dies in m
tvaporation development
THE FALL1
K. C. D. HICKMAN Easfman
THE surface of a liquid--usually
the interface between
exchange of molecules that evaporate and recondense, and in a closed system this process reaches equilibrium; in a partially open system, the liquid distills or evaporates, and the word “evaporation” as ordinarily used should be regarded a,s “net” evaporation, comprising the excess of outgo over return. The open-path, high vacuum still, which reacheb its climax in the molecular still, suffers no return of emergent molecules to occur, and a process of completely nonequilibrant distillation obtains; this we hall call “projective“ distillation. I t has been deduced by Knudsen (19-21) and Langmuir (%, 23) arid confirmed experimentally by them and many others that projective evaporation can occur a t a rate equal to, but not esceeding, the passage of
Evaporation
D. J. TREVOY N. Y
Kodak Co., Rochester,
a liquid and its vapor or other gas-t,here is a continual
=
AND
EAAA TENSlMETE
Glossary of Terms Emergence of molecules from surfacc of liquid. (In common parlance, “evaporation” and “distillation” are used interchangeablv t o describe perinanent conversion of liquid to vapor. Water in a dish has been said to “evaporate” slowly in air. In the present terminology, water is evaporating a t its niaximuin characteristic rate whether in a bottle or open vessel: it is the “mte of distillation” that alt,!rs q i t h external circumstances. Evaporalion 1s a natural phenomenon, distillation a managed event. The verbal limitation is a nuisance, but is considercd necessary for the present purpose.)
Rate of evaporation
= Mass of molecules emerging from unit area
Knudsen rate of evaporation
=
Evaporation coefficient
= Ratio of actual rate of evaporation to niaxi-
in unit time. Seither “evaporation” nor “rate of evaporation” defines the fate or destination of the evaporated riioleculc lIaxirnuin rate at which molecules could emerge from surface, deduced from measuremrnts of saturated vapor pressure at same temperature
mum Knudsen rate = Reverse process to evaporation Condensation Accommodation = Ratio of actual rate of condensation to inaxicoefficient mum Iinudsen rate = Irreversible removal of evaporated molecules Distillation from the syetem, generally iolloffed by condensation elsen-here (see evaporation)
molccules across an imaginary windorv of the same area as tile evaporating surface, situated anywhere within the sat,urated vapcr. It is a matter of history that the idea of high vacuum, nonreturn distillation, which had been examined casually during the previous 50 years, was revealed decisively by Brdnsted and Ilevesy (6) and by the papers and patents of Burch (7-Ql. An industry aro,ce from this foundation without further ncetl o i examining the basic concepts. After nearly 20 years of successful molecular distillation, it has become apparent that technology has outstripped science, and there remain many gaps in our knowledge and many anomalies to be explained. The most troublesome anomaly is the spontaneous refusal of liquids t,o eva.porate ab rates and t,einpcratures appropriate to t,hem-sometimes they do, sometimes thcy do not.
Projective distillation
=
Evaporation with free escape and subitantially no return of molecules to the evapo. rating surface. Uolecular distillation is a. restricted form of projective distillation Nomenclature
a A
= accommodation coefficient
area for distillation. sq. cin. constant mean free path, cm. L JI molecular weight number of molecules per cc. n refractive indes a t 25” C., using the sodium-D line p vapor pressure p,, = vapor pressureaccording to Perry and Weher (ZQ),microns = average rate of distillation in tn-o zones, gramsjsec.jsq, r meter = rate of distillation in Ion-er zone, grams/sec./sq. meter r1 rTn = rate of evaporation calculated from vapor pressure data, grams/sec./sq. meter = rat,e of distillation in upper zone, grams/seo./sq. inetcr ru Re = Iteynolds number t = temperature, O C. T = temperature, A. tu = weight, evaporated, yranlc = mole fraction of EIlP in liquid x = mole fraction of EIlP in vapor y a = relative volatilit’y C Y ~ I = relative volatility for ideal liquid and vapor in equilibrium = relative volatility for liquid and vapor in equilibrium, ae esperiment,al E = evaporation coefficient e = time to fill pipet n-ith condensate, see. p = inicrons u = molecular diameter, em. I;
1882
= = = = = = =
August 1952
INDUSTRIAL AND ENGINEERING CHEMISTRY
Any departure from the optimum or Knudsen rate is spoken of as the evaporation coefficient ( U ) ,somet.imes thought to be synonymous with the accommodation coefficient (2O), and it is apparent that the coefficient can be less than unity and variable. I t has been appreciated, perhaps instinctively, by distillation engineers, that there is a repressive skin on the surfaie of most phlegmatic liquids, hence the falling-film ( 1 7 , 32) and centrifugal stills ( 5 ) which scrub or renew the surface continuously. These devices beg the question by avoiding the problem, anti they leave laboratory chemists struggling with the vacuum pot still, n-here liquids often fail t o evolve any vapor until they suddenly “esplode,” becoming quiet again after the burst of vapor h a s passed t o the pump. The phenomenon repeats itself periodically, annoyingly, and sometimes dangerously. Terms such as “superheat” and “association” have been applied to a situation which they neither explain nor cure. Vacuum stills are not used primarily to demonstrate speed of distillation but are employed t o separate constituents from mixtures, with the least thermal damage consistent with the means. It ie common knowledge t h a t the separatory powers of simple pot stills vary with trifling changes in design and operation, and such stills, fitted with allegedly fractionating columns, may have less separatory power than without the columns. I t has thus been difficult t o assign fixed relative volatilities to binary test mixtures and equally difficult to decide what is the separatory power Qf one unit act of projective distillation. There has been no definition of a “theoretical molecular plate” ( l a ) , and without, this, the design and evaluation of new stills is seriously hampered. As t o the less practical gaps in our knowledge, Raoult’s law, Henry’s law, and such matters as activity coefficients have scarcely been examined in the high vacuum still. The work here planned and reported in three following papers has Rimed to explore such matters and follow whatever leads are found during exploration. Evaporation and Accommodation Coefficients
The most pressing problem is the inhibition of evaporation which causes reduced and variable values of a in the Knudsen equation:
4;
limited way (4,33) for rationalizing meager data. Indeed, some “results” have been produced without recourse to the laboratory a t all (65-27). In reviewing the literature, it has been apparent that evaporation coefficients fall into two classes, those near unitv and those very far from it. Clean mercury was found by Knudsen (21) to be in the first class, dirty mercury in the second. I t is easy to picture mercury covered with it film of dross; theneed forpurifyingmercury is familiar t o every student. That water or glycerol or ethanol could have a film of dross seems t o have escaped attention. The hypothesis that seemed attract,ive to us in planning this work was that new, absolutely clean sur-A faces of any liquid might exhibit a coefficient of unity; B stale contaminated surfaces b would not. The low coefficien ts reported heretofore C could then be classed as unstirred pot-still effects. Our C task thus divided itself into two main parts: Production of the cleanest possible surfaces and study of relatively stagnant and presumablyaontaminated surfaces. Obviously the latter are those of ordinary everyday experience, and it is the former that require special means of production.
A Falling-Stream
-
Rate (weight/time/area)
=
k , a, p
vhere p is the saturated vapor pressure a t temperature, T , of a liquid having molecular weight M , and k is a conversion conRtant dependent on the units of temperature, pressure, weight, time, and area employed. In the present state of knowledge, there is no way of deducing or calculating rates of distillation from atomic constants, but given the “measured vapor pressure,” the optimum rate is deduced a t once. The accommodation coefficient, a, is, of course, the degree of departure from this optimum. The concept of accommodation coefficient arose in connection with the evaporation of metalj, evaporation being the esact reverse of condensation, and condensation in turn being possible only when a site is open t o accommodate the arriving particle. In the situations we shall consider, evaporation iE: not necessarily t h e mirror image of condensation but is rather to be considered as a step in a process which becomes exactly balanced in proceeding through a complete cycle. We shall refer to the evaporation coefficient, e, in place of a in what follows. In recent years there has arisen a small but important school of workers who have studied evaporation rates of many liquids (1, 3 ) and have concluded that water (2) and other polar liquids (4)have low coefficients, while symmetrical molecules (3, 4) evaporate a t the optimum rate. Some of the measurement&have been made with ultrasimple apparatus-the meanb have perhaps been inadequate for the task. A consequence has been a resort to mathematical theory on the grand scale by Frenkel (Is), Eyring (II), Hirschfelder (28), and by others ( I O ) , in a more
1883
Tensimeter
The means chosen for the examination of the evaporation coefficient from clean surfaces is the collection of distillate in a falling-stream molecular still in such a manner that there will be reasonable assurance that rate of collection of distillate = rate of Figure 1. Diagram of Obrervadistillation = rate of evapotional Parts of Falling-Stream ration and t h a t Tensimeter rate of collection ~ _ . _ _ _ _ calculated Knudsen rate emax = best evaporation coefficient ( = a = accommodation coefficient under the conditions then obtaining). The tensimeter thus strives t o produce a pristine distilling surface as suggested later (15)or failing this, a limiting or optimum approach thereto. I t is not known yet what category of surface is produced but that the tensimeter nears its objectives is shown by two characteristics:
c
1. The two fluids, ethyl hexyl phthalate and ethyl hexyl sebacate (EHP and EHS), each distill almost exactly at the maximum Knudsen rate. The same is true with glycerol, as shown in later experiments by Trevoy (not yet published). 2. The relative volatilities of the mixed liquids are reproducible and are the highest that have been reported from an unequivocally unit act of distillation-Le., repetition and reflux are absent.
_
1884
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 44, Na. 8
compositions of distillate remain alike in the two zones for very high rates of distillation, such as 100p saturation pressure or 180 kg. per square meter per hour, so it may be assumed that thc temperature and composition of the surface are the same as the main bulk of liquid and therefo1.p sufficiently well known.
f
Working drawings of the tensimeter are sho\~.n in Figures 2a and 26. The firvt tensimeter \vas designed and built in these laboratories by Hickman. Calculations by Trevoy showed that the jet of 0.78-cm. diameter >\-odd produce a t,urbulent stream of liquid in the common rengcs of viscosity, besides requiring a largc filling. L-i smaller instrument wi-ith a narrow jet (
