4.3. Merger Scenarios
Let us now take a closer look at two alternate models for the construction of this two-component form of the MDF: mergers and accretions.
The idea that an elliptical galaxy could form from the direct merger of two colliding disk galaxies, thus forming more globular clusters out of the gas present within the progenitor galaxies, was forcefully presented by Schweizer (1987), followed by the more detailed modelling of Ashman & Zepf (1992 and subsequent papers). In this picture, most or all of the metal-poor clusters would have come from the progenitor galaxies while most of the metal-rich ones were formed out of the merger. Schweizer, Whitmore, and their colleagues have recently used the HST in a series of imaging studies of merger remnants of different ages (Whitmore & Schweizer 1995; Miller et al. 1997; Whitmore et al. 1997; Schweizer et al. 1996) to show in a very direct way that a large E galaxy indeed emerges out of the ``train wreck'' provided by the collision, and that a new, second population of globular clusters has formed in each merger.
It has been suggested in many papers (cf. the ones cited above) that the merger process is a way to understand the generic ``high specific frequency problem'' that ellipticals generally have larger values of SN than disk galaxies by factors of 2 or 3. Many of these statements are wrong simply because they ignore the field stars that will also form during the merger (the specific frequency is the ratio of cluster numbers to field-star light, so SN might stay the same, increase, or even decrease after the merger depending on the efficiency of cluster formation). In some other papers, this effect is recognized, but it is claimed that mergers ought to be an ideal place for super-efficient cluster formation because of the expected high degree of gas shocking and compression (though it is not obvious that these effects would be any more extreme than in, say, a more conventional Searle-Zinn formation process of merging gas clouds in a large potential well).
An excellent test of these arguments is to refer directly to the observations. The studies of merger remnants listed above show quantitatively that the newly born E galaxies have low specific frequencies, in the range SN ~ 1 - 3 after the age-fading of the field-star light is taken properly into account (cf. the references cited above). That is, globular clusters are indeed formed in the merger, but in more or less the same proportions relative to field-star light as the original disk galaxies had, so there is little change in the net specific frequency. These studies indicate that many of the low-SN ellipticals found in sparse groups populated by spirals and irregulars might indeed be the simple results of earlier mergers of this type.
What about the much higher-SN ellipticals such as in Virgo and Fornax (SN ~ 5), or even the M87-like objects with SN 10? Are there ways in which the merger process can be adapted to allow all ellipticals to have formed this way? In fact, the problem of sheer numbers of clusters in many big ellipticals presents quite a severe challenge to the merger hypothesis, in at least two directions:
(a) Taking NGC 4472 as a typical case, we see that it contains N(MP) 3660 metal-poor clusters (Geisler et al. 1996). If all of them came from the progenitor spirals, then we would require the merger of not just two disk galaxies (at typically ~ 200 clusters per galaxy, like the Milky Way or M31, this would be out of the question) but rather more like 15 to 20 of them. With this many mergers, it is also less clear how such a sharply defined bimodal MDF would still exist in the final product, or how the outer halo could preserve the strong net rotation velocity mentioned above.
(b) In NGC 4472, the number of metal-rich clusters is N(MR) 2440. If all of them were formed during the mergers, at the normal efficiency e0 1 per 2 x 108 M of gas, then the total input gas mass would have to be at least 5 x 1011 M. This enormous amount of gas does not resemble any merger happening today.
Both of these aspects of the problem put very extreme demands on the merger process: many mergers seem to be required, and the amount of input gas is so large that it is equivalent to an entire protogalaxy. Alternately, we must arbitrarily require that the cluster formation efficiency is vastly higher than normal.
Let us evaluate the problem more generally. If the progenitor galaxies have specific frequencies SN1, SN2 and luminosities L1, L2 (appropriately age-faded to ~ 13 Gyr for an old E galaxy population), then it can quickly be shown (see H99) that the specific frequency SN of the resulting elliptical will be
where L3 is the amount of field-star light (again, age-faded)
formed in the merger, and e is the efficiency of
conversion of gas into new globular clusters relative to the
``baseline'' efficiency e0 corresponding to
SN0 = 3.5 (see above).
This equation can be used to predict the outcoming SN for
any pair of input galaxies and input gas mass Mg =
L3 / (M/L).
Typically, in the present-day universe, even very ``gas-rich''
disk galaxies can supply only 109 - 1010
M of
input gas, which is far short of the amount needed to make
large numbers of clusters even if the efficiency e is an
order of magnitude higher than normal. To change SN from
a typical disk-galaxy level of SN ~ 2 up to the gE level
of 6 or more requires both an extremely high e-factor
(ten times e0 or more!) and extremely large amounts of
gas (5 x 1010
M or more). The
one obvious
era of the Universe's history when such large amounts of gas
were available was, of course, in the protogalactic epoch.
Such a merger picture differs significantly from the version
originally presented by
Schweizer (1987) or
Ashman & Zepf (1992),
and it is not clear how it could be distinguished from the
hierarchical amalgamation of gas clouds as in Searle & Zinn.
We must emphasize that mergers, in the mode proposed by Ashman & Zepf,
are a convincing way to interpret some E galaxies - the
ones with low specific frequencies like their progenitor spirals.
Such objects are quite clearly continuing to form
in the low-redshift contemporary universe in sites like the
Antennae galaxies and others.
However, this model is not easily adapted to building
the higher-SN ellipticals found in rich clusters.
Other noteworthy difficulties with the merger hypothesis as
a solution for all types of E formation are discussed at length
by (e.g.)
Geisler et al. (1996),
Forbes et al. (1997),
and Kissler-Patig et
al. (1998).