Observations of the 2142-4420 Cluster

The 2142-4420 cluster lies at redshift 2.38, at coordinates 21:42:30-44:20:30 (J2000). As will be shown, it is a region of the early universe that is highly overdense in Ly $\alpha$ emitting galaxies. It is unlikely to be gravitationally bound, and so would not meet most low redshift definitions of a galaxy cluster.

Francis & Hewett originally identified the cluster as a pair of strong Lyman limit systems at matching redshifts (2.38) in two 19th magnitude QSOs: 2138-4427 and 2139-4434. The QSO sight-lines are 500 $^{\prime \prime}$ ( $\sim 4$ proper Mpc) apart at z=2.38 (both QSOs lie at $z \sim 3.2$ ). Such a pair of Lyman-limit systems at matching wavelengths is unlikely to occur by chance in Francis & Hewett's sample. As we report in Section 2.2, a third QSO has been found behind the cluster, and it too shows strong Ly $\alpha$ absorption at the cluster redshift.

Are there any galaxies associated with this concentration of QSO absorption lines? Many high redshift galaxies show weak Ly $\alpha$ emission (Steidel et al. 1996, Hu, Cowie & McMahon 1998). The field containing the three QSOs, however, contains three strong Ly $\alpha$ emitting galaxies at z=2.38, with Ly $\alpha$ fluxes $F_{\rm Ly \alpha} > 10^{-16}{\rm erg\ cm}^{-2}{\rm s}^{-1}$ ( $> 10^{-19}{\rm W\ m}^{-2}$ , Francis et al. 1996, Francis, Woodgate & Danks 1997); an order of magnitude greater than that of normal galaxies at these redshifts. All three have reliable spectroscopic redshifts.

Do three such sources constitute a cluster? Strong Ly $\alpha$ emitters are rare at these redshifts: Francis et al. surveyed a total volume of 460 co-moving cubic Mpc, but all three sources were found within a 5 cubic co-moving Mpc volume: ie. $\sim$ 1% of the surveyed co-moving volume. Martínez-González et al. (1995) surveyed a co-moving volume of 1400 cubic Mpc at z=3.4 for Ly $\alpha$ emitting sources to a comparable flux limit but detected nothing. The odds of the cluster being an artifact of the coincidental proximity of three such sources is thus < (5/1860)² (the probability of finding two more such objects within one proper Mpc of the first), ie. < 10^-5. Clearly the space density of Ly $\alpha$ emitting galaxies in this region is higher than average.

Could the presence of absorption in all three QSOs be coincidental, or is the cluster really surrounded by Lyman-limit absorption-line systems? The transverse separation of the QSO sight-lines is $\sim 5$ Mpc, which corresponds to a redshift difference of $\sim 20$ Å along the line of sight. The absorption-line systems all have equivalent widths of > 20Å. The probability of seeing an absorption line with an equivalent width this strong within $\pm 20$ Å of any given wavelength is 1.3% (Francis & Hewett 1993). Thus the probability of finding three such absorption-line systems within the cluster by chance is 2 x 10^-6.

This calculation should be regarded with caution: this region was first identified as interesting because of the absorption in the two original QSOs (Francis & Hewett 1993), so the statistics are a posteriori. The third QSO, however, was not involved in selecting this region for study, and its coincident absorption alone makes this region overdense in Lyman-limit systems with 98% confidence.

**Figure 1:** 3D view of the 2142-4420 cluster. All objects have been assigned 3-dimensional positions, based on their location on the sky and their redshift, assuming that all redshifts trace the Hubble Flow. Solid lines are the sight-lines to the three background QSOs. Stars are Ly-limit absorption systems, crosses are the lower column density absorption systems, and solid triangles are the Ly $\alpha$ emitting galaxies. Redshifts increase upwards: galaxy redshifts have been derived, where possible, from the metal emission lines. From left to right, the QSO sight lines are 2139-4434, 2139-4433 and 2138-4427.
$\begin{figure} \psfig{file=f1.eps,height=100mm}\par\end{figure}$

Fig. 1 should make the geometry of the cluster clearer. Due to the low predicted overdensities of clusters at this redshift (Section 3.2), peculiar motions should be very small, so three dimensional positions are plotted assuming that all redshift differences are due to distance. The three Ly $\alpha$ galaxies lie within one Mpc of each other. The absorption-line systems are far more dispersed, extending both to lower redshifts and transversely by $\sim 5$ Mpc.

The spatial extent and overdensity of this cluster are comparable to those of the clusters of Lyman-break galaxies being found by Steidel et al (1998) at 3 < z < 3.5: this cluster may be a representative of the same class of object.

Observations

Véron & Hawkins (1995) searched an area including this cluster for variable sources. In addition to both previously identified QSOs, they discovered a third QSO lying between the two: QSO 2139-4433 at z=3.22 (ie. at the same redshift as the other two background QSOs). We measured a position for QSO 2139-4433 (21:42:22.16-44:19:28.7, J2000) using our R-band image with an astrometric solution bootstrapped from on-line scans of UK Schmidt plates (Drinkwater, Barnes & Ellison 1995). A spectrum was obtained with the Low Dispersion Survey Spectrograph (LDSS, Colless et al. 1990) on the Anglo-Australian Telescope on the nights of 1996 August 13 and 14. The total exposure time was 47,700 sec, and the spectral resolution 700 ${\rm km\ s}^{-1}$ . Part of the spectrum is shown in Fig 2.

**Figure 2:** Spectra of the three background QSOs, showing the Ly $\alpha$ absorption at the cluster redshift (4108 Å). The top and bottom spectra are the original AAT spectra: the middle panel is the new LDSS spectrum.
$\begin{figure} \psfig{file=f2.eps,height=100mm}\end{figure}$

As Fig 2 shows, the new QSO 2139-4433 has a strong absorption-line system close in wavelength to the absorption in the two previously known QSOs at z=2.38. This further confirms the remarkable gas properties of this cluster.

Our original spectra of QSOs LBQS 2138-4427 and 2139-4434 are described by Francis & Hewett (1993). Their resolution was excellent (full width at half maximum height $100 {\rm km\ s}^{-1}$ ) but the wavelength coverage (4000 - 4600 Å) was small. An additional spectrum of QSO 2139-4434 was obtained with the KPGL1 grating in the Blue Air camera of the RC spectrograph on the CTIO 4-m telescope on 1995 August 20. Total exposure time was 12,000 sec, with a spectral resolution of 200 ${\rm km\ s}^{-1}$ . This spectrum, while inferior in resolution to the spectrum of Francis & Hewett, covers 3200 - 6200 Å: this greater wavelength range allows us to study CIV and Ly $\beta$ absorption from the cluster.

**Figure 3:** Voigt profile fits to the Ly $\alpha$ absorption at the cluster redshift in QSOs 2138-4427 and 2139-4434. Fits are shown for the two different velocity dispersions b assumed.
$\begin{figure} \psfig{file=f3.eps,height=100mm}\psfig{file=f4.eps,height=100mm}\end{figure}$

Absorption Line Measurements

Combining the old and new data on the two brighter QSOs, we fit Voigt profiles interactively to the absorption at the cluster redshift, using the Xvoigt program (Mar & Bailey 1995). The low spectral resolution, restricted wavelength coverage and blending in our spectra make this process a difficult and ambiguous one. Nonetheless, certain definite conclusions can be reached. Multiple components are required to obtain adequate fits to the Ly $\alpha$ absorption (Fig 3). A minimum of 2-4 components are required (Tables 1, 2): many more, each with smaller column densities, give equally good fits. The column densities of the subsidiary systems are not well constrained. We could not determine the velocity dispersion b of the metal lines: an upper limit of $\sim 100 {\rm km\ s}^{-1}$ can be placed. The central component of the Ly $\alpha$ absorption in all three QSOs was broader: the flux touches zero over $\sim 100 {\rm km\ s}^{-1}$ or more.

**Table 1:** Possible Absorption Systems in QSO 2138-4427
			Column density $\log(N[cm^{-2}])$
System	Ion	Redshift	$b=50 {\rm km\ s}^{-1}$	$b=100 {\rm km\ s}^{-1}$

A	H I 121.6	2.3825	20.47	20.47
	Si II 130.4	2.3823	14.63	14.72
	Si II 126.0	2.3825	14.63	14.22
	Si III 120.6	2.3824	14.30	13.98
	O I 130.2	2.3820	16.22	15.47
	C I 127.7	2.3821	14.62	14.70
	C II 133.4	2.3822	16.00	15.13
B	H I 121.6	2.3731	13.97	14.10

**Table 2:** Possible Absorption Systems in QSO 2139-4434
			Column density $\log(N[cm^{-2}])$
System	Ion	Redshift	$b=50 {\rm km\ s}^{-1}$	$b=100 {\rm km\ s}^{-1}$

A	H I 121.6	2.3792	19.80	19.67
	Si II 130.4	2.3794	13.63	13.80
	Si II 126.0	2.3792	13.53	13.67
	Si III 120.6	2.3789	13.45	13.55
	O I 130.2	2.3804	16.00	15.47
	C I 127.7	2.3790	14.00	14.17
	C II 133.4	2.3783	16.03	15.13
	C IV 154.8	2.3787	14.30	14.23
	C IV 155.1	2.3787	14.17	14.43
B	H I 121.6	2.3890	14.56	14.37
C	H I 121.6	2.3865	14.07	14.07
D	H I 121.6	2.3724	16.93	14.80

We searched for metal-line absorption at the redshift of the dominant Ly $\alpha$ absorption components. With the exception of C IV, these all lie within the Ly $\alpha$ forest, and hence may be chance coincidences with the forest lines. The strongest line near the expected wavelength was fit, assuming velocity widths of 50 and $100 {\rm km\ s}^{-1}$ , and the results are shown in Tables 1 and 2. Due to the risk of blending or confusion with Ly $\alpha$ forest lines, the metal line column densities should be taken as upper limits. Plots of the metal lines within the forest can be found in Francis & Hewett. Note that in QSO 2138-4427, strong absorption lines were invariably detected at the expected wavelengths, while in QSO 2139-4434 the lines were weaker and at slightly shifted wavelengths. We conclude that the central absorption component in QSO 2138-4427 does contain metals, roughly as measured, while for QSO 2139-4434 some or all of the putative lines (except C IV) may be misidentified Ly $\alpha$ forest lines. Our spectrum of QSO 2139-4433 had too low a resolution to determine anything other than the Ly $\alpha$ redshift (2.366) and column density ( $\log(N_H) \sim 20.7$ ).

Are the absorption-line systems really Lyman-limit systems, or could they just be clusters of lower column density Ly $\alpha$ forest lines? In QSO 2138-4427 the Ly $\alpha$ line shows broad wings, and strong absorption is seen at the expected wavelength of most common metal absorption lines: it therefore seems probable that this is, as modelled, a high column density absorption system, probably lying on the column density borderline between Lyman-limit and damped Ly $\alpha$ systems. The spectrum of QSO 2139-4433 is of too low resolution to say much, but the great width and equivalent width of the Ly $\alpha$ absorption also suggest that its absorption column is large.

In QSO 2139-4434, however, the situation is more ambiguous. It is possible to fit the Ly $\alpha$ absorption either with a single absorption-line system with column density $N_H \sim 10^{19.7}{\rm cm}^{-2}$ (plus three much weaker components in the wings), or with a blend of weaker Ly $\alpha$ lines, spread over $\sim 200 {\rm km\ s}^{-1}$ and with a combined neutral hydrogen column density that can be as low as $N_H \sim 10^{16.5}{\rm cm}^{-2}$ . Two pieces of evidence support this latter fit. Firstly, the redshifts of the supposed metal-lines vary by $\pm 80 {\rm km \ s}^{-1}$ (though some or all may be chance coincidences with Ly $\alpha$ forest lines). This can be explained if they are coming from different subcomponents of the absorption system. Secondly, there is tentative evidence that Ly $\beta$ absorption is weak: the spectrum is poor at this wavelength, and the continuum hard to define, but the Ly $\beta$ absorption can be well fit with column densities as low as $N_H \sim 10^{16}$ (though much greater columns also give acceptable fits). On the other hand, the strength of the metal lines, especially low ionisation lines such as C II, imply that the neutral hydrogen column density is $N_H > 10^{18}{\rm cm}^{-2}$ . Note, however, that with the exception of C IV, these lines could be contaminated by Ly $\alpha$ forest absorption.

Note that the total gas column density in the form of the absorbing clouds is almost independent of the interpretation of the data. Gas with a neutral column density of $\sim 10^{19}{\rm cm}^{-2}$ is predicted to be mostly neutral and hence to have a total gas column density $\sim 10^{19}{\rm cm}^{-2}$ . Gas with a neutral column density of $\sim 10^{16}{\rm cm}^{-2}$ , on the other hand, is predicted to be strongly ionised by the UV background, and hence its total gaseous column density will be $\sim \times 10^3$ greater than the neutral column density. Thus the total hydrogen column would be roughly the same as for the damped Ly $\alpha$ interpretation.

Observations of the 2142-4420 Cluster

Review of the 2142-4420 Cluster Properties

Observations

Absorption Line Measurements