Search

This dissertation develops computational techniques to aid in efficiently studying petrologic systems that would otherwise be challenging. It then focuses on a case study in which the transition from diagenesis to syn-magmatic heating led to a recrystallization and sulfur mobilization. A Markov-chain Montecarlo-based methodology is developed to allow for the assessment of uncertainty in calculated phase assemblage diagrams. Such phase equilibria are ubiquitous in modern petrology, but uncertainties are rarely considered. Methods are discussed for visualizing and quantifying emergent patterns as phase diagrams are re-calculated with input data modified within permitted uncertainty bounds, and these are implemented in a new code. Results show that uncertainty varies significantly across pressure-temperature space and that in some conditions, estimates of stable mineral assemblage are known with very little confidence. A Machine-Learning (ML) based methodology is developed for automatically identifying unknown phases using Electron-dispersion spectra (EDS) in concert with a Random Forest Classification algorithm. This methodology allows for phase identification that it is insensitive to overfitting and noisy spectra. However, this tool is limited by the amount of reference spectra available in the dataset on which the ML algorithm is trained. The approximately 250 EDS spectra in the current training database must be supplemented to make the tool more widely useful, though it currently has an excellent success rate for correctly identifying various sulfide and oxide minerals. An analysis of paragenesis associated with Central Atlantic Magmatic Province (CAMP) intrusions helps to better constrain the dynamics of magma emplacement, while also providing a method for estimating the amount of sedimentary sulfide-sequestered sulfur mobilized as a result of magnetite formation associated with igneous activity. This method demonstrates that dike emplacement can trigger liberation of sedimentary sulfur with no direct cooling impact on climate.

Many of the large igneous provinces during the Phanerozoic have been temporally linked to mass extinction events. The intrusion of magma into country rock has been hypothesized to facilitate the release of carbon and sulfur bearing volatiles and has been proposed as one of mechanisms that drove these mass extinctions. In this study I examine a dike of the Central Atlantic Magmatic Province and its interaction with adjacent sedimentary rocks in the Culpeper Basin of Virginia. Sampling was done at the 0.5 m scale along transects of sedimentary lithologies perpendicular to the ~170 m wide diabase intrusion. The observed mineralogical and geochemical changes in sedimentary rocks occur in a much narrower zone from the intrusion than predicted by the applied thermal model. Carbon isotopes of organic matter within the sedimentary rocks are enriched in 13C toward the intrusion indicating the generation of thermogenic methane within the first meter from the intrusion. Additionally, geochemical and petrologic textures suggest both the addition of magmatic sulfur into the country rock, shown through the isotopic signatures of sulfide minerals with mantle compositions. The possible thermal break down of sedimentary pyrite is evidenced by highly negative isotopic composition of sulfide minerals and general lack of pyrite. I suggest that sedimentary pyrite initially reacted to pyrrhotite, which was then converted to chalcopyrite through reactions with copper in fluids derived from the magma. These reactions also allowed for the formation of magnetite, which is elevated near the dike-sedimentary contact and at the end of the transect. A simple illustrative model of a hypothesis of fluid flow along the bed is used to explain the observed isotopic signatures and mineralogical changes along the transect. This study highlights how models for volatile generation through magma-country rock interaction may have overestimated the volatile fluxes from these environments and the roles that heterogeneity of sedimentary rocks and kinetic factors may have in the variance in these fluxes.

A two-dimensional numerical simulation of oxygen isotope transport in the shallow crust has been developed to examine the mechanisms of fluid flow during the formation of skarns. The Empire Mountain skarn in Mineral King, Sierra Nevada -- the motivation of this study -- has recently been identified as a low-δ¹⁸O skarn and interpreted as indicating the presence surface fluids during the onset of its formation (D'Errico et al., 2012). Rapid and sporadic changes in δ¹⁸O values within a single garnet have been interpreted as alternation between meteoric and magmatic fluid during garnet growth. D'Errico et al. (2012) conclude that multiple events of hydrofracturing of existing rock created transient, high-permeability conduits in the shallow crust, providing a low-resistance pathway for surface fluids to reach skarn-forming depths (~2-5 km). However, hydrofracturing typically occurs during retrograde metamorphic conditions which conflicts with the hypothesized prograde hydrofracturing during the incipient formation of the Empire Mountain skarn. Given this discrepancy, the objectives of this study are twofold: to explore mechanisms that can explain (1) the existence of meteoric δ¹⁸O signatures at skarn-forming depths and (2) the rapid and sporadic changes in fluid δ¹⁸O values throughout skarn formation. Numerical simulations reveal three ways in which a meteoric fluid signature can exist at skarn-forming depths during the onset of skarn formation: (1) convection-driven drawdown of surface waters in uniform, high permeability country rock; (2) existence of a large permeability contrast in the surrounding rock (e.g., extensional faults) that entrains surface water to depth; and (3) existence of pore fluids in isotopic equilibrium with ¹⁸O-depleted minerals prior to the magmatic intrusion. Possibilities (1) and (2) are difficult to substantiate given limited knowledge of the once overlying stratigraphy whereas the third provides the simplest explanation. Large fluctuations in fluid oxygen isotope compositions are observed in all three scenarios. In our model, the largest fluctuations in δ¹⁸O values occur while garnet is thermodynamically stable, in areas nearest to the magmatic body, and downstream relative to the topography-driven flow field. However, low δ¹⁸O and δ¹³C values of carbonates at Empire Mountain indicate infiltration-driven fluid flow during metamorphism, likely caused by the ductile collapse of carbonate pore space followed by brecciation. Ultimately, even though there exist scenarios that demonstrably show the occurrence of low-δ¹⁸O skarns, future simulations must include rock deformation (i.e., ductile closure of pore space and hydrofracturing) and mineral reactions to gain further insight to how low-δ¹⁸O skarns incorporate fluids with meteoric oxygen isotope signatures.

A two-dimensional numerical simulation of oxygen isotope transport in the shallow crust has been developed to examine the mechanisms of fluid flow during the formation of skarns. The Empire Mountain skarn in Mineral King, Sierra Nevada -- the motivation of this study -- has recently been identified as a low-δ¹⁸O skarn and interpreted as indicating the presence surface fluids during the onset of its formation (D'Errico et al., 2012). Rapid and sporadic changes in δ¹⁸O values within a single garnet have been interpreted as alternation between meteoric and magmatic fluid during garnet growth. D'Errico et al. (2012) conclude that multiple events of hydrofracturing of existing rock created transient, high-permeability conduits in the shallow crust, providing a low-resistance pathway for surface fluids to reach skarn-forming depths (~2-5 km). However, hydrofracturing typically occurs during retrograde metamorphic conditions which conflicts with the hypothesized prograde hydrofracturing during the incipient formation of the Empire Mountain skarn. Given this discrepancy, the objectives of this study are twofold: to explore mechanisms that can explain (1) the existence of meteoric δ¹⁸O signatures at skarn-forming depths and (2) the rapid and sporadic changes in fluid δ¹⁸O values throughout skarn formation. Numerical simulations reveal three ways in which a meteoric fluid signature can exist at skarn-forming depths during the onset of skarn formation: (1) convection-driven drawdown of surface waters in uniform, high permeability country rock; (2) existence of a large permeability contrast in the surrounding rock (e.g., extensional faults) that entrains surface water to depth; and (3) existence of pore fluids in isotopic equilibrium with ¹⁸O-depleted minerals prior to the magmatic intrusion. Possibilities (1) and (2) are difficult to substantiate given limited knowledge of the once overlying stratigraphy whereas the third provides the simplest explanation. Large fluctuations in fluid oxygen isotope compositions are observed in all three scenarios. In our model, the largest fluctuations in δ¹⁸O values occur while garnet is thermodynamically stable, in areas nearest to the magmatic body, and downstream relative to the topography-driven flow field. However, low δ¹⁸O and δ¹³C values of carbonates at Empire Mountain indicate infiltration-driven fluid flow during metamorphism, likely caused by the ductile collapse of carbonate pore space followed by brecciation. Ultimately, even though there exist scenarios that demonstrably show the occurrence of low-δ¹⁸O skarns, future simulations must include rock deformation (i.e., ductile closure of pore space and hydrofracturing) and mineral reactions to gain further insight to how low-δ¹⁸O skarns incorporate fluids with meteoric oxygen isotope signatures.

To characterize the metamorphism in the Tres Hermanas Mountains contact–metamorphic aureole in Luna County, New Mexico, I mapped and collected samples of the Paleozoic strata near the quartz monzonite pluton along the northeast flank of North Peak. Stratigraphy in this area includes the Pennsylvanian undivided unit that is composed of thin-medium interbedded limestone, shale, and sandstone and the Hueco Formation that could be broken out into four members: a pebble–cobble conglomerate marks the base, a medium to massive bedded siliceous limestone is next, followed by a pure limestone member and topped by a breccia bed in places marking the fault contact with extrusive andesite. These country rocks were intruded first by andesite sills and stocks and a second time by the large quartz monzonite pluton. The metamorphic assemblages are similar for all the sedimentary rock units and include quartz, calcite, diopside, plagioclase, K-feldspar, clinozoisite, garnet, and wollastonite. Mineral composition often vary within individual samples (e.g. various compositions along the diopsidehedenbergite, grossular-andradite, and albite-anorthite solid solution series). I was able to determine that the Paleozoic stratigraphy endured a lithostatic pressure between 22.5 and 37.5 MPa, at temperatures between 245 and 505 °C, and in the presence of a very water-rich metamorphic fluid having an X(CO2) less than 0.01. High-temperature minerals spurrite, merwinite, and monticellite were not observed in the contact aureole and may have formed as a result of the low pressure, water-rich fluids, and higher temperatures within the surrounding magma.

Very high temperature (>900 °C) contact metamorphism and metasomatism of aluminous schist xenoliths in the mafic to ultramafic Cortlandt Complex, New York, resulted in formation of bodies of unusual Fe- and Al-oxide-rich rock called emery. During contact heating, disequilibrium thermal decomposition of the protolith schists in one closely examined xenolith produced two end-member materials: a quartzo-feldspathic water-undersaturated melt which partitioned much of the silica and calcium and all of the alkalis of the original schist; and a highly aluminous fine-grained emery residuum which contained spinel, magnetite, ilmenohematite, sillimanite, and sporadically corundum. During cooling, melt within the xenoliths was injected as cm-scale veinlets into the silica-poor solid residuum. Local increase in silica activity resulted in progressive silication reactions of spinel-rich residuum to several silicates. A simple model of progressive silication would require that reactions should occur from lower to higher silica content of product silicates in stages, e.g., spinel – sapphirine (Si/O=0.10), sapphirine – garnet (0.25), garnet – orthopyroxene (0.28), rather than directly from spinel to higher-silica minerals which would overstep intermediate reaction steps. However, observed reaction textures indicate the latter more complex behavior in which spinel may have reaction rims of, or occur as inclusions within, any of the three silicate minerals. Statistical analysis of several samples has shown the mode to be the spinel-orthopyroxene reaction rim boundary although orthopyroxene is the highest-silica product mineral, based on Si/O ratio. Chi-square test results are significant and show that the textural relations observed among spinel, sapphirine, garnet, and orthopyroxene are dependent. Increased silica activity therefore cannot be the only factor controlling the reaction sequence. Microprobe data has been collected in an attempt to correlate mineral compositions with the different textural occurrences. The effects of local equilibria appear to be the dominant factors in the overstepping of sequential reactions. Qualitative activity-activity diagrams proved useful for examining the effects of bulk composition on the relative stabilities of spinel and the three silicates, including variations in Fe/(Fe+Mg), bulk Mn and Zn contents, and minor local variation in oxygen fugacity. Matrix spinel compositions (i.e., those not modified by reaction to silicates) fall into two groups: a more magnesian one containing spinels with average Fe/(Fe+Mg) (Fe#) of 0.49 and a less magnesian one, average Fe# of 0.67. With regard to this bulk compositional effect, the more magnesian composition should reduce garnet stability due to the strong fractionation of Fe into garnet, thus favoring the reaction of spinel to orthopyroxene within silica-rich areas. In more aluminous areas, spinel will react to form sapphirine, then garnet, then possibly orthopyroxene. A less magnesian composition would expand the stability of garnet at the expense of sapphirine and, to a lesser extent, orthopyroxene. Zinc has a subtle effect on mineral stabilities. Because Zn is strongly partitioned into spinel, higher zinc contents (concentrations in some spinels are as high as 14.9 mol% gahnite) may expand the stability of that mineral considerably. Consequently, spinel stability may increase relative to the three silicates, but this may be quite variable due to variable reaction stoichiometry and different reaction-boundary slopes in the activity-activity diagram. In general, spinels with the highest Zn content occur next to orthopyroxene (ave. 4.9 mol% gahnite in spinels) for which the stability appears to be only slightly affected by this increase in Zn. The greatest decrease in silicate stability is observed in sapphirine. Spinels adjacent to sapphirine contain no more than 1.3 mol% gahnite. The effects of manganese and oxygen fugacity were also examined. Mn increases the stability of garnet due to strong partitioning of Mn into this mineral. It can be inferred using statistical and chemical data that this has some bearing on textural relations in garnet-bearing samples, but the lack of obvious Mn fractionation by other minerals examined makes it impossible to interpret the effects of Mn in the garnet-free samples. Calculated ferric-ferrous ratios in analyzed minerals were examined in an attempt to study the effect of oxygen fugacity on the stabilities of minerals. In the more magnesian compositions, which may correlate with slightly higher fO2 during reactions, spinels should react to form sapphirine, then possibly garnet or orthopyroxene with further silica activity increase. In lower-fO2 environments (perhaps those with higher bulk Fe#), spinel should react directly to form orthopyroxene. The coexistence of magnetite and ilmenohematite dictates T-fO2 conditions very nearly at those of the Hematite-Magnetite buffer. Minor fO2 variations that might have had an effect on silicate-forming reactions would only be recorded by small variations in magnetite and ilmenohematite solid solutions (ulvospinel and ilmenite contents, respectively). These data were not acquired in this study, however, so no definite conclusions could be made.