Chromatographic Estimation of Vitamin
A
in Mixed
Feeds
MAXWELL L. COOLEY, JAMES B. CHRISTIANSEN, AND CARL H. SCHROEDER General Mills, Inc., Larro Research Farm, Detroit, M i c h .
A chromatographic procedure, using high purity sodium carbonate adsorbent, facilitates the simultaneous separation of vitamin A and carotene from most common noncarotene pigments and other substances occurring in livestock feeds which interfere with the antimony trichloride determination of vitamin A. Certain limitations are indicated. Neither vitamin A nor carotene is retained appreciably b y the adsorbent. When the sample contains an unknown source of vitamin A, saponification may b e necessary, since the reduced activity of cod liver oil toward the antimony trichloride reaction is not changed b y the chromatogram. A combination of saponification with the chromatographic procedure may result in a loss of as much as 9% of the original vitamin A, but recoveries are good when saponification is omitted. A number of unsaponified fish oils, of the type commonly used in commercial feeds, have the same activity as following saponification.
of time, nor is the rate of change always the same as tor caroteiie, as is shown in Table I. The figures in Table I are index numbers, !There the per cent transmittance a t 5 seconds is arbitrarily assigned the value 100. It seemed at the outset that a chromatographic procedure n ould be desirable whereby vitamin ’4 could be separated from t lie undetermined chromogenic interfering substances. The use of sodium carbonate as the adsorbent in the determination of carotene has been reported (4). This adsorbent permits the carotenes to pass through the column virtually unadsorbed, while retaining the xanthophylls. The efficiency of a number of adsorbents a as examined for the chromatographic determination of vitamin A. Most of these either caused the destruction of the vitamin or adsorbed it SO strongly that 1007, elution was impossible to attain. Sodium carbonate was found to be a satisfactory and ineupensive adsorbent. (“Sormal” sodium carbonate may be obtained in small quantities or by the barrel from Frank W. Kerr Co., 422 West Congress St., Detroit, Nich., or in barrel lots from the Wyandotte Chemicals Corporation, Wyandotte, Mich.) It adsorbed the unknown interfering chromogenic substances while passing the vitamin A readily. It offered the additional advantage that neither vitamin A nor carotene mas adsorbed, so that both vitamin A and carotene could be determined in the same solution after passage through the chromatogram. Samples of sodium carbonate from some other sources did not retain well the noncarotene pigments. A spectral transmission curve prepared from an extract of feed passed through the sodium carbonate column was indistinguishable from that of SMA p-carotene. The transmission, a t 620 millimicrons, of the antimony trichloride reaction product with the eluate determined a t 5 seconds was in good agreement with the expected value on the basis of the carotene content of the eluate.
T
H E uncertainties associated with the determination of vitamin A in mixtures of natural feed ingredients have been well recognized. 1
Baird, Ringrose, and Alac;\lillan (8) observed in connection with the early work of Fraps and Kemmerer (5) that much niaterial was present in feed ingredients which absorbed light at a wave length of 328 millimicrons. An appreciable error w a s introduced thereby, even though the same mixture without added vitamin A was used in a blank determination. Oser et al. ( 7 ) ,using the Carr-Price reaction with antimony trichloride, described a procedure for determining vitamin A in mixed feeds, and discussed its difficulties. Advantage was taken of the fact that the blue color formed by the antimony trichloride reaction with vitamin A faded rapidly, while the color from interfering substances faded s l o d y . Consequently, the color from the reaction remaining after 2 hours was considered to represent carotene and compounds behaving like carotene. The reading for vitamin A was then corrected for the amount of carotene represented by this color. Koehn and Sherman (5) proposed a method whereby the amount of interference is calculated from the carotene content of the sample. However, biologic material may contain compounds other than carotene which are also a cause of interference ( 7 ) . Little (6) has reported on the value of destructive irradiation as an aid in determining vitamin A in fish oil, liver, and muscle tissue.
Table 1.
PRELIMINARY INVESTIGATIONS
For several years the authors have used a method designed in cooperation with Tu’. D. Embree and K. C. D. Hickman of Distillation Products, Inc., chiefly for studies of vitamin A stability in mixed feeds. I n this method the fat is extracted from the feed in a Soxhlet extractor, and the extracted fat is treated in accordance with the method of Koehn and Sherman (5) for the determination of vitamin A in fish oil. The above method required the simultaneous preparation of two mixtures, one a blank sample. Apparent vitamin A was determined in both samples and the amount of true vitamin A was determined by difference. The necessity for preparing the feed without added vitamin A (as a blank), and the degree of inaccuracy introduced by interfering compounds, decidedly limited its value. Consequently, investigation was initiated to find suitable adsorbents for a chromatographic procedure. Unsaponifiable interfering substances present in mixed feeds constitute a considerable source of error unless they are removed. In the authors’ experience they alone may be responsible for an antimony trichloride reaction product ranging from 980 t o 4500 units of spurious vitamin A per pound of feed. This may equal amounts of the true vitamin often present in a feed mixture. Furthermore, the amount of blue color developed from the unsaponifiable fraction of mixed feeds is not constant over a period
a
Effect of Time on Intensity of Antimony Trichloride Color Percentage of Original Transmittance Remaining at: Feed 5 seconds 15 minutes 30 minutes 120 minutes 100 92 89 80 A 100 118 110 102 B 100 101 104 105 Carotenea S M A &carotene.
APPARATUS AND PROCEDURE
REAGENTS. Petroleum ether, boiling point 60” to 71” c. (Skellysolve B). Sodium carbonate, normal carbonate (ground to pass 40mesh screen and stored to prevent moisture absorption). Johns-Manville Hyflo Super Cel. Anhydrous sodium sulfate. Chloroform. 30% solution of antimony trichloride in chloroform. PREPARATIOX OF COLUMN.The chromatographic column is 3.8 cni. in diameter and 25 cm. long. A small plug of cotton is placcd a t the bottom, and the sodium carbonate is packed tightly into the tube to a depth of about 18 cm. This is attached by means of a rubber stopper to a side-arm flask which is connected. to a suction pump. PROCEDURE.Twent,y grams of the feed are weighed into a Waring Blendor jar ; 5 to 10 grams of Hyflo Super Cel and 100 ml. of petroleum ether are added. The mixture is stirred at low speed for 5 minutes, after tvhich the content of the Blendor jar is poured onto the column, and suction is applied to the column. The Blendor jar is rinsed with petroleum ether. This and additional petroleum ether are passed through the column, at lemt 100 to 150 ml. of the solvent being used for rinsing and elution. If a distinct xanthophyll band is observed to pass through the column, the eluate should be subjected to a second adsorption treatment. The carotene and vitamin A are contained in the eluate, which is evaporated to slightly less than 100 ml. by mild heat and re689
690
INDUSTRIAL AND ENGINEERING CHEMISTRY
Table II. Effect of lndlvldual Steps of Procedure on Recovery of Vitamin
Sample Plain oil in petroleum ether Plain oil in petroleum ether, Waring Blendor a t slow speed, 5 mtn. Plain oil in petroleum ether run through soda ash column Plain oil in petroleum ether stirred 5 min. in Waring Blendor and passed through column 20 grams of ground wheat plus oil and etroleum ether, stirred 5 min. i n aring Blendor and passed t h ~ o i i d i column 20 grams of mixed feeda plus oil and petroleum ether run in Waring Blendor and passed through column ~
&v
A
International Units of A Added
200
International Units of A Recovered 200
%
Recovery 100
200
195
98
200
205
103
200
195
98
200
195
98
200
198b
99
Mixed feed emtained yel1i.n corn meal, wheat bran, wheat standard middlings, ground oats. ground barley, soybean oil meal, alfalfa meal, meat a n d bone scraps. dried whey, flavin concentrate, ground limestone, and salt. 6 Corrected for carutene. Carotene nould have caused a n error of 10 vitamin .4 units.
ducecl pressure. The solution is passed through anhydrous sodium sulfate in a funnel and washed with petroleum ether, and the solution and washings are combined and diluted to 100 ml. in a volumetric flask. DETERMXdTION OF VITAMINA. Suitable aliquots are placed in a comparison tube of the spectrophotometer (Coleman Universal photoelectric spectrophotometer was used in this work) and the petroleum ether is carefully evaporated almost to dryness, uqing mild heat and reduced pressure. The residue is dissolved in 1 ml. of chloroform and placed in the spectrophotometer, which is adjusted to the 620-millimicron wave-length band. Ten milliliters of the antimony trichloride solution are added from an automatic pipet designed for rapid delivery. Galvanometer readings are taken and applied to a standard curve. The results are corrected for the amount of carotene present in the solution, using for this purpose a curve prepared from the reaction of antimony trichloride with a carotene standard under the same conditions. The amount of carotene in the solution may be so small as t o be inconsequential. Saponified U.S.P. reference cod liver oil is usually used to prepare the standard curve for vitamin A, since the unsaponified fish oils used in the authors' laboratory provide results in good agreement with this curve. However, a standard curve may be prepared using an accurately assayed oil of the type contained in the feed. The procedure outlined by Koehn and Sherman (6) w m used for saponification, omitting use of an inert gas.
Vol. 17, No. 11
I V provide information on this question. The extract or oils were saponified before being passed through the column. The data indicate that some vitamin A was lost during the procedure after saponification; this may be due to the diminished stability of vitamin A alcohol. The loss amounted to 1 to 9%, which was not sufficiently great to invalidate the method for practical application, a t least for livestock feeds. However, saponification should be omitted whenever practical. The bottom line of Table IV shows the recovery of vitamin -4 from a feed mixture. I n this instance it was impossible to determine vitamin A accurately without recourse to the chromatograph, so the figure 340 represents the amount of vitamin -4 added to the feed. The 91% recovery, then, takes into account the loss from saponification as well as that involved in the chromatographic purification of the saponified vitamin A. Saponification alone has been associated with an average loss of about 3% of the apparent vitamin A in the oils other than cod liver oil. Recovery of vitamin A from this sample by the standard chromatographic procedure without saponification was 101% (345 International Units). Hence, the 9% loss in this instance was directly attributable to saponification. The crude carotene fraction from this feed contained 80 units of false vitamin A per 20 grams (antimony trichloride reaction); 24% of the amount added (340 I.U.). The fact that a suppressing effect is separated from vitamin A by saponification of cod liver oil but not by the chromatogram suggests the'possibility that the substance might be in combination with the vitamin. The possibility is not excluded that some saponifiable compound might be present which inhibits the reaction of vitamin A with antimony trichloride but is not retained by the adsorbent. This substance is apparently not present in proportionately large amounts in many fortified oils containing 4000 A units per gram.
Table 111.
Effect of Chromatograph and Saponification on Reference Cod Liver Oil % Transmittance a t 620 Millimicrons
I.U..of
Vitamin A
U.S.P
of SbCla Reaction Product Unsaponified Oil Saponified oil, Direct Chromatogram direct
APPLICATION O F METHOD
All determinations in this study were mnde using the antiniony trichloride reaction without saponification except where otherwise specified. The 400D-4000.4 oils used in these determinations provided essentially the same figure for vitamin A regardless of whether the oils were first saponified or not. EFFECT OF SAPONIFICATION. For routine control work where the kind of vitamin A carrier used is known, saponification has not been necessary. However, some saponified fish oils, notably lower potency oils including the U.S.P. reference oil, require a different standard calibration curve or conversion factor than without saponification. Fish oils were saponified in accordance with the procedure described by Koehn and Sherman (6),omib ting use of an inert gas. The ether solutions were evaporated under diminished pressure and diluted t o volume with petroleum ether for adsorption studies. Inasmuch as cod liver oil increases in activity to the antimony trichloride reagent after saponification, it was possible that a suppressing factor might be separated by the chromatogram. Table I11 shows that although saponification liberated this activity, the chromatographic procedure did not appreciably alter the characteristics of the unsaponified oil. Therefore, it was desirable to determine whether or not saponification of an alcoholic extract of mixed feeds would provide more reproducible results when the vitamin A was supplied by different oils. The results in Table
SPECIFICITY O F SODIUM C A R B O N A T E C H R O M A T O G R A M
The foregoing data indicate that the sodium carbonate column adsorbs interfering substances in the proportion occurring in some practical feed mixtures. Preliminary investigations have been conducted to indicate the compounds retained by the adsorbent and the effect of unadsorbed materials on the antimony trichloride reaction. Micron Brand magnesium oxide No. 2641 is an excellent adsorbent for the separation of carotenoid pigments according to Strain (9). Therefore, the following general procedure was used
Table IV. Effect of Chromatogram on Recovery of Vitamin A from Saponified Oils Saponified Oil Used U.S.P. cod liver oil 4OOD-4000A feeding oil A 400D-4000A feeding oil B 10,000 A feeding oil Shark liver oil Extract of feed mixture 4
Vitamin A Found Without Chromatochromatogram graphed 1,750 1,600
3,900 3,950 10,350
20,850
340"
%
recovery
3,700 3,900 10,150 20,000 310
Amount added t o mixture prior t o extraction and saponification.
92 95 99
98 96 91
ANALYTICAL EDITION
November, 1945
to show whether residual amounts of noncarotene pigments might be unadsorbed by the sodium carbonate column. An appropriate weight of the ingredient was extracted in the Waring Blendor with petroleum ether for 5 minutes. The extract was passed through the sodium carbonate column and the container and column were washed with 150 ml. more of petroleum ether. The solution was evaporated under reduced pressure to less than 100 ml. and diluted t o exactly 100 ml. Aliquots were removed for colorimetric estimation of apparent carotene. Part of the solution, after passage through the sodium carbonate column, was then passed through a column containing 1 part of Micron Brand magnesium oxide t o 3 parts of Hyflo Super Cel. This column was 1.3 cm. in diameter and 15 cm. in length. The chromatogram was developed, and the carotene wm eluted using a 5’% solution of acetone in petroleum ether. This solution was adjusted t o the original volume and subjected to colorimetric estimation of carotene. This procedure was described by Wall and Kelley (10). In the case of one sample of dehydrated alfalfa meal, the solution after passage through the sodium carbonate column and the solution after serial passage through both sodium carbonate and magnesium oxide columns were found to be identical. Their adsorption spectra were the same. There was no loss in amount of material absorbing light of a 440-millimicron wave length after magnesium oxide adsorption. The sodium carbonate eluate contained no substances separated by the magnesium oxide column under the conditions applied. Additional samples were treated in accordance with the foregoing procedure (Table V).
Table
V.
Effect of Treatment of Sodium Carbonate Eluate by Magnesium O x i d e Adsorption % Transmittance
Substance Dehydrated alfalfa Corn gluten meal Yellow corn a Micron Brand.
Amount G./100
do.
2
10 20
a t 450 Millimicrons NazC08 NazCOa and column MgOa columns 58 55 80 93
83 93
\Vith most ingredients and mixtures the bands remain near the Carotene passes through rrithout forming a distinct band. I n the case of tomato pomace, a clear-cut red band was formed which descended rapidly through the column. In view of the infrequency of use of tomato pomace in livestock feeds, no attempt has been made to modify the adsorption technique to separate this band. The behavior of this compound in relation to the other pigments in the sodium carbonate and in the magnesia column, as well as its source and absorption maximum of 480 millimicrons, indicates that this compound is probably lycopene. Strain (8) reported that lycopene is the most strongly adsorbed compound of the carotene series. Compounds containing polar groups, such as the hydroxyl groups of xanthophylls, are likely to be more strongly adsorbed than the carotenes. Therefore, of the carotenoid pigments, lycopene appears to be borderline in its behavior and, if present, would cause an erroneous estimate of the antimony trichloride color to be expected from the carotene in the sodium carbonate eluate. More polar compounds including xanthophylls are apparently removed by this adsorbent. Conversely, the adsorbent does not differentiate between compounds approaching the chemical structure of @-carotenemore closely than lycopene. The data in Table V indicate a slight loss in color intensity in extracts from alfalfa and corn gluten meal after passage of the sodium carbonate eluate through the magnesium oxide column. I n a second test the color intensity of a magnesium oxide eluate from corn gluten meal was the same as that of the sodium carbonate eluate. These differences in results seem to indicate technical limitations of the method, although the possibility exists top of the column.
691
Table VI. Comparison of Calculated and Observed Antimony Trichloride Color in Carotene Eluate, and Vitamin A Recovery
carotene per
comparison Formula Tube, y 1 4.4
Carotene 7 Transmission oi SgCls Color Actual Calculated
58
Vitamin A per Gram of Feed Added Recovered
56 91
89
14: 5
..
8.5
15:o
..
a Discrepancy probably due t o amount of lipid material in large aliquot used.
Table
VII. Purification of A.O.A.C. Crude Carotene Solution by Adsorption
% Transmission of Replicate Aliquots Dehydrated alfalfa Corn gluten meal Yellow corn Formula 1 Formula 2 Formula 3
NaKOa column 72
MgO column
;;
50 91
92
84 91
92
91
92
83
Crude carotenp 64 27 86
86
76
78
that 100% elution of the carotene fraction might not have been attained from the magnesium oxide column. The accuracy of this method depends upon the degree with which the antimony trichloride color developed from the carotene in the eluate can be predicted in eluates containing carotene and vitamin 9. Two feed mixtures were prepared having the following compositions: Formula 1: yellow corn, 18; ground oats, 10; ground barley, 10; dehydrated alfalfa meal, 5; wheat bran, 15; wheat middlings, 15; soybean oil meal, 10; meat scraps, 10; dried skim milk, 5 ; salt, 0.5; limestone, 1.48; manganese dioxide (technical), 0.02. Formula 2: yellow corn meal, 47.4; dehydrated alfalfa meal, 10; wheat bran, 20; wheat middlings, 5; soybean oil meal, 25; >alt,0.5; limestone, 2.08; manganese dioxide (technical), 0.02. Table VI shows the per cent transmission of the antimony trichloride color calculated from the standard curve, compared with that actually found in the sodium carbonate eluate. The actual and calculated color intensities are in good agreement except in the case of the aliquot from formula 2 containing 40.8 micrograms of carotene. Experiments with low-carotene, vitamin A-free formulas indicate that the antimony trichloride reagent causes a darkening of the solution and decreased transmission at 620 millimicrons. I n practice i t has been found that saponification is advisable when the sample contains less than 1 I.U. of vitamin A per gram. One of the considerable number of fish meal samples used in routine applications of this method was found to contain a brownish pigment which was not effectively separable from carotene either by the sodium carbonate or the magnesium oxide columns. This pigment was responsible for a false antimony trichloride reaction, and was not present in the unsaponifiable fraction of this sample. Saponification is necessary in such instances. A number of fish meals have been found to contain colorless material indistinguishable from vitamin A. The biologic activity of these latter preparations has not yet been determined. I n a further study of the efficiency of the sodium carbonate adsorbent (Table VII) the samples were treated in accordance with the A.O.A.C. procedure ( I ) , for the determination of crude carotene. Aliquots from the crude carotene preparations were subjected to sodium carbonate and magnesium oxide adsorption. Both adsorbents remove appreciable amounts of the A.O.A.C. crude carotene. The agreement between results from the two adsorbents is sufficiently close for prediction of the color intensity
692
INDUSTRIAL AND ENGINEERING CHEMISTRY
developed from antimony trichloride. Small errors in the determination of carotene have comparatively little effect on the accuracy of the vitamin A measurement. ACKNOWLEDGMENT
The authors wish to thank James K. Brody and Harold H. Williams of the Research Laboratory of the Children's Fund of Rlichigan for their courtesy in discussing the unpublished procedure which that laboratory is using for the determination of vitamin A and carotene in human foods. Up to that time the authors had been unsuccessful in obtaining a suitable adsorbent. The method herein described was developed later and differs from the procedure developed in the Research Laboratory of the Children's Fund in the adsorbent used as uell as in other respects.
Vol. 17, No. 11
LITERATURE CITED
(1) dssoc. Official Agr, Chem., Official and Tentative Methods of Analysis, 5th ed., p. 369 (1940). (2) Baird, F. D., Ringrose, A. T., and MacMillan, M . J., Poultry Sci., 18,441 (1939). (3) Fraps, G. S., and Kemmerer, A. R., Texas Agr. Expt. Sta., Bult. 557 (1937). (4) Kernchan, G., Science, 90,623 (1939). (5) Koehn, C. J., and Sherman, W. C., J. B i d . Chem., 132, 527 (1944). (6) Little, R. W., IKD. ESG.CHEM.,ANAL.ED.,16, 288 (1944). Ibid., 15,724 (1943). (7) Oser, B. L., Melnick, D., and Pader, M., (8) Strain, H. H., "Chromatographic Adsorption Analysis", p. 25, New York, Interscience Publishers, 1943. (9) Ibid., p. 57. (10) Wall, M. E., Lnd Kelley, E. C., IND.ENG.CHEM.,ANAL. ED., 13, 18 (1943).
Improvements in the Potentiometric Titration of Chlorides ROBERT P. YECK
AND
G.
H. KlSSlN
Amzrican Smelting & Refining Company, Central Research Laboratory, Barber,
A simplified electrode system without a salt bridge is described. The assembly consists of a metallic silver electrode and a reference electrode reversible to chloride ions, mounted with a stirrer in a compact unit. The method makes possible titration of chloride ion, with silver nitrate solutions, to an abrupt potential swing at the equivalence point, detected with an electronic voltmeter, resulting in a simplicity and convenience comparable to conventional redox eledrometric titrations. The use of an electronic voltmeter for the detection of the end point eliminates errors of polarization and makes possible the routine potentiometric titration of chloride in solutions containing heavy metal ions.
I Electrolyte -Ig1 Sample
N. J.
KK03(l K) AgCl(sat.)
1 AgCl 1
Ag
At the beginning of the titration when the chloride ions are at high concentration in the sample they are relatively low at the silver chloride electrode. Additions of silver nitrate decrease the chloride-ion concentration in the sample, causing a corresponding change in the potential of the silver chloride indicator electrode.
THE
potentiometric method for chloride determination has considerable advantage over titrations involving chemical indicators and the gravimetric silver chloride method. The recent paper by Yao (@, proposing the titration of chlorides with silver nitrate by a series of potential measurements, and finally, to a definite potential, is not adaptable to routine work; the equipment required and the time and care necessary in making a series of readings make routine use impractical. The titration of chlorides by use of the polarized electrode systems described by Foulk and Bawden (S), Willard and Fenwick ( 7 ) ,and Clippenger and Foulk ( I ) , while capable of giving excellent results, had the disadvantage of requiring chemical separation of heavy metal ions. Because of the applied electrode potential in this type of circuit, depositable ions must be absent, necessitating chemical separations. Titrations of chlorides R ith adsorption indicators proposed by Fajans ( 2 ) , Kolthoff ( 6 ) , and others required the usual separations and, in addition, close pH control. The older chromate method has these disadvantages plus the personal factor of error in exact reproducibility. ELECTRODE CHARACTERISTICS
For the determination of chlorides, an electrode system not adversely affected by other ions, giving a potential change (Figure 1) abrupt enough for easy reading on an electronic voltmeter (titrimeter) was desirable. Such a system has proved satisfactory for rapid control on a variety of materials without chemical separations and with reasonable accuracy. The electrode system consists of two reversible electrodes: the silver chloride electrode reversible to chloride ions, and the silver electrode reversible to silver ions. The cell may be indicated as follows:
70
0.1 .2 .3
.5 ,6 .7 .8
,.,"ML.TITRANT Figure 1.
-
Potential Change
Until the equivalence point is reached, the potential of the silver electrode is determined by the concentration of silver ions resulting from the solubility of silver chloride, which is approximately constant. At the equivalence point the chloride-ion concentrations in the sample and at the silver chloride electrode are approximately the same, controlled by the solubility of silver chloride, but the silver-ion concentration now changes because of the slight excess of silver nitrate. This results in an abrupt change in the potential of the silver electrode indicated by a meter swing of the titrimeter similar to that encountered in redox titrations. ELECTRODE DESCRIPTION
The reference electrode described here offers maximum simplicity of operation and maintenance, elimination of the salt