Geochemical Analogy Viscosity of Mid-Ocean Ridge Basalt as an Indicator for Determining the Location of Seafloor Hydrothermal Fields?

Many achievements have been made in the exploration of seafloor polymetallic hydrothermal vents, however, it needs to further study how to discovery more potential new hydrothermal or massive sulfide fields. In this study, we collected 9818 major-element geochemical data sets of basalt samples from the Mid-Atlantic Ridge (MAR, 78°N–53°S) and the seafloor topographic data for eleven known hydrothermal vents (e.g., TAG, Snake Pit, MAR 22°30′N, Lilliput). Based on the different spatial distances between basalts and hydrothermal vents, we classified all the basalts into three categories: 1) basalt near known vent (focal objects of the research, 29 samples), 2) basalt around known vent (observed objects, 155 samples), and 3) other basalt of the MAR (as background basalt). Meanwhile, we used major element components of basalts to calculate the analogy viscosity and analyzed the possible correlation between analogy viscosity and the spatial distribution of seafloor hydrothermal vents. We found that the analogy viscosity of basalt near known vent fall into a narrow medium value range (∼46.8–68.5 Pa s) contrasted with the range of background basalt values range (∼4.0–356.0 Pa·s). It infers that these basalts on the Mid-Atlantic Ridge with the medium values of analogy viscosity favor the formation of hydrothermal vents. However, due to the complexities of magmatic activity and fault structure, as well as the multi-stage water-rock reaction, whether the analogy viscosity of basalts can be used as an indicator for determining the location of seafloor hydrothermal field or not needs further verification. Hence, this paper provides a possible rule (the variation of analogy viscosity estimated by major element components in the basalts) for exploration the location of hydrothermal vents rather than a fully developed method.


INTRODUCTION
Hydrothermal fields are rich in biological and mineral resources and provide a window to the Earth's depth. Finding and studying hydrothermal vents is one of the most interesting scientific aims worldwide. However, finding hydrothermal vents, especially those inactive hydrothermal zones on the vast, and dark seafloor, which is covered by several kilometers of water, is an extremely challenging task. The vicinities of seafloor hydrothermal fields exhibit special characteristics: higher anomalies of heat flow on the seafloor (e.g., Chiba et al., 2001;Dias et al., 2010;Schmidt et al., 2011), frequent seismic activity (e.g., Johnson et al., 2000;Seewald et al., 2003;Crone et al., 2010), oceanic core complex outcrop and detachments around some of the vents (e.g., Morris et al., 2009;Cheadle and Grimes, 2010;Macleod et al., 2011), increased turbidity on the seafloor (e.g., Baker et al., 2006), and high magnetic anomalies (e.g., Miranda et al., 2005;Rona et al., 2013). According to Baker and German (2004), hydrothermal activity is positively correlated with the full spreading rate of the mid-ocean ridge. The search for new indicators of hydrothermal vents will therefore assist both the study and exploration of these fascinating appearances.
The geochemical characteristics of basalt near hydrothermal vents differ from those of normal mid-ocean ridge basalt (N-MORB) in the Mid-Atlantic Ridge (MAR). Wang et al. (2014) reported that the basalt of the 49.6°E hydrothermal field on the southwest Indian Ridge is richer in K 2 O, Na 2 O, Fe 2 O 3 , and MnO, and poorer in SiO 2 and Al 2 O 3 , than N-MORB. Gale et al. (2013b) compared the compositions of basaltic glass in the French American Mid-Ocean Undersea Study (FAMOUS) data and North FAMOUS segment. Basaltic glass on the North FAMOUS segment, which is near a hydrothermal vent, is rich in SiO 2 , FeO, TiO 2 , and K 2 O and poor in Al 2 O 3 , CaO, and MgO. The viscosity of basalt might reflect its geochemical characteristics (Zhu et al., 1997). It is an important physical property that affects the rising, crystallization differentiation, degassing, ejection, and other behaviors of magma (e.g., Papale, 1999;Sparks, 2003;Dingwell, 2006) and restricts the occurrence, topography, and other features of basalt. By analyzing the viscosity difference between the basalt around known vents (BNV) and the background basalt of MAR, we might obtain a new identifying indicator of hydrothermal vents.
Seafloor hydrothermal activity was first discovered in 1948 (Bruneau et al., 1953). In 1985, the first active hydrothermal field, the trans-Atlantic geotraverse (TAG) field, was discovered in the MAR (Rona et al., 1986). Since then, 721 hydrothermal seabed vents have been discovered worldwide. Among these, 304 have been confirmed as active, 362 have been inferred as active, and more than 50% are located in the mid-ocean ridge (The InterRidge Database, 2020, Accessed date is 2021. 03. 21). By studying the known hydrothermal vents, we might reveal their FIGURE 1 | Basalt sampling sites and locations of the 11 known seabed hydrothermal vents (for information, see Table 1). Shaded relief bathymetric map from GeoMapApp, 2021.  Haase et al. (2009) spatial distribution rules and control factors. To determine the differences in the major elements and viscosities of the BNV and background basalt on the MAR, we collected and analyzed thousands of basalt datasets sampled from the MAR. This paper reports the findings of the study.

Petro-Geochemistry Data
More than 35,000 datasets of rock/glass samples from the MAR were collected and preliminarily analyzed. Some of these samples were located near 11 known hydrothermal vents. The locations and basic information of the samples near the 11 vents, on which this research focuses, are presented in Figure 1 and Table 1, respectively. The data were mainly collected from the Petrological Database of the Ocean Floor (PetDB, http://www.earthchem.org/ petdb/search), a database of global seabed rock (Lehnert et al., 2000) was downloaded on 26 September 2017. In addtition, we also collected some data from Haase et al. (2009). Gale et al. (2013a) used datasets from PetDB to study the mean composition   (Giordano et al., 2008).  of mid-ocean ridge basalts and found some geochemical problems associated with Earth composition and evolution. According to Gale et al. (2013a), the sequence of steps that are required to eliminate questionable or inapplicable analyses and retain only the highest-quality data can be given as follows: 1) Sample location: To assign each sample to the appropriate MAR, only the samples that were located on MAR and its flanks were retained. 2) Altered samples: According to the descriptions that were obtained from PetDB regarding the samples, we eliminated the altered or mineralized rock samples, including those that were altered by seawater. Further, only fresh basalt/basalt glass remained. 3) Complete set of oxides: For the major element data, each sample was required to contain Fe (as FeO and/or Fe 2 O 3 ), SiO 2 , TiO 2 , Al 2 O 3 , MgO, CaO, Na 2 O, and K 2 O. Duplicate samples, which possess identical location and major element compositions, were eliminated. Fe required special attention to determine the total amount of Fe (FeO T ) due to the diverse forms in which it has been reported in PetDB, which is given by Gale et al. (2013a) as FeO T = FeO + 0.9 × Fe 2 O 3 . 4) Igneous compositions and major-element sums: Data were further filtered to eliminate the samples outside the basic-rock range because the vast majority of known hydrothermal vents are observed to develop on a basalt basement (only eight known hydrothermal vents are located in the ultramafic basement worldwide). We retained the samples having 45-53 wt% composition of SiO 2 . The final filter to ensure data quality involved a check of the major elemental sum (SiO 2 + Al 2 O 3 + MgO + TiO 2 + FeO T + CaO + K 2 O + Na 2 O + MnO + P 2 O 5 ). Since MnO and P 2 O 5 are not analyzed by all the laboratories, we calculated the placeholder values strictly for the sum calculations (MnO = FeOT/54; P 2 O 5 was estimated using the polynomial best-fit equation that is based on the K 2 O−contents). Further, the sum was estimated as mentioned previously, and any analyses depicting a sum that was either less than 98 wt% or greater than 101 wt% were eliminated. 5) Estimating H 2 O: H 2 O was estimated using H 2 O est = 1.5 × K 2 O (Langmuir et al., 2006) that was applicable for open-ocean ridge basalts. Finally, 9818 datasets were retained to be the study materials. Their locations are illustrated in Figure 1. The basalt/basalt glass datasets were classified by their total alkali silica (TAS) (Maitre, 1984). The classification results are shown in Figure 2. Most of the samples are basalt; the remaining minority are trachybasalt, basaltic trachyandesite, tephrite/basanite, or phonotephrite. Herein, all samples are referred to as basalt.

Bathymetry Data
Bathymetric maps were taken from the GeoMapApp of the Marine Geoscience Data System on 21 Jan 2022 (GeoMapApp, 2021). Bathymetric data were extracted from the Global Multi-Resolution Topography database with a 100 m grid (for MAR 22°30′N, Snake Pit, the Lilliput field (vents 7-11), and S oceanographer) (Ryan et al., 2009), the Woods Hole Oceanographic Institute (WHOI) with a 20 m grid (Tivey et al., 2003) and a 0.25 m grid (Roman and Singh, 2005;Roman and Singh, 2007) (for TAG), and the WHOI with a 10 m grid (for Lucky Strike and Evan) (Scheirer et al., 2000).

Analogy Viscosity
The earliest models that were used for estimating the viscosity of geologically relevant silicate melts adopted a strict Arrhenius temperature dependence such as the most commonly used Arrhenius model of Fulcher (1925) and Shaw (1972). Giordano et al. (2008) compared their non-Arrhenius model with the model of Shaw (1972) and demonstrated that the former is better. Hence, viscosity was estimated using the model of Giordano et al. (2008) in the following manner: (1) Here, η is the viscosity of magmatic silicate melt (Pa·s), T is the thermodynamic temperature (K), and A, B, and C are adjustable parameters, including the pre-exponential factor, pseudoactivation energy, and VFT 1 −temperature, respectively.
The optimal values for the 18 model parameters and their corresponding confidence limits are reported in Table 2; and a demonstration calculation is provided in Table 3. The minimum temperature of tholeiite erupting from the Kilauea volcano (Hawaii) is 1,423 K (Wright et al., 1968), and 1,433 K at a vapor pressure of 1,000 × 10 5 kPa (Xu & Qiu, 2010). Therefore, in the viscosity estimates of Eq. 1, we set the melting temperature of oceanic basalt to the middle temperature (1428 K) of the Kilauea tholeiite in 1,000-4,000 m depth range (100 × 10 5 -400 × 10 5 kPa) in which vents may appear. Due to constraints such as the water content in the melting basalts was given indirectly with an empirical formula, the analogy viscosity more like a comprehensive geochemical characteristics of basalt than the viscosity value in the physical sense.

THREE CLASSIFICATIONS OF BASALTS
To compare the characteristics of the BNVs and the background basalt on MAR, we classified the basalt samples into three categories based on the spatial distance and topography coherence of their sampling sites to known hydrothermal vents. 1) Basalt near known vent (BNV): The "No." is the number of basalt sampling site, and a plurality of samples can be found around a site. "Lat." is latitude, and "Lon." is longitude. "Visc." is viscosity. "value" refers to compute value. Bold values are compute values.  sites are located farther from the known hydrothermal vents than BNVs, but may be associated with hydrothermal vents. They are separated by 10 3 -10 4 m from the known hydrothermal vents and all in the same topographic unit. Or, they are separated by 10 2 -10 3 m from the known hydrothermal vents in different topographic units. 3) Basalt of the MAR: The basalt samples in the MAR are far (>10 4 m) from the known hydrothermal vents or no hydrothermal vents were found in the range. The basalt of the MAR is treated as the background basalt of the first two categories (especially of BNV). According to the above BNV classification standard, BNV samples from only 11 hydrothermal vents were collected over the whole MAR. These vents are the focal objects of this paper. The classifications of the sampling sites near each vent are described below.

Trans-Atlantic Geotraverse Vent
The TAG vent was discovered by Rona et al. (1986) in 1985. It is located on the north side of a large hydrothermal sulfide mound at the bottom of the axial valley, which has a maximum depth of 3,670 m. Thirty-nine basalt geochemical datasets have been collected from 36 sampling sites around the TAG vent ( Figure 3). The major-element concentrations and viscosities at each site are listed in Table 4. Sites 24, 25, and 26 occupy the basement basalt under the large hydrothermal sulfide mound containing the TAG vent. Basalt samples from sites 24 and 26 were collected by drilling, and the sample from site 25 was collected by submersible equipment. These basalt samples were classified as BNV. The remaining sites are not beneath the hydrothermal sulfide mound and are approximately 1-15 km from the TAG vent; hence, their basalt samples were classified as BAV.

Snake Pit Vent
The Snake Pit vent was discovered by Kong et al. (1985). It is situated on an intensely fissured neovolcanic ridge along the axial valley, which occupies the south side of the 3500-m-deep Kane transform fault. Thirty-two basalt geochemical datasets from 23 basalt sampling sites have been collected around the Snake Pit vent (Figure 4). Table 5 to continue lists the major-element concentrations and viscosities of the basalt at these sites. The Snake Pit vent and sites 12, 13 are located at the top of the same neovolcanic ridge, belong to the same topography unit, and are less than 300 m apart. Therefore, the basalt samples from these sites were classified as BNV. The remaining sites are located 1-5 km from the vent even in different topographic units. Therefore, samples from these sites were classified as BAV.

Lucky Strike Vent and Evan Vent
The Lucky Strike vent was discovered by Alvin in 1993 (Langmuir et al., 1993), and the Evan vent was discovered by Nautile in 2006 (Escartin, 2009). Both vents are located on the seamount in the central part of the Lucky Strike segment ( Figure 5). Water depths at the Lucky Strike and Evan vents are 1740 and 1775 m, respectively, and the vents are approximately 3.5 km apart. 42 basalt geochemical datasets were collected from the 40 sampling sites around the two  Table 6. The Lucky Strike vent and site 14 occupy the depression between the two high topographies and are observed to be not more than 200 m apart. The Evan vent and sites 31 and 33 are also separated by only 400 m and occupy the same slope of the seamount having a similar water depth. Therefore, the basalt samples at sites 14, 31, and 33 were classified as BNVs. The remaining 37 sites and two hydrothermal vents are located in different topographic units or are separated by more than 1 km; hence, the samples that were collected from these sites were classified as BAVs.

S Oceanographer Vent
The S Oceanographer vent, located on a smaller seamount at the bottom of the axial valley, was discovered by Chin et al. (1998). Its depth is 3,460 m. There are 32 basalt sampling sites around the vent, from which 37 geochemical datasets of basalt were collected ( Figure 6). Their major-element concentrations and viscosities are listed in Table 7. The S Oceanographer vent and site 20 are separated by approximately 500 m at the bottom of the axial valley; hence, the basalt sample at site 20 was classified as BNV. The other sites are 1-20 km from the vent; hence, their basalt samples were classified as BAV.

Mid-Atlantic Ridge 22°309N Vent
The MAR 22°30′N vent was discovered by Rona et al. (1986). It is located on the east of the axial valley scarp, which is 2,800 m deep. Twelve basalt geochemical datasets were collected from four basalt sampling sites around the vent (Figure 7). The majorelement concentrations and viscosities of the basalt at each site are listed in Table 8. Site 2 is only 10 m from the MAR 22°30′N vent; hence, its four basalt samples were classified as BNV. The      other sites are more than 5 km from the vent; hence, their basalt samples were classified as BAV.

Lilliput Field
The Lilliput field was preliminarily discovered by German et al. (2002) in 2002 and confirmed by ROV investigation in 2002 (Haase and Scientific Party, 2009). It occupies the segment between the Ascension and the Bode Verde fracture zones, which are 1,500 m deep. The areal coverage of the Lilliput hydrothermal field is approximately (1,000 × 250) m 2 . Five relatively small hydrothermal vents are found on a large (~5,000 m × 600 m) young (estimated as <100 years old) lava flow. From observations of the Lilliput vent fauna over four consecutive years, researchers have inferred a pulsed hydrothermal activity (Haase et al., 2009). Twenty-two basalt geochemical datasets have been collected from 19 sampling sites around the Lilliput vent ( Figure 8). The major-element concentrations and viscosities at these sites are listed in Table 9. The vents of the Lilliput field and sites 1-8 are separated by <500 m on the same slope (possibly created by the same lava flow); hence, their basalt samples were classified as BNV. The remaining sites were distant (<1 km) from the hydrothermal vents; hence, their basalt samples were classified as BAV.
The above classification grouped the 9818 sampled basalt datasets into 29 BNV datasets and 155 BAV datasets. The differences between both types of basalt samples and the background basalts on the MAR are compared in the following section.

RESULTS
The latitudes and the analogy viscosity provided in the 9818 datasets were plotted in a scatter diagram, with latitude as the ordinate and viscosity as the abscissa (Figure 9). The viscosities of the basalts from the MAR, BAV, and BNV groups were 4-356, 17-117, and 46.8-68.5 Pa·s, respectively. As shown in Figure 9, the viscosity range narrows near the vents (e.g., 46.8-68.5 Pa·s), but many of the BAV samples are likely to be background basalt on the MAR. Among the whole 9818 basalt samples collected from the MAR, 31.1% have viscosities ranging from 46.8 to 68.5 Pa·s. In other words, less than one-third of the basalt samples from the MAR fall within the viscosity range of BNV.
Meanwhile, the viscosities of the BNV, BAV, and MAR basalt samples were statistically analyzed, and the results were plotted as histograms ( Figure 10). The modes and means of the viscosities are 54-56 Pa·s and 50.49 Pa·s for the MAR basalt, 58-60 Pa·s and 56.07 Pa·s for BAV, 56-58 Pa·s and 57.71 Pa·s for BNV respectively. The mean viscosities of the BNV and BAV samples are similar but clearly differ from those of the MAR basalt. The viscosity standard deviations of the MAR basalt, BAV,and BNV are 24.30,18.44,and 5.30,respectively. Note that the standard deviation is much smaller in BNV than in the other basalt groups, indicating a stronger aggregation effect of the viscosity in BNV.
In addition, all basalt samples are classified by the maximum and minimum concentrations of BNV major elements and the proportion of the Mid-Ridge basalt samples belonging to BNV category is listed in Table 10. These results show that the concentration range of major elements in most basalts as the same as that of BNV, while only 31% of basalts had the analogy viscosity similar to BNV which shows that the analogy viscosity may be another suitable indicator for the seafloor hydrothermal vents.

DISCUSSION
The major-element concentrations of basalt have limited ability to identify hydrothermal vents (Table 10). To understand this result, we plotted the analogy viscosity versus oxide content for the three types of basalt ( Figure 11). BNV is divided into two distinct categories along the horizontal axis: Type 1, located in the north MAR (20-40°N), and Type 2, located in the south MAR (9°S). Because the geochemical compositions of BNV differ among the MAR locations, they are unsuitable as indicators of hydrothermal vents. In contrast, the analogy viscosity of BNV is highly consistent among the locations and better identifies the hydrothermal vents. According to statistical analysis, the range of BNV analogy viscosity differs from that of the MAR (46.8-68.5 Pa·s in the former vs. 4-356 Pa·s in the latter). This suggests that a narrow analogy viscosity range (e.g. 46.8-68.5 Pa.s) favors the formation and development of hydrothermal vents. To date, the role of basalt viscosity in restricting the hydrothermal vents is not clear, which require further research. Due to the complex formation mechanism, controlling factors and individual differences of the seafloor magmatic system, tectonic fault system and hydrothermal evolution system, it may not be comprehensive to take the basalt viscosity of the FIGURE 9 | BNV, BAV, and MAR basalt samples (distinguished by pink crosses, cyan diamonds, and black circles, respectively) plotted on a viscosity-latitude scatter diagram. All BNV and BAV, and MAR basalts with viscosities below 160 Pa·s (this excludes 20 samples with viscosities between 160 Pa.s and the maximum viscosity of 574 Pa·s) are shown.  "Max" and "min" means the maximum and minimum of BNV. "Proportion" means the ratio of samples between max and min to the total samples on MAR., The smaller the ratio, the better the recognition rate of the hydrothermal vents. Obviously, the viscosity has the best identification rate for the hydrothermal vents. deep magma, fractional crystallization of magma chamber, different development degree of fault zone, and the timing of water-rock reaction, the indicator of analogy viscosity may only provide a possibility for the exploration of submarine hydrothermal field.
First, the viscosity or composition of the rock formations hosting such systems may be of little influence in their generation (besides having some influence on the elastic properties and therewith the fracture behavior of the material). The fracture process itself is governed by the local stress field and the elastic, response of the solidified magma to this stress. In fact, all of the zones that selected for analyses of hydrothermal vent regions in this study lie on fault zones. These are the result of tectonic processes that induce the fractures requires for the generation of permeability. How recovering samples from the ocean floor, performing compositional analyses and then calculating viscosity should return better results at identifying hydrothermal vents than the first order observations made during the dives required to recover these samples in the first place. Further, identifying thermal or chemical signatures of active vents in the water column maybe drastically faster, easier than analogy viscosity, though it may be a useful tool for exploration for hydrothermal vents as found here.
Second, crustal magma transport systems, independent of their geometry, remain filled with rock after solidification of the magma and fluid pathways (used by sea water during hydrothermal activity) are subsequently induced by fracturing. It is this fracturing process in combination with the temperature differential between the intrusion and the sea water that drives fluid circulation and hydrothermal activity (Baker and German 2004). Hydrothermal activity is correlated to spreading velocity since higher spreading velocity means higher heat flow and more intense fracturing. On the other hand, since the H 2 O content drastically affects melt viscosity, simply assuming H 2 O = 1.5* K 2 O maybe not a suitable approach for global mid-ocean ridge system. Varying the water content from 0 to 0.3 wt% for the melt of Table 3 at constant temperature induces a variation in modelled viscosity of > 100 Pas -much larger than the proposed "identifier viscosity range." Especially since water solubility in melts is extremely sensitive to changes in pressure, water depth would likely be a much better indicator for water solubility.
Third, calculated viscosity from compositional data using an empirical model would introduce further uncertainty, and any apparent differences between the samples are composition driven and that information is already present in the compositional data, so whether viscosity is a better parameter for determining the location of hydrothermal vents than composition is not certain. For example, the uncertainty of the GRD model is~± 5% of the measured viscosity value (in this case that translates to~± 2.5 Pa s, i.e.~± 10% of the range suggested relevant to vent formation).
Fourth, the computed viscosities change only with composition and are assumed to have the same temperature (1,423°C). This may be not suitable because lava viscosity is very sensitive to temperature, and lava composition changes with temperature, due to fractional crystallization. The proposed explanation for finding that vents favoring a limited range of intermediate viscosities is based on the assumption that the computed viscosities were those of the lavas when they erupted and eventually formed as crust. But it is not clear the computed viscosities accurately represent the eruption viscosities, given the lack of temperature information. Alternative methods such as the MELTS algorithm are able to compute viscosity from major element compositions of basalts by also calculating the liquidus temperatures. Using such a method would be more meaningful to this study. Assuming one temperature for all the lavas is an important shortcoming.
Therefore, whether the analogy viscosity of basalts can be used as an indicator for determining the location of seafloor hydrothermal field needs more observations to be verified due to the complexities of magmatic activity and fault structure, as well as the multi-stage water-rock reaction. Hence, this paper provides a possible insight for exploration of hydrothermal vents rather than a fully developed method.

CONCLUSION
According to the calculation results of analogy viscosity, the basalts along the MAR system with analogy viscosity values ranged from 46.8 to 68.5 Pa·s are considered to be suitable for exploration the location of hydrothermal vents. However, because of the complex seafloor magmatic system, tectonic fault system and hydrothermal evolution system, the analogy viscosity parameter may only provide a possible insight for the exploration of submarine hydrothermal field rather than a developed method, which needs more observations to be verified.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.