ci
I
H. C. VAN NESS, 0.E. DWYER2, and W. W. SLOBBE*
University of Rochester, Rochester,
N. Y.
Simultaneous Stripping of Styrene and Methanol from Synthetic Rubber Latex Batch still data are useful in predicting changes in capacity of continuous plant scale equipment for making cold rubber
IN
THE MANUFACTURE of “cold” rubber, polymerization temperatures below the freezing point of water are encountered, and because the reacting charge consists of a water emulsion, an antifreeze must be added. Methanol is satisfactory for this purpose. As the polymerization reaction is not carried to completion, a residual amount of each monomer remains in the latex emulsion, as does all of the methanol added. The latex used was of the GR-S type-Le., the monomers were butadiene and styrene. Butadiene has a boiling point of only -4.4’ C., and is easily removed from the final reaction mixture by venting a t about 40’ C. a t a pressure of 5 inches of mercury absolute. Most of the unreacted styrene and a large part of the methanol, with boiling points of 145.2’ and 64.7’ C., respectively, remain behind, and are subsequently stripped from the latex with steam. I t was with this phase of the recovery process that this work was concerned. The latex used was prepared by the Government Synthetic Rubber Laboratories, University of Akron (Table I).
Present address, Department of Chemical Engineering, Rensselaer Polytechnic Institute, Troy, N. Y . Present address, Brookhaven National Laboratory, Upton, L. I., N. Y . a Present address, Sylvania Electric Products, Inc., Seneca Falls, N. Y.
I t contained G.OY0 unreacted styrene and essentially no butadiene, as this monomer had been removed. Methanol had not been used in the preparation of the latex, and was added in the amount desired for each run. I n this way, methanol concentration could be controlled. As methanol is negligibly soluble in the polymer, the point at which it is added should be immaterial.
Theory The object of this work was to determine rates of removal of styrene and methanol from GR-S latex by steam stripping and to correlate the results for use in the design and operation of plant-scale equipment in the manufacture of cold rubber. A plant-scale stripping column operates continuously; but for gathering rate data, a batch still
Table I. Latex Data Constituent Polymer Soap Inorganic solids Water Styrene
% 25.3 1.8
0.2
66.7 6.0
100.0
W*lW P % solids Density at 27’ C., grams per cc.
0.238
27.3
0.969
was believed to be more convenient. The relationship between batch and continuous operation was developed as follows. Although the presence of methanol influences the operation, removal of styrene is the major problem, for in stripping the styrene, the methanol is essentially completely removed. Thus, the methanol concentration is best ,treated simply as one of the variables affecting the rate of removal of styrene. Physical Picture of Batch Stripping Process. The latex charged to the still is an aqueous emulsion of polymer particles containing dissolved monomer. The methanol is dissolved in the aqueous phase. Steam blown through the mixture removes the styrene from the solid polymer particles and the methanol from the aqueous phase. The transfer of styrene from the solid polymer particles to the vapor bubbles rising through the charge is obviously a diffusional process, but a compjicated one. Presumably, a concentration gradient is set u p within the polymer particles themselves, but this is not taken into account in the mathematical development which follows. From the surface of the polymer particles to the main body of the vapor stream, the usual film theory may be applied, but uncertainty exists as to what area should be used in the rate equations. There would seem to be two limiting possibilities-the area V O L 49, NO. 8
AUGUST 1957
1271
&
STEAM TRAP
8
S T W SOW(0E
0
J
----
j
MAHOMETER
T O VACUUM
Figure 1.
1000 cc. of latex was charged to the still in this setup of equipment
of the polymer particles themselves and the area of the vapor bubbles. As no estimate of the vapor area was possible, the area of the polymer particles was of necessity used. For a differential height of the batch still, the diffusion equation may be written :
where n, is the molal rate of flow of styrene in the kapor stream at height h in the batch still, k is a rate coefficient, a is the effective area across which dif-
fusion takes place per unit volume of charge, ps* and p a are the equilibrium and actual pressures of styrene in the vapor, and S is the cross-sectional area of the still. The problem is to integrate Equation 1 over the height of the charge in the still, to obtain the total rate of removal of styrene from the latex at any time. I t is probably a good assumption that, because of the high degree of turbulence in the still, the composition of the liquid phase is uniform from top to bottom of the charge. At the same time, the concentration of styrene and methanol
in the vapor bubbles continuously increases as the bubbles rise through the charge. The variation of must therefore be taken into account in the integration of Equation 1. At any point in the still, by Dalton's law
where P is the total pressure and n*> and n, are the water and methanol rates, respectively, in the vapor stream. By an energy balance, assuming adiabatic operation and negligible sensible-heta effects and heats of mixing:
Table II.
Time, Min. O?Ssd. Corr. 0.0 3.3 5.7 9.0 12.9 17.3 22.2 27.0 32.3 38.3 44.5 50.9 57.5 63.9
70.5 77.0 83.25
1272
0.0 5.0 7.4 10.7 14.6 19.0 23.9 28.7 34.0 40.0 46.2 52.6 59.2
Still Temp.,
c.
80.6 87.5 88.6 90.0 91.0 93.2 94.0 96.5 97.6 98.6 99.2 99.7 99.9 100.1 100.2 100.5 100.6
Sample Experimental Data
Cumulative Condensate Styrene Water, iMethano1 Gram gram Gram cc. mole moles Grams moles 0.0 8.2 17.9 27.2 35.8 42.9 49.0 53.4 56.8 58.9 60.3 61.3 61.9 62.2 62.4 62.55 62.56
0.000 0.071 0.155 0.24 0.31 0.37 0.43 0.46 0.49 0.51
INDUSTRIAL AND ENGINEERING CHEMISTRY
0.000 '1.071 2.16 3.39 4.69 5.89 7.44 9.25 11.31 13.65
0.0 21.3 40.5 57.6 74.5 94.4 109.9 122.9 132.9 140 3 149.8 152.3 154.1 155.4 I
0 * 000 0.665 1.265 1.800 2.33 2.95 3.43 3.84 4.15 4.38 4.56 4.67 4.76 4.82 4.85
Calcd. Cum. Steam Feed,
Gram Moles 0.00 1.67 3.33 5.07 6.86 8.62 10.62 12.79 15.12 17.67
Styrene, V,, cc. 64.8 54.45 44.75 35.45 26.85 19.75 13.65 9.25 5.85 3.75 2.35 1.35 0.75 0.45 0.25 0.10 0.00
Latex Methanol, W,, grams 157.5 134.1 114.9 97.8 80.9 61 .O 45.5 32.5 22.5 15.1 9.4 5.6 3.1 1.3 0.0
COLD RUBBER M A K I N G
where R, is the stream rate fed to the still and L,, L,, and L, are the latent heats of styrene, methanol, and water, respectively. Elimination of n, between Equations 2 and 3 gives nap
ps =
R,
+ n,(l - L,/L,) + n,(l
- L,/L,)
(4) Because the molal latent heats of styrene and methanol are nearly equal to that of water, the last two terms of the denominator in Equation 4 are negligible compared with R, and may be dropped. Further tending to reduce the importance of these terms is the fact that n, and n, are small compared with R,. Although the latent heats vary with temperature, the ratios change little for the range of temperatures covered in this work. Consider then the case for a temperature of 100° C. L, is 9640, L , is 7720, and L, is 9710 cal. per gram mole. Hence, L,/Lw is 0.994 and L J L , is 0.795, and Equation 4 becomes $8
=
nJ'
R,
+ O.OObn, + 0.205n,
=
n,P/R,
Figure 2.
(OBSERVED), MINUTES
Plot for determining average steam rate and zero time. Average
RW = 0.441 gram-mole per minute latex emulsion, and this area in turn should be proportional for a given latex to the two thirds power of the volume of the solid particles. The solid particles contain both polymer and the monomer, styrene. Assuming additivity of volumes (9)
where k' is a proportionality constant and V, and V, are the volumes of polymer and styrene in the charge, respectively. Combining Equations 8 and 9 and setting kk' equal to X
(5)
The second term of the denominator of Equation 5 is seen to be negligible compared with R,. Although the third term is not so small as might be wished, it is expedient to neglect it, as there is no convenient way to express n, in terms of n,. Thus, for the relation between p s and n, in the integration of Equation 1, we have the approximate expression
p.
e -TIME
(6)
Equations 1 and 6 are now combined to form an equation which can be integrated to give an expression for the rate of removal of styrene from the entire charge. After reduction to its simplest form, this equation becomes
where (p.* - ps)l,m,is the logarithmic mean of the driving forces at the top and bottom of the charge. As n,, and pa, are zero and Sh equals V, the volume of the charge, this equation further reduces to
This equation gives the instantaneous rate of styrene removal from the entire charge a t any time. The total volume of the latex did not change appreciably during a run. For lack of a better procedure, the area term, a, is assumed proportional to the area of the solid particles in the
This equation was used to calculate K from the experimental data, and values of X were then correlated with the operating variables.
kets and stoppers used on the still were made of neoprene rubber. Before the start of a run, 1000 CC. of latex were charged to the still. Methanol was then added to give the desired methanol concentration. The still jacket was heated to a predetermined temperature by passing current through the Nichrome windings. The temperature in the jacket air space was controlled throughout the runs, so that heat losses would be negligible. Heating of the jacket and the lines leading from the still was controlled by the Variacs, P. During the heating-up period, the desired steam flow was established by setting the needle valve to give the proper reading on the orifice meter. Steam was not admitted to the still until it had reached the desired temperature. When all other conditions
Apparatus and Procedure With reference to Figure 1, which shows the apparatus assembly, the steam feed line to the still contained a needle valve, D, an orifice flowmeter, E, and a trap, F, and ended in a three-hole distributer inside the still, J. The orifice meter was not used to measure the steam flow, but only as an aid to regulation. The vapor take-off, K, the line to manometer 0, and the still casing were wound with Nichrome wire and insulation, so that they could be heated electrically to prevent vapor condensation therein. The condensers, L, were cooled with ice water. The apparatus included a vacuum system so that runs could be made at subatmospheric pressures. This system consisted of a vacuum pump, drying tube U, manostat T , surge tank S, and cold trap R, and was connected to the apparatus through lines N . The still proper consisted of a borosilicate glass cylinder 43/4 inches in inside diameter, l / 4 inch thick, and 24
60
W
c
4 z
20
5-
$ 4I
>* 3-
2--
VOL. 49, NO. 8
/ AUGUST 1957
1273
perimental data. The calculations are illustrated for a run made under the following conditions. Latex charge. 1000 cc. c?r 975 grams Methanol added. 157.5 grams Still pressure. 765 mm. of mercury Average steam rate to still. 0.441 grammole per minute The experimental data for this run are given in Table 11. [All original data may be found in a technical report to the Reconstruction Finance Gorp. ( Z ) . ] Most of these data were not observed directly, but were calculated by straightforward methods from observed values. The cumulative steam feed was calculated by Equation 3, except that cumulative quantities were used rather than rates. After substitution of the values of the latent heats, Equation 3 becomes
w E:
2
+
Cum R, = Cum n, 0.994 (Cum
I
I
‘0 Figure 4.
IO
20 30 40 50 60 -TIME (CORRECTED), MINUTES
Plot for determining rate of methanol removal
The end of a run was denoted by the absence of both styrene and methanol in the distillate. The time required for making a run ranged u p to several hours, depending on conditions.
were satisfactory, the run was started by manipulating the clamps, Q . T o take data, zero time was recorded as the instant when the first drop of condensate appeared in buret iM. The condensate separated into two phases-styrene and the water-methanol solution. Both phases were measured in the buret, and were withdrawn periodically. The concentration of the methanol in the water phase was determined by calibration against refractive index, measured with an Abbe refractometer.
Calculations
The object of the calculations was to obtain instantaneous values of K in Equation 10 at various times throughout each run. All other quantities in Equation 10 were determined from the ex-
IO0
98
96
94
92
!ti 90 Q E
u a 88
e
-I
86 I
I
0
1
1
I
I
50 60 V, -STYRENE REMAINING IN LATEX, cc.
IO
20
Figure 5.
1274
70
30
40
Plot of still temperatures
INDUSTRIAL AND ENGINEERING CHEMISTRY
a,)
+
0.795 (Cum n,)
(3a)
I t was necessary to make a correction to the observed time recorded during the run because of uncertainties in determining the exact zero of time due to the nature of the experiment. This was done by plotting the calculated cumulative steam feed against time (Figure 2). The best straight line (steam rate was constant) drawn through the points intersects the abscissa at a value of - 1.7 minutes. Consequently, the observed time readings were increased by 1.7 minutes to give corrected time values, which were used in subsequent calculations. The slope of the line was taken to be the average steam rate. The last two columns of Table I1 give the amounts of styrene and methanol remaining in the latex. In this run, 62.65 cc. of styrene were collected, although 64.8 cc. were known to be present in the original charge. For the methanol, 155.4 grams were collected, although 157.5 grams were charged. For calculation purposes, the amounts of styrene and methanol remaining in the latex a t any time were determined by subtracting the amounts collected from the total amounts collected rather than from the total amounts charged. The values so calculated will probably be slightly in error for the early part of the run, but conditions during this period are somewhat uncertain, owing to operational irregularities in getting the run under way. Usually, the data taken during the first few minutes were not used in calculating the final results. The data of Table 11 were first smoothed by making several graphs. Figure 3 shows a plot of V,, the volume of styrene remaining in the latex, us. time. A semilogarithmic graph was made simply for convenience. Values of n,% were determined by taking slopes of this curve. As the slope is
COLD RUBBER M A K I N G
I where p a is the density of styrene and M , is its molecular weight. Figure 4 is the corresponding graph for methanol. Values of n,, were obtained by a procedure analogous to that used to get styrene rates. The partial pressure of styrene in the vapor at the top of the charge, p,,, was calculated by Equation 5 . For determination of pa*, it was necessary to have accurate values for the temperature of the charge in the still. Therefore, the observed temperatures were smoothed, as shown in Figure 5. Values of P8,the vapor pressure of styrene, were found a t the smoothed temperatures from the data of Stull (3),and values of ps*/Ps were taken from the data of Dwyer and Gleich (7). From these data, appropriate values of p.* were readily calculated. Values of V, and V, for use in Equation 10 were determined by subtracting the volume of the water and the volume of styrene in the latex from the volume of latex charged to the still. This difference was the volume of polymer present in the still throughout the run as measured at room temperature. The volume of styrene remaining in the latex at any time as measured a t room temperature is listed in Table 11. Addition of the styrene volumes to the volume of V, as polymer gave values for V, measured at room temperature. In view of the previous assumptions, taking into account density changes due to temperature seemed unnecessary. The quantities needed for the calculation of X vhlues for the run illustrated are given as a function of time (corrected) in Table 111. In addition, values of the concentration of styrene in the polymer particles, W,/ Wp, are listed, as they are needed for the final correlation. W, represents the weight in grams of styrene remaining in the latex at any time, and W, represents the weight of
+
I
8, min. W,, grams n,,, mole/min. nml,mole/min. paarmm. H g WdW, PS*lP8 T , C.
Pa,mm. Hg pa*, mm. Hg c v p
+
+
VJ2’3
(V P Vs)2’3Ps, In [P,*/(P,* - P I< x 106
~ I
I
1
I
I
Figure 6.
1
I
I
I
I
I
Effect of styrene content on K
values of X , increase appreciably with styrene and methanol concentration. The proportional dependence of K on the steam rate is presumably due to the combined effects of agitation and latexsteam interfacial area. The apparent importance of the latter effect wouId indicate that latex-steam area would be preferable to polymer area in Equation 8. As polymer area is easy to determine and is not dependent on the rate of the stripping operation, it was decided to use‘polymer area in developing Equation 10 and to let the effect of variable latex-steam interfacial area be included in the rate coefficient. The dependence of ET on styrene concentration is undoubtedly due to the fact that a concentration gradient is set u p in the polymer particles during the stripping operation. The equilibrium partial pressure, pa*, was based on the average styrene concentration in the polymer. At time zero, this gave the true A& driving force across the water and steam films, but at all later times it gave a Ap greater than the actual Ap across these two films. It was neces-
polymer in the latex charged to the still. Values of K were calculated from the data of Table I11 in accordance with Equation 10. Values were not determined for the early part of the runs because of uncertainty in the data.
Discussion of Results Under the conditions of the study,
K is a function of steam rate, methanol concentration, and styrene concentration in the polymer particles. The effect of styrene concentration in the run for which sample calculations are shown is illustrated in Figure 6. X increases regularly with increasing styrene concentration. Figure 7 shows graphically an example of the final results for the case of an initial methanol concentration such that W,/W, = 0.079. X is directly proportional to steam rate, other things being equal. The complete results are given in Table IV in the form of proportionality factors relating K and steam rate. These factors, and therefore the
Table ill.
vs, cc.
1
Summary of Sample Calculations
60 1.3 148 0.0304 0.0197 48.3 0.218 0.556 86.5 118 65.7 47.9
50 4... 4 131 0.0262 0.184 41.8 0.182 0.501 87.8 124 62.2 46.9
40 ~. 8.0 111 0.0220 0.167 35.2 0.146 0.435 89.3 130 56.6 45.9
30 12.5 88 0.0173 0.144 28.2 0.109 0.354 91.0 139 49.2 44.9
20 18.4 64 0.0122 0.115 20.1 0.073 0.255 93.1 150 38.2 43.9
1740
1760
1660
1490
1180
730
13.3
11.6
10.3
9.2
~
VOL. 49, NO. 8
10 27.9 35 0.0067 0.072 11.1 0.036 0.138 96.0 168 23.2 42.9
AUGUST 1957 .
1275
’
20
-
18 Ws / W e 16
14
0 x
12 IO
2
8
6
and methanol concentration on the rate of removal of unreacted styrene from a cold-rubber latex. This information should be useful in predicting changes in capacity of plant-scale equipment. Before it can be used for the design of commercial stripping columns, a minimum amount of data must be available on operation of a commercial-size plate; the K values reported here are absolutely applicable only for the still that was used to obtain the data, but they should be applicable on a relative basis for any still.
4 Nomenclature
2 interfacial area across which diffusion occurs, sq. cm. per cc. of latex h = height in still, cm. k = rate coefficient, defined by Equation l k’ = proportionality constant, defined by Equation 9 R = rate coe&icient, defined by Equation 10 L = latent heat of vaporization, cal. per gram-mole n = molal flow rate in vapor stream, gram-moles per minute p s = actual partial pressure of styrene in vapor, mm. of mercury pa* = equilibrium partial pressure of styrene in vapor, mm. of mercury P = total pressure, mm. of mercury P, = vapor pressure of styrene, mm. of mercury R, = steam rate to still, gram-moles per minute S = cross-sectional area of still, sq. cm. V = volume of latex in still, cc. V , = volume of monomer-free polymer in still, cc. V, = volume of styrene in latex, cc. WRL= weight of methanol in latex, grams W , = weight of polymer in latex, grams W , = weight of styrene in latex, grams W , = weight of water in latex, grams 0 = time, minutes a
0 0
0.1
0.2
.
0.3
0.4
R, - STEAM RATE, MQLE/MINUTE Figure 7.
0.5
Effect of steam rate on
W m / W , = 0.079 a t zero time.
sary to be more practical than accurate and to define the driving force as (p,* - p a ) , where p8* was evaluated at the average styrene concentration in the polymer. In effect, this adds increasing polymer resistance to those of the water and steam films as the stripping operation ensues-hence the drop in K with time. I n general, the greater the methanol concentration, the greater the calue of IC. This is presumably due to the fact that the presence of methanol produces a larger latex-steam* interfacial area. The molal latent heat of methanol is only about 80% that of water, which means that the vapor bubbles increase in size as they rise through the latex. The increase in vapor rate due to the presence of methanol may also somewhat increase the degree of turbulence. The values of Kfor the case of W,/ W , = 0.480, the highest methanol concentration studied, were a bit lower than those for the case where W,/W, = 0.348. This reversal effect is apparently due to excessive foaming of the latex at the highest methanol concentration. From a stripping standpoint, foamed latex is immobilized latex. Because zero methanol concentration is approached at or before the time the styrene concentration becomes negligible, extrapolation of K values to zero styrene concentration will give values for
0,6
0.7
K
P = 1 atm.
zero methanol concentration as well. In other words, the charge in the still is the same for all runs as the end of the run is approached. Therefore, values of IT obtained by extrapolation to zero concentration, as shown in Figure 6 , would be expected to result in the same IC values for all runs at the same steam rate. That this is true is shown in Table IV. Most of the runs were made with the still at atmospheric pressure. A few runs made at a still pressure of 500 mm. of mercury showed that a reduction in pressure caused an increase in K, no doubt the result of greater bubble size. However, the increase in K was more than offset by the decrease in driving force caused by lower operating temperatures, and the net result was a decreased rate of stripping. The results of these runs are not included here. Little attention has been given in this report to the removal of methanol, because styrene is the more difficult component to remove. The mechanism of methanol removal would be the same as in an ordinary stripping column for the removal of methanol from a water-methanol solution. Actually. the concentration of methanol in the vapor leaving the still was very nearly the equilibrium value. The work described shows the effects of steam rate, styrene concentration,
=
SUBSCRIPTS
m s LG’
1 2
methanol styrene = water = bottom of still = top of still = =
Liferature Cited (1) Dwyer, 0. E., Gleich, W. A , , Ckem. Eng. Progr. Symposium St7.>No. 2, 48, 80 11952). ,( 2 ) Dwyer, 0. E., Slobbe, W. W., Van - - I
Table IV.
1276
Values of a in Relation K = CYR, W,/W, at Zero Time _
_
WJWP
0.079
0.242
0 348
0 480
0.00 0.05 0.10 0.15 0.20
19.6 20.8 22.4 24.2 26.0
19.6 21.8 24.8 28.0 30.7
19.6 26.3 31.0 34.2 37.2
19.6 24.8 28.9 32.1 34.5
INDUSTRIAL AND ENGINEERING CHEMISTRY
hTess, H. C., “Simultaneous Removal of Styrene and Methanol from Synthetic Rubber Latex,” Technical Report to Reconstruction Finance ~ Gorp., Office of Rubber Reserve, University of Rochester, 1949. ( 3 ) Stull, D. R., IND.EKG.CHEM.39, 517 (1947).
RECEIVED for review September 29, 1956 ACCEPTED January 2, 1957