Figure 1.
Five-Foot
Still
on Assembly Floor (Plumbing, Electric Wiring, Control Panel, and Thermal Lagging Not Yet in Place)
COMMERCIAL MOLECULAR DISTILLATION K. C. D. Hickmsn DISTILLATION pROOUCTS, INC., ROCHESTER 13, N. Y.
vapor barriers betxeen the evaporator and condenser. I t is current practice to use all three devices (a, b, and c ) , Pingly or in combination.
HE thin-film molecular still ( 2 ) is believed to cause the least thermal hazard of any known process of distillation. Many substances Jyhich cannot be handled 1%-ithoutdecomposition in ordinary vacuum stills nevertheless do not demand the extreme advantages of the molecular still. I n such cases the niargin of safety can be expended: (1) in the use of larger molecular equipment] complete with external heat interchangers, vihere the total thermal exposure considerably exceeds that of the laboratory still; (2) in promotion of better separation by ( a ) partial reflux, ( b ) multiple redistillation in the same still or in a succeshion of molecular stills, or (c) fractionation] which is defined for the present purpose as securing a better-than-unit separation in a single distillation unit; fractionation involves the use of niechanical
T
LARGE SCALE MOLECULAR DISTILLATION
The unit which has received intensive developnient during the last feiv years is the &foot centrifugal still (18, 28). Figure 1 is a photograph of the still and pumps on the test floor before installation of thermal lagging. Figure 2 is a sketch of the sectional elevation. Bt commercial throughputs the still iniposes a thermal exposure about tiventy times greater than that given by the experimental 14-inch laboratory cent,rifugal still ( I O ) and ahout a quarter of a million times less than the usual petroleum vacuum
686
June 1947
INDUSTRIAL AND ENGINEERING CHEMISTRY
flash still. These figures are forty to sixty times more and a hundred thousand times less, when efficient external heat exchangers are added. Xhether such exchangers are to be used depends on the nature of the distilland. Without t'he exchangers the yield of labile product is higher, the throughput correspondingly loirer. ROTOR. The evaporator in the &foot still units is an aluminum "flower pot'' made in a single casting, with a diameter of 155 cm. at the top and 110 em. a t the bottom; its gross area is 5 square meters and its probable effective evaporating surface, 4 square meters. The casting is turned to a high finish on the inside and a rough finish on the outside, and is accurately balanced dynamically. A4fterinstallation in the still, the evaporator rotates on a vertical shaft xithin a nest of Calrod electrical resistance heaters which acquire a dull red heat in the vacuum and warm the rotor by radiation. Conventional thermal insulation, aided by the prevailing vacuum, focuses most of the heat onto the rotor. We receive continual suggestions that the rotor be heated by an embedded winding or by induction or a circulating fluid, etc., but the fact remains that for simplicity and ease of maintenance the present scheme has many advantages. The other methods are, of course, receiving attention. K e do not knoiv n-hether the shape of the rotor represents an optimum choice, but the performance is encouraging. A relatively steep cone, n-ider at the top, was chosen because: ( a ) The distilland is lifted up the distilling surface and can flow from the collecting gutter directly into the next still in series. The rotor thus serves as a lift pump. ( 3 ) The component of force pressing the distilland against the rotor is high in comparison to the force driving it up the rotor, although both are much greater than gravity. K i t h the rotor turning at 400 revolutions per minute, the component of force driving the oil upward is about 35 times gravity aiid toward the rotor, 130 times gravity. Phenomenally good thermal contact is ohtained with high turbulence, and distillation can be undertaken at high saturation pressures ( 2 7 ) that n-ould otlien\-ice detach the distilland bodily from the rotor. ( c ) F:iiti~iinme~!ti. reduced to a minimum. Accidental splashing or dripping.. ~~-itliiii the still fall to the bottom and are immediately r.onaolidatrtl \\-it11the climbing nil film. The distilland is allowed to enter at the base of the cone through a pipe pointing in the directioii of r(~tation aud of diameter, or nozzle diameter, sufficient to project thehtreani at the peripheral speed of the rotor. Other, more complicated means secure an even better distribution of
687
T h e design of stills suitable for handling oils and heavy chemicals in quantities of 50-950 gallons per hour per unit now favors a rotating evaporator 5 feet in diameter and shaped like a flower pot.
Cast of aluminum and turned on a preci-
sion lathe, the sides of the evaporator slope upward at an angle of 10-95" to the vertical.
Distilland admitted at the bottom
of the cone flows up the hot sides and passes over the top into a gutter; the undistilled residue passes to a pump and then outside the still. Degree of separation of constituents, although never better than lT.M.P. (theoretical molecular plate) and generally between 0.8 and 0.95 T.M.P. is often two or three times better than that available from a nonfractionating pipe or pot still, owing to the thinness and turbulence of the evaporating film on the centrifugal evaporator. Enhanced separation is routinely obtained b y such alternatives as partial redistillation within a single still, partial redistillation with feedback in multiple still assemblies, or distillation through fractionating barriers. Examples and cost data are presented.
splash-free distilland, but' their description is beyond the scope of this paper. The rotor when first installed has a brightly burnished finish on the inside distilling surface, a rough but bright finish on the out,side (Figure 3). In this condition it' is admimbly fitted to conserve radiation towards the cold internal condenser but quite unsuita.ble for receiving radiation from the Galrods. The matter adjusts itself automatically by a convenient series of events. T h e n a ne\v still, and especially a neiv rotor, is first put into operation under high vacuum, the Calrods rise to a temperature high enough to evaporate some metal from their outer shell, which condenses on the back of the rotor. Both Calrod and rotor become pitch )lack (Figure 3), the exchange of radiation becomes better than 90% black body and the temperature of the heaters a t once falls into safe limits. On the front of the n e x clean rotor the radiation losses are negligible, being a small fraction of nonspecific black-body distribution ( 9 ) . The distilland passing across is in intimate contact with the hot metal, the temperature of which it acquires within a fendegrees. The thickness of the distilland, about 0.05 mm., is in the correct range to emit the characteristic infrared spectrum of the constituents (%), as Figures 4 and 5 show. 4 t this stage, therefore, radiation losses from the distilland are high and from the rotor negligible, and the distilland is warmed by conduction. With a clean distilland such as clarified filtered glyceride oil or a crude plasticizer that has been washed free of corrosive catalysts, the status quo could be maintained indefinitely. In practicr, the surface of the rotor gradually heconies soiled from centrifugal separation of dirt, formation of aluminum salts, deposition of phospholipides which "poach" and then char, etc. Thermal conduction is noiv reduced, but I I V \ the distilland becomes warmed in RESIDUE 1 FEED OUTLET addition by radiation from the soiled rotor. The change of status Figure 2. Diagrammatic Elevation of a 5-Eoot Molecular Still
688
Figure 3.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Unused 5-Foot Still Rotor and Deposit on Outside of a Used Rotor
Vol. 39, No. 6
alloxTed to flow over the rotor at the highest temperature the test dye will tolerate nithout instant evaporation. The dye is introduced from an electrically controlled squirt a t the base of the rotor, and the times are determined for a colored film to appear and vanish at’ the upper rim. Glyceride fat, fed at 750 kg. per hour and 200 O C. required 1.2 seconds for passage in a layer approximately 0.08 mm. thick at the start of the ascent and 0.04 mm.a t the top. COZDCSSER.The folloiving characteristics are generally desirable for the condenser of a large molecular still: There phould be ample opportunity for escape of permanent gas ( 2 5 ) ; the condenser should be cool enough to collect 90-99% of the wanted constituents at one molecular impact; and the condenser should be hot enough to reject most of the unwanted semicondensables which can then be collected by a separate, very cold condenser. These considerations h a r e influenced us to reject the simple and obvious barrel, or inverted top-hat, condensing surfaces of Figure 8 for niultileaf vertical condensers which have inherent selfpumping qualities. These condensers (Figure 9) perform admirably and n-ill be retained until w.e are able t o examine the relative advantages of simpler types. The vertical leaf condenser consists of a nest of heavy vertical t,ubes manifolded a t each end into circular distributors. The leaves, made of sheet aluminum, are fabricated around the pipes and are provided with sets of individual gutters (three to a leaf at present) for the collection of three separate fractions. The condenser is hung xvithin the rotor from stationary supports a t the top of the still. The leaves are inclined slightly to meet oncoming molecules (21), which are flung into the condensing ITedge and thus croxvd the noncondensables through the far end into the centrsl vacuum space (Figure 10). I n spite of the high porosity of this arrangement, the distillate has been found to condense alniod entirely in the appointed zones. Semicondensables are trapped by the center coil which is cooled by cold water. The condenper leaves are “cooled” by warm or even boiling water. Loss of radiant heat is minimized by maintaining the leaves a t the highest t,eniperature compatible vith proper functioning. V.kccni S ~ s T m f .The machinery for producing a sufficiently high “molecular“ vacuum consists of three parts: ( a ) condensers and scrubbers, (b! high vacuunl pumps, and ( e ) forepumps. The line of demarkation bet\Teen ( b ) and ( e ) is movable, and intermc.iliatc stages are used, often referred to 3~ lmoster?. I n the nt instance t i y o schemes are available. One compr 1 ejwtors in series xhich serve as the forepunip (1. boosters and R diffusion pump (Figure I), each operated by oil vapor. The other arrangement (Figure 2) emploj-s a five-st,age steam ejector and one condensation pump, designed to have a
is satisfactory, and distillation continues a t full efficiency. Presently, hon-ever, the layer of dirt on the rotor becomes so thick that a thermal gradient develops m-ithin the layer, charring occurs in the back l&nae, and it is necessary t o dismantle the still for cleaning. A soiled rotor is shon-n in Figure 6. To obviate this, automatic cleaner. have been installed which wipe the rotating surface a t appropri2 0 ate intervals and enable distillation v, v, t o be continued indefinitely a t the 2 highest efficiency. The function of W the cleaners (Figure 7) is not t o LL 0 reproduce the pristine burnished ap> pearance but to maintain the “second t v, condition” indefinitely. z W As to the thickness of the distz tilland layer, it has not been convenient to make tests during service operation. Instead v.e have used the method of dye iniection ( I I ) , applied to the hot rotor Figure 4. Infrared Emission Spectra from with the condenser removed. The inside Surfaces of Clean ( A ) and Soiled (B) distilland a t its usual feed rate i j Rotors
9
8
7
6
5
4 3 M
WAVE LENGTH Figure 5. Infrared Emission Spectra of Clean Rotor (A), Clean Rotor with Mineral O i l Flowing across Surface (61,and Clean Rotor with Vegetable Oil Flowing across Surface (C)
June 1947
INDUSTRIAL AND ENGINEERING CHEMISTRY
Figure
6.
inside of Soiled Rotor
high compression ratio between intake and exhaust. n’ith either scheme the oil in the pumps soon acquires a false vapor pressure which is a composite of the stable volatiles received from the still and the polymers produced from the unstable volatiles. This situation may be adjusted by incorporating dynamic fractionating features (141, consisting of a polymer sink and a still column (24), or by frequent changes of oil, which can be undertaken without interrupting the operation of the still. The latter alternative is tantamount to the use of an external purifier. At present a number of admirable new pump fluids (I, 6, 27) have appeared on the market; their resistance to thermal exposure and their ability to produce very high vacuums tvarrant, in
Figure 7. A V i e w of the H i g h l y Burnished Rotor after Being Recleaned
689
some cases, expensive synthesis and consequent high price. These fluids are suited for dry ( I S ) vacuum systems. The large inolecular still operates in a w&t saturated vacuum, Fyhich contaminates the pump fluid and reduces most oils t o a common denominator in a short time. The pumping system of the &foot still IS designed, therrfore, to operate on the leaet expensive type of oil available-namely, selected fractions of petroleum distillate ( 4 ) which can produce pressures of k s s than a micron and opcxrate into a forevacuum of millimeters. If the oil becomes “burned” or otherwise accidentally spoiled, it is replaced a t trifling cost SIMPLE DISTILL~TIOR.. When vapor is leaving the rotor of the &foot still a t the rate of 10 kg. per hour, the forward or saturation pressure of molecules advancing toward the condenser is about 1 micron (26). High-saturation unobstructed-path distillation has been carried out satisfactorily a t forty to sixty times this rate. The rate of feed of distilland has been varied between 300 and 1000 kg. per hour. Sometimes it is not feasible to distill more than 40-50y0 of the distilland a t one pass if the still is to be kept in operation for extended periods, owing to the tendency of the rotor to become dry near the upper edge. If complete or nearly complete distillation is wanted, the residue may be recirculated-for instance, according to the cycles pictured in Figure 11. The first scheme shows the residue being recombined with the entering distilland, a small quantity, 1-5%, being rejected on the way as a final residue. A variation, which can be applied to all the schemes, is to add a nonvolatile liquid (16) to the incoming distilland at
INDUSTRIAL AND ENGINEERING CHEMISTRY
690
1
OUTLET
I
CONDENSER
+-----
ROTOR
COOLlNO WATER
INLET
Figure 8.
Diagrammatic Elevations of Barrel and inverted-Dome Condensers
approximately the same rate as the blretl-off. This heavy mat,erial then accumulates in the still and constitutes about SOC; of “false residue” which cycles around and around, and k w p b t h e rotor properly wet, Castor oil added during t,he distillation of a phthalate ester is a typical example of t,his procedure. The second scheme is similar to the first except that the residue is collected in a tank until ‘a sufficient quantity has been accumulat,ed to warrant a unit redistillation by itself. There results a second residue which may be discarded or worked up when enough is available to a third or rith residue. This method is indicated when the original crude contains small quantit’ies of less volatile constituents of commercial value. The third method provides complete processing on a succession of stills of decreasing area. The scheme is to be preferred in large molecular plants where the residue from one large still or group of stills can be passed to successively smaller stills in a continuous progression (19).
(T.1I.P.). Complex stills which give enhanced separation will then he assessed a s multiples of the T.M.P. and rated in terms of equivalent molecular plate (E.N.P.). The concept of theoretical plate in equilibrium distillation is eminently practical in that, it stems back to measurements made in an idealized equilibrium still. With an “ideal” binary mixture, adequate stirring, absence of reflux, and slow rate of distillation, thc ratio of the composition of the vapor to that of the distilland is a t a maximum and is direct,ly dependent on the vapor pressur& of the two constituents. As a corollary, whether t h e mixture is ideal or ot,hrrwise, the equilibrium still operated under the conditions specified provides separation of exactly one theoretical plate. The concept of theoretical molecular plate in the unobstructedpath still has an equally practical foundation, although there is a rornplicating factor which is inherent,in both falling film and centrifugal stills, due to the progressive travel of the film across t,he eraporating surface. I n t,hewell-stirred pot still it is assumed that every elrment of evaporating area is of identical composition. I n the progressive film still succeeding areas cannot be identical, exccpt a t infinitely low rat,ios of cvaporation to throughput. One T.1I.P. is considered to be achieved when the condenser is indefinitely close to the evaporator, distillation is so slow that the evaporating surface is truly representative in composition of the main bulk of distilland, and the vapor molecules emerge SO far apart from one anot,her in time and space t,hat there are no collisions in flight. Under such conditions the difference in composition of distilland and molecular distillate is a t a maximum. The reparation of an “ideal” binary mixture will then be proportional t o t.hc ratio of the partial pressure and t o the square root’s of the molecular weights of the constiruents. Tjnder these conditions, the ,still operates with an efficiency of one T.M.P. whether the mixture is ideal or not. The concept of t,heoretical molecular plate requires extended mathcniatical treatment and experimental verification under widely differing conditions, and must consequently be left t o a future paper. The practical point for present consideration is that, thr 5-foot centrifugal still, at commercial throughputs of
.
THEORETICAL M O L E C U L A R PLATE
Before defining means for better separation than are available in a single passage through the still, it is desirable t o define the unit of separation that a single passage can achieve. This will be considered t o represent one theoretical molecular plate
Vol. 39, No. 6
Figure 9.
Vertical Leaf Condenser
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1947
691
viscous liquids of high molecular weight, continues to give separations of 0.804.95 optimum (Le., 0.80-0.95 T.M.P.) whereas the unstirred pot still in high vacuum gives only 0.3 t o 0.5 times the separation indicated by t'he equilibrium still. Therefore, the centrifugal molecular still is itself a means for better-than-unit separation by a factor of tm-o t o three t,imes. The 5-foot rotor, operating at saturation pressures between 1 and 50 microns, provides separations in t,he range of 0.95-0.80 (T.M.P.). Although this is not sufficient t o separate substances of near boiling point, it makes entirely acceptable partitions b e h e e n many substances of high molecular weight because the extended temperature scale affords room for the necessary differences in boiling point. Three established examples typify the performance t,o be expected:
VITAMINX ESTERS FROM FISHLIVER FAT. Average boiling point difference between esters and fat is 70" c'. Degree of separation in single-pass distillation is: vitamin .1 fraction, five times increase in potency, 70Yc yield of vitamin: glyceride fracFigure 10. Plan V i e w , from A b o v e , of Leaf Condenser Installed tion, four times decrease in potency, 2 5 5 yield of vitamin: difInside Rotor ference effected in concentrations, twenty times. SITOSTEROLS FROM VEGETABLE OILS. This refers to free, 1111esterified sterols only. Average boiling point difference between This scheme can be extendcd further by employing a successterols and fat is 120" C.: sterol fraction, eighty times increase in potency, 90% yield; oil fraction, nine times decrease in potency, sion of stills in which each distillate is passed back to the lOC", yield, difference in concentrations, about seven hundred feed of the previous still or t o the feed of t'he still next-but,-one times. behind ( 8 ) . Fawcett (?) discussed various schemes for grouping 2-ETHYLHEXANOL AND h D I P I C h C I D FROM DI-~-ETHYLHEXTL stills in cascade which employ the kind of sequence used in sepaADIP.ITE. l l o s t of the 2-ethylhexanol and some of the adipic ncid pass to the vacuum pump or intercondenser. Diqtillate rations by recrystallixatjon. Fraser pointed out that' each unit contains the rest of the adipic acid with a little ethyl hexyl diptillation is itself the summation of an infinite series of consecuphthalate; the residue of ethylhexyl adipate is virtually free from tive distillations so that the distillate collecting on the evaporator unreacted constituents, and difference in concentration is indefinear the ent,rance of the sttll differs from that near the exit. Franitely high. ser thus prefers to return the reflux or residues from rcdistillat,ion There are, then, certain crude mat,erials forn-hichstraight niclecto points farther apart than those chosen in the Fawcett pattern. ular distillation is properly suited. Inevitably, there arr many Typical Fawcett and Fraser diagrams are reproduced in Figure other mixtures which require distillation in apparatus having 12, -4and B. Since the area of any st.ill in wries should be apimproved separators pcn-ers. .Isan example of what can be done proximately proportional t o t.he total volume of distilland (feed, by multiple redistillation, we may cite the separation of CY-, y - , residue, and reflux) entering that still, the size of still should eviand &tocopherols which differ by one methyl group each, in a dently diminish as the series or cafcade progresses. This detail total molecular weight of 102-403 units. After the fifttran t o .should receive careful atttmt,ion in any grouping of ccntrifugal >eventeen distillations involved in preparing rertain commercial concentrates, the agregattd almost entirely from the other tn'o. FEED S;~./.DlSTILLAl
1
M E A N S F O R BETTER S E P A R A T I O N S
PARTIAL REFLCX. The simplesr espcdicnt to obtain a separation of the lighter con-
itituc,nt better than one T.1I.P. is to cnrirh tlic feed w i t h a portion of the di9tillatc). The eoncontration of lightrr distillate then builds u p in the still until a nvn- cquilibrium is established between the concent rntioris in the distillate and the residue. Thc thermal hazard, H , t o which the lighter constituent is csposed increases in proportion to the fraction. F , of distillate returned:
H
=
1/(1
W
a be
I
la 0
if 1-5XBLEED d
OFF
SCHEME I
- F)
The concentration of lighter constituent passing to the residue increases with inereask in purity of the distillate so that the intrinsic performance of the still is not improved.by this expedient. ~ I C L T I PREDISTILLATIOS. LE The division of the distillate into collected fract,ion and reflux may be made externally or a special reflux fraction may be secured in the still by use of a multizone condenser.
FI'NAL RESIDUE 5 %
Figure 11.
Flow Sheet for Total Distillation (Dirtilland and Residues Shown as Solid Lines, Distillates as Broken Lines)
I. Residue i s returned with bleed-off. II. Residue is collected and redistilled later. 111. Distilland is distilled substantially to dryness in a cascade of stills of diminishing capacity.
INDUSTRIAL AND ENGINEERING CHEMISTRY
692
Vol. 39, No. 6
4.A
A I
D I S TI LL ATE
I I
f-----
I
1
1
I iI
I
-+ RESIDUE -FINAL+ Figure 12.
DISTILLATE
q
s,
p
sc
]
.1 I
FINAL RESIDUE
Flow Sheet for Fractional Distillation According to Schemes Distilland and residues shown
4s
of Fawcett (A), Fraser ( B ) , and Present Practice (C)
solid lines, distillates as broken liner.
collects on condenser 2, drips off into zone 3, and so on. The dicstills. The diameters in the series shown in Figure 12C are artillate from the last zone represents the purest light fraction obranged in the sequence 2, 1.73, 1, on the assumption that 50% of tainable from the apparatus and is withdrawn by pump to an the distilland entering each unit will be distilled. If the feed conexternal receiver. The reader may wonder whether considerablr sists of equal parts of two components, A and B, it will be supplied splashing will result from the haphazard return of drippings to in two unit volumes together with one volume of residue and one of distillate, making four volumes in all to reach the first still, S1. the rapidly moving rotor. The answer is that the splashing 1' minor except where the original distilland is introduced. Since, Similarly, stills Sr and S d will receive three volumes each, while however, any spray produced a t the top of the still mingles Kith St and Ss handle two volumes apiece and provide one volume each distillate that is to be redistilled loner down, the matter is of litof final products, substantially A and B. The optimum sizes of tle consequence. The important point is that the scheme ha.. stills in the Fraser arrangement are more difficult to calculate, been tried and r\ orks well. Evidently the quantity of distillat t L particularly if components A and B are not present in a 1 :1ratio. coming from the smallest zone must always be equal to or larger The two situations of great practical importance are: (1) wherc the Tvanted component is present in large excess and must be separated from one or more volatile contaminants and a nonvolatile residue; (2) where the wanted component is "i - - I present in small or even trace quantities and must be separated (a) as a light fraction, ( b ) as a light fraction from a multicomponent liq'I I I uid, so that the light fraction must itself be I t further fractionated, or (c) as a heavy or DISTILLATE RES. I heaviest fraction. Situations 2a and b are handled sufficiently well for many commercial purposes by the relatively simple linkages shown for the 5-foot still and associated equipment in Figure 13. Since the thermal hazard is proportional, exponentially to the temperature and linearly with the time of exposure, and since the latter is proportional to the volume of the distilland and, hence, to the area of the still (as rough approximations), it is better to concentrate a FRACTION light fraction by redistillation than by reflux. Hence, scheme 5 is preferred to the others of Figure 13. Multiple redistillation in a single unit may be readily secured in a centrifugal still of the cone type (do). Distilland is admitted to a zone of high temperature part way up the cone, as in Figure 14, whence it travels up+VOLATILE I ward to the exit gutter. A distillate, enriched CONTAMINATION MAIN in lighter fraction, drips off the top condenser FRACTION to a portion of the cone precediug the entrance Figure 13. Five Linkages for Simple Fractionation as Used with 5-Foot Stills of original distilland and begins to climb until it reaches the distilland. .4 second distillate Distilland and residues shown as solid lines, distillates as broken lines.
-*-- - -*-
A@-;
I
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1947
than the quantity of light constituents in the original feed, if this feed is t o be robbed of substantially all this constituent as it passes through the still. This relation calls for different areas of condensers according to circumstances or to regulation of temperature of the evaporating zones or both. If there are S constituents in the original distilland, there must, be amininium of S - 1 passes t,hrough the st,ill, or 5 - 1 successive stills to effect separation of constituents.
693
DlST ILLAT E TO VACUUM
FRACTIONATION
If definitions given earlier are t o be adhered t o closely, “fractionation” cannot be achieved in the molecular still since distillation, being more or less obstructed, ceases to meet the “molecular” concept. Severt,heless,fractionating barriers, placed in the molecular still and operated a t high saturation pressures, do form a valuable means of increasing separations due to a unit nianipulation, 1T-e have esperimented with various kinds of barriers; a simple wire mesh screen (121, hanging between the evaporator and condenser in t.he 5foot. still (Figure IS), is particularly useful because it increases the E.3I.P. two t o four times, according to the rate and hence temperature of operation. The temperature of this barrier is self-regulating, unless special heating or cooling features are added, so that Kith slow rates of evaporation from the rotor, condensation on the gauze is virtually complete; the quantit,y re-evaporating t o the condenser is negligible. The gauze condensate falls into the rotor and provides nearly total reflus. At high rate of evaporation, absolute reflux increases but is overtaken by re-evaporat.ion so that t,he separatory power decreases. Means of employing multiple harriers (22) and recirculating reflus (3)suggest themselves. -4simple and practical measure consists of using one barrier, without return of final reflux t o the top of the barrier. This arrangement is particularly useful when a water-white product is desired, absolutely fret. from entrainment. ABSENCE OF QUANTITATIVE DATA
The author n-ould like t o include a series of generalized formulas for calculating the separatory pon-er of the various arrangements outlined here, if such formula were available. Actually, however, the operat,ing conditions vary so markedly with circumstances and are so definitely under the control and preferencr of individual still operat,ors that, the formulas would be of acadcniic interest only. Kevertheless, sufficient quantities of a t r o component, plate mixture have been prepared for Perry and Fuguitt to make measurements for t,he more standardized arrangements; these are reported in another papc’r of this symposium ( 2 9 ) .
Figure
14. Integral Redistillation by Returning Condensate Directly to Rotating Cone (Diagram of Sectional Elevation)
Distilland and residuesshown as solld lines, distillates as broken liner.
that must service their output and, in c.ertair1 cases, their input. The molecular still can be viewed equally well as the progenitor of a chemical manufacturing business or as the modern adjunct of established chemical industry. We have atkmpted t o illust,rate this interdependence of stills and chemical plant by the flow sheet of Figure 16 which deals with a n imaginary plasticizer, “Super C,” produced by t,he “Hypothetical Chemical Company” ( 6 ) . In premolccular-still days the management’ of our hypothetical firm decided against making Super C because the react,ion A B e C does not go to completion and no ready means then existed for removing unreacted A and B. rZlt’hough C could be distilled in existing vacuum stills, a quantity of tar would result which would decrease yields below profitableoperation, anddisposalof the residue would cause difficulties with thr State River Commission.
+
-GAUZE RACTIONATOR - ROTOR
COMMERCIAL CONSIDERATIONS
The operation which established the feasibility of the molecular still-namely, the extraction of vitamin A esters from fish oilsproved to be deceptively simple. Crude oil entered, the hot evaporator rotated, and relatively high values dripped from the exit. As employed today in the preparation of many products and derivatives from natural and synthesized crudes of high molecular weight, the stil1.q are dn arfcd by the chemical plant
VERTICAL LEAF CON DENSER
Figure 15.
Gauze Fractionator in 5-Foot Still
T A B L EI. OPER~ T I S G REQVIREAIESTS FOR FOUR TYPICAL ;\I.ITERIALS
Material Plasticizer Partly reduced petroleum crude Marine f a t Vegetable fat
Coniplete
1
T o natural limit Partial Stripping cir deodoriz-
2
Ing
2
1
251-255 1000
350
700
250*
800-980 35*
7
150 200-400
1-5
20-200
315
315
693
693
...
.
10-100 15; 693*
250* 400*-490
17* 7
040
2 6
4 0
0.46-0.56 5.0
l 1-1.4 38.0
1.0-1.2 150
1.1
1.7
0.17
694
Vol. 39, No. 6
INDUSTRIAL AND ENGINEERING CHEMISTRY
The situation now is entirely different. The initial reaction can be carried to completion by continual removal and recycle of unreacted constituents; formation of tar is prevented, and virtually all the heavy residues are reclaimed, some of the items finding a higher market price than the main product. Encouraged with this outcome, the research department is examining the residues from other processes which have been accumulated during the past fifteen years, with a view to reworking by molecular distillation. The possibility of replacing existing plasticizers and hydraulic oils by materials of much higher molecular weight is being considered. The Hypothetical Chemical Company is worried about the cost of molecular distillation, but they are reassured by the following considerations: ( a ) Although the capital outlay is large, the throughput is high and the investment per pound of product is low. ( b ) The only operating requirements are heat and power, and the relative cost of these continues to decline. (c) The chief expenditure of labor is in receiving, preparing, and admitting the crude and extracting the products from the distillation assembly; increasing the number of stills in the assembly, as for fractionation, does not seriously increase labor requirements. Leaving the hypothetical case for concrete facts and figures, obviously the operating costs nil1 vary with circumstances and must be confidential in any specific case. K e are, however, able to offer data in Table I from which the individual can make a tentative cost sheet for proposed operations. The data cover four representative classes of material, plasticizers, mineral oil, and marine and vegetable oils. as they would be handled in one or two 5-foot molecular stills.
LITERATURE CITED
(1) Brown, G . P., Rev.Sci. Instruments. 16, 316-18 (1945). (2) Burch, C. K.,Proc. Roy. SOC.(London), A123, 271 (1929). (3) Cox, H. L., and Plewes, A. C., U. 8. Patent 2,310,399 (Feb. 9. 1943). (4) Distillation Products, Inc., commercial literature for Myrane
pump fluids.
(5) I b i d . , for Octoil and Octoil-S. (6) Embree, K.D., Oil & Soap, 23, No. 10, 305-10 (1946). (7) Fawcett, E. W., and iMcCowen, J. L., U. S. Patent 2,073,202 (Mar. 9, 1937). (8) Fraser, R. G. J., I b i d . , 2,128,223 (Aug. 30, 1938). (9) Hickman, K., Am. Scientist, 33, 219 (1945). (10) Hickman. K.. Chem. Products, 9, 25-30, esp. Fig. 2 b (1946). (11) Hickman, K., Chem. Revs., 34, 70 (1944). (12) Ibid., 34, 78.(1944). (13) Hickman, K., J . Applied Phys., 11, 311 (1940). (14) Hickman, K., U. 9. Patent 2,080,421 (May 18, 1937). (15j Ibid., 2,150,685 (Mar. 14, 1939). (16) I b i d . , 2,165,378 (July 11, 1939). (17) I b i d . , 2,199,994 (May 7, 1940). (18) I b i d . , 2,210,928 (-4ug. 13, 1940). (19) I b i d . , 2,218,240 (Oct. 15, 1940). (20) I b i d . , 2,234,168 (41ar. 11, 1941). (21) Ibid., 2,308,006 (Jan. 12, 1943). (22) I b i d . , 2,313,548 (Mar. 9, 1943). ( 2 3 ) Hickman, K., and Hecker, J. C., Ibid.. 2,180,052 ( S o v . 14, 1939). (24) Hickman, K., and Kuipers, G. A,, Ibid., 2,379,436 (July 3, 1945). (25) Kapff, S. F., Science, 104, 274-5 (1946). (26) Langmuir, I., Phys. Rev., 8, 149 (1916). [Most calculations of rate of distillation as used in references ( 9 ) and ( 2 3 ,for ~
instance, are derived from Langnluir's equation. See also Hickman, IC.. Chem. Reo., 34, 52, 75 (19441.1 ( 2 7 ) Sational Itesearch Corp., commercial literature for Narcoil fluids. (28) Olive, T . li.. Chem. & Met. Eng., 51, 100-4 (dug. 1944). (29) Perry, E. S.,and Fuguitt, li., ISD. ENG.CHEM.,39, 782 (1947). PRESEXTED as part of the Thirteenth .Innual Chemical Engineering Symposium of the Dirision of Industrial and Enpineerlng Chemistry, -k&lERICAN CHzwcaL FOCIETT. Communication S o . 109 irom the Laboratories of Distillation Prudurr.. Inc.
CHIEF PROOUC
ZCRYSTALS IMPURE Q 3.TRACES SEMI-VOLATILE B - Q TO) p
FILTER
1s
REMOVE Q )MARKETABLE
1
1
FILTER
'tJ
FEED BACK T O RECOVER C
MOLECULARLY D I S T I L L TO LEAVE 8 - P RESIDUE
6
FINAL 1 M A R K E T A B L E TRI TETRA RESIDUE PREMIUM V A L U A
:OMPONENT A
80% PURE
8-Q (MARKETABLE)
Figure 16.
WATER-
MARKETABLE \ L O W VALUE )
Flow Sheet of Molecular Distillation and Chemical Plant Processes of a Hypothetical H e a v y Chemical Manufacture
The main product is Distilled C.
Another main product i s redirtilled water-white Super C. Unreacted constituents A and B are purified and returned to the leacting kettle. impurity, Q, i s rejected, nnd polymers of C are recovered in distilled marketable form.
An
