Photoelectirc Vitamin A Photometer

Balanced galvanometer, Leeds. Northrup No. 2320. K\. Coarse adjustment key. Ki. Fine adjustment key. R. 200 megohms that of the zinc lamp, while the r...
1 downloads 0 Views 440KB Size
374

INDUSTRIAL AND ENGINEERING CHEMISTRY

Vol. 13, No. 6

Summary

Literature Cited

The methods available for the determination of combined acyl in cellulose esters fall into two classes: saponification with aqueous or alcoholic alkalies or hydrolysis in acid solutions. Difficulties inherent in these methods and of establishing the accuracy of a method are pointed out and discussed. The modified Eberstadt method is the most accurate and generally satisfactory for the analysis of cellulose acetate and certain other esters of low molecular weight acids. It involves swelling the sample with aqueous alcohol and saponifying with aqueous alkali a t room temperature for 1 to 2 days. A more nearly universal procedure consists of a saponification with 0.25 N alcoholic alkali for 16 to 24 hours a t not higher than 30” C. However, good accuracy is more difficult to attain by this method than by the Eberstadt method. The acid distillation method of Ost requires more labor per sample, but is the most rapid of the three methods. It has unique advantages which make it useful for special purposes if not for routine use. The effects of the most important variables, limits of applicability, precision, and accuracy are given for these three methods.

(1) Abribat, M., Ann. chim. anal., 15,147-57 (1933). (2) Battegay, M., and Penche, J., Bull. aoc. chim., 45, 132 (1929). (3) Cross, C. F., and Bevan, E. J., “Researches on Cellulose, 18951900”, p. 38, London, Longmans, Green and Co., 1901. (4) Eberstadt, O., dissertation, “Ueber Acetylcellulose”, Heidelberg, 1909. (5) Green, A. G., and Perkin, A. G., J . Chem. Soc., 89, 811 (1906). (6) Knoevenagel, E., 2.angew. Chem., 27,507(1914). (7) Knoevenagel, E., and Koenig, P., Cellulosechem., 3, 119 (1922). (8) Krueger, D., Farben-Ztg., 35, 2032 (1930) ; “Zelluloseazetate”, pp. 218-28, Berlin, Theodor Steinkopff, 1933. (9) Marsh, J. T., and Wood, F. C., “Introduction to the Chemistry of Cellulose”, p. 213, New York, D. Van Nostrand Co., 1939. (10) Mork, H. S.,J . Am. Chem. Soc., 31, 1069 (1909). (11) Murray, T. F., Jr., Staud, C. J., and Gray, H. LeB., ISD. ENG. CHEM.,Anal. Ed., 3, 269 (1931). (12) Ost, H., 2.angew. Chem., 19,993 (1906); 25, 1469 (1912); 32, 67 (1919). (13) Pilgrim, F. D., Tennessee Eastman Corp., unpublished data. (14) Roeper, E., Eastman Kodak Co., unpublished data. (15) Woodbridge, R . G., Jr., J . Am. Chem. SOC.,31,1067 (1909). (16) Zemplen, G., Gerecs, A . , and Hadacsy, I., Ber., 69, 1827-9 (1936). PRESENTED before the Division of Cellulose Chemistry at the 100th Meeting of the American Chemical Society, Detroit, Mich.

Photoelectric Vitamin A Photometer BEAUMONT DEMAREST, National Oil Products Company, Harrison, Iv. J.

W

HEN it was established that vitamin A has a peak of absorption a t 3280 A. which has a high degree of per-

sistence in most fish liver oils, it became possible to make vitamin A assays spectrophotometrically (2, 3, 6). This represented a great saving in time over the animal test. It was soon recognized that an instrument which would measure photoelectrically the absorption of fish liver oils in the neighborhood of 3280 A. would be a valuable contribution, since, compared with the conventional spectrophotometer, it would offer a means of more rapid and objective vitamin A measurements while also representing a smaller initial cost. Several such instruments have been constructed, the most prominent of these being the United Drug Company vitamin A meter (made by the G. M. Laboratories, Chicago, Ill., 6), the Bills and Wallenmeyer electronic photometer (made by the Schoene Equipment Co., Evansville, Ind., I), and the photoelectric photometer of Parker and Oser ( 7 ) . The electronic photometer uses, as a light source, an argon lamp, isolating the two argon bands at 3180 and 3380 A. by means of a Corning KO.986 filter and a nickel chloride solution. The other two instruments use, as a light source, a copmercial sodium vapor lamp, isolating the sodium line at 3303 A . by means of a single Corning No. 986 ater. I n this laboratory, a new photoelectric instrument has been developed for the measurement of vitamin A, using a zinc vapor lamp as light source and a single sodium photocell. The zinc arc spectrum contains six strong lines between 3282 and 3346 8. When used with a sodium photocell, a single Corning KO.597 filter effectively isolates these six lines, since the sodium photocell does not respond to the zinc arc lines above 5000 A. which are transmitted by the filter. The transmission of this filter for the isolated group of six lines is about 50 per cent. This group of zinc lines has certain advantages over the sodium and argon lamps as light sources. As compared with the commercially available sodium lamps, these six zinc lines contain a much higher percentage of the total radiation of the zinc spectrum than the percentage of the total sodium

K.

radiation contained in the single sodium line at 3303 This larger amount of available energy results in a simplification of the amplification problem. Of the three light sources mentioned, the sodium lamp would appear to be superior from the point of view of wavelength characteristics, since its effective radiation is much closer to the peak of vitamin A absorption. That this is not necessarily true is brought out in the following discussion. \Then one is dealing with a fish liver oil or concentrate which exhibits a true vitamin A absorption curve, either type of radiation would yield the same result, provided the proper calibration is applied to each. I n the case of materials which show a departure from the true vitamin A curve, different results may be obtained, depending upon the type of radiation used. I n the absorption spectra of fish liver oils and concentrates, the departures from the true vitamin A curve which are commonly encountered may be divided into two classes. Those which are due simply to interfering absorption, which we may call Type I, are characterized by a decrease in persistence of the peak a t 3280 1. Other departures, which we may call Type 11, show a deformation of the vitamin A absorption curve itself and are characterized by a serjes of absorption peaks on the long wave-length side of 3280 A,, such as reported by Edisbury et al. (4). I n general practice, departures of Type I are encountered much more frequently than those of Type 11. I n the case of Type I departures, the correct result-i. e., the result which would be obtained if the interfering absorption were removed-is equal to or less than the lowest of the several results with each of the light sources, since the effect of interfering absorption is always additive. I n the study of a large number of absorption curves of Type I, it has been noticed that the departures from the true vitamin A curve are such that the zinc lamp will never yield a higher result than either the sodium or argon lamp. Furthermore, when there is great loss of persistence of the vitamin A peak, the result given by the sodium lamp will be somewhat higher than

375

ANALYTICAL EDITION

June 15, 1941

of drift of the plate current, the operation can be quickly repeated, since the potentiometer is now very near the correct setting. In practice, very little difficulty was encountered due to drift of the plate current.

I

1.

" 8, FIGURE 1.

I

The photocell dark current has no effect on the potentiometer setting. If I is the first potentiometer setting made with the cell containing a solution of vitamin A oil in 99 per cent isopropanol in the light path, and IOis the second potentiometer setting made with the cell removed, the density of the cell plus the solution is given by log Io/I. The extinction of the dissolved oil is given by subtracting from this the density of the cell containing pure solvent, which has been previously determined by an average of several measurements. This procedure, instead of the usual practice of refilling the cell with pure solvent, shortens the time interval between the two potentiometer settings, which is an obvious advantage. The E value of the oil is obtained in the usual manner by dividing the extinction by the concentration.

DIAGRAM O F PHOTOMETER

B I . 180 volts B;. B-battery, 22.5 volts Bs. A-battery, 2 volts B I . C-battery, 1.5 volts Bs. 1.5 volts G. Balanced galvanometer, Leede 8- Northrup No. 2320 K I . Coarse adjustment key K2. Fine adjustment key R . 200 megohms

t h a t of the zinc lamp, while the result given by the argon lamp will be considerably higher than that of the sodium lamp. From this we can assert that, for oils and concentrates showing departures of Type I, the zinc lamp is superior t o both the sodium and argon lamps. I n the case of Type I1 departures, the zinc lamp will usually yield a slightly higher result than either of the other light sources, the sodium lamp being more nearly correct. However, bearing in mind that departures of Type I are more commonly encountered than those of Type 11, the zinc lamp is a more satisfactory light, source than the argon lamp for the routine vitamin A measurement of fish liver oils and concentrates, and appears to be a t least as satisfact'ory as the sodium lamp.

Precision Although the single-photocell type of instrument is admittedly simpler in construction than the double-photocell type, it has the inherent weakness that the results obtained in its use are affected by variations in the intensity of the light source. However, the error so introduced can almost always be made negligibly small by a restriction of the operating density range. Consider an instrument of the type shown in Figure 1, assuming that the response of the photocell is constant. Assuming for the moment that the intensity of the light source is constant, Z and ZOwill both be subject to a constant uncertainty, 4Z, inherent in the current measuring system. The effect of the variation in the intensity of the light source may be properly introduced by considering l o subject to an additional uncertainty, AZO,which represents the variation in the intensity of the light source during the time between the measurement of . Z and l o . The resulting fractional uncertainties in log Io - will Z be as follows, grouping together the terms containing AZ:

Details of Instrument Figure 1 is a diagrammatic representation of the new instrument. 2,the zinc vapor lamp, is manufactured by the Osram Lamp Company of Germany. The lamp, in series with a choke coil, operates on 230-volt alternating current through a constantvoltage transformer (Sola Electric Company). It consumes about 50 watts a t 4 amperes. L is a shutter located about 10 cm. from 2 and immediately in front of the 1-cm. Uviol glass absorption cell, P. F is a Corning No. 597 filter, 5 mm. thick. 'The photocell, located immediately behind the filter, is a sodium vacuum photocell manufactured by R. C. Burt of Pasadena, ,Calif. The amplifying tube shown is a 1B4P used as a triode, with the screen as plate, and has the following characteristics: plate current, 100 microamperes; grid current, 5 X 10-11 .ampere; transconductance, 200 micromhos. The current sensitivity of the galvanometer is 1 microampere per mm. R!, the balancing resistor, is made up of three small radio potentiometers of 20,000, 1000, and 100 ohms in series. The internal resistance drop across R is balanced by M ,a Leeds & Northrup student's potentiometer with 12 volts' working voltage. Switch S is connected by a cord to shutter L,so that opening S closes the shutter and closing S opens the shutter. C is a 0.005mf. condenser which by-passes the alternating current components of the photocell current, owing to the modulation of the light source in the frequency of the power supply. The 200megohm resistor, the photocell, the amplifying tube, and the condenser are enclosed in a shielded earthed box.

Operation With the lamp lit and switch S open, the galvanometer is brought to zero by adjusting resistor R1. S is then closed, which opens shutter L, and M is adjusted until the galvanometer again reads zero. Again opening S, the galvanometer reading is checked. If the galvanometer does not return to zero, because

el represents the uncertainty introduced by the current measrepresents the uncertainty introduced by the uring system. variations in the light source. el and the second term of el are Z both inversely proportional to log 2. Z The first term of el has a minimum value a t the point

(2.30?:ol

az Z

XI)

'

=o

which gives l o g 2 = 0.435. It increases as we go from this Z point in either direction. By assigning values of AI and 410appropriate to our instrument, we calculate by means of Equations 1 and 2 the fractional I variation to be expected in log 2 a t different values of the I latter. The uncertainty in the slide wire of the potentiometer is given by the makers as 0.5 millivolt, which is equivalent to 0.000357 ZO. The uncertainty introduced by the reading of the galva-

INDUSTRIAL AND ENGINEERING CHEMISTRY

376

TABLEI. I I

density, the individual determinations were made a t approximately the same potentiometer setting, so that the maximum uncertainty in the potentiometer slide wire would probably not be operative. Accordingly, one might expect the variations to be not much greater than e2,which is indeed true. I n all the other sets of determinations a t intermediate densities the variation is within the predicted value.

Z

VARIdTIONS IN L O G -"

Z

+

log 2

el

el

0.1 0 2 0 4 0 8 1.2 1 5

0.0047 0 0027 0 001s 0,0019 0,0029 0.0045

0.0174

0.0221

00.0087 .0044 0.0022 0.0014 0.0012

0 . 0 01 61 24 0.0041 0.0043 0.0057

el

Vol. 13, No. 6

4

TABLE11. MEASUREMENTS OF U. S. P. REFERENCE OIL Date Concentration, R 1% cm.

Av. Av. extinction

6/12/40 0.1392 1.507 1.493 1.486 1.501 1.473 1.507 1.486 1.493 1.493 1.493 1.493 0.208

6/24/40 0.1315 1.491 1.491 1.491

.... .. ... ..... . ... ...

1.491 0.196

6/21/40 0.270 1.503 1.506 1.506 1.514 1.506 1.503 1.503 1.499

6/21/40 0.398 1.510 1.505 1.610 1.497 1.508 1.505 1.505 1.508

6/25/40 0.5203 1.495 1.498 1.492

... ... ... ...

...

*.. ...

... ... ...

1,505 0.406

1.5C6 0.600

1.495 0.779

...

nometer and the drift of the plate current will be taken to be 0.3 galvanometer division, equivalent t o 0.000125 lo. Summing up, we have AZ = 0.00476 l o . In order to estimate AZO,the output of the lamp was measured once a minute for a continuous period of 1 hour. The maximum change during any 1-minute interval was 0.4 per cent. Since the actual lapse of time between the measurement of Z and ZOis about 1 minute, we take AZO = 0.004 l o . With these values, Table I of fractional variations in log

Z o / l a t different values was calculated. From Table I we see that if e2 were zero we would have a theoretical precision of about +0.25 per cent in the density range 0.2 to 1.2. With e2 as calculated we have, from the last column, a precision of about b0.5 per cent in the density range 0.45 to 1.5. This is better precision than is generally regarded as necessary for vitamin A work. With a variation in the light source three times that which we have, and in the density range 0.45 to 1.5, the theoretical precision would be about * 1.5 per cent, a tolerable figure for this type of work. This justifies the statement made above that, by a proper restriction of the working density scale, we may get results of good precision on a single photocell instrument in spite of sizable variations in the intensity of the light source. The operating range 0.2 to 1.5, which gives an over-all theoretical precision of 1 1per cent, was selected.

Measurements of U. S. P. Reference Oil A series of 74 determinations was made on U. S. P. reference oil dissolved in 99 per cent isopropanol, covering an extinction range of 0.2 t o 1.5 and extending over a period of 18 days. The results are given in Table 11. Ten determinations were made a t an extinction of 0.208 on June 12. The extreme variation of these determinations is ~ 1 . per 1 cent, while the theoretical precision a t that point, from Table I, is also equal to b 1 . 1 per cent. Three determinations made on the following day a t approximately the same extinction also fell within these limits. Tine determinations made on June 24 and 25 a t an extinction of 1.43 show a variation of 10.20 per cent and ten results on June 7 show a variation of b0.14 per cent. These results, a t the two extremes of the density range, are of particular interest because a t the extreme low density e2 dominates el by about 3 to 1, whereas a t the extreme upper density el dominates e2 by about 4 to 1. At the lower density range the agreement with prediction is extremely good. At the extreme upper range the variation is much less than the calculated value. This may be readily explained as follows: I n each of the two sets of determinations made a t the high

6/ 24/ 40 0.526 1.500 1.502 1.500 1.498 1.495 1.495 1.495 1.498

6/20/40 0.673 1.495 1.492 1.494 1.492 1.495 1.491 1.495 1.492

6/25/40 0.796 1.478 1.483 1.475

...

... ...

6/17/40 0.8835 1.486 1.489 1.489 1.489

... ... ...

6/24/40 0.977 1.457 1.456 1.456 1.461 1.458 1.455

6/25/40 0.9826 1.458 1,455 1.456

... ... ... ...

... ... ... ...

...

...

.. .. ..

...

... ... ...

...

1.498 0.788

1.493 1.005

1.479 1.177

1.488 1.312

1.457 1.423

1.456 1.430

...

**.

...

...

...

6/7/40 0,982 1.483 1.485 1.485 1.486 1.485 1.487 1.485 1.487 1.483 1.487 1.485 1.459

Although the reproducibility shown in each set of determinations is within the theoretical value, surprisingly large variations are evident between the averages of the different sets made on different days a t approximately the same extinction-for example, although the determinations made a t the very low extinction on June 12 and 24 are in very good agreement, the averages of determinations made a t the very high density on June 7 and 24-25 differ by 2 per cent. This variation is inconsistent with the agreement within the individual groups. A variation in the response of the photocell a t different times would seem to be the only logical explanation. Such a variation would be more noticeable the higher the density. I n view of this i t was decided to restrict the operating density range to 0.2 to 1.2, which is much greater than the operating range of the conventional spectrophotometer. The complete variation of all the determinations on U. S. P. reference oil within this density range is +1.4 per cent, which compares well with the theoretical value of *l.l per cent. The average E value of the U. S. P. reference oil within the selected density range was 1.495. With the Bausch & Lamb spectrophotometer the author has obtained on U. S. P. reference oil an average E value of 1.555 on the oil and 1.490 on its unsaponifiable, and is using a conversion factor of 2000 for this instrument. I n order to render results obtained on the new instrument comparable with those obtained on the spectrophotometer, he adopted the factor 2080 for the former. Using this conversion factor, determinations were made on 25 miscellaneous samples of fish liver oils and concentrates ranging in potency from 1130 to 326,000 units of vitamin A per gram, both on the new instrument and on the Bausch & Lomb spectrophotometer. The maximum difference between the two instruments on any of these samples was 3 per cent. This is consistent with the accuracy of * 1.4 per cent for the new instrument and *1.5 per cent for the spectrophotometer, which is the precision assigned to this instrument as the result of a large number of determinations.

Literature Cited (1) Bills and Wallenmeyer, J. B i d . Chem., 123,X I (1938). (2) Coward, Dyer, Morton, and Gaddum, Biochem. J . , 25, 1102 (1931); 26, 1593 (1932). (3) Drummond and Morton, Ibid., 25, 785 (1929). (4) Edisbury, Gillam, Heilbron, and Morton, Ibid., 26, 1164 (1932). (5) McFarlan, Reddie and Merrill, IND.ENG.CHEM., Anal. Ed., 9, 324 (1 937). (6) Morton and Heilbron, Nature, 122, 10 (1928); Biochem. J., 22, 987 (1928). (7) Parker and Oser, IND.ENG.CHEM.,Anal. Ed., 13,260 (1941).