separating the detergent suggests increased accuracy by elimination of interferences in the usual procedure. Substances that normally interfere with the colorimetric method-thiocyanate, nitrite, peptone, and urine (5)--would not be expected to pass through the column with the detergent. As the peak effluent volume of this anionic detergent coincides with that of formic acid, peaks previously observed in the fractional area of forniic acid of both the river waters and sewage effluents may be partly or totally due to alkyl acid sulfates. Anaerobic Digester Liquors. Analyses on the supernatant liquor from both the primary and secondary anaerobic digesters of the local treatment plant were of interest. Peaks corresponding t o the commonly occurring fatty acids-butyric, propionic, and acetic-were observed in all samples chromatographed. The presence of these acids ivould be expected and is attributed to the metabolic activity of the organism types present. The repeated appearance in several analyses of a peak in the fractional area characteristic of synthetic detergents and formic acid, however, is not readily explainable. After titration of the fractions, common to )he elution of these acids, the presence of anionic detergents was assayed colorimetrically. The results of one such assay showed 0.32 peq. of detergent present, or 4.3% of the acidity (7.4 peq.) as determined by titration. A similar analysis, which compared favorably in the quantities of acids present, showed 11.3% of the acidity in this fractional area accounted for
as synthetic detergent. Formic acid was not determined in the fractions. Fractionation of the acids present in the distillate from the supernatant liquor resulted in peaks corresponding to butyric, propionic, and acetic acids: however, the peak common to synthetic detergents and formic acid, present in the analysis of the supernatant liquor, It-as not observed. Hence, the absence of formic acid in the chromatograph suggests the presence of another acid or acids that are coincident with the elution of synthetic detergents and forniic acid. Fumaric and a-ketoglutaric acids are unresolvable with formic acid by the solvent system employed; however. thcir presence in such high concentration is improbable.
expected in vien- of the heterogeneity of 11-aste materials and metabolic processes involved in their degradation. The methodology and data presented only begin to illustrate the potentialities of the chromatographic analysis of natural waters. The procedure is not complicated and is quantitative for the acids included in the reference chromatogram. A t present the method is capable of separating many organic acids occurring in natural waters. Further work is desirable to extend the number of organic acids identifiable by column and solvent modifications, and to provide suppleniental tests for confirmatory identification of the acids by colorimetric means, as in the case of synthctic detergents, by selective enzymes, or by ionization product.
DISCUSSION A N D SUMMARY LITERATURE CITED
Preliminary data on river waters though not intended for comparative purposes, show a somewhat uniform correlation in the acid types appearing. These commonly occurring acids are likely to be of biological origin, but may come from other sources. It is possible that acids appear as a result of alkaline hydrolysis of complev materials during concentration of the sample. Also, anionic detergents and other acids are eluted in the same fractional area as formic acid. On the other hand, the chromatographic analyses of the sewage effluents a t various stages of the treatment process are not as readily interpreted as the analyses of the river waters. Several known and identifiable acid peaks are evident in each stage, but complex and erratic peaks of unknown origin are also present. This is not
(1) Bulen, W. A,, Varner, J. E., Burrell, R. C., A x . 4 ~ CHEY . 24, 187 (1952). (2) Isherwood, F. -4., Biochem. J . (London)40. 688 11946). (3) Jones, J. H.,‘ J . kssoc: O$c. d g r . Chemists 28, 398 (1945). (4) Marvel, C. S., Rands, R. D., Jr., 6.-4m. Chem. SOC.72, 2642 (1950). ( 5 ) hloore, TI’. -4., Kolbeson, R. A4., SNAL. CHEW28, 161 (1956). (6) RIueller, H. F., Buswell, A . hf., Larson, T. E., Sewage & Ind. Wastes
28,255 (1956). (7) Yeish, 9. C., Can. J . Research B27, 6 (1949). (8) Neish, A. C., Xatl. Research Council of Canada, Rept . 46-8-3 (Second Revision) 15 (1952). (9) Steele, R. H., White, .4. G. C., Peirce, TV. A,, Jr., J . Bacteriol., 67, 86 (1954). RECEIVEDfor review May 6, 1957. Accepted September 18, 1957. Division of Water, Sewage, and Sanitation. Chemistry, 130th Meeting, ACS, Btla,ntic City, 5’. J., September 1956.
Spectrophotometric Determination of Iron with Ethy lenedia mine Di(0- Hy d roxyphe ny Ia cetic Aci d) A. L. UNDERWOOD Department o f Chemisfry, Emory University, Emory University, Ga.
b The new reagent ethylenediamine di(o-hydroxyphenylacetic acid) can be applied in the spectrophotometric determination of iron. The ferric complex of the reagent is so stable that color development is maximal over a wide pH range, and a large excess of reagent is not necessary. Once formed under proper conditions, the colored solutions are stable for a long time. In the method described, the optimal concentration of iron i s about 5 p.p.m., where the standard devia44
ANALYTICAL CHEMISTRY
tion is about 0.5Q/O. While the method i s not entirely free of interferences, it compares favorably with many others. Determination of iron in an aluminum alloy i s described.
T
synthesis of ethylenediamine di(o-hydroxj-phenylacetic acid) a phenolic analog of ethylenediaminetetraacetic acid, was reported recently by Kroll and his coxvorkers ( 1 ) . This new compound \Vas stated to give a red HE
~
COOH
O \-H
HOOC
-
/w HO
color with ferric iron through the formation of a very stable chelate complex whose formation constant \vas estimated as about 1030. This reagent should be of considerable interest in analytical chemistry, although much iyork will undoubtedly be required to make possible the complete exploitation
----.I
l
l
,
l
I
I
I
1
1
1
0.051 I
l
400 450 500 WAV E L ENG T H (MILL I M I C R 0N S )
Figure 1.
Absorption spectra Figure 2.
Effect of pH
uf its iiitererting properties.
In the study rrported, it is shown t'hat ethyleriedianiiiie di(o - li~-droxyphenylacetic acid) is :i very useful reagent for the spect,roliliotoiii~,ti,i~det~r~iiin:it,ion of iron. APPARATUS A N D REAGENTS
Alll :Ilisorbanc.e nieasurenients n w e performed Ivith a Becknian Model DLspectiol,liotonieter, usiiig 1-cm. Cores cells. pH valuc,s were measured n-ith a Becknian 3Iodel H pH meter. A standard ferric solution was prepared by dissolving 1.000 gram of pure iron wire iii 20 ml. of 1 t o 1 nitric acid, boiling tlie solution for a few minutes, and diluting to exactly 1 liter \\-it11 dist~illrd n-wtc3r. Dilutions of this st'ock solutioii w r t > prepared as required, with the addition of sufficient' nitric acid to prevciit hydrolysis and precipitation of the iron. The rc.:igc,nt ethylenediaminr cli(oIi!dros!-~,henylncetic acid) (Chel 138) was ohiiiied t,lirough tlie courtesy of Gcigj- Industrial Cheriiicals Division of Gcigy Clicmical Corp., .irdsley, S.Y . A solution of t,his reagent was prepared by suspt>ndinyI .8 grams of the solid in about 50 ml. of n-ater and adding dropn i w G N sodium hydroside n M r checking n-ith a pII nicter. A s the solution was stirrcd. tlic reagent s l o ~ l ydissolved. ]Then tile pH reaciird ii stable value of 8.5 to 9.0, solution n-as l)ractic:i!ly complete. m d the volunie \\-as adjusted to 100 i d . with n-ater, after which the solution TKIS filtered through a "slov?' filtcr p:ipcr. For some of the preliniinwy w r k described bctlon-. a more dilute reugrnt solution. prepared by trwtiiig 360 uig. of reagent as drscribcd :il)ovc. \cis usedj but the more coricentratcd solution is rec.oiiimc,ndrd in t'he malyticxl Iiic.tliod. Thv 1.11 acctatc buffer \vas prepared b\. dissolving GO nil. of glacial acetic acid in wat(br, adjusting the pH t'o 5.0 with sodium hydroside,. and diluting the solution to 1 liter. Its final plI vias 3.9. Rcngeiit-grade niaterials wvrre used t'liroughoiit tlip study. VARIABLES AFFECTING COLOR DEVELOPMENT
Absorption
Spectra.
Figure
1
Y Figure
shons t'he absorption spectrum of a solution containing 150 y of iron, 2 mi. of reagent solution: and 5 ml. of acetate buffer diluted to a final volume of 25 ml. The spectrum of a blank solution containing no iron is also shown. The iron solution exhibit's an absorption band with a maximuni a t about 4T0 nip but sufficiently broad so that the \rave length sett'ing is not highly critical. d l l absorbance rneasurenimts reported in this paper Irere made at a wave length of 470 m p . The blank absorption is small but not unappreciable at' this wave length. Presumably the color is due, a t least in part, to oxidation products of the phenolic reagent itself, and it' niiglit be possible to obtain a colorless blank by purifiying the reagent carefully and then excluding air from its solutions. Effect of pH. d series of solutions was prepared containing 50 y of iron, 2 ml. of reagent solution, and 5 inl. of acetate buffer. The pH values of thc solutions were varied by thr addition of hydrochloric acid or sodium liydrosidr. and the final pH values were eheckcd after dilution to 25 ml. Figure 2 shons that the wide pH range over which the absorbance values rcmaiii constarit is an attractive feature in the use of this reagent. The rest of the results reported n-ere obtained a t pH va1ut.s between 3.6 and 5.0. Effect of Reagent Concentration. \-ar\-ing amounts of the more dilute
i
I
i
2
I
i
3 4 REAGENT SOLN.
MILLILITERS OF 3. Effect of reagent concentration
Chel 138 solution were addcd t o a series of solutions containing 50 y of iron and 5 ml. of acetate buffer, and the solutions n-ere diluted to 25 nil. Figure 3 s h o w the absorbance values of these solutions as a function of the aniount of reagent added. The sharp break in the curve reflects the great stability of the ferric-reagent complex. The slightly positive slope of the curve after the break is due to the s~iiall absorption by the excess reagent itself. The fact that a large excess of reagent is not needed for full color deyclopment is another attractive feature in tlie use of this reagent for iron. The curve of Figure 3 is. in principle, a photonietric titration curve. Although this aspect is not emphasized in the present study, it should be possible to titrate ferric iron with Chel 138 making use of photometric detection of the end point. Interferences due to extraneous suhstances which are colored hut which do not react with the reagent could be circunimnted in this n a y , because in a photonietric titration only relatix-e absorbance readings reflecting the progrcof the titration are required. Effect of Time. The red solutions are very stable. However. Irvhen buffer n a s added to a n iron solution before the reagent was added, the color did not quite attain its maximal intensity immediately upon the addition of the reagent. The absorbance VOL. 30, NO. 1, JANUARY 1958
45
of a solution containing 150 y of iron increased by about 1.8% over a period of 1 hour after mixing. On the other hand, when the reagent was added before the buffer, maximal color was attained immediately. Possibly ferrichydrolysis products, formed when the pH is raised to 5 in the absence of reagent, react rather slonly when the reagent is added. Once formed, the red solutions are stable indefinitely; absorbance values are unchanged over a period of 8 hours, and several solutions left standing for 2 weeks appeared unchanged to the eye. The reagent solution itself, originally light yellow-brown, slowly darkens upon standing for several days. The absorbance of a reagent blank solution, then, depends upon the freshness of the reagent solution. It is recommended that fresh reagent solutions be prepared Iyeekly.
At the 10-mg. level (excess of nearly 70 to l), mercury joins the non-interferers, while uranium, thorium, and nickel interfere only slightly. At the 1-mg. level (7 to 1 excess), uranium, thorium, copper, nickel, and zinc are added to the list of noninterferers, and I
46
ANALYTICAL CHEMISTRY
I
I
I
’
“‘I“”
Figure 4. Ringbom plot of calibration data
I
RESULTS
Beer’s Law and Ringbom Plot. Solutions, prepared by adding 2 ml. of Chel 135 solution to aliquots of a standard ferric solution, followed by the addition of 5 ml. of 1M acetate buffer and dilution to 25 ml., adhered to BeerJs law over the iron concentration range studied (0 to 350 y of iron, or 0 t o 14 p.p.m. of iron, corresponding to absorbance values against a reagent blank of 0 t o 1.172). Figure 4 shows a Ringbom plot of these data. The optimal concentration of iron is about 5.2 p.p.m., with extremes of roughly 3.2 and 8.0 p.p.m. delineating the range \\-here the error is very nearly minimal. The per cent relative analysis error per 1% absolute photometric error is approximately 2.8, as evaluated from the slope of the central portion of the Ringbom plot. Reproducibility with Pure Iron Solutions. Three groups, each consisting of 20 solutions, were prepared containing 2, 6, and 10 p.p.m. of iron. The absorbance values against a reagent blank are shown in Table I, along with a statistical summary of the results. It is apparent that, as a result of the preliminary work described above, the variables effecting the color reaction are under control. Interfering Ions. Ten-milligram quantities of 13 metals were tested for interference in the measurement of 150 y of iron (6 p.p.m.). Where no interference was found, the study was repeated using 100 mg. of the foreign metal while, if interference was found, 1 mg. of the metal was tested. I n the two cases where even 1 mg. interfered, 0.5-mg. quantities were tried (Table 11). I t may be seen that 100 mg. (representing a ratio of metal to iron of nearly 700 to 1 on a weight basis) of lead, aluminum, manganese, calcium, or magnesium shows no interference.
I
chromium interferes only slightly. The interferences of cobalt and chromium persist slightly, even a t the lowest level tested (3 to 1). Determination of Iron in an Aluminum Alloy. To demonstrate the utility of the reagent in a practical
I
Table 1.
I
I
I
I
i
1 1 1 l 1 1 l 1 l
Reproducibility at Three Concentrations
2.00 P.P.M. Fe
Range
Median .-.
Mean Av. dev. Std. dev. Table II.
6.00 P.P.M. Fe 10.00 P.P.M. Fe Absorbance Dev. from Absorbance Dev. from Absorbance Dev. from mean us. mean mean vs. us. absorbance blank absorbance blank blank absorbance 0.003 0.000 0,850 0.179 0.001 0.515 0.003 0.850 0.514 0 001 0.001 0,848 0,008 0.001 0.516 0.180 0 855 0,002 0.000 0.515 0.179 n 008 0.845 0.002 0.517 0.182 0.003 0.850 0,004 0.519 0.180 0.003 0.850 0.001 0.000 0.514 0.178 0 855 0.002 0.005 0.520 0.175 0.003 0,003 0.850 0.002 0.516 0.001 0.180 0.017 0.870 0.510 0.005 0.176 0.002 0 003 0.850 0.001 0.514 0.180 0.002 0 007 0,860 0.000 0,515 0 000 0.178 n 007 0.860 0,002 0.517 0.177 0 001 0.003 0,850 0.005 0.510 0.002 0.176 0.860 0.007 0.000 0.515 0,001 0.177 0.003 0.850 0 003 0 002 0.512 0.176 0.855 0.002 0.002 0 513 0.177 0.001 n 003 0.003 0.850 0.518 0 004 0.174 0.860 0.007 0.002 0 001 0.513 0.179 0.003 0.850 0.002 0.513 0.000 0.178 0.025 0.010 0,008 0.850 0.515 0.178 0 i78 0 515 0 853 0 0020 ( 0 3970) 0 0049 (0 5770) 0 0016 (0 90%) 0 0028(0 5370) 0 0061 (0 7270) 0 0020(1 1%) Interference Study of Effect of 13 Metal Ions on Absorbance Value Obtained with 150 y of Iron
(Standard reading for no interference, 0.515, with normal range Ion Present Added as 0.5 llg. 1 Mg. 0.512 ... U@+ 0.513 ... Th4+ Pb++ o:iio Cu++ 0.614 0: is0 eo++ 0.506 Ni++ Al+++
Mn++ Zn++
C r+++ _.
Hg++ Ca++ Mg++
0.533
...
...
0 :5i2
0.547
about 0.510 to 0.520) 10 Mg.
0.560 0.494 0.516 0.410 0.874 0,483 0.515 0.522 0.360 0.748 0.514 0.522 0.516
100 hIg. 0,523 ... ...
0:5k 0.510
Pptn. 0.510 0.517
case, a n aluminum alloy furnished by the National Bureau of Standards was analyzed for iron. The alloy (Standard Sample No. 603, wrought alloy 61s) was certified t o contain 0.21% iron, along with 0.29% copper, 1.Olycmagnesium, O.52yc silicon, o.24Y0 chromium, and o.037y0 titanium. A sample of about 100 mg. of the alloy was covered with about 1 ml. of water in a 5O-ni1. Erlenmeyer flask, and concentrated hydrochloric acid was added dropwise until dissolution began to take place a t the desired rate. After solution was complete (accomplished by warming in the later stages), the sides of the flask were rinsed down with water, about 0.5 ml. of concentrated nitric acid was added, and the solution
was boiled gently for several minutes. The volume of the solution was adjusted to about 10 or 12 ml. with rinsing, and 6M sodium hydroxide solution was added dropwise until a slight permanent precipitate just appeared. Then 2 ml. of Chell38 reagent solution and 5 ml. of acetate buffer were added, and the volume of the solution was adjusted to 25 ml. with water. The solution a t this stage was slightly cloudy, and a portion was poured through a slow filter paper, with collection of the clear filtrate in a Beckman cell begun after the paper had equilibrated with the solution so that no adsorption loss would occur. The solution was read against a blank, and the iron content was obtained by comparison with the reading for a standard iron solution.
The results of three replicate analyses were 0.20, 0.20, and 0.22%, compared with the certified value of 0.21%. The time required for the three analyses, from the weighing of the samples to the calculation of the results, was about 1.5 hours. Considerable improvement in technique and saving of time per analysis should be accomplished if the analysis were placed on a routine basis. LITERATURE CITED
( 1 ) KroJl, H., Knell, &I., Powers, J., Simonian, J., J. Am. Chem. SOC. 79, 2024 (1957).
RECEIVED for review July 9, 1957. cepted September 16, 1957.
Ac-
Carrier Precipitation of Trace Elements Radioisotope Evaluation of Efficiency EDWARD E. PICKETT and BOBBY E. HANKINS Departments of Agricultural Chemistry and Chemistry, University of Missouri, Columbia, M o . ,The efficiency of separation of the trace elements, copper, cobalt, and molybdenum, from the major constituents of agricultural samples by precipitation with aluminum oxinate or indium oxinate carrier, with or without added thionalide or tannic acid, was determined by means of radioisotopes. For practical purposes all are completely precipitated at realistic levels of these elements when all three organic reagents are used. With aluminum oxinate alone, copper is incompletely recovered; with indium oxinate alone, none are completely recovered. Certain experiments were performed to gain some information on the mechanism of the coprecipitation.
I
spectrographic determination of the biologically important trace elements (copper, cobalt, zinc, and molybdenum) in plant materials, animal tissues, soil extracts, and the like, the need usually arises for their preliminary chemical separation from the major constituents. The last three of these metals usually are present in such small amounts as to escape spectrographic detection in the whole wet- or dry-ash preparations. Several methods have been proposed for making this separation. Extraction procedures employing dithizone ( 2 , 16) do not bring molybdenum into the trace element fraction. A precipiN THE
tation or extraction procedure employing sodium pyrrolidinedithiocarbamate (12, 14) separates the molybdenum along with copper, cobalt, zinc, and other heavy metals. An earlier precipitation method, employing 8-quinolinol (oxine) alone, and later in combination with thionalide and tannic acid, was presented by Mitchell (7, 9 ) , and has been used by others (1, 10, IS). A carrier precipitate of aluminum and iron oxinates, with excess oxine and some thionalide and tannic acid, was formed and the solution was buffered a t p H 5.2 with ammonium acetate. Quantitative recovery of cobalt, nickel, molybdenum, copper, and zinc was found when oxine alone was used; the thionalide and tannic acid served to render recovery of chromium, vanadium, beryllium, germanium, tin, lead, and cadmium complete as well. Iron served as variable internal standard. A modification of the method by Heggen and Strock (4) used indium both as carrier and internal standard. I n preliminary trials of Mitchell’s procedure in this laboratory the thionalide and tannic acid were omitted because, as noted above, oxine alone was reported to recover copper, cobalt, zinc, and molybdenum completely. Tannic acid gives high blanks and is difficult to purify. Incomplete recovery was found for copper. With indium oxinate as carrier, and without thionalide and tannic acid, both molybdenum and copper were
incompletely recovered. The results were supported in numerous experiments involving various conditions of precipitation and measurement. I n order to confirm and extend these results with greater certainty and in the hope of gaining some information about the mechanism of carrying by such precipitates, radioactive tracers were employed. INVESTIGATIONS WITH RADIOISOTOPES
Several small batches of copper-64, cobalt-60, and molybdenum-99 in solution were obtained from Oak Ridge National Laboratory over a period of about 9 months. Their specific activities were approximately 3000, 20,000, and 30 mc. per gram, respectively. I n some experiments trace amounts of the individual isotopes were coprecipitated from solution with 15 mg. of aluminum or indium by adding 10 ml. of a 5yc solution of oxine in 2N acetic acid (9), 45 ml. of 2N ammonium acetate solution, and raising the pH to 5.2 with ammonium hydroxide solution. I n other experiments 2 ml. of thionalide solution (1% in glacial acetic acid) and/or 2 ml. of tannic acid solution (lOyc in 2iV ammonium acetate solution) were also added just before final adjustment to pH 5.2 (9). The precipitates, except as noted, were allowed to stand overnight, then filtered through Whatman No. 42 paper and washed several times with small amounts of water. Acid was added to the filtrates to dissolve any oxine which separated during filtration and their volume was made up to 250 VOL. 30, NO. 1, JANUARY 1958
47
