Contrib Mineral Petrol (1987) 97:116-126 Contributions to Mineralogy and Petrology 9 Springer-Verlag 1987 F e - M n partitioning between garnet and ilmenite: experimental calibration and applications Mark I. Pownceby 1, Victor J. Wall 1, and Hugh St.C. O'Neill 2. 1 Earth Sciences Department, Monash University, Clayton, 3168, Australia 2 R.S.E.S. Australian National University, Canberra, A.C.T., Australia Abstract. A new mineralogic geothermometer based on the partitioning of Fe and Mn between garnet and ilmenite has been calibrated by reversal experiments in the P-T range 600-900 ~ C, 2 and 5 kbars and for fO 2 = QFM. The results constitute a sensitive geothermometer applicable over a broad range of composition and conditions. Garnet- ilmenite thermometry has advantages relative to existing geothermometers because of its accurate calibration, marked temperature sensitivity and the chemical and struc- tural simplicity of the crystalline solutions involved. Appli- cation to natural assemblages reveals that the garnet-ilmen- ite geothermometer yields temperatures that agree well with other estimates. The reactivity of, and relatively rapid Fe-Mn diffusion in ilmenite may lead to retrograde reset- ting of high temperature partition values, but these factors may: be useful for estimating rock cooling rates. Analysis of the experimental data indicates minor positive deviations from ideality for Fe-Mn garnets and ilmenites. Absolute magnitudes of interaction parameters (WAB) derived from a regression analysis are subject to considerable uncertainty. The partition coefficient is, however, strongly dependent on the difference between solution parameters. These differ- ences are well constrained with a magnitude of r;~zn,1 VrFeMn gar -- WF~eMn ~-~ 300 cal mol- 1. The accuracy and applicability of garnet-ilmenite thermometry will improve with the avail- ability of better thermodynamic data for garnet crystalline solutions. Introduction One of the basic aims of petrology is to relate observed mineral assemblages to their previous T-P history. Early * Present address: C.R.A. Research A.T.D., P.O. Box 42, Boolaroo, N.S.W., 2284, Australia Offprint requests to: M.I. Pownceby Abbreviations and symbols used in text: R universal gas constant (cal/mol/~ T absolute temperature (~ or ~ P pressure (kbars); AV ~ volume change of reaction (1); AHgl,r) standard state enthalpy change of reaction (1) at 1 bar and the T of interest, in cal/mole; AS ~ entropy change of reaction (1) at T of interest, in cal/mole/~ AG~e,r) standard free energy change of reaction (1) at the T and P of interest, in cal/mole; K gar'ilm distribution coeffi- DFe - Mn cient for Fe--Mn partitioning between garnet and ilmenite; K ap- parent equilibrium coefficient for reaction (1); e{ activity of compo- nent i in phase j; Wn- B binary A - B interaction (Margules) param- eter; gar garnet; ilm ilmenite; blot biotite; ol olivine; opx orthopy- roxene attempts at quantitatively associating tectonic processes with crustal and upper mantle processes (England and Ri- chardson 1977; Wells 1980; Thompson 1982)have demon- strated that a detailed knowledge of crystallisation tempera- tures is essential. Knowledge is required not only of the peak crystallisation temperatures, but also of regional varia- tions in the inferred maximum temperatures, as well as the prograde and retrograde temperature paths of rocks. Conse- quently, field and laboratory studies have played prominent roles in shaping current theories concerning the thermal processes operating within the crust and upper mantle. To assess previous thermal histories, petrologists must rely largely on the availability of well-calibrated geothermome- ters. In recent years considerable attention has been devoted to the development of solid-solid equilibria to estimate the temperatures of crystallisation or recrystallisation of rocks. Of the many geothermometers that have been proposed, those based on the distribution of a pair of cations between coexisting phases (Kretz 1959) have been the most widely studied. Both experimental (Hensen and Green 1973; Ferry and Spear 1978) and thermodynamic (Saxena 1979) or com- bined experimental and thermodynamic (O'Neill and Wood 1979; Harley 1984) approaches have been adopted. The practicability of exchange equilibria as geothermometers is greatly enhanced if one phase strongly fractionates a partic- ular cation over the coexisting phase, so that the magnitude of the distribution coefficient is large and significantly vari- able over geologically reasonable P- T conditions. Garnet strongly concentrates manganese over most coexisting phases (Hollister 1966) with the ratio Mn/ (Mn + Fe) decreasing in the order Gar > Ilm >> Sta > Crd >Bio (Thompson 1976; Woodsworth 1977). This observa- tion, and the comparative abundance of garnet-bearing as- semblages in many bulk rock compositions may, in princi- ple, be used to formulate geothermometers based on the exchange of Mn with a cation of similar charge and ionic radius (e.g., Fe, Mg, Ca - Kretz 1959). An important manganese-bearing phase with which gar- net may stably coexist with, is ilmenite. Ilmenite (FeTiO3) commonly forms a solid solution with pyrophanite (MnTiO~) suggesting the possibility of constructing a geo- thermometer based on the Fe-Mn partitioning between garnet and ilmenite. This mineral pair is particularly suit- able for investigation because the chemistry of the system is simple and the two minerals are associated in a wide variety of igneous and metamorphic rocks formed not only within the crust, but also in the upper mantle. The composi-