Silicate perovskite-melt partitioning of trace elements and geochemical signature of a deep perovskitic reservoir

We have determined the partitioning of a wide range of trace elements between silicate melts and CaSiO₃ and MgSiO₃ perovskites using both laser ablation-ICPMS and ion microprobe techniques. Our results show that, with the exception of Sc, Zr, and Hf, all trace elements we considered are incompatible in MgSiO₃ perovskite, from highly incompatible for U, Th, Ba, La, Sr and monovalent elements to slightly incompatible for heavy rare earth elements. MgSiO₃ perovskite-melt partition coefficients increase slightly with Al content in the perovskite. These observations contrast strongly with partitioning between CaSiO₃ perovskite and silicate melts. In the latter case, all rare earth elements are clearly compatible as are U and Th. Our data also suggest that, contrary to pressure and temperature, melt composition can significantly affect CaSiO₃ perovskite-melt partitioning; partition coefficients for rare earth elements and U and Th increase with decreasing CaO melt content. The presence of ∼0.4 wt% water in melt makes little difference, however. Partitioning of trace elements into the large site of both MgSiO₃ and CaSiO₃ perovskites follows the near-parabolic dependence on ionic radius predicted from the lattice strain model. The peaks of the parabolae are much higher for the CaSiO₃ phase, perhaps suggesting that the mechanisms of charge compensation for heterovalent substitution are different in the two cases. Our partitioning data have been used to assess the potential effect of perovskite fractionation into the lower mantle during early Earth history. Crystallisation of less than 8% of a mixture of CaSiO₃ and MgSiO₃ perovskites could have led to a ‘layer’ enriched in U and Th without disturbing the chondritic pattern of refractory lithophile elements in the primitive upper mantle. The resultant reservoir could have high Sm/Nd, U/Pb, Sr/Rb, Lu/Hf ratios similar to the HIMU component of ocean island basalts, but would not balance the observed depletion of the primitive upper mantle in Si and Nb.     

Solid solution behaviour of CaSiO<Subscript>3</Subscript> and MgSiO<Subscript>3</Subscript> perovskites

Using density functional simulations within the generalized gradient approximation and projector-augmented wave method together with thermodynamic modelling, the reciprocal solubilities of MgSiO₃ and CaSiO₃ perovskites were calculated for pressures and temperatures of the Earth’s lower mantle from 25 to 100 GPa and 0 to 6,000 K, respectively. The solubility of Ca in MgSiO₃ at conditions along a mantle adiabat is found to be less than 0.02 atoms per formula unit. The solubility of Mg in CaSiO₃ is even lower, and most important, the extent of solid solution decreases with pressure. To dissolve CaSiO₃ perovskite completely in MgSiO₃ perovskite, a solubility of 7.8 or 2.3 mol% would be necessary for a fertile pyrolytic or depleted harzburgitic mantle, respectively. Thus, for any reasonable geotherm, two separate perovskites will be present in fertile mantle, suggesting that Ca-perovskite will be residual to low degree melting throughout the entire mantle. At the solidus, CaSiO₃ perovskite might completely dissolve in MgSiO₃ perovskite only in a depleted mantle with <1.25 wt% CaO. These implications may be modified if Ca solubility in MgSiO₃ is increased by other major mantle constituents such as Fe and Al.     

Ca- and Mg-perovskite phases in the Earth’s mantle as a probable reservoir of Al: Computer-simulation evidence

Semi-empirical and quantum chemical studies of Al atom energy in CaSiO₃ and MgSiO₃ with the perovskite-type structure at pressures and temperatures of the Earth’s mantle are reported. The phase diagram for CaSiO₃ is reproduced and refined. Probable mechanisms of Al incorporation in the structures studied are considered. According to the results of the calculations, Al is preferably incorporated into MgSiO₃, rather than into CaSiO₃. Evaluation of the isomorphic capacity of perovskite phases in relation to Al shows that the Al content in MgSiO₃ may reach 2.4 mol % at 120 GPa and 2400 K. CaSiO₃ cannot be a source of Al atoms in the Earth’s mantle.     

Low Mantle Perovskite: Solid Solution,Spin State of Iron and Water Solubility

Silicate perovskites((Mg, Fe)SiO 3 and CaS iO 3) are believed to be the major constituent minerals in the lower mantle. The phase relation, solid solution, spin state of iron and water solubility related to the lower mantle perovskite are of great effect on the geodynamics of the Earth's interior and on ore mineralization. Previous studies indicate that a large amount of iron coupled with aluminum can incorporate into magnesium perovskite, but this is discordant with the disproportionation of(Mg,Fe)SiO 3 perovskite into iron-free MgS i O3 perovskite and hexagonal phase(Mg0.6Fe0.4)SiO 3 in the Earth's lower mantle. MnS iO 3 is the first chemical component confirmed to form wide range solid solution with Ca SiO 3 perovskite and complete solid solution with MgS i O3 perovskite at the P-T conditions in the lower mantle, and addition of Mn Si O3 will strongly affects the mutual solubility between Mg Si O3 and CaS iO 3. The spin state of iron is deeply depends on the site occupation of the Fe3+or Fe2+, the synthesis and the annealing conditions of the sample. It seems that the spin state of Fe2+ in the lower mantle perovskite can be settled as high spin, however, the existence of intermediate spin or low spin state of Fe2+ in perovskite has not been clarified. Moreover, different results have also been reported for the spin state of Fe3+ in perovskite. The water solubility of the lower mantle perovskite is related with its composition. In pure Mg SiO 3 perovskite, only less than 500 ppm water was reported. Al–Mg Si O3 perovskite or Al–Fe–MgS iO 3 perovskite in the lower mantle accommodates water of 1100 to 1800 ppm. Further experiments are necessary to clarify the detailed conditions for perovskite solid solution, to reliably analyze the valence and spin states of iron in the coexisting iron-bearing phases, and to compare the water solubility of different phases at different layers for deeply understanding the geodynamics of the Earth's interior and ore mineralization.     

Equations of state of CaSiO3 Perovskite: a molecular dynamics study

The molar volumes and bulk moduli of CaSiO₃ perovskite are calculated in the temperature range from 300 to 2,800 K and the pressure range from 0 to 143 GPa using molecular dynamics simulations that employ the breathing shell model for oxygen and the quantum correction in addition to the conventional pairwise interatomic potential models. The performance of five equations of state, i.e., the Keane, the generalized-Rydberg, the Holzapfel, the Stacey–Rydberg, and the third-order Birch–Murnaghan equations of state are examined using these data. The third-order Birch–Murnaghan equation of state is found to have a clear tendency to overestimate the bulk modulus at very high pressures. The Stacey–Rydberg equation of state degrades slightly at very high pressures along the low-temperature isotherms. In comparison, the Keane and the Holzapfel equations of state remain accurate in the whole temperature and pressure range considered in the present study. K ₀′ derived from the Holzapfel equation of state also agrees best with that calculated independently from molecular dynamics simulations. The adiabatic bulk moduli of CaSiO₃ perovskite along lower mantle geotherms are further calculated using the Keane and the Mie-Grüneisen–Debye equations of state. They are found to be constantly higher than those of the PREM by ~5%, and also very similar to those of the MgSiO₃ perovskite. Our results support the view that CaSiO₃ perovskite remains invisible in the Earth’s lower mantle.     

Decarbonation reaction of magnesite in subducting slabs at the lower mantle

High-pressure and temperature experiments (28–62 GPa, and 1,490–2,000 K, corresponding to approximately 770–1,500 km depth in the mantle) have been conducted on a MgCO₃ + SiO₂ mixture using a laser-heated diamond anvil cell combined with analytical transmission electron microscope observation of the product phases to constrain the fate of carbonates carried on the subducting basalt into the lower mantle. At these conditions, the decarbonation reaction MgCO₃ (magnesite) + SiO₂ (stishovite) → MgSiO₃ (perovskite) + CO₂ (solid) has been recognized. This indicates that above reaction takes place as a candidate for decarbonation of the carbonated subducting mid ocean ridge basalts in the Earth’s lower mantle.     

Ferropericlase—a lower mantle phase in the upper mantle

Experiments on compositions along the join MgO–NaA³⁺Si₂O₆ (A=Al, Cr, Fe³⁺) show that sodium can be incorporated into ferropericlase at upper mantle pressures in amounts commonly found in natural diamond inclusions. These results, combined with the observed mineral parageneses of several diamond inclusion suites, establish firmly that ferropericlase exists in the upper mantle in regions with low silica activity. Such regions may be carbonated dunite or stalled and degassed carbonatitic melts. Ferropericlase as an inclusion in diamond on its own is not indicative of a lower mantle origin or of a deep mantle plume. Coexisting phases have to be taken into consideration to decide on the depth of origin. The composition of olivine will indicate an origin from the upper mantle or border of the transition zone to the lower mantle and whether it coexisted with ferropericlase in the upper mantle or as ringwoodite. The narrow and flat three phase loop at the border transition zone—lower mantle together with hybrid peridotite plus eclogite/sediments provides an explanation for the varying and Fe-rich nature of the diamond inclusion suite from Sao Luiz, Brazil.     

Diamond precipitation and mantle metasomatism – evidence from the trace element chemistry of silicate inclusions in diamonds from Akwatia, Ghana

Trace element concentrations in the four principal peridotitic silicate phases (garnet, olivine, orthopyroxene, clinopyroxene) included in diamonds from Akwatia (Birim Field, Ghana) were determined using SIMS. Incompatible trace elements are hosted in garnet and clinopyroxene except for Sr which is equally distributed between orthopyroxene and garnet in harzburgitic paragenesis diamonds. The separation between lherzolitic and harzburgitic inclusion parageneses, which is commonly made using compositional fields for garnets in a CaO versus Cr₂O₃ diagram, is also apparent from the Ti and Sr contents in both olivine and garnet. Titanium is much higher in the lherzolitic and Sr in the harzburgitic inclusions. Chondrite normalised REE patterns of lherzolitic garnets are enriched (10–20 times chondrite) in HREE (La_N/Yb_N = 0.02–0.06) while harzburgitic garnets have sinusoidal REE_N patterns, with the highest concentrations for Ce and Nd (2–8 times chondritic) and a minimum at Ho (0.2–0.7 times chondritic). Clinopyroxene inclusions show negative slopes with La enrichment 10–100 times chondritic and low Lu (0.1–1 times chondritic). Both a lherzolitic and a harzburgitic garnet with very high knorringite contents (14 and 21 wt% Cr₂O₃ respectively) could be readily distinguished from other garnets of their parageneses by much higher levels of LREE enrichment. The REE patterns for calculated melt compositions from lherzolitic garnet inclusions fall into the compositional field for kimberlitic-lamproitic and carbonatitic melts. Much more strongly fractionated REE patterns calculated from harzburgitic garnets, and low concentrations in Ti, Y, Zr, and Hf, differ significantly from known alkaline and carbonatitic melts and require a different agent. Equilibration temperatures for harzburgitic inclusions are generally below the C-H-O solidus of their paragenesis, those of lherzolitic inclusions are above. Crystallisation of harzburgitic diamonds from CO₂-bearing melts or fluids may thus be excluded. Diamond inclusion chemistry and mineralogy also is inconsistent with known examples of metasomatism by H₂O-rich melts. We therefore favour diamond precipitation by oxidation of CH₄-rich fluids with highly fractionated trace element patterns which are possibly due to “chromatographic” fractionation processes. Received: 27 January 1996 / Accepted: 5 May 1997     

Magmatic evolution of the material of the Earth’s lower mantle: Stishovite paradox and origin of superdeep diamonds (Experiments at 24–26 GPa)

The ultrabasic–basic magmatic evolution of the lower mantle material includes important physicochemical phenomena, such as the stishovite paradox and the genesis of superdeep diamonds. Stishovite SiO₂ and periclase–wüstite solid solutions, (MgO · FeO)_ss, associate paradoxically in primary inclusions of superdeep lower mantle diamonds. Under the conditions of the Earth’s crust and upper mantle, such oxide assemblages are chemically impossible (forbidden), because the oxides MgO and FeO and SiO₂ react to produce intermediate silicate compounds, enstatite and ferrosilite. Experimental and physicochemical investigations of melting phase relations in the MgO–FeO–SiO₂–CaSiO₃ system at 24 GPa revealed a peritectic mechanism of the stishovite paradox, (Mg, Fe)SiO₃ (bridgmanite) + L = SiO₂ + (Mg, Fe)O during the ultrabasic–basic magmatic evolution of the primitive oxide–silicate lower mantle material. Experiments at 26 GPa with oxide–silicate–carbonate–carbon melts, parental for diamonds and primary inclusions in them, demonstrated the equilibrium formation of superdeep diamonds in association with ultrabasic, (Mg, Fe)SiO₃ (bridgmanite) + (MgO · FeO)_ss (ferropericlase), and basic minerals, (FeO · MgO)_ss (magnesiowüstite) + SiO₂ (stishovite). This leads to the conclusion that a peritectic mechanism, similar to that responsible for the stishovite paradox in the pristine lower mantle material, operates also in the parental media of superdeep diamonds. Thus, this mechanism promotes both the ultrabasic–basic evolution of primitive oxide–silicate magmas in the lower mantle and oxide–silicate–carbonate melts parental for superdeep diamonds and their paradoxical primary inclusions.     

Melting experiments on the enstatite-diopside join at 70–224 kbar,including the melting of diopside

 Melting relations on the enstatite−diopside (En, Mg₂Si₂O₆−Di, CaMgSi₂O₆) join, including the compositions of crystalline phases and melts coexisting along the solidi, were experimentally determined in the pressure range 70–224 kbar with a split-sphere anvil apparatus (USSA-2000). Melting is peritectic in enstatite-rich compositions at 70–124 kbar (1840–2100° C) and eutectic at higher pressures, while the diopside-rich clinopyroxene melts azeotropically at 70–165 kbar and up to 300° C lower temperatures than the eutectic. Orthopyroxene is replaced with enstatite-rich clinopyroxene at 120 kbar and 2090°C. First garnet with 17 mol% Di forms on the solidus at 158 kbar and 2100° C. Two garnets coexist on the solidus at 165–183 kbar and 2100° C, garnet coexists with CaSiO₃ perovskite at 183–224 kbar (2100–2230° C) and two coexisting perovskites are stable at higher pressures. The melting curve of diopside was determined at 80–170 kbar; the slope becomes negative at 140 kbar and 2155° C. At 170 kbar and 2100° C, diopside with 96% Di breaks down to garnet with 89% Di and CaSiO₃ perovskite. The new data were used to calculate an improved temperature-pressure phase diagram for the CMAS system, which can be useful for estimating the mineralogy of the Earth's upper mantle. Received: 15 October 1994 / Accepted: 15 October 1995     

Fe–Mg partitioning between post-perovskite and ferropericlase in the lowermost mantle

Fe–Mg partitioning between post-perovskite and ferropericlase has been studied using a laser-heated diamond anvil cell at pressures up to 154 GPa and 2,010 K which corresponds to the conditions in the lowermost mantle. The composition of the phases in the recovered samples was determined using analytical transmission electron microscopy. Our results reveal that the Fe–Mg partition coefficient between post-perovskite and ferropericlase (K _D^PPv/Fp) increases with decreasing bulk iron content. The compositional dependence of K _D^PPv/Fp on the bulk iron content explains the inconsistency in previous studies, and the effect of the bulk iron content is the most dominant factor compared to other factors, such as temperature and aluminum content. Iron prefers ferropericlase compared to post-perovskite over a wide compositional range, whereas the iron content of post-perovskite (X _Fe^PPv, the mole fraction) does not exceed a value of 0.10. The iron-rich ferropericlase phase may have significant influence on the physical properties, such as the seismic velocity and electrical conductivity at the core–mantle boundary region.     

An alternative interpretation of lower mantle mineral associations in diamonds

Diamonds containing ferropericlase (Mg,Fe)O and other silicate (enstatite [(Mg,Fe)SiO₃], in particular) assemblages are generally believed to be derived from the Earth's lower mantle. On the basis of the observed ratio between ferropericlase and enstatite inclusions and the FeO content of these ferropericlases, it is concluded that most of these minerals entrapped in diamonds may not represent the lithology of the lower mantle itself as has been suggested by many investigators. Instead, ferropericlases in these diamonds represent most likely the disproportionate product of ferromagnesite [(Mg,Fe)CO₃], which underwent a decarbonation reaction to form both diamond and ferropericlase simultaneously in the lower mantle. The wide variation in the Mg# of ferropericlase inclusions in diamonds is attributed to the decarbonation "loop" of the MgCO₃-FeCO₃ solid solutions. Some of the enstatite inclusions coexisting with these ferropericlases in the same diamond may represent the most abundant mineral species of (Mg,Fe)SiO₃-perovskite in the lower mantle. The latter mineral phase experienced a retrogressive transition into enstatite during the transport of diamonds to the Earth's surface.     

Experimental study of reaction between perovskite and molten iron to 146 GPa and implications for chemically distinct buoyant layer at the top of the core

Partitioning of oxygen and silicon between molten iron and (Mg,Fe)SiO₃ perovskite was investigated by a combination of laser-heated diamond-anvil cell (LHDAC) and analytical transmission electron microscope (TEM) to 146 GPa and 3,500 K. The chemical compositions of co-existing quenched molten iron and perovskite were determined quantitatively with energy-dispersive X-ray spectrometry (EDS) and electron energy loss spectroscopy (EELS). The results demonstrate that the quenched liquid iron in contact with perovskite contained substantial amounts of oxygen and silicon at such high pressure and temperature (P–T). The chemical equilibrium between perovskite, ferropericlase, and molten iron at the P–T conditions of the core–mantle boundary (CMB) was calculated in Mg–Fe–Si–O system from these experimental results and previous data on partitioning of oxygen between molten iron and ferropericlase. We found that molten iron should include oxygen and silicon more than required to account for the core density deficit (<10%) when co-existing with both perovskite and ferropericlase at the CMB. This suggests that the very bottom of the mantle may consist of either one of perovskite or ferropericlase. Alternatively, it is also possible that the bulk outer core liquid is not in direct contact with the mantle. Seismological observations of a small P-wave velocity reduction in the topmost core suggest the presence of chemically-distinct buoyant liquid layer. Such layer physically separates the mantle from the bulk outer core liquid, hindering the chemical reaction between them.     

Determination of the phase boundary between ilmenite and perovskite in MgSiO3 by in situ X-ray diffraction and quench experiments

Determination of the phase boundary between ilmenite and perovskite structures in MgSiO₃ has been made at pressures between 18 and 24 GPa and temperatures up to 2000 °C by in situ X-ray diffraction measurements using synchrotron radiation and quench experiments. It was difficult to precisely define the phase boundary by the present in situ X-ray observations, because the grain growth of ilmenite hindered the estimation of relative abundances of these phases. Moreover, the slow reaction kinetics between these two phases made it difficult to determine the phase boundary by changing pressure and temperature conditions during in situ X-ray diffraction measurements. Nevertheless, the phase boundary was well constrained by quench method with a pressure calibration based on the spinel-postspinel boundary of Mg₂SiO₄ determined by in situ X-ray experiments. This yielded the ilmenite-perovskite phase boundary of P (GPa) = 25.0 (±0.2) – 0.003 T (°C) for a temperature range of 1200–1800 °C, which is generally consistent with the results of the present in situ X-ray diffraction measurements within the uncertainty of ∼±0.5 GPa. The phase boundary thus determined between ilmenite and perovskite phases in MgSiO₃ is slightly (∼0.5 GPa) lower than that of the spinel-postspinel transformation in Mg₂SiO₄. Received: 19 May 1999 / Accepted: 21 March 2000     

Deformation of lower-mantle ferropericlase (Mg,Fe)O across the electronic spin transition

Recent high-pressure studies have shown that an electronic spin transition of iron in ferropericlase, an expected major phase of Earth’s lower mantle, results in changes in its properties, including density, incompressibility, radiative thermal conductivity, electrical conductivity, and sound velocities. To understand the rheology of ferropericlase across the spin transition, we have used in situ radial X-ray diffraction techniques to examine ferropericlase, (Mg_0.83,Fe_0.17)O, deformed non-hydrostatically in a diamond cell up to 81 GPa at room temperature. Compared with recent quasi-hydrostatic studies, the range of the spin transition is shifted by approximately 20 GPa as a result of the presence of large differential stress in the sample. We also observed a reduction in incompressibility and in the unit cell volume of 3% across the spin transition. Our radial X-ray diffraction results show that the {0 0 1} texture is the dominant lattice preferred orientation in ferropericlase across the spin transition and in the low-spin state. Viscoplastic self-consistent polycrystal plasticity simulations suggest that this preferred orientation pattern is produced by {1 1 0}<1–10> slip. Analyzing our radial X-ray diffraction patterns using lattice strain theory, we evaluated the lattice d-spacings of ferropericlase and Mo as a function of the ψ angle between the compression direction and the diffracting plane normal. These analyses give the ratio between the uniaxial stress component (t) and the shear modulus (G) under constant stress condition, which represents a proxy for the supported differential stress and elastic strength. This ratio in the mixed-spin and low-spin states is lower than what is expected from previous studies of high-spin ferropericlase, indicating that the spin transition results in a reduced differential stress and elastic strength along with the volume reduction. The influence of the spin transition on the differential stress and strength of ferropericlase is expected to be less dominant across the wide spin transition zone at high pressure–temperature conditions relevant to the lower mantle.     

An Experimental Melting—crystallization Study on Basalt at High Pressures

A series of melting-crystallization experiments on alkali basalt samples from Minqing, Fujian Province was carried out in dry and waterbearing systems at high pressures. A high-pressure melting curve was obtained. The results indicate that clinopyroxene crystallized from basalt melt at 13.5–23.7 kbar. spinel at 23.7–28.6 kbar and garnet at > 28.6 kbar. With increasing pressure, the CaSiO₃ contents of clinopyroxenes increase; and the FeSiO₃ decreases, but the chemical composition of garnet does not show any significant difference. The minerals are larger and euhedral in the water-bearing system. Therefore, we consider that natural megacrysts of the basalt can crystallize from the water-bearing basalt magma at high pressure. So the megacrysts may be derived from the upper mantle as a result of magmatic crystallization-fractionation under high pressure. This project was financially supported by the National Natural Science Foundation of China.     

Micro- and nano-inclusions in a superdeep diamond from São Luiz,Brazil

We report cloudy micro- and nano-inclusions in a superdeep diamond from São-Luiz, Brazil which contains inclusions of ferropericlase (Mg, Fe)O and former bridgmanite (Mg, Fe)SiO₃ and ringwoodite (Mg, Fe)₂SiO₄. Field emission-SEM and TEM observations showed that the cloudy inclusions were composed of euhedral micro-inclusions with grain sizes ranging from tens nanometers to submicrometers. Infrared absorption spectra of the cloudy inclusions showed that water, carbonate, and silicates were not major components of these micro- and nano-inclusions and suggested that the main constituent of the inclusions was infrared-inactive. Some inclusions were suggested to contain material with lower atomic numbers than that of carbon. Mineral phase of nano- and micro-inclusions is unclear at present. Microbeam X-ray fluorescence analysis clarified that the micro-inclusions contained transition metals (Cr, Mn, Fe, Co, Ni, Cu, Zn) possibly as metallic or sulfide phases. The cloudy inclusions provide an important information on the growth environment of superdeep diamonds in the transition zone or the lower mantle.     

Effect of KCl on melting in the Mg<Subscript>2</Subscript>SiO<Subscript>4</Subscript>–MgSiO<Subscript>3</Subscript>–H<Subscript>2</Subscript>O system at 5 GPa

To examine the effect of KCl-bearing fluids on the melting behavior of the Earth’s mantle, we conducted experiments in the Mg₂SiO₄–MgSiO₃–H₂O and Mg₂SiO₄–MgSiO₃–KCl–H₂O systems at 5 GPa. In the Mg₂SiO₄–MgSiO₃–H₂O system, the temperature of the fluid-saturated solidus is bracketed between 1,200–1,250°C, and both forsterite and enstatite coexist with the liquid under supersolidus conditions. In the Mg₂SiO₄–MgSiO₃–KCl–H₂O systems with molar Cl/(Cl + H₂O) ratios of 0.2, 0.4, and 0.6, the temperatures of the fluid-saturated solidus are bracketed between 1,400–1,450°C, 1,550–1,600°C, and 1,600–1,650°C, respectively, and only forsterite coexists with liquid under supersolidus conditions. This increase in the temperature of the solidus demonstrates the significant effect of KCl on reducing the activity of H₂O in the fluid in the Mg₂SiO₄–MgSiO₃–H₂O system. The change in the melting residues indicates that the incongruent melting of enstatite (enstatite = forsterite + silica-rich melt) could extend to pressures above 5 GPa in KCl-bearing systems, in contrast to the behavior in the KCl-free system.     

The influence of Al2O3 on the H2O content in periclase and ferropericlase at 25 GPa

In this paper I present results of IR spectroscopic measurements of water solubility in Al-bearing periclase and ferropericlase (Mg# = 88) synthesized at 25 GPa and 1400–2000 °C. The IR spectra of their crystals show narrow absorption peaks at 3299, 3308, and 3474 cm^?1. The calculated H₂O contents are 11–25 ppm in periclase (Al₂O₃ = 0.9–1.2 wt.%) and 14–79 ppm in ferropericlase (Al₂O₃ = 0.9–2.9 wt.%). Ferropericlase contains more H₂O and Al₂O₃ than periclase at 1800–2000 °C. I suggest that addition of Al₂O₃ does not influence the solubility of water in ferropericlase but can favor the additional incorporation of Fe₂O₃ into the structure. The incorporation of Fe³⁺ into ferropericlase increases water solubility as a result of iron reduction to Fe²⁺. It is shown that water has limited solubility in ferropericlase from mantle peridotite; therefore, ferropericlase cannot be considered an important hydrogen-bearing mineral in the lower mantle.