cipitate in the presence of Hz02under these conditions. General Applications. Below are listed results of severitl titrations with sulfatocerate using Ru(DMBP) as fluorescent indicator: FeC2H4(NH&(SOJ2-4HzB (Oesper's salt); meq. taken 0.4463; found (av. oj' 5 titns.), 0.4464; rel. std. dev., 0.08%). VOSO,; meq. taken, 0.2230; found (av. of 6 titns.), 0.2234; rel. std. dev., 0.47%. H202; meq. taken, 0.2197; found (av. of 3 titns.), 0.2195; rel. citd. dev., 0.1401,. Oesper's salt was E ~ titrated O with standard permangana#te; meq. taken, 0.4197; found (av. of 5 titns.), 0.4204; rel. std. dev., 0.22%.
Application to Microdetermination
of Arsenic. The sensitivity of the fluorescent end point is well suited to microtitrations. Using the procedure previously described, the following results were obtained for the titration of 3 arsenite samples: blank; 0.11, 0.10, 0.11 ml.; total ml. cerate required, 4.795, 4.795, 4.810; peq. arsenite taken, 4.000; found 3.996, 4.005, 4.009. Therefore quantities of arsenic as small as 0.15 Ing. can be titrated with an accuracy of 0.1%. LITERATURE CITED
( 1 ) Dikshitulu, L. S. A,, Rao, G. G., Talanta 9, 289 (1962).
(2) Dwyer, F. P., Proc. Roy. SOC.New South Wales 83, 134 (1949). (3) Geyer, R., Steinmeteer, H., Angew. Chem. 72, 634 (1960). (4) Goto, H., J . Chem. SOC.Japan 59, 1357 (1938); C.A. 33, 2062 (1939). (5) Hume, D. H., Kolthoff, I. M., J . Am. Chem. SOC.65, 1895 (1943). (6) Sagi, S., Rao, G. G., Talanta 5 , 154 (1960). (7) Smith, G. F., Richter, F. P., TND. ENG.CHEM.,ANAL.ED. 16, 580 (1944). (8) Steigman, J., Birnbaum, N., Edmonds, S.M., Zbid., 14,30 (1942). (9) Veening, H., Brandt, W. W., ANAL. CHEM.32, 1426 (1960). RECEIVEDfor review October 17, 1963. Accepted December 19, 1963. Presented in part before the Division of Analytical Chemistry, J45th Meeting, ACS, New York, N. Y., September 1963.
Chemiluminescent Indicator Titration of Cadmium with Potassium Ferricyanide FREDERIC KENNY, RUSSEL 8. KURTZ, ALICE C. VANDENOEVER, CAROLE J. SANDERS, CAROL A. NOVARRO, LUCIA E. MENZEL, ROSEMARY KUKLA, and KATHRYN M. McKENNA Hunter College o f The! City University of New York,
b In this new ond rapid determination of cadmium, the chemiluminescent property of siloxene indicator and a photomultiplier photometer are employed in titrations of cadmium with potassium ferricyanide. The photometer is previously standardized against cadmium solutions of known concentration. Metcils forming precipitates with ferricyanide interfere. Nitrate solutions of cadmium are generally used. However, chloride sobtions can be employed, if the chloride concentration is approximately the same in both the standardization sample and the unknown.
I
titration, cadmium is precipitated as ferricyanide: 3Cd+a 2Fe(CN)6-3 Cd3[Fe(CN)8]* N THIS
+
-
and hence, in an ideal titration, 30.00 ml. of 0.100OM cadmium solution containing 0.3372 gram of cadmium would require the use of 20 00 ml. of 0.1000M ferricyanide solution A 0.1OOOX ferricyanide solution was employed in this work. Siloxene indicator, prepared as directed by Kenny et al. (11) except that the washing with ether was omitted, was added in a 100-mg. portion prior to each titration. This indicator emits light in the presence of exeess ferricyanide because cf the increase of potential in the solution. The effect of potential change on light evolution has been discussed ((?,9). The conditions in the solution, after each increment of ferricyanide and at
695 Park Ave., New York, N. Y.
the stoichiometric point, must be as near to equilibrium as possible. By stirring rapidly and always a t the same rate, and adding the ferricyanide slowly a t a constant rate, an attempt was made to give the reaction the time and opportunity needed to approximate equilibrium conditions closely. Titration rates slightly under 4.0 ml. per minute were found to be satisfactory when the solution was rapidly stirred a t a constant speed with a mechanical stirrer. Samples requiring approximately 20.00 ml. of ferricyanide could be titrated in slightly over 5 minutes. When conditions approximating equilibrium are maintained, the change of intensity of light evolution becomes a function of the change of potential in the solution (10) and, hence, a t the stoichiometric point an end point can be observed. The pronounced change in light intensity a t the stoichiometric point was observed by means of a Photovolt Multiplier Photometer 520A set a t its highest sensitivity. This instrument a t this setting is sensitive to less than 0.002 microlumen. X 0.1OOOM cadmium solution was prepared by dissolving 22.482 grams of C.P. cadmium in approximately 150 ml. of 8 M nitric acid and quantitatively diluting to 2 liters. Since intensity of light emission of the indicator depends in part on the pH of the solution, care was taken to use cadmium solutions having roughly the same pH-approximately 0.5. A 0.10CUM potassium ferricyanide solution was prepared by drying the C.P. salt a t 110' and dissolv-
ing 64.830 grams quantitatively in 2 liters of solution. The photometer was standardized by titrating known amounts of cadmium in three or four samples of 0.1000M solution. The average deflection of the instrument was used as the end point deflection in subsequent unknown titrations. The response of siloxene indicator is not instantaneous, but increases with increase of potential. Under the conditions of the titration a faint light may appear prior to the stoichiometric point. Because of the gelatinous precipitate formed in the reaction, some of this light may be obscured. Furthermore, the potential jump a t the stoichiometric point is relatively small compared, for instance, with the jump for the titration of iron with ceric sulfate. Hence, it was considered preferable to determine the deflection experimentally a t the stoichiometric point rather than attempt to locate with exactness the point of maximum rate of change of light emission. Since, as pointed out (11), the indicator loses light-emitting capacity with time, any batch of siloxene indicator used in a standardization must be used on the same day in the corresponding unknown titrations. The usable life of a batch of indicator has been found in this and previous researches to be a t least 3 days. All samples in this research had a pH, prior to titration, of approximately 0.5. The unknown samples of Table I VOL. 36, NO. 3, MARCH 1964
529
Table I.
Cadmium Samples in 0.1 OOOM Solution Titrated with 0.1 OOOM Potassium Ferricyanide
Cadmium present, g.
0.3372 0.3372 0.3372 0.3372 0.3035 0.3035 0,3038 0.3035 0.3709 0,3709 0.3709
Table II.
(Starting pH 0.5, end point deflection 29.0) Potassium Cadmium Deviation ferricyanide, found, from mean, ml. g. p.p.t. 20.00 0.3372 2.5 19.90 0.3355 2.5 19.99 0.3371 2.0 19.90 0.3355 2.5 Av. 19.95 0.3363 2.4 18.03 0.3040 5.6 17.83 0.3006 5.6 17.93 0.3023 0.0 17.93 0.3023 0.0 Av. 17.93 0.3023 2.8 22.09 0.3724 0.9 22.03 0.3714 3.6 22.20 0.3743 4.1 Av. 22.11 0.3727 2.9 Av. 2 . 7
Interference
(20.00 ml. of 0.1000M potassium ferricyanide added t o 30.00 ml. of 0.1M metal ion solutions) Ion forming Ion producing precipitate clear solution Ag +
m: ;* Pb
Bi +3 c u +2 Hg+i' Zn
NH4 + K+ Ns+ Mg +Z Sr +2 Ca+2 Ba +2 Cr +3
Error, p.p.t.
2.7
3.9
4.8 3.8
concentration of approximately 0.1-If. The resulting solutions were then titrated with 0.1000M potassium ferricyanide, using siloxene indicator (Table 111). To keep the volumes of the standardization samples the same as those of the unknown samples, amounts of distilled water were added to the unknowns, prior to titration, equal to the volumes of solution of the second metal used in the corresponding unknown samples. RESULTS, DISCUSSIONS, AND CONCLUSIONS
SbCls-
range from 10% smaller to 10% larger than the known samples. Any ion which forms a precipitate with ferricyanide under the conditions of the titration would be expected to interfere with the determination of cadmium. Furthermore, ions not forming precipitates may interfere for other reasons and need investigation. When 20.00 ml. of 0.1000M potassium ferricyanide was added to 30.00 ml. of 0.1M solutions of 23 metal ions, precipitates were formed in some cases, and clear solutions resulted in others (Table 11). I n the experiments reported in Table I1 nitrate solutions were used except where chloride is indicated. All metal ions producing a precipitate were considered to be interfering and no further work was done on them. The metal ions which produced no precipitate were added separately in 0.5M concentrations to 30.00 ml. of 0.1OOOM cadmium solutions, in amount sufficient to give the added metal a 530
0
ANALYTICAL CHEMISTRY
The results of Table I indicate that cadmium was determined with an accuracy ranging from 2.7 to 4.8 parts per thousand, and a precision ranging from 2.4 to 2.9 parts per thousand. Although each titration required about 5 minutes, the over all time for each sample was about 10 minutes, Cadmium could be determined in 11 samples similar to those shown in Table I in less than 2 hours. To this must be added about 20 minutes for the preparation of the indicator and 40 minutes for standardizing the photometer. This brings the entire time, exclusive of preparation of the sample, to about 2 hours and 50 minutes. However, a batch of indicator, once prepared, can be used for a t least 3 days. As shown in Table 11, interference due to precipitation was encountered in solutions containing silver, mercurous ion, lead, bismuth, copper, mercuric ion, zinc, ferric ion, cobalt, nickel, manganous ion, and antimony ion. KOfurther work was done on these ions. The ions listed in Table I1 which did not produce precipitates when cadmium
was titrated with ferricyanide yielded successful titrations, with the exception of tin (Table 111). In the titrations of Table 111, the accuracy ranged from 0.3 to 9.8 parts per thousand and the precision ranged from 0.5 to 7.9 parts per thousand, Although the tin ions listed in Table I1 did not produce a precipitate, successful titration was impossible because light emission started long before the stoichiometric point was reached. The tin solution contained a high concentration of chloride, whereas all the solutions successfully titrated up to this point had contained no chloride. Furthermore, this solution was much stronger in acid than any of the other solutions. Consequently, further study of the effect of chloride, acidity, and possibly other ions was considered desirable. The amount of light a t the stoichiometric point may be affected by factors influencing the ferricyanide-ferrocyanide potential and factors influencing the light-evolving properties of the indicator. The effect of increased hydrogen ion concentration on the ferricyanideferrocyanide potential is to increase the potential, probably by repression of the ferrocyanide ion ( l a ) ,bringing about an early end point. Furthermore, the light-evolving property of siloxene indicator is enhanced (10) by an increase in hydrogen ion concentration. The early end point obtained with tin could be due, a t least in part, to a combination of these two effects, and, therefore, it was considered desirable to run all the titrations of Tables I to I11 a t approximately the same acidity. I n the case of tin, the high chloride concentration might well have been a factor in the light-evolving properties of the indicator. It was decided, therefore, to compare the effect of nitrate and chloride on light evolution a t the stoichiometric point. Cadmium chloride, when titrated to the stoichiometric point, produced a photometer deflection of 166, compared with 31.5 for a similar titration of cadmium nitrate. This, of course, could lead to serious error in titrating an unknown cadmium sample containing chloride. Any factor which increases light evolution a t the stoichiometric point in the titration of an unknown but is absent in the standardization, will cause the end point to be reached too soon and will yield low results. As shown in Table 111, cadmium can be determined satisfactorily when the solution is 0.1OM in magnesium nitrate. The light evolution for this titration was examined over magnesium nitrate concentrations varying from 0 to 0.20M. No substantial change in light evolution was observed.
When magnesium (chloride was substituted for the nitrate, progressively greater deflection was obtained, increasing from 15.2 to 54.0scale divisions. This can be attributed to the effect of the chloride ion. As a further means of contrasting the effects of nitrate and chloride ions on light evolution at the stoichiometric point, two series of titrations were carried out, in the pre,jence of increasing concentrations of (1) potassium nitrate and ( 2 ) potassium chloride. In both series the concentralions were varied from 0 to 0.20M. In the nitrate series a gradual increase in dctflection from 13.1 to 25.8 was observed, whereas in the chloride series the vitriation was from 31.5 to 150.0. I t is obvious that chloride ion increases light evolution a t the stoichiometric point. T o avoid error in the determination of cadmium in samples containing chloride, the photometer was standardized with cadmi I m samples containing approximately the same concentration of chloridl: as the unknown sample. The results shown in Table 1V are accurate in spite of the presence of chloride. The chloride content of the unknown must be estimated roughly before cadmium determination can be undertaken. A possible explanation of the difference in behavior of chloride and nitrate ions may be that both anions form saltlike compounds with the siloxene indicator base. Some other salt-forming acids are hydrobromic, hydriodic, sulfuric, and phosphoric (1-7, 9). The nitra:e, because it contains oxygen atoms, inay be reduced by the adjacent Si-Si bonds (2, 3, 6, 7 , Q), the activation energj, for this reduction being supplied by the primary oxidation (electron removal) of the indicator by ferricyanide, to put the chemiluminescent 6-silicon rings into higher energy states. The chemiluminescence would be quenched by using itr; energy to activate nitrate reduction. This differs from quenching of fluorescence, which does not necessarily involve a chemical reaction, The aotiveted Si-Si bonds would, thereafter, 3e oxidized (oxygenated) by the adjacent nitrate group. There would thus be a destruction of the chemiluminescent 6-silicon ring by forming Si-O-Eii bonds and, moreover, energy which would otherwise have been liberated as chemiluminescence now would go to activate nitrate reduction. Chloride ion adjricent to activated Si-Si bonds would not be affected by the energized silicon orbitals, because chloride is not an oxidizing anion like nitrate. In fact, by displacing; the hydroxy group of the indicator “base” in forming a chloride salt, the activated 6-silicon ring would no longer be oxygenated
Table 111.
Titration of
30.00 MI. of 0.1OOOM Cadmium Nitrate with 0.1OOOM Potassium Ferricyanide
(0.3372 gram of cadmium present) Dev. End Potassium from point ferriStarting deflec- cyanide, Cd found, mean, Error, ml. gram p.p.t. p.p.t. Solution added, ml. pH tion Ammonium nitrate 7.5 ml., 0.58 53.0 19.78 0.3339 3.3 0.5M 19.70 0.3321 7.5 20.15 0.3399 15.8 19.75 0.3330 4.8 Av. 19.85 0.3346 7.9 7.7 Potassium nitrate 7.5 ml., 0.51 18.5 19.90 0.3355 1.2 0.5M 19.95 0.3364 1.5 19.90 0.3358 1.2 19.93 0.3360 0.3 Av. 19.92 0.3359 1.1 3.9 20.08 0.3385 3.6 43.3 0.51 Sodium nitrate 7 . 5 ml., 0.5M 19.82 0.3342 9.2 20.02 0.3375 0.6 4.7 20.10 0.3389 Av. 20.01 0.3373 4.7 0.3 Magnesium nitrate 7.5 ml., 0.50 16.9 20.15 0.3397 2.9 0.5M 20.10 0.3389 0.4 20.11 0,3390 0.9 20.02 0.3374 3.4 Av. 20.10 0.3375 1.9 0.8 Strontium nitrate 7.5 ml., 0.50 16.0 19.91 0.3367 0.5 0.5M 19.93 0.3360 0.5 19.91 0.3357 0.5 19.93 0.3360 0.5 Av. 19.92 0.3359 0.5 3.8 Calcium nitrate 7 . 5 ml., 0.5M 0.50 16.9 2u. 15 0.3397 0.4 20.10 0.3389 1.9 20.15 0.3397 0.4 20.15 0.3397 0.4 Av. 20.14 0.3395 0.8 6.8 Barium nitrate” 11.67 ml., 0.50 16.0 20.00 0.3372 1.5 0.33M 20.00 0.3372 1.5 19.90 0.3355 3.5 19.99 0.3370 0.5 Av. 20.00 0.3367 1.7 1.4 Chromium nitrate 7 . 5 ml., 0.45 8.6 19.99 0.3370 7.4 0.5M 20.15 0.3399 1.2 20.28 0.3419 7.1 20.12 0.3392 0.9 Av. 20.13 0.3395 4.2 6.8 Ammonium nitrate 7.5 ml., 0.45 30.0 19.71 0.3323 7.2 0.50M 19.76 0.3332 4.5 20.00 0.3372 7.5 19.94 0.3362 4.5 Av. 19.85 0,3347 5.9 7.4 Arsenic acid 7.5 ml., 0.5M 0.60 51.0 20.12 0.3392 3.5 20.18 0.3402 0.5 20.18 0.3402 0.5 20.30 0.3423 5.4 Av. 20.20 0.3405 2.5 9.8 a Since the solubility of barium nitrate is only 0.33M, 11.67 ml. was employed to obtain approximately 0.1M concentration in sample to be titrated.
Table IV.
Titration of
30.00 MI. of 0.1OOOM Cadmium Nitrate with 0.1000M Potassium Ferricyanide
(0.3372 gram of cadmium present) End Potassium point ferriStarting deflec- cyanide, Solution added pH tion ml. Magnesium chloride 7 . 5 ml., 0.60 71.2 20.10 0.51%f 20.07 20.02 20.00 Av. 20.05
Cd found, gram 0.3389 0.3384 0.3379 0.3372 0,3381
Dev. from mean, Error, p.p.t.
p.p.t.
5.0 3.5 2.0
0.0
2.6
VOL. 36, NO. 3, MARCH 1964
2.9
531
(adversely) by the hydroxy group. Increasing the ’chloride would displace increasing numbers of hydroxy groups and give increasing protection and increasing possibility of greater light emission. Kolthoff and Tomsicek (13), have pointed out that the ferricyanideferrocyanide potential may be affected by the presence of salts. They found that the effect was virtually independent of the type of anion. Thus, potassium nitrate and potassium chloride, and sodium nitrate and sodium chloride have identical effects on the potential a t the same ionic strength. The effects produced by cations, however, differ considerably. For alkali cations it decreases in the order Cs, Rb, K, Na. This is attributed to the degree of dissociation of the corresponding ferrocyanides. A larger effect was observed with divalent cations such as the alkaline earths. Inasmuch as the chloride and nitrate anions had distinctly different effects on
light emission a t the stoichiometric point, and sodium, potassium, and the alkaline earths produced no substantial difference, it must be concluded that in determining cadmium, the lightemitting properties of the siloxene indicator far outweigh the possible changes in ferricyanide-ferrocyanide potential as a determining factor in the extent of light emission a t the stoichiometric point. The results included in Tables I to I11 indicate that this new and rapid method for the determination of cadmium permits the presence of ammonium, potassium, sodium, magnesium, strontium, calcium, barium, chromium, aluminum, and arsenate ions. The titration is best performed in a nitrate solution. Interference by chloride ion can be eliminated by standardization of the photometer in the presence of approximately the same chloride concentration as in the unknown solution.
LITERATURE CITED
(1) Ephraim, Fritz, “Inorganic Chemistry,” p. 835, Interscience, New York, 1947. (2) Kautsky, H., Kolloid 2. 1 0 2 , 9 (1943). (3) Kautsky, H., 2. Anorg. Chem. 117, 217 (1921). (4) Kautsky, H., Gaubatz, E., Ibid., 191, 388 (1930). (5) Kautsky, H., Herzberg, G., Ibid., 139, 142 (1924). (6) Kautsky, H., Hirsch, A., Ber. 64, 1614 (1931). (7) Kautsky, H., Thiele, H., 2. Anorg. Chem. 173, 118 (1928). (8) Kenny, F., Trans. N . Y . Acad. Sci. 16, 394 (1954). (9) Kenny, F., Kurtz, R. B., ANAL. CHEM.22, 693 (1950). (10) Kenny, F., Kurtz, R. B., Ibid., 25, 1550 (1953). (11) Kenny, F., Kurtz, R. B., Beck, I., Lukosevicius, I., Ibid., 29, 543 (1957). (12) Kolthoff, I. M., Furman, W. H., “Potentiometric Titrations,” 2nd ed., p. 318, Wiley, New York, 1931. (13) Kolthoff, I. M., Tomsicek, W. J., J. Phys. Chem. 39, 953 (1935). RECEIVEDfor review July 16, 1963. Accepted November 26, 1963.
Analytical Applications of the Flame Spectra of the Rare Earth Elements and Scandium ARTHUR P. D’SILVA, RICHARD N. KNISELEY, VELMER A. FASSEL, RONALD H. CURRY,’ and ROBERT B. MYERS Institute for Atomic Research and Department of Chemistry, Iowa Stafe University, Ames, Iowa
b The analytical potentialities of the simple spectra emitted by the rare earth elements when ethanol solutions of their perchlorates are aspirated into fuel-rich, oxyacetylene flames are evaluated. The sensitivities of detection of the strongest lines are tabulated and data showing the application of these spectra to the analysis of complex rare earth mixtures are presented. The relative standard deviations for these analyses ranged from k l . 3 to 3.7% of the amount present.
T
relatively intense atomic line spectra emitted by the rare earth group of elements, which include scandium, in accordance with IUPAC recommended nomenclature (6), when ethanol solutions of their perchlorates are aspirated into fuel-rich, oxyacetylene flames were described in a recent communication ( 3 ) . The ultimate analytical utility of these spectra will be determined primarily by the detection sensitivities of the most sensitive lines and the degree of interference they enHE
Present address, Sperry Rand Research Center, Sudbury, Mass. 532
ANALYTICAL CHEMISTRY
counter. Both factors have now been evaluated, and the data are presented in this communication. One of the most appealing analytical applications of these spectra is the analysis of rare earth mixtures, since classical chemical methods cannot in general be used here. Although various emission and absorption spectrometric techniques have been successfully applied to this problem ( Z ) , it is apparent that serious deficiencies exist in their capabilities for providing accurate and complete analyses of all types of mixtures in a simple manner. The line spectra emitted in fuel-rich, oxyacetylene flames possess several distinctive advantages over other spectra heretofore used for the analysis of rare earth mixtures. First, the spectra are striking in their simplicity, compared to the arc or spark spectra. It is therefore possible to achieve adequate spectral dispersion with small table-model spectrometers, whereas large, high dispersion spectrographs are required when arc or spark spectra are observed. Secondly, all of the rare earths except cerium exhibit line or band spectra in the flame of sufficient intensity to possess analytical utility. The group of rare earth elements which
evade detection by the optical absorptiometric technique ( 2 ) are readily observed in the flame spectra. Finally, the results reported in the present communication show that there is no evidence of interelement effects and that line interference, even in small monochromators, is rarely a serious problem. This contrasts sharply with the selective enhancement and absorption effects observed in x-ray fluorescent spectrometric measurements ( 2 ) . EXPERIMENTAL
Apparatus. The experimental facilities and operating techniques have been described ( 3 ) . For both the detection limit determinations and quantitative analyses, the scanning speed was reduced to 5 A. per minute. For this scanning speed, a chart speed of 1 inch per minute was used. The recorder response, photomultiplier tube voltage, and amplifier sensitivity were adjusted to optimal settings for the sensitivity determinations so that maximal signal to noise ratios were obtained. For the quantitative measurements a t higher concentrations, appropriate amplifier sensitivities and photomultiplier tube voltages were selected as discussed below. The sample aspiration rate for the
