LIMNOLOGY

and

OCEANOGRAPHY: METHODS

Limnol. Oceanogr.: Methods 10, 2012, 296303

2012, by the American Society of Limnology and Oceanography, Inc.

Determination of 232Th and 230Th in seawater using a chemical

separation procedure and thermal ionization mass spectrometry
Per S. Andersson* and Hans Schberg
Laboratory for Isotope Geology, Swedish Museum of Natural History, Box 50007, 104 05 Stockholm, Sweden

Abstract
A method for the determination of 232Th and 230Th in seawater has been developed by substantial modifications of available techniques mainly used for other matrixes. We use a well calibrated 229Th spike and iron oxyhydroxides as a carrier phase for the precipitation of Th from about 3.5 L filtered seawater. The chemical separation and purification of Th from the Fe carrier comprise three ion chromatography steps including AG 1-x8
(200-400 mesh, Cl form), TRU Resin, and UTEVA Resin. The separated Th fraction is loaded with a mixture of
graphite/phosphoric acid onto zone-refined single rhenium filaments. These are analyzed in peak-jumping
mode on a thermal ionization mass spectrometer. Due to the low Th concentration in seawater, the 232Th blank
constitute a significant problem for the chemical separation of Th from the seawater matrix and also for the filament loading. By applying a detailed cleaning scheme including a HF-HCl mixture, the total procedural blank
could be substantially reduced to approximately 0.4 pg. This is generally < 1% of a typical 232Th concentration
in surface seawater, which ranges from about 0.01 to 1 picomol/kg. The analysis of standards, artificial seawater, and a series of seawater samples obtained from the GEOTRACES intercalibration cruises in 2008 and 2009
demonstrate that this method can yield a precision for 232Th concentration in seawater of about 1% to 2% (2s)
and 230Th concentration of about 2% to 5% (2s).

Precise and accurate 230Th and 232Th isotope measurements

are important for studying a wide variety of processes in earth
science including scavenging and particle dynamics in the
ocean (Bourdon et al. 2003). The element thorium has several
naturally occurring radioactive isotopes with a wide range of

half-lives stretching from the primordial 232Th (t1/2 = 1.39

1010 years) with a very long half-life compared with the intermediate daughters in the uranium decays series, which
includes 234Th (t1/2 = 24.1 days) and 230Th (t1/2 = 7.5 104 years).
The isotope 230Th is a key parameter in the GEOTRACES program as a tracer of ocean circulation and particle scavenging,
see GEOTRACES science plan ([Link]
The two isotopes 230Th and 232Th have different sources in
seawater where the long-lived 232Th is ultimately derived from
Aeolian- or river-introduced detrital particles, whereas the
230
Th is produced in situ from the uranium dissolved in seawater. In contrast to uranium, Th is highly insoluble and is
readily scavenged onto particle surfaces (Santschi et al. 2006),
which forms the basis for using Th to determine rates of particle transport in the ocean. The concentration of 232Th in seawater shows substantial variation, and in the open ocean, it
varies from about 0.01 to 1 picomol/kg (Moran et al. 1995;
Roy-Barman et al. 1996; Hsieh et al. 2011). The concentration
of 230Th varies between about 3 attomol/kg to 250 attomol/kg.
These low and variable concentrations are a major difficulty
for accurate determination of the isotopic composition of Th
in seawater.
Early studies of U-Th in the ocean used counting methods
for the determination of the radioactive decays (e.g., Moore

*Corresponding author: E-mail: [Link]@[Link]

Acknowledgments
We gratefully acknowledge the many colleagues that have assisted
both in the field and in the laboratory. The chief scientist, Greg Cutter, the
scientific party of GEOTRACES intercalibration cruises in 2008 and 2009
and the crew of R/V Knorr made a tremendous effort obtaining a set of
large volume samples for the intercalibration of Th isotopes. A large effort
has also been undertaken by the GEOTRACES Th intercalibration coordinators Bob Anderson and Brad Moran. Bob Andersson and Martin Fleisher
at Lamont Doherty Earth Observatory at Columbia University, NY, USA
produced and kindly provided the artificial seawater standards and corrected all 230Th data for ingrowth. The discussion with and comments
from Bob Anderson have been instrumental. Karin Wallner made valuable
contributions to the laboratory work. The input provided by an anonymous reviewer improved the manuscript. We also recognize financial support from the Swedish Research Council (VR grant 621-2007-4283). This
paper is part of the Intercalibration in Chemical Oceanography special
issue of Limnology and Oceanography: Methods that was supported by
funding from the U.S. National Science Foundation, Chemical
Oceanography Program (Grant OCE-927285 to G. Cutter).
DOI 10.4319/lom.2012.10.303

296

Andersson and Schberg

Determination of

232

Th and

230

Th in seawater

text), tested for blank levels, and if necessary, further purified.

All sample handling in the laboratory, including evaporation,
are carried out under clean conditions in laminar flow hoods
with HEPA-filtered air.
Sampling of dissolved Th
To evaluate the preconcentration, the chemical separation
and mass spectrometric analysis samples of seawater collected
during the GEOTRACES intercalibration cruises in 2008 and
2009 were used. For the near-surface water, a towed-fish sampler was used to collect trace metal clean samples. The sampling of deep water was done by using the US GEOTRACES
sampling facility trace metal clean system developed from the
method described by Measures et al. (2008). This system comprises 24 GO-Flo sampling bottles (12 L) mounted on a trace
metal-clean CTD carousel, a Kevlar conducting cable, and for
filtration and sub-sampling, a clean laboratory van with
HEPA-filtered positive pressured air. From the towed-fish and
the trace metal clean GO-Flo system water was filled into two
500 L trace metal clean tanks and homogenized for intercomparison (Anderson et al. 2012). From these tanks, the water
was distributed into high density polyethylene (HDPE) bottles
of 10 L volume. The HDPE bottles used in this study were
cleaned with ethanol and water on the outside and leached
with 10% HNO3 (analytical grade) for at least 1 week. The
leached bottles are rinsed several times in Milli-Q water, dried,
and wrapped in plastic bags. The plastic bag is only to be
opened during filling of the filtered water sample. The water
was filtered using an Osmonics cartridge of 0.2 m pore size
and acidified, pH < 2, using ultrapure HCl, 1 mL/L. The water
sample was drawn from the tanks into the sample container,
rinsing each container twice before filling it.
The 229Th spike calibration
An enriched 229Th spike is used for quantification of 232Th
and 230Th. The 229Th solution (1 M HNO3) is stored in a fluorinated ethylene propylene (FEP) bottle weighted before and
after spike preparation. This solution was calibrated by mixing
with a gravimetric Th standard (Alfa Aesar) and analyzing the
mixture using TIMS. The results from the analyses of seven
mixtures gives a 229Th spike concentration of 0.08439
0.00034 picomol/g (Fig. 1). The spike concentration is compensated for a loading blank contribution from the graphite
and the phosphoric acid of about 3.5 pg. For details regarding
the blank (see Blanks section).
Spiking and iron coprecipitation of Th
The volume of water required to measure 230Th and 232Th
will depend on the concentration, which varies in the ocean
and is generally lower in surface waters and increases with
depth (Roy-Barman et al. 1996). About 3.6 kg filtered and acidified water was aliquoted and weighted into a clean 4 L HDPE
bottle. About 0.017 pmol of the calibrated 229Th spike was
added by weight to the sample. An iron solution of 20 mg
Fe/mL in 4 M HCl was prepared from iron(III) chloride hexahydrate (Alfa Aesar) dissolved in HCl (SeaStar). The Fe solution
was cleaned by solvent extraction using diisopropyl ether and

1981; Nozaki et al. 1981; Bacon and Anderson 1982) but significant improvement in mass spectrometric techniques has
largely replaced counting methods for measuring long-lived
U-Th isotopes (Goldstein and Sterling 2003). Two types of
mass spectrometry, thermal ionization mass spectrometry
(TIMS) (Chen et al. 1992) and multiple collector inductively
coupled plasma mass spectrometry (MC-ICP-MS) (Luo et al.
1997; Hoffmann et al. 2007) are mainly used. Advantages with
MC-ICP-MS compared with TIMS includes high ionization
efficiency and faster sample throughput, whereas TIMS has
lower background intensities, fewer potential interferences,
and better abundance sensitivity. Thus, the different advantages and disadvantages make these techniques complementary. A variety of preconcentration and chemical separation
procedures have been used for preparation of brackish and
seawater samples for mass spectrometric determination of
232
Th in picomolar and 230Th at attomolar concentrations (e.g.,
Chen et al. 1986; Andersson et al. 1995; Moran et al. 1995;
Roy-Barman et al. 1996; Moran et al. 2001; Choi et al. 2001;
Shen et al. 2008).
Given the low and variable concentration of 232Th and 230Th
in seawater, the analytical challenges comprise the entire
chain from appropriate cleaning of sampling containers, the
sampling bottle system, preconcentration, and chemical separation to mass spectrometric analysis. During the GEOTRACES
intercalibration exercises, it became evident that discrepancies, larger than expected, between different laboratories and
techniques for the 232Th and the 230Th analyses in filtered seawater and particles existed. A large effort has been undertaken
by the Th intercalibration coordinators and several analytical
laboratories around the world to address these problems and
advise on methods to obtain intercomparable data. It was recognized that a crucial part of the problem is likely to be found
in chemical separation and the blank contribution from
resins, chemicals, and handling. To improve the analytical
protocols, it is necessary to examine in meticulous details all
material and reagents in contact with the sample. In this article, the used method for chemical separation, partly builds
upon the technique given in Pin and Santos-Zalduegui (1997).
We present and evaluate a detailed technique for the chemical
extraction, separation, and measurement of the 232Th and
230
Th isotopes in seawater using thermal ionization mass spectrometry (TIMS).

Materials and procedures

The described method for 232Th, 230Th analysis of seawater
comprises cleaning of sampling and storage containers as well
as cleaning, column set-up and preparation of reagents and
chromatography resins, preparation of filaments for mass
spectrometry, filament loading and mass spectrometric analysis. All of the material in contact with the sample have been
thoroughly cleaned by leaching in 10% HNO3 (analytical
grade) followed by rinsing in ultra-pure water. The chemical
reagents used are of highest quality (brands reported in the
297

Andersson and Schberg

Determination of

232

Th and

230

Th in seawater

decanted and the final fraction, about 500 to 800 mL, was centrifuged in several steps, using a cleaned 250 mL polyethylene
bottle and the supernatant discarded. The final precipitate was
washed once using Milli-Q water. The iron precipitate was dissolved in 7 mL of 1 M HCl (SeaStar) and 20 dr. H2O2 (Merck
Suprapur) added to reduce possible organic content in the
sample. The sample was transferred to a 15 mL Savillex Teflon
(PFA) vial and evaporated to dryness.
Chemical separation and purification of Th
The separation and purification of the Th fraction from the
iron precipitate comprise three polyethylene columns made
from BioRad Bio-Spin disposable chromatography columns of
2.5 mL total volume and with a porous frit of 30 m pore size.
The three columns hold three different types of chromatography resin where column #1 contains 0.75 mL BioRad AG 1-x8
(200-400 mesh, Cl form), #2 holds 0.5 mL TRU Resin from
Eichrom, and #3 contains 120 mg (~0.5 mL) UTEVA Resin
from Eichrom. The principal layout of the three columns is
displayed in Fig. 2.
The overall reason for using a three column set-up for the
extraction of Th from the seawater matrix is that our method
is part of a combined system for simultaneous coprecipitation
and separation of several elements including uranium and the
Rare Earth Elements. The first columns primary task is to
eliminate the Fe from the coprecipitation. After the second
column, the TRU Resin, the Th fraction is in most cases ready
for the mass spectrometer. However, by introducing a third
column with UTEVA Resin, a cleaner fraction, with respect to
uranium and matrix problems mainly derived from the presence of iron in the Th cut, is obtained.
The resin material might be a severe and variable source for

back extraction into a weak HCl solution in a glass separation

funnel. This procedure was repeated several times. About 14
mg iron was added to the acidified and spiked sample, and the
sample was allowed to equilibrate for at least 48 h. Thereafter,
the pH was raised to about 8 by using NH3OH (SeaStar) followed by vigorous stirring during 2-3 d. The formed precipitate was allowed to settle by gravity during 2-3 d. The length
of the stirring and settling times were not specifically tested
but can probably be reduced. The supernatant was gently

Fig. 1. The concentration of 229Th in the spike solution (pmol/g) determined by measuring the 229Th/232Th ratio in a mixture of the spike and a
gravimetric 232Th standard. The experiment gave a 229Th concentration of
0.08439 0.00034 pmol/g (n = 7). The concentration is corrected for a
blank contribution of 3.5 pg Th. The error bars are 2s.

Fig. 2. A schematic diagram showing the three columns, #1-3, used for separation of thorium from the iron precipitation. The columns #1 and #2 are
coupled in tandem. The boxes labeled A-E show the principal sample loading, wash and elution scheme used for the seawater matrix. The Th sample is
eluted into a PFA vial.

298

Andersson and Schberg

Determination of

232

Th and

230

Th in seawater

Aquadag colloidal graphite. However, this type of graphite can

significantly contribute to the blank (see Blanks section) and was
replaced by microcrystal grade, 2-15 micron, graphite powder
from Alfa Aesar (99.9995%). The graphite is cleaned by a H2O2
mixture for 12 h followed by centrifugation. This step is
repeated two times. The colloidal graphite solution is diluted
to a concentration of about 10 ng graphite/L using 0.25 M
H3PO4. The loading of the dried Th sample comprise rewetting
of the sample/phosphoric acid bead in the FEP vial using 1 L
Milli-Q water and addition and mixing of 2 L of the colloidal
graphite solution. This mixture is loaded onto the Re filament
in 1 L portions using a 1 L pipette. The filament is heated
in air at about 2 A, but is not allowed to glow. It is crucial that
the sample is embedded into a large excess of graphite.
Thermal Ionization Mass Spectrometry (TIMS)
The Th was analyzed on a Thermo Scientific Triton TIMS in
peak jumping mode using the RPQ-filter and ion-counting system. The filament was heated slowly, 500 mA/min to 0.8 A,
while carefully watching the pressure increase due to degassing
of the phosphoric acid. After stabilization of the pressure in the
low 108 mBar region, the heating continued. Measurement
usually commence in a temperature interval between 1750C
to 1950C. The count rate for mass m/z = 232 should be in the
range 1 105 to 8 105 cps (counts per second). The ion
counter dark noise is < 0.02 cps. The Peak-center and shape are
monitored and adjusted on m/z = 229 and m/z = 232. The peakjumping measurement routine is carried out in sequence m/z =
229, 230, 232 with 4-s integration time and 3-s idle between
peaks. About 60 cycles are collected for each sample during a
period of about 20 min. No mass fractionation correction is
applied. Typical apparent ionization efficiency for Th is estimated to be in the range 0.01% to 0.06%.
Blanks
A crucial part of the analytical protocol includes the determination and monitoring of blank levels for the different
parts of the protocol. The blank comprise the water sampling,
sampling container, reagents, resins, and filament-loading
blanks.
The background count rate in the Th mass range from the
filament is insignificant but the loading blank for Th was
found to be substantial. After the cleaning of the Aquadag colloidal graphite a series of loading blank analysis, including
both the graphite and the phosphoric acid, gave a loading
blank of 3.5 0.4 pg Th (n = 6). This value was used as a correction for the spike calibration (Fig. 1). The resin material also
seems to be a somewhat erratic but significant contributor to
the blank depending on type of resin and batch. After the first
cleaning of the resin material, we monitored a blank contribution of about 2 pg derived from the resins and reagents. The
iron carrier is also cleaned but has not been tested independently for Th blank contribution. A few tests have been made
using Fe precipitation of Milli-Q water and passing through
the entire chemical separation procedure, and this has not
rendered any significant increase in the Th blank from the

blanks, and to minimize the blank contribution from the resin

material, a rigorous cleaning method was applied. The resin
cleaning comprises bulk cleaning of the resin using a BioRad
Glass Econo-Column with a 250 mL reservoir. The reservoir is
filled to about 1/3 with resin material, the AG 1-x8 or TRU,
taken from the delivery bottles. The first rinsing step comprise
pro analysi quality of 6 M HCl followed by a mixture of 0.3 M
HF and 0.1 M HCl, (SeaStar), hereafter abbreviated FC31, and
Milli-Q water. About 80 mL of each rinse fluid is passed slowly
through the sample reservoir. This is repeated at least 6 times.
Following the initial bulk cleaning of the resin and after
filling the resin into column #1 and #2, these are rinsed with
8 resin bed volumes (BV) of 6 M HCl (SeaStar) and 4 BV of
FC31 (SeaStar). Before loading the sample onto the columns,
they are conditioned using 4 BV (2 times 1 mL) of 6 M HCl
(Fig. 2). The column #3 is rinsed with 4 BV of FC31 and 4 BV
of 0.05 M HNO3 (SeaStar) before usage.
The dried sample, stored in a 15 mL Savillex PFA vial, is dissolved in 4 mL of 6 M HCl. This solution is loaded onto column
#1, which sits directly on top of column #2 (Fig. 2). The Fe and
U binds to the first column, AG 1-x8, (200-400 mesh, Cl from),
and Th binds to the TRU Resin in the second column. Following the sample loading the columns are washed with 2 times 0.5
mL of 6 M HCl and the entire 6 M HCl fraction of 5 mL volume
is discarded. The second TRU Resin column is washed using 1
mL of 6 M HCl and 1 mL of 0.05 M HNO3, this is followed by
0.4 mL of 1 M HCl, which is also discarded. The Th fraction is
eluted into a PFA vial from the TRU Resin column using 2.6 mL
of 1 M HCl. To this HCl solution, 0.8 mL of concentrated HNO3
is added, transferring the Th fraction to a 3.8 M HNO3 solution.
This fraction is loaded onto the third column with UTEVA
Eichrom Resin, which is conditioned using 2 mL of 4 M HNO3.
After the sample loading 1 mL of 2.4 M HNO3, added to the
sample vial, is also loaded on the column followed by washing
with a second 1 mL of 2.4 M HNO3. The Th fraction is finally
eluted into a PFA vial using 2 mL of the FC31 mixture. The
eluted fraction is evaporated to dryness on a hot plate in a laminar flow bench using an Al-block heated to 140C. During
evaporation about 10 dr of concentrated HNO3 and 1 dr of 0.25
M H3PO4 (Sigma Aldrich 99.999%) is added to the solution. The
Th fraction is now ready for mass spectrometry. The total yield
for the Th separation is estimated to be about 95% using ICPMS (Inductively Coupled Plasma Mass Spectrometry) measurements. The columns and resins were discarded and replaced
between samples.
Filament loading
Single filaments made of zone refined rhenium where prepared and outgassed for 1 h at 2.5 A followed by 1 h at 4.5 A.
The background contribution from the outgassed zone refined
Re filament loading was tested by loading 229Th spike on the
filament. This gives a value for the filament contribution of
0.04 pg 232Th. This is insignificant compared with the blank
contribution from the separation chemistry. The purified Th
fraction was loaded onto the outgassed filament using
299

Andersson and Schberg

Determination of

chemical separation. Thus it looks like the cleaned Fe carrier

does not contribute significantly to the Th blank. Assuming
that the total 232Th blank for our intercalibration samples is a
combination of the blank for chemical separation and loading
blank, our total blank was initially about 5.5 pg. The total
amount of 232Th, for a 3.5 kg sample, range from about 60 pg
in the surface water to about 200 pg in the deep water. Thus
the blank contribution varies between 9% and 3% for surface
and deep water, respectively. This can be corrected for but
improving the blank levels would most likely also improve the
precision and reproducibility in the 232Th analyses.
The graphite seems to contribute most to the blank and by
replacing the colloidal graphite with graphite powder from
Alfa Aesar the loading blank could be substantially reduced.
By increasing the number of cleaning steps for the reagents,
primarily the graphite and the chromatography resins, results
in a substantial reduction of the combined chemistry and filament loading blanks with values of about 0.4 pg (range
between 0.30 to 0.38). In total, we have analyzed about 50 Th
blanks to examine where the problem is and how it can be
kept under control. Clearly the blanks in the laboratory,
mainly derived from the reagents, can be substantially lowered and controlled by applying a detailed cleaning procedure
and continuous monitoring of the blank levels.

232

Th and

230

Th in seawater

Fig. 3. The 230Th/232Th isotopic composition of the IRMM-035 standard

(232Th/230Th = 87100) analyzed during a period of 2 weeks. About 200 ng
Th was loaded for each run, and the data show a long term reproducibility of 87500 1000 (1.1 %) 2s (n = 10).

tion (3140N, 6410W) and the second in May 2009 to the

Pacific Ocean SAFe Station (30N, 140W). In total, eight 10-L
HDPE bottles were cleaned, shipped, and sampled for near surface waters and deep waters (Table 1). The samples, homogenized in 500-L trace element clean tanks, were filtered and
acidified on board by the intercalibration crew and subsequently shipped to participating laboratories. We used about
3.6 kg of the filtered water for 232Th and 230Th analyses following the outlined procedures. For the surface and deep water,
this yielded a total amount of 232Th for loading on the filament ranging between about 60 to 250 pg (0.3 to 1 pmol). The
concentrations of 232Th and 230Th, corrected for the blank, are
presented in Table 1 for the eight bottles obtained from the
GEOTRACES intercalibration cruises (all errors are 2s). The
230
Th was also corrected for the ingrowth between the sampling date and the chemical separation date and reported as
230
Th* in Table 1.
At the BATS Station, we were not successful at obtaining reliable results for the shallow water (10 m depth) sample, but for
the BATS deep water (2000 m), three different individual bottles (LIG# 62, 69, 70) were analyzed (Table 1). For the 232Th concentration, the 2s relative errors for the three samples ranged
between 0.93% to 1.2% and for the 230Th concentration, from
3.3% to 4.3% and between 3.4% to 4.0% for the 230Th/232Th
ratio. The results for the BATS deep water, the 232Th and 230Th
concentrations and the 230Th/232Th atom ratios are similar
within the analytical errors. However, there is an exception for
the 232Th concentration in sample LIG# 69, which is about
7.3% higher compared to the samples LIG# 62 and 70.
The samples from the Pacific SAFe Station comprise surface
water (30 m), bottle LIG# 75, and deep waters (3000 m), bottles LIG# 73, 74, and 76. For the 232Th concentration at SAFe

Assessment
Th standard
To establish our accuracy and reproducibility measuring
the 232Th/230Th ratio using TIMS, we used the isotopic reference material IRMM-035. The IRMM-035, stored in 1 M HNO3
solution, has a reported certified value for 232Th/230Th = 87100.
About 200 ng of 232Th was aliquot by weight from the solution
and evaporated to dryness and loaded on zone-refined rhenium filaments mounted in double filament configuration.
Repeated analyses of the IRMM-035 (n = 10) gave a 232Th/230Th
= 87500 1000, 2s (Fig. 3). Although this standard has a
232
Th/230Th ratio about one to two order of magnitude higher
compared with seawater, it serves to demonstrate our ability to
measure high 232Th/230Th ratios.
Artificial seawater standard
An artificial seawater standard, 2010-1, has been prepared
at Lamont Doherty Earth Observatory and provided to the
GEOTRACES intercalibration community. The concentration
in 0.5 g of the standard solution is equivalent to that of 230Th
and 232Th in 10 L seawater from 2000 m at the BATS Station.
The standard was spiked and passed though the column
chemistry, and repeated analysis gave a 232Th concentration of
985 16 pg/g, 1.6 % (2s, n = 7), a 230Th concentration of 249
7 fg/g, 3.0 % (2s, n = 7) and a 230Th/232Th ratio of 2.54 104
0.03 104, 1.2 % (2s, n = 7).
Seawater samples
The described method was evaluated using samples from
the two GEOTRACES intercalibration cruises, the first in June
2008 to the Atlantic BATS (Bermuda Atlantic Time Series) Sta300

Determination of

6
7
7
4.4
12
5
20
19

178
174
178
46.5
961
391
430
936
0.33
0.44
0.38
0.08
0.4
0.4
1.4
0.6

9.49
9.90
9.40
0.81
26.7
26.4
28.6
27.4
0.33
0.44
0.38
0.08
0.4
0.4
1.4
0.6

9.75
10.16
9.66
1.14
27.0
26.8
29.0
27.7
1.4
1.9
1.7
0.4
2
2
6
3

Th and

230

Th in seawater

Discussion
The low and variable concentration of 232Th and 230Th in
seawater makes the blank contribution from the sampling and
preparation a major issue for precise and reproducible measurements. This has been a long standing problem both for
counting techniques, using extraction of Th from large volumes (>103 L) of water (Moore 1981), as well as for mass spectrometric methods using substantially smaller water volumes,
between 1 and 10 L. The results for the intercalibration samples demonstrate that although the 230Th concentration is

230

Th* decay corrected for the ingrowth during the time interval between sampling and separation.
second preparation

42.4
44.2
42.0
5.0
117
116
126
121
0.5
0.7
0.6
0.2
0.2
0.6
0.8
0.2

53.7
57.5
53.4
17.5
28.0
68.2
67.0
29.5
0.002
0.003
0.002
0.001
0.001
0.002
0.003
0.001

0.232
0.248
0.230
0.076
0.121
0.294
0.289
0.127
10
2000
2000
2000
30
3000
3000
3000
3000
BATS shallow
BATS deep
BATS deep
BATS deep
SAFe shallow
SAFe deep
SAFe deep
SAFe deep
SAFe deep
KN-193-6-710
KN-193-6-612
KN-193-6-613
KN-193-6-614
KN-195-8-2019
KN-195-8-2119
KN-195-8-2120
KN-195-8-2120
KN-195-8-2121

61
62
69
70
75
76
73
73
74

Station
LIG #
R/V Knorr #

232

the 2s relative errors range between 0.83% to 1.3%, between

1.5% to 7.0% for the 230Th concentration and from 1.3% to
9.5% for the 230Th/232Th ratio. The errors are similar in the
deep waters at both stations, and the larger errors are associated with the lower surface water 230Th concentration, which
is about 20 times lower compared with the deep water. In bottle LIG# 73 the 232Th concentration is about 2.3 times higher
compared with bottles LIG# 74 and 76 collected at the same
depth. The sample LIG# 73 was prepared and analyzed a second time, and this gave a similar 232Th concentration within
errors (Table 1). However, the 230Th concentrations do not
show this discrepancy between the different bottles, which
indicates a possible contamination of 232Th in bottle LIG# 73.
The results for the deep water samples from both of the
intercalibration stations, BATS and SAFe, demonstrate that the
230
Th concentrations are similar within errors for all samples
from the same depth. The precision (2s ranged between 1.5%
and 4.8%. For the surface water, the error is larger, about 7%,
due to significantly smaller (20 times) concentration of 230Th.
In contrast to the 230Th concentration the 232Th concentration show variations larger than the errors in the deep water
samples from both BATS and SAFe stations. The bottle LIG# 73
shows higher 232Th concentration compared with LIG# 74 and
76. Also at BATS Station, one of the deep water samples shows
a small but significantly higher 232Th concentration compared
with the other two. This contrasting variation between 230Th
and 232Th most likely reflect the different source functions for
the two isotopes where 230Th is being produced in situ by 238U
radioactive decay and the 232Th most likely reflects input from
terrestrial detrital dust sources or in deep waters resuspension
of material from bottom sediments. Due to the different
sources, the 232Th is probably also more prone to contamination compared with the 230Th. The larger 232Th concentration
found in bottle LIG# 73 might be caused by increased blank
contribution from the chemical separation or introduced during sampling, filtration, and transfer to sampling containers
on-board the ship. As the repeated second preparation of bottle LIG# 73, using new chemicals, gave similar 232Th concentration compared with the first preparation, it seems unlikely
that the discrepancy is caused by contamination during sample preparation in the laboratory. However, with the current
available information, we cannot firmly determine where a
possible contamination of bottle LIG# 73 might come from.

230

Th*
(fg/kg) 2s

230

Th
(fg/kg) 2s

230

Th
(amol/kg) 2s

230

Th
(pg/kg) 2s

232

Th
(pmol/kg) 2s

232

Depth
(m)

Table 1. The 232Th and 230Th concentrations and isotope ratios in filtered water from the GEOTRACES intercalibration cruises in 2008 and 2009.

Th/232Th
(106) 2s

Andersson and Schberg

301

Andersson and Schberg

Determination of

232

Th and

230

Th in seawater

234 [doi:10.1016/0012-821X(94)00262-W].
Anderson, R.F., M.Q. Fleisher, L.F. Robinson, R. L. Edwards, J.
A. Hoff, S. B. Moran, M. Rutgers van der Loeff, A. L.
Thomas, M. Roy-Barman, and R. Francois. 2012. GEOTRACES intercalibration of 230Th, 232Th, 231Pa, and prospects
for 10Be Limnol. Oceanogr. Methods 10, 179-213
[doi:10.4319/lom.2012.10.179].
Bacon, M., and R. F. Anderson. 1982. Distribution of Th isotopes between dissolved and particulate forms in the deep
sea. J. Geophys. Res. 87:2045-2056 [doi:10.1029/JC087
iC03p02045].
Bourdon, B., G. M. Henderson, C. C., Lundstrom, and S. P.
Turner (Eds.). 2003. Uranium-series geochemistry, Reviews
in mineralogy and geochemistry, vol. 52. The Mineralogical
Society of America.
Chen, J. H., R. L. Edwards, and G. J. Wasserburg. 1986. 238U,
234
U, and 232Th in seawater. Earth Planet. Sci. Lett. 80:241251 [doi:10.1016/0012-821X(86)90108-1].
, , and . 1992. Mass spectrometry and
applications to uranium-series disequilibrium. In M.
Ivanovich and R. S. Harmon [Eds.], Uranium-series disequilibrium: Applications to earth, marine, and environmental
sciences, 2nd ed. Oxford Univ. Press.
Choi, M. S., and others. 2001. Rapid determination of 230Th
and 231Pa in seawater by desolvated micro-nebulization
Inductively Coupled Plasma magnetic sector mass spectrometry. Mar. Chem. 76:99-112 [doi:10.1016/S03044203(01)00050-0].
Goldstein, S. J., and C. H. Sterling. 2003. Techniques for measuring uranium series nuclides: 1992-2002. Rev. Mineral.
Geochem. 52:23-57 [doi:10.2113/0520023].
Hoffmann, D. L., J. Prytulak, D. A. Richards, T. Elliot, C. D.
Coath, P. L. Smart, and D. Scholz. 2007. Procedures for
accurate U and Th isotope measurements by high precision
MC-ICPMS. Int. J. Mass Spec. 264:97-109 [doi:10.1016/
[Link].2007.03.020].
Hsieh, Y.-T., G. Henderson, and A. L. Thomas. 2011. Combining seawater 232Th and 230Th concentrations to determine
dust fluxes to the surface ocean. Earth Planet. Sci. Lett.
312:280-290 [doi:10:1016/[Link].2011.10.022].
Luo, X., M. Rehkmper, D. C. Lee, and A. Halliday. 1997. High
precision 230Th/232Th and 234U/238U measurements using
energy-filtered ICP magnetic sector multiple collector mass
spectrometry. Int. J. Mass Spec. 171:105-117.
Measures, C. I., W. M. Landing, M. T. Brown, and C. S. Buck.
2008. A commercially available rosette system for trace
metal-clean sampling. Limnol. Oceanogr. Methods 6:384394 [doi:10.4319/lom.2008.6.384].
Moran, S. B., J. A. Hoff, K. O. Buessler, and R. L. Edwards. 1995.
High precision 230Th and 232Th in the Norwegian Sea and
Denmark by thermal ionization mass spectrometry. Geophys Res. Lett. 22:2589-2592 [doi:10.1029/95GL02652].
, C. -C. Shen, S. E. Weinstein, L. H. Hettinger, J. H. Hoff,
H. N. Edmonds, and R. L. Edwards. 2001. Constraints on

between 2 to 4 orders of magnitude lower compared with

232
Th concentration in seawater, the blank problem seems to
be smaller for the 230Th analyses compared with 232Th. This is
most likely due to their different sources in seawater. In terms
of contamination during sampling dust from the ship or the
sampling equipment might also be a possible contributor.
Clearly 232Th is substantially more prone to potential contamination than 230Th. The results show that the variation in 232Th
concentration is larger than from analytical uncertainty only.
The precipitation and chromatographic separation include
several steps and different reagents that are potential contributors to the 232Th blank. Besides using high purity chemicals,
it is also necessary to include a cleaning procedure. The ion
exchange resins are severe sources for 232Th blanks, and it is
necessary to apply a detailed cleaning protocol including a
HF-HCl mixture, which must be repeated several times. The
loading of the sample on the filament include colloidal
graphite, which is also a source for 232Th blank. However, by
replacing the graphite with another brand and applying a
cleaning procedure including H2O2, the blank could be substantially reduced.
During the period of the GEOTRACES intercalibration project, we have been able to reduce the total blank, which
include all steps from the iron carrier, chemical separation
procedure, and filament and instrumental blanks to about 0.4
pg 232Th. This constitute <1% of the total 232Th load from
about 3.5 L of a surface seawater sample. Clearly, it is possible
to reduce and monitor the blanks by careful cleaning.
By analyzing thorium standards, artificial seawater and several seawater samples using our low blank-loading technique
for TIMS, we have demonstrated that it is possible to determine the 232Th concentration in seawater with a precision of
about 1% to 2% (2s) and 230Th concentration to about 2% to
5% (2s). Our results demonstrate that it is possible to obtain
oceanographically consistent data for low level concentrations
of 232Th and 230Th isotopes in samples of about 3 L volume.

Comments and recommendations

The present method builds upon the development of a procedure using a well calibrated 229Th spike and iron oxyhydroxides as a carrier phase for the precipitation of Th in filtered seawater. The chromatography technique for the
chemical separation and purification of Th from the Fe carrier
uses commercially available ion resin material. For determination of Th in low concentrations, such as in seawater, it is necessary to apply a rigorous procedure monitoring blanks and
cleaning of the chromatography resin and reagents used for
separation. Compared to 230Th, the 232Th is more prone to contamination during sampling and sampling handling.

References
Andersson, P. S., G. J. Wasserburg, J. H. Chen, D. A. Papanastassiou, and J. Ingri. 1995. 238U-234U and 232Th-230Th in the
Baltic Sea and river water. Earth Planet. Sci. Lett. 130:217302

Andersson and Schberg

Determination of

232

Th and

230

Th in seawater

Roy-Barman, M., J. H. Chen, and G. J. Wasserburg. 1996. 230Th232

Th systematics in the central Pacific Ocean: the sources
and fate of thorium. Earth Planet. Sci. Lett. 139:351-363
[doi:10.1016/0012-821X(96)00017-9].
Santschi, P. H., and others. 2006. Thorium speciation in seawater. Mar. Chem. 100:250-268 [doi:10.1016/[Link].
2005.10.024].
Shen, C. -C., and others. 2008. Variation of initial 230Th/232Th
and limits of high precision U-Th dating of shallow-water
corals. Geochim. Cosmochim. Acta 72:4201-4223
[doi:10.1016/[Link].2008.06.011].

deep water age and particle flux in the Equatorial and

South Atlantic Ocean based on seawater 231Pa and 230Th
data. Geophys. Res. Lett. 28:3437-3440 [doi:10.1029/
2001GL013339].
Moore, W. S. 1981. The thorium content of ocean water. Earth
Planet. Sci. Lett. 53:419-426 [doi:10.1016/0012-821X(81)
90046-7].
Nozaki, Y., Y. Horibe, and H. Tsubota. 1981. The water column
distributions of thorium isotopes in the western North
Pacific. Earth Planet. Sci. Lett. 54:203-216 [doi:10.1016/
0012-821X(81)90004-2].
Pin, C., and J. F. Santos-Zalduegui. 1997. Sequential separation
of light rare-earth elements, thorium and uranium by
miniaturized extraction chromatography: Application to
isotopic analyses of silicate rocks. Anal. Chim. Acta. 339:7989 [doi:10.1016/S0003-2670(96)00499-0].

Submitted 28 July 2011

Revised 10 February 2012
Accepted 14 March 2012

303