ANALYTICAL CHEMISTRY
1968 by the addition of oxalate but, in pa,rtially denatured or cloudy serum, the precipitation is incomplete. McLean and Hastings (6) have shown that a large fraction ( u p to 50%) of the calcium in serum is protein-bound. Protein which precipitates on standing a t physiological pH takes with i t part of the calcium. When trichloroacetic acid is employed for protein removal, all the calcium is left in solution. Carr (2) has shown with bovine gerum albumin t’hat there is no protein-hound calcium when the pH is lowered to 4.5 and that the binding increases as the p H is raised above this figure. Trichloroacetic acid serves a dual purpose by releasing all the protein-hound calcium while removing the protein. While protein removal ensures complete precipitation of all the calcium in serum, the estimat’ion of the calcium through the oxalate content of the precipitate has several pitfalls. T h e theoretiral composition of the precipitate itself is in doubt (IO),for i t is reported that the oxidizing agent must he standardized against a known calcium solution carried through the Pame precipitation steps as the unknown. A standardization against sodium osalate gives erroneous calcium values. The difficulties of separating calcium and mngnesiuni are discussed thoroughly in a n elementary cpantitative analysis book (11). Smit,li et al. (9) used t,he direct spectrographic metliod of Boyle et a[.(1) for calcium and magnesium to make a thorough study of the precipitate obtained when ammonium oxalate is added to blood serum. They report that as much as 15% of the oxalate content may be due to magnesium. The uncertainties which may be encountered by determining d c i u i n , through the oxalate content of the precipitate, niay be c,liminated by a direct measurement of the calcium with a flame photometer. Nosher et al. ( 7 ) measured the calcium content l)y flame photometer after first wet-ashing the serum and separating the calcium as the phospha,te. On freshly drawn serunl the calcium may be determined directly after proper dilution, iising calcium chloride standard solutions which have been compensated for sodium anti pot ium (3). E’r)i. rloudy serum sam-
ples, complete protein removal v i t h trichloroacetic acid is b r i t to liberate the calcium from the suspended material. The c d cium content of these filtrates mag be determined by flame photometer with standard solutions containing phosphate or hy use of a suitable working curve (3). For the most accurate n.oi,k i t is rrcommended t h a t the calcium be precipitated as the oxalate from the trichloroacetic acid filtrate. The oxalate precipitate need not be washed carefully, as i t has been shown ( 3 )t h a t neithcr ammonium oxalate nor magnesium will interfere in the flame photometric determination of calcium. Since sodium, potassium, and phosphate have been removed, standards containing only calcium chloride may be used. LITERATURE CITED
(1) Boyle, A. J., Whit,ehead, T., Bird, E.
S.,Batchelor, T. AI.. 1sri.i. L. T., Jacobson, 6 . D., and LIyers, G. B., J . Lab. Clin. M e d . , 34, 625 (1949). (2) Carr, C. W., Arch. Biochetn. and Biophys., 43, 147 (1953). (3) Cheri, 12. S., Jr., and Torihara, T. Y., ANAL. CHEM.,25, 1Gt3 (1953). (4) Clark, E. P., and Collip, J. B., J . H i d . Chem., 63, 461 (1925). (5) Kramer, B., and Tisdall, F. F.. Ibid., 47, 475 (1921). (6) IIcLean, F. C., and Hastings, A. B., Ana. J . Med. Sci., 189, 601 (1935).
(7) llosher. 11. E., Itano, LI., Boyle, A. !J., Myers, G. B., and Iaeri, L. T., Am. J . Clin.Pathol., 21, 75 (1951). (8) Sendroy, J., Jr., J . R i d . Chem., 152, 539 (1944). (9) Smith, R. G., Craig, P., Bird, E. J., Boyle, A. J., Iseri, L. T., Jacobson, S.D., and Myers, G. B., Am 263 (1950). (IO) Willard, H. H., and Diehl, H., “Advanced Quantitative Analysis,” p. 296, New York, D. Van Sostrand Co., 1944. ( I 1 ) Willard, H. H., Furman, K. H., and Flagg, J. F., “A Short Course in Quantitative .inalysis,” pp. 211-15, Nen- York, D. Van Nostrand Co., 1944.
R E C ~ I V Efor D review November 30, 1933. -4ccepted J u n e 30, 19.54. Based on work performed under contract with the United States Atomic Energy Commission a t the Vniversirs of Rorhester Atolnir? Energy Project, Rorhester. X. Y.
CoIo rimet ric Dete rmination of CobaIt with 2,2 ’,2 ”-Te rpy ridine RONALD R. MILLER’ and WARREN W. BRANDT Department o f Chemistry, Purdue University, Lafayette,
An extraction adaptation of the colorimetric method for the determination of cobalt using 2,2’,2’’-terpyridine has doubled the sensitivitj and offset the disadvantage of the instability of the color. \-ariatiom in pH between 2 and 10 do not affect the colored species, and Beer’s law is valid for cohalt concentrations from 0.5 to 25 p.p.m. Interference hy most of the common metals except iron and copper above 100 p.p.rn. and nickel above 30 p.p.m. is not serious. Oxidizing agents and cyanide should he absent.
T
HE cobalt-2,2’,2”-terpylidiiie complex was first recognized by it,s interference with the iron terpyridine system (3). The composition of the orange salt \vas established hy Rforgari and Burstall to contain 2 molecules oi ligand per cobalt ion ( 2 ) . Oxidation of the bromide salt with chlorine gave a yellow cobalt( 111)chloride with the same polyaininn coordination. lloss and Mellon ( 4 ) proposed n c~~lorimetric method for cobalt using 2,2’,2”-terpyridine in the pH r:iiige 2 t o 10. Beer’s law is valid for cobalt concentrations from 0.5 to 50 p.p.m., and most common metals do not interfere. Copper, nickel, iron, cyanide, and dichromate interfere. This method suffers from a limited stability of the color which hiiidors its usefulness. I
Present address, Union Oil Co , R r m Calif.
Ind. This investigation proposes a method whereby the stability of the color is no longer a limiting factor, and in some instances allows a larger amount of interfering constituents to be prewiit. EXPERIMEVTAL WORK
Solutions and Apparatus. The csolor-forming reagent Mas a 0.1% aqueous solution of 2,2’,2”-terpyridine containing enough hydrochloric acid to dissolve the reagent. A standard solution of cobalt nitrate was prepared by dissolving the desired amount of reagent grade cobalt nitrate hexahydrate in iron-free distilled water. T h e nitrobenzene mas vacuum distilled a t 1- to 2-mm. pressure. Adjustments of p H were made with 6M sodium hydroxide and 6 M hydrochloric acid. The pH was measured with a line operated Leeds and Sorthrup p H indicator. Standard solutions of the anions studied were prepared from the alkali metal salts in most cases. Sitrate, chloride, and sulfate salts of the cations were used. Each solution contained 10 mg. per ml. of the ion in question. Spectrophotometric curves were obtained with a General Electric recording spectrophotometer with a band width of 10 mp. Individual absorbance measurements were made with a Beckman Model B epectrophotometer with a band width of 5 mp. Extraction Attempts. The success of Margerum and Banks in extracting the iron-phenanthroline complex ( I ) suggested the extraction of this complex into an immiscible organic solvent as a means of making the color more stable and of avoiding some
V O L U M E 2 6 , NO. 1 2 , D E C E M B E R 1 9 5 4 of the interferences present in the aqueous system. Attempted extractions into eight organic solvents in the presence of four anions are summarized in Table I. No extraction was obtained from chloride, cyanide, or acetate solutions. Nitrobenzene appeared to be the most suitable solvent and the perchlorate anion \\as chosen because it is less likely to contribute side reactions and interferences.
Table I .
complex with hydroxylamine. Extracting this complex into isoamyl alcohol might raise this limit even higher. It was al.io necessary to convert silver t o its ammonia complex in order to raise its upper limit t o 50 p.p.m. Sickel and zinc, if present in sufficiently high concentrittions, cause low results apparently due to complex formation ith the reagent. The addition of excess reagent might overcome this interference. Attempts to remove iron, by complexing with fluoiide or tartrate, were unsuccessful. Anions causing the most difficulty are shown in Tahle 11.
Extractability of Cobalt-2,2’,2”-Terpj-ridiiie Complex
Solvent Benzene E t h y l ether Carbon tetrachloride Chloroform Hexyl alcohol Tsoarnyl alcohol Butyl phosphate Nitrobenzene - S o extraction. t Extrartion. 0 Possible extraction
ClOaPpt
-
Ppt 0
-
--
1-
t
Color Reaction. Three to 5 ml. of the reagent were used for 0.25 mg. of cobalt in 25 nil. ( I O p.p.m.). The amount of nitrobenzene was 0.4 of the amount, of aqueous solution. A second extraction yielded no additional color. The hue of the complex in nitrobenzene is orange and the absorption maxima have shifted 5 mp t,o 150 arid 510 mM. .4s with the aqueous system, an unusually wide range of p H is allowable. There is no adverse effect upon the intensity of the color or the extraction of the complex resulting from pH changes h c t m e n 2 to 11. Beer’s law was found to be valid for cohalt concentrations between 0.5 and 25 p,p.m. Fading of the aqueoua solution occurred within 24 hours. In the nitrohenzene l a y ~ r ,there was no measurable change of ahsorbance after 9 days and only a 3% change after 12 days. The molar ahsorptivity of the complex in nitrobenzene is 2900 a t 510 mp. This is almost double the value for the aqueous *?stem (.$). This f:tct coupled with the advantage of concentration by extraction provides a considerable increase in the usef u l sensitivity of the method. Effect of Diverse Ions. The following procedure was followed i l l determining the extent of interference by diverse ions. To 0.24 mg. of cobalt in a 25-ml. volumetric flask was added the desired volume of the ion under consideration. After the addition of 10 ml. of water, 3 ml. of reagent, and 1 drop of 1 to 1 perchloric acid, the p H was adjusted to above 3. The solution was adjusted to the mark and transferred t o a 60-ml. separatory funnel. Ten millilitcrr: of nitrobenzene were added, and the mixture was shaken for about 20 seconds, allowed t o separate, and then filtered through a plug of glass wool in a funnel. Spectrophotometric reading- were taken a t 510 mp using a 1-cm. cell with nitrohenzc‘ticas the blank. The follon.ing ions, when presciit in concentratioiis of 400 p.p.m., were found to cause no more than a 2% error in the determination of 9.6 p.p.ni. of cobalt: aluminum, acetate, aminoniuni, arsenite, barium, bismuth, borate, tetraborate, bromide, c~:iIc.iuin,citrate, chloride, fluoride, format?, iodide, lead, lithium, ni:ignesiuiii, mangsnese, nitr:rte, oxal:tte, perchlorate, hypo~~liosph:itc, phosphate, potassimn, selenate, sodium, strontium, sulfite, sulfate, tartrat,e, thioc‘yanatc,, thiosulfate, and thorium. S:ilts of most mineral acids do not affect the color r e a d o n . The ions which did interfere are listed in Tahle 11. The appwsimate limiting concentrations permissible for the determinat i o i l of 9.6 11.p.m. of cobalt, with :in wror of not more than 2’30, shown. XI1 oxidizing agents and iron must be :tl)sent. The. addition of a reducing agent to the aqueous solution before txltracting the complex helps, if the reduced metal does not itself interfere. Under the ~onditioiisused, a precipitate is 01)tained with gold, mercuric mercury, prrinanganate, and >latinum. Copper(I1) interferes considerably. However, 100 p.p.m. cause no interference if the copper is converted t o the ammonia complex and then reduced to the colorless cuprous ammonia ;i1’(3
1969
DISCUSSION
The use of the nitrobenzene extraction has removed the objection of the instability of the color of [Co(trpy)t] + + in aqueoui medium. This increases the prscticahility of the method and eliminates the necessity of preparing fresh standards d:dy for visual comparisons. The molar absorptivity is doublrd, thus providing a greater sensitivity. In addition, the method loses nolie of the advantages 01 accuracy of the method in the aqueous system. The same nick? range of pH is permissible and the interferences are of the wine order of magnitude, if not less. Oxidizing agents must be a l l v n t :ts well as iron and vanadate. Kickel, if present in mow than 30 p.p.m., and copper above 100 p.p.m. will interfere
Table 11. Effect of Interfering Ions
’
~illlount
Ion
Added as
Amount .idcli:d, P.P.hI. ROO 25 400 100
Fe+++
,50
Ft.++
.i0
Xi++
30 30 100
4g+
UOa + + ZnT+ ZrO +
.io
,? 0
+
Error, 3 42 3 2 40 40 2 2 2
2 4
25
4
100
lii
100 23
11 10
.Ij
100
?no
. 0i 2 .i
70
I’armissihle. P.P.11. 200 10 3on 100
?
36 50 100 00 23
4 10 2 2 42
RECO519IENDEU PROCEDURE
Treatment of Sample. Dissolve the sample by appropriatc means and remove any interfering constituents in accordanw mith the tolerances listed in Table I. Adjust the volume in a volumetric flask so that each 100 ml. contains 1 to 4 mg. cobalt. Measurement of Desired Constituent. TSTithdraw a 25-ml aliquot of the sample solution, add 1 drop of 1 t o 1 perchloric acid, adjust to p H 2 t o 11, and dilute to 50 ml. Transfer to a 120-ml. separatory funnel, pipet in 20 ml. of nitrobenzene, and shake well for a t least 20 seconds. Allow the layers t o separate, and then filter through a plug of glass wool. Measure a t 510 mp or use a blue-green filter such as Corning No. 428 and a 1-em. cell. ACKNOI’LEDGMENT
The authors wish to e s p r ~ s stheir appreciation to the ;Itonlie I.:nergy Commission for the financial support given this ~vorli. LITERATURE CITED
( I ) xlargerum, D., and Banks, C., .\NAL. CHEM., 26, 200 (1954). (2) hlorgan, G. T., and Burstall, F. H.. J . Chem. Soc., 1937, 104!~. (3) Rloss, If.,and Mellon, Il.,IND.EXG.CHEM.,ANAL.E u . . 14, 862 (1942). (4) Ibid., 15,74 (1943). *
RECEIVED for review June
16, 1954.
Accepted August 18. 1954