normally from 70 t o 90% (based on labeled mercury) depending on the degree of heating. The residue normally contains 2 to 6% labeled mercury. Repeated and prolonged heating failed t o remove this mercury residue completely, as measured by the labeled mercury. The volatile losses of mercury were normally 2 to 5% but OCcasionally were as high as 20%. This loss is again dependent upon the rate and time of heating. Both heating the column containing gold a t 105” C. for several minutes to ensure complete dryness and the use of an ether wash in lieu of acetone were tried. No losses were noted on drying; however, the ether rinse did give somewhat poorer precision. The best precision was obtained with a high specific activity radioactive mercury solution. The errors in the isotope dilution technique are larger a t the low concentration. These are primarily due t o errors in weighing from moisture, catalyst and reagent dust, sampling error, and counting errors. Table I1 gives an example of the precision found by the isotope dilution method for one sample. X - Ray Emission Spectroscopy. Since carbon, a light absorber of x-rays, and mercury, a heavy absorber, vary in their relative amounts from specimen t o specimen, the total absorption also changes from specimen to specimen. Thus, direct measurement of samples and standards would not yield a linear working
curve. AS t h e dilution of the sample, with a relatively transparent material, is increased, the differences in composition of the original sample affect total absorption in a decreasing amount. When using sodium borate as the diluent, a tenfold dilution was necessary to obtain a linear working curve over the range of 1 to 18% mercuric chloride. Dilution with sodium borate offers another advantage. Since carbon, a major constituent of the catalyst samples, is a very light absorber of x-rays, it is inconvenient to measure enough sample so that infinite thickness is achieved,-Le., sufficient thickness so that the intensity is independent of the thickness. Sodium borate, although relatively transparent, is a heavier absorber than carbon. Thus, the dilution with sodium borate permits the usage of a sample of a more convenient thickness. Incomplete mixing, uncorrected absorption effects, and ordinary instrumental fluctuations are perhaps the chief sources of error. Assuming mixing to be complete, a precision of =kO.l% and accuracy of =k0.2Y0absolute is estimated for this determination. Table I11 shows some results found. Analyses of a large number of samples of both known and unknown mercury content gave agreement satisfactory for our purposes. Some typical examples are given in Table IV. Part of the early differences found between X R F and isotope dilution values (for example NO. 13) were due to sampling errors,
since the carbon did not contain a uniform mercury content. Quartering the samples minimized this error. The comparison of mercury analyses by various methods is summarized in Table IV. The sulfuric-nitric acid digestion and emission procedures were developed and performed by an independent laboratory; the other analyses were made by company personnel. Both isotope dilution and x-ray emission spectrography have been used for routine analyses of mercury on carbon catalyst containing HgClz and CeC18. The x-ray method is preferred because of speed, accuracy, and precision, once calibrations have been prepared. LITERATURE CITED
(1) Boyd, T. (to Monsanto Chemical Co.), U. S. Patent 2,446,123-4 (July 27, 1948). (2) Brookes. H. E.. Solomon. L. E.. ‘ Analyst 84, 633 (1959). (3) Fahey, J. J., IND. ENG. CHEM., ANAL.ED.9, 477 (1937). (4) Gunn, E. L.. ANAL. CHEXI.29, 184 (1957).’ (5) Hillebrand, W. F., Lundell, G. E., “Applied Inorganic Analyses,” pp. 167 ff., Wiley, New York, 1929. (6) Lacy, J., Anal. Chim. Acta 20, 195 (1959). (7) Scott, W. W., “Standard Methods of Chemical Analysis,” 5th ed., pp. 574 ff., Val. I. W. H. Furman. ed.. Van Nostrahd, New York, 1958. (8) Zarkovskii, F. V., Aptechnoe Delo, 1952, No. 5, 35; C . A . 47,445f (1953). ’
’
I
RECEIVEDfor review April 21, 1961. Accepted January 19, 1962.
Determination of Antimony by the Ring Oven Technique PHILIP W. WEST and ALBERT0 J. LLACER Coates Chemical I aboratories, I ouisiana Sfate University, Baton Rouge, I a.
b A method for the determination of antimony, suitable for use in air pollution studies, is presented utilizing solvent extraction of the Sbld- complex on the ring oven and subsequent development of color with phosphomolybdic acid. The limit of identification of the method is 0.08 pg. with no serious interferences for analyses of air samples. A modification of the Weisz ring oven to permit concentration of a sample and subsequent extraction of one component to the ring zone is also presented.
I
that both antimony and its salts are poisonous. Fairhall and Hyslop ( I ) have reviewed the literature on the toxicology of antimony, and the analytical problems T IS WELL ESTABLISHED
associated with the determination of this metal have been discussed by Jacobs (3). The development of the ring oven technique now offers the possibility of providing relatively simple and rapid methods for separating, concentrating, and determining trace amounts of antimony collected during air pollution and industrial hygiene studies. The maximum allowable concentration for antimony has been given as 0.1 mg. per cubic meter of air. For convenience in sampling, a method sufficiently sensitive for samples of 1 cubic foot would be ideal. In terms of absolute amounts of antimony, this requires a sensitivity of 2.8 pg. A sensitivity of 1 pg. per cubic foot as a minimum sensitivity would be more realistic so that an amount substantially less
than the maximum allowable concentration could be determined easily. Reasonable efficiency can be assumed for the collection of antimony dusts and fumes using paper and Millipore filters, impingers or electrostatic precipitators. Regardless of the method of sample collection, it can be assumed that any antimony isolated will be accompanied by possible interfering substances such as lead, iron, aluminum, copper, bismuth as well as certain nuisance dusts such as silica, silicates, and various carbonaceous materials. Therefore, the new method was designed to separate antimony from possible interferences, followed by a highly distinctive and sensitive discriminatory test utilizing a minimum of equipment and chemical operations. VOL. 34, NO. 4, APRIL 1962
555
the original test solution if it contained no oxidants, may then be used in the following steps. Make a tiny pencil mark in the center of a piece of Whatman No. 41 filter paper. Place the paper on the supplemental aluminum plate resting on a ring oven warmed to approximately 90" C. Hold the paper in place with a brass ring resting beneath the guiding tube. Using the same capillary pipet for both standard and unknown rings, spot on the filter paper as many 5-A portions of the test solution as are required to obtain sufficient antimony for a reliable test. Add 5 A of 4% H2S04 solution, followed by 5 A of 10% K I solution, to the center of the ring. When the spot has almost dried, withdraw the filter paper from the auxiliary plate and place it, as usual, on a second ring oven, maintained a t about 75" to 85" C., using the pencil mark and a capillary pipet passing through the guide tube to ensure correct centering of the ring. The [Sb14]formed in the previous step is then extracted with 5-A portions of the solvent mixture, using a total volume of EXPERIMENTAL approximately 2 ml. (30 to 40 extractions). Care should be taken to avoid Special Reagents. PHOSPHOJIOLYBany diffusion of the solvent mixture DIC ACID. 5% aqueous solution. The beyond 1 mm. from the heating block reagent solution is best when freshly to prevent charring of the ring zone. prepared, but does work satisfactorily After the extraction is completed, let when 3 to 4 days old. the filter paper stand in place for a few SOLVENT.Benzene-ethyl alcohol seconds. Generally a visible ring will mixture; 2 to 1 by volume. be formed due to the solute extracted, ANTIMOW CHLORIDE.Standard solubut it is no cause for concern. tion prepared in approximately 6 N HC1 Heat some phosphomolybdic acid and standardized against standard solution to its boiling point in a caschloramine T solution ( 5 ) . serole and bathe the filter paper in the Procedure. If a test solution is hot solution for about 1 minute. The suspected of containing Sb(V) or any appearance of a persistent blue ring is oxidants, a preliminary treatment of an indication of the presence of antithe solution is required. T o this mony(II1). Wash off the excess reaend, it suffices to add a small portion gent by bathing the filter paper in disof solid sodium sulfite to the test solutilled water and let dry. tion. The reduced test solution, or The limit of identification was determined as 0.08 pg., using the method of establishing a limit described by Table I. Results of Comparing Test Feigl(2). Rings with a Standard Scale M a d e up The stability of the intensity and of a Series of Rings Containing 0.1 to color of the rings was checked by comparing rings which had stood for varying 1 .O pg. Sb(l1l) periods of time with freshly prepared Antimony concentration = 0.9 p g . rings containing the same amount of Ring in Standard Scale antimony(II1). S o significant changes Marched by Test Ring were noticeable, even a month after Test Ring Actual Theoretical the development of a ring. The weak positive blank can be attributed to impurities. The procedure of West and Hamilton ( 7 ) held promise for the selective extraction of antimony(II1) on the ring oven. Because the extraction process is carried out in the interstices of paper with the ring oven technique, certain modifications were required in adapting the original procedure which depends on the specific extraction of antimony iodide or association complexes of antimony using benzene as the extractant. In general, a small volume of the sample solution containing antimony(II1) was placed on a circle of filter paper. The sample spot was centered on the ring oven so that subsequent washing operations would extract the antimony to the ring zone, giving equal distribution of the extract throughout the circumference of the ring. The conditions for effective isolation of antimony were studied , and methods for the subsequent detection and determination of the isolated metal were surveyed and finally evaluated.
RESULTS
VI VI1 Table II.
Sample NO.
0.9
0.9
0.9
0.9
Analyses of Synthetic Unknowns
Actual Concentration, PR.
Found
0,30" 0.20 3 0.2G Average of two determinations: 0.320 28. 1 2
556
0.30 0.20 0 25
ANALYTICAL CHEMISTRY
The quantity of antimony(II1) present was estimated by direct comparison of the intensities of the rings formed from the test solution with those formed from a standard solution. The standard rings were prepared as described above by collecting varying amounts of antimony(II1) ranging from 0.1 to 1.0 pg. A study was carried out t o determine the accuracy of visual comparison of rings for three different ranges of concentration of antimony(II1). Seven test rings containing 0.3 pg. of Sb(II1)
were compared independently with the standard scale, and each mat,ch 0.3 pg. Seven test rings containing 0.5 pg. of Sb(II1) were similarly compared, and each was matched correctly. The results of the comparison of seven test rings containing 0.9 pg. of Sb(II1) are presented in Table I. The reliability of the estimation of antimony(II1) in the presence of Sn(II), Fe(II), and As(II1) was investigated. For each ion, three rings were prepared containing 0.2, 0.5, and 0.8 pgo of antimony(III), with the following ratios of Men+: Sb(II1): Sn(I1):Sb = 250/1 Fe(I1):Sb = 400/1 As(II1):Sb = 380/1 Comparison of the nine rings so obtained with the standard scale resulted in the theoretical matching being obtained in each case. Three synthetic unknown solutions were analyzed by preparing three test rings for each and performing the calculations as verified by Knodel and Weisz ( 4 ) . The ratio of the concentration of the test solution to that of the standard solution is found by dividing the sum of the number of drops (or fiveportions) of test solution used to make the test rings into the sum of the number of drops of standard solution, used to make the standard rings matched by the test rings. The quotient or ratio is multiplied by the concentration of the standard solution to obtain the concentration of the test solution. The results of these analyses, presented in Table 11,show that the method is accurate within the 570 limit suggested ( 4 ) . Further verification of the method mas obtained by statistical evaluation. Several standard scales were prepared, and two solutions having known concentration ratios to standard solutions were made up. The standard scales were obtained using the customary technique in which the rings represented a series of 1, 2, 4, 6, 8, and 10 drops each. The drops were obtained from standard capillary pipets. At the same time, 15 test rings were prepared covering the range from 1 to 10 standard drops. The rings were divided in sets of three according t o an arbitrary pattern chosen prior to the preparation of rings. For the statistical evaluation, each sample consisted of three rings which were treated as a unit. The average value for the ratio of the test solution t o the standard solution was then computed according to the method of Knodel and U'eisz, and the results are shown in Tables I11 and IV. DISCUSSION
Isolation (Extraction) Process. For maximum efficiency, i t n-as desirable to carry out all steps for conditioning,
--GUILE TUBE
R RING ING
Figure 1.
Modified ring oven
separating. and determining the antimony on a single sheet of filter paper. T o keep the reactions confined prior t o the extraction and determination operations, the conventional ring oven was modified by using a n auxiliary aluminum plate to reduce the size of the ring oven bore hole from the normal 22-mm. diameter to a diameter of 10 mm. (Figure 1). By using solutions and reagents in increments of 5-h volumes and hazing the auxiliary plate for confining the area occupied by the reacting system, it was possible to carry out all conditioning operations, leaving the resulting products fixed in a ring having a diameter slightly less than 10 mm. The extraction of antimony from the conditioned system was then accomplished by removing the auxiliary plate and centering the 10-mm. primary ring to make a perfectly concentric circle with the normal bore hole of the ring oven. The final process of extracting the antimony iodide system to the outer ring was then accomplished by successive additions to suitable solvent. The confined spot will contain a n appropriate amount of sample combined with various agents introduced to condition the system for subsequent operations. If the sample used is in the form of a solution, it can be treated with a small amount of sodium sulfite to ensure the reduction of any antimony(V) to antimony(II1). Alternatively, the appropriate amount of sample on the paper can be treated uith 5 A of 1% sodium sulfite, followed by 5 h of dilute sulfuric acid. Because of the proclivity of antimony to hydrolyze, it is necessary to maintain relatively high acidities. The concentration of acid is somewhat critical, however, because of the deteriorating effect of the acid and heat on the paper a t the ring zone. The optimum concentr:ition of acid used was approairnately A S . The formation of the antimony ioditle for the extraction process was readily accomplished by diffusing 5 of IO%,
potassium iodide solution into the initial ring immediately after it had been treated rvith a like volume of 4iv sulfuric acid. The extraction of the antimony iodide was then accomplished by removing the auxiliary plate, centering the initial ring on the ring oven, and then extracting with 5-h portions of solvent (30 to 40 portions). Although benzene was the solvent employed in the original procedure devised by West and Hamilton, it proved to be unsatisfactory for use with the ring oven technique because of its poor diffusion characteristics through filter paper. Various solvents such as carbon tetrachloride, chloroform, propyl alcohol, butyl alcohol, diethylether, and acetone were tried, but none was found to be effective. Because benzene was ideally suited except for its diffusion characteristics, further studies were made. It was found ultimately that the addition of ethyl alcohol to benzene did not appreciably restrict its solvent capabilities, and yet its diffusion characteristics were greatly improved. Various proportions of benzene to alcohol were tried, and a 2 to 1 ratio was finally selected as optimum. Experiments were run under various conditions to determine if there n-as an actual extraction of the [SbIdI- complex. Washing Kith either distilled water or 0.1N HC1 resulted in the formation of weak or spread rings. The significantly more intense coloration of those rings developed by using the solvent mixture for the same concentration of antimony(II1) demonstrated the important role of the solvent. The amount of solvent required was investigated. Too little washing resulted in incomplete collection of antimony in the ring zone, while too much washing was unnecessary. From 30 t o 40 5-1 portions of solvent was optimum. The temperature of the ring oven was quite important, as was the volume of reagents added, particularly during steps in which the auxiliary plate was used. Temperatures of 75" to 85' C. provided rapid evaporation of solvents without being sufficiently high to enhance charring of the paper by acid's being concentrated in the ring. The volume used for reagents and solvents, 5 h, was sufficient for diffusion of the spot to the heated area of the ring, without a t the same time causing an overriding of the ring by flooding. Where successive volumes of solution were added, care n as taken to avoid too rapid addition of solutions n-hich ~ o u l d ,of course, result in flooding. Color Reactions for Estimation of Isolated Antimony. Various spot tests and other color reactions for antimony TI ere investigated for po+ sible use in the quantitative estimation of the iqolated metal. The tormation of the antinionv-rhodamiiie
Table 111.
Determination of Antimony
Ratios of Test Standard Solution Taken Found" 1.08 =I= 0.02b 0.89 f O . O l b a Based on averaging 5 values calculated from 3 rings each according to the method of Knodel and Weisz (4). * Calculated at 90yo confidence level. 1.125 0.875
Table IV. Errors Found in 10 Determinations of Antimony Using the Knodel and Weisz Method of Calculation Absolute Value of Error in a Single
Determination" Maximum 0.065 Minimum 0.005 Average 0.033 The units of measurement are ratio of the test solution concentration to that of the standard solution. Table V. Interferences in Test for Sb(lll) When Present in 1 00-fold Excess Ion Type of Interference
Yellow iodide caused Sb(II1) ring to be off color, although of proper intensity Se(1V) and (1.1) Reduced to Sea giving a reddish ring which obscures the Sb(II1) reaction Te(1V) and Forms a dark ring which obscures the Sb(II1) (VI) and Pt nietals reaction W( VI ) Prevents the Sb(II1) reaction
TI(1)
13 coinples (7) did not lend itself to application with the ring oven. K i t h one exception, other color reactioiis tried either lacked the necessary sensitivity or were not applicable d i e n used with the ring oven technique. Fortunately, the classical reaction of antimony(II1) \\-ith a hot solution of phosphomolybdic acid yielded sharply defined rings (molybdenum blue) which n-ere insoluble and so intensely colorcd that excellent sensitivity was obtained. The reaction with antimony is ordinarily far from specific, but this was no tletcrrent in the present case because selectivity was obtained through the estraction step. The final estimation of antimony was carried out by the statistical method of comparison with standard rings as Jcscribed by Knodel and JTeisz (+). Interferences. The interfercnce studies followed the procedure of Kest ( 6 ) . The study was made with 100 fig. of potential interfering ioiie compared to 1 pg. of antimony(II1). The ions were checked both alone anti when mixed with antimony(II1). VOL. 34, NO. 4, APRIL 1962
557
Under these conditions, the following ions neither gave a test which could be mistaken for antimony nor prevented the test for antimony when it was present: Li, Na, K, Cu(II), Rb, Ag, Cs, Au(III), Be, Mg, Ca, Bn, Sr, Cd, Ba, Hg(I), Hg(II), BO:-, B407-’, Al, Ga, Ce(III), Ce(IV), Ti(IV), G~OS-’,Zr(IV), Sn(II), Sn(IV), Pb, Th, NH(+, NO’-, NOS-, Pod-’, VO+’, OS-, As(III), As(V), Bi(III), S-’, SZOS-~,SOS-’, SO4-’, Cr(III), Mo(VI), UOZ+~, uO4-’, F-, C1-, c104-, Mn(II), Br-, BrOs-, I-, Ios-, Re(III), Fe(II), Fe(III), Co, Ni, CN-. A few ions gave interferences. Their nature is described in Table V.
The oxidizing agents, NOz-, Cr(VI),
c103-, Br03-, IOs-, and Fe(III), do not interfere when present as 100 pg. with 1 pg. of Sb(III), but, because they liberate iodine and may prevent complete complexation of larger amounts of Sb(III), they should be destroyed. Prior treatment of the test solution with hydrazine sulfate is recommended, except for NOz- which can be treated with urea. LITERATURE CITED
(1) Fairhall, L. T., Hyslop, F., U. S. Public Health Repts., Supplement 195 (1947).
(2) Feigl, F., “Qualitative Analysis by
Spot Tests,” 3rd ed., p. 4, Elsevier, New York, 1946;( (3) Jacobs, M. B., The Analytical Chemistry of Industrial Poisons, Hazards and Solvents,” 2nd ed., p. 253, Intencience, New York. 1949. (4) Knodel, ’W,, Weisz, H., Mikrochim. Acta, 1957, 417;(
(5) Vogel, A. I., Quantitative Inorganic Analysis,” 2nd ed., p. 376, Longmans, Green and Co., New York, 1967. (6) West, P. W., J . Chem. Educ. 18, 528 (1952). ( 7 ) West, P. W., Hamilton, W. C., ANAL. CHEM.24, 1025 (1952). RECEIVED for review October 30, 1961. Accepted January 30, 1962. Work was supported by the National Institutes of Health, Grant No. 7481.
Estimation of Beryllium with Eriochrome Cyanine R Using the Ring Oven Technique PHILIP W. WEST and PATRICIA R. MOHILNER Coates Chemical laboratories, Louisiana State University, Baton Rouge, l a .
A method for the microdetermination of beryllium using Eriochrome Cyanine R and the ring oven technique is presented. Determinations can b e made on as little as 0.01 mg. per ml. (0.05 pg.) of beryllium with an average error of 7%. Of the elements likely to b e of significance in air pollution studies, none was found to interfere when present in 10-fold excess, and only Mg, Th, AI, and Cr interfere when present in 1 00-fold excess.
T
of beryllium and its compounds has emphasized the need for improved methods for determining this metal in air pollution studies. This work has been carried out with a view to developing a rapid, reliable, semiquantitative procedure for the determination of beryllium in atmospheric samples utilizing the simplicity of the ring oven method (3) in an adaptation of the spectrophotometric method employing Eriochrome Cyanine R ( I ) . The method presented here permits a rapid evaluation of the concentration of a solution, such as may be prepared from an air sample suspected of containing beryllium as a pollutant, and is applicable to readily soluble forms of beryllium collected from air samples. HE INCREASING USE
EXPERIMENTAL
Reagents. Eriochrome Cyanine R. Solution is prepared by triturating the
558
ANALYTICAL CHEMISTRY
solid material (Matheson, Coleman, and Bell, dry stain) with 2 t o 3 ml. of water. One milliliter of the saturated solution is decanted and diluted to 5 ml. This solution should be prepared fresh daily. The buffer solution is prepared by dissolving approximately one mole of ethylenediamine in 500 ml. of water, adding HC1 to bring the pH to 9.8 as measured with a pH meter, and diluting to 1 liter. The acid solution is approximately 0.1N HC1, while the masking agent is a 2% aqueous solution of sodium nitrilotriacetate (NaNTA) (LaMotte Chemicals, Dallas, Tex.). Paper. Whatman No. 41 filter paper. Circles of 55-mm. diameter are convenient for the ring oven. Collection and Preparation of Sample. The sample of air to be examined for beryllium content may be collected in any of the customary ways and placed in solution by any method which avoids the use of fluorides, which mask the reaction if present in gross excess. Clearly, only the more soluble forms of beryllium can be detected and estimated by this technique. Procedure. Add to the center of the piece of Whatman No. 41 filter paper on the ring oven the following, in order, taking care to wait between additions for a sufficient period of time (20 to 30 seconds) to prevent flooding of the ring, 10 pl. of 2% aqueous NaNTA, sample, 10 pl. of 2% aqueous NaNTA, 3 10-pl. portions of 0.1N HCl, 10 pl. of buffer, 10 pl. of Eriochrome Cyanine R solution, and 10 10-p1. portions of buffer.
Remove the paper from the ling oven and immediately ITash with continuous agitation in distilled water until all traces of yellow, browi, orange, or red color disappear. The washed ring is then dried over an electric hair drier. Analysis of an Unknown. The above procedure is followed for the preparation of each ring. A standard scale is conveniently prepared by making rings where the sample in each case is 1, 2 , 4, 6, 8, or 10 portions containing 5 or 10 pl. of a standard solution containing 0.01 pg. of Be per pl. The size portion chosen for the standard scale should also be chosen for making the rings of the unknown. Generally 3 rings made from different numbers of portions of the solution are sufficient for each unknown. Each of the three rings is then compared visually to the standard scale, and it is decided if the ring matches one of the standard scale or falls between two members. Thus, to each of the three rings made from the unknown solution is assigned one of the numbers: 1, 1.5, 2 , 3, 4, 5, 6, 7, 8, 9, 10. To compute the ratio of the concentration of the unknown to that of the standard solution, the sum of the numbers of the three matching rings of the standard scale is divided by the sum of the number of portions of unknown solution used in making the unknown rings. The concentration is then determined by multiplying the quotient found in the preceding step by the concentration of the solution used to prepare the standard scale. Once the stable standard scale is prepared, the analysis of an unknown, including the preparation of the three