2.1. Geological setting and lithostratigraphy of Hole 801C

The ODP Site 801C is located in the ∼170 Ma-old oceanic crust of the western Pacific Ocean, at a water depth of 5674 m (Figure 1). The crust at this site formed at a fast-spreading ridge having an estimated full spreading rate of 160 km/Ma, and the lithostratigraphy has been described in detail elsewhere [Lancelot et al. 1990; Pringle 1992; Plank et al. 2000; Alt and Teagle 2003]. Hole 801C is penetrated through 461.6 m of sediments and 474 m of igneous basements. The basement section has been divided into eight major units: Unit I consists of 60.2 m of alkali basalt sills that intruded sediments at 157 Ma; Units II and V are 10–20 mm and 1-mm-thick hydrothermal silica–iron deposits at the top of the tholeiite section and at 150 msb (meters sub-basement), respectively; the remaining Units III, IV, VI, and VIII comprise of 400 m tholeiitic basalts including massive, sheet, and pillow flows, with a 10 m thick breccia Unit VII at ∼360 msb. The volcanic basement is similar to other oceanic sites, with the exception of the off-axis alkalic sills and the two hydrothermal silica–iron deposits and associated alteration that formed from low-temperature hydrothermal fluids at the spreading axis [Alt and Teagle 2003; Alt 2003]. Basement porosity averages 11 ± 4% from 0 to 300 msb but is lower (7 ± 1%) below 300 msb [Jarrard et al. 2003]. We refer to these zones as the upper volcanic section (UVS) and lower volcanic section (LVS).

2.2. Petrography of the samples

We studied 11 samples of tholeiitic basalt from 143 msb down to the bottom of the Hole at 467 msb (605–928 m below seafloor). The mineralogy and chemistry of basement alteration in Hole 801C has been documented by Alt and Teagle [2003], and descriptions of samples analyzed in this study are given in Table 1. In general, the basalt samples are slightly recrystallized (5–15%), with saponite ± carbonate ± minor pyrite replacing olivine and interstitial material and filling vesicles. Five samples also exhibit 2–7 mm wide alteration halos that are 5–10% altered into celadonite + iron oxyhydroxide ± saponite, and one sample (43R2 121–124) consists entirely of such alteration halos. All samples contain veins, which range from 40 μm to 2 cm wide, and can be divided into two types: (1) carbonate ± saponite ± chalcedony ± minor pyrite, and (2) celadonite + iron oxyhydroxide ± saponite.

Using the same approach as our previous work on Hole 1256D, we selected zones away from macroscopic veins and associated visible alteration halos in order to characterize alteration of the host basalt “background” alteration. We first removed visible secondary veins by sawing. Thus, we have two sample types: the background basalt samples and a subset of veins and associated alteration halos separated from the basalts. Samples were then crushed and sieved to a grain size fraction smaller than 150 μm in an agate mortar.

3.1. Step-heating extraction of total C and H

Step-heating experiments were used for the extraction of total C and H, as in previous work on Hole 1256D [Shilobreeva et al. 2011]. Approximately 100 mg of sample powder were loaded in a low C and H, blank fused quartz tube, which was connected to a vacuum line. After one-night degassing at room temperature, the samples were then heated at 100 °C under vacuum for 1 h to remove surface contaminants. Volatiles were then extracted during a one-step combustion by heating the samples at 1100 °C for 1 h, with an oxygen gas pressure of 4 mbar (set with hot CuO) because we have shown that most of the C release occurs during this step [Shilobreeva et al. 2011]. During this combustion, all C species were transformed into CO₂ that was quantified using a Toepler pump, with a precision better than 5% using a procedure described by Javoy and Pineau [1991]. Carbon concentrations are thus given in wt% of CO₂. Hydrogen, occurring as OH or H₂O bound in rock minerals, was extracted as H₂O and converted into H₂ over hot uranium for subsequent quantification using a Toepler pump with a calibrated volume.

3.2. Step-heating extraction of reduced C

In order to determine the reduced C content in each sample, about 1 g of powder was loaded in a Falcon 50 mL tube for reaction with ultrapure 6N HCl acid for 12 h at room temperature [Thomazo et al. 2009]. This step removes carbonates from samples that contribute to the production of CO₂ during the online combustion. After the HCl reaction, samples were rinsed several times (n ∼ 5 or 6) with ultrapure deionized CO₂-free water till the solution reached a pH value close to 6. After removing the water, the solid residue was stored in a dry oven at 50 °C for 1 week. Finally, online combustion of approximately 100 mg of powder was performed as described in Section 3.1. Analytical blank, determined for this protocol, is 1.7 ± 0.2 μmol (i.e. 0.07 ± 0.01 wt% CO₂) with δ¹³C ≈−26.0 ± 5.3‰. C contents and isotope compositions were corrected in order to remove this blank contribution. The corrections are smaller than 30% for C content and uncertainties are better than ±2‰ for δ¹³C value.

3.3. Extraction of the carbonate (inorganic) fraction

The concentration and isotope composition of carbonates (CaCO₃ and (Ca,Mg,Fe)CO₃) were measured using gas chromatography coupled to a He continuous flow isotope ratio mass spectrometer (GC-IRMS, AP2003). The GC column (stainless column 60_1/800_2 mm, packed with HayeSep Q 60/80 mesh, Chrompack) separates the CO₂ gas from other residual gas components (e.g., N₂, O₂, NO₂, Ar) [Assayag et al. 2006]. Depending on the sample, between 5 and 500 mg of powder were loaded into exetainer^® vials. After flushing with ultrapure He gas to eliminate atmospheric gases, orthophosphoric acid (H₃PO₄) was introduced in the vial to transform carbonates into gaseous CO₂. After 4 h of reaction at ambient temperature (≈22 ± 3 °C), only CaCO₃ is expected to be transformed into CO₂ [McCrea 1950]. The He–CO₂ gas mixture was thus transferred into a mass spectrometer for analysis of the CaCO₃ component (CO₂ amount and δ¹³C). The vial was flushed again with pure helium gas and heated at 80 °C for 2 h to release CO₂ from other carbonate species ((Ca,Mg,Fe)CO₃) for analysis of C and O isotope composition [Lebeau et al. 2014]. Internal standards of pure calcite and dolomite were run with the samples for calibration of both concentration and isotopic composition. The isotopic ratio ¹³C/¹²C is reported using the conventional δ-notation versus PDB international standard, whereas the SMOW international standard is used for δ¹⁸O. The external reproducibility was ±0.1‰ for δ¹³C, ±0.5‰ for δ¹⁸O and ±10% (relative) for the carbonate fraction content.

3.4. C and H isotope analyses

The isotope compositions of CO₂ and H₂ collected from step-heating experiments were measured on a Thermofisher Delta +XP dual inlet mass spectrometer. For H₂, the H3+ interference factor was measured before any hydrogen isotope measurement. The isotope compositions are reported versus PDB for δ¹³C and SMOW for δD. The external repeatability of H isotopic measurement is about ±2‰ (2σ). Repeated measurements of the international graphite standard USGS24 gave δ¹³C values of −15.96 ± 0.06‰ (2σ; n = 10). This value is indistinguishable from the recommended value [−15.99 ± 0.11‰ 2σ; Stichler 1995]. Thus, the accuracy of the present procedure on δ¹³C measurements is better than ±0.20‰ (2σ).

4.1. C and water content

Total, reduced, inorganic carbon (all expressed as CO₂ content) and water contents in background altered samples are reported in Tables 2, 3, and 5. Total C in the UVS decreases with depth, whereas no trends are observed with depth in the LVS (Table 2 and Figure 2). Although small variations exist, the reduced C content ranging from 0.12 to 0.29 wt% CO₂ is roughly constant throughout the basement. In contrast, if we except 2 samples located around 300 msb (31R1 and 35R1a), inorganic C content exhibits a drastic decrease close to the UVS/LVS boundary, from 0.66 down to ∼0.08 wt% CO₂ (Tables 3, 5).

The relative proportions of inorganic and reduced C thus change at the UVS/LVS boundary, from inorganic/reduced C >1 in the UVS to <1 in the LVS. Samples 31R1 in the UVS and 35R1a in the LVS are not following this general trend probably because the transition is not so sharp, and some variation may occur around that depth. These two samples will thus not be considered for average estimations of the UVS/LVS in Tables 2, 3, and 5 since they are not representative of the sections as we have defined them.

The total water contents vary from 0.42 to 1.15 wt% H₂O (Table 2) but do not show any significant correlation with depth nor difference between the UVS and LVS. The average water content calculated for this portion of the crust is 0.83 ± 0.24 wt% H₂O.

4.2. C, O, and H isotope compositions

Depth profiles of δ¹³C values for the total, reduced and inorganic C in background altered basalts are shown in Figure 3, along with δ¹³C data for carbonate C in the veins. The altered rock depth profiles for δ¹³C values are generally similar to those for C content (Figure 2). In the UVS, δ¹³C values for total carbon in bulk basalt become more negative with depth (from −1.6 at the top to −3.6‰ at the UVS/LVS boundary) whereas those of inorganic C do not (Figure 3). Except for samples 31R1 in the UVS and 35R1a in the LVS, a strong and sharp negative shift in δ¹³C values is observed for total and inorganic C at the UVS/LVS boundary. In contrast, δ¹³C values of reduced C remain at about −26‰ throughout the UVS and LVS.

δ¹³C values of the carbonate fraction in veins vary from −10 to + 2‰ (Table 4), which is in the range for carbonates precipitated from seawater in the altered basaltic crust with the incorporation of oxidized carbon from intercalated sediments for the more negative values [e.g., −11.2‰ to 2.9‰ reported in carbonate veins from Hole 801C by Alt and Teagle 2003; Alt et al. 1996]. However, the very negative δ¹³C values of −9.8 and −10‰ are two silicate veins (801C-43R2, 801C-45R2) that have very low C concentrations, similar to inorganic C in the basalt samples from the same locations in the LVS (cf. Tables 2 and 3). A third silicate + carbonate vein (801C-31R1a) has an inorganic C content of 0.4 wt% CO₂ and δ¹³C = −4.90‰, which are intermediate between those of LVS rock samples and vein carbonates.

Oxygen isotope compositions of carbonate fractions of the altered rock and veins are given in Tables 3 and 4. The transition seen in δ¹³C values at ∼300 msb is also visible in δ¹⁸O values measured in the altered rocks (Table 3; Figure 4). The weighted average δ¹⁸O value of the carbonate fraction in the UVS rocks (27.2‰, n = 5) is clearly higher than that for the LVS (20.5‰, n = 5). These values are within the range determined by Alt and Teagle [2003] for vein carbonates in the UVS, whereas those in the LVS are at the low end of this range (Figure 4). δ¹⁸O value of the vein carbonates, however, show similar variations than those reported by Alt and Teagle [2003], suggesting alteration temperatures increasing downward and always below 100 °C.

δD values in the UVS (mean = −39.7 ± 14‰,) are higher than those in the LVS (−69.3 ± 13.7‰; Table 2), but strong variability appears in both sections.

4.3. C in the oceanic crust: inorganic and reduced C

Inorganic and reduced C are present throughout Site 801C basement. Reduced C contents and isotope compositions show little variations throughout the basement section (∼0.2 wt% CO₂ and −22.6 to −27.8‰, Figures 2 and 3), but observed variations in inorganic C content (Figure 2) lead to differences in total C abundance and isotope composition. High concentrations of inorganic C in the UVS (up to 1.76 wt% CO₂ at the top of the section) dominate total C there (1.92 wt% CO₂), leading to relatively high δ¹³C_total C values of −1.6 to −5.4‰. In contrast, concentrations of inorganic C in the LVS are much lower (<0.1 wt% CO₂), so total C in the LVS is dominated by reduced C, with δ¹³C_total C values of −18.7 to −23.5‰. It is interesting to note that the isotope composition of inorganic C also varies from the UVS to the LVS. The weighted average δ¹³C value of inorganic C in the UVS is + 0.5‰ and shifts to more negative values (−9.8‰) in the LVS. The oxygen isotope composition of inorganic C likewise changes from the UVS to the LVS: from a mean δ¹⁸O value of 27.2‰ to 20.5‰, respectively.

5.1. Possible origins for reduced and inorganic carbon

5.2. Carbon incorporation during aging of the crust

We thus propose a sequence of processes that affect the contents and isotope compositions of carbon in oceanic basalts. Abiotic (FTT and BDR) processes can leave their imprint on reduced and inorganic C throughout the rocks of the volcanic section at the ridge crest, resulting in the presence of reduced C having negative δ¹³C values (−22 to −28‰), and of small amounts (0.07 ±0.02 wt% CO₂, Table 5) of inorganic C having δ¹³C values of −9 to −11‰. This could happen during cooling and crystallization of the basalts and/or through the passage of cooled hydrothermal fluids “leaking” upward from the underlying sheeted dikes at the spreading axis, such as those that produce the cool diffuse hydrothermal venting at the seafloor along ridge crests [e.g., Alt 1995]. Later, as cool seawater solutions percolate through the porous and permeable volcanic section and alter the basement at low temperatures, microbial activity results in the addition of microbial sulfide to the rocks, which can also contribute to organic C having δ¹³C values of −22 to −34‰ to the basalts throughout the lavas. Relatively late seawater solutions also deposit seawater DIC in carbonate veins in the basalt basement and within altered basalts, mainly in the uppermost portion of the volcanic section where high porosity and permeability allow greater fluid circulation [Alt and Teagle 1999, 2003; Alt 1995; Jarrard et al. 2003; Coogan and Gillis 2018].

As the accumulated effect of these processes, reduced C contents and isotope compositions show little variation throughout the Hole 801C basement section ( ∼0.2 wt% CO₂ and −22 to −28‰, respectively), but variations in inorganic C contents lead to differences in total C abundance and isotope composition. High concentrations of inorganic C in the bulk rocks of the UVS (up to 1.76 wt% CO₂ at the top of the section) result from greater flux of seawater through this portion of the crust. This leads to inorganic C dominating total C in the UVS rocks and to relatively high δ¹³C_total C values of −1 to −5‰. In contrast, concentrations of inorganic C in the LVS basalts are much lower ( <0.1 wt% CO₂), so total C in the LVS is dominated by reduced C with δ¹³C_total C values of −19 to −23‰. The contents and isotope compositions of C in the Hole 801C basement thus reflect a relatively uniform component of reduced C, derived from abiotic and biotic reactions, with variable addition of seawater DIC to the rocks, with seawater derived DIC also deposited in carbonate veins (Figure 6).

5.3. C budget at mid-ocean ridges

Except for the upper part of the basaltic basement, where both seawater DIC and biologically derived C are abundant, magmatic CO₂ is a source for C in the background altered basaltic crust at mid-ocean ridges (MORs). This C is stored in the altered basaltic ocean crust as (i) inorganic C retained in the CO₂-outgassed basaltic melt and as (ii) reduced C formed during abiotic reduction reactions or through microbial activity. Integrating the present results with data from Site 1256D obtained in a previous study [Shilobreeva et al. 2011], we can estimate the magmatic C budget at MOR. At Site 801, the reduced C concentration in the volcanic section is 0.19 wt% CO₂ (average value between the UVS and LVS). We should add to this value the inorganic fraction of magmatic C (0.07 wt% CO₂ in the LVS; Table 3), giving a total concentration of 0.26 wt% CO₂. As we discussed above, some fraction of the reduced C is derived from biological activity. To get a rough estimate of each fraction (biological versus abiotic), we can use the C concentrations measured in Hole 1256D. Hole 1256D passed through all lithological units of the basaltic ocean crust, except gabbro, where it penetrated only the uppermost tens of meters of the gabbro section. The estimated temperatures in the crust today reach more than 120 °C at a depth greater than 1000 mbsf (meters below seafloor), indicating that microbial activity is not expected to be sustained at deeper levels [>600 msb: i.e., the transition zone, sheeted dikes, and gabbros, Shilobreeva et al. 2011]. Even if thermogenic carbon from the shallower part can be delivered downward via hydrothermal circulation, because of low permeability, we propose that most of the C measured at depths greater than 600 msb is magmatic carbon. The average value calculated for depths between 600 msb down to 1200 msb [Table 3 in Shilobreeva et al. 2011] is 0.126 wt% CO₂, which is half of the value calculated at Site 801. We can compare this value to the depth-weighted average CO₂ content estimated by Alt and Teagle [1999] from a compilation of different sites. From Table 1, a depth-weighted average CO₂ content of 0.05 wt% CO₂ is calculated for the deeper levels ( >600 msb) of the crust, which again can be assumed to be magmatic carbon at these depths. These two values probably represent the upper (0.126 wt% CO₂) and lower (0.05 wt% CO₂) limit for magmatic carbon stored in the altered crust. A way to explain this difference is that Site 801 was formed in the Jurassic age when atmosphere and ocean were enriched in CO₂ and when CO₂ degassing in association with magmatism at MOR was more active than the present [Gillis and Coogan 2011]. These values are lower than the mean total C content in the altered basaltic crust [cf. Alt and Teagle 1999; Shilobreeva et al. 2011]. The difference between the total and magmatic C contents results from C precipitated seawater DIC and DOC that is stored not only in the background basaltic rock but also in veins and alteration halos along veins. For this reason, C in the altered oceanic crust is dominated by magmatic C in the young basaltic crust where it may exceed 80% of the total carbon (e.g., Holes 504B, 1256D), whereas for an old basaltic ocean crust containing greater amounts of carbonate veins (e.g., Site 801) the magmatic C fraction may be less than 40%.

Considering a crustal production rate of 6.0 ± 0.8 × 10¹⁶ g⋅y⁻¹ [Mottl 2003], the total flux of magmatic C stored in altered basaltic ocean crust ranges between 1.5 × 10¹²–2.0 × 10¹² molC⋅y⁻¹ (using C_mag = 0.126 wt% CO₂). This magmatic flux stored in the crust can be compared with the flux of C degassed from the mantle during the formation of oceanic crust at mid-ocean ridges. The latter ranges from ∼1.0 × 10¹² to 2 × 10¹³ molC⋅y⁻¹ according to various estimations [e.g., Javoy et al. 1982, 1983; Des Marais 1985; Javoy and Pineau 1991; Marty and Tolstikhin 1998; Cartigny et al. 2001, 2008; Chavrit et al. 2014; Le Voyer et al. 2019]. Using the reevaluated value of Cartigny et al. [2008] or the most recent value by Le Voyer et al. [2019], 2.0 × 10¹² molC⋅y⁻¹ or 1.32 ± 0.8 × 10¹² molC⋅y⁻¹ respectively, the flux of magmatic C stored in the altered oceanic crust is similar to the C flux degassed from the mantle. Moreover, and except for the UVS, our data suggest that the dominant form of magmatic carbon that is retained in the crust is reduced C (70%, Table 5). This observation raises the question of its behavior during subduction and recycling in the mantle. The reactivity of carbonates during subduction has been intensively studied, yielding either to a large recycling into the deep mantle, or more recently, to a large transfer back to the surface via mantle wedge [Kelemen and Manning 2015]. Understanding the reactions of reduced forms of C at HP–HT in subduction zones conditions will be necessary in order to refine future global C cycle models.