Detection and impacts of leakage from sub-seafloor deep geological carbon dioxide storage

Fossil fuel power generation and other industrial emissions of carbon dioxide are a threat to global climate1, yet many economies will remain reliant on these technologies for several decades2. Carbon dioxide capture and storage (CCS) in deep geological formations provides an effective option to remove these emissions from the climate system3. In many regions storage reservoirs are located offshore4, 5, over a kilometre or more below societally important shelf seas6. Therefore, concerns about the possibility of leakage7, 8 and potential environmental impacts, along with economics, have contributed to delaying development of operational CCS. Here we investigate the detectability and environmental impact of leakage from a controlled sub-seabed release of CO2. We show that the biological impact and footprint of this small leak analogue (<1 tonne CO2 d−1) is confined to a few tens of metres. Migration of CO2 through the shallow seabed is influenced by near-surface sediment structure, and by dissolution and re-precipitation of calcium carbonate naturally present in sediments. Results reported here advance the understanding of environmental sensitivity to leakage and identify appropriate monitoring strategies for full-scale carbon storage operations.

Detection and impacts of leakage from sub-seafloor deep geological carbon dioxide storage

Fossil fuel power generation and other industrial emissions of carbon dioxide are a threat to global climate1, yet many economies will remain reliant on these technologies for several decades2. Carbon dioxide capture and storage (CCS) in deep geological formations provides an effective option to remove these emissions from the climate system3. In many regions storage reservoirs are located offshore4, 5, over a kilometre or more below societally important shelf seas6. Therefore, concerns about the possibility of leakage7, 8 and potential environmental impacts, along with economics, have contributed to delaying development of operational CCS. Here we investigate the detectability and environmental impact of leakage from a controlled sub-seabed release of CO2. We show that the biological impact and footprint of this small leak analogue (<1 tonne CO2 d−1) is confined to a few tens of metres. Migration of CO2 through the shallow seabed is influenced by near-surface sediment structure, and by dissolution and re-precipitation of calcium carbonate naturally present in sediments. Results reported here advance the understanding of environmental sensitivity to leakage and identify appropriate monitoring strategies for full-scale carbon storage operations.

Global Rates of Free Hydrogen (H2) Production by Serpentinization and other Abiogenic Processes within Young Ocean Crust

The main conclusion of this dissertation is that global H2 production within young ocean crust (<10 Mya) is higher than currently recognized, in part because current estimates of H2 production accompanying the serpentinization of peridotite may be too low (Chapter 2) and in part because a number of abiogenic H2-producing processes have heretofore gone unquantified (Chapter 3). The importance of free H2 to a range of geochemical processes makes the quantitative understanding of H2 production advanced in this dissertation pertinent to an array of open research questions across the geosciences (e.g. the origin and evolution of life and the oxidation of the Earth’s atmosphere and oceans).

The first component of this dissertation (Chapter 2) examines H2 produced within young ocean crust [e.g. near the mid-ocean ridge (MOR)] by serpentinization. In the presence of water, olivine-rich rocks (peridotites) undergo serpentinization (hydration) at temperatures of up to ~500°C but only produce H2 at temperatures up to ~350°C. A simple analytical model is presented that mechanistically ties the process to seafloor spreading and explicitly accounts for the importance of temperature in H2 formation. The model suggests that H2 production increases with the rate of seafloor spreading and the net thickness of serpentinized peridotite (S-P) in a column of lithosphere. The model is applied globally to the MOR using conservative estimates for the net thickness of lithospheric S-P, our least certain model input. Despite the large uncertainties surrounding the amount of serpentinized peridotite within oceanic crust, conservative model parameters suggest a magnitude of H2 production (~1012 moles H2/y) that is larger than the most widely cited previous estimates (~1011 although previous estimates range from 1010-1012 moles H2/y). Certain model relationships are also consistent with what has been established through field studies, for example that the highest H2 fluxes (moles H2/km2 seafloor) are produced near slower-spreading ridges (<20 mm/y). Other modeled relationships are new and represent testable predictions. Principal among these is that about half of the H2 produced globally is produced off-axis beneath faster-spreading seafloor (>20 mm/y), a region where only one measurement of H2 has been made thus far and is ripe for future investigation.

In the second part of this dissertation (Chapter 3), I construct the first budget for free H2 in young ocean crust that quantifies and compares all currently recognized H2 sources and H2 sinks. First global estimates of budget components are proposed in instances where previous estimate(s) could not be located provided that the literature on that specific budget component was not too sparse to do so. Results suggest that the nine known H2 sources, listed in order of quantitative importance, are: Crystallization (6x1012 moles H2/y or 61% of total H2 production), serpentinization (2x1012 moles H2/y or 21%), magmatic degassing (7x1011 moles H2/y or 7%), lava-seawater interaction (5x1011 moles H2/y or 5%), low-temperature alteration of basalt (5x1011 moles H2/y or 5%), high-temperature alteration of basalt (3x1010 moles H2/y or <1%), catalysis (3x108 moles H2/y or <<1%), radiolysis (2x108 moles H2/y or <<1%), and pyrite formation (3x106 moles H2/y or <<1%). Next we consider two well-known H2 sinks, H2 lost to the ocean and H2 occluded within rock minerals, and our analysis suggests that both are of similar size (both are 6x1011 moles H2/y). Budgeting results suggest a large difference between H2 sources (total production = 1x1013 moles H2/y) and H2 sinks (total losses = 1x1011 moles H2/y). Assuming this large difference represents H2 consumed by microbes (total consumption = 9x1011 moles H2/y), we explore rates of primary production by the chemosynthetic, sub-seafloor biosphere. Although the numbers presented require further examination and future modifications, the analysis suggests that the sub-seafloor H2 budget is similar to the sub-seafloor CH4 budget in the sense that globally significant quantities of both of these reduced gases are produced beneath the seafloor but never escape the seafloor due to microbial consumption.

The third and final component of this dissertation (Chapter 4) explores the self-organization of barchan sand dune fields. In nature, barchan dunes typically exist as members of larger dune fields that display striking, enigmatic structures that cannot be readily explained by examining the dynamics at the scale of single dunes, or by appealing to patterns in external forcing. To explore the possibility that observed structures emerge spontaneously as a collective result of many dunes interacting with each other, we built a numerical model that treats barchans as discrete entities that interact with one another according to simplified rules derived from theoretical and numerical work, and from field observations: Dunes exchange sand through the fluxes that leak from the downwind side of each dune and are captured on their upstream sides; when dunes become sufficiently large, small dunes are born on their downwind sides (“calving”); and when dunes collide directly enough, they merge. Results show that these relatively simple interactions provide potential explanations for a range of field-scale phenomena including isolated patches of dunes and heterogeneous arrangements of similarly sized dunes in denser fields. The results also suggest that (1) dune field characteristics depend on the sand flux fed into the upwind boundary, although (2) moving downwind, the system approaches a common attracting state in which the memory of the upwind conditions vanishes. This work supports the hypothesis that calving exerts a first order control on field-scale phenomena; it prevents individual dunes from growing without bound, as single-dune analyses suggest, and allows the formation of roughly realistic, persistent dune field patterns.

A new period of volcanogenic massive sulfide formation in the Yilgarn: A volcanological study of the ca 2.76 Ga Hollandaire VMS deposit, Yilgarn Craton, Western Australia

The Archean Hollandaire volcanogenic massive sulfide deposit is a felsic–siliciclastic VMS deposit located in the Murchison Domain of the Youanmi Terrane, Yilgarn Craton, Western Australia. It is hosted in a succession of turbidites, mudstones and coherent rhyodacite sills and has been metamorphosed to upper greenschist/lower amphibolite facies and includes a pervasive S1 deformational fabric. The coherent rhyodacitic sills are interpreted as syndepositional based on geochemical similarities with well-known VMS-associated felsic rocks and similar foliations to the metasediments. We offer several explanations for the absence of textural evidence (e.g. breccias) for syn-depositional origins: 1) the subaqueous sediments were dehydrated by long-lived magmatism such that no pore-water remained to drive quench fragmentation; 2) pore-space occlusion by burial and/or, 3) alteration overprinting and obscuring of primary breccias at contact margins. Mineralisation occurs by sub-seafloor replacement of original host rocks in two ore bodies, Hollandaire Main (~125 x >500 m and ~8 m thick) and Hollandaire West (~100 x 470 m and ~5 m thick), and occurs in three main textural styles, massive sulfides, which are exclusively hosted in turbidites and mudstones, and stringer and disseminated sulfides, which are also hosted in coherent rhyodacite. Most sulfides have textures consistent with remobilisation and recrystallisation. Hydrothermal metamorphism has altered the hangingwall and footwall to similar degrees, with significant gains in Mg, Mn and K and losses in Na, Ca and Sr. Garnet and staurolite porphyryoblasts also exhibit a footprint around mineralisation, extending up to 30 m both above and below the ore zone. High precision thermal ionisation mass spectrometry of zircons extracted from the coherent rhyodacite yield an age of 2759.5 ± 0.9 Ma, which along with geochemical comparisons, places the succession within the 2760–2735 Ma Greensleeves Formation of the Polelle Group of the Murchison Supergroup. Geochemical and geochronological evidence link the coherent rhyodacite sills to the Peter Well Granodiorite pluton ~2 km to the W, which acted as the heat engine driving hydrothermal circulation during VMS mineralisation. This study highlights the importance of both: detailed physical volcanological studies from which an accurate assessment of timing relationships, particularly the possibility of intrusions dismembering ore horizons, can be made; and identifying synvolcanic plutons and other similar suites, for VMS exploration targets in the Youanmi Terrane and worldwide.

Chemistry of altered basaltic glass shards and zeolites of volcaniclastic sediments from ODP Hole 157-953C

Trace element concentrations of altered basaltic glass shards (layer silicates) and zeolites in volcaniclastic sediments drilled in the volcanic apron northeast of Gran Canaria during Ocean Drilling Program (ODP) leg 157 document variable element mobilities during low-temperature alteration processes in a marine environment. Clay minerals (saponite, montmorillonite, smectite) replacing volcanic glass particles are enriched in transition metals and rare earth elements (REE). The degree of retention of REE within the alteration products of the basaltic glass is correlated with the field strength of the cations. The high field-strength elements are preferentially retained or enriched in the alteration products by sorption through clay minerals. Most trace elements are enriched in a boundary layer close to the interface mineral-altered glass. This boundary layer has a key function for the physico-chemical conditions of the subsequent alteration process by providing a large reactive surface and by lowering the fluid permeability. The release of most elements is buffered by incorporation into secondary precipitates (sodium-rich zeolites, phillipsite, Fe- and Mn-oxides) as shown by calculated distribution coefficients between altered glasses and authigenic minerals. Chemical fluxes change from an open to a closed system behavior during prograde low-temperature alteration of volcaniclastic sediments with no significant trace metal flux from the sediment to the water column.

Strontium and boron geochemistry of ODP sites at Hydrate Ridge, eastern Pacific Ocean

ODP Leg 204, which drilled at Hydrate Ridge, provides unique insights into the fluid regime of an accretionary complex and delineates specific sub-seafloor pathways for fluid transport. Compaction and dewatering due to smectite-illite transition increase with distance from the toe of the accretionary prism and bring up fluids from deep within the accretionary complex to sampled depths (<= 600 mbsf). These fluids have a distinctly non-radiogenic strontium isotope signature indicating reaction with the oceanic basement. Boron isotopes are also consistent with a deep fluid source that has been modified by desorption of heavy boron as clay minerals change from smectite to illite. One of three major horizons serves as conduit for the transport of mainly fluid. Our results enable us to evaluate fluid migration pathways that play important roles on massive gas hydrate accumulations and seepage of methane-rich fluids on southern Hydrate Ridge.

(Table 2) Concentrations and D/L values for three amino acids in bulk foraminifera from ODP Sites 182-1126 and 182-1127

A number of studies have shown that methanogens are active in the presence of sulfate under some conditions. This phenomenon is especially exemplified in carbonate sediments of the southern Australian continental margin. Three sites cored during Ocean Drilling Program (ODP) Leg 182 in the Great Australian Bight have high concentrations of microbially-generated methane and hydrogen sulfide throughout almost 500 m of sediments. In these cores, the sulfate-reducing and methanogenic zones overlap completely; that is, the usual sulfate-methane transition zone is absent. Amino acid racemization data show that the gassy sediments consist of younger carbonates than the low-gas sites. High concentrations of the reduced gases also occur in two ODP sites on the margin of the Bahamas platform, both of which have similar sedimentary conditions to those of the high-gas sites of Leg 182. Co-generation of these reduced gases results from an unusual combination of conditions, including: (1) a thick Quaternary sequence of iron-poor carbonate sediments, (2) a sub-seafloor brine, and (3) moderate amounts of organic carbon. The probable explanation for the co-generation of hydrogen sulfide and methane in all these sites, as well as in other reported environments, is that methanogens are utilizing non-competitive substrates to produce methane within the sulfate-reducing zone. Taken together, these results form the basis of a new model for sulfate reduction and methanogenesis in marine sediments. The biogeochemical end-members of the model are: (1) minimal sulfate reduction, (2) complete sulfate reduction followed by methanogenesis, and (3) overlapping sulfate reduction and methanogenesis with no transition zone.

(Table 1) Determined and theoretical chemical composition of thaumasite of ODP Hole 135-841B

Hole 841B was drilled in the forearc region of the Lau Basin at a water depth of 4810 m. The hole penetrated a roughly 500-m-thick series of Miocene volcanic sediments with a number of basaltic to andesitic units (sills?) varying in thickness between 7 cm and 17 m. The volcanics are slightly to moderately altered and contain analcite, chabazite, natrolite-thompsonite, heulandite (?), prehnite, and quartz as secondary phases. In addition, thaumasite [Ca3Si(OH)6 * 12H2O](SO4)(CO3) was identified in the altered sequence. Sulfur isotope data of two thaumasite separates (+23.5 per mil and +21.1 per mil d34S) indicate a seawater origin of the sulfate sulfur. It is suggested that thaumasite is a product of low-temperature (<60 °C), seawater-derived CaCl2-rich fluids that were almost identical in composition to those presently circulating in the sub-seafloor.

Rare earth element concentrations and geochemistry of carbonate minerals from the Logatchev Hydrothermal Field and the Gakkel Ridge

The ultramafic-hosted Logatchev Hydrothermal Field (LHF) at 15°N on the Mid-Atlantic Ridge and the Arctic Gakkel Ridge (GR) feature carbonate precipitates (aragonite, calcite, and dolomite) in voids and fractures within different types of host rocks. We present chemical and Sr isotopic compositions of these different carbonates to examine the conditions that led to their formation. Our data reveal that different processes have led to the precipitation of carbonates in the various settings. Seawater-like 87Sr/86Sr ratios for aragonite in serpentinites (0.70909 to 0.70917) from the LHF are similar to those of aragonite from the GR (0.70912 to 0.70917) and indicate aragonite precipitation from seawater at ambient conditions at both sites. Aragonite veins in sulfide breccias from LHF also have seawater-like Sr isotope compositions (0.70909 to 0.70915), however, their rare earth element (REE) patterns show a clear positive europium (Eu) anomaly indicative of a small (< 1%) hydrothermal contribution. In contrast to aragonite, dolomite from the LHF has precipitated at much higher temperatures (~100 °C), and yet its 87Sr/86Sr ratios (0.70896 to 0.70907) are only slightly lower than those of aragonite. Even higher temperatures are calculated for the precipitation of deformed calcite veins in serpentine-talc fault schists form north of the LHF. These calcites show unradiogenic 87Sr/86Sr ratios (0.70460 to 0.70499) indicative of precipitation from evolved hydrothermal fluids. A simple mixing model based on Sr mass balance and enthalpy conservation indicates strongly variable conditions of fluid mixing and heat transfers involved in carbonate formation. Dolomite precipitated from a mixture of 97% seawater and 3% hydrothermal fluid that should have had a temperature of approximately 14 °C assuming that no heat was transferred. The much higher apparent precipitation temperatures based on oxygen isotopes (~ 100 °C) may be indicative of conductive heating, probably of seawater prior to mixing. The hydrothermal calcite in the fault schist has precipitated from a mixture of 67% hydrothermal fluid and 33% seawater, which should have had an isenthalpic mixing temperature of ~ 250 °C. The significantly lower temperatures calculated from oxygen isotopes are likely due to conductive cooling of hydrothermal fluid discharging along faults. Rare earth element patterns corroborate the results of the mixing model, since the hydrothermal calcite, which formed from waters with the greatest hydrothermal contribution, has REE patterns that closely resemble those of vent fluids from the LHF. Our results demonstrate, for the first time, that (1) precipitation from pure seawater, (2) conductive heating of seawater, and (3) conductive cooling of hydrothermal fluids in the sub-seafloor all can lead to carbonate precipitation within a single ultramafic-hosted hydrothermal system.

(Table S1) Trace element and isotopic compositions of anhydrites sampled in the Manus Basin, Papua New Guinea

Microchemical analyses of rare earth element (REE) concentrations and Sr and S isotope ratios of anhydrite are used to identify sub-seafloor processes governing the formation of hydrothermal fluids in the convergent margin Manus Basin, Papua New Guinea. Samples comprise drill-core vein anhydrite and seafloor massive anhydrite from the PACMANUS (Roman Ruins, Snowcap and Fenway) and SuSu Knolls (North Su) active hydrothermal fields. Chondrite-normalized REE patterns in anhydrite show remarkable heterogeneity on the scale of individual grains, different from the near uniform REEN patterns measured in anhydrite from mid-ocean ridge deposits. The REEN patterns in anhydrite are correlated with REE distributions measured in hydrothermal fluids venting at the seafloor at these vent fields and are interpreted to record episodes of hydrothermal fluid formation affected by magmatic volatile degassing. 87Sr/86Sr ratios vary dramatically within individual grains between that of contemporary seawater and that of endmember hydrothermal fluid. Anhydrite was precipitated from a highly variable mixture of the two. The intra-grain heterogeneity implies that anhydrite preserves periods of contrasting hydrothermal versus seawater dominant near-seafloor fluid circulation. Most sulfate d34S values of anhydrite cluster around that of contemporary seawater, consistent with anhydrite precipitating from hydrothermal fluid mixed with locally entrained seawater. Sulfate d34S isotope ratios in some anhydrites are, however, lighter than that of seawater, which are interpreted as recording a source of sulfate derived from magmatic SO2 degassed from underlying felsic magmas in the Manus Basin. The range of elemental and isotopic signatures observed in anhydrite records a range of sub-seafloor processes including high-temperature hydrothermal fluid circulation, varying extents of magmatic volatile degassing, seawater entrainment and fluid mixing. The chemical and isotopic heterogeneity recorded in anhydrite at the inter- and intra-grain scale captures the dynamics of hydrothermal fluid formation and sub-seafloor circulation that is highly variable both spatially and temporally on timescales over which hydrothermal deposits are formed. Microchemical analysis of hydrothermal minerals can provide information about the temporal history of submarine hydrothermal systems that are variable over time and cannot necessarily be inferred only from the study of vent fluids.