onto the column to obtain the chromatogram tracing illdstrated in Figure 2. The similar tracing shown in Figure 3 was obtained by injection of an aliquot of the eame mixture of acids diluted with an equal volume of water. Because the longer chain acids are soluble in water to only a limited degree, it was necessary to shake the aqueous mixture vigorously during withdrawal of the sample. Acid mixtures of unknown compoaition are in many cases obtained as water solutions of alkali salts extracted from biological materials. After acidification, these solutions could be sampled and treated as described above. To simulate such a situation, an aliquot of a solution of sodium acetate and propionate containing about 20% by weight of the salts waa chromatographed after it was acidified to p H 2 with 85% phosphoric acid (Figure 4). RESULTS AND DISCUSSION
Figures 2 and 3, illustrating the separation of anhydrous and hydrated organic acids, respectively, are closely similar in that the peaks occur a t very nearly corresponding points in both chromatograms. Admixture of water with the organic acids used does not
significantly affect the resolving power of the column under the conditions described nor does the resolving power of the column noticeably change with continued use. At the 125' C. temperature of separation selected for these runs, the more volatile components emerge as sharp, closely spaced peaks, whereas the high-boiling components are low, broad, and widely separated. This situation could be improved by operating the column a t slightly higher temperatures; however, higher temperatures will cause a decrease in the resolution of the lower-boiling components. For quantitative work two determinations a t different temperatures could be made. One could be run a t a temperature below 125' C. to improve the separations of the Co through CS acids, and another separate determination made at a higher temperature to improve the separation of the acid homologs above CS. Figure 4 illustrates that the separation procedure described is applicable to the identification of simple volatile organic acids derived from plant tissues or other biological materials. Since the liquid phases used in this
work have been applied successfully in the separation of the esters of fatty acids (S), higher acids than caprylic could be analyzed on the same column a t the same operating temperature by converting them to esters. ACKNOWLEDGMENT
The authors thank E. B. Kester of this laboratory for technical advice. LITERATURE CITED
(1) Hardy, C. J., Pollard, F. H., Chromatog. 2, 22 :1959).
J.
( 2 ) James, A. T.,, Martin, A . J. P., Biochem. J. 50, 679 (1952j. (3) Lipsky, S. R., Landowne, R. A., Bzochzm. et Bzowhus. Acta 27, 666 I\A1 aKQ) Q"",.
f
."
4) McInnes, A., in "Va our Phase Chro-
6.
IS'.Desty, ed., tterworths London, 1957. Iartin, A. h,,Smart, J., Nature 175, (1955).
mith Bength, Acta Chem. Scand. A
m iinEn\
RECEIVED for review September 14, 1959. Accepted December 21, 1959. Mention of specific products does not imply recommendation by the Department of Agriculture over others of a similar nature not mentioned.
Separation of PoIycycIic Aromatic Hydrocarbons in Complex Mixtures Chromatog ra phic Determination of Trace Amounts in Petroleum Waxes WILLIAM LlJlNSKY Division o f Oncology, The Chicago Medical School, Chicago, 111.
b Four polycyclic aromatic hydrocarbons-benzo[a]pyrene, dibenz[a,h]anthracene, benz [alanthracene, and chrysene-added to petroleum waxes both individually and combined, have been recovered at as low a concentration as 0.01 p.p.m. Recovery ranged from 35% for benzo [alpyrene to 94% for dibenz[a,h]anthracene. The analytical method was a combination of adsorption chromatography on magnesiaCelite and subsequent paper chromatography using different solvent systems.
M
ANY natural and commercial prod-
ucts are of interest in studies of environmental cancer because they may contain very small quantities of polycyclic aromatic hydrocarbons, some 684
ANALYTICAL CHEMISTRY
of which are carcinogenic. As some petroleum products may contain traces of polycyclic aromatic hydrocarbons (not necessarily carcinogenic), methods for the isolation and estimation of those hydrocarbons added to such largely aliphatic material were studied. As representative of the more difficult type of substance which might be encountered, two petroleum waxes were chosen; it has been difficult to determine trace components in waxy materials, because of bheir low solubility. Five polycyclic aromatic hydrocarbons-dibenz [qhlanthracene, benzo [a]pyrene, benz [a]anthracene, chrysene, and anthracene-were added to wax a t three concentrations-1, 0.1, and 0,Ol p.p.m.-both separately and as a mixture, and their separation and quantitative recovery were attempted. Adsorption chromatography seemed
the best means of separating the polynuclear hydrocarbons from the nonaromatic material. The most suitable adsorbent for such a separation was magnesium oxide (calcined magnesite), mixed with Celite to permit easier filtration ( 9 ) . Magnesium oxide is an extremely powerful adsorbent; benzene is unable to elute tetracyclic or pentacyclic aromatic hydrocarbons from it. This is important because large volumes of benzene had to be washed through the chromatographic columns to remove the wax. The very high adsorption affinity of magnesia for the polynuclear hydrocarbons prevented their quantitative removal from the adsorbent by simple elution with polar solvents (acetone and ethyl alcohol), Thus, from 20 grams of magnesia on which were adsorbed 100 y of dibenz[a,h]anthracene, only 10 to
20% could be eluted with 3 liters of a mixture of ethyl alcohol and acetone. However, by dissolution of the magnesia in dilute acid the adsorbed material mas liberated and could be eluted easily from the residual Celite. Zechmeister and Cholnoky ( 7 ) refer to cases in which such a procedure had to be used. Because of the very small quantities of polycyclic aromatic hydrocarbons to be identified, paper chromatography was emplo!.ed for further resolution of the material adsorbed on the magnesia. Such procedures have been described by Tarbell e2 al. (6) [whose method has been used by Van Duuren (6) to separate the components of tobacco tar] and by Bergmann and Gruenwald ( I ) . The former used N,N-dimethylformamide as the stationary phase and hexane as the mobile phase, while the latter used toluene - methanol water. The dimethylformamide system was superior for resolving synthetic mixtures of hydrociti hons, and was improved further by replacing hexane with 2,2,4-trimethylpentane, as described by Mitchell (4). Much of the colored material in the fr:ictions obtained from a wax chromatogram migrated with the polycyclic hydrocarbons in the dimethylforninmide system, whereas in the toluene-methanol-water system most of the colored material remained in the starting spot. The latter system, therefore, \ \ a s preferable for the initial stage of the paper chromatographic separation. TKO\vases were used: a microcrystalline wax for the hydrocarbons a t the 1- and 0 1-p.p.m. concentrations and a white paraffin wax a t the 0.01-p.p.m. concentration. Analysis of the two waxes without addition of polycyclic hydrocarbons served as controls on the polynuclear hydrocarbons already present in the wases and on the materials and procedures used.
-
REAGENTS AND APPARATUS
The polycyclic aromatic hydrocarbons aere obtained from Eastman Kodak (dibenz [a, hlanthracene), Hoffmann-LaRoche(Lenzo[a]pyrene), and L. Light (benz [u ]anthracene) Anthracene and chrysene were prepared from the commercial products. All were purified by chromatography before use. Melting points (corrected) were: benz [alanthracene 157-8', anthracene 213-14 ', benzo [alpyrene 177-8", chrysene 2titi-6', and dibenz[a, hlanthracene 267-8'. Hexane and 2,2,4-trimethylpentane (iso-ortanc) nere from Phillips Petroleum Co.; acetone, benzene, toluene, methanol, and hydrochloric acid were analytical grade from Fisher Scientific Co. ; iV,S-dimethylformamidewas from E. I. du I'ont de Kemours & Co. Magnesia (Seasorb So. 43) WRS from Food, blachinery and Chemical Corp., I
Table I. Absorptivities of Hydrocarbons
Dibenz [a,h]anthracene Benzo [a]pyrene Benz jalanthracene Chrysene Anthracene
Absorbance
A, MIJ
a'u
of 1 y / M .
298 298
239,000 55 ,500
0.86
288
269 252
and Celite No. 545 was from JohnsManville. The waxes were both commercial samples-yellow microcrystalline wax, melting point 171' F.; and white paraffin wax, melting point 131.5' F. The chromatographic columns were of the two-piece type with perforated glass disk from Scientific Glass Apparatus Co. Spectra were taken on a Beckman DK-1 spectrophotometer. Paper chromatography using Whatman No. 1 paper was carried out in circular glass tanks, using the descending technique. The atmosphere in the tanks was saturated with the developing solvent, contained in a dish at the b o b tom of the tank. Fluorescence was observed with a Mineralight, Model SL 3660, long wave (Ultra Violet Products, Inc., San Gabriel, Calif.). PROCEDURE
Solutions of dibenz[a,h]anthracene, benzo [alpyrene, benz [alanthracene, chrysene, and anthracene in iso-octane (50 y per ml.) were made up. One milliliter of each solution was diluted to 50 ml. with iso-octane and the absorbance curves were taken. The absorbance maxima so obtained (Table I) mere used to determine the recovery of the hydrocarbons added to the waxes. Mixtures of the various polycyclic hydrocarbons with the waxes were made by adding 2 ml. of the above solutions to 100 grams of microcrystalline wax (1 p,p.m.), 0.2 ml. to 100 grams of microcrystalline wax (0.1 p.p.m.), and 0.2 ml. to 1000 grams of paraffin wax (0.01 p.p,m.). I n this way mixtures were made of the five hydrocarbons in microcrystalline wax both individually and combined. The 0.01 p.p.m. experiment was performed only on a mixture of all five hydrocarbons in paraffin wax. Separation of Aromatic from Nonaromatic Material. The mixtures of wax and polycyclic hydrocarbons were dissolved in approximately 1 liter of benzene, kept near the boiling point on a water bath. The clear solution thus obtained was poured through a 24 X 4.5 cm. column of magnesiaCelite (2 to 1 by weight, thoroughly .mixed), which had been packed under vacuum and washed with 500 ml. of boiling benzene. Aspiration was continued throughout the chromatography. Filtration was fairly rapid and the lower part of the column was heated with an infrared lamp to avoid clogging that part of the column with wax. After filtration of all the wax solution, the column was washed with 500 ml. of hot ben-
79,500 167,000 194,000
0.22
0.35 0.69 1.09
zene to remove the residual wax and developed with 500 ml. of a mixture of hexane-benzene-acetone (3 : 1:1). The column was allowed to dry and the adsorbent extruded with a metal plunger. It was cut into three fractions: (1) the upper 3 ern., (2) the next 7 cm., and (3) the remaining 12 em., each of which was made into a slurry with acetone. The division of the column was based on prior standardization of the adsorbent using 0.1 mg. of each of the five polycyclic hydrocarbons on a column of the same size and using the same developing procedure as for the wax solution. In these circumstances benzo[alpyrene and dibenz[o,h]anthracene were in the upper 3 cm., benz[a]anthracene and chrysene were in the next 7 cm., and anthracene could not be located on the column. Fraction 3 was poured into a sinteredglass funnel (600-ml. capacity, medium porosity), filtered and washed successively with benzene (about 150 ml.) and 95% ethyl alcohol and acetone (approximately 1 to 5 ) ; a total volume of 2 liters of filtrate was collected. The combined washings were transferred to a separatory funnel, water was added, and the upper benzene layer was washed continuously until free of water-miscible solvents, using a perforated glass bulb maintained in the benzene layer as described by LeRosen (2). The benzene layer was dried with sodium sulfate. Fractions 1 and 2 of the chromatogram were each poured into a sinteredglass funnel and sucked dry. Acetone was removed by washing with about 50 ml. of water, to avoid any chemical reaction as a result of the heat produced during solution of the magnesia. Dilute hydrochloric acid (1 to 3) was added slowly with stirring until effervescence stopped, all the magnesia having been dissolved. The effervescence was due to the small amount of magnesium carbonate in the adsorbent. This required 200 to 300 ml. of acid. The filtration was continued; acetone and ethyl alcohol ( 5 to 1) with about 100 ml. of benzene were used to elute adsorbed material from the residual Celite, and a total volume of 2 liters of filtrate was collected. The solution was treated in a separatory funnel as described for fraction 3 and the benzene solutions were dried. The volume of each of the three solutions was measured, the absorbance a t 290 mp was taken, and these absorbances were converted to that of the material in 100 ml. of solution. The solutions were evaporated to dryness under nitrogen, leaving 20 to 50 mg. of yellow or brown oil. This residue was dissolved VOL. 32, NO. 6, MAY 1960
685
Table II.
Hydrocarbon Chrysene
Recovery of Hydrocarbons Added to W a x Hydrocarbons Added to Concn. in A in 5 MI. Wax, P.P.M. Iso-octane Wax, Y 1 0.1
Mixture 0.01
Benz [ a ]anthracene
1
0.1
Mixture Dibenz [a$] anthracene
0.01 1
0.1 Mixture
0.01
Benzo[a]pyrene
1 0.1
Mixture Anthracene
0.01 1 0.1
Mixture 0.01
in a small volume of benzene determined empirically according to the absorbance a t 290 mp, If the absorbance was approximately 10, the volume of benzene was 2.0 ml. If the absorbance differed from 10 by a factor of 2 or more, a corresponding adjustment was made in the volume of benzene used-for example, fraction 1 of microwax, absorbance 6 to 8, made up to 2 ml. in benzene; fraction 1 of paraffin wax, absorbance 2 to 3, made up to 1 ml. in benzene. Resolution of Aromatic Mixture. From a micropipet 0.05 ml. of solution was placed within a spot 1 cm. in diameter on each of several 1 X 48 cm. strips of filter paper (many of which were joined together a t top and bottom by cutting 0.3 X 48 em. slots in a large sheet of paper). Solvent was removed in a stream of nitrogen. The number of strips run a t one time varied according to the quantity of hydrocarbon expected in the solution. It was arranged so that 2 to 5 y of hydrocarbon would be present (two strips of the 1-p.p.m. fractions, eight strips of the 0.1-p.p.m. fractions, and 10 strips of the 0.01-p.p.m. fractions). The strips were placed in a trough of toluene-methanol-water (1:lO:l) and the solvent was allowed to descend. The rate of development varied from run to run, but spots containing 2.5 y of benzpyrene and 2.5 y of benzanthracene moved 30 cm. down a control strip in 4 to 5 hours. At this point the strips were removed from the tank, dried for 1 hour or longer a t 37", and cut into sections, first those corresponding roughly in position to the benz [alanthracene and benzo [alpyrene spots, and then those above and below. This ensured that virtually none of the polycyclic hydrocarbons in the wax fractions remained on the paper. Analogous sections of paper from the same wax fractions were combined and eluted four times with approximately 5 ml. of a warm mixture of ethyl alcohol-iso-octane (1 to 4). The combined eluate was 686
ANALYTICAL CHEMISTRY
10 6 6 10 10 6 6 10 10 6 6
10
15 6 6 10 10 6 6 10
1.02 0.42 0.54 0.86 0.29 0.21 0.18 0.29 1.43 0.83 0.68 1.63 0.29 0.095 0.115 0.195
9%
Recovery
...
74 50 65 62 42 48 43 42 83 80 67 94 43 35 43 45 0
...
0 0 0
... ...
evaporated to 5 ml. in a small tube. The absorption spectrum of each fraction was taken. The reference solution in the spectrophotometer was obtained by elution of a strip of paper chromatographed without addition of wax fraction or polycyclic hydrocarbon. The spectrum of this blank solution showed that no material absorbing above 250 mp was eluted from the paper. Those fractions showing evidence (by inflections in the spectrum) of the presence of the hydrocarbons added to the wax were evaporated to dryness under nitrogen, dissolved in a small volume of benzene, and rechromatographed on 1 X 48 em. strips of paper saturated with dimethylformamide, using isooctane as the mobile solvent. Addition of the benzene solutions was timed to take approximately 1 hour, during which the paper became partially dry. The chromatograms were allowed to develop 4 to 7 hours, until a control spot of benzanthracene or benzopyrene had moved about 30 em. down the paper as viewed with an ultraviolet lamp. Separation of dibenzanthracene from benzopyrene was almost complete under these conditions, but benzanthracene and chrysene could not be separated from one another in this system. The fluorescent spots and nonfluorescent portions of each paper strip were cut out and eluted separately. Chrysene (the fluorescence of which was too weak to be detected in these conditions) was located by analogy with a benzanthracene spot on another strip. The spectrum of each fraction was taken after elution with iso-octaneethyl alcohol as described above. Those fractions from the same chromatographic column which, a t this stage, had similar spectra-Le., containing the same polycyclic hydrocarbons-were combined, evaporated and, if necessary, rechromatographed in the dimethylformamide-iso-octane system. Chromatography was repeated in the same way until a spectrum was obtained of the various hydrocarbons corresponding ex-
actly with those of the pure compounds and, finally, the maximal absorbance was measured in a suitable volume of isooctane. The quantity of each hydrocarbon recovered was calculated and expressed as a percentage recovered of the hydrocarbon added to wax. To increase the accuracy of the determination, the final solutions obtained from several paper chromatographic runs of the same column fractions were combined and the total quantity of hydrocarbon was determined, from which an average value for the recovery was obtained. For example, in the 1-p.p.m. experiment, fractions corresponding to 10% of the wax were combined; in the 0.1-p.p.m. experiment, fractions corresponding to 60% of the wax. In the 0.01-p.p.m. experiment, the entire wax fractions were used. Benzanthracene was separated from chrysene, a t the final stage, on strips of filter paper which had been acetylated ( 1 ) in a mixture of acetic anhydride and toluene (3 to 1) for 1 hour a t SO", vvashed free of the reagents with methanol and water, and dried a t 37" for a t least 48 hours. The developing solvent was toluene-methanol-water (1 :10: 1). Development !vas rather slow, benzanthracene requiring 12 hours to move 30 em. down the strip. At this stage the benzanthracene was separated by at least 2 cm. from the upper chrysene zone. Both zones were diffuse and it was necessary to cut the entire strip into portions to ensure elution of all the chrysene (which was nonfluorescent). RESULTS A N D DISCUSSION
The recovery of the various hydrocarbons added to wax is shown in Table 11. None of the five polycyclic hydrocarbons in this study could be detected in either wax itself, showing that, if present, their concentration was below the sensitivity of the method. In a control experiment some colored material was released from magnesia on treatment with acid, but this did not contain polynuclear hydrocarbons. Recovery of benzopyrene and benzanthracene a t all concentrations was not as good as that of dibenzanthracene and chrysene. This is probably due to the considerably lower absorption intensity of the two former compounds, which led to the discarding of some fractions containing small quantities of them which, in the case of dibenzanthracene and chrysene, would have had sufficiently marked inflections to be retained and rechromatographed. The examination of all three fractions from the column chromatograms of the waxes showed that, in each case, dibenzanthracene and benzopyrene were in fraction 1, benzanthracene and chrysene were in fraction 2, and anthracene could not be detected, even a t the highest concentration in the wax. A trace of benzanthracene and chrysene was present in fraction 1 a t the highest
concentration (1 p.p.m.) but not a t the two lower concentrations. It may be concluded, therefore, that under conditions described, dibenzanthracene and benzopyrene are retained in the upper 3 cm. of a magnesiaCelite column, benzanthracene and chrysene are adsorbed in the 4th to 10th cm. of the column, and anthracene is not retained by magnesia-Celite under these conditions. All four adsorbed polycyclic hydrocarbons may be detected and estimated with a minimum of 35% recovery, a t a concentration as low as 0.01 p.p.m. Different batches of magnesia vary in adsorptive power and should be standardized with known hydrocarbons. The behavior of the four polycyclic
hydrocarbons studied suggests that most polycyclic aromatic hydrocarbons with four or more fused rings can be separated from complex mixtures and estimated with reasonable accuracy using the method described. ACKNOWLEDGMENT
The author is greatly indebted to C. R. Raha for his help in developing the paper chromatographic procedures, particularly that using acetylated paper, and for his advice and criticism, and to Philippe Shubik for his constant interest in the work. LITERATURE CITED
(1) Bergmann, E. D., Gruenwald, T., J . Appl. Chem. (London)7, 15 (1957).
(2) LeRosen, A. L., IND.ENG. CHEM., ANAL.ED. 14, 165 (1942). (3) Lijinsky, W., Saffiotti, U., Shubik, P., J. Natl. Cancer Inst. 18, 687 (1957).
(4)hlitchell, L. C., J . Assoc. Ofic. Ayr. Chemists 4 2 , 161 (1959). (5) Tarbell, D. S., Brooker, E. G., Vanterpool, A., Conway, W., Claus, C. J., Hill, T. J., J . Am. Chem. SOC.77. 767 (1955). ’ (6) Van Duuren, B. L., J . Natl. Cancer Inst. 21, l(1958). (7) Zechmeister, L., Cholnoky, L., “Principles and Practice of Chromatography,” trans. from 2nd German ed., 2nd impr., p. 55, Wiley, New York, 1943.
RECEIVED for review August 21, 1959. Accepted March 7 , 1960. Work carried out as part of Project MC-3, American Petroleum Institute.
Chromatographic Separation and Identification of Organic Acids Application to Detection of Organic Acids in River Waters H. F. MUELLER, T.
E. LARSON, and M. FERRETTI
lllinois State Water Survey, Urbana, 111.
b Although organic acids are present in river waters in relatively low concentrations, they can to some extent be separated and identified by chromatographic methods. Preliminary data from 25 river samples illustrate the potential of chromatography for the evaluation of water quality. An alkaline evaporation of the sample,followed by an ether extraction of the acidified aqueous residue, is used to prepare the sample for the separation of individual acids or groups of acids by column partition chromatography. Identity of some acids is conflrmed by a comparison of the R, values obtained for pure acids by partition chromatography on paper.
A
organic acids are seldom responsible for tastes and odors (6), their concentrations in surface water supplies may be several times the concentration of neutral and basic organics present. This report concerns the effectiveapplication of the combined techniques of column and paper partition chromatography for the separation and identification of water-soluble acids. The acids are separated by the column method and are tentatively identified LTHOUQH
by their respective elution patterns. Confirmatory identification of some of the separated acid fractions is made by partition chromatography on paper. EXPERIMENTAL
Sample Preparation. The water samples are collected in a container t o which a few pellets of sodium hydroxide have been added. After filtration, a sample volume of 3 to 4 liters is heat-concentrated t o approximately 30 ml. The concentrate is acidified with mineral acid and clarified by filtration when necessary, and the organic acids are extracted with ethyl ether for 15 hours. After extraction, the acids are titrated with dilute sodium hydroxide using bromothymol blue as the indicator. The neutralized solution is then warmed on a steam bath t o volatilize the ether layer, and the aqueous phase containing the acid salts is concentrated to a volume of 1 ml. This sample extract is acidified and adsorbed on silicic acid for addition to the column. Chromatography, COLUMNPARTITION METHOD. T h e acids are separated initial1 by the method of Bulen, Varner, and Surrell (6)as modified by Mueller, Larson, and Lennare (19) for the quantitative determination of the lower volatile acids. Silicic acid
is used as the adsorbent, onto which is fixed dilute sulfuric acid as the stationary phase. The mobile phase consists of various concentrations of 1butanol in chloroform. The procedure differs from that described only in that effluent fractions of 5 ml. are collected and titrated with 0.02N alcoholic sodium hydroxide using 0.1% bromothymol blue as the indicator. Blank values are determined for each solvent system and subtracted from the re spective effluent titration values. By this technique 20 or more acids could be partially separated. Each acid has a characteristic peak effluent volume, which is that fraction having the highest titration value. The positions of the acids separated by this method are shown in the reference chromatograms in Figures 1, 2, and 3. The locations of these acids were determined by repeated appearance within one fraction of the position indicated. Recoveries in excess of 90% r e r e obtained for all acids. All of the acids are not separated completely by this method, because two or more acids often have the same elution pattern. I n some cases these can be separated by further column treatment modified for the particular acids in question. An example is the peak VOL. 32. NO. 6, M A Y 1960
687