&EPA
United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
P.O. Box 93478
Las Vegas NV 89193-3478
Pre Issue Copy
November 1989
Research and Development
Pre-Concentration Method
for inductively Coupled
Plasma- Mass Spectrometry
Project Report/
Project Summary
-------
PRE-CONCENTRATION METHOD FOR
INDUCTIVELY COUPLED PLASMA-MASS SPECTROMETRY
by
J. T. Rowan
Lockheed Engineering and Sciences Company
1050 E. Flamingo Road, Las Vegas, Nevada 89119
and
E. M. Heithmar
U.S. Environmental Protection Agency
P.O. Box 93478, Las Vegas, Nevada 89193-3478
Contract 68-03-3249
Project Officer
E. M. Heithmar
Quality Assurance and Methods Development Division
Environmental Monitoring Systems Laboratory
Las Vegas, NV 89193-3478
OFFICE OF MODELING, MONITORING SYSTEMS AND QUALITY ASSURANCE
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, DC 20460
-------
NOTICE
The research described in this document has been funded wholly or in part by the United States
Environmental Protection Agency under Contract 68-03-3249 to Lockheed Engineering and Sciences
Company. It has been subject to the Agency's peer and administrative review, and it has been
approved for publication as an EPA document. Mention of trade names or commercial products does
not constitute endorsement or recommendations for use.
11
-------
ABSTRACT
A semi-automated system is used to pre-concentrate Ti, V, Mn, Fe, Co, Ni, Cu, Cd, and Pb.
The pre-concentration system accepts digests with acid concentrations equivalent to 0.8% - 1.4%
nitric acid, neutralizes them and loads them onto a macroporous iminodiacetate resin. The alkali and
alkaline earth metals, along with deleterious anions such as chloride, are washed off the resin before
the concentrated analytes are eluted with nitric acid. Measurement of a total of 13 isotopes of the
analytes, as well as two internal standard elements added to the eluant stream, indicates that the
technique enhances the ICP-MS response of the target metals. Investigation of the nature of the
blank signals suggests that the detection limits of several of the isotopes could benefit by much larger
pre-concentration factors, but those of copper, cadmium and lead are currently limited by reagent
purity and laboratory contamination. Method performance data is presented for several simple
synthetic matrices, synthetic sea water, two waste waters and a natural surface water.
This report was submitted in fulfillment of contract number 68-03-3249 by Lockheed
Engineering and Sciences Company under the sponsorship of the U.S. Environmental Protection
Agency. This report covers a period from October 1, 1988 to September 30, 1989. Work is on-
going.
111
-------
IV
-------
CONTENTS
Notice ii
Abstract iii
Figures vi
Tables vii
1. Introduction 1
2. Conclusions 3
3. Recommendations 4
4. Materials and Methods 5
5. Experimental Procedures 9
6. Results and Discussion 11
References 32
Appendix
A. On-Line Pre-Concentration of Trace Metals Prior to Determination
by Inductively Coupled Plasma-Atomic Emission Spectrometry or
Inductively Coupled Plasma-Mass Spectrometry 33
-------
FIGURES
Number Page
1 Schematic representation of on-line pre-concentration device 7
2 Effect of buffer concentration on pre-concentration profile
of titanium (m/z 48 shown) 12
3 Typical elution profile for cobalt, along with the intensities of
the two post-column internal standards 14
4 Typical blanks and 0.5-fJ.g/L standard responses with the
10-mL sample loop 17
1-A Pre-concentration apparatus 35
VI
-------
TABLES
Number Page
1 ICP-MS CONDITIONS 6
2 INTEGRATED PRE-CONCENTRATION INTENSITIES COMPARED TO
INTEGRATED DIRECT NEBULIZATION INTENSITIES 13
3 RELATIVE ANALYTE INTENSITIES WITH DIRECT NEBULIZATION
OF VARIOUS MATRICES 15
4 DETECTION LIMITS (/ig/L, 3-a CRITERION, N=5) FOR
PRE-CONCENTRATION AND FLOW-INJECTION SYSTEMS 16
5 EFFECT OF CYCLING ACID AND BUFFER POST-ANALYSIS RINSES 24
6 PRECISION AS A FUNCTION OF CONCENTRATION AND SAMPLE SIZE 25
7 SPECTRAL INTERFERENCES OBSERVED IN FLOW-INJECTION DIRECT
NEBULIZATION OF SIMPLE SYNTHETIC MATRICES 26
8 PERCENT RECOVERIES FROM SIMPLE SYNTHETIC MATRICES 27
9 APPARENT ANALYTE CONCENTRATIONS IN NON-SPIKED SYNTHETIC
SEA-WATER MATRIX BY PRE-CONCENTRATION AND BY
FLOW-INJECTION DIRECT NEBULIZATION 28
10 PERCENT RECOVERIES FROM SPIKED SYNTHETIC SEA WATER BY
PRE-CONCENTRATION AND BY FLOW-INJECTION DIRECT
NEBULIZATION 29
11 PERCENT RECOVERIES OF TRACE METALS FROM SPIKED WASTE
WATERS AND WASTE-WATER DIGESTS BY PRE-CONCENTRATION
AND BY FLOW-INJECTION DIRECT NEBULIZATION 30
12 APPARENT ANALYTE CONCENTRATIONS IN NON-SPIKED
WASTE-WATER DIGESTS BY PRE-CONCENTRATION AND BY
FLOW-INJECTION 31
1-A ESTIMATED DETECTION LIMITS (Mg/L) OF ANALYTES OF INTEREST 34
2-A PRE-CONCENTRATION SYSTEM DECONTAMINATION PROCEDURE 39
3-A PRE-CONCENTRATION SAMPLE RUN PROGRAM 41
vn
-------
SECTION 1
INTRODUCTION
The analysis of environmental samples for trace elements poses two major problems. First,
the regulatory action levels of several metals are in the low part-per-billion range, and reliable
quantitation requires analytical methods with detection limits 5-10 fold lower than threshold levels.
Second, the matrices encountered in environmental samples are extremely diverse and often highly
complex. Because of the need for low detection limits, graphite furnace atomic absorption
spectrometry (GFAAS) and, more recently, inductively coupled plasma-mass spectrometry (TCP-
MS) have been employed in environmental analysis1'2. ICP-MS has the advantage over GFAAS of
being a multi-elemental technique, but there is incomplete performance data for ICP-MS analyses
of target elements in all matrices of interest to the U.S. Environmental Protection Agency.
One of the major drawbacks of ICP-MS is the interferences often encountered with the
complex matrices common in environmental analysis. These interferences can be spectral1'3 or
physico-chemical4'5 in nature. The former have generally been treated mathematically, by the use
of various fundamental or empirical correction terms in the calibration function. The physico-
chemical effects can be ameliorated to a certain extent by the use of appropriate internal standards4
or alternative sample introduction techniques (e.g. flow injection analysis5). None of these
approaches is completely adequate for very complex matrices, and all decrease the signal-to-noise
ratio to some extent.
Pre-concentration can be used to separate analytes from interferents prior to analysis.
Separation can be effected by solvent extraction7, precipitation8, or by complexation on an
immobilized form of the chelating agent, such as with a resin. The last approach has become
increasingly popular in the last few years. The resin can be digested to liberate the trace elements9,
or the analytes can be released by changing the ionic form of the resin. The latter technique allows
the development of semi-automated methods that make use of resin-packed columns10"16. Some
methods have employed chelating agents adsorbed on hydrophobic resins10, or acidic14 or basic15
alumina, or ion-pairing and adsorption on non-polar resins16, but the largest number of methods
have used some form of an iminodiacetate-functionalized copolymer11"13. These resins were
-------
previously used in the batch mode for pre-concentration and matrix elimination prior to analysis by
GFAAS1 or neutron activation analysis18.
Pre-concentration techniques have not been widely used with ICP-MS. McLaren19 and
Beauchemin used 8-hydroxyquinoline immobilized on silica. They were able to determine several
trace elements in a coastal sea water and a river water. The immobilized hydroxyquinoline is very
efficient; it allows pre-concentration from large quantities of water, and some separation from
magnesium and calcium can be achieved. Plantz et al.16 developed an on-line method using
complexation in solution and adsorption of the metal bis(carboxymethyl)dithiocarbamate on a non-
polar resin. This approach worked well for moderately high salt concentrations, and the amounts
of four trace elements in sea water and urine were determined.
Although iminodiacetate copolymers are among the best resins for the separation of trace
elements from the alkali and alkaline earth metals, their routine use suffers from two drawbacks.
First, many formulations of the resin exhibit pronounced changes in volume with changing ionic
form21. This can seriously impair concentration and elution efficiencies, as well as the physical and
mechanical integrity of the resin. Second, the resin exhibits its best discrimination against alkali and
alkaline earth metals at pH values from 5.0 to 5.8. Several trace element species, especially Fe3+ and
Cu2+, hydrolyze and precipitate at these pH values. The present work uses a commercially available
highly cross-linked copolymer, with reduced tendency to swell, and a flow-injection approach to
sample neutralization and buffering which minimizes hydrolysis. Post-column internal-standard
addition is employed to examine high apparent pre-concentration efficiency and to correct for
instrumental drift. The suitability of on-line pre-concentration for the analysis of environmental
water samples is demonstrated using several synthetic and natural matrices.
-------
SECTION 2
CONCLUSIONS
On-line pre-concentration of Ti, V, Mn, Fe, Co, Ni, Cu, Cd, and Pb with an iminodiacetate
resin minimizes several spectral and physico-chemical interferences in ICP-MS. Interferences from
sodium, potassium, and magnesium are largely eliminated, and that from calcium is greatly
attenuated. Anions which can cause severe spectral interferences, such as chloride, are also removed.
Native organic chelating agents, which might reduce pre-concentration recoveries, are eliminated
from samples by the use of a nitric acid digestion.
The method allows the direct pre-concentration of environmental digests with a fairly wide
range of acid strengths. Analytes give linear responses over a dynamic range of up to five orders
of magnitude. The analyte response functions for the elements studied, with the exception of that
for lead, are very stable. The response drift of lead can be treated by post-column internal
standardization or by hourly re-calibration.
The nature of the limiting noise in the pre-concentration blanks indicates that the detection
limits for most of the elements, except for copper, cadmium, and lead, are not limited by reagent
purity, and significant improvements could be achieved simply by increasing sample size. Carryover
is a limiting factor in the detection limits for V, Co, and to a lesser degree for Ti and Cu. This
effect has been minimized with a more thorough column rinse procedure.
-------
SECTION 3
RECOMMENDATIONS
The successful combination of this sample introduction technique with ICP-MS suggests that
it will also prove applicable to ICP-AES. While more robust, ICP-AES suffers from lower sensitivity
than ICP-MS, and this technique would benefit from the improvement of detection limits. The
current pre-concentration method (Appendix A) is expected to be readily adaptable to ICP-AES,
while the data reduction process may require minor modification.
The efficiency and cost-effectiveness of this method could be improved with expansion to
an automated, multiple-column system, wherein many samples would be in various stages of analysis
at a given time. Unfortunately, the current status of ICP-MS software makes complete automation
of the pre-concentration techniques problematic. Not only is satisfactory support for transient
signals lacking, but so is the hardware and software which would make interfacing a less daunting
task. Several approaches to this aspect of the problem are being investigated.
Methods for discriminating against the pre-concentration of iron and aluminum should be
developed because the high concentrations of these elements in some samples could lead to spectral
interferences. In addition to these efforts, alternate resin chemistries for the pre-concentration of
other elements, such as arsenic and selenium, should be developed.
A multi-laboratory evaluation of this method should be conducted. This would assess the
routine performance of pre-concentration coupled with both ICP-MS and ICP-AES and would
determine the comparability of pre-concentration with conventional sample introduction.
-------
SECTION 4
MATERIALS AND METHODS
INSTRUMENTATION
The inductively coupled plasma-mass spectrometer is the Perkin-Elmer Sciex Elan Model
250, equipped with mass-flow controllers and upgraded ion optics. The instrumental conditions
employed for pre-concentration (Table 1) are optimized for high sensitivity, rather than for
minimum molecular ion interferences, since many molecular ions would be minimized by matrix
elimination.
The pre-concentration device is shown schematically in Figure 1. The system is based on an
iminodiacetate resin column with a high-pressure, programmable pump, and four-way, high pressure
slider valves (all from Dionex Corporation, Sunnyvale, CA), and a 16-channel peristaltic pump
(Lachat Instruments, Mequon, WI). The column is 0.9 cm i.d. by 25 mm long, and contains
approximately 1.5 mL of resin. Both a 2.5-mL and a 10-mL sample loop (0.8 mm and 1.5 mm i.d.
TFE tubing, respectively) were used in various parts of the study.
REAGENTS
Doubly distilled (sub-boiling, in quartz) concentrated nitric acid and acetic acid, ammonia
in doubly distilled (in quartz) water (Seastar, Sidney, B.C.), and distilled, doubly deionized water
were used throughout the study. Stock analyte solutions of titanium, vanadium, manganese, iron,
cobalt, nickel, copper, cadmium, and lead in 1% (v:v) nitric acid, and an internal standard solution
containing yttrium and bismuth at 1 mg/L in 5% nitric acid, were prepared from commercial ICP
standards.
1-M nitric acid was used as one of the two eluant solutions. The other eluant, a solution of
approximately 2-M ammonium acetate at a pH of 5.5, was prepared from 4 M stock solutions of the
Seastar reagents. A second ammonium acetate solution of approximately 2-M concentration was
prepared, and its pH was adjusted with 4 M ammonia or acetic acid, so that, when mixed on-line
with a 1% nitric acid sample (see Figure 1), the pH of the resulting solution was 5.5.
-------
TABLE 1. ICP-MS CONDITIONS
Torch: Normal "short" type
Gas flows: Plasma - 12.0 L/min
Auxiliary - 1.8 L/min
Nebulizer - 1.1 L/min
Sample flow (by peristaltic or high-pressure pump): 1.0 mL/min
Sampler position: 18 mm above load coil
Sampler/skimmer: platinum
(1.1 mm sampler orifice, 0.9 mm skimmer orifice)
Lens settings: First and third Einzel lenses - -18.1 v
Bessel box plates - -9.7 v
Bessel box barrel - +2.1 v
Bessel box stop lens - -5.3 v
Interface pressure: 2 Torr
Mass spectrometer pressure: 2.5 x 10"5 Torr
Resolution: 0.95 - 1.0 m/z units at 10% peak height
Typical sensitivities at stated conditions: Li - 1000 s'Vg"1!-
Co - ISOOs'Vg^L
In - 3000 s'Vg"1!-
^Tl - 1000 s'Vg^L
Peaks monitored during routine analysis: m/z 46, 48, 49, 51, 52, 54, 55, 57, 60, 62, 63, 65, 89, 111,
208, and 209. m/z 53 substituted for m/z 46 during direct nebulization experiments to allow
correction for C1O+ on 51V.
Measurements/peak: 1
Measurement time: 81 ms
Dwell time: 27 ms (3 dwells per measurement)
Synthetic matrix solutions of 2000-mg/L sodium, potassium, magnesium or calcium were
prepared in 1% nitric acid from the 99.99+ % chlorides (Aldrich Chemicals, Milwaukee, WI).
Similarly, a synthetic sea-water solution was prepared from these chloride salts, sub-boiling,
distilled-in-quartz sulfuric acid (J. T. Baker Chemicals, Phillipsburg, NJ), and nitric acid, so that the
final solution contained 10,560 mg/L sodium, 1,270 mg/L magnesium, 400 mg/L calcium, 380 mg/L
potassium, 21,000 mg/L chloride, and 880 mg/L sulfur. This solution had the same matrix as a sea-
water sample acidified with 0.5% hydrochloric acid and 1% nitric acid.
-------
2 M NH4 Ac
1 M NHO3
Sample
Buffer
Rinse
Blank
Internal_
Standard
1-3
ml/
min.
V1
Waste
VI
4 mL/min.
1 mL/min.
1 mL/min.
0.1 mL/min.
Sample
Loop
Preconcentration
Column
Waste
C V2
\
\
s
s
ICP
Figure 1. Schematic representation of on-line pre-concentration device. Both 4-way valves are
shown in the "on" position; dashed lines indicate flow paths in the "off" position.
-------
SAMPLE PREPARATION
Municipal waste-water samples were obtained from the primary effluent discharge line and
the secondary effluent stream at the Clark County Sanitation District Water Treatment Plant. A
water sample from the Las Vegas Wash was also obtained from a point 5 km south of the Advanced
Water Treatment Plant of the Sanitation District. These samples were filtered through 0.45 jzm
cellulose filters and acidified with 1% nitric acid.
Dissolved organic carbon (DOC) was determined on each sample. DOC values were 13 mg/L,
22 mg/L, and 33 mg/L for the wash water, secondary effluent, and primary effluent, respectively.
The alkali and alkaline earth metal concentrations of the samples were determined by ICP-AES.
These concentrations were all fairly constant among the samples at 210 ± 10 mg Na/L, 22 ± 2 mg
K/L, 52 ± 3 mg Mg/L, and 115 ± 8 mg Ca/L.
A 45-mL aliquot of each filtered (referred to as "raw") sample was placed in a TFE bomb
with 5 mL concentrated nitric acid and digested in a microwave oven (Model 8ID, CEM, Matthews,
NC). The microwave power was maintained at 545 watts for 10 minutes, then lowered to 345 watts
for an additional 10 minutes. The cooled digestate was diluted 10-fold in water.
-------
SECTION 5
EXPERIMENTAL PROCEDURES
PROCEDURE
The pre-concentration procedure begins by placing the sample-uptake line in acidified
sample. The sample, at 4.0 mL/min, is mixed with ammonium acetate at 1.0 mL/min and passed
through the sample loop to waste via valve 1 in the "on" position (refer to Figure 1). During this
time, the resin is being preconditioned to the ammonium form at pH 5.5 with 2-M ammonium
acetate. With valve 2 in the "off" position, the ammonium acetate passes to waste, and 2% nitric acid
rinse solution is delivered to the plasma. The sample does not remain at pH 5.5 for more than 3.4
minutes when the 10-mL loop is used (50 seconds when the 2.5-mL loop is used).
At t = 0.0 minutes (all times are given for 10-mL samples; subtract 3 minutes from all
subsequent times for 2.5-mL samples), the pre-concentration program is initiated. Valve 1 turns off,
and the high-pressure pump sweeps the sample to the resin column with 2-M ammonium acetate at
3 mL/min, pre-concentrating the analytes, and passing the sample matrix, including alkali metals and
anions, to waste. At 4.0 minutes , valve 1 turns on, and the high pressure pump continues to pass 2-
M ammonium acetate through the column to waste. This wash-out period replaces chelated Mg2"1"
and Ca2+ with NH4+. At 6.0 minutes, the high-pressure pump switches to 1-M nitric acid at 3
mL/min. At 6.7 minutes, the pump slows to 1 mL/min and valve 2 turns on, passing the eluted trace
metals to the ICP-MS. Internal standard solution is mixed with the eluate at an approximately 1:10
ratio, yielding 100-/Ltg/L concentrations of Y and Bi internal standards.
At 8.7 minutes, after the analyte peaks have passed, valve 2 turns off and 3 mL/min of 2-
M ammonium acetate is passed through the column for 1 minute. This is followed by successive 3-
mL rinses with nitric acid, ammonium acetate, nitric acid, and finally ammonium acetate, again,
each at a rate of 3 mL/min. The resin is now reconditioned and ready for the next sample, which
generally is already passing through the sample loop. The entire pre-concentration procedure is
conducted with two reagent blanks at the beginning of each day, as the first pre-concentration cycle
generally has elevated backgrounds.
-------
The mass spectra are obtained using the "Multiple Elements" program of the ELAN ICP-MS.
The measurement routine described in Table 1 allows the observation of each peak every 0.67
seconds, and a temporal resolution of 2 seconds. The acquisition of the spectrum is initiated at 6.2
minutes in the pre-concentration program. The analyte peaks generally appear after about 50
seconds and reach near the base line before valve 2 switches off and data acquisition ceases. The
spectrum is converted to an ASCII file and automatically sent via the Kermit communications
protocol to an 80386-based microcomputer for processing. The spectrum is placed in a commercial
spreadsheet by a macro that automatically finds the analyte peaks and integrates each over the
optimal detection period. Because there is some variability in the elution time of the peaks, relative
to the start of data acquisition, the ArC+ intensity at m/z 52 is used as an approximate time
reference for the peak-search algorithm. The exact communications, peak location and calculations
algorithms are available upon request.
10
-------
SECTION 6
RESULTS AND DISCUSSION
EFFECT OF BUFFER IONIC STRENGTH
Initial experiments with off-line buffering involved varying the buffer concentrations from
the minimum amount that would neutralize 0.1% nitric acid solutions (about 0.03 M ammonium
acetate) up to 1-M ammonium acetate. Figure 2 shows the pre-concentration elution profiles for
48Ti at 0.03 M and 0.25 M buffer concentrations. It is apparent that the lower-concentration solution
exhibited very pronounced tailing. This was eliminated by the use of 0.25 M buffer. Concentrations
above 0.25 M had little further effect. Similar results were obtained for isotopes of vanadium, iron,
and lead. Other analytes were less affected. Since the pH values of both solutions were well within
the optimal pH range, the effect was apparently related to ionic strength. Titanium, vanadium, and
iron all have multiple oxidation states with varying affinities for iminodiacetate, but it is unknown
if this was a factor here. Houk noted a similar phenomenon of poor pre-concentration in dilute
solutions and attributed it to a "salting out" effect, but the chemistry of that system was different;
neutral metal chelate was adsorbing on a neutral resin. The reason for the effect of the buffer
concentration in this study is not currently known.
METHOD RUGGEDNESS
The ability to routinely cope with a variety of sample types was a primary consideration in
the development of this method. A fundamental problem was the adequate neutralization of samples
of varying acidity on-line, without contamination from impurities in the buffer. With the sample
digestion techniques used in this work, samples were expected to contain approximately 1% nitric
acid. The on-line buffer was made 2-M, in order to produce a 0.4-M final buffer concentration in
the sample. While this raised the concern of elevated blanks, these proved to be acceptable. In
testing the ruggedness of the method, it was found that the on-line buffering technique adequately
buffered standard solutions, ranging from 0.8% to 1.4% nitric acid, to a pH of 5.4 ± 0.2. Analyte
recoveries from these solutions ranged from 90% to 103% relative to the recoveries from a standard
solution in 1% nitric acid.
11
-------
PRE-CONCENTRATION EFFICIENCY
The efficiency of the pre-concentration system was determined by by-passing the column
and using the sample loop for flow-injection sample introduction directly to the nebulizer. The total
integrated signals for 10-/ig/L solutions of each of the analytes in 1% nitric acid was compared with
the integrated signals after pre-concentration.
Table 2 gives the results of that comparison. It is obvious that the pre-concentration system
is yielding enhanced sensitivity. This phenomenon is confirmed by Figure 3, which shows the pre-
concentrated elution profile of 10-Mg/L cobalt and the corresponding internal standard intensities.
At the front edge of the cobalt elution, the yttrium intensity suddenly increases, followed by a
similar increase in the bismuth intensity. The fact that these effects are not entirely coincidental
indicates that they are not due to mixing anomalies caused by the change in eluant viscosity. It
should be noted that the cobalt elution profile (as well as those of the other analytes) overlaps these
intensity enhancements.
10ppbTi
2.5 mL sample loop
0.25 M buffer
0.03 M buffer
Figure 2. Effect of buffer concentration on pre-concentration profile of titanium (m/z 48 shown).
12
-------
TABLE 2. INTEGRATED PRE-CONCENTRATION INTENSITIES COMPARED TO
INTEGRATED DIRECT NEBULIZATION INTENSITIES0
Element % Relative Signal
Ti 150
V 130
Mn 130
Fe 140
Co 150
Ni 150
Cu 150
Cd 120
Pb 130
Direct nebulization of 1% HNO3 solution = 100%. Relative signals are ±10% (N=2).
In order to establish conclusively that the matrix at the elution front causes an ion-signal
enhancement, several different matrices spiked with lO-jig/L analyte concentrations were directly
nebulized. The first matrix, used as a reference, was 1% nitric acid. The second was 1-M nitric
acid. The third matrix was 0.8 M ammonium nitrate in 0.2-M nitric acid. Finally, a mixture of 0.6
M ammonium nitrate, 0.6 M acetic acid, and 0.2-M nitric acid was analyzed. The results of these
analyses, referenced to intensities in 1% nitric acid, are given in Table 3. The ammonium nitrate,
which is present in the elution front, along with nitric and acetic acids, causes a pronounced
enhancement of signal. The proportions of the various matrix components at the exact time of
analyte elution are not known, and in fact are highly transient. It is apparent, however, that the
matrix of the elution front enhances analyte sensitivities even more than would be expected by pre-
concentration.
DETECTION LIMITS
The detection limits obtained with any pre-concentration system depend on the mode of peak
detection (area vs. peak height), and the integration time (or time constant, in the case of peak-
height measurement). As a result of preliminary studies, it was decided that peak areas would be
used for quantitation. Peak widths, and therefore peak heights, sometimes varied considerably
during the day. Peak areas, conversely, were very reproducible. In order to establish optimal
13
-------
parameters, the integration time was varied from 20 seconds to 60 seconds, with different starting
points for the integration window. The detection limits calculated from the analysis of five blanks
exhibited a broad minimum for nearly every element from 20 seconds integration time, to 50 seconds
integration, with an optimal starting point of 10 seconds before the manganese signal maximum.
The manganese peak was chosen as a reference, because its elution profile always exhibited a distinct
local maximum coincident with the elution time of manganese. Since the full-widths at half-
maximum (FWHM) of the peaks could vary from 14 to 20 seconds, a conservative integration time
of 40 seconds was selected. Table 4 gives the detection limits found for each isotope for both the
2.5-mL loop and the 10-mL loop. The detection limits obtained under identical ICP-MS conditions
by flow-injection analysis, wherein samples were loaded into a 300-/JL sample loop and injected
directly to the nebulizer, by-passing the pre-concentration column, are also presented in Table 4.
Co
Y internal std.
Bi internal std.
i—i—i r—i—i—i
Figure 3. Typical elution profile for cobalt, along with the
intensities of the two post-column internal standards.
14
-------
TABLE 3. RELATIVE ANALYTE INTENSITIES WITH DIRECT NEBULIZATION OF
VARIOUS MATRICES0
Element 1-M HNO3 NH4NO3/HNO3 NH4NO3/acetic/HNO3
Ti
V
Mn
Fe
Co
Ni
Cu
Cd
Pb
" 1% nitric acid matrix =
110
110
110
110
110
100
100
90
110
100%.
140
140
140
140
140
130
120
100
130
Relative signals are ±10% (N=2).
140
140
140
130
130
150
130
80
130
Two complementary approaches can be taken to improve these detection limits. If the blank
noise is predominantly from sources other than analyte signal, increasing the sample size should
provide a virtually linear improvement in signal-to-noise ratio. On the other hand, if the analyte
is the major contributor to blank noise, further clean-up of reagents should yield lower blanks, and
consequently lower detection limits. The latter case was of concern to us, as the reagents were all
used without further purification, and all operations were performed in a general-purpose
instrumentation laboratory. Examination of the elution profiles of the blanks helped to establish the
predominant blank contributions.
Figure 4(a) shows a representative blank for i, along with the signal obtained for 0.5-
Mg/L Ti. Although there appears to be a substantial peak in the blank corresponding to the elution
time of the titanium, the maximum contribution to this by analyte can be seen by multiplying the
49Ti blank profile by the ratio of the natural abundances of ^i and 49Ti, which is 13.4. The result
is shown at the bottom of Figure 4(a), and accounts for no more than 30% of the blank. The rest of
the blank must be molecular in nature. One possible contribution, also shown, can be inferred by
assuming the m/z 46 intensity is entirely due to NO2+ and the m/z 52 peak is due only to ArC+, and
multiplying by the appropriate isotopic ratios. The remainder of the m/z 48 blank is of unknown
molecular origin.
15
-------
TABLE 4. DETECTION LIMITS (Mg/L, 3-a CRITERION, N=5) FOR
PRE-CONCENTRATION AND FLOW-INJECTION SYSTEMS
Pre-concentration
Flow-In iection
Isotope
«Ti
49Ti
V
Mn
54Fe
57Fe
Co
^Ni
62Ni
63Cu
65Cu
ulCd
208pb
a Detection limit
2.5-mL Sample
0.1
0.05
0.08
0.03
20
2
0.002
0.1
0.1
0.05
0.05
0.03
0.08
in high chloride matrices = 4
10-mL Sample
0.04
0.02
0.006
0.006
1
0.3
0.0004
0.01
0.02
0.004
0.007
0.006
0.005
Mg/L.
300-/iL Sample
1
0.7
0.1°
0.3
50
30
0.05
0.2
0.4
0.5
0.3
0.3
0.5
The similarity of blank and 0.5-Mg/L sample profiles in Figure 4(b) shows that, unlike its
sister isotope, pre- concentrated 49Ti apparently accounts for the majority of its blank. Figure 4(c)
demonstrates that the m/z 51 blank is also probably dominated by pre-concentrated analyte. Figure
4(d) for m/z 55, on the other hand, exhibits a blank profile that is dominated by molecular
contribution, possibly by ArNH"1".
The two iron profiles in Figures 4(e), m/z 54, and 4(f), m/z 57, are quite a contrast. 5
of Fe creates just a small shoulder on the m/z 54 blank, which is dominated by the ArN+ surge from
the 1-M nitric acid. The m/z 57 blank shows a small peak corresponding to pre-concentrated iron
on top of the fairly constant ArOH+ background. These two figures clearly demonstrate the reason
for the superior 57Fe detection limit.
16
-------
0.5 ppb Ti
m/z 48 blank
NO++ArC +
2
140
160
180
time (s)
0.5 ppb Ti
m/z 49 blank (x5)
140
160
180
time (s)
Figure 4. Typical blanks and 0.5-/ng/L (ppb) standard responses with the 10-mL sample loop (unless
otherwise indicated in the legend) for (a) ^i, (b) 49Ti, (c) V, (d) Mn, (e) 54Fe, (f) 57Fe,
(g) Co, (h) ^Ni, (i) 62Ni, (j) 63Cu (65Cu similar), (k) mCd, and (1)
17
-------
40 -I
70 -|
(C)
0.5 ppb V
m/z 51 blank (x10)
160
time(s)
(d)
0.5 ppb Mn
m/z 55 blank (x5)
160
180
40 60 80 100 120 140
180
Figure 4 (continued).
18
-------
200 -
190 -
180 -
"in
!
£
| 150 H
3
| 140-
130 -
120 -|
110 -
100
(e)
5 ppb Fe
m/z 54 blank
20
40
60
80 100
time (s)
120
140
160
180
5 ppb Fe
m/z 57 blank
40 60 80 100 120 140 160 180
Figure 4 (continued).
19
-------
60 -i
50 -
1 40 H
I 30 H
c
o
20 H
10 -
(g)
0.5 ppb Co
m/z 59 blank (x20)
20 40 60 80 100 120 140 160 180
time (s)
0.5 ppb Ni
m/z 60 blank
60 80 100 120 140 160 180
Figure 4 (continued).
20
-------
en
•O
I
c
3
C
O
40 60 80 100 120 140 160 180
0.5 ppb Ni
m/z 62 blank
10 ml
m/z 62 blank
2.5 ml
"en"
CO
CO
3
O
.C
a>
ra
c
8
0
26-
24 -
22-
20-
18-
16-
14 -
12-
10-
8 -
6-
4 -
2 -
0)
0.5 ppb Cu
m/z 63 blank
20
40 60 80 100 120 140 160 180
Figure 4 (continued).
21
-------
c
.1
6 -i
5 -
in
TJ
CO
-------
The m/z 59 blank (Figure 4(g)) shows another large contribution from a transient molecular
ion, although there may also be a small analyte contribution as well. The m/z 60 blank (Figure
4(h))is apparently dominated by molecular transients, while the m/z 62 blank in Figure 4(i) has a
small, but perceptible, contribution from analyte, judging from the smaller size of the second half
of the 2.5-mL blank. Blanks for both copper isotopes are dominated by analyte contribution (see
Figure 4(j) for m/z 63), as could be expected for this common contaminant. Figures 4(k) and 4(1)
are also dominated by cadmium and lead contamination, respectively.
It is apparent from these blank profiles that only in the cases of copper, cadmium and lead,
as well as 49Ti, would rigorous reagent purification and stringent clean-room conditions significantly
improve detection limits. Conversely, the detection limits of the other isotopes would definitely
improve with larger sample sizes. Even in the cases of copper, cadmium, and lead, blank areas did
not increase by a factor of four from the 2.5-mL sample loop to the 10-mL loop, so some
improvement with increased sample size is still possible. This is because there are contributions to
the blanks for these elements from the column reconditioning and the wash-out solutions; these
contributions are not dependent on sample size.
MEMORY
Early experiments using the pre-concentration system used a continuous post-analysis rinse
with 1-M nitric acid for 4 minutes at 3 mL/min, prior to the column reconditioning with 1 minute
of ammonium acetate. Considerable memory was apparent for some analytes, particularly vanadium,
in the first blank after a high concentration standard. Successive blanks decreased until a base-line
was reached after 2 or 3 blanks. Since the degree of memory did not appear to be greatly affected
by length of nitric acid rinse or of ammonium acetate reconditioning, it was surmised that the
alternation between ionic forms was the essential factor in decreasing the blank. Various programs
of alternating rinse times were tested, and the regime described in the Procedure was chosen as the
best compromise between speed of analysis and freedom from memory. Table 5 shows the apparent
memories in the first blank analyzed after a 10-/ig/L standard, for the original program of 4 minutes
of acid wash, and for the 4 minutes of alternating 1 minute acid and buffer segments. It is not clear
why vanadium, and to a lesser extent, titanium, cobalt and copper exhibit memory. The effect is not
limited to elements likely to form oxy-acids that would not elute well in nitric acid. The fact that
alternating the ionic form alleviates the problem to such a large extent suggests that the swelling and
shrinking of the resin may be involved. Alternatively, perhaps the chemical properties of the frontal
zones between acid and buffered eluants increase the rate of elution for these analytes.
23
-------
TABLE 5. EFFECT OF CYCLING ACID AND BUFFER POST-ANALYSIS RINSES0
Element 4-min acid rinse Two 2-min acid/buffer cycles
Ti 0.05 ND
V 0.15 0.06
Co 0.03 0.01
Cu 0.12 ND
a Only elements exhibiting detectable memory are listed. Values given are apparent concentrations
in the first 10-mL blank after a 10-mL, 10-Mg/L standard.
LINEARITY AND PRECISION
Calibration curves were determined for each of the analytes from 0.3 M8/L to 30
using the 2.5-mL sample loop, and from 0.3 to 10 Mg/L, using the 10-mL sample loop (the iron
calibration curves were determined using ten times these concentrations). All thirteen isotopic
calibrations were linear, with correlation coefficients ranging from 0.998 to 0.9999. Table 6 gives
the precision obtained at 10 and 0.5 Mg/L with the 2.5-mL sample loop, and at 0.5 Mg/L with the
10-mL sample loop. Generally, the long-term precision of the method was good, but the sensitivity
for lead tended to drift during the day. Variable sensitivity in the high mass range is occassionally
encountered in ICP-MS. Lead calibrations and analyses used the internal standard intensity of
bismuth to correct this drift.
SIMPLE SYNTHETIC MATRICES
The degree of residual interferences from molecular ions of alkali and alkaline earth metals,
as well as chloride, was examined as a function of pre-elution wash-out time. Solutions (2000 mg/L)
of each major alkali and alkaline earth metal were pre-concentrated, and the ammonium acetate
wash-out times were varied from 1 to 6 minutes. Only 2000 mg/L calcium produced residual
spectral interferences greater than the detection limits given in Table 4 at wash-out times greater
than one minute. After 2 minutes wash-out, the calcium solution produced an apparent Ti signal at
m/z 48 corresponding to 75 Mg/L Ti. This is caused by residual ^Ca in the resin, equal to about 1.5%
of the original Ca concentration. The only molecular spectral interferences remaining after two
minutes wash-out were the equivalent of 270 Mg/L iron at m/z 57 (^CaOH"1"), the equivalent of 0.1
Mg/L cobalt at m/z 59 (42CaOH+, ^Ca^OH, etc.), and the equivalent of 1 /ig/L nickel at m/z 60
24
-------
TABLE 6. PRECISION AS A FUNCTION OF CONCENTRATION AND SAMPLE SIZE"
Element
*Ti
4*Ti
V
Mn
54Fe*
5W
Co
<*Ni
62Ni
63Cu
65Cu
inCd
208pb
2.5-mL
10 Mg/L
0.5
2.0
0.7
0.5
2.1
3.3
2.0
2.0
1.9
3.7
3.5
1.1
2.2
Samole
0.5 Mg/L
3.8
4.7
2.5
3.2
NDC
3.6
3.2
3.2
5.9
9.6
10.9
3.4
2.7
10-mL Sample
0.5 Mg/L
2.2
1.9
3.3
0.8
1.2
4.6
2.1
2.7
5.7
5.1
5.4
3.4
2.3
a Percent relative standard deviation (N=4).
b Concentrations of iron were 100 and 5
c Concentration near detection limit at this sample size.
Precision not determined.
). No molecular interferences remained after 6 minutes of wash-out, and only the equivalent
of 1.8 Mg/L Ti remained at m/z 48, corresponding to 0.03% of the original Ca concentration. It
should be noted that the 2000-mg/L synthetic matrices also contained up to 5900 mg/L chloride,
which never produced any molecular interferences.
The 2000-mg/L synthetic matrices were also analyzed by direct nebulization. In these
experiments, the column was again by-passed and a 300-ML sample loop was installed on valve 1.
This flow-injection system minimized the build-up of solids on the sampler orifice6. The calibration
was based on the ratios of signals to the on-line Y and Bi internal standards, as the sodium and
potassium matrices caused significant signal suppression. Table 7 lists the apparent concentrations
of analytes caused by direct nebulization of the four simple synthetic matrices. As expected, all of
25
-------
TABLE 7. SPECTRAL INTERFERENCES OBSERVED IN FLOW-INJECTION DIRECT
NEBULIZATION OF SIMPLE SYNTHETIC MATRICES0
Matrix Element
Analyte
*Ti
49Ti
V
Mn
54Fe
57Fe
Co
*°Ni
62Ni
63Cu
65Cu
Na
ND6
12
15
ND
ND
ND
ND
1.2
62
21
ND
K
ND
8
11
6
ND
52
ND
0.4
ND
1.4
0.8
Mg
2.9
27
40
0.4
ND
51
2
1.2
2.7
0.2
0.5
Ca
c
14
20
0.6
ND
22000
15
270
6.6
ND
13
" Apparent concentration of analyte (MS/L) caused by 2000-mg/L matrix element and associated
chloride.
b Not detected at the detection limit of flow-injection direct nebulization method.
c Not measured - exceeded dynamic range of ICP-MS.
the solutions produce strong interferences on m/z 51, due mainly to 35C1O+, when aspirated directly.
35C1N+ also interferes at m/z 49. The alkali and alkaline earth metals contribute their own
interferences. In short, every isotope between m/z 48 and m/z 65 is subject to a spectral
interference by one of the four matrices, except 54Fe.
Percent recoveries of 10-/ig/L spikes of the simple synthetic matrices for pre-concentration
sample introduction are compared to .the percent recoveries for flow-injection direct nebulization in
Table 8. AH but two of the recoveries for pre-concentration sampling fall between 85% and 115%.
One of these is for manganese in the calcium solution. Recoveries of manganese with iminodiacetate
resins are known to be adversely affected by very high calcium concentrations . If manganese is to
be determined in such matrices, standards similar in composition to the samples should be prepared,
or the method of standard addition can be employed. The isobaric ^Ca interference on ^i caused
26
-------
TABLE 8. PERCENT RECOVERIES FROM SIMPLE SYNTHETIC MATRICES0
Analyte
-Ti
49Ti
V
Mn
54Fe
57Fe
Co
^Ni
62Ni
63Cu
65Cu
nlCd
208pb
Na
PC* FIAC
106
105
104
102
103
103
101
102
104
103
103
102
100
104
103
130
86
23
74
85
77
103
78
69
69
70
Matrix Element
K Me C
PC
103
104
104
101
107
102
100
99
102
94
94
101
92
FIA
105
110
103
88
20
85
92
81
76
64
70
74
65
PC
98
100
110
88
95
99
97
95
100
99
97
96
89
FIA
101
106
104
89
28
82
84
68
68
72
73
73
74
PC
-
96
98
81
99
96
93
90
90
90
89
93
98
a
FIA
189
104
107
94
38
*
80
*
82
75
73
79
79
a 10-/ig/L spike of each analyte (100-Mg/L spike of iron).
b Pre-concentration ICP-MS.
c Flow-injection direct nebulization ICP-MS.
d Percent recovery £ 0, caused by large background intensity.
a negative recovery, due to slight variability in the efficiency of the ammonium acetate wash-out
between the spiked and non-spiked samples. Of the directly nebulized samples, more than three
fourths of the recoveries fall outside a 85% to 115% acceptance window. It was impossible to
distinguish any analyte signals on top of the very intense CaOH+ and CaO+ interferences.
SYNTHETIC SEA WATER
The results of the pre-concentration ICP-MS analysis of non-spiked synthetic sea water are
compared with the results from flow-injection direct nebulization in Table 9. The only spectral
interferences remaining after pre-concentration and a 6-mL wash-out with ammonium acetate are
27
-------
TABLE 9. APPARENT ANALYTE CONCENTRATIONS IN NON-SPIKED SYNTHETIC
SEA-WATER MATRIX BY PRE-CONCENTRATION AND BY FLOW-INJECTION
DIRECT NEBULIZATION"
Analyte Pre-concentration Flow-Injection
T,
4*Ti
V
Mn
54Fe
57Fe
Co
^i
62Ni
63Cu
65Cu
a Concentrations in Mg/L.
39
ND
ND
ND
ND
130
ND
ND
ND
ND
ND
1120
120
ND
ND
ND
3330
1.3
19
500
52
15
relatively small residual 4SCa+ and CaOH+ peaks. A comparison of the percent recoveries in the
spiked sea-water sample by the two methods is shown in Table 10. Again, recoveries for the pre-
concentration technique are far superior to those for flow-injection direct nebulization, even though
the former technique only used internal standardization for lead, while the latter required the use
of internal standardization for every analyte.
WASTE WATER RECOVERIES
A major drawback to pre-concentration methodologies is the low recoveries obtained in the
presence of strong chelating agents, such as certain naturally occuring compounds. In natural waters
and waste waters, colloids often sequester analyte, and must be destroyed by appropriate treatment
prior to pre-concentration17. Table 11 presents the percent recoveries for the pre-concentration of
both the raw and the microwave-digested waste-water samples. It can be seen that, especially in the
case of the primary effluent, titanium, copper, cadmium and lead show very poor recoveries in the
raw material. The simple microwave digestion procedure, however, is sufficient to liberate the
bound metals and restore full recoveries.
28
-------
TABLE 10. PERCENT RECOVERIES FROM SPIKED SYNTHETIC SEA WATER BY
PRE-CONCENTRATION AND BY FLOW-INJECTION DIRECT NEBULIZATION"
Analyte Pre-concentration Flow-Injection
«r.
49Ti
V
Mn
54Fe
57Fe
Co
"Ni
62Ni
63Cu
65Cu
niCd
208pb
110
98
108
85
109
118
98
99
97
95
95
94
97
*»
75
117
68
61
*
54
41
*
47
35
48
72
" 10-/ig/L spike of each analyte (100-/ig/L spike of iron).
* Percent recovery < 0, caused by large background intensity.
The percent recoveries in the spiked waste water and waste-water digests by flow-injection
direct nebulization are also shown in Table 11. As with synthetic samples, recoveries for the pre-
concentration of waste-water digests are superior to those for flow-injection direct nebulization.
The ICP-MS analyses of non-spiked waste-water digests by pre-concentration and by flow-injection
direct nebulization are compared in Table 12. The spectral interferences encountered in the direct
nebulization of the waste-waters appear .to consist largely of sodium and calcium interferences.
These are greatly dissipated by pre-concentration with a 6-mL ammonium acetate wash-out. As
seen in the analysis of synthetic sea water (Table 9), small interferences remain at m/z 48 and 57
after pre-concentration.
29
-------
TABLE 11. PERCENT RECOVERIES OF TRACE METALS FROM SPIKED WASTE WATERS AND WASTE-WATER DIGESTS BY
PRE-CONCENTRATION AND BY FLOW-INJECTION DIRECT NEBULIZATION0
Las Vegas Wash
PC*
Analyte
48Ti
49Ti
V
Mn
*Fe
57Fe
Co
»Ni
62Ni
«Cu
65Cu
niCA
208pb
Raw
106
86
104
97
90
150
99
100
103
78
78
95
83
Digest
95
93
99
94
90
94
97
97
101
97
98
100
95
Raw
68
117
105
88
32
516
91
84
63
77
80
88
86
FIAC
Digest
70
81
96
88
9
59
90
87
79
87
87
102
102
Secondary Effluent
PC
Raw
100
52
110
100
81
139
101
101
94
54
57
94
67
Digest
92
100
102
%
106
97
99
99
101
106
105
96
92
FIA
Raw
2
74
105
90
17
30
91
90
80
77
73
91
84
Digest
78
91
100
91
28
88
91
88
84
89
86
102
101
Primary
PC
Raw
0
27
110
97
97
97
98
103
99
37
38
66
28
Digest
95
96
101
94
90
95
97
95
100
99
99
%
90
Effluent
FIA
Raw
18
104
105
88
19
341
92
99
98
80
81
94
87
Digest
81
89
98
92
13
73
90
88
79
86
86
100
98
" 10-ng/L spike of each analyte (100-/^g/L spike of iron).
* Pre-concentration ICP-MS
c Flow-injection direct nebulization ICP-MS
-------
TABLE 12. APPARENT ANALYTE CONCENTRATIONS IN NON-SPIKED WASTE
WATER DIGESTS BY PRE-CONCENTRATION AND BY FLOW-INJECTION0
Las Vegas Wash Secondary Effluent
Analyte PC* FIAC PC FIA
^i 68
49Ti ND
V 3
Mn 48
54Fe ND
57Fe 304
Co 0.4
"Ni 7
62Ni 7
63Cu 3
65Cu 3
1HCd 0.2
^Pb ND
" Concentrations in M8/L
454
23
ND
47
ND
2555
1
22
30
7
11
ND
ND
79
0.18
2
10
64
374
0.3
5
4
8
8
ND
ND
507
22
ND
11
ND
2699
2
21
20
24
29
ND
ND
Primary Effluent
PC FIA
82
0.13
2
8
143
372
0.4
5
4
ND
ND
ND
ND
494
20
ND
9
ND
2777
2
20
22
11
14
ND
ND
b Pre -concentration ICP-MS
c Flow-injection direct nebulization ICP-MS.
31
-------
REFERENCES
(1) Henshaw, J. M.; Heithmar, E. M.; Hinners, T. A. Anal. Chem. 1989, 61, 335-342.
(2) McLaren, J. W.; Beauchemin, D.; Herman, S. S. Anal. Chem. 1987, 59, 610-613.
(3) Lyon, T. D. B.; Fell, G. S.; Hutton, R. C.; Eaton, A. N. J. Anal. At. Spectrom. 1988, 3, 265-
271.
(4) Beauchemin, D.; McLaren, J. W.; Berman, S. S. Spectrochim. Acta, Part B 1987, 42B, 467-
490.
(5) Douglas, D. J.; Kerr, L. A. /. Anal. At. Spectrom. 1988, 3, 749-752.
(6) Hutton, R. C.; Eaton, A. N. J. Anal. At. Spectrom. 1988, 3, 547-550.
(7) Van Loon, J. C. Selected Methods of Trace Metal Analysis; Wiley-Interscience: New York,
1985; pp. 104-106.
(8) Xiao-quan, S.; Tie, J.; Xie, G. /. Anal. At. Spectrom. 1988, 3, 259-263.
(9) Chung, Y. S.; Barnes, R. M. /. Anal. At. Spectrom. 1988, 3, 1079-1082.
(10) Brajter, K; Dabek-Zlotorzynska, E. Analyst 1988, 113, 1571-1574.
(11) Olsen, S.; Pessenda, L. C. R.; Rfizicka, J.; Hansen, E. H. Analyst 1983, 108, 905-917.
(12) Hirata, S.; Yoshima, U.; Masahiko, I. Anal. Chem. 1986, 58, 2602-2606.
(13) Hirata, S.; Honda, K.; Kumamaru, T. Anal. Chim. Acta 1989, 227, 65-76.
(14) Furuta, N.; Brushwyler, K. R.; Hieftje, G. M. Spectrochim. Acta, Part B 1989, 44B, 349-
358.
(15) Karakaya, A.; Taylor, A. J. Anal. At. Spectrom. 1989, 4, 261-263.
(16) Plantz, M. R.; Fritz, J. S.; Smith, F. G.; Houk, R. S. Anal. Chem. 1989, 61, 149-153.
(17) Kingston, H. M.; Barnes, I. L.; Brady, T. J.; Rains, T. C.; Champ, M. A. Anal. Chem. 1978,
50, 2064-2070.
(18) Greenberg, R. R.; Kingston, H. M. Anal. Chem. 1983, 55, 1160-1165.
(19) McLaren, J. W.; Mykytiuk, A. P.; Willie, S. N.; Berman, S. S. Anal. Chem.l9S5, 57, 2907-
2911.
(20) Beauchemin, D.; McLaren, J. W.; Mykytiuk, A. P.; Berman, S. S. Anal. Chem. 1987, 59, 778-
783.
(21) Werefridus, W. V. B.; Overbosch, A. W.; Feenstra, G.; Maessen, F. J. M. J. /. Anal. At.
Spectrom. 1988, J, 249-257.
32
-------
APPENDIX A
On-Line Pre-Concentration of Trace Metals Prior to Determination bv Inductively Coupled Plasma-
Atomic Emission Spectrometrv or Inductively Coupled Plasma-Mass Spectrometrv
NOTICE
This document is a preliminary draft. It has not been formally released by the U.S. Environmental
Protection Agency and should not at this stage be construed to represent Agency policy. It is being
circulated for comments on its technical merit and clarity. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
1.0 SCOPE AND APPLICATION
1.1 This method improves the sensitivity of inductively coupled plasma-atomic emission
spectroscopy (ICP-AES) and inductively coupled plasma-mass spectrometry (ICP-MS) in the
determination of many trace metals and reduces matrix interferences from anionic species and from
alkali and alkaline earth metals. See Method 6010 (ICP-AES) or 6020 (ICP-MS) for information
about the specific determinative method.
1.2 Elements for which this method is applicable are listed in Table 1-A. Instrument
detection limits, sensitivities, and linear ranges for these elements will vary with the instrumentation
and operating conditions. Restrictions on the use of this method are as stated in the pertinent
determinative method.
1.3 This method is limited to the pre-concentration of dissolved metals from aqueous
solutions. Samples must be digested by a procedure appropriate to the determinative method if total
metal concentration determinations are needed, or if the sample is not an aqueous solution.
2.0 SUMMARY OF METHOD
2.1 This method describes a technique of sample introduction for ICP-AES and ICP-MS in
which samples are pre-concentrated by passing them through a column containing a macroporous
33
-------
TABLE 1-A. ESTIMATED DETECTION LIMITS (Mg/L) OF ANALYTES OF INTEREST
Element ICP-AES ICP-MS
Cadmium
Cobalt
Copper
Iron
Lead
Manganese
Nickel
Titanium
Vanadium
To be determined.
1.0
*
*
*
1.0
*
*
*
*
0.04
0.0004
0.004
0.3
0.005
0.006
0.01
0.02
0.006
iminodiacetate chelating resin. This resin has a high affinity for many metals existing as polyvalent
cations in solution. Matrix modification is achieved by the selective elution of group IA and IIA
metals, as well as anions, for which the resin has a relatively low affinity, prior to the elution of
analytes to the ICP.
3.0 INTERFERENCES
.3.1 The extent of spectral interference from alkali and alkaline earth metals depends largely
on the timing of the procedure. Sufficient time must be allowed for the selective elution of
interfering matrix ions.
3.2 Organic chelators native to the sample may interfere with the resin and attenuate
recovery. The digestion procedure should be adequate to destroy these colloids.
3.3 Interference due to memory must be kept at a minimum by the use of an efficient resin
rinse procedure and by running blanks after highly concentrated samples.
4.0 APPARATUS AND MATERIALS
4.1 The ore-concentration apparatus should be plumbed as shown in Figure 1-A. There
must be no metal in the flow path. Required components are:
34
-------
2 M NH4 Ac
1 M NHO3
Sample
Buffer
Rinse
Blank
Internal _
Standard
1-3
ml/
min.
V1
Waste
Sample
Loop
V1
\ •
\
4 mL/min.
1 mL/min.
Preconcentration
Column
Waste
1 mL/min.
C V2
\
0.1 mL/min.
ICP
Figure 1-A. Pre-concentration apparatus. Valves are shown in the 'ON' position.
VI = valve 1, V2 = valve 2.
35
-------
4.1.1 Pre-concentration Column: Macroporous iminodiacetate chelating resin column
(Dionex MetPac CC-1 or equivalent).
4.1.2 Eluant delivery system (PI):
4.1.2.1 Constant-flow programmable pumping system capable of delivering
eluant from either of two reservoirs. The pump should be capable of operating in the
range 0-1000 psi at 0-3 mL/min.
4.1.2.2 Use of pressurized eluant reservoirs to provide a positive head on the
eluants is suggested to aid in pumping.
4.1.3 Auxiliary pump (P2): A peristaltic pump with line splitter/mixing block is
recommended.
4.1.3.1 One channel of the pump (P2a) will be used to pump the sample to the
sample loop.
4.1.3.2 A second channel (P2b) will be used, with a mixing block, to pump a
buffer solution to adjust the sample pH on-line prior to injection.
4.1.3.3 A third channel (P2c) will be used to provide continuous flow of rinse
blank to the ICP.
4.1.3.4 A fourth channel (P2d) can be used, with a line splitter, for on-line
internal standardization of the effluent from the column.
4.1.4 Sample Loop: A length of tubing sufficient to provide the desired sample
volume.
4.1.5 Valves: Two inert, pneumatically operated four-way valves.
4.1.5.1 Source of regulated gas (100 psi).
4.1.6 Tubing: Teflon or polyethylene.
4.2 The detection system is to be connected to the pre-concentration apparatus by a
minimum length of tubing from the output of valve 2 (V2) to the spectrometer. The detection
36
-------
system will consist of either
4.2. 1 An inductively coupled plasma-atomic emission spectrometer (see Method 60 1 0,
Section 4.0), or
4.2.2 An inductively coupled plasma-mass spectrometer (see Method 6020, Section
5.0).
5.0 REAGENTS
5.1 In the determination of trace elements, containers can introduce either positive errors
by contributing contaminants through leaching or surface desorption, or negative errors by depleting
concentrations through adsorption. Given the high sensitivity of this technique, it is recommended
that all vessels used for containment of samples and solutions be constructed of linear polyethylene
or Teflon. The use of glass is not recommended. The following cleaning sequence is recommended
to minimize contamination of these vessels: detergent, ASTM Type II water, 1+1 hydrochloric acid,
ASTM Type I water, 1+1 nitric acid, and ASTM Type I water. All samples and reagents should be
stored in plasticware cleaned in this manner.
5.2 All reagents used in the preparation of samples, standards, and eluants should be of the
highest purity. They should be purchased or prepared from ultra-high purity grade chemicals or
metals.
(CAUTION: Many metal salts are extremely toxic if inhaled or swallowed. Wash hands
thoroughly after handling.)
5.3 ASTM Type I water is required unless otherwise specified.
5.4 Acids and Bases:
5.4.1 Oxalic acid (0.2 M), C2H2O4: Dissolve 25 g 99% Oxalic acid dihydrate
HO) in a minimum amount of water. Dilute to 1000 mL.
5.4.2 Nitric acid (1.0 M), HNOj: Dilute 97 g 65% (concentrated) HNO3 to 1000 mL
with water. Prepare directly in eluant container.
5.4.3 Nitric Acid (1%), HNOj: Dilute 10 mL 65% (concentrated) HNO3 to 1000 mL
37
-------
with water.
5.4.4 Acetic acid (4.0 M), C2H4O2: Dilute 243 g 99% (glacial) acetic acid to 1000 mL
with water.
5.4.5 Ammonia (4.0 M), NH3: Dilute 272 g 25% (concentrated) NH3 to 1000 mL with
water.
5.5 Ammonium Acetate Buffer can be prepared as follows:
5.5.1 Eluant (approximately 2 M): In a fume hood, combine equal quantities of 4.0 M
acetic acid and 4.0 M ammonia directly in the eluant container. Adjust the pH to 5.5 with
either the acetic acid or the ammonia as necessary.
5.5.2 On-line buffer (approximately 2 M): In a fume hood, combine equal quantities
of 4.0 M acetic acid and 4.0 M ammonia. Adjust the pH with either the acetic acid or the
ammonia as necessary so that, when mixed with 1% nitric acid in a 1:5 ratio, the pH is 5.5.
NOTE: Do not introduce foreign objects (magnetic stirring bars, pH electrodes, pH paper,
etc.) into any solutions to be used in the course of this procedure.
5.6 Three types of blanks are to be used in the analysis. The calibration blank is to be used
in establishing the calibration curve, the reagent blank is used to monitor for possible contamination
resulting from the sample preparation procedure, and the rinse blank is used to flush out the system
between all samples and standards.
5.6.1 The calibration blank should be prepared as instructed in the specific determinant
method.
5.6.2 The reagent blank should be prepared as instructed in the pertinent method.
5.6.3 The rinse blank consists of 2% nitric acid in water. Prepare a sufficient quantity
to flush the sample pump (P2a,b) and sample loop between standards and samples, and to be
pumped continuously on the auxiliary pump (P2c).
5.7 Internal standards, if used, should be prepared as directed in the specific determinative
method.
38
-------
5.8 Any other solution called for by the determinative method should be prepared as
specified in that method.
6.0 SAMPLE COLLECTION, PRESERVATION, AND HANDLING
6.1 Refer to the determinative method.
7.0 PROCEDURE
7.1 Trace metal contamination must be removed from the system prior to use. The following
procedure (summarized in Table 2-A) describes the decontamination of the entire system.
TABLE 2-A. PRE-CONCENTRATION SYSTEM DECONTAMINATION PROCEDURE
Time
(min.)
0
5
10
20
30
35
40
50
65
80
85
a 1 = ON,
VI"
1
0
0
0
0
0
0
0
1
1
Stop
0 = OFF
V2a
1
1
1
1
1
1
1
0
0
0
Pump
P2a,b
P2a,b
El
E2
El
E2
E2
El
El
E2
Eluant
0.2-M oxalic acid
1 % nitric acid
0.2-M oxalic acid
0.2-M oxalic acid
Water
Water
1-M nitric acid
2-M ammonium acetate
Water
Water
Flow
(mL/min)
5.0
5.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
7.1.1 Disconnect V2 from the spectrometer, and configure the system so that VI and
V2 are on. Pump 0.2-M oxalic acid through the sample pump (P2a,b) and sample loop to
waste at 5.0 mL/min. Switch VI off after 5 minutes and rinse with 1% nitric acid at 5.0
mL/min for 5 minutes.
39
-------
7.1.2 Pump 0.2-M oxalic acid from eluant line El through the pre-concentration
column for 10 minutes at 2.0 mL/min. Repeat this procedure from eluant line E2.
7.1.3 Pump water from each eluant line through the pre-concentration column at 2.0
mL/min for 5 minutes each.
7.1.4 Pump 1-M nitric acid through eluant line E2 and the column for 10 minutes at
2.0 mL/min.
NOTE: The pre-concentration column should not be left in the hydronium form for
extended periods of time.
7.1.5 Switch V2 off and reconnect to the spectrometer. Connect El to the 2-M
ammonium acetate reservoir and pump the buffer through the column for 15 minutes at 2.0
mL/min.
7.1.6 Switch VI on and disconnect the eluant lines from VI. Pump water through each
eluant line to waste at 2.0 mL/min for 5 minutes. Connect El to the 1-M nitric acid reservoir
and E2 to the 2-M ammonium acetate reservoir.
NOTE: Although the decontamination procedure described in Section 7.1 need only be
performed in its entirety before the first use of the apparatus and after extended periods of disuse,
equivalent measures must also be taken to decontaminate replacement components.
7.2 Sample solubilization and digestion procedures are presented in the determinative method.
7.3 Set up the spectrometer, as instructed in the specific determinative method, on channel
one of the auxiliary pump with V2 in the 'off position.
7.4 Proceed with the specific determinative method by running standards, samples, and
blanks on the pre-concentration apparatus according to the following procedure. The timing
presented below is specific to the use of a 10-mL sample loop and will vary with loop volume. The
timing will also vary with the volume of the tubing used in setting up the apparatus. The procedure
is summarized in Table 3-A.
7.4.1 Load sample: With VI on and V2 off, pump analyte solution through sample
loop to waste for 2 minutes at 4.0 mL/min with the on-line buffer at 1.0 mL/min. This can
be done concurrently with step 7.4.5.
40
-------
TABLE 3-A. PRE-CONCENTRATION SAMPLE RUN PROGRAM"
Time
(min)
12.8
0.0
4.0
6.0
6.8
8.8
9.8
10.8
11.8
12.8
13.8
VI*
1
0
1
1
1
1
1
1
1
1
Repeat
V2b
0
0
0
0
1
0
0
0
0
0
as necessary
Pump
P2a,b
El
El
E2
E2
El
E2
El
E2
El
(see Section
Solution
Sample + buffer
2-M ammonium acetate
2-M ammonium acetate
1-M nitric acid
1-M nitric acid
2-M ammonium acetate
1-M nitric acid
2-M ammonium acetate
1-M nitric acid
2-M ammonium acetate
7.4.6).
Flow
(mL/min)
3.0
3.0
3.0
1.0
3.0
3.0
3.0
3.0
3.0
a Vary as necessary (see Section 7.4).
b 0=OFF, 1=ON
7.4.2 Inject sample: Switch VI off. Pump 2-M ammonium acetate for 4.0 minutes at
3.0 mL/min.
7.4.3 Elute matrix elements: Switch VI on. Continue pumping 2-M ammonium
acetate for 2.0 minutes. At the same time, rinse the sample pump and sample loop with rinse
blank for at least 2 minutes.
7.4.4 Elute analytes: Switch eluant to 1-M HNO3 and pump at 3.0 mL/min for 0.7
minutes. Decrease flow to 1.0 mL/min and switch V2 on. Continue in this mode for 2.0
minutes.
NOTE: Contamination by carryover can occur if the sample elution time is inadequate.
Whenever an unusually concentrated sample is encountered, it should be followed by the
analysis of a blank to check for cross contamination.
7.4.5 Recondition column: Switch V2 off. Increase flow to 3.0 mL/min. Cycle the
41
-------
eluants through the column in 1.0 minute increments in the following order 2-M ammonium
acetate, 1-M nitric acid, 2-M ammonium acetate, 1-M nitric acid, and finally 2-M
ammonium acetate.
7.4.6 Repeat steps 7.4.1 through 7.4.5 for each solution to be analyzed.
NOTE: Before running samples, or if the analysis is suspended for an extended period of
time, flush the lines and condition the column by analyzing a sufficient number of blanks to achieve
an acceptable background.
7.5 Shut down pre-concentration apparatus.
7.5.1 Switch VI on and V2 off. Pump 2-M ammonium acetate through the column to
waste for 5 minutes at 2.0 mL/min.
7.5.2 Rinse eluant lines with water.
8.0 QUALITY CONTROL
8.1 Refer to the specific determinative method.
9.0 METHOD PERFORMANCE
9.1 In a single-laboratory evaluation, the calibration functions were found to be linear for
concentrations in the range from 0.3 M8/L to 30 Mg/L for a 2.5-mL sample loop, and from 0.3 Mg/L
to 10 M8/L for a 10-mL sample loop, with correlation coefficients ranging from 0.998 to 0.9999.
The relative standard deviation of a 10-/ng/L standard in a 2.5-mL sample loop is 2.0%, while that
for a 0.5-/ig/L standard is 4.7%.
9.2 The method was tested with high alkali and alkaline earth metal solutions, synthetic sea
water, and waste-waters spiked at 10 Mg/L. Recoveries are generally within a 90% to 110% window
except for some elements in high calcium matrices and in undigested natural samples.
9.3 Method performance must be validated in a multi-laboratory evaluation by the EPA.
42
-------
10.0 REFERENCES
1. Kingston, H. M.; Barnes, I. L.; Brady, T. J.; Rains, T. C.; Champ, M. A. Anal. Chem. 1978,
50, 2064-2070.
2. Greenberg, R. R.; Kingston, H. M. Anal. Chem. 1983, 55, 1160-1165.
3. Chung, Y. S.; Barnes, R. M. J. Anal. At. Spectrom. 1988, 3, 1079-1082.
43
-------
ON-LINE PRE-CONCENTRATION OF TRACE METALS PRIOR TO DETERMINATION BY
INDUCTIVELY COUPLED PLASMA-ATOMIC EMISSION SPECTROMETRY OR INDUCTIVELY
COUPLED PLASMA-MASS SPECTROMETRY
Start
7.1
Decontaminate
pre-concentration apparatus
7.2
Prepare samples according
to determinative method
7.3
Set up ICP spectrometer as directed
by specific determinative method
7.4
Proceed with specific determinative
method (Method 6010 or 6020) using
pre-concentration analysis procedure
7.5
Shut down pre-concentration apparatus
Stop
44
------- |