Ne and Fe abundances in the ISM: Archival Study of Fe-L and Ne-K edges in Chandra and XMM-Newton

TL;DR

This study uses archival XMM-Newton and Chandra HETG soft X-ray spectra of luminous LMXBs to measure Milky Way ISM Fe and Ne abundances from Fe-L and Ne-K absorption features. By modeling the spectra with TBVarabs, ISMabs, and dust scattering (xscat) and by exploring Fe depletion into grains, the authors derive robust relative abundances: the fiducial result is [Fe/Ne] = $-0.523 \,\pm\,0.025$, [Fe/H] + 12 = $7.482 \,\pm\,0.016$, and [Ne/H] + 12 = $8.012 \,\pm\,0.022$, consistent with literature while providing tighter constraints. They demonstrate that the abundance measurements are sensitive to the assumed Fe depletion (about ~5%) and to the inclusion of scattering (≈1–7%), and they validate uncertainties with jackknife tests and cross-model comparisons. Overall, the work establishes the strongest observational constraints to date on Fe and Ne abundances in the local ISM and highlights the importance of depletion and scattering in X-ray edge analyses, reinforcing the utility of X-ray spectroscopy for studying ISM composition and dust processes.

Abstract

The abundance of elements in the interstellar medium (ISM) is a key facet for many fields of astrophysical study. In the soft X-ray spectra, absorption by interstellar gas can result in deep absorption features that affect continuum measurements. In this paper, we focus on measuring the abundance of interstellar iron and neon from the column densities observed in soft spectra from XMM-Newton and Chandra for various low mass X-ray binaries (LMXBs), which allows for a direct probe of elemental abundances. As a noble gas, neon will not deplete into solid form, thus providing a benchmark with abundances determined via UV spectroscopy. We find that, when assuming Fe is 90\% depleted into grains, [Fe/Ne]$ = -0.523\pm0.025$, [Fe/H]$ + 12 = 7.482\pm0.016$, and [Ne/H]$ + 12 = 8.012\pm0.022$, which are the tightest observational constraints on these abundances to date, while being consistent with literature which uses protosolar abundances. We also test how depletion into solid grains and scattering affect the results. The choice of depletion fraction can affect the abundance measurement by roughly $5\%$, and that the inclusion of a scattering component can affect abundance measurements by $\sim1-7\%$.

Figures (4)

• Figure 1: A selection of both XMM and Chandra Fe-L edges with varying column densities. These plots are constructed by fitting the data, then evaluating the model with the Fe abundance set to 0 and plotting the resulting ratio of the data to the model. From left to right, these span a low ($N_H = 1.90\times10^{21}$ cm$^{-2}$), moderate ($N_H = 3.01\times10^{21}$ cm$^{-2}$), and high ($N_H = 4.40\times10^{21}$ cm$^{-2}$) column density, as reported by the HEASARC HI4PI tool. this demonstrates how the depth of the edges vary as the column density increases.
• Figure 2: The abundance of Fe compared to Ne (left) and H (center). We also compare Ne to H (right). Black points represent XMM-Newton observations, and red points represent Chandra observations. The data points shown on this graph are only for the fiducial Fe depletion of 0.1. The fit to that data set is shown as a solid black line, with the gray shaded region being the uncertainty. The green and blue dashed lines represent fits to data sets with Fe depletion of 0 and 1, respectively. We do not plot the data sets for these depletion factors, nor do we plot the uncertainties in their fit to maintain readability. The uncertainties reported in this plot are only those from the fit, and do not include additional uncertainty from statistical comparison of models.m The low column source XTE J1118+480 provides only an upper limit on ${\rm N_{Ne}}$
• Figure 3: Here we compare our fit results to those measured in other literature. In order, these points refer to (a) anders89, (b) grevesse98, (c) lodders03, (d) lodders09, (e) asplund09, (f)asplund21 , (g) nieva12, (h) psaradaki24, (i) gatuzz15, (j) gatuzz16, and (k) juett06 Rows (l), (m), and (n) are the results of this work for $1 - \beta_{{\rm Fe}}=0,0.1,$ and 1, respectively. The vertical line is the abundance put forward by wilms00. The error bars displayed here represent a conservative estimate on the uncertainty, which includes the discrepancy between models with and without scattering.
• Figure 4: (Top) The 0.68-0.74 keV region surrounding the Fe-L edge for 4U 1636-53 (XMM Observation 0500350401). (Bottom) The data divided by the fit model, which demonstrate the slight variations between the various models. The leftmost panel demonstrates the fit with no scattering or shielding due to depletion. The center panel is identical, but the depletion fractions are set to physically realistic values. The right panel shows the fit with both scattering and shielding due to depletion. Visually, it appears that scattering improves the fit on the low energy side of the edge, and shielding improves the fit surrounding some of the structures on the high energy end of the fit. This is perhaps more apparent in the residuals along the bottom column, which demonstrates the pre-edge region improving with a scattering component, followed by an underestimate of the edge depth.