Analyst, February, 1983, Vol. 108, pp. 159-165 159 Development Progress in Plasma Source Mass Spectrometry Alan R. Date and Alan L. Gray Institute of Geological Sciences, 64/78 Gray’s Inn Road, London, WClX 8NG Department of Chemistry, University of Surrey, Guildford, Surrey, G U2 6XH Development work in the application of the inductively coupled plasma to plasma source mass spectrometry is described. Preliminary results obtained under continuum (or bulk plasma) sampling conditions are illustrated and compared with previously published boundary layer sampling work. The technique is now shown to be viable for multi-element analysis of complex samples. Keywords : Plasma source mass spectrometry ; inductively coupled plasma ; continuum (or bulk plasma) sampling mode ; multi-element trace analysis Plasma source mass spectrometry is a rapidly developing technique that combines the speed and convenience of sample introduction to a plasma source with the simple spectra and isotope ratio capability of atomic mass spectrometry.In this context, the technique is applied to solution samples, introduced to the inductively coupled plasma source (ICP) using a conventional pneumatic nebuliser with an uptake rate of 1.5 ml min-l. Samples may be processed at a rate of one every few minutes. Two modes of operation have been identified by us, boundary layer sampling and continuum (or bulk plasma) samp1ing.l Previously published work describing development of the ICP as an ion source2-5 has been limited to the boundary layer sampling mode. Boundary Layer Sampling Mode Ions extracted from the tail flame of the ICP into the vacuum system of the mass spectro- meter have to traverse a boundary layer of cool gas which forms over the sampling aperture.The formation of this boundary layer is discussed else~here.~9~ The boundary layer sampling mode is characterised by: (a) high signal to background ratios; (b) excellent detection limits (see Table I) ; (c) operation with aerosol desolvation ; (d) serious ionisation suppression ; (e) formation of oxide and hydroxide ions; and (f) aperture blocking at high salt concentrations. However, the unique configuration of plasma torch, sampling cone, ion lens and detector we have developed for use in the boundary layer sampling mode5 leads to the almost complete removal of stray background. Although the technique is limited to the analysis of simple solutions, there are several possible applications.The rapid determination of lead isotope ratios in galena (natural PbS) samples has been described and the potential of the technique for elements forming volatile hydrides illustrated.1 The high signal to background ratio achieved in this mode suggests that time spent in sample enrichment will be rewarded in terms of the speed and convenience of subsequent mass analysis. Continuum Sampling Mode The two advantages of boundary layer sampling, high signal to background ratios and excellent detection limits, are off-set by the inability of the system to accept solutions with total salt concentrations greatly in excess of 10 pg ml-1. For the analysis of complex samples it is necessary to induce sufficient flow from the plasma gas to break through the cool boundary layer.Simply increasing the sampling aperture diameter in order to effect this, results in the formation of a “pinch” discharge in the aperture mouth caused by compression of the free electron population of the plasma gas. By regulating the pressure immediately behind the sampling aperture it is possible to suppress the discharge and achieve controlled expansion of the plasma gas into the vacuum system. Development work in the continuum sampling mode is described in detail elsewhere.6160 DATE AND GRAY: DEVELOPMENT PROGRESS Aaalyst, Vol. 108 RF power supply Multi-channel analyser and display X - Y Teletype Cassette plotter recorder A Fig.1 . General system diagram : continuum sampling mode. The general system diagram, incorporating a controlled expansion stage, is shown in Fig. 1, and the new plasma sampling interface in Fig. 2. A blank spectrum (1% V/V nitric acid) obtained in the continuum sampling mode is illustrated in Fig. 3, taken with the total trans- mission reduced to avoid saturation in the major peaks. Although broadly similar to the equivalent boundary layer spectrum (see Fig. 3 in reference 5), the 190H,+, 30NO+ and 8oAr,+ peaks are much smaller, and the 81Ar,H+ peak is not detected. P2 = Water-cooled front plate - Extraction electrode Expansion Stage 2 -Torch box U ~ ~ ~ ~ a r t z bonnet C - \ Sampling - cone r coil torch ll - atm ’ cm Fig. 2. Plasma sampling interface : continuum sampling mode.At the present stage of development, the continuum sampling mode is superior to the boundary layer sampling mode in all but two characteristics, (a) and (b) above. Although the signal to background is lower by one or two orders of magnitude, and detection limits in most instances are inferior (see Table I), desolvation of sample solutions is unnecessary,February, 1983 I N PLASMA SOURCE MASS SPECTROMETRY 161 41 (2 968617 counts s-’1 16 Fig. 3. Blank spectrum (1% V/V nitric acid) : continuum sampling mode. ionisation suppression is much less significant (Fig. 4), oxide formation is limited, even for strongly bound oxides such as thorium (Fig. 5 ) , and salt condensation is insignificant. The upper limit in the dynamic range for the system is controlled by the detector and counting chain.The response for cobalt, shown in Fig. 6, is closely linear over six orders of magni- tude. TABLE I DETECTION LIMITS (30, ng ml-l) IN PLASMA SOURCE MASS SPECTROMETRY Element Lithium . . Boron . . Magnesium Aluminium Titanium . . Vanadium. . Chromium. . Manganese Iron .. Cobalt . . Copper . . Zinc .. Germanium Arsenic . . Selenium . . Rubidium . . Silver . . Cadmium . . Indium . . Tellurium . . Caesium . . Barium . . Lanthanum Cerium . . Tungsten . . Gold . . Mercury . . Lead . . Bismuth .. Thorium . . Uranium . . . . .. . . .. .. .. .. .. .. .. .. .. .. . . .. . . . . . . .. .. .. .. .. .. .. .. .. .. .. .. .. Boundary layer mode 0.4 0.2 0.3 0.3 0.1 0.06 0.5 0.1 0.3 2.0 0.1* 0.2 0.1 0.2 - - - - - - 0.2 - - 2.0 0.2 - 0.05 Continuum mode 5.0 2.0 0.7 0.9 0.5 0.6 0.2 1.3 0.8 4.3 1.6 - - 11 23 0.4 0.3 0.7 0.2 0.7 0.2 0.4 3.4 0.3 1.5 0.7 0.2 0.6 0.4 0.3 1.5 0.3 0.5 0.5 * Hydride generation.162 lo7 DATE AND GRAY DEVELOPMENT PROGRESS Analyst, Vol.108 ! - /' '"10 20 50 100 200 500 1000 Sodium concentration/pg m1-l Fig. 4. Effect of up to 1000 pg ml-l of sodium on 10 pg ml-1 of cobalt. I 232Th (86 767 counts s-l) Fig. 5 , Spectrum for thorium at 100 pg ml-1, showing small oxide peak. 10' 0.001 0.01 0.1 1.0 10 100 1000 Cobalt concentration/yg ml- ' Fig. 6. Calibration graph for cobalt, with background correction, and gain adjustment (to overcome counting losses at high count rates). Broken line: corrected for background (1% nitric acid blank). Detection limit (20, blank), 0.001 pg ml-1.Febraary, 1983 IN PLASMA SOURCE MASS SPECTROMETRY 163 In order for plasma source mass spectrometry to find ready application to multi-element trace analysis of complex samples, several criteria must be satisfied.The technique must be capable of simultaneous analysis of a wide range of elements in a variety of matrices. 2 4 6 8 10 12 14 lonisation energylev Fig. 7. Relationship between response (counts s-l a t 1 pgml-') and ionisation energy, Vi (ev), for 25 elements in the range m/z 7-209 (corrected to lOOyo abundance in each instance). Sensitivity is within a factor of three for elements with Vi below 11 eV. Elements: Li, B, Al, C1, V, Cr, Mn, Fe, Co, Zn, Ge, As, Se, Br, Rb, Ag, Cd, In, Te, I, Cs, Au, Hg, Pb, Bi. To supplement detection limit data, presented in Table I, the relationship between response (counts per second at 1 pg ml-l) and ionisation energy (electron volts) for 25 elements in the range 7-209 m/z (corrected to 100~o abundance in each example) is illustrated in Fig.7. Sensitivity lies within a factor of three for elements with ionisation energies below 11 eV. The two elements outside this range, bromine and chlorine, have ionisation energies of 11.85 and 6 3 ~ u + (sampling aperture) 238u + ..a - 3 Fig. 8. Spectrum in 2048 channels over the range rn/z 0-260 for a solution containing 10 pg ml-l each of Al, Co, As, Br, Rb, In, Te, I, Cs, La, W, Au, Pb, Bi and 17 in 1% V / V nitric acid. With a dwell time per channel of 800 ps and only 60 sweeps, the total spectrum was taken in just over 1 min.Each isotope occupies approximately seven channels and was therefore addressed for about 0.2 s.164 DATE AND GRAY: DEVELOPMENT PROGRESS Analyst, Vol. 108 13.02 eV, respectively. The ionisation equilibrium of the plasma flame is controlled princi- pally by the plasma support gas argon, with a first ionisation energy of 15.76 eV. Elements with second ionisation energies below 15.76 eV will be partially doubly ionised. This is illustrated in Fig. 8, which shows a spectrum in 2048 channels over the range 0-260 m/z for a solution containing 10pgml-l each of Al, Co, As, Br, Rb, In, Te, I, Cs, La, W, Au, Pb, Bi and U in 1% V/V nitric acid. Lanthanum and uranium show significant double ionisa- tion, while lead (second ionisation potential, 15.03 eV) is only slightly doubly ionised.With a dwell time per channel of 500 ps and only 60 sweeps, the total spectrum was taken in just over 1 min. Each isotope occupies approximately seven channels and was therefore addressed for about 0.2 s. I 75As+(65 000 I counts s-l) 79Br+ 85Rb+(79133 counts s - l ) 8 1 ~ ~ + Fig. 9. Scale expansion (128 channels) of the region in Fig. 8 covering arsenic, bromine and rubidium. Count rates were calculated from the peak channel count in each instance. J U L Although it is possible to identify most of the isotopes present in Fig. 8, the region o1 the spectrum covering arsenic, bromine and rubidium is subjected to vertical ( x 16) and hori- zontal ( x 16, 128 channels) scale expansion in Fig. 9. All isotopes may be readily identified by this approach. Count rates (calculated from the peak channel integral count in each instance) are shown for 75As+ (65000 counts s-I) ands5Rb+ (79 133 counts s-l).The apparently shortened appearance of the 80Ar-Ar+ dimer peak is caused by peak fold-over at high gain in the data system used (Canberra Series 40). In order to make a preliminary assessment of performance with a complex matrix, the solution used in the above example was diluted with a synthetic geological-matrix solution i (16367 counts s-') Fig. 10. Scale expansion (128 channels) for the same solution as in Fig. 8 diluted to 5 pg ml-1 showing the lead isotopes and bismuth (A), com- pared with a similar dilution in the presence of a synthetic geological matrix containing Na and K at 50 pg ml-l and Mg, Al, Ca and Fe at 100 pg ml-l (B).Fe brzGary , 1983 IN PLASMA SOURCE MASS SPECTROMETRY 165 prepared from BDH atomic-absorption standard solutions.A scale-expanded spectrum of the region covering the lead isotopes and bismuth (at 5 pg ml-l) is illustrated in Fig. 10, and compared with a similar dilution in 1% V/V nitric acid. The concentrations used for the synthetic matrix, 50 pg ml-l of Na and K and 100 pg ml-l of Mg, Al, Ca and Fe, were limited by the available stock solution concentrations. They represent a dilution factor of 1000 (0.1 g in 100 ml) for a solid sample containing 5% Na and K and 10% Mg, Al, Ca and Fe. The negligible effect of this matrix addition on lead and bismuth is shared by all but two of the trace elements. Iodine is a serious contamination in the synthetic matrix and is probably present in one of the BDH standard solutions (probably the potassium solution).Although the system is far from optimised, and much research and development remain, these data suggest that a viable multi-element technique for the analysis of complex samples will soon be more widely available, a technique having far-reaching implications in fields as diverse as geochemical research, the life sciences, pollution monitoring and the nuclear power industry. Gold has disappeared from both solutions (see Fig. 10). This work was supported by the Institute of Geological Sciences and the European Com- munity Research and Development Programme on “Uranium Exploration and Extraction” (contract No. EXU 033-81-UK). The paper is published with the approval of the Director, Institute of Geological Sciences (NERC) . References 1. 2. 3. 4. 5. 6. Date, A. R., and Gray, A. L., Spectrochirn. A d a , Part B, in the press. Houk, R. S., Fassel, V. A., Flesch, G. D., Svec, H. J., Gray, A. L., and Taylor, C. E., Anal. Chem., Houk, R. S . , Fassel, V. A., and Svec, H. J., in Price, D., and Todd, J . F. J., Editors, “Dynamic Gray, A’,?., and Date, A. R., in Price, D., and Todd, J. F. J., Editors, “Dynamic Mass Spectro- Date, A. R., and Gray, A. L., Analyst, 1981, 106, 1255. Gray, A. L., and Date, A. R., in preparation. 1980, 52, 2283. Mass Spectrometry,” Volume 6 , Heyden, London, 1981, pp. 234-251. metry, Volume 6 , Heyden, London, 1981, pp. 252-266. Received July 27th. 1982 Accepted September 14th, 1982