Separation of M and B star components in the spectrum

[Robin Leadbeater]

I have compared my latest AZ Cas spectrum near H alpha with an early M supergiant spectrum from the UVES bright stars catalogue (I chose HD49331)

(Note that I have matched the AZ Cas spectrum to the HD49331 velocity frame)
There is obvious agreement but there is much fine detail that is lost at R ~5000 compared with the much higher resolution of UVES. If we filter the UVES spectrum to R ~5000, the match is excellent except at H alpha where the effect of the B star is seen

By dividing the AZ Cas spectrum by the HD49331 spectrrum we can see the effect of the B star (and the effect of any interraction between the two stars)

Note the strong H alpha emission which agrees with what is seen well outside eclipse. (The outer regions of the M star are already absorbing the light from the Be star) It is redshifted relative to the M star absorption which is as expected at this phase.

The small residual Ha absorption, blue shifted relative to the M star absorption is interesting. Perhaps this indicates transfer of material between the two stars ?

Cheers
Robin

[Andrew Smith]

Interesting work as always Robin. I was just wondering about dividing of one spectrum by the other rather than scaling and subtraction. I think you used the latter when removing telluric line in your previous work. Did you change the approach for a reason?

I can see that what you are doing is equivalent to treating the M star as part of the instrumental response calibration but when I try the sums I don't see how it just leaves the B star component.

I am probably mistaken and did the sums wrong.

Regards Andrew

[Robin Leadbeater]

You are right Andrew, this is a gross oversimplification of the situation.

When the two stars are well separated, the two spectra should indeed add. Because they are so different in temperature and luminosity though, most of the flux in the region around H alpha comes from the M star so subtracting or dividing would give a similar looking result (just scaled differently)

At this point in the orbit though the Be star is partly obscured by the extended outer atmosphere of the M star (ie the M star atmosphere is acting like a filter with a very complex wavelength response) which is why I chose to divide by it so I could then perhaps separate out the underlying H alpha profile from the Be star unaffected by the M star

As we aprroach periastron, changes are expected in the H alpha emission on top of the effect of absorption from the M star atmosphere. It is unfortunate that the periaston and eclipse are so close in the orbit such that the two effects overlap, though this of course makes it all the more interesting This is still probably a gross oversimplification so I offer it just as an interesting observation, nothing too rigorous

Cheers
Robin

[Robin Leadbeater]

Hi Andrew

I was just wondering about dividing of one spectrum by the other rather than scaling and subtraction. I think you used the latter when removing telluric line in your previous work.

You definitely divide when removing tellurics (the atmosphere is a filter) The scaling was to scale the magnitude of the telluric lines in my standard template to match the magnitude of the tellurics seen in the star spectrum. The resulting scaled telluric spectrum (normalised to 1) is then divided into the star spectrum (Vspec handles this all automatically in the telluric line removal function)

Cheers
robin

[Francois Teyssier]

For symbiotic stars, whose composite spectrum is the addition of a hot compact and a late type giant stars (with nebular continuum), the method is to *subtract* the spectrum of the M Star.

Best regards

François

[Andrew Smith]

Thanks Robin & François - I think I have it now. If you have two independent sources then they add so you need to subtract one to get the other. If one source is being filtered then you divide. In this example we have a mixture of the two.

Please correct if I have it wrong.

Regards Andrew