1. health and fitness
2. disease
3. gerd and acid reflux

Effects of sample preparation and calibration strategy on
accuracy and precision in the multi-elemental analysis of soil by
sector-field ICP-MS{
Emma Engstro¨m,a Anna Stenberg,b Douglas C. Baxter,a Dmitry Malinovsky,b
Irma Ma¨kinen,c Seppo Po¨nnid and Ilia Rodushkin*a
a
Analytica AB, S-977 75 Lulea˚, Sweden. E-mail: [email protected]
Division of Applied Geology, Lulea˚ University of Technology, S-971 87 Lulea˚, Sweden
c
Finnish Environment Institute, P.O. Box 140, F-00251 Helsinki, Finland
d
Pirkanmaa Regional Environment Centre, P.O. Box 297, F-33101 Tampere, Finland
b
Received 25th November 2003, Accepted 18th March 2004
First published as an Advance Article on the web 28th May 2004
Soil samples were prepared for multi-element analysis using HNO3 leaching or pseudo-total digestion with
HNO3, HCl and HF in a microwave oven, both methods requiring 70 min heating time. Two calibration
approaches for the soil characterization were also compared: external calibration, combined with internal
standardization, and isotope dilution (ID) after appropriate spiking of the soils with a stable isotope mixture
prior to sample preparation. Analyses were performed using inductively coupled plasma sector field mass
spectrometry (ICP-SFMS). Accurate total elemental concentrations were only obtained for Cd and P using
both sample preparation methods in two certified reference materials, NIST SRM 2709 and CCRMP SO-2, as
well as comparable values for a Finnish inter-laboratory soil. The pseudo-total digestion method also provided
accurate results for As, Be, Co, Fe, Mn, Ni, Pb, Sb, Ti, V and Zn. For Cu in SO-2 and Cr in both certified
reference materials, incomplete recoveries were always obtained. In the case of Cr, this is due to difficulties
associated with the complete solubilization of refractory minerals.
For a given final dilution factor, external calibration provides better limits of detection (LODs) than ID.
As both methods of quantification yield results of essentially equivalent accuracy and precision, external
calibration is to be preferred as a greater number of elements are amenable to analysis in a shorter
measurement time. On the other hand, ID can be combined with matrix separation (NH3 precipitation was
used here), allowing lower dilution factors to be used without deleterious effects on the instrumental
performance. In particular, improved LODs could be obtained for Cd, Cu and Hg, primarily as a result of
being able to introduce ten-fold more concentrated solutions from which the bulk of the matrix had been
removed. For Cu and Ni, matrix separation almost eliminated Ti, and thus the formation of spectrally
interfering TiO1 was completely suppressed. Potentially, the combination of ID and matrix separation would
allow these elements to be determined without resorting to medium resolution measurement mode, again
improving the LODs for the determination by ID-ICP-SFMS.
DOI: 10.1039/b315283a
Introduction
858
During the last decade there has been an increasing need
for reliable measurements of heavy metal concentrations in
environmental samples. Of these, soil is of a complex nature,
with variable amounts of mineral phases, co-precipitated and
sorbed species associated with soil minerals or organic matter,
and dissolved species that may be complexed by a variety of
organic and inorganic ligands.1 Numerous sample preparation
procedures have been developed for soils, bearing in mind
factors such as time and equipment constraints, as well as the
needs of the end-users of the data. Various national and
international standards have also been introduced, dictating
the methods applied in routine analytical laboratories. These
include acid extractable fractions using HNO3 and aqua regia,2
total mineralization applying alkaline fusion,3 high pressure
digestion with HNO3 and HF,4 prolonged digestion/wet ashing
with acids including HNO3 and HF,4 and, in addition, HClO4,5
and HCl,6 and microwave digestion using HNO3 and HF,2 plus
H2O2.7 Thus, to achieve total elemental coverage, several
methods may be required, depending on the chemical nature of
the analytes and the soil matrix.8 Separation of the element of
{ Presented at the 4th International Conference on High Resolution
Sector Field ICP-MS, Venice, Italy, October 15–17, 2003.
J. Anal. At. Spectrom., 2004, 19, 858–866
interest may also be necessary to attain measurements of high
quality.6,8,9
The elevated interest in comparability and traceability of
results has also led to demands for verifiable accuracy and
improved precision. Certified and laboratory matrix reference
materials, with known elemental content, are important tools
for quality control and performance assessment, and may be
used for both inter- and intra-laboratory comparisons and
evaluations.10 Today there are a number of soil reference
materials available11 but, as yet, this quantity does not cover all
future demands12 and the large variation in soil chemistry
worldwide.13 Thus, the needs for soil reference materials continue to increase, in particular regarding their characterization
with respect to a wider range of elements.
The instrumental technique of inductively coupled plasma
sector field mass spectrometry (ICP-SFMS) is an excellent choice
for such characterization, due to its great detection capabilities,
the possibility for interference-free determination and significantly improved isotopic ratio precision (in low resolution mode)
compared with quadrupole-based systems (ICP-QMS).14 For
routine analyses, external calibration with internal standardization is used (almost) exclusively with ICP-SFMS, thus relying on
the aforementioned attractive features of the technique. Should
non-spectral interferences persist, then alternative calibration
strategies, such as isotope dilution (ID),15 may be required.
This journal is ß The Royal Society of Chemistry 2004
Isotope dilution mass spectrometry is often regarded as a
definitive method, since all chemical manipulations are carried
out on a direct weight basis, and the analysis involves isotope
ratio measurements rather than absolute mass determinations.16 As the method is founded on a sound theoretical basis,
and since a complete uncertainty statement can be budgeted,
ID has the highest metrological qualities.17–20 After equilibration of the natural analyte and spike, all sampling handling will
affect the isotopes equally, which eases the complete recovery
requirement for purification and pre-concentration steps, yet
contamination must still be avoided.21 By using an isotope of
the same element as an internal standard, non-spectral interference effects are perfectly corrected, and thus ID is often
applied during the analysis of samples containing complex
matrices like soils.6,7,22
Using ID, however, at least two spectrally interference-free
stable (or in certain cases long-lived radio-) isotopes of each
analyte must be available, and a priori knowledge of concentrations may be desirable to achieve an optimum mixture of the
spike and sample.21 Total dissolution of analyte-bearing phases
in the soil and avoidance of selective losses of the measurand
or the enriched isotope are also vital to ensure complete
equilibration between the original sample and spike. Uncertainties in isotopic abundances and spike concentrations, the
costs of material and the fact that several environmentally
interesting elements are mono-isotopic, may present further
practical limitations.21
The object of this study was to compare the capabilities of
external calibration and ID for the multi-elemental characterization of a candidate soil reference material by ICP-SFMS,
with respect to the accuracy and uncertainty of the results.
Practical aspects, in particular regarding methods of sample
preparation and elemental coverage, as well as labour and
instrument-time requirements, are also considered.
Experimental
Instrumentation
The ICP-SFMS instrument used was the ELEMENT (Finnigan MAT, Bremen, Germany) equipped with an ASX 500
sample changer (CETAC Technologies, Omaha, NE, USA), Ni
sampler (orifice diameter 1.1 mm) and skimmer (0.8 mm)
cones, and a standard torch with 1.5 mm injector diameter. The
instrument provides three fixed resolution settings (m/Dm #
320, 4200 and 10 500). For sample introduction a Teflon micronebulizer (Elemental Scientific Inc., Omaha, NE, USA) was
used with a peristaltic pump (Perimax 12, SPETEC, Erding,
Germany) to control the sample uptake rate at approximately
80 ml min21. Instrumental operating conditions and data acquisition parameters used for external calibration measurements are
reported elsewhere.23,24 Details of the conditions employed for
ID measurements are given in Table 1.
A microwave oven (MDS-2000, CEM, Matthews, NC,
USA), equipped with 12 perfluoroalkoxy-lined vessels with
safety rupture membranes, was used for sample digestion.
The centrifuge used was a Megafuge 1.0 (Heraeus Sepatech,
Hannover, Germany).
Samples and reagents
All calibration and internal standard solutions used for
external calibration were prepared by diluting 1 g l21 single
element standard solutions (SPEX Plasma Standards, Edison,
NJ, USA).
The certified isotopic reference materials 111Cd (92.11%),
65
Cu (99.30%), 53Cr (98.23%), 67Zn (94.60%), 61Ni (86.44%)
and 199Hg (91.09%) in metallic/oxidic forms were purchased
from Oak Ridge National Laboratory (Oak Ridge, TN, USA),
except 206Pb (92.15%), which was Radiogenic Lead Isotope
Standard, NIST SRM 983 (Gaithersburg, MD, USA). The
enriched isotopes were dissolved in appropriate acids and
diluted to approximately 100 mg l21 for storage in pre-cleaned
Nalgene glass bottles. Element concentrations in these solutions were determined by ICP-OES (excluding Hg) with
external calibration and correction for isotopic shifts, and by
reverse-ID using ICP-QMS and ICP-SFMS.18
NIST SRM 981 (Common Lead Isotope Standard) was used
to correct for instrumental mass discrimination during Pb
isotope ratio measurements.
The dilution of samples and standards was performed using
distilled Milli-Q (MQ) water (Millipore Milli-Q, Bedford,
USA). Analytical-reagent grade nitric acid (Merck, Darmstadt,
Germany) was utilized after additional purification by subboiling distillation in a quartz still. The hydrofluoric (40%,
Merck, Suprapure grade) and hydrochloric (30%, Fluka,
Steinheim, Germany, analytical-plus grade) acids, as well as
25% ammonia solution (Merck, Suprapure grade), were used
without additional purification.
The soil reference materials used were NIST SRM 2709 (San
Joaquin Soil) and Canada Centre for Mineral and Energy
Technology Reference Soil Sample SO-2. Additional information about these materials is provided in the certificates of
analysis. The in-house soil control sample, intended for quality
control and performance assessment in an on-going interlaboratory comparison, was collected in southern Finland
(Kangasala) near by an esker. The sample was dried at room
temperature and sieved through a 0.25 mm sieve followed by
dividing into 50 g sub-samples using a rotary sample divider
equipped with a vibratory feeder. The homogeneity of the
sample was tested by replicate analyses at the analytical
laboratory of the Finnish Environment Institute using ICPQMS. Hereafter this sample will be referred to as M1. As the
deadline for submission of analytical results from participating
Table 1 Isotope dilution measurement parameters
Element
Cd
Ratio
Resolution
Acquisition mode
Acquisition window (%)
Integration window (%)
Dwell time/s
Setting timea/s
No. of samples per nuclide
No. of scans
On-line mass bias correction
Bracketing standards
114
Cr
111
Cd/ Cd
Low
E-scan
10
10
0.001
0.001
21
3 6 2500
Sn
Cd standard
52
Cu
53
Cr/ Cr
Medium
E-scan
80
60
0.005
0.3
15
3 6 150
No
Cr standard
63
Hg
65
Cu/ Cu
Low/medium
E-scan
10/80
10/60
0.001/0.005
0.001/0.3
21/15
3 6 2500/3 6 150
No
Cu standard
201
Ni
199
Hg/ Hg
Low
E-scan
10
10
0.001
0.001
21
3 6 2500
Tl
Hg standard
60
Pb
61
Ni/ Ni
Low/medium
E-scan
10/80
10/60
0.001/0.005
0.001/0.3
21/15
3 6 2500/3 6 150
No
Ni standard
208
Zn
206
Pb/ Pb
Low
E-scan
10
10
0.001
0.001
21
3 6 2500
Tl
SRM 981
66
Zn/67Zn
Medium
E-scan
80
60
0.005
0.3
15
3 6 150
No
Zn standard
a
For isotope ratio measurements in medium resolution, 0.3 s settling times were used to ensure stable mass calibration throughout the duration
of the scans. The mass range covered in medium resolution also requires the magnet mass to be changed at least once; all specified Cr, Cu, Ni
and Zn isotopes were measured in each scan.
J. Anal. At. Spectrom., 2004, 19, 858–866
859
laboratories expired in November 2003, this report will not
interfere with the performance evaluation scheme.
Sample preparation
Soil samples were prepared using microwave-assisted (MW)
treatment according to the following procedures.
HNO3 digestion. 0.5 g of sample was digested (2 6 30 min at
600 W power followed by 10 min at 800 W power) with 5 ml of
concentrated HNO3. Although HNO3 alone is incapable of
extracting elements associated with the silicate matrix, this type
of leaching has been generally used in environmental exposure
assessment in Nordic countries, in accordance with US EPA
Method 3050.
Digestion with acid mixture. The soil samples, 0.2 g, were
digested in a mixture of HNO3 (3 ml), HCl (2 ml) and HF (1 ml)
using the same MW parameters as for HNO3 digestion. This
acid mixture has proven to provide quantitative recoveries for
a variety of elements in a suite of solid samples, including
soils.23–26 Analysis of SRM 2709 digests by ICP-OES revealed
recovery of approximately 65–70% of the sum of major
inorganic constituents. Low Ca, Mg and Al recoveries are due
to formation of insoluble fluorides,27 present as a grey residue
after sample digestion. This precipitate can be brought into
solution by either repeated evaporation of the digest with
HNO3 or by addition of H3BO3.26 As the excess of HF is
unlikely to cause formation of precipitate for elements under
consideration in this study, neither of these approaches was
tested. It should be stressed that some of the most refractory
minerals, such as zircons and chromites, may not be completely
solubilized using the proposed method. The use of alternative
sample preparations (e.g., fusion, Carius tube digestion or high
pressure ashing) or longer MW digestion procedures at higher
pressures and temperatures should be selected for complete
dissolution.28
For ID quantification, weighed amounts of isotopic spike
mixture were added to half of the samples prior to digestion. In
order to minimize the relative error in the measured isotope
ratios in the spiked samples, concentrations of analytes in the
mixture as well as the added amounts were optimized in
accordance with the procedure proposed by Garcı´a Alonso,29
taking into account concentrations in sample M1 obtained by
external calibration and the sample mass for each procedure.
All operations for preparation of the spike mixture were performed gravimetrically. No separate optimization was performed for soil reference materials. To avoid risks of diluting
digestion media excessively, less than 500 ml of the mixture was
used.
After sample treatment, all solutions were transferred to precleaned polypropylene auto-sampler tubes and the volume was
adjusted to 10 ml with distilled MQ-water. For reproducibility
assessment, each sample digestion was repeated on at least two
different days. At least three replicates of sample M1 were
prepared for each digestion and each preparation day, both
with and without spike addition. At least two procedural
blanks were prepared together with each digestion batch using
just the reagent solutions.
860
relatively high sample dilution is important to minimize these
matrix effects, keeping response variations within 10% even for
long (w5 h) analytical sequences. Further discussion on the
selection of appropriate sample dilution for ICP-SFMS can be
found elsewhere.24
Samples were analysed using In and Lu for internal
standardization (added to all solutions at 25 mg l21) and
external calibration using a set of multi-element standards in
the expected concentration range.
Analysis by isotope dilution
For Cr isotope ratio measurements, digestion solutions were
diluted as described for solutions intended for analysis by
external calibration. For Pb isotope ratio measurements,
digestion solutions were diluted to obtain approximately
5 mg l21 of total Pb in the measuring solutions, followed by
addition of Tl at 5 mg l21 (for on-line mass discrimination
correction). For the rest of the analytes, matrix separation was
performed as follows: 2 ml of the digested spike–sample blend
solution was added dropwise to a pre-cleaned polypropylene
auto-sampler tube containing 2 ml NH3 solution. The solution
was carefully swirled and immediately (within 10 min from
digest addition) centrifuged at 4000 rev min21 for 4 min. The
supernatant was thereafter transferred into another precleaned polypropylene auto-sampler tube. The resulting solution was diluted 10-fold further with distilled MQ-water thus
providing significantly lower dilution factor compared to that
required for external calibration. For on-line mass bias correction, an aliquot intended for Hg analysis was spiked with Tl at
5 mg l21.
Prior to each isotope ratio measurement session, thorough
mass calibration was performed in ‘manual’ mode for all
isotopes monitored. Each measurement sequence starts with
analysis of a synthetic blank, followed by a standard with
known isotopic composition, synthetic blank, procedural
blanks, samples, synthetic blank and standard. Dead time
correction was performed on-line by the instrumental software.
Measured isotope ratios were transferred to a spreadsheet
program for blank and mass discrimination corrections. The
latter were performed using the bracketing standards approach. For Pb, Hg and Cd on-line mass discrimination correction
was also possible (Table 1) according to the linear model30
using external element ratios. The mass concentration of
analyte in the sample, Cx (mg g21), was calculated as:17–20
Mx .my .A2,y . Rx Ry {Rb
.
Cx ~Cy
(1)
My .mx .w.A1,x
Rb {Rx
where Cy represents the mass concentration of the spike
(mg g21), M x and M y the atomic weights (g mol21) of the
sample and the spike, mx and my the masses (g) of soil sample
and spike solution in the blend, A2,y and A1,x the atom fractions
of the enriched isotope in the spike and the reference isotope in
the sample, and Rx, Ry and Rb are the mass discriminationcorrected isotope amount ratios in the sample, spike and blend,
respectively. The term w is a correction factor for sample
moisture content.19
All measurements were performed in at least two independent sessions using two different ICP-SFMS instruments.
Analysis by external calibration
Uncertainties
Digestion solutions were diluted with 0.28 M HNO3, resulting
in dilution factors of approximately 2000 and 4000 for HNO3
leaching and acid mixture digestion, respectively. Though
samples with lower dilution can be run by ICP-SFMS, high
concentrations of major matrix elements (namely Al and Fe)
in measuring solutions will gradually clog the orifice of the
skimmer cone, affecting ion sampling into the mass spectrometer with analyte responses suffering as a result. Therefore,
Most of the uncertainty components involved in analytical
measurements are accounted for by replicating the entire procedure, as was done here. Most importantly, uncertainty
contributions from sample heterogeneity as well as digestion,
weighing and dilution operations will be included.7
J. Anal. At. Spectrom., 2004, 19, 858–866
Isotope dilution. By performing measurements on at least two
different days, inclusion of variations in mass discrimination
correction factors, as required for the ID calculation,19 will also
be achieved. Additional factors that must also be included are
the uncertainties in isotopic abundances of analyte in the
samples and spike materials and, most importantly, the concentrations of the latter. Other than Pb, it is assumed that the
analytes in the samples have the natural isotopic abundances
recommended by IUPAC.31 Of the elements determined using
ID in this work, it is considered that only the IUPAC recommended isotopic abundances for Zn are inapplicable to the
SPEX standards employed in this work; separate measurements made by multi-collector ICP-SFMS in this laboratory
(not shown) suggest that the 66Zn/67Zn isotope amount ratio in
the SPEX Plasma Standard is about 1.3% higher than the
IUPAC value.31 Similar findings for other commerciallyavailable Zn standard solutions have recently been reported.32
For this reason, data appropriate to the Zn standard employed
here, as well as the soil samples, were selected accordingly for
use in eqn. (1). Data for the spikes were taken from the
accompanying documentation, after experimental verification.
For the uncertainty components originating from the spike
concentrations, more detailed consideration is required. Concentrations were determined by ICP-OES (excluding Hg) with
external calibration and correction for isotopic shifts, and by
reverse-ID18 using ICP-QMS and ICP-SFMS. Data from these
analyses were then hcombined
weighted mean con to yield
i0:5
e y zs2 C x
centrations, UID ~k. s2 C
zBy , in the manner
33
described by Schiller and Eberhardt. To account for possible
systematic differences between individual concentration estimates, a bias allowance term, By, defined as the maximum
absolute deviation from the weighted mean, is also calculated.
The total combined uncertainty for ID, U ID, is expressed as
h i0:5
e y zs2 C x
UID ~k. s2 C
zBy
(2)
to yield
where k is the coverage factor,19,34 chosen an approe y the standard
ximately 95% confidence level, i.e., k ~ 2, s C
uncertainty of the weighted mean spike concentration and
¯ x) that of the mean sample concentration estimated from
s(C
replicate analyses. It should be noted that, unless all determined
spike concentrations agree perfectly, By can never be zero.
External calibration. Unless the behaviour of the internal
standard and analytes is identical during the analysis of solid
samples by ICP-SFMS, changes in normalized intensities will
result. To account for such differential drift, a series of
standards was run before and after the samples, and the change
in slope of the linear calibration function was calculated. It was
assumed that half of the deviation in the slope from the initial
value corresponds to the maximum systematic error in concentration for any sample. This was used as a bias allowance
term, Bcal, in computing the expanded uncertainties for determination by external calibration:
¯ x) 1 Bcal
U EC ~ k?s(C
(3)
Again, it is assumed that all additional sources of uncertainty
¯ x), by replication of the entire analytical
are included in s(C
process.
Limit of detection
For external calibration, LODs were defined as three times the
standard deviation (s) for procedural blanks, while those for ID
were calculated according to the modified equation by Yu et al.;35
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
L2Da zR2y L2Db DF
ID
LD ~ (4)
Ax {Ry Bx DFEC
where LDa is the linear LOD obtained from external calibration
(typically the major isotope), and LDb that calculated as
LDa ~
3Ax
LDa Bx
(5)
for isotopes A and B (enriched isotope), Ax and Bx are the atom
fractions in the sample, Ry is the isotope amount ratio in the
spike, and DFID and DFEC represent the dilution factors for
isotope dilution and external calibration, respectively.
Results and discussion
Matrix separation
For the routine analysis of soil samples by ICP-SFMS, large
dilution factors (w2000) are necessary to prevent problems
with carryover contamination, clogging of the cones, etc. One
approach to circumvent such problems is to perform some
form of matrix separation prior to the instrumental measurement, which may also permit the use of lower dilution factors,
and hence, improve the limits of detection of the method.
Inagaki et al.9 employed NH3 precipitation to separate matrix
from the total digestion of sediment samples and determine Cd
by ID-ICP-QMS. In this way, isobaric interferences from Sn1
and ZrO1 were removed by co-precipitation of Sn and Zr
species with the major matrix elements, Al and Fe. Owing to
the inherent simplicity of this means of matrix separation
and the low blank values obtained,9 NH3 precipitation was
evaluated for the present application.
Table 2 exemplifies the results obtained from multi-element
analyses of a soil extract after matrix separation. (Note that
some liquid remained in the precipitate after centrifugation and
decanting the supernatant, thus partly explaining the nonquantitative recoveries of elements such as Na.) Precipitation is
clearly efficacious for the removal of the major components
(Al, Fe), but also for some potentially troublesome oxideforming elements, such as P, Ti and Zr.36–38 Sn is almost
completely precipitated, but since Mo is not efficiently
separated, as noted by Inagaki et al.,9 mathematical correction
for MoO1 interferences on Cd isotopes will still be
required.23,39,40 In the soils analysed, the concentrations of
both Mo and Sn were comparable to, or much lower than,
respectively, those of Cd, thus requiring minimal levels of
correction. The fact that a substantial proportion of W persists
Table 2 Percentages of elements remaining in the decanted solution
phase following NH3 precipitation of an aliquot of HNO3 extract of
soil sample M1a
Element
Solution (%)
Element
Solution (%)
Ag
Al
As
Ba
Be
Bi
Ca
Cd
REEb
Cl
Co
Cr
Cs
Cu
Fe
Ga
Hf
Hg
I
K
Li
84
3.3
1.4
72
0.8
v0.1
65
91
v0.04
86
89
1.3
91
91
0.05
7.7
0.1
97
92
92
50
Mg
Mn
Mo
Na
Nb
Ni
P
Pb
Rb
S
Sb
Sn
Sr
Th
Ti
Tl
U
W
V
Zn
Zr
13
72
75
93
0.3
91
0.3
0.3
91
69
9
0.8
75
0.1
v0.1
85
0.2
17
0.2
89
0.1
a
Relative standard uncertainties were typically about 2%.
earth elements.
b
Rare
J. Anal. At. Spectrom., 2004, 19, 858–866
861
in the dissolved phase also implies a similar requirement for
WO1 correction of Hg isotopic measurements. Other oxideforming elements are rarely present in soils at concentrations
likely to interfere with Hg measurement (Re, Rh, Os, Ta), or
are completely removed by precipitation (Hf; see Table 2). It is
important to perform phase separation soon after NH3 addition; after standing overnight, larger fractions of the soluble
elements are adsorbed on the precipitated phases, resulting in
lower recoveries.
Unfortunately, essentially all Cr and Pb is lost from solution
and thus isolated together with the bulk of the matrix. In
principle, it may be possible to further separate these elements
from the matrix by work-up of the precipitate, such as dissolution and ion exchange chromatography.41 Applying the anion
exchange method for Fe separation described previously,42,43
preliminary results showed that it was possible to recover 74%
of Pb and 93% of Cr, while eliminating w99% of Fe.
Spectral interferences
Limit of detection
The limits of detection (LODs) for the two calibration approaches and both digestion methods are summarized in Table 3.
Except for Pb, LODs for the acid mixture digestion are
significantly higher compared with those for the HNO3 leaching, which is perhaps expected considering the use of less pure
reagents, lower sample to reagent ratio and higher sample
dilution.
When comparing the two calibration approaches, poorer
Table 3 Limits of detection for the analysis of soils by ICP-SFMS
using external calibration and isotope dilution
Limit of detection/mg g21
Elementa
Method
External calb
Isotope dilutionc
As (HR)
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
0.01
0.3
0.007
0.02
0.002
0.006
0.007
0.14
0.01
0.05
0.009
0.6
0.1
12
0.01
0.04
0.02
0.2
0.04
0.1
0.04
0.07
0.3
0.7
0.003
0.003
0.001
0.03
0.03
0.4
0.006
0.2
0.0003
0.003
0.1
2
—
—
—
—
0.001
0.004
0.008
0.17
—
—
0.003
0.2
—
—
0.008
0.03
—
—
—
—
0.06
0.1
—
—
0.006
0.006
—
—
—
—
—
—
—
—
0.1
3
Be
Cd
Cr (MR)
Co (MR)
Cu (MR)
Fe (MR)
Hg
Mo
Mn (MR)
Ni (MR)
P (MR)
Pb
Sb
Ti (MR)
V (MR)
U
Zn (MR)
a
Measurements performed in medium (MR) or high (HR) resolution
mode, otherwise low resolution. b Calculated using the 3s criterion.
c
Calculated according to Yu et al.,35 eqn. (4)
862
LODs for ID are seen for Cr, Ni, Pb and Zn. At best, the ID
LOD is improved three-fold for Cu and about two-fold for Cd
and Hg. Eqn. (4) results in conservative estimates of LODs,
but accounts for the degree of isotopic enrichment of the spike
and the uncertainties in the measurement of the spiked
isotope.35 Results from an evaluation of ID made by Tibi
and Heumann,44 also demonstrate higher LODs for a number
of trace elements when using eqn. (4) compared with the
common 3s criteria. As a rule, LODs are below 1% of
the concentrations found in the investigated samples for the
majority of elements. However, As, Cu and Hg concentrations
in SO-2 are uncomfortably close to the corresponding LODs
for acid mixture digestion. For Hg, relatively high instrumental
blanks (corresponding to 10–15 ng l21 concentration levels in
solution, with Ar supply lines being the most probable source),
rather then reagent impurities and contamination during
handling, are responsible for the poorer LODs.
J. Anal. At. Spectrom., 2004, 19, 858–866
High resolution mass spectra were acquired for the elements
Cr, Ni, Cu and Zn to elucidate the sources of spectral interferences arising from the prepared soil samples. Some selected
examples are depicted in Fig. 1, showing the effects of matrix
separation by NH3 precipitation on an HNO3 extract of soil
sample M1. Since only minor amounts of Cr remain in solution
(see Table 2), matrix separation is ineffective and, in fact, the
ratio of interfering 40Ar12C1 to 52Cr1 increases considerably,
from 0.03 to 0.87 (Fig. 1(a)), suggesting that a larger fraction
of the C-containing species do not precipitate. For 67Zn1, the
spectral interference at m/z # 66.952 is attributable to
134
Ba21,44 as confirmed by exact mass measurements and
isotopic abundance ratios (measured 136Ba21/134Ba21 ~ 3.17;
natural abundance31 ratio 136Ba/134Ba ~ 3.24). The fractions
of Ba and Zn remaining in solution are similar (Table 2) and so
matrix separation by NH3 precipitation is not expected to
markedly decrease this spectral interference, as can be seen in
Fig. 1(b).
On the other hand, matrix separation was found to completely eliminate the spectral interferences observed on both Cu
isotopes, Fig. 1 (c,d). The most commonly cited interferences in
the recent literature are probably 40Ar23Na1 on 63Cu1 and
SiCl1 on both Cu isotopes.19,34,46,47 However, Table 2 shows
that precipitation would not substantially attenuate the Na
concentration; hence ArNa1 formation can be excluded as the
source of the interference on 63Cu1. In addition, the mass
difference between 63Cu1 and ArNa1 (#0.023 u) is greater
than that actually observed (#0.017 u). Microwave-assisted
HNO3 extraction will not bring significant amounts of Si into
solution, and the total digestion procedure would largely
eliminate Si as volatile SiF4.19,46,47 Thus, SiCl1 species are
unlikely candidates to explain the observed spectral interferences evident in Fig. 1 (c,d). Furthermore, the integrated
intensity ratio of the interferent species at masses 63/65 is 1.32,
whereas the natural and theoretical abundance ratios of Cu1
and SiCl1, respectively, are actually very similar, about 2.24,
which also argues against assignment of the interferences to
SiCl1 species. Consideration of the mass differences and
isotope ratios suggests that TiO1 is responsible,36–38,45 as substantiated by further high-resolution measurements at m/z 64
(major TiO1 isotopomer) and 66. This conclusion is also verified
by the fact that the interfering peaks are eliminated by matrix
separation, i.e., quantitative precipitation of Ti (Table 2).
Accuracy and precision
Of the analytes that are routinely determined in soil samples,
several are mono-isotopic (As, Be, Co, P) or else enriched
isotopes were unavailable in this laboratory at the time of this
study. Results for these elements are listed in Table 4: note that
Fig. 1 High resolution mass spectra acquired for (a) 52Cr (with 40Ar12C1 interferent on the high mass side), (b) 67Zn (134Ba21), (c) 63Cu (47Ti16O1)
and (d) 65Cu (49Ti16O1) isotopes in HNO3 extract of spiked soil sample M1 before (broken lines) and after (continuous lines) matrix separation
by NH3 precipitation. The scale-expanded right-hand axes highlight the spectrally-interfered regions on the higher mass sides of the vertical
dashed lines. Note that the 67Zn and 65Cu peaks in panels (b) and (d), respectively, contain major contributions from the added spike mixture. The
measured 47Ti16O1/49Ti16O1 isotopomer ratio is 1.32 (interferent peaks to the right in panels (c) and (d)), in agreement with the theoretical value
of 1.33.
Table 4 Comparison of mass fractions and recoveries obtained using two sample preparation methods and external calibration only
SO-2
SRM 2709
Cert (ref)a/
mg g21
Found/
mg g21
Element
Method
As
HNO3
Acid mixture
HNO3
Acid mixture
HNO3
Acid mixture
0.84
1.04
0.43
1.85
2.75
7.18
HNO3
Acid mixture
2.58 ¡ 0.21
5.24 ¡ 0.39
HNO3
Acid mixture
145 ¡ 13
685 ¡ 65
Be
Co
Fed
Mn
Mo
P
Sb
Ti
U
V
¡
¡
¡
¡
¡
¡
0.13
0.94
0.03
0.31
0.14
0.29
HNO3
Acid mixture
HNO3
Acid mixture
0.69
1.60
3170
3190
¡
¡
¡
¡
0.03
0.07
170
120
HNO3
Acid mixture
HNO3
Acid mixture
0.020
0.058
317
8130
¡
¡
¡
¡
0.004
0.016
37
880
HNO3
Acid mixture
HNO3
Acid mixture
0.420
0.655
26.9
57.8
¡
¡
¡
¡
0.044
0.026
1.1
1.6
(0.78 ¡
—
(0.49 ¡
(1.78 ¡
(2.60 ¡
9¡
(6.92 ¡
(2.61 ¡
5.56 ¡
(4.58 ¡
(137 ¡
720 ¡
(659 ¡
—
—
(3230 ¡
3000 ¡
(2940 ¡
—
(0.1)
—
8600 ¡
(7540 ¡
—
—
(26.9 ¡
64 ¡
(43.5 ¡
Recovery
(%)
0.12)
94
2.95 ¡ 0.19
3.52 ¡ 0.19
—
3.50 ¡ 0.11
100
3.10 ¡ 0.30
4.12 ¡ 0.24
95
518 ¡ 53
559 ¡ 41
(360–600)
538 ¡ 17
104
352 ¡ 36
580 ¡ 39
—
150)
200
90)
106
—
200
140)
95
—
2.0)
10
1.9)
90
—
17.7 ¡ 0.8
—
—
(10–15)
13.4 ¡ 0.7
Found/
mg g21
80
—
0.8
3.5
0.04
0.62
1.5
1.0
Recoveryc
(%)
15.3
18.0
0.72
2.94
12.5
13.0
—
0.07)
0.03)
0.23)
2
0.28)
1.30)
0.16
0.09)
6)
20
3)
Cert (ref)b/
mg g21
Found/
mg g21
¡
¡
¡
¡
¡
¡
M1
0.40
2.10
570
625
¡
¡
¡
¡
0.06
0.21
90
15
—
(2.0)
—
620 ¡ 50
0.026
7.7
53
3150
¡
¡
¡
¡
0.015
1.3
22
350
—
7.9 ¡ 0.6
—
3420 ¡ 240
1.58
2.86
48
118
¡
¡
¡
¡
0.03
0.24
2
11
—
(3)
(51–70)
112 ¡ 5
102
—
97
(105)
101
97
92
(95)
105
9.80
10.2
0.52
1.76
9.46
12.16
¡
¡
¡
¡
¡
¡
0.39
2.1
0.02
0.30
0.73
0.77
0.73
0.94
767
799
¡
¡
¡
¡
0.04
0.06
31
41
0.11
7.86
1640
4290
¡
¡
¡
¡
0.04
0.47
96
35
3.19
4.25
60.4
106.9
¡
¡
¡
¡
0.24
0.21
4.1
7.1
a
Reference values in parentheses in the HNO3 rows are mean values ¡1 s for routine measurements of SO-2 performed for quality assurance
purposes in this laboratory over the last two years using HNO3 leaching. Values in parentheses in the acid mixture rows are sums of sequential
extraction data (¡2 s) from Li et al.5 except for Sb, where an information value from the certificate of analysis is given. b Reference values in
parentheses are non-certified data (single values) or ranges of results obtained by HNO3 leaching according to US EPA Method 3050. c Recovery values in parentheses are calculated using non-certified total concentrations. d Fe concentrations are given as mass fractions in %. e Uncertainties are 95% confidence limits unless noted otherwise.
J. Anal. At. Spectrom., 2004, 19, 858–866
863
Table 5 Comparison of mass fractions and recoveries obtained using external calibration and isotope dilution for two sample preparation methods
SO-2
SRM 2709
b
Elementa
Method
Found /
mg g21
Cd (LR)
HNO3
HNO3 (ID)
Acid mix
Acid mix (ID)
HNO3
HNO3 (ID)
Acid mix
Acid mix (ID)
HNO3
HNO3 (ID)
HNO3 (ID)
Acid mix
Acid mix (ID)
Acid mix (ID)
HNO3
HNO3 (ID)
Acid mix
Acid mix (ID)
HNO3
HNO3 (ID)
HNO3 (ID)
Acid mix
Acid mix (ID)
Acid mix (ID)
HNO3
HNO3 (ID)
Acid mix
Acid mix (ID)
HNO3
HNO3 (ID)
Acid mix
Acid mix (ID)
0.066 ¡
(0.066)
0.140 ¡
0.133 ¡
5.31 ¡
(4.82)
9.05 ¡
8.65 ¡
3.84 ¡
—
(3.88)
3.76 ¡
4.36 ¡
4.27 ¡
0.080 ¡
(0.100)
0.073 ¡
0.085 ¡
3.38 ¡
—
(3.54)
4.67 ¡
4.70 ¡
4.94 ¡
5.64 ¡
(5.39)
18.2 ¡
18.3 ¡
49.8 ¡
(54.4)
113 ¡
114 ¡
Cr (MR)
Cu (MR)
(LR)
(MR)
(MR)
(LR)
(MR)
Hg (LR)
Ni (MR)
(LR)
(MR)
(MR)
(LR)
(MR)
Pb (LR)
Zn (MR)
c
Cert (Ref) /
mg g21
0.004
Recovery
(%)
(0.069 ¡ 0.009)
0.022
0.011
0.56
—
—
—
(5.28 ¡ 0.64)
0.48
0.54
0.58
16 ¡ 2
(11.7 ¡ 0.3)
(3.51 ¡ 0.30)
57
54
0.72
0.09
0.16
0.041
7¡1
(6.42 ¡ 0.20)
54
62
61
(0.095 ¡ 0.018)
0.016
0.018
0.22
0.082 ¡ 0.009
89
104
0.26
0.46
0.45
0.31
8¡2
(5.32 ¡ 0.45)
(5.43 ¡ 0.37)
0.6
0.8
2.5
21 ¡ 4
(18.2 ¡ 5.4)
(52.2 ¡ 2.5)
87
87
6
4
124 ¡ 5
(97.4 ¡ 1.3)
91
92
(3.18 ¡ 0.31)
58
59
62
b
Found /
mg g21
0.372 ¡
(0.366)
0.372 ¡
0.365 ¡
62.8 ¡
(61.9)
104 ¡
96.8 ¡
30.4 ¡
(31.1)
(30.8)
34.2 ¡
32.1 ¡
32.9 ¡
1.48 ¡
(1.57)
1.425 ¡
1.481 ¡
77.9 ¡
(73.8)
(71.6)
83.9 ¡
81.4 ¡
81.4 ¡
13.1 ¡
(12.7)
19.0 ¡
18.5 ¡
89 ¡
(89)
103 ¡
99.4 ¡
M1
d
Cert (Ref) /
mg g21
0.019
0.018
0.014
8.0
10
6.6
2.1
—
0.38 ¡ 0.01
130 ¡ 4
0.092
0.069
4.8
1.40 ¡ 0.08
8
5.2
80
75
(26–40)
34.6 ¡ 0.7
1.4
1.1
2
98
96
(60–115)
3.5
1.1
0.6
0.12
4.4
3.4
3.3
0.5
Recovery
(%)
99
93
95
—
102
108
(65–90)
88 ¡ 5
95
93
93
(12–18)
18.9 ¡ 0.5
100
93
(87–120)
106 ¡ 3
97
94
Found/
mg g21
0.779
0.780
0.774
0.787
53.9
53.0
81.5
78.6
121.2
120.0
122.9
122.3
120.7
120.1
0.222
0.274
0.264
0.287
20.5
20.7
19.3
25.0
24.3
23.3
26.7
25.6
36.5
36.2
68
73
81
80
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
¡
0.021
0.031
0.037
0.036
2.4
3.1
4.7
6.0
4.8
1.9
15.7
6.4
2.4
9.4
0.042
0.030
0.024
0.017
1.7
0.8
1.8
0.9
1.1
1.3
1.2
1.2
1.8
1.7
6
11
5
7
a
Measurements made in low- (LR) or medium-resolution (MR) mode. b Values in parentheses are for a single treated sample. c Reference
values in parentheses in the HNO3 rows are mean values ¡1 s for routine measurements of SO-2 performed for quality assurance purposes in
this laboratory over the last two years using HNO3 leaching. Values in parentheses in the acid mix (ID) rows are sums of sequential extraction
data (¡2 s) from Li et al.5 (Note that these values were obtained by ICP-OES and not ID!) d Reference ranges given in parentheses are results
obtained by HNO3 leaching according to US EPA Method 3050.
all data given in Tables 4 and 5 have been corrected for
moisture content determined on separate aliquots of the soil
samples. The reference material SO-2 is utilized for quality
assurance purposes, having been prepared and analysed in
parallel with routine samples in accordance with the Swedish
standard method (analogous to US EPA Method 3050) for
HNO3-leachable elements in soil over a long time. Data derived
from control charts for the HNO3-leachable fractions are
therefore included in Tables 4 and 5, as these values may be
useful for comparison with results obtained in other laboratories. For SRM 2709, corresponding data have been included
in an addendum to the original certificate of analysis, and the
present data are seen to fall within specified ranges, where
available.
Considering all the results for certified elements in SRM 2709
included in Tables 4 and 5, it is immediately apparent that
accurate data are readily obtained using external calibration
and mixed acid decomposition. Sample preparation using the
latter protocol differs only from the standard HNO3 leach in
terms of the digestion medium, and so can be routinely applied
with only marginally increased labour intensity. The only
element for which a significant difference is obtained is Cr (as
revealed by t-tests;48 results not shown), as may be expected
given that Cr is often present in environmental materials in the
form of refractory minerals, such as chromite.28 As such, more
aggressive acid mixtures or higher temperatures and pressures
are required to ensure complete dissolution.
The situation for SO-2 is less satisfactory, with recoveries
averaging at only 85% over the certified elements, excluding Cr
for reasons cited in the previous paragraph. Further neglecting
864
J. Anal. At. Spectrom., 2004, 19, 858–866
analytes for which the mean values found lie within the
confidence limits (Co, Hg, P, Pb, V) and those whose expanded
uncertainties encompass the certified values (Fe, Mn, Ti),
discrepancies remain for Cu, Ni and Zn. For this reason,
additional literature data5 for SO-2 have been included in
Tables 4 and 5 for comparison.
Li et al.5 applied a sequential extraction scheme to a variety
of reference materials prior to multi-elemental analysis by ICPOES. Total concentrations were presented as the sums of
values obtained in each of the five stages of extraction, the final
step in this scheme involving prolonged digestion at elevated
temperatures in a mixture of HNO3, HClO4 and HF. The total
values thus obtained for Ni and Zn were significantly lower
than the certified concentrations, but in agreement with our
data, as verified by t-tests at the 95% confidence level48 (not
shown). The only remaining concern is with Cu, as our data are
clearly at odds with the certified value as well as the result
reported by Li et al.,5 and no reasonable explanation for this
deviation can be found. Problems with spectral interferences
and gross calibration errors can be ruled out, however, as no
unexpected artifacts were revealed in the high-resolution mass
spectrum of the calibrant, and our Cu data for SRM 2709 are
commensurate with the certified value.
For the analytes to which both calibration strategies could be
applied (Table 5), ID gave higher results for Zn in the HNO3
leachate, as well as Cu in the mixed acid digest, of SO-2.
Further discussion of Cu is unwarranted, given that the
certified value for this element in SO-2 could not be reproduced. For Zn, which has a fairly high ionization potential
and a mass that is only half that of the internal standard,
non-spectral interferences, perhaps caused by such rudimentary effects as differences in acid strength,49–51 may not be
adequately corrected for in this particular case. Otherwise no
significant differences were obtained at the 95% confidence
level.
One important observation made during the analysis of the
unknown sample M1 using ID and HNO3 leaching should be
mentioned. Three replicates of M1 were extracted on one day
together with nine other samples and blanks, filling the
microwave oven carousel. The Pb concentration obtained
was 25.473 ¡ 0.035 mg g21 (1 s; n ~ 3), the standard
uncertainty being that for the three replicates only, i.e., not
including contributions from uncertainties in spike concentrations, etc. On another occasion, only five samples in total were
extracted using the same standard microwave program,
resulting in a Pb concentration of 25.727 ¡ 0.032 mg g21
(1 s; n ~ 3), which is significantly higher at the 99.9% confidence level. The difference is primarily ascribable to variations in the degree of Pb extraction, since the lower total sample
mass in the latter experiment would result in a higher final
temperature,27 and hence more efficient release of Pb from the
soil. Therefore, one of the prerequisites for application of ID,
viz., equilibration between incipient analyte and spike, is clearly
not fulfilled by HNO3 leaching.
Regarding the time required for instrumental analysis, 18
elements could be determined in 5.5 min (including uptake and
wash-out times) using external calibration with two internal
standards. For ID-ICP-SFMS, the time required per sample
was about six times longer using the conditions given in
Table 1. If some sacrifice in isotope ratio measurement precision were acceptable, then it would be possible to shorten the
data acquisition period.
As the virtues of ID in terms of improving precision are
often expounded, F-tests48 were used to compare uncertainties.
Although ID yielded better precision in most cases, as is
evident in Table 5, significant improvements were observed in
only a few isolated examples, e.g., Cu in SO-2 and SRM 2709.
It should be remembered that matrix separation was applied
prior to measurement of Cu and Ni, allowing their determination in less diluted solutions. Thus much of the gain in precision
is to be expected from the improved counting statistics.52,53
These elements were also determined by ID-ICP-SFMS in low
resolution mode, as spectrally interfering sample components
were effectively eliminated by matrix separation (Fig. 1). The
results for Cu showed improved precision, though not statistically significantly so, whereas those for Ni were independent of
resolution. Although more precise isotope ratios were generally
obtained for both elements in low resolution, other factors are
decisive for the uncertainty budget. Concentrations of Cu
and Ni derived from the isotope dilution equation, eqn. (1),
displayed similar variability between sub-samples, ruling out
differences in heterogeneity as a probable cause. Instead, the
uncertainty budget for Ni is highly affected by the bias
allowance term, eqn. (2),33 for the concentration of the spike.
separation, is mandatory for the determination of these
elements in soils. The only element studied for which spectral
interferences could not be completely eliminated using matrix
separation or ICP-SFMS was Cd. For the soils analysed here,
MoO1 formation was minimal and could be readily corrected
mathematically,23,38,39 allowing accurate quantification of Cd
by both ID and external calibration.
For the multi-elemental analysis of soils, no clear-cut benefits
of ID over external calibration using internal standardization
with respect to accuracy and precision were actually observed
(Table 5). In combination with matrix separation, however, the
advantages of ID become more apparent, since complete
recoveries of the analytes were not obtained following NH3
precipitation,9 as seen in Table 2. For elements remaining
principally in the liquid phase (Cd, Cu, Hg, Ni, Zn), reduction
of the matrix loading allows more concentrated solutions to be
introduced to ICP-SFMS, which can be used to compensate for
the poorer detection limits of ID compared with external
calibration for a fixed dilution factor35 (see Table 3).
Considering that much more effort and instrument time is
required, the use of ID-ICP-SFMS is not an economically
viable option for routine multi-elemental analyses of soils. The
results presented in Table 5 suggest that ICP-SFMS, in combination with external calibration and internal standardization,
can actually provide results of comparable quality, with much
less effort for a greater range of elements.
Conclusions
13
High resolution mass spectra revealed TiO1 and Ba21 as
prominent spectral interferences in the mass region 60–68 u
during the analysis of HNO3 leachates (Fig. 1) and mixed acid
digests of soils. Although common sources of concern in the
infancy of ICP-QMS,35–37,45 these spectral interferents appear
to have fallen into neglect in much of the more recent
literature,19,34,46,47 which is surprising considering the ubiquity
of Ba and Ti in soils, sediments and geological materials. It may
be prudent to note that TiO1 will interfere with all Cu and Zn
isotopes,37 as well as 62Ni, requiring an instrumental mass
resolution of at least 3200. Therefore exploitation of the high
resolution capabilities of ICP-SFMS, or the use of matrix
14
Acknowledgements
This work was supported by EU’s structural fund for Objective
1 Norra Norrland and the Centre for Isotope and Trace
Element Measurement, Lulea˚ University of Technology.
Analytica AB provided invaluable technical and financial
assistance.
References
1
2
3
4
5
6
7
8
9
10
11
12
15
16
17
18
19
20
M. V. Ruby, R. Schoof, W. Brattin, M. Goldade, G. Post,
M. Harnois, D. E. Mosby, S. W. Casteel, W. Berti, M. Carpenter,
D. Edwards, D. Cragin and W. Chappell, Environ. Sci. Technol.,
1999, 33, 3697.
J. Sastre, A. Sahuquillo, M. Vidal and G. Rauret, Anal. Chim.
Acta, 2002, 462, 59.
N. H. Suhr and C. O. Ingamells, Anal. Chem., 1966, 38, 730.
W. Diegor, H. Longerich, T. Abrajano and I. Horn, Anal. Chim.
Acta, 2001, 431, 195.
X. Li, B. J. Coles, M. H. Ramsey and I. Thornton, Chem. Geol.,
1995, 124, 109.
K. E. Murphy, E. S. Beary, M. S. Rearick and R. D. Vocke,
Fresenius’ J. Anal. Chem., 2000, 368, 362.
O. Mestek, R. Koplı´k, H. Fingerova´ and M. Sucha´nek, J. Anal.
At. Spectrom., 2000, 15, 403.
M. Hoenig, Talanta, 2001, 54, 1021.
K. Inagaki, A. Takatsu, A. Uchiumi, A. Nakama and
K. Okamoto, J. Anal. At. Spectrom., 2001, 16, 1370.
R. Walker and I. Lumley, Trends Anal. Chem., 1999, 18, 594.
J. Namies´nik and B. Zygmunt, Sci. Total Environ., 1999, 228, 243.
B. de Guillebon, F. Pannier, F. Seby, D. Bennink and
Ph. Quevauviller, Trends Anal. Chem., 2001, 20, 160.
Ph. Quevauviller and O. F. X. Donard, Trends Anal. Chem., 2001,
20, 600.
I. Feldman, W. Tittes, N. Jakubowski, D. Stuewer and
U. Giessmann, J. Anal. At. Spectrom., 1994, 9, 1007.
H. Klinkenberg, W. Van Borm and F. Souren, Spectrochim. Acta,
Part B, 1996, 51, 139.
J. R. Moody and M. S. Epstein, Spectrochim. Acta, Part B, 1991,
46, 1571.
A. G. Adriaens, W. R. Kelly and F. C. Adams, Anal. Chem., 1993,
65, 660.
R. L. Watters, Jr., K. R. Eberhardt, E. S. Beary and J. D. Fassett,
Metrologia, 1997, 34, 87.
I. Papadakis, P. D. P. Taylor and P. De Bie`vre, J. Anal. At.
Spectrom., 1997, 12, 791.
C. S. J. Wolff Briche, C. Harrington, T. Catterick and B. Fairman,
Anal. Chim. Acta, 2001, 437, 1.
J. Anal. At. Spectrom., 2004, 19, 858–866
865
21 J. D. Fassett and P. J. Paulsen, Anal. Chem., 1989, 61, 643A.
22 Ph. Quevauviller, E. A. Maier, B. Griepink, U. Fortunati,
K. Vercoutere and H. Muntau, Trends Anal. Chem., 1996, 15, 504.
23 I. Rodushkin, M. D. Axelsson and E. Burman, Talanta, 2000, 51,
743.
¨ hlander, Analyst,
24 M. D. Axelsson, I. Rodushkin, J. Ingri and B. O
2002, 127, 76.
25 R. A. Nadkarni, Anal. Chem., 1984, 56, 2233.
26 T. Prohaska, M. Watskins, C. Latkoczy, W. W. Wenzel and
G. Stingeder, J. Anal. At. Spectrom., 2000, 15, 365.
27 H. M. Kingston and L. B. Jassie, in Introduction to Microwave
Sample Preparation: Theory and Practice, eds. H. M. Kingston and
L. B. Jassie, American Chemical Society, Washington, DC, 1988,
ch. 6, pp. 93–154.
28 W. R. Kelly, K. E. Murphy, D. A. Becker and J. L. Mann, J. Anal.
At. Spectrom., 2003, 18, 166.
29 J. I. Garcı´a Alonso, Anal. Chim. Acta, 1995, 312, 57.
30 P. D. P. Taylor, P. De Bie`vre, A. J. Walder and A. Entwistle,
J. Anal. At. Spectrom., 1995, 10, 395.
31 J. R. De Laeter, J. K. Bo¨hlke, P. De Bie`vre, H. Hidaka, H. S. Peiser,
K. J. R. Rosman and P. D. P. Taylor, Pure Appl. Chem., 2003, 75, 683.
32 T. F. D. Mason, D. J. Weiss, M. Horstwood, R. R. Parrish,
S. S. Russell, E. Mullane and B. J. Coles, J. Anal. At. Spectrom.,
2004, 19, 218.
33 S. B. Schiller and K. R. Eberhardt, Spectrochim. Acta, Part B,
1991, 46, 1607.
34 T. Prohaska, C. R. Que´tel, C. Hennessy, D. Liesegang,
I. Papadakis, P. D. P. Taylor, C. Latkoczy, S. Hann and
G. Stingeder, J. Environ. Monit., 2000, 2, 613.
35 L. L. Yu, J. D. Fassett and W. F. Guthrie, Anal. Chem., 2002, 74,
3887.
866
J. Anal. At. Spectrom., 2004, 19, 858–866
36 M. A. Vaughan and G. Horlick, Appl. Spectrosc., 1986, 40, 434.
37 S. H. Tan and G. Horlick, Appl. Spectrosc., 1986, 40, 445.
38 L. L. Burton and G. Horlick, Spectrochim. Acta, Part B, 1992, 47,
E1621.
39 S. J. Jiang, M. D. Palieri, J. S. Fritz and R. S. Houk, Anal. Chim.
Acta, 1987, 200, 559.
40 J. L. M. de Boer, J. Anal. At. Spectrom., 2000, 15, 1157.
41 J. Korkisch, Handbook of Ion Exchange Resins, CRC Press, Boca
Raton, 1989.
42 A. Stenberg, D. Malinovsky, I. Rodushkin, H. Andre´n,
¨ hlander and D. C. Baxter, J. Anal. At. Spectrom.,
C. Ponte´r, B. O
2003, 18, 23.
43 D. Malinovsky, A. Stenberg, I. Rodushkin, H. Andre´n, J. Ingri,
¨ hlander and D. C. Baxter, J. Anal. At. Spectrom., 2003, 18,
B. O
687.
44 M. Tibi and K. G. Heumann, Anal. Bioanal. Chem., 2003, 377,
126.
45 V. Balaram, C. Mankyamba, S. L. Ramesh and K. V. Anjaiah, At.
Spectrosc., 1992, 13, 19.
46 R. F. J. Dams, J. Goosens and L. Moens, Mikrochim. Acta, 1995,
119, 277.
47 F. Vanhaecke, L. Moens, R. Dams, I. Papadakis and P. Taylor,
Anal. Chem., 1997, 69, 268.
48 J. C. Miller and J. N. Miller, Statistics for Analytical Chemistry,
Ellis Horwood, Chichester, 1988, 2nd edn., ch. 2, pp. 33–52.
49 J. J. Thompson and R. S. Houk, Appl. Spectrosc., 1987, 41, 801.
50 F. Vanhaecke, H. Vanhoe, R. Dams and C. Vandecasteele,
Talanta, 1992, 39, 737.
51 E. Bjo¨rn and W. Frech, J. Anal. At. Spectrom., 2001, 16, 4.
52 J. M. Hayes and D. A. Schoeller, Anal. Chem., 1977, 49, 306.
53 I. S. Begley and B. Sharp, J. Anal. At. Spectrom., 1994, 9, 171.