1392 I N D UXTRIAL AND ENGINEERING CHEMISTRY Vol. 20, No. 12

Figure 1 contained 2 oleate to 1 stearate, while Figure 2 contained ester in the ratio of 1 oleate to 1 stearate. The backgrounds in Figures 1 and 2 appear black. They were actually white and clear as Figure 3, but the lighting was arranged to give this effect in order to show the crystals by contrast.

When ethyl oleate is present in the proportion of 4 parts oleate to 1 part stearate, this crystallization of the stearate from the film is not observed; the ratio of 3 oleate to 1 stearate was not studied.

The authors believe that this experiment throws light on the blooming phenomena (often observed in varnishes, in certain simple oil-nitrocellulose lacquer coatings, and in patent leather finishes) which appear after repeated elevation and dropping of the temperature incidental to climatic changes. The development of tacky, odorous, and dark- colored produets by the films composed of the ethyl esters of linseed mixed fatty acids also furnishes evidence for the source of the development of odors and color in varnished insulation fabrics as well as varnished papers used for bottle caps and for other purposes.

In comparison with the films of ethyl esters, one was pre- pared containing polymerized oil. This film remained flexible and dry and showed no crystallization phenomena. It yel- lowed on 10 months’ storage slightly more than the ethyl oleate films, but by no means to the extent that the esters of unpolymerized linseed oil did.

When exposed outdoors on the roof for 1 month (Septem- ber) polymerized linseed-oil films remained flexible, while those of all the esters mentioned first became sticky and then brittle, thus demonstrating the added advantage of lowered iodine number of the oleaginous component of the films brought about through polymerization.

Conclusions

The esters of the less unsaturated fatty acids of the type of oleic are very stable in films. The esters of the more unsaturated acids of linseed oil when present in the unpolym- erized form are not so stable, as evidenced by rapid tend- ency to become sticky, odorous, and dark-colored. It is therefore of advantage to compose films of lower iodine number fatty constituents which may be effected by use of oleic derivatives 01 substances or similar iodine number range obtained by polymerization of linseed oil, or, certain mixtures of oleic and stearic derivatives obtained by hydrogenation of linseed oil. This eliminates the tendency to become sticky and dark in color.

The darkening in color of certain of the films as contrasted with the failure t o darken in others affords support to the current belief that yellowing of drying-oil films arises from oxidation of the highly unsaturated fatty acid groups and that it and drying of linseed films are not interdependent. It is also independent of the presence of glycerol and is a necessary consequence of such oxidation.

Action of Ultra-Violet Light upon Ferric Citrate Solutions’”

H. Shipley Fry and Elmer G. Gerwe

DEPARTMENT OF CHEMISTRY. UNIVERSITY OF CINCINNATI, CINCINNATI, OHIO

HIS quantitative study of the action of ultra-violet light upon solutions of citric acid in the presence of T ferric salts was suggested and fostered by John Uri

Lloyd,3 a pioneer in the manufacture of pharmaceutical preparations containing compounds of iron. Such solutions, containing citrates in the presence of ferric salts, when unprotected from the action of sunlight, suffer decomposi- tion to such an extent that stoppers may be blown from their glass containers by the accumulation of the evolved carbon dioxide. Accordingly, the purpose of this study was to determine whether or not carbon monoxide or any other deleterious decomposition products of citric acid are formed, and to ascertain not only the specific nature of the involved photochemical changes but particularly to what extent, quantitatively, these changes might be dependent upon the concentrations of the ferric salts present when standardized solutions containing citric acid and varying multiple molar quantities of ferric sulfate were exposed to the action of ultra-violet light.

Previous Work

Eder4 and Vries5 found that ordinary light affected the re- duction of ferric salts to ferrous in the presence of oxalic, citric, tartaric, and malic acids. Vries stated that the ferric

1 Received July 28, 1928. 2 Synopsis of a section of a thesis presented by Elmer G. Gerwe to the

faculty of the Graduate School, University of Cincinnati, June, 1927. in partial fulfilment of the requirements for the M.A. degree.

8 Proc. Am. Pharm. Assocn., 27, 741, 743, 744 (1879). 4 Bn., 82, 606 (1880). 6 Chem. Zcnlr., 66, 219 (1885).

salts act as catalysts, “carriers” of oxygen, since a small amount of ferric chloride oxidized a great excess of acid. This was due to the fact that the reactions were conducted with access to air, which continuously re-oxidized the ferrous salt to ferric and thereby led to the complete oxidation of the carbon compound.

Wisbare and Seekamp’ conducted photochemical reactions in the presence of uranium salts. Seekamp observed that uranic oxide, similarly to ferric salts, suffered reduction.

Neuberg8 has made an extensive biochemical study of the action of both sunlight and ultra-violet light upon sixty- odd substances, some in the presence of ferric salts, others in the presence of uranic salts. He also exposed drugs con- taining iron to the action of sunlight and measured the extent of the reaction by quantitative determinations of the yields of certain products, but did not establish any stoichiometrical ratios for possible specific reactions. His work embodies some excellent ~ummar ies .~

None of the earlier investigators attempted to establish exact stoichiometrical relationships between the quantities of the ferric salts employed and the yields of the various prod- ucts of oxidation. However, Berthelot and Gaudechon,’” the first to employ ultra-violet light as the source of energy, determined quantitatively the proportions of carbon dioxide and hydrogen involved when solutions of oxalic acid were

6 Ann., 262, 232 (1891). 7 I b i d . , 278, 373 (1894). 3 Bioclrem. Z., 44, 495 (1912). 9 I b i d . , 13, 305 (1908); 17, 270 (1909); 27, 271, 279 (1910); 99, 158

(1912); 44, 495 (1912); 71, 219 (1915). 10 Compt. rend., 168, 1791 (1914).

December, 1928 INDUSTRIAL AND ENGINEERING CHEMISTRY

exposed and noted that the relative volumes of these gases varied with the wave lengths of the rays used. The continua- tion of this work by Allmand and Reeve” was directed to a study of the energetics invohed as the most accurate method of determining to what extent a secondary or tertiary reaction takes place. They made no attempt to isolate any decom- position products other than gaseous ones, and their results are given with the understanding that they hold for the “initial stages” of the photolytic reactions only.

As to the photochemical decomposition of citric acid in presence of ferric chloride, Benrathlz noted that acetone and ca rbon dioxide were the final products of oxidation, d i ca rboxy l i c acetone and acetoacetic acid be ing t h e i n t e r m e d i a t e p roduc t s . Euler and Ryd,lJ and like- wise Ciamician and Silber14 recorded the formation of acetone and carbon dioxide.

It is of historical interest to note that Liebig15 in 1859 found that acetone and car- bon dioxide were products of the oxidation of c i t r i c acid by manganese per - oxide.

Preliminary Experiments

A review of the literature, as briefly noted, furnishes no quantitative data to es- tablish any stoichiometrical relationships between t h e initial quantities of fe r r ic salts present and the corre- sponding yields of carbon d iox ide when citric acid

1393

The function of the ferric sulfate may be regarded as the oxidation of the hydrogen formed in equation (1) in conformity with the following equation:

Fe2(S04)3 + HZ + 2FeS04 + HzSOi (4)

From these points of view the photochemical change, equation (l), is promoted by the oxidizing action of the ferric sulfate in removing the hydrogen, as noted in equation (4), and reactions (1) and (4) are followed by the concurrent reactions (2) and (3), which are not, per se, oxidation-re- duction processes. The summation of equations (1). (2),

Pharmaceutical solutions containing citric acid and ferric compounds are readily decomposed when ex- posed to light. Carbon dioxide is evolved, acetone is formed, and the ferric compounds are reduced.

This quantitative study of the action of ultra-violet light upon solutions containing unit concentrations of citric acid in excess and varying concentrations of ferric sulfate shows that three molecules of carbon dioxide are liberated for every molecule of ferric sulfate present.

It was assumed that the photochemical change in- volves the oxidation of citric acid, with the liberation of one molecule of carbon dioxide, to the unstable acetone-dicarbonic acid. This also loses a molecule of carbon dioxide, thereby yielding acetoacetic acid. The latter readily decomposes forming acetone and a third molecule of carbon dioxide. The summation of the equations for these assumed intermediate reactions gives a complete equation (CHzCOOH)&(OH)COOH + Fe~(S0dr --f CHaCOCHa + 3COs

which calls for the stoichiometrical ratio Fe2(SO&: 3C0,. The quantitative data obtained confirm this ratio and the proposed reaction mechanism.

+ 2FeS01 + HzSOl

is decomposed by the action of ultra-violet light. To this end the experimental

. ,, . . , (3), and (4) gives:

+ Fe2(S04)3 --f CH3CO- CH2 + 3COn $- 2FeSOd + (CHzCOOH)tC(OH)COOH

HzSO4 ( 5 )

The establishment of the h i t h e r t o u n d e t e r m i n e d stoichiometrical ratio, Fez- (SO&: 3C02, which is the i m m e d i a t e object of this quantitative study, will not only confirm the proposed s u m m a t i o n equation (5), representing the final result of the action of ultra-violet light upon solutions of citric acid in the presence of ferric sulfate, but will also lend direct support to the theory of the reaction mechanism as involving the several re- ac t ions r ep resen ted b y equations (I), (2), (3), and (4) *

Experimental

ULTRA-VIOLET LIGET- procedure involved the use

Preli&ary experiments, using first sunlight and then ultra-violet light, confirmed the qualitative observations of the previously noted investigators. As the reactions pro- ceeded to completion the green solutions became colorless and the ferric salt (ferric sulfate) was completely reduced to the ferrous state.

Since acetone dicarboxylic acid and acetoacetic acid are the products intermediately formed, the complete reaction mechanism may be assumed to involve four distinct changes. The first, a photochemical change, is presumably the con- version of a molecule of citric acid to acetope dicarboxylic acid with the elimination of one molecule each of hydrogen and carbon dioxide according to the following equation:

(CHzCOOH)zC(OH)COOH + (CHZCO0H)zCO + Hz + COS (1)

The instability of acetone dicarboxylic acid immediately leads to the formation of acetoacetic acid and a second mole- cule of carbon dioxide, thus:

(CHtC0OH)zCO --f CHsCOCHzCOOH + COz (2)

Lastly, the instability of acetoacetic acid yields acetone and a third molecule of carbon dioxide according to the equation:

CHaCOCHzCOOH + CHaCOCHa + COS (3)

1 1 J. Chem SOC , 1926, 2834. 12 Z. p h y s i k . Chem., 74, 118 (1910); Ann., 382,222 (1911). 18 Biochem. Z., 51, 97 (1913). 14 Bcr., 46, 1558 (1913). 15 A n n . , 113 (1859).

of a quariz mercury vapor lamp, spectra range 1850 A. to 14,0000 A., two-thirds of the total radiation being less than 4500 A. No attempt was made to employ specific radiation by means of interposed filters. In all experiments quartz tubes, 2 X 13 cm. of 20 cc. capacity, containing the stand- ardized reaction mixtures, were placed practically adjacent and parallel to each other, directly beneath and equidistant from the source of light, and inclined a t an angleof 45 degrees.

runs (A, B, and C) were conducted and in each two quartz test tubes were used-one designated as “sample,” the other as “blank.” The contents of the standardized reaction mixture in the “sample” tubes were made up according to this tabulation of concentrations of reacting components:

PREPARATION O F REAcTIoN-MIxTUREs-Three sets Of

RUN CITRIC ACID FERRIC SULFATE WATER TOTAL VOLUME (1 M soln.) (0.1725 M soln.)

cc. cc. CC. CC. A 9 B 9 C 9

5 7 . 5

10

5 2.5 0

19 19 19

In juxtaposition to each sample tube in each of the runs there was placed a blank tube containing 9 cc. of 1 M citric acid solution and 10 cc. of distilled water-likewise a total volume of 19 cc. Thus the concentration of citric acid was identical in each sample and in each blank tube; but in the sample tubes, in runs A, B, and C, the molar concentrations of the ferric sulfate stood, respectively, in the ratio 1: 1.5: 2, and the molar concentration of the citric acid was many times that required by the molar proportions of citric acid

1394 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 20, No. 12

and ferric sulfate, as indicated ia the complete summation equation ( 5 ) .

The blank tubes were used in determining the quantity of carbon dioxide eliminated through the action of ultra- violet light upon citric acid alone. Subtraction of the weight of carbon dioxide evolved from the blanks from the weight of that evolved from the samples gave the weight of carbon dioxide eliminated by the action of the light upon the citric acid in the presence of ferric sulfate. Accordingly, in order to verify the stoichiometrical ratio, Fes (so&:3co2, required by equation (5) , the quantities of carbon dioxide so eliminated in runs A, B, and C should conform to the respective molar concentrations of the ferric sulfate-namely, 1: 1.5: 2. In other words, run B should yield 1.5 times, and run C, 2 times as much carbon dioxide as is obtained in run A. DETERMINATION OF CARBON DIoxrDE-Each sample and

each blank tube, used concurrently in each run, practically filled with their respective reaction mixtures (thus excluding any air and thereby preventing excessive oxidation through the regeneration of ferric sulfate), was securely fitted with a two-holed rubber stopper bearing an inlet and an exit tube. The inlet tube, of capillary dimension, extended to the bottom of the quartz tube and served for the introduction of a slow, steady, continuous current of pure nitrogen, which kept the reaction mixture well stirred and carried over all of the evolved carbon dioxide through the exit tube to a train of apparatus comprising two U-shaped drying tubes filled with calcium chloride, a Liebig bulb containing concentrated sulfuric acid, and lastly, a weighed Stesson-Norton glass- stoppered carbon dioxide absorption bottle filled with As- carite (a proprietary sodium hydrate asbestos absorbent mixture), for the fixation and ultimate determination of the evolved and dried carbon dioxide from each blank and from each sample tube, respectively.

Table I--Yields of Carbon Dioxide i n Runs A. B. and C WT. ASCARITE WT. WT. ASCARITE WT.

BULB COZ BULB co2 RUN TIME (Sample) (Sample) (Blank) (Blank) COa

As A b As-Ab Hours Grams Gram Grams Gram Gram

A-0 0 93.0520 0 91.6342 0 0 B-0 0 90.4628 0 95.0324 0 0 c-0 0 92.7562 0 90.8140 0 0 A-2 R-2 c-2 A-4 B-4 c - 4

93.0869 90.5137 92.8213 93.1520 90.5718 92.8920

0.0349 0,0509 0.0651 0.0381 0.0581 0.0707

91.6378 95.0362 90.8165 91.6419 95.0405 90.8193

0.0036 0.0038 0.0025 0.0041 0.0043 0,0028

0.0313 0.0471 0.0626 0.0340 0.0538 0.0679

A-6 6 93.1567 0.0317 91.6464 0.0045 0.0272 B-6 6 90.6138 0.0420 95.0457 0.0052 0.0368 C-6 6 92.9503 0.0583 90.8231 0.0038 0.0545 A-8 8 93.1698 0.0131 91.6502 0.0038 0.0093 B-8 8 90,6320 0.0182 95.0506 0.0049 0.0133 C-8 8 92.9730 0.0227 90.8288 0.0057 0.0170

B-10 10 90,6385 0.0065 95.0542 0.0036 0.0029 C-10 10 92.9816 0.0086 90.8334 0.0046 0.0040

€3-12 12 90.6422 0.0037 95.0587 0.0045 -0.0008 (2-12 12 92.9879 0.0063 90.8379 0.0045 0.0018

A-IO IO 93.1778 o.oo80 91.6555 0.0053 0.0027

~ - 1 2 12 93.1823 0.0045 91.6604 0.0049 -0.oon4

A-14 14 Completed Completed Completed Completed Completed B-14 14 Completed Completed Completed Completed Completed C-14 14 92.9926 0.0047 90.8430 0.0051 -0.0004

PRocEDuRE-In each run one sample and one blank tube containing its standardized contents of reaction mixture were simultaneously exposed to the same intensity of ultra-violet radiation for a total period of 12 to 14 hours-that is, until the difference between the yield of carbon dioxide evolved from the sample (As)-and that evolved from the blank (Ab) was practically constant. Thus when As - Ab reached its approximate minimum value-that is, just before this difference gave a minus reading-it was naturally assumed that the rates of the decomposition of the citric acid, both in the sample and the blank, were practically the same or, in other words, all the ferric sulfate in the sample had been re- duced to ferrous sulfate and the concomitant oxidation of

the citric acid to acetone and carbon dioxide, in conformity with the proposed summation equation (5) , had been com- pleted. Table I embodies all the data of the three runs.

Discussion of Results

At the end of each 2-hour period during the runs the blank tubes, each containing the same concentration of citric acid and no ferric sulfate, evolved fairly uniform quantities of carbon dioxide-that is, the Ab values varied from 0.0025 gram to 0.0051 gram; but from the sample tubes at the end of each corresponding 2-hour period, the quantity of carbon dioxide (As) evolved in each C run was greater than that of each corresponding B run, and in like manner the B-run quantities were greater than corresponding A-run quantities. In other words, with increasing molar quantities of ferric sulfate, increasing quantities of carbon dioxide were evolved. These values are noted as As - Ab in the last column.

The total quantities of carbon dioxide thus evolved during the total time required for the completion of the reaction according to the proposed summation equation ( 5 ) were found by adding all the increments, As - Ab, possessing a plus value, for when As - Ab attained a minus value no carbon dioxide was being eliminated through the action of ferric sulfate which was entirely reduced to ferrous sulfate. At this stage As - Ab would necessarily have a minus value, since the concentration of citric acid in the blank was greater than that in the sample. Accordingly, the total quantity of carbon dioxide evolved by the action of the light and ferric sulfate upon citric acid in run A is closely approximated and estimated as the sum of the increments As - Ab from A-0 to A-10, inclusive. For run B the summation includes increments As - Ab from B-0 to B-10, and likewise run C includes the increments from C-0 to C-12.

Table I1 embodies the molar quantities of ferric sulfate used in runs A, B, and C; the corresponding yields of carbon dioxide evolved according to the proposed summation equa- tion ( 5 ) ; the theoretical yields of carbon dioxide calculated in conformity with the stoichiometrical ratio Fe2(SO& : 3C02 of equation (5); and the per cent theoretical yield of carbon dioxide based upon the same ratio.

Table 11-Summarized Data of Runs A, B, and C

RUN Fez(S0i)s Found Theory Per cent theory COZ FOUND 0.1725 M CARBON DIOXIDB

cc. A 5 0.1045 0.1138 91.8 B 7.5 0.1539 0.1707 90.1 C 10 0.2078 0.2277 90.3

The actual yields of evolved carbon dioxide in runs A, B, and C stand in the ratios of 1:1.5:2, which are the ratios of the quantities of ferric sulfate initially present in the respec- tive sample tubes of these runs. This result, also expressed in Table I1 as the practically constant per cent theoretical yield of carbon dioxide, shows that, under the present ex- perimental method of procedure, the effect and the extent of the action of ultra-violet light upon solutions of citric acid and ferric sulfate are directly proportioned to the molar con- centrations of the ferric sulfate employed, in conformity with the stoichiometrical ratio Fe2(S04)3 : 3C02 indicated in the proposed summation equation (5 ) . The quantitative veri- fication of this stoichiometrical ratio also lends support to the proposed reaction mechanism involving the intermediate reactions represented by the equations (l), (2), (3), and (4), the summation of which gives the established equation (5).

Analysis of the gas evolved in separately conducted runs showed that carbon dioxide was the only gaseous product. No deleterious carbon monoxide was found.

The authors gratefully acknowledge the valuable sug- gestions and assistance in fellowship grants of John Uri Lloyd which have made this study possible.