Scientific Justification

Using QSO spectra to trace the early chemical evolution in galaxies. Understanding the history of star formation and the associated metal enrichment in galaxies, and through it in the universe, is one of the central themes in astrophysics today. Quasars are key since they are the highest redshift objects for which spectra can be obtained. Hamann & Ferland (1993; see also Hamann & Ferland 1999 and references therein) showed how the NV emission line in QSO spectra can be used to measure the chemical abundances in the broad emission-line region (BELR) gas. The abundance of nitrogen relative to the other heavy elements (C,O, etc.) is a good marker of the degree of chemical enrichment of the interstellar medium in galaxies because N is built up as a secondary element in the CNO cycle, so that at higher metallicities N/Z µ Z, or N/H µ Z2. The N/Z abundance is what is most directly measured; it is then related to the total metallicity Z through a chemical enrichment model (typically a closed box with infall). With the further assumption that the BELR gas is being accreted from the surrounding galaxy, this offers a way to trace the history of metal enrichment in the centers of galaxies at large lookback times. This method is now starting to see widespread use (cf. Dietrich & Wilhelm-Erkens 2000; Thang et al. 2002). This body of work has shown that the quasar phenomenon only occurs after star formation has been ongoing for xxx years so that structure had to form well before the highest redshift quasars are seen, and that there is a correlation between metallicity and luminosity perhaps due to the known metallicity - galaxy mass correlation.

Figure 1. Our optical-uv spectrum of Q0353-383. The region to the red of Lya is also shown with the flux scale blown up by a factor of 5. The inset shows the overlap between the HST spectrum (heavy line) and our ground-based spectrum (light line).



The unusual spectrum of Q0353-383. This approach to determining metallicity relies on the single NV emission line, since only very strong lines can be detected in most objects. The z=1.96 quasi-stellar object Q0353-383 was discovered by Osmer & Smith in 1980. It was quickly seen to have a very unusual emission-line spectrum, in which the normal ultraviolet lines are present but are supplemented by exceptionally strong NIII] 1750 and NIV] 1486 intercombination lines. Osmer (1980) concluded that this object probably has a large underabundance of oxygen. However, we would have expected the very unusual N/O intensity ratios to instead be due to an overabundance of N.

Although photoionization simulations suggest that NV is a good Z indicator, it is important to validate it in a few well-understood cases. We are re-investigating Q0353-383, using new spectra and up-to-date models. This object is fundamental since it is one of the few in which many N lines can be measured, and so test whether the NV method works. Our goal is to determine whether the emission-line spectrum of Q0353-383 really does indicate exceptionally high nitrogen abundances, or whether some other mechanism is responsible for the unusual strengths of the nitrogen lines. This will tell us whether or not we can use the relative strengths of the nitrogen lines in the spectra of normal QSOs to trace the early chemical evolution of galaxies.

We have already taken new high-quality ground-based spectra and a HST STIS spectrum, and the data are fully analyzed. The combined data set is shown in Figure 1.

Figure 3. Metallicity determination for Q0353-383, using the models shown in Fig. 2. The model results for the different intensity ratios are represented by the separate lines on the plot (dotted lines indicate metallicites beyond the model grid). The symbols show where the measured intensity ratios (see key) fall on the corresponding lines on the plot. The heavy lines connecting pairs of the same symbol show the metallicity range due to the uncertainty in the NV strength.



Hamann et al. (2002) have recently calculated models for a series of different metallicities and substantially different ionizing continuum shapes, and used the results to pick out emission-line intensity ratios that are sensitive to metallicity but not to the continuum shape. Our results are summarized in Figure 2. Ratios of intercombination lines that involve the NIII] and NIV] lines are found to be good abundance indicators, as are NV 1240/HeII 1640 and NV 1240/CIV 1549.

Figure 2. Emission-line intensity ratios as a function of metallicity Z, computed by Hamann et al (2002) summing over clouds distributed throughout a model BELR, and for different ionizing continuum shapes: : solid line = Mathews & Ferland (1987) continuum; dashed line = hard (a = -1) power law; dotted line = broken power best fitting current estimates of average continuum shape.

Our ground-based spectrum of Q0353-383 gives high-quality results for these lines. Figure 3 shows their intensity ratios plotted onto the model results as a function of metallicity. All of these ratios point to a perplexingly high metallicity, Z>10Z. If the N abundance really is as high as it appears to be and is due to the general evolution of the stellar population averaged over the inner regions of the host galaxy then a huge starburst, far larger than anything previously observed, would be required to induce the type of rapid star formation this would imply. A slightly different model is that the environment has been polluted by localized enrichment and is not representative of the whole central region of the host galaxy. In this case, only ~50 Mof highly CNO-processed material (such as we actually find in the ejecta of objects such as h Carinae) could produce the observed nitrogen emission lines. This analysis has shown that NV is a good Z indicator, but raises other questions - the emission line spectrum is unusual in other ways - compared to the average quasar, the N III] and N IV] lines are far stronger relative to NV and Lya is stronger relative to the continuum.

The need for X-ray data. What is going on in Q0353-383 and what does it tell us about the early chemical evolution of massive galaxies? Our new HST and ground-based spectra show that NV is a valid indicator of Z, but raises other questions because of the differences mentioned above. The fact that lower ionization lines are stronger relative to NV than in most quasars suggests that the ionizing continuum may be softer. Is this the case? Does the SED become dramatically softer as the star cluster ages and Z increases?

Figure 5. Observed SED for Q0353-383, compared to broken power law continuum representing typical QSO SED. The "observed" X-ray data are from the ROSAT upper limit, assuming Fn µ n-1.6 below 1 kev, and Fn µ n-0.9 above 1kev.

Figure 4. Curves show predicted intensities of several strong emission lines relative to CIV, as a function of ionizing continuum shape, for our model in which XX% of C and O has been converted into N. The ionizing continuum is parameterized as a simple power law fn µ n-a(o-x). The observed ratios are marked with heavy O symbols. Even in this simple model, choosing a steep continuum with a(o-x) ~ 2.5 simultaneously explains the strengths of the most of the lines, including (not shown on the plot) their equivalent widths.

Figure 4 shows the results of a very simple BELR model in which we take gas in which XX% of the C and O have been converted into N (the h Carinae formula), at an arbitrarily-chosen point on the ionizing flux-gas density plane, and subject it to a power law ionizing flux. For an x-ray/optical flux ratio about 1000 times lower than normal (aox ~ 2.5), the observed Lya, NV, NIV], CIV, NIII], and MgII intensities ratios and equivalent widths can simultaneously be fitted to within a factor of two. A similar plot made using solar abundances does not simultaneously fit any of the line strengths for any SED. But do the data support the idea that Q0353-383 has both unusual abundances and an unusual continuum shape.

Figure 5 shows what we presently know about the SED of this object. Over the optical-UV range (rest-wavelength range 700-3300Å) it is very typical of other QSOs. However, we need to know if the ionizing spectrum, at shorter wavelengths, is abnormal. The most direct observational constraints will come from measuring the x-ray flux, so that we can at least interpolate across the energy range of interest rather than have to extrapolate across it.

The only available x-ray data are from the ROSAT archives, and give an upper limit of F(0.2-2.4kev) < 1.5x10-13 erg cm-2s-1. This is consistent with a normal aox. What we propose to do here is to push to a 10x fainter x-ray limit, to see whether we can convert this upper limit into a detection or whether Q0353-383 has a much steeper than normal aox. If in fact the spectrum is sharply cut off between the optical and x-ray passbands, a possible explanation will be that the ionizing continuum radiation is from a starburst, rather than being a non-thermal AGN spectrum. This would then lead us back to a simple and consistent picture of what is happening in this very unusual object, while at the same time giving us further confidence in the use of the nitrogen line strengths as abundance indicators in high-redshift objects.



References

Dietrich, M. & Wilhelm-Erkens, U. 2000, Astron. Astrophys., 354, 17.

Green, P., Aldcroft, T.L., Mathur, S., Wilkes, B.J., & Elvis, M. 2001, ApJ, 558, 109

Hamann, F. & Ferland, G. 1993, ApJ 418, 11.

Hamann, F. & Ferland, G. 1999, Ann.Rev.Astron. & Astrophys., 37, 487.

Hamann, F., Korista, K., Ferland, G., Warner, C. & Baldwin, J. 2002, ApJ, 564, 592.

Mathews, W.G. & Ferlan, G.J. 1987, ApJ. 323, 456.

Mathur, S., Wilkes, B. & Ghosh, H. 2002, astro-ph/0202202.

Osmer, P.S. 1980, ApJ 237, 666.

Osmer, P.S. & Smith, M.G. 1980, ApJ Suppl. 42, 333.

Thang et al. 2002 (need to get correct name here)



Proposed Observations

We are requesting Chandra time to use ACIS to measure aox and to at least crudely estimate the x-ray photon index G, down to either a detection or an upper limit 10 times fainter than the existing ROSAT upper limit. The limit from the ROSAT archives is from the 3s limit of 0.011 counts per second through a 90 arcsec radius aperture. Assuming a typical radio-quiet quasar slope of G = 2.5 gives Flux(0.1-2.4keV) < 7.8E-14 erg cm-2 s-1, or Unabsorbed-Flux(0.1-2.4keV) < 1.5E-13 erg cm-2 s-1. The power-law fit corresponding to this unabsorbed flux limit is shown on Fig. 5.

We then used PIMMS to estimate the ACIS-S count rates both for an object with this upper limit, and for an object 10 times fainter. Assuming only Galactic absorption at the position of Q0353-383 (1.68E20 cm-2), the predicted ACIS count rates are 2.3E-2 and 2.3E-3 cts/s. The corresponding pileup fractions are 3% and 0.3%, respectively, so pileup mitigation is unwarranted.

We propose to observe Q0353-383 for 20 ksec, which will result in 460 total counts if this QSO is as bright as the ROSAT upper limit, yielding G with <5% error, whether the object has normal G and only a Galactic absorbing column, or G=1.8 and as much as 1E23 intrinsic absorption (like broad absorption line BALQSOs; Green et al. 2001). If instead the quasar is 10 times fainter than the current flux limit, the resulting ~50 counts still constrains G to about 25%. However, since this error already encompasses most of the G dispersion in the quasar population, 20ksec represents a reasonable minimum request. For a 20ksec observation, the 3s detection limit for calculating aox would be 2.2E-15 erg cm-2 s-1, corresponding to aox > 3.3, indeed an unusual quasar akin to the most extreme X-ray weak BALQSO.

This observation will therefore either detect the x-ray flux from Q0353-383 and measure the important slope of the high energy continuum, or at worst show that the x-ray/optical flux ratio is at least 50,000 times smaller than normal. Either result will be an important new constraint on our interpretation of the very unusual emission-line spectrum from this object. This will provide a key piece for solving the puzzle of whether or not the nitrogen emission line strengths in QSOs can tell us about the evolution of chemical abundances early in the lives of massive galaxies.