Aug. 5 , 1960
ETHYLENE POLYMERIZATION
WITH A SOLUBLE
CATALYST
3883
ml.) were withdrawn and added t o 10 ml. of cold 10% aqueMercuric oxide formed was filtered, washed twice with cold water, dissolved in 10 ml. of 5 N nitric acid and titrated with 0.2 N potassium thiocyanate. In cases where the concentration of mercuric chloride was lower, lower values for mercuric oxide were obtained because of the solubiliy of mercuric oxide in water. Rate Measurement.-Aliquots (10 ml.) were withdrawn DETERMINATIONS OF MERCURICION CONCENTRATIONS IN at intervals and added t o 10% aqueous sodium hydroxide. 75% ETHANOL CONTAIXINC VARIOUS AMOCNTS O F HYDRO- The amount of mercuric oxide formed was determined by the method described above. CHLORIC ACID Conversion of 8-Acetoxyethylmercuric Chloride to 8HgCIz, M HCI, M HgO obtained, % Ethoxy-ethylmercuric Chloride.-In the absence of per0.025 0.00 99.0 chloric acid, 8-acetoxyethylmercuric chloride was recovered unchanged from ethanol solution after standing for one week. .10 98.7 However, in the presence of perchloric acid, the acetoxy .15 98.7 compound was converted t o the ethoxy compound. 8-Ace.20 98.3 toxyethylmercuric chloride (3.2 9.) was dissolved in 95 ml. .25 96.6 of ethanol containing 30 ml. of 20y0 perchloric acid. After 5 days, the solution was neutralized with 0.5 N aqueous .50 94.5 sodium hydroxide and cooled to - 10". The crystals formed .035 .05 99.8 Mixed m.p. with authentic (1.59 g.) melted a t 89-90'. .15 99 8 sample of 8-ethoxyethplmercuric chloride showed no depres.Ol .05 94.4 sion. .10 93.6 Acknowledgment.-We wish to express our .20 93.5 thanks to Sumitomo Chemical Co. for financial .05 1.00 88.7 assistance.
p-Hydroxyethylrnercuric chloride was prepared according t o the method of Cotton and Let0 by the reaction of ethylene with aqueous mercuric nitrate; m.p. 152-153" (reported'3 153-155'). Determination of Mercuric Ion Concentrations in Hydrochloric Acid.-Out of mercuric chloride solutions in 75% ethanol containing hydrochloric acid, aliquot samples ( 10
(13) F. A. Cotton and J. R. Leto, THIS J O U R N A L , 80, 4824 (1958).
[CONTRIBUTION FROM
THE
ous sodium hydroxide.
YOSHIDA,KYOTO,JAPAN
RESEARCH DEPARTMEXT, UNIONCARBIDEPLASTICS Co., DIVISIONOF UNIOSCARBIDECORP.]
Transition Metal Catalysts. I. Ethylene Polymerization with a Soluble Catalyst Formed from an Aluminum Halide, Tetraphenyltin and a Vanadium Halide' BY WAYNEL. CARRICK, RUDOLPH W. KLUIBER,EUGENE F. BONNER, LLOYDH. WARTMAN, FRANK hZ. R U G G AND JOSEPH J. SMITH RECEIVED AVGUST24, 1959
A mixture of an aluminum halide, tin tetraphenyl arid a trace of a vanadium halide in cyclohexane forms a clear solution which contains an active catall-st for the low pressure polymerization of ethylene. I n this catalyst the optimum weight concentration of the vanadium halide is only 0.1 to 5 parts in IO3 parts of total components, and polymerization will take place when the concentration is as low a s one part in lo6. Each molecule of the vanadium compound can catalyze the formation of as many as 3000 polymer molecules. The polyethylene obtained is linear in molecular structure, high in moleculai weight and narrow in molecular weight distribution. The narrow distribution suggests t h a t catalysis is homogeneous and due t o a single catalytic species.
Introduction Organometallic mixed catalysts for ethylene and a-olefin polymerization were first disclosed in a publication by Ziegler, Holzkamp, Breil and Martin2 describing the polymerization of ethylene a t low pressures with a mixture of titanium tetrachloride and an aluminum alkyl. Since then many other catalyst mixtures have been reported which polymerize ethylene a t low pressures to a high molecular weight, linear p 0 1 y m e r . ~ ~ Some ~ of these also polymerize a-olefins to stereoregular polymers. These catalysts ordinarily are formed by combining a reactive organometallic compound with a transition metal compound of groups TV through V I of the periodic table. The catalyst-forming reaction is very complex, and attempts t o define the active species generally have led to ambiguous re(1) This work was presented before t h e 133rd Meeting of t h e American Chemical Society, San Francisco, Calif., April 13 to 18, 1958. Address inquiries t o W . I,. Carrick. (2) K . Ziegler, E. Holzkamp, H . Breil and H . Martin, Angew. C h e m . , 67, 541 (1955). (3) SOC.Chem. Ind.. Reports on t h e Progress of Applied Chemistry, 42, 436 (1957) (4) J. K. Stille. Ciiem. Revs., 68, 541 (19.58).
sults. At present, there is no general agreement as to whether the reaction is ionic6 or heterogeneous6 or homogeneous,s which transition metal valence is o p t i m ~ r n , ~ -or~ lif the polymer molecule grows from a transition metal ~ e n t e r 6 ~ ~ ~ ~ " ~ ~ ~ or some other site.8112 However, most investigators do agree that a transition metal compound is a necessary part of the active catalyst species. The general problem of the mechanism of olefin polymerization by organometallic mixed catalysts has been under investigation in this Laboratory for several years, and numerous catalyst combinations have been examined. Studies on the organic derivatives of transition metals, interactions of cata(5) G. N a t t a , Chim. e ind. ( M i l a n ) , 8 7 , 888 (1955). (6) H . N . Friedlander and K. Oita, I n d . Eng. Chem., 49, 1885 (1957). (7) C. D. Nenitzescu and A . H . Ciresicahuch, Angew. Chem., 68, 438 (1956). (8) G. N a t t a , P. Pino, G. Mazzanti, U.Gianini, E. Mantica and M. Peraldo, J , Polymer S c i . , 26, 120 (1957). (9) D. S. Breslow and N. R . Newburg, THIS J O U R N A L , 81, 81 (1959) (IO) D. B. Ludlum, A . W. Anderson and C. E. Ashhy, ibid., 80, 1380 (195s). (11) w. L . Carrick. ioid., 8 0 , 6456 (1958). (12) F. P a t a t and H . Sinn, A n g e w . C h e m . , 70, 40G (195s).
3884
CARRICK, KLUIBER, BONNER,WARTMAN, KCGG AKD SMITH
lyst components, valence determinations, polymerization and copolymerization kinetics, and measurements of polymer structural units will be reported in this series of papers. This initial paper describes the novel catalyst formed from the interaction of a n aluminurn halide: tetraphenyltin and a small Zmount of a vanadium halide. Experimental
Vol. b%
in the reference beam for methyl determinations. This barnple showed less absorption at 7 . 2 5 @ than a polyxnetllyleiie prepared from diazomethane. Linear polyethylenes prepared and isolated so t h a t no phenyl groups were present were used in the reference beam for measuring phenyl content. Polymer Fractionation.-Sixty grams of polyethylene, 0.0 g. of di-l-butyl-)-cresol antioxidant and 1500 ml. of distilled ethylbenzene was maintained at 115'. .After the resin dissolved, about 1575 mi. of n-amyl alcohol was added, with agitation, to reach a cloud point. This quantity of non-soh ent usually was not sufficierit t o precipitate the first, highest molecular \wight, fraction of polymer. X nitrogen blanket was kept over the solution, and about 25-ml. incrcments of n-amy1 alcohol were added, successively with stirring. When roughly 5 g. of polymer had precipitated, the bath temperature was raised to 120' t o redissolve the resin, and reprecipitation then was allowed to proceed slowly, overnight, by cooling to 115'. After each fraction was precipitated, the soluble polymer was siphr~nedoff into a second flask and another increment ( i f precipitant added. The fractioii in the first flask was dissolvctl i n 1 1. of etliylhenzene and siphoned off into a large beaker ciiiit;rining 2500 i d . of methanol a t room temperature. The preciiiitatctl resin was removed by filtration, dricrl, weighed, :tiid it:, intrinsic viscosity in tetralin a t 130" i v a 5 obtnincd usiiig Ubbelohde viscometers. Successive fractions w r c collected by repetition of tlle same technique. I f , toward the end of a fractionation, more than 300 ml. of n-amyl alcohol was required t o precipitate a fraction, the bath temperature was dropped atxJut 5" instead. The initial temperature of 115" \ m s ellosen bccause this is about 20" above tlle crystallization teiiiperaturc of completely linear polyethylene i n this solvent systcm, thus permitting fractionation according t o molecular \vciglit rather than crystallinity.
Materials.-The ethylene used in most of this work was a C.P. grade obtained from the Matheson Co. I t s purity is >997; with the major contaminants being 0.2Yb butanes, 0.37; butenes, about 300-600 p.p.m. of osygen and 100 p.p.ni. of water. The water content was reduced further bypassage through a column of Drierite or Linde molecular sieves. The cyclohexane was a 99yb pure grade (Shell) t h a t was further refined by treatment with concentrated sulfuric acid, followed by a water wash, drying over sodium hydroxide and a final distillation. Tin tetraphenyl was obtained from Metal and Thermit Corporation and was recrystallized from a mixture of benzene and cyclohexane before use. \.anadium tetrachloride was supplied by the Cnion Carbide Merals Co., Division of Union Carbide Corp., and was used without further treatment. Several differcnt samples of aluminum chloride were used, but each was an anhydruus, resubliined, reagent grade. Aluminum bromide was prepared by the direct bromination of aluminun1,~3and the product was triply distilled over aluminum foil t o remove vanadium impurities before use. The final product was snow white. Cyclohexane solutions of aluminum bromide and vanadium tetrachloride were prepared and stored under nitrogen in bottles equipped with serum bottle stoppers. Xliquots mere removed with hypodermic syringes. Polymerization Conditions.-A mixture of 1500 ml. of dry cyclohexane and 1.5 millinloles of tin tetraphenyl was heated Results to boiling, purged with dry nitrogen t o remove volatile imReaction Conditions.--A mixture of an aluiniiium purities, and cooled t o 65". The nitrogen flow was stopped, halide (XIClaor -I1Br3),tin tetraphenyl and a minor and ethylene was introduced a t 2 l./min. Three millimoles amount of a vanadium halide (VCld, VOCl.+ etc.) of aluminum chloride (90 nil. of a saturated solution in boiling cyclohexane) and 0.025 millimole of vanadium halide in an inert hydrocarbon, such as cyclohexane, forms then were added as dilute solutions in cy-clohexane by way of a highly active catalyst for the low pressure polyhypodermic syringes. Upon addition of the third catalyst merization of ethylene. The ratios of the three component, rapid polymerization commenced, as evidenced components can be varied a t least 20-fold with by the appearance of a milky precipitate of colloidal polymer particles. As the polymerization progressed, the polymer retention of good catalytic activity; however, the particles became larger and at the end of the reaction were composition used in this work consisted of a mixture coarsegrainsthesizeofsand. Attheend of two hours, thereof the aluminum, tin and vanadium compounds in action was quenched by the addition of isopropyl alcohol, eyen though polymerization was still in progress b u t at a di- the mole ratios of approximately 1-3: 1 :0.02. In minished rate. The polymer was removed from the quenched contrast t o the usual organometallic mixed cataslurry by filtration, washed three times with isopropyl lysts which are heterogeneous, this mixture is alcohol or acetone, and dried. The yield of fluffy white soluble in cyclohexane. \Then the total conceiitrnpolymer was 80 g. riuinerous polymerizations were carried out by this same tion of the three components was less than 20 niilligeneral procedure using total catalyst concentrations ranging moles per liter, the solution did not show a Tyndall from 0.5 t o 25 millimoles per liter. polymer yield per unit Ream effect and was filtered through a 1 p bacterial of catalyst increased with decreasing catalyst concentration, filter without loss of catalytic activity. Almiiiiuiii increasing reaction time and increasing purity of reagents. chloride and aluniiiiuni bromide are iiiterchaiipeT h e highest yield obtained was 300 g. of polymer per gram of catalyst mixture. able, and other organometallic compounds can be Infrared Measurements.-Determinations of the various substituted for tin tetraphenyl14; however, the structural units were carried out on pressed polymer films (10-20 mils) using a Perkin-Elmer model 21 infrared spec- vanadium compound is essential, and its function cannot be replaced by compounds of other transitrophotometer. Methyl groups were determined using the 7.25 p band, phenyl groups at 14.35 p , vinyl groups at 11.0 p , tion metals without destroying or radically changpendent methylene a t 11.25p, and internal unsaturation a t ing the catalytic behavior. 10.35 p . The absorption coefficient for the phenyl group Table I describes a representative series of polymeasurement was the essentially constant ( i 3 i : ) value obtained from reference solutions of a number of n-alkylben- merizations carried out a t 63' with an ethylerie flow zenes in cyclohexane. of 2 l.?/niin.for two hours a t atniospheric pressure. For the measureincnt of some of these groups, pcily-cthylAlthough titaniutii tetrachloride is a comliioli C Y ene films of appropriate structure and comparable thickness catalyst in orqanometallic systems and polytilc r were placed in t h r reference beam of the infrared instrument t o compensate for absorptions which were immediately ad- melt index is increased by 311 increase in its relative jacent t o the analytical wave length. For example, the concentration,l;l the first five experirllents in the highest molecular weight (Jfn> 120,000) linear pr~lyethylene table show t h a t the addition of this halide to t h e prepared with the catalyst described in this paper was used
AIX:j-VX,-Sn(C~HS)~system causes no sigriificallt
(13) L. F. Audrieth. "Inorganic Syntheses," Vnl, 111. AfcGraiv-Ilill Book C o , I n ? , Nrwr York. N . Y . , 1950, p . 30.
(1-1) TY. I>. C a r r i c k . I t ~ l i n nP a t e n t .ii (1.5) E. J Badin, TITIS J O U R N A L , 80,
ETHYLENE POLYMERIZATION WITH A SOLUBLE CATALYST
Aug. 5 , 1960
change in either the yield or melt index of the resin. Under these conditions, the titanium valence is essentially unchanged, and the titanium tetrachloride is obviously inert in the system. Experiment 8 shows that aluminum bromide is also a good cocatalyst, and experiment 7 again shows that titanium tetrachloride is not an adequate substitute for the vanadium halide. The small amount of polymer obtained in experiment 6 without the external addition of a vanadium halide is due to the fact that this particular reagent grade sample of aluminum chloride contained a few parts per million of vanadium compounds. This latter effect is more pronounced if certain commercial grades of aluminum bromide, which contain O.O1-O.lyovanadium halides, are used without purification. However, rigorous purification of these materials by double distillation gives an aluminum bromide that is inactive as a cocatalyst with tin tetraphenyl, either alone or in combination with titanium tetrachloride under the conditions described here (expt. 7). External addition of a small amount of a vanadium halide regenerates the catalytic activity (expt. 8). TABLE I Millimoles per I500 mI.---TiCla Sn(C6Hd.l
7--
Expt.
AlCh
1 2
3 3 3 3 3 3
3 4
5 6
VCh
0.0 1.2 3.6 6.0 12.0
1.5 1.5 1.5 1.5 1.5
0.025
..
1.5
...
,025 ,025 ,025 025
Polymer yield,a g .
6585 58 71 81 71 6
A I B ~ ~
7 3 1-12 1.5 ...
