Rapid multi-element analysis of groundwater by high-resolution inductively coupled plasma mass spectrometry

Abstract

A rapid and sensitive method was developed to determine, with a single dilution, the concentration of 33 major and trace elements (Na, Mg, Si, K, Ca, Li, Al, P, S, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Sr, Mo, Cd, In, Sn, Sb, Cs, Ba, Re, Hg, Pb, Bi, U) in groundwater. The method relies on high-resolution inductively coupled plasma mass spectrometry (HR ICP-MS) and works across nine orders of magnitude of concentrations. For most elements, detection limits for this method are considerably lower than methods based on quadrupole ICP-MS. Precision was within or close to +/-3% (1 sigma) for all elements analyzed, with the exception of Se (+/-10%) and Al (+/-6%). The usefulness of the method is demonstrated with a set of 629 groundwater samples collected from tube wells in Bangladesh (Northeast Araiharzar). The results show that a majority of tube well samples in this area exceed the WHO guideline for As of 10 microg L(-1), and that those As-safe wells frequently do not meet the guideline for Mn of 500 microg L(-1) and U of 2 microg L(-1).

Full text

Abstract A rapid and sensitive method was developed to
determine, with a single dilution, the concentration of 33 ma-
jor
and trace elements (Na, Mg, Si, K, Ca, Li, Al, P, S, Ti,
V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Sr, Mo, Cd, In, Sn,
Sb, Cs, Ba, Re, Hg, Pb, Bi, U) in groundwater. The
method relies on high-resolution inductively coupled plasma
mass spectrometry (HR ICP-MS) and works across nine
orders of magnitude of concentrations. For most elements,
detection limits for this method are considerably lower
than methods based on quadrupole ICP-MS. Precision
was within or close to ±3% (1
σ
) for all elements ana-
lyzed, with the exception of Se (±10%) and Al (±6%).
The usefulness of the method is demonstrated with a set
of 629 groundwater samples collected from tube wells in
Bangladesh (Northeast Araiharzar). The results show that
a majority of tube well samples in this area exceed the
WHO guideline for As of 10 µgL
–1
, and that those As-
safe wells frequently do not meet the guideline for Mn of
500 µgL
–1
and U of 2 µgL
–1
.
Keywords HR ICP-MS · Groundwater · Multi-element
analyses · Bangladesh
Introduction
Rarely is the measurement of a single parameter sufficient
to properly address a question in environmental research
or monitoring. Consequently, there is a considerable inter-
est in developing rapid, yet accurate analytical methods
that measure a wide suite of parameters. Analysis of mul-
tiple constituents of aqueous samples that range over sev-
eral orders of magnitude of concentrations often have to
rely on several different instruments and sample prepara-
tions (e.g., colorimetry, titration, and atomic absorption/
emission spectrometry). There are some widely used
methods for multi-element quantification that are based
on inductively coupled plasma optical emission spectrom-
etry (ICP-OES), or quadrupole ICP mass spectrometry
(ICP-MS), such as US EPA method 200.8 [1], but these
often require considerable correction for spectral interfer-
ences in emission or mass from other elements and poly-
atomic molecules. With HR ICP-MS, which relies on a
magnetic sector to separate ions with greater discrimination
according to their mass/charge ratio, many of these inter-
ferences can be eliminated. A growing number of multi-
element methods that rely on HR ICP-MS have therefore
been developed in recent years, but they typically involve
a limited array of elements and a well-defined sample ma-
trix (e.g., high-purity metals [2], sediments [3], marine
particulate material [4], coastal seawater [5], Mn–Fe nod-
ules [6], and fresh surface water [7]).
In this paper, we describe a rapid and sensitive analyt-
ical method based on HR ICP-MS for the precise quan-
tification of 33 elements in groundwater that requires only
a single dilution step. The analytes include all major
cations present in groundwater (Na, K, Ca, Si, Mg) and
most minor, trace, and ultra-trace elements that are of
common environmental interest (Li, Al, P, S, Ti, V, Cr,
Mn, Fe, Co, Ni, Cu, Zn, As, Se, Sr, Mo, Cd, In, Sn, Sb,
Cs, Ba, Re, Hg, Pb, Bi, U). The list includes 13 of the 19
inorganic constituents for which health concerns have led
the World Health Organization (WHO) to set guideline
values [8, 9] We also show that the developed method can
be applied to the analysis of wastewaters. We conclude with
a demonstration of the method with a set of 629 ground-
water samples collected from tube wells in Bangladesh.
Z. Cheng · Y. Zheng · R. Mortlock · A. van Geen
Rapid multi-element analysis of groundwater
by high-resolution inductively coupled plasma mass spectrometry
Anal Bioanal Chem (2004) 379 :512–518
DOI 10.1007/s00216-004-2618-x
Received: 27 January 2004 / Revised: 17 March 2004 / Accepted: 21 March 2004 / Published online: 20 April 2004
ORIGINAL PAPER
Z. Cheng (✉) · Y. Zheng · R. Mortlock · A. van Geen
Lamont Doherty Earth Observatory of Columbia University,
Palisades, NY 10964, USA
Tel.: +1-845 365 8649, Fax: +1-845 365 8154,
e-mail: czhongqi@ldeo.columbia.edu
Y. Zheng
School of Earth and Environmental Sciences, Queens College,
CUNY, Flushing, NY, 11367, USA
©Springer-Verlag 2004

513
Experimental
Reagents
High-purity water (denoted as MQ hereafter) with a resistivity
>18
MΩwas obtained from a Millipore MQ ion-exchange system
fed with deionized water (US filter). Nitric acid (1%) was prepared
from high-purity concentrated (16 M) HNO
3
(Optima, Seastar
Chemical). A diluent containing 200 µgL
–1
Sc, 50 µgL
–1
of Ge,
and 1 µgL
–1
Tl internal standard spikes in 1% HNO
3
was prepared
and used throughout for sample and standard dilutions for drift
correction. Hydrochloric acid (Optima, SeaStar Chemical) was di-
luted to 6 M and then used to acidify all groundwater samples to
1% HCl in the field.
Major and trace element standards were prepared as two sepa-
rate primary stock solutions each in 5% HNO
3
. This is required to
prevent the precipitation of some insoluble salts (e.g., Ca
3
(PO
4
)
2
)
and because the solids used to make major ion standards could
contain appreciable amounts of trace elements. The major element
stock solution was prepared by dissolving CaCO
3
(Puratronic, Alfa
Aesar), KNO
3
, and NaNO
3
solids (Fisher brand), and mixing with
Mg and Si single standards (SpecPure, Alfa Aesar). A trace ele-
ment stock solution was prepared first by mixing single element
standards (for Mn, Fe, Al, Ti, Zn, As, Ba, Sr, Cs, and Pb,
SpecPure, Alfa Aesar) and an existing mixed metal solution in the
lab containing 1 mgL
–1
of most metals (SpecPure, Alfa Aesar);
small amounts of KH
2
PO
4
and K
2
S
2
O
8
solids (Fisher brand) were
added later to obtain the desired concentrations of P and S. The
mixed multi-element standards were then calibrated against single
element standards to determine the exact concentrations. The ma-
jor element stock contained 100 mgL
–1
Na, Mg, Si, 200 mgL
–1
K,
and 400 mgL
–1
Ca. The trace element stock had 50 mgL
–1
Fe,
25 mg L
–1
Mn, 20 mgL
–1
S, 10 mgL
–1
P and Ba, 5 mgL
–1
As, Sn,
and Sr, 1 mgL
–1
Al, Zn, Se, Pb, and 0.5 mgL
–1
Ti, Cr, Co, Ni, Cu,
Cd, In, Hg, Li, V, Mo, Sb, Cs, Re, Bi, and U.
Samples and stock standards were diluted with the diluent in
6-mL Omni* polyethylene vials (Wheaton) that fit a Gilson Model
221–222 autosampler. Early tests found that unwashed vials gave
low blanks and they were therefore used as purchased. A leaching
with acid would be preferable, especially for some ultra-trace metals.
Analysis takes about 10 min per sample and requires a total vol-
ume of 1–1.5 mL of diluted sample. For both major and trace ele-
ments, standard series were prepared by diluting 0.5, 1.0, and 2.0 mL
of primary stock solution with 20 mL of diluent. For trace elements,
standard series were prepared by diluting 100, 200, and 400 µL of
stock solution with 20 mL diluent. For unknowns, 0.5mL of sam-
ple was diluted 10-fold with diluent directly in the vials.
Instrumental settings
We used an Axiom Single Collector HR ICP-MS (Thermo Ele-
mental, Germany) which combines a double-focusing, magnetic
sector mass spectrometer with an optimized ICP ion source. This
specific model is no longer produced, but it is not fundamentally
different from other HR ICP-MS instruments on the market. The
VG Axiom features a range of resolution settings, with a maxi-
mum resolving power (RP) of 16,000 that is sufficient to eliminate
virtually all potential spectral interferences for groundwater analy-
sis. The detector system combines an on-axis Faraday cup and a
fast switching electron multiplier, and provides nine orders of
magnitude of working dynamic range. For sample introduction, we
used a PFA-100 micro-nebulizer (Elemental Scientific, Inc.) at an
actual flow rate of about 150 µLmin
–1
.
Instrument settings and operating conditions that were rou-
tinely employed for groundwater analysis are listed in Table 1. Iso-
topes
23
Na,
39
K, and
45
Sc were acquired in Faraday cups, while all
other isotopes (
7
Li,
26
Mg,
27
Al,
30
Si,
31
P,
32
S,
43
Ca,
47
Ti,
51
V,
52
Cr,
55
Mn,
57
Fe,
59
Co,
60
Ni,
63
Cu,
66
Zn,
74
Ge,
75
As,
78
Se,
88
Sr,
98
Mo,
111
Cd,
115
In,
120
Sn,
121
Sb,
133
Cs,
137
Ba,
187
Re,
202
Hg,
203
Tl,
207
Pb,
209
Bi, and
238
U) were acquired using the multiplier. To limit the time required
to reset the slits, only three resolution settings with RPs of about
400, 4,000, and 12,000 were used. These are referred to hereafter
as low-, medium-, and high-resolution settings, respectively. Al-
though high resolution was mainly employed to resolve interfer-
ences (
32
S,
55
Mn,
57
Fe,
75
As,
78
Se, and
88
Sr), we also use high re-
solving power to reduce sensitivity and avoid saturation of the
multiplier for high concentration elements S, Mn, Fe, and Sr. Iso-
tope
74
Ge was also acquired at the same setting for drift correction.
Medium resolution was used for
23
Na,
26
Mg,
27
Al,
30
Si,
31
P,
39
K,
43
Ca,
47
Ti,
51
V,
52
Cr,
59
Co,
60
Ni,
63
Cu,
66
Zn,
74
Ge,
121
Sb, and
137
Ba.
Low resolution was used for all other isotopes. The interfering
masses and remaining unresolved interfering molecules are listed
in Table 2. Typical peak width was set at 2.5, and for isotopes mea-
sured at low-resolution settings it was limited to 0.1 to expedite
data acquisition. Dwell time was set to 40 ms for most isotopes,
but for isotopes acquired at high resolution it was increased to
70 ms to improve counting statistics and compensate for the sig-
nificantly lower sensitivity (reduced to approximately 2% that of
low resolution).
Each analysis for one sample consisted of three replicate mea-
surements (i.e., three sweeps) and each measurement scans over all
isotopes. For isotopes acquired with a narrower window (0.1 peak
width), the number of scans per measurement was increased to 3 to
improve precision.
Analytical procedure
Prior to analysis of samples, the instrument settings are first opti-
mized with a 1 µgL
–1
mixed-metal (Be, Mg, Co, Y, In, Pb, Bi, and
U) solution by adjusting the torch position and nebulizer and aux-
iliary gas flow rates. A 1 µgL
–1
mixed-metal solution typically
gives 1 millioncountss
–1
for
115
In at 400 RP; obtaining a steady
and low-noise beam at this point is critical for the quality of sub-
sequent analyses. The tuning procedure is then followed by a mass
calibration performed at medium resolution.
Following acquisition, the data were transferred to a spread-
sheet for drift correction based on counts obtained for
45
Sc,
203
Tl,
and
74
Ge. Sensitivity drift for isotopes acquired with the Faraday
collector (
23
Na and
39
K) at medium resolution were corrected with
45
Sc counts acquired at the same resolution. All isotopes acquired
at high resolution were corrected with
74
Ge counts obtained at high
resolution. High mass isotopes (
187
Re,
202
Hg,
207
Pb,
209
Bi, and
238
U)
were corrected with
203
Tl counts acquired at low resolution. All
other isotopes were corrected with
74
Ge counts acquired at medium
resolution. Since the Axiom software does not permit the scanning
of an isotope at more than one resolving power,
74
Ge was ficti-
tiously identified as ArS at medium resolution.
Quality control samples
and natural groundwater samples analyzed
In addition to blanks determined for 1% HNO
3
, 1% HCl, and the
diluent used for drift correction, an artificial groundwater sample
Table 1 Instrument settings and measurement conditions
Nebulizer Microconcentric, pneumatic
Spray chamber Temperature-controlled impact
bead type, quartz
RF power (W) 1,350
Magnet settling time (s) 0.5
Gas flow rates (L min
–1
)
Nebulizer 0.6–0.8
Auxiliary 1.5–2.0
Cooling ≈14
Sample uptake rate (µLmin
–1
) 150
Sample uptake time (s) 60
Wash time (s) 60
Bench temperature (°C) 32–34

514
(LDEOGW) with all elements of interest served as a lab consis-
tency standard and was analyzed with each batch of groundwater
samples. A natural water standard (SRM 1640, Trace elements in
natural water) from the National Institute of Standards and Tech-
nology (NIST, Gaithersburg, MD) was used for external quality
control. An array of standard addition experiments were also con-
ducted for additional quality control, particularly for the accuracy
of a few elements not referenced in SRM 1640, and to examine po-
tential matrix and interference effects. The samples used for stan-
dard addition experiments included two groundwater samples from
Bangladesh, one sample of tap water, a bottled commercial drink-
ing water, and two wastewater samples (sample Waste-1 is the gas
condensate from a landfill and sample Waste-2 is from a cooling
tower where pressure-treated wood was used). Finally, 629 ground
-
water samples collected from tube wells, a subset of samples col-
lected from 6,000 wells within a 25-km
2
area of Bangladesh [10],
were analyzed by HR ICP-MS.
Results and Discussion
Blanks and detection limits
Blank levels were measured for two solutions: the diluent
solution containing internal standard elements and 1% HNO
3
,
and a 0.1% HCl in diluent solution mimicking “real” di-
luted groundwater sample (Table3). Overall, the level of
blanks and their variations are insignificant compared to
mean concentrations of most elements reported for
groundwater [11, 12]. There are some small but noticeable
differences between solutions and among elements, and in
many cases the blank variations were larger than the blank
Resolved Interferences Remaining interferences
a
Equivalent inter
-
fered-element
conc. (mgL
–1
)
b
Isotopes acquired at low resolution (400)
98
Mo Ar-Fe(10,047), Ar-Ni(12,755), Br-OH,
Br-O (9,776) 0.45
111
Cd Ar-As(7,372), Br-S(8,051), MoO(32,343) 0.08
133
Cs BaH(17,879),SnO(17,567) 0.02
115
In ArAs(5,775), Br-S(6,223) 0.1
Isotopes acquired at medium resolution (≈4K)
23
Na Ti++, Li-O
26
Mg Cr++, C-N, C-C, B-O, BeO
27
Al Cr++, Fe++, C-N, B-O, C-N, Be-O
30
Si Ni++, N-O, C-O, N-N-H
31
PNi++, N-O, N-O-H
39
KNa-O ArH(5,690) 80
43
Ca Al-O, Ar-Li, Mg-O, Mg-O-H Sr++(10,391) 53
47
Ti P-O, S-N, P-O, Ar-Li, P-O, N-O-O Mo++(60,335) 0.2
51
VCl-O, Cl-N, S-O-H, Ar-B, Ar-N-H, S-O-H
52
Cr Ar-C, Ar-O, Ar-N, Cl-N, Cl-O, Cl-O-H
59
Co Sn++, Ca-O, Ar-Na, Ar-F, K-O, Ar-O-H
60
Ni Ca-O, Sn++, Ar-Mg, Na-Cl
63
Cu Ti-O,Ar-Al, Ar-Na, Sc-O, Ti-O-H, P-O-O
66
Zn S-S, Ar-Mg, Ti-O, Ti-O-H, Ba++, V-O, S-O-O Ar-Si(4,839), Ti-O(4,418), Cr-O(4,111) 0.11
121
Sb Ar-Br(4,806) 0.06
137
Ba Sb-O(19,341), Sn-O(40,915), BaH(20,855) 0.5
Isotopes acquired at high resolution (≈12 K)
74
Ge ArS, Ni-O, Ar-Ar, Cl-Cl, K-Cl, Fe-O, Ca-O-O,
Sm++, Nd++ Fe-O(12,936) 0.002
32
SZn++, S-H, O-O, O-O-H, N-O-H
55
Mn Cd++, Sn++, Ar-O, Ca-O, K-O, Ar-N-H, Cl-O-H
57
Fe Sn++, Cd++, K-O, Ar-F, Ar-O, Ca-O, K-O, Ar-O-H
75
As Co-O, Ar-Cl, Ar-K, Ni-O, Ca-O-O, Sm++, Nd++
78
Se Ar-Ar, Ar-Ca, K-Cl, Ni-O Ni-O(13,081) 0.2
88
Sr Ar-Ca, Ge-O Ar-Cr(36,175), Ar-Ti(18,684) 0.03
Table 2 Interference molecules for isotopes used for acquisition.
The equivalent interfered-element concentration was determined
with 0.1% HCl–1% HNO
3
solutions containing internal standard
elements and possible levels of remaining interfering elements in
groundwater
a
Numbers in parentheses are minimum resolving power needed for
that specific molecule
b
Concentrations for interfering elements are 10 µgL
–1
for Cr, Ti,
Ni, Sn, Sb, and Mo, 500 µgL
–1
for Ba and Sr, 1,000 µgL
–1
for As,
2mgL
–1
for Fe, 40 mg L
–1
for Br. All solutions were in 0.1%
HCl–1%HNO
3
diluent solution

level itself. This suggests that in these cases the major
contributors to the blank were not the water or reagents,
but the test tubes and pipettor tips used for dilution, the
sample introduction, and instrument systems (i.e., rinse
blank), or perhaps minor spectral interferences.
It is possible that for some elements the blank levels are
dominated by isobaric interferences. The magnitude of po-
tentially
significant interferences are expressed as equiva-
lent interfered-element concentrations and given in Table2.
Interferences are insignificant (e.g., <1% that of sample
concentrations) for normal groundwater samples and there-
fore do not require correction. However for some samples
with abnormally high Fe, S (reducing groundwater), Br
(brines), Ni, Cr, Ti, As, Mo, Sn, Sb, and Ba (wastewater)
concentrations, the interferences cannot be ignored.
Table3 also lists typical detection limits, calculated as
3
σ
standard deviations of 10 measurements for the 0.1%
HCl in diluent solution for isotopes with no unresolved in-
terferences. Residual interferences for some isotopes (98Mo,
111
Cd,
133
Cs,
115
In,
43
Ca,
47
Ti,
66
Zn,
121
Sb,
137
Ba,
78
Se,
88
Sr)
were determined with 0.1% HCl–1% HNO
3
solutions con-
taining internal standard elements and possible levels of
interfering elements in diluted groundwater (Table2). The
detection limits for individual real samples may be some-
what higher, probably closer to those listed in Table 2, due
to sample-to-sample concentration variations for interfer-
ing elements and other matrix effects that are often diffi-
cult to evaluate.
The high sensitivity and resolving power of the instru-
ment result in detection limits of the order of µgL
–1
for
major elements and sub-µgL
–1
to sub-ng L
–1
level for trace
elements. In comparison, the detection limits of the method
for many elements (U, As, Co, V, Se) are several orders of
magnitude lower than for US EPA method 200.8. Other
515
Table 3 Blanks, detection lim-
its of the HR ICP-MS method.
The WHO guideline values and
US EPA maximum contami-
nant levels (MCL), maximum
contaminant level (MCLG) are
given for comparison
Blanks Detection limits Regulations
0.1% HCl Diluent This US EPA WHO EPA EPA
+diluent method (200.8) MCL MCLG
Major elements (mg L
–1
)
Na 0.06 0.07 0.01
Mg 0.002 0.007 0.001
Si 0.009 0.004 0.007
K 0.1 0.1 0.01
Ca 0.09 0.06 0.06
Trace elements (µgL
–1
)
P 0.7 0.9 0.04
S 3.6 3.9 0.5
Mn 0.08 0.13 0.05 0.1 500
Fe 2.5 1.7 0.4
Al 0.5 0.3 0.04 0.05
Ti 0.08 0.07 0.1
Cr 0.08 0.05 0.03 0.07 50 100 100
Co 0.008 0.05 0.005 0.03
Ni 0.11 0.09 0.09 0.2 20
Cu 0.09 0.06 0.02 0.03 2,000 1,300 1,300
Zn 0.3 0.2 0.2 0.2
As 0.08 0.08 0.07 0.9 10 50 0
Se 0.15 0.17 0.15 7.9 10
Cd 0.01 0.02 0.03 0.1 3 5 5
In 0.04 0.02 0.06
Ba 0.15 0.03 0.3 0.5 700 2,000 2,000
Hg 0.06 0.07 0.15 1 2 2
Li 0.01 0.01 0.002
V 0.05 0.006 0.02 20
Sr 0.02 0.01 0.01
Mo 0.06 0.07 0.03 0.1 70
Sn 0.01 0.01 0.01
Sb 0.007 0.004 0.06 0.1 5 6 6
Cs 0.0004 0.0006 0.02
Re 0.001 0.0002 0.001
Pb 0.1 0.06 0.05 0.1 10
Bi 0.005 0.001 0.002
U 0.001 0.0006 0.00007 0.02 2 30

detection limits are considerably lower (Ba, Mo, Sb, Ni,
Cd, Pb, and Mn) or at least comparable (Al, Zn, Cr, and
Ba). The detection limits of the method for elements of
health concern are also several orders of magnitude lower
than WHO guideline values, EPA maximum contaminant
level (MCL), or even the EPA maximum contaminant
level goal (MCLG).
Precision
Typical drift in instrument sensitivity over a 6-hour period
is of the order of 10–15% for
45
Sc, 15–25% for
74
Ge,
5–10% for ArS and
203
Tl. Superimposed on this gradual
drift are sample-to-sample variations of up to 5% due to
fluctuations in the sample aspiration rate. Three internal
standards spanning the mass range were monitored at dif-
ferent resolutions because the drift typically increases with
resolution.
The effectiveness of drift correction using internal stan-
dards
to improve precision is demonstrated by results
from repeated analysis of LDEOGW every 30–40 samples.
In-run precision as 1
σ
standard deviations of 7 replicate
analyses of LDEOGW during a run (Table4) shows that
correction based on internal standards improves the in-run
precision to generally better than 3%. Comparable preci-
sion was observed for repeated analysis of LDEOGW
over a 3-month period, with the exception of Al (≈5%)
516
Table 4 Summary of measurement results for a laboratory control sample (LDEOGW) and a certified standard (NIST 1640)
LDEOGW NIST 1640
LDEOGW In-run
a
In-run
a
Long-term Ref ± HR ICP-MS ± %
(uncorr) (corr) (n=26) Values (n=26)
Major elements (mg L
–1
)
Na 20.8 10.7 3.4 1.8 29.35 0.31 29.0 0.6 1.9
Mg 18.3 10.3 3.1 3.5 5.819 0.056 5.79 0.12 2.1
Si 25.3 8.9 1.8 2.6 4.73 0.12 4.80 0.15 3.1
K 19.8 8.3 1.1 2.2 994 27 1002 44 4.4
Ca 71.3 6.7 0.5 1.9 7.045 0.089 7.15 0.11 1.5
Trace elements (µgL
–1
)
P 1,380 7.4 2.1 0.9
S 1,980 4.8 0.8 1.4
Mn 1,896 4.6 0.4 0.9 121.5 1.1 120.5 1.7 1.4
Fe 1,947 5.4 3.0 1.0 34.3 1.6 33.4 1.5 4.5
Al 2,204 7.9 1.3 5.5 52.0 1.5 52.3 0.5 1.0
Ti 3.8 6.7 3.7 4.5
Cr 7.8 6.4 1.6 3.5 38.6 1.6 37.7 1.4 3.7
Co 5.8 5.5 2.9 2.1 20.28 0.31 20.16 0.33 1.6
Ni 26.9 4.2 5.2 3.8 27.4 0.8 26.8 1.4 5.2
Cu 10.6 6.7 1.3 3.3 85.2 1.2 84.9 2.0 2.4
Zn 45.6 6.7 2.6 1.6 53.2 1.1 53.9 0.9 1.7
As 320 6.0 4.8 2.0 26.67 0.41 26.3 0.5 1.9
Se 5.2 9.0 8.2 10.0 21.96 0.51 20.9 1.8 8.6
Cd 2.50 2.9 4.0 3.5 22.79 0.96 21.4 1.1 5.1
In 10.1 1.6 1.5 1.2
Ba 465 5.5 3.2 3.2 148.0 2.2 149.0 3.1 2.1
Hg 3.4 4.6 3.5 3.0
Li 10.0 1.8 3.8 2.2 50.7 1.4 50.9 1.4 2.8
V 5.6 9.5 1.5 1.8 12.99 0.37 13.10 0.20 1.5
Sr 191 3.0 2.4 2.3 124.2 0.7 123.4 2.5 2.0
Mo 4.82 2.7 1.9 1.5 46.75 0.26 46.8 0.9 1.9
Sn 10.5 9.0 6.8 4.3
Sb 0.97 7.2 3.4 1.7 13.79 0.42 13.3 0.5 3.4
Cs 1.20 3.1 7.6 2.5
Re 1.50 5.5 2.7 2.0
Pb 3.57 4.1 2.2 3.5 27.89 0.14 27.60 0.30 1.1
Bi 1.50 8.0 4.2 3.0
U 2.60 4.6 3.2 2.9
a
1
σ
relative standard deviation (%) of three measurements for a sample. “Uncorr” is without internal standard drift correction, while
“corr” is with internal standard drift correction

and Se (10%). The slightly higher variability in Al is prob-
ably
due to a more variable Al blank in the system. Our
analyses of groundwater samples suggested that Al con-
tent could vary by 4 orders of magnitude, which some-
times generates memory effects that are difficult to elimi-
nate. The precision of Se measurements is limited by the
low count rate (because of lower ionization efficiency of
Se, high-resolution acquisitions, and low concentrations
of Se in analyzed groundwater) and some residual inter-
ference from the very large Ar–Ar peak.
Accuracy
The accuracy of the measurements was determined by re-
peated analysis of external standard SRM 1640. This stan-
dard solution is rather different in composition from that
of typical drinking water (e.g., Pb, Se, Cd, Cr exceed EPA
MCL) or groundwater but is probably the closest that is
available. Twenty-six replicates of SRM1640 were ana-
lyzed during a 3-month period. For all 24 elements our re-
sults within uncertainty are in excellent agreement with
certified or reference values, which were determined with
11 different methods excluding HR ICP-MS (Table 4).
The largest differences occur for Cd (6%), Se (5%), and
Sb (4%); all other elements agree within 3%. The accu-
racy of all elements (in particular the 9 elements not refer-
enced in SRM 1640) is also demonstrated by results of the
standard addition recovery experiment. Most recoveries
of the standard additions are close to 100%, and typically
within 3% (Fig. 1). This clearly suggests that despite sig-
nal suppression from groundwater matrix, the matrix ef-
fect can be sufficiently corrected by internal standard iso-
topes in the same sample solution. In general, groundwa-
ter, tap water, and drinking water samples show more con-
sistent recoveries than the two wastewater samples. This
is probably due to a larger matrix effect and higher con-
centrations of interfering elements in the wastewater sam-
ples.
Applications to the analysis of Bangladesh groundwater
Groundwater is increasingly used without treatment as a
source of potable water, particularly in developing coun-
tries where surface water is frequently contaminated with
microorganisms. The rapid acquisition of data for many
elements by HR ICP-MS greatly facilitates groundwater
studies, monitoring, and remediation efforts. Soon after
the development of the method, a batch of 629 tube well
samples from Araihazar upazila, Bangladesh, was ana-
lyzed for 31 elements (Se and Ti was not targeted initially
for these samples). The analytical work was completed
within about one week. A summary of the concentrations
is presented in Fig. 2. The concentration of each analyte
varies by at least one, and frequently by several, orders of
magnitude. In the case of As, Al, S, V, Cr, Co, Ni, Cu, Zn,
Mo, Cd, In, Sn, Sb, Cs, Re, Hg, Pb, Bi, and U, minimum
concentrations are the same as the detection limit of even
this sensitive method. Our results also show that in addi-
tion to As, a large number of tube wells that tap into shal-
low aquifers of Araihazar, Bangladesh, do not meet WHO
guidelines for drinking water [8, 9]. In the case of Mn and
U, 84% and 5% of the 629 groundwater samples that were
analyzed did not meet the WHO guideline values of 500
and 2 µgL
–1
, respectively. On the other hand, the concen-
trations of 10 other elements (Cr, Ni, Cu, Se, Cd, Ba, Hg,
Mo, Sb, and Pb) measured for the same set of samples
were well below their respective WHO guideline values.
Our observations based on this regional study are gener-
ally consistent with recent nationwide surveys [12, 13].
The fractions of wells that meet the As and Mn guidelines
set by WHO guidelines in Araihazar are significantly
lower than the equivalent statistic for the entire country
(Table5). Only 25 (4%) of the 629 tube wells that were
tested simultaneously meet the WHO guideline limits for
As, Mn, and U.
517
Fig. 1 Recovery of standard additions into real groundwater
(BGW-1 and BGW-2), tap water (Tap), commercial drinking water
(Drink), and two waste water samples (Waste-1 and Waste-2).
Note that the data points are for all 33 elements in 6 samples
Fig. 2 Summary of concentrations of 31 elements in 629 ground-
water samples from Bangladesh

518
Conclusions
This HR ICP-MS method provides a convenient means of
routinely measuring the concentrations of many elements
in aqueous samples with good precision and accuracy. The
analysis needed only a single dilution, and fast data col-
lection takes less than 10 min per sample. The method has
lower detection limit and better precision than quadrupole
ICP-MS methods, especially for some elements of health
significance, such as As and Se. The lower detection lim-
its are important for studying the subsurface behavior of many
metals and metalloids whose environmental levels can be
very low. Through careful selection of resolution, isotopes,
and collectors, the method covers all major and most trace
elements of health significance or geochemical interest.
This is an improvement from existing HR ICP-MS meth-
ods that only target a small number of elements whose
concentrations usually do not vary by many orders of
magnitude [1, 2, 3, 4, 5, 6, 7].
Although developed primarily for groundwater analy-
sis, the method can also be applied to a wide range of ma-
trices such as drinking water, surface water, and even
some wastewater. The good recovery rates from the stan-
dard addition experiments suggested that sensitivity loss
due to matrix effect can be corrected with internal standards.
In addition, the list of analytes could be expanded for a
few other elements of potential health concern, such as
Be, Ag, and Th; however, we have not done so because
their concentrations in groundwater are generally too low
to be of sufficient interest.
The method does not provide the best detection limit
and precision for some elements, however, compared to
the capability of the instrument when a single element is
targeted. A much lower detection can be reached for As
and Se, for instance, by combining HR ICP-MS with hy-
dride generation [14]. But the method reported here gives
approximately 10% precision with a detection limit at
0.2 µgL
–1
for Se and approximately 5% precision with a
detection limit of 0.10 µgL
–1
for As, both of which are
sufficient for environmental analysis under the current WHO
and US EPA MCL standards. Mercury (Hg) is an element
that is typically analyzed with cold vapor methods ap-
proved by US EPA. Our standard addition experiments with
fresh prepared Hg solution and analyzed by HR ICP-MS
also show excellent recovery rates, suggesting that Hg
analysis with HR ICP-MS at detection limits lower than
0.2 µgL
–1
may be possible, baring memory effects which
have been noted by many researchers. The accuracy of
such Hg analysis by HR ICP-MS needs to be verified with
other methods, however.
Acknowledgments The development of this method was sup-
ported by the US NIEHS/Superfund Basic Research Program
(Grant P42ES10349). NSF grant OCE 9977429 contributed to the
purchase of the VG Axiom. We thank Xiaoguang Meng (Stevens
Institute of Technology, Hoboken, NJ) for providing the two waste
-
water samples. This is Lamont Doherty Earth Observatory contri-
bution number 6600.
References
1. US EPA (1992) Methods for the determination of metals in en-
vironmental samples. US Environmental Protection Agency,
Cincinnati, Ohio
2. Matschat R, Czerwensky M, Hamester M, Pattberg S (1997)
Fresenius J Anal Chem 359:418–423
3. Schnetger B (1997) Fresenius J Anal Chem 359:468–472
4. Cullen JT, Field P, Sherrel M (2001) J Anal Atom Spectrom
16:1307–1312
5. Field PM, Cullen JT, Sherrell RM (1999) J Anal Atom Spec-
trom 14(9):1425–1432
6. Axelsson MD, Rodushkin I, Ingri J, Ohlander, N (2002) Ana-
lyst 127:76–82
7. Field PM, Sherrell RM (2003) J Anal Atom Spectrom 18(3):
254–259
8. WHO (1996) Guidelines for drinking-water quality, 2nd edn,
vol 2, health criteria and other supporting information. World
Health Organization, Geneva, p 940
9. WHO (1998) Guidelines for drinking-water quality, 2nd edn,
addendum to vol 2. World Health Organization, Geneva, p 281
10. Van Geen A, Zheng Y, Versteeg R (2003) Water Resour Res
39(6):1140–1153
11. Allard B (1995) Groundwater. In: Salbu B, Steinnes E (eds)
Trace elements in natural waters. CRC, Boca Raton, p 151
12. BGS/DPHE (2001) Arsenic contamination of groundwater in
Bangladesh, Final Report. British Geological Survey, Key-
worth, UK
13. Frisbie SH, Ortega R, Maynard DM, Sarkar B (2002) Environ
Health Persp 110:1147–1153
14. Klaue B, Blum JD (1999) Anal Chem 71(7):1408–1414
Table 5 Percentage of wells that meet the WHO guideline limits (µgL
–1
) for As, Mn, and U in this study (Northeast Araihazar) and in
3other datasets for Bangladesh nationwide survey
Analytical Method This study BGS survey
a
BWDB
a
Frisbie et al.
b
HR ICP-MS HG-AFS, HG-ICP-AES, ICP-AES ICP-AES, ICP-MS ICP-MS
Total samples analyzed 629 3,534 101 112
% of samples As<10 11 58 69 51
% of samples Mn<500 16 65 68 50
% of samples U<2 95 – 88 –
% of samples As<10 and Mn<500 and U<2 4 – 44 –
a
BGS report (2001), Phase 2. “BGS Survey” is national hydrochemical
survey data, and “BWDB” is Bangladesh Water Quality Monitoring
Network data (only samples with complete As-Mn-U data were used
b
Ref. [13]