Download Report

Transcript 5X_Measuring_galaxy_redshifts

Spectra of galaxies from E  Im:
Continuum: sharp break and rise
after H and K lines (hence brighter
Kennicutt (1992)
in red than blue); dominated by red
star light, strong absorption lines;
E and S0 have K-star type spectrum
Galaxies are shown
Flattening of continuum
in order of
(reduction of red light compared to
increasing Hubble
blue light)
type from top to
Spirals: F-K
stars dominate spectrum
Decrease in strength of
absorption lines
Increase in emission lines and
strength of emission lines;
A-star dominate spectrum
but also
- lots of SF, many O, B stars:
- abs. lines of He (typical of O, B stars)
- em. lines of ionized gas
AST 3003
Notes from Anthony Fairall – based on chapter 2 on Measuring Galaxy
redshifts of his book “Large-Scale Structures in the Universe”, 1998, WileyPraxis Series in Astornomy and Astrophysics
Most galaxies are distant, small angular size, only slightly larger than
the ‘seeing’, with light centrally concentrated – that light is captured
and fed to the slit of a spectrograph.
The telescope optics direct the light of the distant galaxy onto the slit.
The width of the slit is a compromise between the seeing and the
resolution of the spectrograph. A very narrow slit gives better
resolution, but may only capture a fraction of the galaxy’s light. Usually
the slit width is matched to the general seeing.
The Collimator then converts light passing through the slit into a parallel
beam. A miniature Cassegrain telescope, used in reverse mode, is a
favoured design for a collimator. The focal ratio of the collimator must
match that of the telescope.
Many spectrographs use reflection gratings, which work by the constructive interference
of light:
nλ = a(sinα + - sinβ)
where n is the order number
Angle θ is the blaze angle, which concentrates the light in a general direction, favouring
one or more of the orders.
The beams reflected off the grating are intercepted by a camera, which reverses the role
of the collimator by bringing the parallel beams back to focus.
The image of the slit is smeared sideways to form a spectrum. Slit width and dispersion
determine the resolution Δλ, which can also be expressed as
Resolution: R = λ/Δλ
The final image is focussed on a CCD detector –> a high quality version of those used in
everyday cameras.
CCDs have a matrix of potential wells within a silicon chip. These wells – each a pixel trap photo-electrons released by incoming light. By moving the wells like a conveyor belt,
the pixels are led to the corner of the CCD where the content of each is assessed.
Special elongated CCDs accommodate extended (effectively one-dimensional) spectra.
CCD’s are very efficient: 75% efficiency of light compared to 5% of photographic plates
BUT; low sensitivity in blue
Multi-fibre spectrographs enable the spectra of many neighbouring
galaxies to be recorded simultaneously.
(SALT uses mulitple slits to allow for MOS-Mulit-object spectroscopy)
They use optical fibres to isolate light from different galaxies in the focal
plane of the telescope.
One technique is to locate the image in the focal plane in register with
that of a metal plate, prepared with drilled holes. Fibres are plugged
into the holes.
The 2dF/6dF systems (British-Australian) have the fibres connected to
magnetic buttons (with miniature prisms). A robot sets up the field
beforehand by drawing out each fibre from its storage and placing the
button on a metal plate exactly where the light of a particular galaxy will
fall. One field can be set up whilst another is being observed.
Galaxies are relatively faint objects and integrations have to be
extended so that enough photons are gathered. At low photon counts,
photon noise is significant – a case of signal versus noise.
Determine Dark Counts (or bias) at beginning or end of night
In reality, the situation is also complicated by cosmic rays, which
dapple the image. Fortunately their damage is confined to single pixels,
and can be tidied up by software routines.
 easily removed by having 2 (or more) exposures
Response of pixels in CCD: not uniform  measure flats
The pixel elements of a CCD do not have identical efficiencies in
recording photons. Flat field exposures (e.g. normal white light) enable
the relative sensitivities to be determined.
Calibration arcs – from low-density argon, neon or helium lamps – bracket the galaxy
observations. The wavelengths of the arc lines are known precisely.
Wavelength and pixel number are reconciled in a computer reduction, normally by an
iterative fit to a polynomial relation.
After that the galaxy spectrum can be re-binned in wavelength intervals.
Radial velocity standard stars, often observed in the twilight, also serve
as a check. Spectra have good signal to noise and reveal the principle
absorption features.
Sky subtraction is very necessary for faint objects. the night sky – even
without moonlight – has significant continuum towards the blue and
prominent emission features (forbidden Oxygen lines) at 5577, 6309
and 6364 Angstroem
The galaxy spectra following also exhibit modest redshifts, spectral
features have been shifted to slightly longer wavelengths (Doppler
While stellar content may be assessed, the main purpose in obtaining
such spectra is to extract a redshift:
z = (λ / λ0) – 1
when this is small, it can be converted into a velocity of recession via
V = cz
However, if the velocity is relativistic, the correct relationship is
(z + 1) 2 -1
(z + 1) 2 + 1
The convention however is to publish cz, but confusion exists as some
software packages apply the relativistic correction.
The other convention is to publish heliocentic velocities (i.e. correcting
for the Earth’s orbital motion of 30 km/s).
Further corrections may be made. In his ‘Reference Catalogues’, de
Vaucouleurs advocated correcting to the local frame of rest within the
Local Group of galaxies via
V0 = V + 300 sin l cos b
where l and b are the Galactic longitude and latitude, though there is
some debate as to this correction.
By contrast, radio astonomers are fond of correcting to the ‘Local
standard of rest’ after removing the Sun’s motion relative to
neighbouring stars, which is believed to be 19 km.s towards RA 18h,
Decl. = +300
The modern day alternative to making manual measurements of the
wavelengths of redshifted spectral features is to conduct a mathematical
‘cross-correlation’ .
To do this, the spectrum is first re-binned into channels of equal width in ‘log
wavelength’, such that the Doppler shift is the same for all channels. Largescale variations are then filtered out (i.e. only spectral lines left). The spectrum
is then ‘slid’ against a template prepared from a known bright galaxy or star,
and a correlation function derived.
The peaks in the function show where the spectrum best fits the template. The
highest peak is almost certainly the correct redshift.
Emission-line galaxies have wonderfully sharp spectral features that
give precise redshifts. Again the correlation technique can be used, but
with an emission-line template.
Except for very nearby galaxies, the redshifts normally reveal the cosmological
expansion of the Universe, as reflected in Hubble’s law.
(Licence has been taken in labelling it a ‘Doppler’ shift as that is not true, except it is equivalent to a
Doppler shift, as the galaxies are moving away from our Galaxy.)
However, peculiar velocities may add to or subtract from the cosmological
The spectra also reveal the nature of the dominant optical emission from the
galaxy. Usually it is stellar, i.e. absorption lines. Average spectral types are
usually late G-early K, though a few ‘early-type’ spectra do appear. But traces
of weak emission may show up in <50% of galaxies.
About 10% of galaxies show strong emission lines, that light originating from
nebular, not stellar, sources. But these usually have strong blue continua from
hot young stars – the source of excitation.
About 1-2% of galaxies reveal broad emission lines – of active galactic nuclei,
the topic covered in a following section.