Coaxial cathode ion source for solids mass spectrometry - Analytical

Improvement of ion source performance in glow discharge mass spectrometry. N. Jakubowski , D. Stuewer , G. Toelg. International Journal of Mass Spectr...
3 downloads 0 Views 388KB Size
M. D . Denton, H. S. Glazer, T.Walle,

D. C. Zellner, and F. G. Smit, Ref. 1, p 373. (22) T.Walle, Ref. 1, p 355. (23) E. Kovats, Helv. Chem. Acta, 41, 1915 (1958). (24) E. Kovats, Helv. Chem. Acta, 46, 314 (1964).

(21)

RECEIVEDfor review August 11, 1975. Accepted October 31, 1975. This work was supported in part by a Chemical Sciences Grant from the Italian National Research Council (CNR 73.0098403).

Coaxial Cathode Ion Source for Solids Mass Spectrometry W. A. Mattson,' B. L. Bentz, and W. W. Harrison* Department of Chemistry, University of Virginia, Charlottesville. Va. 22903

A tubular metal anode coupled with a coaxial cathode forms the basis of a gas discharge ion source operated at approximately 1-5 Torr. The sputtering of the cathode from ion bombardment allows analysis of conducting solids or of solution residues dried to a surface film on a high purity graphite or metal cathode. Metal rods or wires may be used as cathodes directly. Sensitivities to ppb levels can be obtained. Multielement precisions and accuracies of several percent (RSD) are achieved in exponential scans using multielement isotope dilution techniques.

The gas discharge, one of the oldest known methods of ionization, has recently been revived as a promising ion source for the mass spectrometric analysis of trace elements in solids (1-6). The simplicity and stability of the discharge make it an interesting alternative to the conventional rf spark source normally used for solids analysis. The sensitivities observed with the gas sources have been in the ppm to ppb range with precisions of 8-25%. Previous gas discharge work in this laboratory has been with the hollow cathode ion source (HCIS) (3, 5, 6 ) , a sensitive means to ionize solids or solution residues for elemental analysis, operated in the dc (3, 5 ) , pulsed dc ( 7 ) , or pulsed rf (6) modes. However, it is often inconvenient to adapt solid samples to the hollow cathode configuration and the evaporation of solutions into the cavity also requires special handling for cathodes which have an ion extraction orifice in the bottom. This study describes the design and operation of a coaxial cathode ion source (CCIS) (8)which maintains sensitivities comparable to those from the HCIS, but which allows convenient analysis of solids and solutions. Multielement isotopic dilution (9, 10) is used with the CCIS to obtain precisions and accuracies of 2-5%.

EXPERIMENTAL Apparatus. The AEI MS-702 mass spectrometer, discharge gas train, and electrical detection readout, including modifications added in our laboratory, were previously described ( 5 , 111. In addition, a modular V/F converter (Teledyne Philbrick, Model 4705) has been added. Because the total ion beam is very stable, the standard AEI ratio system is not used. After amplification, the signal from the electron multiplier is input directly to the V/F converter. The output pulses from the V/F converter (-50 kHz volt-I) are directed to a 4096 channel MCA (EDAX, Model 706),operated in the multiscale mode, for spectral accumulation and identification. A visual readout module (EDAX, Model 871N) which intensifies and integrates operator selected channel bands of the MCA display is used to obtain peak areas for quantitative analysis. A pulse generator (Hewlett-Packard Model 8003A) in combination with a programmable power supply (Kepco Model BHK) all

Present address, Stockton College, Pomona, N.J.

lows either a dc or pulsed dc discharge to be switch selected. The generator and power supply are floated at the 10-kV acceleration potential by means of an isolation transformer (Del Electronics Corp., Mt. Vernon, N.Y.). Alternatively, the rf spark circuit supplied with the MS-702 mass spectrometer can be used to drive the CCIS. CCIS Design. The CCIS is shown in Figure 1. Tantalum is used for the tubular anode (d) and ion extraction slit (0.040 inch). The metal or graphite cathode (e) is attached to a mounting rod (g) by a small set screw. A glass sleeve (f) shields the mounting rod, which is attached to the glass body (c) of the source by a Swagelok connection. The discharge gas (argon) enters (b) the source through a Cajon Ultra-Torr fitting. The cathode is made negative relative to the anode and held a t the acceleration potential by electrical connections at (a) and (g). Procedures. Metal samples (NBS steels and oxygen free copper) were surface cleaned by rinses in ethanol, HC1, or "03, and deionized-distilled water. Solution samples were deposited dropwise (usually 1 ml total) onto 1 inch length, ?/*-inch diameter electrodes and dried under an infrared lamp. Both graphite (high density and conventional) and copper electrodes were used for solutions; blanks were run before sample applications. T h e CCIS mounts into the mass spectrometer source chamber, modified for fast pumping (6). The discharge is operated in the 1-5 Torr range, with currents of 0.5-5 mA a t voltages of 400-1000 volts. Critical parameters such as the acceleration/ESA ratio, beam centering voltage (Y-deflector), and discharge pressure are optimized while monitoring the matrix ion intensity. Stable isotopes were obtained as nitrates from Union Carbide Corporation, Isotope Sales, Oak Ridge, Tenn.

RESULTS AND DISCUSSION Sample Considerations. The CCIS allows considerably more flexibility in sample type and handling than does the HCIS. Basically, all that is required is that the sample be a conductor and be attachable to the cathode holder. While the present model uses a set screw mount, a more universal mount (such as a small spring clip) is planned for grasping the sample in a succeeding model. Metals in the form of rods, wires, or irregular pieces can be analyzed directly. The discharge fills the entire glass body and produces what, by visual surface examination, appears to be uniform sputtering. Metal powders could be pressed into an electrode form and nonconducting powders could be mixed with a suitable matrix, such as graphite, for electrode formation in a like manner. For solutions, we have used, in addition to the described preferred method, a spiral wire dipped into solution and dried. The total sample surface area exposed to the discharge is important in sensitivity considerations. In selecting a cathode for solution samples, a nonporous material offers sensitivity advantages, in that the sample residue film remains entirely on the surface. Thus, our first choices were metal (copper) and impervious (high density) graphite, both of which worked well. However, CCIS cathodes are not easily reusable because some of the sample beANALYTICAL CHEMISTRY, VOL. 48,

NO. 3,

MARCH 1976

*

489

Table I. Precision and Accuracy of Isotopic Ratios for Solution Residue* on a Graphite Cathode R Error, av. from

7c RSD Ratio a

SrIa Sr

Ag/ a Ag Ba/’ Ba

I

’ a

Figure 1. Coaxial cathode ion source, showing (a) extraction slit plate of mass spectrometer, (b) gas inlet, (c) glass body, (d) tantalum anode and exit slit, (e) cathode, (f) glass sleeve, and (9) cathode connector rod

comes imbedded under the surface from sputter deposition, leading to a background spectrum contribution in subsequent runs, even after surface cleaning. The cost of the impervious graphite thus led to the use of standard spectroscopic graphite rods, which show only a slight reduction in sensitivity and are sufficiently inexpensive as to be disposable. Discharge Modes. A gas discharge may be powered by several different types of sources. The simple dc discharge has worked well in the HCIS, but there are literature reports (12, 13) of greatly enhanced optical emission from pulsed hollow cathodes, which led us to examine the ionization properties under pulsed conditions. As a third mode, the rf source of the MS-702 offered a somewhat hybrid approach of an rf voltage which is pulsed a t a selected pulse length and repetition rate, allowing a comparison of pulsed dc vs. pulsed rf. Pulsed studies with the HCIS (7) have shown advantages over continuous mode for some materials. Certain matrices, such as iron, exhibit a higher matrix/ discharge gas ion ratio for pulsed than for continuous, leading to higher sensitivity. Pulsed operation results in a lower discharge resistance, comparing net average voltages for a given discharge current. A useful, although unexplained, feature of pulsed discharges is the different energy range for matrix (sample) ions vs. argon ions as compared to continuous operation wherein the matrix and argon ions show similar energies. This characteristic allows instrumental optimization for the matrix species and partial discrimination against the argon ions in the pulsed mode. A more thorough comparison of the different discharge modes will be undertaken in a subsequent report; the major thrust of this study is the description of the CCIS and its capabilities for trace element analysis. Sensitivity. Sensitivities were studied for metal samples and for solutions deposited onto metal or graphite electrodes. Pulsed vs. continuous operation yielded generally the same sensitivities, with certain exceptions. A S/N was obtained from the MCA display by taking the ratio of the peak height in counts to the variation in background counts on each side of the peak. A S/N of 2 was taken as minimum detectability. For the metals tested (copper and steels), sensitivities for isotopes unobstructed by interferences generally showed sensitivities in the 10-100 ppba range. For example, in NBS 461 steel, sensitivities of 11 (119Sn+),10 (184W+),and 10 (208Pb+)ppba were calculated from observed signal and background intensities. As a typical solution example, 1 ml of a solution containing 9 wg/ml 490

ANALYTiCAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976

Accepted value

11.8 0.95

_____ Set 1 b

3.6 2.4 2.2

dccpt. value ~

Set 2

Set 1

Set 2

2.6 1.8

6.6 3.3 6.6

1.9 1.5

10.9 5.0 Average: 2.2 1 ml of 100 pg/ml. b Set of 8 scans.

7.2 4.5

of Sr, Mo, and In deposited onto a graphite electrode showed sensitivities of 5 (88Sr+), 21 (9sMo+), and 15 (l151n+)ng/ml. The differences in sensitivities between elements have been similar to those observed for the HCIS, about a factor of 3-5. Interference lines arise from all gas discharge sources, particularly in the mass region below 60, affecting sensitivity considerations. Argon species Ar+, ArH+, Ar2+,and Arz+ are prominent. The matrix, M, can contribute M+, M2+, MH+, MO+, and MAr species, and various gas impurities such as N+, O+, HzO+, Nz+, CO+, 0 2 + , and CO2+ are observed. In the analysis of solution residues, impurities in the cathode matrix can be significant for trace analysis. A major factor in the selection of graphite for solution analysis is the availability of high purity graphite which provides a “clean” blank spectrum. Graphite sputters very poorly (14) and contributes little in the way of matrix interferences. Copper, which sputters very well, yields excellent solution sensitivity, but also contributes more matrix species which can interfere with certain isotopes. Reproducibility. The precision of the CCIS was investigated by taking sets of 8 consecutive scans from NBS standards for 15-20 isotopes. The precision appears to be better than that obtained for the HCIS. The continuous and pulsed dc modes yield average precisions of 10-15% RSD, depending upon isotopic concentration and spectral background. This can range from 5% or better % RSD for isotopes a t hundreds of ppm to 20-25% for isotopes a t the low or sub-ppm level. As in the case of the HCIS (6),the pulsed rf has yielded better precision (-5-10% RSD), but the sensitivity is not as good. That, coupled with the increased spectral contribution from the anode in this mode, has led to relatively little use of the pulsed rf, pending the construction of a modified CCIS specifically for the rf mode. The precision and experimental convenience of the pulsed rf make it of considerable interest, however, for further development. For solution samples, one expects an initially high signal, followed by a drop in intensity as the solution residue is sputtered off the cathode. However, a t the low currents (0.5-1.0 mA) normally used with the CCIS, this process is moderated. There is an initial signal peaking, but within a few minutes a “plateau” is reached which has only a slight negative slope, where perhaps redeposition tends to counter the sputter effect. Analysis of solutions a t the 1-100 pg/ml level has been run for up to an hour and retained typically 40-50% of original peak counts after that time. Over a 7-minute mass scan, the sample loss is relatively small, but it could be significant if the internal standard is considerably removed in mass from the analysis isotope. Application to Multielement Isotopic Dilution. Because of the very stable ion flux obtained from a gas discharge source, excellent isotopic ratio measurements can be obtained from an exponential mass scan. This suggested an advantageous use of the CCIS for trace element analysis by isotopic dilution. Carter and coworkers ( 1 0 ) have shown

Table 11. Multielement Isotopic Dilution Analysis with the CCIS Re1 std dev, 5% Concn, Cathode

a/ml

E

Sr /Sra

38Ba/"'Ba

g/, Error, Av., from true value

zOapb/. *06Pb

Av.

a8Sr/. 87Sr

Graph it e 100 1.7 3.4 0.9 3.5 2.9 Graphite 10 3.9 1.7 1.8 3.0 2.9 Graph it e 1 28.5 5.8 10.4 14.9 24.6 1.6 3.3 Copper 10 3.3 1.4 2.7 apreparedratios: 88Sr/87Sr= 0 . 9 4 6 ; 138Baj13sBa = 0 . 9 2 4 ; z o s P b / z o 6 P=b1.03.

that multielement isotopic dilution can produce better than 5% precision for SSMS with a photoplate detector to average out ion fluctuations. We have applied the same multielement principle, on a limited scale, for the CCIS with electrical detection. Our initial experiments were with normal (non-enriched) isotopic patterns. One ml of a solution containing 100 pg/ml of Sr, Ag, and Ba nitrates was applied to a graphite cathode. Table I shows the results for sets of 8 consecutive scans for isotopes that varied by as much as an order of magnitude. In isotopic dilution, it is advantageous to adjust the measured isotopic ratio to a value approaching unity ( 9 ) ,if possible, to improve the accuracy of measurement. Enriched stable isotopes (85.9% 87Sr, 78.8% I3jBa, and 99% *06Pb) were used to spike standard prepared solutions, producing a net sample with isotopic ratios near unity. One ml of 1, 10, and 100 pg/ml solutions was deposited on graphite electrodes; the data are summarized in Table 11. Both graphite and copper cathodes were used. At 1 pg/ml, fluctuating background contributions caused problems in obtaining accurate data. Otherwise, comparison of the prepared ratio to the experimentally determined ratio consistently yielded results in the 2-5% RSD range. All of the data in Table I1 were taken in the continuous discharge mode. Recent work has shown that pulsed dc operation with a copper cathode yields enhanced sensitivities for solution residues. However, graphite appears to offer considerable advantage for solutions, particularly its nonreactivity with acidic solutions. The reduced sputtering activity is also useful in maintaining the residue film.

CONCLUSIONS The CCIS provides a means of trace element analysis in solids and solutions which shows better precision and es-

138Ba/.

135Ba

zo6pb/. z06Pb

Av.

4.0 3.0 9.0 3.4

2.1 2.4 6.2 2 .o

2.3 2.4 13.3 2.3

sentially similar sensitivity when compared to spark source mass spectrometry. However, more interferences are encountered with the CCIS. Compared to the HCIS, the CCIS allows much more convenient handling of both solids and solutions and operates at low current conditions, which reduce heating effects and allow solution residues to be maintained on the electrode surface. Multielement isotopic dilution with the CCIS offers promise for analysis of solutions where high accuracy is desired, thus justifying the additional measurements and sample preparation.

LITERATURE CITED (1) (2) (3) (4) (5) (6) (7)

(8) (9) (10) (11) (12) (13) (14)

J. W. Coburn and E. Kay, Appl. Phys. Lett., 19, 350 (1971). J. W. Coburn, E. Taglauer, and E. Kay, J. Appl. Phys., 45, 1779 (1974). W. W. Harrison and C. W. Magee, Anal. Chem., 46, 461 (1974). B. N. Colby and C. A. Evans, Anal. Chem., 46, 1236 (1974). E. H. Daughtrey, Jr., and W. W. Harrison, Anal. Chem., 47, 1024 (1975). D. L. Donohue and W. W. Harrison, Anal. Chem.. 47, 1528 (1975). W. W. Harrison and W. A. Mattson, 23rd Annual Conference on Mass Spectrometry and Allied Topics, Houston, Texas, May 1975, paper N-11. W. A. Mattson and W. W. Harrison, 23rd Annual Conference on Mass Spectrometry and Allied Topics, Houston, Texas, May 1975, paper N-10. H. Farrar in "Trace Analysis by Mass Spectrometry", A. J. Ahearn, Ed., Academic Press, New York, 1972, pp 283-291. J. A. Carter, D. L. Donohue, J. C. Franklin, and R. W. Stelzner, 23rd Annual Conference on Mass Spectrometry and Allied Topics, Houston, Texas, May 1975, paper N-2. W. W. Harrison and W. A. Mattson, Anal. Chem., 46, 1979 (1974). J. B. Dawson and D. J. Ellis, Spectrochim. Acta, Part A, 23,565 (1967). H. Prugges, R. Grosskopf, and R. Torge, Spectrochim. Acta, Part8, 26, 191 (1971). G. Carter and J. S. Colligan, "Ion Bombardment of Solids", Heineman, London, 1968.

RECEIVEDfor review August 25, 1975. Accepted November 3, 1975.

Use of Trapped Water as a Chemical Ionization Agent in Mass Spectrometric Analysis of Environmental Air Samples 1. C. Wang, H. S. Swofford, Jr.,* P. C. Price, D. P. Martinsen, and S. E. Buttrill, Jr. Department of Chemistry, University of Minnesota, Minneapolis, Minn. 55455

The advantages of cryogenic preconcentration of volatile trace components in air samples are exploited in chemical ionization mass spectrometry using water as the reagent gas. A Du Pont 21-490B instrument was modified for chemical ionization (CI). Known environmental air samples were drawn through a trap at -78 O C for 1 min at a flow rate of

400 mllmin. As the trap was warmed, the contents were swept into the mass spectrometer with methane for analysis in the CI mode. Water vapor served as a secondary reagent gas. The limit of detection of ethyl butyrate in air was 1.5 ppb.

ANALYTICAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976

491