Application of Mohr-Westphal Balance to Rapid Calibration of Wide Range DensityGradient Columns F. M. RICHARDS AND T. E. THOMPSON University Laboratory of Physical Chemistry Related to Medicine and Public Health, Harvard Unicersity, Boston, Mass, Wire Immersion Effect. In the depign described, the buoyancy effect, W b , is very small-of the order of 0.005 mg. per cm. for a liquid of unit density. Thus the length of the immersed wire, h, need be measured only to the nearest centimeter, and the average density of immersing liquid, dl, known only approximately.
gradient column technique has been developed for AMODIFIED measurement of the densities of solids Various indirect (1).
methods of calibrating gradient columns have been used (3)or suggested (1)in order to eliminate the necessity of maintaining a large number of standard solutions. These methods require considerable apparatus and/or extensive standardization. To maintain the advantages of speed and simplicity inherent in the modified gradient tube procedure, the balance described below was developed. The Westphal balance, employing Archimedes’ principle, measures the change in weight of a solid object of known volume on immersion in an unknown liquid. I n a gradient column, the density is a function of the vertical position in the tube. The Westphal balance method may be applied to the gradient column, provided that the sinker be very small and symmetrical, so that the measured density can be assigned to a given level. DESIGN AND OPERATION
A schematic diagram of the apparatus is given in Figure 1 The basic instrument was a commercially available doublehook torsion balance of range 0 to 10 mg. with a sensitivity of 0.005 mg. (2). A 0.75-inch hole was drilled through the base directly under the weighing hook, and a vertical rack and pinion supporting a holder for a 10-ml. test tube were mounted under this hole. The suspension consisted of a solid glass sphere of 5 to 10-ml. volume sealed onto a straight length of 0.001-inch diameter tungsten wire. Operation. After the solid or liquid unknown had attained its equilibrium level in the column, the gradient tube w s placed in the holder. Its position was adjusted by means of the rack and inion until the equator of the sinker reached the level of the unnown substance. The change in weight of the sinker was then noted. From this the equatorial density was computed.
K
Figure 1. S c h e m a t i c Diagram of App a r a t u s for G r a d i e n t Tube Calibration a.
b. b’. e. d.
ACCURACY OF METHOD
The standard Westphal balance is designed so that the surface tension and wire immersion effects are virtually negligible. The volume of the sinker used in this modification is so small that these effects are of the order of f 0 . 0 5 grams per ml. and, therefore, cannot be neglected. Regions of abrupt density changes in the gradient column may be eliminated by careful preparation of the tube. Secondorder effects in the neighborhood of the sinker may then be neglected, and the system described by the following equations :
where dl = density of liquid at equator of glass sinker V G = volume of glass sinker, cu. mm. A W = observed change in weight, mg., on immersion of sinker T t 7 b = buoyant effect, mg., due to immersed wire W,t = surface tension effect on suspension wire, mg. V , = volume of suspension wire, ml. per cm. h = length of immersed suspension wire less length of immersed compensation wire d, = length-average density of liquid surrounding immersed wire c = circumference of wire, cm. y = surface tension of liquid, dynes per cm. e = wetting angle g = gravitational acceleration
e.
Double-hook torsion balance Tungsten suspension wire Tungsten oompensation wire Glass sinker Density-gradient tube Rack and pinion for tube position adjustment
Surface Tension Effect. The surface tension effect, W , t , is of the order of 0.2 to 0.6 mg., depending on the liquid involved. If the wetting angle, e, is either measured or assumed and y is known, W8v,tmay be calculated. The effect is more satisfactorily eliminated by compensation. A small length of wire, identical with that used in the suspension, is hung on the counterweight hook. This compensating wire passes through an interface of the same composition as that of the solution to be measured. Identical interfaces for both suspension and compensating !vires were assured by immersing the compensating \Tire in a pure liquid (the pure light column component for organic columns) and placing a small layer of the same liquid on top of the gradient tube just before measurement. The wetting of the tungsten wire appeared to be uniform for the organic liquids used. For an air-water interface, the wetting was very erratic. This difficulty was overcome by layering about 0.5 em. of ethyl or n-propyl alcohol over the nater surface. The alcohol-Kater interface formed disappeared almost immediately, xhile the wetting a t the alcohol-air interface again appeared uniform. Diffusion of the alcohol into the deeper portions of the aqueous solution was too slow to be troublesome. Displacement Effect. With the design indicated, a displacement of the density gradient of about 0.04 mm. occurs upon introduction of the sinker into the region of the sample. However, 1052
1053
V O L U M E 2 4 , NO. 6, J U N E 1 9 5 2 Table I. Densit from StanJard Westphal Balance. G./Ml.
Results on Test Solutions War, Mg.
Calod. for e =0
Observed
Measured Densities, G./Ml. BY BY correction compensation
Benzene-Bromobenzene Mixtures 1.4832 1.3650 1,2478 1.1213 1.0160 0.8760
0.28 0.27 0.26 0.24 0.23 0.22
0.9977
0.54
0.30 0.27 0.26 0.25 0.23 0.20
1.484 1.363 1.246 1.117 1.016 0.872
1.487 1,363 1,246 1.119 1.015 0.878
Water without Alcohol Layer 0 . 5 0 f 0 . 2 0 1.OOiO.05 0.95ztO.05
Water-Phosphoric Acid Mixtures with n-Propyl Alcohol Layer 0.9977 1,1574 1,3027 1.4367 1,5536
0.18 0.18 0.18 0.18 0.18
0.19 0.20 0.20 0.20 0.18
0.995 1.155 1.302 1.434 1.557
0.997 1.158 1,303 1,438 1.557
if there is no mutual mechanical interference between sinker and sample, no error can result; for to obtain an accurate measurement, it is only necessary to bring the sinker to rest a t the final level of the specimen after displacement of the gradient liquid. Actually, since in this application positions were estimated to only about 0.3 mm., the effect was unimportant. Results on Test Solutions. Series of organic and aqueous mixtures, calibrated on a standard Westphal balance (Table I, column l), were used to test the method. All measurements on the torsion balance were corrected for the wire immersion effect. Surface tension effects were calculated from the known surface tensions of the pure liquids assuming a zero wetting angle (Table I, column 2). The observed effects are recorded as the weight change occurring on immersion of the compensating wire (Table
I, column 3). The errors due t o irregular wetting of the wires amount to h0.02 mg. The volume, Vo, of the sinker used was 6.14 cu. mm., as measured by immersion in a liquid of known density. Thus, the computed density is subject to an error of about 10.003 gram per ml. The densities, measured without compensation, were corrected for the calculated surface tension error (Table I, column 4). The densities, measured by compensation, recorded in the last column, are somewhat more accurate than those determined by correction. The deviation from the true value is not more than 10.003 gram per ml. by conipensation and ~k0.004gram per ml. by correction. CONCLUSIONS
The method described for the calibration of the wide range density-gradient column is sufficiently accurate for this purpose. I t would be possible to improve the accuracy by a slight increase in the size of the sinker and a decrease in the diameter of the suspension wire, and by paying scrupulous attention to the cleanliness of the wire and liquid surfaces. The method is not recommended if accuracies of better than fO.OO1 gram per ml. are required. It has proved to be very rapid and convenient for routine column calibration. LITERATURE CITED
(1) Low,B. W.,and Richards,F. M.,J.-4m.Chem. S O C . , 1660(1952). ~~, (2) Roller-Smith, Inc., Bethlehem, Pa., Catalog KO.2405D. (3) Tessler, S., Woodberry, K. T., and Mark, H., J. Polymer Sci., 1, 437 (1946). RECEIVED for review December 4, 1951. Accepted March 3, 1952. Woik eupported by the Eugene Higgim Trust, by grants from the Rockefeller Foundation and the National Institutes of Health, by contributions from industry, and b y funds of Harvard University, and carried out during the tenure by F. M. Richards of predoctoral research fellowship in the biological sciences, Atomic Energy Commission.
Determination of Small Amounts of 1,2=Propylene Glycol in Ethylene Glycol W. A . CANNON
AND
L. C . JACKSON, Wyandotte Chemicals Corp., Wyandotte, Mich.
for rapid, routine determination of small quantiA METHOD ties of 1,2-propylene glycol in ethylene glycol was desired for control analysis. The method of Reinke and Luce ( 5 ) was investigated, but was found unreliable for extremely small concentrations of I,%propylene glycol. Moreover, the experimental requirements for accurate work are somewhat critical and unsuited for rapid, routine analysis. The method of Hoepe and Treadwell ( 2 ) for determination of mixtures of glycols involves separate determinations of total aldehydes and formaldehyde and is obviously unsuitable for the purpose in mind. A more recent paper by Nogare, Norris, and Mitchell ( 4 ) has described a procedure for determination of small amounts of l,>propylene glycol based upon the reaction of acetaldehyde, derived from oxidation of the glycol, with hypoiodate to form iodoform which is determined spectrophotometrically a t 347 mp. The details of this procedure appeared in print after the experimental work described in this paper was completed. Karshowsky and Elving (6) have published a method for simultaneous determination of ethylene glycol and 1,2-propyIene glycol which involves an oxidation of the mixture of glycols with periodic acid, removal of the aldehydes by distillation, and polarographic determination of the formaldehyde and acetaldehyde in the distillate. Because the reduction potential of formaldehyde is more positive than acetaldehyde, there is a lower limitation on the concentration of 1,2-propylene glycol that can be detected in the presence of appreciable amounts of ethylene glycol by this method. When the concentration of formaldehyde in the distillate exceeds approximately ten times the concentration of acetaldehyde, the determination of l,2-pro-
pylene glycol is inaccurate. Compensation techniques for enhancing the acetaldehyde wave can be used to good advantage, but even so, the authors have found it impossible to detect less than 1 part of 1,2-propyIene glycol in 50 parts of ethylene glycol by the Warshowsky and Elving procedure. By a suitable modification of this method, it has been found possible to make fairly accurate determinations of as little as 1 part of 1,2-propylene glycol in 500 parts of ethylene glycol, EXPERIMENTAL
The work of Rhitnack and Moshier ( 7 ) on the polarographic determination of formaldehyde has indicated that the diffusion current of formaldehyde is markedly affected by pH, being reduced a t lower p H values. On the other hand, the diffusion current of acetaldehyde is little affected over the p H range 6.8 to 12.7 (1). By carrying out the polarographic analysis in a supporting electrolyte near a p H of 7, the interference of the large excess of formaldehyde is to a considerable extent minimized. In addition, partial separation of acetaldehyde and formaldehyde can be made during the distillation step. Essentially all of the acetaldehyde is contained in the first portion of the distillate and only a part of the formaldehyde.
A 250-ml. aqueous solution containing 250 mg. of ethylene glycol, 2.5 mg. of 1,2-propylene glycol, and 1.5 grams of periodic acid was distilled, in the distillation apparatus described below, and successive fractions of 25 ml. were collected. A polarographic examination, by the method given herein, showed that essentially complete transfer of the acetaldehyde was obtained in the first 25-ml. fraction. KO detectable amount of acetalde-
