Master of Science / Department of Geology / Matthew W. Totten / Sedimentary beds rich in volcanic ash have been reported throughout Kansas. It is believed the source of these ashes are the large-scale eruptions from the Yellowstone Calderas. Very few of these ash units have been dated, however, and the vast majority simply reported as “Pearlette Ash.” The objective of this research was to investigate the potential of trace element geochemistry in correlating individual ash outcrops in Kansas to their eruptive source. Thirty-six previously reported ash occurrences of unknown age in Kansas were reoccupied and sampled. In addition, three unreported ash deposits were discovered and sampled. Two ash units previously identified as Huckleberry Ridge-aged and three as Lava Creek B were also collected. The samples were processed using the method of Hanan and Totten (1998) to concentrate ash shards. These ash concentrates were analyzed for specific trace and rare earth element (REE) concentrations using inductively coupled mass-spectrometry (ICP-MS) at the University of Kansas. The ash samples from known eruptions have distinct trace and REE signatures, allowing comparison to the unknown ash units. Most of the unknown ash samples correlate with specific Yellowstone eruptions. The majority of the undifferentiated “Pearlette Ash” samples correlate with the most recent Lava Creek B eruption and several unknown ashes correlate to the Huckleberry Ridge eruption. The distribution of ash units in Kansas being dominated by Lava Creek (0.60 ma) is expected because it is the most recent of the Yellowstone eruptions. The abundance of the older Huckleberry Ridge (2.10 ma) over the more recent Mesa Falls (1.27 ma) is likely the result of the much larger Huckleberry Ridge eruption.

A fracture and permeability analysis was performed on the Santana Tuff because of its similarity to the Topopah Springs unit at the Yucca Mountain site. The Topopah Springs unit is the proposed horizon for the spent nuclear fuel repository. Because of the impossibility of completely characterizing the flow properties of the unit without destroying the characteristics that make it desirable as a repository, other ash flow tuffs must be studied. The Santana Tuff and the Topopah Springs tuff both are rhyolitic in composition, nonwelded to densely welded and fractured. Fractures were examined at six outcrop locations spanning a five mile area. Stereonets and rose diagrams were constructed from over 312 fracture orientations. Although the composite data showed two major orientations of nearly vertical fractures, fracture trends at individual outcrops showed a variety of preferred orientations. Over 900 surface permeability measurements were taken using a mini-permeameter. The samples were categorized by three observed types of surface weathering: fresh, weathered, or varnished. Fracture surfaces were generally classified as weathered. The average permeabilities for the samples are 55.33 millidarcies, 5.03 millidarcies, and 3.31 millidarcies, respectively. The one-way statistical analysis performed on the data indicated that the permeability of fresh tuff surfaces is significantly different than both the permeabilities of the weathered and varnished tuffs, using both a least significant difference and greatest significant difference test. However, no difference was shown to exist between the weathered and varnished tuff permeabilities. Samples of fresh, weathered, and varnished tuffs were examined by X-Ray Defraction, the Scanning Electron Microscope, and in thin section. The SEM analysis showed surface differences between the three weathering classifications. The weathered and varnished samples were similar, exhibiting a platy, lamellate texture. The fresh surfaces were irregular and jagged. In thin section, a thin rind of dark minerals (FE-oxides) is observed on the edges of the varnished samples and in microcracks. This fills surface pores and causes the reduction in permeability. Two other zones of weathering have been identified in some of the samples, which may also cause changes in permeability. Tuff permeabilities were also analyzed for directional dependence. After an ash flow tuff is deposited and cooled, it may undergo flattening of pumice fragments and glass shards. These flattened fragments can be identified in handsamples, and are indicative of the direction of flow emplacement. The analysis showed that permeability is enhanced parallel to the emplacement direction, which is generally horizontal. Cut surfaces showed a 30% decrease in permeability perpendicular to flow direction. On varnished surfaces, this trend is still evident, although decreased in magnitude. This is expected because of the clay particles which make up the desert varnish. This study indicates that the formation of low permeability weathering rinds in association with vertical fractures may inhibit infiltration at the surface. It may accelerate infiltration at depth and allow more fluid to penetrate vertically into the tuff. In the event that fluid is absorbed into the matrix, it will travel horizontally, along the enhanced permeability parallel to the emplacement direction. / text

Theoretical and experimental studies of the chemical and physical factors which affect molecular diffusion of dissolved substances from fractures into a tuffaceous rock matrix have been made on rocks from G Tunnel and Yucca Mountain at the Nevada Test Site (NT8). Although a number of physical/chemical processes may cause nonadvective transport of dissolved species from fractures into the tuff matrix, diffusion in these rocks is controlled by the composition of the groundwater through multicomponent effects and several rock properties. The effective molecular diffusion coefficient of a particular species in the tuff can be related to its free aqueous diffusion coefficient by Dₑ = θ(m)(α/τ²)D₀ where bm is matrix porosity, α is the constrictivity, and τ is the tortuosity. The porosities of the samples studied ranged from 0.1 to 0.4. The parameter (α/τ²) ranged from 0.1 to 0.3, and effective matrix dif— fusion coefficients were measured to be between 2 to 17. x 10⁻⁷ cm²/s for sodium halides and sodium pentafluorobenzoate. Total porosity was found to be the principle factor accounting for the variation in effective diffusion coefficients. The constrictivity— tortuosity factor was found to have a fair correlation with the median pore diameters measured by mercury intrusion. Measurements of bulk rock electrical impedance changes with frequency indicate that the constrictivity factor, a, has a maximum value of 0.8 to 1, but may be smaller. If the larger values are correct, then the diffusion paths in tuff are more tortuous than in granular media. The diffusion coefficient matrix computed for various tracers in J-13 well water from the NTS indicates coupling of the diffusion fluxes of all ionic species. Multicomponent diffusion is a second order effect, however, which does not significantly affect experimental results. The results of a bench—scale fracture flow experiment revealed that the transport of ionic tracers (SCN ⁻ and pentafluorobenzoate) was affected by diffusion into the tuff matrix. The transport of a particulate tracer did not appear to be affected by diffusion.

Reconnaissance work and high-density sampling of volcanic rocks at the Quaternary Tatara-San Pedro complex (TSPC) in the Southern Volcanic Zone (SVZ) of Chile has yielded one of the most complete eruptive chrono-stratigraphies and comprehensive geochemical datasets of any arc volcano on Earth. The TSPC is a large frontal arc stratovolcano within the SVZ that exhibits a wide compositional diversity of lavas from basalt to rhyolite, covering most of the ranges in major and trace element contents across the SVZ. The TSPC occupies a pivotal position within the SVZ, where it is "intermediate" in terms of geophysical and geochemical characteristics between northern and southern SVZ volcanoes. The large TSPC dataset and stratigraphic control provides a unique opportunity to elucidate magma source heterogeneity and distinguish between contributions from upper mantle, subducted slab, and crust in a volcanic complex overlying relatively thick continental crust. Furthermore, the results of this investigation provide important constraints about the role of various recycled materials in generating the SVZ mantle and implications for along-arc magmatism and geochemical variability. TSPC magmas least impacted by crustal contamination (evolved lavas are filtered out on the basis of > 56 wt.% SiO2 and Rb/Y >1.75) have compositions bounded by three chemically and isotopically distinct mantle-derived end-members. The `prevalent TSPC mantle' end-member, which includes the largest number of analyzed lavas, is interpreted to represent melts of the upper mantle below TSPC that has been modified by long-term subduction. A second end-member shows extreme depletions in incompatible high field strength elements (HFSE) and the lowest concentrations of fluid-immobile incompatible elements, but has the highest aqueous fluid-mobile/immobile element ratios at the volcano (e.g., Sr/Nd and Pb/Ce). The source of these `low HFSE' magmas is `prevalent TSPC mantle' that experienced previous melt extraction, followed by more recent melting due to infiltration of solute-rich fluid from the subducting basaltic Nazca oceanic crust. A third end-member is enriched in incompatible elements and has the lowest Nd-Hf and highest Sr isotope ratios. This `TE enriched' end-member has common chemical characteristics with behind-the-arc basalts, indicating derivation from trace element-enriched behind-the-arc South American mantle that has been advected trenchward into the convecting mantle wedge. Determining the composition and relative input of slab-derived components to the SVZ mantle wedge has remained elusive for the last 2+ decades because of inadequate datasets and the controversial role of crustal contamination within the thick Andean continental crust. The `prevalent TSPC mantle' magmas, which best represent melts of the subduction-modified mantle wedge composition beneath the TSPC, provide important constraints on both the composition of the "pre-subduction" mantle and geochemical modifications by way of subduction. Mass-balance modeling suggests that the source of `prevalent TSPC mantle' magmas has been generated by a two-stage, three component mixing process. Isotopic and trace element evidence indicate that ~7-11% bulk subducted Chilean trench sediment has been added to an `E-MORB-like' pre-subduction mantle composition. This mantle mixture is further infiltrated by ~4% solute-rich fluid derived from the subducted Nazca basaltic oceanic crust. Trace element patterns of end-member `prevalent TSPC mantle' magmas are best fit by a two-stage partial melting model whereby the residual mantle, after a small degree melt extraction (F = 0.1%), undergoes F = 22% partial melting. The high melt fraction appears to potentially correspond with large volumes of solute-rich fluid released from the subducted Mocha Fracture Zone (MFZ). Although geochemical attributes of mafic TSPC magmas suggest that they are all derived from the same general mantle framework operating below the complex, one lava sequence appears to deviate. The mantle origins of the Upper Placeta San Pedro Sequence (UPSPS) have remained elusive since first being studied. It is a well-characterized basaltic lava series that erupted over a short interval at ~235-240 ka with highly variable incompatible element abundances and a large xenocrystic cargo. The new comprehensive chemical and radiogenic isotope (Sr-Nd-Pb-Hf) dataset, along with stratigraphic control and understanding of the larger-scale geochemical variability at the TSPC, provides fresh perspectives about the mantle sources and evolution of UPSPS magmas. While the UPSPS magmas are derived from the same sources as other TSPC magmas, they have undergone a unique petrogenetic evolution. This is evident from decoupled trace element-isotopic trends that are difficult to reconcile with other mafic TSPC magmas. Based on the Nd-Hf isotope ratios and trace element ratios, the two UPSPS unit magmas are derived from depleted-`TE enriched' and `prevalent TSPC mantle' sources. High ratios of aqueous fluid-mobile/immobile elements, such as high Pb/Ce and Sr/Nd, as well distinctive Sr and Pb isotope ratios, indicate that the UPSPS magmas were generated through fluxing of the mantle wedge, already depleted by melt removal by an solute-rich fluid derived from the subducted Pacific oceanic crust as well as the overlying trench sediment, which caused it to melt.

The model of rejuvenation of a near-solidus crystal mush to produce large volumes of crystal-rich magma is tested here by analyzing the mineralogical, chemical, modal, and physical characteristics of the 31.1 Ma super-eruptive (2000 km3) Cottonwood Wash Tuff. It is the oldest in a series of three so-called "monotonous intermediate" ignimbrites from the Indian Peak-Caliente volcanic field in southern Utah and Nevada. A crystal-rich (~50% Pl > Qz ≈ Hbl ≈ Bt > Mag ≈ Ilm > Cpx + Zrn + Ap+ Po) dacite (62 - 69 wt% SiO2), the Cottonwood Wash Tuff is similar in age, volume, mineralogy, crystallinity, and elemental composition to the 28.0 Ma, ~5000 km3 Fish Canyon Tuff (~45% Pl + Kfs + Qz + Hbl + Bt + Ttn + Mag + Ilm + Ap + Zrn + Po, 66 - 68 wt% SiO2), used as the basis of the rejuvenation model, which suggests that magma chambers remain in a near-solidus state until a late heating event melts the magma enough to allow eruption. The Cottonwood Wash magma chamber was compositionally varied, as shown by the composition of mineral and juvenile clast compositions. Most of the whole-rock compositional variations are likely due to the variation of mineral proportions induced by shear in the magma chamber. A volumetrically minor component with evolved mineral compositionss, is represented by "evolved" juvenile clasts. Mineral compositions and experimental phase relationships show the pre-eruption magma crystallized at 800°C, 2.3 kb under water-undersaturated but oxidized conditions (delta QFM = 2.1). The majority of plagioclase and amphibole grains exhibit small-scale oscillatory zonation; where systematic compositional zonation exists, normal and reverse zonation are equally present. Cathodoluminescence of quartz reveals typically normally zoned phenocrysts with late resorption, considered to be the result of eruptive decompression. Many of the characteristics used to identify the warming of a near-solidus mush for the Fish Canyon Tuff are not present in the Cottonwood Wash Tuff [i.e., reversely zoned hornblende or plagioclase, partially remelted mineral aggregates, evidence of fluid saturation, resorption textures not related to decompression, rapakivi mantles, and hybrid andesite inclusions]. The Cottonwood Wash Tuff magma system did not undergo rejuvenation from a near-solidus state. Instead, the magma was apparently cooling and crystallizing just prior to eruption.

Deuterium ratios (δD) of hydrated volcanic glass have been used to reconstruct paleoenvironments, although the reliability and proper sample preparation protocol have been debated. In this study, hydrated volcanic ash samples from the lee of the Cascades were prepared using two separate methods. Method 1 involves sonicating and rinsing samples with hydrochloric acid (HCl) followed by hand-selection of glass shards (125-212µm). Method 2 requires hydrochloric acid (HCl) and hydrofluoric acid (HF) abrasion as well as heavy liquid separation of shards (70-150µm). Method 2 produced more consistent results with decreased intra-replicate variability in both water content (-0.92 wt. %) and deuterium values (-2.5‰ δD). Method 2 δD values of ≥99% isotropic glass were also 2.5-10 % more negative relative to Method 1 values, with an increasing discrepancy with age (3.68-32.66 Ma). Method 2 results suggest volcanic glass did not re-equilibrate with modern water, based on 1) < 2‰ discrepancies between samples of the same ash flow taken from unique sample localities and 2) a ~20‰ difference between samples of different ages (~8 Ma apart) from the same locality. These results support the specified use of HF abrasion and heavy liquid separation on 70-150 µm glass shards to minimize the impact of contaminants on reconstructed paleowater δD values.

The model of rejuvenation of a near-solidus crystal mush to produce large volumes of crystal-rich magma is tested here by analyzing the mineralogical, chemical, modal, and physical characteristics of the 31.1 Ma super-eruptive (2000 km3) Cottonwood Wash Tuff. It is the oldest in a series of three so-called “monotonous intermediate” ignimbrites from the Indian Peak-Caliente volcanic field in southern Utah and Nevada. A crystal-rich (~50% Pl > Qz ≈ Hbl ≈ Bt > Mag ≈ Ilm > Cpx + Zrn + Ap+ Po) dacite (62 – 69 wt% SiO2), the Cottonwood Wash Tuff is similar in age, volume, mineralogy, crystallinity, and elemental composition to the 28.0 Ma, ~5000 km3 Fish Canyon Tuff (~45% Pl + Kfs + Qz + Hbl + Bt + Ttn + Mag + Ilm + Ap + Zrn + Po, 66 – 68 wt% SiO2), used as the basis of the rejuvenation model, which suggests that magma chambers remain in a near-solidus state until a late heating event melts the magma enough to allow eruption. The Cottonwood Wash magma chamber was compositionally varied, as shown by the composition of mineral and juvenile clast compositions. Most of the whole-rock compositional variations are likely due to the variation of mineral proportions induced by shear in the magma chamber. A volumetrically minor component with evolved mineral compositionss, is represented by “evolved” juvenile clasts. Mineral compositions and experimental phase relationships show the pre-eruption magma crystallized at 800°C, 2.3 kb under water-undersaturated but oxidized conditions (delta QFM = 2.1). The majority of plagioclase and amphibole grains exhibit small-scale oscillatory zonation; where systematic compositional zonation exists, normal and reverse zonation are equally present. Cathodoluminescence of quartz reveals typically normally zoned phenocrysts with late resorption, considered to be the result of eruptive decompression. Many of the characteristics used to identify the warming of a near-solidus mush for the Fish Canyon Tuff are not present in the Cottonwood Wash Tuff [i.e., reversely zoned hornblende or plagioclase, partially remelted mineral aggregates, evidence of fluid saturation, resorption textures not related to decompression, rapakivi mantles, and hybrid andesite inclusions]. The Cottonwood Wash Tuff magma system did not undergo rejuvenation from a near-solidus state. Instead, the magma was apparently cooling and crystallizing just prior to eruption.

Easter Island and Sala y Gomez are part of the Sala y Gomez Ridge, a broad band of high topography and scattered seamounts extending ESE from the East Pacific Rise. It has been proposed that the Sala y Gomez Ridge results from the movement of the Nazca Plate over a fixed melting spot in the mantle. To test this hypothesis volcanic rocks from Easter Island and Sala y Gomez were analyzed for their K-Ar ages and major element abundances. Subaerial Easter Island was constructed in three distinct episodes, occurring at 2.5 m.y., 0.9 m.y., and 0.4 m.y. ago. The youngest rocks on the island are the Roiho olivine basalts, and are probably less than 50,000 years old. Eruptive activity on Sala y Gomez was essentially contemporaneous with the early volcanism on Easter Island. No migration of volcanism with time is apparent along the Sala y Gomez Ridge, thus a major criterion of the melting spot hypothesis is not fulfilled. Volcanic rocks from Easter Island constitute a tholeiitic differentiation series; they are chemically similar to those from other islands situated near mid-ocean rise crests. The wide compositional spectrum is most likely the result of fractional crystallization from a basaltic parent liquid, though the data is ambiguous for the highly silicic differentiates. The youngest basalts possess more alkaline affinities which are probably not related to fractional crystallization from the earlier basalts. The alkaline nature of these rocks may be the result of a downward migration of the fusion zone with time, as the island moved eastward over a progressively thickening lithosphere. Volcanic rocks from Sala y Gomez belong to an alkali olivine basalt series. The fundamental chemical differences between the Easter Island and Sala y Gomez suites suggest that the two islands were not derived from a common source, as predicted by the melting spot hypothesis. The evidence does not support a melting spot origin for Easter Island, Sala y Gomez, and the Sala y Gomez Ridge. An alternative model involving diapiric intrusion and decompression melting of asthenosphere material along a major fracture in the Nazca Plate provides a better explanation for the data. Synchronous volcanism along the eastern extension of the Easter Island transform fault has given rise to the islands and seamounts on the Sala y Gomez Ridge. / Graduation date: 1975

Graduation date: 1999

Thesis (M.A.)--University of Texas at Austin, 1990. / Vita. Includes bibliographical references (leaves 96-101).