Comment on “Subduction factory: 1. Theoretical mineralogy, densities, seismic wave speeds, and H2O contents” by Bradley R. Hacker, Geoffrey A. Abers, and Simon M. Peacock

[1] Hacker et al. [2003a] presented an interesting work on metamorphic influence on the physical properties (density, H2O content, seismic wave velocities) of mafic and ultramafic rocks, and in the companion paper [Hacker et al., 2003b] they presented the consequences on the localization of earthquakes in subducting slabs. While this work is very interesting and well presented, this paper appears to ignore similar works for densities calculations [Bousquet et al., 1997; Le Pichon et al., 1997; Henry et al., 1997; Gerya et al., 2002], for H2O content in rocks [Bousquet et al., 1997; Kerrick and Connolly, 2001], and for the seismic waves velocities calculations [Sobolev and Babeyko, 1989, 1994]. This absence of discussion might be without major consequence if all papers agreed. However, this is not the case. The results of densities calculations for high-grade conditions are in some temperature-pressure (TP) conditions in contradiction with previous [Hacker, 1996; Bousquet et al., 1997] and newest [Goffé et al., 2003; Vasilyev et al., 2004] works as well with measured densities of natural rocks [Austrheim, 1987] or experimental works [Green and Ringwood, 1967; Ito and Kennedy, 1971] with similar compositions. Density should vary with physical parameters according to the thermodynamic laws. [2] For example, for mid-ocean ridge basalt (MORB)-type rocks we notice a strong increase of density with temperature increasing around 800°C for pressures between 10 and 15 kbar and a decrease of density with pressure increasing around 27 kbar for temperatures between 500 and 700°C (Figure 1). These results seem to follow from mineral compositions of metamorphic facies chosen by Hacker et al. [2003a], in which some assumptions are questionable: [3] 1. In high-temperature conditions the increase of density depends mainly on the amount of garnet in rock [Bousquet et al., 1997]. In the granulite facies, slopes of reaction of garnet breakdown into plagioclase are all positive in a P-T diagram [Ringwood, 1975]. Knowing that garnets are denser than plagioclases, rocks should become less dense with an increase of temperature and not denser as claimed by Hacker et al. [2003a]. [4] 2. They claim (paragraph (7)) "there is no stability field for chloritoid in Figure 1, consistent with its absence in naturally metamorphosed rocks of MORB composition." However, among the many mountain belts where mafic rocks were metamorphosed under blueschist and eclogite conditions, chloritoid has been described in several of them. For example, chloritoid is well known in mafic rocks of the European Alps from east [Miller, 1986] to Liguria [Messiga et al., 1995] through the central and the western Alps [Bearth, 1973; see Droop et al., 1990] as well as in rocks of Greece [Schliestedt, 1986]. Their assumption also contradicts experimental work [e.g., Schmidt and Poli, 1998] and previous modeling [e.g., Peacock and Wang, 1999]. [5] It is also unusual to assume that the Na-amphiboles (glaucophane and Fe-glaucophane) are stable in amphibolite facies from 600 to 800°C at 0.8–1.2 GPa [Hacker et al., 2003a, Table 3]. Such stability field of Na-amphibole is in contradiction with their own assumption presented in their Table 4 in which they assume that in “amphibolite, granulite, and related facies…all Na is in albite and pargasite.” Since the extensive work of W. G. Ernst in the Franciscan mélange, it is well known that Glaucophane-bearing rocks derive from low-temperature gradient [e.g., Ernst, 1973], and this assumption is used in many modeling works [e.g.,Peacock, 1993; Bousquet et al., 1997; Kerrick and Connolly, 2001]. Experimental works have shown that stability field of Na-amphibole is strongly temperature-dependent [Maresch, 1977]. The stability field is below 600°C for pressures up to 1.5 GPa [Evans, 1990] and goes up to 800°C for pressures higher than 2.5 GPa [Tropper et al., 2000]. [6] Uncertainties in the densities of rocks presented by Hacker et al. [2003a] also have a great impact on the prediction of seismic wave velocities (e.g., for comparison, see Sobolev and Babeyko [1994]). Similar to the density, the water content of a rock is strongly dependent on the mineral composition. Inconsistencies in the mineral assemblages also explain the discrepancies of water-content phase diagrams with previous studies [Hacker, 1996; Bousquet et al., 1997; Kerrick and Connolly, 2001]. [7] In conclusion, we regret that these innovative papers showing the dependency between seismic wave velocities of rocks and metamorphic reactions do not mention previous published works with which they have significant disagreements. Also, we do not think their paper would have been hurt by noting that Sobolev and Babeyko [1994] had already modeled the evolution of seismic wave velocities in function of metamorphic reactions for different bulk composition of rocks and Bousquet et al. [1997] had already modeled the water contents of rocks and their consequences on the localization of crustal earthquakes in convergent zones.