# Why does Gaia use only calcium NIR lines for stellar radial velocity measurements?

I was reading this overview article about the Gaia spacecraft and I saw the following statement:

These spectra provide radial velocity information that are used to study the kinematic and dynamic evolution of the Milky Way. Radial velocities are derived from three isolated Calcium lines at 849.8, 854.2 and 855.2nm. Other lines in the 847 to 874nm range can provide data on star composition, surface gravity, and metal abundance.

note: as pointed out in the comments below, the third line is at 866.2nm, not 855.2nm - this is suggested to be a typo - the '855' number also shows up on this ESA page.

The high resolution spectroscopy only seems to operate between 847 and 874nm, and "three isolated Calcium lines" are used to measure radial velocity.

Do all stars have enough calcium in their atmosphere to produce strong enough features to measure the radial velocity so accurately? I had thought that there are some stellar populations have very little besides hydrogen and helium in their atmosphere.

Are these always emission lines or absorption lines, or will there be some stars with one and some of the other? What fraction of stars simply won't have significant amounts of calcium?

above: Gaia's Radial Velocity Spectrometer from here, credit: ESA.

above: Gaia's Imaging system, including mirrors 4, 5 and 6, prisms, diffraction gratings, and CCD array, from here, credit: EADS Astrium.

above: Gaia's Optical Module, including Ravial Velocity spectrometer (gratings) and afocal field corrector, from here, credit: SAS Astrium.

• Other sources have the third Ca II line at 866.2 nm, not 855.2 nm. – Mike G Sep 16 '16 at 21:16
• Correct, it's a typo. 866.2 nm. – Rob Jeffries Sep 17 '16 at 0:00
• @RobJeffries the '855' value shows up on this ESA page as well, as shown in this answer below. I've added a note in the question (I don't want to help propagate the number if it's incorrect). I wonder how far it goes! A quick google search shows 849.8 nm, 855.2 nm and 866.2 nm which contains '855' in a different location. – uhoh Sep 17 '16 at 0:48
• In order to accommodate a very large number of stellar spectra simultaneously superimposed on the RVS CCD array at high dispersion, it seems to have been necessary to choose only a narrow wavelength range. So far there are three good answers here that explain that the Ca II triplet is present in a wide range of stars and usually narrow, the Paschen hydrogen series is close by for the hotter stars, and it is near the "energy-distribution peaks of G- and K-type stars which are the most abundant RVS targets." In this case I can't choose a single "accepted" answer & encourage up-voting them all! – uhoh Sep 17 '16 at 5:38
• The Ca IR triplet is at 849.8, 854.2 and 866.2 nm en.m.wikipedia.org/wiki/Calcium_triplet See also any picture of a spectrum! – Rob Jeffries Sep 17 '16 at 6:40

The Ca triplet in the near infrared are extremely strong resonance absorption lines. They are by far the strongest features in the near infrared spectra of cool G,K,M type dwarfs and giants, which will be the majority of the stars observed by the Gaia RVS. The Ca triplet lines are so strong that even in low metallicity halo stars, that have little Ca in their photospheres, these lines are still strong enough to measure radial velocities.

The lines are much weaker and much broader for hotter O, B and A stars, and measuring radial velocities for these will be difficult and much less precise.

You can have a look at an atlas of the Gaia Ca triplet region for stars of different spectral types in Figure 2 of Munari et al. (2001). http://cds.cern.ch/record/531022/files/0109057.pdf

I should also add that these three lines are not the only features used to determine the velocities, they are just the strongest features in the spectra of most stars.

The ESA states it pretty clearly (although their figure of 855.2 nm is incorrect; it should be 866.2 nm):

The RVS wavelength range, 847-874 nm, has been selected to coincide with the energy-distribution peaks of G- and K-type stars which are the most abundant RVS targets. For these late-type stars, the RVS wavelength interval displays, besides numerous weak lines mainly due to Fe, Si, and Mg, three strong ionised calcium lines (at around 849.8, 854.2, and 855.2 nm).

Using Wien's law, we can see that stars with these as peak wavelengths in this interval correspond to effective temperatures in the range of 3000-3500 K: $$T=\frac{b}{\lambda_{\text{max}}}$$ $$\begin{array}{|c|c|} \hline \text{Wavelength (nm)}&\text{Temperature (K)}\\ \hline 847 & 3431\\ \hline 849.8 & 3409\\ \hline 854.2 & 3392\\ \hline 866.2 & 3345\\ \hline 874 & 3315\\ \hline \end{array}$$ In reality, the majority of the stars Gaia studies have the most intense emissions at effective temperatures higher than this; these peaks correspond to hot M-type stars, not K- or G- type stars. The Sun, for instance, has an effective temperature of about 5800 K, and many K-type stars have effective temperatures around 4000 K. However, the target stars still guarantee intense emissions in the relevant parts of the spectrum, and thus noticeable calcium lines.

According to Cropper and Katz 2011 part 2.2, the RVS working group considered other bands, but the ~850 nm band is relatively unaffected by absorption in the Earth's atmosphere, facilitating ground-based preparation and follow-up. In addition to the strong Ca II triplet, this band is rich in lines enabling study of astrophysical quantities other than radial velocity, adding to the science return on the spectrometer investment.

For type B and hotter stars, a small minority of the population, they hope to get radial velocity from the Paschen hydrogen series, which accounts for the wide troughs at 854.3, 859.6, and 866.3 nm in the top of Munari 2001 figure 2.

• Thanks - this is very helpful to understand better the various considerations involved in selecting the final wavelength band for the RVS. – uhoh Sep 17 '16 at 1:13