1952
ANALYTICAL CHEMISTRY
purity than any of the other samples. It was a liquid IThile all the others were solids. I n addition, it would appear that some of these samples contain an impurity n-hich is not any of those considered, as all the latter compounds would give values for F which ~ o u l dbe either equal to or greater than E. It is probable that these unknown impurities are xeaker bases than those being considered. Determination of Water. Kater may be determined in the various pyrazine samples by the Karl Fischer reagent in the usual manner. A stable end point could be obtained. LITERATURE CITED
(1) Beilstein, F., “Handbuch der organischen Chemie,” 4te Auflage,
Bd. XXIII, p. 91, Beilin, Julius Springer, 1936; -4nn Arbor Mich., Edwards Bros., 1944. Hillebrand, T. F., and Lundell, G. E. F., “Applied Inorganic Analysis,” p. 172, S e w York, John Wiley &- Sons, 1929. Pfann, H. F., J . Am. Chem. SOC.,66, 155 (1944). Stoehr, J . prakt. Chem., 51, 459 (1895). RECEIVED for review April 21, 1952. Accepted September 4 , 1952. Presented before the Division of dnalytical Chemistry a t the 121st Meeting of the AXERICAN CHEMICAL SOCIETY. Buffalo, K. Y., a n d before the Meetingin-Miniature of the North Jersey Section, AMERICAACHEMICAL SOCIETY, J a n u a r y 9, 19.50.
Paper Chromatographic Separation and Determination of Some: Water-Soluble Vitamins ‘
JAMES A. BROWN AND RlAX 41. MARSH Eli Lilly and Co., Indianapolis, Ind. With the large number of multiple vitamin preparations now being manufactured, the problem of adequate routine analytical control of these mixtures is becoming increasingly complex. As a potentially simpler general technique, paper chromatographic separation procedures for various vitamins were studied. It was found that four of the commonly encountered B complex vitamins-thiamine salts, riboflavin, nicotinamide, and pyridoxine hydrochloridewere readily separable on paper strips in a butanol-acetic acid-water system. By means of an automatic scanning device adapted for the Cary
E
XAMIKATIOS of the literature reveals the work of many on paper chromatography. It has become an accepted tool of the analytical chemist for the separation of closely related materials. K i t h particular reference to vitamins, Hais and Pec&kovB( 7 ) have used this procedure for separating riboflavin, Beran and Sicho ( 1 ) for thiamine, Kodicek and Leddi (9) and Huebner (8) for nicotinamide, and Wollish, Schmall, and Shafer (13)for nicotinic acid. I n this paper it is shoTYn how four synthetic vitamins of the B complex-thiamine salts (hydrochloride or monitrate), riboflavin, nicotinamide, and pyridovine hydrochloride-can be simultaneously separated, identified, and quantitatively estiniated by means of the paper chromatographic technique, the apparatus for automatically scanning paper strips with the Cary recording spectrophotometer described by Parke and Davis (IO), and the correlation of area measurements of the peaks on the resulting graphs. Somewhat similar methods for the quantitative evaluation of components separated on a paper strip have been described ( & - 6 , I I , I 2 ) . All of these make use of a relation between concentration and a measure of zone dimension-Le., maximum absorbancy, total absorbing area, or length. I n all cases, precision has been rather unsatisfactory and the discontinuous process of measurement very tedious The present method for plotting the absorbancy of the zones continuously in order to obtain accurate zone measurements is rapid and the precision obtainable is much greater than that outlined in other procedures. EXPERIMENTAL
The paper chromatographic portion of the work involved the development of chromatograms on 17/32 X 22.5 inch strips of Whatman S o . 1 filter paper cut parallel to the grain of the paper. Bscending technique vias used, the solvent being placed in the
recording spectrophotometer, curves were plotted of total ultraviolet absorption us. position on the strip. Quantitative results were obtained by measuring the areas of the discrete zones of absorption corresponding to each component and comparing them to those obtained from standard solutions. Preliminary studies indicate that the quantitative technique devised is sufficiently reproducible to be useful as a control method for certain types of pharmaceutical preparations. A single procedure may be utilized to analyze mixtures of many different and not necessarily related compounds.
bottom of 12 X 24 inch cylindrical glass jars having flat, closefitting glass lids. The jars were wrapped with opaque paper to prevent the destructive action of light. Jars made of nonactinic red glass have subsequently been obtained and are more convenient to use. The solvent system finally chosen was the upper phase obtained by shaking together 40 parts of reagent grade n-butyl alcohol, 5 parts of reagent grade glacial acetic acid, and 55 parts of distilled water (by volume). This solvent is similar to that of Hais and Pecitkovit ( 7 ) , but contains less acetic acid. The vitamins (purest commercial grade available)-individually and in mixtures-were dissolved in 50% (v./v.) acetic acid in water (heating, if necessary) and finally diluted so the resulting solution was 10% (v./v.) acetic acid. A 0.0175-ml. sample was applied 1.5 inches from the end of each strip. It was measured with a capillary pipet made from a common 0.09-ml. antibiotic pipet. The chamber of an antibiotic pipet was drawn out to a small volume and the outside dimension of the tip was reduced to about 1.5 mm. The chamber was filled by capillary action and discharged by the capillary action of the paper strip when the strip was touched to it. I t was calibrated with distilled water, using a semimicrobalance to weigh the water delivered, and Tvas found to discharge 0.0175 =k 0.0002 ml. The strips were dried in a 50’ C. oven after the sample wae applied, 10 minutes’ time in the oven being sufficient. They were then suspended in the jar with the end containing the sample dipping into the solvent in the bottom of the jar to a depth of approximately 0.25 inch. The lid was lubricated with stopcock grease and weighted to make a tight fit. The strips were developed overnight (I5 hours) a t room temperature and then dried in a 50” oven for 15 minutes. The principal scanning to provide measurable areas was performed at predetermined wave lengths; absorbancy us. a linear function of strip length was recorded as outlined by Parke and Davis ( I O ) . The instrument was set near its maximum sensitivity, the slit setting being about 2.75 mm. The slit width changed somewhat during the course of a run to compensate for voltage or light intensity changes. Figure 1 shon-s the graphs of a typical strip both before and after development.
1953
V O L U M E 24, NO. 12, D E C E M B E R 1 9 5 2
W U
z
4
m IY
0 ul
a
< H
W
> I-
4 W E
G F
E
0 C
0
30
20
25
15
10
5
0
RELATIVE P O S I T I O N ALONG PAPER S T R I P Figure 1. Scan of Filter Paper Strip before and after Chromatography A. E. C. D. E F: 6. H.
Before, a t 370 mp Before, a t 297 m p Before, a t 264 mp Before, a t 240 m p After a t 370 m p (riboflavin) Afte; a t 297 m g (pyridoxine hydrochloride) After: a t 264 m p (nicotinamide) After, a t 240 m p (thiamine mononitrate)
I n this work, individual known solutions of a vitamin nere compared to known solutions of a mixture of the four. I n practice, unknown solutions of a vitamin preparation would be compared to known solutions of a mixture of the same vitamins. The area of each peak in the series was measured, aftrr thp base line had been inserted (see discussion), with a Keuffel and Esser polar planimeter readable to 0.01 sy. inch. Table I11 shows the resulting data. DISCUSSIOh
The paper strips ":a2 inch wide were used because the dimensions of the scanner are such that this width permits the maximum exposure of the zone to the light beam of the instrument. The beam width of the spectrophotometer was found to be about 0.5 inch a t the phototube end of the cell compartment. It was further found that the paper must extend a minimum of 1/64 inch beyond each side of the aperture to afford a guiding surface; the result was a 17/a2-inch strip which alloffed 94% of the strip width to be scanned. When a strip wider than "/a2 inch Kas used, a correspondingly lower percentage of the strip was scanned and the accuracy was decreased because of a variability in the
lateral distribution of the material in the band. Subsequently i t seemed desirable to redesign the scanner mechanism to incorporate an aperture 1j/a2 inch wide in order to use a 0.5-inch strip, since O.5-inch rolls of Khatman filter paper are a stock item. The reduction in sensitivity due to the slightly smaller aperture is not significant. The cutting of 18 X 22.5 inch filter paper sheets parallel to the grain of the paper gave strips of a convenient length; the final results were as good as or superior to those obtained when strips cut perpendicular to the grain were used. Whatman S o . 1 paper was chosen because i t has the most uniform structure of several types examined, thus giving the least variability in the background absorption. It was found that other types of paper sometimes contained impurities which seemed to destroy some of the vitamins. As is shown in Figure 1, there is considerable variation in background absorption of the paper strips. I t is this variation which a t present is one obstacle to the automatic integration of the curves and the resultant automatic evaluation of area. Improvements in this technique an-ait the development of a more uniformly light-absorbing material possessing the capillary properties of paper.
ANALYTICAL CHEMISTRY
1954
The wave lengths used to obtain the graphs of a typical strip in Figure 1 are those of peak abSystem (Percentages Given b y Volume) Reason for Rejection sorption of each substance and were determined Ethyl acetate 48%-acetic acid 9%-water 43% Too little travel of BI and Bz originally on a similar strip by scanning a t an Too much travel of all vitamins Benzyl alcohol 40%-acetic acid 10%-water 50% approximate wave length to locate the bands and Butyl alcohol 48%-concd. HC1 Z%-water 50% Too little travel of all vitamins Butyl alcohol 40%-acetic acid 20%-water 4070a Too much travel of all vitamins then positioning each band in the spectrophotomButyl alcohol 40%-acetic acid 15%-water 45%a Too much travel of all vitamins Butyl alcohol 40%-acetic acid 13%-water 47% Too much travel of all vitamins eter beam and recording absorbancy L I S . wave Butyl alcohol 49%-acetic acid l%-water 50% Too little travel of B1 and Bz length in the conventional manner. The absorpButyl alcohol 60%-water 50% BI and B I ran together. a These systems were miscible a n d were used a s such. The rest were not miscible and tion spectra for the typical strip are shown in the water poor layer was used in all cases. Figure 2 . The peak wave lengths are constant for the solvent system used, but may change in other systems. The solvent system chopen seemed to give the best band separaI n all the spectrophotometry in this work it proved necessary tion for the conditions of concentration and component limitato use a piece of untreated paper in the reference beam rather tions with n-hich this paper is concerned. Table I gives a number than to utilize the internal alternating current reference feature of the Cary instrument. The iris diaphragm attachment (10) of other systems tried and the reasons for their rejection. was used for this purpose. The dissolving procedure and solvent for the vitamin samples The characteristic peaks along the strip for each vitamin were necessary because of the difficult solubility of riboflavin. are very evident in Figure 1. The R f value (location on the strip) Table I1 gives some other possible solvents, of 11hich only the pyridine-water system seems feasible. It would be desirable of a band is constant in any given system and serves t o help identify the substance. The peaks a t the extreme left are located to have a solvent volatile enough to dry quickly, if drying of the a t the solvent front and are due to impurities in the system. strips before development is desired Drying is not absolutely These impurities are not detrimental and the peak serves the usenecessary with the dilute acetic acid; however, when not dried, ful purpose of marking the front. The pattern resulting from there is a pronounced tendency for the thiamine to appear as the scanning is shown to be highly reproducible on the average a double peak and there is some tendency for thiamine and riboexcept for the low amplitude “wiggle” of the line due to backflavin to run together. ground “noise” in the instrument. This noise is characteristic The four vitamins were used in the chosen proportions because of the Cary instrument when operating a t the highest sensitivity those ratios are commonly encountered in multivitamin preparations. The proportions vary among various products. but in nearly all cases 11here both nicotinamide and pyridoxine hydrochloride are present, the proportion of the former to the latter iE of the order of 15 to 1. Table I.
Table 11.
Solvent Systems Tried for Vitamin Chromatography
Solvents for Vitamin Mixtures Containing Riboflavin
Solvent (Percentages Given by Volume) Water 10% acetic &id in water 50% acetic acid, then diluted t o 10% Acetone 80% acetone in water 80% dioxane in water 80% pyridine in water
Characteristics with Respect t o Riboflavin Dissolves too slowly Dissolves too slowly Dissolves in hot 50% acetic acid Maximum solubility too l o x Solution O.K. when w’arm, supersaturated when cold Riboflavin destroyed Apparently satisfacton
The size of the samples used n a s near the maximum thought to be allowable. It seemed desirable to keep the area of the sample spot as small as possible, A length of about 6/8 inch a t the point of application of the sample to the strip appears to be about maximal; this corresponds to a sample size of about 0.02 ml. At the same time the volume of the sample must be kept as large as possible because of the limited solubility of riboflavin and the small amount of pyridoxine hydrochloride present. The highest concentrations shown in Table IV are maximal for the conditions described. If the absolute amount of nicotinamide on a strip were much higher than that given, the recorder pen would be driven off the chart. The area resulting from pyridoxine hydrochloride is low, but seems to be adequate. I t has been possible to obtain higher concentrations of riboflavin in the dilute acetic acid, but such solutions are supersaturated The drying of the strip after development is, of course, necessary. Because acetic acid absorbs in the ultraviolet region, even small amounts may affect the akjsorption pattern. However, trouble from solvent recidues was never a problem if the strips were dried for 15 minutes a t SO” C. It was found to be necessary to protect the strips from light a t all times. The riboflavin band fades rapidly if exposed to daylight. When being scanned, the exposed length of strip was always covered.
h 200
1
I
I
250
300
350
WAVE LENGTH Figure 2.
400
-ap
Ultraviolet Absorption Spectra of Zones on Paper Chromatogram A . Pyridoxine hydrochloride B . Riboflavin C. Nicotinamide D . Thiamine mononitrate
V O L U M E 2 4 , NO. 1 2 , D E C E M B E R 1 9 5 2
1955
quires very little additionaI effort to run a number of rep(0.3331 g. nicotinamide per 100 ml. h a t e strips per sample, adeg . pyridox/ne hydrochloride per 100 ml. Solution A , { 0.0222 quate data can be obtained 0.0530 g. riboflavin per 100 ml. 10.1112 g. thiamine mononitrate per 100 ml. which will permit statistical Solution B. 0.3331 g. nicotinamide per 100 ml. Solution C. 0.0222 g. pyridoxine hydrochloride per 100 ml. evaluation of the results with Solution D. 0.0530 g. riboflavin per 100 ml. Solution E. 0.1112 g . thiamine mononitrate per 100 ml. almost any desired accuracy. Table IV and Figure 3 show Solutions of Individual Vitamins Solution B Solution C Solution D Solution E Xmax 264 mp Xmax 297 mp Xmsx 370 m p Xmax 240 m p the results of a run a t several 1st d a y 2nd day 1st d a y 2nd d a y 1st d a y 2nd d a y 1st day 2nd day concentrations of a given proZonalareas,sq.inches 5.16 5,24 1.01 1.11 1.82 1.78 5.17 4.65 portion of the four vitamins. 5 47 5.19 1.06 1.11 1.85 1.73 5.07 4.72 4.89 5.24 1.01 1.04 1.80 1.71 4.84 5.00 The graphs of log concentra5.69 5,57 1.03 1.09 1'82 1'73 5'06 4'96 tion us. log area show that the Average 5.30 5.31 1.03 1.09 1.82 1.74 5.04 4.83 Dev. from av., % +7.4 i4.9 +2.9 +1.8 +1.6 f2.3 4-2.6 $3.5 relationship gives a straight or -7.7 -1.3 -1.9 -4.6 -1.1 -1.7 -4.0 -3.7 nearly straight line when thus plotted. This corresponds to Mixture of Four Vitamins (Solution A) Kicotinamide Pyridoxine HCI Riboflavin Thiamine h-08 an equation of the type A = Xmax 264 mp Xmax 297 mp X m a x 370 mp Xmsx 240 mp 1st day 2nd dav kCn, where k and n are conI t t day 2nd d a y 1st d a y 2nd da, 1 s t day 2nd d a y Zonal areas, sq. inches 5.42 4.94 1.01 1.05 1.66 1.57 4.86 4.60 stants. Through further re5.41 4.93 1.10 0.98 67 63 '.O1 4 , 68 finements it may ultimately be 5.57 5.15 0.98 1.05 1.66 1.72 4.82 5.01 5.37 5.36 0.96 1.05 1.78 1.75 4.90 4.94 possible to establish a standard 1.03 Bverage 5,44 5.10 1.01 69 " 67 4. 4. 86 curve which would apply to all Dev. f r o m a v . , yo +2.4 +5.1 '9.0 +1.9 +5.3 +4.8 i2.2 +4.2 determinations under a given -1.3 -3.3 -50 -4.9 -1.8 -6.0 -1.6 -4.4 set Of conditions. Four area values for each solution on each d a y represent replicate strips. Entirely new solutions were made for the second day's r u n , t h e concentration of both being identical. The Cary recording spec__trophotometer has proved to be an ideal instrument for this application. The sensitivity is (maximum dynode voltage), but has not so far been detrimental high and the response of the recorder pen extremely rapid. With it, the ultraviolet spectrum of a given band can be rapidly deterto the results of this work. The background pattern other than mined. However, where such an instrument is not available, the noise makes the prescanning of the strips to be used absolutely necessary for quantitative work in order that a valid base it may still be possible to apply the method. The authors have determined that a Beckman DU spectrophotometer with a Beckman line can be traced in for each peak. However, in practice it photomultiplier attachment can be used to measure the absorbhas not been found necessary to prescan a t each wave length ancies of bands on the paper strips ~ i t hadequate sensitivity. Because the noise characteristics of the paper do vary somewhat A scanning mechanism similar to the present one has been with wave length, it was found desirable to run a base line a t 370 designed for attachment to the Beckman DU monochromator and mp and one a t 264 mp, the latter being used for the 240 and 297 it is proposed to measure the per cent transmittance by using a mh traces as well. I t was demonstrated experimentally that development of a "blank" strip in the solvent system did not alter the background absorption appreciably. Table IV. Concentration-Area Data for Four \'itamins at Various Table I11 shows among other things the reproConcentrations ducibility of the area measurements, It can be Solution 1 2 3 4 0 seen that the spread is on the average below 12%. Sicotinamide g per 100 ml. 0.0800 0.1600 0.2400 0.3200 0.4000 Pyridoxine H ~ ;'Ig. per 100 ml. 0.0080 0.0160 0.0240 0.0320 0.0400 For small areas the spread tends to be greater Riboflavin, g. p& 100 ml. 0.0120 0 0240 0.0360 0.0480 0 0600 on a percentage basis than for large ones and ocThiamine NO*, g. per 100 ml. 0 0280 0.0360 0.0840 0,1120 0.1400 casionally the spread for areas of any size goes Zonal Areas, Square Inches above the 12% value. Some trouble has been exSicotinamide, Pyridoxine, Riboflavin Riboflavin Thiamine, max. 264 mp max. 297 mp max. 370 m; max. 275 m; max. 240 mu perienced with occasional apparent variations in Solution 1 2.12 0.48 0.52 0.97 1.37 the speed of the paper through the scanner which 1,96 0.39 0.44 0.89 1.41 2.00 0.41 0.49 0.89 1.46 have affected the areas a t any time a speed varia2.05 0.41 0.51 0.95 1.36 tion occurs while a band is passing the aperture. AV. 2.03 0.42 0.49 0.93 1.40 The occasional instances where one area is Solution 2 3.35 0.77 0.91 1.58 2.57 0 95 1,45 2.49 3.36 0.75 markedly different from the rest have been attrib3.41 0.71 0.84 1.39 2.72 uted partly to the speed variation and partly to 3.55 0.78 0.91 1.68 2.99 .4v. 3.42 0.75 0.90 1.53 2.69 variations in the lateral distribution of the vitamin in a band. The causes of the speed variations have not been fully determined, but a t least part seems to be due to the soft character and the slightly varying thickness of the rubber on Solution 4 5.50 1.22 1.61 2.26 4.53 6.05 1.28 1.57 2.30 4.80 the rolls. h somewhat different design for the 5.92 1.33 1.56 2.15 4.58 5.80 1.40 1.51 2 20 4. 8.5 .. roller, which it is believed R-ill eliminate the sv. 5.82 1.31 1.56 2.23 4.69 trouble from this source, is being developed. Often Solution 5 6.18 1,75 1.89 2.30 5,47 there is remarkably good agreement among rep6.37 1.61 1.88 2.37 5.60 6.75 1.61 1.97 2.37 q.03 licates, which leads to the thought that ulti6.26 1.62 1.94 2.59 n.24 mately as techniques and equipment improve and BY. 6.39 1.65 1.92 2.41 .5.34 the causes of the variations are determined, the Each set of four values represents replicate strips precision could increase considerably. As it rp-
Table 111. Concentration-.4rea Data for Four Vitamins Individually and Mixed Together
-
I
7
'' ''
''
1956
ANALYTICAL CHEMISTRY analytical chemist has involved only the four vitamins discussed for the sake of expediency. Work on several other vitamins as well as on other substances is in progress. It is believed that with special precautions t o prevent deterioration, the method can be extended to include applications to many nom difficult separation problems. The quantitative application of this technique may prove useful in evaluating natural vitamin sources.
6.0 8.0
4.0 ,
I
.
51
I
3.0 2.0
-
ACKNOWLEDGMENT
I
o
;
0.4
c
'
/
1
/
0
RIBOFLAVIN
Xcknon-Iedgment is made to W. W.Hilty of these laboratories for valuable suggestions and assistance and to Beckman Instruments, Inc., Chicago Office, for the loan of a photomultiplier attnchment.
- 275
0 T H I A M I N E NO,
LITERATURE CITED
0.3 0.005
0.01
0.02
CONCENTRATION
Figure 3.
0.05
- GM.
0.1
0.2
P E R 100
Log-Log Plot of Concentration
0.4
ML. cs. Area
photomultiplier tube and nn amplifier to drive a fast-acting recorder. Thus, after the penks for a given system have been determined (which can be done manually with the assembly described above), all the information needed can be obtained with a setup which is considerably less expensive than the Cary instrument. However, it is not possible a t this time to report data obtained on such a setup, because of the slow delivery of suitable instrumental components. Plotting of per cent transmittance rather than density 21s. time would seem to have some advantages, as the areas of the less intense bands would tend to be greater in relation to those of the more intense bands. The arca-concentration relationships would no longer produce a substantially straight line on log-log paper but might do so on a semilog chart. The development of this application of a new tool for the
(1) Beran, IT.,and Sicho, V., Chem. L i s t y , 45, 134-6 (1931). (2) Block, R. J., Science, 108, 608 (1948). (3) Bull, H. B.. Hahn, J. IT., and Baptist, V. H., J . Am Chem. SOC., 71,550 (1949). (4) Fisher, R. B., Parsons, D. S., and Morrison, G. A , , S a t w e , 161, 764-5 (1948). (5) Fosdick, L. S.,and Blackwell, R. Q., Science, 109, 314-15 (1949). (6) Fowler, H. D., Sature. 168, 1123-4 (1951). 17) Hais. . . -~ ~, I. AI.. and PecLkovL. L.. Ibid.. 163. 768 (19491. (8) Huebner, C. F., Ibid., 167,' 167, 119-20 (1951 (1953). j. (9) Kodicek, E., and Leddi I