Anal. Chem. 1980, 52, 2168-2173
2168
(11) Klimish, H. J.; Reese. D. J . Chromatogr., 1972, 6 7 , 299. (12) Karger. B. L.; Snyder, L. R.; Eon, C. J . Chromatogr. 1976, 125, 71. (13)Karger, B. L.; Snyder, L. R.; Horvarth, C. "An Introduction to Separation Science"; Wiley-Interscience: New York, 1973; p 50. (14) Gordon, A. J.; Ford, R. A. "The Chemist's Companion"; Wiley-Inter-
science: New York, 1972; p 432. Hansler, D. W.; Hellgeth, J. W.; McNair, H.
M.; Taylor, L. T. J. Chromatogr. Sci. 1979, 17, 617. (16) Snyder. L. R. In "Modern Practice of Liquid Chromatography"; Kirkland, J. J., Ed.; Why-Interscience: New York, 1971; p 138.
(15)
(17) Karger, B. L.; Snyder, L. R.; Eon, C. Anal. Chem. 1978, 50,2126.
RECEIVEDfor review May 7,1980. Accepted August 11,1980. This research was supported by U.S. Department of Energy, Contract No. W-7405-Eng-82, Office of Health and EnvironResearch* Pouutant Characterization and Safety Research Division (GK-01-02-04-3).
Computer-Controlled Scanning Monochromator for the Determination of 50 Elements in Geochemical and Environmental Samples by Inductively Coupled Plasma-Atomic Emission Spectrometry M. A. Floyd, V. A. Fassei,' and A. P. D'Silva Ames Laboratory and Department of Chemistty, Iowa State University, Ames, Iowa 500 1 1
The application of a computer-controlled, scanning monochromator to the determination of 50 elements in geachemlcal and environmental matrices is described. The monochromator Is combined wlth an Inductively coupled plasma excitation source so that elements at major, minor, and trace levels may be determined in sequence without changlng experimental parameters other than the spectral line observed. A single set of spectral lines was found to be applicable to a broad range of sample compositions and to show negllglble spectral Interferences regardless of the sample matrlx.
In years past, elemental determinations at the major, minor, a n d sometimes at the trace level in ores, minerals, soils, manufactured products, e.g., glass and pigments, and sediments and ashes have been performed by classical gravimetric, titrimetric, and spectrophotometric techniques. During the past several decades, the classical approaches have been largely replaced by a variety of instrumental techniques, which include X-ray fluorescence, atomic absorption spectrometry, arc emission spectroscopy, neutron activation analysis, and proton-induced X-ray emission analysis. I n recent reviews ( I - @ , several authors have adequately documented that inductively coupled plasma-atomic emission spectroscopy (ICP-AES) offered several advantages as an alternative approach for the analysis of the sample types discussed above. I n view of these advantages, i t is not surprising that ICP-AES has been applied, to a rapidly increasing extent, to the analysis of the sample types discussed above (7-21). Most of the applications discussed to date have involved t h e use of a polychromator for the multielement determination of selected elements of interest in matrices of closely similar composition. When the analyst is faced with the determination of a broader range of elements a t various concentration levels in samples of widely varying composition, the fixed array of exit slits used for isolating the spectral lines in a polychromator becomes restrictive. We have recently described (22) a computer-controlled, scanning monochromator system that provides the capability for the rapid sequential determination of the elements a t the major to ultratrace levels without changing the experimental 0003-2700/80/0352-2168$01.OO/O
conditions, other than the elemental lines that are observed. In this communication, we describe the application of this instrument to the rapid determination of major, minor, and trace constituents in sample types as diverse as ores and minerals, coal and fly ash, urban particulates, and sediments. A single sample dissolution procedure is used, a n d a single set of analytical calibrations apply to the broad range of sample compositions examined. EXPERIMENTAL SECTION Instrumentation. The programmable scanning monochomator utilized in this investigation and the operating conditions for the ICP have been previously described (22). Sample Dissolution. A 0.20-g sample was accurately weighed into a 15-mL graphite crucible (Vitrecarb, Anaheim, CA) after which 2 g of NaOH pellets was added to the crucible. The crucible was gently heated with a Fisher blast burner to melt the contents, care being taken to assure that no spattering occurred. After the moisture was evaporated from the NaOH, which usually required 2 to 3 min, the sample was brought to red heat until a clear melt was obtained. The red hot fused melt was carefully and quickly poured into a 150-mL platinum dish. The upright crucible was also set into the dish. After the sample cooled, 5.0 mL of concentrated HCl was added to both the fused sample and the graphite crucible. Heat was gently applied on a hot plate to dissolve the melt, including the portions adhering to the crucible. The crucible was then thoroughly washed with deionized water and the contents were added to the platinum dish. The sample was then diluted to approximately 40 mL with deionized water and heat applied gently until the solution was clear. Samples containing high silicon content were filtered or centrifuged to remove flocculent hydrated silicon oxide. The sample solution was then transferred to a 100-mL polypropylene volumetric flask and diluted to volume with deionized water. The time required for dissolution of a sample was approximately 30 min. Prior to sample fusion, the graphite crucibles were processed by the fusion-dissolution procedure described above to minimize contamination. Crucibles preprocessed in such a manner were reusable for several fusions. During the analysis of samples, a blank solution prepared from the NaOH pellets and HC1 utilized in the fusion-dissolution procedure was analyzed, and the appropriate blank corrections were stored in memory and were automatically subtracted, if significant,from the determined total concentrations. Reference Solutions. Stock solutions were prepared by dissolution of pure metals or reagent grade salts in dilute (1% ) 0 1980 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 13, NOVEMBER 1980
2169
~~
Table I. Detection Limits
A1 As Ba Be Bi Ca Cd Ce
cr co cu
DY
Er Eu Fe Ga
Gd
Ge Hf Ho In Ir La Lu Mg
Mn Mo Nb
wavelength, nm
detection limits, rg/g
308.2 197.2 455.4 313.0 223.1 315.9 214.4 413.8 267.7 228.6 324.8 353.2 337.3 420.5 238.2 294.4 342.3 265.1 217.3 389.1 230.6 224.3 394.9 261.5 279.6 257.6 202.0 316.3
1 5 0.25 0.03 2.5 0.05 0.5 2 0.5 1 0.2 0.5 0.5 0.25 0.25 3 2 10 1 0.5 10 0.3 1 0.1 0.02 0.05 3 0.5
wave- detection limits, length, nm rglg Nd Ni Pb
Pd Pr Re Rh Ru Sb sc
Se Sm Sn Sr
Ta Tb
Te Th Ti T1 Tm U V W Y Yb
Zn Zr
430.4 231.6 220.4 340.5 390.8 221.4 233.5 240.3 217.6 361.4 196.0 359.3 235.5 407.8 226.2 350.9 214.3 283.7 334.9 351.9 346.2 386.0 310.2 207.9 371.0 328.9 213.9 343.8
1.5 1 10 0.5 0.1 2 5 4 15 0.05 35 6 5 0.01 8 0.5 2 5 0.2 4 2 6 0.1 5 0.04 0.05 0.1 0.5
acid or deionized water. Reference solutions were prepared by the incremental addition of the stock solutions to an appropriate blank solution. Reference solutions were prepared and stored in class A volumetric glassware which had been leached for 72 h with a 10% HCl/deionized water solution. A complete set of reference solutions was prepared that provided multielement calibrations throughout specified analytical ranges. Each reference solution contained the equivalent of 2 g of NaOH (same as the dissolved samples) and was confined to only one concentration level of the analytes so that any cross contamination due to impurities in the stock solutions was insignificant. With this approach, cross contamination from impurities in the primary stock solution was insignificant. Selection of Analysis Lines. The analyte lines were selected on the basis of their net and background intensity as well as their freedom from spectral interferences. The line selection process was facilitated by the list of prominent lines recently published by this laboratory and by the associated atlas of wavelength scans, which is now being prepared for publication (23). The wavelengths of the single set of selected analyte lines used throughout the course of our studies are shown in Table I. R E S U L T S AND DISCUSSION Spectral Line Interferences. In general, environmental and geochemical samples provide diverse matrices that are exceptionally complex, both chemically and physically (24). T h e complexity of these samples raises the possibility that interelement effects of various types or spectral interferences may introduce errors in the analytical determinations. Observations on several classical solute vaporization interference systems (25, 26) have shown t h a t most interelement interference effects are reduced to negligible proportions in the ICP. However, a t analyte concentrations in the microgram per milliliter to nanogram per milliliter concentration range, spectral line interferences may arise from high concentrations of concomitants. In these cases the interfering emission usually lies within 0.04 nm of the analyte wavelength. The potentiality of these interferences was investigated by obtaining profiles of emission intensity vs. wavelength for the four strongest emission line for each element of interest (27). Each profile covered a small wavelength region (0.3 nm) that
IlI
1
$'
1
0 -
et220.10
22C.30
22c. 0 WAVELENGTH
223.3C
NN,
Figure 1. Wavelength profile of the Pb 220.35-nm spectral line demonstrating shifts in the spectral background caused by the presence of 1000 Kg/mL AI: 0, 10 pg/rnL Pb; A, 1000 pg/mL AI.
encompassed one of the analytical lines of interest. Additionally, the spectra of single element solutions containing lo00 wg/mL of Al, Ca, Fe, and Mg and 200 wg/mL of Cr, Cu, Mn, Ni, Ti, and V were profiled over the same wavelength region. All profiles were obtained by stepping the monochromator across the wavelength region and taking intensity measurements every 0.01 nm. By superimposing these 11profiles onto one graph, we could determine potential spectral line interferences arising from these elements. In terms of interferences the collected intensity data ranged from the one extreme, where an analyte line had no observable interference, to the other extreme, where the analyte line was lost in the background under the interfering line. We circumvented these limitations and maximized the amount of information obtainable by plotting the profile data on both linear and logarithmic intensity axes. The linear scaled plots were best suited for the profiles in which the analyte and interfering line peaks were of approximately the same magnitude. For the linear plots the plotting routine was set up so that the peak intensity of the analyte line fell between half and full scale, thus ensuring that the analyte peak could always be seen. Other peaks, that were significantly more intense than t h e analyte peak, were therefore truncated a t the full-scale reading of the intenstiy axis. The linear plots allowed the degree of interference to be estimated easily by a simple comparison of interference intensity (at the center of the analyte peak) with the peak analyte intensity. T h e log-scaled plots were best suited for the profiles in which the analyte and interference peaks were of significantly different magnitude. For example, sometimes interfering spectral line peaks of relatively low intensity were observable on the log plots that were not observable on the linear plots. For the log plots, the range of the intensity axis was automatically selected so that all analyte and interference intensities were shown. Several examples of typical graphs observed during the course of our studies are shown in Figures 1 and 2. These profiles illustrate several problems that may arise when complex samples are analyzed. In Figure 1, 1000 pg/mL of Al, which would be equivalent to an A1 content in a sample of 50%, increases the intensity of the spectral background significantly from that when no A1 is present in t h e sample. Without a n accurate background correction scheme, the measurement of the emission intensity a t the analytical wavelength would result in an analytical bias. Figure 2A shows an example of a direct spectral overlap among three analytical wavelengths. Both the A1 309.27-nm and Mg 309.30-nm spectral lines interfere with the V 309.31-nm wavelength. Because the interferences due to 1000 pg/mL of A1 and Mg were equivalent to 100 and 8 pg/mL, respectively, of V, this line was not chosen for analytical purposes. Figure 2B shows
2170
ANALYTICAL CHEMISTRY, VOL. 52, NO. 13, NOVEMBER 1980
Table 11. Comparison of Observed and Calculated Maximum Intensities wavelength, nm Mn 257.6
Y
371.0
Ba 455.4
nominal concn
concn, pg/mL
LQD 5 X LQD 10 X LQD 1 O O x LQD LQD 5 X LQD l o x LQD 1 O O X LQD LQD 5 X LQD l o x LQD 1 O O X LQD
0.0005 0.0020 0.0040 0.0625 0.0005 0.0040 0.0080 0.0625 0.0020 0.0080 0.0160 0.2500
1
a tracing of V 310.23 nm, which was chosen for analytical purposes. The absence of any significant spectral interference from the elements studied makes this analytical wavelength a good choice for analytical work. The small hump underneath the analytical line is part of an OH band. Because this OH band is also present in the blank spectra, the software will automatically make corrections for its presence. In the current investigation a judicious effort was expended in selecting interference-free spectral lines, occasionally to the detriment of the detection limit. When a polychromator is utilized for the analysis of complex materials, the option of choosing appropriate analytical lines for trace analysis is usually not available. Detection Limits. When the above procedure for dissolution of the samples is followed, all constituents which occur in these samples are diluted 500-fold. Thus, the quantitative determination of trace elements in these samples requires the best available in powers of detection. The overall performance of the ICP with reference t o this figure of merit has been adequately documented (22,28). Detection limits for 54 elements were determined under the operating conditions defined in our earlier communication (22) for solutions containing 2.990NaCl (same as the dissolved samples). These “solution” detection limits (micrograms of analyte/milliliters of sample) were then converted to “solid” detection limits (micrograms of analyte/grams of sample) by multiplying each calculated value by a factor of 500. Detection limits obtained with the computer-controlled monochromator are given in Table I. T h e detection limits reported correspond to the concentration of analyte required to give a net signal equal to three times the standard deviation of the background. The average and standard deviation of the background were obtained a t a preselected wavelength adjacent to each analyte line while a multielement reference solution was nebulized. T h e average and standard deviation of the gross signal were obtained a t the analyte wavelength. Q u a n t i t a t i v e D e t e r m i n a t i o n s N e a r Detection Limit Concentrations. When analyte concentrations are determined near the detection limit concentrations, the spectral background will normally comprise a large fraction of the total signal. Because quantitative determinations a t detection limit concentrations are not feasible, it is useful to estimate the lowest quantitatively determinable concentration (LQD). If the reasonable assumption is made that a signal 10-15 times greater than the standard deviation of the background scatter (noise) is desirable for quantitative determination, then a concentration five times greater than the experimentally determined detection limits may be considered as a reliable estimate of the LQD. It is then of interest to inquire how reliably the “maximum intensity” routine can sense t h a t a peak exists a t LQD concentration levels and how accurately the net intensities may be measured under these conditions.
photocurrents obsd from calcd by wavelength profiles “max int” routine 5.34 5.93 6.72 2.98 1.57 2.03 2.80 1.05 1.43 2.63 4.14 5.20
x 10-9
x 10-9 x 10-9
X X
5.21 5.88 6.80 2.96 1.52 2.07 2.76 1.06 1.46 2.59 4.10
x 10-7
5 . 2 5 x 10.’
x 10-9 x 10-9 X
lo-*
X X
x
x 10-7 X lo-’
309.10
x 10-9 X X X X
x X X X
309.30
lo-’ lo-’ lo-@ lo-’
309.10 WAVELENGTH
B
% error
2.43 0.84 1.19 0.68 3.18 1.97 1.79 0.95 2.10 1.52 0.97 0.57
309.30 (NM)
V
310.00 310.10
310.20 310.30
30.00 310.10
WAVELENGTH
310.20 310.30
(NMI
Flgure 2. (A) Wavelength profile demonstrating the effects of 1000 pg/mL of AI and Mg on the emission signal in the region of the V 309.31-nm line and the interferences due to the AI 309.27 nm and the Mg 309.30 nm spectral lines: 0, 1 pg/mL V; A , 1000 pg/mL AI, 0 , 1000 lg/mL Mg, X, 1000 Mg/mL Fe; Z, 200 Mg/mL Ti. (B) Wavelength profile demonstrating the effects of 1000 pg/mL AI and Mg on the emission signal in the region of the V 310.23-nm line and the general absence of interferences from the elements studied: 0,1 g/mL V; A, 1000 pg/mL AI; 0 , 1000 pg/mL Mg; X, 1000 pg/mL Fe; 2,200 pg/mL Ti.
To provide an answer to these questions, we designed and conducted the following experiment. First, the assumption was made that the most reliable measurement of net relative intensities is made by background stripping from a complete wavelength scan profile, obtained in the manner described in our earlier communication (22). Second, the net relative intensities were measured and calculated via the “maximum
ANALYTICAL CHEMISTRY, VOL. 52, NO. 13, NOVEMBER 1980
2171
-
Table IV. Analytical Results of USGS Coal Samples“ PAS 4 4 ALb USGSC Zn Co Ni
I z
455.20
Mn Cr Ga V
455.40
455.20
WAVELENGTH
Be Cu Yb La
455.40
(NMI
Flgure 3. Wavelength profile of the Ba 455.40-nm line. The lowest Ba concentration is approaching a concentration level of five times the detection limit, the least quantitatively determinable concentration (seetext): 0, background; A,LQD concentrationfor Ba (0.0005 wg/mL); +, 5 times LQD concentration; X , 10 times LQD concentration.
Table 111. Analytical Results of NBS SRM No. 1633 (Coal Flyash)a element A1 Ca
Fe Mg Ti As Ba Be Ce
co Cr cu Eu
Ga La Mn Ni
Pb Pr
sc
Sr V Y
Yb Zn Zr
A L ~ 12.8% 4.5% 6.0% 1.8% 7330 56 2800 11 140 41 123 120 2 72 78 510 94 81 24 20 1620 23 3 65 7 21 2 28 8
NBS
61
OndovC
Fisherd
12.7% 4.7% 6.2% 1.8% 7400 58 2700
12.7% 4.3% 6.4% 1.2% 6800
12
38 131 128
493 98 70 1380 214 210
146 42 127 25 77 82 496 98 75 27 1700 235 62 7 216 301
2210 12
Sc
Y Eu
Ba Pr Sr
141 40 85 105 104 68 136 28 61 9
140 36 96 103 99 52 152 27 54 7
