in such practical cases and values of G should be accompanied by a statement of the particular plotting technique used for the determinations. Reed and nerkson made a suggestion that is currently very pertinent to the nomenclature of emulsion response plots. They gave several examples, none from the field of spectrography, of forms of the logistic function in use under a variety of specialized names, and strongly urged that scientists should learn to recognize forms of the logistic function and rcfcr to them by the generic term “logistic.” The technique of making a logarithmic-logistic response plot of E and T by plotting log E and the ordinate log (1/T - 1) was first presented by Baker ( 1 ) . Common use of the symbol A for log (1/T - 1) by both I3aker and Seidel has suggested strongly to the author (4) that Seidel, who frequently is given credit for the technique, merely retrieved Ijakcr’s innovation from the literature and did not make even an independent rediscovery. The equations given by Hull (@) for the ion-response of Q-2 plates provided, for the first time, forms of the logistic function that gave a good fit
with response data at low transmittances approaching a T, that differed appreciably from zero. If densitometer sensitivity controls are arbitrarily adjusted to refer transmittances to the processing fog level as unity, then these equations become equivalent to integrated forms of Equation 2 and the parameter R of Hull’s work and G are algebraically the same. The author supports Fked and Herkson on terminology and recommends adoption of the usagcs logarithmic-logistic (or loglogistic) plot and logarithmic-logistic function where these entities occur in spectrographic work. Mr. Hunt is to be congratulated for the investigation of the vapordeposited silver bromide film as an ion detector. This detector shows values of T o , T,, and G that are preferable to those obtained with conventional, emulsionbased detectors. The author’s paper (7) contained only a verbal descrip tion of relative transmission without formal presentation of its algebraic definition, Equation 1 ; and confusion between the parameter G and the ded l o i ( l / T - l) rivative unfortunately d log E
has resulted. This communication is submitted to prevent propagation of confusion and to detail explicitly some prolarties of G in relation to other contrast measures for spectrographic emulsions. LITERATURE CITED
( 1 ) Baker, E. A., Proc. Roy. SOC.Edin. 45, 166 (1925). ( 2 ) Hull, C. W., Mass Spectrometry Conf., New Orleans, Paper No. 72, June 3-8, 1962. (3) Hunt, M. H., ANAL. CHEM.38, 620 (1966). ( 4 ) McCrea, J. M., “Developments in
Applied Spectroscopy,” Vol. 4, E. N. Davis, ed., p. 501, Plenum, New York, 1465 -I--.
(5) McCrea, J. M., 4th National Meeting, Society for Applied Spectroscopy, Denver, Paper No. 157, Aug. 3GSept. 3, 1965. (6) McCrea, J. M., Speclrochem. Acta 21, 1014 (1965). ( 7 ) McCrea, J. &I., 12th Annual Conf. on Msss Spectrometry and Related Topics, Montreal, Paper No. 92, June
7-12, 1964; A p p l . Speclr. 20, 181 (1966). (8) Reed, L. J., Berkson, J., J . Phys. C h a . 33, 760 (1929). J. M. MCCREA
Applied Research Laboratory United States Steel Carp. Monroeville, Pa. 15146
Polarographic Determination of Iron in Concentrated Phosphoric Acid SIR: Wet process phosphoric acid produced from Florida rock contains between 0.5 and 1.5’% FeOs, with both ferrous and ferric iron usually present. I n order to study the relationship between the iron couple and various kinetic properties of the system, it is advantageous to measure simultaneously the concentrations of both oxidation states in situ, without significantly sffecting the Fe(II)/Fe(III) ratio or the reaction of either ion with the substrate. The standard methods of analysis generally measure only Fe(II), Fe(III), or total iron concentration. A polare graphic method, originally described by von Stackelberg and von Freyhold (S) and later by Lingane (2), was applied by Doumas ( 1 ) to the determination of iron in phosphoric acid. Although the method measures the concentrations of both ions, the procedure involves a dilution of the sample, complexation of the iron with oxalate, and addition of potassium chloride to reduce the migration current. Concentrated phosphoric acid, per se: is in fact an ideal solvent for the polarographic analysis of dissolved iron in that it provides a self-contained s u p porting electrolyte and a complexing
medium in which the anodic wave for ferrous iron is as well defined as the cathodic wave for iron 111. These reactions are reversible at the dropping mercury electrode, and the contaminants which commonly occur in wet process phosphoric acid, including aluminum, do not interfere with the polarographic waves for iron. EXPERIMENTAL
Apparatus. A Sargent Model X X I Recording Polarograph was used with a 10-ml. Heyrovsky cell containing a mercury pool anode. The open circuit characteristics of the D.M.E. were: t = 6.14 and 6.64 sec. in 41.5 and 74y0 H3Y04, respectively, and m = 0.941 mg. set.-' at h = 41.5 cm. No special precautions were taken to
Table 1.
thermostat the solutions; ambient temperature was 22” f 1” C. Reagents. Aqueous 0.5.U stock solutions were prepared from reagent grade FeS04(NH4)2S04. 6 H 2 0 and FeNHI(SOI)S.12H20. Ferrous salt solutions were maintained under a nitrogen atmosphere until final dilution with aqueous solutions of phosphoric acid. Reagent grade 85% H3P04was used in preparing the calibrating solutions. Procedure. Calibration data were obtained on a series of solutions containing 1 X 10-5 to 5 X 10-2M Fe+a and 1 X 10-5 to 5 X 10-*M Fe+2 in both 41.5 and 74% HSPO,. Solutions were deaerated with tank nitrogen and polarograms scanned from +0.5 to -1.5 volts us. the mercury pool. Anodic ferrous iron diffusion currents were measured at +0.1 to f0.2 volt
Diffusion Current Constants of Ferrous and Ferric Iron in 4 1.5 and Phosphoric Acid
Iron concn. range 0. ooo2-0.01 0. ooo2-0.05 0.00054.20 0 . 0002-0.20
-
Oxid. state +2 +2 +3 +3
HaPo,, % 41.5 74 41.5 74
74%
Av. id/C f std. dev.
(amp.-L/mol) 1441 z!L 100 531 f 18 931 f 24 479 f 24
VOL 38, NO. 10, SEPTEMBER 1966
1403
100
1
1
1
1
1
500
1
1
1
1000 DlRuiion Cuncnt Constant, m.-l/mol.
1
1
1
1
~
1500
Figure 1 . Diffusion current constants of Fe(ll) and Fe(lll) as a function of phosphoric acid concentration
to =k8% and the relative error *lo%, The method is applicable to the measurement of concentrations up to 2% total iron: 1.7% ferric iron and 0.3% ferrous iron. Obviously, higher concentrations can be determined by diluting samples with the appropriate concentration of phosphoric acid; for this work, however, no dilutions were made on test samples because of the possible effects on the ferrousjferric ratio or on the solubility of various complex ferric iron salts which can form in this medium. Because solution viscosity affects the diffusion coefficients of the iron salts, acid concentration was varied in 5% increments; and the diffusion current constants were determined for both ferrous and ferric iron a t several concentration levels. These results, shown in Figure 1, indicate that the accuracy of the measurement is essentially unaffected by a variation of less than =t2oJ, in the phosphoric acid concentration. LITERATURE CITED
(1) Doumas, B. C., Ph.D. thesis, Vir-
and the cathodic diffusion currents of ferric iron at approximately -1.2 volts. The latter potential was selected to avoid the rather severe maximum which appeared a t ferric iron concentrations greater than 0.01M. [As an alternative, the maximum can be suppressed with 1 to 2 drops of a cationic 0.5% FC-134-obsurfactant-e.g., tained from Minnesota Mining and Manufacturing Co., in 85% HaPO4 containing 1% isopropanol.]
RESULTS AND DISCUSSION
Diffusion current constants, &/C, are listed in Table I for 41.5 and 74% phosphoric acid solutions of both ferrous and ferric iron. The analyeable concentration ranges can be extended to 0.1M for Fe(I1) and 0.5M for Fe(III), with some loss in accuracy, by preparing calibration curves. The average relative standard deviation of the measured i,/C was &3%
ginia Polytechnic Institute, Blacksburg, \’a,, 1961. (2) Lingane, J. J., Chem. Revs. 29, 1 (1941). (3) von Stackelberg, M., von Freyhold, H., Z.Electrochem. 46, 120 (1940). EDWARD W. BILINSKI DONALD E. DIEBALL THOMAS P. WHALEY
Researeh and Development Division International Minerals and Chemical Corp. Libertyville, Ill. 60048
Oxygen-Tube Cornbustion Method for Liquid Scintillation Assay of Carbon-14 and Tritium SIR: In order to take advantage of the convenience and sensitivity of liquid scintillation counting (LSC), organic substances, which are insoluble or otherwise incompatible with counting fluids, are frequently converted to COZ and HzO which are compatible with some counting fluids. Oxidation trains, with slight modifications of the sample collection method, have been adapted from gravimetric and volumetric assay to radiometric assay with GeigerMuller counters. However, the limitations on daily capacity and strong “memory” effects for tritium led to searches for other methods for LSC. Two general approaches have been most successful, the sealed tube combustion of a sample at high temperatures with metal oxides and the oxygen-flask method where the sample is ignited in a large flask which has been purged of air 1404
*
ANALYTICAL CHEMISTRY
with oxygen gas. These methods are applicable to both carbon-14 and tritium and are more convenient than train methods. The sealed-tube method is less sensitive to sample preparation, and to the composition of the sample, but product collection requires more apparatus and overall requires more labor than the flask method. The accuracy of the flask combustion depends on evaluating the extent of combustion. This in turn is affected by how well the sample holds together in the region of incandescence. If parts of the sample or partial combustion products migrate from the flaming region their temperature may drop and combustion will stop. These losses are different for samples of different physical properties--e.g., polymer films compared to crystalline powders-and care must be taken to prevent losses and
ensure reproducible combustion. In the sealed-tube method the entire environment is a t combustion temperature so dispersion is not important and sample preparation is not critical. At room temperature materials having a large surface, rich in oxide, will tend to bind water firmly so that it cannot be removed easily. Sealed tube methods are subject to loss of tritium because of this. The flask method is free from this defect if sufficient time is allowed and a thorough rinsing procedure followed in collecting the combustion products. The subject of this pawr is a combination of the two methods which minimizes the disadvantages of each. EXPERIMENTAL
Apparatus. Vycor combustion tubes are made from g-mm., 17-mm., 25-mm., and 30-mm. tubing by form-