Abstract: Magmatic deposits are generally regarded as the product of magmatic differentiation for lack of strong near-ore wall rock alteration. Ophiolitic podiform chromite deposits are typical magmatic deposits. Popular genetic models of podiform chromitities include the magma channel model and melt-rock interaction model. However, the discovery of a massive deep seated crystal population, indicated the chromite ore system is not an ideal but complex dynamic system with at least two subsystems. A different genetic model, therefore, is needed to account for the new evidences. One characteristic feature of chromite deposits is the zoning from harzburgite wall rock, through shell dunite, to chromitite, with progressive contacts between dunite and chromitite, and abrupt, progressive or intrusive contacts between shell dunite and harzburgite. Thus, the origin of dunite is key to the understanding of formation mechanism of podiform chromitities. The recent observations of two typical ultramafic complexes in Luobusa and Zedang in the Yarlung Zangbo suture zone revealed: (1) the width of shell dunite varies from centimeters to hundred meters, but the rocks uniform fine-grained structure can not be explained by the popular models; (2) shell dunite can be classified into massive and schistose structures, which implies a multi-stage process in dunite formation; (3) shall dunite mainly consists of blastic olivine, with few olivine crystals crystallized from melt or fluid; (4) in contrast to harzburgite, olivine crystals in shell dunite are characterized by high MgO, Cr2O3, CaO and low MnO, Al2O3 contents-two contradictory features; and (5) the anti-podiform texture, commonly found in dunite near the chromite ore body, resembles the mesoscale texture formed by turbulence in the multi dilute fluid flow. These seemingly contradictory observations suggest that dunite formation involves a deep seated fluid mass, hence fluid process could be a basis for building a new model. Based on this assumption, we proposed here a melt-fluid flow model, its mechanism is the dissolution-precipitation reaction: Opx+Fluid→Ol±Sp±Cpx±Pl±SiO2 (fluid); the basic preconditions for the mechanism are continuous supply of deep fluid and rapid rise of melt-fluid flow. In this paper we also demonstrated that evidences from igneous geology, petrology and crystal chemistry of nominally anhydrous minerals can be used to rebuild the fluid state of a magma.

Key words: shell dunite, podiform chromitites, olivine, melt-fluid flow, Luobusa ophiolite, Tibet