Next Contents Previous

4. RESOLVING THE STELLAR POPULATIONS IN A LENSED GALAXY

Gravitational lensing can be used as a tool to increase the resolution attainable in studies of distant galaxies. Although morphological information is hard to disentangle because of the geometric distortions and uncertainties in lens modelling, the amplification afforded by strong lensing can allow the stellar populations to be mapped on the sub-kpc scale, as well as magnifying the total flux.

4.1 Optical/Near-Infrared Imaging of Lensed Arcs at z = 4.04

Combining archival HST/WFPC 2 data with deep near-infrared imaging taken with Keck/NIRC (Matthews & Soifer 1994) in good seeing, we have measured the spatially-resolved colours in a z = 4.04 galaxy, gravitationally lensed by the rich cluster Abell 2390 (z approx 0.23) into a pair of highly-magnified near-linear arcs 3-5" in length (Frye & Broadhurst 1998). At the redshift of these arcs, the H (lambdacent approx1.65 µm) and k (lambdacent approx 2.2 µm) near-infrared pass-bands straddle the age-sensitive rest-frame 4000 Å+ Balmer break (Fig. 6). Comparison of the optical and near-infrared photometry with a suite of spectral evolutionary models (the latest version of Bruzual & Charlot 1993) has enabled us to map the underlying stellar populations and differential dust extinction (Bunker et al. 1998a). The WFPC2 images clearly reveal several knots, bright in the rest-ultraviolet, which correspond to sites of active star formation. However, there are considerable portions of the arcs are significantly redder, consistent with being observed > 100 Myr after star formation has ceased, with modest dust extinction of E(B - V) approx 0.1m. There is degeneracy in the models between dust reddening and age for the optical/near-infrared colours, but the most extreme scenario where the colour gradients are solely due to heavy dust reddening of an extremely young stellar population are strongly ruled out by upper limits in the far-infrared/sub-mm from ISO/SCUBA (Lémonon et al. 1998; Blain et al. 1999).

Figure 6a
Figure 6b

Figure 6. Top: An illustration of the unreddened rest-frame optical spectra of two galaxies, one observed only 3 Myr after the end of an instantaneous burst of star formation (long-dash curve) and the other seen after 400 Myr have elapsed (solid line). We also show the 3 Myr model with dust extinction of AV = 0.5m, typical of high-z star-forming galaxies (e.g., Pettini et al. 1998; Steidel et al. 1999). Note the strong Balmer + 4000 Å break due to the older stars. Also plotted (dotted lines) are the H and k filters in the rest-frame of a z = 4.04 galaxy, straddling the break. The SEDs come from the latest Bruzual & Charlot models. Bottom: The evolution of the (H - K) colour of a galaxy at z = 4.04 as a function of the time elapsed since an instantaneous burst of star formation. The solid curve is the Bruzual & Charlot model for Solar metallicity (z = 0.020), with the dashed line showing lower metallicity, 1/5 solar (z = 0.004). For this redshift, the (H - K) colour is an excellent tracer of the time elapsed since the end of star formation. The dotted curve is the solar-metallicity model with dust reddening of AV = 0.5m.

4.2 Keck/LRIS Spectroscopy

We have obtained optical spectroscopy from Keck/LRIS at moderate dispersion (lambda / DeltalambdaFWHM approx 1000) with a long slit aligned along the major axis of the arcs (Fig. 7). Our 4 ksec spectrum shows regions with Ly-alpha in emission that are adjacent to some of the bright knots seen in the optical HST images which sample the rest-frame ultraviolet (Figs. 7 & 9). The non-detections of N V 1240 Å, C IV 1549 Å & He II 1640 Å strongly favor the interpretation that the Ly-alpha arises from the Lyman continuum flux produced by OB stars, rather than the harder ultraviolet spectrum of an AGN. We see the Ly-alpha line morphology extending approx 1" beyond the ultraviolet continuum, which we attribute to resonant scattering from H I (Bunker, Moustakas & Davis 2000). In the bright knots, the SEDs are consistent with a very young stellar population (< 10 Myr) or ongoing star formation.

Figure 7

Figure 7. Left: the F814W image with elliptical galaxy model subtracted (note the counter arcs perpendicular to the axis of the main arcs, predicted by the lens model of Frye & Broadhurst 1998). The area covered by the long-slit optical spectroscopy is shown (slit axis is vertical). The right panel is this elliptical-subtracted image, smoothed to approx 0.6" seeing. Center: the LRIS spectrum, with the long-slit aligned along the arcs. The dispersion axis is horizontal, with wavelength increasing from left to right. The image has been smoothed by convolving with a Gaussian kernel of sigma = 1 pixel. Note the spatial range of Ly-alpha emission, which extends well beyond the detectable continuum of the arcs. The positions of the closest continuum knots N4 (-7.6") and S1 (4.3") are indicated by the horizontal bars. The knot N4 (top) lies on the edge of the slit, hence the slight spatial offset of the Ly-alpha line centroid and the lower flux compared to the line from S1 (bottom). Also indicated on the left panel is the z = 1.129 [O II] lambda 3727 Å galaxy also falling on our spectroscopic long-slit (see Fig. 9).

Figure 8

Figure 8. The top panels show archival HST/WFPC 2 imaging of the cluster Abell 2390. The z = 4.04 galaxy is the arclet at PA = +23° that is bisected by the elliptical. Top left is the HST V-band (F555W, 8400 s) which encompasses Ly-alpha, with the HST I-band (F814W, 10500 s) top right. The knots which are bright in the rest-ultraviolet (and so are presumably sites of recent star formation) are indicated. Our Keck/NIRC images were obtained in good seeing (0.4-0.5" FWHM) and are shown lower left (H, 2280 s) and lower right (K', 2880 s).

4.3 Evolutionary Status of the z = 4 Galaxy

We have evidence for both ongoing star formation and regions of older stellar populations in the lensed arcs. It is therefore unlikely that this z = 4 system in a true 'primæval' galaxy, viewed during its first major burst of star formation. Rather, our results suggest that the star formation history of this system has not been coeval, with current activity concentrated into small pockets within a larger, older structure. Correcting for the gravitational amplification (estimated to be approx 10 from lens models), the intrinsic properties of the z = 4.04 galaxy are comparable to the Lyman-break selected z approx 3-4 population of Steidel et al. (1996b, 1999). The current extinction-corrected star formation rate (approx 15 h50-2 Msun yr-1 for q0 = 0.5) may be adequate to 'build' an L* galaxy over a Hubble time, but a more likely scenario may be the creation of a sub-unit which will undergo subsequent merging with nearby systems (such as the other z = 4.04 galaxy identified in this field by Pelló et al. 1999) to assemble hierarchically the massive galaxies of today.

Figure 9a Figure 9b

Figure 9. Left: one-dimensional spectral extraction of the [O II] lambdalambda 3726.1,3728.9 Å galaxy at z = 1.1293, 3" below the southern arc and intercepted by our spectroscopic long-slit (Fig. 7). The extraction width is 7 pixels (1.5"). Note that the emission-line doublet is clearly resolved. We do not see this structure in the emission lines from the arcs, implying that their origin is not [O II] 3727 Å at z = 0.64. Right: one-dimensional spectral extraction of the southern arc, showing the region around Ly-alpha at z = 4.04. The extraction width is 8 pixels (1.7"), and encompasses both line- and continuum-emission regions. The asymmetric emission line profile is readily apparent, with the sharp decline on the blue side due to absorption by H I within the galaxy - a blueshifted Ly-alpha absorption trough is visible from the outflowing H I.

Figure 10

Figure 10. The broad-band optical/near-infrared flux from the entire northern arc. Also plotted are reddened instantaneous-burst stellar population models viewed at various ages, arbitrarily normalized to the flux measured from the HST/WFPC 2 F814W image. The flux in F555W is severely attenuated by the opacity of the intervening Ly-alpha forest. The colours are best reproduced by a stellar population ~ 50 Myr old, with in situ dust reddening of E(B - V) approx 0.1m. Note that at z = 4.04, the strong Balmer + 4000 Å break due to the older stars lies between the H- and K'-filters.

Next Contents Previous