A Petrologic Model for Hydrothermal Metamorphism in the Portage Lake Volcanic Series, Northern Michigan
We developed a petrologic model to assess the timeline of hydrothermal metamorphism of the Portage Lake Volcanics series during the Proterozoic Eon. Altered basalts and conglomerates include a variety of mineral deposits that record the pressure, temperature, and compositional history of this hydrothermal fluid over one billion years ago. Mineralogical optical properties and microtextures are examined using petrographic microscopy in six samples from the Calumet and Hecla mines in Michigan, as well as one sample from Minnesota. Micro-Raman spectroscopy is used for the first time to confirm mineral identities in this geologic series. The results yield a three-stage paragenetic sequence that is identified by distinct equilibrium mineral assemblages. Grain size distributions depict the kinetic evolution of this system, where temperatures ranged between 150°C-340°C. An initial period of prograde metamorphism and heating was followed by periods of declining temperature until hydrothermal activity subsided. Compositional changes in the minerals indicate two significant changes, including an initial increase in the activity of Fe3+ in the hydrothermal solution followed by an equilibrium shift favoring calcite deposition. This petrologic model presents a framework for future experiments and calculations to constrain the reaction kinetics and heat flow in this ancient hydrothermal system.
The hydrothermal metamorphic process produces many of Earth’s economic mineral deposits through subsurface fluids that infiltrate and alter porous lithic bodies at elevated pressures and temperatures. While precipitating minerals throughout Earth’s crust and interior, hydrothermal fluids also transport heat and aid in the sequestration of volatile compounds, such as H2O and CO21. However, many of the ancient hydrothermal systems became dormant long ago, and only through the remnant mineralogical compositions can we evaluate their paleocharacteristics.
This hydrothermal system in Northern Michigan originated after the Midcontinent Rift began to split the North American continent about 1.1 billion years ago (Ga), forming the Lake Superior Basin along the northern boundary of the present-day Keweenaw Peninsula in Michigan. Upwelling magma in this rift deposited the basalts and inter-bedded conglomerate beds constituting the present-day Portage Lake Volcanic series (PLV) during a 2 to 3 million-year time span (~1.095 Ga). After magmatic activity ceased (~1.08 Ga.), the PLV were buried underneath eroding sediments in the subsiding Lake Superior basin. A period of regional contraction associated with Grenvillian compression finally closed the rift while reversing the southern margin fault (~1.06 Ga) and exhuming the PLV2.
The Keweenaw Fault now truncates the base of the PLV on the Keweenaw Peninsula as a northeast-striking 25-70° thrust fault sitting at the southern slope of the Lake Superior syncline3. At the northern end of the syncline, the PLV beds are exposed on Isle Royale in Lake Superior (Figure 1). Similar igneous beds extend for over 1,000 km around Lake Superior .
Since prehistoric times native copper has been mined in the PLV and the region was the U.S.A.’s principle source of copper during the late 19th and early 20th centuries. Field relationships and radiometric techniques date the native copper deposits between 1,060 and 1,050 Ma2, approximately the same time as a thrust fault was activated and hydrothermal fluids infiltrated the region5,6. The bulk of the native copper occurs in pores, fissures, and structures along a 45 km portion of the PLV, on the hanging wall of the Keweenaw Fault3. The deposits occur adjacent to other secondary minerals in the permeable parts of rocks, markedly in amygdules in basalts and in the inter-clast matrix in conglomerates.
Attempts to propose hydrothermal model have begun to resolve the complex variety of mineral assemblages found in the PLV. Analyses of the regional distribution of the metamorphic minerals confirmed that copper deposits and other minerals originated in a regional hydrothermal solution7. The PLV constitute a low-grade metamorphic terrain, ranging from a lower-temperature Zeolite facies to a higher-temperature Prehnite-Pumpellyite facies. A dehydration reaction following the fluid path plausibly explains this metamorphic gradient, where dehydrated epidote [Ca2 (Fe3+,Al)3(SiO4)3(OH)] formed in amygdules at deeper stratigraphic levels, whereas hydrated pumpellyite [Ca2(Mg,Fe2+)(Fe3+2,Al2)(SiO4)(Si2O7)(OH)2•(H2O)] formed in the middle section and chlorite [hydrous aluminosilicate] formed at the top8. Livnat (1983) proposed a series of metamorphic reactions, which yielded P-T equilibriums consistent with zeolite to prehnite-pumpellyite facies, with temperatures estimated between ≥ 150°C ± 10° at the top to <320°C at the bottom. However, different mineral assemblages appear laterally within the stratigraphy, and indicate that the fluid’s thermochemistry varied with its migratory pathway. Accounting for copper precipitation, a cooling fluid that circulated up-dip through the PLV cooling would have rehydrated and oxidized surrounding rocks, creating the necessary conditions for copper deposition [9,10].
Whereas previous studies considered the paragenesis of the PLV as a single event with regional variations in mineral composition, this study describes the changes in the fluid’s composition over time. The timeline of mineralization is depicted by visibly distinct stages in thin section, which define the sample’s stable mineral assemblages. Optical microscopy and micro-Raman spectroscopy wee used to examine this crystallization sequence and analyze the evolution of the system. In addition to micro-textural observations from the polarizing petrographic microscope, Raman spectroscopy helped to identify the unique spectral signals of several indistinguishable minerals (Figure 2). The confocal micro-Raman spectrometer in the Mineral Physics Laboratory at Northwestern uses a 457 nm solid-state diode-pumped laser, interfaced with an Olympus BX microscope. Raman scattered light is collected through a rectangular confocal aperture into a 0.3 meter focal length spectrograph (Andor Newton 303i). In the current measurements, a 1200 lines/mm grating was used to achieve ~2 cm-1 spectral resolution. The spectra were recorded using an Andor Newton DU970 electron multiplying CCD camera, chilled to -90C by thermoelectric cooling.
We analyzed six rock samples from the PLV, including three amygdulated basalts and three felsic conglomerates gathered from the Calumet and Hecla Mines outside of Calumet, MI (Figure 2). The three basalts originated at the Red Jacket Shaft, one of which (R93) was reportedly collected from one mile below the surface. The three conglomerate samples come from a separate site located less than a kilometer to the north in the Allouez conglomerate layer.
An intriguing pattern emerges in both the basalts and conglomerates, whereby the size of secondary hydrothermal minerals increases with distance from wall spaces in amygdules and veins. Although it appears in the polycrystalline aggregates in conglomerates, this pattern is most easily observed in amygdules, where concentric mineral sequences differentiate between regions with different mineral compositions and grain sizes. Our petrologic model correlates observed grain sizes, mineralogical composition, and distance from the wall surfaces with thermodynamic developments over time (Figure 3). Each mineralogical zone corresponds to a period of time when specific dissolved components crystallized into minerals inwards from the wall surface. The sequence therefore provides the basis for deciphering the thermodynamic conditions and solution chemistry during the hydrothermal regime.
Three amygdules were modeled in R91 and grain size was plotted against the distance from the wall (Figure 4). Larger calcite [CaCO3] and quartz [SiO2] grains predominantly occur closer to the center, although there are numerous exceptions. The incremental increase in average grain size corresponds to transitions from a pumpellyite-dominated outer region, to an epidote-dominated inner region, and finally a calcite-dominated interior. The five amygdules modeled in R93 showed a similar increase in grain size, corresponding to the transition from an outer spherulitic epidote population to an inner euhedral epidote population (Figure 4).
Grain size diagrams confirm the observed increase in grain size with distance from amygdule walls (Figure 4). The increase in size reflects longer uninterrupted crystallization periods associated with slower cooling rates11. The composition of the secondary minerals evolves with temperature. In all six samples a calcium-aluminosilicate mineral, either pumpellyite or spherulitic epidote, occurs first, followed by increasing amounts of epidote, while last and largest is calcite.
Evolution of Composition and Temperature
The mineralogical composition results from a chemical reaction between the unknown dissolved ionic content of the initial solution and the known bulk composition of the rocks through which the fluid moved. Using the changing chemical composition of the secondary minerals and the bulk composition of the surrounding rocks, the changes in the dissolved ionic composition of the fluid can be inferred.
Jolly and Smith’s electron microprobe data on basalts from this region suggest a similar CaO-FeO-MgO-Fe2O3-Al2O3-SiO2-H2O solution8. We observed minor albite precipitation in the basalts, implying that an additional Na2O component was present and late calcite precipitation again implies late CO2 activity. During the P-T shift from the lower-grade zeolite facies to the higher-grade prehnite-pumpellyite facies in a simplified CaO-Al2O3-SiO2-H2O-CO2 system, laumontite and prehnite undergo a phase transition with increasing temperature to wairakite and zoisite. However, epidote becomes stable during the transition to the prehnite-pumpellyite facies with the addition of Fe3+. However, if the Fe2+ is not oxidized then pumpellyite and chlorite may form12. Livnat’s microprobe data showed that many pumpellyite crystals contain both Fe2+ and Fe3+, but that regions richer in pumpellyite had higher levels of Fe2+, which suggests that the availability of Fe2+ controlled the preferential precipitation of calcium-aluminisilicates9.
A simple three-stage paragenetic sequence can approximate the hydrothermal system’s P-T-X evolution. Stage 1 of this sequence, closest to the wall, is characterized by small grain sizes and the presence of opaques in all but one sample. Alteration of the parent rock’s igneous feldspar, pyroxene, and olivine minerals probably began during diagenesis, when rising temperatures and pressures during burial would have induced the following reaction in feldspars8:
CaAl2Si2O8 (anorthite)+Na++Si4+èNaAlSi3O8 (albite)+Ca2++Al3+
As temperature and pressure further increased during burial, pyroxene and olivine grains would have broken down and supplied Fe2+ and Mg2+. The addition of a fluid at this point therefore explains the extensive serpentinization in the main bodies of the basalt samples. Stage 1 alteration continued in both the main rock bodies of basalts and conglomerates and along the rims of their pore spaces, where very fine-grained pumpellyite and spherulitic epidote began to replace albite and crystallize along the wall surfaces. Diagenesis in the conglomerate samples emplaced a siliceous fine-grained polycrystalline grain aggregate in between clasts, indicating that the solution was locally enriched in silica.
In sample R93, the unique relationship between pumpellyite and epidote crystals highlights the transition from stage 1 to stage 2 alteration, where two distinct epidote populations intergrow across a continuous transition. The first epidote population displays a spherulitic habit that characterizes other chlorite, iddingsite, and chalcedony deposits in these samples and indicative of certain viscosity and saturation levels at the time of nucleation13. Epidote, chalcedony, and chlorite crystals form radial spherulites, while iddingsite assumes a remarkable dendritic form in R93. However, the radial morphology is anomalous for epidote crystals, and it is therefore likely that spherulitic epidote pseudomorphed earlier lower-grade metamorphic minerals, such as zeolites, as P-T-X conditions evolved14.
The second epidote population in R93 formed euhedral columnar crystals, which are typical of epidote. This population also displayed a dramatic increase in grain size, indicating that nucleation kinetics stabilized and temperatures probably decreased (Figure 4). The optical properties of the euhedral epidote crystals distinguish them further. Their green color deepens while birefringence increases across the spherulitic-euhedral transition, reflecting increased partitioning of Fe3+ into the crystals14 (Figure 5). Additionally, the euhedral crystals are concentrically zoned in cross-polarized light, and EBSD data from the PLV found that epidote cores are more Fe-poor than their rims in the PLV15. Hence, the euhedral epidote crystals mark the evolution of iron in the local hydrothermal fluid.
A pumpellyite to epidote transition between stages 1 and 2 parallels the timing of the spherulitic-euhedral transition in some samples, which may relate to a proposed metamorphic reaction:
Ca4Al4Fe3+Fe2+Si6O23(OH)3·H2O (pumpellyite)èCa4Al4Fe3+2Si6O24(OH)2 (epidote) +2H2O+1/2H2
This oxidation and dehydration reaction is dependent upon the fugacity of O2 (fO2) in the solution, which would have oxidized Fe2+ to Fe3+ in favor of precipitating pistacitic epidote instead of pumpellyite8. If this reaction occurred, then the dramatic increase in epidote deposition could reflect a temporal increase in fO2. Such an increase in fO2 could be due to oxygen-depletion deeper beneath the surface, the influx of an oxygen-enriched fluid at this stage8,9. Alternatively, it is possible that pumpellyite precipitation was limited because the bulk Fe2+ content partitioned into serpentine and pumpellyite during stage 1. The cause of the epidote shift is further obscured in a less amygdulated and darker section of sample R93 (Figure 5). Here, stage 1 serpentinization indicates that fluid initially passed through ubiquitously, but the darker color and relative absence of spherulitic epidote indicate a compositional difference and the lack of pores indicates the restriction of fluid flow after stage 1. During stage 2, when fluids deposited euhedral epidote along with a brown-green residue in most of this sample’s amygdulated channels, pumpellyite was deposited instead in the dark zone. There is only one amygdule, especially large, in the dark zone, and it is filled with a major chlorite deposit, as well as calcite. The absence of epidote seems to be tied to the transportation capacity of the fluid. This indicates that the rehydration reactions that formed chlorites during stage 1 depleted the fluid in water content (XH2O) , effectively increasing its viscosity. Finally, stage 3 alteration is dominated by the deposition of large calcite crystals. The size reflects a further temperature decline and an increase in the activity of CO2, possibly related to the depletion of XH2O16.
The distinct mineralogical assemblages represent evolving equilibriums of phase stability in the hydrothermal portions of the samples. However, the rocks never fully equilibrated with the hydrothermal fluid and therefore the equilibriums are not perfectly described compositionally. Furthermore, variable porosity within the samples distributed secondary minerals non-uniformly throughout the samples. Accordingly, equilibrium assemblages are best described for the simplest observable system, which are the amygdules in the basalt samples.
When mineral abundances are compared between the 3 samples inside the amygdules, stage 1 equilibrium assemblages range from (1) pumpellyite + quartz to (2) epidote + quartz. Stage 2 equilibrium assemblages range from (1) pumpellyite + quartz to (2) epidote + quartz to (3) epidote + pumpellyite + quartz, but display a remarkable increase in the amount of epidote and decrease in pumpellyite. Stage 3 equilibrium assemblages consist of mostly calcite, but include quartz and minor chlorite. In addition to the amygdules, the parent rock bodies are also affected by stage 1 metamorphism, and their equilibrium assemblage ranges from (1) pumpellyite + chlorite to (2) pumpellyite + chlorite + epidote + albite. Most of these assemblages fit within the prehnite-pumpellyite facies, and the most profound transitions are the increase epidote during stage 2 and the increase in calcite in stage 3. The general shift in conglomerates from quartz deposition during stages 1 and 2 to calcite and quartz during stage 3, along with the dendritic iddingsite and chlorite inside the large anamalous amygdule in R93, reassure us that retrograde metamorphism occurred through decreasing P-T space17.
While previous studies calculated the temperature range of this system between 150°C and 320°C7.9, mineralogical geothermometers from a comparable and active hydrothermal basalt system in the Philippines compares these theoretical values to real-world values18. Using these geothermometric minerals, Figure 6 indicates a similar temperature range to the theoretical calculations, lying between 80°C and 350°C. The three paragentic stages display a broad cooling trend, associated with the observation of increased grain size. While this model depicts the temperature trend, it does not capture the local temperature discontinuities that corresponding with abrupt second stage grain size increases in R91 and third stage grain size increases in R93 in Figure 4. Rather, it may be interpreted to approximate temperatures in the PLV region during the hydrothermal regime.
Although the old age of this hydrothermal system makes it challenging to constrain the origin of this fluid, the inferred thermochemistry offers some constraints. Native copper occurred only in the conglomerate samples, where it is deposited as a small inclusion inside some quartz grains, but is found more frequently adjacent to quartz, calcite, or epidote in pore spaces, and thus formed during stages 2 and 3. The formation of native copper helps constrain the thermochemistry of the hydrothermal solution, because it requires a decrease in the temperature and reduction potential of the solution9,10. The conditions for native copper deposition could be related to epidote and pumpellyite if fO2 controls their formation. This would imply that the oxidizing hydrothermal solution traveled through a gradient from a reduced environment into to an oxidized environment, where its reduction potential lowered and native copper was deposited along with epidote. If this scenario is accurate, then it constrains the origins of the hydrothermal solution to below the surface, where oxygen levels are generally low.
The typical architecture of a hydrothermal network involves a heated deep-seated fluid transporting material into a shallower and cooler depositional environment18. While a magmatic fluid source might explain the heating, isotopic evidence does not support the theory9,15. Contact metamorphism from the south could have provided some heating, which would explain the north-to-south trend line of minerals increasing in metamorphic grade15. However, this trend can also be explained by the tilting of the stratigraphy during thrust faulting.
A regional metamorphic scheme explains the regional mineral distribution and its relationship to the stratigraphy through burial during diagenesis and upwelling during the reversal of the Keweeenaw Fault. If fluid infiltrated the rocks during diagenesis, exothermic serpentinization reactions may have produced some of the heat for the subsequent hydrothermal fluids. The currently exposed sections of PLV all lie in the hanging wall of the fault, but footwall section of the PLV is still buried today beneath the Jacobsville Sandstone. When the Keweenaw Fault bisected the original PLV bed over one billion years ago, dehydration reactions in the footwall, which are typically associated with prograde metamorphism in down-going slabs, may have released hot fluids, which travelled upwards to mix and heat meteoric fluids, resulting in the hydrothermal deposits observed today.
In order to assess the validity of this theory, more precise chronology must determine the timespan of the hydrothermal system, and whether alteration occurred before, during, or after thrust faulting. In order to more precisely describe the mineralogical composition and P-T evolution, Mossbauer spectroscopy could verify oxidation state of iron in the epidote populations and surrounding minerals. Metamorphic reactions that fully account for the observed equilibrium assemblages could provide the basis for calculating the Gibbs Free Energy, and in turn constructing phase diagrams through which to monitor the evolution of the system through P-T space. Imposing precise constraints upon the system’s P-T-X parameters will elucidate the burial depth and help eliminate ambiguities surrounding the origin and pathway of this hydrothermal fluid. Ultimately, this petrologic model can be applied to other amygdulated hydrothermal deposits throughout the earth’s crust, and may help to better understand the way that mineral resources, and in fact, life on Earth, have evolved over time.