Oxidation States of Fe in Constituent Minerals of a Spinel Lherzolite Xenolith from the Tariat Depression, Mongolia: The Significance of Fe³⁺ in Olivine

Abstract

The oxidation states of Fe within olivine, orthopyroxene, clinopyroxene, and spinel in a spinel lherzolite xenolith from the Tariat Depression, Mongolia were investigated using ⁵⁷Fe Mössbauer spectroscopy to evaluate the redox condition of the upper mantle from which the Tariat spinel lherzolite xenolith was derived. The purity of separated minerals for the Mössbauer spectroscopic analysis was examined using X-ray powder diffraction, Raman spectroscopy, and transmission electron microscopy. Average Fo and Fe contents of olivine at the core part of the xenolith are 89.9(4) mol % and 0.195(3) atoms per formula unit, respectively. The Fe³⁺/ΣFe values of the olivine, orthopyroxene, clinopyroxene, and spinel, determined by Mössbauer spectroscopic analysis, are 0.027(2), 0.15(1), 0.26(3), and 0.34(5), respectively. The Mössbauer spectrum of olivine consists of two doublets assigned to Fe²⁺ at the octahedral sites and one doublet, with I.S. of 0.40(2) mm/s and Q.S. of 0.69(3) mm/s assigned to Fe³⁺ at the octahedral site. Since the Tariat spinel lherzolite xenolith in this study shows no evidence of metasomatism or thermal alteration, the existence of a small amount of Fe³⁺ in olivine and the fairly high Fe³⁺ contents of clinopyroxene, orthopyroxene, and spinel imply that the upper mantle under the Tariat area was in a rather oxidized condition.

1. Introduction

Oxidation states of Fe within constituent minerals of mantle peridotite can be used as an indicator for the evaluation of redox conditions in the upper mantle, and have been studied repeatedly [1,2,3,4,5,6,7,8,9,10,11,12,13,14]. In some regions of the upper mantle, oxygen fugacity (fO₂) is expected to be higher than that defined by the fayalite + magnetite + quartz buffer (e.g., [2,15]), and a non-negligible amount of ferric iron is contained in orthopyroxene, clinopyroxene, and the spinel of the upper mantle [1,2,3,4,5,6,7,8,9,10,11,12,13,14]. Pyroxene, spinel, and garnet are the most important carriers of Fe³⁺ in the upper mantle [16]: clinopyroxene, orthopyroxene, and spinel in mantle xenoliths have Fe³⁺/ΣFe values of 0.12–0.33, 0.04–0.09, and 0.05–0.40, respectively [7].

In contrast, olivine, the dominant mineral in the upper mantle, is not allowed to contain trivalent cations such as Fe³⁺ in its structure, in terms of its stoichiometry. In fact, although Banfield et al. [6] detected Fe³⁺ in olivine from metasomatized mantle xenoliths, a source of Fe³⁺ was attributed to submicroscopic intergrowths of laihunite in the olivine. However, Ejima et al. [13] proved the occurrence of Fe³⁺ at the octahedral sites within olivine in a mantle lherzolite xenolith from Oki Island. Ejima et al. [17] suggested that a part of Fe²⁺ within the olivine might have changed to Fe³⁺ by the annealing of the xenoliths by heat from the host magma. Therefore, when we apply the existence of Fe³⁺ in olivine from the mantle xenoliths for the estimation of the redox condition of the upper mantle, an effect of high temperature oxidation during the transportation from the upper mantle to the near surface on the mantle xenoliths has to be taken into consideration.

The Tariat Depression, Mongolia, is a well-known mantle xenolith locality in Central Asia [18]. The spinel in spinel lherzolite xenoliths from Tariat has been known to contain ferric iron (Fe³⁺/ΣFe = 0.16–0.24 [2]), suggesting a rather high fO₂ condition in the source region of the Tariat spinel lherzolite xenolith. However, the oxidation states of Fe within olivine and other constituent minerals of these xenoliths have not been reported.

In this study, the chemical composition and oxidation states of Fe of olivine, orthopyroxene, clinopyroxene, and spinel in a spinel lherzolite xenolith from Tariat were determined using electron microprobe analysis and ⁵⁷Fe Mössbauer spectroscopy, respectively, to examine the redox condition of the upper mantle from which the Tariat spinel lherzolite xenolith was derived. In order to evaluate the purity of the separated minerals for the Mössbauer spectroscopic analysis, X-ray powder diffraction, Raman spectroscopic, and transmission electron microscopy (TEM) methods were also used.

2. General Geology and Sampling Site

The Tariat Depression is one of the most typical areas of deep-seated megacrystic xenoliths and mantle-derived xenoliths in the Baikal-Mongolia rift. It is located approximately 500 km southwest of Lake Baikal. The Tariat Depression is mainly characterized by alkalic volcanic rocks of Pliocene, Pleistocene, and Holocene in age. Abundant upper mantle and crustal xenoliths occur in this area [19]. Mantle xenoliths consist of spinel-garnet-bearing websterite, garnet lherzolite, and spinel lherzolite [20]. The sampling site in the present study is located at southeast of Tarkhin Tsagaan Lake (Figure 1).

3. Experimental Methods

3.1. Sample Preparation and Purity Evaluation

One thin section from the core of the xenolith and two thin sections from the contact part of the xenolith and the host basalt were prepared for petrographic observations and chemical analysis of constituent minerals.

Olivine, orthopyroxene, clinopyroxene, and spinel grains were separated from the lherzolite xenolith for Mössbauer spectroscopic analyses. Grains of each mineral were dissected out from the core of the xenolith and examined under a binocular microscope to make sure that other minerals are not included. The purities of the separated grains were checked by electron microprobe analysis (EMPA), X-ray powder diffraction (XRD) analysis, and Raman spectroscopy. The separated grains were manually crushed into powder using an agate pestle and mortar.

3.2. Electron Microprobe Analysis

Chemical compositions of minerals were determined using JEOL JXA-8530M electron microprobe analyzer (JEOL, Tokyo, Japan) at Kyushu University and National Institute of Advanced Industrial Science and Technology of Japan (AIST), operated at an accelerating voltage of 15 kV, with a beam current of 20 nA and a beam diameter of 5 μm. The standard materials were diopside for Si (K_α), Mg (K_α), and Ca (K_α); and rutile for Ti (K_α), sanidine for Al (K_α), Cr₂O₃ for Cr (K_α), almandine for Fe (K_α), bustamite for Mn (K_α), NiO for Ni (K_α), and anorthite for Na (K_α).

3.3. Raman Spectroscopy

Raman spectra were measured using the Ram532 micro-Raman spectroscopy system (JASCO, Tokyo, Japan) at Kyushu University. Excitation was by means of the 532 nm line of a solid-state ion laser with an output power of 20 mW. The instrument was equipped with a microscope with a focal spot size of 2 μm.

3.4. Transmission Electron Microscopy (TEM)

To evaluate purity of the olivine, nanoscale observation of an olivine grain in the thin section was performed using high-resolution TEM (HR-TEM). A TEM foil, about 5 µm × 11 µm × 0.1 μm in size, was prepared from an olivine sample using a JOEL JEM-9310 (JEOL, Tokyo, Japan) focused ion beam (FIB) system at Ehime University. The detailed FIB milling procedure for TEM foil preparation is described elsewhere [21]. TEM and HR-TEM observations were carried out with a JEOL JEM-2010 (JEOL, Tokyo, Japan) at Ehime University, operated at 200 kV. Selected-area electron diffraction (SAED) and bright-field imaging were employed to characterize the microstructure of the sample.

3.5. X-ray Powder Diffraction Measurement and Rietveld Analysis

XRD data of olivine, orthopyroxene, and clinopyroxene were collected at Shimane University using a Rigaku RINT automated Bragg-Brentano diffractometer system equipped with a curved graphite diffracted-beam monochromator (RIGAKU, Tokyo, Japan). The Cu X-ray tube generator for CuK_α radiation was operated at 35 kV and 25 mA. The profile was taken in a 2θ range between 10° and 120°, using the step-scan method with a step interval of 0.02° and a count time of 5 s per step. XRD data of spinel was collected at AIST using a Rigaku RINT-automated Bragg-Brentano diffractometer system at AIST, equipped with a curved graphite diffracted-beam monochromator. The rotating anticathode Cu X-ray tube generator for CuK_α radiation was operated at 40 kV and 30 mA. The profile was measured in a 2θ range between 10° and 120° with a step interval of 0.02° and a count time of 5 s per step.

The measured XRD data of olivine were analyzed by the Rietveld method using the RIETAN-FP program by Izumi and Momma [22]. The modified split pseudo-Voigt function was used as the profile function, and correction for preferred orientation was done by using the March-Dollase function [23].

3.6. ⁵⁷Fe Mössbauer Spectroscopy

Powdered samples of the separated olivine, clinopyroxene, orthopyroxene, and spinel grains from the core of the xenolith were analyzed by ⁵⁷Fe Mössbauer spectroscopy (TOPOLOGIC SYSTEMS, Tokyo, Japan) to determine the oxidation state and distribution of Fe in each crystallographic site. The Mössbauer spectra were measured at room temperature in the Doppler velocity range of −4 to 4 mm/s at Shimane University, using 370 MBq ⁵⁷Co in Pd as a source. The absorber was about 20 mg of finely ground sample. The Mössbauer data were obtained with a constant acceleration spectrometer fitted with a 1024 multichannel analyzer. The isomer shift is relative to metallic iron foil. The QBMOSS program of Akasaka and Shinno [24] was used for computer analysis. The quality of the fit was judged from the χ² value and standard deviations of the Mössbauer parameters.

4. Results

4.1. Petrography

The studied spinel lherzolite xenolith sample is about 10 cm in diameter. It has a porphyroclastic texture and consists of olivine, orthopyroxene, clinopyroxene, and spinel (Figure 2). The olivine is euhedral to subhedral in form and about 1–4 mm in size (Figure 2a–d). Kink-bands are present in the olivine crystals. The orthopyroxene is subhedral in form and about 0.2–6 mm in size (Figure 2a). The spinel is anhedral in form and about 0.5–3 mm in size (Figure 2c). Clinopyroxene is subhedral to anhedral in form and about 0.2–5 mm in size, and has the rim consisting of diopside and glass (Figure 2a,c). Near the surface of the spinel lherzolite xenolith that is in contact with the host basalt, orthopyroxene has a reaction rim, and the clinopyroxene has been completely broken into diopside and glass (Figure 2e,f).

The host basalt consists of olivine, clinopyroxene, and plagioclase phenocrysts (Figure 2h) and groundmass composed of euhedral plagioclase and magnetite, exhibiting intersertal texture. Olivine and clinopyroxene phenocrysts are euhedral to subhedral, and 10–200 µm in size. Plagioclase phenocrysts are euhedral to subhedral, and about 10–300 µm in size.

4.2. Chemical Compositions of Constituent Minerals

Chemical compositions of the constituent minerals of the spinel lherzolite are shown in

Table 1. The average chemical compositions of constituent minerals in the xenolith and basalt.

	Xenolith								Basalt
	Olivine		Spinel		Orthopyroxene		Clinopyroxene		Olivine		Clinopyroxene
	Av. (n = 48) *	s.d. *	Av. (n = 36) *	s.d. *	Av. (n = 22) *	s.d. *	Av. (n = 18) *	s.d. *	Av. (n = 18) *	s.d. *	Av. (n = 23) *	s.d. *
SiO2	40.84	0.20	-	-	55.06	0.33	52.02	0.38	38.38	0.59	50.81	1.17
TiO2	-	-	0.13	0.02	0.12	0.02	0.61	0.05	-	-	1.75	0.57
Al2O3	-	-	57.54	0.29	4.67	0.08	7.32	0.20	-	-	3.23	1.14
Cr2O3	-	-	11.52	0.20	0.40	0.03	0.97	0.07	-	-	-	-
FeO **	9.53	0.14	9.83	0.12	5.97	0.09	2.69	0.13	23.76	2.80	7.37	0.75
MnO	0.14	0.03	-	-	0.16	0.05	0.09	0.04	0.32	0.06	0.14	0.04
NiO	0.19	0.02	0.19	0.02	0.05	0.01	-	-	0.05	0.03	-	-
MgO	49.30	0.28	20.79	0.48	32.56	0.54	14.77	0.14	37.62	2.37	14.40	0.85
CaO	0.05	0.01	-	-	0.66	0.04	18.51	0.61	0.30	0.05	21.67	0.56
Na2O	-	-	-	-	0.19	0.03	2.53	0.20	-	-	0.51	0.06
Total	100.03		100.00		99.83		99.50		100.54		99.98
Number of Cations on the Basis of 4 Oxygens					Number of Cations on the Basis of 6 Oxygens				O = 4		O = 6
Si	0.999	0.004	-	-	1.902	0.012	1.884	0.010	1.000	0.005	1.887	0.042
Ti	-	-	0.003	0.000	0.003	0.001	0.017	0.001	-		0.049	0.016
Al	-	-	1.751	0.006	0.190	0.003	0.312	0.009	-	-	0.142	0.050
Cr	-	-	0.235	0.005	0.011	0.001	0.028	0.002	-	-	-	-
Fe2+	0.195	0.003	0.212	0.003	0.172	0.003	0.082	0.004	0.519	0.068	0.229	0.025
Mn	0.003	0.001	-	-	0.005	0.001	0.003	0.001	0.007	0.001	0.004	0.001
Ni	0.004	0.000	0.004	0.000	0.001	0.000	-	-	0.001	0.001	-	-
Mg	1.798	0.007	0.800	0.014	1.677	0.024	0.797	0.007	1.460	0.071	0.796	0.041
Ca	0.001	0.000	-	-	0.024	0.002	0.718	0.025	0.009	0.002	0.863	0.026
Na	-	-	-	-	0.013	0.002	0.178	0.014	-	-	0.037	0.004
Total	2.999		3.005		4.000		4.018		2.999		4.011
Fo mol %	90

* Av.: Average; s.d.: standard deviation. ** Total Fe as FeO.

, in which total Fe is represented as FeO, and average compositions with the standard deviations of the minerals are listed. The EMPA data indicate that olivine, orthopyroxene, clinopyroxene, and spinel are quite homogeneous in chemical composition, as shown by the standard deviations of the average values. Average Fo value and Fe content (n = 48) of olivine at the core part of the xenolith are 89.9 mol % and 0.195(3) atoms per formula unit (apfu; O = 4), respectively. Since Ejima et al. [13,17] discussed possible influence of heat from basalt magma on xenolith, the authors examined the chemical composition of olivine in contact with host basalt. EMPA analysis of olivine at the positions of 10, 15, and 30 µm from the surface resulted in Fo_78.3, Fo_88.7, and Fo_89.8 mol %, respectively, implying that, even in the olivine in contact with the basalt, the influence of the basaltic magma is limited to the very surface of olivine. The orthopyroxene is enstatite with the formula (Mg_0.84Fe²⁺_0.09Al_0.05Ca_0.01)_Σ0.99(Si_0.95Al_0.05)_Σ1.00O₃ on average (n = 22). The clinopyroxene is a Na-bearing diopside: (Na_0.18Ca_0.72Mg_0.80Fe²⁺_0.08Cr_0.03Ti_0.02Al_0.19)_Σ2.02(Si_1.88Al_0.12)_Σ2.00O₆ on average (n = 18). By assuming electronic neutrality for the stoichiometric clinopyroxene, Fe²⁺:Fe³⁺ = 0.027:0.054 (in atoms per formula unit) is calculated, which gives the chemical formula of (Na_0.177Ca_0.715Mg_0.794Fe²⁺_0.027 Fe³⁺_0.054Cr_0.028Ti_0.017Al_0.186)_Σ2.000(Si_1.875Al_0.125)_Σ2.000O₆. The chemical formula of spinel derived from the EMPA data is (Mg_0.800Fe²⁺_0.212Ni_0.004)_Σ1.016(Al_1.751Cr_0.235)_Σ1.986O₄ on average (n = 36). Charge balance calculation results in 0.012 Fe³⁺ per formula unit, and, thus, the chemical formula is rewritten as (Mg_0.800Fe²⁺_0.200Ni_0.004)_Σ1.004(Al_1.751Cr_0.235Fe³⁺_0.012)_Σ1.998O₄.

Olivine phenocrysts in the host basalt have normal zoning from Fo₇₉ at the core to Fo₆₅ at the rim. Clinopyroxene phenocrysts are on the average composition of (Na_0.04Ca_0.86Mg_0.80Fe²⁺_0.23Al_0.03Ti_0.05)_Σ2.01(Si_1.89Al_0.11)_Σ2.00O₆ (n = 23) (Table 1).

4.3. Purity of the Samples Separated for Mössbauer Analysis

The minerals for Mössbauer analysis are separated from core of xenolith to avoid contact part with the host basalt. XRD patterns of each mineral dissected out from the core of the xenolith indicate no impurities, as described below. Further observation of five grains chosen from the purified grains of each mineral, using a reflection microscope and EMPA, and Raman spectroscopic analysis detected no impurities.

4.4. X-ray Powder Diffraction and X-ray Rietveld Analysis

The XRD patterns of orthopyroxene, clinopyroxene, and spinel are shown in Figure 3, indicating that the separated orthopyroxene and clinopyroxene samples are single phases, and that the spinel sample was contaminated with a very small amount of quartz, which was used for cleaning of the agate mortar.

Details of the X-ray diffraction data collection and the crystallographic data of olivine are shown in

Table 2. Details of data collection and crystallographic data of the separated olivine grains from the core of the xenolith.

	Olivine
Radiation	CuKα (λ = 1.541769 Å)
Monochromator	Graphite
2θ scan range (°)	5.00–120
Step interval (°2θ)	0.02
Max. intensity (counts)	5750
No. of phases refined	1
Space groupe	Pbnm
a (Å)	4.7631(1)
b (Å)	10.2284(1)
c (Å)	5.9941(1)
V (Å3)	292.022(7)
Z	4
Dcalc (g/cm3)	3.32
Rp (%) *	9.55
Rwp (%) *	12.86
Re (%) *	10.18
S *	1.26

* R_p, R-pattern; R_wp, R-weighted pattern; R_e, R-expected; S, Goodness-of-fit [28].

, and the results of the Rietveld analyses using the XRD data of the olivine are shown in Figure 4. To refine the site occupancies of Mg and Fe at the octahedral M1 and M2 sites in olivine with the structural formula ^VIM2^VIM1^IVTO₄, Mg + Fe at each site was constrained to be 1.0. Site occupancy of the tetrahedral site was fixed to 1.0Si. In the structural refinement, the weighted-pattern reliability factor R_wp and goodness-of-fit were reduced to 12.86 and 1.26, respectively. The refined Mg and Fe occupancies are Mg = 0.902(4) and Fe = 0.098 at the M1 site, and Mg = 0.926(4) and Fe = 0.074 at the M2 site (

Table 3. Site occupancies (g) *, atomic coordinates (x, y, z) *, and isotropic displacement parameters (B_eq in Ų) * of the olivine.

Olivine
Site	n **	g	x	y	z	Beq
M1	4	Mg0.902(4)Fe0.098	0	0	0	1.58(5)
M2	4	Mg0.926(4)Fe0.074	0.9860(4)	0.2784(1)	1/4	1.58(6)
T	4	1.0Si	0.4288(4)	0.0949(1)	1/4	1.77(5)
O1	4	1	0.7636(7)	0.0927(3)	1/4	1.8(1)
O2	4	1	0.2159(7)	0.4482(4)	1/4	1.12(9)
O3	8	1	0.2767(5)	0.1627(2)	0.0325(5)	1.81(8)

* Numbers in parentheses represent standard deviations. ** n_eq = multiplicity of sites.

), in which standard deviations, 1σ, are shown in the parentheses. The total Fe population per 4 oxygens is 0.172 apfu, which is close to 0.195(3) apfu given by EMPA (Table 1). The refined atomic positions are listed in Table 3.

4.5. ⁵⁷Fe Mössbauer Spectroscopic Analysis

The Mössbauer spectra of the separated olivine, orthopyroxene, clinopyroxene, and spinel are shown in Figure 5. The Mössbauer hyperfine parameters of the doublets, such as isomer shift (I.S.) and quadrupole splitting (Q.S.), and their assignments are presented in

Table 4. ⁵⁷Fe Mössbauer hyperfine parameters and area ratios of the Mössbauer doublets of olivine, orthopyroxene, clinopyroxene, and spinel separates.

Mineral	Doublets	I.S. * (mm/s)	Q.S * (mm/s)	FWHH * (mm/s)	Area Ratio (%)	Assignments	χ2/Freedom	Fe2+:Fe3+
Olivine	AA′	1.165(2)	3.081(3)	0.228(3)	49.8(14)	M2Fe2+ in olivine	1.65	97.3(14):2.7(2) **
	BB′	1.141(2)	2.899(4)	0.228(3)	46.1(14)	M1Fe2+ in olivine
	CC′	0.40(2)	0.69(3)	0.228(3)	2.7(2)	M2Fe3+ in olivine
	DD′	1.13(3)	1.48(6)	0.228(3)	1.4(2)	Fe2+ in spinel
Orthopyroxene	AA′	0.39(2)	0.86(3)	0.38(1)	15(1)	M1Fe3+	1.14	85(8):15(1)
	BB′	1.19(3)	2.21(5)	0.38(1)	27(6)	M1Fe2+
	CC′	1.10(5)	2.05(10)	0.38(1)	58(6)	M2Fe2+
Clinopyroxene	AA′	0.43(1)	0.88(2)	0.51(4)	16(2)	M1Fe3+	1.21	74(4):26(3) ***
	BB′	1.06(1)	2.00(2)	0.25(4)	11(2)	M2Fe2+
	CC′	1.09(1)	2.54(2)	0.45(3)	34(3)	M1Fe2+
	DD′	0.32(1)	0.28(2)	0.51(4)	30(3)	Unidentified
	EE′	0.36(1)	2.28(2)	0.51(4)	8(1)	Unidentified
Spinel	AA′	0.88(2)	1.30(3)	0.57(9)	36(3)	Fe2+	1.41	66(6):34(5) ***
	BB′	0.94(2)	1.99(4)	0.37(5)	18(7)	Fe2+
	CC′	0.19(1)	0.71(3)	0.50(4)	28(5)	Fe3+
	DD′	−0.1(2)	1.9(3)	1.2(2)	18(4)	Unidentified

* I.S., isomer shift relative to a metallic iron absorber; Q.S., quadrupole splitting; FWHH, full width at half peak height. ** Fe²⁺ in spinel is excluded. *** The Fe²⁺:Fe³⁺-ratios of clinopyroxene and spinel were calculated based on the area ratios of the doublets assigned to Fe²⁺ and Fe³⁺ in each mineral.

.

The Mössbauer spectrum of olivine consists of four doublets (Figure 5a), since Shinno et al. [29] found I.S. = 1.1–1.2 mm/s, Q.S. = 3.0–3.2 mm/s and I.S. = 1.1–1.2 mm/s, and Q.S. = 2.8–2.9 mm/s for the doublets by Fe²⁺ at the octahedral M2 and M1 sites of olivine, respectively. The doublets AA′ (I.S. = 1.165(2) mm/s; Q.S. =3.081(3) mm/s) and BB′ (I.S. = 1.141(2) mm/s; Q.S. =2.899(4) mm/s) are assigned to Fe²⁺ at the M2 and M1 sites, respectively. The doublet CC′ with I.S. = 0.40(2) and Q.S. = 0.69(3) mm/s is assigned to Fe³⁺ at M2 site of olivine, based on the results of Ejima and Akasaka [30] (I.S. = 0.38–0.40 mm/s; Q.S. = 0.57–0.93 mm/s). The Fe³⁺/ΣFe value, determined using the area ratios of the doublets, is 0.027(2). From this result and the average Fe content obtained by EMPA, 0.195(3) apfu, Fe³⁺ in the olivine is derived as 0.005(2) apfu on average. The doublet DD′ (I.S. = 1.13(3) mm/s; Q.S. = 1.48(6) mm/s) is assigned to Fe²⁺ in spinel, as explained below. Although spinel inclusion in olivine was not detected by EMPA observation, X-ray diffraction analysis, and Raman spectroscopic analysis, the Mössbauer sample is considered to contain olivine grains including spinel inclusions.

The spectrum of orthopyroxene is fitted to three doublets (Figure 5b). Based on the hyperfine parameters of the doublets by Fe³⁺ and Fe²⁺ at the octahedral M1 site and of Fe²⁺ at the octahedral M2 site after Annersten et al. [31] (I.S. = 0.45 and Q.S. = 0.70–0.86 mm/s; I.S. = 1.10 and Q.S. = 2.77 mm/s; I.S. = 1.12–1.13 and Q.S. = 2.22–2.25 mm/s, respectively), the doublets AA′, (I.S. = 0.39(2) mm/s; Q.S. = 0.86(3) mm/s) BB′ (I.S. = 1.19(3) mm/s; Q.S. = 2.21(5) mm/s), and CC′ (I.S. = 1.10(5) mm/s; Q.S. = 2.05(10) mm/s) are assigned to Fe³⁺ at the M1 site, Fe²⁺ at the M1 site, and Fe²⁺ at the M2 site, respectively. The Fe³⁺/ΣFe value is 0.15(1).

The spectrum of the clinopyroxene sample consists of five doublets (Figure 5c). Three doublets are assigned to Fe in clinopyroxene: the doublet AA′ (I.S. = 0.43(1) mm/s; Q.S. = 0.88(2) mm/s) corresponds to that attributed to Fe³⁺ at the octahedral M1 site (I.S. = 0.32–0.37 and Q.S. = 0.88–0.90 mm/s after Akasaka [32,33]; BB′ (I.S. = 1.06(1) mm/s; Q.S. = 2.00(2) mm/s) is assigned to Fe²⁺ at the 8-coordinated M2 site based on the data (I.S. = 1.10–1.14 mm/s; Q.S. = 1.76–1.91 mm/s) by Dowty and Lindsley (1973); and the doublet CC′ (I.S. = 1.09(1) mm/s; Q.S. = 2.54(2) mm/s) to Fe²⁺ at the M1 site after the data of Akasaka [33], Dowty and Lindsley [34], and Williams et al. [35] (I.S. = 1.14–1.19 mm/s; Q.S. = 2.05–2.57 mm/s). The DD′ and EE′ doublets, with I.S. = 0.32(1) and Q.S. = 0.28(2) mm/s and I.S. = 0.36(1) and Q.S. = 2.28(2) mm/s, respectively, are not consistent with ⁵⁷Fe Mössbauer hyperfine parameters of Fe in clinopyroxene and other minerals reported by now. As described already, clinopyroxene has the rim consisting of diopside and glass. Therefore, we assume that the doublets DD′ and EE′ are attributed to Fe in the glass of the rim of the clinopyroxene. The Fe³⁺/ΣFe value of the clinopyroxene calculated from the area ratios of the doublets AA′, BB′, and CC′ is 0.26(3).

Four doublets are fitted to the Mössbauer spectrum of the separated spinel sample (Figure 5d). The doublets AA′ (I.S. = 0.88(2) mm/s; Q.S. = 1.30(3) mm/s) and BB′ (I.S. = 0.94(2) mm/s; Q.S. = 1.99(4) mm/s) are assigned to Fe²⁺ at the octahedral sites based on the data of hercynite (FeAl₂O₄) (I.S. = 0.91 mm/s; Q.S. = 1.57 mm/s) after Osborne et al. [36] and of ulvöspinel (Fe₂TiO₄) (I.S. = 0.83 mm/s; Q.S. = 1.91 mm/s) after Malysheva et al. [37], and CC′ (I.S. = 0.19 mm/s; Q.S. = 0.71 mm/s) is assigned to Fe³⁺ at the octahedral sites after the data (I.S. = 0.29 mm/s; Q.S. = 0.92 mm/s) by Warenborgh et al. [38]. The rather broad line width of the doublet AA′, FWHH = 0.57(9) mm/s (

Table 5. Published data of Fe³⁺/ΣFe of minerals from spinel lherzolite xenoliths, determined by Mössbauer spectroscopic analysis.

Reference	Occarrence	Rock Type	Fe3+/ΣFe
			Olivine	Clinopyroxene	Orthopyroxene	Spinel
Wood and Virgo [2]	Kilbourme Hole, New Mexico	Spinel-lherzolite xenolith	-	-	-	0.18–0.25
Wood and Virgo [2]	Massif Central, France	Spinel-lherzolite xenolith	-	-	-	0.25–0.28
Wood and Virgo [2]	Ichinomegata, Japan	Spinel-lherzolite xenolith	-	-	-	0.26–0.28
Wood and Virgo [2]	Tariat, Mongolia	Spinel-lherzolite xenolith	-	-	-	0.16–0.24
Wood and Virgo [2]	Vitim Plateau, Mongolia	Spinel-lherzolite xenolith	-	-	-	0.15–0.27
Canil et al. [3]	Southeastern Australia	Spinel-lherzolite xenolith	0	0.16–0.19	0.06	0.15–0.22
Canil et al. [3]	Massif Central, France	Spinel-lherzolite xenolith	0	0.19	0.05	0.26
Woodland et al. [11]	Beni Bousera, Morocco	Orogenic lherzolite massifs	0	0.03–0.14	0.02–0.06	0.02–0.13
Woodland et al. [14]	Ronda, Spain	Orogenic lherzolite massifs	0	0.06–0.19	0.04–0.08	0.08–0.26
Woodland et al. [11]	Pyrenees, France	Orogenic lherzolite massifs	0	0.19–0.32	0.05–0.07	0.10–0.27
Perinelli et al. [52]	Northern Victoria Land, Antarctica	Spinel lherzolite xenolith	-	-	-	0.17–0.23
Hao and Li [14]	Eastern China	Spinel lherzolite xenolith	0	0.22–0.31	0.05–0.13	0.14–0.23
This study	Tariat, Mongolia	Spinel lherzolite xenolith	0.03	0.26	0.15	0.34

), suggests strong overlapping of a couple of doublets. It is considered that the doublet AA′ in our study corresponds to two doublets Fe²⁺-(I) and Fe²⁺-(III) of Hao and Li [14]. The doublet DD′ is not assigned to Fe in spinel, because the I.S. and Q.S. of the doublet DD′ do not fit to those of Fe²⁺ nor Fe³⁺ in spinel group minerals. It may be attributable to Fe in some inclusion that was not detected in the XRD pattern. The Fe³⁺/ΣFe value of the spinel is 0.34(5).

4.6. Raman Spectroscopy, Transmission Electron Microscopy, and Electron Diffraction Analysis

Raman spectra were measured at 10 positions on the five separated olivine grains, and at 15 positions on olivine in a thin section of the core part of the xenolith. All of the Raman spectra measured match forsteriteR060551 with a composition of (Mg_1.82Fe_0.18)SiO₄ in the Raman database (Figure 6) and no additional phases are detected.

As shown by a bright-field TEM image (Figure 7a) and a HR-TEM image of the SAED pattern of the olivine foil (Figure 7b), none of inclusion and impurity are observed in the olivine foil.

5. Discussion

5.1. Construction of Structural Formula of Olivine, Orthopyroxene, Clinopyroxene, and Spinel

In this study, we showed that separated olivine grains from the core of the spinel lherzolite xenolith have a small amount of ferric iron (0.027(2) in Fe³⁺/ΣFe value and 0.005(2) apfu, on average). Although the Mössbauer hyperfine parameters indicate existence of Fe³⁺ cations at the M2 site in the olivine, laihunite and laihunite layer have not been detected by HR-TEM observation and Raman spectroscopic analysis, as described already. Thus, the structural formula of olivine determined by EMPA and Mössbauer analysis is finally constructed as ^M2[Mg_0.893Fe²⁺_0.098Mn²⁺_0.003Ca_0.001Fe³⁺_0.005]^M1[Mg_0.905Fe²⁺_0.091Ni_0.004]Si_0.999O₄, in which Mn²⁺ and Ca ions are assigned to the M2 site based on the X-ray structural study by Ottonello et al. [39], and Ni ions to the M1 site after Nord et al. [40].

According to the chemical formula of enstatite, derived from the EMPA data, Fe³⁺ is not calculated by charge balance calculation, implying all Fe is ferrous iron. However, Mössbauer data shows that 15% of total Fe is Fe³⁺, and Fe²⁺ at the M2 and M1 sites are 58(6) and 27(6)% of total Fe, respectively. The structural formula of enstatite, based on the recalculate FeO and Fe₂O₃ contents (5.07 and 1.00 wt %, respectively), is given as ^M2[Na_0.013Ca_0.024Mg_0.849Fe²⁺_0.100Mn_0.003]_Σ0.989 ^M1[Mg_0.825Fe²⁺_0.046Fe³⁺_0.026Cr_0.011Ti_0.003Al_0.089]_Σ1.000^T[Si_1.899Al_0.101]_Σ2.000O₆, in which the assignments of Ca and Mn to the M2 site and of Al, Fe³⁺, Cr, and Ti to the M1 site are based on the X-ray structural study by Nestola et al. [41].

As explained already, the Fe²⁺:Fe³⁺-ratio of the clinopyroxene can be estimated by charge balance calculation. However, it is well known fact that the Fe²⁺:Fe³⁺-ratios calculated from EMPA are generally unreliable and commonly are not consistent with those determined by Mössbauer spectroscopic analysis. The Fe²⁺:Fe³⁺-ratio of the clinopyroxene in this study, determined by the Mössbauer spectroscopy, is 74(4):26(3) in atomic % (Table 4), which results in the recalculated FeO and Fe₂O₃ contents of 1.99 and 0.78 wt %, respectively, and the structural formula of the clinopyroxene is calculated as ^M2[Na_0.177Ca_0.717Mg_0.090Fe²⁺_0.025]_Σ1.009^M1[Mg_0.706Fe²⁺_0.035Fe³⁺_0.021Cr_0.028Ti_0.017Al_0.193]_Σ1.000^T[Si_1.881Al_0.119]_Σ2.000O_6.

Although charge balance calculation of spinel gave 0.012 Fe³⁺ per formula unit (O = 4), the Mössbauer spectroscopic analysis resulted in Fe²⁺:Fe³⁺-ratio of 66(6):34(5) in atomic %. According to the calculation of the chemical formula using recalculated FeO and Fe₂O₃ contents, 6.49 and 3.71 wt %, respectively, the chemical formula is given as (Mg_0.793Fe²⁺_0.139Ni_0.004)_Σ0.936(Al_1.735Cr_0.233Fe³⁺_0.071)_Σ2.039O₄.

5.2. Significance of Oxidation States of Fe in Olivine, Clinopyroxene, Orthopyroxene, and Spinel

5.2.1. Olivine

In the published studies of the oxidation state of Fe in olivine from mantle-derived xenoliths, only olivine in metasomatized spinel lherzolite from Dish Hill [5,6] and Oki Island [13] have been reported to contain a small amount of Fe³⁺ (0.01–0.06 Fe³⁺/ΣFe). However, in the former, the Fe³⁺ detected in olivine was attributed to that within laihunite layers in the olivine, which might have formed by metasomatism [42]. On the other hand, in the latter, Fe³⁺ within olivine has been proved to be located at the octahedral sites [13], but a part of Fe²⁺ within the olivine may have changed to Fe³⁺ by annealing of the xenoliths due to heat from the host magma [13,17]. In contrast, olivine from the spinel lherzolite xenolith in this study does not contain detectable inclusions, such as laihunite or precipitates, and there is no evidence of alteration by metasomatism or high-temperature oxidation. Therefore, it is concluded that the Fe³⁺ within olivine in the Tariat xenolith was generated at rather oxidized conditions in the upper mantle, where Fe³⁺ was stable.

As reviewed by Ashworth and Chambers [43], one possible mechanism by which Fe²⁺ oxidizes to Fe³⁺ in olivine is diffusion of electrons and cations [44]. The reaction of 3Fe²⁺ ⇔ 2Fe³⁺ + □ (vacancy) shows that one Fe²⁺ ion has to move from an octahedral site to outside of the olivine structure. In fact, Fe²⁺ ions diffuse to the edge of the olivine grain, as Mackwell [45] demonstrated in an annealing experiment of fayalite. This mechanism seems to apply reasonably to the generation of Fe³⁺ within olivine in the upper mantle if the redox conditions are suitable for oxidation of Fe²⁺ to Fe³⁺ in the olivine, as well as the case of high temperature oxidation of olivine in lavas and scoria [46,47,48,49].

Another model on the formation of Fe³⁺ in terms of diffusion of H₂O, OH, or H in olivine has been proposed by Hwang et al. [50]. If the release of H from the mantle olivine containing OH takes place during exhumation through a dehydrogenation–oxidation process [51], Fe³⁺ may be formed in the following manner: Fe²⁺ + (OH)⁻ → Fe³⁺ + O^2– + 0.5H₂. This process also requires movement of one Fe atom per 3Fe atoms to the outside of olivine to maintain the charge balance. In this case, owing to the very fast diffusion rate of hydrogen, the generated Fe³⁺ would not be localized in the olivine.

The first model seems to be more suitable to our conclusion that the Fe³⁺ within olivine in the Tariat xenolith was generated at rather oxidized conditions in the upper mantle.

5.2.2. Clinopyroxene, Orthopyroxene, and Spinel

The Fe³⁺/ΣFe value of the clinopyroxene in the Tariat spinel lherzolite xenolith (Fe³⁺/ΣFe = 0.26) is close to the highest value in clinopyroxene (Fe³⁺/ΣFe = 0.03–0.32) of spinel lherzolite xenoliths from other localities (Table 5). The Fe³⁺/∑Fe value of the orthopyroxene (Fe³⁺/ΣFe = 0.15) is also close to the highest value reported from unmetasomatized spinel lherzolite xenoliths (Fe³⁺/ΣFe = 0.05–0.13).

The Fe³⁺/ΣFe value of spinel in this study is 0.34, which is higher than values of spinels in Tariat spinel lherzolite (Fe³⁺/ΣFe = 0.16–0.24) reported by Wood and Virgo [2] but rather similar to values of spinel from a metasomatized spinel lherzolite xenolith reported by McGuire et al. [5] and from pyroxenite by Dyar et al. [7].

Because the Tariat spinel lherzolite xenolith in this study show no evidence of metasomatism or thermal alteration, the existence of a small amount of Fe³⁺ in olivine and the fairly high Fe³⁺ contents in orthopyroxene, clinopyroxene and spinel indicate more oxidized condition at the source upper mantle of the Tariat spinel lherzolite xenolith than other districts. However, the results in the present study and of Wood and Virgo [2] indicate variable Fe³⁺/ΣFe values of spinels in the Tariat spinel lherzolite. Thus, the redox condition of the source area in the upper mantle is not regarded as uniform.

5.3. Evaluation of Redox Condition of the Source Area of the Tariat Spinel Lherzolite Xenoliths

Hao and Li [14] concluded that the calculated fO₂ values on the olivine-orthopyroxene-spinel and olivine-orthopyroxene-clinopyroxene assemblages in their spinel lherzolite data set lie within the range of data provided elsewhere, mostly slightly below the fayalite-magnetite-quartz buffer. On the other hand, as concluded above, the Tariat spinel lherzolite in this study suggests rather oxidized condition of the upper mantle than other district. Thus, the Tariat spinel lherzolite in this study is expected to lie above the fayalite-magnetite-quartz buffer. Based on their calculation using geothermo-barometers on the garnet websterite and garnet lherzolite from the Tariat Depression, Osanai et al. [20] resulted in ca. 1100 °C and 17–23 kbar, which is very close pressure-temperature conditions of the previous studies (e.g., Harris et al. [53]). If we assume that the pressure-temperature condition of the Tariat spinel lherzolite xenolith in the present study is same as that of garnet-bearing mantle xenoliths from the same locality, Δlog(fO₂)^FMQ (fO₂ relative to the FMQ buffer) of 3.4–3.6 in log unit is given by applying an oxygen geobarometer by Ballhaus et al. [15], Δlog(fO₂)^FMQ = 0.27 + 2505/T − 400P/T − 6log($X_{\mathrm{Fe}}^{\mathrm{olv}}$) − 3200(1 − $X_{\mathrm{Fe}}^{\mathrm{olv}}$)²/T + 2log($X_{{\mathrm{Fe}}^{2+}}^{\mathrm{sp}}$) + 4log($X_{{\mathrm{Fe}}^{3+}}^{\mathrm{sp}}$) + 2630($X_{\mathrm{Al}}^{\mathrm{sp}}$)²/T, where P is in GPa, T in K, $X_{{\mathrm{Fe}}^{3+}}^{\mathrm{sp}}$ and $X_{{\mathrm{Fe}}^{2+}}^{\mathrm{sp}}$ the Fe³⁺/ΣFe and Al/ΣR³⁺ ratios in spinel, and $X_{\mathrm{Fe}}^{\mathrm{olv}}$ and $X_{{\mathrm{Fe}}^{3+}}^{\mathrm{sp}}$ the Fe²⁺/(Fe²⁺ + Mg) ratios in olivine and spinel. According to the relationship between Δlog(fO₂)^FMQ and Cr/(Cr + Al) in spinel, shown by Ballhause et al. [15], Δlog(fO₂)^FMQ values of the primitive peridotite xenoliths and metasomatized ones are ca. −3 to −1 and ca. −1 to 1 in log unit. Thus, the Δlog(fO₂)^FMQ value of the Tariat spinel lherzolite xenolith in the present study indicates higher fO₂ conditions (above FMQ buffer) than those of the primitive or metasomatized peridotite xenoliths reported so far, and is consistent with the oxidation state of Fe and high Fe³⁺ contents in the constituent minerals.

However, as discussed above, the source area of the Tariat spinel lherzolite xenoliths in the upper mantle is regarded as not uniform in redox condition. Therefore, further study for the additional Tariat xenoliths is required to clarify the variation of redox conditions in the upper mantle of the Tariat Depression area.