Volume 110, Issue 11

Mudrocks and mud-sized sediments (i.e., silt to clay) dominate the surface of Earth and Mars. These fine-grained sediments preserve a rich history of sedimentary processes from source to sink and shed light on ancient climates. However, both the physical and chemical nature of these materials make them difficult to fully characterize with traditional laboratory techniques. Here, we explore a cross-disciplinary and high-resolution approach using synchrotron radiation for X-ray diffraction, pair distribution function analysis, and submicrometer-scale X-ray fluorescence, combined with transmission electron microscopy, to better understand the nanostructure and composition of mud-sized sediments from a glacio-fluvial watershed in southwest Iceland. Our results demonstrate that sediments in the cold and wet climate of Iceland are more altered than previously thought, as evidenced by the identification of kaolinite and mixed-layer kaolinite-smectite. Additionally, sediments are enriched in amorphous materials and nanocrystalline phases, as determined from grain morphologies and compositions consistent with allophane, hisingerite, ferrihydrite, and halloysite. These alteration products are present as intimate mixtures that vary across depositional sites, demonstrating the dynamic nature of the secondary assemblage from source to sink. This work has implications for Mars, where, for example, basalt-sourced sedimentary rocks from Gale crater are abundant in clay minerals and amorphous materials. Finally, this work underpins the importance of using high-resolution techniques, a coordinated methodology, and developing innovative approaches for future planetary sample return missions (e.g., Mars sample return).

Granitic pegmatites are renowned as significant sources of rare metals (e.g., Li, Ta, Nb, Be, Cs). However, the origins and mechanisms underlying the enrichment of rare metals in granitic pegmatites remain subjects of intense debate. This study provides comprehensive petrography, major and trace element analyses of muscovites and isotopic data from pegmatite dikes associated with the Early Cretaceous Mufushan composite batholith in South China to elucidate geochemical fractionation processes and the mechanisms responsible for rare metals mineralization. Rare metal-bearing pegmatite and barren pegmatite dikes are found within the metasedimentary strata and the granitoid batholith, respectively. Both types of pegmatite dikes exhibit internal zoning, featuring three distinct structural zones with varied mineral assemblages, where muscovite is ubiquitously present across all zones. Both rare metal pegmatites and barren pegmatites show a limited monazite ε Nd (t) range between −9.0 and −7.6, which is aligned with the apatite ε Nd (t) values of −9.8 to −7.8 from the monzogranites. Progressing from the wall zone to the core zone in both pegmatites, muscovite exhibits increasing enrichment in incompatible elements (Li, Rb, Cs, Ta, Mn, and F) but a depletion in Ti, Zr, and Pb, indicative of a gradual differentiation trend. The consistent Nd isotope characteristics but variable fractionation in muscovite chemistry for the pegmatites and monzogranites indicate that they were derived from a similar magmatic source. The most primitive units of both pegmatites show differentiation degrees of muscovite comparable with the two-mica monzogranite but lower than the muscovite monzogranite, implying that the pegmatites and monzogranites might represent evolutionary products independent of two-mica monzogranite and muscovite monzogranite. Melt compositions (Li, Rb, and Cs) in equilibrium with muscovites for pegmatites are quantitatively calculated, evaluated, and aligned with a Rayleigh fractional crystallization model. Our modeling indicates that the formation of barren pegmatite requires merely 50% Rayleigh fractional crystallization degree, whereas the intermediate zone, marking the initial conspicuous occurrence of rare metal minerals in rare metal pegmatites, necessitates ∼90% Rayleigh fractional crystallization degree. Notable variations in Li, Rb, and Cs contents across different zones underscore the progressive enrichment of rare metals due to the fractionation of pegmatitic magmas within the cooler metasedimentary rocks. We propose that pegmatites in the Mufushan batholith predominantly concentrate rare metals by evolutionary processes through Rayleigh fractional crystallization and constitutional zone refining (CZR).

Pyrite (FeS 2 ) is a common host mineral of Pt in many types of deposits of late magmatic and hydrothermal origin, including the Aguablanca Ni-Cu ore deposit in southwestern Spain. Yet the exact state of Pt in the pyrite from Aguablanca, including its oxidation state and local atomic geometry, still lacks spectroscopic evidence. Here, we investigated the state of ∼30 ppm of Pt in a natural pyrite sample from Aguablanca using the high-energy resolution fluorescence detection X-ray absorption near edge structure (HERFD-XANES) technique. Our analysis of the Pt L 3 -edge HERFD-XANES experimental spectra and theoretical spectra simulations performed via the ab initio FDMNES code revealed that Pt is incorporated in the cationic subshell of the natural pyrite sample. The presence of Pt-bearing micro-or nano-sized inclusions disseminated in the pyrite matrix was ruled out. The Pt-S interatomic distance ( R Pt-S ) of 2.34 ± 0.01 Å in the Aguablanca sample is almost identical to that in the Pt-doped synthetic pyrite sample ( R Pt-S = 2.35 ± 0.01 Å). The Pt-S interatomic distance is ∼3.5% larger than the Fe-S interatomic distance in the pure pyrite structure, primarily due to a 3% difference between the covalent radii of Pt and Fe, rather than differences in their ionic radii or electronegativities. The Pt “formal” oxidation state of +3.5 and partial atomic charge of +0.4 e in the natural sample substantially exceed those of Fe in the pure pyrite structure (+2 and +0.2 e, respectively). By employing available literature data on the solubility of Pt in synthetic pyrite samples, we demonstrated that the Pt enrichment of the natural pyrite specimen took place at a temperature above 360 °C.

Coexistence in natural samples of zircon (ZrSiO 4 ) and reidite (a high-pressure polymorph of ZrSiO 4 ) is attributed to the effects of hypervelocity impact events. The grains and intergrowths in those minerals can be merely a few nanometers in size, which makes phase identification by standard methods of structure analysis difficult. However, analytical scanning transmission electron microscopy (STEM) utilizing electron energy loss spectroscopy (EELS) can provide important information on phase transition mechanisms and pressure-temperature conditions associated with the shock event at the nanoscale. Here we demonstrate that the valence as well as oxygen core-loss EELS can be employed for nanoscale mapping of zirconreidite distributions in zircon-reidite aggregates. Moreover, other accompanying phases, e. g., baddeleyite, could also be identified and mapped by this method. We further compare the EELS maps with a 4D-STEM nanobeam precession electron diffraction data, and demonstrate the advantages of the EELS mapping, which provides spatial resolution down to the nanometer scale and is independent of crystal orientation.

This study describes the incorporation mechanism of Ni, Co, and Cu in zoned marcasite-pyrite (FeS 2 ) crystals from vein-type hydrothermal mineralization in the Sierra de Orihuela (Betic Cordillera, SE Spain). Single-spot analyses and X-ray mapping by electron probe microanalyzer (EPMA) of zoned marcasite-pyrite grains reveal positive correlations among Ni (up to 12.8 wt%), Co (up to 4.42 wt%), and Cu (up to 7.33 wt%), as well as a negative correlation between these elements and Fe. In contrast, nanoscale analysis using high-resolution transmission electron microscopy (HRTEM) reveals that these compositional variations arise from a combination of solid solutions and mineral nanoparticles of Ni, Co, and Cu. The cores of the zoned grains are marcasite, which is depleted in Ni, Co, Cu, and other trace metals. Cores are surrounded by zoned rims that contain inner portions with nanoscale zoning defined by alternating layers (∼500 nm thickness) of pyrite (FeS 2 ) with cattierite (CoS 2 ) and/or carrollite (CuCo 2 S 4 ). This transitions to outer portions exhibiting Pa 3 pyrite with frequent discrete mineral nanoparticles, including 80–120 nm sized cubic Fd 3 m carrollite (CuCo 2 S 4 ), tetragonal I 4 2 d chalcopyrite (CuFeS 2 ), and cubic Fd 3 m violarite (FeNi 2 S 4 ). The shift in the mode of Ni, Co, and Cu occurrence, from incorporation via solid solution to nanoparticles, reflects variations in the physicochemical conditions of the crystallizing system rather than the specific nature of the crystalline mineral matrix forming the zoned structure. The formation of the zoned structure begins with the initial crystallization of trace-element-poor cores of marcasite. This is followed by pyrite rims displaying an outward progressive enlargement of the crystal lattice due to the incorporation of Ni, Co, and Cu in solid solution, reaching characteristic d -spacings of carrollite and cattierite. Finally, direct precipitation of pyrite and Ni-Co-Cu-rich nanoparticles occurs when pH increases and salinity decreases. Nanoscale insights from Sierra de Orihuela mineralization provide an unprecedented view of the partitioning of Ni, Co, and Cu into Fe-sulfides. These findings reveal mechanisms of metal enrichment in hydrothermal systems that are invisible to conventional tools used in the study of mineral deposits.

The identification of the genetic type of a mineral deposit, based on the compositional characteristics of specific minerals, has long been a focus of interest for economic geologists and mining companies. Traditional binary plots, due to the limitation of their dimensions, fail to encompass the whole element information, potentially introducing bias to the discrimination results. This is particularly the case in classifying Zn-Pb deposits. The current study employs four widely used machine learning algorithms (random forest, extreme gradient boosting, support vector machine, and multi-layer perceptron) to train 4908 sets of element data for sphalerite compiled from five distinct Zn-Pb deposit types (VMS, SEDEX, MVT, skarn, and epithermal). The data are then visualized and interpreted through principal component analysis and t-distributed stochastic neighbor embedding, which indicate that reducing sphalerite element data to a two-dimensional projection leads to the loss of significant feature information, hindering the ability to effectively distinguish the genesis of the deposit. The machine learning results show that all four models have macro F1-scores above 0.95 on the test set, demonstrating robustness and excellent generalization ability, which reflects the reliability of using sphalerite geochemistry to distinguish Zn-Pb mineralization types. The SHAP value analysis highlights the key role of Mn, Fe, Ge, Cd, and Co concentrations in facilitating the differentiation of deposit types through machine learning algorithms. Our models show an accuracy rate of 83% in predicting the combined results on an external, independent data set. The models have also been applied to classify three Zn-Pb deposits of unknown types, and the results are consistent with geological observations. The models’ parameters have been further exported and programmed into an Excel macro program and a user-friendly software application, which can be accessed via https://sdeakii.github.io/machine-learning.

Te-Bi melts have been recognized as crucial agents for scavenging gold in gold deposits related to magmatism, often characterized by a notable Au-Te-Bi association. This association is also recognized in the gold deposits of the Jiaodong district, although the specific roles of Te and Bi in gold enrichment in this region have not been thoroughly evaluated. In this study, we investigate the mineralogy and geochemistry of Au-Ag-Te-Bi mineral assemblages and pyrite at the Taishang gold deposit to better understand the influence of Te-Bi melts on gold enrichment and refine the ore genesis models for the Jiaodong district. The gold mineralization at Taishang comprises three stages: phyllic alteration (Stage I), pyrite-quartz veins (Stage II), and polymetallic sulfide veins (Stage III). In Stage I, gold mainly occurs as inclusions of native gold and electrum within pyrite, alongside hessite, petzite, and minor tetradymite. Stage II features multiphase Au-Ag-Te-Bi-(S)-(Pb) assemblages, including native gold, electrum, hessite, petzite, tetradymite, tsumoite, joséite-B, and sub-micrometer Te-Bi-(Pb)-(Au)-(Ag)-rich phases as inclusions and fracture-fillings in pyrite. In Stage III, gold mineralization is less intense, marked by sparse native gold and electrum grains intergrown with pyrite, galena, and chalcopyrite. The mineralogical patterns reveal a strong relationship between Au, Ag, Te, and Bi, corroborated by trace element compositions of pyrite, which exhibit strong positive correlations among these elements. Furthermore, the texture of Au-Ag-Te-Bi assemblages in Stage I and II (i.e., unsharp and curvilinear boundaries) suggests crystallization from melts. Supported by relevant experimental evidence, the low temperature of Au-Ag-Te-Bi systems enables Te-Bi melts to effectively scavenge gold from fluids at the temperature conditions of Taishang (246–336 °C). The presence of Au-Ag-Te-Bi association in both Taishang and other Jiaodong gold deposits underscores the importance of Te-Bi melts in enhancing gold enrichment. This finding introduces vital insights into the established model of gold mineralization in the Jiaodong district.

Mineralization of Nb and Ta associated with metaluminous-weakly peraluminous (MWP) magmatic systems has gained increasing attention in recent years, particularly with the recognition of several major Nb-Ta deposits in China originating from MWP granites. Columbite-group minerals (CGMs), the key ore minerals for Nb-Ta, serve as crucial indicators for understanding the mechanisms underlying this mineralization. This study presents a comprehensive analysis of major and trace elements, as well as Hf isotopes, of CGMs from seven Nb-Ta deposits in China, related to MWP systems. We compared these CGMs with those from deposits associated with strongly peraluminous (SP) magmatic systems to uncover the distinctions in their mineralization processes. Statistical analyses, including elemental plots and partial least squares discriminant analysis, revealed significant variations in CGMs between the MWP system and the SP system. The earliest crystallized CGMs from MWP systems exhibit higher Nb 2 O 5 contents resulting from higher Nb concentration in the initial magma and therefore do not need to undergo as high a degree of fractional crystallization as those in SP systems do. Additional geochemical markers further differentiate MWP magma sources, including higher Sc, total REE, and Y contents, as well as Th/U atomic ratios, along with lower Li contents and Y/Ho atomic ratios. As magma evolves, the Zr/Hf atomic ratios and total REE contents decrease, while Ta/(Nb+Ta) atomic ratios increase. Additionally, the influence of melt-crystal-fluid interactions on REE characteristics in late-stage CGMs cannot be ignored. The Hf isotopic compositions of CGMs indicate that the Nb-Ta mineralization in South China during the Early Cretaceous (∼140–120 Ma), associated with MWP systems, is attributed to contributions from a deep material source within an extensional tectonic setting. The differences in chemical compositions, U-Pb ages, and Hf isotopes of CGMs between the two systems offer valuable insights for heavy mineral prospecting to locate Nb-Ta deposits related to MWP granites in regional exploration.

Spodumene and petalite are the primary minerals mined from pegmatites, which are the most important sources of hard-rock lithium resources. In Li-mineralized pegmatites, the subsolidus reaction of petalite→spodumene+2 quartz (1) and its reverse reaction: spodumene + 2 quartz →petalite (2) produce secondary spodumene, usually in the form of spodumene and quartz intergrowths (SQI), and secondary petalite, respectively. The pressure-temperature ( P-T ) conditions under which these reactions occur have been investigated using quenching-type methods (e.g., with cold-seal pressure vessels). However, the kinetic parameters of the reactions are difficult to measure in these experiments and are thus frequently overlooked, even though their determination can significantly improve our understanding of the formation of spodumene and petalite in pegmatites. In this study, we employed a hydrothermal diamond-anvil cell to monitor the reaction process and crystallization dynamics in situ. The reaction rates ( V r ) of reaction 1 in sample systems with 5–95 wt% H 2 O ranged from 10 −6 to 10 −11 cm 3 /day at 355–753 MPa and 500–600 °C. V r is driven by the Gibbs free energy change (Δ G r ) of the reaction, which is dependent on pressure and temperature. As Δ G r (<0) decreases, V r rises. V r also correlates positively with the growth rate ( V g ) of secondary spodumene, though V g is lower than that of primary spodumene formed in pegmatite-forming melts. Hence, by extrapolating experimental Δ G r – V r data to pressure-temperature ( P-T ) conditions of pegmatite formation, we can estimate V r and thus assess V g for secondary spodumene in natural SQI. For example, the formation of secondary spodumene in the Highbury pegmatite (South Africa) is attributed to reaction 1, which occurred at a V r of 10 −8 to 10 −9 cm 3 /day under conditions of 370 MPa and 400–500 °C. This V r could correspond to a high V g , leading to secondary spodumene crystals reaching up to 0.5 cm in size. In contrast, the fine-grained secondary spodumene (e.g., <0.2 mm) in the Tanco pegmatite (Canada) is possibly attributed to a lower V g under P-T conditions of ∼290–260 MPa and below 400 °C. Reaction 2 at 244 MPa and 600 °C yielded V r of ∼10 −7 cm 3 /day, approaching the growth rate of petalite crystallized from pegmatite-forming melts. Thus, secondary petalite is easy to form via reaction 2 during pressure drops in pegmatite formation.

Hydrated iron sulfate minerals have received considerable attention from the standpoint of environmental science, as well as due to extensive studies on the mineralogy of Mars. In this paper, we report on the thermal evolution of coquimbite AlFe33+SO46H2O12⋅6H2O $\mathrm{AlFe}_3^{3+}\left(\mathrm{SO}_4\right)_6\left(\mathrm{H}_2 \mathrm{O}\right)_{12} \cdot 6 \mathrm{H}_2 \mathrm{O}$by single-crystal X-ray diffraction (SCXRD) from −173 to 77 °C. Powder X-ray diffraction (PXRD) was performed in the temperature range of −180 to 740 °C and low vacuum of 600 Pa. Magnetic properties for coquimbite are reported in the range of −271 to 7 °C. It was observed that coquimbite is stable between −180 and + 145 °C and at a low vacuum of 600 Pa. We observed a gradual transition from coquimbite to the amorphous phase at 150 °C, followed by a transition to mikasaite at 225 °C, and a second amorphization at 575 °C, with afterward crystallization to hematite. SCXRD shows that the behavior of coquimbite with increasing temperature can be divided into two stages, with negative and strongly anisotropic thermal expansion at Stage I (−173 to −143 °C) and only positive thermal expansion at Stage II (−133 to 77 °C). All the O-H···O bonds remain virtually intact during Stage I, except for Ow2-H3···O2. The negative thermal expansion observed along the c -axis in the LT range is a result of the simultaneous reduction of several bond lengths and angular distortions: (1) decrease of Ow2-H3···O2 hydrogen bonds oriented approximately along the c -axis; and (2) shrinkage of M 3O 3 (H 2 O) 3 octahedra, evidenced by the decrease in M 3-O3 and M 3-Ow3 bonds. The nature of the expansion of the coquimbite structure during Stage II is better understood in terms of the orientation of [ M 2 M 32(SO 4 ) 6 (H 2 O) 6 ] 3− clusters along the c -axis. M -O and S-O bonds are only slightly affected by the temperature rise at Stage II, whereas O-H···O angular transformations seem to be the main driving force for the expansion of the coquimbite structure along the α 11 direction upon heating. Coquimbite exhibits distinct magnetic properties compared to other iron sulfates, driven by antiferromagnetic interactions within its M3- M 2- M 3 trimeric clusters of Fe 3+ . The presence of Al 3+ -Fe 3+ site mixing in coquimbite introduces structural disorder, partially disrupting its magnetic ordering and contributing to magnetic entropy and magnetization features, such as a 1/3 magnetization plateau.

Struvite (MgNH 4 PO 4 ·6H 2 O), a magnesium phosphate mineral, is a promising recoverable phosphorus (P) and nitrogen (N) source and is employed as a slow-release agricultural fertilizer. However, struvite is unstable in air and transforms into two distinct crystalline phases, newberyite (MgHPO 4 ·3H 2 O) and dittmarite (MgNH 4 PO 4 ·H 2 O), before breaking down into an amorphous phase. The transformation products of struvite have been indicated in the literature. However, published studies quantifying these processes have focused on struvite decomposition in solutions, and investigations in air have been conducted only at high temperatures, far from the conditions under which this mineral would be stored for use as a fertilizer. Furthermore, the kinetics are unknown, including transformation rates and mechanisms. Thus, we present complementary X-ray diffraction (XRD), vibrational spectroscopic, and light to electron microscopic data from which we determined the mechanisms for the transformation at 22–60 °C and breakdown of struvite in air. Our results revealed that its transformation into both newberyite and dittmarite follows a coupled dissolution-reprecipitation mechanism, with the reaction kinetics dominated by the diffusion of H 2 O and NH 3 out of the struvite structure. The reaction pathways proved to be temperature dependent, with newberyite being the only transformation product at room temperature, even after 10 months. This transformation resulted from the loss of ammonia and three of the six water molecules. On the contrary, at higher temperatures, dittmarite was the major transformation product of struvite, formed through the release of five of the six water molecules while retaining the ammonia. At 60 °C and after 3 months, dittmarite breaks down into an amorphous magnesium phosphate phase. This study provides comprehensive insights into the kinetics and the underlying mechanisms governing the transformation of struvite at atmospheric conditions. Our findings have significant implications for the long-term storage of struvite and its subsequent utilization as a slow-release fertilizer.

A find of native tin in deeply weathered material at the Neogene epithermal Au-Ag Pukanec deposit in Slovakia raises the question of transport and reduction to native metal, especially when such a metal requires very strongly reducing conditions. The deeply weathered material contains powdery Mn oxides and clays, with the heavy fraction made of cerussite, smithsonite, nuggets of Au-Ag alloys, and oval nuggets of native tin. During weathering, Au and Ag are mobile and can reprecipitate as native metals. Native tin persists, although it shows signs of weathering to cassiterite. Electron microprobe analyses of native tin document Ni (up to 7 wt%) in a few spots as the only other abundant component. The association of tin with pyrite indicates that tin is primary, but it had to be transported in an inert form through the fluid. The isotopic composition of native tin (expressed as δ amu Sn 3161A ) shows very limited variations between 0.01 and 0.07‰ (±0.014‰), which is close to the expected composition of parent rhyolites, suggesting that the latter were likely the source of Sn. The oval shape of the tin grains and their physical properties (e.g., the low melting point of tin) suggest that tin is transported as metal-melt droplets in the hydrothermal fluid of the epithermal system. Tin must have been reduced before entering the space of the hydrothermal veins, perhaps in the underlying carbonaceous Carboniferous sediments. Reduction of Sn(II) from hydrothermal fluid is unlikely because under such conditions, Fe(II) would also be reduced to native iron. The most feasible explanation for these facts is the local reduction of Sn(II) deep within the hydrothermal system (perhaps in the carbonaceous sediments) and transport as metal droplets to the deposition site.

Wangyanite (IMA 2024-008a), ideally PdNi 8 S 8 , is a Pd end-member mineral of the pentlandite group that was discovered in the J-M Reef of the Stillwater Complex, Montana, U.S.A. Wangyanite occurs as anhedral to subhedral granular crystals, 200–400 μm in size, associated with isoferroplatinum, braggite, pentlandite, and chalcopyrite, interstitial to plagioclase grains within anorthosite. Wangyanite exhibits a yellowish-brown color with a black streak and a metallic luster. It is brittle, with uneven fractures, and has a calculated density of 5.14 g/cm 3 . The mineral does not show discernible pleochroism, bireflectance, or anisotropy. It has an average composition of 9.95 wt% Pd, 31.95 wt% Ni, 25.02 wt% Fe, 0.57 wt% Co, 31.74 wt% S, totaling 99.23 wt%. The empirical formula, based on eight sulfur atoms per formula unit, is (Pd 0.76 Co 0.08 ) Σ0.84 (Ni 4.39 Fe 3.60 ) Σ7.99 S 8 . Wangyanite has a cubic unit cell with a space group of Fm 3 m (#225), with lattice parameters of a = 10.1167(12) Å, V = 1035.4(4) Å 3 , and Z = 4. Its crystal structure has been solved by single-crystal three-dimensional electron diffraction study. The strongest X-ray diffraction lines of wangyanite are calculated at [ d in Å ( I %)( hkl )]: 5.841(14.03)(111), 3.050(100)(311), 1.947(29.16)(115,333), 1.264(11.66)(800), 3.577(8.79)(220), 2.920(20.82)(222), and 2.321(9.34)(331). Wangyanite shares the similar crystal structure as pentlandite, but the octahedrally coordinated site is mainly occupied by Pd in wangyanite. Based on the textural features and previous experiments in the Pd-Fe-Ni-S phase system, wangyanite could form by a peritectic reaction between braggite, pentlandite, and a sulfide liquid. These mineral associations are stable in a Ni-Pd-rich sulfide melt system at about 550 °C or even lower temperatures. Therefore, wangyanite can potentially serve as an indicator of the presence of Pd-rich residual melts. The mineral is named in honor of Christina Yan Wang, a renowned researcher of platinum-group element (PGE) occurrences and enrichment mechanisms in mafic-ultramafic intrusions, particularly those related to the Emeishan large igneous province in China.

High-quality single crystals of Al-bearing stishovite and Al-rich phase D were analyzed using X-ray diffraction, high-temperature Raman, and FTIR spectroscopy. The samples were synthesized in the MgO-SiO 2 -Al 2 O 3 -H 2 O system at 25 GPa, 1350–1600 °C. The results demonstrate that Al-bearing SiO 2 glass formed by the amorphization of stishovite under high-temperature conditions can incorporate water in the form of hydroxyl groups, with a concentration of at least several hundred ppm H 2 O by weight. Al 3+ prefers to enter the crystal structure of phase D by substituting for Si 4+ in the M2 site. Since Al 3+ can be charge-balanced by H + , the Al-rich phase D may have a higher water storage capacity than pure Mg-phase D. Raman and infrared spectral features of Al-rich phase D at ambient temperature can still be observed up to 350 °C. This indicates that, compared to pure Mg-phase D, Al-rich phase D maintains stability at higher temperatures. Al-rich silicate glass formed by the amorphization of phase D under high-temperature conditions can also retain water in the form of hydroxyl groups. In the deep mantle, Al-rich silicate glasses and Al-bearing SiO 2 glasses may serve as potential reservoirs of water (hydrogen) in high-temperature regions beyond the ideal stability fields of dense hydrous magnesium silicates and high-pressure SiO 2 phases.