Rare earth elements in the Pacific and Atlantic …kbruland/Manuscripts/...different positions while...

17
Genchrmrca CI Cosmochrmrca Acra Vol. 49, pp. 1943-1959 0 Pewmon Press Ltd. 1985. F’rintcd tn U.S.A. Rare earth elements in the Pacific and Atlantic Oceans* HEIN J. W. DE BAAR,’ MICHAELP. BACON. PETER G. BREWER Woods Hole Oceanographic Institution, Woods Hole. MA 02543. U.S.A. and KENNETH W. BRULAND Center for Coastal Marine Studies. University of California at Santa Cmz. Santa Cruz, CA 95060. U.S.A. (ReceivedAugust 27, 1984: accepted in ~~ised~orm June 2 1. 1985) Abstract-The first profiles of Pr, Tb, Ho, Tm and Lu in the Pacific Ocean, as well as profiles of La. Ce. Nd, Sm, Eu, Gd and Yb, are reported. Concentrations of REE (except Ce) in the deep water are two to three times higher than those observed in the deep Atlantic Ocean. Surface water concentrations are typically lower than in the Atlantic Ocean, especially for the heavier elements Ho. Tm, Yb and Lu. Cerium is strongly depleted in the Pacific water column. but less so in the oxygen minimum zone. The distribution of the REE group is consistent with two simultaneous processes: ( 1) cycling similar to that of opal and calcium carbonate, and (2) adsorptive scavenging by settling particles and possibly by uptake at ocean boundaries. However. the first process can probably not be sustained by the low REE contents ofshells, unless additional adsorption on surfaces is invoked. The second process. adsorptive scavenging. largely controls the oceanic distribution and typical seawater pattern of the rare earths. INTRODUCTION THE RARE EARTH ELEMENTS (REE) include the element lanthanum (La) and the fourteen elements that follow La in the periodic table (the lanthanid~). They form a very coherent group, although two elements, Ce and ELI, may develop anomalies due to changes in their oxidation states. The most stable oxidation state of all Ianthanides is the REE(II1) form. The gradual filling of the inner 4f shell with increasing atomic number leads to gradual changes in chemical properties. For instance. the heavier REE(II1) were predicted to be more strongly complexed in seawater (GOLDBERG et al., 1963: TURNER et al., I98 f 1. Such gradual changes of prop erties in aqueous solution make the REE series a unique group of elements in marine chemistry. The La3’, Gd3+ and Lu3+ cations with exactly empty, half-filled and completely filled 4,felectron shell are particularly stable, and this may lead to excursions ofthese elements from the gradual trends in shale normalized distribu- tion patterns (DE BAAR et al., 1985a). The Ce“’ and Eu2+ cations represent the only other geochemicaJly important oxidation states. These dif- ferent valences lead to distribution anomalies com- pared to the strictly trivalent REE. Reduction of Eu is thought to occur often in magmatic processes. The larger ionic radius of Eu(I1) then leads to its fraction- ation from the rest of the series. Anomalies of Eu have been reported in various types of igneous and sedi- mentary rocks (HENDERSON. 1984). However. reduc- tion of Eu does not seem to occur within the ocean basins except by hydrothermal systems (MICHARD et al., 1983). Present address: Department of Earth Sciences, University of Cambridge, Cambridge CB3 OEZ, U.K. * W.H.O.I. Contribution Number 5750. Cerium may exhibit anomalies due to oxidation to the Ce(IV) state. Development ofCe anomalies during formation of igneous rocks can be largely ruled out (SCHREIBER et a/., 1980). with one exception ascribed to recycling of oceanic lithosphere (HEMING and RANKIN, 1979). Otherwise, Ce anomalies have been found exclusively within the ocean basins. Seawater is typically depleted in Ce by comparison with neigh- boring La and Pr, whereas fe~omangane~ nodules often exhibit a Ce enrichment. Many other trace ele- ments in seawater (e.g., Mn, Fe, Cu. As, Sb, Se and I) are also affected by their multiple oxidation states. However, Ce and Eu are uniquely interesting because anomalies can be defined easily and quantitativeIy by comparison with the strictly trivalent neighbors in the REE series. Thus, in principle, one can single out ox- idation-reduction reactions of Ce and Eu from all other processes affecting their oceanic distributions. The analysis of picomolar REE concentrations in seawater is more difficult than the determination of the much higher REE concentrations in marine de- posits. However, the resulting values for REE dissolved in seawater represent a single phase of great significance for unde~~ding their marine ge~hemist~. The REE contents of bulk marine deposits, on the other hand, often represent an assemblage of several geochemical phases. For example, the REE signal of calcareous oozes (SPIRN, 1965) may be masked by ferromanganese coatings (PALMER, 1983) or a dominant te~genous (detrital) component with its flat shale type REE pat- tern. The sedimentary geochemistry of the REE was recently reviewed by FLEET (1984). In this paper we discuss mainly the modern seawater results. Until recently little reliable data were available on the REE in seawater. Values reported by GOLDBERG et al. (1963). HOGDAHLet al. (1968) and MASUDA and IKEUCHI(1979) fall within the range reported recently 1943

Transcript of Rare earth elements in the Pacific and Atlantic …kbruland/Manuscripts/...different positions while...

Genchrmrca CI Cosmochrmrca Acra Vol. 49, pp. 1943-1959 0 Pewmon Press Ltd. 1985. F’rintcd tn U.S.A.

Rare earth elements in the Pacific and Atlantic Oceans*

HEIN J. W. DE BAAR,’ MICHAEL P. BACON. PETER G. BREWER Woods Hole Oceanographic Institution, Woods Hole. MA 02543. U.S.A.

and

KENNETH W. BRULAND

Center for Coastal Marine Studies. University of California at Santa Cmz. Santa Cruz, CA 95060. U.S.A.

(Received August 27, 1984: accepted in ~~ised~orm June 2 1. 1985)

Abstract-The first profiles of Pr, Tb, Ho, Tm and Lu in the Pacific Ocean, as well as profiles of La. Ce. Nd, Sm, Eu, Gd and Yb, are reported. Concentrations of REE (except Ce) in the deep water are two to three times higher than those observed in the deep Atlantic Ocean. Surface water concentrations are typically lower than in the Atlantic Ocean, especially for the heavier elements Ho. Tm, Yb and Lu. Cerium is strongly depleted in the Pacific water column. but less so in the oxygen minimum zone. The distribution of the REE group is consistent with two simultaneous processes: ( 1) cycling similar to that of opal and calcium carbonate, and (2) adsorptive scavenging by settling particles and possibly by uptake at ocean boundaries. However. the first process can probably not be sustained by the low REE contents ofshells, unless additional adsorption on surfaces is invoked. The second process. adsorptive scavenging. largely controls the oceanic distribution and typical seawater pattern of the rare earths.

INTRODUCTION

THE RARE EARTH ELEMENTS (REE) include the element lanthanum (La) and the fourteen elements that follow La in the periodic table (the lanthanid~). They form a very coherent group, although two elements, Ce and ELI, may develop anomalies due to changes in their oxidation states.

The most stable oxidation state of all Ianthanides is the REE(II1) form. The gradual filling of the inner 4f shell with increasing atomic number leads to gradual changes in chemical properties. For instance. the heavier REE(II1) were predicted to be more strongly complexed in seawater (GOLDBERG et al., 1963: TURNER et al., I98 f 1. Such gradual changes of prop

erties in aqueous solution make the REE series a unique group of elements in marine chemistry. The La3’, Gd3+ and Lu3+ cations with exactly empty, half-filled and completely filled 4,felectron shell are particularly stable, and this may lead to excursions ofthese elements from the gradual trends in shale normalized distribu- tion patterns (DE BAAR et al., 1985a).

The Ce“’ and Eu2+ cations represent the only other geochemicaJly important oxidation states. These dif- ferent valences lead to distribution anomalies com- pared to the strictly trivalent REE. Reduction of Eu is thought to occur often in magmatic processes. The larger ionic radius of Eu(I1) then leads to its fraction- ation from the rest of the series. Anomalies of Eu have been reported in various types of igneous and sedi- mentary rocks (HENDERSON. 1984). However. reduc- tion of Eu does not seem to occur within the ocean basins except by hydrothermal systems (MICHARD et al., 1983).

’ Present address: Department of Earth Sciences, University of Cambridge, Cambridge CB3 OEZ, U.K.

* W.H.O.I. Contribution Number 5750.

Cerium may exhibit anomalies due to oxidation to the Ce(IV) state. Development ofCe anomalies during formation of igneous rocks can be largely ruled out (SCHREIBER et a/., 1980). with one exception ascribed to recycling of oceanic lithosphere (HEMING and RANKIN, 1979). Otherwise, Ce anomalies have been found exclusively within the ocean basins. Seawater is typically depleted in Ce by comparison with neigh- boring La and Pr, whereas fe~omangane~ nodules often exhibit a Ce enrichment. Many other trace ele- ments in seawater (e.g., Mn, Fe, Cu. As, Sb, Se and I) are also affected by their multiple oxidation states. However, Ce and Eu are uniquely interesting because anomalies can be defined easily and quantitativeIy by comparison with the strictly trivalent neighbors in the REE series. Thus, in principle, one can single out ox- idation-reduction reactions of Ce and Eu from all other processes affecting their oceanic distributions.

The analysis of picomolar REE concentrations in seawater is more difficult than the determination of the much higher REE concentrations in marine de- posits. However, the resulting values for REE dissolved in seawater represent a single phase of great significance for unde~~ding their marine ge~hemist~. The REE contents of bulk marine deposits, on the other hand, often represent an assemblage of several geochemical phases. For example, the REE signal of calcareous oozes (SPIRN, 1965) may be masked by ferromanganese coatings (PALMER, 1983) or a dominant te~genous (detrital) component with its flat shale type REE pat- tern. The sedimentary geochemistry of the REE was recently reviewed by FLEET (1984). In this paper we discuss mainly the modern seawater results.

Until recently little reliable data were available on the REE in seawater. Values reported by GOLDBERG et al. (1963). HOGDAHL et al. (1968) and MASUDA and IKEUCHI (1979) fall within the range reported recently

1943

1944 H. J. W. de Baar et al.

by PIEPGRAS et al. ( 1979), PIEPGRAS and WASSER~URG ( 1980. 1982, 1983), ELDERFIELD and GREAVD ( 1982), DE BAAR et al. (1983a. 1985a). KLINKHAMMER et al. (1983) and in this paper. However. both GOLDBERG et al. ( 1963) and MASUDA and IKEUCHI ( 1979) report values for only one sample, which cannot be verified for internal oceanographic consistency. The results of HOGDAHL et al. (1968) exhibit considerable scatter when plotted as vertical profiles. Early work by BAL- ASHOV and KITROV (196 1). HOD ( 1966), SHIGE-

MATSU et al. (1967), HAYES ( 1969). NAGATSAKU et al. ( 197 1) and KOLESOV et al. ( 1975) is mostly of his- torical interest. In this paper we present new data on the REE distributions in the Pacific Ocean based on modem, clean sampling techniques and analysis. A comparison of the new results with data previously obtained from the Atlantic Ocean sheds new light on the geochemical behavior of this important group of elements. The first results of this study were presented earlier (DE BAAR et al., 1983b).

METHODS

The Pacific Ocean water samples were obtained at about 18”N, 108”W during the VERTEX II cruise in November 198 1. Our samples were taken on different days at slightly different positions while the ship tracked a free bloating sedi- ment trap array. Seawater samples were collected with 30- liter, Tetlonxoated, modified Go-Flo bottles (General Ckean- its) on Dacron-sheathed, Kevlar hydroline. Upon recovery the bottles were placed outside a clean air laboratory van and connected to filter assemblies inside the van. The water was filtered through acid&at&, 142 mm diameter, 0.3 micron pore size Nuclepore (double pore density) polycarbonate membrane filters held in Millipore Teflon filter sandwiches. with 0.Cmicron-filtered nitrogen overpressure. Water samples were stored in hot-acid-cleaned, polyethylene contakrs and acidified to pH 2 with double quartz distilled 8 N hydrochloric acid. All acidifications, filter changing and other sample ma- nipulations were performed inside a class-100. laminar flow bench inside the clean van.

Details of the analytical methods have been reported else- where (DE BAAR, 1983, 1984). Briefly, extractions and puri- fications were done in a class-100 laminar flow bench in a clean air laboratory using double quartz distilled reagents and hot acid cleaned nolvethvlene and Teflon (PTFE: FEP) lab ware. After a&&on- ani equilibration of ‘carrier-free ““Ce intemal standard spikes, the seawater was neutralized, buffered at pH 6 and pumped through a CHELEX-100 chromatog raphy column for extraction of REE and separation of major ions. Subsequently co-extracted U and remaining traces of Na were removed by cation exchange chromatography (STRELOW, 1963). Any remaining U traces were removed by a final anion exchange step (KLAUS and NELSON, 1955). Complete elimination of U is important, because neutron ir- radiation of 235U would produce the interfering fission products ““La, “Ice, ‘*Cc and “‘Nd. For example, one mole U gives rise to a “Ice signal which equals that produced from about 0.5 mole Ce. Some ofthe neutron activation analyses of marine particles, ferromanganese nodules and apatite reported earlier (KNAUSS and Ku, 1983; DYMOND et al.. 1984; MURPHY and DYMOND, 1984; WRIGHT, SEYMOUR and SHAW, 1984) may be suspect, fot they would have required pre-separation of U to be sure ofeliminating significant interferences. The apparent U/Cc mole ratios reported in those samples ranged from very low to as high as -3. Interestingly enough the first reported negative Ce anomalies in nodules were determined with other techniques (ELDERFIELD and GREAVES. 198 1). The problem of potential interferences is even more serious for seawater.

for which U/Cc molar ratios in the Iti- IO4 range necessitate rigorous U separation.

Radiotracer studies with YZe. IJ2Eu and “‘Gd estabhshed quantitative recoveries for each of the three consecutive sep- arations as well as the overall procedure. Mixtures of REF standards were made up gravimetricaily from pure REE metals (Ames Laboratory). Standard mixtures were intercalibrated with those of Elderfield and Greaves (DE BAAR et 01.. 1985a). Irradiation of samples and standard mixtures for 8 hours at a high flux of 5 X 10” neutrons. cm *. set ’ at the MIT Nu- clear Reactor Laboratory was followed by gamma spectrom- etry for final determination of REE concentrations. The ele- ments Dy and Er were not determined because of the short half-life (2.3 hours) of ldsmDy and major ‘@‘Yb interferences with the “‘Er peak. Spectra were monitored to assure the absence of 239Np, a sensitive indicator of residual traces of 1’.

RESULTS

The hydrography of the VERTEX II sate (iX”h. 108”W) has been described by BROENKOW and KRENZ ( 1982). It is an area of merging surface waters from (I) the California Current. (2) the Equatorial Coun- tercurrent, and (3) the Gulf of California. il is In this region of surface convergence that the North Equatorial Current is formed ( WYRTKI, 1967). The most notable aspect of this area is the well developed oxygen mini- mum zone extending from 100 m to 800 m depth (see Fig. 7). Between these depths the oxygen concentration is less than 10 micromoles kg-’ (about 4% saturation) but was never observed to be below about 1.5 rmol * kg-’ during the VERTEX II cruise. At depths between 150 m and 700 m there was evidence of de- nitrification with a 10 pmol kg-’ nitrate anomaly and corresponding nitrite maximum centered at 250-300 m. These features (oxygen, nitrate, nitrite) also hare an advective component. The region is characterized by nitrate reduction in the sub-oxic zone (CLINE and RICHARDS, 1972).

All REE except Ce exhibit increasing concentrations with depth (Table 1; Fig. 1). Such an increase was also observed for the REE in the North Atlantic Ocean (ELDERFIELD and GREAVES. 1982: DE BAAK ~‘r al., 1983a), but it is much more pronounced in the Pacific, especially for the heavier elements like Yb and Lu. In the deep water these heavy REE are very strongly en- riched (Fig. 2). Concentrations of Ce decrease with depth, and Pacific Ce levels are always below those found at comparable depths in the Atlantic Ocean (Fig. 1). The negative Ce anomaly observed over the whole water column increases with depth, with verq’ strong Ce depletions occurring in the bottom waters. Smaller anomalies of Gd and Tb are also observed systemati- cally throughout the water column. These latter anomalies are probably due to solution chemistry shifts in the transition from an exactly half-filled rlj‘electron shell and are discussed fully in a separate paper (Dt BAAR et al., 1985a). For all elements except La (see below). the concentrations in the upper water column are equal to or less than those in the Atlantic Ocean (Fig. 1). This trend is consistent with the results of KLINKHAMMER et al. (1983), who found even lower surface water values at a station in the western Pack

REE in the oceans 1945

(18”N, 145’E). The interoceanic differences are very apparent from plots of the Atlantic/Pacific concentra- tion ratios at comparable depths (Fig. 3).

The profiles versus depth appear to be rather smooth in the upper 750 m, with the exception of pronounced maxima at 150 m. These maxima coincide with the salinity maximum which characterizes the core of the Subtropic Surface Water (WYRTKI, 1967), and they lie only about 20 m below the very pronounced oxygen minimum (see Fig. 7). At the same station MARTIN et al. (1983) found broad maxima for Cu and Ag in the same depth range and also minima at about 60 m. The 150 m depth range is also marked by a distinct Fe maximum (LANDING, 1983; GORDON et al.. 1982) and by sharp increases in Mn (LANDING, 1983; MARTIN and KNAUER, 1984) and Co (KNAUER et al., 1982) which reach maxima at about 300 m depth. Indepen- dent model estimates of MARTIN and KNAUER ( 1984) and LANDING ( 1983) both suggest that the Mn maxi- mum is maintained by a combination of in situ regen- eration (reduction) of Mn from sinking particles and lateral transport of dissolved Mn from other regions. MARTIN and KNAUER (1984) believe that lateral transport from source regions as far away as the con- tinental shelves is the more important process. Of course, the high Mn levels in such source regions are ultimately generated by chemical reactions similar to those operating in situ in the suboxic zone ofthe VER- TEX II station.

The limited REE data set for this station does not allow assessment of the relative importance of lateral transport versus in situ chemical reactions. However. the surface depletion (Fig. 1) could well be explained by some degree of scavenging on particulate ferro- manganese oxides or oxide coatings on settling parti- cles. Subsequent release of REE upon dissolution of the Fe/Mn oxide phases in the oxygen depleted zone may lead to the REE maxima at 150 m. Similar ob- servations were made in the Cariaco Trench. where

most REE exhibit a pronounced increase just below the C&/Hz!5 interface (DE BAAR, 1985b), very much hke Fe and Mn (BACON et al., 1980; JACOBS, 1984). The suggested correlation with Fe and Mn is also sup- ported by elevated REE concentrations in ferroman- ganese coatings (PALMER, 19X3), nodules (ELDERFIELD et al., 198 I) and crusts (FLEET, 1984). Of course. the upper water column at the VERTEX II station also exhibits intense biological activity, and more direct biological controls cannot be fully excluded.

From the 600-1000 m range downward the Pacific REE concentrations begin to exceed the Atlantic val- ues. Pacific deep water values are two to four times as high as in the western North Atlantic Ocean. The origin of the REE maxima at 1000 m (Fig. I) is uncertain. Similar maxima were observed in the same depth range in the North Atlantic Ocean (DE BAAR ef ul.. 1983a). Possibly they are associated with the Antarctic Inter- mediate Water, but there are too few data to be certain of it.

DISCUSSION

Biogeochemical qxiing

The elevation of the REE concentrations in deep Pacific compared with deep Atlantic waters resembles that of the micronut~en~ (BROECKER and PENG, t 982) and suggests some degree of biogeochemical cycling. The deep sections (> 1000 m) of the profiles (Fig. I) resemble profiles of silicate or alkalinity more than they do those of nitrate or phosphate. We earlier reported the linear relationship

Lu [pmol . kg-‘]

= 0.0306 Si [pmol - kg-‘] + 0.435 ( I)

with r = 0.99 in the North Atlantic Deep Water (NADW) over the 98 1-4427 m depth range (DE BAAR et al., 1983a). In the western North Atlantic Si is gen-

Table 1. Concentrations l1O~“mol.kg~~I of REE in filtered seawater sapnples

collected at IB’N, 108*W during YERTEX II*.

Depth [ml La Ce Pr Nd Srn Eu Cd Tb Ho Trn yb !_!I

15 19 11 3.2 13 2.7 0.70 4.0 0.54 0.97 0.35 2.2 0.35 45 22 10 3.5 16 2.8 0.69 3.7 0.56 0.71 0.40 1.9 0.30 100 32 10 3.3 15 2.6 0.76 4.0 O.S8 0.83 0.52 2.8 0.44 150 47 25 4.3 24 4.0 1.23 6.3 0.91 1.50 0.86 5.75 0.96 200 17 17 2.5 13 2.6 0.71 3.7 0.55 l.lf 0.57 3.5 0.60 300 19 18 3.0 15 2.6 0.77 4.3 0.61 1.02 0.57 3.7 400

0.63 22 13 2.3 14 2.6 0.71 4.0 0.54 1.20 0.62 4.0 0.68

500 20 13 3.1 15 2.5 0.75 4.2 0.58 1.50 0.66 4.0 0.71 750 34 8.4 4.2 17 3.1 0.82 4.1 0.70 1.40 0.78 5.5 0.98

1000 35 7.4 7.6 3b 6.4 1.56 8.6 1.41 3.52 1.84 13.2 2.44

1250 33 4.2 4.5 25 4.5 1.25 7.1 1.13 2.36 1.5 9.1

1750

1.63

49 4.2 7.4 27 6.0 1.47 8.6 1.33 3.3 1.9 13 2.4

2000 46 5.3 5.6 24 5.2 1.30 7.2 1.12 2.8 1.5 11 2.0

2250 67 3.3 8.5 33 6.7 1.68 9.4 i.47 3.75 2.0 14 2.6

2750 63 2.9 a.9 42 9.0 2.32 13 2.01 4.4 2.5 17 3.1 3000 51 3.4 9.2 49 8.8 2.43 13 2.11 4.8 2.4 15 2.7 3250 67 2.9 7.0 41 7.7 2.15 12 1.81 4.0 1.95 13 2.3

Shales*‘: 295 592 72 263 50 10.6 40.4 7.7 8.1 3.7 20.4 3.5

*Precision (one standard deviation) for each of the 12 elements was estimsled to he better than 52 (DeEasr. 1984).

**Average concentrstione l1O~‘mol.kg“l in shales BS determined from BD arithmetic mean of REE levels

in North American Shales Composite, 1966).

European shales and gussian platform sheles (HASKIN Blld HASKIN,

1946 H. J. W. de Baar ei al

Concentfatton /pi0 fnol kg-1

0 20 40 60

- PACIFIC -----.ATLANIIC

FIG. 1. Vertical profiles of 12 rare earth elements in the Eastern Equatorial Pacific Ocean. The dotted lines represent REE profiles in the Northwest Atlantic Ocean taken from DE BAAR er al. C I983a,, 1985al

erally consideted an approximately conservative tracer. and by analogy one would expect similar behavior for Lu. In contrast, plots of the lighter REE (Sm, Eu and Tb) versus Si (or Lu) exhibit negative curvature at mid- depth, indicative of their preferential removal either in situ or at another location earlier in the history of the NADW (DE BAAR et al., 1983a). KLINKHAMMER et al. (1983) found a similar linear relation between Er, another heavy REE, and silicate at two Pacific sta- tions. They also demonstrated negative curvature in plots of light REE (Nd, Sm) versus Er or silicate. Both trends were also observed by PALMER ( 1983) in the deep water of a station in the Eastern Equatorial At- lantic Ocean. Negative curvature of light versus heavy REE indicates preferential scavenging of the lighter elements. The heavy REE are indeed more stabilized in solution due to their stronger inorganic complexa- tion (GOLDBERG ef al.. 1963; TURNER ef ul.. I98 I ).

For the deep Pacific water (1250-3250 m) linear regression (r = 0.8) gives the relationship

Lu [pmol - kg-‘] = 0.027 Si [pm1 - kg ‘] I.82 (21

but the data are more scattered than are the Atlanttc data. Although the intercepts differ markedly. the slopes in the Atlantic and Pacific are similar with ALU/Z& 2 0.03 X lO-6 in both deep water masses. In contrast with the NADW (DE BAAR et al., 1983a; PALMER, 1983) and with the Pacific stations of KLINKHAMMER EZ al. (I 983) there is no distinct negative curvature of the lighter REE versus Lu or Si in the deep Pacific water at our own station,

Alkalinity in the deep ocean is affected partly by dissolution of CaC03 and by nitrate formation but is largely a conservative tracer. After correction for nitrate formation (BREWER, 1978)and normalization to 35% salinity, the resulting alkalinity variations 3re due

REE in the oceans

Concentration fiicomo,! kg-7

0 4 8 12 I60 I 2 30 2 4 6

1947

012 30 4 812 160 1 2 3

- PACIFIC c----e ATLANTIC

FIG. I. (Continued)

mainly to variations in CaCO, dissolution. The rela- tionships with Lu are similar in appearance to those of silicate (Fig. 4) at both sites:

Atlantic

Lu = 0.023 A&,,-,& - 52.65 (r = 0.98) (3)

Pacific

Lu = 0.022 Alb,& - 5 1.6 (r = 0.8). (4)

The offsets between oceans for Lu/Si and Lu/Alk relationships are compatible with the general circula- tion. Assuming that Lu (and other REE) are delivered mainly to the Atlantic Ocean, deep scavenging of the REE (see below) would cause them to increase less

than silicate and alkalinity as the deep water is trans- ported from Atlantic to Pacific Ocean. This scavenging of the REE also implies that they are recycled only a few times, say on average five times (ELDERFIELD and GREAVES, 1982) compared with Si, which is thought to be recycled about 20 times before final burial (BROECKER and PENG, 1982). The correlation of REE with both silicate and alkalinity is further demonstrated by the comparable values for Pacific/Atlantic ratios of concentration differences between surface and deep water (Table 2).

The observed correlations of REE with the biogeo- chemical cycle of skeletal material may result from two related mechanisms:

1948 H. J. W. de Baar e[ ui

La ce Pr Nd Sm Eu Gd fb

0.06

006

,r” L.

05

04

- OGe

GE

04

02

01

02

01

cl2

01

01

02

01

0 01

008

006

0.004 ~w-T7----T

LO Cc Pr Nd Sm Eu Gd ‘t! ‘K ‘(

FIG. 2. Distribution patterns of REE normalized versus shales (Table I) at selected depths in the Pa&c Ocean. Offset vertical logarithmic scales are used. Note the strong Ce depletions and heavy REE enrichment< at greater depths and the Gd anomalies at all depths

(i) true incorporation of REE(III) cations within the crystal lattice of opal or calcium carbonate

(ii) scavenging of REE by adsorption on settling particles, especially adsorption on surfaces or surface coatings (Fe/Mn oxides, organic films) of settling shells.

Both mechanisms would share the same skeletal ma- terial as REE carrier. Most of the regeneration of car- bonate and opal occurs shortly after arrival at the sea- floor (BROECKER and PENG, 1982). The associated REE would also be released into the bottom waters. It is likely that a small portion of the REE would be re- tained within the benthic boundary layer, due to scav-

enging in the nepheloid layer or adsorption on exposed surfaces, notably ferromanganese nodules and phos- phatic fish debris. From the major portion escaping into the deep water, a small fraction would be lost again due to scavenging in the water column. As a result the REE would not quite match the buildup ofsilicate and alkalinity levels with ageing of the deep and battom waters. This would explain the Pacific/Atlantic offsets discussed above (Fig. 4).

Unfortunately, the analogous oceanic distributions of Si and alkalinity (EDMOND, 1974) do not allow a clear distinction between opal or CaCQ as possible REE carriers. With respect to the possible uptake within

REE in the Oceans 1949

La Ce Pr Nd Sm Eu Gd Tb Ho Tm Yb Lu

LO Cs PI Nd Sm Eu Gd lb Ho Tm Yb LU

FIG. 3. Pacjfic/Atiantic con~ntmtion ratios at selected depths. pacific surface water is depleted relative to the Atlantic Ocean. The strong Ce depIe:ion at these depths mostly results from the high positive Ce anomalies at the Northwest Atlantic station, which are probably not representative of average North Atiantic surface waters. A crossover occurs at about 750 m depth, below which Pacific deep water is highly enriched in atl REE except Ce, which is strongIy depleted. ~MMER et al. (f983) reported similar trends for the deep water.

crystal lattices, the properties of the elements can at least serve to guide speculation. Incorporation of REE in opal cannot be ruled out, but the REE(II1) cations would fit very uncomfortably, if at all, in the silicon- oxygen matrix. In the oceanic water column the ele-

ments Zn (BRULAND, 1980) and Ba (CHAN et al., 1977) show strong correlations with silicate. Yet the Zn/Si and Ba/Si ratios in carefully cleaned opahne frustules taken from plankton tows are at least one order of magnitude lower than the water column ratios (COL- LIER and EDMOND, 1984). For the REE, which exhibit a relatively poor correlation with silicate (Fig. 41, one would expect the levels in opal to be even less signifi- cant.

On the other hand, the ionic radii of the trivalent REE cations, ranging from about 1.03 A (La) to 0.86 A (Lu) in sixfold ~~rdination, are very close to the 1.00 A radius of the sixfold coordinated Ca2’ ion (SHANNON, 1976). Solid solution theory would also favor incorporation of REE)+ cations in a calcium car- bonate lattice (PALMER, 1983). Of course all shells, whether opaline or calcareous, are formed by living organisms, so expectations based on equilib~um ther- modynamic principles may not hold.

PALMER (1983) recently reported REE concentra- tions in foraminifera. These were collected from sed- iment cores and carefully cleaned by physical methods and selective leaching techniques for removal of the detrital phase (clays) and ferromanganese coatings. Without the leaching procedure, the coatings would dominate the REE contents. The REEfCa ratios of clean forams were found to be low and in themselves can probably not sustain the high REE levels in deep waters (Table 2). However, the bulk REE/Ca ratio of shells settling through the water column may be higher due to adsorption of REE, either directly on the shell surfaces or on fe~oman~n~ coatings. In the latter case, the REE/Ca ratios of sedimentary foraminifera with coatings might serve as upper limits (Table 2). These ratios indeed exceed the AREE/AOS Alkalinity ratios in the water column. Thus a lesser amount of fe~omangan~ coatings on settling shells could pro- vide REE/Ca ratios exactly matching the required water column ratios. The single point REE maximum (Fig. 1) coinciding with a well defined sharp Fe maximum at 150 m depth (GORDON et al., 1982: LANDING, 1983) is compatible with the existence of such a coating. In this case the coating happens to dissolve again in the local suboxic zone (see above), while normally it would remain intact throughout the oxygenated water column.

It is irn~~ant to consider whether inco~oration of the REE in the uptake/dissoIution cycle may be wholly or partly responsible for the typical HREE enrichment of seawater. At first glance the REE patterns in surface waters and deep waters of both the Pacific (Fig. 2) and Atlantic (DE BAAR et nl., 1983a) are quite similar. Thus the inevitable fractionations with the REE series due to uptake in (calcareous) shells in themselves cannot fully explain the typical HREE enriched pattern. This is partly due to the inadequate REE/Ca ratios, but the actual fractionations are also minor given the ‘typical’ seawater pattern of cleaned foraminifera (PALMER,

1983). It is concluded that strictly biological cycling of the

REE by incorporation in shell crystal lattices cannot

1950 H. J. W. de Baar el al

‘1 . ATLANTIC

. PACING

.

l ,

7. ‘i

. ’ I c

_’ *

.’

_’

0 r-l-77-7-

50 100 ---77 T1 ,;~_,.>

St /&I kg-‘] Corrected Al&a/inity [/A equw tq-;:

FIG. 4. Lu VHXU silicate (left) and alkalinity (right) in the Northwest Atlantic Ocean (34”N, 58”W) and the Eastern Equatorial Pacific Ocean. Data for silicate and alkalinity from GEOSECS stations 120 (Si. Alk ). 344 (Si) and 345 (Alk) (BAINBRIDGE, 1981; BROECKER et al., 1982). Alkalinity data corrected for nitrate formation and normalized to 35% salinity. The effects of phosphate and silicate on the titration alkalinitie\ were deemed negligible

account for either the nutrient type distribution or the heavy REE enriched pattern in seawater.

Scavenging

If biological uptake is ruled out as the principal mechanism of REE transport, one has to invoke ad- sorptive scavenging for removal of REE from the water column. The apparent involvement with the cycling of skeletal material, then, results mostly from REE scavenging by biogenic particles. Two more or less in- dependent observations also favor control of REE by adsorptive scavenging onto settling particles (GOLD- BBRG, 1954; CRAIG. 1974): 1) the striking agreement of the shale normalized patterns (Fig. 2), including the Gd anomaly (DE BAAR ef al.. 1985a), with the predicted speciation in seawater (TURNER d al., t 98 1); and 2) the observed negative curvature in plots of light REE versus heavy REE in the deep water at various stations (DE BAAR et al, 1983a; KLINKHAMMER e[ al., 1983: PALMER, 1983). The element Cu is a particularly good analog to consider, because it too appears to be affected

both by a nutrient-like regeneration and by scavenging in deep water (BOYLE ef al., 1977: BRULAND, 1980). The similarity can be seen by comparing the VERTEX II Cu profile shown in Fig. 5 with the REF profiles (Fig. I).

Application of a one-dimensional scavenging modei (CRAIG, 1974) to the NADW yielded, as expected, :i trend in relative REE scavenging rate constants which is the inverse of the typical seawater concentration pattern (DE BAAR, 1983). The product of this rate con- stant + of each REE and its shale normalized concen- tration in seawater equals the shale normalized scav- enging rate $Rrr(CpcPv&Cshaia)REE of each element. This rate ideally has the same value for all REE. In other words, the average downward flux of’ REE ad- sorbed on settling particles is expected to exhibit a flat shale pattern (see also DE BAAR et al., 1985a). Unfor- tunately, the complex hydrography of the North AI- lantic invalidates some of the assumptions ofthe scav- enging model, and the results are somewhat uncertain. This problem also arises from stations in the Eastern Equatorial Atlantic (ELDERRELD and GREAVES, 1982: PALMER, 1983). The two South Pacific stations where

Table 2. PecificlAtlentlr ratios of concentration differences A = (surface - deep)

for REE, silicate and alkalinity+.

La Ce Pr Nd sm Eu Gd Tb HO Tm Yb I." S, kik -I

A./A. 0.7 -- I._-- 7 2.3 4 4.8 5 2.7 5.5 5 6 *

[AItgE/O.5AAlkl, 0.55 --- 0.6L 0.27 0.06 0.013 0.09 0.017 0.036 0.014 0.2 0.036 bREE/O.5Mlkl, 2.6 -- O.,7 -- 0.3 0.017 0.06 0.011 0.022 0.017 0.114 0.02 IREE/C~ (1) 0.19 0.13 -- 0.16 0.03 0.01 0.04 -- -- -- 0.02

[REEKal (2) 1.7 2.05 -- 1.6 0.34 0.08 0.36 -- -- -- 0.14

IREE/Cal (3) 4.5 *5 -- 4.8 4.2 I.> L.0 -- -_ -_ 1.2 _--I

WEE ratios scaLter considerably bur are. except for La and Pr. in the same rmgo a* ratios for Si end AIL. Titration alkalinity was corrected for nitrate formation using AOU normalized to 35'100 salinity.

- P&401 (BRMER. 1978) and Aleo shown are .-atios AREE/O.SMlk (x lOa) in esch ocean (after

ELDERFIELJI and GREAVES. 1982). Ratios (1) for cleaned forms (PAlHER, 1983) suggest that celcareoub skeletona cannot, on their am, account for the high REE levels in the. deep rater. Ratioa (2) for form lattice + Fe/M coating are mwh higher. Earlier data (3) of non-chemically leached forms ~DPL likely also includes some coatina material (ELDERFIELD et al., 1981).

REE in the oceans 1951

f :. \ l \

‘! \ i

3560 1

FIG. 5. Copper profile at VERTEX II.

rare earth data are available are not suitable either, because one is situated exactly over a ridge crest and one over a trench (KLINKHAMMER et al., 1983). The Pacific data set reported in this paper did not yield clear trends upon application of the scavenging model. It is hoped that analyses of a profile in the central North Pacific Ocean, currently in progress, will provide a data set more suitable for this type of modelling.

It is concluded that oceanic REE distributions are largely controlled by the interaction of scavenging and seawater speciation. The lightest element La (38% free ion; TURNER et al.. 1981) should then be scavenged the most efficiently and show the largest depletion in REE patterns of seawater. This holds true for the upper few hundred meters of the Atlantic water column (DE BAAR et al., 1983). but below that depth and through- out the Pacific water column La is strongly enriched (Fig. 2). For the combined data sets La and Sm are not as strongly correlated as some of the other REE, and the average La/Sm ratio is greater than the value for average shales (Fig. 6). Apparently La behaves rather independently from the other REE. Preferential ad- sorptive scavenging of the light REE, combined with release (desorption) of the light REE enriched fraction upon dissolution of the carriers at the seafloor, was proposed to explain the local La enrichments in the bottom waters at the Atlantic site (DE BAAR et al., 1983a). The large ionic radius of the La(III) cation (SHANNON, 1976) would also cause fractionations upon

incorporation into shells. if it were not for the fact that the latter process was deemed negligible (see above).

Ce oxidation and reduction

The behavior of Ce in the oceans is largely controlled by its oxidation/reduction chemistry. Exact reaction mechanisms are not known. but the equation

4 CeOz(s) + 12 H+ = 3 Ce3+ + Oz(g) + 6 Hz0 (5)

is useful for this discussion. Just like the other. strictly trivalent REE, the Ce3+ cation is affected by processes such as adsorptive scavenging, biogeochemical cycling and water mass circulation. However. the effects of its unique oxidation-reduction reaction can be singled out by definition of the Ce anomaly versrds neighboring elements La and Pr:

C&e* = 2 (Ce/Ce,hal,YCa/Ls~c + Pr/h.d (6)

where Ce* is the hypothetical concentration a strictly trivalent Ce would have. In contrast to the other REE. the concentrations of Ce are extremely low over the complete Pacific water column. Levels are less than half of those found in the Atlantic Ocean (Fig. 1). The Ce depletion is most developed at greater depth and is always more extreme than in the Atlantic Ocean (Figs. 3 and 7). The depletions in the Pacific range from about twofold (200 m) to almost 25-fold (3250 m) relative to shales. At the Atlantic station the change in oxidation state (Eqn. 5) approximately doubles the removal ef- ficiency for Ce in the deep water (DE BAAR et al., 1983a). It is not surprising that the Pacific Ocean. with less input of aeolian or rivet-me terrestrial matter. is more depleted. Global circulation may lead to deple- tions of Ce and simultaneous enrichments of the other REE in the deep Pacific versrls deep Atlantic waters (Fig. 3). The deep Ce depletion due to oxidative scav- enging is analogous to the dissolved Mn depletion in the deep waters (Fig. 8).

The Ce fractionation in well oxygenated waters would yield a positive Ce anomaly, or at least less of a Ce depletion, in the authigenic fraction of particles settling towards the seafloor. Upon regeneration of this relatively Ce enriched authigenic phase, the released REE fraction would produce higher values of Ce/Ce* in the water. This is exactly what is observed in the strong 02-minimum zone (200-750 m) at the Pacific station (Fig. 7). The effect of this regeneration would roughly correspond to the hatched area in the Ce/Ce* profile. This phenomenon is consistent with the posi- tive Ce anomalies in Atlantic surface waters (DE BAAR et al., 1983a), the sharp gradient of the Ce anomaly near the Atlantic seafloor (Fig. 3) and the enhanced Ce levels in anoxic waters of the Cariaco Trench (DE BAAR et al., 1985b). The Ce anomaly maximum roughly coincides with a nitrate anomaly at 250-300 m, and the onset of Ce reduction may be related to denitrification.

The broad Ce/Ce* peak is strikingly similar to the maximum of dissolved Mn in the same 02-minimum

1952 H. J. W. de Baar et al

Lo-9.lSm-51 (r = 0.87) Pr -i.iSm + 0.34 (f’O.91)

sholes La/ Sm = 5 84 sholes: Pr/Sm=143

I, ! / I, I *

0 2 4 6 8 & 0 2 4 6 a 40 Sm/mol/kg/ Sm(mo//kg]

- -- Nd j 4.9Sm + i.6 (r = 0.98) __- Tb=0.2iSm + 0016 tr = 098)

- - shales Nd/Sm= 5.26 - - shales Tb/Sm = 0 155

A ATLANTIC

l PACIFIC

FIG. 6. Concentrations of La, Pr, Nd and Tb plotted versus those of Sm for the Pacific (0) and Atlanttc (A) stations. Sm was chosen because it is roughly in the middle of the series, does not exhibit anomalies (in contrast to Ce, Eu and Cd) and is determined more precisely than Nd and Pr. The corresponding Eu-Sm relation is shown in Fig. 9.

zone (Fig. 8). It is conceivable that Mn oxide serves as a REE carrier. Upon dissolution of this carrier phase under low oxygen conditions, the REE are released. This interpretation is supported by observations in the anoxic Catiaco Trench, where most REE, and espe- cially Ce, exhibit a sharp increase just below the 02/ HtS interface (DE BAAR et al., 1985b). On the other hand, regeneration of a particulate organic carrier phase cannot be ruled out at either location. Yet, the high REE concentrations, usually with a positive Ce anom- aly, in ferromanganese nodules (ELDERFIELD et al.. I98 1; FLEET, 1984) also support uptake of REE by a Mn oxide phase.

In defining the Ce anomaly Ce/Ce* the neighboring elements La and Pr were chosen as reference elements to represent the strictly trivalent REE. As discussed above, however, La seems to act somewhat indepen- dently of the other REE, so it is desirable to examine comparisons with additional reference elements. Fig. 8 shows that plots of the Ce/Nd ratio and Ce/Sm ratio

yield essentially the same features as the Ce/C‘e* profiles (Fig. 6).

Very little is known about the exact mechanisms and thermodynamic constraints for the overall reaction (5). CARPENTER and GRANT (I 967) reported the dis- appearance of free Ce3+ during the oxidation reaction, with dramatically faster kinetics at pH 8 relative to pi-l 7 in seawater. Most probably an adsorption of the free Ce3+ cation precedes the oxidation reaction.

HIRANO and KOYANAGI (I 978) proposed hydro- lyzed intermediates in solution like Ce(OH)‘+ Ce(OH); and Ce(OH)! to be the dominant species in seawater, in contrast with the more recent model of TURNER et al. (198 1). This was supposed to explain the reported lower oxidation rates at lower pH (CAR- PENTER and GRANT, 1967) and the absence of Ce anomalies from river waters, as indeed no hydrolyzed Ce species occur in fresh water (pH 6) (TURNER et ai., 198 1; the limited data base for river water does indeed have a flat shale type pattern: MARTIN r>t rc! !976:

REE in the oceans 1953

FIG. 7. The Ce anomaly Ce/Ce* = 2 (Ce/CeW)/(La/& + Pr/Prw) as a function of depth at the Pacific and Atlantic stations. AIso shown is the profile of dissolved O2 at the Pacific site. Horizontal scales are logarithmic.

KEASLER and LOVELAND, 1982; and HOYLE et al., 1984). Yet this would also imply hydrolyzed inter- mediates for the adsorption of all other REE from sea- water unless Ce oxidation takes place in solution, a process which is unlikely from a thermodynamic point of view (see below). As a result, the heavy REE, which are more strongly hydrolyzed in seawater (TURNER et al., 198 I), would be removed more rapidly than the light REE.

In contrast. the generally held belief that heavy REE have longer oceanic residence times than the light REE

is consistent with adsorption of free REE3+ cations in- stead. Also the overall oxidation reaction (5) of Ce pro- duces protons and would still be favored at higher pH. The solid Ce(lV) phase is probably not simple CeOz but could very well be a hydroxide Ce(OH), , a hydrated crystalline form or, possibly, a solid solution.

The oceanic water column is an open system. often approaching a steady state, but definitely not in ther- modynamic equilibrium. Nevertheless simple ther- modynamic considerations may serve as a guide for predicting the direction in which reactions tend to go. Unless it is very strongly complexed. the Ce4+ ion is capable of oxidizing water to oxygen:

4 Ce4+ + 2 Hz0 = 4 Ce’+ + 4 H+ + O*(g)

log K = 8.6 (BAES and MESMER. 1976). (7)

In other words dissolved Ce4’ falls outside the stability field of water (GARREIS and CHRIST, 1965) and is vir- tually non-existent in aqueous solution. The above corresponds to

log (Ce3+/Ce4+f = 8.6 f pH - 0.25 fog ftoz. (8)

At pH 8.2 and a p02 of 0.2 atm, this ratio amounts to Ce3+/Ce4+ = 10”. This value may vary somewhat with ionic strength. At lower Qo2 (e.g.. in anoxic waters) the ratio would be even higher. For acidified seawater (pH = 2) with a *#Ce internal standard spike (as in our analyses) the ratio would still be around 10”.

From thermochemical data (BAES and MESMER, 1976) pure cerium dioxide should be very insoluble:

CeOz(s) + 4 H+ = Ce4’ + 2 Hz0

log KS = -8.16. (9)

Combination with Eqn. (7) yields

log [Ce3’] = 0.44 - 3pH - 0.25 log p02 ( 10)

LEGEND

0 Pocifii

A Atlantic

FIG. 8. The anomalous behaviour of Ce relative to Nd and Sm in the Pacific and Atlantic Oceans. The profiles are of similar shape as for the more conventionally defined Ce anomaly Ce/Ce* (Fig. 6). Note the similarity with the profile of dissolved Mn taken from LANDING (1983).

1954 H. J. W. de Baar et al.

for the overall reaction (5). The constants K and KS

were determined at very high Ce concentrations, in solutions of high acidity with different anionic com- position than seawater. Attempts could be made to take these effects, as well as the seawater speciation, into account. More important, however, is the notion that pure crystalline CeOz is almost certainly not the solid phase on marine suspended particles. Neverthe- less Eqn. (10) predicts that oxygenated seawater with [Ce3’] = I-100 pmol/kg would be oversaturated, in agreement with the observed preferential removal of Ce from the oceanic water column (Table 3). These numerical values are of limited validity. More signif- icant is the conclusion that lower pH and lower p0~ tend to bring Ce3+ into solution. It is interesting to note in Eqn. (10) that changes in pH have a larger effect than variations in p0,. This strong pH depen- dence may explain the apparent lack of Ce anomalies in fresh waters which are just as well oxygenated but generally have a lower pH than seawater.

Within the oceans, it appears that Ce is fractionated between marine environments with pO2 above a certain threshold level, probably about 0.00 l-0.0 10 atm, and marine environments with p0, below that threshold level. At higher pOz dissolved Ce(II1) is removed by formation of insoluble Ce(IV). As a result Ce tends to be depleted in open ocean waters and normal or en- riched in marine waters below the p02 threshold. Sur- faces of constant Ce anomaly (i.e.. constant Ce/Ce*) can be envisioned. In a hypothetical equilibrium ocean the no-anomaly surface (Ce/Ce* = I ) would coincide with the pO&teshold surface, the latter generally lying below the sediment water interface. However, the ocean is an open non-equilibrium system. Transport which is fast relative to redox kinetics will cause decoupling of the no-anomaly surface and the p02-threshold sur- face. For instance positive Ce anomalies were found in well oxygenated Atlantic surface waters (Fig. 7).

Matters are seemingly complicated by the observed stronger Ce depletion in the Pacific compared to the Atlantic Ocean. However, the aforementioned trans- port terms may also be seen as the ultimate driving force behind this interoceanic difference. In other words, going from the Atlantic into the Pacific basin the O2 threshold surface tends to intersect with lower and lower (more depleted) Ce anomaly surfaces (Fig. 9). With the Pacific water column generally more de-

Table 3. Predicted equilibrium conc~ntrarionr of ce” from equation (LO).

state of

S.¶t”rati0” DB DOZ loa (CFX.) in scaxater

8.2 1 atm -24.16 oversatur*t,cd 8.2 O.Zlatm -24.0 Wcr‘atur.tcd a.2 O.Olatm -23.66 Ovccsaturetcd 7 O.Ol~rm -2O.bb Ovcr‘aturatcd 2 0.21atm - 5.39 Undersaturated*

pleted it is more difficult to generate ‘normal’ (Ce/Ce* = 1) or enriched Ce levels by regeneration in reducing zones.

This continuous, kinetically controlled redistribution of Ce in the water column of the modem ocean. com- bined with the likelihood of active diagenesis and the strong pH dependence of the reaction ( IO). complicates (see also GRAF, 1978; HOLLAND. 1984) the use of Cc anomalies as paleoindicators of oxic v~‘r.ru.\ anoxic conditions in ancient oceans (FRYER, 1977: WRIGHT ef al.. 1984; LKJ and SCHMITT. 1984).

If the observations so far are indeed characteristic of the Pacific and Atlantic Oceans. then the Atlantic Ocean, with less of a Ce depletion, would appear to be the major recipient of the external terrestrial KEE input which is presumed to have a flat shale pattern (I.L’.. with no Ce depletion). This. combined with the fact that most of the world’s river water discharge drains into the Atlantic Ocean (MILLIMAN and MEADE. 1983). suggests that the dissolved river load is an importanr source of REE to the ocean. However. the relative im- portance of riverine versus aeolian sources of REE in the various ocean basins needs further exploration.

Eli

Under strongly reducing conditions Eu can he re- duced to the Eu2+ cation which has an ionic radius (SHANNON, 1976) very similar to that of the Ba*+ ion. This leads to preferential uptake of Eu in barites formed under reducing conditions. Continental barites closely associated with hydrothermal events indeed exhibit strong positive Eu anomalies; however most deep sea barites are not Eu enriched and therefore not neces- sarily of hydrothermal origin (GUICHARD ei ai, 1979). Fractionations of Eu during hydrothermal circulation have been inferred (DE BAAR ef al., 1983a) from such Eu anomalies, both in bar&es and in some, but not all, metalliferous deposits (COURTOIS and TREIIII.. I977. CORLISS el al.. 1978).

Recently strong positive Eu anomalies were indeed observed in hydrothermal fluids collected at the 13”N vent field on the East Pacific Rise (MICHARD PI al. 1983). Concentrations of REE were reported to bc about 1000 times those found in seawater. with AEu/ ASm ratios around 2.6 versus Eu/Sm = 0.25 m the open ocean (Fig. 10). Linear regression of the Atlantic data set alone leads to a slightly lower Xu/ASm = 0.22. very close to Eu/Sm = 0.21 for shales. the latter thought to be representative of average terrestriai inputs into the ocean basins. The Pacific ratio AEu/ ASm = 0.26 is slightly higher, although there is quite some overlap in the data. Although this difference is small and may not be significant, it is consistent with two major processes: 1) slower scavenging of the heavier element Eu, which in seawater is complexed more strongly than Sm. This would lead to a Eu enrichment in Pacific waters, the latter assumed to have a less direct terrigenous input, and 2) a relatively larger hydrother- mal input of REE into the Pacific Ocean due lo the generally faster spreading rates in that basin

REE in the oceans

FIG. 9. Sketch depicting the decoupling of Ce/Ce* surfaces from a p0, threshold level (typically at or below the sediment/water interface) due to transport terms. A major continental source (e.g., rivets with Ce/Ce* = 1) in Atlantic surface waters is assumed. Following the general circulation of the oceans the water becomes more and more Ce depleted as the oxidative removal of Ce exceeds its reductive regeneration. The west-east transect in the western North Atlantic exhibits positive Ce anomalies in surface waters resulting from a Ce enriched regenerative flux from reducing shelf sediments (DE BAAR et al., 1983).

The latter process is supported by the PacificfAtlan- tic comparisons (Fig. 3). At all depths Eu exhibits a small but distinct enrichment in the Pacific Ocean, relative to both its neighbors Sm and Gd in the REE series. There is some uncertainty due to the anomalous behaviour of Gd and due to shifts in speciation ex- pected from the lower carbonate concentrations in the deep Pacific Ocean (BROECKER and PENG, 1982). Nevertheless the ratios Eu/Gd and Eu/Tb also appear to be higher in the Pacific than in the Atlantic Ocean, rather than lower as would be expected from the sce- nario given in (I) above. KL~WHAMMER et al. ( 1983) also show a distinct positive Eu anomaly in a com-

parative Pacific/Atlantic plot for samples at 2500 m depth (their Fig. 5b). Both their Pacific stations are exactly at sites of hydrothermal activity, with the 2500 m sampling depth close to ridge crest depth. Of course this Eu enrichment is not necessarily representative of their samples at other depths or, for that matter, the Pacific Ocean in general. Also the Atlantic sample at 2500 m which they used as a reference happens to exhibit a fairly distinct Eu depletion (Eu/Sm = 0.19; ELDERF~ELD and GREAVES, 1982) versus shales (Eu/ Sm = 0.21). This tends to enhance the apparent Eu enrichment of the Pacific samples. Most likely such variability of the Eu/Sm ratio is a reflection of varia-

1956 H. J. W. de Baar PI al

LEGEND

NW Attbntlc 8 EE Pacific - Eu=025Sm-003 k=O964)

C) NW Atlantlc --- Eu = 022Sm - 0.02 (r'O.996)

l EE F'ncific -- Eu = 0.26Sm + 0.05 (r'O.9691

a Carioco Trench

v EE Atlonhc

0 2 4

Sm/pmo/ skg.,j 8 d0 12

FIG. 10. The Eu/Sm relationship in the Atlantic and Pacific Oceans and the Cariaco Trench (prelimmal? results taken from DE BAAR, 1983). The two data points with low Eu/Sm ratio in the Eastern Equatonal Atlantic Ocean (ELDERFIELD and GREAVES, 1982) probably arise from a similar Eu depletion of the original terrestrial source (RAHN. 1976) rather than from a fractionation within the ocean basin.

tions in terrestrial source material around the average shale ratio. Such variations obscure the small hydro- thermal signal that may be present.

MASS BALANCE AND RESIDENCE TIMES

The riverine (and aeolian) input of terrestrial ma- terial into the oceans is expected to exhibit an average crustal abundance, i.~. shale-type. REE pattern. The limited data base for river waters and aerosols does. by and large, exhibit a flat shale pattern (MARTIN ei ai., 1976; RAHN, 1976: KEASLER and LOVELAND,

1982; HOYLE et a/., 1984). but this needs to be verified by further studies. Removal of REE from the water column to the sediment is dominated by scavenging on settling particles, although a minor portion may also be. removed by incorporation in the crystal matrix of skeletal material. Most all skeletal material is regen- erated either in the water column or from surface sed- iments (BROECKER and PENG, 1982). Only a very small fraction of the skeletal material escapes dissolution and is buried permanently. Concurrent burial of REE truly incorporated within the crystal matrix. with a typical seawater pattern, is probably negligible in a mass bal- ance. It is worth noting, however. that this phase would by and large record the seawater pattern and Nd iso- topic signature at the time of deposition (PALMER. 1983).

We consider the net removal of REE from the oceans to be dominated by adsorption on settling particles. This process may in fact consist of several removal- regeneration cycles. The effect of one such cycle can be expressed by a mean oceanic scavenging rate con- stant $J. The trend in REE scavenging rate constants was suggested to be roughly the inverse of the typical seawater pattern (see above and DE BAAR (31 al.. 1985a).

If this is the case, then for all REE the shale normalized scavenging rate #REE(CPawnur/C-)mE would be equal. For the strictly trivalent REE the constant $ would be proportional to the percentage free ion in solution. For instance the percentage free La” ion IS about twice that of the free Nd3+ ion (TURNER c't ui. i r)X I J, and La is expected to be removed twice as fast as Nd. Fol Ce the scavenging constant $ is enhanced considerabl? by oxidation, yet Ce is also about one order- of mag- nitude depleted in seawater and the shale normahzed removal rate $c,(C,,,.,&~&, would agaIn equal that of the other REE. Thus the Ce depletion in sea- water does not need to be complemented h) C’e en.. riched biogenic or authigenic deposits (PIPER. 1974.

HOLLAND, 1984; SHAW and WASSERBURC;. 1985 ). This issue has also been discussed by DE BAAR (‘I ui ( 1985aj.

After a few scavenging-dissolution cycles each REE atom will eventually be permanently removed from the water column and buried in an authigemc deposit. The corresponding mean oceanic residence time foi the water column reservoir

T [years] = Reservoir [pmol/kgl = Reservoir/Shales

Flux [pmol/kg y] Flux/Shales

111:

is comprised of a term Flux/Shales with the same value for each REE, while the term Reservoir/Shales. and hence z, would be about five times higher for heavy REE than for La and 5-10 times lower for Ce than for La.

It is concluded that the net sedimentation of authi- genie REE deposits would, on average, match the flat shale pattern of continental inputs. This does not nec- essarily imply that the REE are evenly distributed within marine sediments. In analogy with the behavior

REE in the oceans 1957

ofZ3*Th and *3’Pa (ANDERSON et 01.. 1983a,b) the light REE. which are scavenged more rapidly, may even- tually be buried preferentially in pelagic sediments, as indeed observed by THOMSON et al. ( 1984). The heavy REE. more stable in solution, are more affected by lateral transport and may become enriched in other depositional environments, for instance hemipelagic sediments. Preferential deposition in pelagic sediments as compared with hemipelagic sediments may be very pronounced for Ce, which is scavenged very rapidly in the open ocean and most likely regenerated in ocean margin sediments (DE BAAR el al., 1983a). Of course matters are complicated by local variations in REE continental inputs, rather different from the simple uniform source distributions of ““rh and “‘Pa. On a more local scale the sedimentary imprint may depart even further from the average distribution as a result of enhanced scavenging due to a nepheloid layer or exposed surfaces, notably ferromanganese nodules (ELDERL~ELD et al., 198 1) and phosphatic ftsh debris (SHAW and WASSERBURG, 1985). Postdepositional diagenesis would lead to further REE fractionation and redistribution, probably in favor of phases rich in iron or phosphorous (ELDERFIELD el al., 198 I). The differ- ent types of biogenic and authigenic minerals do indeed exhibit a wide variety of REE patterns, with both pos- itive and negative Ce anomalies and both enrichments and depletions of the heavy REE (ELDERF~ELD et al., 1981: ELDERP~ELD and GREAVES, 1981; PALMER. 1983; FLEET, 1984; THOMSON et al., 1984). The global average net ~imentation rate should nevertheless match the presumed flat shale pattern of the terrestrial input. Construction of meaningful mass balances for authigenic sediments would prove the point, but is he- yond the scope of this paper.

The above outline of a mass balance for the water column is quite simple, but it is much more difficult to provide meaningful estimates of 7, the corresponding mean oceanic residence time. In order to maintain the interoceanic ‘43Nd/‘44Nd difference (PIEPGRAS and WASSERBURG, 1982) one must invoke a ?Nd not greatly exceeding the interoceanic mixing time, which has an upper limit of about 1000 yrs. (BROECKER and PENG, 1982; STUIVER et al.. 1983). This gives an upper limit of 500 yrs. for TL~ and 4000 yrs for 7Lu. The latter places the heavy REE in the same range as Cu (BOYLE

et al., 1977). Because Ce is roughly an order of mag- nitude depleted in seawater relative to La, a value of 50 yrs. is the upper limit on TV. On the other hand, the REE correlations (except Ce) with silicate and al- kalinity suggest that the REE are recycled at least once and possibly several times. From this it follows that TREE will be somewhat longer than the 500-1000 yr. oceanic mixing time.

The above residence time in principle matches the residence time relative to external sources, mostly rivers and aerosols. (The hydrothermal source is deemed negligible on a global scale (MICHARD ef al., 1983).) Exact estimates are difficult because of great uncer- tainties in such inputs. ELDERFIELD and GREAVE~

(1982) made an attempt and arrived at values in the 200-2000 yrs. range, not far removed from the limits estimated above.

Additional data are needed to constrain the above mass balance for the modern ocean, assumed to be in a steady state over time scales of 104-IO5 years. Also, at the longer time scales often assigned to the formation of various deposits, notably ferromanganese nodules (HUH and Ku, 1984) and crusts (SEGL et al., 1984) the assumed steady state does not necessarily hold.

CONFUSIONS

I) Distributions of all dissolved REE, except Ce, in ocean waters are dominated by their internal cycling within the ocean basins. The ultimate external sources (rivetine, aeolian, hydrothermal) and sinks (authigenic minerals) generally have little impact on the spatial distributions. Analogies with distributions of other properties within the oceans suggest that the REE as a group are affected by two simultaneous processes: I ) cycling similar to that of opal and calcium carbonate; and 2) adsorptive scavenging on settling particles. The latter mechanism is also supported by the parallels be- tween REE(II1) speciation in seawater and the seawater REE abundance pattern.

II) Oxidation and reduction reactions dominate both Ce and Mn in the ocean basins. Ce appears to have much shorter residence times than the other REE. Its oceanic dist~bution, with Ce depleted in the Pacific versus the Atlantic Ocean, is more strongly affected than the other REE by external tenigenous sources, which for all REE are predominantly into the Atlantic rather than the Pacific Ocean.

III) The results so far strongly suggest that the REE in seawater are affected by bioge~hemi~ mechanisms for removal and regeneration. This is also demonstrated by the interoceanic ‘43Nd/‘MNd isotopic differences (PIEFGRAS and WASSERBURG, 1982) which can only be maintained by removal processes. Therefore, neither the shale-normalized REE pattern nor the ‘43Nd/‘44Nd ratio are expected to be reliable as water mass tracers.

IV) By analogy with other elements (Si, Fe, Mn) the open ocean distributions of the REE suggest that the reactive pool (about 10% ?: PIPER, 1974) of REE in marine sediments is strongly affected by diagenesis, in contrast with the conclusions of FLEET (1984). En- hanced REE levels in anoxic waters of the Cariaco Trench (DE BAAR et al., 1985b) also suggest the mo- bilization of REE in reducing marine sediments. Stud- ies of REE diagenesis, supported by careful physical and chemical separation methods (PALMER, 1983) are required for understanding the incorporation of REE in many authigenic deposits.

~c~ffow~~~ge~enf~-The authors are greatly indebted to Bill Landing for assistance during sampling and for previews of his data and manuscripts. Without the loyal cooperation of John Bernard, William Fecych and Kwan Kwok at the MIT Nuclear Reactor Laboratory, this work would not have bee.n possible. Rebecca Belastock, Aian Fleer, Fred Frey, Hugh

1958 H. J. W. de Baar Ed al

Livingston, Pieter Nella, Peter Sachs and Deborah Shafer provided help, advice and encouragement during various stages of the project. Discussion with Harry Elderheld. Gary Klink- hammer and Ed Boyle led to major improvements in the manuscript. We gratefully acknowledge the constructive crit- icisms of all three reviewers. This research was supported by National Science Foundation Grant GCE79-23322. Depart- ment of Energy Contract DE-ASO2-76EV03566 and Office of Naval Research Contract NO00 14-82-C-00 19 NR 083-004.

Edirorlal handling: S. E. Calvert

REFERENCES

rare-earth elements. Ph.D. Thests. MlT/WHCil Jotnt PFO- gram in Oceanography. 278 p.

DE BAAR H. J. W. (1984) Neutron activation analysis of rare earth elements in seawater. Pr’roc. o/‘In//. Srmposrrttn oti rhe Use and Development c?/ LOM and Medjum Nu\- Rc- search Reaclors. Special Volume Atomkemenergre-Kerr]-- technik. supplement 44, 702-709.

DE BAAR H. J. W.. BACON M. P. and BREWER I’. G. I lY83a) Rare-earth distributions with a positive Ce anomaly in the Western North Atlantic Ocean. hhrurc 301. 324-327.

DE BAAR. H. J. W.. BACON M. P. and BREWER P. G. (I 983b) Ram earth elements in the eastern equatorial Pactfic Ocean. EOS 64, 1030.

DE BAAR H. J. W., BREWER P. G. and BACON hl. P. ( 1985a Anomalies in rare earth distributions in seawater: Cd and

ANDERSON R. F.. BACON M. P. and BREWER P. Cr. (1983a) Tb. Geochim. Cosmochrm. .Icrcr 49, I955- 106 ; Removal of 2)orh and *“Pa from the onen ocean. Ear//r DE BAAR H. J. W.. GERMAN C. R.. ELDERRELD H. and Bn- Planet. Sci. LeIt. 62, 7-23.

ANDERSON R. F.. BACON M. P. and BREWER P. G. (1983b) Removal of *qh and *rIPa at ocean margins. Eurfh Planef. Sci. Leu. 66, 73-90.

BACON M. P.. BREWER P. G., SPENCER D. W.. MURRAY T. W. and GODDARD T. (1980) Lead-210, polonium-210. manganese and iron in the Cariaco Trench. Deep-Sea Res. 27A, 119-135.

BAES C. F. and MEISMER R. E. (1976) The Hydrolysis ofCu~- ions. Wiley. 489 p.

BAINBRILIGE A. E. ( 198 I ) GEOSECS Allantlc Expedition, Vol. I, IDGE/NSF. US Govt. Printing Office. Washington. 121 p.

BALASHOV Y. A. and KHITROV L. M. ( I96 1) Distribution 01 the rare earths in the waters of the Indian ocean. Geochem Inr. 9, 877-890 (translation).

BOYLE E. A., SCLATER F. R. and EDMOND J. M. (1977) The distribution of dissolved copper in the Pacific. Earth Planei. Sci. Lell. 37, 38-54.

BREWER P. G. (1978) Direct observation of the oceanic CO: increase. Geophys. Res. Lett. 5,997-1000.

BROECKER W. S.. CRAIG H. and SPENCER D. ( 1982) tiEO- SECS Pacific Expedition. Vol. 3. BIDE/NSF, US Govt. Printing Office. Washington. 137 p.

BROECKER W. S. and PENG T. H. (1982) Tracer.s IIJ the Sea. Eldigio Press of Columbia University. Palisades, NY. 690 p.

BROENKOW W. and KRENZ R. (1982) Oceanographic results from the VERTEX II particle interceptor trap experiment off Manzanillo, Mexico. MLML Technical Publ. 82- 1, Moss Landing, Calif.

BRULAND K. W. (I 980) Oceanographic distributions of cad- mium, zinc, nickel and copper in the North Pacific. Eurrh Planet. Sci. Letl. 47, 176-198.

CARPENTER J. H. and GRANT V. E. (1967) Concentration and state of cerium in coastal waters. J. Mar. Res. 25(3),

228-238. CHAN L. H., DRUMMOND D.. EDMOND J. M. and GRANT B.

(1977) On the barium data from the Atlantic GEOSECS expedition. Deep-Sea Res. 24, 6 13-649.

CLINE J. E. and RICHARDS F. A. (1972) Oxygen dehcient conditions and nitrate reduction in the eastern tropical North Pacific Ocean. Limnol. Oceanogr. 17, 885-900.

COLLIER R. and EDMOND J. (1984) The trace element geo- chemistry of marine biogenic particulate matter. Prog Oceanog. 13, 113-199.

CORLW J. B., LYLE M.. DYMOND J. and CRANE K. (1978) The chemistry of hydrothermal mounds near the Galapagos Rift. Earth Planet Sri. Len. 40, 12-24.

COURTOIS C. and TREUIL M. ( 1977) Distributton des terres rates et de quelques elements en trace dans les sediments recents des fesses de la Mer Rouge. Chem. Geol. 20, 57- 72.

CRAIG H. (1974) A scavenging model for trace elements in the deep sea. Earfh Planer. Sci. Leti. 23. l49- 159.

DE BAAR H. J. W. (1983) The marine geochemistry of the

CON M. P. (1985b) Rare earth distributions In the Canaco Trench. Terra Cognila. 5, 188.

DYMOND J,, LYLE M., FINNEY B.. PIPER D. Z.. MURPHY k.. CONARD R. and PISIAS N. ( 1984) Ferromanganese nodules from MANOP Sites H. Sand R-Control of mineralogical and chemical composition by multiple accretionan: pro- cesses. Geochim. Cosmochim. Acra 48, 93 I-950.

EDMOND J. M. (1974) On the dissolution of carbonate and silicate in the deep ocean. Deep&u Re.v 21, 4555480.

ELDERF~ELD H. and GREAVES M. J. ( I98 1) Negative ceriunr anomalies in the rare earth element patterns of oceantc ferromanganese nodules. Earth Plunel. .Sci I.cvt 55, I63 170.

ELDERFIELD H. and GREAVES M. J. ( 1982) ‘I he rare earth elements in seawater. Nafure 296, 2 14-Z 19,

ELDERFIELD H.. HAWKESWORTH C. J.. GREAV~S M. J. and CALVERT S. E. (1981) Rare earth element geochemrstry of oceanic ferromanganese nodules and associated sediments. Geochim. Cosmochim. Acto 45, 5 13-528.

FLEET A. J. (1984) Aqueous and sedimentary geochemrstr: of the rare earths. In Rare Earth Elemenr Georhemisrrt~ (ed. P. HENDERSON), pp. 343-373. Elsevier. Amsterdam

FRYER B. J. (1977) Trace element geochemistry of the So- koman Iron Formation. Can. J Eurrh Su. 14, 1 i38- I6 IO

GARREL~ R. M. and CHRIST C. L. ( 1965) .SoIukvu. Mineruic und Equilibria. Harper and Row. New York.

GOLDBERG E. D. (1954) Marine Geochemistry 1. (‘hemrcal scavengers of the sea. J. Geol. 62, 249-265.

GOLDBERG E. D., KO~DE M.. SCHMITT R. A. and SMIIII R. H. (1963) Rare-earth distributions in the marme envr- ronment. J. Geophys. Res. 68, 4209-42 I?.

GORDON R. M.. MARTIN J. H. and KNACJ~R C. .2. ( IY82) Iron in Northeast Pacific waters. Nmrre 299, 6 1 ! -6 12

GRAF J. L. (1978) Rare earth elements, iron formatrons and seawater. Geochim. Cosmochim. .4czu 42. 1845- 1850.

GUICHARD F., CHURCH T. M.. TREUIL M. and JAI-~EZI( H. (1979) Rare earths in barites: distribution and effects on aqueous partitioning. Geochim. Cwno&m~ . IL-/U 43. 9P3- 997.

HASKIN M. A. and HASKIN L. A. (1966) Rare earths m bu- ropean shales: a redeterminatton. Screncc 154, 507-509.

HAYES D. W. (1969) A study of the distribution of the lan- thanide elements in the Gulf of Mexico using neutron ac- tivation analysis. Ph.D. thesis. Texas A&M Unrversity. viir + 168 p.

HEMING R. F. and RANKIN P. C. ( 1979) Ce-anomalous lavas from Rabaul caldera, Papua New Guinea. Gcochrm (‘III- mochim. Acta43, 1351-1355.

HENDERSON P. ( 1984) Rare Earth Elemenr Gcwhem~.srr~~ Elsevier. Amsterdam. 5 10 pp.

HIRANO S. and KOYANAGI T. (1978) Study on the chemical forms of radionuclides in seawater-t. Chloride, sulfate and hydroxide complexes of IaCe. J Ocwn Sot, Jup 34,269- 275.

HOCDAHL 0. T., MELXIN S. and BOWEN V. T. i i%)o Neu- tron activation analysis oflanthanide elements rn seawater

REE in the oceans 1959

ln Truce Inorgunics in Wuter (ed. R. A. BAKER), pp. 308- 325. Adv. Chem. Series 73, ACS, Washington. DC.

HOLLAND, H. D. (1984) The Chemical Evolution sf rhe At- mosphere and Oceans. pp. 405-406. 498-506. Princeton University Press.

PALMER M. R. (1983) Rare earth elements and Nd and Sr isotopes in the Atlantic Ocean. Ph.D. Thesis, Univ. of Leeds. 284 pp.

HWD D. W. (1966) Rare earth distributions in waters of the Gulf of Mexico. In The Chemistry and Anal.vsis qf Trace Mefals in Seawater. Section IV. Texas A&M Project 276; AEC Contract No. AT-(40-I)-2799.

HOYLE J., ELDERF~ELD H., GLEDHILL A. and GREAVES M. (I 984) The behavior of the rare earth elements during mix- ing of river and sea waters. Geochim. Cosmochim. Acra 48, 143-149.

PIEPGRAS D. J. and WASSERBURG G. J. (1980) Neodymium isotopic variations in seawater. Earth Planet. Sci. Lefr. 50. 128-138.

P~EPGRAS D. J. and WASSERBURG G. J. (1982) Isotopic com- position of neodymium in waters from the Drake Passage. Science 217, 207-2 14.

PIEPGRAS D. J. and WASSERBURG G. J. (1983) Influence of the Mediterranean outflow on the isotopic composition of neodymium in waters of the North Atlantic. J. Geophvs. Res. 88, 5997-6006.

HUH C.-A. and Ku T. L. (1984) Radiochemical observations on manganese nodules from three sedimentary environ- ments in the North Pacific. Geochim. Cosmochim. Acfa 48, 951-963.

JACOBS L. A. (1984) Metal geochemistry in anoxic marine basins. Ph.D. Thesis. Univ. ofWashington. Seattle. pp. 99- 121.

KEASLER K. M. and LOVELAND W. D. (1982) Rare earth element concentrations in some Pacific Northwest rivers. Earth Planet. Sci. Lett. 61, 68-72.

KLINKHAMMER G., ELDERFIELD H. and HUDSON A. (1983) Rare earth elements in seawater near hydrothermal vents. Nature 305, 185-188.

KNAUER G. A., MARTIN J. H. and GORDON R. M. (1982) Cobalt in Northeast Pacific waters. Nature 297, 49-5 1.

KNAUSS D. and KU T. L. (1983) The elemental composition and decay-series radionuclide content of plankton from the East Pacific. Chem. Geol. 39, 125-145.

KOLESOV G. M.. ANIKYEV V. V. and SAVENKO V. S. (1975) On the geochemistry of rare earth elements in the waters of the Black Sea. Geochem. Int.. 82-88 (in English).

KRAUS K. A. and NELSON F. (1955) Anion exchange studies of the fission products. Proc. 1st U.N. Intern. Conf Peacejiil Uses Atomic Energy 7, 113- 125.

LANDING W. M. (1983) The biogeochemistry of manganese and iron in the Pacific Ocean. Ph.D Thesis, Univ. Calif.. Santa Cruz. 202 p.

PIEPGRAS D. J., WASSERBURG G. J. and DASCH E. J. (1979) The isotopic composition of Nd in different ocean masses. Earth Planez. Sci. Len. 00(2), 223-236.

PIPER D. Z. (1974) Rare earth elements in the sedimentary cycle: a summary. Chem. Geoi. 14,285-304.

RAHN K. A. ( 1976) The chemical composition of the atmo- spheric aerosol. Graduate School Oceanography. Univ. Rhode Island, Technical Report: xi + 265 p.

SCHREIBER H. D.. LAUER H. V. JR. and THANAYASIRI T. (1980) The redox state of cerium in basaltic magmas: an experimental study of iron-cerium interactions in silicate melts. Geochim. Cosmochim. Acla 44, 1599-16 12.

SEGL M., MANCINI A., BONANI G.. HOFMANN H. J.. NESSI M., SUTER M., WOLFLI W., FRIEDRICH G.. PLUNGER W. L.. WIECHOWSK~ A. and BEER J. (1984) “Be-dating of a manganese crust from Central North Pacific and impli- cation for ocean paleocirculation. Nature 309. 540-543.

SHANNON R. D. (1976) Revised effective ionic radii and sys- tematic studies of interatomic distance in halides and chal- cogenides. Acfa Crvsr. A32, 75 l-767.

SHAW H. F. and WASSERBURG G. J. (1985) Sm-Nd in marine carbonates and phosphates: Implications for Nd isotopes in seawater and crustal ages. Geochim. Cosmochrm. .4cta 49,503-5 18.

Lru Y. G. and SCHMITT R. A. (1984) Chemical profiles in sediment and basalt samples from DSDP Leg 74. Hole 525A. Walvis Ridge. Init. Repts. DSDP 74, 7 13-730.

MARTIN J. H., KNAUER G. A. and GORDON R. M. (1983) Silver distributions and fluxes in north-east Pacific waters. Nature 305, 306-309.

MARTIN J. H. and KNAUER G. A. (1984) VERTEX: man- ganese transport through oxygen minima. Earth Planef. Sri. Len. 67, 35-47.

MARTIN J. M., HOCDAHL 0. and PHILIPP~T J. C. (1976) Rare earth element supply to the ocean. J. Geophys. Res. 81(1X), 3119-3124.

MASUDA A. and IKEUCHI Y. (1979) Lanthanide tetrad effect observed in marine environment. Geochem. J. 13, 19-22.

MICHARD A., ALBAREDE F., MICHARD G., MINSTER J. F. and CHARLOU J. L. (I 983) Rare-earth elements and ura- nium in high-temperature solutions from East Pacific Rise hydrothermal vent field ( 13”N). Nature 303, 795-797.

MILLIMAN J. D. and MEADE R. H. (1983) Worldwide delivery of river sediment to the oceans. J. Geol. 91, I-2 I.

MURPHY K. and DYMOND J. (1984) Rare earth element fluxes and geochemical budget in the eastern equatorial Pacific. Nature 307,444-447.

SHIGEMATSU T., TABUSHI M.. AOKI T., FUJINO 0.. NISHI-

KAWA Y. and GODA S. (1967) Activation analysis of Lan- thanum and Europium in seawater and lake water. Bull. Chem. Res. Kvoto Univ. 45. 307-317.

SPURN R. V. (1965) Rare earth distributions in the marine environment. Ph.D. Thesis, MIT. Cambridge, MA. 165 p.

STRELOW F. W. E. (1963) The separation of uranium from scandium, yttrium. rare earths. thorium, beryllium. mag- nesium, copper, manganese. iron. aluminium and other elements by cation-exchange chromatography. Joemoal van die Suid-Afiikaanse Chemiese Instituut XVI, 38-47.

STUIVER M. P., QUAY P. D. and OSTLUND H. G. (1983) Abyssal water carbon-14 distribution and the age of the world oceans. Science 219, 849-85 I.

THOMSON J.. CARPENTER M. S. N.. COLLEY S., WILSON T. R. S., ELDERF~ELD H. and KENNEDY H. (1984) Metal accumulation rates in northwest Atlantic pelagic sediments. Geochim. Cosmochim. Acta 48, 1935-1948.

TURNER D. R., WHITRELD M. and DICKSON A. G. (1981) The equilibrium speciation of dissolved components in freshwater and seawater at 25°C and I atm pressure. Gee- chim. Cosmochlm. Acfa 45, 855-88 I.

WRIGHT J.. SEYMOUR R. S. and SHAW H. F. (1984) REE and Nd isotopes in conodont apatite: variations with geological age and depositional environment. Cieol. Sot. Amer. Spec. Pup. 196, 325-340.

NAGATSAKU S.. SUZUKI H. and NAKAJIMA K. (1971) Acti- WYRTKI K. (1967) Circulation and water masses in the eastern vation analysis of lanthanide elements in natural water. equatorial Pacific Ocean. Inl. J. Oceunol. Llmnol. l(2), I 17- Radiorsoropes 20(7), 305-309. 147.