I
F. NORMAN GRIMSBY' and EDWIN R. GlLLlLAND Massachusetts Institute of Technology, Cambridge, Mass.
Continuous Oxygen-lnitia ted Ethylene Polymerization This research adds to available knowledge on synthesis of polyethylene, presenting data on the polymerization reaction in a continuous system
THE
tough, flexible white plastic known as polyethylene was first synthesized by Imperial Chemical Industries, Ltd., in 1933, from ethylene at high pressures and elevated temperatures according to the reaction : AH
=
nC2H4 = ( C2H4)n -22,000 cal./mole C2H4
Traces of oxygen, consumed during the reaction, served as catalyst. Rapid removal of the large amount of heat generated by the polymerization was necessary to avoid an explosive decomposition of the ethylene, especially a t high oxygen concentrations (5, 6, 74). Emerging from its wartime application as a superb insulator for high-frequency wiring, polyethylene enjoyed an enormous expansion in both application and production. Still the synthesis data in the literature did not enable prediction of the yield, molecular weight, or other property to be expected from a polymerization carried out a t a given temperature, pressure, oxygen concentration, and residence time (though a few qualitative relations had been established). Even the most thorough studies were based on data from batch polymerizations, in which the variables cannot be individually controlled. These data neither led to a good correlation of the variables nor gave rise to theory for the reaction mechanism. I t was felt that a laboratory study of the polymerization of ethylene in a continuous system would permit control of variables, so that the data could be more readily interpreted, and that analysis of these data would lead to a theory for the oxygen-catalyzed polymerization of ethylene.
valve provided with electrical heating. A glass vial held in a steel shell collects the polymer on the low-pressure side of the valve; the exit gas is metered through a wet-test meter. The synthesis gases are supplied from cylinders storing ethylene which contain the desired concentrations of oxygen, as analyzed by a modification (75) of the. micro-oxygen technique of MacHattie and Maconachie (73). Molecular weights of the polymers were determined by measuring the viscosities of their xylene solutions a t 85' C. and calculating the numberaverage molecular weights according to the correlations of Harris ( 8 ) . Further product evaluation included infrared analysis and measurement of the electrical power factor, dielectric constant, density, tensile strength, and tensile elongation; analysis of test results indicated the effects of changes in the synthesis conditions on the oxygen content of the polymer, degree of olefinic unsaturation, and degree of short-chain branching, and effects on the measured physical properties themselves.
Results and Discussion Early runs indicated that stainless steel, or other corrosion-resistant material of construction, was necessary to avoid a badly discolored product. All
-
BY-PASS RUPTURE
T , c,
t
Apparatus and Procedure The GE compressor delivers the ethylene a t pressures up to 1000 p s i . to the third-stage compressor, which in turn delivers the ethylene to the reactor at pressures up to 30,000 p.s.i. (Figure 1). The temperature of the boiling Dowtherm is controlled by adjusting the pressure of the Dowtherm reflux system. The polymer and unreacted ethylene are continuously removed through a high-pressure throttling needle Present address, Shell Development Co., Emeryville, Calif.
parts of the apparatus in contact with the hot ethylene or polymer were replaced by stainless steel counterparts. The pump lubricant (motor oil) was replaced with glycerol, as the solubility of the former in compressed ethylene led to a yellow polymerization product. After these and other modifications of the apparatus, runs were made investigating pressure, 6000 to 30,000 p.s.i.; temperature, 155" to 255" C. (mean jacket temperatures) ; oxygen concentration, 10 to 2170 p.p.m.; and residence time, 0.37 to 14.8 minutes. Conversions of ethylene to polyethylene u p to 33% were obtained and average molecular weights from 7270 to 49,200 were measured. The effect of pressure on per cent conversion is shown for several oxygen concentrations in Figure 2. I n all runs residence time was long enough to attain essentially the maximum conversion. Polymer is not formed below about 9000 p s i . , but conversion in; creases linearly with increases in pressure above 9000 p.s.i. That higher conversions are obtained with higher oxygen concentrations is also evident from Figure 2. Figure 3 is a replot of the relations shown in Figure 2 using pressure as the parameter. Slight corrections have been applied (according to the slopes of Figure 2) to some of the original data points in order to plot
\ ACTIVATED CHclRCOAL
@- PRESSURE
&
OOWTHERM REFLUX CONOENSER (GLASS)
REACTOR TUBE I8 - 8 STAfNLESS STEEL
1
GAUGE
TO ATMOSPHERE
{O.O. X //d'/.O. X 48''
1
@-REGULATOR
o'sc
ELECTRICALLY HEATED
CALCIUM CHLORIDE
t"PIP=Llfo~~AlNlMG BOILING OOWTHERM
8 Figure 1 .
Ethylene was continuously polymerized in this laboratory setup VOL. 50, NO. 7
0
JULY 1 9 5 8
1049
Figure 2. Pressure has a linear effect on conversion at various oxygen concentrations
/oooo
5000-
/5000
20000
25 000
where k,, kb, and k , are rate constants, 0 represents time, and n and c are the number and the concentration, respectively, of the particles indicated by their subscripts. If k , is sufficiently large, in a time very short compared to the total time of the reaction a concentration of free radicals will be reached, such that free radicals are removed according to Equation 3 at the same rate as free radicals are being created by step A. This permits solving for a steady-state free-radical concentration, which, when substituted in Equation 2, leads to the following expression for the rate of polymerization :
PRESSURE, P 5 1 A
them a t the exact parametric pressures, and data a t 684 p.p.m. oxygen have been added. The lines were drawn a t a slope of 0.5 and provide a good correlation of the data. [For comparison, the dotted curves represent some of the best batch data in the literature ( 7 4 . 1 Thus conversion is proportional to the square root of the concentration of oxygen in the ethylene. Several peroxide-catalyzed polymerizations-e.g., styrene-are reported to have rates of reaction that vary as the square root of the catalyst concentraticn. Theories for the mechanism of these reactions have been proposed and the kinetics involved have been studied to some extent (7-3, 7, 9-77, 76, 77). I t was hoped that modification of these theories would permit their application to the present work. The kinetics proposed for these re-
actions assume a three-step free-radical mechanism : A.
B.
Initiation. Propagation.
X M I * 4- M I -+ Mi
+ Mi*
Mz”
Mz” + M i + Ma*
.....
+ +
Mi*+ M j I * Termination. M, + M1 is a “dead” monomer molecule, X is the catalyst, Mi* is a free radical of the monomer, and the other subscripts indicate the number of monomer units in the growing free-radical or terminated polymer molecule. Assuming initiation first-order with respect to both catalyst and monomer concentration, propagation first-order with respect to both free-radical and monomer concentration, and termination second-order with respect to free-radical concentration, the rates for these steps may be expressed by : C.
Mi* My5
d+,
where V is the volume of the system. Monomer concentration, cM1, is included in Equation 1 to represent the general case of an initiator which functions through reaction with monomer. While oxygen is assumed to be such an initiator, the nature of the intermediate reaction product of oxygen and ethylene is not known. Had cMl not been included in Equation 1, thus representing the general case in which the initiator functions by dissociation rather than by reaction with monomer, cAM1would have appeared to the first power in Equation 4, giving the expression usually derived for peroxidecatalyzed polymerization. The exponent of monomer concentration in various ethylene polymerization systems has been discussed (4, 72). To apply these relations to the oxygeninitiated polymerization of ethylene, in which the oxygen catalyst is being consumed during the reaction, it is postulated rhat the depletion rate is first-order with respect to the oxygen concentration. The resulting relation expresses F, the fractional conversion 01’ ethylene to polyethylene, in terms of fo, the original mole fraction of oxygen, and J , the mole fraction of oxygen at the time corresponding to the conversion F (both f andfo are based on the original number of moles of ethylene) : 1
OXYGEN
Figure 3. ethylene
1050
C O N C E N T R A T I O N , P . P . M.
Conversion is proportional to the square root of oxygen concentration in
INDUSTRIAL
A N D ENGINEERING CHEMISTRY
-
(1
- F)’/2 = k(fO”2 -
f’i’)
(5)
where k = kb/k,1’2k,1’2. Substituting for the total conversion, FT, when all the oxygen is consumed, and noting that for small conversions (1 - F)”’
E t H Y L E N E POLYMERIZATION IO 08
SO
06
2
e5
0 4
a
8. g
eo
-
c2 8
c k?
I5
t
B
2 2
B
02
01
0 08
IO
b
;P
Figure 5. Correlation according to Equation 7 gives a linear relation
5
006
oo4
0 0
4
2
AVERAGE
6
d
IC
IO
14
RESIDENCE T I M E , MINUTES
0 02
Figure 4. Effect of residence time on conversion at various temperatures. Higher conversions are possible at higher ternperatires
o o / 0l
l
I
t
0l 5
l
/ lo
l
Il 5
l
2 l0
l
l
2 5
l
3 l0
l
3J
e , MINUTES
may be approximated by 1
I
- Z F we
have :
FT
= 2kf0'l2
(6)
which is the relation indicated by Figure 3 . Integrating the oxygen depletion expression and substituting from the relation between F and f lead to an expression between conversion and residence time : In +(F,FT) = -kk,CM@/2
x
(7)
in which @(F,FT) represents an expression approximately equal to 1 - F/FT. Obtaining reaction rate data required making runs of shorter residence times (Figure 4). Such residence times could not be maintained with the desired degree of uniformity because of difficulties in handling the higher flow rates through the product valve, even with an improved valve design. Only a t 190" C. and 455 p.p.m. oxygen were sufficient data obtained to make any sort of a test of Equation 7. According to Equation 7, data a t a given temperature plotted as in Figure 5 should produce a straight line from the origin. The line drawn for 190" C. represents the data fairly well. The curve a t 190' C. in Figure 4 was drawn according to this correlation line. Figure 4 also shows data at several other temperatures. At the lower jacket temperatures long preheat times before the reaction begins prevent this analysis from being applied to the points connected by the dotted curves in this plot. At the two temperatures higher than 190" C. the solid curves are drawn in according to Equation 7, which requires but two points to de-
termine constants k , and FT. The severa1 runs a t long residence times a t 255" C. sumort the raDid leveling off *. toward a maximum conversion indicated by the theory and also show that higher conversions are possible a t the higher temperatures. Because of the difficulties mentioned above it was felt that the present apparatus warranted only one run of very short residence time a t each of the higher temperatures to show roughly the rate of reaction. Molecular weights increased with increases in pressure, but decreased with increases in temperature or oxygen concentration. As oxygen concentration decreases during passage through the reactor, one would expect runs of short residence times to yield polymer or lower molecular weight than comparable runs in which the reaction was given sufficient time to reach completion. As shown in Table I, the experimental data confirmed this conclusion and checked fairly well quantitatively, especially a t the higher temperatures, with the theoretically derived
relation. ( M W ) T I ( M W )= 2
- (F/FT)
(8)
v
Table 1.
This may be considered further support of the assumptions used in deriving the several expressions. The product evaluation tests revealed several additional effects of changes in the synthesis conditions. The oxygen content of the prsduct, as indicated by the power factor and infrared data, was lowest when the polymer was made under conditions of high conversion and low oxygen concentration. Olefinic unsaturation, as determined by infrared measurements, and short-chain branching, as determined both by a n inverse relation with density reported in the literature (78) and by infrared measurements, were independent of oxygen concentration used but were increased markedly by evaluation of synthesis temperature and decreased by increases in pressure. The tensile strengths of the polymers increased with their densities, in good agreement with a correlation in the literature (78), thus es-
Molecular Weight vs. Conversion (455 p.p.m.
0 2 )
Press., P.S.I.
Res. Time, Min.
Conv., F
Molecular Weight Calcd. by Measd. Eq.8
17,800 17,600
4.03 0.66
0.192 0.089
27,600 21,500
18,000
18,000 18,000
8.74 0.94
0.317 0.194
16,230 11,250
12,430
17,900 18,000
3.83 0.38
0.290 0.205
9,410 7,270
I
Temp., O
c.
190 220 255
VOL. 5 0 , NO. 7
...
... ... 7,270
JULY 1958
1051
sis of the data supports a proposed theory for the mechanism of the reaction through consistency of the correlated data with the derived theoretical expressions for the relationship between conversion and oxygen concentration, conversion and residence time, and molecular weight and residence time. Literature Cited
$1 E
8
Figure 6. Three-dimensional schematic representation of reaction variables and extreme reaction conditions. More efficient heat transfer (higher velocities in longer tubes) would give greater conversion
*
Black polymer encountered
tablishing a definite similarity between these and polyethylenes generally known commercially. Single measurements of the tensile elongations did not give results upon which conclusions could be based. Table I1 summarizes qualitatively the effects of changes in synthesis conditions on dependent variables. Some polymer was made from a synthesis gas containing 10.570 ethane. There was no drastic effect on the nature of the polymer, the conversion and rate of reaction were only moderately reduced. In several runs pressure, jacket temperature, and oxygen concentration were increased individually to the point where black, sometimes sooty, products were obtained. Figure 6 is a threedimensional schematic representation of the ranges of variables included in this study. Thus, at 190” C. jacket temperature and 455 p.p.m. oxygen
Table II. Effects of Changes in Synthesis Conditions 0
Press. Temp. Concn. Conversion Molecular weight Oxygen content Olefinic unsaturation Short chain branching Tensile strength Density
+ Direct
++ -
-
-
+
+
+++ ++ ---
++
Conclusions
0
0 0 0
relationship. - Inverse relationship. 0. No marked effect. Double symbols. Very pronounced effect.
J 052
concentration white polymer was obtained a t all pressures up through 24,000 p s i ; at 27,000 p s i . black polymer was formed. At 190’ C. and 18,000 p.s.i. white polymer was obtained a t oxygen concentrations up through 684 p.p.m.; a t 2170 p.p.m. the product was black. At 190” C. and 12,000 p.s.i. changing from 455 p.p.m. oxygen to 2170 p.p.m. introduced a yellow color into the product. At 455 p.p.m. and 18,000 p.s.i. white polymer was produced throughout the range 140” to 220’ C., but at 24,000 p d . black product was encountered a t 220’ C. In each case the black color was due to carbon and indicated that ethylene decomposition (5, 74, 79, 20) was taking place. However, no explosions occurred in the small-bore reactor used in the present work. The temperatures necessary for this reaction to occur (over 400” C. according to literature accounts) were reached through the high rates of reaction at the high level of each variable. More efficient heat transfer, such as would be brought about by higher velocities in longer tubes, would allow higher levels of these variables to be used, and greater conversions to be obtained.
Operation of a continuous polymerization system has given data for the oxygen-initiated ethylene polymerization which readily show the effects of changes in synthesis variables on the extent of conversion and the nature and properties of the product. Analy-
INDUSTRIAL AND ENGINEERING CHEMISTRY
(1) Abere, J., Goldfinger, G., Mark, H., Naidus, H., Ann. N . Y.Acad. Sci. 44. 267-96 11943’1. (2) Bodinstein, MI, Z.-El~trochem.42, 443 (1936). (3) Bodenstein, M., Z. physik. Chem. 85, 329 (1913). (4) Ehrlich, P., Cotman, J. D., Jr., Yates, W. F., J. Polymer Sci. 24, 283-5 11957). (5) Fawcett, ‘ E. ’W., Gibson, R. O., Perrin, M. W.(to Imperial Chemical Industries), U. S. Patent 2,153,553 (April 11, 1939). (6) Fawcett, E. Mi., Gibson, R. O., Perrin, M. W., Paton, J. G., Williams, E. G., Imperial Chemical Industries, Brit. Patent 471,590 (Sept. 6, 1937). (7) Gee, G., Melville, H. W., Trans. Faraday SOC. 40, 217-80 (1944). (8) Harris, I., J . Polymer Scz. 8, 353-64 (1952). (9) Herington, E. F. G., Robertson, A., Trans. Faraday SOC. 38, 490 (1942). (10) Hulbert, H. M., Harman, R. H., Tobolsky, A. V., Eyring, H., Ann. iV. Y.Acad. SCZ.44, 371-418 (1943). (11) Kallal, R. J., “Reactions of Ethylene with Alcohols,” Sc.D. thesis, MIT, Cambridge, 1949. (12) Laird, R. K., Morrell, A . G., Seed, L.,Dascussions Faraday Soc. 22, 12637, 147-53 (1956). (13) MacHattie, I. J. W.,Maconachie, J. E., IND.ENG.CI-IEM., ANAL. ED. 9,364-6 (1937). (14) Perrin, M. W., Paton, J. G., Williams, E. G. (to Imperial Chemical Industries), U. S. Patent 2,188,465 (Jan. 30, 1940). (15) Poynton, 3. A., “Natural RubberCarbon Black Adsorption Studies,” Sc.D. thesis. MIT. CambridPe. 1952. (16) Price, C. C., Ann. ik Y.AcadySci. 44, 351-70 (1943). (17) Price, C. C., Kell, R. br., J. Am. Chem. Sot. 63,2798--801 (1941). (18) Sperati, C. A., Franta, W. A., Starkweather. H. W.., Jr.., Ibid.. 75. 612733 (1953). (19) Tani, Hisaya, Chem. High Polymers ( J a j a n ) 4, 82 (1947). (20) Torrans, D. J., “High Pressure Polymerization of Ethylene,” Sc.D. thesis, MIT, Cambridge, 1941.
RECEIVED for review September 3, 1957 ACCEPTED April 10, 1958 Divisions of Industrial and Engineering Chemistry and Paint, Plastics, and Printing Ink Chemistry, Symposium on Chemical Engineering Aspects of Polyethylene and Urethane Production, 132nd Meeting, ACS, New York, September 1957. Abstracted from the thesis submitted by F. Norman Grimsby, December 3, 1953, at Massachusetts Institute of Technology in partial fulfillment of the requirements for the degree of doctor of science in chemical engineering. Work supported financially by a Dow Chemical Co. fellowship for research in high pressure chemical reactions.