The observations of the diffuse ultraviolet background at high galactic latitudes are collected in Figure 1. The figure caption contains references to the sources of the various observations. We will, for now, assume that this radiation is extragalactic.
Figure 1. The plotted observations of (assumed) extragalactic diffuse cosmic ultraviolet background radiation (letters in Figure) have been reviewed by Henry (1991). The intensity in units is plotted against the wavelength of observation in Å. Superposed on the observations are predicted spectra of the recombination radiation from ionized intergalactic clouds (solid line, clouds that are expanding as the universe expands; dashed line, gravitationally bound clouds). Neutrino decay radiation would also produce a spectrum shaped like the dashed line. The observation V, shortward of Lyman (1216 Å), is the Voyager upper limit of Holberg (1986; solid line: calibration of Holberg et al. 1982, dashed line: calibrations of Brune et al. 1979 and Cook et al. 1989). The observations longward of L are all positive detections, not upper limits. They include spectroscopy (boxes, and filled point): A (Anderson et al. 1979), M (Martin and Bowyer 1990), T (Tennyson et al. 1988), H (Hurwitz et al. 1990), and photometric observations (open circles): W (Weller 1983), S (Paresce et al. 1979), J (Jakobsen et al. 1984), K (Onaka 1990), A (Anderson et al. 1979), D (Joubert et al. 1983), F (Fix et al. 1989). We have omitted the Apollo 17 observation of Henry et al. (1978), which agrees with the others at 1500 Å, but which shows a decline at longer wavelengths that is surely spurious.
First, we note that the extragalactic background shortward of L (1216 Å) is less than 100 photons cm-2 s-1 sr-1 Å-1 (units that will be called "units" hereafter) if the calibration of Holberg et al. (1982) is used (solid line in figure), or still less (dashed line) if the calibration of Brune et al. (1979) or Cook et al. (1989) is used. In contrast, all of the observations in Figure 1 longward of 1216 Å are positive detections of 200 units, not upper limits.
We stress that the scatter that appears in the observations of Figure 1 demonstrably reflects a spatial variation in overall intensity of the ultraviolet background on the sky, rather than either uncertainties in the observations or gross (or detailed) spectral variations. This is demonstrated most clearly by examining individual spectra, for example Figure 10 of Anderson et al. (1979). Such examination makes clear that the ultraviolet background longward of 1216 Å (at a given location) shows no trace of a decline toward shorter wavelengths. The decline is abrupt, and occurs near 1216 Å.
What is the interpretation of the remarkable spectrum of Figure 1? The striking feature of course is the ledge: no detectable radiation short of about 1216 Å; strong, well-observed radiation longward of that wavelength. Now, a ledge of just this character has been searched for ever since de Rújula and Glashow (1980) suggested that the dark matter might be massive neutrinos that might decay with the emission of an ultraviolet photon. A ledge of exactly this character was predicted, and has been vigorously searched for. For example Stecker (1980) identified a ledge in the fragmentary observations of the time, and suggested the possibility that neutrino decay radiation had actually been observed. While that particular ledge is no more, interest in the general subject remains keen (Kimble et al. 1981; Henry and Feldman 1981; Murthy and Henry 1987).
However, while we have now located an excellent ledge, the shape of the spectrum toward longer wavelengths does not fit an origin in neutrino decay radiation. The expected spectrum is given in Equation 1, as a function of redshift z (for the same equation in terms of wavelength, see Kimble et al. 1981).
where nv is the local density of neutrinos of the type that decay with the emission of a photon, eÅ is the wavelength of the emitted radiation, is the lifetime of the neutrinos against radiative decay, c = 3 x 1010 cm s-1, the Hubble parameter H0 = 50 h50 km s-1 Mpc-1, 1 + z = / 1216 Å is the redshift due to the expansion of the universe, and = 1 (from inflation, but see also Dekel et al. 1993).
The spectrum of Equation 1 is shown in Figure 1 as a dashed line. The decline to longer wavelengths is ~ -5/2 and clearly such a decline is not supported at all by the observations. The only hope of preserving this origin would be to hypothesize that some other (galactic?) source was filling in at longer wavelengths, but if this were the case, the observations (see the figure) would provide no evidence at all for the neutrino decay radiation itself. There is therefore no support whatever for the idea that neutrino decay radiation has been detected.
Had we been able to conclude that the observed radiation actually was neutrino decay radiation, we would have had to attribute the fact that the ledge occurred at a wavelength very close to that of Lyman to chance. Let us next look at another potential origin where again we would ascribe the wavelength of the ledge to coincidence: an origin in the redshifted light of galaxies. In particular, could the ledge be the Lyman limit to the light from a tremendous burst of star formation in starburst galaxies?
To have the ledge occur near 1216 Å requires that the starbursts occurred at a redshift of 0.33, and the sharpness of the ledge requires that the activity have begun and ended rather abruptly. None of this is very palatable. Even more important, Martin, Hurwitz, and Bowyer (1991) have looked very carefully and critically at the idea that the diffuse background spectrum observed longward of 1216 Å could originate in starburst activity, and do not succeed in reconciling the observations with a dominant contribution from unclustered starburst galaxies at low redshift.
Finally, we consider the idea that we are seeing redshifted extragalactic hydrogen recombination radiation and that the fact that the ledge occurs near 1216 Å is not a coincidence. The expected spectral shape for the case of gravitationally-bound clouds of ionized hydrogen is identical to that of neutrino decay radiation, and so that case may be rejected immediately on observational grounds.
The spectrum for intergalactic clouds that are expanding as the universe expands is given in Equation 2.
where = 2.8 x 10-13 cm3 s-1 is the effective recombination coefficient (recombination to n = 1 generates no L radiation), x = 0.746 is the fraction of baryons that are hydrogen nuclei, the local density of ionized hydrogen nuclei n0 = 2.83 x 10-6 h250 g, eÅ = 1216, C = the clumping factor (which is independent of z in this model). We take the gas temperature (which affects ) to be 8000 K, for reasons that are given below; our results are very insensitive to the temperature.
The spectrum of Equation 2 appears in Figure 1 as the solid line, which fits the data acceptably well, considering that the data are obtained at many different locations at high galactic latitudes. The parameter g is the contribution to that is due to ionized intergalactic gas. If we attribute the jump at z = 0 of 300 units that is shown by the observations (Figure 1) to intergalactic recombination radiation, we find that
describes the observations.
At this point, we recapitulate facts that bear directly on the possibility of detecting redshifted Lyman recombination radiation from ionized intergalactic gas.
The very small spatial fluctuations observed by COBE indicate that the dark matter was an essential ingredient in the formation of structure among the baryons following recombination. The structured dark matter was already there, and following recombination the neutral hydrogen and helium fell into the potential wells, creating the structure we observe today.
Intergalactic space was left free of neutral hydrogen. Indeed, intergalactic space is astonishingly free of neutral hydrogen, the density being < 4.5 x 10-14 h50 cm-3 (Steidel and Sargent 1987). This means that in a volume of 50 cubic megaparsecs, where there is on average one galaxy, of mass 8 x 1010 solar masses (Allen 1973), there are < 50,000 solar masses of (smoothly distributed) neutral intergalactic gas. Galaxy formation gathered up all of the baryons except a fraction < 6 x 10-7, an efficient process indeed - unless intergalactic hydrogen is highly ionized. Observations of the cosmic microwave background (Mather et al. 1990) show that such intergalactic hydrogen cannot be at very high temperatures (Rogers and Field 1991). At lower temperatures, recombination becomes more efficient, especially if the intergalactic gas is clumped. The clumping C = < n2e > / < ne >2 = 1.5 x 107 for galaxies, and is unknown for ionized intergalactic gas.
The final fact that bears directly on the search for recombination radiation is that only a small part of the expected baryonic matter is accounted for by matter that has already been detected. For example Persic and Salucci (1992) estimate the baryon mass density of the universe due to the stars in galaxies and hot gas in clusters and groups of galaxies. They find b = 0.003, which is less than 10 percent of the lower limit predicted by standard primordial nucleosynthesis which implies that the great majority of the baryons in the universe are as yet unseen.
We are now prepared to explore the consequences of equation 3. Persic and Salucci quote Kolb and Turner (1990) and Peebles et al. (1991) in giving b h250 = 0.06 as the most probable value for the baryon density from nucleosynthesis. We take g = b, that is, essentially all of the baryons are intergalactic ionized hydrogen. Insertion into equation 3 then gives C / h50 = 50,000. Notice that this clumping is vastly less than for the visible matter (galaxies) in the universe. The clumps of ionized baryons in the universe are much larger in relation to their separation than are galaxies.
We have seen that the intergalactic medium is unquestionably highly ionized. What causes this ionization? There is some controversy over this. Meiksin and Madau (1993) provide several models in which the observed QSOs can provide the required ionizing photons at early epochs, and we accept their conclusion.
However, our highly clumped ionized intergalactic medium has much more severe problems. For h50 = 1, the recombination time of our clouds is only 1.3 x 107 years, and for h50 = 2 the recombination time is 6.3 x 106 years. Thus, if this interpretation of the diffuse high galactic latitude diffuse background is correct, a strong additional source of ionizing photons is required. Just such a source, radiative decay of neutrinos, has been proposed by Sciama (1993, e.g., in which Sciama references defenses of his neutrinos against the conclusions of Davidsen et al. 1991, who failed to observe a neutrino decay line from the cluster of galaxies A665). Sciama's neutrinos decay with the emission of photons that are just capable of ionizing hydrogen (hence our assumed temperature, above, of ~ 8000 K). Sciama's neutrinos, in our present picture, would be the dark matter into which the baryons all fell following recombination. In every potential well, most (or rather, in most cases, all) of the hydrogen was re-ionized by Sciama's neutrinos. In exceptional cases dissipation occurred and quasars and galaxies formed. In most cases, in contrast, the hydrogen simply re-ionized and expanded out of the well, forming our present clouds that are expanding with the universe. Such clouds would be extremely hard to detect, either in emission or absorption, because of the very large velocity dispersion that is expected.
In the present picture there is no way of avoiding Sciama's neutrinos (or their equivalent). If we assume that all of closure density ( = 1) is ionized gas (that is, we ignore the nucleosynthesis argument), we obtain present-day recombination times that are of the order of the Hubble time, but the recombination time earlier (when the clouds were denser) would still be too short.
We conclude that if the high-galactic latitude diffuse ultraviolet background is extragalactic in origin, then its observed spectrum implies all of: a) detection of the baryonic dark matter; b) detection of the effects of the non-baryonic dark matter (Sciama neutrinos); and c) evidence for new physics beyond the standard model of elementary particle physics (Sciama neutrinos). As that would be an important set of discoveries indeed, we now turn to the observations at lower galactic latitudes in an effort to account for the high galactic latitude signal as being galactic, rather than extragalactic, in character.