Particle geochemistry in the Rainbow hydrothermal plume, Mid-Atlantic Ridge1

Particle geochemistry in the Rainbow hydrothermal plume, Mid-Atlantic Ridge1

Geochimica et Cosmochimica Acta, Vol. 68, No. 4, pp. 759 –772, 2004 Copyright © 2004 Elsevier Ltd Printed in the USA. All rights reserved 0016-7037/04...

751KB Sizes 0 Downloads 8 Views

Geochimica et Cosmochimica Acta, Vol. 68, No. 4, pp. 759 –772, 2004 Copyright © 2004 Elsevier Ltd Printed in the USA. All rights reserved 0016-7037/04 $30.00 ⫹ .00



Particle geochemistry in the Rainbow hydrothermal plume, Mid-Atlantic Ridge HENRIETTA N. EDMONDS1,2,* and CHRISTOPHER R. GERMAN1 1


Southampton Oceanography Centre, European Way, Southampton S014 3ZH, UK The University of Texas at Austin, Marine Science Institute, 750 Channel View Dr., Port Aransas, TX 78373-5015 USA (Received August 2, 2002; accepted in revised form July 18, 2003)

Abstract—We report the analysis of 18 large volume (500 –1500 L) in situ filtered samples of particulate material from the largest hydrothermal plume on the Mid-Atlantic Ridge, overlying the ultramafic-hosted Rainbow hydrothermal field at 36° 14⬘N. Measured particulate iron concentrations reach 614 nM. High concentrations of particulate Fe oxyhydroxides result from the extremely high Fe concentration (⬃24 mM) and Fe/H2S ratio (⬃24) of the vent fluids, and persist to at least 10 km away from the vent site due to the advection of plume material with the ambient along-axis flow. Two of the nine pairs of pump deployments appear to have intercepted the buoyant or otherwise very young portion of the hydrothermal plume. These samples are characterized by anomalously (compared to neutrally buoyant plume samples) high concentrations of Mg, U, and chalcophile elements, and low concentrations of Mn, Ca, V, Y, and the rare earth elements (REE). Within the neutrally buoyant plume, elemental distributions are largely consistent with previously observed behaviors: preferential removal of chalcophile elements, conservative behavior of oxyanions (P, V, and U), and continuous scavenging of Y and the REE. This consistency is particularly significant in light of the underlying differences in fluid chemistry between Rainbow and other studied sites. Chalcophile elements are preferentially removed from the plume in the order Cd⬎Zn⬎Co⬎Cu. Phosphorus/iron and vanadium/iron ratios for the neutrally buoyant plume are consistent with global trends with respect to the concentration of dissolved phosphate in ambient seawater. Comparison of buoyant and neutrally buoyant plume ratios with data from hydrothermal sediments underlying the Rainbow plume (Cave et al., 2002) indicates, however, that while P/Fe ratios are indeed constant V/Fe ratios increase progressively from early stage plume particles to sediments. REE distributions in the buoyant and neutrally buoyant plume appear most consistent with a continuous scavenging process during dispersion through the water column. Copyright © 2004 Elsevier Ltd 2002). The underlying basement consists of ultramafic rocks, and the vent fluid compositions differ from those in basalthosted systems due in part to serpentinization of the host rocks at Rainbow. Key characteristics of the Rainbow fluids include high chlorinity (750 mM), low pH (⬃2.8), high methane, and extremely high Fe concentrations (24 mM), resulting in a Fe/H2S molar ratio of ⬃24 which is more than 10 times higher than at TAG (e.g., Charlou et al., 1997; Douville et al., 2002). Rare-earth element (REE) patterns of the Rainbow fluids exhibit particularly strong enrichments in the light REE and pronounced Eu anomalies compared to TAG (Douville et al., 2002). A study of the geochemistry of the Rainbow plume comparable to the previous comprehensive studies conducted at TAG (e.g., Trocine and Trefry, 1988; German et al., 1990; 1991a; 1991b) is therefore important to documenting and understanding potential differences in chemical behavior and fluxes resulting from ultramafic-hosted versus basalt-hosted hydrothermal systems. The dispersion of the Rainbow neutrally buoyant plume, as determined using hydrographic, nephelometer, dissolved tracer, and current profiler data, and verified by year-long mooring deployments, is controlled by interactions between a persistent along-axis flow and the local topography (German et al., 1998; Thurnherr and Richards, 2001; Thurnherr et al., 2002). The buoyant plume rises from the vent field to a depth of ⬃2000 –2100 m, whereupon the neutrally buoyant plume moves to the north, rounding Rainbow Ridge in a clockwise fashion and continuing up the rift valley along its southern and eastern sides (see Fig. 1). During our sampling cruise, the


Hydrothermal plumes play an important role in 1) modification of the chemical signal of crust-seawater interaction imparted to the oceans; 2) marine geochemical budgets (particularly for particle-reactive elements); 3) determining the chemical signatures recorded by metalliferous sediments; 4) exploration for high-temperature activity on the world’s ridge crests; and in some cases 5) deep ocean circulation (e.g., Baker et al., 1995). Hydrothermal plumes form as hot, buoyant vent fluid rises through the water column, entraining ambient seawater until the resulting mixture attains neutral buoyancy and disperses laterally. The Rainbow hydrothermal site at 36° 14⬘N on the Mid-Atlantic Ridge (MAR) was first located on the basis of the water-column signature of its neutrally buoyant plume (German et al., 1996). With nephelometry signals of up to 0.75 V and an estimated heat flux of 1 to 5 GW (Thurnherr and Richards, 2001), the Rainbow plume is comparable in size and strength to that overlying the TAG site (26°N MAR), but exhibits several key differences which make important a detailed examination of the geochemical processes it hosts. The Rainbow hydrothermal site consists of at least 10 high temperature (⬃365°C) black smokers in a field approximately 100 m by 250 m in area, situated at ⬃2300 m depth on the western flank of a non-volcanic ridge in an offset between the South AMAR (“American Mid-Atlantic Ridge”) and AMAR segments of the MAR (Fouquet et al., 1998; Douville et al., * Author to whom correspondence ([email protected]).



addressed 759


H. Edmonds and C. German

Fig. 1. Map of the Mid-Atlantic Ridge rift valley in the vicinity of Rainbow Ridge, indicating the locations of SAPs sampling stations occupied during the FLAME cruise in 1997 (numbers 1–10), as well as two stations occupied during the HEAT cruise in 1994 (squares) for which data have been presented by Ludford (1996). During SAP01, weather conditions resulted in the “towing” of the pumps to the northwest across the plume path for ⬃1.5 km along the line indicated. The white star marks the location of the active vent field, on the west flank of Rainbow Ridge. The contour interval is 200 meters, with color boundaries at 2000 and 2400 meters (darker greys indicate greater depth). Arrows indicate the circulation pathway of the Rainbow neutrally buoyant plume (German et al., 1998; Thurnherr and Richards, 2001).

Rainbow plume was traced to more than 50 km from its source. In contrast, the dispersion of the TAG plume at 26°10⬘N on the MAR takes place in a tidally-dominated flow regime (Rudnicki et al., 1994), and hydrothermal plume anomalies are restricted to within a few km of the source (Rudnicki and Elderfield, 1993). We present a comprehensive data set for the elemental geochemistry of particulate material within the Rainbow hydrothermal plume. Particles form in hydrothermal plumes by two major processes: initial precipitation of iron and other chalcophiles as sulfides in the buoyant plume, followed by oxidation of residual iron and the resulting formation of particulate iron oxyhydroxides (e.g., Rudnicki and Elderfield, 1993). Previous studies (which have largely focussed on the particles dispersing in neutrally buoyant plumes, because of the relative ease of sampling them compared to buoyant plume particles) have shown that elements typically exhibit one of three types of behavior in hydrothermal plume particles when viewed relative to Fe, the primary particle-forming element. Chalcophile elements and others which precipitate as sulfides are rapidly lost from the plume due to either preferential settling or oxidative dissolution; elements that exist in seawater primarily as oxyanions appear to coprecipitate with iron oxyhydroxides early in plume formation and exhibit constant element to iron ratios; and particle-reactive elements such as Be, Y, Th, and the REE exhibit positive curvature with respect to Fe, suggesting continuous uptake from seawater onto circulating and settling oxyhydroxide particles (e.g., Feely et al., 1991, 1998; German et al., 1990, 1991a, 1991b; Trocine and Trefry, 1988). In a more recent study of REE in a hydrothermal plume at 9°N on the East Pacific Rise, Sherrell et al. (1999) suggested that the curvature of REE-Fe plots at TAG may in fact be due

to an equilibrium rather than a kinetic effect, i.e., to draw-down of the local dissolved REE pool at high particulate Fe concentrations. Our data, in addition to providing an assessment of geochemical behaviors in a particularly large hydrothermal plume with heretofore unique chemical and physical oceanographic characteristics, also enable further evaluation of the

Table 1. Description of SAP samples collected in the Rainbow hydrothermal plume during the FLAME cruise. Sample ID

Latitude (N)

Longitude (W)

depth (m)

Volume (L)

SAP01_1 SAP01_3 SAP03_1 SAP03_3 SAP04_1* SAP04_3* SAP05_1 SAP05_3 SAP06_1 SAP06_3 SAP07_1 SAP07_3 SAP08_1 SAP08_3 SAP09_1* SAP09_3* SAP10_1 SAP10_3

36° 15–16⬘ 36° 15–16⬘ 36° 14.46⬘ 36° 14.46⬘ 36° 14.29⬘ 36° 14.29⬘ 36° 13.71⬘ 36° 13.71⬘ 36° 12.29⬘ 36° 12.29⬘ 36° 15.83⬘ 36° 15.83⬘ 36° 15.36⬘ 36° 15.36⬘ 36° 14.42⬘ 36° 14.42⬘ 36° 14.01⬘ 36° 14.01⬘

33° 53.0–53.6⬘ 33° 53.0–53.6⬘ 33° 55.04⬘ 33° 55.04⬘ 33° 54.07⬘ 33° 54.07⬘ 33° 49.81⬘ 33° 49.81⬘ 33° 46.42⬘ 33° 46.42⬘ 33° 52.83⬘ 33° 52.83⬘ 33° 43.06⬘ 33° 43.06⬘ 33° 53.92⬘ 33° 53.92⬘ 33° 54.55⬘ 33° 54.55⬘

2150 2100 2050 2025 2135 2110 2025 2000 1940 1915 2150 2125 2200 2175 2100 2075 2100 2075

920.9 1541.9 773.2 734.4 610.6 545.2 582.4 662.2 693.2 772.8 900.2 753.2 681.6 817.0 829.4 514.2 783.2 1196.0

Both filters deployed at station SAP02 tore during pumping, therefore no particulate data are available. In this and the following tables, the samples believed to have intercepted the buoyant plume are indicated with asterisks.

Table 2. Summary of particulate concentration data (except REE) for Rainbow SAP samples. Fe (nM)

SAP01_1 SAP01_3 SAP03_1 SAP03_3 SAP04_1* SAP04_1r* SAP04_3* SAP05_1 SAP05_3 SAP06_1 SAP06_3 SAP07_1 SAP07_3 SAP08_1 SAP08_3 SAP09_1* SAP09_3* SAP10_1 SAP10_3 typical error

11.0 10.7 9.3 8.1 69.2 233.5 278.8 420.7 26.4 35.3 18.0 8.3 22.4 20.9 128.4 614.1 5.7 7.9 ⬍5 nM

Mn (nM)

Al (nM)

Ca (nM)

Mg (nM)

0.185 0.160 0.157

2.3 0.9 0.8 1.2 0.9

66.8 67.9 63.3 47.2 77.1

12.5 9.6 26.6 10.9 23.0

0.216 0.184 0.199 0.144 0.185 0.216 0.190 0.211 0.181 0.045 0.181 0.182 0.388 ⬍0.02 nM

2.1 0.3 1.1 1.0 2.0 3.4 1.8 0.8 1.7 0.9 1.8 1.0 1.9 ⬍1 nM

108.5 83.6 106.8 51.0 62.1 72.2 71.5 70.6 63.7 38.6 120.6 58.7 78.3 ⬍5%

42.0 33.9 42.8 14.5 27.8 12.0 22.0 20.8 10.0 86.9 40.6 8.3 13.5 ⬍5%

Fe/ (Fe⫹Mn⫹Al)

P (nM)

0.91 0.85 0.99

1.0 1.1 1.0 0.7 7.0 23.5 30.0 42.8 3.2 4.9 2.1 0.8 2.5 2.4 13.0 59.3 1.3 1.2 10%

0.99 1.00 1.00 0.96 0.94 0.83 0.81 0.96 0.92 0.99 1.00 0.83 0.78

V (pM)

Co (pM)

Cu (pM)

Zn (pM)

Y (pM)

Ag (pM)

Cd (pM)

Pb (pM)

U (pM)

59 60 52 46

3.2 1.8 1.6 1.8 43.5

62 34 34 121

102 48 121 324

880 1389 1741 143 209 98 55 135 126 504 2084 40 44 5%

126.7 47.2 70.2 4.1 5.3 5.0 2.1 2.9 2.8 43.5 173.3 1.1 1.6 5%

4813 2386 3251 134 196 153 106 78 76 1781 7719 18 27 5%

2875 287 383 178 268

2.1 2.3 1.9 1.6 3.0 3.1 7.8 13.0 18.0 2.3 3.6 2.4 2.2 2.7 2.8 4.1 13.5

0.16 0.08 0.02 0.18 2.00 1.81 3.77 2.89 3.81 0.29 0.26 0.25 0.13 0.12 0.17 1.44 7.15

0.44 0.24 0.17 0.25 5.99 5.79 18.75 0.39 0.63 0.24 0.16 0.41 1.79 0.22 0.22 6.42 31.35

13.9 7.7 7.2 6.7 20.9 21.0 20.0 24.5 28.8 19.4 12.4 24.6 10.0 15.6 10.3 7.2 41.8

0.020 0.017 0.019 0.015 0.104 0.106 0.104 0.121 0.168 0.044 0.031 0.024 0.021 0.060 0.024 0.055 0.396

1.7 5%

0.08 7%

0.71 10%

8.1 5%

0.036 8%

451 194 751 3391 95 524 6%

Particle geochemistry in the Rainbow hydrothermal plume


Errors (2 s.d.) for Fe, Mn, Al, Ca, and Mg were estimated from the regression statistics for each sample with standard additions. Errors for remaining elements, analyzed by ICPMS, are conservative estimates based on standard reproducibility and regression statistics. Sample SAP04_1 was run twice in the 150⫻ dilution ICPMS run. SAP04_1 was inadvertently omitted from the 15⫻ run, and SAP10_1 from the 150⫻ run. Manganese concentrations at Station SAP01 are believed to have been contaminated from the Mn cartridges used for thorium sampling, due to a plumbing error.



H. Edmonds and C. German

Table 3. Particulate rare earth element concentrations, Ce anomalies, and Eu anomalies for Rainbow SAP samples. Typical analytical errors are 5% for La-Sm and 10% Eu-Lu. Sample

La (pM)

Ce (pM)

Pr (pM)

Nd (pM)

Sm (pM)

Eu (pM)

Gd (pM)

Tb (pM)

Dy (pM)

Ho (pM)

Er (pM)

Tm (pM)

Yb (pM)

Lu (pM)

Ce/ Ce*a

Eu/ Eu*b

SAP01_1 SAP01_3 SAP03_1 SAP03_3 SAP04_1* SAP04_1r* SAP04_3* SAP05_1 SAP05_3 SAP06_1 SAP06_3 SAP07_1 SAP07_3 SAP08_1 SAP08_3 SAP09_1* SAP09_3* SAP10_3

1.39 1.49 1.22 1.18 1.71 1.58 5.07 6.83 9.84 1.18 2.08 1.54 1.35 1.32 1.67 2.30 6.61 0.93

2.27 2.34 2.12 2.11 1.95 1.83 4.40 3.63 5.22 1.29 2.25 2.38 2.27 1.76 2.18 1.38 4.06 1.74

0.368 0.367 0.307 0.309 0.396 0.383 1.014 1.330 1.934 0.272 0.508 0.392 0.336 0.347 0.408 0.439 1.339 0.241

1.574 1.431 1.227 1.233 1.688 1.621 4.269 5.190 7.836 1.117 2.037 1.563 1.330 1.570 1.582 1.788 5.565 1.046

0.326 0.269 0.225 0.229 0.304 0.310 0.796 0.951 1.466 0.217 0.388 0.293 0.249 0.304 0.304 0.330 1.082 0.190

0.086 0.072 0.064 0.063 0.154 0.150 0.463 0.379 0.635 0.071 0.113 0.083 0.067 0.093 0.083 0.180 0.595 0.059

0.265 0.208 0.173 0.193 0.365 0.331 0.754 0.823 1.438 0.203 0.323 0.225 0.206 0.328 0.230 0.294 1.128 0.230

0.041 0.035 0.029 0.030 0.053 0.053 0.123 0.150 0.233 0.035 0.053 0.041 0.032 0.048 0.040 0.050 0.197 0.031

0.239 0.199 0.170 0.175 0.321 0.312 0.737 0.917 1.426 0.194 0.313 0.229 0.193 0.308 0.235 0.307 1.172 0.188

0.043 0.039 0.032 0.032 0.066 0.064 0.143 0.196 0.311 0.038 0.060 0.044 0.035 0.055 0.047 0.064 0.246 0.035

0.115 0.102 0.082 0.090 0.167 0.174 0.405 0.543 0.842 0.108 0.161 0.121 0.096 0.153 0.128 0.174 0.713 0.098

0.014 0.013 0.010 0.011 0.021 0.022 0.049 0.072 0.105 0.013 0.020 0.015 0.014 0.017 0.016 0.022 0.090 0.011

0.090 0.079 0.071 0.063 0.143 0.123 0.282 0.418 0.666 0.093 0.121 0.091 0.073 0.114 0.098 0.137 0.551 0.080

0.012 0.011 0.009 0.009 0.018 0.017 0.040 0.061 0.093 0.013 0.017 0.012 0.010 0.017 0.013 0.018 0.078 0.009

0.69 0.71 0.77 0.78 0.51 0.51 0.41 0.26 0.26 0.50 0.48 0.68 0.75 0.55 0.59 0.29 0.29 0.79

1.35 1.41 1.50 1.39 2.08 2.14 2.75 1.97 2.01 1.54 1.47 1.49 1.37 1.34 1.44 2.67 2.46 1.28

(a) Ce/Ce* was calculated as 3*(Ce/Ceshale)/(2*(La/Lashale) ⫹ (Nd/Ndshale)). Values less than 1 are defined as negative anomalies. (b) Eu/Eu* was calculated as 2*(Eu/Eushale)/((Sm/Smshale) ⫹ (Gd/Gdshale)).

two competing hypotheses regarding the behavior of REE (and other particle reactive species) in hydrothermal plumes. 2. SAMPLES AND METHODS

filter heads were brought to the van where they were connected to a vacuum pump to remove excess water. After removing the top of the filter head the filter was rinsed under vacuum with deionized distilled water. The filter was folded, double-bagged, and frozen until analysis.

2.1. Sample Collection

2.2. Elemental Analyses

Eighteen large-volume samples of suspended particulate material from neutrally buoyant plume height at the Rainbow site were collected by in situ filtration using Challenger Oceanic Stand-Alone Pumps (SAPs) during the FLuxes At AMAR Experiment (“FLAME”) cruise (RRS Discovery 228) in May–June, 1997. Sample sites and depths (Table 1, Fig. 1) were selected based on nephelometer and dissolved tracer data from CTD casts and from tow-yo deployments of BRIDGET, the BRIDGE deep-tow instrument (see, e.g., German et al., 1998). SAP deployments occurred between 30 hours and 14 days after collection of the data used to target them. While short-term tidal variability has been shown to influence plume characteristics in some locations (e.g., Rudnicki et al., 1994; Rudnicki and German, 2002), the consistent along-axis flow at Rainbow dominates tidal flow and results in relatively consistent plume features such as particle concentration (Thurnherr and Richards, 2001). In the largest observation of variability during the FLAME cruise, two CTD casts conducted 1 week apart near the sites of SAPs 1 and 7 revealed a strong plume in the first instance, and almost no signal the second time. However, in the absence of real-time nephelometry data during SAP sampling, the wealth of data from BRIDGET and CTD deployments provides excellent context for sampling and interpretation of particle chemistry. At each station, pumps were deployed in pairs at separations of 25 to 50 m, on a plastic coated wire suspended below the conducting cable used for hydrocasts. The pumps were loaded with 293 mm diameter 1 ␮m Nuclepore polycarbonate filters for particulate trace metal and radionuclide sampling. In addition, two 3.3-inch MnO2-impregnated filter cartridges were connected in series between the filter head and the flow meter, for sampling dissolved thorium isotopes (Buesseler et al., 1992). The volumes pumped ranged from 514 to 1500 L, and particle loadings on most filters were high, based on visual inspection of the samples. We do not have data for total suspended matter (i.e., by mass) because of the difficulties inherent in accurately and cleanly weighing the 293 mm diameter filters. All shipboard filter handling (loading and sampling) was done in a laminar flow bench in a clean lab van, and the filter heads were covered with plastic bags whenever they were outside the van. On recovery, the

For elemental analyses, filters were heated to reflux with concentrated nitric acid (doubly distilled in Teflon) in 30 mL Teflon PFA vials (Savillex Corp.) for ⬃72 h. This procedure results in complete digestion of hydrogenous phases and minimizes potential interference from detrital aluminosilicate material (German et al., 1991a). Two unused filters were handled in the same way as samples for use as procedural blanks. At the end of the digestion period, no residual particles were visible. The filter material was brittle and fragmented, but not dissolved. Aliquots were withdrawn for analysis from the digestion vessels with care taken to avoid including filter fragments. Digest aliquots were analyzed for a suite of 29 elements (Tables 2 and 3) by inductively coupled plasma atomic emission spectrometry (ICP-AES) and mass spectrometry (ICP-MS). Fe, Mn, Mg, Ca, and Al were analyzed on 10:1 dilutions of the leach solutions by ICP-AES with standard additions. Phosphorus was analyzed in a later run by ICP-AES using external standard calibrations. Yttrium, Ag, Cd, Pb, U, and the REE’s were analyzed on 15:1 dilutions, and V, Cu, and Zn on 150:1 dilutions, by quadrupole ICP-MS, using internal Ga, In, and Re standards, and external standard calibration. Co was also added as an internal standard to the first set of runs, but was found to exist in significant concentrations in the samples. Co was rerun on 50:1 dilutions using external standard calibrations. 3. RESULTS

3.1. General Characteristics of Plume Particles The Rainbow plume particles are dominated by iron oxyhydroxide material, with the Fe/(Fe⫹Mn⫹Al) ratio exceeding 0.9 for most of the samples, and not falling below 0.78 (Table 2). This ratio and that of Al/(Fe⫹Mn⫹Al) are used as an indication of hydrothermal versus detrital sediment inputs (e.g., Bostro¨ m et al., 1969). Sample SAP10_3 (Fe/(Fe⫹Mn⫹Al) ⫽ 0.78), which was collected slightly to the west of the vent field

Particle geochemistry in the Rainbow hydrothermal plume


Fig. 2. Particulate Mn, Ca, and Mg in Rainbow plume particles as a function of particulate Fe concentration. In these and the following figures, solid squares indicate Rainbow neutrally buoyant plume particles and open squares are postulated Rainbow buoyant plume samples (SAPs 4 and 9).

(Fig. 1), is the only sample in which particulate Mn concentrations exceed 0.22 nM and has the lowest Fe/Mn ratio (20) of all the FLAME samples. Even this sample, however, appears to be predominantly fresh hydrothermal material. Particulate aluminum concentrations in all samples were close to the detection limit for our method, resulting in large relative errors, but the concentrations are nearly all less than 3.3 nM, the general background level determined for Atlantic particles in the Sargasso Sea (Sherrell and Boyle, 1992). The Al and Mn data are also in good agreement with eight SAP samples collected at two stations (HEAT 1 and 2, Fig. 1) in this area during a previous survey cruise in August– September 1994 (Ludford, 1996). This predominance of Fe oxyhydroxide material and relatively low apparent contribution of resuspended or pelagic detrital material (as inferred from the low abundance of particulate Mn and Al), along with the range of geochemical measurements made, render the Rainbow dataset (including the HEAT samples) most directly comparable to that reported for the TAG site by German et al. (1990, 1991a, 1991b). Particulate iron concentrations (Table 2) exceeded 60 nM in six of the FLAME samples— both pumps at each of Stations 4, 5 and 9 —reaching as high as 614 nM. Stations 4 and 9 were located very close to the vent site, and thus might be expected to have among the highest Fe concentrations. Station 5, however, was located on the eastern flank of Rainbow Ridge, almost 10 km downstream of the active vents (see Fig. 1). It appears that relatively little Fe is lost to near-field fallout as sulfides compared to other plumes because of the high Fe/H2S ratio of the Rainbow vent fluids. The formation of a large amount of fine-grained Fe oxyhydroxides, and the rapid advection of plume particulate material in the ambient flow field (German et al., 1998; Thurnherr and Richards, 2001) results in high concentrations of suspended particulate material well away from the source. Indeed, large nephelometer anomalies were observed as far as 50 km away from the Rainbow vent site during the FLAME cruise (German et al., 1998; Thurnherr and Richards, 2001).

3.2. Possible Interception of the Buoyant Plume The elemental data (Tables 2 and 3) contain several indications that we sampled the buoyant or otherwise very young portion of the hydrothermal plume at Stations 4 and 9. Sample SAP09_1 in particular (128.4 nM particulate Fe) shows several marked differences from the other samples, including the lowest particulate Mn concentration (0.045 nM), lowest Ca (38.6 nM), and highest Mg (86.9 nM) measured (Table 2, Fig. 2). The coincident observations of low Ca and high Mg suggest a hydrothermal origin for these anomalies as these two elements generally behave in a complementary manner in hydrothermal systems. One possible explanation for the high Mg is that during early entrainment of seawater into buoyant hydrothermal fluids, temperatures are still high enough to precipitate seawater Mg in a phase which is rapidly removed or otherwise not normally captured in more evolved neutrally buoyant plume samples. Bischoff and Seyfried (1978) found that heating of seawater alone above 250°C resulted in the precipitation of a magnesium-hydroxide-sulfate solid, later identified in chimney deposits at 21°N EPR and named caminite (Haymon and Kastner, 1986). Bowers et al. (1985) also predicted through thermodynamic modeling that talc and saponite should precipitate during mixing of an EPR-like endmember fluid with seawater. Additional removal of seawater Mg in buoyant hydrothermal plumes would enhance the low dissolved Mg signature of advecting neutrally buoyant plumes reported by de Villiers and Nelson (1999). As discussed below, the FLAME neutrally buoyant plume data show elemental distributions generally consistent with previous studies (e.g., Feely et al., 1991, 1998; German et al., 1990, 1991a, 1991b; Trocine and Trefry, 1988). In conjunction with this, the young plume particles sampled at Stations 4 and 9 exhibit predictable deviations from the overall element/Fe trends. Specifically, the Station 4 and 9 samples have anomalously high concentrations of chalcophile elements, low concentrations of “scavenged elements” such as the REE, and congruent concentrations of oxyanion elements when compared to the neutrally buoyant plume


H. Edmonds and C. German

Fig. 3. Distribution of chalcophile elements in Rainbow plume particles. Solid squares indicate Rainbow neutrally buoyant plume particles; open squares are Rainbow SAPs 4 and 9.

samples. These samples are highlighted with the use of contrasting symbols in Figs. 2 through 8. 3.3. Elemental Distributions Particulate Ca concentrations increase with increasing Fe (Fig. 2), with the exception of the low concentration in sample SAP09_1 mentioned above. Calcium concentrations are higher than those measured at Rainbow during the HEAT cruise. As with previously observed variations in the TAG plume (German et al., 1991a; Trocine and Trefry, 1988), this difference most likely reflects seasonal variations in biogenic calcium carbonate production in the overlying surface ocean between the May–June (FLAME) and August–September (HEAT) cruises. The increase in particulate Ca with increasing particulate Fe that overlies this biogenic background has been attributed to scavenging onto Fe oxyhydroxides (Feely et al., 1994a) and to formation of a Ca-rich phase (Feely et al., 1992) as well as to solid solution between Fe oxyhydroxides and a Ca-Si-FePO4 phase or CaHFe(PO4)2 (Feely et al., 1994b; Lilley et al., 1995). Particulate Mg also generally increases with Fe (Fig. 2) as observed at TAG (German et al., 1991a; Trocine and Trefry, 1988). In addition to the higher temperature magnesium-rich phases discussed in section 3.2 above, it has been suggested

that Mg enrichments in neutrally buoyant plume particles result directly from scavenging onto Fe oxyhydroxides (Trocine and Trefry, 1988). Chalcophile elements (e.g., Cu, Co, Cd, and Zn) have been observed to exhibit negative curvature with respect to particulate Fe in neutrally buoyant plumes, and decreasing element/iron ratios with decreasing particulate Fe, indicative of preferential removal by settling and/or oxidative dissolution of sulfides (e.g., Trocine and Trefry, 1988; German et al., 1991a; Feely et al., 1992, 1994a, 1994b; Metz and Trefry, 1993). The influence of those samples which intercepted the buoyant plume dominates the plotted distributions of the chalcophile elements in the Rainbow SAP samples (Fig. 3). When the data from SAPs 4 and 9 are removed from the plots, the chalcophile distributions in the neutrally buoyant plume at Rainbow (Fig. 4) more closely resemble those observed at TAG (Trocine and Trefry, 1988; German et al., 1991a), illustrating the generally positive correlation of chalcophile and Fe concentrations. The lack of any neutrally buoyant plume samples at Rainbow with Fe concentrations between ⬃50 and 280 nM, as well as scatter in the Zn and Cd data at low Fe concentration (which may result from sediment resuspension or sample contamination), precludes observation of the expected negative curvature in the distributions.

Particle geochemistry in the Rainbow hydrothermal plume


Fig. 4. Chalcophile element distributions, as in Fig. 3 but with the removal of Rainbow SAPs 4 and 9, in order to facilitate comparison with TAG data (German et al., 1991a; crosses in this and following figures) and highlight the distributions in the dispersing neutrally buoyant plumes. Also shown are Cu data for the HEAT stations (small squares). Coincident high values of Zn and Cd in low-Fe samples may indicate sediment resuspension or sample contamination.

The high concentrations of the chalcophile elements in the four samples from Stations 4 and 9 (Fig. 3) are consistent with the interpretation that these samples represent the early stages of plume formation. Chalcophile elements precipitate rapidly from vent fluids as sulfides upon cooling and mixing with seawater (e.g., Feely et al., 1987; Mottl and McConachy, 1990). The subsequent settling and/or dissolution of buoyant plume particles containing chalcophile elements is apparent in the much lower concentrations observed in the two high-Fe neutrally buoyant plume samples at Station 5 compared to Stations 4 and 9 (Fig. 3). A comparison (Fig. 5) of element/iron ratios in the Rainbow vent fluids (Douville et al., 2002), young plume particles (Stations 4 and 9), and “downstream” neutrally buoyant plume particles (represented by Station 5) indicates that the relative reactivity of these four elements (i.e., extent of loss from the plume as sulfides) decreases in the order Cd⬎Zn⬎Co⬎Cu. This conclusion is based on the decrease in the element/Fe ratio: for cadmium, near-field plume ratios are ⬃45 times higher than the ratios at Station 5, while for copper the corresponding decrease is only a factor of ⬃2. Cave et al. (2002) reported hydrothermal (i.e. corrected for detrital and other components) Cu/Fe and Zn/Fe ratios for four sediment

cores collected along the path of the Rainbow plume, which are consistent with our ratios in overlying plume samples and also indicate that Zn is preferentially lost from the plume very soon after venting. This order of reactivity is generally consistent with several previous studies of chalcophile elements in buoyant and neutrally buoyant plume particles. Mottl and McConachy (1990) found that buoyant plume particles at 21°N EPR have higher element/Fe ratios than the endmember vent fluids, with the order of enrichment being Cu⬎Co⬎Cd⬎Zn. This order nearly exactly mirrors our removal order, suggesting that those sulfides which form most rapidly also persist the longest. Note that the order of enrichment in our SAP 4 and 9 samples cannot be directly compared to the results of Mottl and McConachy (1990) because their samples, which were collected by submersible within 22 m above the vents, reflect an even earlier stage of plume formation. In a study of the dissolution of black smoker particles, Feely et al. (1987) found that sphalerite dissolves more rapidly than chalcopyrite, and in a laboratory study of natural plume particles Metz and Trefry (1993) found that Cd is released from sulfide particles more rapidly and thoroughly than is Cu. Trocine and Trefry (1988) reported that in the TAG plume Cd/Fe and Zn/Fe ratios decrease more rapidly


H. Edmonds and C. German

Fig. 5. The variation of chalcophile/Fe ratios in Rainbow vent fluids (Douville et al., 2002), near-field/buoyant plume particles (SAPs 4 and 9), and far-field, high-Fe neutrally buoyant plume particles (SAP 5). All element/Fe ratios have been normalized to the vent fluid ratio, i.e. the vent fluid ratio is set equal to 1.

than Cu/Fe. The enrichment and apparent persistence of Cu in particular has been attributed to direct association of Cu with Fe oxyhydroxides in addition to the lower solubility of chalcopyrite compared to other sulfides (e.g., Feely et al., 1987; Trocine and Trefry, 1988). Particulate silver and lead concentrations (Fig. 6) are both positively correlated with iron in the Rainbow plume, and do not show the same strong association with buoyant plume

particles as the chalcophiles discussed above. Silver concentrations are lower, and more tightly defined with respect to iron, than in the TAG plume (Trocine and Trefry, 1988; German et al., 1991a), while lead concentrations appear to be similar or slightly lower in the Rainbow plume than in the TAG plume at a given Fe concentration (German et al., 1991a). These differences in Ag/Fe and Pb/Fe ratios in the two plumes appear to reflect to some degree the Ag/Fe and Pb/Fe ratios of the vent

Fig. 6. Particulate concentrations of silver and lead in the Rainbow plume (squares; open squares for SAPs 4 and 9), with TAG data (crosses).

Particle geochemistry in the Rainbow hydrothermal plume


Fig. 7. Particulate concentrations of phosphorus, vanadium and uranium in the Rainbow plume (squares; open squares for SAPs 4 and 9), with TAG (crosses) and HEAT (small squares) data where available. Solid lines indicate linear regressions for the FLAME neutrally buoyant plume samples. The P/Fe ratio of Rainbow plume particles determined by linear regression (forced through zero) of all the FLAME SAP data is 0.100 ⫾ 0.002 (r2 ⫽ 1.00), while the V/Fe ratio determined by forced regression of the neutrally buoyant plume samples (all except SAPs 4 and 9) is 0.0043 ⫾ 0.0002 (r2 ⫽ .99). The regression of U/Fe for the neutrally buoyant plume gives a slope of 3.5 ⫻ 10⫺7 mol/mol (r2 ⫽ .94).

fluids, which are ⬃5 and ⬃3.5 times higher at TAG than at Rainbow, respectively (Douville et al., 2002). The concentrations of P, V, and U—all of which exist as oxyanions in seawater—in the Rainbow plume particles correlate positively with particulate Fe (Fig. 7). Phosphorus and vanadium exhibit tight linear correlations with iron in the neutrally buoyant plume, although the young plume samples appear to have slightly lower V/Fe. The P and V data are discussed in further detail below. Uranium concentrations also increase nearly linearly with iron, but two of the buoyant plume samples (SAP04_1 and SAP09_3) have substantially higher concentrations than would be expected from this trend, suggesting enrichment of U in early plume precipitates. This enrichment cannot be derived from vent fluids (as was the case for the chalcophiles) because U is quantitatively removed from seawater during hydrothermal circulation (Chen et al., 1986). The behavior of U in hydrothermal precipitates, and specifically the U/Fe ratio, is more variable than those of the other

oxyanions due to redox transformations between U(IV) and the more soluble U(VI). Previous studies (e.g., German et al., 1993; Mills et al., 1993) have documented ten-fold higher U/Fe ratios in near-field sediments compared to neutrally buoyant plume particles at TAG, due to uptake of U from seawater as a result of the establishment of reducing conditions during the diagenesis of sulfide-rich sediments. Mills et al. (1994) have documented even higher localized, apparently temporary enrichments of U on sulfide grains in metalliferous sediments which they hypothesized were due to microbially mediated oxidation of sulfides. Our buoyant plume samples may have captured a similar enrichment of U in the buoyant plume, although the possible entrainment of some chimney material cannot be ruled out. At the same time, our results from the neutrally buoyant plume confirm the relatively low U/Fe ratios observed for plume oxyhydroxide material compared to hydrothermal sediments (German et al., 1991b). Yttrium and the REE’s exhibit positively curving relation-

Fig. 8. Particulate Y, Nd and Er concentrations in Rainbow plume particles (squares; open squares for SAPs 4 and 9), with HEAT (small squares) and TAG (crosses) data where available. Nd and Er were plotted as representative light and heavy REE, respectively.


H. Edmonds and C. German 4. DISCUSSION

4.1. Phosphorus and Vanadium Behavior

Fig. 9. Shale-normalized REE patterns in Rainbow plume particles. Open symbols indicate SAPs 4 and 9 (buoyant plume) and solid symbols indicate samples from the neutrally buoyant plume, with SAP 5 (the highest Fe concentrations in the dispersing plume) highlighted as ⫹ and ⫻.

ships with iron in the Rainbow plume (e.g., Fig. 8). These relationships are defined by both the FLAME neutrally buoyant plume samples and the HEAT samples and are in good agreement with observations in the TAG plume (German et al., 1990). Concentrations in the younger, potentially buoyant plume samples fall below these curves, consistent with their young age and the hypothesis that hydrothermal precipitates continue to adsorb REE from seawater, both as they disperse in the hydrothermal plume and after they settle to the seafloor (German et al., 1990; Olivarez and Owen, 1989). Shale-normalized REE patterns (Fig. 9) exhibit both negative Ce anomalies and positive Eu anomalies, indicating contributions of both hydrothermal and seawater sources of REE’s to the plume particles. The samples from SAP 4 and 9 have both stronger Eu anomalies and less pronounced Ce anomalies than the other samples (particularly obvious for SAP 5), consistent with their relative immaturity, and with a stronger influence of hydrothermally-derived REE. The REE patterns of the near-field samples also retain some of the strong enrichment in the light REE reported for the Rainbow vent fluids (Douville et al., 2002).

Phosphorus and vanadium are believed to be incorporated from seawater into hydrothermal iron oxyhydroxides during their formation in buoyant plumes (Trefry and Metz, 1989; Feely et al., 1991, 1992, 1994b, 1998; Ludford et al., 1996). A compilation of hydrothermal plume P/Fe and V/Fe ratios in the Atlantic and Pacific Oceans (Feely et al., 1998) has demonstrated that plume P/Fe ratios are positively correlated with the concentration of dissolved phosphate in ambient seawater, consistent with a simple coprecipitation of seawater phosphorus with Fe oxyhydroxides in the buoyant plume. Transmission electron microscopy of thin-sectioned plume particles has shown strong covariation of P and Fe, in support of coprecipitation rather than adsorption as the mechanism of phosphorus enrichment in plume oxyhydroxides (Feely et al., 1990). In contrast, plume V/Fe ratios decrease linearly with increasing dissolved phosphate, suggesting a competitive inhibition of V scavenging onto iron oxyhydroxides by the presence of phosphate (Feely et al., 1994a). The importance of these global relationships lies in the potential to use these ratios in oxic metalliferous sediments to reconstruct past variations in oceanic phosphate distributions (Feely et al., 1998), which are relevant to estimating paleo-deepwater circulation patterns. Our data from the neutrally buoyant plume at Rainbow are consistent with and extend these correlations to lower phosphate concentrations (Fig. 10). The consistency of the Rainbow V data is particularly important because the previously available data for this area (Ludford, 1996) indicated V/Fe ratios more than five times that predicted by Feely et al. (1998) which the latter authors attributed to a possible localized V enrichment. It is now clear that these earlier data were affected by an analytical problem, and that the V/Fe relationship does indeed hold true throughout the range of sampled hydrothermal plumes. The predictable behavior of P and V in the Rainbow plume, despite the underlying geochemical differences of the vent system compared to others studied, indicates that this aspect of hydrothermal plume geochemistry can be predicted on a global scale. This predictability is important to calculation of the role of hydrothermal precipitates in the global marine

Fig. 10. Global trends of plume P/Fe and V/Fe with seawater dissolved phosphate, after Feely et al. (1998) with the addition of the FLAME data (solid symbols). The Rainbow V/Fe ratio is calculated from neutrally buoyant plume samples only. Solid lines indicate regression lines through the data. For P/Fe, the regression equation is P/Fe ⫽ 0.0492[PO3⫺ 4 ] ⫹ 2 0.0476, r2 ⫽ 0.95, and for V/Fe the regression yields V/Fe ⫽ ⫺0.001[PO3⫺ 4 ] ⫹ 0.0056, r ⫽ 0.92.

Particle geochemistry in the Rainbow hydrothermal plume

geochemical budgets of these elements (Feely et al., 1990; Trefry and Metz, 1989; Wheat et al., 1996). Previous studies have argued that P/Fe and V/Fe ratios do not vary following particle formation and emplacement in the neutrally buoyant plume (e.g., Feely et al., 1990, 1991, 1994a, 1998; Metz and Trefry, 1993). The P/Fe ratio of Rainbow plume particles reported here— 0.100 ⫾ 0.002—is consistent throughout the plume and equal to that reported for the hydrothermal component of surface sediments underlying the Rainbow plume (0.08 – 0.10, Cave et al., 2002), confirming that this ratio is “locked in” to hydrothermal oxyhydroxides early in plume formation and invariant thereafter. Despite the agreement of the neutrally buoyant plume data for V with the global relationship of Feely et al. (1998), however, the data from Rainbow indicate that the V/Fe ratio of hydrothermal precipitates may not in fact be set as early in plume formation, nor remain constant throughout dispersal and settling. The V/Fe ratio determined for the younger samples from Stations 4 and 9 (0.0035 ⫾ 0.0002) is significantly lower than that for the neutrally buoyant plume samples (0.0043 ⫾ 0.0002). Cave et al. (2002) report still higher V/Fe ratios of ⬃0.007– 0.008 for the hydrothermal component of their surface sediments. Previous studies of V in buoyant plumes (Campbell, 1991; Feely et al., 1994b) have also shown increases in particulate V/Fe during buoyant plume rise. The study of Feely et al. (1994b) showed a concomitant increase of P/Fe in the buoyant plume. It is unclear whether these variations are due to progressively increasing P/Fe and V/Fe ratios on Fe oxyhydroxides in the buoyant plumes or to progressively lower dilution by Fe sulfides with presumably low P/Fe and V/Fe ratios. The Rainbow data are interesting in that they show no variation in P/Fe ratios in conjunction with the rise in V/Fe from young particles to the neutrally buoyant plume and into surface sediments. This argues against dilution by other particle phases, and suggests a decoupling of P and V behavior. This difference is consistent with the suggestion that the P/Fe relationship results from coprecipitation of P and Fe, perhaps as solid solution of CaHFe(PO4)2 with Fe oxyhydroxides (Lilley et al., 1995), while V may be scavenged onto particle surfaces. These observations have important implications regarding the potential for reconstructing past phosphate distributions from elemental ratios in oxic metalliferous sediments (e.g., Schaller et al., 2000), despite encouraging initial results (e.g., German et al., 1997; Feely et al., 1998). The Rainbow data suggest that V/Fe cannot be considered constant, and while P/Fe ratios are apparently constant for hydrothermal oxyhydroxides they are less easy to discern in sediment records because of significant and variable non-hydrothermal inputs of phosphorus to marine sediments. 4.2. Scavenging of Rare-Earth Elements Rare earth elements have been measured previously in neutrally buoyant plume particles at the TAG (German et al., 1990) and 9°N EPR (Sherrell et al., 1999) hydrothermal sites, and in the buoyant plume (both dissolved and particulate phases) at TAG and Snakepit on the MAR (Mitra et al., 1994). Additional data have been reported for Atlantic hydrothermal plumes by Ludford et al. (1996). German et al. (1990) first reported that the concentrations of the REE exhibited positive curvature with


respect to particulate iron in the TAG plume. They attributed this distribution to continuous uptake of REE from seawater onto iron oxyhydroxide particles as the plume disperses, following an initial coprecipitation of both vent- and seawaterderived REE with the newly formed particles. The behavior of REE in hydrothermal plume oxyhydroxides was formalized in the model of Rudnicki and Elderfield (1993), and further supported with the buoyant plume measurements of Mitra et al. (1994). Dissolved REE concentrations are drawn below those of ambient seawater in both the buoyant plume (Mitra et al., 1994) and in the mid-depth water column near ridge crests (Klinkhammer et al., 1983). Further uptake is evident in hydrothermal sediments, which show increased REE/Fe ratios and increasingly seawater-like distribution patterns when compared to hydrothermal plume particles (e.g., German et al., 1997; Cave, 2002), and increasing REE/Fe ratios with distance from the ridge crest (Ruhlin and Owen, 1986; Owen and Olivarez, 1988; Olivarez and Owen, 1989). A major implication of all of these studies is that hydrothermal activity serves as a net removal mechanism for the REE from seawater. Our Rainbow plume data are consistent with this previous body of work in showing a) increasing REE/Fe at decreasing Fe concentration in the neutrally buoyant plume, which presumably reflects increased dilution and greater age of the particles; b) higher REE/Fe ratios in neutrally buoyant plume particles than in the buoyant plume/near field particles (SAPs 4 and 9) at similar particulate Fe concentrations; c) an evolution in the REE patterns from a more vent fluid-influenced pattern (stronger Eu anomalies) to a more seawater-like pattern (stronger Ce anomalies) from the near field samples to the neutrally buoyant plume samples (Table 3, Fig. 9). Sherrell et al. (1999) have recently proposed a second explanation for the curvature observed in REE vs. Fe plots in the neutrally buoyant plume at TAG, based on their analysis of lower-Fe samples at 9°45⬘-N EPR which did not exhibit curvature. Although they affirm that continued REE adsorption must occur in surface sediments due to their higher REE/Fe ratios than suspended particles, they contend that on the time scale of plume advection, adsorption reflects a near-instantaneous equilibrium of hydrothermal precipitates with ambient seawater. The samples analyzed on the EPR reach maximum particulate Fe concentrations of 26 nM (Sherrell et al., 1999), and the authors demonstrate that their elemental distributions are dominated by mixing with resuspended local sediment material. Because such mixing of two particle populations cannot directly produce the curvature seen in REE vs. Fe plots at TAG (and now Rainbow), Sherrell et al. (1999) suggest that the curvature is due to a saturation effect, whereby at high particulate Fe concentrations the particles have scavenged all available REE’s from solution. According to this model, the low REE/Fe ratios at high Fe reflect a limit in the available REE, and as these particles subsequently mix into “fresh” background seawater they can adsorb additional REE, with the particulate/dissolved concentration ratio (Cp/Cd) determined by a constant distribution coefficient (Kd) and the concentration of suspended particulate material (SPM), as Cp/Cd ⫽ Kd ⴱ SPM. The Rainbow data do not directly address the differences between the continuous uptake and equilibrium draw-down scenarios in that they do not include dissolved REE measurements, nor measurements of SPM concentration, which to-


H. Edmonds and C. German

gether with the particulate REE concentrations would allow the determination of Kd. They do however provide important support of the continuous uptake model, in the differences in REE/Fe ratios and in Ce and Eu anomalies between the nearfield high-Fe samples (SAPs 4 and 9) and the far-field samples at SAP 5. The fact that the REE concentrations in the Rainbow neutrally buoyant plume are comparable to or slightly higher than those in the TAG plume at similar particulate Fe concentrations (see Fig. 7) is also relevant. Because the REE have nutrient-like depth profiles in seawater (Elderfield and Greaves, 1982), ambient seawater REE concentrations are lower at the shallower Rainbow site than at TAG and a draw-down effect, if operative, would be expected to be even more pronounced. Consideration of the Rainbow data in the light of two competing models to explain them necessitates reexamination of the merits of the two hypotheses. First, it is important to note that both models will result in additional uptake of seawater REE onto plume particles as they mix away from the vent site and thus imply changing REE/Fe ratios on the time-scale of plume advection. One difficulty with the discussion of Sherrell et al. (1999) is their contention that German et al. (1990) rely on the assumption of a continually varying Kd to explain continuous uptake. This assumption is not explicit in the earlier papers, and all that is required is that the kinetics of REE uptake are comparable to or longer than the time-scale of plume advection. In fact, laboratory studies of REE adsorption from seawater onto amorphous FeOOH particles indicate that adsorption continues for at least 1–2 weeks (Koeppenkastrop and De Carlo, 1992), which is comparable to the time-scales considered in neutrally buoyant plumes (Rudnicki and Elderfield, 1993). In addition, Kd’s are empirical parameters that are known to vary with the aqueous chemical environment, total suspended particulate material concentration, and evolution of the chemical characteristics of the particle surface (e.g., Li et al., 1984). All of these parameters are expected to vary from particle formation in the buoyant plume through dispersion in the neutrally buoyant plume, and therefore the assumption of a varying Kd, even though not explicit in the earlier TAG models, does not seem unreasonable. The most direct evidence for continuous uptake on the timescale of plume formation and advection comes from the study of Mitra et al. (1994), who measured dissolved and particulate REE in the buoyant plumes at the TAG and Snakepit sites on the MAR. Their samples represented mixing ratios of ⬃100 – 700 parts seawater to endmember hydrothermal fluid, with total REE concentrations still significantly in excess of ambient seawater but dissolved concentrations drawn down well below ambient levels. In these samples, particulate REE concentrations account for up to 99% of the total, particularly for the light REE. The particulate REE/Fe ratios are in good agreement with those predicted for fresh Fe oxyhydroxide precipitates using the plume mixing and chemistry model of Rudnicki and Elderfield (1993), and lower than those in the TAG neutrally buoyant plume samples of German et al. (1990). Further examination of the scavenging removal of REE in hydrothermal plumes will require similarly detailed measurements of paired dissolved and particulate samples in the high SPM portion of neutrally buoyant plumes.


The geochemistry of particulate material in the Rainbow hydrothermal plume, which is comparable in scope to the TAG plume but different in underlying vent fluid chemistry (ultramafic rather than basalt-hosted) and in the nature of its dispersion (advection rather than tidal), enables us to draw several conclusions both about this specific hydrothermal system and about the general nature of chemical processes in hydrothermal plumes. Particulate iron concentrations in the Rainbow plume are very high due to the elevated Fe/H2S ratio of the hydrothermal fluids. These particles are dispersed, and the high concentrations maintained, over large distances due the advection-dominated circulation of the Rainbow plume. Other aspects of the vent fluid chemistry that are reflected in the plume particles include the concentrations of silver and lead, and the LREE enrichment of the near-field plume samples. The behaviors of chalcophile, oxyanion, and “scavenged” elements in the Rainbow hydrothermal plume are consistent with those demonstrated previously in plumes overlying basalthosted systems. Thus, despite underlying differences between this and other sampled plumes, the geochemical processes active in all are similar. This observation is crucial to our ability to make global inferences about the influence of hydrothermal processes on marine geochemistry. The Rainbow data extend the global relationship between seawater dissolved phosphate concentrations and the P/Fe ratio of hydrothermally-derived sediments, and confirm the constancy of this ratio throughout particle formation, dispersal, and settling. On the other hand, sampling of near-field plume particles at Rainbow indicate that the V/Fe ratio of plume-derived iron oxyhydroxides increases through this sequence and thus that this ratio in oxic metalliferous sediments cannot be used to reconstruct past seawater composition. The buoyant/near-field plume data at Rainbow are also important in addressing two competing models for the behavior of rare earth elements in hydrothermal plumes. Combined with the neutrally buoyant plume samples, and particularly those with very high particulate Fe, the REE concentrations and distribution patterns of these samples support the hypothesis of continued scavenging of REE from seawater as hydrothermal plume particles form and disperse.

Acknowledgments—This work was funded by the EC as part of project “AMORES” (contract no. MAST3-CT95-0040). The FLAME cruise was funded by the BRIDGE programme of NERC (BRIDGE Grant 85). Edmonds’s participation was supported by the North Atlantic Treaty Organization under a grant awarded in 1996 (NSF-NATO Postdoctoral Fellowship). We thank the captain and crew of the RRS Discovery, and the RVS technicians, for their expert assistance. Initial sample digestions were conducted in Brad Moran’s laboratories at URI. Peter Statham allowed us to use his clean lab facilities at the SOC. Martin Palmer and Bob Nesbitt provided access to the ICPMS facilities. Darryl Green, Andy Milton, and Damien O’Brien provided additional assistance at sea and in the laboratory. We also acknowledge consultations with Meg Tivey, and the helpful comments of four anonymous reviewers and Associate Editor David Burdige on this manuscript. Associate editor: D. J. Burdige

Particle geochemistry in the Rainbow hydrothermal plume REFERENCES Baker E. T., German C. R., and Elderfield H. (1995) Hydrothermal plumes over spreading-center axes: Global distributions and geological inferences. In Seafloor Hydrothermal Systems: Physical, Chemical, Biological, and Geological Interactions (eds. S.E. Humphris, et al.), pp. 47–71. American Geophysical Union. Bischoff J. L. and Seyfried W. E. (1978) Hydrothermal chemistry of seawater from 25° to 350°C. Am. J. Sci. 278, 838 – 860. Bostro¨ m K., Peterson M. N. A., Joensuu O., and Fisher D. E. (1969) Aluminum-poor ferromanganoan sediments on active oceanic ridges. J. Geophys. Res. 74(12), 3261–3270. Bowers T. S., Von Damm K. L., and Edmond J. M. (1985) Chemical evolution of mid-ocean ridge hot springs. Geochim. Cosmochim. Acta 49, 2239 –2252. Buesseler K. O., Cochran J. K., Bacon M. P., Livingston H. D., Casso S. A., Hirschberg D., Hartmann M. C., and Fleer A. P. (1992) Determination of thorium isotopes in seawater by non-destructive and radiochemical procedures. Deep-Sea Res. 39(7/8), 1103–1114. Campbell A. C. (1991) Mineralogy and chemistry of marine particles by synchrotron x-ray spectroscopy, Mo¨ ssbauer spectroscopy, and plasma-mass spectrometry. In Marine Particles: Analysis and Characterization (eds. D.C. Hurd and D.W. Spencer). American Geophysical Union. Cave R. R. (2002) A geochemical study of hydrothermal signals in marine sediments: The Rainbow hydrothermal area, 36°N on the Mid-Atlantic Ridge. Ph.D. Thesis, University of Southampton. Cave R. R., German C. R., Thomson J., and Nesbitt R. W. (2002) Fluxes to sediments underlying the Rainbow hydrothermal plume at 36°14⬘N on the Mid-Atlantic Ridge. Geochim. Cosmochim. Acta 66 (11), 1905–1923. Charlou J. L., Donval J. P., Douville E., Knoery J., Fouquet Y., Bougault H., Jean-Baptiste P., Stievenard M., and German C. (1997) High methane flux between 15°N and the Azores Triple Junction, Mid-Atlantic Ridge. Hydrothermal and serpentinization processes. Eos, Trans. AGU 78, 831. Chen J. H., Wasserburg G. J., Von Damm K. L., and Edmond J. M. (1986) The U-Th-Pb systematics in hot springs on the East Pacific Rise at 21°N and Guaymas Basin. Geochim. Cosmochim. Acta 50, 2467–2479. de Villiers S. and Nelson B. K. (1999) Detection of low-temperature hydrothermal fluxes by seawater Mg and Ca anomalies. Science 285, 721–723. Douville E., Charlou J. L., Oelkers E. H., Bienvenue P., Colon C. F. J., Donval J. P., Fouquet Y., Prieur D., and Appriou P. (2002) The Rainbow vent fluids (36°14⬘N, MAR): the influence of ultramafic rocks and phase separation on trace metal content in Mid-Atlantic Ridge hydrothermal fluids. Chem. Geol. 184, 37– 48. Elderfield H. and Greaves M. J. (1982) The rare earth elements in seawater. Nature 296, 214 –219. Feely R. A., Gendron J. F., Baker E. T., and Lebon G. T. (1994a) Hydrothermal plumes along the East Pacific Rise, 8°40⬘ to 11°50⬘N: Particle distribution and composition. Earth Planet. Sci. Lett. 128, 19 –36. Feely R. A., Lewison M., Massoth G. J., Robert-Baldo G., Lavelle J. W., Byme R. H., Von Damm K. L., and Curl H. C., Jr. (1987) Composition and dissolution of black smoker particulates from active vents on the Juan de Fuca Ridge. J. Geophys. Res. 92 (B11), 11347–11363. Feely R. A., Massoth G. J., Baker E. T., Cowen J. P., Lamb M. F., and Krogslund K. A. (1990) The effect of hydrothermal processes on midwater phosphorus distributions in the northeast Pacific. Earth Planet. Sci. Lett. 96, 305–318. Feely R. A., Massoth G. J., Baker E. T., Lebon G. T., and Geiselman T. L. (1992) Tracking the dispersal of hydrothermal plumes from the Juan de Fuca Ridge using suspended matter compositions. J. Geophys. Res. 97(B3), 3457–3468. Feely R. A., Massoth G. J., Trefry J. H., Baker E. T., Paulson A. J., and Lebon G. T. (1994b) Composition and sedimentation of hydrothermal plume particles from North Cleft segment, Juan de Fuca Ridge. J. Geophys. Res. 99(B3), 4985–5006. Feely R. A., Trefry J. H., Lebon G. T., and German C. R. (1998) The relationship between P/Fe and V/Fe ratios in hydrothermal precipi-


tates and dissolved phosphate in seawater. Geophys. Res. Lett. 25 (13), 2253–2256. Feely R. A., Trefry J. H., Massoth G. J., and Metz S. (1991) A comparison of the scavenging of phosphorus and arsenic from seawater by hydrothermal iron oxyhydroxides in the Atlantic and Pacific Oceans. Deep-Sea Res. 38(6), 617– 623. Fouquet Y., Barriga F., Charlou J. L., Elderfield H., German C. R., Ondre´ as H., Parson L., Radford-Knoery J., Relvas J., Ribeiro A., Schultz A., Apprioual R., Cambon P., Costa I., Donval J. P., Douville E., Landure´ J. Y., Normand A., Pelle´ H., Ponsevera E., Riches S., Santana H., and Stephen M. (1998) FLORES diving cruise with the Nautile near the Azores—First dives on the Rainbow field: hydrothermal seawater/mantle interaction. InterRidge News 7, 24 – 28. German C. R., Bourle`s D. I., Brown E. T., Hergt J., Colley S., Higgs N. C., Ludford E. M., Nelsen T. A., Feely R. A., Raisbeck G., and Yiou F. (1997) Hydrothermal scavenging on the Juan de Fuca Ridge: 230 Thxs, 10Be, and REEs in ridge-flank sediments. Geochim. Cosmochim. Acta 61(19), 4067– 4078. German C. R., Campbell A. C., and Edmond J. M. (1991a) Hydrothermal scavenging at the Mid-Atlantic Ridge: Modification of trace element dissolved fluxes. Earth Planet. Sci. Lett. 107, 101–114. German C. R., Fleer A. P., Bacon M. P., and Edmond J. M. (1991b) Hydrothermal scavenging at the Mid-Atlantic Ridge: Radionuclide distributions. Earth Planet. Sci. Lett. 105, 170 –181. German C. R., Higgs N. C., Thomson J., Mills R., Elderfield H., Blusztajn J., Fleer A. P., and Bacon M. P. (1993) A geochemical study of metalliferous sediment from the TAG hydrothermal mound, 26°08⬘N, Mid-Atlantic Ridge. J. Geophys. Res. 98(B6), 9683–9692. German C. R., Klinkhammer G. P., Edmond J. M., Mitra A., and Elderfield H. (1990) Hydrothermal scavenging of rare-earth elements in the ocean. Nature 345, 516 –518. German C. R., Klinkhammer G. P., and Rudnicki M. D. (1996) The Rainbow hydrothermal plume, 36°15⬘N, MAR. Geophys. Res. Lett. 23(21), 2979 –2982. German C. R., Richards K. J., Rudnicki M. D., Lam M. M., Charlou J. L., and and the FLAME Scientific Party. (1998) Topographic control of a dispersing hydrothermal plume. Earth Planet. Sci. Lett. 156, 267–273. Haymon R. M. and Kastner M. (1986) Caminite: a new magnesiumhydroxide-sulfate-hydrate mineral found in a submarine hydrothermal deposit, East Pacific Rise, 21°N. Am. Mineral. 71, 819 – 825. Klinkhammer G. P., Elderfield H., and Hudson A. (1983) Rare earth elements in seawater near hydrothermal vents. Nature 305, 185–188. Koeppenkastrop D. and De Carlo E. H. (1992) Sorption of rare-earth elements from seawater onto synthetic mineral particles: an experimental approach. Chem. Geol. 95, 251–263. Li Y.-H., Burkhardt L., Buchholtz M., O’Hara P., and Santschi P. H. (1984) Partition of radiotracers between suspended particles and seawater. Geochim. Cosmochim. Acta 48, 2011–2019. Lilley M. D., Feely R. A., and Trefry J. H. (1995) Chemical and biological transformations in hydrothermal plumes. In Seafloor Hydrothermal Systems: Physical, Chemical, Biological, and Geological Interactions (eds. S. E. Humphris, et al.), pp. 369 –391. American Geophysical Union. Ludford E. M. (1996) The Geochemistry of Submarine Hydrothermal Fluids and Particles. Ph.D. Thesis, University of Bristol. Ludford E. M., Palmer M. R., German C. R., and Klinkhammer G. P. (1996) The geochemistry of Atlantic hydrothermal particles. Geophys. Res. Lett. 23(23), 3503–3506. Metz S. and Trefry J. H. (1993) Field and laboratory studies of metal uptake and release by hydrothermal precipitates. J. Geophys. Res. 98 (B6), 9661–9666. Mills R. A., Elderfield H., and Thomson J. (1993) A dual origin for the hydrothermal component in a metalliferous sediment core from the Mid-Atlantic Ridge. J. Geophys. Res. 98(B6), 9671–9678. Mills R. A., Thomson J., Elderfield H., Hinton R. W., and Hyslop E. (1994) Uranium enrichment in metalliferous sediments from the Mid-Atlantic Ridge. Earth Planet. Sci. Lett. 124, 35– 47. Mitra A., Elderfield H., and Greaves M. J. (1994) Rare earth elements in submarine hydrothermal fluids and plumes from the Mid-Atlantic Ridge. Marine Chem. 46, 217–235.


H. Edmonds and C. German

Mottl M. J. and McConachy T. F. (1990) Chemical processes in buoyant hydrothermal plumes on the East Pacific Rise near 21°N. Geochim. Cosmochim. Acta 54, 1911–1927. Olivarez A. M. and Owen R. M. (1989) REE/Fe variations in hydrothermal sediments: Implications for the REE content of seawater. Geochim. Cosmochim. Acta 53, 757–762. Owen R. M. and Olivarez A. M. (1988) Geochemistry of rare earth elements in Pacific hydrothermal sediments. Marine Chem. 25, 183– 196. Rudnicki M. D. and Elderfield H. (1993) A chemical model of the buoyant and neutrally buoyant plume above the TAG vent field, 26 degrees N, Mid-Atlantic Ridge. Geochim. Cosmochim. Acta 57, 2939 –2957. Rudnicki M. D. and German C. R. (2002) Temporal variability of the hydrothermal plume above the Kairei vent field, 25°S, Central Indian Ridge. Geochem. Geophys. Geosyst. 3(2), 10.1029/2001GC000240. Rudnicki M. D., James R. H., and Elderfield H. (1994) Near-field variability of the TAG non-buoyant plume, 26° N, Mid-Atlantic Ridge. Earth Planet. Sci. Lett. 127, 1–10. Ruhlin D. E. and Owen R. M. (1986) The rare earth element geochemistry of hydrothermal sediments from the East Pacific Rise: examination of a seawater scavenging mechanism. Geochim. Cosmochim. Acta 50, 393– 400. Schaller T., Morford J., Emerson S. R., and Feely R. A. (2000) Oxyanions in metalliferous sediments: Tracers for paleoseawater

metal concentrations? Geochim. Cosmochim. Acta 64(13), 2243– 2254. Sherrell R. M. and Boyle E. A. (1992) The trace metal composition of suspended particles in the oceanic water column near Bermuda. Earth Planet. Sci. Lett. 111, 155–174. Sherrell R. M., Field M. P., and Ravizza G. (1999) Uptake and fractionation of rare earth elements on hydrothermal plume particles at 9°45⬘ N, East Pacific Rise. Geochim. Cosmochim. Acta 63(11/12), 1709 –1722. Thurnherr A. M. and Richards K. J. (2001) Hydrographic and hightemperature heat flux of the Rainbow hydrothermal site (36°14⬘N, Mid-Atlantic Ridge). J. Geophys. Res. 106(C5), 9411–9426. Thurnherr A. M., Richards K. J., German C. R., Lane-Serff G. F., and Speer K. G. (2002) Flow and mixing in the rift valley of the Mid-Atlantic Ridge. J. Phys. Oceanog. 32, 1763–1778. Trefry J. H. and Metz S. (1989) Role of hydrothermal precipitates in the geochemical cycling of vanadium. Nature 342, 531–533. Trocine R. P. and Trefry J. H. (1988) Distribution and chemistry of suspended particles from an active hydrothermal vent site on the Mid-Atlantic Ridge at 26°N. Earth Planet. Sci. Lett. 88, 1–15. Wheat C. G., Feely R. A., and Mottl M. J. (1996) Phosphate removal by oceanic hydrothermal processes: An update of the phosphorus budget in the oceans. Geochim. Cosmochim. Acta 60(19), 3593– 3608.