calcincs, sinters, residues, and othcr niiscellaneous smelter materials. Treatmerit of such samples n-ith H N 0 3 before evaporation with HClO, is recommended to avoid explosion hazards. Large amounts of iron may be extracted by 25 nil. of isopropyl ether from the perchlorate residue dissolved in 7 V HC1. Following the iron separation, evaporate residual ether froin the aqueous layer and add 0.5 ml. of HClO,. Evaporate to remove HC1 and excess HCIOl and proceed with the HBr extractions. One gram of ascorbic acid added to the HBr phase 10 minutes prior to adding the isopropyl ether reduces the interference of iron. For high-lead or high-copper materials, use two separate 5-ml. HBr washes rather than a single 10-ml. portion. Any of these modifications is likely to change the over-all efficiency of extraction. Consequently, for accurate results, it is neccssary to prepare the standard curve by (wrying standard samples through the w t i r e procedure with 11hirh the unlrnonn samples are to be treated. Accuracy and Reproducibility. Suitable standard samples are not a v d a h l c for checking the accuracy of this ni(Jt1iod. A series of production sam1)les was run in quadruplicate by the dithizone eytraction procedure. The results. shown as samples 5 to
~
Table II.
NO.
1
2 3 4 5 6 m
;3
9 10
Accuracy and Reproducibiiity of Indium Determination s,
70,
% In,
Relative Error
% In, Polarographic
0,000675 0.00116 0.00476 0.00938 0.00595 0.00468 0,00192 0,00190 0.00225 0,00894
4.75 0.52 0.90 1.39 1.50 0.77 1.50 0.63
...
Estimat,ed std. dev.
2.57
Sample Type Metal Metal Metal Metal Metal Metal Oxide Oxide Oxide Oxide
Dithizone
10 of Table 11, indicate a satisfactory correlation with the indium concentration as obtained by polarographic means. The standard deviation for the relative error of all the samples in Table I1 is 2.6%, a figure which includes any nonhomogeneity of the samples as well as errors of measurement by the method. -4 scries of eight synthetic samples containing 20 pg. of indium and 1 gram of zinc had a coefficient of variation of 1.6. LITERATURE CITED
(1) Hudgrns, J. E., ?;elson, L. G., Ax.4~.
CHEK24, 1472 (1952).
5.40
2.40
...
... 0.00545 0.00465 0.00187
0.00213 0,00228 0.00594
% Diff. ... ... ... 0,00050 0.00003 0,00005 0.00023 0.00003 0.00000
Av. 0.00014
(2) Kosta, L., Hoste, J., Mikrochim. Acta 416, 790 (1956). (3) May, I., Hoffman, J. I., J . Wash. A c a d . Sci. 38.329 11948). (4) Xorrison, G. H.,‘Freiser, H., “Solvent Extraction in Analytical Chemistry,” pp. 131-3, Wiley, New York, 1957. (5) Sandell, E. B., “Colorimetric Determination of Traces of Metals,” pp, 87112, Interscience, New York, 1950. ( 6 ) Sunderman, D. K.,Ackerman, I. B., Meinke, R. W.,AXAL.CHEM.31, 40-3 (1959). RECEIVEDfor review June 20, 1960. Accepted October 5, 1960. Presented in part at the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 2, 1959.
Spectrochemical Determination of Boron in Saline Waters R. C. REYNOLDS, JR.,’ and JOHN WILSON Research Cenfer, Pan American Pefroleum Corp., Tulsa, Okla.
b A spertrochemical method is described for the determination of trace amounts of boron in saline waters. High voltage spark excitation of the natural liquid is accomplished by the rotating disk method; beryllium is used as an internal standard. The technique covers the range of 0.4 to 8 p.p.m. of boron in sea water; the standard error of the method is computed to b e about *3%. The method offers the advantages of simplicity and rapidity while maintaining adequate sensitivity and precision.
T
HE
SPECTROCHEMICAL
METHOD
described here !vas developed to provide a rapid, sensitive, and precise method for determining the boron contents of natural saline waters. Other analytical methods of adequate precision and sensitivity have been used for the determination of boron in trace
quantities. Of these, colorimetric methods are the most widely reported (8, 9, I d , 15’). Colorimetric techniques are capable of greater precision than the usual spectrochemical methods. However, spectrochemical techniques are usually more rapid and are not subject to interference by trace elements and organic materials; matrix effects are a variable, but inasmuch as these effects are associated with major constituents, suitable calibration procedures can be developed that allow the matriv effects to be evaluated. Other spectrochemical procedures (11) for the determination of boron in brines are designed to provide for determination of additional elements. It is difficult to select a n internal standard that is suitable for such a multipurpose technique; one element must serve as a n internal standard for several elements, all of which have different excitation and
emission characteristics. Furthermore, the excitation conditions used must produce usable spectra of several elements. These conditions do not necessarily represent optimum stability and intensity of the emission of any one of the elements being determined. The technique reported here is concerned only with the determination of boron. The internal standard (beryllium) was selected because i t \Tas known to be particularly suited to boron analysis (IO). The excitation conditions 17-ere adjusted to piovide optimum stability and intensity of boron emission. The resulting technique provided a standard error of 2.8%, which is comparable to colorimetric methods (9).
Present address, Department of Geol-
ogy, Dartmouth College, Hanover, N. H. VOL. 33, NO. 2, FEBRUARY 1961
247
Table I.
Operating and Photographic Conditions
Electrode rotor speed Electrodes Electrode spacing Source Exposure Transmission Grating doors Slit width Film
30 r.p.m. UCPa 106 disks, 2021 mandrels, and 100U counters 3 mm.
High voltage a x . spark Et 240 volt input and 120 Kh. 40 sec. 40 c7,
W J e open
50 microns Kodak Spectrum Analysis X o . 2 3 min. at 20" C. with Development automatic agitation Developer Kodak D-19 4 min. a t 20' C. in Fixer Kodak rapid liquid fixer ARL film drier; 2 min. Drying with heat lamp and blower, 1 min. with blower only. a United Carbon Products Co., Inc.
Table II. Calculation of Standard Error for 3.0 P.P.M. of Boron Standard
DeviIBZ~B/ ations, IB~Z~W d d z x lo3 0,740 0.019 0.361 0.775 0.016 0.256 0.002 0.004 0.757 0.734 0.025 0.625 0.764 0.005 0.025 0 . 720 0 039 1 ,521 0 768 0 009 0 081 0 765 0 006 0 036 0 780 0 021 0 441 0 749 0 010 0 100 0 047 2 209 0 806 0 . 766 0.007 0.049 0.744 0 015 0 225 0 755 0.004 0.016 .4v. 0 759 E d 2 = 5 949 X
=
~ ~ 1x . 10-3 4
=
0.021
Std. error = 0.021/0.759 X 100 a Given by Ahrens ( 1 ) .
=
2.8Q0
INSTRUMENTATION
Applied Research Laboratories equipment was used throughout. The analyses were performed using a 2-meter grating spectrograph with a PaschenRunge-type mounting. The 24,400 lines-per-inch grating provided a dispersion of 5.2 A. per mm. in the first order. First order spectra were used; all spectra were recorded on film. Excitation was accomplished by a High Precision source unit. Because of the high potential required to excite boron, the high voltage section of this unit was used. The source provided an alternating current spark (four discharges per cycle-Le., 240 discharges per second) damped by an inductance of 720 ph.; the input was adjusted to 240
248
ANALYTICAL CHEMISTRY
2 PARTS PER M',LIOL
1
1
1
4
6
8
I
-BORCl
Figure 1 . Analytical curve for determination of boron in sea water
volts. These conditions provided a peak output of 20,000 volts and 8 amperes. The film was developed and fixed in an automatically agitated developing tank that had temperature stability to within = t 0 . 5 O C. Drying was carried out in a film drier. Line densities were measured with a projection comparitordensitometer. PROCEDURE
Working Curve Standards. Standards for t h e analysis of sea n a t e r were prepared t h a t contained artificial sea water (7) as a matrix. Boron, as boric acid, was added to the matrix to provide a solution containing 4.0 p.p.m. of boron. The 4.0-p.p.m. boron stock solution was then diluted with distilled water to provide standards containing 0.4, 0.75, 1.0, 1.5, and 3.0 p.p.m. of boron. A solution of the matrix with twice the salt content of sea water was prepared and boric acid was added to give 8.0 p . p m of boron. The solutions were acidified 15-ith a few drops of concentrated HCl; BeClz was added to serve as the internal stnndard (IO). Care should be taken to allow for the extreme toxicity of soluble beryllium compounds. Two milliliters of 50-p.p.m. beryllium chloride solution were added for each 10 nil. of standard. The preparation procedure described above yields a series of qtandards in which the salt content (matrix) varies directly n i t h the boron content. The matrix was varied with the boron because the boron in sea water was expected to follow the salinity (4). Because the boron did follow the salinity, the accuracy of the technique n a s improved; the standards 11-ere almost identical to the sea n-ater unknonns in every respect. Sample Analysis. Place 10 ml. of untreated sea water in a polyethylene bottle. Add 2 ml. of BeClz solution (containing 50 p.p.ni. of beryllium) and 1 drop of concentrated HC1.
The HC1 IS added to prevent hydiolysis and precipitation of t h e beryllium. Transfer 1 ml. of t h e sample to a Coo1 S o . 2 porcelain combustion boat and analyze it on the emission spectrograph by the rotating disk method (3). The operating and photographic conditions are li-ted in Table I. Use the first-order boron line a t 2497.7 A. as the analysis line, and an unresolved beryllium triplet a t 2494 ab the internal standard line. Adjust trnnmission and exposure conditions so that the transmission of the beryllium triplet is about 2OYo.(see Table I). Run all samples in duplicate and average the ratios 1B2498 ' I ~ ~ 2 4 9for 1 the tmo to provide the final intensity ratio. Refer this average ratio to the analytical curvc to convert to parts per million of boron. Approxiniately l ' ' ? hours are required for the analysis in duplicate of eight samples. Sixteen individual samples may be analyzed in the same time with a slight loss in precision. RESULTS
Figure 1 is a plot of the analytical curve constructed from analyses of the standards. The numbers above individual points on the curve refer to the number of individual analyses that were averaged to provide each point. The curve is linear and has a slopc of one from 0.75 to 8.0 p.p.ni. of boron. The average values for the standards definr the curve closely without scatter. The analytical procedure provides a close relation between the boron contents of the standards and the experimentally measured parameter I B Z ~ ~1 ~~ S~ 2 4 ~ 4 . The precision of the method was coniputed from analyses of the 3.0-p.p.111. boron standard. Fourteen separate analyses of this standard, obtained on seven filnis over a 6-month intprval, were used in computing the standard error. The standard error was computed to be
x 2.S yo for a single determination. The data and calculations are presented in Table 11. The standard error, calculated from rc,plicatc analyses of one standard, is asslimed to represent the standard error for the analysis specimens. This assumption is not warranted in powder spectrochemical procedures. However, the authors believe that in the solution twhnique discussed here, the standards a r t virtually identical t o the unknowns. Therefore replicate analyses of the 3.0-p.p.m. boron standard are considered to provide as accurate a statement of precision as could have been provided by replicate analyses of one analytical specimen. The boron contents of 34 water samples from the Gulf of hIexico are given by Frederickson and Reynolds ( 2 ) . The salinities were estimated by referring measured water densities corrected
for temperature, to published densitysalinity curves (6). Boron data were taken using the technique described here. The data make it possible to assess the accuracy of the boron determinations. Harvey (5) gives 4.55 p.p.m. of boron as the accepted value for sea water of a salinity of 36.00 parts per mil (%OO). The data of Frederickson and Reynolds ($) indicate a value of 4.65 p.p.m. boron a t 35.00 %OO salinity. Apparently the accuracy of the method is as good as the precision, both providing a standard error of approximately =k 3%. LITERATURE CITED
(1) Ah,yns, L. H., “Spectrochemical Analysis, p. 107, Addison-Wesley Press, Cambridge. 1950. (2) FrederiTkson, A. F., Reynolde, R. C , 0 2 1 a n d Gas J. 5 8 , 154-8 (1960). (3) Harvey, C. E.. “Spectrochemical
Procedures,” pp. 323, 346, Applied Research Laboratories, Glendale, Calif ., 1950. (4) Harvey, H. W., “Chemistry and Fertility of Sea Waters,” p. 3, Cambridge University Press, London, 1955. (5) Ibid., p. 4. (6) Ibid., p. 128. ( 7 ) Ibid., p. 147.
(8) Hatcher, J. T., Wilcox, L. V., ANAL. CHEM.22,567-9 (1950). (I)) Kuemmel, D. F., Mellon, hl. G., Ibzd., 29, 378-82 (1957). (10) Melvin, E. H., O’Connor, R. T., IND.ESG. CHEX.,ANAL. ED. 13, 520 (1941). (11) Russell, R. G., ANAL. CHEIl. 22, 904-7 (1950). (12) Sandell, E. B., “Colorimetric Determination of Traces of Metals,” pp. 264-5, Interscience, New York. 1950. (13) Welche;, F. J., “Organic Analytical Reagents, Vol. 2, p. 336; Vol. 4, pp. 432, 465, 539, Van Nostrand, Princeton, N. J., 1947. RECEIVEDfor review August 4, 1950. Accepted September 23, 1960.
Precise Difference Method for the Spectrophotometric Determination of Cerium(lV) at Micromolar Concentrations 1. A. BLATZ fos Alamos Scientific laboratory, University o f California, 10s Alamos,
b In a cation resin exchange study of the sulfate ion complexes of cerous ion, a precise analytical spectrophotometric method for determination of cerium in the range 5 to 12 X 10-6M was necessary. Existing methods were modified by introduction of a difference method involving reduction of cerium (IV) to cerium(ll1) and studies were made of the factors influencing precision, such as suspended material, presence of organic (reducing) material, purification of reagents and solutions, cell constants, stray light of the spectrophotometer, interfering substances, optimum concentrations of all reagents, time factors, and blank corrections. The molar absorptivities a t 320, 3 4 0 , and 3 6 0 mp were 5.92 X lo3,5.12 X lo3, and 3.52 X lo3. Reproducibility checks revealed a probable error of a single determination of about 0.25% a t each wave length. Iron, thorium, uranium, neodymium, zirconium, nitrate ion, and excess persulfate ion do not interfere. Other possible interferences are discussed. Beer’s law is obeyed within the limits of the precision obtained.
A
for the spectrophotometric determination of cerium (IV) was suggested by Sandell (4) in 1944. It was later developed by PROCEDURE
N. M.
Freedman and Hume (1) and then, with modifications, by Medalia and Byrne (a). I n this paper the procedure has been further developed and modified to give a highly reproducible and precise method for the spectrophotometric determination of cerium in the micromolar concentration range. MATERIALS AND APPARATUS
Absorption measurements were made Kith a Beckman Model DU spectrophotometer that had a n exit slit width of 0.3 mm., a hydrogen lamp, and 10-em. silica cells. The distilled water used throughout was 0.1 to 0.3 expressed as “p.p.m. as NaC1,” as measured by a Barnstead Model PM-2 uuritv meter. Analytical grade, concentrated sulfuric acid was heated with sodium dichromate for about 1 day, distilled under reduced pressure, and stored in previously cleaned bottles. Analytical grade 70% perchloric acid was heated for 2 days and distilled under reduced pressure. The analytical grade concentrated nitric acid was used as received. Cerous perchlorate was prepared as described (3, 8). The final stock solution was 0.3M in perchloric acid and 0.1265 rfi 0.0003M in cerous perchlorate as determined by precipitation from measured volumes with saturated ammonium oxalate and heating to a constant weight of ceric oxide over Meker
burners. All solutions were prepared from this stock solution by dilution using 50-ml. Greiner, Teflon plug burets, precision Norniax pipets, and volumetric flasks. The stock sodium perchlorate solution was obtained by neutralization of Mallinckrodt analytical grade sodium carbonate with 70y0perchloric acid. Baker analyzed reagent grade ammonium persulfate was used. Generally, 250 ml. of a stock solution containing 0.20 gram per milliliter were prepared every 2 months, since it could be stored in a refrigerator a t 4’ C. for this length of time with practically no change in concentration. Solutions were purified by the use of the absorbent carbon Spheron Grade 6 (Godfrey L. Cabot, Inc., Boston, bIass.) from which fine particles mere removed by 30 to 40 swirlings with distilled water. This carbon was first tested with both distilled and conductivity water. Tests indicated that all the optical absorbers whichcould be removed by the carbon were removed in the first hours and that the Spheron-6 itself did not contribute any optical absorbance. The same results were obtained with a very pure (6) graphite, special spectroscopic graphite grade SP-1 (National Carbon Co., New York, N. y.)) which has a much smaller surface area (6) per gram than Spheron-6 and removed the optical absorbing impurities more s l o ~ l y . The stock salt solutions were usually kept in contact with the Spheron-6, VOL 33, NO. 2, FEBRUARY 1961
249
