Genesis of ferromanganese crusts from the TAG hydrothermal field

Ž.Chemical Geology 1762001283–293 r locate r chemgeoGenesis of ferromanganese crusts from the TAGhydrothermal fieldRachel ls ),Deborah M.Wells,Stephen RobertsSchool of Ocean and Earth Science,Southampton Oceanography Centre,Uni Õersity of Southampton,Southampton,SO143ZH,UKReceived 9March 2000;accepted 27September 2000AbstractŽ.The rare earth element REE geochemistry and Nd isotopic composition of marine ferromanganese crusts allow insight into their environment of deposition.We present REE along with major and trace element data from a suite of samples collected near an active hydrothermal site from the Mid-Atlantic Ridge and demonstrate extreme REE variability over short spatial scales.Both positive and negative Ce anomalies are observed,along with varied fractionation of the REEs.Some of the Fe–Mn crusts are formed directly from low-temperature hydrothermal fluids,channelled over 3km from any high-temperature vent site.A significant hydrothermal component is observed in one Fe–Mn crust sampled.Other samples,collected from the same area,are predominantly hydrogenous and form at a wide range of accumulation rates depending on the hydrothermal plume input.The identification of extreme heterogeneity in source and depositional environment within Ž.one hydrothermal field suggests that 1this variability can be resolved both spatially and temporally within a fossil Ž.hydrothermal deposit and 2the integrated record provided by bulk analysis should be interpreted with caution.q 2001Elsevier Science B.V.All rights reserved.Keywords:Rare earth elements;Neodymium isotopes;Ferromanganese crusts;Trans-Atlantic geotraverse;Low-temperature hydrothermal fluids1.IntroductionIron–manganese oxyhydroxide encrustations are found in many oceanic settings.Growth rates of these deposits vary by many orders of magnitude depending on type;from 0.5mm Ma y 1for hydroge-nous crusts to 105mm Ma y 1for hydrothermal Ž.deposits Hein et al.,1997.The Fe r Mn ratio and trace element content of such crusts also varies)Corresponding author.Tel:q 44-23-8059-2678;fax:q 44-23-8059-3052.Ž.E-mail address:ram1@ ls .widely.The scope for reconstructing past oceanic conditions from the trace element and isotopic com-position of hydrogenous deposits was recognized in Ž.the 1970s O’Nions et al.,1979and has recently Žbeen widely exploited e.g.von Blanckenburg et al.,.1996;Burton et al.,1997;Christensen et al.,1997.Interpretation of isotopic signatures from Mn crusts depends on assumptions of how these crusts reflect Žthe local geochemical environment Chabaux et al.,.1997.The preferred model for Fe–Mn crust formation involves precipitation of colloidal phases from sea water followed by extensive scavenging of dissolved0009-2541r 01r $-see front matter q 2001Elsevier Science B.V.All rights reserved.Ž.PII:S 0009-25410000404-6()ls et al.r Chemical Geology 1762001283–293284trace metals into the mineral phases during formation Ž.Koschinsky and Halbach,1995.The main differ-ence between hydrogenous and hydrothermally de-rived Fe–Mn deposits is the rate of growth and Ž.associated scavenging processes Kuhn et al.,1998.The trace element content of hydrothermal Fe–Mn deposits reflects local hydrothermal inputs and the complex particle scavenging reactions that occur in the near vicinity of vents as well as the ubiquitous Ž.sea water scavenging reactions Hein et al.,1997.In general,hydrothermal Fe–Mn deposits have lower trace metal contents and more varied Fe r Mn ratios Žcompared with their hydrogenous counterparts Hein .et al.,1997.We have studied Fe–Mn deposits from an active hydrothermal site on the Mid-Atlantic Ridge.Early Ž.work at the Trans-Atlantic Geotraverse TAG hy-drothermal field at 268N was carried out on Fe–Mn deposits demonstrating the first evidence of hy-Ždrothermal activity in the Atlantic Scott et al.,1974,1978,1979;Shearme et al.,1983;Rona et al.,1984;.Thompson et al.,1985.It was not until 1986that high-temperature hydrothermal venting was observed Ž.directly at this site Rona et al.,1986.Fig.1is a map of the TAG hydrothermal field,showing the sampling areas relative to the various sulphide de-posits in the area.Although Fe–Mn deposits represent areally exten-sive expressions of hydrothermal activity at mid-oc-ean ridges,they account for a small proportion of the annual hydrothermal flux of Mn to the oceans Ž.Schultz and Elderfield,1996.The distribution of Mn in the oceans reflects its sources and the resi-dence time of Mn in the oceans is inferred to be Ž.extremely short ;50years .Therefore,there must be significant Mn scavenging away from sites of active venting to remove this hydrothermal input.Many of the processes studied at ridge axes must also be important in these distal areas of Mn accu-mulation.Ž.Fig.1.Map of the TAG vent field adapted from Karson and Rona 1990.The active and inactive hydrothermal mounds are shown.Sampling sites are indicated by dive or dredge site number.Sites 1243–1247are Alvin dive areas during which samples were collected Ž.Thompson et al.,1985.Sites 3386and 3387are BRAVEX dredge areas during which samples were collected.Site 3423is a Mir dive site,Fe–Mn crusts and a sediment sample were collected.()ls et al.r Chemical Geology1762001283–2932852.Rare earth element geochemistryWhile Fe–Mn deposits around hydrothermal ventsare not significant sinks of crustally injected Mn, they do play an important role in scavenging of trace metals,especially the REEs from sea water.The REEs are present in sea water at10y10molar levels, and are enriched106-fold in hydrothermal vent fluids Ž.Klinkhammer et al.,1994.Sea water exhibits chon-Ždrite normalised REE patterns negative Ce anomaly,Ž..heavy REE HREE enrichment that arise from the processes controlling their complexation and redox Žchemistry Turner and Whitfield,1979;Cantrell and .Byrne,1987.The extremely low REE content of sea water arises from the short residence time of these elements in the ocean and the limited sources to the system.The anomalously low Ce content of sea water arises from its autocatalytic oxidative removal ashighly insoluble CeO.This process can lead to2Ž.further fractionation of Ce IV from the trivalent REEs in the solid phase,positive Ce anomalies areŽoften observed in hydrogenous deposits e.g.Fleet,.1983;Hein et al.,1997;Bau,1999.Negative Ce anomalies are sometimes observed in Fe–Mn de-posits indicating that they have accumulated moreŽrapidly,incorporating less Ce e.g.Kuhn et al., .1998.The sign of the Ce anomaly in Fe–Mn de-posits can be used to interpret the rate of crustal growth and the palaeoenvironmental conditions Ž.Elderfield,1988.End-member hydrothermal fluids exhibit lightŽ.REE LREE enrichment;the relative distribution of REEs is inferred to be controlled by equilibrium withŽplagioclase phases at depth Klinkhammer et al., .1994.There is no Ce anomaly as this element is trivalent in vent fluids under reducing conditions.A significant positive Eu anomaly is inherited from plagioclase reaction and readily transported as a divalent chloride complex in the high temperature,Ž. reducing conditions present Sverjensky,1984. Rapid scavenging of REEs from solution during mixing of hydrothermal fluids with sea water de-pletes the mixture to sub-sea water levels near to theŽ. hydrothermal source Klinkhammer et al.,1983. Significant hydrothermal signatures are only ob-served in or close to active hydrothermal mounds Ž.ls and Elderfield,1995where details of the REE patterns can be used to infer mixing and reac-tion.Neodymium isotopes can be used to identifyŽ.sources of Nd and therefore REEs in addition to mixing,transport and depositional processes Ž.Piepgras and Wasserburg,1987.Present day NorthŽ.Atlantic Deep Water NADW that bathes the TAGŽ.area has a well defined´0value of y13.5"0.5NdŽ.Piepgras and Wasserburg,1987.The Nd isotopic composition of TAG vent fluids is close to that of ŽŽ..wMORB´0sq11.9T.Masuzerawa,personal Ndxcommunication,1995.Deposition of mixtures of basalt-derived fluid and sea water will be manifest in the Nd isotopic composition of the precipitate.Here we present REE and Nd isotope data for Fe–Mn deposits collected across the TAG hydrother-mal field and use these data to constrain the pro-cesses leading to Fe–Mn crust formation within the TAG field.3.Sampling and methodologySample locations are shown in Fig.1.Metallifer-ous oxide crusts were collected by DSRV Alvin in 1982during dives1243,1244and1247.The miner-alogy and geochemistry of these samples has beenŽpreviously described in detail Thompson et al.,1985;.Ravizza et al.,1996;Table1.Basalt and indurated carbonate sediment with mm-thick Mn and Fe-rich oxide coatings were recovered from the same area of the eastern valley wall by dredge hauls during the British–Russian Atlantic Vents Expedition Ž.Ž. BRAVEX in1994sites3386and3387;Fig.1. Siliceous samples and basalt with mm-thick Mn and Fe-oxide coatings were also collected from the inac-tive Alvin zone during the BRAVEX submersible Ž.programme site3423;Fig.1.The geochemistry of these samples has been described by Bogdanov et al.Ž.1995and is reproduced in Table1.A sediment sample was also collected from site3423by the Mir submersible in an area of carbonate ooze with Fe–Mn staining.Samples of Mn crust were hand picked.The-2 m m fraction of the sediment sample3423was sepa-Žrated and purified by standard techniques Jackson, .1969with minor modification.The samples wereŽ. characterised using X-ray diffraction XRD and()ls et al.r Chemical Geology1762001283–293286Table1Mineralogy and major element geochemistry for selected Fe–Mn crustsŽ.Ž.Ž.Ž.Ž.Ž.Ž. Sample Mineralogy Fe%Mn%Fe r Mn atomic Co ppm Ni ppm Cu ppm Zn ppm a1243-1Nontronite32.350.6747.4––––a1244-1B Birnessite-0.152.30-0.002––––a1244-1C Birnessitea1247-2-2D Amorphous41.52 2.4316.8––––a1247-2-2B Amorphousb3387-R1X-ray amorphous oxides14.510.78 1.32172018526701100 b3423-R-33X-ray amorphous oxides 6.9 5.0 1.3623068212002200 with minor carbonate3423-nontronite Nontronite40.00.013930––––Mean Pacific d-MnO19260.728110492611877732chydrogenous crustsValu Fa ridge Todorokite,Birnessite 1.351.60.03404411295 chydrothermal crustsŽ.–Below detection limit for method used.aŽ.Data from Thompson et al.1985.bŽ.Data from Bogdanov et al.1995.cŽ.Data from Hein et al.1997.Ž. transmission electron microscopy TEM.Approxi-mately100mg Mn crust was totally digested asŽ. described by Mills and Elderfield1995.Two repli-cates were digested for samples3387-R1-A1and 3387-R1-A2;REE determinations are within errorŽ.for these analyses Table2.REE separation wasŽ. carried out as described by Greaves et al.1989 with minor modification.Separated REEs were anal-ysed using a VG Elemental PlasmaQuad PQ2q Inductively Coupled Plasma Mass Spectrometer.Ac-Ž.curacy was better than4%2s SE,while externalŽ. precision of the data is better than3%2s SE.Neodymium isotopes were separated by HDEHP r Teflon reverse phase cation exchange chro-matography.Isotope ratios were evaluated by analy-sis using a VG Isomass54E thermal ionisation mass spectrometer.Isotope ratios were normalised to 146Nd r144Nd s0.7219.Standard JMC321was anal-ysed repeatedly giving143Nd r144Nd s0.511121"9Ž.ppm2s SE,cf.0.511123quoted.2s error bars are plotted in Fig.6.Data are expressed using the´Ž.Ž. notation,relative to the CHUR0value0.512638 of143Nd r144Nd.4.ResultsThe published mineralogy and major element geo-chemistry of the sample suite is presented in Table1.Ž.Thompson et al.1985describe samples as mix-tures of nontronite and amorphous Fe–Mn oxides Ž.Ždive1243and1247and laminated birnessite dive .1244.XRD analysis identified all BRAVEX sam-ples,both dredged and submersible collected,as X-ray amorphous oxides.TEM energy dispersive X-ray spectroscopy identified that samples described as amorphous consisted of Fe–Si particles and Fe–Mn particles,which are micro-crystalline.The mi-cro-crystalline BRAVEX samples are much finer grained than the samples described by Thompson et Ž.al.1985.The-2m m fraction of the sediment sample3423was characterized as Al-poor,Fe-rich nontronite.The Fe and Mn content of the suite of samplesŽ.studied varies widely Table1.Fe content varies from-0.01%to41.52%,Mn content from0.01% to52.3%and the Fe r Mn ratio from-0.002to 3930.The trace element content of the Fe–Mn crusts is also highly variable across the TAG field.Samples 1243,1244,1247and3423-nontronite all have trace element levels-0.01%.Conversely,samples3387 and3423exhibit high levels of Co,Ni,Cu and Zn.REE data are presented in Table2and normalisedŽ.to chondrite values of Evensen et al.1978in all subsequent figures.The Ce anomaly is defined as )Ž.Ce r Ce s Ce r La q Pr r2,where the subscriptn n nn refers to chondrite normalised values.Similarly the)ŽEu anomaly is defined as Eu r Eu s Eu r Sm qn n()ls et al.r Chemical Geology 1762001283–293288.Gd r 2.Fractionation of the LREEs from the HREEs n across the group is expressed here as the Nd r Yb n n ratio.The data show an extremely wide range in REE Ž.content 0.5–300ppm Nd and varied cross group Ž.fractionation Nd r Yb ;Fig.2.Samples from dives n n Ž1244and 1247have low REE contents 0.5–1.5ppm .Ž.Nd and are HREE enriched Nd r Yb s 0.8–1.1.n n These samples exhibit a clear negative Ce anomaly Ž).Ce r Ce s 0.4–0.6and no Eu anomaly relative to chondrite.The nontronite rich sample from dive Ž.1243-1exhibits LREE enrichment Nd r Yb s 4.8,n n which is similar to the separated nontronite sample Ž.3423;Nd r Yb s 5.31,and these two samples n n Žhave intermediate REE contents 23.8and 7.63ppm,.respectively with negative Ce anomalies.Samples from 3386,3387and 3423exhibit elevated REE Ž.contents )230ppm Nd and also show LREE Ž.enrichment Nd r Yb s 3.8–4.6.The Ce anomaly n n is positive for all samples collected from the eastern Ž)valley wall sites 3386and 3387;Fig.1;Ce r Ce s .1.8–2.2.Negative Ce anomalies are observed in the two Mn crusts collected from the inactive Alvin zone Ž.site 3423;Fig.1.No sample collected displays any positive Eu anomaly.The Nd isotopic composition of TAG Mn crusts varies from y 12.3to y 2.13´units,with values generally similar to those of other Atlantic ferroman-Žganese deposits and sea water Fig.3;Piepgras et al.,.1979;Piepgras and Wasserburg,1987.Only one sample,1247-2-2D,exhibits a Nd isotopiccomposi-Fig.2.Chondrite normalised REE patterns for the TAG Fe–Mn crusts and separated nontronite.Sample numbers correspond to locations in Fig.1.Fig.3.Histogram of Nd isotopic composition for the TAG Fe–Mn crusts.Data for an Atlantic hydrothermal crust and ferroman-Ž.ganese nodules are shown for comparison Piepgras et al.,1979.Ž.Data for NADW are from Piepgras and Wasserburg 1987.ŽŽ.tion significantly different from sea water ´0s Nd .y 2.13.5.DiscussionMn oxidation is kinetically limited,whereas Fe precipitation occurs very rapidly under oxic condi-tions and there is extreme fractionation of these two hydrothermally derived elements on the scale of metres.Hydrogenous deposits tend to exhibit a more Žconsistent Fe r Mn ratio ;0.7for Pacific Fe–Mn .crusts;Table 1and indeed the amorphous samples collected here have Fe r Mn ratios of ;1.3,which is consistent with hydrogenous accumulation from sea water.The wide range of Fe r Mn ratios observed in this suite of samples reflects the sharp geochemical gradients present in hydrothermal settings and the wide range of processes occurring over the scale of the TAG field.Generally,hydrothermal Fe–Mn deposits exhibit much lower trace element levels than hydrogenous Ž.deposits Hein et al.,1997.Only the X-ray amor-phous oxide samples 3387and 3423contain elevated trace element contents,all other samples are depleted in trace metals.The Co content of samples 3387and Ž.3423is relatively high 230–1720ppm and is inter-mediate between Pacific hydrothermal and hydroge-Ž.nous deposits Table 1.The Co and Ni enrichment of these samples is indicative of slow hydrogenous()ls et al.r Chemical Geology1762001283–293289accumulation and oxidative scavenging from sea wa-ter.The Cu and Zn contents of these amorphousŽsamples is extremely high up to21200ppm Cu and.2200ppm Zn in sample3423.The input of metals from TAG high temperature venting is a potential source of Cu and Zn to these deposits.Cu and Zn sulphides have been observed to fall-out from the neutrally buoyant plume above the active TAG field and could be a source of chalcophile elements toŽ.sample3423Trocine and Trefry,1988.The trace element geochemistry and mineralogy of these sam-ples suggests that samples3423and3387are domi-nantly hydrogenous with some hydrothermal input of Cu and Zn evident in3423.The geochemistry and mineralogy of samples1243,1244and1247sug-gests that they are formed by hydrothermal processes Ž.Thompson et al.,1985.The variability in REE pattern is extreme overŽ.short spatial scales Fig.2.Samples from the east-ern valley wall display two orders of magnitude variation in REE content,both positive and negative Ce anomalies and variable HREE enrichment.Sam-ples from the inactive Alvin zone have intermediate REE contents and inter-element fractionation.The two indices of REE behaviour of relevance here areŽ).the Ce anomaly Ce r Ce and the cross groupŽ.fractionation Nd r Yb.These two parameters aren nplotted in Fig.4where a wide range of behaviour is apparent in the samples analysed.Samples from sites3386,3387,3423and1243-1 show REE patterns indicative of extensive uptake of REEs from sea water.Similar patterns have been previously reported for both hydrothermal and hy-Ždrogenous ferromanganese deposits Toth,1980;. Fleet,1983;Bau et al.,1996;Kuhn et al.,1998. Uptake of REEs from sea water results in fractiona-tion during preferential removal of the middle REEs ŽCantrell and Byrne,1987;German et al.,1990; Koeppenkastrop and de Carlo,1992;Rudnicki and.Elderfield,1993;Bau,1999.This is demonstratedby comparison of the Nd r Yb measured in then nŽ.crusts Nd r Yb s3.8–5.3with that of TAG sea n nŽ.water Nd r Yb s 1.1;Mitra et al.,1994.These n nŽ.crusts3386,3387and3423are typical of deposits found in a variety of tectonic settings from differentŽ.oceanic areas Bau et al.,1996both in REE abun-dance and overall pattern.The Nd r Yb fractiona-n ntion during uptake is similar for sample1243-1Fig.4.Plot of Nd r Yb against Ce r Ce)for the TAG Fe–Mnn ncrusts and separated nontronite.Variations in the Ce anomaly aredetermined by either hydrothermal or hydrogenous processes.Variations in the Nd r Yb ratio are controlled by plume orn nlow-temperature fluid input as discussed in the text.Ž.Nd r Yb s4.8and the separated nontronite sam-n nŽ.ple from site3423Nd r Yb s5.3though the REEn ncontent is much lower than for samples3386and Ž.3387Nd r Yb s4.1–4.6.There is no correlation n nbetween REE content and Fe r Mn ratio or any appar-ent mineralogical control on the bulk REE patterns. It is proposed that the two-order of magnitude differ-ence in REE content arises from the more crystalline nature of the nontronite samples leading to decreased effective scavenging area compared with the X-ray amorphous crusts.The Ce anomaly and rate of accumulation of crusts relates to proximity of plume derived sources of Mn.Hydrogenous crusts with positive Ce anoma-lies are inferred to accumulate at-10mm Ma y1Ž.Kuhn et al.,1998.The shallowest samples on theŽ.eastern valley wall3386and3387are dominantly hydrogenous in nature suggesting that the influence of any hydrothermal plume at TAG is restricted to below the2900m contour on the eastern valley wall for at least the last few100000years.This is consistent with the distribution of the active hy-drothermal plume,which rises to a maximum eleva-Ž. tion of3200m Rudnicki and Elderfield,1992. Samples collected from deeper than the2900m contour exhibit hydrothermal signatures manifest as negative Ce anomalies.()ls et al.r Chemical Geology1762001283–293 2906.Evidence for low-temperature hydrothermal ventingCrusts1244and1247have distinct REE composi-tions and low Nd r Yb ratios compared with then nhydrogenous and plume-derived crusts discussed above.Crusts from these two sites are dominated by birnessite and Fe-oxides.Fig.4demonstrates the extreme fractionation of the REEs relative to the modes of formation in the suite of samples studied here.As shown in the hydrogenous uptake of REEs in samples3386,3387and3423,REE fractionation during scavenging uptake from sea water leads to light to middle REE enrichment in the solid phase, therefore the HREE enrichment of these samplesŽ.cannot be from sea water Fig.4.Trace metal complexation in solution is strongly dependent on pH,temperature and ligand content. The REEs are particularly sensitive to changes in pH as their solution chemistry is dominated by carbonate complexation in sea water,the proportion of free REE ions in solution is predicted to increase fivefoldŽ. for a0.5pH unit decrease Byrne et al.,1988. Increasing the free REEs in solution is one way to increase the scavenging uptake.However,the four-fold HREE enrichment of the hydrothermal Fe–Mn deposits is unlikely to arise solely from changes in fluid complexation,cross-group variation in chloro-complexation is not predicted to exceed a factor of2Ž.Wood,1990;Haas et al.,1995.The REEs are fractionated during non-conserva-tive subsurface mixing processes and mineralisa-Žtion r dissolution reactions Mills and Elderfield, .1995.Very few measurements of the REE content of low-temperature fluids from TAG are available, those published indicate subsurface mixing of black smoker fluids with sea water with some modificationŽbut no significant cross-group fractionation James.and Elderfield,1996;Bau and Dulski,1999.Pore fluids collected from one core from atop the activeŽmound do show HREE enrichment Nd r Yb valuesn n.as low as0.33along with a small positive Eu Ž.anomaly Rudnicki and Mills,1997.This HREE enrichment is remarkably similar to that observed inŽ.the Fe–Mn crusts analysed here Fig.4.We con-clude that the HREE enrichment observed in the samples analysed results from changes in fluid com-plexation during subsurface precipitation and reac-tion.The nearest area of high-temperature mineralisa-Ž.tion is the Mir mound Fig.1.The fluids that are inferred to source the low-temperature deposits ex-amined here must be channeled at least3km from the source of high temperature activity to vent as low-temperature fluids on the eastern valley wall. Structural controls on fluid flow through the TAG segment have previously been inferred from high resolution imaging of the sea floor and the distribu-Žtion of mineralisation across the field Thompson et.al.,1988;Kleinrock and Humphris,1996.Further evidence for a direct hydrothermal source for crusts from sites1244and1247comes from published Os isotopic analyses for the same samples Ž.187186Ravizza et al.,1996.Os r Os ratios for sam-ples1244and1247fall in the range 4.23–4.77, representing some of the least radiogenic Os isotopicŽ. signatures observed at TAG Ravizza et al.,1996. Sample1243-1exhibits a more sea water like OsŽ187186isotopic ratio Os r Os s6.8,cf.sea water s .Ž8.5,which is consistent with the REE patterns Figs..2and4.However,there is no significant correlation Ž.187186between´0and Os r Os values for the sam-Ndple suite highlighting the different controls on the marine geochemistry of these two elements.7.Nd isotope systematicsExcluding sample1247-2-2D,all samples anal-Ž.ysed here have´0values between y10.7andNdŽ.y12.3Fig.3.These values are within the range of compositions of bulk Atlantic ferromanganese nod-Ž. ules and hydrothermal crusts Piepgras et al.,1979. More recently,high precision subsampling across growth laminae in crusts has allowed the Nd isotopic composition of NADW to be determined through Ž. time Burton et al.,1997;Abouchami et al.,1999.Ž.´0has shifted towards less radiogenic values NdŽ.over the last13Ma,with a range in´0values ofNdy10.77to y11.84recorded in hydrogenous crustsŽfrom258N in the Eastern Atlantic Basin Abouchami .et al.,1999.Since all of the TAG deposits form onŽcrust that is-1Ma old based on half spreadingy1.rates of1.3cm year,there is no detectable varia-Ž.tion in NADW´0composition during the timeNd()ls et al.r Chemical Geology1762001283–293291Ž.Fig.5.Plot of´0against Nd r Yb for the TAG Fe–MnNd n ncrusts.The dashed line indicates conservative mixing of the TAG black smoker source with NADW in the proportions indicated.taken for the hydrogenous deposits to form.This isŽ.reflected in the uniform´0composition of theNdhydrogenous and plume derived crusts sampled from Ž.TAG Fig.3.One sample from the suite measured has a signifi-Ž. cant radiogenic component reflected in the´0Nd value of y2.2which must contain significant MORB derived Nd.Fig.5demonstrates that the radiogenicŽNd isotopic composition of sample1247-2-2D Fig. .3is consistent with mixing of a residual hydrother-Ž.Ž.mal fluid0.29%with sea water99.71%with fractionation of Nd from Yb that is similar to theŽother low-temperature derived crusts Nd r Yb sn n .Ž.0.8–1.1.While such MORB dominated´0val-Ndues have been previously observed in metalliferous Ž.sediments ls et al.,1993,this is the first observation of significant MORB derived Nd in a marine Fe–Mn crust.This signature must be related to high-temperature venting at TAG,either the cur-rent venting from the active mound or previousŽ. events associated with the inactive mounds Fig.1. The hydrothermal component in the low-temperature crusts is inferred to be channelled to the eastern valley wall over distances of over3km via perme-Žability and structural controls Kleinrock and.Humphris,1996.8.ConclusionsFerromanganese deposits collected from the TAG hydrothermal field exhibit a wide range of REE contents.Samples dredged from the eastern valley wall exhibit hydrogenous REE patterns with positive Ce anomalies,high REE contents and sea water dominated REE patterns.These observations are consistent with the relatively constant Fe r Mn ratios and the elevated Co and Ni content of these crusts. Samples from the Alvin zone are derived from a hydrothermal plume and exhibit negative Ce anoma-lies along with enrichment in plume derived metals such as Cu and Zn.The distribution of these samples confines any hydrothermal plume impinging on the eastern valley wall to depths)2900m over the last few100ka.The majority of the samples are domi-nated by sea water scavenging processes,the range in REE content being controlled by the nature of the substrate and effectiveness of scavenging.Four sam-ples from the eastern valley wall demonstrate evi-dence for input of a residual hydrothermal fluid,with significant HREE enrichment and more radiogenic Nd isotopic composition.One Fe–Mn crust contains up to2.9%of a residual hydrothermal component, which is the first observation of significant MORB derived Nd in a marine Fe–Mn deposit.REE and Nd isotopic data confirm that some of the Fe–Mn deposits from the low-temperature zone on the eastern valley wall have a direct hydrothermal origin.The extent of low-temperature diffuse flow at TAG is well established around the active mound; here,we provide geochemical evidence for low-tem-perature venting away from the active mound with a significant input of radiogenic Nd to a Fe–Mn crust forming)3km from any high-temperature source. The wide range in REE behaviour observed over short spatial scales highlights the need for extreme care in interpreting the fossil ferromanganese record to reconstruct palaeoenvironments.The strong vari-ability within one hydrothermal field opens up the possibilities to make detailed recontruction of hy-drothermal venting during the growth period of the deposits.AcknowledgementsThis research was carried out with support from Ž.NERC GST r02r976to RAM and a University of Southampton Studentship to DMW.We are grateful。