INDUSTRIAL A N D ENGINEERING CHEMISTRY
1682
train 2 converter. A summary of the operating and analytical data is given in Table 111. During the 30-hour period required to stabilize the catalyst, the temperature rose from 100" F. t o a maximum of 222" F. and then dropped to 98" F. The total oxygen consumed by the catalyst was 124 pounds, or 2.0% of the oxygen present in the catalyst when it had been charged into the converter. When the catalyst basket in train 1 was opened, samples of catalyst taken from the surface of the bed were found to be nonpyrophoric. However, when the heat exchanger tubes embedded in the catalyst were pulled out there was some evidence of pyrosis of the catalyst. This was attributed to breakage of the catalyst and its exposure to air. Although there was some burning of the catalyst, it was subsequently removed from the converter without difficulty, and the stabilization was considered to be of benefit in the removal operation.
Vol. 45, No. 8
of the catalyst. However, this was not determined on a large scale a t the TVA plant because the activity of the catalyst had been impaired before it was stabilized. Nevertheless, stabilization of the spent catalyst served the useful purpose of rendering the operation of removing the catalyst from the converter less hazardous than it would have been with active, pyrophoric catalyst. Acknowledgment
The authors wish to acknowledge the aid of E. J. O'Brien and J. L. Snyder in helping to adapt the procedure to the ammonia plant operations. G. D. Barnes, F. E. Lancaster, M. K. iClcConnell, H. Nash, and W. H. Stowers assisted in carrying out the activity tests and in the development of the procedure. E. D. Frederick assisted in the preparation of this paper, and A. Hastings assisted in the design and installation of the test equipment.
Conclusions
literature Cited
The small scale experiments demonstrated that ammonia synthesis catalyst can be stabilized by exposing the active catalyst to a gas mixture containing a small percentage of air. This procedure is described in the literature, and it was adapted to the large ammonia plant converters. At the two large converters, space velocities of about 300 were employed and the times required t o stabilize the catalyst were 24 and 30 hours, respectively. Greater space velocities might be employed with a corresponding decrease in time required for stabilization. Apparently the stabilization process does not affect the activity
L., Pole, G. R., Beinlich, A. W., Jr., and Thompson, H. L.. Chem. Eng. Progr., 43, 291-302 (1947). (2) Hein, L. B., Ibid., 48,412-18 (1952). (3) Khrizman, I. A., Ber. Inst. Phys. Chem., Akad. Wiss. Ukr. S. 8.R.,
a a a
(1) Bridger, G.
12, 15-20 (1940). (4) Miller, A. M., and Junkins, J. h-.,Chem. &: N e t . Eng., 50, 119-25 (1943 ). (5) Temkin, M. T., and Pyzhev, V. M., Rnss. Patent No. 33544/332, 137 (April 30, 1945). ( 6 ) Tour, R. S.,ILD. ENG.CHEM.,13, 298-300 (1921). RECEIVED for review December 12. 1952. ACCEPTED April 15, 1963. Presented before the Southwide Chemical Conference at Wilson Dam, Ala., October 20, 1951.
Cresol Separation by Fractional Solvent Extraction
a
a
Countercurrent Contacting Apparatus EDGAR L. COMPERE'
AND
ADA RYLAND2
Louisiana Sfafe Universify, Bafon Rouge, l a .
U
SE of liquid-liquid extraction for the separation of mate-
rials has been divided ( 3 ) into two general categories: extraction for removal and extraction for fractionation. The latter class has two subdivisions. If a single increment, or a small limited number of increments, of feed is added to the counterflowing solvents, the process is called countercurrent distribution. If feed is added continuously, or in inultiple increments, so that a steady state is reached in the system and large quantities of material can be processed, the term "continued-feed" fractional solvent extraction is used. I n recent years much interest has been exhibited in the use of countercurrent distribution and continued-feed fractional solvent extraction for the separation of closely related materials. The wide range of applications is well indicated in review articles by Craig ( 2 ) and Treybal (10). I n previous work done in these laboratories on fractional solvent extraction ( 1 ) the batchwise countercurrent pattern shown 1 Present address, Chemistry Division, Oak Ridge National Laboratory, Oak Ridge, Tenn. 2 Present address, Polychemicals Department, Du Pont Experimental Station, Wilmington, Del.
in Figure 1 was developed. This is the same as the double withdrawal pattern of Craig and Craig (3),except that feed increments are added a t each cycle. The illustration shows a five-stage process since each portion of one solvent is contacted five times with the other solvent before emerging as product after the pattern is established. The feed point indicated in the figure is the third, or center, stage, but feed can be added a t any desired stage. The pattern can be expanded to any number of stages and will produce a desired quantity of product a t constant purity once a steady state is reached. The operations required by this procedure were originally performed using a single separatory funnel and a series of flasks, but this involved considerable time, and no separation which required more than five to seven stages could be considered feasible using such methods. I t was desirable to develop apparatus for fractional solvent extraction experiments which could be operated according to the basic pattern and could provide a large number of stages. The all-glass apparatus of Craig and Post ( 4 ) seemed most readily adaptable t o these purposes. A modification of their design has been developed and used in the present work.
August 1953 Table 1.
m
a
1683
INDUSTRIAL AND ENGINEERING CHEMISTRY
Five-Stage Operation Using Extraction Machine
( T = tube; L = light solvent; H = heavy solvent) Position after Operation -1 0 +1 Product Product Operation Feed in center Ti TI LI Ha HI Ta Tz TI Shake settle decant LI La Shift lubes t b right Ha Ha Hi Add L2 Ts T2 TI Shake, settle, decant La L3 LI Feed in center Hz HI Ha Add La T4 Ts Ta TI Shake, settle, decant L2 La Shift tubes t o right Ha Ha HZ L1 Tz T4 Ta Shake, settle, decant Ls L4 LZ Feed in center Ha Ha Ha Bdd L4 Ta Tz Ts T4 Shake, settle, decant La L4 Shift tubes t o right H4 Hs HZ Ha Lz
position, the light solvent flows from the reservoir into E', the entry tube of the next unit. The transfer tube, D. from the reservoir is made with a double bend for ease of adjustment. The side arm tube, F , provides a means of introducing feed as required by the pattern, without disturbing the apparatus. During operation the entry tube and the feed tube are provided with rubber caps t o prevent evaporation of the solvents. The portion of the mixing tube containing the heavy solvent will hold about 30 ml. It has been found unnecessary t o control this dimension too accurately, since glass beads can be used t o adjust the level of the heavy solvent. The use of glass beads also makes it possible to operate with less than 30 ml. of heav solvent, if desired. Light solvent volumes from 10 t o 50 m? can be used conveniently. Il1aintenanc.e of a constant volume of heavy solvent necessitates introducing the feed only in the light solvent. The assembled machine, containing the seventeen tubes present1 available, is shown in Figure 3. The lywood bed, essentia{y a box, 3 feet X 14 X 16 inches witgout top and front, is suspended about a foot above desk level on a Flex% frame mounting. The shaking mechanism consists of a Lightnin stirring motor controlled by a Powerstat and connected t o a cam through a gear box. The cam is attached t o the front underside of the plywood box by a heavy rubber tube. A second rubber tube, connected from the back of the bed to a fixed osition at the desk level, provides a balancing spring force. for 2 or 3 minutes is generally sufficient t o establish equilibrium between the phases, although a somewhat longer time is required if the tubes are nearly full.
Asking
l u b e Arrangement Allows Numerous Extraction Stages This apparatus consists of a series of horizontal glass tubes in which equilibrium can be established between the two solvent phases by shaking. A reservoir above each tube provides for transfer of the light solvent from one tube to the next during the course of the extraction. The tubes are individually mounted on L-shaped oak strips. During the extraction the tubes are placed on a plywood bed so that they may be manipulated simultaneously.
-4sifigle tube is shown in Figure 2. The solvents are placed in the mixing tube, A , which is rocked through an angle of about 20' in either direction by the shaking mechanism, causing thorough mixing of the two phases. After the phases have separated, the bed is tilted through an angle of 90" so that the reservoir, C, is lower than B, the transfer tube from the mixer. The light solvent flows from the mixer into the reservoir. The heavy solvent remains behind since its volume is adjusted t o fill the mixing tube just to the outlet when the apparatus is in the decanting position. As the device is returned t o its normal FEED SOLUTlON
HEAVY 5OLVE N l
1
LIGHT SOLVENT
Figure 2.
J
A. E. C. IN
Figure 1.
Single Tube from Fractional Solvent Extraction Apparatus Mixing tube Transfer tube Reservoir
D. E.
F.
Transfer tube Entry tube Tube for introducing feed
IN
Operating Pattern for Five-Stage Batchwise Fractional Solvent Extraction
The arrangement of the tubes on the bed permits their movement, an operation for which no provision is made in the original Craig apparatus. The movement of the heavy solvent with the tube plus the transfer of the light solvent by decantation allows operation of the machine according to the basic extraction pattern in Figure 1. The machine can also be operated according to any of the Craig patterns (3) with single feed. It has been pointed out that the basic extraction pattern of Figure 1might be produced on the original Craig apparatus by suitably moving feed input and product removal points along the train. The manner in which a three-tube operation is performed is shown in Table I. Any number of tubes can be used, although The positions of the EO far only odd numbers have been used. three tubes on the plywood bed are arbitrarily indicated in Table I as -1, 0, and +l. The center position, 0, is the one in which feed material is added a t each cycle, although the feed
INDUSTRIAL AND ENGINEERING CHEMISTRY
1684
position is not restricted to the center. If more tubes are used, the positions are numbered consecutively plus and minus from the feed tube, which always occupies position 0. This pattern of operation uses t tubes to provide 2t - 1stages. The illustration is, consequently, for a five-stage process. Table I indicates the operations which are perfoimed and the contents of the tubes for each step in the process. The appropriate quantities of heavy solvent, HI, light solvent, L,and feed solution are placed in tube 1, which is put in position 0. Tube 2 containing only the heavy solvent, H p ,is placed in position -1. The apparatus is shaken and the phases allowed to separate. The bed is then tiltedandL1is decanted into tube 2, which ahead)contains H p . The tubes are then moved one position to the right. Tube I, still containing HI, is in position fl; tube 2 holding Hp and Ll is in position 0; and tube 3 containing H 3 is put in position - 1. A fresh portion of light solvent, Lp, is introduced into tube 1 with HI. The positions and contents of the tubes are then as indicated in the second row. The machine is again shaken and tilted, and the light phases are decanted. L1 is transferred into tube 3 with Ha; Lz into tube 2 with Hz; and fresh light solvent, L3, is pipetted into tube 1 with H I . The second volume of feed solution is added to the
Vol. 45, No. 8
center tube. After shaking and settling, decantation produces L, as product a t the left, while tube 1 is withdrawn from the right to collect the heavy product, HI. Tubes 2 and 3 are moved one position to the right, and tube 4 containing HA is put in position -1. The operation is continued as before until the desired number of products has been collected. A cycle is considered completed v i t h the removal of the products. A cycle, consequently, consists of two steps, one of which includes the addition of feed eolution, the other, a shift of the tubes to the right. This parallels precisely the extraction pattern shown in Figure 1. Theoretically, the machine should provide a complete separation of the solvent phases a t each transfer. It has been found, however, that in actual operation there is a slight amount of carry-over involved in its use. That is, each light product collected contains about 1 to 2 ml. of the heavy solvent, vihile each heavy product contains approximately the same amount of light solvent. This effect tends to make the observed results slightly different from the calculated ones, but it does not, in general, lower the degree of separation. Calculations and Extraction Method Used for Cresol Separation
il previous paper ( 1 ) described a graphical calculation for predicting the results of continued-feed fractional solvent extraction experiments under various steady-state operating conditions. The graphical method has the advantage of being applicable to systems for which the distribution ratios are not constant, but it is essentially trial and error, and, consequently, may require considerable time when a number of variations in operation are to be examined. It was considered advantageous to develop an algebraic equation with which the results could be predicted more rapidly. This equation, which is a steady-state solution, is based on equations presented by Scheibel (8). Scheibel's equations assume a constant distribution ratio and no solute injection except a t the feed stage. These equations are: y n -
Figure 3.
Table II.
Seventeen-Tube, 33-Stage Fractional Solvent Apparatus
Separation of m- and p-Cresol b y 33 Stages of Fractional Solvent Extraction Solvent Val., h l l 2 5 35 30
Benzene solution to feed stage, Lf Pure benzene to first stage Lo Aqueous 0.25 M phosphat; buffer t o last stage, H o
-
p Cr eso1
m-Cresol
Concn. in Feed Solution, Moles/L. 0.052 0.050
Av.
Distribution Ratio 0.9: 0 73
Results after 70 Cycles Yield Puritya Observed 0bFeed O u t q u t served Calcd. basis basis 85 89 94 94 Benzene, % para 93 89 74 80 Buffer, % meta moles isomer in phase a yo Purity = moles total solute i n phase moles isomer in phase x 100. % = moles isomer in feed (or output)
Calod. feed basis 89 89
EHn-l
Ex - 1
Y,; X, = Y,/D
The general equation mas developed from these by calculating the concentration of a solute in the feed stage, assuming that the feed stage concentration must be the same whether it is calculated from the light or heavy solvent input end of the sy-stem. The resultant equation is:
(3) This equation represents the ratio of the product concentrations of one solute in the light and heavy solvents in terms of the various operating variables. For use in predicting a separation, two such equations must be used, one for each solute. I t is possible to simplify the equation for use in special caseFfor example, when there is no feed input in the heavy solvent, as with the machine used here, and for no solvent input a t all with the feed. In the latter case a particularly simple result is obtained if the system is center fed, that is, if r = s. I n this case, as indicated by Rometsch ( 7 )and Scheibel (Q), the i e w l t is: (4)
The general equation is strictly applicable only for substances with constant distribution ratios, but i t gives a good approxima-
August 1953
INDUSTRIAL AND ENGINEERING CHEMISTRY
.030
.025
I
W
z
.01?
N
z W
m
r ,010
*.
d
z 0 0
Figure
4.
Distribution of Cresols between Benzene and Aqueous Phosphate Buffer
1685
benzene as a blank. The aqueous phase was acidified, extracted with benzene t o remove the cresols, and the benzene extract analyzed i n this same fashion. All measurements were made on a Beckman Model D U quartz spectrophotometer. Table I1 shows the operating conditions and the calculated and observed results for the 33-stage (seventeen-tube) operation. According to calculations presently being done by the authors and t o work done by Scheibel (9),the approach to a steady state is rather slow for a process involving this many stages. This effect is primarily due to the necessity for building up the equilibrium concentrations in the tubes. To produce the steady state more rapidly, the first nine feed increments were made approximately ten times as concentrated as the values listed, with the same ratio of meta to para isomer. The tenth feed increment and all succeeding ones had the concentrations shown. I n making the calculations, average distribution ratios were determined from the distribution curves and these figures were used in the algebraic equation. Material balance checks based on the output concentrations indicated that a steady state had been reached by the seventieth cycle. The results of the analysis of the seventieth products are shown in Table I1 in terms of the purities of the products and the yields of the isomers in the two phases. The values shown include the solute in the solvent carried over into each product. There is fairly good agreement between the observed and calculated results, with the purity of p-cresol somewhat lower than calculated b u t the yield compensatingly higher, and the reverse for m-cresol. These discrepancies may, in part, be attributed to the solvent carry-over. The over-all separation observed is better than predicted, although it is not symmetrical.
Approximately 25' C.
tion when the distribution ratios are not constant if a n average ratio, calculated over the concentration range to be used, is assumed. Since T and s are equal for the center feed operation, equal yields and purities of two solutes can be obtained from a 50-50 mixture if EL for one solute is made equal t o EH for the other solute. The separation is improved by making the two extraction factors as large as possible. The use of the extraction machine i n a n actual separation has been demonstrated by the fractionation of m- and p-cresol. These isomers provide a good test of the extraction method and, in addition, are of practical interest since distillation fails to yield a high degree of purification of these compounds. D a t a were available on the cresols, primarily from Golumbic and his coworkers (5, 6 ) who investigated the influence of p H on the separations of phenols from coal tar. Their results indicated t h a t benzene and a n aqueous phosphate buffer of p H about 11 would yield a satisfactory separation. Walker (11) also separated mand p-cresol using aqueous sodium hydroxide and benzene as solvents. The cresols used in this study were practical grade Eastman chemicals and were distilled before use. Freezing point estimates of purity indicated t h a t each cresol was approximately 96% pure. The samples were used in the investigation without further purification. The phosphate buffer used as a solvent was made 0.25 M i n phosphate, and the p H was adjusted by additionof sodium hydroxide until a distribution ratio of approximately 1 was obtained for p-cresol using benzene as the countersolvent. The p H of this buffer was read on a Beckman p H meter after dilution 1 t o 100, and was found to be 10.8. Distribution ratios were determined for the cresols between benzene and the 0.25 M phosphate buffer at different concentrations. The results of these determinations are shown in Figure 4. The m- and p-cresols were determined in the benzene phase by measurement of the light absorption a t 280 and 286 mm., using
Machine Provides Versatile Tool for Extraction Studies
It is believed that the extraction machine has great utility as a device for the separation of closely related compounds. It is a versatile instrument since it can be used for single feed operation in any one of the styles recommended by Craig (5), or for multiple feed operation with feed increments being added at any selected stage. For example, the machine can be operated according t o the Craig single withdrawal pattern t o determine the distribution ratios of the materials i n a complex mixture. The separation can then be carried out using the multiple feed pattern so t h a t a quantity of material suitable for further laboratory operations can be separated. The machine can also be expanded t o any desired size by the addition of tubes so t h a t separations which require large numbers of stages can be achieved. The performance of the machine is entirely parallel t o a continuous (discrete-stage) column or mixer-separator unit having the same number of stages as the machine (1). This must be true since the material balance expressions around a stage, in terms of volumes for the multiple transfer machine or i n terms of rates of flow for the continuous units, are the same. This correspondence of performance indicates that the multiple transfer extraction machine should prove useful i n studying the behavior of continuous units under various operating conditions and in predicting or evaluating the suitability of such a unit for a particular separation. The relative ease of adding or removing stages further increases the flexibility of the machine for this purpose. The equation which has been developed provides a rapid method for calculating the results of fractional solvent extraction experiments under various operating conditions. Although it assumes a constant distribution ratio, the calculated results also agree well with those obtained when a n average value is used. The equation should, consequently, prove useful t o those who are interested in fractional solvent extraction as a separation process.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1686 Nomenclature
Y
Vol. 45, No. 8
Literature Cited
= concentration in light solvent
(1) Compere, E. L., a n d Ryland, A., IXD. ENG.CHEM.,43, 239 (1951). (2) Craig, L. C., Anal. Chem., 23, 41 (1951). (3) Craig, L. C., and Craig, D., in Weissberger, A., “Technique of
X = concentration in heavv solvent D = distribution ratio, Y / x EL = extraction factor on light solvent input (heavy product) side of feed. L,D/(H, 4-H i ) E H = extraction factor on’heavy sol&t input (light product) side of feed, H,/(L, Lj)D Lo, Ho = input volumes of pure light and heavy solvent, respectivelv Lf,Hf = volumb of feed solution in light and heavy solvent, respectively T = number of stages to feed stage from light solvent input, including feed stage S = number of stages t o feed stage from heavy solvent input, including feed stage 1 r S s = total number of stages
(7) Rometsoh, R., H e h . Chim. A d a , 33, 184 (1950). (8) Scheibel, E. C., Chem. Eng. Progr., 44, 681 (1948). (9) Scheibel, E. C., IND.ENG.C H E M 43, , 242 (1951). ( I O ) Treybal, R. E., Ibid., 43, 79 (1961). (11) TT‘alker, C. A., Ihid., 42, 1226 (1950).
Subscripts n indicates any stage on heavy solvent input side of feed m indicates an stage on light solvent input side of feed p indicates va?Le in product
RECEIVED for review October 22, 1051. ACCEPTED February 27, 1963. Taken from the dissertation of Ada L. Ryland, submitted in partial fuifillment of the requirements f o r the degree of doctor of philosophy, Louisiana State University, Baton Rouge, La., August 1961.
.+
+
Organic Chemistry,” Vol. 111, Chap. IV, New York, Intelscience Publishers, 1950. (4) Craig, L. C., and Post, O., Anal. Chem., 21, 500 (1949). ( 5 ) Golumbic, C., J . Am. Chem. Soc., 71, 2627 (1949). (6) Golumbic, C., Orchin, M., and Weller, S., Ibid., 71, 2624 (1949).
0
e 0
e
Heat Transfer Coefficients of Pseudo-Plastic Fluids JU CHIN CHU, FRANK BROWN’, AND K. G. BURRIDGE’ Polyfechnic lnrtifofe o f Brooklyn, Brooklyn 2, N. Y.
A
COSSTANTLY increasing number of industrially iinpoi taiit liquids show non-Newtonian flow behavior ( 1 , 9,8, 19-21, 24-27). The viscoaity coefficients of these liquids are dependent on shear conditions. The anonlalous behavior may take vmious forms, but the most significant kind to date from the industIial standpoint is pseudo-plastic behavior. A pseudo-plastic fluid is one whose viscosity has a finite value at zero rate of shear, falling off, as the rate of shear increases, asymptotically to a lower value. Specific illustrations in industrial processes are: emulsion polymerization of butadiene and styrene to GR-S synthetic rubber; emulsion polymerization of vinyl acetate to polyvinyl acetate; solution polymerization of styrene to polystyrene; the compounding of natural and synthetic rubber latices and high polymer latices such as polyvinyl chloride latex; preparation of starch by extraction from potatoes; film casting and dipping processes involving solutions of plastics such as vinyl chloiide-vinylidene chloride copolymers and rubber hydrochloride; the processing of cellulose acetate solutions in the rayon industry; the preparation of nitrocellulose lacquers; and the preparation of adhesives such as polymethyl methacrylate solutions. The object of the present work is to investigate the heat transfer characteristics of pseudo-plastic fluids whose viscosities fall into the medium to low viscosity range-Le., less than 15 centipoises. Pseudo-plastic fluids, such as plastics ( l 7 ) ,which fall into the high viscosity range, are not included in the present work, as they are not in general handled in the standard types of heat transfer equipment used for liquids. The viscosity range covered includes the important fields of synthetic rubber latex and natural rubber latex, most of the vinyl plastic emulsions, and many of the more dilute high polymeric solutions in industrial use. Attempts to widen the range of viscosity too much in the investigation might have led to confu1
Present address, Dunlop Rubber Go., Birmingham, England.
sion if the region where rodlike flow obtained had been infringed upon. This practice of division of results according to the viscosity ranges appears to be in accordance with the practice of many investigators of the flow behavior of Newtonian liquids. The temperature range considered in the investigations is 30 to 80 ’ C. Lower temperatures are usually avoided, if possible, in industrial processes to eliminate the need for refrigeration equipment. Similarly, higher temperatures are not too frequently encountered because the emulsions and high polymer solutions which comprise the majority of the industrially occurring pseudoplastic fluids are either chemically or physically unstable a t higher temperatures. While it is not pertinent to review here all the work on heat transfer characteristics of the Newtonian fluids (6, 7 , 11, 14-16, 62), the fundamentals of the approach to the problem of the heat transfer characteristics of R pseudo-plastic fluid are similar. Applying the concept of the film coefficient of heat transfer to a fluid flowing in a tube (2, 12, 23), the following relation holds: O
In practice it is always found that the temperature difference across the fluid film varies along the length of the tube. The log mean temperature difference can be used in Equation 1, if the value of the film coefficient of heat transfer is a constant over the range of the change in temperature of the fluid during its passage through the exchanger. Heat transfer data on fluids deviating from simple Newtonian behavior are presented by Winding, Dittman, and Kranich (26), Bonilla ( 4 ) , Hoopes et al. (IO),and hlacLaren and Stair ( I S ) . The data of Winding ( 2 6 ) are for the heating and cooling of pseudo-plastic GR-S latices of several types. They are correlated by a log-log plot of
