Distribution of Rare Earth Elements plus Yttrium among Major Mineral Phases of Marine Fe – Mn Crusts from the South China Sea and Western Pacific Ocean : A Comparative Study

Marine hydrogenetic Fe–Mn crusts on seamounts are known as potential mineral resources of rare earth elements plus yttrium (REY). In recent years, increasing numbers of deposits of Fe–Mn crusts and nodules were discovered in the South China Sea (SCS), yet the enrichment mechanism of REY is yet to be sufficiently addressed. In this study, hydrogenetic Fe–Mn crusts from the South China Sea (SCS) and the Western Pacific Ocean (WPO) were comparatively studied with mineralogy and geochemistry. In addition, we used an in situ REY distribution mapping method, implementing laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) and a sequential leaching procedure to investigate the partitioning behavior of REY in the Fe–Mn crusts. The typical Fe–Mn crusts from SCS were mainly composed of quartz, calcite, vernadite (δ-MnO2), and amorphous Fe oxides/hydroxides (FeOOH). The Fe–Mn crusts from the Central SCS Basin and the WPO contained quartz, δ-MnO2, FeOOH, todorokite, and phillipsite. Furthermore, geochemical analysis indicated that the typical SCS crusts had a higher growth rate and lower REY concentrations. The LA-ICP-MS mapping results showed that the δ-MnO2 and FeOOH dominated the occurrence phases of REY in the SCS crusts. Four mineral phases (i.e., easily exchangeable and carbonate, Mn-oxide, amorphous FeOOH, and residual aluminosilicates) in these Fe–Mn crusts were separated by a sequential leaching procedure. In the SCS and WPO crusts, the majority of total REY (ΣREY) was distributed in the Mn-oxide and amorphous FeOOH phases. The post-Archean Australian shale-normalized REY patterns showed that light REY (LREY) and heavy REY (HREY) were preferentially adsorbed onto δ-MnO2 and FeOOH, respectively. It is noteworthy that ~27% of ΣREY was associated with the residual aluminosilicates phase of the WPO crusts. The La/Al ratios in the aluminosilicates phase of the typical SCS crusts were the values of the upper crust. We conclude that large amounts of terrigenous materials dilute the abundance of REY in the SCS crusts. In addition, the growth rates of Fe–Mn crusts have a negative correlation with the FeOOH-bound and aluminosilicate-bound REY. As a result of the fast growth rates, the SCS crusts contain relatively low concentrations of REY.

The South China Sea (SCS), one of the largest marginal seas, is strongly subject to large terrestrial inputs and has very fast rates of sediment accumulation.Generally, the fast sediment accumulation rate is unfavorable for marine Fe-Mn precipitate mineralization.Yet, increasing numbers of deposits of Fe-Mn crusts and nodules were discovered in SCS in recent years and hotly studied [19,20,22].Some mineral and geochemical data of the SCS Fe-Mn crusts were reported, showing that these SCS Fe-Mn crusts are hydrogenetic in origin [19,20].Compared to the Fe-Mn crusts in the Pacific Ocean, the SCS ones have much higher contents of Li, Al, Si, Fe, Pb, and Th, but lower contents of Co, Ni, Cu, Zn, REY, and other heavy metals [19,20,23].However, the interelement correlations and phase associations in the SCS crusts are yet to be sufficiently addressed.Previous studies were mostly concerned with the bulk composition of the deposits, which is not enough to understand how REY is scavenged from seawater in the continental marginal environment.
In this contribution, the Fe-Mn crusts from the SCS and WPO were comparatively studied with mineralogy and geochemistry.The samples were examined using in situ laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) mapping in order to observe the occurrence phase of REY in the layer structure.Furthermore, major mineral phases (i.e., easily exchangeable and carbonate, Mn-oxide, amorphous FeOOH, and residual aluminosilicates phases) of these Fe-Mn crusts were separated using a sequential leaching experiment which was proven to be an effective method for understanding the incorporation of elements in mineral phases of Fe-Mn crusts [8,14,[24][25][26][27].These results are expected to provide better information on the process of REY incorporation into Fe-Mn crusts in a marginal sea.

Geological Setting and Samples
The submarine topography of the SCS appears as a ladder-like form from the edge to the center (Figure 1) [28].Located near the focus of the convergence among three plates (i.e., Eurasian, Pacific-Philippine Sea, and Indo-Australian) with two superconvergent zones (the circum-Pacific and Tethyan zones) [29], the SCS has a very complex developmental process and basement rocks from multiple periods.The basement of the SCS mainly consists of Cenozoic metamorphic rock in the center and Paleozoic-Mesozoic basement on the edge.The edge of the SCS is a continental slope with elevated seafloor features, while the center of the SCS is a deep-sea basin with many seamounts and sea hills.On the seafloor of the SCS, terrigenous sediment, hemipelagic clay, and siliceous ooze distribute from the continental slope to the Central SCS Basin [30].The Western Pacific Ocean (WPO) has a deep basin with a depth of over 6000 m, where there are a large number of islands and chain-shaped seamounts (such as Magellan seamounts and Marcus-Wake seamounts).
Fe-Mn crusts from the SCS and WPO were studied (coordinates as shown in Table 1).The SCS crusts were collected by trawling on the research vessel "Haiyangsihao" of the Guangzhou Marine Geological Survey (GMGS) in 2011 and 2012, and the sampling depths ranged from 1150 m to 1160 m.The WPO crusts were collected by shallow core drilling on the research vessel "Haiyangsihao" of the GMGS during the 27th cruise in 2012.Sample ST1 involves the northern continental slope of the SCS; ZSQD42A involves the north-western (NW) continental slope of the SCS; ZSQD253A grows on the seamount in the eastern SCS; HYD66-2 involves the seamount in the Central SCS Basin; and NASD44-1 and MP4SD24-1 involve Magellan seamounts and Marcus-Wake seamounts, respectively.The substrates of these Fe-Mn crusts were limestone, altered basalt, and breccia (Table 1).The detail petrographic study was reported in Reference [19].

Mineralogical Analysis of Bulk Samples
Each sample was washed with ultra pure water (Millipore, MA, US), dried in a blower oven at 30 °C for 24 h, and then ground into 200-mesh powders in an agate grinder.Then, 200 mg of each of the powders was analyzed by X-ray diffraction (XRD).XRD data were collected on a two-dimensional (2D) image-plate detector using a Rigaku D/MAX Rapid II instrument (Rigaku, Tokyo, Japan) taken with a 0.1-mm collimator using Mo-Kα radiation at Guangdong Provincial Key Laboratory of Marine Resources and Coastal Engineering (GPKLMRCE), Sun Yat-sen University (SYSU, Guangzhou, China).All samples were analyzed twice under the same conditions; then, the signals were merged in order to improve the signal-to-noise ratio.Identification of major minerals was completed via PDXL2 software (Rigaku, Tokyo, Japan).

Mineralogical Analysis of Bulk Samples
Each sample was washed with ultra pure water (Millipore, MA, USA), dried in a blower oven at 30 • C for 24 h, and then ground into 200-mesh powders in an agate grinder.Then, 200 mg of each of the powders was analyzed by X-ray diffraction (XRD).XRD data were collected on a two-dimensional (2D) image-plate detector using a Rigaku D/MAX Rapid II instrument (Rigaku, Tokyo, Japan) taken with a 0.1-mm collimator using Mo-Kα radiation at Guangdong Provincial Key Laboratory of Marine Resources and Coastal Engineering (GPKLMRCE), Sun Yat-sen University (SYSU, Guangzhou, China).
All samples were analyzed twice under the same conditions; then, the signals were merged in order to improve the signal-to-noise ratio.Identification of major minerals was completed via PDXL2 software (Rigaku, Tokyo, Japan).

Qualitative Element Mapping by LA-ICP-MS
In situ element mapping was conducted at GPKLMRCE following the analysis procedure described by References [31,32] The basic principle involves the laser running along several parallel, adjoining, and equally long lines to generate an element map.In this study, laser sampling was performed using an ArF excimer laser ablation system (GeoLasPro, Coherent Inc., Dieburg, Germany) and the ion-signal intensities were acquired using an Agilent 7700× inductively coupled plasma mass spectrometer (ICP-MS) (Santa Clara, CA, USA).A beam size of 24 µm was employed with an energy density of 5 J/cm 2 and a repetition rate of 5 Hz.A six-pulse ablation was done for each spot, and the next spot was 20 µm away, which eventually ablated a small overlap to avoid gaps.Once a line of spots was completed, the laser was immediately moved to the beginning of the next parallel line with a line spacing of 20 µm, and the protocol was repeated.After the data acquisition, data reduction was performed using the ioGAS software [31] and, eventually, a series of qualitative mapping images of the elements was produced.

Geochemical Analysis of Bulk Samples
For each sample, 50 mg of 200-mesh powder was treated with a 20-mL acid mixture (HF:HNO 3 :HClO 4 = 7:3:1) in Teflon beakers for 2 h, and then evaporated to dryness on a hot plate at 200 • C. The dryness was dissolved in 10 mL of 50% HNO 3 and diluted to different concentrations with 5% HNO 3 before the measurement.The concentrations of major elements were measured using a PerkinElmer (Waltham, MA, USA) Optima 8300 inductively coupled plasma atomic emission spectrometer (ICP-AES) at Instrument Analysis and Research Center (IARC, SYSU) using an incident power of 1300 W, plasma gas flow of 12 L/min, and atomized gas flow of 0.55 L/min.The trace elements were determined using an Agilent 7700× ICP-MS at GPKLMRCE.Cerium anomaly (Ce/Ce*), europium anomaly (Eu/Eu*), yttrium anomaly (Y n /Ho n *), and light and heavy REY (LREY/HREY) enrichment (La n /Yb n ) were calculated using the following equations: where the PAAS data referred to Reference [33].

Sequential Leaching Experiment
The four major mineral phases of all crust samples were chemically separated following a sequential leaching procedure [27]; the flow chart of the procedure is shown in Figure 2. In brief, 0.5 g of the powdered samples was put in a 100-mL plastic centrifuge tube, mixed with the leaching reagent, and shaken at room temperature.The leaching solution was filtered through 0.45-µm Millipore filter paper, and the filtrate was stored for analysis.The residues on the filter paper were washed twice with Milli-Q water to clear the remaining dissolved fraction of the previous step and then immediately transferred to the next leaching step.The following four mineral phases were separated: The leaching experiments were carried out in a Class-100,000 clean room at GPKLMRCE.Concentrations of major and trace elements in four leaches were determined in the same way as the bulk geochemical analysis.The concentrations of each element in the four leachates were summed up and compared with the corresponding bulk data to check the recovery (within the range of 80-120%; see Table S1, Supplementary Materials).The leaching experiments were carried out in a Class-100,000 clean room at GPKLMRCE.Concentrations of major and trace elements in four leaches were determined in the same way as the bulk geochemical analysis.The concentrations of each element in the four leachates were summed up and compared with the corresponding bulk data to check the recovery (within the range of 80-120%; see Table S1, Supplementary Materials).

Distribution of REY in Fe-Mn Crusts
The photographs of samples ST1, HYD66-2, and NASD44-1 and their corresponding images of elemental distributions determined by LA-ICP-MS are shown in Figure 4.In the images of elemental distributions, the concentration of each element is qualitatively represented by the counts per second (CPS).

Distribution of REY in Fe-Mn Crusts
The photographs of samples ST1, HYD66-2, and NASD44-1 and their corresponding images of elemental distributions determined by LA-ICP-MS are shown in Figure 4.In the images of elemental distributions, the concentration of each element is qualitatively represented by the counts pe second (CPS).
As shown in the photograph of ST1 (Figure 4), calcite and quartz were common in the Fe-Mn crusts.In the elemental images, distributions of Ca and Si matched well with calcite and quartz, respectively.Furthermore, the distribution of Mn was almost the same as the Fe distribution, indicating that the Mn minerals were intergrown with the amorphous FeOOH.The distributions of REY (e.g., La, Ce, Gd, and Y) were similar to each other.Sample ST1 showed a clear positive correlation between Fe, Mn, and REY, probably linked to the precipitation of Fe-vernadite or intergrowth of vernadite and Fe-oxyhydroxides instead of calcite and quartz.Nevertheless, the samples HYD66-2 and NASD44-1 showed an inverse correlation between Fe and Mn.In HYD66-2, Mn was clearly linked with Ca, probably indicating the todorokite phase.Fe was associated with REY and Si, indicating (1) the presence of aluminosilicates and Fe-oxyhydroxides and (2) that the REY mainly occurred in the aluminosilicates and/or Fe-oxyhydroxides.It is noticeable that the Ce distribution relatively differed from the other REY, showing Ce decoupling.In NASD44-1, REY showed some correlation with Mn, Fe, and Si.As shown in the photograph of ST1 (Figure 4), calcite and quartz were common in the Fe-Mn crusts.In the elemental images, distributions of Ca and Si matched well with calcite and quartz, respectively.Furthermore, the distribution of Mn was almost the same as the Fe distribution, indicating that the Mn minerals were intergrown with the amorphous FeOOH.The distributions of REY (e.g., La, Ce, Gd, and Y) were similar to each other.Sample ST1 showed a clear positive correlation between Fe, Mn, and REY, probably linked to the precipitation of Fe-vernadite or intergrowth of vernadite

Validity of LA-ICP-MS Mapping
LA-ICP-MS mapping is an "only slightly" destructive technique and was previously applied to Fe-Mn crusts and Fe-Mn nodules [14,35].The accuracy of the LA-ICP-MS is mainly influenced by a few critical factors.The first and most obvious shortcoming of the line scanning technique is the signal smearing.Both the accumulation of ablated material in linear ridges at the margins of raster trenches and the affection of washout characteristics of the sample cell of the laser machine contribute to the signal smearing, which is more pronounced for low-abundance refractory elements than major elements [36].Reducing the beam size or the scanning speed [36] or implementing pre-ablation [36] could effectively reduce the effect of accumulation.In addition, previous studies confirmed that the ablation direction of LA-ICP-MS analysis cannot affect the real element distribution patterns [37].Our experimental results showed that the distribution of the major elements (Ca, Si, Fe, and Mn) matched very well with the minerals (Figure 4), that is to say, the data arose simultaneously with laser ablation spots [31].Furthermore, there was no pronounced signal smearing in the distribution of the trace elements, and even the refractory elements (e.g., REY) compared with the major elements (Figure 4), and the correlation of the elements coincided with the sequential leaching experiment.
Secondly, the internal standard has a strong influence.Mg was used as the internal standard in the LA-ICP-MS quantitative mapping because it is more homogeneously distributed than other elements in the Fe-Mn oxides [35].However, it differs among the different minerals and is obviously lower in detrital minerals such as calcite and quartz.Thus, the Mg internal standard may cause too The residual aluminosilicates phase (L4) hosted between 1% and 46% (average 10%) of ΣREY.There were significant differences among the respective fractions in the Fe-Mn crusts.In samples ST1, ZSQD253A, and ZSQD42A, the residual aluminosilicates phase carried a low proportion (1%-10%, average 3%) of ΣREY, and the proportions of LREY, MREY, and HREY were 4%, 4%, and 2%, respectively.The PAAS-normalized REY patterns (Figure 6c, blue patterns) were flat with positive Ce anomalies (average Ce/Ce* = 1.85) and positive Eu anomalies (average Eu/Eu* = 1.30).In samples HYD66-2, NASD21-1, and MP4SD24-1, 2%-40% (average 17%) of ΣREY was leached in L4 and the average extraction proportions of LREY, MREY, and HREY were 31%, 22% and 3% in L4, respectively.The normalized REY patterns (Figure 6c, yellow patterns) decreased from the LREY over the MREY to the HREY.There were positive Ce anomalies (Ce/Ce* = 2.65) and negative Y anomalies (Y n /Ho n = 0.56).

Validity of LA-ICP-MS Mapping
LA-ICP-MS mapping is an "only slightly" destructive technique and was previously applied to Fe-Mn crusts and Fe-Mn nodules [14,35].The accuracy of the LA-ICP-MS is mainly influenced by a few critical factors.The first and most obvious shortcoming of the line scanning technique is the signal smearing.Both the accumulation of ablated material in linear ridges at the margins of raster trenches and the affection of washout characteristics of the sample cell of the laser machine contribute to the signal smearing, which is more pronounced for low-abundance refractory elements than major elements [36].Reducing the beam size or the scanning speed [36] or implementing pre-ablation [36] could effectively reduce the effect of accumulation.In addition, previous studies confirmed that the ablation direction of LA-ICP-MS analysis cannot affect the real element distribution patterns [37].Our experimental results showed that the distribution of the major elements (Ca, Si, Fe, and Mn) matched very well with the minerals (Figure 4), that is to say, the data arose simultaneously with laser ablation spots [31].Furthermore, there was no pronounced signal smearing in the distribution of the trace elements, and even the refractory elements (e.g., REY) compared with the major elements (Figure 4), and the correlation of the elements coincided with the sequential leaching experiment.
Secondly, the internal standard has a strong influence.Mg was used as the internal standard in the LA-ICP-MS quantitative mapping because it is more homogeneously distributed than other elements in the Fe-Mn oxides [35].However, it differs among the different minerals and is obviously lower in detrital minerals such as calcite and quartz.Thus, the Mg internal standard may cause too high corrections in the elemental concentrations when the crust contains a lot of detrital minerals.Here, we discarded the calibration of standards and used CPS to represent the relative level of content of each element directly.Eventually, the LA-ICP-MS mapping method could supply qualitative information of the elemental distribution in a straightforward way.Compared with the sequential leaching experiment, the LA-ICP-MS mapping technique is an in-situ, simpler, rapider, and highly reproducible technique regarding sample preparation, laboratory experiments, and data reduction [26,37].Furthermore, the mapping results mutually corroborated the results of the sequential leaching experiment.

Occurrence Phases of REY
The mineralogical analysis of all samples showed that the δ-MnO 2 and X-ray amorphous FeOOH were the dominant minerals, which is the typical mineral combination of hydrogenetic Fe-Mn crusts [10].Since these Fe-Mn crusts are typically hydrogenetic in origin, the todorokite in some samples is more likely to indicate that they experienced lower-oxidation-potential seawater conditions [17] rather than a diagenetic input.The results of the sequential leaching experiment showed that most of the REY was extracted in the Mn-oxide and amorphous FeOOH phases (Figure 5), suggesting that the REY mainly occurred as δ-MnO 2 and amorphous FeOOH in the Fe-Mn crusts.The same conclusion could be obtained from the LA-ICP-MS mapping results (Figure 5; Table S3, supplementary materials).
δ-MnO 2 is composed of one or two water layers within Mn-O-octahedral sheets [38].These sheets contain vacancies and abundant isomorphic substitution of Mn 4+ , both of which can induce charge deficits [38].The incorporation of hydrated interlayer cations can compensate for the negative surface charge of these sheets and, thus, the δ-MnO 2 is able to accommodate REY [38,39].In contrast, todorokite contains a large three-dimensional (3D) tunnel structure [40], in which the Ce(IV) and REY(III) are unable to be incorporated.Other factors can affect the crusts during early diagenesis in some microlayers at the base of crusts, precipitating new minerals (e.g., todorokite).These processes were observed in nodules and crusts from the Pacific Ocean and the Atlantic Ocean [8].Therefore, crystal structures of Mn minerals are another considerable factor in the proportion of Mn-bound REY.
Amorphous FeOOH has a large specific surface area and high chemical reactivity [41,42]; thus, it has a greater ability to adsorb heavy elements.Clay minerals also have surface charges and highly specific surfaces, and these characteristics are crucially important for the adsorption of REY [43].The WPO crusts contain phillipsite (Figure 3) and have a high proportion of aluminosilicate-bound REY, which implies the clay might constitute an additional REY source.Instead, crystalized quartz and felspar account for most of the aluminosilicate material in the SCS crusts.These minerals carry some of the REY (Figure 5) and lead to low concentrations of aluminosilicate-bound REY in the SCS crusts.

Enrichment Process of REY
As mentioned above, δ-MnO 2 and FeOOH are the most important occurrence phases of REY, that is to say, the process of REY participation in the δ-MnO 2 and FeOOH is suggested to be a controlling factor for REY enrichment in hydrogenetic Fe-Mn crusts from the SCS and WPO [44][45][46][47].To illustrate the REY distribution between δ-MnO 2 and FeOOH [24], we calculated Lu-normalized ratios of Fe-bound REY (REY-FeOx) to Mn-bound REY (REY-MnOx) for the members of the REY series.
The values of the (∆ REY-FeOx/REY-MnOx ) Lu ratios were close to the values of the ([REY(CO 3 ) 2 − ]/[REYCO 3 + + REY 3+ ]) Lu ratios (Figure 7).The similarity demonstrates that the preferential adsorption of REY carbonates by Mn minerals and Fe minerals controls the REY enrichment in the Fe-Mn crusts [24].Moreover, the [REY(CO 3 ) 2 − ]/[REYCO 3 + + REY 3+ ] Lu ratios increased along the REY series (Figure 7).This indicates that LREY preferentially occurs as LREYCO 3 + and LREY 3+ , while HREY prefers to form HREY(CO 3 ) 2 − in seawater [53,54].The zero-point charge (ZPC) of δ-MnO 2 is 2.3 [55,56], while that of amorphous FeOOH is 8.8 [57].Therefore, δ-MnO 2 has an excessive negative surface charge, while the surface of FeOOH may be electrically neutral or weakly electropositive in seawater.In this case, the δ-MnO 2 has a tendency to adsorb positively charged REY complexes (e.g., LREYCO 3 + and LREY 3+ ), whereas amorphous FeOOH prefers to adsorb negatively charged REY complexes such as HREY(CO 3 ) 2 − [48].This can explain the LREY enrichment in the Mn-oxide phase (Figure 6a) and HREY enrichment in the amorphous FeOOH phase (Figure 6b).The yttrium, unlike other HREY, prefers to form YCO 3 + in seawater [24,26,49], which explains why there were negative Y anomalies in the amorphous FeOOH phase (Figure 6b).The values of the (Δ REY-FeOx/REY-MnOx )Lu ratios were close to the values of the ([REY(CO3)2 -]/[REYCO3 + + REY 3+ ])Lu ratios (Figure 7).The similarity demonstrates that the preferential adsorption of REY carbonates by Mn minerals and Fe minerals controls the REY enrichment in the Fe-Mn crusts [24].Moreover, the [REY(CO3)2 -]/[REYCO3 + + REY 3+ ]Lu ratios increased along the REY series (Figure 7).This indicates that LREY preferentially occurs as LREYCO3 + and LREY 3+ , while HREY prefers to form HREY(CO3)2 -in seawater [53,54].The zero-point charge (ZPC) of δ-MnO2 is 2.3 [55,56], while that of amorphous FeOOH is 8.8 [57].Therefore, δ-MnO2 has an excessive negative surface charge, while the surface of FeOOH may be electrically neutral or weakly electropositive in seawater.In this case, the δ-MnO2 has a tendency to adsorb positively charged REY complexes (e.g., LREYCO3 + and LREY 3+ ), whereas amorphous FeOOH prefers to adsorb negatively charged REY complexes such as HREY(CO3)2 - [48].This can explain the LREY enrichment in the Mn-oxide phase (Figure 6a) and HREY enrichment in the amorphous FeOOH phase (Figure 6b).The yttrium, unlike other HREY, prefers to form YCO3 + in seawater [24,26,49], which explains why there were negative Y anomalies in the amorphous FeOOH phase (Figure 6b).Ce is a redox-sensitive element that is easily oxidized to Ce(IV) when the other REY elements remain strictly trivalent, which usually causes Ce decoupling [4].In hydrogenetic Fe-Mn crusts, Ce is tetravalent [25].The Ce oxidation absorption by δ-MnO2 is widespread because Mn ions are a strong oxidant in seawater [25,58].Ce oxidation absorption by FeOOH, however, tends to get overlooked because of the slow rate of Ce(IV) oxidation [44].In our experimental results, an average of 50% of Ce was leached in the FeOOH phase and the values of δCe (average 3.75) were even higher than the Ce is a redox-sensitive element that is easily oxidized to Ce(IV) when the other REY elements remain strictly trivalent, which usually causes Ce decoupling [4].In hydrogenetic Fe-Mn crusts, Ce is tetravalent [25].The Ce oxidation absorption by δ-MnO 2 is widespread because Mn ions are a strong oxidant in seawater [25,58].Ce oxidation absorption by FeOOH, however, tends to get overlooked because of the slow rate of Ce(IV) oxidation [44].In our experimental results, an average of 50% of Ce was leached in the FeOOH phase and the values of δCe (average 3.75) were even higher than the values of the Mn-oxide phase.This suggests that the Ce oxidation absorption by FeOOH is an essential part of the enrichment and fraction of Ce in the Fe-Mn crusts [13,24,59] and may be more important than Ce oxidation absorption by δ-MnO 2 in seawater.Moreover, this may be the reason why sample HYD66-2 (with high Fe content) contained a high Ce concentration.The SCS crusts (except sample HYD66-2) grow within the continental marginal environment and they are significantly affected by terrigenous materials [60].The eolian dust and the huge river systems (e.g., the Pearl, Red, and Mekong Rivers) transport a large amount of terrigenous detrital minerals into the SCS [61,62], and these detrital minerals are mainly quartz and felspar.Quartz and felspar are the last to be leached into the residual aluminosilicates phase (L4) due to their insolubility [27].In the SCS crusts (except sample HYD66-2), the PAAS-normalized REY patterns of the residual phase were relatively flat with minor positive Eu anomalies (Figure 6c).The felspars are responsible for the Eu anomalies because Eu 2+ is more strongly stabilized in the structure of the residual felspars than other trivalent lanthanides [63].The La/Al ratios (8.3 × 10 −4 ) and Lu/Al ratios (10.6 × 10 −6 ) of the residual aluminosilicate phase are close to the values of the upper crust (3.8 × 10 −4 and 3.8 × 10 −6 , respectively) [64].This indicates that the REY in the residual aluminosilicates phase can be explained by the presence of terrigenous aluminosilicates [65].The terrigenous aluminosilicates contain very low concentrations of REY (Figure 5) and eventually dilute the abundance of REY in the SCS crusts [66,67].
The sample HYD66-2 was obtained from the Central SCS Basin, which is much less affected by the continent than other Fe-Mn crusts from SCS.The mineralogy (Figure 3 and Section 3.1), geochemistry (Table 2 and Section 3.3), and REY distribution in major mineral phases showed that the Fe-Mn crusts in the Central SCS Basin had similar characteristics to the WPO crusts.Thus, crusts of HYD66-2 represent mixed signatures between the SCS and WPO Fe-Mn crust, considering Fe-Mn crusts that formed in the open oceans for the convenience of subsequent comparison.In these open-ocean crusts, a large part of the REY was leached into the residual fraction (L4).The Lu/Al ratios (17.0 × 10 −6 ) in L4 were close to the upper crust value, whereas the La/Al ratios (142 × 10 −4 ) were much higher than in the upper crust.This indicates the existence of excess LREY, which cannot be explained by terrigenous quartz and felspar.It was found at numerous sites in the Pacific Ocean that deep-sea mud contains high concentrations of REY [68]; however, in the SCS, the REY concentrations in the sediments are much lower [66].It can be inferred that, in the open oceans, the clay may constitute an additional REY source in the Fe-Mn crusts.In addition, oceanic phillipsite was marked by a positive Ce anomaly and the Ce concentration could reach 140 ppm [69]; thus, the phillipsite was a likely cause of the positive Ce anomalies in L4 (Figure 6c).Therefore, it can be suggested that REY adsorption by clay minerals may be a considerable factor in REY enrichment in the open-ocean Fe-Mn crusts [70,71].

Growth Rate
The SCS crusts (except sample HYD66-2) had a higher growth rate (Table 2) than the open-ocean crusts (including samples HYD66-2, NASD44-1, and MP4SD24-1).It is known that the SCS Fe-Mn crusts are subjected to large amounts of terrigenous inputs (see Section 4.4.1).The terrigenous materials eventually promote the growth rate of the SCS Fe-Mn crusts [11,17].Apart from the terrigenous inputs, the growth rate of the hydrogenetic Fe-Mn crusts is significantly influenced by the development of the oxygen minimum zone (OMZ) [12,72,73].Figure 8 shows the vertical profiles of dissolved oxygen in different sea areas and the water depth of the Fe-Mn crusts [74][75][76].The core of the OMZ in the SCS (e.g., Zhongsha Islands) is generally located at a depth of 600−800 m [74].The Central SCS Basin has a very wide OMZ at a depth of 600−1600 m, and the vertical change of dissolved O 2 is more complex because of the influence by the North Pacific Intermediate Water through the Philippine Sea [77].Obviously, sample HYD66-2 grows in the OMZ, while the other SCS crusts are below the OMZ (Figure 8).Generally, Fe-Mn crusts grow faster below the OMZ than in the OMZ [11,17].Hence, sample HYD66-2 had a lower growth rate than the other SCS crusts.In the WPO crusts, the situation was the same.The depth of OMZ in the WPO (including the Magellan seamount and Marcus-Wake seamount areas) extends to approximately 1500 m, and the sample MP4SD24-1 below the OMZ had a faster growth rate than sample NASD44-1 in the OMZ (Figure 8).The OMZ may influence the growth rate and then influence the adsorption/enrichment of metals [1,78].As shown in Figure 9 (see Table S2 for correlation coefficients, Supplementary Materials), the growth rate of the Fe-Mn crusts was negatively correlated with the concentrations of REY (R 2 = 0.7481), FeOOH-bound REY (L3, R 2 = 0.5099), and aluminosilicate-bound REY (L4, R 2 = 0.5948).Instead, there was no obvious correlation between growth rate and Mn-bound REY (L2, R 2 = 0.0037).This may suggest that the slow growth rate conduces more adsorption of REY from the surrounding seawater into the FeOOH and aluminosilicates rather than the δ-MnO2 [70,78].The SCS Fe-Mn crusts have fast growth rates, and this may be the reason why they contain lower concentrations of REY than the open-ocean crusts [1].[74].Data from the Central SCS Basin refer to Reference [75].Data for the Magellan seamounts and Marcus-Wake seamounts are from Reference [76].The blocks represent the Fe-Mn crusts formed at different water depths in different sea areas.ZSQD253A and ZSQD42A formed in the Zhongsha Islands, HYD66-2 grew on a seamount in the Central SCS Basin, NASD44-1 grew on Magellan seamounts, and MP4SD24A-1 grew on Marcus-Wake seamounts.
The OMZ may influence the growth rate and then influence the adsorption/enrichment of metals [1,78].As shown in Figure 9 (see Table S2 for correlation coefficients, Supplementary Materials), the growth rate of the Fe-Mn crusts was negatively correlated with the concentrations of REY (R 2 = 0.7481), FeOOH-bound REY (L3, R 2 = 0.5099), and aluminosilicate-bound REY (L4, R 2 = 0.5948).Instead, there was no obvious correlation between growth rate and Mn-bound REY (L2, R 2 = 0.0037).This may suggest that the slow growth rate conduces more adsorption of REY from the surrounding seawater into the FeOOH and aluminosilicates rather than the δ-MnO 2 [70,78].The SCS Fe-Mn crusts have fast growth rates, and this may be the reason why they contain lower concentrations of REY than the open-ocean crusts [1].
FeOOH-bound REY (L3, R 2 = 0.5099), and aluminosilicate-bound REY (L4, R 2 = 0.5948).Instead, there was no obvious correlation between growth rate and Mn-bound REY (L2, R 2 = 0.0037).This may suggest that the slow growth rate conduces more adsorption of REY from the surrounding seawater into the FeOOH and aluminosilicates rather than the δ-MnO2 [70,78].The SCS Fe-Mn crusts have fast growth rates, and this may be the reason why they contain lower concentrations of REY than the open-ocean crusts [1].

Conclusions
Based on our investigations into the mineralogy of the Fe-Mn crusts from the SCS and WPO and the partitioning behavior of REY among the major mineral phases, the following can be concluded:

1)
Mineral composition is a considerable factor in the REY enrichment.Primary occurrence phases for REY in the SCS crusts are δ-MnO 2 and amorphous FeOOH, both of which have advantageous crystal structures for REY enrichment.The WPO crusts have a large proportion of aluminosilicate-bound REY.We suggest that REY adsorption by clay minerals may play a role in the REY enrichment in the open-ocean Fe-Mn crusts.2) In the hydrogenetic Fe-Mn crusts from the SCS and WPO, most of the ΣREY is bound to Mn-oxide and amorphous FeOOH phases, with LREY and HREY preferentially adsorbed onto δ-MnO 2 and FeOOH, respectively.REY-carbonate complexes are the dominate REY species in seawater.Thus, the selective adsorption of major REY complex species from seawater by δ-MnO 2 and FeOOH is the predominant mechanism for REY enrichment in the hydrogenetic Fe-Mn crusts.
3) The SCS crusts grow in the continental margin environment and they are significantly affected by terrigenous materials.The terrigenous aluminosilicate minerals in the SCS crusts dilute the abundance of REY.In contrast, Fe-Mn crusts in open oceans are rarely affected by the continent, and they contain a higher concentration of REY. 4) The terrigenous inputs and the development of OMZ determine the growth rate of the hydrogenetic Fe-Mn crusts.A fast growth rate is unfavorable for the adsorption of REY from the surrounding seawater into the FeOOH and aluminosilicates.Therefore, the SCS crusts with a faster growth rate contain lower REY enrichment.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2075-163X/9/1/8/s1:Table S1: Concentrations of REY in the leaches and in bulk crusts investigated in this study; Table S2: Correlation coefficient matrix for the growth rate and REY in different mineral phases; Table S3: LA-ICP-MS mapping data (CPS) of three Fe-Mn crusts samples;

Figure 1 .
Figure 1.Locations of the Fe-Mn crust samples in the South China Sea (SCS) and West Pacific Ocean (WPO).

Figure 1 .
Figure 1.Locations of the Fe-Mn crust samples in the South China Sea (SCS) and West Pacific Ocean (WPO).

Figure 2 .
Figure 2. The flow chart of the sequential leaching procedure.

Figure 2 .
Figure 2. The flow chart of the sequential leaching procedure.

Figure 4 .
Figure 4. Microscope photograph and the images of elemental distributions in samples ST1, HYD66-2, and NASD44-1 determined by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS).The size of each image is 500 μm × 500 μm.In images of elemental distributions, red and blue colors represent the maximum and minimum determined values, respectively.

Figure 4 .
Figure 4. Microscope photograph and the images of elemental distributions in samples ST1, HYD66-2, and NASD44-1 determined by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS).The size of each image is 500 µm × 500 µm.In images of elemental distributions, red and blue colors represent the maximum and minimum determined values, respectively.

Figure 7 .
Figure 7.The pattern of Lu-normalized distribution of the REY between dominate speciation ([REY(CO3)2 -]/[REYCO3 + + REY 3+ ])Lu in seawater at pH 8.2 and the patterns of (Δ REY-FeOx/REY-MnOx )Lu values for hydrogenetic Fe-Mn crusts.The similarity demonstrates that the REY speciation in seawater can basically describe the distribution and fractionation of the REY between the Mn oxides and hydrous Fe oxides in Fe-Mn crusts.They further suggest that there are other factors in the fractionation of light REY (LREY) and middle REY (MREY) between the Mn oxides and hydrous Fe oxides.

Figure 7 .
Figure 7.The pattern of Lu-normalized distribution of the REY between dominate speciation ([REY(CO 3 ) 2 − ]/[REYCO 3 + + REY 3+ ]) Lu in seawater at pH 8.2 and the patterns of (∆ REY-FeOx/REY-MnOx ) Lu values for hydrogenetic Fe-Mn crusts.The similarity demonstrates that the REY speciation in seawater can basically describe the distribution and fractionation of the REY between the Mn oxides and hydrous Fe oxides in Fe-Mn crusts.They further suggest that there are other factors in the fractionation of light REY (LREY) and middle REY (MREY) between the Mn oxides and hydrous Fe oxides.

4. 4 .
Factors of the Difference in REY Enrichment in the SCS and WPO Crusts 4.4.1.Terrigenous Input

Minerals 2018, 8 , 20 Figure 8 .
Figure 8. Vertical profiles curve of dissolved oxygen in different sea areas.The curves of the Philippines Sea and Zhongsha Islands in the South China Sea are modified from Reference[74].Data from the Central SCS Basin refer to Reference[75].Data for the Magellan seamounts and Marcus-Wake seamounts are from Reference[76].The blocks represent the Fe-Mn crusts formed at different water depths in different sea areas.ZSQD253A and ZSQD42A formed in the Zhongsha Islands, HYD66-2 grew on a seamount in the Central SCS Basin, NASD44-1 grew on Magellan seamounts, and MP4SD24A-1 grew on Marcus-Wake seamounts.

Figure 8 .
Figure 8. Vertical profiles curve of dissolved oxygen in different sea areas.The curves of the Philippines Sea and Zhongsha Islands in the South China Sea are modified from Reference[74].Data from the Central SCS Basin refer to Reference[75].Data for the Magellan seamounts and Marcus-Wake seamounts are from Reference[76].The blocks represent the Fe-Mn crusts formed at different water depths in different sea areas.ZSQD253A and ZSQD42A formed in the Zhongsha Islands, HYD66-2 grew on a seamount in the Central SCS Basin, NASD44-1 grew on Magellan seamounts, and MP4SD24A-1 grew on Marcus-Wake seamounts.

Figure 9 .
Figure 9. Bivariate diagram of growth rate and REY in different mineral phases of the hydrogenetic Fe-Mn crusts.Data are from six crusts in this study and from two hydrogenetic crusts from the Central Indian Ocean Basin (CIOB) [13].

Figure 9 .
Figure 9. Bivariate diagram of growth rate and REY in different mineral phases of the hydrogenetic Fe-Mn crusts.Data are from six crusts in this study and from two hydrogenetic crusts from the Central Indian Ocean Basin (CIOB) [13].

Figure S1 :
Microscope photograph of sample ST1 and the elemental maps determined by LA-ICP-MS; Figure S2: Microscope photograph of sample HYD66-2 and the elemental maps determined by LA-ICP-MS; Figure S3: Microscope photograph of sample NASD44-1 and the elemental maps determined by LA-ICP-MS.Author Contributions: Y.R., X.S., and Y.G. conceived the experiments.Y.R. analyzed the data and wrote the paper.X.S. and Y.G. substantially revised the manuscript.Z.X. and Y.L. operated the XRD and LA-ICP-MS analysis, respectively.J.L. and Z.G. conducted the sequential leaching experiment.

Table 1 .
Sample descriptions and bulk mineralogy of Fe-Mn crusts in the South China Sea (SCS) and West Pacific Ocean (WPO).XRD-X-ray diffraction; E-east; N-north.

Table 1 .
Sample descriptions and bulk mineralogy of Fe-Mn crusts in the South China Sea (SCS) and West Pacific Ocean (WPO).XRD-X-ray diffraction; E-east; N-north.

Table 2 .
Bulk geochemical compositions and relative growth rates of the Fe-Mn crusts in the South China Sea and West Pacific Ocean.Major elements are in wt.% and trace elements are in parts per million (ppm).Co n indicates cobalt concentration normalized to Fe and Mn contents.