Volume 108, Issue 6

Understanding past and present aqueous activity on Mars is critical to constraining martian aqueous geochemistry and habitability, and to searching for life on Mars. Assemblages of minerals observed at or near the martian surface include phyllosilicates, sulfates, iron oxides/hydroxides, and chlorides, all of which are indicative of a complex history of aqueous activity and alteration in the martian past. Furthermore, features observed on parts of the martian surface suggest present-day activity of subsurface brines and at least transient liquid water. Terrestrial analogs for younger and colder (Hesperian–Amazonian) martian geologic and climatic conditions are available in the McMurdo Dry Valleys (MDV) of Antarctica and provide opportunities for improved understanding of more recent aqueous activity on Mars. Here, we study the VXE-6 intermittent brine pond site from Wright Valley in the MDV region and use coordinated spectroscopy, X-ray diffraction, and elemental analyses to characterize the mineralogy and chemistry of surface sediments that have evolved in response to aqueous activity at this site. We find that brine pond activity results in mineral assemblages akin to aqueous alteration products associated with younger sites on Mars. In particular, surficial chlorides, a transition layer of poorly crystalline aluminosilicates and iron oxides/hydroxides, and a deeper gypsum-rich interval within the upper 10 cm of sediment are closely related at this Antarctic brine pond site. Activity of the Antarctic brine pond and associated mineral formation presents a process analog for chemical alteration on the martian surface during episodes of transient liquid water activity during the late Hesperian and/or more recently. Our results provide a relevant example of how aqueous activity in a cold and dry Mars-like climate may explain the co-occurrence of chlorides, clays, iron oxides/hydroxides, and sulfates observed on Mars.

The solubility, speciation, and local atomic environment of chlorine have been determined for aluminoborosilicate glasses equilibrated with various sources of chlorine (NaCl and PdCl 2 ) at high pressure (0.5–1.5 GPa) and high temperature (1350–1400 °C). The Cl solubility reaches up to 11 mol% in borosilicate glass and appears to be strongly influenced by the concentration of network-modifying cations (Ca and Na) and increases with increasing CaO + Na 2 O content. The Cl solubility is enhanced in Ca-bearing rather than Na-bearing borosilicate glass, suggesting a higher affinity of chlorine for alkaline-earth cations. Cl K -edge XANES and Cl 2p XPS spectra reveal that chlorine dissolves in glasses only as chloride species (Cl – ) and no evidence of oxidized species is observed. Using PdCl 2 as a chlorine source leads to a pre-edge signal for PdCl 2 in the XANES spectra. The EXAFS simulations show that the Cl – local environment is charge compensated by Na + or Ca 2+ at a distance to first neighbor on the order of 2.7 Å, which is comparable to the observed distances in crystalline chloride compounds. The coordination to charge compensating cation is lower in the case of Ca 2+ (~1.1) than Na + (~4.3).

Apatite containing 14 wt% TREO (total rare earth oxide) and coexisting with calciobritholite with 37.2 wt% TREO has been synthesized at 800 °C and 10 kbar from a felsic melt with the addition of NaCl. The analysis of the experimental products with regression analysis of time-resolved (RATR) laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) data allowed to estimate the composition of the coexisting phases. The results suggest that equilibrium has been established during the run and both apatite and calciobritholite contained REE in [Si 4+ REE 3+ ] to [Ca 2+ P 5+ ] solid solution, whereas the coupled substitution [Na 1+ REE 3+ ] to [2Ca 2+ ] was insignificant despite crystallization from an alkaline, Na-rich melt. The coexistence of the apatite and calciobritholite and available experimental data allowed the miscibility gap to be constrained between apatite and calciobritholite, and suggest complete miscibility between apatite and britholite above 950 °C. The melt that produced coexisting apatite and calciobritholite was characterized by a significant Cl content of (0.51 wt%) and elevated REE (526 ± 19 ppm Ce) and low-P content (112 ± 49 ppm). The change of the accessory mineral association from monazite to apatite and calciobritholite with the addition of NaCl illustrates the importance of halogens for mineral associations. The partition coefficients of britholite are similar to those of apatite and are distinguished mainly by a higher preference for REE and Th. Henry’s law was not acting for the total REE content in the melt because of the buffered system; however the partition coeficients could still be used for the prediction of the relative REE patterns for melts that generated high-REE apatite and/or calciobritholite. These results have implications for the interpretation of the phosphate associations in alkaline volcanic and plutonic rocks.

The thermal behavior of 15 natural tourmaline samples has been measured by X-ray powder diffraction from room temperature to ~930 °C. Axial thermal expansion is generally greater along the c crystallographic axis (α c 0.90–1.05 × 10 –5 /K) than along the a crystallographic axis and the symmetrically equivalent b axis (α a 0.47–0.60 × 10 –5 /K). Ferro-bearing samples show lower expansion along a than in other tourmalines. In povondraite the thermal expansion along the c axis is higher than in other tourmalines, whereas along a it is lower [α a = 0.31(2) and α c = 1.49(3) × 10 –5 /K]. Volume expansion in the tourmaline-supergroup minerals is relatively low compared with other silicates such as pyroxenes and amphiboles. Volume also exhibits a relatively narrow range of thermal expansion coefficients (1.90–2.05 × 10 –5 /K) among the supergroup members. An interpretation for the small changes in thermal expansion in a compositionally heterogeneous group like tourmaline is that all members, except povondraite, share a framework of dominantly Z AlO 6 polyhedra that limit thermal expansion. Povondraite, with a framework dominated by Z Fe 3+ O 6 polyhedra, displays thermal expansion that is different from other members of the group. Unit-cell dimensions of tourmalines having significant Fe 2+ deviate from linearity above 400 °C on plots against temperature ( T ); along with the resulting substantial reduction in unit-cell volume, these efects are likely the result of deprotonation/oxidation processes. Lithium-rich and Fe 2+ -free tourmalines deviate similarly at T > 600 °C. In Li- and Fe 2+ -free tourmalines, no such deviation is observed up to the highest temperatures of our experiments. It is not clear whether this is due to cation order-disorder over Y and Z sites that occurs during the highest temperature measurements, a phenomenon that is apparently inhibited (at least in the short term) in Li-free/Mg-rich samples. If so, this must occur at a relatively rapid rate, as no difference in unit-cell values was detected at 800 °C after heating in both one- and 12-h experiments on Na-rich rossmanite.

Diffusivity in iron (Fe) alloys at high pressures and temperatures imposes constraints on the transport properties of the inner core, such as viscosity. Because silicon (Si) is among the most likely candidates for light elements in the inner core, the presence of Si must be considered when studying difusivity in the Earth’s inner core. In this study, we conducted diffusion experiments under pressures up to about 50 GPa using an internal-resistive-heated diamond-anvil cell (DAC) that ensures stable and homogeneous heating compared with a conventional laser-heated DAC and thus allows us to conduct more reliable difusion experiments under high pressure. We determined the coeficients of Fe–nickel (Ni) interdiffusion in the Fe–Si 2 wt% alloy. The obtained difusion coeficients follow a homologous temperature relationship derived from previous studies without considering Si. This indicates that the efect of Si on Fe–Ni interdifusion is not significant. The upper limit of the viscosity of the inner core inferred from our results is low, indicating that the Lorentz force is a plausible mechanism to deform the inner core.

The environment model is used to describe the location of carbonate in nine carbonated apatites containing varied percentages of carbonate and Na + , K + , or NH 4 + ions. Unlike the traditional model for carbonate substitution, which identifies different locations and orientations of the carbonate ion in the apatite structure, the environment model utilizes the different structural surroundings to describe the different types of carbonate. The A-type carbonate environment is assigned to channels lined only with calcium ions (A-channel configuration = Ca6) or to channels containing one Na + or a vacancy (A′-channel configuration = Ca5Na or Ca5▢), and the B-type carbonate environment is the surroundings of the replaced phosphate ion. The assignments are made by peak-fitting the carbonate asymmetric stretch region (ν 3 ) of the IR spectrum, following previously published criteria. These assignments lead to the conclusion that the percentage of channel carbonate (A- and A′-environments) is greater than that of B-type for each of these carbonated apatites. In general, the use of triammonium phosphate as the phosphate source in the synthesis produces apatites with larger amounts of channel carbonate (A- and A′-environments), while the use of sodium-containing phosphate reagents produces smaller amounts of channel carbonate. The environment model provides explanations for the differences within IR and NMR spectra obtained for apatites containing a range of total carbonate content. The B-type appearance of the carbonate ν 3 region of the IR spectrum is found primarily in apatites containing sodium, which allows increased amounts of carbonation via co-substitution of Na + with carbonate and creation of A′-environments with populations equal to that of B-type carbonate. The presence of ammonium or alkali metal salts with cations larger than Na + results in the utilization of a charge-balance mechanism that produces vacancies rather than cation substitution in the channel. The carbonated apatites formed with primary utilization of the vacancy mechanism generally contain greater percentages of carbonate in the A-environment and carbonate IR spectra that contain an obvious high-frequency peak at about 1550 cm –1 . The multiple peaks in the solid state 13 C NMR spectra previously observed for carbonated apatite are attributed to substitution in the A-, A′-, and B-environments rather than different stereo-chemical orientations of the carbonate ion.

Rozenite (FeSO 4 ·4H 2 O) is a candidate mineral component of the polyhydrated sulfate deposits on the surface and in the subsurface of Mars. To better understand its behavior at temperature conditions prevailing on the Martian surface and aid its identification in ongoing and future Rover missions, we have carried out a combined experimental and computational study of the mineral’s structure and properties. We collected neutron powder difraction data at temperatures ranging from 21–290 K, room-temperature synchrotron X-ray data and Raman spectra. Moreover, first-principles calculations of the vibrational properties of rozenite were carried out to aid the interpretation of the Raman spectra. We found, in contrast to a recent Raman spectroscopic study, that there are no phase transitions between 21 and 290 K. We confirm the heavy atom structure reported in the literature (space group P 21/ n ) to be correct and present, for the first time, an unconstrained determination of the H atom positions by means of high-resolution neutron powder diffraction, and report the complete crystal structure at 290 and 21 K. The anisotropy of the thermal expansion of the lattice vectors is α a :α b :α c = 1.00:2.19:1.60 at 285 K. Subsequent analysis of the thermal expansion tensor revealed highly anisotropic behavior as reflected in negative thermal expansion approximately ||〈101⟩ and ratios of the tensor eigenvalues of α 1 :α 2 :α 3 = −1:3.74:5.40 at 285 K. Lastly, we demonstrated how combining Raman spectroscopy and X-ray difraction of the same sample sealed inside a capillary with complementary first-principles calculations yields accurate reference Raman spectra. This workflow enables the construction of a reliable Raman spectroscopic database for planetary exploration, which will be invaluable to shed light on the geological past as well as in identifying resources for the future colonization of planetary bodies throughout the solar system.

Hydrothermal rutile (TiO 2 ) is a widely distributed accessory mineral in hydrothermal veins or alteration assemblages of porphyry deposits and provides important information for further understanding hydrothermal fluid signatures. This study determines the geochemical composition and U-Pb dates of hydrothermal rutile from the Yulong porphyry Cu-Mo deposit in east Tibet, China. Three types of TiO 2 polymorphs have been identified based on their Raman spectroscopic, textural, and chemical characteristics. (1) Brookite and anatase pseudomorphs after titanite in a fine-grained matrix, indicating low-temperature hydrothermal fluids destabilizing primary Ti-bearing minerals during argillic alteration (type-I). (2) Elongated and prismatic rutile present in hydrothermal veins or in clusters in accompanying alteration envelope characterized by weak zoning (type-II). And (3) rutile intergrown with sulfides in hydrothermal veins, characterized by well-developed patchy and sector zoning (type-III). In contrast to the type-I and type-II TiO 2 polymorphs, tungsten is enriched in backscattered bright patches and sector zones in type-III rutile, which is due to the substitution of W 6+ in the Ti 4+ octahedral site. The mechanism of the enrichment of tungsten is efectively driven by the halogen-rich (F, Cl) aqueous fluids during hydrothermal mineralization. In situ U-Pb dating of the type-III rutile yields a lower intercept age of 41.8 ± 1.2 Ma, which brackets the timing of the Cu-Mo mineralization. The relationship between rutile textures and composition indicates that W-bearing rutile can serve as a recorder of hydrothermal processes in porphyry Cu deposits.

Magnetic contributions have the potential to significantly influence predicted phase stability within alloy and mineral mixing phase diagrams, yet have been historically challenging to incorporate due to a significant increase to phase space sampling. In this work, we employ a computational protocol that includes spin orientation as an additional configurational component within multi-component cluster expansions between magnetic and non-magnetic metal oxide alloys [calculated using density functional theory (DFT) and the generalized gradient approximation]. This approach was used to determine the efect of magnetic contributions to corundum-eskolaite and corundum-hematite phase equilibria from first principles. Two-component cluster expansions of the magnetic components of eskolaite and hematite were first performed showing the ability of this method to properly calculate their respective magnetic properties. Two-component cluster expansions were then performed for non-magnetic Al(III) and ferromagnetic Cr(III) and Fe(III), and phase diagrams were calculated for later comparison. Finally, a non-magnetic Al(III) and “up” and “down” magnetic configurations for anti-ferromagnetic Cr(III) and Fe(III) were performed. Magnetic contributions to the calculated phase diagram for the corundumeskolaite system were shown to be inconsequential but are vital for accurate determination of the corundum-hematite solvus.

New petrographic and microanalytical studies of mineral inclusions in the Purang ophiolitic chromitites (SW Tibet) are used to scrutinize the evolution of the associated Cretaceous sub-oceanic lithospheric mantle section. Silicate inclusions in the chromite grains include composite and single-phase orthopyroxene, clinopyroxene, amphibole, and uvarovite. Most inclusions are sub-rounded or globular, whereas a few inclusions exhibit cubic/octahedral crystal morphologies. The latter are randomly distributed in the large chromite grains, though discrete aggregates are consistently confined to the grain centers. Abundant micrometer-scale, clinopyroxene inclusions are topotaxially aligned along crystallographic planes. Less-abundant sulfide, wüstite, apatite, and uvarovite inclusions are observed in some samples. The trace element geochemistry of the Purang chromitite evoke parental MORB- and boninitelike melts, consistent with the supra-subduction zone setting. The δ 26 Mg values of the high-Cr and high-Al chromitites range from –0.25 to –0.29‰ and –0.05 to –0.32‰, respectively. The associated harzburgite has nearly overlapping δ 26 Mg values of –0.13 to –0.37‰, but pyroxenite sills show distinct δ 26 Mg values (–0.61 to –0.67‰). The variable Mg isotope signatures, combined with abundant exotic, ultrahigh-pressure and super reduced (UHP-SuR) mineral inclusions in the chromite grains, suggest that recycling and recrystallization under different mantle conditions played an important role in the genesis and evolution of these rocks. Furthermore, discrete silicate, sulfide, and metal alloy inclusions in the Purang chromitites are comparable to those reported in other Tethyan ophiolites, and collectively suggest a common geodynamic evolution.

Stratabound ore has been recognized as an end-member of porphyry copper systems, but pyrite chemistry has not been widely applied to linking stratabound ore with the related porphyry and skarn system. Stratabound ore is commonly developed around porphyry-skarn systems in eastern China, and is characterized by abundant colloform pyrite; however the origin of the colloform pyrite has been contentious. Xinqiao deposit is ideal for study of pyrite geology and geochemistry with the aim of elucidating formation of the stratabound ore and to decipher the evolution of pyrite compositions in a porphyry-skarn environment. The colloform pyrite paragenesis and S isotopes indicate that it formed during early skarn mineralization, based on its occurrence in stockwork veins cutting skarn minerals, and that it is replaced by later hypogene sulfides; the δ 34 S of colloform pyrite (average 6.12‰) is close to the δ 34 S value of both porphyry- (average 5.06‰) and skarn-type pyrite (average 4.65‰). The colloform texture formed as an aggregate of nanometer- or micrometer-sized (<0.2 μm) pyrite cubes produced by rapid crystallization from a high- f S2 , low-temperature, and supersaturated fluid. Supersaturation of the fluid was probably produced by rapid decompression that triggered fluid boiling and cooling when the magmatic-hydrothermal fluid (derived from Cretaceous magma) flowed along the Devonian-Carboniferous unconformity. Subsequently, the colloform pyrite was replaced by later stage pyrite with distinctive trace elements (Co, Ni, and Se), indicating that the stratabound ore at Xinqiao formed from multiple pulses of magmatic-hydrothermal fluids derived from an Early Cretaceous stock. Co, Ni, and Se enrichment in porphyry- and proximal skarn-type pyrite suggests they formed at relatively high temperature, whereas the colloform pyrite shows trace element contents (Cu, Pb, Zn, Ag, and Bi) similar to those in distal skarn pyrite, suggesting that they may have formed in the same fluid environment. The trace element variations in pyrite from stratabound, skarn and porphyry ore probably resulted from decreasing fluid temperature and increasing pH away from the source. Our data, combined with previous studies, show that Co and Ni in pyrite increase toward porphyry and skarn ore, whereas As, Sb, Pb, Ag, and Bi are enriched in pyrite in distal stratabound ore, which extends for 1–2 km away from the intrusion. A plot of As/Se vs. Co discriminates the three ore types that occur associated with porphyry-skarn Cu systems in the Middle and Lower Yangtze belt (MLYB). These results indicate pyrite chemistry can be effective in discriminating the genesis of different deposit types related to porphyry-skarn systems and can potentially be used as a vectoring tool during exploration in the MLYB and elsewhere.

Aluminosilicate garnet is an excellent phase to research solid-solution behavior in silicates. Natural almandine-pyrope, F e 3 3 2 + M g 3 − 3 x A l 2 S i 3 O 12 , $\left\{\mathrm{Fe}_{3 \mathrm{3}}^{2+} \mathrm{Mg}_{3-3 \mathrm{x}}\right\}\left[\mathrm{Al}_{2}\right]\left(\mathrm{Si}_{3}\right) \mathrm{O}_{12},$and almandine-spessartine, F e 3 x 2 + , M n 3 − 3 x 2 + A l 2 S i 3 O 12 , $\left\{\mathrm{Fe}_{3 \mathrm{x}}^{2+}, \mathrm{Mn}_{3-3 \mathrm{x}}^{2+}\right\}\left[\mathrm{Al}_{2}\right]\left(\mathrm{Si}_{3}\right) \mathrm{O}_{12},$crystals were measured by UV/Vis/NIR (~29 000 to 10 000 cm –1 ) optical absorption spectroscopy using a microscope. The spectra and changes in energy of several Fe 2+ and Mn 2+ spin-forbidden electronic transitions of diferent wavenumber were analyzed as a function of garnet composition across both binaries. The spectra of Alm-Pyp garnets are complex and show several Fe 2+ and Fe 3+ transitions manifested as overlapping absorption bands whose intensities depend on composition. There are diferences in energy behavior for the various electronic transitions, whereby lower wavenumber Fe 2+ transitions decrease slightly in energy with increasing pyrope component and those of higher wavenumber increase. The spectra of Alm-Sps solid solutions show both Fe 2+ and Mn 2+ spin-forbidden bands depending upon the garnet composition. The variations in energy of the different wavenumber Fe 2+ transitions are unlike those observed in Alm-Pyp garnets. The three lowest wavenumber electronic transitions appear to vary the most in energy across the Alm-Sps join compared to those at higher wavenumber. Four narrow and relatively intense Mn 2+ spin-forbidden bands between 23 000 and 25 000 cm –1 can be observed in many Sps-Alm garnets. Their transition energies may increase or decrease across the join, but scatter in the data prohibits an unequivocal determination. A consistent crystal-chemical model and Fe 2+ -O bond behavior, based on published diffraction and spectroscopic results, can be constructed for the Alm-Pyp binary but not for the Alm-Sps system. The spectra of the former garnets often show the presence of high-wavenumber spin-forbidden bands that can be assigned to electronic transitions of Fe 3+ occurring at the octahedral site. The most prominent band lies between 27 100 and 27 500 cm –1 depending on the garnet composition. Fe 3+ -O 2– bonding is analyzed using Racah parameters. State-of-the-art electronic structure calculations are needed to understand the precise physical nature of the electronic transitions in garnet and to interpret better UV/Vis/NIR spectra.

The UV/Vis single-crystal absorption spectra of two almandine-bearing and several spessartine garnets were measured and their respective Fe 2+ and Mn 2+ spin-forbidden electronic transitions analyzed. Spin-forbidden bands of Fe 3+ are also considered, because many aluminosilicate garnets contain some Fe 3+ . The spectra of the almandine-bearing garnets were recorded at room temperature between about 10 000 and 30 000 cm –1 . The spectrum of a nearly end-member spessartine (97 mol% M n 3 2 + A l 2 S i 3 O 12 $\left.\mathrm{Mn}_{3}^{2+} \mathrm{Al}_{2} \mathrm{Si}_{3} \mathrm{O}_{12}\right)$was measured between about 15 000 cm –1 and 30 000 cm –1 at room temperature and 78 K, the latter for the first time. The 78 K spectrum shows absorption features not observed at room temperature. Five additional spessartine-rich garnets with diferent Mn 2+ /(Mn 2+ + Fe 2+ ) ratios, and two with unusual chemistries, were recorded up to 26 000 cm –1 . The spectra of the two almandine-bearing garnets agree well with published results and show several overlapping Fe 2+/3+ bands located between about 14 000 and 25 000 cm –1 . The spectra were deconvoluted to gain more insight into the electronic transition behavior. These results, together with an analysis of other measured spectra, reveal several absorption features that were previously unrecognized or misassigned. The spectrum of spessartine shows several Mn 2+ bands, and most are clearly spaced from one another. A synthesis of various UV/Vis spectroscopic results is made and assignments for the Fe 2+/3+ and Mn 2+ bands are attempted. The intensities of the Mn 2+ spin-forbidden bands and the ligand → metal charge edge observed in the various spessartine spectra are discussed. Spectra of almandine and spessartine have been interpreted using Tanabe- Sugano diagrams that are constructed for cations in octahedral coordination, point symmetry O h . However, such analysis does not appear to be fully successful because Fe 2+ and Mn 2+ in garnet have triangular dodecahedral coordination with point symmetry D 2 . The interpretation of the spectrum of spessartine is especially problematic. An analysis shows that published model calculations of Fe 2+ electronic transition energies in garnet are not in good agreement with each other and are also not in full agreement with experimental spectra. First principles calculations are needed to better understand the spin-forbidden transitions of Fe 2+ , Fe 3+ , and Mn 2+ in garnet.

The various intervalence charge transfer (IVCT) mechanisms that can occur in silicate garnet, general crystal-chemical formula {X 3 }[Y 2 ](Z 3 )O 12 , are not fully understood. The single-crystal UV/Vis/NIR absorption spectra of two different almandine-rich, spessartine-rich and grossular-rich garnets, as well as an intermediate almandine-pyrope garnet, were measured. Absorption was observed from roughly 15 000 to 30 000 cm –1 . The spectra were deconvoluted and a very broad band with FWHM values ranging from 5000 to 7000 cm –1 (except in the case of one grossular where the FWHM is 8700 cm –1 ) and having an intensity maximum located between about 20 000 and 22 000 cm –1 in the visible region could be fit. Small weaker features located on this broad band were fit as well. The broad band is strongest in a nearly end-member composition almandine and weakest in a very grossular-rich iron-poor crystal. It is assigned to {Fe 2+ } + [Fe 3+ ] → {Fe 3+ } + [Fe 2+ ] IVCT. This is the first recognition of this type of electronic transition mechanism in diferent aluminosilicate garnet species. Photon-induced electron transfer probably occurs through an overlap of the d orbitals of Fe 2+ and Fe 3+ in their edge-shared triangular dodecahedral and octahedral coordination polyhedra, respectively. The two Fe cations with different formal charges should have markedly diferent energy potentials giving rise to asymmetric IVCT behavior. This, together with the relatively long Fe 2+ -Fe 3+ distances (greater than 3.2 Å), could explain the higher energy of the IVCT in garnet compared to Fe 2+ + Fe 3+ → Fe 3+ + Fe 2+ IVCT mechanisms observed in other minerals. The latter typically have iron cations in octahedral or quasi-octahedral coordination. The IVCT in aluminosilicate garnet can occur in different species that grew under dissimilar P - T - X conditions. The resulting electronic absorption band afects color markedly, because it is centered at higher energies in the blue visible region. It remains to be determined why IVCT is observed in the spectra of some garnets but not others. The various proposed IVCT mechanisms in Ca-Ti-bearing and aluminosilicate garnets are reviewed and analyzed.

We present a conceptually simple method to perform high-temperature experiments in a one-atmosphere box furnace that has a negligible cost of materials. The experimental setup consists of two commercially available materials and can be customized to sample or furnace size with few limitations. Furthermore, the design allows easy extraction of samples in one piece, making them eligible for textural analysis. The setup comprises a graphite capsule and a fireclay shell, the latter of which acts as a heat-resistant protective shield. Containers must be individually hand-crafted, but each can hold multiple samples. The setup can be reliably used in temperature conditions below the heat tolerance limit of the commercial fireclay, commonly ~1400 °C. Moreover, the graphite capsule buffers the oxygen fugacity to strongly reducing conditions during the experiment. The main advantage of our method lies in the utilization of easily accessible and low-cost materials that provides a widely applicable experimental setup easily used at larger scales. The method was developed during an experimental study of magmatic crystal-liquid suspensions and was reliable for experiments lasting for up to 36 h.