INDUSTRIAL A N D ENGINEERING CHEMISTRY
October, 1923
I n ordinary beverages like near-beer the specific gravity may be taken as 1.025, but if the beverage is somewhat thick it is best to determine it and apply correction accordingly. The specific gravity, however, can generally be neglected in the calculation, as the error incurred is quite inappreciable.
Sample 1
The results of this method have been repeatedly checked against those obtained by the Crampton and Trescot method, and the checks are reasonably within the limits of experimental error. The following table shows eight analyses with a comparison of the gravimetric and the rapid methods:
-PER CENT-CARBONDIOXIDGravimetric Rapid Method Method Difference
2
3
0.386
0,385
0.410
0.402
4
0.399 0.300
0.395 0.306
8
0.450 0.376 0.350
0.444 0.366 0.358
5 6 7
RESULTS AND COMPARISON WITH GRAVIMETRIC METHOD
1075
0.420
0.412
0.001 0.008 0.004
0.006
0.008 0.006 0.010 0.008
The f i s t six samples were near-beer and the last two popular brands of soda water. The same method, with slight modifications, is found in the Official Methods of the Association of Official Agricultural Chemists4 for determining carbon dioxide in mineral waters. 4
Assoc. Official Agr. Chem., Methods, 1920, p. 28.
Applications of Vapor Pressure Measurements' By H.S. Davis and Mary D. Davis ARTHURD. LITTLE,INC., CAMBRIDQE, MASS.
N PREVIOUS publica-
I
tions* the writers have described a differential vapor pressure apparatus. Further experimentation and the introduction of the device into commercial work have led to improvements in its design along the lines of durability and ease of manipulation. APPARATUS
flask. If any trouble is encountered in breaking off the capillary ends, it may be entirely eliminated by scratching them, before filling, with a file a t the desired points. FILLING THE CONTAINERS-In previous descriptions detailed instructions have been given for filling small containers by alternate heating and cooling or by alternate lowering and raising of the air pressure. However, these methods are open to the following objections:
A convenient vapor pressure apparatus consists of two similar glass flasks connected to a manometer tube and means whereby sealed glass containers full of liquids may be broken inside the flasks. Full details for the manipulation of such a deoice are given. In the recovery of light oils from coke-ooen gas by oil scrubbing, the light oil vapors are to a1certain extent absorbed as a whole. and selectioe absorption plays a smaller part than might be expected. This fact may be partly due to di8erences in the rates of diffusion of the oapors, the lighter ones tending to outstrip the heavier in their race to the absorbent oil. I t is possible io check up the efficiency of a light-oil recovery plant by measuring the total parfial.pressure of the light oils in the gas at various points in the system, and also their tension from the absorbent oil.
The apparatus (Fig. 1) consists of two similar glass flasks, each with a groundglass stopper and a capillary side tube, with stopcocks, connected to a manometer tube. Each flask is provided with a metal device whereby a small, sealed, glass container may be broken inside it. The following improvements are included in the design shown in the figure: MANOMETER TuBE-This was formerly connected to the flask to form one rigid piece. The writers have found, however, that joints of rubber tubing a t the points shown in the figure do much to relieve strain on the apparatus, and also permit manometer tubes of different lengths to be used on the same apparatus. CONTAINERS FOR LIQuIDs-Formerly the liquid whose vapor pressure was being measured was put into small glass bulbs which were attached to movable rods so that when desired the bulb could be broken against the bottom of the flask. The containers now used are made from pieces of thin-walled tubing of 1to 1.5 cm. outside diameter. The tubing is drawn to a capillary a t one end and sealed off 1.5 cm. from the bulb. At the other end it is drawn to a somewhat larger capillary, which is cut off about 2 cm. from the bulb and left open for filling. The metal breaking device has a holder to support the container in a vertical position inside the flask. When it is desired to release the liquid into the flask, the lower part of the breaking device is grasped with the left hand while the upper part is twisted with the right. This movement, acting on a simple contrivance, breaks off bohh the slender ends from the glass container and allows the liquid to flow evenly into the 1 Received February 12, 1923. 1 THIS JOURNAL, 10, 707 (1918); Advisory Council for ScientificTand Industrial Research of Canada, Report 2 (1918),
1-In the case of high-boiling liquids it is necessary to heat them so hot, in order to obtain a vapor pressure sufficient to drive the air from the bulb, that decomposition may begin. On the other hand, their vapor tensions at ordinary temperatures are so low that filling the bulb by changes of pressure is a slow process. 2-In dealing with a liquid mixture there is a tendency t o fractional distillation of the liquid in the bulb. 3-Occasionally a bulb bursts through a too rapid change of pressure. Sometimes the loss of material in this way is a serious matter.
During recent investigations such methods of filling have been completely discarded and instead a capillary pipet, made by pulling out a piece of narrow tubing at one end, has been used. An outside diameter of 0.5 mm. for the capillary will enable ordinary liquids to be drawn with ease through a piece several centimeters long, and the inside diameter of the stem of the container through which it must pass need be very little larger. Such pipets made from glass tubing of 3-mm. inside diameter, and provided with a graduation mark, have been a great convenience for introducing accurate amounts of liquids into small containers of various kinds.3 SEALING THE CONTAINERS-TOseal successfully the open stem of the container, filled with a volatile liquid, it is first heated evenly in a small flame, about 1 cm. from the bulb, and drawn out to a minute capillary. The container is now put aside and allowed to cool completely, after which the capillary 3 A capillary pipet of this kind should be of great use in many physicochemical operations where it is necessary to introduce measured quantities of liquids into small glass containers which are to be sealed up afterwards in the blowpipe. For instance, measurements of vapor density in the Victor Meyer apparatus, of compressibility by certain methods, and of heats of combustion in the Bertholet bomb are made OQ liquids preferably after enclosing them in such containers.
1076
INDUSTRIAL A N D ENGI NEERING CHEMISTRY
end may be sealed off without disturbing the air pressure inside. Except in the case of very volatile liquids, it is not necessary or even desirable to seal off this final capillary a t all, in which case the breaking device may be arranged to break off only the lower stem of the container. QUAXTITY OF LIQUIDI N CONTAINER-when dealing with a homogeneous liquid such as pure benzene, the actual vapor pressure developed in the apparatus is independent of the quantity of liquid used; but in the case of a mixture of two or more liquids, the vapor pressure developed is seriously affected by this factor. This is due to the fact that, the partial evaporation into the flask, which takes place after the bulb is broken, somewhat changes the composition of the residual liquid. The part FIG.1 which evaporates is richer in volatile constituents than the remainder, so that the final vapor tension of the residual liquid is somewhat lower than the true vapor tension of the original liquid in the bulb. This error is present in all measurements of vapor pressure by the static method. When the composition of the vapor is known, the necessary correction can easily be calculated. It will be found convenient, for all measurements of the vapor tension of liquids in this apparatus, to have the volunie of liquid used one one-hundredth that of the volume of the flask. PUTTING I N THE STOPPERS-The sealed containers are placed in the supports on the lower ends of the rods through the stoppers. The stoppers are slightly greased, put in place, and secured by small, stout rubber bands over the lugs. It is not necessary to wire in the stoppers. WATERBATH-The flasks of the apparatus are immersed in a liquid bath until half their necks are covered, while the manometer tube remains outside. CLEANIXG THE APPARATUS-After the determination is completed the stopcock in the manometer is closed, the other stopcocks are opened, and the stoppers removed. The flasks are cleaned by wiping all grease from the insides of the necks and sucking out any remaining liquid through a small glass tube connected with the suction pump. They are then rinsed with a little benzene or alcohol, which is afterwards sucked out. Between the apparatus and the suction pump there should be a small safety bottle to catch fragments of glass. Each of the flasks is then throughly cleaned of vapor by drawing air through it for at least 15 minutes in the following manner: A cork stopper, through which has been fitted a glass tube reaching nearly to the bottom, is placed in the flask. Suction is now applied to the stopcock and air is drawn in through the glass tube to the bottom of the flask and out at the stopcock.
(c)
(d)
(e)
cf)
(g)
(12)
Vol. 15, No. 10
Put the apparatus in the water bath. Open both stopcocks to the air in order to bring the pressures in both flasks to atmospheric, then open the manometer stopcock and close the other two. Read the manometer levels, estimating with the eye t o tenths of millimeters, at 5-minute intervals until the readings are constant. Break the container. Read the manometer levels a t intervals until they are constant. Close the manometer stopcock, open the other stopcocks, remove the stoppers, and clean the apparatus. Read the temperature of the bath and the barometer pressure,
The difference between the final and initial readings.gives the vapor tension from the liquid in terms of the manometer liquid. If the absolute value of this pressure is desired, certain obvious corrections must be applied. The largest of these arises from changes of air pressure in the two flasks caused by the movement of the manometer liquid. This correction is of special importance when a light liquid is employed and, in some cases, may amount to 25 per cent of the total reading. The following measurements carried out on the tension of light oil from a sample of stripped oil from a well-operated American plant mill illustrate the method: Time Min. 0 (Put apparatus in bath) 5
10 16
20
’
25 Pressure developed Correction for changes in air pressures in flasks Total pressure Pressure in millimeters of mercury
Pressure in Millimeters of Oil (Sp. Gr 0.56) No. 1 No: 2 3.1 1.5 1.7 1.2 0.7 0.6 0.4 0.4 0.2 0.1
3 9 3.9 0.9 4.5 0.30
4 2 4 2 0.9 5.1 0.32
DIFFUSIONOF VAPORSAND ITSRELATION TO ABBORPTION
PROCESSES The rates at which vapors are disseminated by true diffusion and by convection currents are very different, as the following striking experiment will illustrate: Fix firmly in one of the supports a container completely open and unrestricted at t h e top. Fill this with a volatile liquid t o within a few millimeters of the brim and leave it unsealed. Then lower the stopper gently into place in the flask. It will be found that the development of pressure is surprisingly slow. Now break the lower stem of the container, thus allowing the liquid to flow over the bottom of the flask. The usual rapid development of pressure, caused mainly by convection currents, will result.
It is well known that the evaporation of a liquid into a gas a t rest is determined principally by the rate of diffusion of its vapor through the gas which lies above it, and similar considerations must apply to the absorption of its vapor into a nonvolatile medium. It would reasonably follow that the rates a t which two vapors are simultaneously absorbed from a gas into another medium must be somewhat influenced by the difference in their rates of diffusion in the gas. This difference is greater than that between their Caution-It is of the utmost importance that the apparatus be kept clean from any liquid of the kind whose vapor pressure rates of diffusion separately in the gas, according to the is being determined. The stoppers must be removed from the curious law, first pointed out by Graham, that the moveapparatus as soon as the experiment is completed. If they are ment of the faster is accelerated and that of the slower left for long periods in the flask containing a volatile liquid, its retarded in simultaneous diffusion. vapor may penetrate the connecting tubes and rubber connecOf course, if equilibrium were reached a t each point in the tions, causing trouble. absorption system, Raoult’s law and the law of partial presSPECIFIC DIRECTIONS FOR MEASURING THE VAPORTENSION sures would hold. On the other hand, if equilibrium is not OF A LIQUID attained, there should be a tendency for the heavier and less ( a ) Fill a container with a measured quantity of liquid. diffusible vapor to be retained in the gas. It will be readily ( b ) Place it in the support attached to the right-hand stopper seen that such an effect would oppose selective absorption. and insert the stoppers in the flasks.
INDUSTRIAL Ah’D ENGINEERING CHEMISTRY
October, 1923
1077
the volumes of benzene and of the flasks in separate experiTESTING EFFICIEXCY OF BENZENE (LIGHTOIL)SCRUBBERS~
If selective absorption of the various light oil vapors took place to any great extent in an ordinary scrubber, then the relative proportions of the constituents of the light oil in the absorbent oil would vary a t different points, the higher boiling constituents predominating in the first part of the washing ~ y s t e m . ~ The selective absorption effect is much less than might be expected, and perhaps this can be explained by the diffusion effect pointed out above. But whatever the explanation, the fact remains that in an ordinary light oil scrubber the benzene and toluene, to a certain extent, are absorbed, not in inverse proportion to their volatility, but as a whole. It requires prolonged contact between the absorbing medium and the gas to effect a separation of the two vapors.6 For this reason vapor pressure measurements can be successfully applied to the control of Iight oil scrubbers in the following ways : 1-The relative efficiencies of the different units can be compared by measuring the tension of light oil from samples of wash oil taken from various points in the system. 2-The measurements in No. 1 can be checked by measuring the pressure of the residual light oil in the gas at these same points. 3-The efficiency of the stripping still can be checked by measurements of the tension of light oil from the debenzeneized oil. This is an important measurement because no amount of scrubbing will ever reduce the benzene content in the gas below a pressure equal to the tension of light oil from the debenzeneized oil.
The measurement of the vapor tension of light oil from an absorbent oil is a comparatively easy matter and has already been described. Comparative measurements of the total pressure of light oil in the gas at various points in the system are carried out in the following maimer: Into each flask of the vapor pressure apparatus is put a sealed container filled with an accurately measured quantity of liquid benzene. (It is recommended that the volume of benzene used be one one-hundredth of the volume of each flask.) One flask of the apparatus is then filled wiih the gas to be analyzed and the other is left full of air. When the two containers are broken simultaneously, the differential pressure which develops between the two flasks is proportional to the vapor pressure of light oil originally present in the gas. The theory of these measurements hm already been dealt with in previous publications.2 The writers failed, however, t o lay enough stress on the importance of measuring equal volumes of benzene into the containers in each experiment and of preserving the same ratio between 4
6
Cantello, Cas. Chem. Met., 6, 196 (1922). Spert , Gas Age, 41, 393 (1918)
ments. In 1917 tests were carried out with this apparatus on a light oil recovery plant operating under wax conditions on coke-oven gas in Canada. The absorption system comprised six large washing towers of the hurdle type in series, through which the gas and wash oil were passed in countercurrent contact. In 1918 further tests were carried out. During the intervening period centrifugal sprays had been introduced before each of the first three towers in a vain attempt to increase the recovery. The experienced operator of a modern benzene-recovery plant will a t once perceive that this was a very inefficient recovery unit, and indeed, in view of the results obtained from these tests, radical changes were planned in it which were only cut short by the armistice. A summary of the results from these tests is given in Table I. DISCUSSION OF RESULTS Through the greater part of the absorbing system, the vapor tension of light oil from the wash oil was nearly equal to the partial pressure of light oil in the gas at the corresponding point. Indeed, the wash oil in the third tower actually gave up benzene to the gas, a condition which was traced to its relatively higher temperature there. In spite of the superabundance of washing, a large part of the benzene passed through the towers unabsorbed. This condition was caused by the inefficiency of the stripping (Hirzel) still, which left a large percentage of light oil in the poor or debenzeneized oil. No amount of washing w ill ever compensate for inefficiency in the still. The quality of coke-oven gas delivered to this plant varied much in quality. On May 13 the superintendent was annoyed by the small output of light oil, whereas on the 17th the yield was excellent. As will be evident from the table, these discrepancies were c’aused, not by any change in the efficiency of the benzene-recovery unit, but by variations in the quality of the gas. GASOLINE AND SOLVENT RECOVERY Many of the principles of benzene recovery are embodied in processes for the recovery of gasoline from natural gas or still vapors. There is, however, this difference, that the benzene is rather sharply defined from the other constituents of cokeoven gas in chemical and physical properties, whereas the dividing lines between the constituents of natural gas are much less clearly marked, and the properties of the gasoline are greatly influenced by the methods of recovery. However, a study of the vapor pressures involved cannot fail to throw light on any system for gasoline or solvent recovery.
TABLE I-VAPORPRESSURE ~IEASUREMIXTS OK THE SCRUBBING TOWERS (SIX TOWSRS AND THREE SPRAYS, IN SERIES) OF 4. LIGHTOIL-RECOVERY PCANT LASTPOINTIN SCRUBBERS SAMPLE PASSED BY GAS September, 1917 May 6, 1918 May 13, 1918 May 14, 1918 May 17, 1918 May 18, 1918 ... ... Rich oil 6.1 Rich gas Entrance 8.4 6.1 1 0 . 3 (5.3) 5.4
......
...
Gas from Tower 1 Gas from Tower 2
1st spray Tower 1 (2nd spray)
... 6.2
Gas from Tower 3
Tower 2 (3rd spray)
4.6
Gas from Tower 4
Tower 3
5,O
from Tower 5
Tower 4
4.8
Gas from Tower 6
Tow-er 5
3.7
Poor gas
Tower 6
3.3
Gas
...
...
::!}6.8 5.7
3”::
;:E
13.5 13.5
;:: 13.3 ;:;
12.6
...
... ... ... ...
7.0 5.2
...
(4.3)
...
(4.2)
... ...
5.0
... ...
...
4.8
...
...
...
...
...
...
4.3
4.2
5,s (3.9) ...
)]:;
...
5.0
...
j2.3
... Poor oil From Hirzel still ... 3.0 I n the case of a gas, the figures represent the total partial pressures of the light oil vapors in millimeter of mercury; in the case of an oil they refer to vapor tensions of light oil in millimeter of mercury. The figures in parenthesis indicate partial pressures of pure benzene (CeHd, The table shows: 1--That the light oil content of the coke-oven gas varied from day to day. 2--That no absorption was taking place in Towers 3, 4, and 5. 3--.That about one-third of the light oil passed through the scrubbing system unabsorbed.
