The clustering and X-ray properties of faint galaxies

Abstract

The clustering properties of faint (23 ≤ B ≤ 27) galaxies were investigated, with the aim of obtaining information about their redshift distribution, N(z), and therefore about the processes of galaxy evolution causing the steep rise in the galaxy number counts at B > 23. We calculated the angular correlation function, w(θ), for galaxies detected on a CCD survey reaching a blue magnitude of B(_eed) = 25.0 and a red magnitude of R(_eed) = 23.5, and on a smaller area reaching a very faint limit of B(_eed) = 27.0. The w(θ) amplitude of galaxies was found to decrease much more steeply between magnitude limits of B(_eed) = 23 and B(_eed) = 25 than predicted by models in which galaxy clustering is stable in proper co-ordinates and the galaxy redshift distribution possesses a no-evolution form. At a B = 24.5 limit the w(θ) amplitude was found to be only ~ (^1)(_4) that expected from a non-evolving model, enabling us to reject such a model at the 4σ level. The red-limited sample, R(_eed) = 23.5, also gave a w(θ) amplitude lower than the non-evolving prediction, although the difference was less significant (~ 2σ). A pure luminosity evolution model, which accounts for the excess in the galaxy number counts by enabling galaxies undergoing rapid star-formation at 1 ≤ z ≤ 4 to become visible at B > 23.5, gave a good fit to the w (θ) results without requiring any departure from stable clustering. The B(_eed) ≤ 24.5 galaxies with redder (B - R ≥ 1.5) colours gave a significantly higher w(θ) amplitude than the bluer galaxies at the same limit, consistent with a stable clustering model with no luminosity evolution. As these red galaxies would lie at z < 1, within an approximately non-evolving N(z), this result suggests that galaxy clustering is indeed stable out to at least z ~ 0.5. The low uj{9) amplitude of the B(_eed) ≤ 24.5 sample was found to be associated only with the bluer galaxies, supporting the PLE model interpretation of the drop in w(6) amplitude at B > 23.5 as being caused by the appearance of blue star-forming galaxies at high redshifts. If all the blue galaxies seen at B ~ 24.5 are instead at z < 1, they must belong to a separate population of dwarf starburst galaxies, which must be intrinsically clustered extremely weakly compared to normal galaxies. The combination of our w(θ) results at B(_eed) = 25.0 and B(_eed) = 27.0 limits with those obtained by Efstathiou et al. (1991) at B = 26 gave some indication that the steep drop in w(θ) amplitude at 23 ≤ B ≤ 25 levels out at 25 ≤ B ≤ 27. This would be expected if galaxies with L* luminosities are seen out to some maximum redshift at B ~ 25, so that N(z) becomes no more extended at fainter magnitudes. The amplitude of w(θ) at this limit would depend only on the value of the maximum redshift, on q(_0), and on the rate of evolution of clustering. The observed amplitude of w(θ) ~ 4 x 10(^-4)(deg)(^-0.8) would be consistent with 3 ≤ Z(_maz) ≤ 4, stable clustering and any value of Ω from 0 to 1. If galaxies are seen out to their maximum redshifts at B ~ 25, the gradient of the differential galaxy number counts at even fainter magnitudes would follow the faint end slope of the galaxy luminosity function. The galaxy number counts from our deep CCD survey are much higher at B > 24.5 than the predictions of q(_0) = 0.5 PLE models, and appear to rise more steeply at 25 ≤ B ≤ 27.5, with (^d(logN))(_dm) ~ 0.3, than even our q(_0) = 0.05 PLE model would predict. The observed number count gradient at B > 25, in combination with the levelling out of the w(θ) scaling, would then suggest that the luminosity function steepens to a ~ -1.75 at high redshifts, in addition to undergoing the strong brightening of the L* luminosity predicted by PLE models. We also investigated the X-ray properties of faint (18 ≤ B ≤ 23) galaxies, by cross- correlating their positions with those of X-ray sources (without identifications as either QSOs or stars) detected on deep ROSAT images. The results indicated that ~ 1% of all B ≤ 21 galaxies were detected above a 4σ threshold on the ROSAT X-ray images, with the cross-correlation between sources and galaxies being of 3σ significance. As the detected galaxies would possess L(_X)/L(_B) ratios higher than those typical of local galaxies by factors of ~ 100, these results suggest a wide dispersion of σ ~ 0.8 in the distribution of log (L(_X)/L(_B)) for galaxies. Spectroscopic investigation of the galaxies identified as probable X-ray sources indicated that they included both star-forming and early-type galaxies, with a mean redshift of z ~ 0.22, and were mainly of L ~ L* luminosity in the blue band, but possessed very high X-ray luminosities of L(_X) ~ 10(^42) ergs s(^-1) On one ROSAT image we also detected a rich galaxy cluster at a high redshift of z = 0.561, and estimated its X-ray luminosity as L(_X) - 5.4 X 10(^43) ergs s(^-1), which is only about ~ 30% of that expected for a cluster of the same richness seen locally. The number counts of galaxies in the 0.5-2.0 keV band were estimated as 27.18 ± 10.35 deg(^-2) at a flux limit of 10(^-14.4) ergs cm(^-2)s(^-1) The number counts of detected QSOs appear to level out at ~ 10(^-14) ergs cm(^-2)s(^-1) suggesting that QSOs faintward of our detection limits would produce only ~ 30% of the unresolved X-ray background (XRB).The autocorrelation function (ACF) of the unresolved 0.5-2.0 keV XRB was measured on 3 ROSAT images (totalling 1.05 deg(^2)), and was found to show no significant signal - we obtained a 2σ upper limit of w(θ) ≤ 1.4 x 10(^-3)(deg)(^-0.8) on its amplitude. This would be consistent with normally clustered galaxies as the origin of most of the XRB, if their X-ray luminosity increases with redshift sufficiently that a large proportion of the X-ray flux is produced by galaxies at z > 1. A very significant (> 5σ) cross-correlation was detected between 18 ≤ B ≤ 23 galaxies on A AT photographic plates and the fluctuations in the unresolved XRB on the same 3 ROSAT images. The cross-correlation amplitude, after applying corrections for the effects of galaxy clustering, indicated that these 18 ≤ B ≤ 23 galaxies produced 16.8 ± 2.1% of the unresolved 0.5-2.0 keV background, corresponding to a volume X-ray emissivity of p(0.5-2.0 keV) = (5.27 ± 0.65) x l0(^38) ergs s(^-1)Mpc(^-3). This is about an order of magnitude higher than the emissivity expected from the L(_X)/L(_B) ratio typical of normal galaxies, but a σ ~ 0.8 dispersion in log L(_X)/L(_B) ratios could account for this. The number counts of X-ray detected galaxies and the galaxy/XRB cross-correlation amplitude are both consistent with a slightly smaller dispersion in log (L(_X)/L(_B)) of σ = 0.7, if combined with an exponential increase with look-back time of the X-ray luminosity of galaxies, with a timescale of r = 0.4H(_o)(^-1). Extrapolation of our estimate of the galaxy X-ray emissivity to higher redshifts, predicts that non-evolving galaxies would account for ~ 35% of the unresolved 0.5-2.0 keV background, with most of the galaxy contribution produced at z < 1. However, with a r = 0.4H(_0)(^-1) rate of X-ray luminosity evolution, the predicted contribution of high-redshift (1 ≤ z ≤ 4) galaxies to the XRB is greatly increased, accounting for the lack of signal in the ACF. Galaxies out to z ~ 4 would then produce in total ~ 70% of the unresolved XRB, accounting for the entire non-QSO component. No part of this work has previously been submitted for a degree in any University. Some of the results described in this thesis have been published elsewhere in the following papers [brace not closed]