Kinetics of Hydrochlorination of lauryl Alcohol
-
development
HENRY A. KINGSLEY, JR.~,AND HARDING BLISS Yale University, New Haven, Cann.
A
MONG the types of heterogeneous reactions carried out in
for flows over t,he range used here by absorbing in sodium hy-
chemical manufacture, one is that in which a gas is brought into intimate contact with a liquid by passage through a body of the liquid. The hydrochlorination of lauryl alcohol t o yield lauryl chloride is representative of this sort of reaction and is of industrial interest also, since lauryl chloride is a raw material for the manufacture of dodecyl mercaptan, a modifier for synthetic rubber. In this case, as in most problems of this sort, the rate of reaction and how it is affected by process variables are of interest. In addition to the more commonly met variables such as pressure, temperature, degree of conversion, and catalyst nature, condition, and amount, there are in these cases gas passage rate, gas purity, and method of dispersion. Also, certain dependent variables as well as the rate of reaction, such as composition of effluent gas, and particularly, the economy of utilization of the gas are of interest. I n short, the aim of this work was t o discover the mechanism by which this particular reaction occurs and to learn the effects of important variables on the appropriate rate constants. The utility of such data in predicting means of accomplishing reasonable conversion rates and high gas utilization a t the same time will be demonstrated.
droxide the hydrogen chloride passed. The calibration was repeated periodically without change. The off-gases from the reactor were sent to an ordinary Liebig condenser, Q, sealed to a 250-ml. receiver, R, and thence t o a bubbler, S, filled with saturated aqueous hydrogen chloride, where the rate of hydrogen chloride t o discharge could be observed. Ordinarily the hydrogen chloride effluent was sent to a water aspirator for disposal. The reactor, E, was a standard borosilicate glass jar of &inch outside diameter and 12-inch height with a wall thickness of about 3/16 inch. The impeller, H , is illustrated in Figure 2. It was modelled after the Arrowhead gas disperser of the Mixing Equipment Co., Rochester, N.Y., and was fabricated of 25 gage. Hastelloy Bsheet by die forming the blades and spot-welding them to the disk. Heat treatment for 20 minutes a t 1150" C. completed the construction The impeller was maintained a t 21/4 inches from the bottom of the jar in use. It was driven by means of the I/r-inch Hastelloy B shaft (G, Figure 1). The baffles, F, consisted of an assembly of four borosilicate glass plates, '/z inch wide, and the depth of the jar which was fused totwo glass rings a t top and bottomfor rigidity. With the agitator operating a t 1800 r.p.m. a minimum liquid depth of 61/2 inches was required t o prevent formation of a deep vortex and the pumping of gas into the liquid. The reactor top, L, was made of '/*-inch steel plate, 10 inches in diameter. It was provided with openings for the impeller shaft, for charging, for gas introduction, for venting, for anchoring the baffles and permitting the introduction of a small tube carrying carbon dioxide in certain runs to minimize fire hazard; and for glass thermowell, J, and sampling tube, K. The latter was big enough to permit the introduction of a pipet for sampling and prevented the escape of hydrogen chloride gas during sampling. Openings for charging, for the glass thermowell, and for the sampling tube were used for insertion of the immersion heaters in order to heat the batch prior t o a run. The reactor top was gasketed and clamped onto the reactor jar with three vertical rods. Protection of the top from corrosion and finding suitable materials for sealing the various holes in the top gave considerable difficulty. No suitable coating for the top was ever found, and instead the top was thermally insulated t o keep it hot and prevept condensation. Despite the ever-present water vapor, in the absence of liquid water, corrosion was not a t all excessive. Neoprene stoppers were quite satisfactory for the holes in the top, although after four or five runs they had swollen considerably. Heating them in an oven t o drive off the lauryl alcohol and chloride restored them t o their original shape fit for re-use. Hycar gasketing material, fabricated by the Connecticut Hard Rubber Co., New Haven, was successful between top and jar
Scope Hydrochlorination of lauryl alcohol was accomplished in batches of from 1.67 to 3.79 liters (depths of 4.5 to 9.5 inches.) The catalyst was zinc chloride from 0 t o 2oJ0, although aluminum chloride was used in one run. Some studies of catalyst life were made. The gas was dispersed either by introduction directly beneath an impeller or by a jet-type mixer in which the gas was pumped by the recirculating li uid mixture. Thus, later references will often be made t o impe8er runs and jet mixer runs. Temperatures of 110°, 130°, 140", and 150' C. were used. The gas rate was varied from 0.13 to 10.2 gram mole per hour per liter of charge. The gas composition was varied from 22 to 100 mole per cent hydrogen chloride, the diluent being steam. The impeller speeds were varied from 850 to 1800 r.p.m. Apparatus
FORIMPELLER RUNS. This is illustrated in Figure 1. Anhydrous hydrogen chloride was delivered from the cylinder, A , superheated in B , and then throttled to 1 atmosphere by the needle valve, C, which was used t c control the rate. The flow rate was measured with the meter, D. Copper tubing and brass fittings were used. The superheater was a simple resistance wire, sufficient to heat the gas t o 40" C. before throttling, and it was very effectivein eliminating the fluctuations in rate experienced before it was installed. The flowmeter was a simple capillary, the pressure drop across which and the downstream pressure of which were measured in manometers filled with saturated aqueous hydrogen chloride. This meter was calibrated 1
Present address, Shell Development Co., San rranciuco, Calif.
2479
INDUSTRIAL AND ENGINEERING CHEMISTRY
2480
for four or five runs. The packing for the stuffing box was l/&xh square Alpax packing. Preheating of the reactor charge mas accomplished by two 2OO-m~att tubular Chromalox immersion heaters. Since the reaction was slightly exothermic, heating during the course of the reaction was not necessary. The heaters were withdrawn before catalyst or hydrogen chloride was introduced. There was no evidence of attack on the copper sheaths of these heaters after all the runs were completed.
i C
Figure 1.
R
Vol. 44, No. 10
standard spherical joint, the inner member being stainless steel threaded to the piping and the outer glass fused to the jet mixer. The jet mixer, U,was a laboratory-size borosilicate glass aspirator purchased from Eck and Krebs Co., New York. The discharge from the jet mixer was accomplished beneath the liquid surface of the reactor. This system was used in t x o ways. First, i t was used only as a means of bringing gas and liquid into intimate contact. I n such a case the feed gas was introduced t o the jet mixer and the off-gas sent to the Condenser and receiver as before. Second, it was used as a true recycling device in which case the second condenser, Y , was required to remove as much water as possible and the stopcock, 2, was opened to permit the return of unreacted hydrogen chloride to the jet. I n this case hydrogen chloride vias fed from the cylinder only to supply that consumed in reaction. The motors to drive the pump and impeller were I/, hp., class I, group D, explosion-proof because the flash point of lauryl alcohol is very near the higher temperatures used here. Procedure
Apparatus for Impeller Runs
I n order to minimize heat loss, the whole reactor jar was inclosed within a sheet metal cylinder, M , (Figure 1) 11 inches in diameter by 13 inches high, lined with asbestos paper and provided with two sight glasses for observation. I n the air space between jar and shell was placed a helical heater, N , wound on a hexagonal form. This was controlled with the thermostat, 0, to maintain the reactor at the desired temperature. An important modification was made to this apparatus for some runs so that hydrogen chloride diluted with water vapor could be the feed gas. -4vaporizer for aqueous hydrochloric acid was made as follon~s: A top preheater section of glass tubing, 12 inches long and with an outside diameter of 17 mm. was connected t o a boiler section of glass tubing 20 inches long and with an outside diameter of 28 mm. and this to a superheater section of tubing 20 inches long and with an outside diameter of 15 mm. All sections were packed with '/(-inch ceramic Berl saddles. The preheater, boiler, and superheater had 130-, 600-, and 230-watt heater windings, respectively, Aqueous hydrochloric acid was fed to the top of this vaporizer through a calibrated meter and the superheated wet acid vapor was taken from the bottom through a heated glass tube to the gas-introduction tube of the reactor. This device was also used merely to boil water to supply steam used to dilute the normal anhydrous hydrogen chloride for some runs when compositions other than that of aqueous hydrochloric acid were desired.
FORJETMIXERRUNS. This apparatus is illustrated in Figure 3. The reactor is essentially the same, although the various holes in the top were used for purposes more appropriate to this type of operation, The impeller and shaft remained, although they are not shown in Figure 3. The liquid was drawn into the pump through a gas-liquid disengaging device, V , which was simply an open glass tube with a diameter big enough to permit gas to separate from the liquid as it passed downward. The pump, T , was a Series 50, Size 2 bronze gear pump made by Oberdorfer Foundries, Syracuse, N. Y . It was capable of delivering 1.1 gallons per minute a t 20 pounds per square inch gage a t 600 r.p.m. The pump discharge pressure was measured in a simple Bourdon gage, W . Brass piping and fittings were used and they were insulated and provided with an electric winding to minimize heat loss. The metal piping was connected to the jet mixer by means of a
FORIMPELLER Ruiw Since it was necessary to remove and clean the reactor top after each run, the first step was to put the top in place and insert the feed line, thermowell, etc. The assembly was given a brief pressure test with carbon dioxide at R few inches of water so that leaks could be detected and eliminated. The required volume of lauryl alcohol was introduced into the reactor and brought up to the desired temperature by the immersion heater with the agitator on. Carbon dioxide was introduced in the vapor space during this period, and the heaters were removed when 2" C. below the proper temperature was reached. A weighed portion of zinc chloride from a freshly broken ampoule was added and final adjustments were made on the thermostat controlling the temperature in the annulus between reactor and shell. Hydrogen chloride flow was then begun and adjustment to the proper rate was made with this gas discharged to the hood. When dissolution of zinc chloride T a s completed, the reactor contents were sampled, the flow of carbon dioxide was stopped, and the hydrogen chloride was mitched to the reactor. Introduction of hydrogen chloride increased the temperature about 2 " C. Samples were withdrawn at intervals for subsequent analysis, and these were usually allowed to cool slowly to room temperature. I n later runs, however, they were chilled quickly by immersion in cold water. The progress of a run could be followed in a semiquantitative manner by observing the volume of water collected in the receiver, R, Figure 1. At high conversion (96%) the catalyst was largely precipitated out, indicating very low solubility in lauryl chloride. At the end of a run the various heaters were turned off, the flow of carbon dioxide was resumed, and the flow of hydrogen chloride stopped. To facilitate disposal of the charge, it was cooled quickly by cautious addition of liquid water with evolution of steam. A simple water-cooled bayonet exchanger was also used in this connection. The apparatus was disassembled and cleaned after each run with particular attention paid to the top, which was steamed, dried, and stored in a warm place. When the vaporizer for aqueous hydrochloric acid or water was used, the liquid flow was started and the heaters were turned on about 20 minutes before the start of the run proper, discharging the gas to exhaust. At the proper time the gas was switched to the reactor. FORJET RUNS. When the jet apparatus was used merely as a device for bringing gas and liquid into contact, the procedure was essentially the same as that for impeller runs. The charge was 1.50 ml. larger, however, to fill the pump and lines and still have the same liquid depth in the reactor as for comparable im-
INDUSTRIAL AND ENGINEERING CHEMISTRY
October 1952
peller runs. When the batch had reached the desired temperature, the pump ( T,Figure 3) was started and hydrogen chloride gas was introduced via the jet instead of by the tube under the impeller. The impeller was operated a t 1300 r.p.m. in these runs.
'
2481
this gas was measured in the differential manometer. This was considered to be the same flow as recycle and make-up as in the run proper. This test was of short duration, and wfen completed valve 2 was opened and hydrogen chloride from supply reduced to its make-up value. Loss of hydrogen chloride in the condensate from the off-gas was minimized by running the first condenser rather warm.
Materials
;i: D R I L L 7
The hydrogen chloride gas was supplied by Matheson Co., East Rutherford, N. J., and was 99.5% pure, containing 0.3% chlorinated hydrocarbons and 0.2% acetylene. The lauryl alcohol was that sold by Du Pont Co. as Lorol with the following approximate analysis : n-Deoanol Yo 2.6 n-Dodecaiol % n-TetradecaAol % n-Hexadeoanol,' % n-Ootadeoanol, 7%
61.0
23.0
11.2 2.2
The average molecular weight was 204. The zinc chloride was granular, anhydrous, reagent grade from Baker and Adamson. It was repackaged in small ampoules t o maintain the anhydrous condition, each ampoule of suitable amount for one run. Aluminum chloride was powdered, analytical reagent grade, anhydrous from Mallinckrodt.
Analysis
i cnlt
I
Figure 2.
t
I
Detail of Blade
When the jet apparatus was to be used as a recycling device, the operation was first as above, and when hydrogen chloride appeared in the gases leaving the second condenser ( Y ,Figure 3), the stopcock Z was opened, allowing the unused hydrogen chloride to be recycled. The flow of hydrogen chloride from the storage cylinder required constant manual adjustment in these runs, so that only a small amount of it escaped through the bubbler. This small purge served to eliminate any inerts which might have leaked into the system. I n order to correlate these data by the methods later described, it was necessary to know the rate of flow of recycle gas. This was measured as follows: When the gas was being recycled properly, the pressure at the inlet to the jet mixer (measured on the static manometer) was about atmospheric. Then the valve, 2,was closed which caused a drop in pressure. Flow of gas from the tank was increased to bring the pressure back up to atmospheric and the flow rate of
The acidity caused by zinc chloride and dissolved hydrogen chloride in the sample was determined by adding a known weight (approximately 2 grams) of sample to a flask containing approximately 25 ml. of distilled water and 10 ml. of toluene and titrating with approximately 0.01875 N sodium hydroxide (freshly made) to phenolphthalein end point. The hydroxyl content of the reaction mixture was determined by a modification of the method of Smith and Bryant (3). After 10 ml. of approximately 1.5 M acetyl chloride in toluene in a 250-ml. glass-stoppered flask was chilled by immersion in crushed ice, about 7 ml. of pyridine in toluene (29% by volume) was added. Upon removal from ice, a known weight (approximately 2 grams) of sample was added and the flask held for 20 minutes a t 60" C. The flask was then cooled and about 25 ml. of distilled water was added to hydrolyze the excess acetyl chloride in 30 minutes. The liberated acetic and hydrochloric acids were titrated with approximately 0.75 N sodium hydroxide to phenolphthalein end point. Blanks were always run. First samples and blanks were run in quadruplicate and subsequent samples in duplicate. The weight fraction alcohol on a catalyst-free basis, P, was computed with the equation
in which VI is the milliliters of base of normality N to neutralize the blank, Vg that for the sample (corrected for the presence of zinc chloride), M is the molecular weight of the alcohol, W is the grams of sample, and f is the weight fraction catalyst. The mole fraction alcohol on the catalyst-free basis (1 z), was computed with the equation
-
(1 - z ) = P +
(PU M
M
- P ) [I - M
+ 18.5 + p [I
It has been estimated that the various experimental errors in this analysis may reach 0.41% on unconverted alcohol and 0.37% on 95% converted alcohol. Results
VARIABLES.A few of the experimental results are presented in Table I. The complete Table I is on B e with the American
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
2482
Table 1.
Run 5 (Impeller)
Stirrer Speed, R.P.M. 1800
25 (Impeller)
29 (Jet)
Temp., 0
c.
Representative Experimental Data and Run Conditions Catalyst Conon.,
%
Lorol Vol., Liters, V 2.56
Gas Rate, G. Moles/ Hr., G 5.89
Q v
Pump Press Lb.)' Ssuare I n . Gage
2.30
..
150
0.99
1800
130
0.99
2.52
8.70
3.46
..
1300
150
1.0
2.73
8.89
3.25
16 11 18
20 23 11 11 31 (Jet recycle)
6
1300
150
Vol. 44, No. 10
0.9Y
8.45" 3.84 3.12 4.55 4.20 4.99 3.18 2.46
2.73
..
10
8 8 13 15 18 8 8
Time,
Iioiirs, t
0 0.25 0.70 1.20 1.70 2.20 3.00 4.50 0 0.75 1.75 2.76 4.10 5.25 6.5 0 0.5 1.0 1.5
2.0 2.5 3.0 0 0.5 1.17 2 .o 2.15 2.5 2.82 3.0
Conversion,
Mole % _____ 0 13.82 34,02 53.92 72.58 87.65 96.29 99.51 0 20.04 40.88 57.93 76.66 89.49 97.55 0 29.18 55.04 73.66 89.43 95.75 98.51 0 22.38 51.60 76.41
0
14.28 34.47 54.29 72.99 87.78 96,69 '$9.60 0 20.28 41.16 58.34 76.99 90.17 98 00 0
29.56 57.49 74.19 89.48 96.09 98.81 0 22.41 51.69 76.54
Sample Aoidity Me./G.,
0.146 0.263 0,257 0.231 0.212 0.186 0.149 0.034 0.147 0.408 0.345 0.315 0.263 0.222 0.149 0.148 0.336 0.286 0 2.52 0.198 0.111 0.050 0.146 0.281 0.278 0,249
70.90
ei.'is
o.'i92
95.39
95.85
0.098
Fresh ga6 rate.
Documentation Institute. While it was realized that this reaction carried out in this way is a very complicated one because of the possible importance of diffusion as me11 as the chemical reaction velocity, nonetheless a preliminary interpretation on the basis of reaction velocity alone was attempted. I t was found that the time-conversion data of all runs at constant conditions could be excellently fitted with an nth order (with respect to lauryl alcohol) irreversible equation. Thus the rate equation is (3)
and the integrated form (4)
The great majority of runs \?"ere of apparent half order, in which case Equation 4 becomes
The irreversibility of this reaction is best shown by run 8 which was made in the usual way except that steam was passed through a lauryl alcohol-lauryl chloride mixture, all conditions being approximately the same as a normal run. The reaction did not reverse, but rather reacted in the fora ard direction a bit more, probably at the expense of some dissolved hydrogen chloride or with some hydrogen chloride resulting from the hydrolysis of zinc chloride, since thi- run was made with the product of run 7. This apparent half-order reaction fits the dataexcellentlyup to 96 to 97% conversion, where it begins to break down, presumably because of the catalyst precipitation. Less catalyst is left in solution, and it will be shown later that the catalyst concentration IS a variable of much importance to the rate.
Although more \vi11 be said later on the matter of order and mechanism, the half-order reaction presents a suitable framework within which to discuss t,he effects of the principal variables. In Tables I1 to VI1 the values of the half-order reaction velocity constant are presented. grouped in such a way t'hat the effects of these variables may be readily seen. Runs 2, 3, 4, and 8 being exploratory in nature are omitted from the treatment to follow, TEWERATURE.The runs in Table I1 show a rather marked temperature dependence of k l l z . These may be treated in the customary way to yield an energy of activation of 16,000 calories per gram mole for series 1; 14,000 for series 2; and 16,000 for series 3. It is probable that the somewhat lower fiLvre at the lowest values of G/V (series 2) point's to a greater importance of diffusion at lorn gas rates, as would be expected. FLOW RATEAND LIQUID VOLUME. The runs in Table I11 show this effect. Runs 12, 5, and 10 at markedly different values of G and V but the same G/V yielded approximately the same
Z Y
R
Figure 3.
Apparatus for Jet M i x e r Runs
INDUSTRIAL AND ENGINEERING CHEMISTRY
October 1952
k1,2 = 0.43 Table 11.
Effect of Temperature on Reaction Constant G, G .
Run No.
Moles/
Hr.
(1% ZnCla catalyst)
V,L.
U/V
TZmz.,
kua
k/C2
Constant gas flow rate; constant weight of Lorol
18 5
5.89 5.89 5.89 5.89
2.48 2.52 2.54 2.56
2.38 2.34 2.32 2.30
110 130 140 150
0.282 0.430 0.671 1.080
0.084 0.41 0.86 2.20
9
U / V ratio constant and equal to 1.56 3.96 2.52 1.57 130 0.374 5.89 3.79 1.55 150 0.869
0.39 2.86
25 11
U / V ratio constant and equal to 3.43 8.70 2.52 3.46 130 0.510 8.70 2.56 3.40 150 1.305
0.41 2.11
23
Table 111. Run NO.
Effect of G / V Ratio on Reaction Constant U,G.
Moles/
Hr.
(1 % ZnClz catalyst)
V, L.
U/V
k/Ca
kil a
2483
+ 64f
(9)
which does not apply i f f < 0.005. CATALYST CONDITION.In order to try to learn more of the mechanism, speculation was made that the reaction might be governed or importantly affected by a reaction between lauryl alcohol and the catalyst. I n run 3 the zinc chloride was charged to the alcohol 10 seconds before the start of the run and in run 4, they were brought together for 47 hours before hydrogen chloride feed was started. The two runs were practically identical. Speculation was also made that the catalyst might undergo change or decomposition during the run. However, Table VI1 shows that the addition of fresh catalyst during the run causes a jump in kl12 to 1.65, whereas 1.67 would be predicted solely on the basis of the increased catalyst weight. Similarly, part B of Table VI1 shows that a run interrupted-i.e. hydrogen chloride passage stopped, cooled, and held overnight-may be reheated and resumed to yield the same value of kl/*. Thus the catalyst seems very stable and no rate-controlling reaction of it with the alcohol seems indicated.
Temperature, 150' C.
12 5 10 1
3.84 5.89 8.70 1.98
1.67 2.56 3.79 2.56
2.30 2.30 2.30 0.77
1.026 1.080 1.068 0.565
2.11 2.20 2.29
9 11
5.89 8.70
3.79 2.56
1.55 3.40
0.869 1.305
2.86 2.11
Inadequate HC1
Temperature, 130' C.
23 6 25
3.96 5.89 8.70
2.52 2.52 2.52
1.57 2.34 3.46
0.374 0.430 0.510
,
0.39 0.41 0.41
k112,showing that the ratio and not the individual members is important. With increasing G / V , kl/Z increased and quantitatively
Table IV.
Effect of
Inlet Concentration of Chloride
(Temperature, 150' C.; Moles/
No.
Hr. V,L. 5.89 2.56 1.0 2.56 0,614 9.6 2.56 0.224 26.2 go
5 17 16
1% ZnCh catalyst)
G, G.
Run
Hydrogen
U/V 2.30 3.76 10.2
* V
2.30 2.30 2.30
ki/r 1.080 0.850 0.453
';'Y'O'/
k/Q
1.080 2.20 1.385 2.24 2.02 1.79
Table V. Effect of Inlet Concentration of Hydrogen Chloride and Ratio of G/V on the Value of kl/:/yo (Temperature, 150' C.; Run No.
Moles/ Hr.
V,.I
1.98
2.56
G/V 0.77
5.89 1.0 3.84 1.0 5.89 1.0 8.70 1.0 8.70 1.0 0.614 9.6 0,224 26.2
3.79 1.67 2.56 3.79 2.56 2.56 2.56
1.55 0.869 0.869 2.30 1.026 1.026 2.30 1.080 1.080 2.30 1.068 1.068 3.40 1.305 1.305 3.76 0.850 1.385 10.2 0.453 2.02
YO
1.0
1
1% ZnCla catalyst)
a, G.
kiia
k/Ca
ki/a/go
0.565 0.565
Ina$pate llUI
Equations 6 and 7 are for reactions at 150' C. and 1%catalyst, and at 130" C. and 1% catalyst, respectively. The dependence of kl/$ on the quantity G/V certainly points to a contribution of diffusion. This is shown INLET HYDROGEN CHLORIDE CONCENTRATION. in Table IV, where it is seen that the effect of low yo's is t o decrease kl/s more than the simultaneous increase in G/V (G being total gas rate) tends to increase it. These runs were made at a constant ratio of hydrogen chloride feed to alcohol volume. The quotient IC112 / yoincreased as G / V increased. INLET CONCENTRATION AND GAS RATE. This is best illustrated in Table V. These runs may be correlated by the equation
for 150" C. and 1% zinc chloride. There is excellent agreement between 0.53 of this equation and 0.55 for Equation 6. CATALYSTCONCENTRATION EFFECT ON APPARENTORDER AND CONSTANT.The data in Table VI show that for catalyst concentrations from 0.52 to 1.97% the apparent order is one half. However, for no catalyst the apparent order is 2.5 with respect to lauryl alcohol, and between 0 and 0.52% the order apparently changes from 2.5 to 0.5. This is an unusual result, but it may well be that one function of the catalyst is to reduce the order as well as to increase the constant. Table VI also shows the effect of catalyst concentration on k1/2 a t 150" C. and constant G/V of 2.30. Over the range of concentrations investigated these data may be correlated by the linear equation
9 12 5 10 11
17 16
Table VI.
2.86 2.11 2.20 2.29 2.11 2.24 1.79
Effect of Catalyst Concentration
(Temperature, 150' C; G / V = 2.30 for all runs) Catalyst Ap arent Concn., OrBer of %, Reaction, Run No. 100f n k/ Ca
14 20 38 13 5 10 12 19-1 21 19-2 16 R u n No.
13 21 12 10 5 19.1 19-2 15
Effect on apparent order, n 0.0 0.0
0.1 0.52 0.99 0.98 0.97 0.99 0.97 1.94 1.97 Catalyst Reaction Concn., %, Constant, 100 f ki/a Effect on value of k 11~
0.52 0.97 0.97 0 98 0 99 0.99 1.94 1.97
0.740 1.036 1.026 1.068 1.080 1.03 1.65 1.67
k/ C1 1.11 2 06 2.11 2.29 2.20 2.24 5.5 6.92
INDUSTRIAL AND ENGINEERING CHEMISTRY
2484
Effect of Other Variables
Table VII.
(Temperature, 150' C.) Adding Catalyst during a R u n Catalyst Concn., %, k/Cz R u n No. 100 f ki/2 19-1 0.99 1.03 2.24 19-2 1.94 1.65 5.5 B. Shutting a Run Down Overnight a t Partial Completionb
A.
R u n No. 21-1 21-2
ki/2
k/cz 2.06 1.85
1.04 1.04
AlCls as a Catalyst n kal z k/C2 2.5 0.0226 0,0860 2.5 0,0234 0.102 2.5 0.0276 0.143 D. Stirrer Speedc Stirrer Speed, R.P.M. kl/2 k/Cz 1.422 0.860 0 2.18 1.032 850 2.20 1.062 1300 2.20 1.080 1800 2.29 1.068 1800 2.11 1.026 1800 2.24 1.03 1800 1.036 1800 Using a J e t Mixer t o Contact the Two Phases
C.
AlClx, % 0.0 0.0 1.0
R u n No. 14 20 27
R u n No 26 24 22 5 10 12 19 21
No. 28 29 30
F. Run
No. 31 32
V,
Hr. L. G/V 1OOf ki/i k/Cz 2.36 0.96 1.10 2.73 2.16 5.89 2.33 1.00 1.27 2.73 3.28 8.89 3.16 1.92 1.85 2.73 2.16 5.89 Recycling HC1 Gas with J e t Mixer Used as a Gas P u m p
V ,L
G / 'y
5.27 10.0 4.3 9.3
2.73 2.73 2.73 2.73
9.3 >9.9 8.35 8.408
3.96 3.96 2.73 2.73
1.93 3.66 1.57 3.40 2.30 2.34 >2.50 3.06 3.08
G
..
33 34
15% reduction in the rate constant. The lack of effect is probably due to the vigorous agitation of the gas bubbles themselves. USE OF THE JETMIXER. The data given in part E of Table VI1 were observed on runs in which the jet mixer was used only as a means of contact, i.e. hydrogen chloride gas was supplied in the full amount as in other runs and not recycled. The values of kl/z are practically the same, as would be predicted from the conditions of G/V, catalyst amount, and temperature from the other runs. While the impeller was operated a t 1300 r.p.m. during these runs, it could have exerted no effect, since the gas was introduced in the jet and not beneath the impeller as in the other runs. It must be concluded, therefore, that the jet mixer is quite comparable to but no better than the impeller as a means of bringing the gas and liquid together. A recent article ( 1 ) describing the work of Haddeland and Kennedy states that they found mixing in a closed impeller to be better than dispersion with a porous tube or sparger for the reactions of chlorine with paraldehyde or acetic acid. EMPIRICAL CORRELATIOK OF k1/2. I t may therefore be stated that all runs with either the impeller or the jet mixer and with zinc chloride catalyst in excess of 0.5% may be interpreted as half-order irreversible reactions with the equation
...
E. G , G. Moles/
Run
Vol. 44, No. 10
..
100 f 0.99 0.99 1.91 1.91 1.91 1.07 1.07 2.04 2.04
kv? 1.01 1.46 1.27 2.13 1.65 1.12 1.23 2.05 2.05
CATALYST RE-USE. I n view of the noted stability run 37 was made with the catalyst precipitated at the end of run 36. Only about 85% of the catalyst used in run 36 was recovered for run 37, and the value of k112for this run was 0.925, while 0.975 would have been predicted by Equation 9. The amount of recovered catalyst was not determined by direct weighing but by titration of the solution of it in the lauryl alcohol for run 37. Thus a small physical carry-over of acid could explain the slight discrepancy above. I n any event, a t least one re-use of the catalyst seems perfectly reasonable. ALuMIxunr CHLORIDE.Run 27, compared with no catalyst in Table VII, shows that this material neither reduces the order nor appreciably increases the value of k5/2 for this reaction. It is therefore practically useless as a catalyst. This confirms the observations of Guyer, Bieler, and Hardmeier ( 2 ) with the corresponding reaction of cetyl alcohol. STIRRERSPEED. Since certain previous effects have pointed t o some importance of diffusion, it was thought that this variable might be important. However, the results of part D of Table VI1 show that above 850 r.p.m. it exerts practically no effect. However, without any mechanical agitation there was about
The rate constant, k l l z , is influenced principally by temperature, ratio of gas rate to liquid volume, gas purity, and catalyst amount, The most thorough investigation has been at 150" C., and it is believed that kl/2 may be correlated at that temperature by the equation
bYo2 = 0.645(0.43 + 6 4 f )
(8)0'55
I t must be recognized, however, that the effects of yo, ( G I V ) , and f were separately determined with the other variables at only one constant value and that the ranges of these variables for determining Equation 10 were: G / V , 0.77 to 10.2 gram moles per hour per liter: f, 0.0052 to 0.0197 weight fraction; and yo, 0.224 to 1.0 mole fraction of hydrogen chloride. At 130" C. and 1% catalyst one may write
It is probable that the effects of catalyst concentration and inlet hydrogen chloride composition are about as at 150" C., but it was not established in this work. Values of k1/2a t other temperatures can be computed using an activation energy of 16,000 calories per gram mole. The exponent for the G / V term is temperature dependent, presumably because of diffusion effects. These values should be true of either impeller or jet mixer operation. Below 0.56% catalyst the above equations fail because the whole mechanism is undoubtedly altered, as is evidenced by the change of apparent order. Utility of Results The data and results herein presented are useful in themselves, for they should permit calculation of conversion versus time for a variety of conditions. These results, however, are particularly helpful in solving the problem of efficient gas utilization, always a difficult matter in processes of this sort. It is obvious that in the late stages of such a process the discharge of hydrogen chloride gas is very large. This material must either be lost or reclaimed, an expense in either case. The present authors present three methods to accomplish high gas utilization: 1. Recycling the off-gas after a simple purification (a condensation of the water in this case) 2. Adjusting feed gas to an ever-decreasing rate such that, it is
2485
INDUSTRIAL AND ENGINEERING CHEMISTRY
October 1952
always present to a certain and small allowable excess; this means a gas rate adjusted to the reaction rate, which becomes very low indeed toward the end of a run 3. Provision of additional vessels in series t o utilize the off-gas in subsequent conversion The first two of these methods have been subjected t o experimental verification in this work. Runs 31, 32, 33, and 34 were those involving the use of the jet mixer with recycling, and were effective in attaining good conversion in a reasonable time and with very little loss of hydrogen chloride. The reducing feed rate method was devised as follows. A figure of 20% excess hydrogen chloride was chosen arbitrarily as permissible. The flow rate, therefore, must always be 1.2 times the rate of reaction, which in turn depends on the flow rate. Specifically, the rate of reaction in terms of fractional conversion, is
and therefore the gas rate, G, must be
dx 4-1.2 N A O- = G dt
I
(12)
Applying this to a run a t 150" C. with 1,07y0catalyst and with a volume of 2.56 liters, Equation 6 reduces t o
2
3
4
TIME H O U R S Figure 4.
Calculated Conversions for Vessels 1 and 9 of Two in Series
k1/2 = 0.42 Go.''
The use of series converters was not tried experimentally, but calculations have been made based on the present data. Assuming two reactors, each of 2.56 liters charge, with 1% catalyst and a feed rate of 7.5 gram moles per hour (25% in excess of initial reaction rate in the first converter), the time conversion equation of the first vessel is
This may be incorporated into Equation 11 t o obtain
which may then be equated to Equation 12 -dx= -
dt
G 1.2 N A O
1
to obtain
-- dl - x
=
0.318 t
(16)
The rate of reaction in the first converter, in terms of moles converted, is
This gives a schedule of the required values of G for various values of 2. Integrating Equation 13 yields
= 6.0
-
1.91 t
from Equation 16. The rate of hydrogen chloride discharged from the first vessel and fed t o the second is the right-hand side being integrated graphically with the appropriate value of G as derived previously. Actually when this was done a slight error in determining G was made so that the excess was 20% until 80% conversion and 25% thereafter. Therefore the values of G actually used were put into Equation 15 for integration. Run 35 was made in this manner, and a comparison between observed and calculated conversions versus time is given in Table VIII. Table VIII.
Reducing Gas Flow Rate Run
(Comparison of calculated and observed conversion of r u n 35) Conversion, Mole yo Time, Hr. Obsvd. Calod.
0.5 1 .o 1.5
2.0 2.5 3.0 4.0
25.3 45.1 61.0 73.6 83.5 89 0 94.5
27.0 46.9 61.4 72.6 81.0 87.4 91.9
The agreement here is remarkable when it is considered that an extrapolation of the effect of G is made to one-fourth the minimum experimental value. Thus the constant flow rate data may be used to predict the behavior with variable flow rates.
7.50
- 6.0 + 1.91 t
=
1.50 f 1.91 t
This expression may then be substituted in Equation 6 for kl, 2
and this into the differential equation for conversion
which may be readily integrated t o give conversion versus time for the second vessel. The removal of water vapor between the two vessels, which is easy t o accomplish with this system, has been assumed. The loss of hydrogen chloride in the condensate is not large and it is easily recoverable. Figure 4 indicates the conversion in the first and second vessels versus time according to these calculations. At 97% conversion in the first vessel, the efficiency of gas utilization in the first vessel is only 47%. At 97% conversion in the second vessel the efficiency of gas utilization in both vessels is 73%, a significant increase. Probably a third vessel would be required t o get as high utilization of gas as can be reached in one vessel by either gas recycling or the reducing feed rate method. However, this method has the advantage over
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
2486
Vol. 44, No. 10
-
I.
G(yo YI) = kVCw (21) That is, the hydrogen chloride lost from the gas stream is exactly equal to that consumed by the first-order reaction. The Iefthand term in Equation 21 is correct for such a case as this because water is given off mole for mole for the hydrogen chloride consumed-Le., G is constant. Equation 21 may be solved for yl to yield
0.
y1 = yo
-k v CB
(22)
G
This may then be put into the equilibrium expression to yield
E cw
cw
= yo - kV -q
c 2
and CW solved for to yield
X
-I C
CW
0.0.
YO = ___
(24)
V
This may be put into the rate equation
and integrated to yield
which may be transformed by substituting 0.c s(fractiona1 conversion) = 1
C -A
CAO
Figure 5.
Typical Plot for Proposed Mechanism
the first two in that the hydrogen chloride is generated a t a constant rate. Fundamental Kinetics
There is no theoretical reason to explain such an order as onehalf. It is satisfactory as an empirical method, but it should not be used beyond that. It was mentioned that diffusional effects are undoubtedly important, and the following development takes into account such effects or at least the phase equilibrium which results from such a phenomenon. The following assumptions are made:
1. The li uid phase is lauryl alcohol, lauryl chloride, dissolved hydrogen chqoride, and catalyst. 2. The liquid mixture is saturated with hydrogen chloride a t all times after a very short initial period. 3. The hydrogen chloride is soluble only in the lauryl alcohol and not in the chloride. (This is confirmed in a ualitative manner by observing the last column of Table I. T%e acidity before gas introduction (no conversion) is appreciable because of the presence of zinc chloride. The acidity jumps suddenly a t gas introduction and thereafter decreases as the conversion increases in all cases.) 4. The phase equilibrium between the hydrogen chloride in the gas phase yl and the mole fraction of hydrogen chloride in the liquid phase, CW,is defined by
At constant alcohol concentration, C A , this reduces to the usual vapor-liquid equilibrium ratio. This expresses the observed fact that hydrogen chloride solubility is less as C A is less. 5. The reaction is irreversible and first order wlth respect to the dissolved hydrogen chloride concentration, Cw, being unaffected by alcohol concentration until very near completion (this being so high relative to C W )or the chloride concentration. A material balance may then be mittel?
to yield log(1 - x) = - __ -. Cao 2.3 CLG
By plotting log (1 of slope -
jIGV cYO,,t - x]
- 2) against
’
- Cao should result. Such straight lines did 2.3 C, G result, a typical one being shown in Figure 5, confirming the mechanism assumed. The ratio of the constants k / C z may be determined from these slopes and the consistency among all the runs, grouped into the same series as before, is shown in the last column of Tables I1 through VII. Considering the number of variables and the range for each of them, the consistency of the ratio k/Cz is very good. It is probably not possible to separate k and CZon the basis of the data here presented, but the data of Table I1 show that the temperature effect on k must be quite large, since Cz is probably not greatly affected by temperature. Table I11 shows, as would be expected, that k/Cz is not affected by G/V. Tables IV and V show that the entering concentration of hydrogen chloride does not influence k/C2, although run 16 does differ somewhat, attributed by the authors to experimental error. Table VI shows an almost linear dependence on catalyst amount as would be expected, and it is apparent in runs 14 and 20 that this mechanism is equally good for the uncatalyzed reaction. Part A of Table VI1 shows the entirely expected effect of catalyst addition and part B the unimportance of interrupting a run. Part C with aluminum chloride exhibits values of k/Cz essentially the same as the uncatalyzed reaction. Part D shows a slight effect of stirrer speed on the value of k / C z , and part E shows the jet mixer to be only a slightly better means of bringing about contact than the turbine. The results of these last two series probably point to the assumption of the attainment of phase equilibrium as being somewhat oversimplified. However, it appears that all these results may be ~
October 1952
INDUSTRIAL AND ENGINEERING CHEMISTRY
explained on the basis of this mechanism with a constant IC/C*, essentially affected only by temperature and catalyst nature and amount. However, one serious shortcoming to the use of this development is that the equations for the prediction of conversion for an arbitrary schedule of gas rates or for reactors after the first one in a series cannot be integrated, either in a closed form or graphically. Numerical integration can be performed, but it is tedious and does not give any improvement over the results which can be obtained quite easily with the empirical approach. Thus the empirical half-order mechanism has been retained because of its greater utility. Conclusions
These results show that the hydrochlorination of lauryl alcohol with zinc chloride catalyst may be expressed as a nonreversible reaction, half order with respect t o the alcohol. The rate constant of such an interpretation is greatly influenced by temperature, gas rate per unit volume of liquid, and gas purity; effects of these variables are given in the text. They also show that a very effective way of carrying out this reaction is the use of a jet mixer as a recycling device to bring the discharged and unreacted hydrogen chloride back into contact with the lauryl alcohol. Other methods of increasing the fractional utilization of hydrogen chloride are discussed, and the effectiveness of feeding this gas a t a constantly reducing rate was demonstrated experimentally. A more fundamental kinetic interpretation is presented in which the importance of diffusion is incorporated. This shows that the mechanism is probably one of a diffusional equilibrium between hydrogen chloride in the gas and hydrogen chloride dissolved in the liquid, followed by a reaction &st order with respect to this dissolved hydrogen chloride. Such an interpretation leads to a constant, dependent only on temperature and ratalyst nature and amount. Acknowledgment
The first author wishes to acknowledge the financial assistance of the AMERICAN CHEMICALSOCIETYthrough its fellowship program.
2487
Nomenclature
C
= =
f
G k
= = =
kl/2
=
Cz
kS12 =
M = N = NA = P = t
=
z y
=
V = Vl = Vz = W = =
molar concentration, gram moles per liter distribution constant for hydrogen chloride between vapor and liquid phases weight fraction of catalyst in a sample total gas flow rate, gram moles per hour reaction rate constant reaction rate constant obtained by interpreting data as a half-order reaction reaction rate constant obtained by interpreting data as a 2.5-order reaction average molecular weight of alcohol normality of sodium hydroxide solution gram moles of alcohol weight fraction of alcohol in a sample time, hours volume of liquid phase at reaction temperature, Biters titer for a blank in acetylation procedure, ml. titer for a sample in acetylation procedure, ml. weight of sample, grams mole fractional conversion mole fraction of a component in the gas phase
Subscripts
A B
0 1
= component in general or lauryl alcohol = component in general or hydrogen chloride = initial value =
exit value
Superscripts
n
= apparent order of reaction
Literature Cited (1) Chem. Eng. News, 29, No. 28, 2856 (July 9, 1951). (2) Guyer, A,, Bider, A., and Hardmeier, E., Xelv. Chim. acta, 20,
1462-7 (1937). (3) Smith, D. M., and Bryant, W. M. D., J . Am. Chem. floc., 57,61-5 (1935).
RECEIVED for review October 3, 1951. ACCEPTED M a y 19, 1952. For material supplementary t o this article, order Document 3598 from American Dooumentation Institute, 1719 N Street, N. W., Washington 6, D. C., remitting $1.00 for microfilm (images 1 inch high on s t a n d a r d 35 mm. motion picture film) or $1.35 for photocopies (6 X 8 inohes) readable without optical aid. This paper is based on a dissertation presented by Henry A. Kingsley, Jr., to the faculty of the Yale School of Engineering in partial fulfillment of the requirements for the degree of Doctor of Engineering.
laboratory Continuous Distillation Column For Concentration of Aqueous Solutions of Volatile Flavors
Engrnyring Process development
K. P. DlMlCK AND MARION J. SIMONE Western Regional Research Laboratory, Albany, Calif.
I
N THE study of the chemistry of fresh-fruit volatile flavors, a
fractionating column was required for concentrating the large amounts of very dilute aqueous solutions of the flavors which were obtained by flash evaporation (3) of fruit purees. For this purpose an all-glass laboratory fractionating column of high plate efficiency and high throughput was desired. Furthermore, the material should not be subjected to prolonged heating, which might accelerate decomposition or alteration of the fruit flavors. A modified, continuous-feed, Oldershaw bubble plate column met these requirements. This type of column as described by
Oldershaw (6) and improved by Collins and Lantz ( 2 ) consists of a series of perforated glass plates sealed into a tube. Each plate is equipped with a weir, t o maintain the proper liquid level on the plate, and a reflux return tube directing the liquid to the next lower plate. The conventional column, which was designed for nonaqueous distillations, is not suitable for the distillation of aqueous solutions ( 2 ) because of the high surface tension of water, which causes flooding and abnormal pressure drop across the column. This limitation was overcome by increasing the distance between plates and diameter of the plate perforations.