ANALYTICAL EDITIOiY
410
Vol. 2, No. 4
Rapid Determination of Total Fat’,’ Clarence P. Harris 522 FIFTHAvE., NEW YORK,S. T.
T
H E well known, if not classical, method for the determination of total fat described by Lewkowitsch (13) consists of mixing the material with four to six times its weight of either sand or finely powdered gypsum, in the case of hygroscopic substances, drying the treated sample a t 100’ C., and extracting with a suitable solvent in a Soxhlet apparatus. Petroleum ether he considers advisable as solvent “for strictly accurate results.” as this liquid does not dissolve the oxidized acids of sulfur, olive oil, dCgras, etc., nor theobromine occurring in cacao beans. Ethyl ether readily extracts this alkaloid (3, 8),but is likely to extract animal fats incompletely owing to the presence of lecithin (11). Many variations and improvements of this general method have been developed, all of them based upon the principle of iso!ating and weighing the actual fat present. This principle is quite sound with materials which have been so treated that the oil or fat is “free”-that is, not occluded by vegetable or animal tissues. Finely ground cacao nibs (chocolate liquor) or cottonseed press cake (cottonseed meal) are completely extracted without difficulty, but substances like chocolate expeller cake and either raw cottonseed or cottonseed meats are more troublesome. Cottonseed meats, for example, are extracted twice, being removed from the extraction thimble and ground between times. Furthermore, almost all of these complete extraction methods are F!OW, the average elapsed time necessary being about 5 to 6 hours. But their greatest disadvantage lies in the fact that a laboratory and either a chemist or a specially trained technician are required for their operation. The writer believes that analyses of Gil- and fat-bearing substances would be made much more frequently if methods were available which, while accurate, would be much less difficult in operation, and above all if sufficiently rapid so that the result could be made known to the plant operator in time for him to make use of the information. Methods which are rapid, accurate, and sufficiently simple so that non-technical persons can operate them have been devised. The time required varies from 10 minutes with cottonseed meal to 35 minutes d h finely ground chocolate liquor or ground beef scrap (“cracklings”). Their accuracy is a t least as great as that claimed for the official methods and their simplicity is such that a laboratory boy can become adept after a few hours’ practice. Importance of Time Factor Fats and oils of vegetable origin do not occur naturally in a pure state and analysis is necessary to assay the value of the raw material. As it is usually possible to wait a t least seyeral hours without inconvenience for analyses of products not yet in process, the time required for such a test is ordinarily not of paramount importance. But during the processes for isolation of the desired product-as, for example, by pressure methods-prompt knowledge of the fat content of the material, including the expressed residues, is necessary if efficient operation is to be maintained. The efficiency of a press must be known quickly if avoidable losses are to be stopped. I n certain cases, where it is desired to press to a definite fat content of non-fatty residue, as in the case of breakfast cocoa Received September 8, 1930 T h e methods described in this paper were devised by t h e writer while associated with t h e Schwarz Laboratories, Inc , New York, h- Y 1 2
with its specified standard of 22 per cent cocoa butter, it is essential that a certain hydraulic pressure result in a press cake of slightly above this percentage. Too high a percentage of cocoa butter in the breakfast cocoa is not only needless waste to the manufacturer but it is of no advantage to the consumer. as the digestibility of cocoa decreases slightly with increase of fat content. P e t the author has known of cocoa manufacturers obtaining cocoa press cakes with a butter content of 25 to 30 per cent, supposing that the standard of 23 per cent was being maintained. What is true of cocoa i j also true of other products and illustrates the need of a method of analysis sufficiently rapid to be of use to the plant operators. Older Methods for Determination of Total Fat The desirability for speed in this operation has long been recognized, and occasional efforts have been made with this object in mind. hIost of them are variations of the standard method mentioned above. Some are very slow, a few are quite rapid, but all are subject to the objection that a skilled chemist is required t o obtain significant results. With few exceptions the idea has always been to isolate the actual fat present and to weigh it. The most exact results are said to be given by Heller’s (6) method, which consists of a 24-hour extraction in a Soxhlet apparatus. Kreutz’s (10) variation involves a digestion with chloral hydrate previous to extraction with ether. Kooper (9) modified the well-know1 method of Gerber for milk, a centrifugal method similar t o the Babcock, and by using sodium salicylate solution to dissolve the protein and butanol as a clarifying agent he applied this rapid method to cocoa. Butyrometers somewhat similar to the Babcock apparatus are used. This method works well with dairy products such as milk, cream, cheese, etc., but, possibly because the proteins other than casein are less easily soluble in the reagent? used, the results on chocolate products do not appear to be entirely reliable. Kelman (16) suspends the sample in mater and then extracts with wet ether. The ether layer clarifies on standing from 6 to 24 hours, an aliquot of which is taken to complete the analysis. Hanus ( 2 ) uses a dilute ammoniacal and alcoholic suspension before the ether extraction, a mixture of petroleum and ethyl ethers being employed. Gephardt3 speeded up these extraction methods by finding that good results are obtained by the use of only 1 gram of sample and by a very short extraction and washing of the material on a Gooch crucible. The method is rapid and probably accurate, but involves considerable skill. A method devised by Hughes ( 7 ) is popular in some circles because of its rapidity. It consists of successive extraction3 of the fat-bearing substance with ether, centrifuging the ether layer after each extraction. Low results are obtained if the extraction has not been complete; on the other hand, high results are likely if the centrifuging of the finely divided particles is less than perfect, d variation of the extraction method consists of the S. B. P. procedure, using trichloroethylene. h larger sample (8 to 10 grams) is taken and an aliquot of the extract is evaporated and weighed in the usual manner. The method is fairly rapid, but the technic is involved. 3 This method is in use by certain manufacturers, b u t so f a r as the writer is a n a r e it is not described in t h e literature.
October 15, 1930
INDUSTRISL ALVDEJiGISEERIXG CHEMISTRY
Lepper and Waterman (12) have recently modified the official method described above and use a Knorr extraction apparatus. The sample is introduced into the Knorr tube, stirred with petroleum ether, and washed ten times. The method has been made official and appears to be more rapid than the earlier official procedures. Heiduschka and X u t h (4) have devised glass apparatus for extraction and evaporation methods which shorten the number of manipulations and transfers of solution. The Gooch crucible is designed to serve also as a weighing flask for the sample, and the distilling flask is also fitted with a cover and is sufficiently small to act as a weighing flask for the extracted fat. Richter (14) and Herty (6) broke with tradition, so far as total fat estimation is concerned, and devised methods which do not depend upon weighing the fat actually present in the sample. They recognized that solution of a second substance in a liquid alters the physical properties of the liquid, usually in direct proportion to the amount of substance dissolved. Therefore, if the change in that property can be sufficiently accurately measured, the amount of added substance can be determined. Richter‘s method depends upon the change in refractive index, Herty’s upon the change of specific gravity. Richter used as dissolving liquid a mixture of ether, alcohol, and trisodium phosphate. The substance the fat content of which was to be determined was extracted with this mixture, the clear solution, obtained after warming, adjusted exactly to a definite temperature (17.5” C.), and examined under a Zeiss refractometer. From the refractive indices of solvent, solution, and the pure fat, the amount of fat present in solution, and consequently the amount present in the original sample, may be calculated. One difficulty with this method lay in the very great volatility of the ether and alcohol of the solvent. Evaporation of these substances during the operation will render this method inaccurate. To overcome this objection, Wesson (16) proposed the use of Halowax, a trade name for chlorinated naphthalenes. As certain of these products (impure monochloro compounds) are liquids, they can be used instead of the original mixture of Richter. Halowax is non-volatile and otherwise suitable for this method, but the commercial product varies in composition and properties. The refractometer method has been further elaborated by Coleman and Fellows (1) to include a number of fat-bearing materials. Before using this method the Halowax must be standardized-i. e., a table showing the variation of refractive index with increasing percentages of the oil or fat must be developed before the particular batch of solvent may be used. According to Coleman and Fellows, “it is necessary to prepare for each new lot of the solvent a standard conversion table for each kind of oil-bearing material on which the test is to be made.” Furthermore, for certain materials, such as chocolate liquor, an involved process of preparation is necessary. This substance must be grated, melted, chilled, and grated again before the analysis can be run. The temperature of the solution must be carefully adjusted before the refractometer reading is taken. However, when all these precautions are taken, a skilled operator can obtain results in very short intervals. Herty’s (6) method consists of solution of the fat by means of carbon tetrachloride and analysis of the solution by the use of a Westphal balance. Owing to the volatility of this solvent, its evaporation during the operation will give high results. Here again the specific gravity of the solvent will T ary over an appreciable range, and in contact with various materials will hydrolyze and form various degradation products which very much alter the gravity. Consequently, the solvent requires frequent checking and standardization, the operation requires considerable skill to avoid a change in specific gravity during the interval, and the Westphal balance is not without objection.
411
Experimental Method
Quantitative solution of a substance is considerably more rapid in many circumstances than quantitative isolation. The latter involves the washing of a solution from solids mixed therewith. Theoretically this is a n endless process approaching infinity as a limit. Practically a definite number of washings will prove sufficient for most purposes, but one source of error is insufficient washing. Another and more important one is the removal of the solvent. This is supposed to be done until constant weight is reached, but materials which absorb oxygen from the air, as is the case with all vegetable oils and fats in varying degree, often exhibit difficulty, even to the point of increasing in weight. The time of drying is therefore specified in the methods which isolate and weigh the extracted fat, but the writer has frequently seen an analytical extract of a liquid vegetable 011 weighed as a solidified, apparently oxidized solid. The objection to most of the older methods, then, lies not only in their slowness and difficult technic, but in their inherent possibility of errors. Given a substance containing a fat all of which is on the surface and not occluded in tissue cells, quantitative solution with a good solvent is extremely rapid. For example, without the application of heat all of the cocoa butter present in a sample of cocoa will enter solution in a chlorinated hydrocarbon in 3 minutes or less. It should be remembered in this connection that solution by extraction with petroleum ether in a Soxhlet apparatus is also effected cold, the ether dripping from the cold condenser. With a solid fat the temperature of the ether is below the melting point of the fat and solution is effected by erosion from the surface. When the same sample is mixed with a large excess of solvent, solution is effected simultaneously from every particle of the substance. The fact that solution of a substance alters the physical properties of the solrent is well known. Herty’s ( 6 ) method is based upon this principle. An aliquot of the solution is sufficient to examine the change in physical property effected, and this change is a measure of the amount of solute present in the original sample. It is necessary to measure the degree of this change with sufficient accuracy so that the Figure 1-Hydromsubstance, in this case oil or fat, may be ep:ebddftt8 estimated as closely as the best of the methods now existing. Specific gravity was selected as the physical property, because it can be measured by hydrometers which are simple to use and which can be manufactured to be accurate to 0.0001. o-Dichlorobenzene was selected as the solvent. It has the advantages of being non-inflammable, very slightly volatile a t ordinary temperatures, and not too expensive. Further, the difference in gravity between it and a fat is so great that small additions of fat cause a marked change in the gravity of the solution, thus permitting accurate results. Solutions of fat in this solvent may be permitted to stand exposed t o the atmosphere for 24 hours with no observable change in specific gravity. As commercial o-dichlorobenzene is not uniform in properties, it is especially adjusted by the manufacturers of the analytical equipment so that uniform results may be obtained. The analytical balance was discarded, as men accustomed
A N A L Y T I C A L EDITIOM
412
t o handling tons of product in the plant or laboratory boys unfamiliar with careful technic could not be expected to use so delicate an instrument. Accordingly a torsion balance accurate to 0.1 gram was employed, using a sample of 100 grams in order to maintain the accuracy of weighing to 0.1 per cent. The methods as developed consist of the soluiion of the fat with a definite weight of solvent, filtration of an aliquot from the insoluble residue, and determination of the change of specific gravity by means of special hydrometers accurate t o 0,0001 specific gravity, and reading directly the percentage of fat present. Upon this principle are based methods for the determination of total fat in chocolate products, such as liquor, sweet and milk coatings, cocoa powder, shells, expeller cake, and nibs. Methods have also been dereloped for the determination of the oil in cottonseed meal and the tallow in ground meat scrap. Other developments are in process.
similar. These substances must be ground to 60-80 mesh before being analyzed. Methods for Cottonseed Meal, Meat Scrap, Etc. The method developed for cottonseed meal is simpler. The sample is ground so that 80 per cent passes a BO-mesh sieve. All of the solvent is then weighed into the beaker at once. It is unnecessary to wet the filter paper and this procedure is omitted. The operation subsequent t o the grinding requires 10 minutes. The method for ground meat scrap (cracklings) is similar, but in this case it is necessary to allow the mixture of sample and solvent to stand for 15 minutes, decanting onto the funnel, in order to secure a sufficiently rapid filtration. Results
A comparison of the results obtained by these and other methods is given in the accompanying table.
Procedure for Chocolate Products
A special aluminum beaker and
M,-
stirring rod is placed on the right-hand pan of a torsion balance. A brass weight equal to the weight of the beaker and rod plus 100 grams is placed on the left pan and the sample introduced into the beaker until the scale is in equilibrium. A dash pot is used on the newest balances which damps the vibration and brings the pointer to rest very quickly. -After 100 grams Figure %-Filtering the of sample, either cocoa powder or Solution molten chocolate liquor or coating, have been thus weighed out, about 2 inches (5 cm.) of the beaker are filled with the solvent and the mixture is stirred until it is homogeneous. The beaker is now rep!aced on the right-hand balance pan, the weight replaced by a larger one (weight of beaker and rod plus 100 plus 584.11 grams), and the solvent run in from a glass siphon until the balance is again in equilibrium. The contents of the beaker are now thoroughly stirred and allowed to stand 2 to 3 minutes with occasional stirring. This procedure dissolves all of the cocoa butter present. A Buchner funnel is now prepared with a filter paper and wet with 3 cc. of solvent, measured in a small calibrated vial, Two tablespoonfuls (6 grams) of shredded asbestos are stirred into the beaker t o assist filtration, the filter flask shown in Figure 1 connected to a source of suction, and the mixture filtered. The time of filtration will vary with the fineness of division of the cacao particles, but should not exceed 8 to 10 minutes. The funnel is removed from the filtering cylinder, and one of the special hydrometers (Figure 2) is inserted in the solution of cocoa butter in the solvent. The percentage of cocoa butter present in the sample is read directly from this hydrometer. These hydrometers are standardized at 20" C., but the temperaturecorrection scale makes accurate readings possible at any point between 17' and 23" C. Only a very approximate temperature adjustment is therefore necessary. If the temperature is too low, the mercury will not appear on the lower scale, and the solution in the cylinder can easily be warmed by the heat of the hand to the proper range. If the temperature is too high, the solution is cooled by a special jar fitted with water inlet and outlet. I n summer, if the room temperature is high, it is advisable to stand the cylinder in the cooling jar during the filtration. The technic used for cacao nibs, shell, and expeller cake is
VOl. 2, No. 4
SAMPLE
T o t a l Fat D e t e r m i n a t i o n s by X e w and O l d M e t h o d s N E W .4.0. c.s. h f F R . ' S NEW A. 0. A. C. MFR.'S M E T H O D M E T H O D ANALYSISS A M P L E METHODh f E T H O D ANALYSIS
%
70
%
%
C O T T O S S E E D MEAL
51 C-14 C-15
5.75 5.95 6.1
.,. . .
5.76 6.25 6.17
...
CHOCOLATE LIQUOR
9 10
A. 0. A. c. 53.9 53.1 53.84 52.45 52.25 53.5 53.3 53.3 ... 52.6 52.64 53.2 ... 53:22 52.7 52.72 54.2 5i:23 ,.. 55.9 55.3 , , . 55.1 , . . 55.2 ... 53.6 53.6
11 12 13 14 15 16 17
26.4 22.7 22.9 9.9 13.5 22.4 20.5
1 2 3 4 5 6 7
8
COCOA P O W D E R
26.6 22.9 22.87
... ... ... ...
%
%
S W E E T CHOCOLATE COATING
...
22.9 23.88 9 ' 73 13.5 22.04 20.2
MI1>K CHOCOLATE COATINGa
26 27 28 29 30 31 32 33 34
35.3 34.5 33.5 35.3 34.8 41 6 33 6 33.3 32.9
35.5 34.1 33.9 35.53 34 93 41.69 33.24 32.93
34: il
... ...
35.6
33:& 33.19
...
E X P E L L E R CAKEb
35 36 37 38
11.5 11 5 9 5 11.4
11.43 11.33 9.69 11.26
CACAO S H E L L S
. . .
.. ,
Accra
5.6
5.66
"l$Za Superior Bahia Lagos Sanchez
5.7
5.51
,,,
5.8 10.2 4.3
5.93 10.20 4.10
, , ,
. .. ...
M E A T SCRAP
4 10.8 10.75 ... 2 10.4 10.41 ... 1 10.5 10.78 a Owing t o t h e difference in gravity between milk f a t and cocoa butter, a correction must be applied t o t h e results on this material, usually 0.3 per cent, but sometimes higher if a large amount of milk solids has been incorporated. This correction which is added t o the result is determined b y analysis of t h e product and remains constant for t h a t particular grade of coating. b T h e expeller cake must be ground t o 60 mesh before being analyzed. . . I
Acknowledgment The painstaking and capable assistance of Xanuel Horwitz in working out many of the details is gratefully acknowledged. Literature Cited (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Coleman and Fellows, U. S . Dept. Agr., Bull. 71 (1928). Hanus, Z. X a h r . Genussm., 2, 738 (1906). Hartley, J. Physiol., 36. 17 (1907). Heiduschka a n d Muth, Chem.-Ztg., 62, No. 90 (1928). Heller, Agoth. Ztg., 31, 330 (1916). Herty, Stem, a n d Orr, J . IND.ENG.CHEM.,1, 76 (1909). Hughes, Chem. News. 119, 1041 (1919). Koch, C h e m - Z t g . , 36, 1040 (1911). Kooper, Z . A'ahr. Genussm., 30, 461 (1915). Kreutz, Ibid., 16, 680 (1908); 16, 584 (1908). Kumagawa and Suto, Biochem Z., 8, 212 (1908).
October 15, 1930
INDUSTRIAL AAVD ENGINEERISG CHEAZIISTRY
(12) Lepper and Waterman, J . Assocn. OPicial A E Y . Chem., 9, 46, 461 (1926). (13) Lewkowitsch, “Chemical Technology and Oils, Fats, and Waxes,” 6th edition, Vol. I . p. 288.
413
(14) Richter, 2 . S n h r . Genussm., 24, 312 (1912). (15) Welman, Z . ofentl. Chern., 6, 304 (1900). (16) Wesson, Cotton Oil Press, 4 , KO.3, 70 (1920)
Estimation of Aldose Sugars b y Titrating with Standard Iodine and alkali'^' Modified Method G . M . Kline and S. F. Acree BUREAU OF STANDARDS, WASHIXGTON, D. C.
T
HE determination of
aldose sugars by titration with iodine-alkali reagents is based on the following reaction:
+ ++
RCHO 12 3NaOH -+ RCOONa 2hTaI HzO
+
Iodine also reacts with the alkali to form sodium iodate 31
+ 6KaOH +NaI03 + 5NaI
+ 3Hz0
In a recent article ( I ) experiments were described in which aldoses were titrated, first with alkali and then with iodine, to give the aldonic acid. The reagents were added comparatively rapidly in single portions and in considerable excess, and an accuracy of about 2 per cent was obtained. Further studies are now reported on various factors affecting the titration, such as the mode of addition of the reagents, iodate formation, hydrogen-ion concentration, oxidation of the aldonic acid and of ketones and non-reducing sugars, and the time of oxidation. By adding the standard iodine first and then the alkali, each in small fractions of the total volumes required, the reaction is completed rapidly and a precision of 0.2 to 0.3 per cent is obtained. Iodate formation is an index of the completion of the oxidation of the aldoses.
sugar calculated on the basis of the alkali c o n s u m p t i o n agrees with the iodine value. ( C o m p a r e t h e t h i r d and fourth columns under Xylose in Table I.) By the use of an excess of pure dextrose, therefore, this reaction can be used to standardize the iodine s o l u t i o n against the alkali solution or vice versa and hence furnishes a convenient method for a rapid check on the strength of the reagents. T h e e r r o r introduced by titrating the free iodine in the alkaline solution before acidifying to liberate iodine from iodate has been studied. This error is the result of the oxidation of the thiosulfate reagent to sodium sulfate by iodine in alkaline solution, viz.:
which can take no part in the sugar-oxidation reaction. The customary procedure has involved the use of 100 to 200 Der cent excess of iodine and alkali, resulting in general in overoxidation and therefore high values. hIodification of the procedure is recommended, based on the observation that alkali added slowly t o a solution of a n aldose sugar containing a small amount of iodine will tend, not t o form sodium iodate, but to react with the sugar. By addition of the iodine and alkali successively in small portions the concentration of sugar relative to the iodine-alkali or sodium hypoiodite is kept a t a level favoring the sugar-oxidation reaction. The rapid formation of iodate, therefore, indicates the complete oxidation of the aldose to the monobasic acid. By this procedure only 2 ml. excess of 0.1 iV iodine are required. (Compare the first, third, and sixth columns under Xylose in Table I.) Previous methods have specified empirical time periods of various extent up to 30 minutes for the completion of the oxidation. However, using the proposed modification, the reaction may be taken as complete 2 minutes after the last of the reagent is added. Iodate formation takes place more rapidly than the very slow oxidation of dilute solutions of ketoses and non-reducing sugars. (Compare the third and sixth columns under Levulose and Sucrose in Table I.) These non-aldehydic sugars, therefore, do not interfere with the determination of aldoses. The slow oxidation of the non-aldehydic sugars consumes iodine and alkali in approximately the 4 : 5 ratio required for the oxidation of a primary alcoholic group to a carboxyl. (See experiments in Table I in which concentrated solutions of sugars were used.) It has further been demonstrated that the percentage of
K i t h xylose an apparent overoxidation of about 10 per cent is obtained with 20 ml. excess of iodine and 30 ml. excess of alkali, whereas if the solution is acidified before titrating with thiosulfatethetrueoveroxidationis found tobe about 2per cent. Levulose and sucrose give about 7.5 and 6 per cent apparent oxidation, respectively, and 3.5 and 2 per cent true oxidation calculated on the basis of four equivalents of iodine required for the oxidation of one equivalent of sugar as found experimentally. If the percentage oxidation is calculated on the basis of two equivalents of iodine required for oxidation, as found experimentally for aldoses, the above values are d ~ u b l e d . ~The free iodine and iodine combined as hypoiodite may be determined with about 1 per cent error by saturating the alkaline solution with carbon dioxide and titrating with thiosulfate; the iodine combined as iodate is then obtained by acidifying with hydrochloric acid and completing the titration with thiosulfate. The use of both indicators and the glass electrode has shown that a t about pH 6.4 the reaction is very slow and requires 18 hours for only 34 per cent completion. Khen the pH is kept a t about 9 to 10, the end point is reached within about 2 minutes after the last portions of the reagents are added. The regulation of the pH is therefore of prime importance in both the analytical method and in the practical application of these reactions in the production of sugar acids now under way.
1 Received September 6, 1930. Presented before t h e Division of Sugar Chemistry a t t h e 80th Meeting of t h e rlmerican Chemical Society, Cincinnati, Ohio, September 8 t o 12, 1930. 2 Publication approved by t h e Director of the Bureau of Standards of the U S Department of Commerce
3 T h e preliminary oxidation data reported by Slater and Acree ( I ) are too high owing t o these t w o factors-namely, t h e titration of free iodine in alkaline solution, and the calculation of the per cent oxidation of sucrose and levulose on t h e basis of two equivalents of iodine instead of four required per mol of sugar
NazS203
+ 412 + 10NaOH
-
2Ka2SO4
+ 8NaI + 5H20
