428 standardization. Heating the solution to 65 O during the titration caused only a slight improvement. Therefore, the standardization against potassium dichromate was carried out by adding a slight excess of ferrous ion to the dichromate solution in 1 N sulfuric acid, and titrating the resulting ferric ion with the chromous solution. The titration of ferric ion is an excellent one, steady potentials are quickly established, and the titration curve a t the equivalence point is symmetrical, so that the end point can be determined accurately from the maximum value of A E / AV. A typical titration curve is shown as curve 2 in Figure 3. The equivalence point potential is 0.00 volt versus the saturated calomel electrode ($0.24 volt against the standard hydrogen electrode). The ferrous salt which is used must, of course, be free from ferric ion. The authors used Mallinckrodt reagent quality ferrous ammonium sulfate hexahydrate, which had been very carefully assayed in connection with another investigation and found t o have a purity factor of 99.90%. A 0.2200-gram portion of pure, dried potassium dichromate was dissolved in approximately 100 cc. of 1 N sulfuric acid in the titration vessel, and, after the flow of carbon dioxide was started, 2 grams of ferrous ammonium sulfate hexahydrate were added. The titration was made after carbon dioxide had been passed through the solution for 30 minutes. I n two titrations, 44.81 and 44.82 cc. of the chromous sulfate solution were required, corresponding to an apparent normality for the chromous sulfate solution of 0.1001, which agrees very well with the theoretical 0.1000.
Equally satisfactory results were obtained in the preparation of a 0.02 M chromous sulfate solution. LITERATURE CITED Asmanow, A, Z . anorg. allgem. Chern., 160, 209 (1927). Brennecke, E., in W. Bottger’s “Newer Methods of Volumetric Chemical Analysis,” pp. 131 et seq., New York, D. Van Nostrand Co., 1938. Buehrer, T. F., and Schupp, 0. E., Ind. Eng. Chem., 18, 121 (1926). Dimroth, O., and Frister, F., Ber., 55, 3693 (1922). Forbes, G. S., and Richter, H. W., J. Am. Chem. Soc., 34, 1140 (1917). Grube, G., and Breitinger, G., Z . Elektrochem., 33, 112 (1927). Grube, G., and Schlecht, L., Ibid., 32, 178 (1926). Irving, G. W., and Smith, N. R., IKD.EKG.CHEM.,ANAL.ED.,6, 480 (1934). Jablceynski, K., 2. phusik. Chem., 64, 748 (1908).
Kolthoff, I. M., and Furman, N. H., “Potentiometric Titrations,” pp. 380 et seq., New York, John TViley and Sons, 1931. Lingane, J. J., ISD.ENG.CHEW,AKAL.ED.,17, 640 (1945). RienRcker, G., “ S e u e potentiometrische Titrationsmethoden Bur Bestimmung von Schwermetallen,” dissertation, Munich, 1926.
Stone, H. W., and Beeson, C., IND. ENG.CHEM.,ANAL.ED.,8. 188 i l 9 3- 6-, ) -. ( 1 4 ) Thornton, W. M., and Sadusk, J. F., Ibid., 4 , 240 (1932). (15) Thornton, W.hl., and Wood, A. E., Ind. Eng. Chem., 19, 160 (1927). (16) Traube, W., Burmeister, E., and Stahn, R., 2. anorg. allgem. Chem., 147, 50 (1925). (17) Zintl, E., and Rienlicker, G., Ibid., 161,374 (1927). (18) Zintl, E., and Zaimis, P., Z . angew. Chem., 40, 1286 (1927). RECEIVED April 26, 1947. \ - -
Apparatus for Measurement of Vapor Pressure R. R. LEGAULT, BENJAMIN MAKOWER,
AND W. F. TALBURT Western Regional Research Laboratory, Albany, Calif.
A compact, portable glass apparatus has been designed for the determination of aqueous vapor pressure of dehydrated agricultural products. The apparatus includes a mercury manometer of the Dubrovin type, which has a sensitivity about seven times that of an ordinary U-tube mercury manometer. The apparatus will be found useful for vapor pressure measurements in general.
N STUDIES on food dehydration and related problems, equilibrium vapor pressure measurements are frequently required. Makower and Myers ( 4 ) proposed that aqueous vapor pressure rather than per cent moisture content be used as the index of degree of dehydration, and described a manometric apparatus for such vapor-pressure measurements. A slightly modified form of this apparatus has been described by Fischbach ( 2 ) . I n both units, the manometric fluid is an oil of low vapor pressure. The first-mentioned apparatus has proved entirely satisfactory, both as to accuracy and precision, provided the precautions outlined by RIakower and Myers (4)are folloxed. Hoxever, it is inconvenient for routine measurements because of its fragility and lack of compactness, and because of the necessity of continuous pumping on the reference arm of the oil manometer. A device described by Vincent and Bristol (6) employs a U-tube mercury manometer and, in consequence, has not the sensitivity required for precise work. These disadvantages have been eliminated in a new design, incorporating a mercury manometer of the Dubrovin type (1) that has a sensitivity about seven times as great as a Utube mercury manometer. The new apparatus, shown in Figure 1, is rugged, compact, easily portable, and identical in principle to the previously described unit (4). It consists essentially of a sample flask, A , connected through the trap, T,with the modified Dubrovin manometer, M . The manometer rests on the ring seal, V ,at the bottom of tube 2.
OPERATIOS OF THE GAGE
The procedure for making a measurement is as follows: Flask A containing the material-for example, ground dehydrated vegetable-the vapor pressure of which is t,o be det’ermined, is attached t,o the apparatus by means of the ground joint and immersed in a constant-temperature bath. The flask containing the sample and the rest of the apparatus are evacuated through t’he open stopcocks, D and C, to less than 0.1 mm. of mercury pressure lvith an oil pump. During the evacuation, which requires about 2 minutes, trap T is kept, cold by means of a solid carbon dioxidealcohol mixture contained in a Dewar flask. Stopcock D is then closed, the cold bath is removed from trap T , and the small amount of ice collected is evaporated by n-arming the trap slightly above room temperature. Approximately 0.5 hour is allon-ed for temperature and pressure equilibrium to become established, as indicated by a constant gage reading. If the evacuation is not thorough, or if the material in the flask releases adsorbed gases or volatile decomposition products during the measurement, the pressure reading will not represent JTater vapor alone. The pressure of noncondensable gases (air or carbon dioxide) may be easily determined by closing stopcock C to isolate the sample, then freezing out the water vapor in the gage by immersing trap I’ in a solid carbon dioxide-alcohol bath. The aqueous vapor pressure may then be taken as the difference between the first and second pressure readings. When the apparent pressure of noncondensable gases exceeds 0.5 mm. of mercury, the rate of diffusion of water vapor into the cold trap will be greatly retarded. To avoid errors in measuring the pressure of noncondensable gases, it is best in those cases to re-evacuate the whole system in the manner previously described. When evacuation is applied to samples containing finely ground materials, the fine powder is frequently sprayed from the sample bottle into the trap and the
V O L U M E 20, NO. 5, M A Y 1 9 4 8
429 cury. Since a change of 0.3 mm. in the position of the float could readily be perceived with the eye, the sensitivity of the gage was of the order of 0.04 mm. of mercury. The construction of an all-glass manometer requires attention to details, some of which have not been previously described. Float F must be kept in a vertical position and must be free t o move up and down with minimum friction. To guide the tube, three glass prongs were affixed to the top of the float and a constriction was made in tube M a t point H . Guide prongs were not used near the bottom of the float because of excessive friction arising from glass-to-glass contact under a mercury surface. For the same reason the mercury level was maintained below H in M . The choice of the dimensions of the float was in some measure governed by observations made during calibration of earlier models. Some gages exhibited a phenomenon of “hysteresis”-that is the position of the float was different (lower) xhen approached from the high-pressure side than when the same pressure was attained from the lovi-pressure side. This effect was reduced by the use of float tubes of large diameter and by maintaining a \Tide annular space between the float and M (at the point where the float enters the mercury). With a float having a 5-mm. inside diameter, the maximum hysteresis effect was equivalent to 1.0 mm. of mercury; with a larger float, 10 mm. in diameter, it was reduced to a maximum of 0.5 mm. Hysteresis was practically eliminated (less than 0.05 mm. of mercury) by coating both walls of the float tube and the inside wall of M with a thin layer of colloidal graphite, applied by wetting the meticulously cleaned walls with a dilute water suspension (approximately 1 to 100) of the colloidal graphite, draining, and allowing the adhering fdm to dry on the glass surface. The function of the graphite is not clearly understood. I t is not known whether it functions by virtue of its electrical conductivity (eliminating any static charges) or whether it has the effect of equalizing the advancing and receding contact angles of mercury on glass. I n assembling the Dubrovin manometer, it is important that F be thoroughly evacuated. Any air that is left, or that may accumulate in the float when the gage is in use, will be compressed when the float falls, and consequently the sensitivity of the gage will decrease with increasing applied pressure. To conduct the evacuation, a special all-glass adapter was constructed, as shown in Figure 2.
Figure 1. Vapor Pressure Apparatus (Left), Dubrovin Manometer (Right) KO scale
manometer. T o prevent this occurrence, it is convenient to place a plug of glass wool in the tube immediately above the joint that connects the sample bottle to the gage. CONSTRUCTION AND ASSEMBLY O F GAGE
Detailed construction of the Dubrovin manometer is shown in Figure 1. I t consist,sof an open tube, X,containing mercury and an evacuated tubular float, P , which is free to move up and down with variation in pressure. The position of the top of the float is the indes of pressure. This position is read on an arbitrary scale engraved on tube M, and the reading is interpreted in pressure units from a calibration described below. A convenient waj- of securing a scale is by the use of the graduated part of either a buret or a pipet for the top portion of AM. A complete description of the t,heory of the gage has been given by Gerniann and Gagos ( S ) , who prefer metal floats because of their mechanical strength. The present authors found it expedient to use glass floats, inasmuch as metal ones of the desired characteristics were not readily available. Furthermore, glass float,s were found to be very st,urdy and are simple to construct. T o avoid breakage of the float when the gage is brought to atmospheric pressure, the reservoir tube, M , is made long enough to allow for about, 2-em. clearance between the base of M and the bottom of F when the latter is completely full of mercury. Most of the authors’ manometers were designed to cover a range from zero to about 30 mm. of mercury; in a few cases the range n-as estcnded to about 60 mm. by increasing the length of the float. Float F was constructed from soft-glass tubing of fairly uniform diameter (obtained from the IGmble Glass Co.); its approximate dimensions Tyere: length, 33 cm.; outside diameter, 10 mni. ; wall thickness, 0.3 mm. or less. The float moved a distance of about 7 mm. for a change in pressure of 1 mm. of mer-
SIDE V I E W OF ADAPTER A T B
Figure 2. Adapter for Evacuation of Manometer
The reservoir tube, M , was attached t o the adapter by means of ground joint f. At the same time the float, F,was kept out of the mercury by supporting it, under the prongs, with a glass rod attached eccentrically to ground-glass stopper I . The adapter n-as attached to a vacuum system (manifold shown in Figure 3) through its joint, a. The whole manometer was then rotated to a nearly horizontal position around the ground-glass swivel joint,
430 Table I. Pressure, Mm. Hg 4.58 12.94 17.86
Precision of Gage at Various Pressures s o : of Dubrovin Gage Readings0
10 10 5
Standard Deviation from Average of Pressure Valuee, RI m. Hg t0.03 * O . 04 t0.01
Readings made alternately from high- and low-pressure sides.
L, in order to boirthe mercury under vacuum without excessive bumping. .4fter thorough evacuation, boiling, and cooling, the float was released by turning g and allowed to slide into the mercury as the manometer was slowly rotated to a vertical position.
CALIBRATION OF THE GAGES
The gages were calibrated against a U-shaped oil manometer. For this purpose one or more units similar to that illustrated in ~i~~~ 1 were attached through joint E to a glass manifold as indicated in Figure 3. A low-vapor-pressure oil of measured density (Octoil, obtained from Distillation Products, Inc.) was used as the reference manometric fluid. Preliminary to the calibration, the system (Figure 3) was pumped down overnight to a pressure of about s microns, measured with a gage. Corresponding readings of the Dubrovin gages were considered the zero pressure points. Stopcock d (Figure 3) was closed and pumping was continued on the reference limb of the oil manometer during the whole calibration process. Comparative readings of the Dubrovin gages and of the oil manometer were made following each successive addition of a small increment of gas to the system through stopcock D (Figure I), until the pressure range of the gages was covered. Subsequently, another series of readings was obtained through progressive decrease of the pressure down to the zero point with an auxiliary pump connected to D (Figure 1). In most cases the empirical calibration curve obtained in that manner followed a linear relationship within the limit of error of the calibration. Occasionally, a deviation from linearity appeared a t the high-pressure end of the calibration curve. I t could usually be shown that this deviation was caused by improper evacuation of the float tube. The calibration may be expressed in terms of the equation:
g=A+Bp where g is the gage reading, p is the pressure in millimeters of mercury, and A and B are constants. The results for one gage, calibrated with dry air, carbon dioxide, and water vapor, are given as an illustrative example. From the reading errors of the oil manometer and of the Dubrovin gage (+0.04 mm. each) the maximum reading error of the calibration was estimated to be +0.08 mm. of mercury. The value found for slope B was 6.81 for the three gases. The magnification factor of the gage, 6.96 (the change in gage reading in millimeters per millimeter of mercury change in pressure) was obtained by multiplying B by 1.02, the factor for converting to millimeters the arbitrary unit intervals of the scale of the Dubrovin gage. The values of intercept A (calculated reading of the gage a t zero pressure) were 16.5, 16.9, and 17.1 for dry air, water vapor, and carbon dioxide, respectively. I t is not possible t o say whether the slight differences in the A values signify real differences in the behavior of the gage toward the three gases. Practically they are not important, as they would give rise to differences in calculated pressures of not more than *0.01 mm. of mercury if the average value A = 16.8 were used. VAPOR PRESSURE APPARATUS
ments a t O.OOo, 15.18", and 20.30" C. The summarized results in Table I show that the precision of the gage, as expressed by standard deviation, is kO.04 mm. of mercury or less. The gage readings a t known vapor pressures of water, referred to in Table I, constitute an independent calibrat'ion. The three points fall on a straight line whose intercept., 16.3, is in excellent agreement with the experimentally determined value, 16.5. The slope, 6.90, is about 1% higher than the value obtained from calibration against the oil manometer. The discrepancy is apparently due to errors inherent in the oil manometer used for the calibration. Although calibration against the the more accurate is known vapor pressure Of procedure, it has the disadvantage that it is conveniently a p plicable over a limited pressure range, up to the saturation pressure of water vapor, or about 20 mm. of mercury a t room temperature. For this reason, calibration against an oil manometer is more convenient for routine work. It is of interest to know how the calibration of the gage is affected by changes in ambient temperature of the room. An increase in temperature causes the level of the mercury to rise in the reservoir and raises the position of F with respect to the graduations on tube M (Figure 1). This rise is partially offset by the fact that F sinks deeper because of decreased density of the mercury. The net effect for a rise in temperature of 5' c. is an error of about 0.03 mm. of mercury in the pressure reading. Dimensional changes, with temperature, of the various glass units are negligibly small and have not been included in these calculations. However, in addition t o this correction, the preasure readings obtained with the Dubrovin gage are subject to the same temperature corrections as those for any mercury manom-
The assistance of Elisabeth ~ gratefully
iin making ~ the lcalibrations ~ is ~
LITERATURE CITED (1) Dubrovin, J., Instrurnents,,6, 194 (1933). (2) Fischbach, H., J . Assoc. Oficial Agr. Chem., 28, 186 (1946). (3) Germann, F. E. E., and Gagos, K. A., IND. ENG.CEEM.,ANAL. ED.,15,285 (1943). (4) Makower, B., and
Myers, S., Proc. Inst. Food Tech.,
TO PlRANl QAUGE
1943, 156. (5) Vincent, J. F., and
Bristol, K. E., IND. ENG.CHEM., ANAL. ED.,17,465 (1945). REcEIvED
(FIG 1 )
Reproducibility of vapor-pressure values obtained with the gage was determined by repeated measurements at each of three definite pressures. The pressures used were those of water vapor in equilibrium with liquid water. The water was contained in flask A (Figure 1) and was held within +0.02" C. during three different experi-
Manifold for Calibration of Gages
TO VACUUM PUMP