Geochimica et Cosmochimica Acta, Vol. 63, No. 3/4, pp.

363–372, 1999
Copyright © 1999 Elsevier Science Ltd
Pergamon Printed in the USA. All rights reserved
0016-7037/21801 $20.00 1 .00
PII S0016-7037(98)00279-8

Rare earth elements in seawater: Particle association, shale-normalization,

and Ce oxidation
DIA SOTTO ALIBO and YOSHIYUKI NOZAKI*
Marine Inorganic Chemistry Division, The Ocean Research Institute, University of Tokyo, Nakano-ku, Tokyo 164, Japan

(Received June 2, 1998; accepted September 2, 1998)

Abstract—Dissolved (,0.04 mm, not ,0.4 mm) and total acid-soluble concentrations of rare earth elements
(REEs) and yttrium were measured by using ICP mass spectrometry in the seawaters obtained from various
depths in the western North Pacific near Japan. The difference, i.e., acid-soluble particulate fraction, was found
to be small, 2–5% for all tri-valent light and middle REEs and less than 1% for heavy REEs and yttrium. The
high particulate fraction of 31% for Ce is consistent with its predicted oxidation state of tetra-valence.
Elevated particulate fraction of all REEs was found within ;80 m above the bottom due to contribution of
flocculated resuspended particles. The vertical profiles of REE(III)s show smoothly increasing convex curves
with depth similar to those reported previously. Dissolved Ce concentration decreases from ;6 pmol/kg near
the surface to a minimum at 2.5 pmol/kg around 400 m where the North Pacific Intermediate Water penetrates,
and then approaches to nearly constant value of ;4 pmol/kg below 800 m. Particulate Ce concentration
significantly increases from the surface to 200 m depth suggesting oxidation of Ce(III) to Ce(IV) and
subsequent scavenging in the upper water column. However, there is no evidence in our data showing that Ce
oxidation is continuously taking place even in the deep sea.
Shale-normalized patterns of dissolved REEs were examined in detail, based on three datasets of composite
shales available in the literature. Distinctively positive La and only slightly positive Gd anomalies were
identified together with well-documented negative Ce-anomaly as common features of seawater. These
anomalies systematically change with depth. Rapid changes occur in the upper several hundred meters
suggesting that their distributions are largely governed by ocean circulation and biogeochemical
cycling. Copyright © 1999 Elsevier Science Ltd

1. INTRODUCTION probe in understanding scavenging processes of particulate

The purpose of this paper is to present the complete and precise
dataset of “dissolved (,0.04 mm)” rare earth elements in
seawater and to extract characteristic features from them. Dur-
ing the last few decades, there have been quantum leap of
knowledge on the concentrations and distributions for trace
elements in the ocean (e.g., Nozaki, 1997). Meanwhile, to
obtain reliable data, marine geochemists have currently devoted
their major effort in eliminating contamination from numerous
sources during sampling, handling, storing and analysis. Prior
to 1970 when the GEOSECS program started, those basic
requirements were seldom met satisfactorily and real features
of oceanic elemental distribution largely remained vague. In
those days, filtration of water samples could induce contami-
nation and, therefore, samples were often analyzed without
filtration. For most elements with their mean oceanic residence
times longer than 103–104 years, this is permissible because the
results with and without filtration agree within the analytical
uncertainty of measurements (typically a few to 5%). However,
for heavy metals like Al, Mn, Fe, Pb, Ce, and Th which have
short residence times of less than 103 years, filtration obviously
becomes important. We present here the vertical profiles of
both dissolved and total acid-soluble REEs and Y in the west-
ern North Pacific near Japan.
Increasing attention has recently been paid on the marine
geochemistry of REEs, because they serve as; (1) a useful
*Author to whom correspondence should be addressed (nozaki@ori.u-
Tokyo.ac.jp).
363
matter (Elderfield and Greaves, 1982; Sholkovitz et al, 1985;
Byrne and Kim, 1990); (2) natural analog of actinides and
transuranics such as Ac, Th, Pu, Am, Cm and Cf in the marine
environment (Nozaki, 1984; Bowen et al., 1980; Livingston
and Anderson, 1983); and (3) tracers of water masses and
ocean circulation (Piepgras and Wasserburg, 1982; German
et al., 1995; Zhang and Nozaki, 1996). In this region, Zhang
and Nozaki (1998) have previously reported 5 vertical pro-
files of REEs and discussed behavior of REEs at the ocean
margin. In the earlier work, only unfiltered and acidified
samples were analyzed. Blank control for Ce measurement
was difficult for that time. These problems were overcome in
the present study.
2. METHODS
Seawaters were obtained at the location of 34°419N, 139°549E in the
slope of Sagami Trough about 35 miles south of Boso Peninsula, Japan,
during the R/V Tansei-Maru Cruise KT 97-2 on March 12, 1997. The
station is located in the Kuroshio Current regime (Fig. 1) and the
bottom depth was 2584 m. Sampling was made by using a CTD/multi-
rossete array mounted on ordinary 10 L Niskin bottles. The inside wall
of the bottle was coated with Teflon and the inner spring was replaced
by Dow Corning Silastic tube. After retrieval of the sampling bottles,
waters were transferred to pre-cleaned 5 L plastic cubic containers and
within 4 h after the sampling, seawaters were filtered through a hollow-
fiber membrane (Millipore HF-400) with an effective surface area of
400 cm2 and a pore size of 0.04 mm. The water samples (both
filtered and unfiltered) were acidified to pH , 1.5 with ultrapure HCl
(TAMAPURE-AA-100) and then transferred to the laboratory at the
Ocean Research Institute, University of Tokyo.
364 D. S. Alibo and Y. Nozaki

3. RESULTS AND DISCUSSION

3.1. Comparison of Filtered and Unfiltered Data

The data on filtered and unfiltered waters are given in Table

Fig. 1. Temperature-salinity diagram. KSW; the Kuroshio surface
water, NPIW; the North Pacific intermediate water, and NPDW; the
North Pacific deep water.
Rare earth elements and Y were measured with an ICP mass spec-
trometer (Yokogawa PMS-2000) according to the method described in
Shabani et al. (1992) and Zhang and Nozaki (1996). The method
includes a quantitative 200-fold preconcentration of REEs and Y from
1L seawater by solvent extraction using a mixture of HDEHP 1
H2MEHP in heptane and a flow injection ICPMS measurement of the
concentrates using indium as an internal standard. The precision of
replicate measurements is less than 4% for light and middle REEs and
2% for heavy REEs and yttrium. The method was calibrated with the
standard solutions provided by Spex Industries Inc. which are accurate
to 60.5%. The isobaric interferences of BaO on 151Eu and 153Eu were
suitably corrected using the isotopic abundunce of 135Ba and 137Ba and
were found to be negligibly small in our solvent extracts. The PrO
interference on 157Gd was ;1.5% of 141Pr signal but this correction is
also small since the Gd/Pr ratios in the water samples vary by only 20%
from 1.3 to 1.6. Our ICPMS measurements show excellent agreement
with those by isotope dilution method of Piepgras and Jacobsen (1992)
as described in Zhang and Nozaki (1998).
1 and 2. Note that “dissolved” concentration is defined here as
those passed through a filter of 0.04 mm in pore size. When
compared with previous data based on 0.4 mm-membrane fil-
tration, submicron particles between 0.04 and 0.4 mm in diam-
eter are removed by the present procedure. Thus, the data
obtained here are closer to “truly dissolved” concentration. The
results based on unfiltered waters should be regarded as “acid-
soluble” total concentrations. The vertical profiles of “dis-
solved” and “acid-soluble” Y, La, Ce, Pr, Gd and Lu are shown
in Fig. 2. The tri-valent REEs and Y show smooth convex
curves which increase with depth. The profiles are similar to
each other, but, in detail, systematically different according to
atomic number. The distribution patterns are in agreement with
those reported previously by Zhang and Nozaki (1998).
The difference of filtered and unfiltered data can be ascribed
to the contribution of particulate matter which can liberate
“dissolved” REEs and Y by acidification to pH 5 1.5. REEs
adsorbed on the surface of particles and contained in biogenic
carbonates (though small) are presumably included in this
fraction, whereas those contained in detrital minerals may be
excluded. This acid-soluble particulate (ASP) fraction is gen-
erally small, less than 5% (Fig. 3), for REE(III)s and Y except
for two samples at 2500 m and 2576 m which are close to the
bottom (2584 m; Fig. 2). The high particulate contribution near
the bottom can be ascribed to the high abundance of flocculated
aggregates which were visually seen by video camera obser-
vation preceding the CTD/multi-rosette cast at the same site.
The particulate fraction amounted to 31 6 17% (Table 3)
which is exceptionally higher than those of other REEs. This is
because Ce(III) is oxidized to Ce(IV) in seawater presumably
through bacterial mediation (Moffet, 1990) as will be discussed
later. The particulate fractions of light and middle REE(III)s are
systematically higher than those of heavy REEs, suggesting

Table 1. Dissolved concentrations (in pmol/kg) of Y and rare earth elements in seawater.

Depth Temp.
(m) (°C) Salinity Y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

5 — — 71.4 7.29 5.55 1.27 6.64 1.49 0.42 2.09 0.39 2.54 0.69 2.07 0.31 1.75 0.27
49 16.48 34.677 76.2 8.44 6.31 1.39 7.12 1.56 0.41 2.17 0.39 2.80 0.71 2.23 0.31 1.81 0.27
99 14.85 34.619 83.3 10.57 6.11 1.54 7.65 1.69 0.47 2.51 0.42 2.96 0.76 2.37 0.34 2.09 0.33
150 13.27 34.531 88.2 10.40 5.66 1.57 7.93 1.78 0.49 2.57 0.43 3.09 0.82 2.51 0.37 2.24 0.36
200 12.27 34.480 93.5 11.80 3.89 1.66 8.16 1.84 0.48 2.64 0.46 3.35 0.90 2.83 0.41 2.54 0.40
299 9.48 34.319 109.6 14.98 2.65 1.92 9.49 1.95 0.52 3.03 0.51 3.76 1.01 3.31 0.48 3.04 0.51
400 7.74 34.232 121.1 18.31 2.49 2.36 11.45 2.21 0.61 3.36 0.57 4.37 1.19 4.00 0.59 3.82 0.63
458 6.75 34.198 130.2 20.02 2.68 2.56 12.62 2.38 0.66 3.72 0.61 4.48 1.25 4.10 0.61 3.94 0.67
599 5.02 34.158 151.8 24.09 3.31 2.98 14.41 2.75 0.75 4.25 0.68 5.14 1.41 4.62 0.69 4.51 0.78
800 3.88 34.278 175.7 29.09 4.67 3.58 17.09 3.27 0.89 4.75 0.86 5.98 1.69 5.67 0.86 5.66 1.01
1000 3.37 34.361 193.4 31.85 4.70 3.87 18.19 3.65 0.91 5.19 0.83 6.33 1.78 6.07 0.92 6.07 1.06
1199 2.80 34.443 200.6 32.17 3.95 4.07 18.83 3.65 0.98 5.51 0.87 6.76 1.91 6.52 1.00 6.73 1.19
1502 2.42 34.512 216.0 35.72 4.67 4.37 20.65 3.92 1.08 5.70 1.06 7.50 2.15 7.38 1.13 7.66 1.38
1748 2.21 34.552 220.0 36.46 4.10 4.50 21.25 3.86 1.05 6.33 0.99 7.71 2.18 7.65 1.15 7.95 1.40
1997 1.99 34.590 227.3 36.09 3.74 4.53 21.47 4.14 1.11 6.47 1.03 7.91 2.23 7.67 1.17 8.13 1.42
2248 1.87 34.611 227.7 37.59 3.94 4.87 22.33 4.19 1.19 6.58 1.06 8.17 2.30 7.93 1.22 8.21 1.45
2400 1.78 34.628 236.6 38.17 4.08 4.99 23.20 4.47 1.22 6.79 1.11 8.25 2.31 7.80 1.22 8.23 1.45
2500 1.70 34.640 237.8 38.69 4.32 5.10 23.66 4.42 1.22 6.76 1.14 8.43 2.33 7.95 1.22 8.40 1.46
2576 1.68 34.644 234.5 39.25 3.55 5.23 24.53 4.63 1.27 6.95 1.14 8.46 2.37 8.05 1.24 8.50 1.48
 Rare earth elements in seawater 365

Table 2. Total acid-soluble concentrations (in pmol/kg) of Y and rare earth elements in unfiltered seawater.

Depth
(m) Y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

5 71.1 7.64 5.65 1.27 6.58 1.52 0.45 2.19 0.35 2.63 0.68 2.02 0.28 1.73 0.26
49 75.3 8.99 6.88 1.47 7.23 1.58 0.43 2.16 0.44 2.76 0.72 2.17 0.30 1.90 0.27
99 79.3 9.24 6.63 1.52 7.56 1.69 0.47 2.39 0.42 2.91 0.75 2.35 0.34 1.97 0.32
150 88.8 11.06 6.67 1.69 8.59 1.86 0.50 2.70 0.45 3.42 0.84 2.64 0.38 2.36 0.38
200 94.3 12.64 6.82 1.86 9.21 1.88 0.54 2.76 0.49 3.38 0.89 2.92 0.44 2.61 0.43
299 110.1 15.56 4.65 2.10 9.93 2.11 0.57 3.15 0.53 3.83 1.02 3.35 0.51 3.15 0.52
400 123.9 18.04 4.02 2.35 11.44 2.25 0.60 3.49 0.56 4.21 1.13 3.79 0.55 3.43 0.58
458 134.7 20.45 4.67 2.66 12.75 2.43 0.64 3.75 0.61 4.55 1.26 4.09 0.58 3.78 0.67
599 155.7 25.30 5.29 3.30 15.76 2.91 0.78 4.45 0.75 5.30 1.44 4.74 0.70 4.57 0.78
800 177.3 30.01 5.90 3.78 17.91 3.45 0.94 4.91 0.88 6.22 1.70 5.69 0.83 5.68 0.97
1000 192.0 30.54 5.49 3.89 17.85 3.41 0.95 5.29 0.81 6.23 1.68 5.82 0.86 5.62 1.00
1199 201.4 32.77 6.51 4.15 19.32 3.76 1.01 5.74 0.92 7.00 1.90 6.60 0.99 6.74 1.20
1502 218.9 36.76 7.58 4.68 21.68 4.14 1.11 6.03 1.11 7.61 2.17 7.42 1.13 7.75 1.39
1748 222.4 36.82 6.78 4.62 21.57 4.13 1.16 6.63 1.03 7.85 2.20 7.52 1.17 7.80 1.40
1997 236.5 37.02 6.09 4.74 22.43 4.24 1.17 6.60 1.05 8.13 2.23 7.65 1.14 7.85 1.39
2248 237.4 37.99 6.46 5.03 23.20 4.56 1.25 6.81 1.11 8.30 2.27 7.62 1.17 8.02 1.42
2400 233.4 38.76 7.54 5.22 24.44 4.72 1.27 6.92 1.l4 8.52 2.33 7.94 1.21 8.07 1.43
2500 244.1 43.26 12.72 6.20 28.35 5.65 1.47 7.96 1.29 9.58 2.53 8.43 1.29 8.71 1.54
2576 252.2 44.66 15.51 6.70 30.36 5.84 1.54 8.47 1.34 9.59 2.58 8.62 1.31 8.73 1.53

fractionation is taking place during scavenging. Although direct
comparison is difficult due to different procedures employed,
the pattern (Fig. 3) is similar to those obtained by acetic acid
digestion of marine particles (Sholkovitz et al. 1994;
Tachikawa et al., 1997; Lerche and Nozaki, 1998).
3.2. Shale-Normalized REE Pattern
The shale-normalized REE patterns of the acid-soluble par-
ticulate fraction for the bottom two samples are shown in
Figure 4 together with those of ambient waters at 1200 m and
1500 m depths. In the figure, the REE concentrations of Post
Archean Australian Sedimentary Rocks (PAAS) compiled by
Taylor and McLennan (1985) were chosen for normalization,
because it might be more relevant for samples collected in the
Pacific Ocean than those compiled from the North American
and European continents. The bottom two samples show a
pattern with marked middle REE enrichment and no negative
Ce anomaly. The pattern is clearly different from the flat
pattern in the ambient waters (Fig. 4) and closer to that of the
underlying sediments around this region (Lerche, 1997). How-
ever, the decreasing trend from middle to heavy REEs (Fig. 4)
is not seen in the surface sediments which show an almost flat
pattern (Lerche, 1997). This suggests that some REE fraction-
ation may be taking place in the acid treatment of the sample,
or naturally at the sediment-water interface and during early
diagenesis of sediments.
Shale-normalized patterns of dissolved REEs are shown in
Fig. 5 which are similar to those published in other oceanic
areas (see, Byrne and Sholkovitz, 1996 for recent review).
Since the present study has provided the reliable data for all

Table 3. Acid-soluble particulate fraction (%) of Y and rare earth elements in unfiltered seawater.

Depth
(m) Y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

5 20.3 4.6 1.7 0.1 20.9 1.7 5.6 4.8 29.6 3.7 21.6 22.5 29.7 20.8 27.1
49 21.2 6.1 8.4 5.5 1.6 1.1 3.7 20.4 11.2 21.2 1.4 22.6 21.7 4.6 21.1
99 25.1 214.5 7.9 21.3 21.2 0.1 20.6 25.1 20.2 21.7 20.8 20.6 20.3 26.0 21.6
150 0.7 6.0 15.2 7.2 7.7 4.2 2.4 4.8 4.6 9.6 2.1 5.1 2.6 5.0 5.6
200 0.8 6.7 42.9 10.8 11.4 2.6 9.9 4.4 6.5 1.0 21.4 3.4 7.3 2.5 6.5
299 0.4 3.7 43.1 8.4 4.4 7.3 9.9 4.0 3.8 1.8 0.7 1.0 5.3 3.5 2.9
400 2.2 21.5 38.0 20.6 0.0 2.0 22.3 3.8 22.0 23.7 25.2 25.8 27.5 211.3 27.9
458 3.3 2.1 42.5 3.7 1.0 2.3 22.5 0.7 0.2 1.5 0.3 20.2 24.8 24.1 21.4
599 2.5 4.8 37.3 9.5 8.6 5.4 4.6 4.6 9.4 2.9 2.3 2.7 1.3 1.2 0.4
800 0.9 3.1 20.8 5.2 4.6 5.1 5.2 3.4 1.5 4.0 0.6 0.4 23.9 0.3 24.0
1000 20.8 6.1 22.9 10.6 8.3 3.6 14.6 11.7 8.1 8.6 4.8 6.1 4.0 2.9 4.6
1199 0.4 1.8 39.3 1.9 2.6 2.9 2.8 4.0 5.6 3.5 20.8 1.2 20.3 0.2 0.7
1502 1.3 2.8 38.3 6.6 4.7 5.4 2.7 5.4 5.1 1.5 1.3 0.5 0.4 1.2 1.1
1748 1.1 5.9 42.5 7.5 6.4 11.3 13.9 9.2 8.7 6.8 5.9 3.3 6.2 3.2 5.1
1997 3.9 2.5 38.6 4.4 4.3 2.3 5.2 2.0 1.7 2.7 0.1 20.2 22.4 23.6 21.8
2248 4.1 1.0 39.0 3.1 3.7 8.0 5.2 3.4 4.2 1.5 21.2 24.1 25.0 22.4 22.1
2400 21.4 1.5 46.0 4.4 5.1 5.2 4.2 1.8 2.3 3.2 0.9 1.7 20.2 21.9 21.7
2500 2.6 10.5 66.1 17.7 16.6 21.9 17.4 15.1 11.7 12.0 8.0 5.7 5.9 3.5 5.0
2576 7.0 12.1 77.1 21.9 19.2 20.7 17.4 14.8 11.7 8.1 6.6 4.9 2.6 3.3
366 D. S. Alibo and Y. Nozaki
Fig. 2. Vertical profiles of Y, La, Ce, Pr, Gd, and Lu.
REEs including mono-isotopic Pr, Tb, Ho and Tm which can
not be obtained by the isotope dilution method, more detailed
arguments on the REE pattern may be possible. However, note
that different shale values have currently been used for nor-
Fig. 3. Mean acid-soluble particulate fraction (%) and its standard
deviation (1s) for REEs and Y. Two samples from 2500 m and 2576 m
were omitted from the calculation due to contribution of resuspension
from the bottom.
malization in the literature. Although the major features are
similar to each other, the normalized values vary as much as
;20% particularly for heavy REEs and therefore, detailed
arguments on the REE pattern need special caution on the
choice of shale values.
Following the previous workers (DeBaar et al., 1985; Ber-
tram and Elderfield, 1993), we calculated anomalies based on
the equation given by,
REE n/REE *n 5 2@REE# n/~@REE# n21 1 @REE# n11! (1)
where [REE] is shale-normalized REE and n 5 1, 2, 3 . . . in
the order of the lanthanide series. Anomalies for the lightest La
(n 5 1) and heaviest Lu (n 5 15) were also calculated from
the nearest neighboring REE(III)s based on the family equa-
tions,
La/La* 5 @La#/~3@Pr# 1 2@Nd#! (2)
and
Lu/Lu* 5 @Lu#/~3@Yb# 1 2@Er#! (3)
Thus, either positive or negative anomaly is defined as the
REE/REE* value greater or less than unity, respectively. Since
REEsn21 for Pr (n 5 3) and Sm (n 5 6) are tetra-valent Ce
showing strong negative anomaly itself and naturally non-
 Rare earth elements in seawater 367

here. However, they are not very different from those given
above.
The calculation indicates distinctively positive La and neg-
ative Ce anomalies irrespective of seawater samples and dif-
ferent shale-normalization (Fig. 6). The anomalies for REEs
heavier than Pr are small and largely scatter over both positive

Fig. 4. PAAS-normalized pattern of acid-soluble particulate (ASP)

REEs in the water near the bottom and ambient seawater.

existent Pm, respectively, we similarly employed the following

equations,

Pr/Pr* 5 2@Pr#/~3@Nd# 2 @Sm#! (4)

and

Sm/Sm* 5 3@Sm#/~@Nd# 1 2@Eu#! (5)

These REE anomalies were calculated by using normalized

values relative to averages of numerous North American, Eu-
ropean and Russian shale composite (Av. shale) compiled by
Piper (1974) and adopted by DeBaar et al. (1985), of 40 shales
mainly from North America (NASC; Haskin et al., 1968;
Gromet et al., 1984), and of PAAS (Taylor and McLennan,
1985). Other REE datasets such as those reported by Goldstein
and Jacobsen (1988) for NASC and by Haskin and Haskin
(1966) for European shales are incomplete and hence not used

Fig. 6. Anomalies calculated based on three different shale values;

PAAS (a), Average shale (b) and NASC (c). See text for their sources
Fig. 5. PAAS-normalized pattern of dissolved REEs. and the method of calculation.
368 D. S. Alibo and Y. Nozaki
Fig. 7. Vertical profiles of La-, Ce-, and Gd-anomaly.
and negative side depending upon samples and the shale values
used for normalization, casting doubt on their significance. For
example, the positive Ho anomaly in the NASC-normalization
(Figure 6c) is not seen in other normalizations, and such is the
case for Lu/Lu*. Nevertheless, as already pointed out by De-
Baar et al. (1985), the positive Gd anomaly appear to be a
common feature irrespective of normalization problem. The
depth profiles of La/La*, Ce/Ce* and Gd/Gd* are shown in Fig.
7. The values of La/La* may have large uncertainty because
they were calculated by extrapolation from Pr and Nd. This is
a reason why the magnitude of La-anomaly varies considerably
depending upon the choice of the shale values. Nevertheless,
they are very pronounced at greater than 2 throughout the water
column, showing profiles of either decreasing or constant with
depth. Anomalous behavior of La in seawater has been noted
for some time. For example, La, like heavy REEs, correlates
much better with Y and dissolved Si than other light REEs
(Zhang et al., 1994; Zhang and Nozaki, 1996). The positive La
anomaly may be caused by unusually high stability of La in
seawater in the lanthanide series which may be related to
absence of 4f electron, leading to it’s solution chemistry being
different from others. Byrne and Kim (1993) suggests that the
solubility of La phosphate is significantly greater than those of
other REE-phosphate compounds, although REE-phosphate
complexation may not be dominant in seawater (Byrne and
Sholkovitz, 1996). Interestingly, La/La* is positively correlated
with Ce/Ce* (Fig. 8a).
The Ce/Ce* profiles do not change significantly by the
choice of the shale values. The profiles show a sharp decrease
from 0.4 at the surface to less than 0.1 below 400 m and agree
with those obtained previously in the North Pacific (Piepgras
and Jacobsen, 1992; Zhang and Nozaki, 1998). The Ce defi-
ciency in seawater is ascribed to oxidation of Ce(III) to Ce(IV)
which is subjected to preferential scavenging over REE(III)s.
This aspect will be discussed in more detail later.
The Gd/Gd* profiles based on PAAS and average shale are
similar to each other, whereas that of NASC is systematically
shifted to more positive side than the other two (Fig. 8). All
show systematic increase with depth. The presence of positive
Gd anomaly in seawater is somewhat controversial. DeBaar et
al. (1985) first pointed out positive Gd and negative Tb anom-
alies based on instrumental neutron activation analysis (INAA)
of Atlantic waters. However, their analysis appeared to be not
very precise, since some Ce values obtained for the same
waters were found to be incorrect by Sholkovitz and Schneider
(1991) who employed more precise isotope dilution thermal
ionization mass spectrometry (ID-TIMS). Although it is not
clear in the NASC-normalization (Fig. 6c), the positive Gd
anomaly is generally accompanied by negative Tb anomaly
(Zhang and Nozaki, 1998; Fig. 6). Therefore, if one ought to
calculate from Eu and Dy instead of Tb (which can not be
determined by ID-TIMS), the Gd anomaly can hardly be seen.
Obviously however, the positive Gd anomaly obtained here
is not an artifact depending upon the choice of the shale-
normalization (Fig. 6). It may be due to improper use of
average Gd and Tb values in shales compiled by previous
workers because it is likely that they have uncertainties as
much as 20%. Therefore, we increased Gd and reduced Tb
values in PAAS by 20% and then calculated Gd/Gd* and
Tb/Tb* in the same manner described above. The results are
compared in Fig. 9. Although the Gd and Tb anomalies are
shifted toward 1.0 (normal value), their vertical profiles show-
ing increasing (for Gd) and decreasing (for Tb) trends with
depth remain the same. There is a positive correlation between
Gd/Gd* and Ce/Ce* suggesting that their depth variations may
be controlled by the oceanographic processes. Thus, we con-
clude that the Gd and Tb anomalies in seawater are not an
artifact of shale-normalization but real.
Masuda and Ikeuchi (1979) and DeBaar et al. (1985) sug-
gested that the positive Gd anomaly is related to the unusually
high stability of Gd in seawater resulting from the half-filled 4f
electron in its orbital shell which is the so called “tetrad effect”.
This would result in relatively passive scavenging of Gd over
neighboring Eu and Tb, leaving Gd behind in solution. Al-
though the acid-soluble particulate fraction (Fig. 3) does not
show any significant anomaly at Gd, it is not surprizing because
 Rare earth elements in seawater 369

Gd anomaly is only 20% compared to a factor of ;5 for Ce ;5000 m by scavenging and settling of particulate matter (see
anomaly. The accompanied negative Tb anomaly (Fig. 7c and Nozaki et al., 1981 for detail discussion).
8) is also difficult to explain by the tetrad effect. Rather, it On the contrary, the oceanic distributions of REE(III)s are
seems that the Gd and Tb anomalies, together with La and Ce largely controlled by biogeochemical cycling and ocean circu-
anomalies, are controlled by some kinetic mechanism of ocean lation as indicated by their good correlation with dissolved Si.
circulation (water mass effect) and biogeochemical cycling of Thus, La and Pr (or Nd) concentrations in the water increase
nutrients and some trace metals (Nozaki, 1997). due to regeneration from particles along the pathway of deep
water much like the case of dissolved Si. As a consequence,
even without any removal of Ce from the deep water, Ce-
3.3. Does Ce Oxidation Proceed in the Deep Sea? anomaly would evolve with time due to preferential regenera-
tion of REE(III)s over Ce in the deep water. Thus, there is no
Since the first measurements of REEs in seawater by Gold-
berg et al. (1963), the negative Ce anomaly has long been
recognized to occur due to oxidation of Ce(III) to insoluble
Ce(IV). This Ce oxidation is taking place relatively rapidly in
the shallow water environment through bacterial mediation in
the absence of dissolved Mn(II) (Moffet, 1990, 1994). Re-
cently, Tachikawa et al. (in press) have shown that the positive
Ce anomaly in suspended particulate phase may be developed
by adsorption of REE(III)s on to the particulate MnO2 surface
and then oxidation of Ce(III) to Ce(IV) at shallow depths, and
subsequent preferential release of the light REEs over Ce(IV) at
deeper depths by desorption. Our data obtained here are con-
sistent with this scheme. The dissolved Ce concentration
sharply decreases from 5.6 pmol/kg at the surface to 2.6
pmol/kg at 300 m (Fig. 2). In contrast, the particulate Ce
increases from 0.1 pmol/kg at the surface to 2.9 pmol/kg at
200 m and then decreases to less than 2 pmol/kg down to
1000 m (Fig. 10a). This pattern suggests that the adsorption and
oxidation of Ce is largely taking place in the surface water
where biological activity is high. As a consequence, the nega-
tive Ce anomaly is developed with increasing depth (Fig. 7).
These observations strongly support that Ce oxidation is indeed
taking place in the upper ocean (Sholkovitz and Schneider,
1991). Below 1000 m, the acid-soluble particulate Ce concen-
tration ranges between 2 and 3 pmol/kg with no systematic
trend except for the samples close to the bottom where resus-
pended sediments become significant.
Recently, German et al. (1995) have suggested that Ce
oxidation gradually takes place even in the deep sea. Their
argument is based on the progressive evolution of Ce anomaly
in the deep sea as dissolved Si concentration increases along
the pathway of the ocean circulation. However, as first dem-
onstrated by Tachikawa et al. (in press), it seems unlikely. First
of all, available data do not show systematic decrease of dis-
solved Ce concentration along the trajectory of the deepwater,
as required by German’s hypothesis. Furthermore, in analogy
to 230Th(IV), the oceanic distribution of Ce(IV) is more likely
controlled by vertical reversible scavenging process (Nozaki et
al., 1981; Bacon and Anderson, 1982) with a mean oceanic
residence time much shorter than the oceanic mixing time
(;103 y). 230Th has a mean oceanic residence time of only 30
yrs (Nozaki et al., 1981). The ratio of acid-soluble particulate
Ce to dissolved Ce is nearly constant at ;0.6 below 1000 m
(Figure 10b) as expected by equilibrium exchange of Ce be-
tween seawater and particles. The corresponding particulate
fraction (;40%; Table 3) of Ce is approximately equal to that
of 230Th. These suggest that the mean oceanic residence time of
Ce is of the order of 50 –100 yr assuming that all Ce is supplied
to the surface ocean and transported vertically to the bottom of Fig. 8. Correlation diagrams for La, Ce, Gd, and Tb anomalies.
370 D. S. Alibo and Y. Nozaki

Fig. 9. Vertical profiles of Gd and Tb anomalies. Open symbols in (a) and (b) are based on calculations when PAAS-Gd
value is increased by 20%, and PAAS-Tb value is reduced by 20%.

Fig. 10. Vertical profile of particulate Ce (a) and particulate Ce/dissolved Ce ratio (b).
 Rare earth elements in seawater 371

Table 4. Mean oceanic residence times of REEs estimated from particular, t(REE) for heavy REEs are either unreliable or
particulate fraction. could not be estimated since the calculated mean particulate
fraction is close to zero or negative. Nevertheless, t(Ce) 5
Particulate Mean oceanic residence time
fraction* 50 –130 yr shows an excellent agreement with previous esti-
Element (%) (y) (y) mates from analogy with 230Th given above. Since the uncer-
tainty of the particulate fraction for Y is smaller than Ho, the
Y 0.8 6 2.2 5100** — 2030 values based on Y may be preferred in comparison with those
La 2.5 6 4.9 1630 — 650
Ce 30.9 6 14.7 130 — 50 of Ho. The mean residence times of light REE(III)s such as La
Pr 5.1 6 3.7 800 — 320 and Pr are considerably longer than that of Ce. This is consis-
Nd 4.3 6 3.5 950 — 380 tent with the conclusion regarding the cause of Ce-anomaly
Sm 4.2 6 2.8 970 — 390 given earlier. The mean residence time of La (1630 yr), which
Eu 5.0 6 4.9 820 — 320
is the longest among those of the light and middle REEs, is also
Gd 3.7 6 3.6 1100 — 440
Tb 3.6 6 5.0 1130 — 450 consistent with positive La-anomaly in the seawater.
Dy 2.7 6 3.4 1510 — 600
Ho 0.6 6 2.5 6800 — 2700**
Acknowledgments—We would like to thank Captain Hayashikawa,
Er 0.6 6 3.1 6800 — 2700
officer and crew of R/V Tansei-Maru for their help in the sampling. We
Tm 20.5 6 4.7 Not estimated
also thank the scientific party of the KT 97-2 cruise for their collabo-
Yb 20.3 6 4.3 Not estimated
ration. The manuscript was significantly improved by the comments
Lu 20.1 6 4.2 Not estimated
provided by two anonymous reviewers. This work is supported by the
Ministry of Education, Science, Sports, and Culture, Japan through
** Excluding the two samples near the bottom.
Grant-in Aid No. 07404051 to the University of Tokyo (Y. Nozaki,
* According to Nozaki et al. (1997).
Principal Investigator).

plausible evidence to support that Ce oxidation continues in the
deep sea.
3.4. Mean Oceanic Residence Times of REEs
Nozaki et al. (1997) have estimated the mean oceanic resi-
dence times of Y and Ho to be 5100 y and 2700 y, respectively.
Although the amounts of those elements in the ocean can be
accurately determined by measuring their oceanic profiles,
quantification of riverine and atmospheric fluxes is not an easy
task due to complex estuarine reactions and kinetic solubiliza-
tion of aerosols. Thus, the mean residence times given above
probably have an uncertainty of factor of 2–3. Nevertheless,
using those values and the data obtained here, we can estimate
the mean residence times of other REEs as follows.
First, we assume that REEs are thoroughly removed by
reversible scavenging and vertical settling of particulate matter
(Bertran and Elderfield, 1993; Jeandel et al. 1995) analogous to
that of 230Th (Nozaki et al., 1981). Zhang and Nozaki (1998)
have shown that removal of REEs at the oceanic boundary is
negligible. We further assume that particle association by re-
versible exchange is equal to the acid-soluble particulate frac-
tion obtained here (Table 3 and 4). In reality, suspended, fine
particles do not settle effectively by themselves, but it is also
assumed that they are exchangeable with large, fast sinking
particles through aggregation and disaggregation (Nozaki et al.,
1987). If these conditions are met for REEs, then the following
relationship holds,
t ~REE! 5 @F p~Y!/F p~REE!# 3 t ~Y!
where Fp is the acid-soluble particulate fraction. Similar equa-
tion holds for holmium as well. Using t(Y) 5 5100y and
t(Ho) 5 2700y and the mean particulate fraction (Table 4), we
estimated the values for t (REE). The values are given in Table
4. Since the particulate fraction has considerable uncertainties,
the calculated residence times must also have large errors in
association with the standard deviation of the mean values. In
REFERENCES
Bacon M. P. and Anderson R. F. (1982) Distribution of thorium
isotopes between dissolved and particulate forms in the deep sea. J.
Geophys. Res. 87, 2045–2056.
Bertram C. J. and Elderfield H. (1993) The geochemical balances of the
rare earth elements and neodymium isotopes in the oceans. Geochim.
Cosmochim. Acta, 57, 1957–1986.
Bowen V. T., Noshkin V. E., Livingston H. D., and Volchok H. L.
(1980) Fallout radionucleotides in the Pacific Ocean: vertical and
horizontal distributions, largely from GEOSECS stations. Earth
Planet Sci. Lett. 49, 411– 434.
Byrne R. H. and Sholkovitz E. R. (1996) Marine chemistry and geo-
chemistry of the lanthanides. In Handbook on the Physics and
Chemistry of Rare Earths. Vol. 23 (ed. K. A. Gschneider Jr. and L.
Eyring) 497–592, Elsevier.
Byrne R. H. and Kim K.-H. (1990) Rare earth element scavenging in
seawater. Geochim. Cosmochim. Acta 54, 2654 –2656.
Byrne R. H. and Kim K.-H. (1993) Rare earth precipitation and
coprecipitation behavior: the limiting role of PO32
4 on dissolved rare
earth concentrations in seawater. Geochmim. Cosmochim. Acta 57,
519 –526.
DeBaar H. J. W., Brewer P. G., and Bacon M. P. (1985) Anomalies in
rare earth distribution in seawater: Gb and Tb. Geochim. Cosmo-
chim. Acta 49, 1961–1969.
DeBaar H. J. W., Brewer P. G., Bacon M. P., and Bruland K. W. (1985)
Rare earth elements in the Atlantic and Pacific Oceans. Geochim.
Cosmocnim. Acta, 49, 1943–1959.
Elderfield H. and Greaves M. J. (1982) The rare earth elements in sea
water. Nature 296, 214 –219.
German C. R., Masuzawa T., Greaves M. J., Elderfield H., and Edmond
J. M. (1995) Dissolved rare earth elements in the Southern Ocean:
Ce oxidation and the influence of hydrography. Geochim. Cosmo-
chim. Acta, 59, 1551–1558.
Goldberg E. D., Koide M., Schmitt R. A., and Smith R. H. (1963)
Rare-earth distributions in the marine environment. J. Geophys. Res.
(6) 68, 4209 – 4217.
Goldstein S. J. and Jacobsen S. B. (1988) Rare earth elements in river
waters. Earth Planet. Sci. Lett. 89, 35– 47.
Gromet L. P., Dymek R. F., Haskin L. A., and Korotev R, L. (1984)
The “North American Shale Composite”: Its compilation, major and
trace elements characteristics. Geochim. Cosmochim. Acta 48,
2469 –2482.
Haskin M. A. and Haskin L. A. (1966) Rare earths in European shales:
A redetermination. Science 154, 507–509.
Haskin L. A., Frey F. A., Schmitt R. A., and Smith R. H. (1968)
372 D. S. Alibo and Y. Nozaki
Meteoritic, solar, and terrestrial rare earth distributions. In Physics
and chemistry of the Earth. Pergamon Press.
Jeandel C., Bishop J. K., and Zindler A. (1995) Exchange of ND and
its isotopes between seawater, small and large particles in the Sar-
gasso Sea. Geochim. Cosmochim. Acta 59, 535–547.
Lerche D. (1997) Fractionation of yttrium and holmium in the marine
environment. Master thesis, University of Tokyo.
Lerche D. and Nozaki Y. (1998) Rare earth elements of sinnking
particulate matter in the Japan Trench. Earth Planet. Sci. Lett. 159,
71– 86.
Livingston H. D. and Anderson R. F. (1983) Large particle transport of
plutonium and other fallout radionucleotides to the deep ocean.
Nature 303, 228 –231.
Masuda A. and Ikeuchi Y. (1979) Lanthanide tetrad effect observed in
marine environment. Geochim. J. 13, 16 –22.
Moffett J. W. (1990) Microbially mediated cerium oxidation in seawa-
ter. Nature 345, 421– 423.
Moffett J. W. (1994). The relationship between cerium and manganese
oxidation in the marine environment. Limnol. Oceanogr. 39, 1309 –
1318.
Nozaki Y. (1997) A fresh look at element distribution in the North
Pacific Ocean. Eos Trans. 78, 221.
Nozaki Y. (1984) Excess 227Ac in deep ocean water. Nature 310,
486 – 488.
Nozaki Y., Horibe Y., and Tsubota H. (1981) The water column
distributions of thorium isotopes in the western North Pacific. Earth
Planet. Sci. Lett. 54, 203–216.
Nozaki Y., Yang H.-S., and Yamada M. (1987) Scavenging of thorium
in the ocean. J. Geophys. Res. 92, 772–778.
Nozaki Y., Zhang J., and Amakawa H. (1997) The fractionation be-
tween Y and Ho in the marine environment. Earth Planet Sci. Lett.,
148, 329 –340.
Piepgras G. J. and Wasserburg G. J. (1982) The isotopic composition
of neodymium in water flowing through the Drake Passage. Science
217, 207–214.
Piepgras D. J. and Jacobsen S. B. (1992) The behavior of rare earth
elements in seawater: Precise determination of variations in the
North Pacific water column. Geochim. Cosmochim. Acta 56, 1851–
1862.
Piper D. Z. (1974) Rare earth elements in the sedimentary cycle: A
summary. Chem. Geol. 14, 285–304.
Shabani M. B., Akagi T., Shimizu H., and Masuda A. (1990) Deter-
mination of trace lanthanides and yttrium in seawater by inductively
coupled plasma mass spectrometry after preconcentration with sol-
vent extraction and back extraction. Anal. Chem. 62, 2709 –2714.
Sholkovitz E. R. and Schneider D. L. (1991) Cerium redox cycles and
rare earth elements in the Sargasso Sen. Geochem. Cosmochim. Acta
55, 2737–2743.
Tachikawa K., Jeandel C., and Dupre B. (1997) Distribution of rare
earth elements and neodymium isotopes in settling particulate ma-
terials of the tropical Atlantic Ocean (EUMELI site). Deep-Sea Res.
44, 1769 –1792.
Tachikawa K., Jeandel C., Vangriesheim A., and Dupre B. (in press)
Distribution of rare earth elements and neodymium isotopes in
suspended particles of the tropical Atlantic Ocean (EUMELI site).
Deep-Sea Res.
Taylor S. R. and McLennan S. M. (1985) The continental crust: Its
composition and evolution. An examination of the geochemical
record preserved in sedimentary rocks. Blackwell.
Zhang J. and Nozaki Y. (1996) Rare earth elements and yttrium in
seawater: ICP-MS determinations in the East Caroline, Coral Sea,
and South Fiji basins of the western South Pacific Ocean. Geochim.
Cosmochim. Acta 60, 4631– 4644.
Zhang J. and Nozaki Y. (1998) Behavior of rare earth elements in
seawater at the ocean margin: A study along the slopes of Sagami
and Nankai Troughs near Japan. Geochim. Cosmochim. Acta. 62,
1307–1317.
Zhang J., Amakawa H., and Nozaki Y. (1994) The comparative be-
haviors of yttrium and lanthanides in the seawater of the North
Pacific. Geophys. Res. Lett. 21, 2677–2680.