Paper Review: Discovery of an ultra metal-poor star

I chose to review this paper, written by Else Starkenburg et al., because of its relevance to our own project. It details the discovery of an especially metal-poor star – poetically named Pristine_221.8781+9.7844 – and it explains the methods used to identify it and determine its metallicity.

This ultra metal-poor star was discovered using the Pristine survey, which is a narrow-band Ca H & K survey based in the Northern Hemisphere. Essentially, that means it applies a photometric filter to the light coming from the sources within the survey region, and this filter only allows through light with a very specific, very narrow range of wavelengths. In this case, that range of wavelengths is centred around the so-called calcium H and K lines, which are two absorption lines (sharp dips in brightness) that appear in a star’s spectrum if that star has an abundance of singly ionised calcium (Ca II) in its atmosphere.

Screen Shot 2018-08-02 at 17.06.54.png
Figure 1: A segment of the normalised spectrum for Pristine_221.8781+9.7844. The top panel shows the spectrum as observed with WHT/ISIS, while the bottom panel shows the spectrum when normalised by a ‘running mean’ implemented using the FERRE code (Allendo Preito et al. 2006), along with the normalised best-fit synthetic spectrum overplotted in red. The sharp dips in flux are the absorption lines, each of which indicate the presence of some chemical element or compound in the star’s atmosphere (note that multiple lines can correspond to the same element/compound). Here we see lines for neutral hydrogen (H) and singly ionised calcium (Ca II), where the two calcium lines are the Ca H & K lines.


Having previously identified Pristine_221.8781+9.7844 as a potential candidate for an extremely metal-poor star, the authors go into some detail in this paper about the investigative techniques used to determine just how metal-poor it really is.

First, they explain how they used photometry to derive stellar parameters for Pristine_221.8781+9.7844, such as its size, its surface gravity, its surface temperature, its distance from us, and its 3D velocity vector. I shan’t go into the details of this (partly because I don’t fully understand them myself), but ultimately they claim that this star is a sub-giant with a surface gravity of log(g) ~ 3.5 and a surface temperature of 5792K, located approximately 7kpc away from us, with a velocity consistent with that of a star located in the galactic halo.

They then follow this up with medium-resolution spectroscopy, which they use to focus on the Ca H & K lines (see figure 1 above), from which information about the star’s metallicity can be inferred. Generally speaking, the smaller the absorption line, the more metal-poor the star will be. An important note is that the Ca II H line is much larger than the Ca II K line to its immediate left, but the authors point out that this is because the Hℇ line (one of the infamous hydrogen Balmer lines) is blended with it. The actual abundance of Ca II in the star’s atmosphere is therefore quite low, and as a result they estimate the star’s metallicity to be [Fe/H] = – 4.66. What this essentially means it that, for Pristine_221.8781+9.7844, the ratio of atmospheric iron (Fe) to hydrogen (H) abundances is around 46,000 times smaller than that of the Sun. That’s pretty darn metal-poor! Furthermore, they also state that due to the absence of an absorption band at around 4300Å (see figures 1 and 2), this star could be very carbon-poor in comparison to stars with similar [Fe/H] values, implying that Pristine_221.8781+9.7844 is a particularly special case.

Screen Shot 2018-08-06 at 16.45.30.png
Figure 2: Synthetic spectra for different stars, all of which have a surface temperature of 5792K, surface gravity of log(g) = 3.5, and metallicity [Fe/H] = – 4.66, but with different carbon abundances. The black line shows the low-resolution spectrum of Pristine_221.8781+9.7844, as viewed with the William Herschel Telescope (WHT), which clearly bears the most resemblance to the synthetic spectra with the lowest carbon abundances.
High-resolution spectroscopy is then used to measure the abundances of lithium (Li), sodium (Na), and the so-called ∝-elements (one of which is carbon), via four different methods. Again, I won’t go into detail about the methods, but suffice to say that they are all in strong agreement with each other, and all give results consistent with an extremely metal-poor star, which is assuredly a good thing. Even at high resolutions, there is a notable absence of any specific carbon-related features, which reinforces the idea that this star is not only metal-poor, but also remarkably carbon-deficient for a star with such low metallicity.

The authors then compare Pristine_221.8781+9.7844 with the most metal-poor star currently detected: SDSS J102915+172927. They conclude that while SDSS J102915+172927 shares many similarities with Pristine_221.8781+9.7844 (e.g. similar surface temperatures, similar colours, similar Ca II and Fe absorption line strengths), but there is a striking difference in the strengths of the magnesium (Mg) absorption lines for the two stars, with Pristine_221.8781+9.7844 displaying much stronger lines than J102915+172927 (see figure 3 below). They say that this implies an elevated abundance of ∝-elements in the former, which makes sense since magnesium is indeed one of these ∝-elements, although they seem to make no comment regarding the significance of this.

Screen Shot 2018-08-07 at 13.13.21.png
Figure 3: A comparison of the spectra of Pristine_221.8781+9.7844 (black) and J102915+172927 (red), focusing in particular on the regions of wavelengths in which the iron (Fe) and magnesium (Mg) absorption lines appear. Both stars have similar Fe lines, but Pristine_221.8781+9.7844 has much stronger Mg lines, implying a higher abundance of magnesium.

The paper concludes by stating that Pristine_221.8781+9.7844 is, indeed, an ultra metal-poor star, with an estimated metallicity of [Fe/H] = – 4.66 ± 0.13 and an unusually low carbon abundance (they give an upper limit of A(C) = 5.6, but at no point do they explain what that means). While the [Fe/H] value is impressively low on its own, it’s the low carbon abundance that makes this star special, as stars with a similar [Fe/H] tend to have high carbon abundances to compensate. The existence of a star with low iron and carbon abundances implies that there must have been several different channels through which long-lived, low-mass stars could have formed in the early Universe.