Astron. Astrophys. 327, 890-900 (1997)

2. Observations

2.1. Discovery and IUE observations

The bright high redshift QSO HE 2347-4342 was discovered as part of the Hamburg/ESO Survey for bright QSOs a wide-angle survey based on objective-prism plates taken with the ESO 1 m Schmidt telescope. The plates are digitized in Hamburg and automatically searched for QSO candidates which are subsequently observed spectroscopically with ESO telescopes (Reimers 1990; Wisotzki et al. 1996; Reimers et al. 1996). HE 2347-4342 was confirmed as an extremely bright (), high-redshift () QSO in an observing run at the ESO 2.2 m telescope in October 1995. The coordinates are 23^h 50^m 34.3^s (2000). The optical spectrum, spectral resolution Å, is shown in Fig. 1. Because of the absence of damped Ly lines and Lyman limit systems, HE 2347-4342 was immediately recognized as a candidate suitable for UV follow-up observations.

Fig. 1. Combined low-resolution optical spectrum observed with the ESO 2.2 m telescope and ultraviolet spectra obtained with the FOS and GHRS onboard the HST. Flux is given in units of 10-15 erg s-1 cm-2 Å-1. The feature at 2800 Å is an artefact.

Another low resolution spectrum (20 Å) was taken in October 1996 with the ESO 1.5 m telescope covering the wavelength range 3200-9000 Å with the aim to obtain an improved QSO redshift. We find 0.005 from the C III and C IV lines. However, we find a higher redshift of 0.005 in both low- and high-resolution spectra of the O I line. Since it is well known that the higher ionization lines underestimate the intrinsic QSO redshift, we shall use 2.885 hereafter.

HE 2347-4342 was observed twice with the Short Wavelength Prime (SWP) camera onboard IUE (SWP 56218, 580^m, 56228, 712^m). The QSO was detected in both images, although with very low signal-to-noise ratio at the 1-2 10^-15 erg cm^-2 s^-1 Å^-1 level.

With a flux more than a factor of 10 higher at the expected wavelength of the redshifted He II 304 Å line than in Q 0302-003 and PKS 1935-692 (cf. Jakobsen 1996), HE 2347-4342 offered the chance to observe the He II Ly forest with the GHRS and the hope of resolving the He II forest.

2.2. UV observations with HST

The ultraviolet spectra have been taken in three visits between June 7 and June 14, 1996 with both the FOS in its high-resolution mode (R=1300) and the GHRS in its low resolution mode. The log of observations is given in Table 1. The standard pipeline processing provided flux calibrated data together with the 1 error in the flux of each pixel as a function of wavelength. The maximum signal-to-noise ratio achieved for the GHRS observation is 14. With the GHRS first-order grating the background is mainly due to counts from particle radiation. Since the background is usually very low the average over all diodes is subtracted from the science data. Inspecting the raw science and background data we find a mean background level of 1-2 % of the mean quasar countrate.

Table 1. The log of HST observations

The overall spectral energy distribution is shown in Fig. 1, together with a low resolution optical spectrum.

The strong break in the spectrum near 3500 Å is due to a Lyman limit system at with an optical depth from which the flux recovers () and rises to a maximum continuum flux at Å of erg s^-1 cm^-2 Å^-1 just longward of the He II break. HE 2347-4342 is nearly a factor of 2 brighter than HS 1700+6416 (Reimers et al. 1992) at the shortest wavelengths making it the most UV-bright high redshift quasar discovered so far.

In this paper we discuss only the strong He II absorption shortward of 1186 Å. A detailed analysis of the rich QSO absorption line spectrum will be deferred to a future paper in which also the high-resolution optical spectra will be included.

Fig. 2 shows the relevant part of the GHRS spectrum. The absolute wavelength scale was checked using the strong interstellar lines Si II 1260 and C II 1335. We applied a wavelength zero point offset of -0.15 Å to shift the interstellar absorption lines to their rest wavelengths.

Fig. 2. Section of the GHRS spectrum and error spectrum (dotted curve) of HE 2347-4342 with the flux given in 10-15 erg s-1 cm-2 Å-1. The expected position of the He II 303.78 Å edge for a QSO redshift of is indicated by the vertical dashed line.

The first question we ask is whether the observed strong He II 303.78 Å edge is at the expected position. With an observed QSO redshift from the O I 1302 emission line in both low and high resolution spectra of , we expect the He II edge at 1180.2 1.5 Å. The observed break is Å to the red of this.

The explanation for this discrepancy is the presence of a strong multicomponent associated () absorption system with redshifts 2.8911, 2.8917, 2.8972, 2.8977, 2.8985, 2.8989, 2.9023, 2.9028 and 2.9041. This associated system is extremely strong in H I and in O VI and strong in N V, C IV and O V (cf. Fig. 3). In He II 303.78 two absorption complexes can be seen, the broad one ranging from 1184 to 1187 Å and the unresolved line pair at 1182 Å (Fig. 4). The former is not saturated in He II, in contrast to H I and O VI, which means that He II is highly ionized close to the QSO. The apparent He II edge at 1186.5 Å and the absorption between 1182 and 1186.5 Å is exclusively due to the partially resolved He II lines of the associated system. Besides the highly redshifted associated systems with to 2.904 (up to 1500 km s^-1 relative to the QSO at ) there are two further systems with highly ionized species: (N V and C IV, unsaturated Ly ) and 2.863 (O VI, weak C IV, no N V, saturated Ly ). The latter again appears to show an unsaturated He II 303.8 Å line, like the redshifted associated system.

Fig. 3. Velocity profiles of absorption lines arising in the associated system as seen in the normalized high-resolution optical data. The velocity km s-1 corresponds to a QSO redshift . At least 14 individual components can be identified by heavy element and/or H I Lyman series absorption lines. The complex associated system is responsible for the observed He II 303 Å absorption in HST data (cf. top panel) redward of the expected He II 303 Å edge at 1180 1.5 Å (corresponding to v km s-1).

Fig. 4. He II 304 Å forest as seen in the normalized GHRS spectrum overlaid with the corresponding H I Ly forest in the normalized high-resolution optical data (dotted curve) scaled in wavelength according to 303.78/1215.67. The expected position of the He II 303.78 Å edge according to a QSO redshift of is indicated by the dashed lines.

In summary, if the associated He II absorption is taken into account, the true observed He II edge is estimated to be around , cf. Fig. 3, close to the QSO emission redshift .

2.2.1. Continuum definition

The GHRS data were first corrected for interstellar reddening according to Seaton's law (Seaton 1979) with , corresponding to N(H I)= 2.01 10 cm^-2 (Stark et al. 1992). A local continuum definition is difficult due to the high absorption line density. We searched for regions in the data apparently free of absorption lines, where we calculated the mean flux and the error in the mean flux to check for consistency with the noise. The continuum for the GHRS data was then constructed by fitting a low-order polynomial to these mean flux values. As is obvious from the GHRS data longward of the He II edge there are still numerous absorption lines presumable from heavy elements, since the number of Ly clouds expected in this range is small at these low redshifts (Bahcall et al. 1993). Thus the derived continuum level might be underestimated due to unresolved line blending. This effect, however, should not influence the analysis of the He II absorption since it is expected to be of the same amount longward and shortward of the He II edge.

2.3. Optical observations and data reduction

HE 2347-4342 was observed on two nights (4-6 October 1996) using CASPEC with its Long Camera on the ESO 3.6 m telescope at La Silla. We obtained three individual 2h45m exposures in the spectral range from 3550 to 4830 Å and a single 2h45m exposure in the range from 4870 to 6180 Å. The slit was aligned with the parallactic angle in order to minimize light losses due to atmospheric dispersion. The data reduction was done with the ECHELLE software package available in MIDAS, supplemented by own software programmes for an optimal extraction of the echelle orders (kindly provided by S. Lopez).

For wavelength calibration, a Th-Ar comparison spectrum was obtained immediately before and after each QSO observation. For each order the wavelength scale was determined by fitting a third-order polynomial to the automatically identified lines. The resulting rms residuals were 0.003/0.005 Å for the two spectral regions. The individual observations were rebinned to the same wavelength scale and each observation was scaled by its median and then coadded, weighting by the inverse variance. After correction for the blaze function using observations of the standard star Columbae, the continuum for each order was determined by fitting low-order polynomials to regions free of absorption lines. Wavelengths have been corrected to vacuum, heliocentric values. The final resolution achieved is R=21 500 with a signal-to-noise ratio of 34 per pixel at =4700 Å. For the single exposure longward of 4870 Å we reached R=24 500 and a S/N=14.

For a quantitative analysis of the He II edge we need to identify the corresponding H I Ly clouds in the optical data. All absorption lines were fitted with Voigt profiles convolved with the instrumental profile using the software package FITLYMAN available in MIDAS to derive z, and b values. For the Ly absorber clouds responsible for the observed part of the He II edge the optical data cover also Ly , Ly and Ly absorption lines allowing either an independent derivation of z, and b or at least a consistency check with the fit results for Ly . With the available optical data we can detect Ly absorption lines by neutral hydrogen down to column densities log .

© European Southern Observatory (ESO) 1997

Online publication: April 6, 1998
helpdesk.link@springer.de