Multielement Analysis of Basaltic Rock Using Spark Source Mass Spectrometry G . H. Morrison and A. T. Kashuba Department of Chemistry, Cornell University, Ithaca, N . Y . 14850 Spark source mass spectrography has been used for the multielement analysis of the U. S. Geological Survey standard basalt BCR-1. U. S. Geological Survey standard diabase W-1 was used as a multielement concentration standard to obtain sensitivity factors. It has been possible to determine 60 elements using graphite and silver blended rock mixtures as electrodes at several magnet settings and exposure sequences. When the analysis is compared with those of other investigators, including a neutron activation analysis performed on a sample of BCR-1 from the same bottle, agreement is found to be good.
Soniple
Figure 1. Electrode arrangement during sparking
AMONGthe most important types of information desired from the lunar material returned to earth by the Apollo mission is the characterization of its chemical composition. Of the various analytical methods available for the determination of the average abundance of elements in geological materials, spark source mass spectrometry offers the possibility of the most complete survey of the elements present in a given sample. Taylor ( I ) , Brown and Wolstenholme (2), and Nicholls (3) have used the technique to examine a variety of rock types, and more recently Graham and Nicholls ( 4 ) have studied the lanthanide elements in basalts. This laboratory, in preparation for the lunar analysis, has investigated the capabilities of the technique for the quantitative multielement analysis of terrestrial rock samples, including sampling, sample preparation, and measurement, as well as a study of comparative standards for quantitation. Based on the results of the a scattering analysis of the top layer of the lunar surface at the landing sites of Surveyors V, VI, and VII, the chemical composition resembles that of terrestrial basalts (5-7). Consequently, a basaltic sample has been examined by spark source mass spectrometry and the capability of determining 60 elements has been demonstrated. Spectral interferences and/or insufficient limits of detection in this complex rock matrix prevents the determination of additional elements; however, those that can be determined include many of great potential geochemical interest. For accurate quantitative measurements to be made, a comparative standard must be employed to obtain elemental sensitivity factors to correct for differences in behavior of the various elements in a given matrix. Therefore, experiments have been run using the U. S. Geological Survey standard diabase, W-1, as the source for obtaining sensitivity factors. A second Geological Survey standard, BCR-1 Basalt, has also been analyzed and treated as the “unknown” in order to assess the capability of the spark source mass spectrometric (1) S. R. Taylor, Geochim. Cosmochim. Acta, 29, 1243 (1965). (2) R. Brown and W. A. Wolstenholme, Nature, 201, 598 (1964). (3) G. D. Nicholls, A. L. Graham, E. Williams, and M. Wood, ANAL.CHEM., 39, 584 (1967). (4) A. L. Graham and G. D. Nicholls, Geochim. Cosmochim.Acta, 33, 555 (1969). ( 5 ) A. L. Turkevich, E. J. Franzgrote, and J. H. Patterson, Science, 158, 635 (1967). (6) Ibid., 160, 1108 (1968). (7) A. L. Turkevich, J. H. Patterson, and E. J. Franzgrote, ibid., 162, 117 (1968). 1842
method. Neutron activation analysis of these samples has also been performed and provides a check on the validity of the results (8). At present the mass spectrometric method results in a relative standard deviation of 5 to 25 about the mean and an average accuracy of 1 4 z for major and trace elements.
z
EXPERIMENTAL Mass Spectrograph. The Nuclide Analysis Associates
GRAF 2 double-focusing spark source mass spectrograph previously described (9) was used in this study. Fixed experimental conditions are given in Table I. Sample Preparation. The diabase W-1 and the basalt BCR-1 were dried at 110 “C for 1 hour. Two hundred and fifty ppm (wt) In as In2O3was added as an internal standard. The mixture was blended with graphite (National Carbon Co. spectroscopic powder) or silver (Cominco American 5-9’s grade Ag powder, lot HPM 7942). Mix ratios and variable experimental conditions are given in Table 11. Electrode Preparation. Sample and conductor in the proportions given in Table I1 were hand-blended for 20 to 30 minutes in an agate mortar and pestle. A polyethylene slug of 11-mm diameter from the AEI pellet press was halved radially with a stainless steel blade. Sample-conductor mix was pressed at -5 X lo5lb/in2to obtain a disk of 1 to 2 mm thickness. The disk was quartered axially with the stainless blade. The quarter disks were mounted in Ta electrode chucks as shown in Figure 1 and presparked for a charge collection of 30 to 60 nC. The sparking occurred uniformly along the freshly exposed surfaces. N o Fe, Ni, or Cr contamination appeared during analysis which could be traced to the press. Analysis of the graphite and silver powders was performed by visual estimation of line intensity. The analysis indicated that those trace elements present as a blank were also present in the rock samples at concentrations orders of magnitude higher, and they do not pose any problem in the rock analysis. The only exceptions were Ni and Cu present in relatively high concentrations in the silver; however, Cu was determined using rock blended with the purer graphite. Interferences precluded the determination of Ni.
(8) G. H. Morrison, J. T. Gerard, A. Travesi, R. L. Currie, S. F. Peterson, and N. M. Potter, ANAL.CHEM., 41, 1633 (1969). (9) C. A. Evans and G . H. Morrison, ibid., 40, 869 (1968).
ANALYTICAL CHEMISTRY, VOL. 41, NO. 13, NOVEMBER 1969
Data Reduction. Per cent transmittance of the maxima of appropriate lines and backgrounds immediately adjacent to these lines were read from a Jarrell-Ash console densitometer. One hundred per cent transmittance was set on a fixed undeveloped plate from the same batch as that used in recording the spectra. A series of programs were written for use with a HewlettPackard H P 9100 electronic calculator as an aid for data reduction. Program 1 gave a least squares linear fit to the equation
ln[(l/T) - 11 = m 1nE
+b
(1)
where T is the transmittance of the line maximum, E the exposure monitor reading in nanocoulombs for the internal standard line 1*5In+,and rn and b are the slope and intercept, respectively. Program 2 solved Equation 1 for E using line maximum Ti and background Tb, then back-solved Equation 1 for corrected T, using the arithmetic difference El - E*. Using T, and the original E, a new fit was obtained using Program 1. The iteration continued in this manner until the differences between successive fits produced less than 1 change in the slopes, expressed in degrees. Program 3 computed the background corrected intensity ratio for each isotope line read, and the average and standard deviation of the body of ratios for the isotope. It also computed the concentration of the element, correcting for isotopic abundance, atomic weight and line broadening, using: line broadening a (rn/e)1'2, rn = mass, amu; e = charge. Sensitivity factors for the individual elements were obtained from the W-1 standard using: SF = literature value/experimental concentration, where the literature values were obtained from recent compilations of analyses of W-1 (IO, l l ) ,as listed in the work of Morrison et al. (8). RESULTS AND DISCUSSION Sensitivity Factors. Earlier work has shown the possibility
of using solids mass spectrography for the analysis of geological materials. As Brown and Wolstenholme indicated, however, lack of sensitivity factors had limited accuracy of their determination to =t300~. To provide a comparative standard in spark source mass spectrography, the criterion of similarity of standard to sample must be fulfilled. Because of the complexity of rock matrices, the preparation of a synthetic standard containing the various trace, minor, and major elements in the appropriate chemical form and concentration i s a formidable task. Whenever possible, the use of a standard rock similar in composition to the samples to be analyzed is preferred. Fortunately, the availability of a standard diabase, W-1, together with a large amount of analytical data from the compilations of Fleischer (10, 11) greatly simplified obtaining sensitivity factors for basaltic samples. Optimization of Experimental Conditions. In order to obtain the highest possible sensitivity for the majority of elements, certain experimental parameters were systematically varied. For ease of data reduction, those variables not expected to intereact with each other were tested separately. A test of electrode configuration used (see Figure 1) contrasted with the normal end-to-end sparking of cylinders or rods of rectangular cross section showed both enhancement of M+ and improved precision. Ratios of rock-to-conductor mix in Table 11 are the highest possible rock fractions that preserved mechanical strength during sparking. (10) M. Fleischer, Geochim. Cosmochim. Acta, 33, 65 (1969). (11) Ibid., 29, 1263 (1965).
Table I. Fixed Mass Spectrographic Conditions
Vacuum conditions : Magnet analyzer Electrostatic analyzer Source chamber Beam monitor: Instrument parameters : Gap voltage Pulse duration Pulse repetition frequency Proportionality" Primary slit Detector Developing conditions: Developing
torr torr -lo-? torr Cary vibrating reed electrometer and Cary Gold-plated turret head -60 kV 100 psec 320/sec 8.9 2 . 5 X 10-3in. 2 X 16 in. Ilford 4-2 photoplate
Eastman D-19 (full strength) at 20 "C for 3.5 min Stop bath Running water for 30 sec Fixing Eastman Rapid Fixer with hardener for 1 min Running water for 5 min Washing Warm air for 3 min Drying a Accelerating voltage = proportionality x electrostatic analyzer voltage. Table 11. Electrode Cornpasition and Variable Experimental Conditions
Rock-uaohite Long Short exposure exposure Sample/conductor ratio Internal standard Magnet current Mass range Electrostatic analyzer voltage No. of exposures Exposure range, nC
1 :1 250 ppm In
175 mA 3-140 1050 V 32 30-10-3
1 :1 250 ppm In 400 mA 20-300
Rock-silver 1 :9 250 pprn In
275 m A 6-240
1050 V
1550 V
27
31
103-1 0- 1
103-1 0- 3
Table 111. Trace Element Lines Observed in Fogged Areas on the Photoplate
Analytical lines IOBi-, llB+ lSF+
31p+ 35Cl+ 37Cl+ 42Ca+ 43Ca+ 5lV+, 5 2 c r + 53Mn+ 59Co+
Lines causing fogging lZC+, 13Cf lQ+, 23NaC 28Si+ 28Si+, 39K+ 39K+,40Ca+ 48Ti+ 48Ti, 56Fef, ;07&+2 l0DAgi-2 56Fe+ 107&+2, lOSAg+Z
Three parameters were suspected initially of possible interaction: duty cycle, the fraction of time that the spark is on; electrostatic analyzer voltage; and the proportionality setting, which determines the ratio of electrostatic analyzer (ESA) and accelerating voltages. A standard analysis of variance experiment was chosen to study the possible interactions and the values of the parameters affording the maximum sensitivity. The only interaction observed was for ESA and proportionality setting. Duty cycles of 3 were observed to give the maximum sensitivity irrespective of other variables. A final study of ESA voltage was performed in the case of the silver matrix because of the appearance of two maxima. The 15.5-kV setting provided the higher sensitivity for representative metallic elements.
ANALYTICAL CHEMISTRY, VOL. 41, NO. 13, NOVEMBER 1969
1843
~
~~~
~~
~
Table IV. Analysis of BCR-1 Basalt (ppm) No.
Element
Mass spec.
Neutron activation
Li B
13 2.4
... ... ... ...
0
41 %
F Na Mg A1 Si
460 1.6% 1.4% 6.3% 24 1730 72 1.1% 4.9% 1.09% 310
z
P
C1 K Ca Ti V
Cr Mn Fe co
10.5
1420 7.5% 45 24 145 24 2.2 0.2 37 270 39 180 19 0.9 (1)b (0.2)b 0.03 0.02 0.3
cu
Zn Ga Ge Se Rb Sr
Y Zr
Nb Mo Ru Rh Pd Ag Cd Sn Sb Te
3
0.9 (0.4)b (0. 8)b 1.1 730 23 45 8 21 6 1.5 6
I
Cs Ba La Ce Pr Nd Sm Eu Gd Tb
1
7 1 5 0.3 3.3 0.5 3.4 0.7
