Updated: 6 March 2011
Conference presentation (pdf) from 213th AAS Meeting, Long Beach, California, January 4−8, 2009
I am only going to talk briefly about the metals-in-galaxies (and IGM/halo) pilot study at the end. The majority of the talk is about the IGM.
This is work I do with my advisor Xavier; my collaborators at Chicago−Chris Thom, who's in the audience, and Hsiao-Wen Chen; and Jenny Graves, a fellow grad student at UCSC and my galaxy guru.
The science driver is one we can all appreciate: enrichment of the IGM over cosmic time. We are addressing this topic in two regimes: the IGM, via a C IV survey, and in the IGM/halo within 150 kpc of galaxies. These are my punch-line plots to-date, and at the end of the talk you will appreciate them more.
High-redshift surveys have found that the C IV mass density Ω(C IV) has not evolved from redshifts z = 5.5 to z = 1.5. A recent pre-print by Becker et al (2008) places an upper limit on the C IV mass density at z = 5.7 that is a factor of two smaller than the C IV mass density at 1.5 ≤ z ≤ 5.5. This is an indication of evolution at high redshift. The name of the game now is to determine what the C IV mass density is at low redshift and to compare. With our study, there have been three measurements. Frye et al (2003), a conference proceeding, and Danforth and Shull (2008), a refereed publication, are two surveys of the STIS E140M spectra, Danforth and Shull being the larger survey that encompasses the Frye et al (2003) data set. The current study encompasses all the Danforth and Shull data.
For the galaxy pilot study, the name of the game is to compare the abundances in the stars in early-type galaxies that have known intergalactic absorbers within 150 kpc. In the plot, you can see that in one system, we're seeing the same [C/N] abundance in the galaxy as in the IGM (or halo).
To compile the spectra, we searched the HST STIS and Goddard High Resolution Spectrograph (GHRS) for any quasar, QSO, AGN, Seyfert, or object that has been used for an IGM study. We found 69 objects, only 48 of which had high enough S/N and resolution and C IV wavelength coverage. The number of spectra that cover each redshift bin is shown in the histogram in the upper right. The STIS E140M spectra covers the range about z ≤ 0.1. The STIS E230M spectra covers the range about 0.4 < z < 1, though with different tilts it can cover lower redshifts. This and medium-resolution gratings and GHRS data fill in the in-between redshift range.
We conducted a blind survey for C IV, relying on the characteristics of the C IV doublet: characteristic separation and equivalent width ratio, profile shape, and any other lines like Lyα that may be part of the system, though, emphasizing again, we first search for C IV. The ultimate goal is to have a C IV catalog with both lines of the doublet detected at ≥ 3 sigma in equivalent width. We visually inspected the C IV candidates. We determine we definitely have 69 C IV systems, 42 of which are detected at ≥ 3 sigma and 27 of which are un-saturated in at least one line of the doublet. We measure column densities by the apparent optical depth method, which yields only lower limits if a line is saturated. So when we analyze the column density frequency distribution and sum the mass density, we use the un-saturated sample.
To push the limits of the data, we also define a "likely" group, these 19 C IV doublets are weaker with non-descript profiles and no coverage of other diagnostic lines, such as Lyα or Si IV. Only five of them are ≥ 3 sigma and un-saturated. The redshift distribution is shown by the hash marks in the top part of the top plot. The upper-most row show the un-saturated "definite" C IV systems; the second row show the saturated "definite" C IV systems; and the grey hashes are the "likely" C IV systems.
We repeat analyses including the "likely" group in order to see how the weaker systems affect fits, since they are more numerous and affect the frequency distribution fits. For example, shown in the bottom plot is the cumulative dN/dX for equivalent widths greater than the lower axis, where X is like redshift. The weakest system we have is 52 mÅ (C IV 1548), and dN/dX = 4.2 at that equivalent width. The fits to the equivalent width frequency distributions of the two C IV groupings are shown as the dashed lines, with which we can extrapolate dN/dX to lower equivalent width limits. Shown in the blue stars is the redshift densities dN/dz from Frye et al (2003); their survey covers such low redshift that the redshift and the pathlength X are nearly identical. Shown in green is the measured dN/dz for Danforth and Shull (2008), which is cited for a lower equivalent width limit that we probe.
The next four slides show actual data, because everyone loves data.
We recover C IV systems also found by Frye et al and Danforth and Shull. Since I'm going to show several velocity plots, let me take a minute to orient everyone. Ignore the bottom panel for now but note the axis. For a velocity plot, the regions of spectra around each absorption line are aligned based on the rest wavelength and the redshift of C IV 1548. The bottom panel shows the apparent optical depth profiles, which we use to objectively rate the match between the C IV doublet line profiles. Also, at the top is indicated whether the C IV doublet is in the Lyα forest.
To emphasize again, we conducted a blind search for C IV and the other lines came along for the ride.
Milutinovic et al (2007) also conducted a low-redshift survey for C IV in the lines of sight towards seven quasars with STIS E230M spectra. We also recover systems that he found.
We are the largest survey for C IV at low redshift. We include lines of sight not examined in any of the other aforementioned studies. So we of course find C IV systems that have not been published.
For my own mental well-being, I am gratified to find that including the GHRS spectra has been fruitful. These two systems are known systems with published analyses (see Chris Thom's poster downstairs). Unfortunately, the 1550 line is not detected at 3 sigma.
As you may have noticed, several of these example C IV systems had C IV doublet in the Lyα forest. For those systems, the detection of other associated lines such as Lyα or Si IV lends credibility to the C IV identification. For some of our C IV doublets, we do not have coverage of other lines; for example, this doublet in MARK132. The profile is non-descript, so that diagnostic is not definitive.
Instead of excluding such systems, we account for the rate at which coincident Lyα forest lines will be included as C IV doublets. Chris Thom spear-headed the simulations of the Lyα forest in synthetic E230M spectra. As you can see on the right, the rate that coincident Lyα lines are recovered as C IV is a function of signal-to-noise. So, for example, the redshift density of Lyα contamination is about 0.25, which is about 6% of the redshift density (rather, the dN/dX) for our C IV sample, which was 4.2.
We fit the column density frequency distribution with a power-law function, and we fit to the coefficient and exponent simultaneously. The fits to the two groupings of our C IV sample result in almost the same exponent (see the top plot). We also divide the sample into two redshift bins z < 0.6 and 0.6 < z < 1. We find that our exponent from the fit to the lower-redshift bin is consistent with the Danforth and Shull (2008) exponent. However, the higher redshift bin has a rather shallow slope compared to high-redshift studies.
Now we return to the punch-line plot I showed earlier. Instead of just one measurement of the C IV mass density, I break up the value into the two redshift bins. The stars are the summed mass density based on the 32 un-saturated C IV absorbers. The black bars show the range allowed by varying the sample and the integration limits for the fit.
More important to emphasize is the time probed by the low-redshift surveys; it's the majority of cosmic time. And our measured C IV mass density is consistent with high-redshift studies. This informs our understanding of the ionizing background, so that maybe more of the intergalactic metals are not in C IV, or about the production of carbon.
Moving on briefly to the galaxy-IGM pilot study. We are examining the galaxies close, within 150 kpc and 500 km/s, to published strong H I absorbers. We're first focusing on the early-type galaxies, where it's surprising to detect any what might be called "halo" gas. We use CLOUDY to measure the abundances in the intergalactic absorbers and EZ_Ages, by Jenny Graves, to measure the abundances in the stars in the galaxies.
The image shows our wider field and how we're focusing on the inner-most region. FJ2155 is also known as PHL1811.
We've only made three measurements of the abundances in the galaxies so far. In the PG1116 system, the [C/N] abundance of the stars is greater than the [C/N] abundance of the IGM absorbers. Perhaps this is due to the intergalactic H I absorber is not very strong and/or because the galaxy is farther away.
In the other system, there are two galaxies close to the FJ2155 sightline. In this system, the [C/N] abundances of the stars in the galaxies and the intergalactic Lyman limit system are equal. Maybe the stars formed and ejected the intergalactic absorber concurrently and not enriched the gas since, Or perhaps some of the intergalactic gas collapsed into the galaxy to trigger the formation of stars and no enrichment has occurred since.
We're going to make more measurements and see if we can tease out any trends.
In summary, we are seeing little to no evolution in the C IV mass density at low redshift; the mass density is consistent with high-redshift studies.
Also, for one case, we see the [C/N] abundance in the stars in the close galaxy (or galaxies) equal the [C/N]1 abundance in the intergalactic absorber.