DISTII,I,ATION
W I T H VAPOR
B Y C. H. BENEDICT
If the vapor pressure of a substance at a given temperature is less than the external pressure, the substance will not boil at that temperature if a liquid, nor sublime if a solid. Since distillation by diffusion is too slow a process to appeal to anyone, it is necessary to reduce the external pressure until it equals that of the substance under consideration or to add a volatile substance in such quantity that the vapor pressure of the system is equal to the atmospheric pressure. T h e first method has been used by Ramsay and Young’ to determine the vapor pressures of liquids and of solids. T h e second method is used very largely in purifying organic compounds. When the pressure of the systeni is made equal to the atmospheric pressure by addition of air, the process is spoken of as distillation with an air current. When the volatile component is water vapor we speak of distilling with steam. There are two cases to be distinguished in distilling with vapor ; when the substance to be distilled is a liquid and when it is a solid. T h e first case presents no difficulty. In order to get the best yield the volatile component added should be a vapor which is not soluble in the liquid to be distilled. Under these circumstances the vapor pressure of the system is the sum of the pressures of the two pure components ; and the same amount of the liquid to be distilled passes over per volume of vapor as if the distillation had been carried on with an air current or at the same temperature by means of diniinislied external pressure. If the vapor added dissolves freely in the liquid to be distilled, the partial pressure of the latter is reduced and the quantity of this component in the unit volume of vapor is less than if the other component was not present. \ When the substance to be distilled is a solid at the temperature of the experiment, matters are very different. There is present the ’Phil. Trans. 175, 37 ( 1884).
C. H. Benedz'ct
398
moiiovariant system, solid, solution and vapor ; and the question at once arises as to the concentration in the vapor phase of the substance to be distilled. Since the solid phase is in equilibrium with the vapor, Ostwald' says that the partial pressure of the solid coniponent in the vapor phase must be equal to the vapor pressure of that component in the pure state ; but Bancroft' has shown that this is not necessary theoretically. Margueritte-Delacharlonnaq has found that strong solutions of sodium hydrate, sodium carbonate and ferric sulfate yield vapors at 60-70" in which the presence of the solute can easily be detected. I t is always possible that some of the solution is carried off mechanically, especially since McLeod has shown that a tube filled with glass beads is not sufficient to remove potassium chlorid carried oyer mechanically by the oxygen formed from potassium ~ h l o r a t e . ~I n this case the amount of solid salt was so great that it could be detected optically in the gas. On the other hand the evaporation from a strong solution at 60-70" is not so rapid as to cause an enormous amount of liquid to'be carried off in the vapor and it is therefore safe to say that Margueritte-Delacharlonnay has called attention to a phenomenon which is worthy of careful study. Very recently niuch the same result has been obtained by Bailey.5 H e investigated the volatility of some alkaline clilorids in presence of water. From a triple-normal solution there distilled some forty-three milligrams of cesium chlorid per liter of water. While the vapor pressure of cesium chlorid has not been determined, there can be little question that this result is much in excess of what was to be expected and it should also be kept in niind that the solution was not saturated. Gooch's experiments on the volatility of boric acid are still more remarkable.6 H e found that boric acid distilled much more rapidly when heated with a given volume of methyl alcohol than ~~
'Analytische Chemie, 33 ; Cf also Wildermann. Zeit. phys. Chem. 1 1 , 407 (1893) and Nernst. Theor. Cheni. 376. *Jour. Phys. Chem. I , 147 (1896). Comptes rendus 103, 1128 (1886). 4Jour. Chem. SOC.69, 101.5(1596). SIbid. 6 5 , 445 ( 1894). 6Proc. Am. Acad. 22, 167 (18S6).
Distiildation with Vapor
'
399
with the same volume of water. Since a given volume of water furnishes a larger volume of vapor than an equal volume of methyl alcohol and boils at a higher temperature it should be doubly effective for the distillation of boric acid whereas the reverse is the case experimentally. This volatility of boric acid is often accounted for in an off-hand manner by assuming the formation of a volatile, instable ester. I am not aware that any experimental evidence has been offered in behalf of these curious hypothetical compounds and it does not seem probable that any will be forthcoming. While boric acid might easily form an ester with methyl alcohol, such a compound with acetone needs justification and yet boric acid distills readily with acetone. A much simpler explanation would be that of an increased vapor pressure of the solute due to the nature of the solvent. Villard' finds that iodin is soluble in carbonic acid vapor and Hannay has shown that salts are distinctly soluble in vapors near the critical temperature and pressure while Rose2 has found that gold volatilizes much more rapidly in an atmosphere of carbon monoxid than in one of coal gas and that its volatility is increased by the presence of copper or platinum. From these facts it seems safe to conclude that the nature of the solvent has an effect on the partial pressure of the solute even when the latter is present in the solid form and it becomes desirable to test this conclusion experimentally. To this end a few measurements on the volatility of naphthalene in the presence of ether have been made. These determinations are . only first approximations and will be repeated with more care in this laboratory during the coming winter. An excess of the solid was always present so that the solution might not become supersaturated. A n ordinary 250 cc distilling flask was used. The distillate was analyzed by evaporating to dryness at room temperature with an air current. There are two sources of error here. T h e solvent may not be entirely driven off and some of the solid may evaporate. Luckily these two errors are in opposite directions and counterbalance each other to some extent. A test ICornptes rendus 120, 182 (1895). ZJour. Chem. SOC.63, 714 (1893).
400
C. H. Benedict
experiment with a known solution of naphthalene in ether showed that this method of analysis was accurate to about two percent. Two different distillations of naphthalene and ether gave for the composition of the distillate 1.181 grams of naphthalene to 13.06 grams of ether and 1.404 grams of naphthalene to 16.4 grams of ether. From these data it is possible to calculate the possible pressure of each component provided we assume that the variations from the Gas Theorems are of the same order and sign for the two components. If x and y denote the number of reacting weights of the two components in any quantity of vapor and the barometer stands at z nim of mercury, the partial pressures will be xz/x+y and yz/x +y millimeters of mercury because the vapor pressure of the system at the boiling point is equal to the external pressure. Taking 128 as the reacting weight of naphthalene, 74 as that of ether and 760 mm for the height of barometer, the above data give for the partial pressure of naphthalene 37.7 and 36.0 nini of mercury. A variation of the barometer of 30 mm would produce a change of less than four percent so that no serious error is introduced here. It remains to be seen what temperature this refers to and whether the result is too high or too low. T h e system, solid naphthalene, naphthalene in ether and vapor, must have a constant boiling point so long as the external pressure remains unchanged. This teniperature is about 62'. There were some fluctuations owing to there being nothing in the flask to ensure steady boiling ; but this may be neglected. On the other hand the vapor began to condense before it passed through the side tube and the temperature at this point was of course much lower since the vapor is chiefly ether. A thermometer in the vapor stood at about 40'. It is however only at the surface of the liquid that the pressure of the system is equal to the external pressure. T h e temperature there is the one which must be taken. This great temperature difference could be eliminated by having the side arm of the distilling flask close to the surface of the liquid. This was deliberately avoided in the present set of measurements because it increased the liability of the solution being spattered over which would have vitiated the experiment. Also, the partial condensation of the vapor decreases the amount of naphthalene in the vapor and therefore if the distillate is more concentrated than one
Distillatioiz with Vapor
401
would expect from the vapor pressure of naphthalene, one could be certain that the real difference was even greater. T h e conditions of the experiment tend to make the quantity of naphthalene in the distillate less than in the vapor. On the other hand any loss of ether through evaporization would tend to reverse this. This loss was made as sinall as possible by surrounding the receiver with ice. T h e value 37 mm is to be considered as the provisional partial pressure of naphthalene at a temperature of 62'. There may easily be an error of twenty percent in this determination. This is of no importance since the vapor pressure of pure naphthalene a t its melting point, 7 g . 2 " , has been found to be g mm by Naumann' and this value has been accepted by Ramsay and Young.' There is therefore an enormous increase in the volatility of naphthalene due to the presence of ether.' Distillation with an ether current would give not less than four times as much naphthalene as if the process were carried on with an air current or under diminished external pressure. With alcohol or benzene it is impossible to make measurements under atmospheric pressure because naphthalene is miscible in all proportions with either of these liquids before the boiling point is reached. Some rough measurements with camphor in different solvents gave similar results, the amount of camphor in the distillate being in excess of that calculated from the vapor pressure determinations ; but the great solubility of camphor in organic solvents makes this substance a difficult one to work with so that these results are not yet in a state suitable for publication. There seems to be no question, however, that the solvent does affect the vapor pressure of the solute and this opens up a new field of research, for the relation between the amount of this effect and the nature of the solvent has to be determined. T h e most plausible assumption would be that IBer. chem. Ges. Berlin, 4, 74 (1871). Thil. Trans. 175, 44 (1884). 3The great difficulty experienced by all chemists i n trying to purify gases from impurities which are ( ( carried over mechanically ) ) is a strong argument i n favor of the view that there may be something more than a mixture ) ) i n the vapor phase.
((
mechanical
C. H. Benedict
402
the more soluble the substance is in the liquid, the more soluble it is in the vapor. This is only an assumption and lacks experimental confirmation as yet. It should be remembered that if we consider the liquid and gaseous states as passing continuously one into the other, the solubility of a substance in a vapor is a necessary consequence. Some day we shall have solubility curves for solids in vapors as well as in liquids. While the difficulties of measurepent will be much greater in the former case than in the latter, we shall have much more opportunity to study mass action than is now possible. T h e concentration of the solvent vapor can be varied from zero up t'o that in equilibrium with the saturated solution whereas a change of pressure has very little effect on the concentration of the liquid solution phase. This research has been carried on under the direction of Professor Bancroft in the chemical laboratory of Cornel1 University.
J u b I, 1896.