'' silicate feature (see broad+sil classification; Fig. 1); this does not happen in the supergiant spectra. Only one supergiant in the sample, S Per (see Fig. 14), exhibits the 12.5 - 13.0 $\mu $m feature seen in some AGB star spectra and this may be due to this star's unusual characteristics. It is believed that S Per is a relatively new supergiant and that its dust shell is thinner and more spherically symmetric than is usually found for supergiants (Richards 1997; Richards et al. 1999). These characteristics make the dust shell more like that of a semiregular red giant than a supergiant and may explain the appearance of the 12.5 - 13.0 $\mu $m feature in its spectrum.

2.4 The featureless spectra

Those AGB and supergiant 8 - 13 $\mu $m excess spectra which have been labeled as "featureless'' are all very weak compared to the local stellar photospheric continua. Signal to noise considerations may therefore be largely responsible for the inability to classify these excess spectra. To test this, we co-added the five AGB star 8 - 13 $\mu $m excess spectra labeled as "featureless'' in Table 5. The resulting mean spectrum had better S/N and resembled the broad AGB star feature discussed above. It may therefore be the case that the respective broad features are responsible for the low-contrast "featureless'' spectra of AGB stars and supergiants.

next previous
Up: Dust features in the

Copyright The European Southern Observatory (ESO)