The accurate determination of potassium and calcium using isotope
dilution inductively coupled ‘‘cold’’ plasma mass spectrometry
Karen E. Murphy,*a Stephen E. Long,a Michael S. Rearicka and Özlem S. Ertasb
a
National Institute of Standards and Technology, 100 Bureau Drive, Stop 8391, Gaithersburg,
MD 20899, USA. E-mail: [email protected]
b
Ege University Faculty of Pharmacy, Analytical Chemistry Department, 35100 Bornova,
Izmir, Turkey
Received 3rd January 2002, Accepted 7th March 2002
First published as an Advance Article on the web 9th April 2002
We describe an accurate method for the determination of K and Ca using isotope dilution inductively coupled
‘‘cold’’ plasma mass spectrometry (ID ICP-MS). Measurements were applied to the certification of K and Ca in
a new oyster tissue standard reference material, SRM 1566b, and the determination of Ca for an international
measurement comparability exercise for serum. Measurements were made using an ICP-MS operated in the
cold plasma mode. Molecular ion interferences arising from the plasma were reduced below 1000 counts per
second (cps) for the isotopes of interest, while maintaining a sensitivity of better than 9 6 106 cps for a
1 mg L21 Ca solution. Detection limits of 1.3 ng L21 and 2.5 ng L21 were obtained for K and Ca, respectively.
Isotope ratio measurement repeatability of the 40Ca/42Ca and 39K/41K ratios for spiked samples was better than
0.2% relative (n ~ 5, 1 s). Though interference from background peaks was reduced, molecular ions arising
from the oyster tissue and serum matrices caused spectral interference. Reduction of the matrix induced
interference was successfully accomplished using cation exchange chromatography. Data that demonstrate the
reproducibility and accuracy of the ID ICP-MS measurements are presented. A split sample comparison with
thermal ionization mass spectrometry (TIMS) was performed.
Introduction
To be useful, analytical measurements must be accurate. When
relied upon for medical diagnosis, or to determine the nutritional content of food, they must also be rapid and inexpensive.
Indeed, for clinical measurements, accuracy and expense are
related. The German Health Report of 1998 states that the cost
of performing repeat measurements on patient samples in
Germany alone amounted to 1.5 billion US dollars.1 Normalized to the US GDP for that year, it is estimated that this would
cost the US health care consumer $7.4 billion.2 The clinical
chemistry community has long recognized the importance of
reference methods to validate the results of rapid, routine test
systems.3–7 The recent European directive on in vitro diagnostics also mandates such a scheme, requiring that the results
of routine methods in laboratory medicine be traceable to the
highest available reference material or method.8 Reference
methods play a critical role in establishing traceability, not only
in the validation of routine methods, but also in the certification of reference materials.
The accurate determination of K and Ca is important. Both
elements are essential in our diet. The Ca content must appear
on the labels of all foods processed in the US, as required by the
Nutrition Labeling and Education Act of 1990.9 An excess or
deficiency of either element will result in adverse health consequences. Calcium is needed for bone strength to combat
osteoporosis, for blood clotting, for proper nerve and muscle
function and for strong, healthy teeth. Too much calcium may
be risky because calcium can inhibit the absorption of iron, zinc
and magnesium.10 Potassium is also required for proper nerve
and muscle function, as well as for maintaining the fluid
balance of cells and heart rhythm. High levels of potassium in
the body can cause heart problems and even death.11 As a
result of nutrition labeling requirements and the rapid growth
of the in vitro diagnostic market, the number of methods to
determine K and Ca abounds. However, the influx of new and
DOI: 10.1039/b200029f
rapid techniques provides no assurance that the accuracy of
results has kept pace.
Isotope dilution thermal ionization mass spectrometry
(TIMS) has traditionally provided the accuracy base for the
determination of K and Ca.12–14 Isotope dilution mass spectrometry has powerful advantages over conventional methods of
quantitation. First, use of an enriched isotope of the element of
interest as the internal standard is an ideal situation because the
chemical behavior of the standard and analyte is identical.
Once equilibrated (which is easily accomplished for inorganic
analytes by complete dissolution), the analyte can be isolated
from the matrix without fear of systematic error due to
loss. Second, intensity ratios rather than absolute signals are
determined in the mass spectrometer. The measurement of
ratios can be accomplished very precisely. Inductively coupled
plasma mass spectrometry (ICP-MS) provides an attractive
alternative to TIMS because of its rapid measurement ability
and subsequent wide usage. The repeatability (precision) of
ratios measured by ICP-MS is typically 0.05–0.3%, relative.15–17
However, high background from the Ar plasma has hindered
the use of ICP-MS for the measurement of K and Ca isotopes.
Difficulty in the determination of K and Ca by ICP-MS
stems from intense background peaks at masses 39, 40, 41
and 42 from 38Ar1H1, 40Ar1, 40Ar1H1 and 40Ar1H21. Various
means have been used to reduce the magnitude of these
interferences in an ICP-MS, including desolvation, high resolution, and dynamic reaction and multipole collision cells.18–26
Stürup et al. used high resolution ICP-MS to determine
44
Ca/42Ca and 43Ca/42Ca ratios in urine.19 Polyatomic ions
could be resolved from the 42, 43, and 44 Ca isotopes at a
resolution greater than 2700, but 40Ca could not be resolved
from 40Ar. They observed interference from Sr21. This also
could not be resolved, and a correction was applied.19 For the
determination of K by high resolution ICP-MS, Becker and
Dietze obtained 0.7% repeatability (1 s, n ~ 5) and 2.3%
accuracy for the 41K/39K ratio measured in SRM 985, a KCl
J. Anal. At. Spectrom., 2002, 17, 469–477
This journal is # The Royal Society of Chemistry 2002
469
isotopic and assay standard.20 At a resolution of 9000, separation of 40Ar1H from 41K could be achieved, but was not
complete.
The use of a quadrupole or multipole pressurized with a
reactive gas has been shown to reduce plasma based interferences for K and Ca as well as a number of other analytes prone
to interference.21–26 Tanner et al. achieved detection limits of
0.4 ng L21 for 40Ca and 0.35 ng L21 for 39K.22 Reaction with a
collision gas serves not only to reduce plasma induced
interference, but is believed to cause collisional damping of
ion beam fluctuations, allowing for precise isotope ratio
measurement.23 Boulyga and Becker obtained repeatability
(1 s, n ~ 6) of 0.26% and 0.40% for 40Ca/44Ca and 42Ca/44Ca
ratios measured in NIST SRM 915, Calcium Carbonate, at a
Ca concentration of 10 mg L21.24 Repeatability better than
0.01% for 42Ca/40Ca has been obtained using a multicollector
ICP-MS with a collision cell.25 Though collision cells appear as
a very promising alternative to high resolution ICP-MS for the
determination of analytes with spectral interference, applications to date have focused primarily on analysis of high-purity
acids and matrix free samples. Feldmann found that use of H2
as a reaction gas was effective in reducing polyatomic ions
caused by Ar, but matrix induced polyatomic ions were
unaffected.26 Operation of the dynamic reaction cell can be
achieved under a single set of plasma conditions; however some
elements are better analyzed in a non-pressurized mode which
requires switching between pressurized and vented modes for
full elemental coverage. The need for additional high-purity
gases and the inability to upgrade current instrumentation add
to the expense of using reaction cell technology. Use of cold
plasma to reduce Ar-based interference, however, requires only
slight modification of existing instrumentation. Recent generation instruments usually come ready for operation in cold
plasma mode.
Low power (‘‘cold plasma’’) conditions along with a
configuration to attenuate the secondary discharge has been
shown to significantly reduce Ar-based interferences.27,28
Sakata and Kawabata used a forward power of 900 W, a
higher carrier gas flow (1.1 L min21), and a grounded metal
shield inserted between the torch and the load coil to significantly reduce the polyatomic argide ion background.28 They
obtained detection limits of 1.1 ng L21 for K and 20 ng L21 for
Ca. Use of a large mechanical vacuum pump to lower the
pressure in the expansion chamber serves to recover any loss in
sensitivity due to cold plasma conditions, resulting in further
improvement of detection limits.29 Instrumentation which
allows for operation in the cold plasma mode has been
commercially available for several years, yet few publications
have appeared until recently.29–33 One reason for limited use
is that the formation of matrix induced polyatomic ions is
enhanced under cold plasma conditions. As a result, application of cold plasma primarily occurs in the semiconductor
industry for analysis of high purity reagents. However,
reduction of matrix effects can be achieved by chemical
separation. Separations are more easily applied when quantifying by isotope dilution because once isotopic equilibration
between the sample and the enriched isotope internal standard
is achieved, total analyte recovery is not required for accurate
results. Cold plasma ID ICP-MS determinations of Fe and Cr
have been made after matrix separation.34 Patterson et al.
applied an ammonium oxalate precipitation at pH 8 to separate
Ca in serum, urine, feces and breast milk for measurement of
Ca ratios in nutritional studies using stable isotope tracers.29
We describe a reference method for the determination of
calcium as well as potassium in oyster tissue and serum
matrices using cold plasma ICP-MS. Application of cation
exchange chromatography completely removed matrixinduced interference and allowed for the determination of
both elements in a single sample aliquot. We demonstrate the
comparability of cold plasma ID-ICP-MS methodology with
470
J. Anal. At. Spectrom., 2002, 17, 469–477
the previously applied ID-TIMS definitive method used in this
laboratory to provide traceability to the SI for clinical
measurements. This demonstration is made directly using the
two techniques for ratio measurements on the same isotope
dilution samples. For calcium, it is also made indirectly
through the analysis of SRM 956a, which was certified in 1996
by ID-TIMS. In addition, the serum calcium results have been
compared with those from other National Measurement
Institutes using different isotope dilution methodologies and
instrumentation. The results for the oyster tissue are compared
with NIST nuclear methods that were also applied in the
certification of this material. The critical evaluation and
application of new methodology capable of accurate measurement, as is described here, will be required to satisfy the
increased need for traceability and global comparability in
chemical measurement.
Experimental
Materials and reagents
Concentrated, high-purity nitric and hydrochloric acids
(Optima grade) were obtained from{ Fisher Scientific,
Pittsburgh, USA. Concentrated perchloric and hydrofluoric
acids and de-ionized water were purified in-house by subboiling distillation. Solutions containing 0.3 mol L21,
0.5 mol L21, 1.0 mol L21 and 1.75 mol L21 HCl were prepared from the purified reagents. Bio-Rad AG 50W-X8,
100–200 mesh cation exchange resin in the H1 form was used
for the separation of K and Ca. The resin was bulk-cleaned in
300 mL batches and cleaned again in individual columns by
washing alternatively with water and 5 mol L21 HCl. Resin was
discarded after a single use. Enriched isotopes, 41K (99.18%,
series #149401) as KCl, and 42Ca (94.42%, series #139601) as
CaCO3 were obtained from Oak Ridge National Laboratory
(Oak Ridge, Tennessee, USA). Stock solutions containing
28.6 mmol g21 and 6.3 mmol g21 total K and Ca, respectively,
were prepared in 0.5 mol L21 HNO3. For the determination of
Ca in serum, a working 42Ca spike solution was prepared by
gravimetric dilution of the master stock solution to yield a
concentration of approximately 1.0 mmol g21. The exact
concentration of enriched isotope in each spike solution was
calibrated against fresh, gravimetrically prepared stock solutions of SRM 918a Potassium Chloride and SRM 915a
Calcium Carbonate primary standards. For Ca, SRM 3109a
Calcium Standard Solution (lot #000622) was used as well. For
accuracy assessment, one vial of SRM 956a Electrolytes in
Frozen Human Serum, Level 2, was analyzed for Ca, and one
bottle of SRM 1566a Oyster Tissue was analyzed for K as well
as Ca.
Instrumentation
ICP-MS measurements were made using a VG PlasmaQuad
3 (Thermo Elemental, Winsford, Cheshire, England). The
instrument was operated with the plasma screen and
S-option pump under low power conditions. Operating parameters are listed in Table 1. Solution was introduced to the
plasma via a concentric nebulizer and impact bead spray
chamber water cooled to 4 uC. Solution was pumped at rate of
0.6 mL min21. The auxiliary gas flow rate and torch position
had a significant effect on the intensity of the background
plasma-based molecular ions and these parameters were
optimized to achieve the best signal to background ratio for
{Certain commercial equipment, instruments, or materials are
identified in this report to adequately specify the experimental
procedure. Such identification does not imply recommendation or
endorsement by the National Institute of Standards and Technology,
nor does it imply that the materials or equipment identified are
necessarily the best available for this purpose.
Table 1 Instrument parameters
Parameter
K
Ca
Forward power/W
Gas flows/L min21
Coolant flow
Auxilliary flow
Nebulizer flow
Torch position
X
Y,Z
Lens voltages/V
Extraction
Collector
Lens 2
Lens 3
Lens 4
Pole Bias
630
630
12.6
1.4
0.92
12.6
1.0
0.94
100 units back from typical hot plasma setting
Optimized for maximum signal to background
263
21.8
0.1
26.6
2107
26.5
243
21.0
20.8
27.1
2120
26.5
each element. For Ca the auxiliary gas flow rate was typically
1.0 L min21 and for K it was 1.4 L min21. Mass calibrations
and lens voltages were optimized to yield the best sensitivity for
each element.
TIMS measurements were made using an in-house designed
solid source mass spectrometer with a 90u single magnetic
sector of 30.5 cm radius of curvature.35 A dc filament supply
having current control was used. Ion intensities were measured
by Faraday cup collection. The temperature of the filament was
measured using an optical pyrometer. Both Re and Ta zone
refined flat filaments of 0.0025 cm thickness and 0.076 cm width
were employed. Prior to use, the filaments were outgassed for
30 min at 1500 uC at a pressure of 1.3 6 1025 Pa.
Sample description and preparation
Oyster tissue. One sample from each of six statistically
selected bottles of the candidate SRM 1566b was analyzed.
SRM 1566b is composed of oysters collected in the Gulf of
Mexico. These were shucked, washed, ground, freeze-dried, reground and sterilized before bottling. Subsamples of approximately 0.25 g were weighed by difference into clean Teflon1
beakers. Loss of mass upon drying was determined for each
bottle using separate samples removed at the same time as the
analytical samples. Accurately weighed aliquots of enriched
41
K and 42Ca spike solutions were added to the analytical
samples by mass difference via capped plastic syringes. The
approximate concentration of K and Ca in the new oyster
tissue was known, and spike was added to form a 39K/41K ratio
of 0.95 and a 40Ca/42Ca ratio of 1.1. These ratios represent a
compromise between the optimum ratio for measurement on
pulse counting systems and the minimization of uncertainty
due to error propagation.36 Subsamples of SRM 1566a, SRM
918a, and SRM 915a were treated in a similar manner.
Samples were dissolved by wet-ashing with nitric (HNO3),
hydrofluoric (HF), and perchloric (HClO4) acids. Hydrofluoric
acid was used because it is possible that some sediment is
present with the oyster tissue. We find HClO4 to be the most
effective acid for the decomposition of organic matter in
biological matrices and this acid was used in the decomposition
of the serum samples as well. Our intent is always to attain the
most complete dissolution of the sample and to achieve the
highest oxidation state for the metals present. This ensures
equilibration of the sample and spike isotopes. However, great
care must be exercised when using HF and HClO4. It is
important to decompose the sample with HNO3 prior to the
addition of HClO4. Eight grams of nitric acid were added, and
the samples refluxed in a class 10 laminar flow hood until fumes
of NO2 were no longer observed. The solution volume was
reduced and the samples cooled. Four grams of HF and 3 g of
HClO4 were added, and the samples gently heated. Heating was
increased and the samples allowed to reflux again. Solution
volumes were reduced to fumes of HClO4 and samples were
gently taken to dryness. The residue was washed several times
with H2O to remove excess perchlorates and then converted to
the chloride by evaporating to dryness several times with
hydrochloric acid (HCl).
Serum. The Ca content of an unmodified human liquid
serum was determined as part of a study to compare existing
measurement capabilities of national metrology institutes for
clinical analytes in blood serum. The comparison is an activity
of the Inorganic Analysis Working Group of the Consultative
Committee for Amount of Substance (CCQM). The CCQM is
one of the consultative committees of the Bureau International
des Poids et Mesures (BIPM), whose task is to ensure
worldwide uniformity of measurements and traceability to
the SI.37 The study, titled ‘‘CCQM-P14, Calcium in Serum’’
was piloted by the Institute for Reference Materials and
Measurements (IRMM), Belgium, and SP Swedish National
Testing and Research Institute, Sweden.38 The same material is
in use for another, ongoing inter-laboratory comparison.
One sample from each of five vials of serum was analyzed.
Prior to use, the vials were allowed to thaw for 12 h to
equilibrate to room temperature. The ampoules were gently
rotated and inverted twice to thoroughly mix the contents and
subsamples of approximately 2 g were weighed by difference
into clean Teflon1 beakers. Each aliquot was then spiked with
an accurately weighed aliquot (2 g) of 42Ca spike solution. Subsamples of SRM 956a, Level 2, and SRM 915a were prepared
in a similar manner. The beakers were transferred to a hot plate
in a class 10 laminar flow hood and 4 g of HNO3 were added to
each sample, control, and blank. The samples were then
refluxed until NO2 fumes were no longer observed, following
which 4 g of HClO4 were added, the samples refluxed again and
then heated to dryness. The resulting crystalline residues were
rinsed with quartz-distilled water and HCl and taken to
dryness. This process was repeated several times to drive off
residual HClO4 and convert the salts to the chloride in
preparation for separations using cation exchange chromatography.
Blanks and spike calibration mixes. Process blanks were
prepared. A small aliquot (#100–500 ng) of 41K and 42Ca spike
was added to clean Teflon beakers and processed in a manner
identical to the samples. In addition to the samples and process
blanks, spike calibration mixes were prepared. The purpose of
these mixes is to exactly calibrate the concentration of enriched
isotope in the spike solution against primary standards by
reverse isotope dilution mass spectrometry. Two separate,
gravimetrically prepared primary standard solutions were
prepared for each element. Standard solutions were prepared
as previously described.39 Subsamples (#0.25 g) of the primary
standards were weighed to ¡0.01 mg on a microbalance. Two
weighed aliquots of each standard solution were added to
aliquots of the spike solution to form four calibration mixes per
element.
Cation exchange separation. Samples were separated using
columns (0.7–1 cm id) which were hand-packed with 5 mL of
AG 50W-X8, 100–200 mesh cation exchange resin. The packed
columns were cleaned by washing alternatively with water and
5 mol L21 HCl, and conditioned with water. Samples were
loaded in 2 mL water. The Na fraction was eluted with 60 mL
of 0.3 mol L21 HCl, and discarded. The K fraction was eluted
with 40 mL of 0.5 mol L21 HCl, followed by Mg, which was
eluted with 30 mL of 1 mol L21 HCl. The Ca fraction was
eluted last with 20 mL of 1.75 mol L21 HCl. The elution
progress of the Na, K and Ca fractions was monitored by flame
J. Anal. At. Spectrom., 2002, 17, 469–477
471
emission using a Bunsen burner. The K and Ca fractions were
taken to dryness, treated with concentrated HNO3 to remove
chloride, and diluted to the appropriate concentration with a
volume fraction of 2% HNO3 in water.
Mass spectrometric procedures
ICP-MS. Measurements were made using peak jump data
acquisition with one point per peak. Five blocks of data, each
1 min in duration, were acquired per sample, and the mean
ratios used for computations. Total time for uptake, stabilization, and measurement was about 8 min per sample. For spiked
potassium samples, masses 39 and 41 were measured at dwell
times of 10 ms each. Sample concentrations were calculated
from the 39K/41K ratio. For spiked calcium samples, masses 39,
40 and 42 were measured at dwell times of 5, 10 and 10 ms,
respectively. Mass 39 was measured to determine the correction
for isobaric interference on 40Ca from 40K but no correction
was required for either the oyster tissue or serum samples.
Sample concentration data were calculated from the 42Ca/40Ca
ratio.
Measured ratios were corrected for mass bias, drift, and
detector dead time. A mass bias correction factor was obtained
at the start of the analysis by normalizing the experimental
ratio to the true ratio for isotopic standards of natural
composition. This factor was then used to correct the measured
ratio of a spike calibration sample measured immediately after.
This mass bias corrected spike calibration sample was labeled
the ‘‘working isotopic standard’’.40 The working isotopic
standard had an isotopic ratio similar to the analytical samples
and was re-measured throughout the analysis and used to
correct the remaining calibration samples, analytical samples,
and blanks for mass bias and any subsequent instrument drift.
Any change from the mass bias factor established at the start of
the analysis was assumed to be drift. A correction was applied
every three samples, assuming the drift to be linear with time.
Detector dead time was experimentally determined using
natural Mg and Ti solutions diluted so that the major isotope
spanned the count rate range of 1 6 105 to 9 6 105 counts per
second. The dead time that resulted in the least variability of
the measured ratios (24Mg/26Mg, 48Ti/49Ti) across the concentration range was chosen. To minimize the correction for
detector dead time, samples were diluted so that the count rate
of the major isotope was below 300 000 counts per second (cps).
Two dilutions were required to reach the analysis concentration for each element. The final dilution was prepared fresh
on the day of the analysis. Potassium was diluted with a volume
fraction of 4% HNO3 in water to an analysis concentration of
9.5 ng g21, yielding a count rate of 240 000 cps 39K for a natural
composition standard. Calcium was diluted to an analysis
concentration of 28 ng g21, yielding a count rate of 250 000 cps
40
Ca for a natural composition standard. It was discovered that
the use of LDPE bottles caused contamination problems for K
and Ca at analysis concentrations, so the final dilution was
prepared and analyzed using Teflon1 containers. Correction
for instrument background was more significant for K than for
Ca. Background at mass 39 for the worst case was 950 cps ¡
11 cps (1 s, n ~ 5), amounting to 0.4% of the signal. However
the background was stable, changing by no more than 100
counts during a 4 hour analysis. The background at mass 41
was 492 ¡ 5 cps, indicating that there was some residual
40
Ar1H1. This too remained stable throughout an analysis. For
Ca the background at mass 40 for the worst case was 290 cps ¡
5 cps (1 s, n ~ 5) and at mass 42 it was 9 ¡ 0.5 cps.
TIMS. For the measurement of potassium a triple-filament
procedure was used. Ten mg of potassium was loaded onto each
of two Re side filaments using a micro-pipette and dried using
an infrared lamp in a laminar flow clean-air enclosure. The
assembly was then mounted into the mass spectrometer and the
472
J. Anal. At. Spectrom., 2002, 17, 469–477
center ionizing filament set at 1150 uC. The side filaments were
slowly heated at a current of approximately 0.6 A to produce a
stable ion beam of approximately 2 nA. Four sets of 41K/39K
isotope ratios were acquired for each sample and the average
ratio used for the computations. Mass fractionation of the
41
K/39K ratio was assessed by measurement of a potassium
isotopic standard.
Calcium measurements were made using a single-filament
procedure employing a Ta ionizing filament. Approximately
5 mg of calcium was loaded onto the filament using a micropipette and heated to dryness under an infrared lamp in a
laminar flow clean-air enclosure. Five mL of phosphoric acid
was then added to the filament and heated to near dryness. The
filament was next covered with a glass trough and the filament
heated electrically at a current of 1 A for a few seconds. The
filament was mounted into the mass spectrometer and initially
heated at 1150 uC for 20 min to burn off any residual
potassium, which would otherwise interfere with 40Ca measurements. Finally, the filament was ramped to a temperature
in the range 1350 to 1400 uC, at which point an extremely stable
ion beam of approximately 0.1 nA was produced. Multiple
blocks of 40Ca/42Ca isotope ratios were then acquired. All of
the ratios were internally normalized by concurrent measurement of the 44Ca/40Ca isotope ratio. This normalization
procedure was very effective in correcting for fractionation
effects, which tended to be more severe than the more traditionally used method employing a Re triple-filament assembly.
Results and discussion
Concentration and between sample variability
The cold plasma ID ICP-MS determination of K and Ca in six
bottles of SRM 1566b, Oyster Tissue, is presented in Table 2.
Our results for the determination of Ca in CCQM-P14 serum
samples are presented in Table 3 and are normalized to the
average ICP-MS result determined in the first run. The results
are presented in this manner in order not to reveal the absolute
concentration because the same material is being used in an
ongoing interlaboratory study. The concentration is within the
range of calcium in the serum of healthy adults, 2.25 to
2.65 mmol L21. Concentrations were calculated using the ID
equation described by Fassett and Paulsen.41 Results for K in
oyster tissue and Ca in serum highlight the repeatability
obtainable with cold plasma ICP-MS. The relative standard
deviation of six determinations of K in SRM 1566b was 0.078%
(1 s). For five determinations of Ca in CCQM-P14, the relative
standard deviation was 0.16% (1 s). ICP-MS ratio measurement repeatability (% relative standard deviation of five, 1-min
measurements) for both the 39K/41K and 40Ca/42Ca ratios was
typically between 0.1 and 0.2%, similar to the between sample
repeatability obtained for K in oyster tissue and Ca in serum.
This indicates that the variability in sample preparation did not
exceed variability due to ICP-MS ratio measurement. Completeness of dissolution is also indicated by agreement in
concentration results between samples of different mass. For
the oyster tissue samples the mass of S# 1327 is 0.06 g greater
than the mass for S #5, an 18% difference, yet the measured
concentrations differ by less than 0.1%. Complete dissolution
ensures that equilibration between sample and spike isotopes
occurred because potassium and calcium each have only one
oxidation state in solution.
Results for Ca in oyster tissue, however, do not show the
good between-sample agreement obtained for K. The percent
relative standard deviation (1 s, n ~ 6) for Ca in oyster tissue is
1.1%, primarily due to one high result for sample # 196. Since
Ca was determined from the same sample aliquot as the K, it is
unlikely that an error in sample preparation is the cause of the
high result. Ca is an abundant element and contamination
during preparation may have occurred. However, no blank of
Table 2 Determination of potassium and calcium in SRM 1566b, Oyster Tissue, by cold plasma-ID-ICP-MS
K concentration/mg kg21
Ca concentration/mg kg21
Bottle#
Dry sample mass/g
Day 1
Day 2a
Day 1
Day 2a
5
196
1327
2759
2926
3605
0.24808
0.25232
0.30308
0.25410
0.24830
0.25251
Avg
s
n
RSD (%)
6528.9
6528.5
6533.0
6523.7
6538.3
6534.1
6531.1
5.1
6
0.078
6534.2
6520.2
6527.9
6518.1
6523.6
6527.7
6525.3
5.9
6
0.09
829.6
853.4
832.7
837.2
829.3
833.7
836.0
9.0
6
1.1
831.9
850.8
833.2
835.0
829.8
835.9
836.1
7.5
6
0.90
a
Day 2 results are provided to indicate ICP-MS measurement repeatability only. Day 2 data were not used to calculate the overall average concentration.
the magnitude required to explain the result was measured, and
results for validation samples show no indication of contamination (see below). It was concluded, therefore, that the result
is indicative of heterogeneity for Ca in this material. It is
entirely possible that shell fragments may have become
accidentally homogenized with the oyster tissue, creating the
observed heterogeneity for Ca.
Also listed in Tables 2 and 3 are results for replicate mass
spectrometric measurements. These data were collected on
different days and in a different run order. Instrument
conditions were optimized each day. Average between day
results differ by 0.09% for K in oyster tissue, and 0.01% and
0.1% for Ca in oyster tissue and serum, respectively. The
magnitude of mass bias and the pattern of drift were different
between days. The good between day agreement verifies that
corrections were accurately applied.
Procedure blank
Blank resulting from the analytical procedure was assessed by
adding small amounts of spike to clean Teflon beakers and
manipulating and measuring them in a manner exactly the
same as the samples. Sample preparation was performed in a
class 10 clean environment. The average procedure blank
determined for K was 29 ng ¡ 13 ng (n ~ 3). The average
procedure blank determined for the Ca in oyster tissue analysis
was 101 ng ¡ 25 ng (n ~ 2), and for the serum analysis it was
280 ng ¡ 160 ng (n ~ 4). For oyster tissue samples the
correction for blank was less than 0.02% for K and 0.05% for
Ca. For serum samples, the correction for Ca blank was less
than 0.15%. A determination of the amount of Ca contributed
from the cation resin yielded a result of less than 12 ng. It is
Table 3 Determination of calcium in international intercomparision
CCQM-P14 Serum Samples by cold plasma-ID-ICP-MS
Normalized Ca concentrationa
Vial#
Sample mass/g
Day 1
Day 2b
P
P
P
P
P
2.02345
2.01469
2.01908
2.02490
2.02297
Avg
s
n
RSD (%)
1.0015
0.9990
0.9979
1.0000
1.0017
1.0000
0.0016
5
0.16
1.0009
1.0010
1.0011
1.0011
1.0010
1.0010
0.0001
5
0.01
a
14-1
14-2
14-3
14-4
14-5
Results are normalized to the average day 1 value. This is done so
as not to reveal the absolute value determined. The Ca in the serum
is at a normal physiological level. bDay 2 results are provided to
indicate ICP-MS measurement repeatability only. Day 2 data were
not used to calculate the overall average concentration.
believed that the beakers, which were cleaned, but had been
used for previous analyses, were the primary source of Ca
blank.
Spike calibrations
The concentration of the enriched isotopes in each spike
solution was calibrated by reverse isotope dilution for each
analysis, regardless of whether a calibration had been previously performed. The reason for this is two-fold. First, the
concentration of the enriched isotope solution could change
between analyses due to evaporation or adsorption. An undetected change in the concentration of the calibrating solution
would result in a bias. Second, mass bias of the instrument can
change dramatically from day to day due to changes in optimum operating conditions. By analyzing the spike calibration
samples and analytical samples under the same operating
conditions, uncertainty in the correction for mass bias is
minimized and only instrument drift between samples needs to
be monitored. Additionally, when preparing spike calibration
samples, they are spiked so the resulting ratio is similar to the
spiked ratio of the analytical samples, and they are diluted to
the same count rate. This reduces uncertainty in the correction
for detector dead time.
Calibrations were performed using two separately prepared
solutions of primary materials. Two aliquots of two gravimetrically prepared solutions were added to aliquots of the
enriched isotope solution, forming four calibration mixes.
Variation between the results for the four mixes is a measure of
the uncertainty associated with gravimetric sample preparation
and ICP-MS (or TIMS) ratio measurement without complications caused by sample matrix. Relative standard deviations of
0.086% for K and 0.062% for Ca were obtained for the
calibration of the enriched isotope solutions used for the oyster
tissue analysis. For the calibration of the 42Ca spike solution
used in the serum analysis, an average Ca concentration of
1.0319 mmol g21 ¡ 0.0006 mmol g21 (0.057% relative; 1 s, n ~
4) was obtained by cold plasma ICP-MS, and an average Ca
concentration of 1.0317 mmol g21 ¡ 0.0012 mmol g21 (0.11%
relative; 1 s, n ~ 4) was obtained by TIMS. Results are 0.02%
different and the ICP-MS repeatability compares favorably
with the TIMS repeatability. There is no difference, within
measurement repeatability, between average results for the two
primary standard solutions used in each calibration, verifying
the completeness of their preparation, and the accuracy of the
spike solution calibrations.
Spectral interference
As already discussed, operation in the cold plasma mode
greatly reduces spectral interference from the plasma: however,
matrix-induced interference must still be evaluated. One way to
assess the presence of interference at the isotopes of interest is
J. Anal. At. Spectrom., 2002, 17, 469–477
473
to compare isotope ratios measured in a matrix-free standard
with those measured in an unspiked sample of the oyster tissue
and serum. The results of this comparison for an unseparated
oyster tissue sample showed the measured 39K/41K and
40
Ca/42Ca ratios to be 1.3% and 30% lower than the same
ratios measured for the pure standards. These results indicate a
larger interference at masses 41 and 42, though in both cases
these masses are more sensitive to interference because of their
lower relative abundance. It is possible that interference occurs
at all the isotopes of interest. The same comparison performed
on unspiked oyster tissue samples that were separated using
cation exchange chromatography showed agreement with the
pure standards to within 0.3%, well within the repeatability of
the measurement, thereby verifying the successful removal of
interferences with the applied separation scheme. Measurement
repeatability (1 s, n ~ 5) for natural composition ratios
averaged 0.3% for the 39K/41K ratio and 0.6% for the 40Ca/42Ca
ratio. Unseparated serum samples showed a negative bias of
7% relative to pure standards for the 40Ca/42Ca ratio. After
cation exchange separation, measured ratios agreed within
0.3%.
Identification of the exact source of matrix-induced polyatomic interferences is difficult. The most probable sources are
the oxides and hydroxides of Na and Mg. Analysis of the
separated fractions for both matrices showed that separation of
Na, Mg, K and Ca was complete. Strontium is difficult to
separate from Ca on cation exchange resin and some Sr was
observed in the Ca fraction. However, doubly charged Sr ions
were not observed under cool plasma conditions in agreement
with the finding of Patterson et al.29 Based on signal intensity of
the separated fractions, recovery of K and Ca was estimated at
over 95%.
Fig. 2 Split sample comparison of cold plasma-ID-ICP-MS and IDTIMS for the determination of Ca in six bottles of SRM 1566b, Oyster
Tissue. Concentration is shown on the left axis and percent difference
from the mean is shown on the right axis. Error bars are 1 s of the ratio
measurement, multiplied by the error multiplication factor (1.04). The
pattern of variance between bottles is replicated by the two techniques,
indicating that material heterogeneity is the source of variance in the
results. Average ICP-MS and TIMS results are less than 0.06%
different.
Split sample comparison with thermal ionization mass
spectrometry
The same sample preparations were split and analyzed by both
ICP-MS and TIMS. The TIMS method for both K and Ca has
been accepted as definitive for clinical analyses.12–14 These mass
spectrometric methods are indeed distinct—in ionization, mass
filtering and detection. A sample-to-sample comparison of
ICP-MS and TIMS results for K and Ca in oyster tissue and Ca
in serum is presented in Figs. 1–3, respectively. Concentration
is shown on the right axis and percent difference from the
TIMS average is shown on the left axis. Error bars are 1 s of the
ratio measurement, multiplied by the error multiplication
factor, 1.05 for K and 1.04 for Ca.36 In general the TIMS
measurement repeatability is a factor of two better than ICPMS measurement repeatability: excellent agreement between
Fig. 3 Split sample comparison of cold plasma-ID-ICP-MS and IDTIMS for the determination of Ca in five vials of international
intercomparision CCQM-P14 serum samples. Percent difference from
the mean is shown on the left axis. Error bars are 1 s of the ratio
measurement, multiplied by the error multiplication factor (1.04).
Average ICP-MS and TIMS results are less than 0.02% different.
the methods was obtained. The ICP-MS/TIMS ratio was
1.0005 ¡ 0.0017 (1 s, n ~ 6) for K in oyster tissue, 1.0005 ¡
0.0010 (1 s, n ~ 6) for Ca in oyster tissue, and 0.9999 ¡ 0.0016
(1 s, n ~ 5) for Ca in serum. The pattern of variability for the
two methods is exactly replicated for the determination of Ca
in SRM 1566b, further indicating that material heterogeneity,
and not measurement uncertainty, is the source of variance
in the results. In all cases agreement better than 0.1% was
obtained, validating not only the absence of interference and
the correction for mass bias, since the ionization processes were
completely different, but also the ICP-MS correction for dead
time since Faraday cup detection was used in TIMS.
Total analytical uncertainty of the cold plasma-ID-ICP-MS
procedure
Fig. 1 Split sample comparison of cold plasma-ID-ICP-MS and IDTIMS for the determination of K in six bottles of SRM 1566b, Oyster
Tissue. Concentration is shown on the left axis and percent difference
from the mean is shown on the right axis. Error bars are 1 s of the ratio
measurement, multiplied by the error multiplication factor (1.05).
Average ICP-MS and TIMS results are less than 0.06% different.
474
J. Anal. At. Spectrom., 2002, 17, 469–477
The expanded analytical uncertainty for the ID-ICP-MS
determination of K and Ca in SRM 1566b and Ca in
CCQM-P14 is composed of individual Type A and Type B
uncertainty components combined according to ISO guidelines.42 Individual uncertainty components, expressed as
relative uncertainties, and the expanded uncertainty, expressed
as a 95% confidence interval, are given in Table 4. Coverage
factors were determined based on the effective degrees of
freedom calculated using the Welch–Satterthwaite formula.42
Minimization of potential systematic uncertainty in sample
preparation, isotope ratio measurement, calibration of the
spike and correction for analytical blank was discussed above.
Uncertainties in these components were treated as Type A, and
based on the standard uncertainties of the sample, spike calibration and blank results. Type B components were assumed to
have a rectangular distribution of stated magnitude. Standard
uncertainties for Type B components were derived by dividing
by the square root of 3. The Type B component estimating
uncertainty in instrument factors affecting ratio measurement
includes uncertainty due to drift, dead time, and mass bias.
Though experimental design to minimize uncertainties in these
parameters was discussed, they are difficult to completely
eliminate. Based upon consideration of count rate and isotope
ratio differences, range of drift, and the difference in calculated
concentration results for the range of experimentally determined dead times, this uncertainty component was estimated to
be 0.3% relative. Uncertainty in the calibrant assay was based
on the largest uncertainty reported on the certificates for SRMs
918a, 915a and 3109a. Weighing uncertainty was estimated to
be of magnitude 0.06% relative, based on the smallest sample
mass, the balance readability, and its linearity. The remaining
Type B uncertainty component, an estimate for the uncertainty
in the correction to dry mass, pertains only to the oyster tissue
samples and would be common to any analytical procedure.
The dry mass correction averaged 4% for the oyster tissue.
The magnitude of the correction depended strongly on the
drying procedure used and an uncertainty estimate of relative
magnitude 1.5% was derived based on the differences measured
for samples dried using the vacuum oven procedure and the
desiccator procedure.43 The uncertainty for the correction to
dry mass is the largest uncertainty component for the oyster
tissue determinations, but its magnitude based on experimental
data is in keeping with what is observed for biological reference
materials.44 Without the uncertainty component for the dry
mass correction, the expanded uncertainties for the cold plasma
ID ICP-MS determinations of K and Ca in oyster tissue
are 0.38% and 1.1% relative, respectively, the latter being
dominated by sample heterogeneity. When the uncertainty
component for correction to dry mass is included, the expanded
uncertainty becomes 1.7% and 1.9% relative for K and Ca,
respectively. The expanded uncertainty for the determination
of Ca in CCQM-P14 by cold plasma ID ICP-MS is 0.41%
relative.
Table 5 Analysis of gravimetric standards by cold plasma-ID-ICP-MS
Element
K
Ca
Ca
SRM 915aa
Standard
SRM 918aa
Gravimetric
998.050
390.953
value/mg kg21
997.50 ¡ 0.71 390.912 ¡ 0.042
Measured
value/mg kg21
(1 s, n ~ 2)
Difference (%) 20.055
20.010
a
Prepared and measured during the certification
Oyster Tissue. bPrepared and measured during
CCQM P-14, serum.
SRM 915ab
79.858
79.783 ¡ 0.089
20.093
of SRM 1566b
the analysis of
Results for control samples
Two types of accuracy validation samples were processed,
gravimetrically prepared primary standards and previously
certified SRMs with matrices similar to the analytical samples.
The gravimetric samples do not provide a measure of matrixrelated uncertainties, but method accuracy can be calculated by
comparing the measured values to the gravimetric concentration. Uncertainty on the gravimetric concentrations is small
(less than 0.05%) based on the linearity and readability of the
six-place microbalance and the certified assay. Gravimetric
controls for K were prepared from SRM 918a and those for Ca
were prepared from SRM 915a. Gravimetric control samples
were treated exactly as the analytical samples and similar
amounts were processed. Results are given in Table 5. Samples
were measured on two different days to assess reproducibility.
The average result and 1 s for the two determinations is
reported. In all cases the measured concentrations agree with
the gravimetric concentration to better than 0.1%.
ID-ICP-MS determinations in matrix matched control
materials, SRM 1566a, Oyster Tissue, and SRM 956a, Level
II Serum, are presented in Fig. 4. Results are plotted as percent
difference from the certified value. The certified interval is
depicted as the line on the left of each plot. The average
measured result is depicted by the symbol. Error bars on the
measured values for the oyster tissue samples are the 95%
confidence interval for 3 separate sample determinations (2
degrees of freedom). The coverage factor for the 95% confidence interval was calculated based on the effective degrees of
freedom for the combined standard uncertainties outlined in
the Uncertainty section. The K and Ca results for SRM 1566a
are both within the certified intervals. One determination was
made for the serum sample. The Ca content of SRM 956a was
certified in 1996 using ID-TIMS. The ID-ICP-MS determination reported here is 0.25% different from the certified value,
Table 4 Uncertainty components for the determination of K and Ca in Oyster Tissue and Ca in Serum by cold plasma ID-ICP-MS
Source
Type A uncertainties—
Sample/measurement
Spike calibration
Blank correction
Combined type A
Type B uncertainties—
Moisture correction
ICP-MS factors (drift, dead time, mass bias)
Calibrant purity
Weighing uncertainty
Combined type B
Combined type A and B
Coverage factor
Expanded uncertainty
Potassium, SRM 1566b
Calcium, SRM 1566b
Calcium, CCQM P-14
ui (% rel.)
d.f.
ui (% rel.)
d.f.
ui (% rel.)
d.f.
0.035
0.043
0.0011
0.055
5
3
2
0.44
0.031
0.036
0.44
5
3
1
0.072
0.028
0.050
0.092
4
3
2
0.87
0.17
0.058
0.035
0.89
0.89
1.96
1.7a
‘
‘
‘
‘
0.87
0.17
0.058
0.035
0.89
0.98
1.98
1.9a
‘
‘
‘
‘
—
0.17
0.049
0.060
0.19
0.21
1.97
0.42a
—
‘
‘
‘
a
95% Confidence Interval.
J. Anal. At. Spectrom., 2002, 17, 469–477
475
reference material SRM 1566b was also demonstrated. Use of
the ICP-MS is advantageous because of its wide availability,
element coverage, sensitivity and measurement repeatability.
Accurate application of the ‘‘cold’’ plasma conditions did
require matrix separation, but separations do not need to be
quantitative with isotope dilution. Removal of the sample
matrix makes the technique generally applicable to any sample
type. Cation exchange separation is efficient and allows for the
determination of K and Ca in the same sample; in fact, Li and
Mg could be determined as well. Cold plasma ID-ICP-MS
results were comparable with ID-TIMS results both within and
outside our laboratory, demonstrating the critical role ID-ICPMS can play in establishing traceability.
Fig. 4 Cold plasma ID-ICP-MS determinations in control materials,
SRM 1566a, Oyster Tissue, and SRM 956a, Level II Serum. The error
bar for the oyster tissue samples is the 95% confidence interval for 3
separate determinations (see text). One determination was made for the
serum sample. All results are within the certified interval.
well within the ¡0.6% relative uncertainty on the certified
value. Our ability to achieve comparability with a well-defined
national standard validates the accuracy of our protocol and
demonstrates that ID TIMS and cold plasma ID-ICP-MS
results for the determination of Ca in serum are commutable.
Interlaboratory comparison CCQM-P14
The NIST result in the CCQM-P14 inter-laboratory comparison for Ca in serum was in excellent agreement with those of
four other institutes that also applied the isotope dilution
technique, including three that used TIMS. Our laboratory was
the only one to use cold plasma, quadrupole ICP-MS.
Certified concentrations for SRM 1566b
The certified values for SRM 1566b are the weighted means of
two or more analytical methods. Results of the different
analytical methods and the final certified values for K and Ca in
SRM 1566b are given in Table 6. For Ca the certified result is
the weighted mean of the NIST cold plasma-ID-ICP-MS result
and the NIST instrumental neutron activation result. For K the
certified result is the weighted mean of the NIST cold plasma
ID ICP-MS result and NIST instrumental neutron activation
and prompt gamma activation results. For both elements,
excellent agreement was obtained between the mass spectrometric and nuclear methods. The stated uncertainty on the
certificate for K is a 95% confidence interval; for Ca, due to
heterogeneity, the stated uncertainty is a 95% prediction
interval.43
Conclusion
Isotope dilution inductively coupled ‘‘cold’’ plasma mass
spectrometry is an accurate method for the quantitation of
K and Ca. An expanded ID-ICP-MS uncertainty of 0.4%
relative was achieved for the determination of K in oyster tissue
(excluding the uncertainty in the dry mass correction) and Ca in
serum. The level of heterogeneity of Ca in the oyster tissue
Table 6 Results of independent analytical methods and certified values
for K and Ca in SRM 1566b
Analytical method
K concentration ¡
expanded u/mg kg21
Ca concentration ¡
expanded u/mg kg21
Cold plasma ICP-MS
INAA
PGAA
Certified value
6530
6490
6540
6520
836 ¡ 16
840 ¡ 28
a
¡
¡
¡
¡
114
140
200
90a
838 ¡ 20b
b
95% confidence interval. 95% prediction interval.
476
J. Anal. At. Spectrom., 2002, 17, 469–477
Acknowledgement
The authors gratefully acknowledge Robert Greenberg of the
NIST Nuclear Methods group for providing data used in this
manuscript. The authors also acknowledge J. D. Fassett and
U. Örnemark for their comments.
References
1 German Health Report 1998, www.gbe-bund.de, accessed July
2001.
2 W. E. May, Before the Senate Committee on Commerce, Science,
and Transportation Subcommittee on Science, Technology and
Space in a hearing on ‘‘E-health and Technology: Empowering
Consumers in a Simpler, More Cost Effective Health Care
System’’ [Testimony]. Washington, DC, Chemical Science and
Technology Laboratory, NIST, July 23 2001.
3 J. Büttner, Eur. J. Chem. Clin. Biochem., 1991, 29, 223.
4 F. D. Lasky, Clin. Chem., 1992, 38, 1260.
5 R. Schaffer, G. N. Bowers Jr. and R. S. Melville, Clin. Chem.,
1995, 41, 1306.
6 L. M. Thienpont, D. Stöckl and A. P. De Leenheer, J. Mass
Spectrom., 1995, 30, 772.
7 M. M. Müller, Clin. Chem., 2000, 46, 1907.
8 Directive 98/79EC of the European Parliament and of the Council
of 27 October 1998 on In vitro Diagnostic Medical Devices.
9 Nutrition Labeling and Education Act of 1990, 21 CFR 101.9.
10 The Oregon State University Linus Pauling Institute. http://
osu.orst.edu/dept/lpi/infocenter/minerals/calcium/calcium.html,
accessed July 2001.
11 The Oregon State University Linus Pauling Institute. http://
osu.orst.edu/dept/lpi/infocenter/minerals/potassium/potassium.html,
accessed July 2001..
12 The National Reference System For The Clinical Laboratory
NRSCL. http://www.nccl.org/nrscl.htm, accessed July 2001.
13 L. J. Moore and L. A. Machlan, Anal. Chem., 1972, 44, 2291.
14 J. W. Gramlich, L. A. Machlan, K. A. Brletic and W. R. Kelly,
Clin. Chem., 1982, 28, 1309.
15 K. E. Murphy, E. S. Beary, M. S. Rearick and R. D. Vocke,
Fresenius’ J. Anal. Chem., 2000, 368, 362.
16 E. S. Beary, P. J. Paulsen, L. B. Jassie and J. D. Fassett, Anal.
Chem., 1997, 69, 758.
17 I. S. Begley and B. L. Sharp, J. Anal. At. Spectrom., 1997, 12, 395.
18 L. Halicz, A. Galy, N. S. Belshaw and R. K. O’Nions, J. Anal. At.
Spectrom., 1999, 14, 1835.
19 S. Stürup, M. Hansen and C. Mølgaard, J. Anal. At. Spectrom.,
1997, 12, 919.
20 J. S. Becker and H. J. Dietze, J. Anal. At. Spectrom., 1998, 13,
1057.
21 I. Feldmann, N. Jakubowski and D. Stuewer, Fresenius’ J. Anal.
Chem., 1999, 365, 415.
22 S. D. Tanner, V. I. Baranov and U. Vollkopf, J. Anal. At.
Spectrom., 2000, 15, 1261.
23 D. R. Bandura, V. I. Baranov and S. D. Tanner, J. Anal. At.
Spectrom., 2000, 15, 921.
24 S. F. Boulyga and J. S. Becker, Fresenius’ J. Anal. Chem., 2001,
370, 618.
25 J. S. Becker and H. J. Dietze, Fresenius’ J. Anal. Chem., 2000, 368,
23.
26 I. Feldmann, N. Jakubowski, C. Thomas and D. Stuewer,
Fresenius’ J. Anal. Chem., 1999, 365, 422.
27 S. D. Tanner, J. Anal. At. Spectrom., 1995, 10, 905.
28 K. Sakata and K. Kawabata, Spectrochim. Acta, 1994, 49B, 1027.
29 K. Y. Patterson, C. Veillon, A. D. Hill, P. B. Moser-Veillon and
T. C. O’Haver, J. Anal. At. Spectrom., 1999, 14, 1673.
30 D. Wollenweber, S. Strassburg and G. Wünsch, Fresenius’ J. Anal.
Chem., 1999, 364, 433.
31 A. A. Pupyshev, N. L. Vasil’eva and S. V. Golik, Ind. Lab., 1999,
65, 489.
32 M. G. Minnich and A. Montaser, Appl. Spectrosc., 2000, 54,
1261.
33 G. H. Tao, R. Yamada, Y. Fujikawa, R. Kotima, J. Zheng,
D. A. Fisher, R. M. Koerner and A. Kudo, Int. J. Environ. Anal.
Chem., 2000, 76, 135.
34 C. J. Park and J. K. Suh, J. Anal. At. Spectrom., 1997, 12, 573.
35 W. R. Shields, in Analytical Mass Spectrometry Section: Summary
of Activities, Technical Note 426, National Bureau of Standards,
Washington, DC, 1967, p. 53.
36 K. G. Heumann, in Inorganic Mass Spectrometry, eds. F. Adams,
R. Gijbels, R. Van Grieken, Wiley, New York, 1988, p. 309.
37 The Bureau International des Poids et Mesures. http://
www.bipm.fr, accessed July 2001.
38 U. Örnemark, Y. Aregbe, L. Van Nevel, I. Papadakis, P. D. P.
Taylor and A. Uldall, Internal Report GE/R/IM/27/01, IRMM,
Geel, Belgium, December 2001.
39 C. S. Phinney, K. E. Murphy, M. J. Welch, P. M. Ellerbe,
S. E. Long, K. W. Pratt, S. B. Schiller, L. T. Sniegoski,
M. S. Rearick, T. W. Vetter and R. D. Vocke, Fresenius’
J. Anal. Chem., 1999, 64, 433.
40 E. S. Beary and P. J. Paulsen, Anal. Chem., 1995, 67, 3193.
41 J. D. Fassett and P. J. Paulsen, Anal. Chem., 1989, 61, 643A.
42 B. N. Taylor and C. E. Kuyatt, Technical Note 1297, NIST,
Gaithersburg, MD, 1994.
43 NIST Certificate of Analysis, SRM 1566b, Oyster Tissue, January
2001.
44 S. Ruckold, K. H. Grobecker and H.-D. Isengard, Fresenius’
J. Anal. Chem., 2001, 370, 189.
J. Anal. At. Spectrom., 2002, 17, 469–477
477