ANALYTICAL CHEMISTRY,
1026
diluted with one part of impropy1 alcohol to four parts of sample. Effech due to viecasity changes have been indicated (4). The effects have not been studied in this laboratory but the use of the alcohol mixture has been continued and all data reported in this paper have been obtained, using the alcohol dilution technique. Eaatman I-N plates were the only type found srttisfactory. Other plates, including Type I-L, did not have sufficient sensitivity in the spectral region used. The Type I-N, which has been designed for use in the infrared region of the spectrum ( 8 ) , apparently decreases in sensitivity in the wave-length region between the sodium and potassium lines. Thia probably accounts for the differences in sensitivity Observed for these two elements, which are not in accord with those reported for arc and spark spectra elsewhere (6). Plate development seemed more critical with Type I-N plates than with others. This waa partly due to the relatively high gamma ( 2 . 6 a t 7699.0 A.) produced by the platea and the developer used. These factors also account in part for the precision achieved in these determinetione, aa well aa the narrow range of concentrations for which the method is suitable. The procedure aa deacribed and evaluated appears to be a promising method for the determination of sodium and potassium in several typea of samples. Accuracy and sensitivity are good, ma11 samples are sufficient (2 to 5 ml.), and existing equipment in many laboratories may be utiiieed. Further study should lead to applications of the procedure to other types of samples
and possibly to the determination of other metals which may respond su5ciently to this form of excitation. LITERATURE CITED
h c . 0 5 c . Agr. Chemists, “05cial bnd Tentative Methods of Analyaia,” 6th ed., p. 121, 1945. Barner, R. B., Richardson, D., Berry,J. W., and Hood, R. L., I m . ENQ.CHEM.,ANAL.ED.,17, 806 (1946). Beckmcm BuU. 1 6 7 4 , National Technical Laboratories, South Paaadena, Calif., 1948. Ibid.. 193-B (19481.
Be*, J. W:, Ch;rppell, D. G., and Barnes, R. B.,IND. ENO. CHBM.,ANAL.ED.,18, 19 (1946). Brode, W. R., “Chemical Spectroscopy,” 2nd ed., New York. John Wiley & Sow, Inc., 1943. Cholak, J., and Hubbard, D. M., IND. ENO.CEEM.,ANAL.ED., 16,728 (1944).
Eaatman Kodak Co., “Photographic Plates for Scientific and Technical Urn’’, 6th ed., 1948. HeL, A. W., and Scribner, B. F., J. Reeearch Natl. Bur. Standa r L , 38, 439 (1947).
Lundeghrdh, H., 2.Phyeik, 66, 108 (1930). Parka, T. D., Johnson, El. O., and Lykken, L., ANAL CHEM., 20,822 (1948).
Peech, M., and English, L., Soil Sci., 57, 172 (1944). Weinbach, A. P., J . Biol. Chem., 110,BS (1935). Wolf, B., IND.ENO.CHEM.,ANAL.ED.,16, 121 (1944). REC~CIVED September 28, 1949. Prwented before the Division of Analytical CHEMICAL and Micro Chemistry at the 116th Meeting of the AM~CRICAN SOCIETY, Atlantic City, N. J. Contribution 391, Department of Chemistry, Kansaa Agriaultural Experiment Station.
Mineral Analysis of Biological Material by Flame Spectroscopy A ppa ratus and A ppl ication ABNER R. ROBINSON, KATHERINE J. NEWMAN, AND ERNEST J. SCHOEB Research Luboratory, Children’s Fund of Michigan, Detroit, Mich. The construction is described of an inexpensive, simple burner with an all-plastic atomizer and glass chamber similar to those described by Griggs (5). With the use of a Bausch & Lomb medium quartz spectrograph 15 elements can be determined directly or upon the dissolved ash of biological materials with an accuracy of *S%.
L
U N D E G b D H (6, 7) in Sweden published in 1929 a text on the use of a flame as the excitation mume for spectral identification and quantitative determination of alkalies, alkaline earths, and a number of other elements. However, in 1941 the flame technique waa being used by only two or three workers in the United States (3,6). Since 1944 many modifications of the flame-producing burner neceesary for thia method of analysis have been described in the literature and eeveral are available through commercial inetrument channels. After extensive experimentation the authors have devised a simple, inexpensive burner with which eatisfactory results can be obtained from analyses of biological materials for certain minerals. BURNER
The mskrials used in constructing the burner Shown in Figura 1 are considerably less expensive than those employed in earlier
bumere. The tip is a , s h d a r d , stainless steel ppduct (Jarrell-Ash Company), eciall rhodrum Iakd and connected to the borosilicate t u L with a ruiber sieeve. The acetylene inlet
3)aas
is a piece of thick-walled capillary tube, sealed into the burner with Plicene. Used with a Bausch & Lomb medium quartz spectrogra h the burner is mounted with the tip 4 cm. from the spectro a fiislit. A flat, stainless steel mirror is mounted directly behinrtge flame, the optical path of the spectrograph, to intensify the emmion. BURNER OPERATION
The burner is operated with comprewd air supplied a t a pressure of 30 ounds and acetylene from a tank a t a water gage preenure of cm. Both air and eoetylene are water-saturated before enterin the burner. An all- lastic atomizer (JarrellAah Company? and lssa atomizer c%amber similar to those described by Grigga except that the. chamber and burner are connected by a standard-taper ground joint, are employed to introduce the sample. The atomizing unit and burner are eaaily cleaned after use. Wetting agents are not required in the operation of the burner and s screen below the burner tip, used in the LundegCdh and later modificatjons, ia unnecessary.
&
b),
SPECTROGRAPHIC TECHNIQUE
In analysis for alkali metals the region 1 am. above the blue cone of the flame is wed, the region LundegCdh designates aa most
V O L U M E 22, NO. 8, A U G U S T 1 9 5 0 Table I. Line Ag 8281 Ba bb8b. b
1027
Speotral Lineo and Scnritivitier
$:j4o' ~$i*
Ca 4227 Co 8127 Cr 8178 Cu 8147. b Fe 8810 K 4044
Line
0.0b 1 0 01004
Li 6708
0.004 0.00b 0,Ob 0.08
Na b880
0.01
p)yo~gi* oma
0.11 0.002 0,l 0.002
Ni 8blb Br 4607
Internal Bkndard
co 8406
t%Eium Calalum
Co 8408 co 8878 co 8878 co 8878 Co 840b c o 8878
Potudum
Manranwe Copper Iron
Analytioal
Range
Mg./lO h i . 0.064. 6 0.2b-2. b 0.02-0.12 0.47-4.7 0.01-0.8 0.02-0.6 0.06-0.6
sensitive for these elements. Two-minute exposures are satisfactory with Eaetman emulsion 88 plafes. Wratten Wainwright panchromatic plates are used to investigate the red range of the spectrum. The s ectral lines wed for anal si8 aad their rwnsitivitiesunder the ana&ticrl conditiono are Iistedrin Table T. Cobalt is u y d ae the internal standard, lines 8406 and 8874 being the control Imes. Standard solutions are analyaed on each plate and standard curves are drawn for each element from each plate. The analytical ranges found for the elements are given in Table 11. Attem te are not made to calibrate the plate emuleiona BLI su4 este8 by Cholak I), but exposure and developing times are rig&y controlled. P ates are proceseed with mechanical agitetion throughout, immersed in Eaetman D 1 9 developer for 8
I
n
APPLICATION OF METHOD
0.001 0.06
Mg 28b2 Mn 4081 Na 8802
Table 11. Analytioal Ranger Element
minutes, in 5% acetic acid solution for 30 aeconda, then in Eaetman x-ra finlaher and h e r . Line densities are meaeured with a b e d s & Korthrup denaitorneter.
YITAL UTAINLEW STEEL TIP RUBBER SLIIEVVL
Using the flame spectroffiopy technique deffiribed, quantitative analytical valu- b v e been obtained for mdium, potaseium, magnesium, caloium, silver C4), copper, iron, strontium, and lithium in human and aow's milk, blood, urine, feces, and tissue. The method is eatisfactory for analysis of the readily soluble auh from biological materials and, under the conditions described, provides reproducible results within an accuracy of 6%. For duplicate spectra, 8 ml. of solution are required. Bemple material must be completely in solution and the total selt concentration must not be so high as to c a m precipit4ationin the atomizer. As ,with other burnem, the presence of strong oxidising acide in final dilutions of a h must be avoided to prevent oxidation of the burner tip. The utandard solutions for each eample must be prepared carefully ta contain amounts of the elements to be determined consistent with the amounts known to be present in the sample. Background effects result when magnesium, iron, and aluminum concentrations vary widely. Chemical methods of analysie are employed to check rerults of &me spectroscopy determinations : the gravimetric procedure (11) for calcium, the aino uranyl acetate gravimetric mebbod (@> for sodium, and the Peeoh ( 8 )colorimetrio procedure for magna sium. Results with spectrographic and chemical determinationa check within the limits of error of both methods. PREPARATION OF SAMPLES
Milk samples are dried from the froaen state with Desivac e uipment. One gram of dry milk is reduced ta u h in a p?atinum crucible a t 460° C. The esh is diesolved in and di!uted to 26 ml. with, 10% redietilled hydrochlorio acid contru$ng 0.76 mg. of cobalt per milliliter for dete+nation of sodium, potaseium, and magnesium. Calcium anal ais requires additional dilution of 1 to 100 and adjustment o f the cobalt concentration to 0.376 mg. per milliliter. Attempte to use fresh milk for the determinations with the internal standard are unsucceeaful because precipitated protein clogs the atomiaer.
Table 111. Effect of Chandng Concentrationo on Reeultr of Flame Analyuae (Valuw in milligrams per 10 ml.) Magnwiumb Pokssium A B A B 0.1b8 0.162 2.8 2.2
ivas
Solution" Milk diluted 1-10 Milk' diluted 2-10 Milk: diluted 1-10 O 8 m Kadded Miik dikted 2-10 &m Na adddd uted 2-10 0 . i mi. M addld Magnesium, t . 2 mg./10 ml . 0.4 mg./10 ml.
Sodium A B 0.8s 0.40 0.74 0.81 0.26 0.4b
0.280 0.286 0.14b 0.1b8
4.4 4 . 2 8.0 2.8
0.288 0.884
4.b 4.6
1.20
4.7
0.80 0.88
...... 0.20 0.40
0.21
.. .. .. ..0.40 .. .. .. .. .. ... ...
4.8
......
1.27
.. .. .. .. .. ..
nl. i : i i:e Potsmiurn 1 6 m sodium,2-10 ig./Td% i.70 i.76 0.6 mg./lO ml. 0.87 O.M 1.1 mg./10 ml. 1.26 1.28 " All rwultl obtained b rwding Btandard in nhioh oonomtrationg ap roxidnuted thole of n d k and oontdning oobalt Y internal #t.ndard. f A, aalouhted from ratio# of dement to cobalt. B, oaloulrt.d from ratios of (dement minui baokground) to (cobalt minu baokground).
...... . . I
I . .
. I ,
I , ,
Samples of blood plasma, t b u e , and feces were webeshed with nitric-perchloric acid mixture, evaporated to dryneee, dissolved in and diluted to correct concentration with 10% h drochloric acid containing the proper concentration of cobalt. [Eve milliliters of laema provide sufficient material for magnesium, p o t a s aim, o a k u m , and sodium determinationa.) Samples of urine are diluted with the proper amounte of cobalt solution and the alkali metals are determined without aahing, but for other cationa concentration of the urine ia necwary. Figure 1. Burner for Flame Spectroocopy
Recovery experiments (Table 111) were made with milk eamplea of different dilutions, milk samples of different dilutions to
1028
ANALYTICAL CHEMISTRY
which known amounts of elements to be determined had been added, and solutions of the elements only, to determine whether methods of calculation are justified or greater care of plate calibration is necessary. The results emphasize the necessity of considering background corrections in calculating h a 1 data from flame analysis spectrograms, especially with respect to sodium and magnesium. Recoveries with pure sodium solutions are not so complete as with sodium in diluted milk; this indicates that a. standard is required which approximates the composition of the sample as shown by the values obtained with pure solutions of sodium in contrast to those obtained with dilute milk solutions. The curves of ratio of element to cobalt and (element minus background) to (cobalt minus background) are drawn on semilog paper, the backgrounds being subtracted directly and not as recommended by Pierce and Nachtrieb (IO). Plate calibrations using the two-line method of Churchill ( 8 ) were made, but the increase in accuracy does not warrant routine use of this procedure, despite the advantage of being able to read the concentrations of elements from straight-line calibration curves.
LITERATURE CITED (1)
Cholak, J., and Hubbard, D. M., IND.ENO.CHEM.,ANAL.ED.,
(2) (3) (4)
Churchill, J. R., Ibid., 16, 653 (1944). Ells, V. R..J. Optical SOC.Am., 31, 534 (1941). Gerbasi, F. S., and Robinson, A. R., Am. J. Clin. Path.,
16, 728 (1944). 19, 668
(1949).
Griggs, M. A., Johnstin, R., and Elledge, E. E., IND.ENQ. CHEM.,ANAL.ED.,13, 99 (1941). (6) Lundegkdh, H., “Die quantitative Spektralanalyse der Elemente,” Vol. I, Jena, Gustav Fischer, 1929. (7) I b d . , Vol. 11, 1934. (8) Peech, M., IND. ENO.CHEM.,ANAL.ED., 13, 430 (1941). (9) Peters, J. P., and Van Slyke, D. D., “Quantitative Clinical Chemistry,” Vol. 2, p. 732, Baltimore, Williams & Wilkins Ca.., ~1932. - - (10) Pierce, W. C., and Nachtrieb, N. H., IND. ENQ.CHEM.,ANAL. ED., 13, 774 (1941). (11) Willard, € H., I. and Furman, N. H., “Elementary Quantitative Analysis,” p. 314, New York, D. Van Nostrand Co., 1939. (5)
R E C E I V ~December D 27, 1949. Presented a t the 53rd Annual Meeting of, the Miohigan Academy of Sciences, Arts and Letters, Sanitary and Medioal Science Section, Detroit, Mioh., April 2, 1949.
Titration of Bases in Nonaqueous Solvents JAMES 9. FRITZ Wayne University, Detroit 1, Mich.
Organic bases may be accurately titrated in glacial acetic acid, benzene, chlorobenzene, nitrobenzene, chloroform, ethyl acetate, diethyl ether, petroleum ether, and acetonitrile, using methyl violet as the indicator. The titrant employed is a solution of perchloric acid in glacial acetic acid. Potentiometric titrations in most of the above solvents may be conveniently made with a pH meter using a glasrrsilver electrode combination without a salt bridge. Small amounts of water do not interfere, but larger amounts of water, alcohol, dioxane, and acetone should be absent.
I
T WAS first pointed out by Conant (8) that the strength of basic compounds is much greater in glacial acetic acid than in aqueous solution. In this work curves were published for the potentiometric titration of a large number of bases with perchloric acid. Nadeau and Branchen (6) titrated amino acids in glacial acetic acid, using any one of several visual indicators to detect the end point. Blumrich and Bandel (1) determined several amines in acetic acid solvent, and Nerd (6) applied thjs method to the determination of quinine. Small amounts of basic impurities in benzene (8) and in hydrocarbon oils (9) have been determined by first mixing acetic acid with the sample and then titrating with an acetic acid solution of perchloric acid. Palit (7) titrated certain bases in glycol-hydrocarbon solvent mixtures. It has recently been found (4) that many bases may be advantageously titrated in dioxane using a dioxane solution of perchloric acid as the titrant. The purpose of this investigation is to demonstrate that both strong and weak bases may be conveniently titrated in benzene, chlorobenzene, nitrobenzene, chloroform, diethyl ether, ligroin (petroleum ether), and other relatively inert organic solvents. The titrant wed is perchloric acid in either glacial acetic acid or acetic acid-chlorobenzene, Methyl violet serves as the indicator. I n each case the end point is a t least as sharp as when the base is titrated in glacial acetic acid, The chief advantage gained, however, is in the analysis of damples in which a base is dissolved in any of the above listed solvents. The determination of aniline in nitrobenzene, estimation of amines in hydrocarbon
polymerization feed, and titration of bases extracted by chloroform from a mixture are but a few examples of practical applications. In each case the base may be titrated directly without fist evaporating the solvent and redissolving in acetic acid. PROCEDURE
A sample of suitable size is dissolved in 25 to 50 ml. of the chosen solvent. Two or three drops of methyl violet indicator (methyl violet in chlorobenzene) are added and the solution is titrated to a green color with 0.1 N perchloric acid in acetic acid. The perchloric acid solution is prepared b dissolvin 8.5 ml. of 70 to 72% perchloric acid in 1liter of laciaracetic acif. If desired, the water ma be removed by acfdmg about 14 grams per liter of acetic anhy&de, and allowin the solution to stand overnight. This solution is standardizefagainst diphenylguanidine, prepared according to the directions of Carlton ( 8 ) . Results obtained by this method a y e e with sodium carbonate standardization; diphenylguanidine 18, however, preferred because it has a much higher equivalent weight and gives a somewhat sharper end point than sodium carbonate.
Data given in Table I indicate that the results obtained from titrating several representative bases in benzene, chlorobenzene, etc., are in agreement with those obtained from titration in glacial acetic acid. INTERFERENCES
It is not possible to titrate bases in dioxane, alcohols, or acetone when perchloric acid in acetic acid is used as the titrant.
