The electron microprobe analysis (EMPA) data for olivine, orthopyroxene, clinopyroxene, spinel, and serpentine-group minerals of the sample E22 are listed in Table 1. Olivine (Fo90–91) relicts contain small amounts of Cr (≈0.03 wt.% Cr2O3) and significant Ni contents (≈0.36 wt.% NiO). Orthopyroxene (En90–91) compositions correspond to Al- and Cr-bearing enstatite, with Mg# (=100×Mg2+/(Mg2++Fe2+)) similar to that of olivine. Clinopyroxenes (En50–53, Wo43–45, Fs5–6) are poor in Ti (≈0.10 wt.% TiO2) and rich in Cr (≈1.28 wt.% Cr2O3). Spinels generally have high Mg# (≈68) and low Cr# (=100×Cr3+/(Cr3++ Al3+); ≈30) values. Bastite after orthopyroxene is higher in Al (≈1.59 wt.% Al2O3) and Cr (≈0.92 wt.% Cr2O3) and lower in Ca (≈0.08 wt.% CaO) and Ni (≈0.10 wt.% NiO) than serpentinite after olivine (Table 1). The line-scanning of the EPMA analysis on sample E22 shows that the calcite veins are characterized by the high relative intensity of Ca and are almost free of Mg and Fe (Fig. 3).
Minerals (wt.%) SiO2 TiO2 Al2O3 Cr2O3 MgO CaO MnO FeO NiO Total Mg# Ol (n=25) Average 40.99 0.01 0.02 0.03 49.77 0.06 0.13 8.98 0.36 100.31 90.81 Min 40.85 < l.o.d < l.o.d < l.o.d 48.53 0.03 0.06 8.46 0.28 98.50 91.09 Max 41.81 0.02 0.04 0.06 50.60 0.10 0.18 9.65 0.46 101.98 90.34 Opx (n=7) Average 54.79 0.06 4.05 0.93 32.05 2.23 0.14 5.86 0.09 100.19 90.70 Min 54.54 0.00 3.29 0.81 31.31 1.43 0.12 5.63 0.04 99.72 90.84 Max 55.14 0.09 4.25 1.06 32.83 3.45 0.18 6.27 0.14 100.83 90.32 Cpx (n=6) Average 51.31 0.10 4.79 1.28 17.38 21.62 0.11 3.06 0.06 99.72 91.01 Min 50.42 0.06 4.49 1.18 16.35 18.48 0.05 2.63 < l.o.d 98.35 91.72 Max 52.04 0.13 5.08 1.34 19.38 23.20 0.14 4.04 0.10 100.21 89.53 Spl (n=3) Average 0.04 0.06 41.48 26.36 17.33 0.01 0.20 14.34 0.25 100.09 68.30 Min < l.o.d 0.05 40.88 25.95 16.98 0.00 0.16 14.03 0.15 99.95 68.33 Max 0.07 0.07 41.88 27.03 17.65 0.04 0.25 14.64 0.30 100.23 68.24 Srp (n=2) Average 41.59 0.05 0.69 0.07 37.12 0.15 0.09 4.51 0.29 84.46 94.62 Min 41.48 < l.o.d 0.69 0.06 36.89 0.06 0.08 4.48 0.11 83.91 93.62 Max 41.71 0.05 0.70 0.08 37.35 0.30 0.10 4.55 0.47 85.01 93.60 Bas (n=4) Average 40.43 0.04 1.59 0.92 36.02 0.08 0.15 5.65 0.10 84.74 91.91 Min 39.86 0.02 1.30 < l.o.d 34.87 0.01 0.10 5.19 0.09 83.85 92.29 Max 41.26 0.05 1.79 0.99 37.23 0.13 0.19 6.33 0.15 85.84 91.29 Averages are reported for analyses above the limit of detection (l.o.d); Mg#=100×Mg2+/(Mg2++Fe2+); Ol. olivine; Opx. orthopyroxene; Cpx. clinopyroxene; Spl. spinel; Srp. serpentine; Bas. bastite.
Table 1. Average and ranges of electron microprobe analyses of minerals in harzburgite
Figure 3. EMPA line analyses results (see Fig. 2f). (a)–(b) EMPA line analyses results across the veins in serpentinized harzburgite sample E22. The carbonate veins are distributed among the mesh texture. They are characterized by the high Ca relative intensity and are almost free of Mg and Fe in the middle of the vein (the relevant data are given in supplementary Table B1).
XRD analysis of the whole rock powder of the sample E06 confirmed the presence of secondary minerals of serpentine (lizardite), magnetite, hematite, calcite, bornite, pyrrhotite, and pentlandite. By contrast, XRD analysis of the whole rock powder of the sample E22 confirmed the presence of serpentine (lizardite), magnetite, hematite, pyrrhotite, and galena (Fig. 4).
Figure 4. The XRD profiles of serpentinized harzburgite samples from the Tianxiu hydrothermal field. Lz. lizardite; Gn. galena; Bn. bornite; Pn. pentlandite; Po. pyrrhotite; Hem. hematite; Cc. chalcocite; Cal. calcite (Kretz, 1983).
The δ13CPDB and δ18OPDB signals of the calcite veins Ⅰ are +0.58‰ and +4.46‰, respectively. By contrast, those of the calcite veins Ⅱ are +0.54‰ and -16.67‰. The equation proposed by Coplen et al. (1983) is used to convert δ18OV-SMOW to δ18OPDB (R1). These two calcite samples have similar δ13CPDB signals, but their δ18OPDB signals are very different. Similar isotope signals of calcite veins were also reported by Schroeder et al. (2015) (-4.66‰ to +3.29‰ δ13CPDB; -19.34‰ to +5.45‰ δ18OPDB) during their studies on an oceanic core complex near the 15°20'N fracture zone (ODP Leg 209) along the Mid-Atlantic Ridge (MAR).
3.1. The Geochemical Characteristics of Serpentinized Harzburgite
3.2. Carbon and Oxygen Isotope Chemistry
The geochemical isotope data for calcite samples provide important clues for understanding the origin of precipitating fluid. The δ13C values for calcite of both types (+0.54‰ and +0.58‰) are close to that of seawater (≈1‰; Kump, 1989), which implies that dissolved inorganic carbon in the seawater is the main source of calcite. The δ18OV-SMOW values for carbonate reflect the combined effect of temperature and δ18OV-SMOW of the precipitating fluid (Urey, 1947). If the precipitation temperature of calcite is known, the δ18OV-SMOW of the fluid can be calculated based on the oxygen isotopic fractionation factor between calcite and water, and vice versa. Temperature calculations are performed using the composition of modern Indian Ocean deep seawater (-0.18‰ δ18OV-SMOW; Schmidt et al., 1999) and the equation of Kim and O'Neil (1997) (R2).
The formation temperature of calcite with -16.67‰ and +4.46‰ δ18OPDB is calculated to be 117 and -6 ℃, respectively. The negative temperature is inconsistent with the temperature of the seafloor bottom water, which can be attributed to the fact that its precipitation occurred under higher seawater δ18OV-SMOW rather than modern values (δ18OV-SMOW= -0.18‰), such as during a glacial maximum (Schroeder et al., 2015). The bottom water temperature of the sampling site (water depth: 3 500 m) is 2 ℃, according to the conductivity-temperature- depth investigation during the DY38th cruise (unpublished data). The seasonal variation of deep water temperature is insignificant. Considering that the samples were collected on the seafloor, we assume 2 ℃ as the precipitation temperature of the calcite vein with δ18OPDB of 4.46‰. Utilizing the equation of Kim and O'Neil (1997), the equilibrium δ18OV-SMOW of calcite- precipitating fluid is calculated to be 1.78‰.
The δ18OV-SMOW values of hydrothermal vent fluids range from 0.28‰ to 2.39‰, with an average value of 1.00‰ according to the statistics of more than 150 vent fluid samples (supplementary Table B2; James et al., 2014; Schmidt et al., 2011; Gamo et al., 2001; Shanks, 2001; Jean-Baptiste et al., 1997). By contrast, vent fluids from the high-temperature and ultramafic- hosted hydrothermal systems, such as the Kairei, Logatchev, and Nibelungen sites along the Central Indian Ridge and the MAR normally have relatively higher δ18OV-SMOW values, ranging from 1.23‰ to 2.39‰ (Schmidt et al., 2011; Gamo et al., 2001), with an average value of 1.74‰ (Table 2). This value is close to the calculated δ18OV-SMOW value (1.78‰) of calcite-precipitating fluid in this study, which indicates that hydrothermal fluid exerted an influence. Hence, the formation temperature of calcite with -16.67‰ δ18OPDB is inferred to be approximately 134 ℃. Moreover, calcite veins in the serpentinized harzburgite samples may also precipitate from the fluid with higher δ18OV-SMOW values. The precipitation temperature of calcite veins would be higher by 6 ℃ if a fluid δ18OV-SMOW of +2.39‰ was adopted, which corresponds to the most 18O-enriched fluids reported thus far (Schmidt et al., 2011; Table 2).
Vent field Temperature (℃) Sample ID δ18O Endmember (‰) Logatchev Ia 350b 253ROV-10 1.42 255ROV-3 1.23 255ROV-17 1.31 259ROV-25 1.45 271ROV-11 1.32 275ROV-5 1.36 275ROV-7 1.42 Nibelungena 372 314ROV-2 1.42 314ROV-3 2.39 314ROV-4 2.24 314ROV-5 2.2 314ROV-6 2.16 Kaireic 360 ROV-Kaiko 1.90 Average 1.74 aSchmidt et al. (2011); bSchmidt et al. (2007); cGamo et al. (2001).
Table 2. Endmember fluid oxygen isotopic composition from the Logatchev I, Nibelungen, and Kairei vent fields.
Significant differences in carbonate precipitation temperatures are found in various ultramafic-hosted hydrothermal systems (Schwarzenbach et al., 2013). Calcite veins in serpentinite from the Mid-Atlantic Ridge Kane (MARK) Fracture Zone area have formation temperatures between 1 and 235 ℃ (Alt and Shanks, 2003), while calcite veins near the 15°20'N Fracture Zone show temperatures from near ambient water to 175 ℃ in detachment fault rocks (Schroeder et al., 2015; Bach et al., 2011). Both hydrothermal systems are located along the MAR, where the young mantle has been exposed by detachment fault (Schroeder et al., 2015, 2007; Bach et al., 2011; Alt and Shanks, 2003; Shanks, 2001; Karson and Lawrence, 1997).
In addition to calcite veins, minerals in harzburgite could also give insights into the possible formation conditions. Marques et al. (2007) suggested that spinels with Mg/Fe > 1 and Al/Cr > 1 indicated mineralized conditions through the studies on spinels from early and evolved stockwork, steatite, and semi-massive sulfide samples. The ratios of Mg/Fe and Al/Cr in spinels of the sample E22 are higher than 1, which indicates that the harzburgite locates in the mineralized zone. Hence, the calcite veins and spinel in the harzburgite samples may indicate the possible influence of hydrothermal fluids from the discharge zone below the Tianxiu hydrothermal field, although the samples are approximately 400 m away from the vent site.
Considering that the sample E06 was highly altered, the primary mineral components of the sample E22 were chosen as the reactant. According to the petrology and geochemical analyses, the reactant contains 80 vol.% olivine (Fo90), 15 vol.% orthopyroxene (Mg#≈90), and 5 vol.% clinopyroxene (Mg#≈90). Hence, the geochemical models were conducted under 35 MPa and 135 ℃, which correspond to the sampling depth and the formation temperature of calcite in E22, respectively. The predicted mineral assemblages and fluid compositions are shown in Fig. 5. The mineral assemblage of reaction path models is stable and almost consistent with the petrological observation. Serpentine, magnetite, and olivine appear under various w/r ratios, while calcite only occurs at w/r ratios ≤3 (Fig. 5a). It is suggested that the instability of calcite under higher w/r ratios is caused by the low Ca2+ and HCO3- concentrations in the fluid (Fig. 5b). Actually, the concentrations of Ca2+, Fe2+, SiO2(aq) all decrease toward higher w/r ratios. This is consistent with the decreasing amount of rocks in the reaction. The increase of Mg2+ toward higher w/r ratios is attributed to the instability of olivine at higher w/r ratios. Alteration of harzburgite drives down the fluids' oxygen fugacity and increases pH, thereby decreasing the carbonate solubility toward lower w/r ratios. The mineral assemblage of magnetite, galena, bornite, pyrrhotite, and pentlandite in both samples indicates a low oxygen fugacity and reduced condition (Fouquet et al., 2010). This result suggests that the variations in mineral assemblage may be due to local variations in the degree of fluid fluxing and rock interaction (Schroeder et al., 2015). Our model supports the idea that lower w/r ratios, alkalinity and reduced conditions promote the precipitation of calcite. This idea also supplements the conclusion given by Schroeder et al. (2015), who adopted troctolite and CO2-enriched fluid as the reactant and noted that the compositional effects may influence the mineral assemblage of the reaction. Additionally, although calcite precipitation temperatures of the MARK area and the 15°20'N Fracture Zone reach up to 235 and 175 ℃ (Schroeder et al., 2015; Bach et al., 2011; Alt and Shanks, 2003) respectively, their median value is only 59 and 3 ℃, with most of the temperatures below 100 ℃. Studies conducted by Allen and Seyfried (2004), Proskurowski et al. (2006) and Klein et al. (2009) have shown that pH values for fluid from serpentinization reactions range from alkaline under moderate to low temperatures to acid under high temperatures. This provides a possible explanation for why high- temperature carbonate minerals in serpentinized ultramafic rocks are seldom discovered.
Figure 5. Reaction path model simulating the serpentinization of harzburgite at 135 ℃ and 35 MPa with w/r ratios ranging from 0.2 to 10 in a titration system. (a) predicted equilibrium mineral assemblages and pH during serpentinization under different w/r ratios; (b) predicted fluid composition and logfO2 during serpentinization under different w/r ratios (see the abbreviations above).
Along the mid-ocean ridge, the formation of secondary minerals in ultramafic rocks is partly explained by the creation of fluid pathways through tectonic processes. The fractures caused by changes in the stress regime, faults during extensional unroofing, and mechanical weathering all provide pathways for fluid (Liu et al., 2018; Boschi et al., 2009; Andreani et al., 2007; Kelley et al., 2001). In the absence of external forces, two mechanisms have been proposed to explain the microfractures: anisotropic thermal contraction and reaction-driven cracking (Rouméjon and Cannat, 2014). The former is capable of producing microcracks in peridotite, particularly in its primary mineral constituent olivine at the slow-spreading ridge (Demartin et al., 2004). The latter produces intense self- propagating fractures (Kelemen and Hirth, 2012; Plümper et al., 2012; Jamtveit et al., 2009). During serpentinization, decomposition of olivine and pyroxene to serpentine is accompanied by a volume increase of up to 40% (Coleman, 1997), which creates microfractures and propagates cracks (Andreani et al., 2007; Macdonald and Fyfe, 1985; Martin and Fyfe, 1970). Thus, the ultramafic rocks are characterized by complex permeability structures and fluid pathways.
Based on the petrological and geochemical considerations above, we propose a schematic model to explain how the alteration processes occurred, and particularly addresses how low- and high-temperature calcite might form during the alteration (Fig. 6). It is assumed that harzburgite consists of olivine, orthopyroxene, and clinopyroxene (Fig. 6a). When the hydrothermal fluid is discharged toward the seafloor surface along the fractured-porous ultramafic rocks, it may mix with the infiltrated seawater, resulting in the alteration of orthopyroxene and olivine. Allen and Seyfried (2004) reported that the rate of serpentinization of olivine is lower than that of orthopyroxene when the temperature is greater than 300 ℃. Hence, it is inferred that the serpentinization of samples with relic olivine occurs at temperatures > 300 ℃. When hydrothermal fluid mixes with infiltrated cold seawater, the mixing fluid drives the alteration of olivine and promotes the precipitation of calcite (Kelley et al., 2001) at 134–140 ℃ along microfractures developed in olivine (Figs. 6b, 6c). This is also consistent with the studies that showed that the majority of carbonates in oceanic peridotite formed during the mixing of seawater and hydrothermal fluid (Klein et al., 2015; Bach et al., 2011; Eickmann et al., 2009). Moreover, the altered sample with high- temperature calcite veins might also indicate the role of tectonic denudation. The low-temperature calcite veins (≈2 ℃) that filled the fractures in the almost completely altered sample (Figs. 6d, 6e) indicate conducive cooling processes that hydrothermal fluid undergoes during the slow ascent from deeper sources along fractures or faults in the mid-ocean ridge. This model provides an insight into how the alteration processes occur and how the alteration minerals form under the fluid interaction with the samples from the Tianxiu hydrothermal field.
Figure 6. Schematic representations of the alteration process of serpentinized harzburgite samples. (a) unaltered harzburgite with olivine, orthopyroxene, and clinopyroxene as the major constituents of two harzburgite types; (b) serpentinized harzburgite with magnetite in the middle of serpentine veins; (c) calcite veins appearing among mesh texture and along the margin of olivine, exhibiting the alteration of the sample E22; (d) totally serpentinized harzburgite with magnetite in the middle of serpentine veins of the sample E06; (e) serpentine and calcite veins filling the fractures in the sample (see the abbreviations above).
Recently, there has been increasing interest in serpentinization due to its potential for sequestering CO2 through the formation of carbonate veins and replacive carbonate (Grozeva et al., 2017; Schroeder et al., 2015; Klein and McCollom, 2013; Kelemen and Matter, 2008). The developed microfractures could act as fluid pathways for the precipitation of secondary minerals or veins. However, the formation of serpentine and/or carbonate can be self-limiting. They potentially result in the closure of fluid pathways and reduce fluid migration, preventing further precipitation of secondary minerals (Bankole et al., 2019; Ma et al., 2018; Schwarzenbach, 2016; Plümper et al., 2012; Macdonald and Fyfe, 1985). This might partly explain the scarcity of calcite veins in our sample. The limited contribution of carbonate veins for carbon sequestration was reported by Bach et al. (2011), who suggested that the general scarcity of low-temperature aragonite veins in the serpentinized mantle indicates their minor role as a sink for CO2. Similarly, the calcite veins in our samples probably do not indicate a globally important sink for CO2 in the Tianxiu hydrothermal field, due to their scarcity in the samples. However, the estimates of CO2 uptake may increase if fracture and carbonate veins are more developed in rocks beneath the area. This possibility can be verified by drilling holes.