ISSN:
1432-0967
Source:
Springer Online Journal Archives 1860-2000
Topics:
Geosciences
Notes:
Abstract The interdependence of the Fe(Mg)−1 (e.g., FeO-MgO in silicate melt; CaFeSi2O6-CaMgSi2O6 in pyroxene) and TiAl2(MgSi2)−1 exchange reactions between silicate melts and coexisting Ca-pyroxene has been examined. High-calcium clinopyroxenes were grown in 1 atmosphere melting and crystallization experiments on rock powders spanning the composition range tholeiite to melilitite (1,092〈T〈1,150° C: $$f_{{\text{O}}_{\text{2}} } $$ ∼QFM). Only microprobe analyses of the crystal rims were used in conjunction with those of the quenched glasses. Assuming equilibrium, relationships between apparent distribution coefficients for Fe2+⇄Mg2+ exchange and $$(X_{{\text{Ti}}^{{\text{4 + }}} }^{M1} )^{{\text{CPX}}} $$ suggest that at given values of $$\mu _{{\text{Fe(Mg)}}_{ - {\text{1}}} } $$ extent of Fe(Mg)−1 substitution is strongly coupled with the TiAl2(MgSi2)−1 substitution in pyroxenes near the five-component space CaMg(Si2O6-CaFe(Si)2O6-CaTi(Al)2O6-CaFe‴(Al,Si)2O6-CaAl(Al,Si)2O6. The inferred stabilization of Ti in iron-rich relative to magnesium pyroxene is consistent with the operation of Fe2+⇄Ti4+ intervalence charge transfer interactions (e.g., Rossman 1980) and observations on zoning in natural titanaugites (e.g., Tracy and Robinson 1977). Although the rims of some pyroxenes grown in some melting experiments exhibit prominent zoning in TiAl2(MgSi2)−1, the average values of $$(X_{{\text{Ti}}^{{\text{4 + }}} }^{M1} /X_{{\text{Mg}}^{{\text{2 + }}} }^{M1} )$$ inferred from the compositions of these pyroxenes, together with those of the relatively homogeneous pyroxenes produced in crystallization experiments, exhibit a 1∶1 correlation with values of $$\mu _{{\text{TiAl}}_{\text{2}} ({\text{MgSi}}_{\text{2}} )_{ - 1} } $$ derived from the solution model of Ghiorso et al. (1983) with a standard error of 750 calories. The Ti contents of Ca-rich pyroxenes crystallizing from a wide range of natural silicate liquids can therefore be predicted.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1007/BF00371701
