1.0 Light can be dispersed into colors
White light is really the sum of many different
We can use a dispersing element to analyze the amount/intensity
light at each wavelength (this is called a spectrum)
a nice image of prism from here
The reason a spectrum is made is due to light slowing down in
denser glass so it must take a more direct route. This is know
Snell's Law: n1sinθ1
so as we can see how Snell's law applies to a prism. from
This shows that since n is a function of lambda
we get a
this site has nice images for basic optics
2.0 Dispersion grating
While prisms are fine for very low resolution spectra in
astronomy we usually need higher spectral resolutions.
a diffraction grating is a reflecting or transparent element, whose
properties are periodically modulated. Most commonly the
gratings are realized as fine parallel and equally spaced
rulings on material surface. When light is incident on a
diffraction grating, diffractive and mutual interferencediffraction orders.
Because of their dispersive properties,
commonly used in monochromators and spectrometers.
a cartoon of a transmission grating from here
A typical reflective diffraction grating from here
what are these orders physically? -- constructive interference!
one way of thinking about these orders from
2.1 The grating Equation
where: n is the order of diffraction, λ is the diffracted
is the grating constant (the distance between successive
grooves) θi is
the angle of incidence measured from the normal and θd is the
diffraction measured from the normal.
The diagram above shows the orders of the diffracted wavelength.
well as positive orders, light can also be diffracted in the
direction (i.e. n = -1, -2 etc.) Higher orders may also appear
these decrease in intensity. Usually the first order lines (n=1
n=-1) are the most intense.
Once we have a grating we need to use a spectrograph to
produce a spectrum
A more "real" design looks like that below from the Carbon
SLITS --- why do we need them?
By use of slit we can keep light from the seeing disk from
The end result is a nice spectrum
site for background on stellar spectra
the spectra of stars look like
This is how astronomers classify different stellar types (in
decreasing effective temp)
Usually astronomers only plot the spectra as lines
What does the spectra really look like on our CCD?
Clearly a lot of processing is required before we can extract
the spectrum. Compare to the calibrated B5V spectrum above...
RESOLUTION of a Spectrograph
R = Lambda / Delta_Lambda = c/Delta_v
where Delta Lambda is the smallest difference the spectrograph
and c = speed of light, and Delta_v is the smallest velocity
difference that can be measured.
(STIS) can distinguish features 0.17 nm apart at a wavelength of
1000 nm, giving it a resolution of 0.17 nm and a
resolving power of about 5,900
The STIS example above then
has a spectral resolution of 51 km/s.
5.0 SPECTROGRAPH DESIGN (lecture 15)
see here for the rest of the
on spectrograph design
from Dennis Zaritsky's lecture.
6.0 SPECTRAL DATA REDUCTION (lecture 16)
See here for notes on how IRAF can
(lecture notes from M. Bote lick obs.)