Distinct Geochemical Behavior of Water in Olivine between Silicate and Carbonatite Metasomatism

Abstract

Metasomatism by silicate or carbonatite melts can significantly modify the mineral physical and chemical properties, and the hydrogen geochemical behavior during the interaction of olivine and melt remains unclear. In this study, olivine wate content was measured by SIMS on two peridotite xenolith suites. The water content in the silicate-metasomatized olivine increases along with the metasomatism intensity, whereas the opposite trend is present in the carbonatite-metasomatized ones. These results reflect that H enters olivine via silicate metasomatism, whereas it is extracted by carbonatite melts due to its high-water solubility. In addition, as ⁷Li (more hydrotropic than ⁶Li) have more similar geochemical behavior to H, ⁷Li preferentially enter olivine during silicate metasomatism, but remain in the melt during carbonatite metasomatism, resulting in distinct Li isotope behavior during the two different metasomatism. Therefore, the water content of olivine can be used to distinguish metasomatic melts, as well as to explain the δ⁷Li differentiation.

1. Introduction

Metasomatism is an important mechanism that generates mantle heterogeneity and plays a key role in the Earth’s lithological formation and evolution [1]. Silicate melt and carbonatite melt are two major metasomatic agents. Both of them are important carriers of material and energy in the Earth [2]. When these melts have a certain amount of water, their capacity to carry material/energy is much greater due to their lower partial molar density and viscosity, and higher buoyancy than dry melts [3]. Considering the key role of water in the mantle evolution and the genesis of basaltic and carbonatitic magmas, it is important to understand the water geochemical behavior during metasomatism.

As hydrogen is very active, the water content in upper-mantle minerals change readily even by cryptic metasomatism [4]. A previous study found that olivine water content correlates negatively with the olivine forsterite (Fo) content during silicate metasomatism [1]. This indicates that silicate metasomatism would increase the water content in olivine. Another petrological experiment shows that the olivine water content decreases with increasing CO₂ proportion in the (CO₂ + SiO₂) equilibrium melt [5], which means a lower partition coefficient of olivine during carbonatite metasomatism. This process may not increase the olivine water content, or even extract water from nominally anhydrous minerals (NAMs) into the melt. This agrees that water solubility in carbonatite melts is two to three times higher than that in silicate melts [6]. Therefore, water content in mantle minerals can be used as a metasomatism indicator, and also as a discriminator of metasomatic types. However, this is yet to be well investigated due to the lack of water content data from natural carbonatite metasomatized samples.

In this study, the secondary ion mass spectrometry (SIMS) water content analytical method, which has high spatial resolution, simple sample preparation and no need to consider mineral anisotropism, is used to measure the water content of two suites of olivine samples that are metasomatized by silicate or carbonatite agents, respectively. The aim is to reveal the water geochemical behavior between olivine and melt during metasomatism.

2. Samples

Our mantle peridotite xenolith samples were collected from the Hannuoba (North China Craton) and Haoti (western Qinling Orogen) areas. The grain sizes range from 200 to 2000 μm. Detailed geology and geochemical information has been described by Su et al. [7].

The Hannuoba mantle xenoliths is constrained to have a depth of 45–65 km [8] and has undergone varying degrees of silicate metasomatism, as evidenced by: (1) gradual change of mineral contents and reaction texture in composite xenoliths (e.g., peridotite with pyroxenite veins); (2) Abundant orthopyroxenite (Si-rich ultramafic rock) and zoned olivine [9] showing reaction between minerals and Si-rich melt [10]; (3) Metasomatic overprinting suggested by whole-rock and mineral trace elemental and Sr-Nd-Pb-Hf-Os isotope features [11]; (4) Abundant Si-rich melt inclusions in olivine and clinopyroxene [9]; (5) Distinctly positive δ⁷Li values (+3.0 to +41.9‰) in olivine of the Hannuoba peridotites (Figure 1) [7].

P-T estimations revel that the Haoti xenoliths given pressure in the range of 18.5–26 kbar correspond to a depth of 56–78 km [23], and the carbonatite metasomatism in the Haoti peridotite xenoliths is evidenced by: (1) Common occurrence of veins and discrete grains of carbonatites [24]; (2) Clinopyroxene with high CaO and Na₂O contents, and Ti/Eu and LREE/HREE ratios, enriched in Ba, Th, U, Pb and Sr, and negative Ti, Hf and Y anomalies [25]; (3) The Haoti peridotites with LREE enrichments, positive Sr and Ba anomalies, carbonatite-like multi-element patterns, and Sr-Nd-Pb isotope mixing trend between the DM (depleted mantle) and EMII (enriched mantle II) end-members [24]; (4) Distinctly negative δ⁷Li values (−29.1 to + 19.9‰) in olivine of the Haoti peridotites (Figure 1) [7].

3. Analytical Methods

3.1. Electron Microprobe

Sample mineral compositions were determined with a JEOL JXA-8100 electron probe micro-analyzer (EPMA) at the State Key Laboratory of Isotope Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (SKLaBIG-GIGCAS). Analysis conditions include 15 kV accelerating voltage, 20 nA beam current, and 5 μm beam diameter. The ZAF correction procedure was adopted for data reduction.

3.2. Water Content Analysis

Water contents in the olivine were measured with a CAMECA IMS 1280-HR SIMS at the SKLaBIG-GIGCAS, following the method described by Zhang et al. [26]. Before the analysis, all samples were placed on a double-sided adhesive tape and enclosed in a tin-based alloy. The alloy mount was photographed under the microscope (reflected light) and then gold-coated (~30 nm thick). This novel sample preparation method can significantly improve the vacuum condition in the sample chamber and yield a smooth surface, giving lower detection limit and better precision.

When SIMS was maintained at the optimum condition after thorough baking, an automatic liquid nitrogen refilling system was used to maintain ultra-high vacuum in the sample chamber. The sample mount was loaded into the storage chamber overnight and further pumped down (for 1 to 2 h) before the analysis [27]. A Cs⁺ primary beam of 4–5 nA and ϕ~15 μm (10 kV impact energy) was used to sputter secondary ion from the samples. The size of the analytical area was 30 μm × 30 μm (15 μm spot size + 15 μm rastering). The normal-incidence electron gun and the nuclear magnetic resonance (NMR) controller were used to ensure charge compensation throughout the measurement session and to stabilize the magnetic field, respectively. Other instrument parameters include 400 μm contrast aperture, 60 μm entrance slit, 80 μm max area, 173 μm exit slit, and 50 eV energy slit with 5 eV gap. The Faraday Cup detectors, located on L’2 trolley with 10¹⁰ Ω amplifiers, and an electron multiplier of mono-collector were used to measure the ¹⁶O⁻ and ¹⁶O¹H⁻, respectively. For ¹⁶O⁻, 500 mm collector slits were used to yield a ~2500 MRP with sufficiently flat plateau, while a ~173 μm exit slit was used for electron multiplier (yielding ~7000 MRP) to obtain a sufficiently flat plateau and separate isobaric interference of ¹⁷O⁻ on ¹⁶O¹H⁻. Water content was calculated according to the calibration curves established by analyses of the olivine standards of San Carlos, KIB-1, ICH-30, and Mongok [26].

4. Results

4.1. Major Element Compositions

Olivine samples have similar major element contents, characteristic of most peridotites worldwide. Olivine grains from Hannuoba and Haoti have forsterite content of 88.4–90.1 and 90.1–91.6, respectively. The major element compositions are given in

Table 1. Summary of major elements, water content, Li content, and δ⁷Li.

Sample Name	Na2O	NiO	K2O	SiO2	FeO *	CaO	Al2O3	MnO	P2O5	MgO	Cr2O3	TiO2	Total	Fo a	Water Content (ppm)	Std (%)	δ7Li b	Error (1σ) b	Li (ppm) b
HT15	0.02	0.52	0.01	41.49	9.24	0.07	0.02	0.13	0.01	48.57	0.05	0.03	100.12	90.4	13.65	9.53	−6.80	0.40	6.30
HT16	0.03	0.54	0.01	41.46	9.59	0.06	0.02	0.12	0.01	48.35	0.03	0.02	100.21	90.1	7.66	13.18	−15.23	2.38	6.43
HT08-7-2	0.02	0.54	0.01	41.18	9.82	0.01	0.01	0.13	0.01	48.59	0.02	0.04	100.31	89.9	1.78	32.86	−11.30	1.80	13.06
HT08-4-1	0.02	0.52	0.01	41.07	9.10	0.08	0.02	0.13	0.01	48.95	0.04	0.03	99.94	90.6	7.94	21.42	−10.40	2.63	4.14
HT24	0.02	0.54	0.01	41.26	8.54	0.07	0.02	0.13	0.02	49.23	0.05	0.03	99.88	91.2	13.84	23.37	−2.70	2.05	7.33
HT32	0.02	0.37	0.01	41.23	8.67	0.08	0.02	0.11	0.01	49.90	0.04	0.02	100.44	91.2	1.85	9.83	−11.83	2.24	3.05
JSB10-47	0.08	0.13	0.01	55.69	7.09	0.66	4.47	0.18	0.01	31.18	0.34	0.12	99.95	88.8	5.52	17.17	12.98	1.99	2.33
JSB10-16	0.02	0.45	0.01	41.28	10.70	0.06	0.03	0.15	0.03	47.68	0.03	0.02	100.41	89.2	16.65	14.37	21.11	2.84	2.45
JSB10-43	0.01	0.59	0.01	41.26	10.12	0.02	0.01	0.15	0.02	48.20	0.02	0.02	100.38	89.6	3.18	8.62	6.69	1.97	4.95
JSB10-2	0.02	0.48	0.01	41.07	10.74	0.05	0.02	0.14	0.01	47.48	0.03	0.03	100.06	89.0	16.77	22.37	28.31	2.22	3.11
JSB10-41	0.02	0.46	0.01	41.51	9.64	0.06	0.02	0.14	0.02	48.51	0.02	0.03	100.41	90.1	3.26	34.46	22.27	2.18	3.32

Note: ^a Fo = 100 × Mg/(Mg + Fe). FeO *: all Fe as Fe²⁺. ^b Li content and δ7Li are reference from Su et al. [7].

.

4.2. Water Contents

As the ¹⁶O⁻ signal intensity is much higher than that of ¹⁶O¹H⁻, uncertainties in ¹⁶O¹H⁻/¹⁶O⁻ ratio are expected from the counting statistics of ¹⁶O¹H⁻ signal intensity as Poisson error [28]. Most analytical error data are below 1%, while some are much larger, for instance, spot 5 (sample HT16), spot 10 (sample HT08-4-1), and spot 1 (sample HT24). As these analytical results are prone to yield higher ¹⁶O¹H⁻/¹⁶O⁻ ratios, we considered that impurities such as water-rich micro-inclusions/-cracks were encountered during the secondary ion sputtering, which was also reported even for gem-quality samples [29]. These data with high error were discarded from further discussion.

Five samples from Hannuoba were analyzed. Eighteen and ten SIMS measurements were conducted on JSB10-16 and JSB10-43, respectively, yielding confined water content of 13.8–22.3 ppm (mean 16.65 ppm, 1SD: 14.37%) and 2.6–3.6 ppm (mean 3.18 ppm, 1SD: 8.62%), respectively. JSB10-47 yielded 3.3–6.4 ppm water content (mean 5.5 ppm, n = 8, 1SD: 17.17%), whilst JSB10-2 and JSB10-41 yielded 16.8 ± 3.8 and 3.3 ± 1.1 with larger standard deviation of 22.37% (n = 18) and 34.46% (n = 17), respectively.

SIMS measurements conducted on the Haoti samples including HT15 (n = 5), HT16 (n = 7), and HT32 (n = 9), yielded narrow water content range (1SD < 15%) of 12.0–15.5 ppm (HT15), 6.3–9.1 ppm (HT16), and 1.6–2.2 ppm (HT32). Nineteen analyses were conducted on HT24 and sixteen on HT08-4-1, giving measured water content of 8.9–20.4 ppm (mean 13.8 ppm, 1SD: 23.37%, 1 spot rejected) and 5.1–10.9 ppm (mean 7.9 ppm, 1SD: 21.42%, 1 spot rejected), respectively. The results of HT08-7-2 yielded higher standard deviation than other Haoti samples due to its lower water content (1.1–2.6 ppm, mean 1.8 ppm, 1SD: 32.89%). These results are summarized in Table 1.

5. Discussion

5.1. Water Diffusion during the Ascent of Olivine-Bearing Magma

When studying the water systematics of NAMs, a key question is whether the water in the xenolith mineral was degassed or not during/after the xenoliths was transported to the Earth’s surface. Solubility of water in olivine varies with pressure and temperature, and its diffusivity is possible to enable rapid re-equilibration with their host magma, even at the timescale of an individual eruption [30,31]. Thus, we need to determine how much water was diffused in the studied samples. Here, we consider three criteria to determine if the olivine water contents were altered.

Previous studies show that the water content in the core is possibly four to six times or even more than that in the rim, and displays a diffusive profile if the olivine was degassed during ascent [30,31]. In this study, the results along single sample profiles have no such degassing trend (Figure 2). Moreover, most results yield standard deviation <23%, near to the acceptable analytical uncertainty, <19% [32]. This indicates that the olivine is homogenous in terms of water content.

Moreover, heterogeneity in water content can be observed around fractures in individual olivine grains [31]. However, in this study, the water content standard deviation for HT08-4-1 < 21.4% (n = 15) and spot No. 6–9 (located close to fractures; Figure A1) yielded similar results (within error) to those at other spots. The same occurs in HT15 too. Spots No. 4 and 5 (Figure A1) show consistent results with other results distal from fractures. This argues against a degassing event.

We also believe that these small standard deviation and consistent results within a sample cannot be explained by complete degassing. If the olivine samples are severely degassed, which will allow all the samples to have consistent low water content. However, in this study, we found that the water content was different among different olivine sample, either in Hannuoba olivine or Haoti olivine. Hannuoba olivine yields relatively higher water content than Haoti olivine, and the sample (JSB10-6) as high as ~22 ppm. Although the samples of Haoti have a relative low water content (~1.78 ppm in HT08-7; ~1.85 ppm in HT32), the water content of HT24 is as high as ~20ppm. Additionally, the variation of water content is well correlated with Fo (Discuss in the Section 5.3), indicating that the variation cannot be controlled by degassing.

Finally, if the water was modified during ascent, finer-grained olivine grains may have lower water content than coarser ones [31]. In this study, analyses on olivine grains of different sizes from the same sample yielded almost the same water content within uncertainties, as demonstrated by their low standard deviation (8.6%~3.5%) of results from different size of grains. For example, the average water contents for three olivine grains from the xenolith samples JSB10-43 and HT32 are 3.2 and 1.8 ppm, respectively, with standard deviation <10%.

A previous study found that different hydration states of peridotites are distinguished according to the water contents in the Hannuoba orthopyroxene (47–130 ppm) [33]. Other mineralogical phases such as garnet and spinel which are very few in the studied samples and their contribution to bulk water concentration can be ignored. Additionally, we can calculate the water contents in olivine from 5–15 ppm according to the hydrogen partition coefficient during olivine and pyroxene ($D_H^{olivine/orthopyroxene}=~0.11$) [34], consistent (within error) with our results (3.2–16.8 ppm). Accordingly, we considered that the olivine preserved their primary water contents and have no alteration during and after the xenolith ascent.

5.2. Hydrogen Species and Water Content

Point defects are considered to be the main type of hydrogen reservoir in NAMs, and hydrogen always decorates point defects in crystalline structure. Theoretically, H can occupy a metallic site in the octahedron, replacing Si⁴⁺ in the tetrahedron, or have coupled-substitution with defects [35,36]. The incorporated H usually bonds with oxygen to form H₂O or OH⁻ hydroxyl groups. The major elemental substitutions in silicate and carbonatite metasomatism are dominated by Fe–Mg and Ca–Mg exchange, respectively [7]. In this study, results show that the overall low MgO (Figure 3a) and high FeO (Figure 3b) contents in olivine from the Hannuoba xenoliths verify that silicate metasomatism is dominated by Fe-Mg exchange. In contrast, the higher MgO content in the studied Haoti olivine (Figure 3a), and the high CaO content in orthopyroxene [7] suggest weak Fe-Mg exchange in carbonatite metasomatized olivine. In this study, the correlations of MgO versus water content in the Haoti and Hannuoba olivine, and the weak correlation with FeO (Figure 3b) and CaO (Figure 3c) and SiO₂ (Figure 3d) imply that Mg²⁺ exchange may facilitate water incorporation into olivine during metasomatism. Additionally, the water content is related to Fo and water content (Figure 4a) either in Hannuoba or Haoti olivine. From this perspective, the stronger Fe-Mg exchange of silicate metasomatism would lead to more water content change in olivine. The results also demonstrate that the silicate-metasomatized Hannuoba olivine grains have much more varied water contents. Sample JSB10-2 (highest water content: 16.77 ppm) has lower Fo (89.0) than that of sample JSB10-41 (lower water content: 3.26 ppm; Fo = 90.0). The carbonatite-metasomatized Haoti olivine grains have lower and less-varied water contents. Sample HT08-7-2 (lowest water content: 1.78 ppm) has low Fo (90.3), whereas sample HT24 (highest water content: 13.84 ppm) has high Fo (91.2).

5.3. Water Geochemical Behavior in Silicate and Carbonatite Metasomatism

The water storage capacity of olivine will increase when the pressure increase. The increasing hydrogen concentration in olivine with increasing depth has been reported in some area which is likely controlled by the increase in H partitioning into olivine at the expense of orthopyroxene as imposed by a decrease in Al content in opx with depth [35]. In this study, we do not think this is reason for the variation of water content in different samples because the studied samples from one same area, either Hannuoba or Haoti have a similar depth according to previous studied [8,23]. Therefore, the variation of water content from same area cannot be related to sample depth. As the variations are well correlated with the Fo (Figure 4a), the index of intensity of the metasomatism, it is reasonable to relate the variations of water to a different degree of metasomatism and distinct of water geochemical behavior in silicate and carbonatite metasomatism.

Sokol et al. [5] shows that water partitioning between olivine and carbonate/silicate melts has distinct geochemical behavior. The partition coefficient between olivine and melt becomes ~6.5 times lower (0.006 to <0.0009) as the CO₂/(CO₂ + SiO₂) molar ratio in the equilibrium melt increases from 0 to 0.4–0.8. In this study, five Hannuoba peridotite samples show a negative correlation between the Fo and water content. This correlativity indicates that the olivine water content gradually increases as the metasomatism intensifies. A similar phenomenon was also reported by Doucet et al. [1]. Meanwhile, the six Haoti peridotite samples show a positive correlation between the Fo and water content (Figure 4a). Their overall positive trend with Fo values (Figure 4a) is in clearly contrast with that of the silicate-metasomatized Hannuoba samples. We interpret that the gradual water content drops when the metasomatism intensifies, suggesting that water was extracted from olivine into the carbonatite melts during carbonatite-metasomatism. This interpretation also agrees with experimental petrological study findings that carbonatite melts can effectively extract water from the peridotite it intruded [5].

5.4. Relations of Water and Li Elemental Contents and Its Isotope Composition during Mantle Metasomatism

Lithium is a fluid-mobile and moderately-incompatible element [37]. It is thus sensitive to melt-/fluid-rock interactions [36,38,39], and has similar geochemical behavior to H [34]. Moreover, Li content in olivine is normally below 20 ppm (Figure 1) [40], which is of the same magnitude as the water content. Furthermore, as ⁷Li and ⁶Li partition preferentially into the fluid and solid, respectively [41], and previous studies showed that Li isotopes tend to be lighter in carbonatite metasomatism and heavier in silicate metasomatism. The large mass difference between ⁶Li and ⁷Li can lead to strong isotopic fractionation from −52.8 to 41.9‰ (Figure 1) [7,12,16,17]. Thus, the properties of the mineral and metasomatic agent (esp. geochemical behavior of water) would affect the Li isotope fractionation.

The water content and Li content both shows an increasing trend when Fo decreasing in the silicate-metasomatized olivine samples; However, distinct trends (decreasing and increasing) have been displayed for water and the element Li in the carbonatite-metasomatized samples (Figure 4), respectively. This difference indicates that both H and Li have similar geochemical behavior during silicate metasomatism, but Li behaves differently when metasomatized by a different agent. Considering that water is extracted from olivine during carbonatite metasomatism, other monovalent elements are more likely to enter olivine during metasomatism and yield higher Li content than silicate-metasomatized olivine. As the ⁷Li is more hydrotropic than ⁶Li, ⁷Li preferentially enters olivine during silicate metasomatism. As for carbonatite metasomatism, water is extracted from olivine and ⁷Li tends to stay in the carbonatite melts, whilst ⁶Li preferentially enters olivine. This process can be explained by the water geochemical behavior model in Figure 5. The different variation trends between water content and δ⁷Li suggest that the incorporation or extraction of water between olivine and the metasomatic agent would affect the Li isotopes in olivine, which induces a wide δ⁷Li range of −52.8‰ to +41.9‰ for mantle olivines (Figure 1) [39,42].

6. Conclusions

Our results reveal that the silicate-metasomatized olivine samples have higher and wider-range water contents than the carbonatite-metasomatized ones. Olivine water contents increase with silicate metasomatism and decrease with carbonatite metasomatism. Such a distinct process indicates that water likely enters olivine during silicate metasomatism, but is extracted from olivine during carbonatite metasomatism. Moreover, the water geochemical behavior has influenced the Li isotope differentiation during silicate- or carbonatite metasomatism. Our findings indicate that water content can be a robust proxy for mantle metasomatic agents.