INDUSTRIAL
AND
ENGINEERING CHEMISTRY
ANALYTICAL EDITION PUBLISHED
BY
THE
AMERICAN
CHEMICAL
SOCIETY
0
HARRISON
E.
HOWE,
EDITOR
Determination of Total and Inorganic Bromide in Foods Fumigated with Methyl Bromide S. A. SHRADER, A. W. BESHGETOOR, AND V. A. STENGER The Dow Chemical Company, Midland, Mich.
methyl bromide escapes fairly quickly from a fumigated food is in accord with their experience, the authors do not feel that his data alone are sufficient to establish this finding conclusively.
The direct determination of methyl bromide present in a food after fumigation is difficult, owing to the ease with which the fumigant is converted to inorganic bromide. In the method proposed, methyl bromide is removed under conditions which lead to the minimum decomposition and is determined as the difference between total bromide and inorganic bromide. -4pplication to several food products indicates that no methyl bromide remains as such for more than 4 days after fumigation.
McLaine and Munro (5) demonstrated that water-soluble bromide is formed during the fumigation of plants. Dudley and eo-workers (3) showed t h a t most of the retained bromide is nonvolatile when a fumigated food is boiled with water. These observations indicate that little methyl bromide is retained as such. According to various authors, bromide conceivably may be retained by a fumigated product in the following ways: 1. As inor anic bromide produced either by hydrolysis of methyl bromi8e to methyl alcohol and hydrobromic acid or by direct reaction of methyl bromide with chemically active constituents of the food, such as sulfur or nitrogen compounds. The former process has usually been suggested as a cause c& bromide retention, but in the authors’ opinion the latter mechanism is a more important factor. In either case, the liberated hydrobromic acid is likely to be fixed as alkali or alkaline earth bromides and should therefore be soluble in water, insoluble in nonpolar organic solvents, and nonvolatile a t moderate temperatures. 2. As methyl bromide adsorbed on the material or dissolved in oily or fatty constituents. This should be soluble in organic solvents and volatile at moderate temperatures, provided it is not allowed to decompose into inorganic bromide. 3. As organic bromide produced by addition of methyl bromide or hydrogen bromide to an unsaturated compound or other attackable group. This type of bromide should be soluble in organic solvents, probably insoluble in water, and robably nonvolatile at moderate temperatures. S o evidence o r t h e formation of such compounds in foods has been encountered in the authors’ work.
M
ETHODS for the determination of total bromide in foodstuffs fumigated with methyl bromide have been published by Dudley (a) and by the authors ( 7 ) . Cncertainty concerning the nature of this retained bromide has led t o the fear t h a t free methyl bromide might be present; in fact, some investigators have considered the total bromide t o be methyl bromide, though without experimental justification. Aside from a few qualitative observations, there seems to be no published information on the state in which bromide occurs in fumigated products. Since the experiments described in this paper were completed, a paper on “Bromide Residues in Foodstuffs” by Laug has appeared (4). Laug determined volatile and nonvolatile bromide by a method proposed by F. L. Hahn. The authors do not feel that Hahn’s procedure is well suited for the determination of methyl bromide in foods, because it involves conditions under which some of the methyl bromide may react during analysis and thus be converted to inorganic bromide. In a comparison of the Laug-Hahn method with theirs on freshly fumigated ground walnut meats, the nonvolatile bromide was found to be from 5 to 10 per cent higher by the former method. Laug fumigated with eight times the usual dosages of methyl bromide in a deliberate attempt to secure high bromide residues, so that most of his dat,a are not representative of commercial conditions and hence are not directly comparable with the authors’. Nevertheless they believe that his values for volatile bromide during the first 48 hours of airing tend to be too low, while those for nonvolatile bromide are too high. Furthermore, his results in some cases show erratic variations which he attributes to nonuniformity of the small samples analyzed. I t seems more likely that they are due to losses during ashing, since the authors had the same difficulty until the conditions recommended in this paper were employed. Although Laug’s conclusion that
hlartinek and Marti (6) employed a method for the determination of methyl chloride in foods, which consists in passing air through the sample a t 100” C. and analyzing t,he vapors. Preliminary experiments showed that this method cannot be applied t o samples containing methyl bromide, for when the temperature is raised methyl bromide reacts more rapidly with some substance in the food and produces inorganic bromide. The same type of reaction takes place if the food sample is boiled with water, although the rate of hydrolysis of methyl bromide in water alone is so low that practically all the halide may be boiled out. This reaction of methyl bromide with food constituents prevents one from preparing known mixtures for testing proposed analytical methods. Attempts were made t o extract methyl bromide from foods by means of organic solvents, as a result of which solvents with small molecules were found t o be more effective t h a n 1
2
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 14, No. 1
those with larger molecules. Methylene chloride was the most promising compound tested, but even successive treatments with it failed to extract the last small amounts of methyl bromide from some materials. Therefore volatilization by boiling with this solvent was resorted t o after removal of most of the methyl bromide by extraction. The boiling point of methylene chloride (40" C.) is low enough so that not much reaction takes place. The problem becomes one of treating the sample under conditions as mild as possible at first, then more and more drastic, until most of the methyl bromide is removed before it has a chance t o decompose. Since the methyl bromide (or other organic bromide) thus comes off in several fractions, it may be determined most easily as the difference between the total bromide originally present and the inorganic bromide remaining after treatment. However, to show that the bromide left is actually inorganic, the sample is finally extracted with water and the solution is analyzed for bromide.
Nickel hydroxide and other insoluble hydroxides are removed by filtering through a No. 2 Whatman paper, collecting the filtrate and washings in a 500-ml. wide-mouthed Erlenmeyer flask. The filtrate is made slightly acid with 6 N hydrochloric acid, then neutralized with sodium hydroxide solution, adjusting to the color change of methyl red. The volume at this point should be approximately 150 ml. About 2 grams of sodium acid phosphate and 5 ml. of hypochlorite solution are added and the mixture is heated to boiling. After a minute or so 5 ml. of sodium formate solution (50 grams 100 ml.) are introduced and boiling is continued for 2 minutes. he sample is cooled and treated with a few drops of 1 per cent sodium molybdate solution, 0.5 gram of otassium iodide, and 25 ml. of 6 N sulfuric acid. Titration sgould be made immediately with standard 0.01 N sodium thiosulfate solution, starch indicator being added just before the end point. A blank on all the reagents should be carried through the entire procedure and subtracted. One milliliter of 0.01 N thiosulfate is equivalent to 0.1332 mg. of bromide ion.
Reagents
1
Methylene chloride. Commercial material usually contains a trace of hydrolyzable bromide which may be removed by shaking 2 liters of the solvent with 15 grams of potassium hydroxide dissolved in 300 ml. of 95 per cent ethyl alcohol and allowing the mixture to stand for several days. The alcoholic potassium hydroxide is washed out with water and the methylene chloride is filtered, dried over anhydrous calcium sulfate, and distilled. Alcoholic potassium hydroxide, 2.5 grams of potassium hydroxide per 100 ml. of 95 per cent ethyl alcohol. Sodium hydroxide, analytical reagent grade. Sodium peroxide, analytical reagent. Hydrochloric acid, about 6 N . This should be as free from bromide as possible. c . P. acid may be diluted to 6 N and distilled, the first and last fractions (each about 10 per cent of the total) being discarded to eliminate most of any free bromine or hydrobromic acid. The other reagents are as described previously ( 7 ) .
Procedure TOTAL BROXIDE.For most samples the procedure previously described ( 7 ) is satisfactory, but in certain cases of foods high in bromide and also in oil or protein, losses during ignition have since been noted. The same probably applies, in these cases, to the somewhat similar ashing method described by Dudley (e). A procedure modified from that of Brodie and Friedman ( I ) has been found preferable. A sample of 5 to 10 grams is treated in a 100-ml. nickel crucible with 40 ml. of alcoholic potassium hydroxide, allowed to stand for an hour, and evaporated to dryness on a steam bath. It is then dried for a short time at 110" C. and is covered with 10 grams of sodium hydroxide pellets. The crucible is kept for an hour or two on a hot plate until the bubbling or smoking diminishes, after which it is placed in a muffle at 600' C. Fusion should be carried out without excessive burning or foaming; if the charge becomes ignited, the crucible should be removed from the muffle until the flame is extinguished. It is then returned to the muffle and this rocess repeated until the volatile gases have been removed. {odium peroxide is added to the melt, a few milligrams at a time, to complete the oxidation of the remaining carbon or organic mat,ter. The peroxide must be added cautiously while the crucible is removed from the furnace; bromide is lost if the charge burns with a flare when too much peroxide is added. Complete combustion of the organic matter can be effected best by returning any organic matter that has raised above the sodium hydroxide t o the bottom of the crucible, where it mixes with the melt and is easily destroyed by addition of the peroxide. This is accomplished by carefully rotating the hot crucible to wash down the organic matter and adding 0.5 gram more of peroxide. If no burning or bubbling takes place, the oxidation is complete. A few carbon particles which may remain after the final addition of sodium peroxide do not affect the accuracy of the results. The crucible is rotated t o allow the melt t o solidify on the sides, and cooled, and the contents are dissolved in 75 ml. of water. Solution of the sodium compounds is hastened by placing the crucible on a hot plate for several minutes. The solution is transferred to a 400-ml. beaker and partially neutralized nrith about 50 ml. of 6 N hydrochloric acid. The solution is boiled t o destroy peroxides and to reduce the volume to 100 t o 125 ml.
,
'
' 0
TOTAL BROMIDt-
0
JNORG. B R O M I D f
a
ORGANIC BROMIDEI
I ~
'
n
I
i
I
IO0
I
125-
- 1
I I
l
I
I50
IL.5
HOURS AIRED AFT€/? FUM/GAT/ON
FIGTJRE 1. RETESTIONO F
BROMIDE BY WHITE
FLOVR
SEPARATION OF ORGAWIC BROMIDE.A sample of 5 to 10 grams in a 100-ml. beaker is treated with 15 ml. of methylene chloride and filtered immediately on a Gooch crucible with a dry asbestos ad, rinsing with three 5-ml. portions of solvent. I n filtering y! suction, the sample should not be allowed to become so cold from evaporation that moisture condenses on it. Most of the sample is transferred back t o the beaker without disturbing the asbestos pad and is allowed to stand for 5 minutes with 15 ml. of methylene chloride. If the sample is lumpy it should be ground with the solvent in a mortar at this stage, then filtered and rinsed as before. The solid is again returned to the beaker and treated with 15 ml. of methylene chloride, this time for a 15-minute period, followed by a third filtration and rinsing- in the same crucible. The filtrate. which is ordinarilv discarded, contains most of any methyl bromide or other solLble organic bromide, but not all, since the extraction may have been incomplete and since volatile compounds may have escaped. The presence of a soluble bromide compound may, if desired, be detected by catching the filtrates in alcoholic potassium hydroxide and determining the jnorganic bromide formed after evaporation to dryness and ashing. Whether or not the bromide so found is methyl bromide or a nonvolatile organic halide may be ascertained by making a duplicate set of extractions and evaporating the extracts to about half of their original volume before addition of alcoholic alkali. Methyl bromide is expelled during this process and nonvolatile bromide may be determined on the remainder after hydrolysis and ashing. In this way the authors have found that within experimental error no organic bromide is formed by reaction of methyl bromide with the products that have so far been tested. To remove any incompletely extracted volatile bromide compounds, the sample remaining after extraction is returned to the original beaker and treated with 15 ml. of methylene chloride. This is evaporated to dryness while being stirred to prevent bumping, but it should not be overheated. When the sample appears dry, the beaker is laid on its side in a warm place, such as on top of an oven, until all the odor of methylene chloride is gone. INORQANIC BROMIDE.Inorganic bromide may now be determined by the same procedure as described above for total
ANALYTICAL EDITION
January 15, 1942
bromide, practically all of the organic bromide having been removed. However, in order t o have further confirmation of the inorganic nature of the remaining bromide, it has been shown in all the authors' tests t o be water-soluble. The beaker and its contents are cooled, 30 ml. of water are added with thorough mixing, and then the mixture is filtered on the original crucible. In the case of some foods, such as flour, filtrations are very slow and preliminary separations by centrifuging are desirable. After each separation the solid is mixed with 30 ml. of water and allowed t o stand for 15 minutes, then filtered or centrifuged again, until four extractions have been made. The combined atrates are treated with 3 ml. of saturated sodium chloride solution and evaporated nearly t o dryness in a silica dish, then 30 ml. of 2.5 per cent alcoholic potassium hydroxide are added and evaporated and the bromide is determined after ashing as described in the previous paper ( 7 ) . UNACCOUNTED BROMIDE.The residue after water-extraction may be placed in a 100-ml. nickel crucible and analyzed for bromide in the manner described under Total Bromide. Generally only a very small fraction of the total bromide will be found in this residue and the amount will be less the more thorough the water washing. Therefore this bromide is usually considered t o be inorganic. Only in some experiments in which methyl bromide was adsorbed upon charcoal has the unaccounted bromide been thought t o be organic. No food adsorbs methyl bromide as tenaciously as does charcoal.
Application to Fumigated Products Since one cannot readily prepare standard mixtures for testing proposed separation methods, the procedure outlined above has been tested indirectly: 1. By considering the data obtained when the procedure is applied to a fumigated product at various intervals after fumigation. Since methyl bromide may escape upon airing or may decompose into inorganic bromide, whereas the latter cannot esca e, one should always find that the total bromide decreases towar$* limit while the inorganic bromide increases toward the same limit. If the inorganic bromide should appear t o decrease during aeration, the proposed method would be unsuccessful, since the only way this could occur would be through a change of methyl bromide into inorganic bromide during the separation. 2. By allowing a reactive product such as flour t o air for a week or more after fumigation and before analysis. In this time the methyl bromide will be completely volatilized or hydrolyzed. If the anal ical method then gives the same values for inorganic and total romides, one can be sure that it does not introduce negative errors in the determination of inorganic bromide. (An error of 0.10 ml. of 0.01 N thiosulfate, or about 0.0003 per cent bromine on a 5-gram sample, is considered allowable.) 3. By fumigating a nonreactive material such as pure dry sugar charcoal and analyzing for inorganic bromide. None of the latter should be found if no decomposition occurs during the separat,ion.
6"
TABLEI. LABORATORY FUMIQATION OF WHEAT FLOUR(WHITE) (1.98 pounds of CHaBr per 1000 au. feet for 12 hours a t 80-90° F.)
Aired after Fumigation Hour8 0.5 4 24 48 96
168 Control, not fumigated
Total Bromide
Inorganic Bromide
Organic Bromide
%
%
0.0224 0.0175 0.0159 0.0159 0.0154 0.0149
0.0135 0.0135 0.0151
0.0040 0.0008
0.0149 0.0148
0.0001
0.0010
0.0009
0.0001
....
%
0.0089
....
0.0005
TABLE11. LABORATORY FUMIQATION OF AMERICANPROCESS CHEESE (2.00 pounds of CHaBr per 1000 cu. feet for 12 hours a t 76-77' F.) Aired after Bromide in Outer O.25-Inch Layer Fumigation Total Inorganic Organio Hours 47, V " 07," ," 0.0114 0.5 0.0047 0.0067 4 0.0093 0.0041 0.0052 24 0.0075 0.0013 0.0062 4s 0.0077 0.0072 0.0005 0.0071 96 0.0076 .... I"
168
Control, not fumigated
0.0079
0.0080
....
0.0005
0.0004
0.0001
3
1
Q
?
3 3
*
.0100
1
I
0
25
TOTAL BROMIDE
0
INORG. BROMIDE
i
1
CRGAN/C8ROM/Df
I
FIGURE2.
0
,
50 75 100 125 150 -175 HOURS AIRFD A f T F R FUMIGATION
RETENTION OF BROMIDE BY AMERICANC R E ~ S E
Any procedure to be satisfactory must conform in all the above points, as well as give consistent results in the analysis of duplicate samples. However, there still might be a possibility that the method allows some reaction of methyl bromide with the food constituents. To minimize this chance, one can only compare various "satisfactory methods" on duplicate portions of a reactive food taken a t the same time shortly after fumigation. The method which under these conditions gives the lowest value for inorganic bromide is the one which allows the least conversion of methyl bromide into inorganic bromide. This has been the final criterion in the choice of the above procedure. To show the types of airing curves obtained, fumigations of some typical foods as well as of sugar charcoal have been carried out, and the products analyzed after definite intervale. I n Table I and Figure 1 are presented data obtained in the laboratory fumigation of white wheat flour. The retention of bromide is greater in fumigations performed on a laboratory scale than in those made commercially, as will be shown in a separate publication; hence these values are higher than would normally be obtained, but the general shape of the airing curves is the same. It is evident that most of the methyl bromide disappears within a day and that it is completely gone within a week. The decrease in methyl bromide content is due mainly to volatilization, only a small part of it decomposing into inorganic bromide after removal from the fumigation chamber. Table I1 and Figure 2 represent a laboratory fumigation of a sample of American process cheese. Half a pound of the unwrapped cheese in the form of a single cube was fumigated and the portions for analysis were taken from the outer layer, to a depth of 0.64 cm. (0.25 inch). It is of interest t o note that methyl bromide does not penetrate process cheese, no more bromide being found in the interior portions than in unfumigated controls. During aeration the cheese dries out, hence the bromide results reported in Table I1 have been corrected back to the as-received basis. The final value of 0.0080 per cent bromine in the outer layer corresponds to 0.0043 per cent based on the entire sample and would be even lower if a larger piece of cheese had been fumigated. Relatively more of the methyl bromide taken up is eventually decomposed to inorganic bromide in the case of cheese than with flour, but the total retention is lower. To test the method on a sample containing adsorbed methyl bromide relatively free from inorganic bromide, sugar charcoal was fumigated and analyzed a t intervals while airing. Because the charcoal adsorbs so much more fumigant than do foods, a greatly reduced dosage of methyl bromide
4
Vol. 14, No. 1
INDUSTRIAL AND E N G I N E E R I N G CHEMISTRY
TABLE 111. LABORATORY FUMIGATION OF SUGAR CHARCOAL (0.0267 pound of CHaBr per 1000 cu. feet for l i hours a t 73-75' F.) Aired after Total Inorganic Unaccounted Organic Fumigation Bromide Bromide Bromide Bromide Hours % % % % 0.5 0.0197 0.0002 0.0036 0.0159-0.0195 4 0,0188 .... 24 0.0188 0.0002 .... 96 0.0183 0.0001 0.0025 O.Olbil0.0182 192 0.0180 .... .... ....
....
326 ...
0.0174
.... .... ....
....
.... ....
720 1008
0.0144 0.0120
2400
0.0113
0.0001
0.0037
0.0002
0.0000
0.0000
Control not fumidated
....
.... ....
0.
O O i i l 0 . 0112 0.0002
was employed. The results, shown in Table 111, demonstrate that within experimental error no methyl bromide is falsely reported as inorganic bromide. Some of the bromide is found in the residues after separation and is therefore labeled as unaccounted; these amounts are not large when the difficulty of removing adsorbed methyl bromide from charcoal is realized. Thus 720 hours of airing removed only 27 per cent of the original methyl bromide, whereas 81 per cent was quickly removed by the analytical procedure outlined in this paper. At the end of the 720-hour period, a duplicate sample was evacuated for 26 hours a t 2 mm. of mercury, air being admitted ten times during the evacuation
to wash out liberated methyl bromide. This treatment lowered the total bromide percentage only from 0.0144 to
0.0122. Comparison of the data obtained on charcoal with those on flour and cheese emphasizes the fundamental differences in the ways in which bromide is retained by a food and by an efficient adsorbent. Methyl bromide taken u p by an adsorbent remains as such for a long time, while free methyl bromide present in a food can escape readily by volatilization and only the inorganic bromide produced by decomposition remains. The analytical method here proposed is capable of being applied to a wide variety of foods and has been found of value for indicating the time of airing which should be allowed after fumigation.
Literature Cited (1) Brodie, B. B., and Friedman, M .M.,J . B i d . Chem., 124, 511 (1938). (2) Dudley, H. C., INDENG.CHEM.,ANAL.ED., 11, 259 (1939). (3) Dudley, H. C., Miller, J. W.,Neal, P. A,, and Sayers, R. R., Pub. Health Rept., 55, 2251 (1940). (4) Laug, E. P., IND.EKG.CHEM.,33, 803 (1941). (5) McLaine, L . S., and Munro, H. A. U., 67th Ann. Rept. Entomol. SOC. Ontario, p. 15 (1936). (6) Martinek, 11. J., and Marti, W. C., IND.ENG.CHEM.,ANAL. ED., 3, 408 (1931). (7) Stenger, V. 9., Shrader, S. il., and Beshgetoor, A. W., Ibid., 11, 121 (1939).
Determination of Citric Acid In Pure Solutions and in Milk by the Pentabromoacetone Method EDGAR F. DEYSHER AND GEORGE E. HOLM Bureau of Dairy Industry, U. S. Department of Agriculture, Washington, D. C.
0F
THE methods that have been used for the quantitative determination of citric acid, the pentabromoacetone method has produced the most satisfactory results. It is based upon the fact that when citric acid is oxidized with potassium permanganate in the presence of bromine, under controlled conditions, the acid is converted quantitatively into pentabromoacetone, CBra.CO.CBrzH, the amount of which can be determined gravimetrically. Althou h the detection of citric acid through the formation of a bromine ferivative of a product of its oxidation with potassium permanganate was noted by Stahre (9) in 1895, Kunz (6) was the first to utilize the reaction to determine the citric acid content of milk, wines, and other fruit products. Hartmann and Hillig (1-4) Reichard (g), Lampitt and Rooke (?), and others also studied the reaction t o determine the conditions that would result in a quantitative conversion into and isolation of the pentabromoacetone. In general, the procedures adopted have been those used by Kunz, with particular attention being given to the temperatures used during the oxidation and bromination procedure, the formation of the precipitate, and the isolation, washing, and drying of the product. The results obtained were generally low in value, especially with solutions containing sugars, and corrections were usually made to account for the losses incurred. The variations in percentage recovery in most cases were also greater than should be expected. After a study of different methods and various modifications of the pentabromoacetone method, Lampitt and Rooke recommended the following procedure with solutions of citric acid and with milk serums. It is given in detail because i t forms the basis for the work herein described. To a solution containing citric acid or to 50 ml. of milk serum are added 10 ml. of sulfuric acid (1 to 1 by volume) and 5 ml. of
potassium bromide solution (37.5 per cent). Potassium per-
manganate solution (5 per cent) is added dropwise from a buret with constant shaking until a brown precipitate persists, 10 ml. usually being required for 0.10 gram of citric acid and 25 ml. for a milk serum. The mixture is allowed to stand at room temperature for 1 hour, further additions of permanganate being made if the brown precipitate disappears. Sufficient ferrous sulfate solution (20 per cent in 1 per cent sulfuric acid solution) is then added slowly till a pale yellow solution containing a white precipitate is obtained and the mixture is cooled in an ice chest overnight (16 hours). The precipitate is removed by filtration through a sinteredglass crucible (size 10 G 4), the reaction flask being washed out with the filtrate to remove the last traces of precipitate, and the washings passed through the crucible. The precipitate is then washed with portions of 10, 10, and 5 ml. of cold water. The crucible is dried to constant weight in a vacuum desiccator (about 16 hours). The precipitate is dissolved out of the crucible with industrial spirits followed by 20-, lo-, and 10-ml. portions of ether. The crucible is again dried in a vacuum desiccator and weighed, the loss in weight being taken as pentabromoacetone.
(
Citric acid (anhydrous) = 0.424 W
+ 0.005v> 100
where W represents the difference in weight of the crucible before and after treatment with industrial spirit and ether, and V the original volume of filtrate from reaction mixture, less the total volume of washings.
In Pure Solutions I n the application of the method as modified by Lampitt and Rooke to citric acid solutions, the authors obtained resuits which were not unlike those reported by these and other workers in the percentage recovery of citric acid (Table I). I n several experiments wherein the excess of reagents was not discharged prior to storage, i t was noted that where the excess of potassium permanganate was slight the manganese