would reduce the total analysis time to about 12 minutes per sample. Consideration has been given to the construction of a completely automated NMR device for performing these determinations. This would be greatly facilitated by the fact that a non-spinning NMR sample tube still allows sufficient resolution for this type of work (at least a t 100 MHz). With such a device, the analysis time could be reduced to about 2 minutes per sample.
LITERATURE CITED
(3) (4)
(5) (6) (7)
(1) ASTM test method designation D1319-70 in "1971 Annual Book of ASTM Standards", American Society for Testing and Materials, Philadelphia, Pa., 1971. (2) L. I. Grindstaff, L. A. Bryan, and M. P. Whittaker, "Characterization of Pe-
troleum Coker Feedstocks (by) Nuclear Magnetic Resonance" (in Symposium on Advances in Analysis of Petroleum and its Products presented before the Division of Petroleum Chemistry, 164th National Meeting, American Chemical Society, New York, N.Y., August 27-September 1, 1972). D. R. Clutter, L. Petrakis, R. L. Stenger, Jr., and R. K . Jensen, Anal. Chem., 44, 1395 (1972). R. B. Williams and N. F. Chamberlain, "New Developments in Hydrocarbon Type Characterization Using Nuclear Magnetic Resonance", (Proc. 6th World Pet. Congr., Germany, Section V, Paper 17, 1963). D. W. Mathieson, Ed., "Nuclear Magnetic Resonance for Organic Chemists", Academic Press, New York, N.Y., 1967. F. A. Bovey, "Nuclear Magnetic Resonance Spectroscopy". Academic Press, New York, N.Y., 1969. L. M. Jackson, "Applications of Nuclear Magnetic Resonance Spectroscopy in Organic Chemistry", Pergamon Press, New York, N.Y., 1959.
RECEIVEDfor review March 17, 1975. Accepted July 7, 1975.
NOTES
Determination of Low Concentrations of Cobalt in Small Samples of Plant Material by Flameless Atomic Absorption Spectrophotometry W. J. Simmons Department of Soil Science and Plant Nutrition, Institute of Agriculture, University of Western Australia, Nedlands, Western Australia 6009
The accurate measurement of Co in plants is of considerable importance since Co deficiency occurs in animals when their feed contains less than about 80 nglg (dry weight) ( 1 1. Concentrations can drop to as little as 10 ng/g. Therefore, very sensitive analytical methods are required. The two most commonly used methods for measuring Co, colorimetry (2, 3 ) and flame atomic absorption spectrophotometry (4-6), require at least 20-g and 5-g samples, respectively when extremely low concentrations of Co are encountered. In two studies (a survey of the trace element content of a wide range of pasture and cereal species (7) and an investigation of the role of CO in legume nutrition) currently being conducted a t this Institute, it was not feasible to collect 5 g of material for analysis. For these studies, a more sensitive procedure was required which could measure Co in as little as 0.5-g of plant material. The great sensitivity of the graphite furnace developed by Massman (8) suggested that it could be applied to the development of a suitable procedure. While Segar (9) has described two procedures using the furnace for measuring Co in marine plants, its application to the determination of Co in terrestial plants does not appear to have been described in the literature. (Since submitting the manuscript of this paper, another method for determining Co in terrestial plants using the furnace has come to my attention (N. Kr. Sorensen, Tidsskr. Planteavl, 78, 156 (1974)). Sorensen's method is more complex and less sensitive than the one described in this paper.) The procedure described here is simpler than either of Segar's methods and more sensitive than his simplest procedure. This paper, which examines the application of the proposed method to eleven different plant species, gives preci-
sion and accuracy data as well as describing the effects of high concentrations of eight of the most common plant elements on the accuracy of the method.
EXPERIMENTAL Reagents. All reagents were tested for Co content. Only "03 was a potentially serious source of contamination. A batch of "03 supplied by one manufacturer contained 0.5 to 1 ng Co/ml. A different batch of acid supplied by another manufacturer and containing no detectable Co was used for the plant digestions described in this paper. Apparatus. A Perkin-Elmer model 403 atomic absorption spectrophotometer, in conjunction with an Hitachi Perkin-Elmer model 165 recorder was used to measure the atomic absorption of samples injected into the special grooved tube of a Perkin-Elmer model 72 heated graphite atomizer (HGA). Procedure. The procedure used is described below. Digestion. Digest 0.5-g samples of plant material (oven dry 70 "C) contained in 50-ml Erlenmyer flasks with 15 ml of concentrated " 0 3 and 2.5 ml of 72% w/v HClO4 (IO) by heating the flasks on an electric frypan. (Carry two blanks through the procedure.) and the sample to proceed Allow the reaction between the "03 for a t least 40 minutes ( 1 1 ) . When the reaction between the concentrated and organic residue is completed,'continue heating a t 180 "C to dehydrate any silica and to remove most of the excess HC104. (If a flask becomes dry, allow it to cool a little, then add 0.5 ml of 72% w/v HC104 and replace it on the frypan. When the HClO4 fumes appear, swirl the flask occasionally to disperse the residue. After 5 to 10 min, remove the flask. Add 0.5 ml of HClO4 to the spare blank.) Extraction. After allowing the flasks to cool, rinse the wall of each flask with 4 to 5 ml of deionized water. Warm the solutions on the frypan, with intermittent swirling, to dissolve any KC104 precipitate. While the flasks are cooling, make up standards containing 0, 10, 30, 60, 90, 120, 150, 180, and 200 ng of Co by adding each of these amounts together with 2.5 ml of 15 HC104/deionized water to
ANALYTICAL CHEMISTRY, VOL. 47, NO. 12, OCTOBER 1975
2015
5 2
IO
;*
:,tr . -C
0
nD 2~NITROsO->-NAPT*LII
Figure 1. (0-0)Effect of amount of 2-nitroso-I-naphthol on the reading of a single extraction of 194 ng of Co. (X-X) Effect of innial amount of 2-nitroso-I-naphthol on % of original 194 ng of Co found in the second extraction All eXtTacUons were performed in the presence Of 1 ml Of 30%
W I V H202
50-ml Erlenmyer flasks. Adjust the volume of each standard to 6 ml with deionized water. When the digests have cooled to room temperature, add 1ml of 40% w/v sodium citrate solution, 1drop of hromthymol hlue and 1 drop of methyl red, and adjust the pH of the digests and standards to be in the range 5.3-5.7 (orange to golden orange color), with concentrated ammonia solution. Correct any overshoot by adding 3M HCI or 0.5M HC1 or 2M NHdOH. Add 1 ml of 30% w/v Hz02 followed by 0.3 ml of 1%wlv ethanolic solution of 2-nitroso-lnaphthol. Transfer each solution to a 15-ml capacity stoppered cylinder. Rinse the walls and any precipitate present in each flask with two lots of about 1.5 ml of deionized water, adding each rinse to the solution in the cylinder. Make the final volume of the solution in the cylinder into the range 11to 12 ml. Add 1ml of isoamyl acetate to each cylinder and vigorously shake the contents for 30 see. (Each solution should he well mixed at each step of the procedure.) Measurement. After inserting the special grooved tube into the HGA, set the drying, ashing, and atomization times and temperatures to 0.6 min at 92 O C , 33 see at 1075 "C, and 12 sec at 2175 "C, respectively. Set the full scale of the recorder to 0.5 absorbance, the hollow cathode lamp current to 20 mA, and the spectral bandpass of the atomic absorption spectrophotometer t o 0.2 nm and measure the absorbance of 50-fil aliquots of the isoamyl acetate extract at the 240.7-nm Co line, The presence of B very small peak, possibly due to a little 2-nitroso-1-naphthol Condensing on the outer cooler parts of the tube during the ashing step, preceding the main Co peak, has no effect on the reading. Safety. The safety precautions for using HClOa have heen described in detail by Everett and Graf (12). A fume extractor (see Figure 2) fitted to each end of the furnace and connected to a water tap suction pump removed nearly all the fumes evolved during the operation of the HGA. Any remaining fumes were removed by an extractor canopy mounted above the spectrophotometer. RESULTS AND DISCUSSION Preliminary Experiments. pH a n d Reaction Time. T h e reaction between 2-nitroso-1-naphthol and Co proceeds rapidly in the cold in the pH range 4-9 (13). The indicator mixture which changes color from orange to yellow over the pH range 5.3 to 6.2 gave adequate control of pH. Amount of 2-Nitroso-1-naphthol. T h e effect of the amount of reagent on efficiency of extraction was evaluated by initially complexing 194 ng of Co with 0.2 to 5 mg of 2nitroso-1-naphthol. After measuring the absorbance of each sample, the isoamyl acetate layer was removed and a second extraction performed using 5 mg of the reagent. The amount of Co present in the second extract was measured using standards prepared with 5 mg of Z-nitroso-l-naphthol. Figure 1 shows that when at least 2 mg of 2-uitroso-lnaphthol was present, only a further 1%. at most, of the original amount of Co could he extracted when the aqueous phase was treated with a further 5 mg of the reagent. 2016
ANALYTICAL CHEMISTRY, VOL.
47, NO. 12,
OCTOBER
Flgi amount of reagent necessary to remove virtually all the Co in a single extraction. While the 13% increase in the reading of the initial ex. tract, as the amount of 2-nitroso-1-naphthol rises from 2 to 5 mg, is difficult to explain, the effect is not important: first, because analysis of the second extract shows that virtually all of the Co has heen removed in the first extraction; second, the plant analysis data (discussed later) show that, provided 3 mg of reagent is used for complexing the Co, the precision and the accuracy of the procedure are adequate for most plant analyses. Extraction Time. Vigorous shaking of the cobalt 2-nitroso-1-naphtholate solution with 1 ml of isoamyl acetate for 10, 20, 30, 60, and 120 seconds gave the same ahsorhance when 48.5 ng of Co was taken for extraction. Thirty seconds was chosen as a reasonable length of time. Effect of Iron. Iron interfered in two ways but its effects were easily overcome. First, the reading of a solution containing 48.5 ng of Co was depressed; the percentage depression increasing from 8.7 to 30% as the amount of added F e rose from 13 to 132 Fg. Second, a black precipitate was present in the extract when 396 Sg or more of Fe was added to the solution. HzOz has heen used to prevent interference from F e in the colorimetric determination of Co with 2-nitroso-1naphthol (14). The interferences were absent when 0.2 to 2.0 ml of 30% wlv H202 was added to a solution containing 48.5 ng of Co and 1380 ua of Fe. Therefore, no interference from this elemf3nt shouli be encountered with 0.5-g samples provided t kie concentration of Fe in the plant material is less than 1 PPk. . 2760 . 0thn.l -P.m r i r i n n n n d A r r r r r n_r vj . Tnrli~rirlnrl Proposed M L I.1_ analyses of a wide range of plant materials comprising eleven different species from four different botanical groups (herbs, legumes, grasses, and cereals) were performed over a period of five months. Consequently the data represent the kind of precision and isccuracy which one could expect from the routine use of the method. PRECISION. Table I shi)ws the relative standard devia. . tions (RSD) at the lowest and nignest sample weignzs or the four plant materials analyzed to obtain precision data. The RSD for a single determination ranges from 2.9 to 7.1%. The precision of the furnace method using 0.3- to 1-g ___.
~
....
1975
__
__
.,
Table I. Effect of Sample Weight on Precision a n d Result Mean
Plant m a t e r i a l
KO, of
conceneation,
kVelght, g
analyses
ng C o l g
RSD
0.5 1.o 0.5 1.o 0.3 1.o 0.3 1.o
6 7 6 12 5 7 5 24
10.7 11.2 36.0 34.3 72.1 70.9 144.9 142.2
6.2 5.6 6.4 6 .O 2.9 3.4 6.8 7.1
Wheat Serradella Erodium Lucerne
samples is similar to that obtained using the flame method ( 4 ) with 1-to 6-g samples. ACCURACY.The accuracy of the method was evaluated in four different ways. First, by running recoveries of Co added to the plant material before digestion; second, by comparing the furnace results with those obtained four years ago for the same plant material using flame atomic absorption spectrophotometry; third, by comparing the results obtained a t two quite different sample weights; finally, by examining the effect on Co recoveries of large amounts of other elements which had been added to wheat to increase the concentration of these elements to the highest levels likely to be encountered in analysis. RECOVERIES.Table I1 shows that the recovery of Co added in amounts approximately equal to that already present in plant material was satisfactory. The difference of the mean recovery from 100% was due to chance for each different plant material ( P > 0.1). COMPARISON WITH FLAMEANALYSIS. The furnace measurements were in satisfactory agreement with the flame results (Table 111). While the mean percentage difference of 2.76% was due to chance ( P > 0.4), the 5.9% dif-
ference between the lucerne results was real ( P < 0.02). A large number of determinations has been necessary to reveal this small difference, the magnitude of which is not important for most purposes. EFFECTOF SAMPLE WEIGHT ON RESULT. Youden (15) has pointed out that the presence of constant bias in a method is not detected by the use of recoveries but is revealed when different sample weights are analyzed. The results for the four different plant materials (Table I) examined in this way were satisfactory since each difference is due to chance ( P > 0.1). EFFECTOF HIGH CONCENTRATIONS OF OTHER ELEMENTS. Although a wide range of species had been examined in testing the accuracy of the method, it was unlikely that extremely high concentrations of other elements present in plants had been encountered. Therefore, the levels of these elements were increased by adding various amounts of extremely pure solutions of each element to 1- or 0.5-g samples of wheat or wheat straw, which originally contained 16 and 19 ng Co/g, respectively, prior to digestion. Ninety-seven ng of Co was added as a recovery to the plant material. It can be seen from Table IV that, of the trace elements, only Mn, when present in amounts greater than 75 pg per digest (150 pg/g for a 0.5-g sample), had a serious effect. At worst, a 20% enhancement could be expected when 608 to 1215 wg was present in a digest (1216 to 2430 pg for a 0.5-g sample). While the reason for the enhancement was not investigated, there are two ways in which its effect may be minimized or eliminated. The first, which has not been tested, is simply to ensure that all samples and standards contain a t least 608 wg of Mn. The second is to restore an acid p H in the solution after the stable Co complex has been formed. This was done by adding 1.5 ml of concentrated HC1 to the solutions after allowing the reaction between Co and 2-nitroso-l-naphthol to proceed in the coni~~
~
Table 11. Recovery of C o Added to P l a n t Material Standard
Plant m a t e r i a l
Range of
Amount of C o
sample weights, g
added, ng
Wheat Ser r ad el la Erodium Lucerne
ho. of recoveries
0.972-1.037 9.7 0.956-1.070 48.5b 0.489-0.9 7 5 48.5b 0.2 96-0.5 13 48.5' a S/v'ii where S is the standard deviation and n the number of determinations. 47 ng of Co was added for one of the recoveries and 97 ng for another.
\lean recover),
error
01
reco, e q a
3
101.4 8.2 98.2 2.5 5 107.1 5.9 5 108.5 5.3 6 ng of Co was added for one of the recoveries
5
47 6
Table 111. Comparison of Furnace Results with Flame Results Fumace
Plant material
KO. of
Sample
Mean,
Flame
determinations
weight, g
ng c o i g
ng C o l g
Difference, 3
11 16 25 25 30 36 57* 62 80 114 151.8' 141
0 +19.4 -20.8 -11.2 -20.7 -3.3 +5.6 +0.2 -9.9 -1.8 -5.9 +15.3
Wheat 13 0.51 11.0 Wheat straw 1 1 19.1 Oats straw 2 1 19.8 Cereal rye seed 1 0.5 22.2 White lupin stems 2 0.5 23.8 S e r r ad e 1la 18 0.5-1 34.8 Wimmera ryegrass 2 0.5 60.2 Purple vetch 2 0.5 62.1 Erodium 25 0.3-1 72.1 Subterranean clover 2 0.5 112.0 Lucerne 35 0.3-1 142.9 W. A. blue lupin 2 0.4 162.6 a Single assay except where indicated. Mean of 2 determinations. Mean of 12 determinations.
ANALYTICAL CHEMISTRY, VOL. 47, NO. 12, OCTOBER 1975
2017
Table IV. Effect of High Concentrations of Trace Elements on Co Recovery u g 31 Element in digest Fe
cu
Zn
Mn
Recovery, %
1526 836 146 73 73 3 76 146 836 73
102 52 2 1 1 17 7 52 1
405 205 205 2 02 2 65 25 5 202
1215 615 615 608 608 195 75 15
115.7 113.0 111.9 117.2 119.6 111.1 94.6 100.5 105.8
a
ed graphite atomizer has been used for the development of this method, the new procedure, or a modification of it, should be applicable to other alternative furnaces. In a study of cobalt distribution in plants, Wilson and Hallsworth ( 1 7 ) were unable to measure the Co in various plant parts because their analytical technique did not have sufficient sensitivity. The proposed procedure should allow such experiments to be carried out since only a small amount of plant material is needed for analysis. ACKNOWLEDGMENT The author thanks A. D. Robson for helpful, constructive discussion in both the development of the method and the preparation of the manuscript. LITERATURE CITED
cal flask for 15-20 min. Duplicate recoveries of 102.2 and 105.7% were obtained when this was done in the presence of the highest levels of trace elements listed in Table IV. A disadvantage of this modification was that the slow liberation of 0 2 which occurred after about llh hr, if a precipitate was present, might be a safety hazard. Replacement of H202 with ascorbic acid (16) might be a satisfactory way of avoiding this hazard. None of the macro elements interfered seriously with the proposed procedure. Recoveries ranging from 101.3 to 106.7% were obtained when Ca (0-25000 pup) and Mg (04000 pg) were added together in various proportions to 0.5 g of wheat. The recovery was 100.2% when 50000 pg of K and 10000 pg of P were added to 1g of wheat straw. CONCLUSION The greater sensitivity of the furnace has allowed the development of a faster, simpler procedure for measuring Co in plant material than either flame atomic absorption or colorimetry because it needs only 0.5-g samples. At the same time, there has been no loss in accuracy or precision when compared with these other methods. While the heat-
(1) E. J. Underwood, "Trace Elements in Human and Animal Nutrition", 3rd ed., Academic Press, New York. N.Y., 1971, p 150. (2) HR : . Marston and D. W. Dewey, Aust. J. Exp. Biol. M e d . Sci., 18, 343 (1940). (3) D. W. Dewey and H. R. Marston, Anal. Chirn. Acta., 57, 45 (1971). (4) W. J. Simmons, Anal. Chem., 45, 1947 (1973). (5) J. Jago, P. E. Wilson, and B. M. Lee, Analyst, 96, 349 (1971). (6) A. L. Gelman, J. Sci. Food Agric., 23,299 (1972). (7) J. S. Gladstones and J. F. Loneragan, Aust. J. Agric. Res.. 18, 427 (1967). (8)H. Massman, Specfrochirn. Acta., Part 8, 23,215 (1968). (9) D. A. Segar, lnt. J. Environ. Anal. Chern., 3, 107 (1973). (10) C. M. Johnson and A. Ulrich, "Analytical Methods for Use in Plant Analysis", Bull. Calif. Agr. Exp. Stn., 768, 32 (1959). (11) J. E. Alien, "The Preparation of Agricultural Samples for Analysis by Atomic Absorption Spectroscopy", Varian AerographNarian Techtron Publication, 1969, pp 7 and 13. (12) K. Everett and F. A. Graf, Jr., "Handbook of Laboratory Safety", 2nd ed., The Chemical Rubber Co., Cleveland, Ohio, 1971, pp 265-276. (13) E. Boyland, Analyst, 71, 230 (1946). (14) E. B. Sandell. "Colorimetric Determination of Traces of Metals", 3rd ed.. Interscience, New York, N.Y., 1959, p 420. (15) W. J. Youden, "Statistical Methods for Chemists", 1st ed., Wiley, New York, N.Y.. 1957, p 40. (16) T. C. Z.Jayman and S.Sivabramaniam, Analyst, 99, 296 (1974). (17) S.B. Wilson and E. G. Hallsworth, Plant Soil, 23,60 (1965).
RECEIVEDfor review May 5 , 1975. Accepted June 30,1975. Work supported by the Western Australian State Wheat Industry Research Committee.
Simultaneous Multielement Determination of Trace Metals by Microwave Induced Plasma Coupled to Vidicon Detector: Carbon Cup Sample Introduction Fred L. Fricke Cincinnati District Food and Drug Administration, 114 1 Central Parkway, Cincinnati, Ohio 45202
Oliver Rose, Jr., and Joseph A. Caruso' Department of Chemistry, University of Cincinnati, Cincinnati, Ohio 4522 1
In numerous laboratories, there is a need for simultaneous multielement analysis, because the time and cost of single element analyses are prohibitive. It .has been shown that inductively coupled plasma-optical emission spectographic systems are useful for simultaneous multielement analysis ( I , 2). The use of low power microwave induced argon plasma for spectral excitation of the elements also has been well documented (3-17). When coupled to a multichannel spectrometer, the microwave induced plasma has potential for multielement trace analysis. Author to whom correspondence should be addressed. 2018
Morrison and coworkers (18) developed a multielement flame emission system using a vidicon tube detector and an optical multichannel analyzer (OMA). They provide an excellent description of this particular instrument. Nixon et al. (19) have developed a tantalum strip sampling introduction device for use with an inductively coupled plasma. This paper describes the combination of a microwave induced plasma with a carbon cup sampling device. A comparison is made between this apparatus and a tantalum strip sampling device coupled to a microwave induced plasma. In addition, a preliminary study of the potential appli-
ANALYTICAL CHEMISTRY, VOL. 47, NO. 12, OCTOBER 1975