Rock Magnetic Signatures of Hydrothermal Mineralization in the Trans‐Atlantic Geotraverse (TAG) Hydrothermal Field

The Ocean Drilling Program Leg 158 drill holes from the Trans‐Atlantic Geotraverse hydrothermal field are investigated to understand the rock magnetic signatures of hydrothermal mineralization. A composite columnar section has been constructed through hole correlation to understand the stratigraphic variation of magnetomineralogy within the stockwork. Isothermal remanent magnetization components unmixing, first‐order reversal curve diagrams, low‐temperature magnetic signatures, and electron microscopic analyses disclose magnetic minerals of disparate occurrences related to predominating hydrothermal mineralization reactions in three broad zones: For basaltic basements, serpentinization of olivine phenocrysts during preliminary hydrothermal alteration produces magnetite, in addition to primary titanomagnetite; Chloritized and silicified zone samples contain relict titanomagnetite and exsolved magnetite that survived hydrothermal dissolution; Anhydrite and sulfide zone samples are dominated by magnetite and hematite, likely from oxidation of polymetallic sulfides due to exposure in oxidative seafloor environments during drilling. Our findings suggest that seafloor oxidation potentially modifies the magnetic properties of polymetallic sulfides in hydrothermal deposits, which applies to magnetic tomography of sophisticated subseafloor vent structures and prospecting seafloor massive sulfides (SMS) deposits therein. Meanwhile, we alert future deep‐sea mining that drilling may promote physicochemical alteration of SMS deposits, causing environmental risks. The established magnetic signatures ultimately contribute to understanding the in situ geological preservation of SMS deposits and optimizing exploitation procedures in the future.


10.1029/2022GC010368 2 of 16
The Trans-Atlantic Geotraverse (TAG) hydrothermal field (Figure 1a; Rona et al., 1986), located in the rift valley of the Mid-Atlantic Ridge near latitude 26°08'N, is one of the best-studied hydrothermal fields to date (Humphris et al., 2015). Ocean Drilling Program (ODP) Leg 158 delineated that fluid-rock reactions beneath the TAG active mound include coupled MORB hydrothermal alteration (i.e., chloritization and silicification) and repeated precipitation of anhydrite and polymetallic sulfides as hydrothermal deposits (Humphris et al., , 2015. Post-cruise rock magnetic analyses identified minor magnetite and pyrrhotite in the TAG hydrothermal deposits (Zhao et al., 1998), preserving weak but complex remanent magnetization of chemical origins. However, detailed magnetomineralogy and magnetic property variations among different hydrothermal products in TAG were not established by Zhao et al. (1998). Consequently, although the low concentration of magnetic minerals within the stockwork explains the reduced crustal magnetization in the TAG active mound (e.g., Gehrmann et al., 2019;Tivey et al., 1993), utilizing magnetic surveying data for sophisticated subsurface stockwork geometry inversion and tonnage estimation of SMS deposits remain challenging (e.g., Galley et al., 2021;Szitkar et al., 2021).
In this study, based on the findings of Zhao et al. (1998), we revisit the rock magnetic properties of samples recovered during ODP Leg 158. A composite columnar section from multiple holes is constructed to demonstrate the stratigraphic variations of magnetic parameters with different lithology zonations. Through comprehensive rock magnetic and electron microscopic analyses, we show that magnetic minerals in SMS deposits and fully-altered MORBs can have different occurrences, despite their similar bulk magnetic parameters. We also provide evidence that the magnetic minerals in SMS deposits may form through the oxidation of polymetallic sulfides in suboceanic conditions.

Samples
In September-November 1994, ODP Leg 158 drilled 17 holes into the TAG active mound within a radius of 200 m (Site 957) to understand its internal structures (Humphris et al., 1996). Drilling recovered materials down to 125.7 mbsf (Hole 957E), producing a composite transection of the TAG mound, which from bottom to top can be divided into basalt zone, chloritized zone, silicified zone, anhydrite zone, and sulfide zone (Humphris et al., , 1996. The chloritized, silicified, anhydrite, and sulfide zones define the stockwork. Detailed lithological descriptions can be found in Humphris et al. (1996) and are briefly summarized here. Hydrothermal alteration dominates the chloritized and silicified zone, where basalt, chloritized basaltic breccia, silicified wallrock breccia, and pyrite-silica breccia represent a progressive alteration pathway Figure S1 and Text S1 in Supporting Information S1). The presence of anhydrite and sulfide marks hydrothermal mineralization, which prevails as veins and breccias in anhydrite (pyrite-silica-anhydrite breccias and pyrite-anhydrite breccias) and sulfide zones (massive pyrite or pyrite breccias), though some sulfide and anhydrite veins are also identified in chloritized and silicified zone samples (Knott et al., 1998; Figure S1 and Text S1 in Supporting Information S1).
Representative lithologies of different zonations were sampled (∼10 cm 3 unoriented chips) from 10 holes ( Figure 1a) to avoid biases from spatial inhomogeneity. A total of 55 samples were collected. Sample density was measured using a solid densitometer. As Holes 957E-G have the deepest penetration and most complete zonation (Humphris et al., 1996), a composite columnar section is constructed through lithostratigraphic correlations with Holes 957E-G ( Figure S2 in Supporting Information S1). Five basalt samples from Hole 957M are exceptions because they represent underlying basement, and thus they are extended below Hole 957E-G (Figure 1b). It is worth noting that there are some uncertainties in the exact zone boundaries due to the overall low recovery (13.3% on average) of the drill holes (Humphris et al., 1996). Consequently, we only use this columnar section to demonstrate rock magnetic property variations of different zonations with no implications for actual stockwork structures.

Rock Magnetic Measurements
Three ∼0.5 g chips were cut from each sample for magnetic measurements and the results were averaged. Powders were also prepared for specific magnetic measurements. The volume of chips and powders was calculated from their actual weight and density, enabling magnetization data to be volume-normalized. Susceptibility was measured with a 200 A/m applied field and 976 Hz operating frequency using an AGICO MFK-1FA Kappabridge at the Paleomagnetism Laboratory, Peking University (PKU). Natural remanent magnetization (NRM) was measured in a magnetically-shielded room at PKU using a 2G-755 cryogenic superconducting quantum interference device (SQUID) rock magnetometer. The Q-ratio, which is the remanent to induced magnetization ratio (Koenigsberger, 1938) (Thébault et al., 2015). Rock samples with Q > 1 can retain stable remanent magnetization, while those with Q < 1 are dominated by induced magnetization (Koenigsberger, 1938). Anhysteretic remanent magnetization (ARM) was imparted to powder samples packed in gelatin capsules using a D2000 alternating-field (AF) demagnetizer with 120-mT AF, 0.0075 decay rate, and 0.05-mT direct-current bias field. ARM intensity was measured using the 2G-755 magnetometer. Zhao et al. (1998) reported that sulfides in stockwork samples ( Figure S1 and Text S1 in Supporting Information S1) could complicate interpretation of magnetic components when heated above 300-400°C. We therefore conducted AF demagnetization and low-temperature magnetization measurements instead of thermal demagnetization or high-temperature measurements. The NRM was demagnetized by in-line AF coils attached to the 2G-755 magnetometer with 2 mT (0-10 mT), 5 mT (10-50 mT), and 10 mT (50-120 mT) steps. Low-temperature magnetic measurements were performed on powder samples packed in gelatin capsules using a Quantum Design Magnetic Property Measurement System (model MPMS3) at the School of Physics, PKU. Data average time of 5 s was used, which provides a sensitivity of ∼10 −11 Am 2 (∼10 −8 emu). Nine selected samples of representative lithology were both zero-field cooled (ZFC) and 2.5-T field-cooled (FC) with a cooling rate of 12 K/min from 300 to 10 K. A low-temperature remanence was imparted after each cooling procedure by a 2.5-T field at 10 K, and then magnetization was tracked in 2 K intervals or continuously when heating the specimen from 10 to 300 K with a 2 K/min rate. Subsequently, a room-temperature remanence was imparted by a 2.5-T field at 300 K, and then magnetization was tracked in 2 K intervals or continuously during low-temperature cycling (LTC) of 300 K -10 K -300 K with a 2 K/min rate.
Hysteresis properties were measured using a MicroMag 3900 vibrating sample magnetometer (VSM) at the Institute of Geophysics, China Earthquake Administration in Beijing. For hysteresis loops, a 0.5-T saturation field was used. Hysteresis parameters, including coercivity (B c ), saturation magnetization (M s ), and saturation remanent magnetization (M rs ), were obtained after paramagnetic correction using the program HystLab (Paterson  (Humphris et al., 1996), and only those from which samples were collected for this study are shown. (b) A composite columnar section constructed through correlations to Holes 957E-G ( Figure S2 in Supporting Information S1). Based on shipboard descriptions (Humphris et al., 1996), samples with different lithologies are indicated. MS = massive sulfides (breccia), PAB = pyrite-anhydrite breccia, PSAB = pyrite-silicaanhydrite breccia, PSB = pyrite-silica breccia, SWB = silicified wallrock breccia, CBB = chloritized basaltic breccia, BAS = basalt. et al., 2018). First-order reversal curves (FORC; e.g., Roberts et al., 2000) were measured for selected samples of representative lithologies. FORC data were processed using the software FORCinel version 3.06 (Harrison & Feinberg, 2008). Isothermal remanent magnetization (IRM) acquisition curves measured using VSM have noisy data due to weak magnetization of stockwork samples. Therefore, the IRM acquisition curves were remeasured using an ASC Scientific impulse magnetizer and the 2G-755 magnetometer at PKU. A total of 60 data points with a logarithmical distribution from 1 mT to 2.5 T were measured. The IRM data were both unmixed as individual samples using the application MAX Unmix (Maxbauer et al., 2016) based on skewed generalized Gaussian function (Egli, 2003) and as a collection along the composite columnar section using the endmember modeling algorithm of Heslop and Dillon (2007). The IRM acquired at 20 mT, 300 mT, and 2.5 T were extracted to quantify the magnetic contribution from low-coercivity and high-coercivity minerals using ratios of IRM 20mT /IRM 2.5T and IRM 300mT /IRM 2.5T , respectively.

Electron Microscopic Analyses
Rock thin sections were prepared for two or three selected samples of each representative lithology. They were carbon-coated and observed using an FEI QUANTA-650 FEG field-emission scanning electron microscope (SEM) at PKU operating at 10-15 kV accelerating voltage and 10 mm working distance. Mineral identification was based on energy-dispersive spectra (EDS).

Bulk Magnetic Parameters
Basalt basement has NRM and susceptibility values of 34.1 A/m and 0.0238 SI (5-sample average), while samples in stockwork (i.e., chloritized, silicified, anhydrite, and sulfide zones) have averaged NRM and susceptibility values of 23.2 mA/m and 1.39 × 10 −4 SI (50-sample average; Figure S3 in Supporting Information S1).  and 3f), respectively. These patterns suggest that fine-grained low-coercivity magnetic minerals are more enriched in the sulfide zone, though the overall concentration of magnetic minerals decreases from chloritized to sulfide zones. Basalt basement samples have IRM 300mT /IRM 2.5T ratio above 0.95, while most stockwork samples generally have IRM 300mT /IRM 2.5T around 0.90-0.95 (Figures 2h and 3h), confirming the presence of high-coercivity minerals such as hematite, but their variation pattern is unclear. Spikes in the depth profile (e.g., at ∼34-35 mbsf in Figure 2) may be related to isolated chloritized or basaltic clasts in overlying silicified, anhydrite, and sulfide zones transported by hydrothermal fluids Humphris et al., 1996). It is noteworthy that most samples are more or less inhomogeneous (error bars in Figures 2a-2f).
Fifty three of fifty five samples have Q-ratio above 1 with no noticeable variation trend (Figures 2c and 3c, and Figure S3 in Supporting Information S1), indicating that all lithologies in the stockwork can preserve stable remanent magnetization. The AF demagnetization of NRM reveals that nearly all samples contain strong secondary components that can be easily erased by 20-35 mT AF (Figures 4a-4c). 50 mT AF is sufficient to unblock or demagnetize >90% of the NRM, regardless of sample lithology (Figures 4d-4f). The declination and inclination here are not informative because our studied samples are unoriented. However, these secondary components were also observed in the AF demagnetization results of Zhao et al. (1998) with nearly subvertical inclinations.  Figure S2 of Supporting Information S1. The blue boxes and black segments give 1 and 1.5 interquartile ranges, respectively, while the red lines indicate median values. Red crosses are outliners. The data producing this figure can be found in Data Set S1 of Supporting Information S2. Therefore, analogizing with Zhao et al. (1998), these secondary components are interpreted as drilling-induced remagnetization. In this case, the Q-ratio can be deceptive because it overestimates the in situ NRM retained in stockwork samples.

Unmixed Magnetic Components From IRM Acquisition Curve
Single-sample-based IRM unmixing reveals a three-component regime for all analyzed samples, regardless of their lithologies (Figures 5a-5c). The low-field (blue) components have mean coercivity (B h ) around 20-25 mT (log B around 1.3-1.4) and dispersion parameter (DP) of 0.25-0.30. The median-field (violet) components have B h values of 40-65 mT (log B around 1.6-1.8) and DP values of 0.25-0.30. The high-field (green) components have B h values above 100 mT (log B above 2.0) and DP values above 0.35 but are barely saturated under a 2.5-T field. There are abundant low-field (blue) components in the basalt zone (Figure 5a), while the median-field (violet) component becomes dominant for stockwork samples (Figures 5b-5c). Furthermore, the high-field (green) component is more abundant in stockwork samples than in basalt samples.
We use endmember(EM)-based IRM unmixing to verify the observed single-sample-based pattern. We choose three endmembers for calculation based on the coefficient of determination (r 2 ) (Heslop & Dillon, 2007) and considering the single-sample-based unmixing results. The Four-endmember regime only provides minor improvements in the r 2 value ( Figure 5d) and can produce endmembers with significant overlaps in coercivity spectra ( Figure S4 in Supporting Information S1). The calculated EM1-EM3 (Figures 5f-5h) resemble the blue, violet, and green components from single-sample unmixing in coercivity spectra (Figures 5a-5c), respectively,  (Figure 5e): EM1 is most abundant in the basalt zone and becomes sporadic in stockwork; EM2 has a trend opposite to that for EM1; EM3 depletes in basalt zone, but gradually increases in chloritized zone and widely occurs in silicified, anhydrite, and sulfide zones.

Low-Temperature Characterization of Magnetomineralogy
All measured samples show typical magnetite Verwey transition (Verwey, 1939) in LTC and ZFC/FC curves ( Figure 6). Verwey transition temperature is 102-104 K for basalt zone and 113-120 K for chloritized, silicified, anhydrite, and sulfide zone samples, indicating magnetite particles in basalt samples are less stoichiometric than those in stockwork samples (Jackson & Moskowitz, 2021). In LTC cooling curves of basalt, chloritized, and silicified zone samples, magnetization gradually decreases towards the Verwey transition, producing a broad shoulder  Corresponding FC/ZFC curves of (a-e). The plots in the background give the first derivative of each measured low-temperature curve with related colors. Magnetite Verwey transition (Verwey, 1939) and hematite Morin transition (Morin, 1950) are indicated as the peak values in the first derivatives at reported temperature ranges. above the Verwey transition temperature in first derivatives (Figures 6c-6e). This is due to low-temperature demagnetization (LTD) caused by kinematic domain wall reorganization in multidomain magnetite (Muxworthy & McClelland, 2000). The LTD effect is less evident for anhydrite and sulfide zone samples (Figures 6a and 6b), indicating finer magnetite grains therein. The hematite Morin transition (Morin, 1950) is also recognized in 5 of 9 measured samples (e.g., Figure 6a and Figures 6c and 6d), which corroborates that the high-field IRM components are related to hematite ( Figure 5). Endmember-based IRM unmixing results suggest that hematite is likely ubiquitous in all stockwork samples, except for two samples from silicified and sulfide zones (Figure 5e). Samples in Figures 6a-6d and Figures 6f-6i all contain prominent high-field IRM components. Consequently, the absence of hematite Morin transition in some stockwork samples may be due to the overwhelming contribution of magnetite towards the overall magnetization (e.g., Lagroix & Guyodo, 2017). The observed Morin transition temperature range is 230-240 K, indicating that these hematite particles may have submicron sizes (Özdemir et al., 2008). Zhao et al. (1998) suggest that there are small amounts of pyrrhotite in sulfide samples based on a low-temperature transition of around 40 K. But in our analyses, the "40-K transition" has a much lower transition temperature (<25 K) than the Besnus transition at 30-34 K (Rochette et al., 1990). It is evident from ZFC/FC curves that such transitions are likely related to the blocking/unblocking of superparamagnetic or room-temperature paramagnetic mineral phases (Figures 6f-6j).

Grain Size and Domain State Variations Revealed by FORC Diagrams
In FORC diagrams, basalt samples have vertically dispersed distributions toward the B u axis ranging approximately −30 to 60 mT and a ridge along B u = 0 axis (Figure 7a), indicating contributions from vortex state and non-interacting single domain (SD) particles (Roberts et al., 2014;Lascu et al., 2018), respectively. Both signals are amplified in chloritized zone samples: vertical dispersion extends to approximately −90 to >70 mT along the B u axis, and the B u = 0 ridge becomes prominent (Figure 7b). The strongly interacting signals may be related to the intra-particle vortex "superstate" (e.g., Harrison et al., 2002). From chloritized to anhydrite zone samples, the vortex state vertical dispersion has shrunk to the range of about ±60 mT along the B u axis (Figures 7b-7d). The B u = 0 ridge is comparatively strengthened (Figures 7b-7d). The sulfide zone samples are dominated by non-interacting SD particles with a predominating B u = 0 ridge (Figure 7e). From chloritized to sulfide zones, the coercivity distribution gradually shifts toward the B u axis (Figures 7b-7e), reconciling with the upward decreasing trend observed for bulk coercivity (Figures 2e and 3e). Overall, the FORC diagrams reveal decreasing grain size for dominating low-coercivity magnetic minerals from chloritized to sulfide zones, consistent with the interpretations of the low-temperature magnetic measurements ( Figure 6) and bulk magnetic parameter variations (Figures 2 and 3).  (Egli, 2013) is applied to (b-e) with smoothing factors (S c0 , S c1 , S b0 , S b1 ) of (9, 9, 5, 9), (9, 9, 7, 9), (9, 10, 7, 10), and (7, 8, 4, 8), respectively. Horizontal and vertical lambda values are 0.1 for all samples in (b-e). Dashed lines indicate the 0.05 significance level (Heslop & Roberts, 2012).

Occurrences and Origins of Magnetic Minerals
Basalt zone samples have experienced slight hydrothermal alteration, as evidenced by the serpentinization of olivine phenocrysts (Figure 8a) and the replacement of interstitial glasses by chlorite (Figure 8b). However, the alteration is spatially inhomogeneous: chlorite and interstitial glass coexist on the sides of one plagioclase crystal (Figure 8b). Serpentinization-produced magnetite explains the Verwey transition in low-temperature measurements for the studied basalt zone samples (Figure 6a), while the dominant magnetic minerals in basalt samples are quenching-related titanomagnetite dendrites with Ti substitution (x value) of 0.65 ± 0.06 (averaged for 11 EDS measurements; e.g., spectrum #4 in Figure S5 and Table S1 of Supporting Information S1). Most primary nanoscale titanomagnetite inclusions may have been altered during hydrothermal alteration (e.g., Wang et al., 2020), though some survivors may exist in relatively fresh interstitial glasses (Figure 8b).  Figure S5 and Table S1 in Supporting Information S1. Different mineral phases are as indicated by abbreviations: Anh = anhydrite, Ccp = chalcopyrite, Chl = chlorite, Cpx = clinopyroxene, IG = interstitial glass, Mt = magnetite, Ol = olivine, Pl = plagioclase, Py = pyrite, Q = quartz, Rut = rutile, Ser = serpentine, Sp = sphalerite, Swf = silicified wallrock fragment, Tmt = titanomagnetite.
The overall magnetic mineral concentration has significantly decreased in chloritized zone samples due to hydrothermal dissolution, similar to the findings of Wang et al. (2020) from the Southwest Indian Ridge Longqi and Yuhuang hydrothermal fields. Some silicified wallrock fragments are found enclosed in chlorite matrix, where two groups of magnetic minerals with different grain sizes, corroborating the interpretations of FORC diagrams (Figure 7b), are embedded in between silicified plagioclase laths (Figures 8c and 8d). The micron-sized group is cratered titanomagnetite with variable titanium content (x values from 0.4 to 0.8; e.g., spectrum #12 in Figure S5 and Table S1 of Supporting Information S1), probably due to their different thermal histories before brecciation (O'Reilly, 1984). The other group is submicron magnetite, and it can be recrystallized from dissolved and exsolved Fe (e.g., Wang et al., 2021). Comparatively, relict Ti produces rutile clusters in silica matrix during enhanced silicification (Figures 8e-8f). Anhydrite and sulfide zone samples contain abundant sulfides. These sulfides are mostly massive pyrite with chalcopyrite and sphalerite inclusions in size of 10-50 μm (Figures 8g-8i). Although FORC diagrams and low-temperature magnetic measurements confirm the existence of fine-grained magnetite and hematite in anhydrite and sulfide zone samples, no ferrimagnetic minerals are directly observed under electron microscopes.
We perform detailed rock magnetic analyses on sulfide sample 158-957I1N1W28-31 to investigate possible sample-scale inhomogeneity of magnetomineralogy (Figures 3a-3f). This sample is chosen because (a) it is an intact corner of a drilled piece and (b) it has relatively strong saturation remanence (5.950 A/m, compared to ∼10 −1 A/m of other sulfide samples). The sample was first cut in half using a diamond wire saw and then subsampled using an electric diamond hand mill every 0.2 cm from piece exterior (0 cm) to interior (2 cm) along the central line ( Figure 9a). Each subsample was milled into powder with an approximate depth of ∼0.5 cm. Hysteresis measurements produce smooth loops ( Figure S6 in Supporting Information S1), which show a two-to three-fold decrease in saturation remanence from 0.6 to 1.8 cm, though the outermost 0.6 cm has some scattering (Figure 9b). For crosscheck, the same sample set with 500-mT IRM imparted by VSM was subsequently remeasured using the 2G-755 superconducting magnetometer. Results show a consistent trend with the VSM data (Figure 9b), though magnetic relaxation (in ∼48 hr) produced slightly decreased M rs values from the 2G-755 measurements (Table S2 in Supporting Information S1). The coercivity and squareness show no obvious pattern, but FORC diagrams show that the exterior has a more prominent vortex state signal than the interior of the sample (Figures 9c and 9e). It is worth noting that sister specimens for bulk magnetic parameter measurements were cut from the innermost part (∼2.0-2.5 cm) and show a dominant non-interacting SD state signal (Figure 7e). The vortex state may explain the scatter of the outmost 0.6 cm in the hysteresis profile (Figure 9b) since contrasting magnetic states are involved during vortex nucleation and annihilation (Lascu et al., 2018). Both near stoichiometric and oxidized magnetite exist in this sample because the identified Verwey transition has a temperature (T v ) of 114-118 K (Figures 9d and 9f) and the LTC curves show a hump between 300 K and T v (Özdemir & Dunlop, 2010). Together with the decreasing concentration and grain size of magnetic mineral from the piece exterior to the interior, we propose that the sample inhomogeneity may originate from an inward-propagating oxidation gradient (e.g., Fabian & Shcherbakov, 2020).

Discussion
The high-temperature hydrothermal alteration model of MORB  explains most of the magnetic property and magnetomineralogy variations in the TAG basalt, chloritized, and silicified zones. Nevertheless, it is noteworthy that the TAG mound has a slightly different alteration precursor compared to the Southwest Indian Ridge basalts: in addition to micron-scale titanomagnetite dendrites and submicron-scale titanomagnetite inclusions (Figure 8b), the TAG MORB contains olivine phenocrysts which were serpentinized during initial chloritization (Figure 8a), producing magnetite Verwey transition in basalt zone samples (Figures 6e and 6j). The chloritized basaltic breccia samples are petrographically comparable to the fully chloritized samples of Wang et al. (2020Wang et al. ( , 2021, and therefore a dual Verwey transition pattern is not expected. Nonetheless, cratered titanomagnetite residues in chloritized basaltic breccia samples (Figure 8d) can be evidence for Fe-Ti exsolution (e.g., Wang et al., 2021).
We did not observe in situ ferrimagnetic minerals directly under electron microscopes in anhydrite and sulfide zone samples due to their low concentrations and inconspicuous contrast with pyrite. Magnetic separation has been tried but is also inefficient to extract magnetic remanence carriers because gravitational torque always overrides magnetic torque in the suspension due to the large density of pyrite (∼4 g/cm 3 , Humphris et al., 1996).
However, it can be inferred from SEM analyses that magnetic minerals in anhydrite and sulfide zone samples must be related to sulfides, based on the fact that sulfides constitute up to 70%-100% of these samples and hydrothermal alteration has little contribution to the anhydrite and sulfide zone samples (Knott et al., 1998;Figure S1 in Supporting Information S1). Post cruise magnetic analyses by Zhao et al. (1998) have also interpreted that the magnetic minerals in sulfide samples are of chemical origins. Pyrrhotite can be an important magnetic carrier  Table S2 of Supporting Information S1. Sampling positions are as indicated in (a). The outermost and innermost subsamples have been used to perform first-order reversal curves (FORC) (c and e) and low-temperature measurements (d and f). FORC diagrams were obtained using VARIFORC smoothing (Egli, 2013) with smoothing factors (S c0 , S c1 , S b0 , S b1 ) of (5,7,5,9) and horizontal and vertical lambda of 0.1. Dashed lines indicate the 0.05 significance level (Heslop & Roberts, 2012). For low-temperature cycling measurements, blue and red curves are cooling and warming curves, as indicated by arrows. The plots in the background indicate the first derivatives, where the magnetite Verwey transition temperature (T v ) is marked respectively for cooling and heating curves. in sulfides (e.g., Honsho et al., 2016), but this can be excluded here since both petrographic (Knott et al., 1998) and rock magnetic analyses ( Figure 6) do not support the ubiquitous presence of pyrrhotite in the TAG mound.
Our sample-scale rock magnetic inhomogeneity test (Figure 9) shows that these sulfide-related magnetic minerals can originate from oxidation. Our rock magnetic analyses reveal similar magnetomineralogy and magnetic properties of sulfide samples to the post-cruise magnetic analyses by Zhao et al. (1998), implying that oxidation during sample storage can be minor. Water is a key oxidation reactant (Fallon et al., 2017) but is kept away from drilled pieces by the constant temperature and humidity in the core repository. Oxidation under seafloor conditions is therefore more likely. It may be related to the drilling process because the decreasing magnetic mineral concentration and grain size from the exterior to the interior of a drilled piece are unlikely to be in situ (Figure 9). Also, drilling-induced secondary remanent magnetization components have been observed during AF demagnetization of NRM (Figure 4; Zhao et al., 1998). One of the major aftermaths of drilling is the entrainment of seawater into the stockwork (Humphris et al., 1996). At temperatures between ∼150 and 300°C (considering the presence of anhydrite and its role in hydrothermal mineralization; Guo et al., 2020), the addition of sulfate from seawater will oxidize Fe 2+ to Fe 3+ (Shen & Buick, 2004): At the same time, Fe 2+ and Fe 3+ can form magnetite through the following series of reactions (Iwasaki et al., 2012): Pre-drilling oxidation may also be involved since almost all samples in the stockwork have high-coercivity hematite components (Figure 5e), which can form under ambient seafloor conditions (Fallon et al., 2017): 2FeS2 + 7.5O2(aq) + 4H2O → Fe2O3 + 4SO 2− 4 + 8H + Seafloor oxidation prior to drilling is also supported by the presence of red chert in other ODP Leg 158 holes (Humphris et al., 1996) and the fact that oxidative seawater can circulate to tens of meters beneath the TAG active mound (Pontbriand & Sohn, 2014;Tivey et al., 1995). It should be pointed out that little is known regarding the reaction rate of the aforementioned two oxidation paths (Fallon et al., 2017), but both pathways can certainly modify the magnetic properties of SMS deposits.
Our findings suggest that the physicochemical properties of SMS deposits are susceptible to disturbance from external environments. Drilling-induced oxidation can be a major concern because it releases heavy metal contents (Meng et al., 2021;Hu et al., 2022), which may have potential environmental impacts on hydrothermal ecosystems (Van Dover, 2019; Orcutt et al., 2020). On the other hand, attention should be paid to the potentially widespread seafloor oxidation during geological preservations of SMS deposits. From the perspective of magnetic properties, although oxidation produces subtle variations in bulk rock magnetic parameters in the TAG active mound (Figures 2 and 3), it can play an essential role in inactive and extinct mounds with longstanding exposure to oxidative seafloor environments. Neoformation of magnetite and hematite may complement the decreased magnetization in seafloor hydrothermal fields, complicating the interpretations of magnetic surveying data. For example, sulfides from inactive mounds of Longqi and Yuhuang hydrothermal fields in the Southwest Indian Ridge have enhanced NRM compared to fully chloritized basaltic breccia samples, partly due to oxidation-produced magnetite and hematite (Wang et al., 2020). In the TAG inactive mounds, gossan has been found as the caprock of SMS deposits . Although its magnetic property has not been systematically studied, there are cases where these iron-oxides and iron-oxyhydroxides contribute to the vent field magnetic anomalies (e.g., Gee et al., 2001).
Despite its significance, our knowledge of the potential oxidation of SMS deposits is still limited. One of the overlooked factors is the galvanic reactions between different polymetallic sulfides, as they have different rest potentials and seawater can serve as the required electrolyte (Fallon et al., 2017). Migrating electrons produce current in SMS deposits, creating spontaneous "geobattery" (Sato & Mooney, 1960). Seafloor self-potential surveys have shown that SMS deposits in inactive hydrothermal mounds have strong electric anomalies Zhu et al., 2020) and therefore confirm the existence of galvanic reactions in SMS deposits. Laboratory simulations have verified that galvanic reactions can accelerate the oxidation rates of sulfides (Knight et al., 2018). Although applying such observations to seafloor conditions remains challenging, these findings call for attention to the galvanic reactions. Overall, the galvanic reaction is only one of those controlling factors. More efforts, including but not limited to combined laboratory simulations and field observations, are needed to delineate how others affect the oxidation of SMS deposits. As SMS deposits are considered potential targets for mining Van Dover, 2019), such knowledge is instructive for understanding the complicated physicochemical properties of SMS deposits, which may ultimately contribute to designing future exploitation strategies.

Conclusions
A composite stratigraphic columnar section covering the basaltic basement and stockwork of the TAG active mound was constructed from multiple ODP Leg 158 drill holes. Bulk magnetic parameter measurements reveal an overall low magnetic mineral concentration in the stockwork. Comprehensive rock magnetic and electron microscopic analyses reveal that the hydrothermal alteration model of Wang et al. (2020Wang et al. ( , 2021 explains the occurrence of titanomagnetite and magnetite in chloritized and silicified zone samples. Although no ferrimagnetic minerals are directly observed using an electron microscope in anhydrite and sulfide zone samples, we provide rock magnetic evidence that they contain magnetite and hematite originated from oxidation of sulfides during drilling or under ambient seafloor conditions. Our results disclose that the magnetic properties of SMS deposits are sensitive to seafloor oxidative environments. Oxidation-related magnetite and hematite can potentially serve as magnetic fingerprints for distinguishing polymetallic sulfides from other hydrothermal products using magnetic surveying methods, which is essential for probing the complex structures beneath seafloor hydrothermal fields and prospection of SMS deposits. However, on the other hand, we recognize that our current knowledge remains insufficient in diagnosing the physicochemical mechanism of polymetallic sulfide oxidation both under ambient seafloor environments and during drilling or mining activities. These findings arouse the necessity of multi-disciplinary studies on the physicochemical alteration of SMS deposits before designing future exploration and exploitation strategies for SMS deposits.

Data Availability Statement
All rock magnetic data in this study are available at Peking University Open Research Data Platform (https://doi. org/10.18170/DVN/VXHZSP). Text S2 in Supporting Information S1 describes the data structure and instructs on usage.