INDUSTRIAL A N D ENGINEERING CHEMIt3TR.Y
506
Vol. 17, No. 5
Suitability of Various Solvents for Extracting Vanilla Beans' Part I11 By J. B. Wilson and J. W. Sale WATERA N D BEVERAGE LABORATORY, BUREAUOF CHEYISTRY, WASHINGTON, D. C
HE preparation and
T
analysis of a series of extracts made from M e x i c a n , B o u r b o n , and Tahiti vanilla beans and from Tonka beans by extraction with 65 per cent ethyl alcohol, 95 per cent ethyl alcohol, alkaline 65 per cent ethyl alcohol, 91 per cent isopropyl alcohol, U. S. P. ether, U. S. P. acetone, and 99.5 per cent carbon tetrachloride, have been described in Parts I2 and IL3
Experiments Using Carbon Tetrachloride for Preliminary Treatment
It is recognized in the hade that it is not practicable to extract vanilla beans with alcohol weaker than about 45 per cent, on account of the difficulty of percolation. As the previous experiments had shown that carbon tetrachloride removes the flavorless, oily material from vanilla beans and extracts very little of the real flavoring constituents, it seemed that if the beans were first treated with carbon tetrachloride to remove the oily matter, they might later be extracted SUCcessfully by means of alcohol of lower strength than 45 per cent. Accordingly, 500 grams of Bourbon vanilla beans were macerated for 48 hours with 1 liter of carbon tetrachloride, with frequent shaking. The solvent was separated by decantation, and evaporated under diminished pressure. When only traces of carbon tetrachloride remained, the residue was transferred to a weighed flask and left near the steam bath. After standing some time it was weighed, and it was found that 31.5 grams, corresponding to 6.3 per cent, of oily matter had been extracted. As a slight odor of carbon tetrachloride still remained, the flask was left near the steam bath overnight. A direct determination of the vanillin in the oily matter was not made, since some of it may have been lost. From the amount of vanillin in the original beans and that found in the extract richest in vanillin, it is estimated that from 0.1 to 0.2 gram of vanillin (corresponding to 0.02 to 0.04 per cent in the extract) was removed from the beans with oily matter. Vanillin can be readily recovered from the oily matter by sublimation and returned to the extracts. After extraction with carbon tetrachloride the beans were dried in air until the odor of the solvent was no longer perceptible. Such quantities (41.2 grams) of these dried beans as were equivalent in solid matter to 50 grams of the Bourbon beans used in the experiments described in Part I were then placed in Erlenmeyer flasks and extracted with three successive portions of solvent and made up to 500 cc. each. The solvents used were mixtures of alcohol, glycerol, and water Received April 1, 1925. Presented before the Divison of Agricultural and Food Chemistry a t the 6 i t h Meeting of the American Chemical Society, Washington, D. C., April 21 to 26, 1924. 2 THISJOURNAL, 15, 782 (1923). * I b i d . , 16, 301 (1924). 1
in varying proportions. No trouble was experienced in separating the solvent from the marc, even when the extract contained only 10 per cent of alcohol. The partial composition of these extracts is set forth in Table I. The data in Table I, when compared with the data in Table I of Part 11,show that preliminary extraction of the beans with carbon tetrachlo* ride did not materially affect either the vanillin content or the lend number. Howeber, the apparent high quality, as judged by the data, of some of the extracts referred to in Table I was not confirmed by the organoleptic tests described below, so that, although it is practicable to employ carbon tetrachloride for the purpose of reducing the alcohol content of the menstruum, the resulting extracts are of inferior quality.
Parts I and I1 of this series dealt with the composition of vanilla and tonka extracts made from oleoresins that had been prepared by the use of various solvents. This paper deals chiefly with organoleptic tests conducted on the vanilla extracts. The most suitable solvent for the vanilla extracts is neutral 65 per cent ethyl alcohol. The other solvents, in the order of their suitability, placing the most suitable first, are alkaline 65 per cent ethyl alcohol, 95 per cent ethyl alcohol, 91 per cent isopropyl alcohol, U. S. P. acetone, U. S. P. ether, and carbon tetrachloride. No advantage is gained by the preliminary removal of oily matter from the beans.
T a b l e 1-Composition of B o u r b o n Vanilla E x t r a c t s (Beans previously extracted with CCh) Ethyl alcohol PER BY found VOLUME per cent VANILLIN LEADNUMBER Ethyl Glycerol by volFolinWichalcohol U.S.P. Water ume A.o.A.c.~ Denisb Winton mannc 0.59 0.78 0.23 0.21 65 35 60.9 0.55 0.86 0.18 0.20 30 70 28.6 0.52 1.05 0.18 0.17 20 80 20.4 0.75 0.18 0.58 0.18 60 18.2 20 20 0.79 0.18 0.55 0.18 70 18.8 20 10 0.55 0.89 0.16 Lost 10 10 80 10.3 a Assoc. Official Agr. Chem., Methods, Revised t o November 1, 1919. b THISJOURXAL, 4, 670 (1912). c I b t d . , 13, 414 (1921). T a b l e 11-Organoleptic Tests o n Vanilla E x t r a c t s Aroma Flavor SOLVENTS Mexican Ethyl alcohol 95%" Isopropyl alcihol, 91%4 Ether U S. P." Acetohe 'U. S. P." Carbon ietrachloride, 99.5%' Ethyl alcohol, 65% Alkaline 65% ethyl alcohol Bourbon 75 90 Ethyl alcohol, 95%" 75 70 Isopropyl alcohol, 91%" 50 50 Ether U. S P." 50 25 Off flavor 80 Acetohe U ' S . P.a Carbon letiachloride, 99.5%" 100 100 Ethyl aclohol, 65% 75 110 Alkaline 65T0 ethyl alcohol Tnhili
Ethyl alcohol 95%" Isopropyl alcAho1, 91%" Ether U. S. P.a Acetohe U. S. P.' Carbon ietrachloride, 99.57OG Ethyl alcohol 65% Alkaline 65%,'ethvl ._ . alcohol Bourbon (double extraction) b 2nd solvent 65% ethyl alcohol 2nd solvent 30Y0 ethyl alcohol 2nd solvent 20% ethyl alcohol 2nd solvent 20% ethyl alcohol and 20% glycerol 2nd solvent 2 0 7 ethyl alcohol and 10% glycerol 2nd solvent ethyl alcohol and 10% glycerol
log
75 75
50 50
60
40
30 Off flavor 40
100 75
80 60
75 50 25 50
70 60
50 40
50
50
50
33
" Beans extracted with the solvents as listed, solvents evaporated off and extractive matter dissolved in 65 per cent ethyl alcohol. b First solvent was carbon tetrachloride which eliminated oily matter.
. May, 1925
I N D U S T R I A L A N D ENGIXEERING CHEMISTRY
Results of Organoleptic Tests on All Vanilla Extracts Described in Parts I, 11, and I11
Two sets of experiments were conducted to test the quality of the extracts. At first, 1 cc. of extract was added to 50-cc. portions of milk sweetened with sugar, but these tests were not so delicate as those carried out by adding 3 cc. of extract to 50-cc. portions of distilled water. This latter procedure was therefore adopted, the results of the organoleptic tests being given in Table 11. The extract made from Mexican vanilla beans with a menstruum of 65 per cent alcohol was employed as a standard of comparison and was given a rating of 100. The numerical ratings in Table I1 indicate the relative flavoring quality of other extracts as compared with the standard. Three people tasted the extracts, the averages only being reported. Discussion of Data in Table I1
These data show that, from the standpoint of aroma and flavor, the most desirable solvent of those tested is neutral 65 per cent alcohol. It is of interest to note that the Research Committee of the Flavoring Extract Manufacturers' Association found 47 to 50 per cent of alcohol (corrected by the addition of alcohol for moisture in the bean) to be the best
507
menstruum for extracting vanilla b e a m 4 The addition of alkali increased the strength, as judged by the flavoring, of the Mexican and Bourbon beans, but made the aroma less pleasant. Also, it affected both the aroma and flavor of the Tahiti beans unfavorably. The extracts made from the oleoresins prepared with carbon tetrachloride possessed an objectionable foreign flavor, which it is believed would render them unmarketable. From the standpoint of flavor the suitability of the other solvents is as follows, the most suitable being given first: 95 per cent ethyl alcohol, 91 per cent isopropyl alcohol, U. S. P. acetone, and U. S. P. ether. For the manufacture of vanilla oleoresins 95 per cent ethyl alcohol would probably be preferable to neutral 65 per cent ethyl alcohol because it could be more easily removed. Xeutral 65 per cent ethyl alcohol is, of course, more desirable than 95 per cent ethyl alcohol when the product to be manufactured is an extract. None of the extracts made from the beans after they had been subjected to a preliminary extraction with carbon tetrachloride equaled the standard in aroma and flavor. Therefore] the preliminary removal of the oily matter from the beans is not recommended. The extracts made with a low-alcohol content were decidedly inferior to the standard. 4
Tea Cofee T r a d e J . , 47, 121 (1924).
Crystallizing Point of Paraffin Wax' By A. P. Bjerregaard EMPIRE REFINERIES,INC., OKMULGEE, OKLA.
OMMERCIAL paraffin wax is a mixture of a large number of homologous hydrocarbons whose melting and boiling points are not widely different. When melted oil-free wax is cooled, a temperature is reached at which crystals separate and over a short range below which the process of crystallization continues until the wax is solid. Obviously, the composition of the wax will determine the temperature a t which crystallization starts and the extent of the solidification range. Two methods have been used for determining the so-called melting point of paraffin wax. The older or American method consists in reading the temperature of the melted wax when the first crystals separate. The wax under the test is heated to about 5' C. above the expected melting point. A hemispherical metal cup 9.5 cm. (33/4inches) in diameter is filled three-fourths full with melted wax. A special waxtest thermometer with a globular bulb and graduated in half degrees Centigrade or quarter degrees Fahrenheit is used to take the temperature. The thermometer bulb is two-thirds immersed in the melted wax. The 'surface of the liquid is closely watched and when the crystals of solid paraffin first appear on the surface and extend to the bulb of the thermometer, the thermometer is read and this temperature recorded as the melting point of the wax. It is a common practice to read the thermometer when three crystals of wax have formed. The English method, now adopted as a standard by the American Society for Testing bIaterials, is based on determining the cooling curve, and, by observation of the slope of this curve, finding the mean temperature of the wax during the solidification of the main bulk of the wax. This method has been fully described under the designation D87-22.2
C
1 Received February 20, 1925. Presented before the Division of Petroleum Chemistry at the 69th Meeting of the American Chemioal Society, Baltimore, Md., April 6 to 10, 1925. 2 A. S. T. M. Standards, 1924, p. 882.
These two methods not only give different figures, but the relationship between the melting points varies according to the composition of the wax under investigation. A sample of wax, the melting points of whose components lie close together, will naturally show a smaller difference between the crystallizing temperature of the first portions and the crystallizing temperature of the main mass. On the other hand, a wax composed of substances covering a wide range of melting points will show a larger difference between the crystallizing point of the first portion and the crystallizing point of the main mass. The accompanying tables present the American and English melting points of paraffin waxes made from two varieties of crude oil. All these waxes came from two independent refineries, one using Kansas crude and the other Oklahoma crude. Waxes f r o m a High-Sulfur Crude f r o m Butler C o u n t v_. .K a n s a s (21 determinations) -American-A. s.T. >-.: c. F. c. F. 130.9 133.5 54.9 56.4 127.0 53.6 128.5 52.8 129.1 130.8 53.9 54.9
Table I-Paraffin
Max. Min. Av.
P e t r o l e u m Oil --Difference-
=
c.
1.5 0.4 0.9
' F.
2.6 0.7 1.6
Table 11-Paraffin Waxes f r o m L o w - S u l f u r Crude P e t r o l e u m Oil f r o m Okfuskee C o u n t y , O k l a h o m a (24 determinations) -American-A. S. T. M.-Difference--.
c.
Max. Min. Av.
54.7 37.8 50.6
F.
130.4 100.0 123.0
c.
' F.
53.9 36.4 49.9
129.1 97.5 121.7
c.
2.2 0.1 0.8
' F. 3.9 0.2 1.4
There is no consistent difference between the melting points as determined by the two methods. The variations are due to the differences in the composition of the waxes with regard to the relative amounts of higher melting homologs and intermediate melting homologs, and with regard to the spread between the melting points of the higher melting point waxes and of the medium melting point waxes actually present.
