Apparatus for the Fractional Distillation of Liquefied Gases ARTHURROSE The Pennsylvania State College, State College, Pa.
L
OW-temperature fractionating columns of numerous designs have been described by Frey and Yant (IO), Oberfell and Alden ( I 7 ) , Podbielniak (a@, Davis ( 2 ) , Bosschart ( I ) , and others (3,7,9, I d , 13,14,18,19, 21,23). These columns are all characterized by a partial condenser, and fractionation takes place in a closed system. McMillan (16) has recently patented an improved form of such apparatus. I n order to obtain a smooth boiling point curve, close and continuous attention must be given to the pressure inside the column and to the condenser temperature because of the effect of these factors on the boiling point observed. The successful operation of the columns requires considerable skill and experience, especially when the mixture being analyzed contains components boiling near one another. The apparatus described here is such that its operation can be easily mastered and it gives a boiling point curve that is smooth and independent of all reasonable variations in condenser temperature and heat input to the still. At the same time the apparatus is relatively inexpensive and simply constructed.
A packed glass fractionating column 310 mm. (12 inches) long and 6.3 mm. (0.25 inch) in inside diameter, fitted with a 25-cc. still and a condenser, is kept cold by placing it in a large Dewar tube 50 by 620 111111. (2 X 24 inches) in inside dimensions (Figure The condenser of the column is cooled by passing liquid air cold air through it. The cold air coming from the condenser is passed through the length of the Dewar tube 'and thus keeps
kk
the entire apparatus in an environment suitable for operation. The acked section is surrounded by a closely fitting but not actua7ly sealed outer jacket, which creates a dead air space and thus partially insulates the column from its environment. Since the column has no vacuum jacket, its construction involves only simple glass blowing. The large Dewar tube can be purchased complete at a cost only about three times that of a widemouthed quart-size Pyrex Dewar. Several different columns, each with certain special characteristics highly developed, can be inexpensively constructed and used with the same large Dewar tube and other accessory apparatus. The column is designed for total condensation, the product being taken off as a gas through a side arm just below the condenser, while the top of the condenser may be left open to the atmosphere. Variations of pressure in the column and the resulting effects on boiling points are thus entirely eliminated, even though the condenser temperature may vary widely. The condenser is purposely constructed with a large heat capacity, so that its temperature cannot change suddenly or vary rapidly to any great extent. It follows that the apparatus may be operated with only slight fluctuations in pressure and boiling point, even when the top of the condenser is shut off entirely from the atmosphere, as in vacuum or pressure distillation, or in the preparation of a pure substance when it is desirable to be absolutely certain that no contamination occurs. A simple semi-automatic device is used to control the rate a t which liquid air is supplied t o the condenser, so that the --x)
PRODUCT RECEIVER
F L E X I B L ~LEAD TUBING MANOMETER 7J OW^ DAESSdRE IU CON-!
OIL TO INDICATE WHEN AIR IS B E I N G DRAWN INTO COLUMN
hECTlhG TLBING AhD .h PRODUCT PECE VERI
CONDENSER-
MANOMETER TO SHOW PRESSURE INSIDE COLUMN
DETAIL OF UDPER PART OF CONDENSER
a GLASS TUBE THAT CAN BE RAISED M1 LOWERED -CONDENSER LARGE DEWAR' TUBE W I T H FRACTIONATING COLUMN INSIDE J
EXIT
CLOSE FITTING BUT NOT ACTUALLY
FIGURE 1. DIAGRAM OF APPARAT~JS 478
ON LlOUlD AIR IN
NOVEMBER 15, 1936
ANALYTICAL EDITION
479
temperature of the latter is kept between narrow limits and need be observed and regulated only occasionally. The column has about ten theoretical plates, a holdup of 0.6 to 1.2 cc., and will not flood until about 1 cc. per minute of liquid is passing down the column and a n equivalent amount of vapor is ascending. The product is usually taken off at rates of 5 t o 25 cc. of gas per minute, but with certain mixtures rates of 75 t o 100 cc. of gas per minute are possible without disturbing the equilibrium of the column. The liquid air requirement is about 0.5 liter per hour.
Details of Construction and Assembly The large Dewar tube is entirely silvered except for two longitudinal strips 15 mm. wide on opposite sides which make it possible t o observe the packed section and the still. The Dewar tube is enclosed in a box with glass front and back which is sealed and kept filled with dry air. This box protects the large Dewar from breakage and prevents condensation and frosting which would destroy visibility of the packed section. The still of the column is heated by a 100-watt internal heater of resistance wire sealed to copper wires carried through the glass with thin-walled platinum tubes. The packing may consist of small glass triangles (Figure 2) or single-turn glass ($4) or wire (6) helices. The former was used in all experiments described here because experiments in this laboratory (8) indicated its superiority as t o efficiency, holdup, and throughput. To revent condensation in the side arm of the column it is wounfwith a heater of resistance wire and this covered with a loosely fitting asbestos sleeve. The side arm is connected by flexible capillary lead tubing to the diaphragm valve, G (Figure l), which in turn is connected to amanometer and through diaphragm valves D and E to the product receivers and to the vacuum line and inlet gas sample drier, respectively. Connections to the top of the column are made through the !lass T into the top of which thermocouple wires are first careully sealed with de Khotinsky cement to make a gas-tight seal. After the condenser is in place enough packing is placed in the column to reach J . To keep the thermocouple junction centered in the column it is fitted with a wire spiral, the last turn of which is the same diameter as the inside of the column. All other turns of the spiral are of small diameter, just large enough to be slipped over the thermocouple wires to which it is tied with thread. The end of the thermocouple wire is then pushed into the top of the column about 2.5 cm. (1 inch) and more packing placed around it. Another spiral is then fastened in place to keep the packing from rising out of place. The thermocouple wire is then ushed down until its end touches the main body of packing at J. f n order to get correct boiling point readings it is essential to have an inch of packing immediately above the end of the thermocouple wire and to have the wire centered in the column. When this is achieved the Tis sealed to the top of the column with de Khotinsky cement. The side of the T-tube is connected to a manometer which measures the pressure in the column and also by means of flexible capillary lead tubing to the diaphragm valves, A and B, leading to the atmosphere and to the product receiver, respectively. Diaphragm valves are very advantageous because after they are once assembled and tested they can be depended upon to be gas-tight even under considerable pressure until the seat becomes worn. Stopcock grease troubles are also avoided. The diaphragm valves on the market a t resent, however, have more dead space than is desirable for t t e distillation of small samples. Two thermocouples are located in the packing just below the side arm, J, and another about 2.5 cm. (1 inch) above the side arm, 1. The former are used to measure the boiling point of the product, being taken off, and the latter to make sure that the reflux coming from the condenser is not too cold. Thermocouples are also located inside and outside the air jacket, G, H , so that the temperature differential between the material in the column and the air in the large Dewar tube can be estimated and kept at a minimum. With this arrangement the environment of the column is maintained within a few degrees of the temperature of the column itself. All temperatures are measured with copperconstantan thermocouples and a high-resistance millivoltmeter which can be read to 0.01 millivolt (0.4' C.). The Condenser is made from two concentric pieces of thinwalled steel tubing of proper dimensions, to which ends and
FIGURE 2. GLASS PACKING FOR COLUMN Packed section of column, and individual pieces of irregular glass triangles used as packing
connecting tubes are brazed or soldered. The inner diameter of the inner tube of the condenser and the outer diameter of the glass part of the column over which it fits are made of such a size that there is very little clearance between them. Further to improve the heat transference across this small air gap, it is filled with a low-melting wax. There is a baffle in the upper part of the condenser to prevent the incoming cold air (or liquid air) from passing directly to the lower end of the condenser. The condenser is packed with lead or copper shot to give it a relatively large heat capacity and to baffle the cold air further. The large heat capacity makes it very easy to kee the condenser near a desired temperature. Liquid air is su pyied to the condenser directly from a 5- or 15liter supply flas! through a siphon which connects to the top of the condenser by means of as short a section of connecting tubing as possible. That part of the siphon that extends into the liquidair supply flask should be of glass or other material of low heat conductivity to avoid undue consumption of liquid air. The slow current of cold air issuing from the mouth of the large Dewar tube causes formation of some frost. This frost is prevented from dropping back into the Dewar tube by packing the topmost 5 cm. (2 inches) of the tube loosely with cloth. At the end of a distillation the frost should be blown away with a stream of air, and the cloth removed and dried. To prevent frost and liquid water from condensing inside the Dewar tube after a distillation, a slow stream of dry air should be passed through the condenser until it warms up to room temperature. The use of small liquidair supply flasks is subject to the disadvantage that occasional sudden ebullition (bumping) of the li uid air greatly increases its rate of transfer to the condenser an8 thus the temperature of the latter is caused to vary. In small liquid-air supply flasks the level of the liquid air dro s rather rapidly and frequent adjustment of the air pressure krcing liquid air into the condenser is required to maintain condenser temperature constant.
Control of Condenser Temperature The rate at which liquid air is supplied t o the condenser and thus the temperature of the condenser is controlled (Figure 1) by maintaining a suitable air pressure (equivalent t o 30 to 60 cm., 12 t o 24 inches, of water) above the liquid air in the supply flask. This is best done by connecting t o a compressed-air line through a needle valve and allowing excess air t o bubble out through a tube, containing a suitable height of water which can be easily varied. By providing the column manometer with a n extra arm (not shown in diagram) leading t o the bottom of the tube through which the excess air is bubbling, the temperature control of the column can be made nearly automatic.
INDUSTRIAL AND ENGINEERING CHEMISTRY
480
BOILING POINT CURVE OF ETHYLENE-ETHANE MIXTURE
a p CF ETHYLENE
a
I
rn
I
I
I I
BDD 800 PRODUCT DISTILLED OVER -CC OF GAS do0
I
I
1ooo
i I
FIGURE 3. BOILING POIKTCURVEOF ETHYLENE-ETHANE MIXTURE
minute, rate of boiling 10 drops per minute, equivalent to.0.5 cc. of liquid or 100 cc. of gas f t -looo, C.) per mi?ute. Air outside air jaoket was approximately 5' 001 er than air inside air jacket.
Operation of the Column The sample is allowed t o enter the previously evacuated column through its side arm by way of the inlet sample drier and valves E and C ( A , B , and D being closed) until a pressure of about one atmosphere is reached. Liquid air is forced into the condenser rapidly until it reaches a tem erature about 60' C. below the boiling point of the lowest boiEng condensable gas in the sample, as measured by thermometer at P. The pressure on the liquid air in the supply flask is then adjusted so that approximately this temperature is maintained. As the sample condenses in the condenser and runs into the still, more sam le is allowed to enter. This process can be hurried in the case oflower boiling gases (ethylene, methane, etc.) by placing a small amount of liquid air in the bottom of the large Dewar tube. In case there are in the sample appreciable amounts of gases not condensable by liquid air, the accumulation of these will cause the pressure in the column to rise. When enough of such gases have accumulated to give a pressure greater than atmospheric, they are slowly removed through valve B into the product receiver and then examined further by conventional means in order to determine their composition. When all the condensable portion of the sample is in the column, valves B and C are closed and the heat input t o the still is adjusted until the desired rate of boiling is reached. If it is desired to distill methane, it is necessary to keep the condenser so cold that drops of liquid air occasionally fall from the condenser exit. Under such conditions there is no difficulty in refluxing methane. If the pressure in the column rises t o greater than atmospheric, the noncondensable gases causing this are removed as before. If the pressure becomes constant at some value below atmospheric, air is allowed to enter slowly through valve A until the desired pressure is attained. If the distillation is to be carried out at atmospheric pressure, valve A may be left open. The liquid-air supply t o the condenser should be adjusted so that the temperature of air in the large Dewar is within 15" of the boiling point of the product being taken off. This is most easily done by noting the condenser temperature corresponding to the correct liquid-air supply and then maintaining the condenser at this temperature. From this point on the operation of the column is very similar to that of a n ordinary packed column, the product being taken off through valves C and D into the product receiver. The usual readings of boiling point and amount of product taken off are made and recorded at suitable intervals. The rate of take-off (along a plateau of the boiling point curve) may be 75 t o 100 cc. of gas per minute without disturbing the equilibrium in the column, but rates of 5 t o 25 cc. per minute are generally more satisfactory. When the tem-
VOL. 8, NO. 6
perature a t the top of the packing shows signs of increasing, the rate of take-off may be decreased to give a higher reflux ratio. If desired, the column may be allowed to operate under total reflux for a suitable period. As the temperature of the top of the packing rises and approaches the boiling point of the next component, the temperature of the condenser may also be allowed to rise t o about the same extent. The size of the intermediate fraction between two pure substances can be decreased if 25 t o 7 5 cc. of product are removed through the top of the column after the boiling point is about midway between those of the two pure substances. This results in the removal of any of the lower-boiling component that may be trapped in the condenser. If the column floods a t any time the trouble is either too much heat input to the still, or too high a temperature in the large Dewar, so that reboiling is taking place in the packed section. In the latter case the remedy is to supply a little more liquid air to the condenser, thus lowering the temperature of the entire apparatus. If there is a tendency to draw in air through the condenser, or if the pressure in the column drops when the product is taken off, the rate of boiling is too low and heat input to the pot should be increased, or the temperature of the air in the large Dewar is too low and should be allowed to rise by decreasing liquid air supplied t o the condenser.
Typical Results Obtained with the Column Mixtures of ethane (boiling point -89" C.) and ethylene (boiling point -104' C.) were chosen to illustrate the results that can be obtained with the column (Figure 3). The boiling points of these two hydrocarbons are close enough t o one another (a = 2.25) so that a real test of the apparatus is afforded, and they are also near the middle of the temperature range for which the column is intended. Results of other distillations are also shown (Figure 4). In examining these boiling point curves it is to be emphaeized that the same smooth curve would be obtained even if the readings were taken and recorded by automatic devices so as t o give a continuous series of points. Experiments with ethylene-ethane mixtures of known composition have shown that the distillation curves can be used t o calculate the volume of either component in such a mixture t o within *40 cc. of its true volume, if a minimum of 400 cc. of that component is present. If the vapor pressure ratio of the components being separated is more favorable, the results will be correspondingly improved. The equal volume method of determining the cut point was used. The equivalent of about 150 cc. of gas remains in the
-120
-140
PRODUCT
DISTILLED OVER
- CC
OF GAS
NOVEMBER 15, 1936
ANALYTICAL EDITION
column a t the end of a distillation after reflux stops. Figure 5 shows the results of a distillation in which condenser temperature, heat input to pot, rate of boiling, and rate of take-off were purposely varied over a considerable range. The observed boiling point varied only slightly. The correct determination of the boiling point of the material a t the top of a small fractionating column is a difficult matter. The total mass and therefore the total heat capacity of the material around the thermocouple junction are quite small and so even a small gain or loss of heat from the surroundings destroys the vapor-liquid equilibrium. There is need for the development of some different method for determining the composition of the material a t the top of a small fractionating column. The method should preferably give continuous readings on very small amounts of material, and be such that extended empirical calibrations are not required. Viscosity, thermal conductivity, refractive index, and optical and electrical properties have not yet been thoroughly investigated.
Distilling Capacity, Efficiency, and Holdup The maximum distilling capacity was estimated from the number of drops per minute falling back into the pot. This was done with both benzene-carbon tetrachloride and lowboiling hydrocarbon mixtui.es and the results ranged around 1 cc. of liquid per minute. The efficiency was determined with a mixture of benzene and carbon tetrachloride after the column had been slightly modified so that samples could be removed from the pot. In a typical experiment the refractive index of the pot sample was n*i = 1.4901, while that of the side arm was 1.4779. In another experiment the corresponding figures were 1.4889 and 1.4782. Reference to the vapor-liquid equilibrium diagram for benzene-carbon tetrachloride indicates approximately ten theoretical plates (6,16). An attempt was also made t o determine the number of theoretical plates with a mixture of ethane and ethylene. After equilibrium had been reached with a mixture of ethane and ethylene in the column, small samples were removed and analyzed by absorption with fuming sulfuric acid. I n a typical experiment the vapors from the pot analyzed 7.7 per cent ethylene, while the sidearm sample analyzed 88.0 per cent ethylene. This gives a value of approximately 88.0 for the enrichment ratio. The vapor-liquid equilibrium relations of ethane-ethylene mixtures are not known, and since the meager evidence (1.2) available indicates that Raoult's law may not hold true for the above system, the equivalent number of theoretical plates was not calculated (4). The holdup was estimated by collecting separately as a gas the ethane in the column a t the end of a distillation when the pot was just dry. The holdup naturally varied with the vapor velocity and ranged from 0.6 to 1.2 cc. of liquid. It is t o be emphasized that a decrease in holdup by itself cannot be equivalent to an increase in equivalent number of theoretical plates. Some confusion has probably resulted because it has not been recognized that as the rate of boiling and therefore the liquid holdup in a column are decreased the number of theoretical plates may actually increase markedly (2.2). Often the improved fractionation is accredited entirely to the decreased holdup, while actually it is due in a large measure to increased number of plates. It is not low holdup alone, but low holdup per plate that is desirable. The disadvantage of large holdup can often be overcome by increasing the size of the sample being distilled, but a column with a small number of theoretical plates will be satisfactory only for separations in which the vapor pressure ratio is large (C, from Ca, or Cs from C, hydrocarbons). Serious difficulties are encountered when the ratio is small and boiling points are close together (isobutane from iso-
481
butene, butene-1 from n-butane, etc.). These difhulties will persist regardless of any changes in holdup or other characteristics, unless the equivalent number of theoretical plates is increased. This is illustrated by the figures in Table I. In part A of Table I it has been assumed that fractionation of a butene-1-n-butane mixture has reached the equilibrium stage when the molal ratio of these components in the still is unity. The per cent composition is given in the first column of the table. The second column gives the vapor pressure ratio for the mixture being distilled. The fourth and 6fth columns give, respectively, the enrichment factors and the percentage compositions of products a t the top of the column for columns of different numbers of theoretical plates as indicated in the third column. Parts B, C, D, and E of Table I give corresponding data for other commonly encountered mixtures. These relationships will hold regardless of any changes in holdup in the column since the conditions are those of total reflux. It has been recognized that the separation of mixtures like D and E is much easier than the separation of those like A and B. Table I gives a quantitative measure of this difference, and relates it to the equivalent number of theoretical plates in a column. Table I1 gives the vapor pressure ratios for other common mixtures. TABLEI. EFFECTOF EQUIVALENT NUMBEROF THEORETICAL PLATES
Cornposition of Liquid in Still
Vapor Reaaure Ratio
Number of Plates 7"
Ci-
umn
Enrichment Factor
Composition of Product Butene-1 n-Butane
%
%
A 50 butene-I 5 0 2 n-butane Isobutane
n-Butane
Z I I isobutane ~ n-butane
1 50% ethane 50% propane
lo
2 5
10
20
30 E 50% ethane 50% methane
50
1
2
10 100
10,000
10,000,000
Propane
Iaobutane
Ethane
Propane
90 99 99.99
10
1 0.01
.... ....
50 2500
Methane 98 99.96
Ethane 2 0.04
TABLE11. VAPORPRESSURE RATIOFOR HYDROCARBON PAIRE Components Methane-ethane Ethane-propane Propane-iaobutane Ieobutane-n-butane n-Butane-isopentane Isopentane-pentane Methane-ethylene Ethylene-ethane Ethane-propylene Propylene-propane Isobutane-iaobutene Isobutene-butene-1 Butene-l-n-butane n-Butane-cis-butene-2 cis-Butene-2-trans-butene-2
At B. P. of More Volatile Component
At B. P. of Less Volatile Component
482
INDUSTRIAL AND ENGINEERING CHEMISTRY
VOL.8, NO. 6
t i o n t h a t a l l of t h e residue is the same composition as the last portion on which a boiling point could be obtained. Errors 1, 2, and 4 are e a s i l y corrected for by measuring the volume of the various parts of the column. UEAT INPUT TD STILL IN AMPERES ( C W E C) Several methods have been proposed for determining the exact volume of a given component from the d i s t i l l a t i o n c u r v e , i n c l u d i n g the equal volume, the equal area (ZO), and the equal mole fraction of vapor methods (7). Only the 4ArE OF TAKE M F IN latter takes into account C C OF GAS PER MINUTE (CURVE A ) the fact that the first half of the intermediate fraction contains more than half of the lighter comTIME IN MINUTES ponent. With the apFIGURE 5. DETAILS OF OPERATION p a r a t u s described here Shows smooth nature of boiling point curve, even when there are large variations in condenser temperature, heat input to the equalvolumemethod pot, and rate of take-off. Top of column was open to atmosphere during this ex eriment so there was no variation in presawe. During interval from 80 to 110 minutes on time saale, apparatus operates with& any attention whatever. seems to give values a s n e a r l y c o r r e c t as the precision of the resultEi The uncertainties in these calculations, due t o lack of comwarrants and this method has been used in the cases displete knowledge of vapor pressure and vapor-liquid equilibcussed here. In such calculations, it is not necessary to rium relations, are not great enough to have any marked effect have an extremely sharp break in the distillation curve if on the results. the boiling points used in plotting the curve are the correct ones. The determination of the boiling points by means of a Limitations and Errors thermocoude in the mckinn below the side arm rather than It is not intended t o represent that the apparatus makes in the conienser give; correct boiling points, since this causes sharper separations and requires smaller samples or less time the temperature of the thermocouple to be determined solely for analysis than others that have been described. I t s utility by the equilibrium mixture of liquid and vapor passing over it, depends upon its ease and simplicity of operation, flexibility, and not by the temperature of the condenser. relatively simple construction, and low cost, and upon the Modifications fact that it gives a correct boiling point curve even in separations where the vapor pressure ratio is unfavorable. The apparatus is described here in its simplest and most Certain errors are unavoidable in any distillation analysis. generally useful form. Modifications of many kinds may be With the apparatus described here these are as follows: made to suit the requirements of particular types of distillation. The air jacket may be replaced by a strip-silvered 1. At the beginning of a distillation some of the lowest boiling com onent is required to fill the dead space in and above the vacuum jacket and the operation of the column then becongnser, the line to the column manometer, and the side arm. comes even easier and efficiency will increase, but the cost of 2. A further ortion of the lowest boiling component is reconstruction is increased. The use of this vacuum jacket is auired to fill dealspace between the diaphragm . - valve, C, and the recommended in cases where the column is to be used by enGroduct receiver. 3. The me of the equal volume method of determining the cut tirely inexperienced operators without supervision. The point may lead to error because the first half of the intermediate column may be operated as a partial condenser type, all raction may not contain half of each component. products being taken off through the top of the column. 4. At the end of a distillation, some of the highest boiling If it is necessary to transfer samples to the still as liquid component remains in the column. phase, a suitable connection of capillary tubing may be made Errors 1 and 2 affect the analysis in two ways-i. e., they directly to the bottom of the column. decrease the percentage of lowest boiling component below its true value and they decrease the total volume of the sample Distillation of Complex Butane-Butene Mixtures below its true value and therefore cause an increase in the The problem of the analysis of mixtures containing isopercentage of all components. Error 3 affect5 the volume butane (b. p. -11" C.), isobutene (b. p. -7" C.), butene-1 and percentage of each component present in a different way, (b. p. -6.5" C.), normal butane (b. p. -0.5" C.), cis-butene-2 since the magnitude of the error depends upon the nature of (b. p. $1" C.), and trans-butene-2 (b. p. +3.7" c.) has not the vapor pressure-composition curve of the pair of comyet been satisfactorily solved and will continue to be of imponents being separated. This error enters twice for each portance because of the availability and reactivity of these component except highest and lowest boiling components. substances. A successful analysis of such mixtures depends Error 4 decreases the percentage of highest boiling component upon efficient fractionation with cuts between isobutane and below its true value and decreases the total volume of the isobutene, and between butene-1 and n-butane. Acid absorpsample below its true value. It also necessitates the assump-
NOVEMBER 15, 1936
ANALYTICAL EDITION
tion or some physical property can then be used for analysis of the fractions, if the cuts are sufficiently sharp. Since the vapor pressure ratio for isobutane-isobutene or butene-l-nbutane is approximately 1.2, i t is obvious that columns having on the order of ten theoretical plates will not be satisfactory since the enrichment factor is less than 7. However, a column like that described in this paper, but with a packed section about 90 cm. (3 feet) long, would have 30 or more plates and an enrichment factor of about 240. This would give the required sharpness and smooth operation, along with a reasonable throughput. Experience with the column described above and other experiments on fractionation in small-diameter columns (22) suggest t h a t a t least some previously described columns, such as those referred to in the first paragraph of this paper, can be considerably improved for purposes of butane-butene distillation by modifying them as follows: 1. Use single-turn glass or wire helices as packing. 2. Relocate the thermocouple junction about 2.5 cm. (1 inch) below the bottom of the condenser. 3. Increase the heat capacity of the condenser by packing it
with lead or copper shot. 4. Sup ly liquid air t o the condenser at a constant rate by a device suegas that described above. 5. Supplement the manually controlled take-off valve by an automatic sintered-glassmercury valve (f 6).
The combined effect of these modifications will be to increase the total time required, but the equivalent number of theoretical plates will be increased a t practical rates of throughput, and irregular variations in the boiling point will be minimized because the direct transfer of heat from the thermocouple junction and its immediate surroundings is made more regular, pressure variations in the column are de-
483
creased, and effjciency of fractionation is stabilized a t a high level.
Acknowledgment The author wishes to acknowledge the suggestions and advice of M. R. Fenske and G. H. Fleming in the development of this apparatus.
Literature Cited Bosschart, IND.ENQ.CHEM., Anal. Ed., 6 , 2 9 (1934). Davis,Ibid., 1,61 (1929). Davis and Daugherty, Ibid., 4,193 (1932). Fenske, IND.ENG.CHEM.,24, 482 (1932). Fenske, Quiggle,and Tongberg, Ibid., 24,410 (1932). Fenake, Tongberg, and Quiggle, Ibid., 26,1169 (1934). Fitch, Nat. Petroleum News, 23, No. 25, 66 (1931). Fleming, G . H., personal communication. Freyand Hepp, IND.ENG.CHEM.,25,441 (1933). Frey and Yant, Ibicl., 19,492 (1927). Hunter, J . P h w . Chem., 10, 350 (1906); International Critiosl Tables, Vol. 3, p. 333. Kistiakowsky et al., J.Am. Chem. Soc., 57, 65,876 (1935). Leslie, “Motor Fuels,” p. 555, New York, Chemical Catalog Co., 1923. Lucaa and Dillon, J . Am. Chem. Soc., 50,1460 (1928). McCabe and Thiele, IND.ENQ.CHEM.,17,205 (1925). McMillan, U.S.Patent 2,005,323 (June 18,1935). Oberfell and Alden, Oil Gas J.,27, No. 22, 142 (1928). Peters and Baker, IND.ENQ.CHEN.,18,69 (1926). Podbielniak,Ibzd., Anal. Ed., 5, 135 (1933). Podbielniak, Ibid., 3, 177 (1931); 5, 119,172 (1933). Rogers and Brown, IND. ENG. CHEM.,22, 258 (1930). Rose, Ibid., 28, 1210 (1936). Schaufelberger, Oil Gas J., 29, No. 16,46 (1930). Wilson, Parker, and Laughlin, J . Am. Chem. SOC.,55, 2795 (1933). RNCEIVED July 29, 1936. Presented before the Division of Gaa and Fuel Chemistry at the 92nd Meeting of the American Chemical Sooiety, Pittaburgh, Pa.,September 7 t o 11, 1936.
Oil Acidity Pipet W. F. DAVIDSON Brooklyn Edison Company, Inc., Brooklyn, N. Y.
T
HE author’s inspection and salvaging of transformer oil often required acidity determinations of many samples. The normal procedure included the weighing of a 2 0 - g r a m sample, the addition of 150 ml. of benzene-methyl alcohol solvent, and titration with standard alcoholic potassium hydroxide, using alkali blue as a n i n d i c a t o r . Evans and Davenport (1) showed this method to be superior to the contemporary tentative A. S. T. M. procedures. The apparatus described below, however, may be adapted for use in the current tentative A. s. T. M. Designation D18827T by making it with the proper capacities for oil and solvent. When there was a large number of samples the weighing seemed tedious and time-consuming. An investigation of the specific gravities of a considerable number of used transformer oils revealed slight differences, and suggested a simpler means of measuring the desired amount for analysis. At
25” C. the average weight of 23 ml. of samples from 20 different transformers was 20.041 grams. The maximum deviation from the mean was 0.133 gram. Since the probable error involved in measuring the sample by volume was considerably less than the error in titration, there could be no advantage in using the more precise weight method. A convenient way of measuring both the sample and solvent may be accomplished with the combination pipet shown in the diagram. The particular feature of this apparatus is its self-rinsing property. The oil sample is drawn into the lower pipet and the oil adhering to the outer side of the tip may be wiped off before introducing the solvent, which is drawn in until the upper pipet is filled. The oil does not mix spontaneously with the solvent, but remains in the lower part of each pipet. When the contents are drained into the titration flask, the oil is washed completely from the pipet by the supernatant solvent. The apparatus may be used subsequently without further cleansing, and contamination is negligible. The use of the special pipet in conjunction with a standard self-filling buret has reduced the time required for acidity determinations to 25 per cent of that required by the old procedure.
Literature Cited (1) Evans and Davenport, IND.ENG.CHEM.,Anal. Ed., 3, 82 (1931). RECEIVED August 29, 1936.