Chemicaa Geology I 19 ( B995, ) I- 12

Islam-Planck-bttstirt!s j& Chetnir, Postjach 3060, D-55020 ,Wainz, Cmnan_v

Received 30 January 1994; revision accepted 12 August 1994

A new high-performance liqui cbromatogra~~~ic (HPLC > proc re has been developed and applied to the
determination of Th and U cone trations in geological materials. ck samples are first acid-digested in screw-
cap Savillex@ bcakcrs with HF 104 and I-ICI. Th and U are t isolated from the geological matrix by a
quantitative two-step ion-exchange procedure. Determination of the elements by FIPLC then follows. Th
are separated from one another and residual matrix elements as complexes with a-hydroxyisobutyric acid (
on a reversed-phase column using gradient elution in - 10 min. Detection of Th and U is accomplished by
spectrophotometry after an on-line postcolumn reaction with Arsenazo HI. Detection limits are - 30 and - 100
pg for Th and U (or - 0.75 and - 2.5 ppb using 100-mg rock samples ), respectively.
The HPLC technique has been applied to the analysis of nine international and three well-characterize
standard rocks to eva!uate ana!ytica! erf@r_mance.T?! and U concentrations in the analyzed samples spanned the
range of c 10 ppb to > 10 ppm. The PLC data show excellent agreement with values obtained by spark source-,
inductively coupled plasma- and isotope dilution thermal ionization mass spectrometry. The I sextcrnal precision
of the method is better than I!I2% and + 5% for the determination 3% and U concentrations higher than 50 and 5
ppb, respectively.

and U ab~~da~ces oft

jar subreservoirs form
A detailed knowledge of the geochemistry of
Th and U is necessary for understanding terres-
trial evolution. Models of the thermal history of
thz Earth are highly dependent upon assump-
tions about the geochemical distribution of Th
and U (e.g., Christensen, 1985; Galer and Gold-
stein, 199 I ). This is so, because heat production
is solely governed by the abundances of naturally
occurring radioactive elements. Therefore con-
siderable &x-t has been made to estimate the Th
INAl
. (e.g., Jochum et al., 1983; Taylor and
nan, 1985). Such data also permit a bet-
ter understanding of the physical processes gov-
erning the fractionation of incompatible trace
elements between the silicate reservoirs
Earth (e.g., Galer and O’Nions, 1985).
work by White ( 1993) has shown that U-
elemental systematics of the mantle may
similar information. Abundances of T
and abundance ratios with other trace elements,
such as Nb, have been used to constrain the mass

0009~2541/95/$09.50 0 I995 Elsevier Science B.V. All rights reserved

SSDIOOO9-2541(94)00105-7
2 M. Rehkiimper 1 Chemical
balance of the Earth’s internal differentiation
(e.g., Hofmann et al., 1986).
Th and U are extremely depleted in some geo-
logical samples (e.g., mid-ocean ridge basalts: Th
-., 150 ppb, U - 60 ppb) and at low concentra-
tion levels in many rock types such as ocean is-
land basalts (Th - 4 ppm, U - 1 ppm) and or-
ogenic andesites (Th - 5 ppm, U - 1.5 pm)
(Taylor and McLennan, 1985; Hofmann, 1988;
Sun and McDonough, 1989). Thus, resolution of
fine-scale abundance variations requires the use
oftechniques with high analytical sbnsitivity and
precision, such as isotope dilution-thermal ion-
ization mass spectrometry (ID-TIhlS), spark
source mass spectrometry (SSMS) or induc-
tively coupled plasma-mass spectrometry (ICP-
MS). The high instrumentation and operating
costs as weli as limited instrument access are sig-
nificant drawbacks, however.
Two HPLC techniques for the determination
of trace elements in rock samples have previ-
ously been developed at this institute (Cassidy,
1988; Rehkgmper, 1994). Long-term experience
with one of these methods has shown that HPLC
can provide reliable trace-element data on a rou-
tine basis, while instrumentation and running
costs are relatively low. Several HPLC tech-
niqucs have been developed in the past for the
determination of actinide elements in a variety
of sample types relevant to the nuclear or min-
eral processing industry (e.g., Barkley et al., 1986;
Kerr et al., 1988). These studies demonstrated
the suitability ofHPLC for precise, routine anal-
ysis of trace Th and U. Here a new, highly sensi-
tive HPLC technique for the quantitative deter-
mination of Th and U concentrations in
geological samples is presented.
The analytical procedure involves acid-diges-
tion of rock samples and the separation of Th and
U from bulk rock components by ion-exchange
chromatography. HPLC analysis then follows.
Gradient elution is used to effect a fast separa-
tion ( - 10 min) of Th and U from one another
and residual matrix elements on a reversed phase
column. With on-line postcolumn addition of
Arsenazo III and spectrophotomctric detection
of Th and U, detection limits in the sub-nauti-
gram range are achieved for both elements.
Geology 119 (1905) i-12
2.1. HPLC system
A schematic diagram of the EC apparatus
is shown in Fig. 1. For eluent delivery a Syk
low-pressure gradient system consisting of an
2000 pump with PEEK (polyetheretherketo
pumpheads, an S 2000 gradient controller and
an S 8110 low-pressure gradient module (with
eluent mixer) was used. Two
were employed during this
125 x 4.6 mm (filled with 3-p
C18 reversed-phase material
125 x 4 mm with 4-pm Superspher@ ZOORP-18
resin. A metal-free filter with 2-pm polyethylene
frits (Upchurch Scientific@) was used for col-
umn protection. Delivery of the postcolumn re-
agritt (PCR) waj accomplished Wifh a second
metal-free PEEK S IO00 pump. The pulse dam-
pener behind the PCR pump consisted of a
125 x4.6 mm PEEK column (Gram@ ) filled
with 3-pm resin. The PCR mixer, constructed
from a PEEK-tee (Upchurch Scientific
similar to the screen-tee reactor descr
Cassidy et al. ( 1987). The remaining compo-
nents of the HPLC apparatus consisted of an in-
ert Rheodyne@’9125 injection valve with 500~~1
PEEK sample loop, a Linear Instruments@’ 204
variable wavelength detector and a Shimadzu*
C-R6A computing integrator. PEEK capillary
tubing and fittings were used throughout the
chromatographic system.
2.2. Reagents and materials
HPLC eluents were prepared with deionized
and double-distilled water from a quartz still,
HPLC grade methanol (Merck@’) and a 1 M (x-
hydroxyisobutyric acid (HIBA) stock solution.
The HIBA stock solution was prepared by disso-
lution of purissima grade HIBA ( - 99%, Fluka@ )
in purified water. Insoluble organic residues were
then separated from this solution by filtration
using 0.45~pm filters (Millipore@ ). Further pu-
rification was achieved by ion-exchange liltra-
tion of Zhe solution through a Teflon@ column
filled with - 200 ml of Bio-Rad AG5OWX8@
 Fig. 1. Schematic
e eluents were p
Th and U standard solutions ( - 1000 pg ml- ’)
HN03 matrix) were obtained from Aldrich@’and
Alfa@. Concentrated gravimetric ( - 100 pg g- ’,
matrix i M HN03) standard solutions contain-
ing both Th and U were prepared from these
commercial solutions after their density had been
determined accurately by repeated weighing of
known volumes. Dilute Th, U standard solu-
tions ( - 1 pg g- I9matrix 2 M HN03 ) were pre-
pared by further dilution of the 100-ppm
solutions.
diagram ofthe
Cl. After refluxi
heated strongly ( > 140’C ) until completely dry.
A second high-temperature evaporation step was
carried 01’~after ad ition of N 1 ml of 6
The samples were t en dissolved in 6 m
and heated overnight in closed beakers to
re a complete solution prior to the ion-ex-
change separation of Th and U.
The dissolution procedure effkiently removes
HF from the rock solutions following mineral
digestion and thus prevents loss of Th by c
cipitation with insoluble
4 M. Rehktimper / Chemical Geology 119 (1995) I- 12

rides. Decomposition of samples in $avillex* fraction was collected in a

beakers is convenient and completely adequate evaporated to dryness on a
for the mafic rocks analyzed in this study. High- The second step of the p
pressure bombs, however, must be used for the quantitative separation of
digestion of samples containing resistant min- fraction collected as desc
eral phases (e.g., zircon, baddeleyite ), since these fraction was dissolved in 1
may carry significant amounts of Th and U. then loaded onto a 5-ml b
resin. Table 2 summarizes the column elution se-
2.4. Ion-exchangeseparationof Th and U quence. After the sample had been carefully
washed into the resin (with 3 x 1 ml of
A two-step ion-exchange procedure was used HN03) the column was rinsed with 8 ml of
for the isolation of Th and U from matrix ele- HN03. This eluent achi
ments. The first step, a cation-exchange separa- tiou of most matrix
tion of Th from rock solution, is summarized in amounts of ferric iron and R
Table 1. The sample, dissolved in 6 ml of 6 M column along with U, which
HCl, was added to a 5-ml bed of cation-exchange titatively up to this point. U was completely
resin and carefully rinsed into the column with washed off the resin by addition of 30 ml of 1 M
3x ! ml of 6 M HCl. The bulk sample and U HCl. The U fraction was collected either in the
were then eluted with 30 ml of 6 M HCl. Collec- beaker containing the (dried) Th fraction or in
tion of the eluate up to this point, in a 50-ml Te- a fresh Teflon* beaker and evaporated to
flon@ beaker, assured quantitative recovery of U. dryness.
This bulk rock fraction was dried down on a hot
plate and saved for further processing. 2.5. Preparationof samples and calibration
The column was then washed with 15 ml of 6 standardsfor HPLC analysis
M HCl for further removal of ,the rare-earth ele-
ments (REE) and 10 ml of 3 M HCl to elute re- For quantitative determination of Th and U
sidual Fe3+, which could not be rinsed off effr- by HPLC an external calibration with pure ele-
ciently with 6 M HCl. Th was now stripped from mental standard solutions was carried out. The
the resin with 30 ml of an acid mixture that was HPLC calibration standards were pre
both 6 Min HCl and 1 Min HF (6 XL-Xl-1 M 1 ~+a combined Th, U standard SO-
weighing a di,u.,
HF; this nomenclature is used hereafter). The Th lution into Teflon@ beakers. After evaporation

Table 1
Elution sequence used for the separation of Th from rock solutions
Eluent Volume Kd Eluted
(ml)
Th U(VT)

Rock solution in:

6MHCI -6 60a-b I-2b bulk rock + U

6MHCI @-j&b
33 l-2b bulk rock + U
6MHCI 15 60”eb REE
3MHCI IO 170”-b Fe3+
6MHCl-I MHF 30 -0.Y Th
Column: 5-ml cation-exchange resin AGSUWX12,200-400 mesh. The distribution coeffncients (K,,-values) are compiled from:
‘Strelow and Boshoff ( 1972 ); bNelson et al. ( 1964) [extrapolation of the data by Strelow ( 1960) gives an identical K,-value];
and cMurase et al. ( 1964) and this study.
Table 2

Eluent

-I - 15” - 300”JJ
I-4 -’ 500‘ < Id)
8 M I-IN& 3x 1 -15” - 300aJJ 6uPk rock
8 M WN03 8 - 15” - 300=JJ bulk rock
1 li!fHC~ 30 w I’ < Id u, (Tb)
Column: 5-ml anion-exe ,200-400 mesh. The ~~s~~~u~io~coefficients (&-values) are compiled from: “Faris
andBuchanan (1964); bDanon (1956);’ unneyetal. (1959) and Wish (1959); anddKrausetal. (1956).

dierit atzlysis was m B

2.6. HPLC determination

The determination of Th and U was effected

by gradir.ut
eluent (10.2
were set to 1.5 and 0.5 ml min- I, respectively,
3.1. Separation of7X and U
and the chromatographic system was flushed un-
til fully equilibrated. A reagent absorbance of
w 0.0 1 absorbance units (AU ) vs. the eluent was
obtained1 when the detector was set to the analyt- Initial studies with several ion-exchange an
ical wavelength of 658 nm. reversed-phase chromatograp
After injection of 25-2509.d sample or stan- that it is not possible to separ
dard aliquots, the uent was programmed to 2 5% titatively from rock sample
methanol, 0.2 M IBA (at pH 4.2) over 1 min. for subsequen
Delivery of this eluent followed for a further 4.5 , a two-step proc

min to complete the elution of Th and U. The Section 2.4, was adopted
gradient was then returned to the initial condi- chemical behavior of Th and U during ion-ex-
tions within 0.7 min to re-equilibrate the col- change chemistry was investigated in detail due
6 M, Rehkiinzper / Chemical Geology 119 (I 995) I- 12
to the necessity of obtaining a quantitative
chemical yield on a routine basis.
The high kn-value of Th in 6 M HCl permits
large-volume sample solutions to be loaded onto
the cation-exchange columns (primary separa-
tion step, Table 1). This is important since ma-
jor-element precipitates may form in more con-
centrated sample solutions, possibly scavenging
Th and U. All samples that were analyzed during
this study showed perfectly clear solutions in
which no precipitate could be detected after
completion of acid-digestion.
Th is efficiently rinsed from the first-step cat-
ion column using 6 M HCl-1 A4HF after selec-
tive elution of major elements and U with 6 A4
and 3 A4 HCl. Although use of HF in ion-ex-
change chemistry is generally not desirable, a
strong complexing agent, such as fluoride, is nec-
essary to remove Th erfficientlyfrom cation-ex-
change resin (Kolarik, 1990). Other possible
complexing agents include oxalic and citric acid,
but these are solids and therefore problematic to
purify and difficult to separate from Th follow-
ing the ion-exchange separation.
U is collected with the bulk rock fraction dur-
ing the first-step cation-exchange separation.
Even though the distribution coeflicient of U for
the anion-exchange system is not particularly
high ( - 15, Table 2 ), quantitative recovery of U
( s 99.5%) was routinely achieved, Elution ex-
periments with rock samples revealed that more
than 15 ml of 8 M HN03 had to be passed
through the column before detectable amounts
of U could be found in the emuent. Addition-
ally, potential losses were monitored by the col-
lection of 8 M HN03 effluent fractions before the
start of U elution. Very small amounts of U
( eO.5016of total) were detected in these frac-
tions in some cases.
The performance of the first-step cation-ex-
change columns was found to be strongly depen-
dent upon sample size, however. Initial analyses
showed that Th was not quantitatively recovered
in the 6 M HCl- 1 M HF wash if the sample size
was > 125 mg. A detailed investigation of the Th
ehtion pattern was undertaken to investigate this
problem. These experiments revealed that fol-
lowing sample loading, some Th was eluted pre-
maturely in the first IO- 15
from the column. No Tb could be
fluent fractions collected Curing farther elution.
Early loss of Th was never detecte
sample size was kept < 115 mg. This su
view that premature elution of Th
cause the capacity of the resin bed is exceeded
for a limited amount of time during and directly
after sample loading. This explanation is not
unique, however. It is also possible that some of
the Th forms stable complexes in the sample so-
lutions (e.g., with trace
tained by the resin. The
distribution coefficients
graphic; system (Table 2) were otherwise con-
firmed by elution experiments. More than 75 ml
of 6 M HCl had to be passed through the column
before any.
trace Th from rock samples could be
detected m the cffluzr~.
Quantitative recovery of Th from rock sam-
ples > 115 mg is nevertheless possible on a rou-
tine basis. If a large rock sample is loaded on the
cation column, the early-eluting Th will be col-
lected quantitatively together with U in the bulk
rock fraction. This fraction is further processed
using anion-exchange chromatography (Table
2 ) . The second column procedure was designed
to permit quantitative recovery of U and Th from
the bulk rock fractions. The sa 21~sloaded must
be completely dissolved in 8 M WN03 prior to
column chemistry, however, since any precipi-
tate is likely to contain Th. This was generally
possible, particularly since bulk rock fractions
were found to be more soluble in 8 M nitric acid
than the original rock samples. A comparison of
the respective distribution coefficients (Table 2)
reveals that Th is retained much more strongly
than U during elution of the rock matrix. For
collection, Th can be efficiently rinsed from the
anion resin along with the U using 1 MHCl.
An alternative procedure is also possible. Rock
samples with sizes > 115 mg are generally only
required for the analysis of U concentrations
< 20 ppb (Section 3.2). In such cases, the solu-
tions can be split prior to column chemistry. The
majority ( - 60-80%) of the rock solution ic then
used solely for the isolation of U, while Th is sep-
arated from the smaller split, containing < 115
are eluted. The so

Th 0.001 AU

Ii
Ji fined by the repetitive injection of an
solution, was N ?8.5-1.5% (la) for S
containing 1-5 ng of Th and LJ.
I , I I
The calibration curve for quantitative ana
0 2 4 6 8
usually spanned a c entration range of 1-2 or-
Time / min ders of magnitude. owever, the effective ana-
Fig. 2. HPLC chromatogram of Th and U after separation lytical range is far greater than this, because di-
from a basaltic rock. Chromatographic conditions are de- lution factors and injected sample volumes could
scribed in the text (see Section 2.5). The prominent concave be varied and chosen according to needs.
shape of the baseline is caused by the gradient profile. Sam-
ple: 100 ,ul of a 500~~1sample solution obtained by separa-
of
areas are essentially unaffected ( w Eo internal
precision) by variations in sample vc”:a~~~ ~XXD
tion of Th and U from w 90 mg of Hawaii basalt K88-30. The
injected sample contains _ 17 ng Th and _ 5.5 ng U. 25 to 250 ~1. The Th and ‘U concentrations can
S M. Rehkdrnper / Chemical Geology 119 (199s) I- 12
be determined together or independently de-
pending upon how the ion-exchange separation
is performed. One advantage in performing the
analyses separately is that the dilution factors can
be chosen independently. Due to the lower de-
tection sensitivity of U this may be advanta-
geous for low concentration samples with high
Th/U ratios.
3.3. Analysisof rock samples
Nine international geochemical reference ma-
terials and three well-documented Max-Planck-
Institut fur Chemie (MPI) in-house rock stan-
dards were analyzed by HPLC for Th and U to
evaluate the analytical performance of the new
technique. The highly depleted reference sam-
ples BIR-I, BTHO and JP-I were used in this
study to demonstrate the high sensitivity of the
HPLC method. Concentrations varied between
0.012 and 19.2 ppm for Th, and 0.008 and 4.76
ppm for U in the samples. Ten rock standards
were analyzed repeatedly using separate disso-
lutions to quantify the analytical precision.
The procedural blank was carefully monitored
to avoid an analytical bias on low-concentration
samples. No accurate total blank values could be
determined, however, since the analysis of blank
samples consistently revealed no detectable con-
tent of Th and U. The same result was obtained
on three procedural blanks for U that were pooled
to increase detection sensitivity. This constrains
blanks to be < 30 pg for Th and U, thus ruling
out any significant blank contribution to the
measured concentrations.
The analytical results obtained on standard
rocks are summarized and compared with refer-
ence values in Table 3. The standard rocks JA-2,
JB-2, JB-3 and K88-30 were each analyzed 4-6
times to evaluate the analytical precision. Based
upon separate dissolutions, relative standard de-
viations ( la) of between + 0.7% and + 3.4% for
Th and + 0.9% and + 3.7% for U were obtained
for these particular samples. The analyses of
other samples with fewer repeat measurements,
and Th and U concentrations of > 50 ppb ex-
hibit a similar precision. The individual anal;+-
cal results for all samples with repeat analyses are
plotted in Fig. 3. Eat individual value for a
sample is normalized to
HPLC value in order to h
scatter. The data show th
proximately constant for
tions of > 50 ppb. These results were then use
to calculate an overallvalue for the anaIyticaI re-
producibility of the technique. This is -- Ir 1.7%
( lo external precision) for samples containin
Th and U concentrations of > 50 ppb.
cibility is poorer for samples
Th and U concentrations. A
of + 4.9% ( lcr) for s
centrations of < 50 ppb is inferred from the an-
alytical results on BIR-2 and JP-1.
The HPLC results are compared with com-
piled literature values and data acquired by mass
spectrometry (ICP-
and neutron activati
der to demonstrate the analytical accuracy of the
new method. Agreement is seen, in general, to be
good for samples with Th and U concentrations
of > 50 ppb, with deviations being < 1O”/o in most
cases. Excellent agreement is achieved for the
well-characterizedreference sample BHVG-I and
the two MPI in-house standarZ
lZOB previously measured by
HPLC results for the standard rocks BE-N and
W-2 are both very similar to the compilation val-
ues and some mass spectrometric results. Other
literature values for these samples, however, dif-
fer significantly from the results obtained in this
study. There are no clear systenmaticdifferences
between the HPLC and reference data sets for the
Geological Survey of Japan standard rocks. In
general, deviations are smaller if the HPLC val-
ues are compared with mass spectrometric data
as opposed to the reference values of Govinda-
raju ( 1989). HPLC Th values are generally
higher than ICP-MS and SSMS results for JA-2,
lower for JB-2 and very similar for JB-3. The
HPLC U data confirm most mass spectrometric
results on JA-2 and JB-2 but are significantly
lower for JB-3. This sample, however, was ana-
lyzed four times by HPLC and the relatively low
U value was confirmed in all analyses. The
agreement of HPLC U in DNC-I with the ICP-
MS data of Jenner et. al. ( 1990) is excellent, but
 %-
d
i-1
PI
I ci
10 M. Rehkiinrper/ Chemical Geology I1 9 (1995) I- 12

script. Pormal reviews

(4 mond and C. Roddick resulte
improvements and clarificatio
thanked for editorial handling.
the authors Ph. thesis at the University of
ainz.

B.,,I
0.01 0.1 1 10 100 eferences
Th (wm)
Barkley, D.J., Blanchette, M., Cassidy, R.M. and Elchuk, S.,
1986. Dynamic chromatographic systems for the deter-
mination of rare earths and ~~or~~rnin samples from ura-
nium ore refining processes. Anal. Chem., 58: 2222-2226.
Barkley, D-J., Bennett, L.A., Charbonneau, J.R. and Pokra-
E
jac, L.A., 1992. Applications of high-performance ion
E chromatography in the mineral processing industry. J.
.._._v0
.._.__ GO 0
__.._.._.......__
_...A.._ Chromatogr., 606: 195-201.
z 0 t 0
Bedard, L.P. and Barnes, S.-J., 1990. Instrumental neutron
‘&
=
.I -5
0 0 8 activation analysis by collecting only one spectrum: Re-
& sults for international geochemical reference samples.
P t O
-10 ’ “” I -&I Geostand. News].. I4: 479-484.
0.001 0.01 0.1 1 10 100 Bunney, L.R., Ballou, N.E., Pascual, J. and Foti, S., 1959.
Anion exchange behavior of several metal ions in hydro-
U kvm) chloric, nitric, and sulfuric acid solutions. Anal. Chem.,
31: 324-326.
Fig. 3. Percent relative deviation of individual analytical rc- Cassidy, R.M., 1988. Determination of rare-earth elements
sults (based upon repeat dissolutions) for (a) Th and (b) U in rocks by liquid chromatography. Chem. Geol., 67: I85-
from the respective mean concentrations. All ten standard 195.
rock samples with replicate analyses listed in Table 3 arc Cassidy, R.M. and Fraser, M., 1984. Equilibria effects in the
plotted in this figure. The la external precision of the HPLC dynamic ion-exchange separation of metal ions. Chro-
method for Th and U concentrations of > 50 ppb based upon
matographia, 18: 369-373.
this is w -t 1.7%.
Cassidy, R.M. and Karcher, B.D., 1986. Post-column reac-
tion detection of inorganic species. In: I.S. Krull (Edi-
peridotite reference sample J&l, HPLC results tor), Reaction Detection in Liquid Chromatography.
for Th and U are very similar to the data ob- Marcel Dekker, New York, N.Y., pp. 129-194.
Cassidy, R.M., Elchuk, S. and Dasgupta, P.K., 1987. Per-
tained by mass spectrometry but much lower than
formance of annular membrane and screen-tee reactors
the compiled values. for postcolumn-reaction detection of metal ions sepa-
rated by liquid chromatography. Anal. Chem., 59: 85-90.
Chauvel, C,, Hofmann, A.W. and Vidal, P., 1992. HIMU-
EM: The French Polynesian connection. Earth Planet. Sci.
Acknowledgements Lett., 110: 99-l 19.
Christensen, U.R., 1985. Thermal evolution models for the
This work would not have been possible with- Earth. J. Geophys. Res., 90: 2995-3007.
Class, C., 1994. A geochemical study of Grande Comore Is-
out the interest and support of Al Hofmann. I land: Indian Ocean hotspot source compositions and
am grateful to C. HCmond and K.P. Jochum for plume-lithosphere interactions. Ph.D. Thesis, Universi-
giving me access to unpublished analytical data. tat Karlsruhe, Karlsruhe, 2 17 pp.
C. HCmond and C. Chauvel provided the MPI Danon, J., 1956. Adsorption of thorium by anion-exchange
reference samples. I would especially like to thank resins from nitric acid media. J. Am. Chem. Sot., 78:
5953-5954.
I. Raczek for advice and technical assistance.
Faris, J.P. and Buchanan, R.F., 1964. Anion exchange char-
A.W. Hofmann and Steve Galer read and con- acteristics of elements in nitric acid medium. Anal. Chem.,
siderably improved early versions of the manu- 36: 1157-l 158.
 oldstein, S.L., B99 1. Early mantle differ-
eochim. Cos-
.J.G. and G’Nions, R.K., 1985. Residence time of
thorium, uranium and lead in the mantle with implica-
tions for mantle convection. Nature (London ), 3 16: 778-
782.
Garbe-§ch~~berg, C.- ., 1993. Simultaneous dctcrmination
of thirty-seven trace element in twenty-eight interna-
tional rock standards by IICP- S. Geostand. Newsl., 17:
81-97.
Gladney, ES. and Roelandts,
elemental concentration d
I, QLO-1, RGM-I, SCo-
Geostand. Newsl., 12: 253-362.
Gladney, ES. and Roelandts, I., 1988b. 1987
elemental concentration data for USGS
and W-2. Geostand. Newsl.. 12: 63- 118.
Govindaraju, K., 1989. 1989 compilation of working values
and sample description for 272 geostandards. Geostand.
News!., 13: S-1 13.
Govindaraju, K. and Woelandts, I., 1993. Second repor:
(1993) on the first three GIT-IWG rock r
ples: Anorthosite from Greenland, AN-G:
sey-la-C&e, BE-N; Granite de Beauvoir, MA-N. Geo-
stand. Newsl., I7: 227-294.
allett, R.B. and Kyle, P.R., 1993. XRF and JNAA deter-
minations of major and trace elements in Geological Sur-
vey of Japan igneous and sedimentaq rock standards.
Geostand. Newsl., 17: 127- 133.
Hemond, C.. Amdt, N.T., Lichtenstein, U. and Hofmann,
A.W., 1993. The heterogeneous Iceland plume: Nd-Sr-Q
isotopes and trace element constraints. J. Geophys. Res.,
98: 15833-I 5850.
Hemond, C., Hofmann,A.W., Heusser, G., Condomines, M.,
Raczek, I. and Rhodes, J.M., 1994. U-Th-Ka systcmatics
in Kilauea and Mauna Loa basalts, Hawaii. Chem. Geol.,
116: 163-180.
Hofmann, A.W., 1988. Chemical differentiation of the Earth:
The relationship between mantle, continental crust, and
oceanic crust. Earth Planet. Sci. Lett., 90: 297-3 14.
Hofmann, A.W., Jochum, K.P., Seufert, M. and White, WM.,
1986. Nb and Pb in oceanic basalts: New constraints on
mantle evolution. Earth Planet. Sci. Lett., 79: 33-45.
Ionov, D.A., Savoyant, L. and Dupuy, C., 1992. Application
of the ICP-MS technique to trace element analysis of per-
idotites and their minerals. Geostand. Newsl., 16; 3 1l-
315.
Jenner, G.A., Longerich, H.P., Jackson, S.E. and Fryer, B-J.,
1990. ICP-MS - A powerful tool for high-precision trace-
element analysis in Earth sciences: Evidence from analy-
sis of selected U.S.G.S. reference samp!es. In: P.J. Potts,
C. Dupuy and J.F.W. Bowles (Guest-Editors), Microan-
alyticall Methods in Mineralogy and Geochemistry. Chem.
Geol., 83: 133-148 (special issue).
Jochum, K.P. and Jenner, G.A., 1994. Trace element analy-
sis of Gi-otogical Studey Df Japan silicate reference mate-
rials: CO~lpauison ofSSMSwjEk 1W--FJ$ &la and a Crz~-
ical discussion of Compiisd values. Frescnius J. .&Jai,
Chem.. 350: 310-318.
Jochum. K.P., Hofmarm. 4.W.3 ito, E:., Seufert? H.M. and
White, W.M.. I983. K, U and Th in mid-ocean ridge ba-
sail ghsscs and heat production, K/U and KjRb in the
mantle. Nature (London ), 306: 43 1-436.
stand. News!., 14: 469-473.
Trace
C and
LlMS. Geostand. News]., 18: 43-5 1.
Kerr. A., Mupferschmidt, W. and Attas, ., 1988. Determi-
nation of uraniiim in natural groun aters using high-
performance liquid chromatography. Anal. Chem.. 60:
2729-2133.
Kolarik, Z., 1990. Cation exchange of thorium. In: Gmelin
andbook of Inorganic Chemistry: Th Supplement Vol-
ume D3. Springer, Berlin, pp. 123-305.
Kraus, K.A., Moore, G.E. and Nelson, F., 1956. Anion-ex-
change studies, XXH. fh( BV) and U (IV) in hydrochloric
acid - Separation of thorium, protactinium and ura-
mum. J. Am. Chem. Sot., 78: 2692-2694.
Loock, G., 1992. Character and distribution of the Indian
Ocean mantle domain: A study of the mantle source com-
positions of Lau Basin (SW Pacific) and Indian Ocean
mid-ocean ridge basahs. Ph.D. Thesis, Universitat Ktiln,
Kohl, 125 pp.
Murase, T., Lind, EL. and Nelson, F., 1964. Jon-exchange
procedures. 1V. Separation of ““Tb (UX, ) from cranyl
nitrate solutions. J. Chromatogr., 14: 478-481.
Nelson, F.. Murase, T. and Kraus. K.A.. 1964. Ion-exchange
procedures, 1. Cation exchange in concentrated HCl and
HC1G4 solutions. J. Chromatogr., 13: 503-535.
Potts, P.J. and Rogers, N.W., 199 I. Determination of trace
elements in selected geological reference materials by in-
strument:.! neutron activation analvsis. Geostand. Newsl.,
15: I1 l-l 16.
Rehkamper, M., 1994. A new low-level HPLC-technique for
quantitative determination of niobium in rocks. Chem.
Geol., 113: 61-69.
Strelow, F.W.E., 1960. An ion exchange selectivity scale of
cations based on equilibrium distribution coefficients.
Anal. Chem., 32: 1185-l 188.
Strelow, F.W.E. and Boshoff, M.D., 1972. Selective separa-
tion of thorium from rare-earths, zirconium and other
elements by cation-exchange chromatography in hydro-
bromic acid media. Anal. Chim. Acta, 62: 351-359.
Sun, S.-s. and McDonough, W.F., 1989. Chemical and iso-
topic systematics of oceanic basalts: Implications for
mantle composition and processes. In: AD. Saunders and
M.J. Norry (Editors), Magmatism in the Oceanic Basins.
Blackwell, Oxford, pp. 3 13-345.
Taylor, KS. and McLennan, S.M.. 1985. The Continental
12 M. Rehkdmper / Chemical Geology 119 (I 995) I- 12

Crust: Its Composition and Evolution. Blackwell, Oxford, 115: 21 l-226.

312 pp. Wish, L., 1959. Anion exchange behavior in mixed acid so-
White. W.M., 1993. 23EU/204Pbin MORB and open system lutions and development of a sequential separation
evolution of the depleted mantle. Earth Planet. Sci. Lett., scheme. Anal. Chem., 3 1: 326-330.