HIGHLIGHT The Discovery of Metallocene Catalysts and Their Present State of the Art WALTER KAMINSKY Institute for Technical and Macromolecular Chemistry, University of Hamburg, Bundesstr. 45, D-20146 Hamburg, Germany Received 23 April 2004; accepted 3 May 2004 DOI: 10.1002/pola.20292 Published online in Wiley InterScience (www.interscience.wiley.com).

ABSTRACT: Metallocene and other transition metal complexes activated by methylaluminoxane are highly active catalysts for the polymerization of olefins, diolefins, and styrene, which was discovered at the University of Hamburg about 25 years ago. These catalysts allow the synthesis of polymers with a highly defined microstructure, tacticity, and stereo-

WALTER KAMINSKY

regularity, as well as new copolymers with superior properties such as film clarity, tensile strength, and lower extractables. A better understanding of the mechanism of olefin polymerization leads to findings of other new single site catalysts. The development of the metallocene/MAO-catalysts from their discovery to their present state of the art is presented. © 2004

Wiley Periodicals, Inc. J Polym Sci Part A: Polym Chem 42: 3911–3921, 2004

Keywords: metallocene catalysts; methylaluminoxane; olefin polymerization; zirconocene complexes; polyolefins; single site catalysis

Dr. Kaminsky received his Ph.D. and degree in chemistry from Hamburg University, Germany in 1971 for investigating the side reactions of biscyclopentadienyl-zirconiumdichloride and triethylaluminum. He then served as lecturer and continued his research on the determination of zirconocene/aluminumalkyl-complexes and recycling of polymers by pyrolysis in a fluidized bed process. During that time he and Hansjo¨rg Sinn discovered that the activity of metallocenes and other transition metals can be increased extremely by the addition of MAO as a cocatalyst. In 1979 he became full professor at the University of Hamburg for Technical and Macromolecular Chemistry and has served as director, dean, and president of German chemical society, Hamburg section. Dr. Kaminsky has received the following honors and rewards: elected to the Royal Society of Chemistry, London as honorary member in 1996; Honorary Professor of Zhejiang University, Hangzhou, China in 1998; Ralf Milkovic Memorial Lecturer at Akron University, 1999; recipient of European Foundation Prize, Koerber Foundation 1988; Karl Heinz Beckurts Award 1991; Alwin Mittasch Medal, Dechema 1995; Carothers Award of American Chemical Society, Delaware Section 1997; Walter Ahlstro¨m Prize of Finland Academy of Technology, 1999; Gold Medal of Benjamin Franklin Society, 1999; Outstanding Achievement Award of SPE Division, 1999; Hermann-Staudinger Prize, 2003.

Correspondence to: W. Kaminsky (kaminsky@chemie. uni-hamburg.de). Journal of Polymer Science: Part A: Polymer Chemistry, Vol. 42, 3911–3921 (2004) © 2004 Wiley Periodicals, Inc.

3911

3912

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 42 (2004)

INTRODUCTION Metallocene/Methylaluminoxane (MAO) catalyst systems for olefin polymerization were discovered about 25 years ago in our laboratory in Hamburg.1–3 These catalyst systems have only one kind of active metal center (single site catalysts) and were the start-up for the introduction of a new family of transition metal complexes for olefin, diene, and styrene polymerization. Up to now, some 10,000 articles and more than 4000 patents have been published on the subject of polymerization with these single site catalysts. The interest has been and still is high because a method was presented to chemists, material scientists, and engineers to tailor the microstructure, that is, comonomer distribution, stereo-, and regioregularity of polymers, which was not possible before.4 – 8 At the same time, the production of polyolefin materials, using mainly Ziegler–Natta or Phillips catalysts or the high pressure process, and also more and more the new metallocene catalysts, has increased from 25 million tons/year (1980) to 90 million tons (2003)—a huge economic potential.9 A great step forward was made in understanding the mechanism of olefin polymerization by metal-organic complexes. In this review, I will recount how metallocene/MAO catalysts were discovered and briefly discuss the present state of the art.

THE DISCOVERY Up to 1975, a large number of homogeneous, soluble Ziegler–Natta-type catalysts had been investigated, primarily to understand the elementary steps of the polymerization reaction in heterogeneous systems.10 –15 Kinetic studies and application of various methods had helped to define the nature of the active centers to explain aging effects of analogous heterogeneous catalysts and to establish a model for catalyst-olefin interactions. Homogeneous catalysts based on biscyclopentadienyl-titanium or zirconium dialkyls, first introduced by Breslow,16 were assumed to be inactive for olefin polymerization. In the research group of Hansjo¨rg Sinn at the University of Hamburg, we investigated the reactivity of biscyclopentadienyl-titanium and zirconium complexes towards triethylaluminium and ethylene to isolate intermediates and the formation of olefin complexes.17 Zirconium complexes were preferentially used to study alkyl exchange and ␤-hydrogen transfer since zirconium is less easily reduced compared to titanium. Zirconium-aluminumalkyl complexes with unusual bonding angles (75.9 °) between the bridging carbon atoms were obtained (Fig. 1).18,19

Figure 1. X-ray structures of bis(cyclopentadienyl)zirconium ethylaluminum complexes.

To decrease the reduction in the titanium systems caused by slow ␣-hydrogen transfer, we also investigated the reaction of biscyclopentadienyl-titanium dimethyl with trimethylaluminum, mainly by NMR analysis at low temperature. To start the reaction, a toluene solution of the titanium complex and a solution of trimethylaluminum were mixed together at –78 °C by Schlenk technique and filled into an NMR tube. This NMR tube was then sealed with a torch, introduced into the NMR machine at – 40 °C, and slowly warmed to RT. This procedure was very time consuming, and the Ph.D. student involved with the measurements simplified the preparation method by filling the reaction mixture directly into a simple NMR tube and putting a plastic lid on.20 Comparing the NMR spectra obtained with the simple method with those obtained with the sealed NMR tubes, he discovered a small peak in the area of CH2-bonds in the NMR spectrum that had never been seen before. Fortunately, the student did not ignore these results, but discussed the effect with us and we decided to scale-up the experiment and repeat it in a 1 L glass autoclave that was cooled down to ⫺40 °C and subsequently warmed to ⫺14 °C. After pressing on ethylene, the pressure decreased very slowly. We then opened a valve of the autoclave to take a sample of the solution for NMR control. After the addition of new ethylene to a pressure of 8 bars, the pressure decreased much faster than in the beginning. This procedure was repeated several times (Fig. 2) and the reactivity increased each time. What was

HIGHLIGHT

Figure 2. Original ethylene pressure measurements of the first polymerization experiment by 1 mmol Cp2Ti(CH3)2, 10 mmol Al(CH3)3 in 600 mL toluene at ⫺15 °C; short opening of the reactor after 1, 8, 21, and 24 h. The autoclave was refilled several times at 8 bar ethylene pressure and then closed.

the reason for this increasing polymerization rate of ethylene by a system known before to be inactive? We discussed the results with our supervisor, Hansjo¨rg Sinn, and came to the conclusion that traces of chloride, oxygen, or water could be responsible for the activity. In the following experiments, using titanium complexes [Cp2Ti(CH3)2] containing traces of chloride or oxygen, no polymerization of ethylene was observed. Finally we added small amounts of water, as a last possibility since water has been discussed to be a strong poison for olefin polymerization catalysts. Surprisingly, we observed a high polymerization rate depending on the amount of water. The polymerization rate reached a maximum when the ratio of water to trimethylaluminum was equimolar. A patent application was written and presented to the BASF AG. This patent application was later withdrawn, because the results were considered to be too similar to discoveries of Reichert,21 who had found an increase by small amounts of water to the halogen-rich homogeneous system Cp2TiEtCl/EtAlCl2

3913

two years before, and the work of Breslow22 with the system Cp2TiCl2/Me2AlCl. But we were fascinated by these results and continued to investigate this new, very active homogeneous catalyst system for ethylene polymerization. The next step was therefore to isolate the product formed in the 1:1 mixture of water and trimethylaluminum. At higher water levels, explosions could happen, so we used inorganic salts containing bonded water, first CuSO4 ⫻ 5 H2O and later Al2(SO4)3 ⫻ 14 H2O in toluene suspension23 to control the amount of water. After 20 h, the reaction mixture was filtrated and the solvent evaporated. The white powder obtained was dried and analyzed. The composition was approximately AlO0,7(CH3)1,5. We named the compound methylaluminoxane (MAO) in analogy to siloxane. The analogous ethylaluminoxane was not very active as a cocatalyst.24 MAO was soluble in toluene and other aromatic solvents. The use of this separately produced MAO together with Cp2Ti(CH3)2 further increased the activity by a factor of 100 compared to the system of Cp2Ti(CH3)2/ Al(CH3)3/H2O.25 Because of the problems with the first patent application, we did not write another one. But for the first time, we were able to polymerize propylene with a soluble biscyclopentadienyl-titanium catalyst to obtain atactic polypropylene and to generate ethylene-propylene copolymers. Pino and Mu¨lhaupt,26 who analyzed a sample, found that the polypropylene synthesized with this catalyst was the purest atactic PP they had ever seen. In 1979, we used biscyclopentadienyl-zirconium dimethyl and MAO for ethylene and propylene polymerization—and in contrast to some earlier experiments with the complex from Figure 1, we obtained extremely high activities, higher than for the titanium system. Up to this date, biscyclopentadienyl-zirconium complexes, activated by aluminum alkyls, were described to be totally inactive for olefin polymerization. After informing Hansjo¨rg Sinn, who had been Minister of Science and Research of the City State Hamburg since 1978, we decided

Table 1. Polymerization Activity of Cp2ZrCl2/MAO at 95 °C and 8 Bar Ethylene Pressure Activity Zirconocene concentration MAO concentration (molecular weight 1200 g/mol) Molecular weight PE Polymerization degree Time for formation of one PE chain Turnover time (insertion) of ethylene

39.8 ⫻ 106 g PE/g Zr ⫻ h 6.2 ⫻ 10⫺8 mol/L 7.1 ⫻ 10⫺4 mol/L 78,000 g/mol 28,000 0.087 s 3.1 ⫻ 10⫺5 s

3914

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 42 (2004)

to file a patent application covering these exciting results.1 Later, we discovered that also the simple Cp2ZrCl2, which is much more stable than Cp2Zr(CH3)2, is an active catalyst precursor in combination with MAO. In 1980 and 1981, I reported on zirconocene/MAO catalysts for the first time at the IUPAC Polymer Congress in Firenze27 and the Macromolecular Meeting in Midland.28 The zirconocene/MAO catalyst system was 10 to 100 times more active than the classical Ziegler catalyst. A complex containing 1 g zirconium produced 40 million grams of polyethylene in 1 h at 95 °C and 8 bar ethylene pressure (Table 1).29 Nearly every zirconium atom is an active center, as shown by Peter Tait30 and Jimmy Chien,31 and produces about 46,000 polymer chains per hour. The insertion time of one ethylene unit is only 3 ⫻ 10⫺5 s. This is similar to the insertion time observed for the synthesis of enzymes. The analogy can be seen in many other fields (influence of substitution, regioselectivity, stereospecificity) as described later. For the first time, we could show that a soluble catalyst, the highly active Cp2ZrCl2/MAO system, is able to produce polyethylene with molecular weights up to 1 million g/mol and a narrow molecular weight distribution of approximately two. All active sites are similar and form polymers with the same average chain length (single site catalysts). Only traces of low molecular weight oligomers are formed. The zirconocene/MAO catalyst polymerizes propylene to a perfectly atactic polymer. Copolymers of ethylene with propylene or higher ␣-olefins such as 1-butene, 1-hexene could be obtained as examples for metallocene based linear low density polyethylene (LLDPE). These results obtained with the Cp2ZrCl2 complex led to a second patent application in cooperation with the Hoechst company.32 The relatively high aluminoxane concentration (Al(MAO) : Zr ⫽ 5000:1) can be explained by the fact that MAO acts as a scavenger and in a lot of side reactions. In the following years, we continued kinetic measurements of ethylene and propylene polymerization, copolymerized ethylene with 1,7-octadiene and butadiene,33 and investigated the mechanism of olefin insertion into the zirconium-carbon bonds. The metal atom in titanocene and zirconocene complexes is linked to two rings of five carbon atoms and to two other groups (methyl, chlorine). The angles between the rings and their substitution pattern play a key role for the activity and stereospecificity. If pentamethylcyclopentadienyl-zirconium dichloride (Cp*2ZrCl2/MAO) is used instead of Cp2ZrCl2/MAO, polyethylene with much higher molecular weight is formed, but at lower activity.34 This means that chain

transfer reactions are much slower in this substituted zirconocene complex.

ISOTACTIC POLYPROPYLENE I remembered the hypothesis of Giulio–Natta that the formation of isotactic polypropylene depends on chiral titanium centers on the surface of TiCl3 crystals. To achieve the same with homogeneous catalysts, it is necessary to substitute the zirconocene in a way that it becomes chiral. We therefore synthesized the complex Cp*Cp ZrCl(CH3). The polymerization of propylene with this complex, however, was disappointing. Again, only atactic polypropylene could be obtained. Discussion with colleagues from the inorganic institute indicated that the movement of the four ligands around the metal center is much faster than insertion of an olefin. Such a complex does not have one stable enantiomer and therefore yields atactic polypropylene. Then I remembered that Hans-Herbert Brintzinger (University of Konstanz, Germany), whom I had met at a scientific program of the German research foundation (DFG), had tried to synthesize ansa titanocenes for the asymmetric hydrogenation of organic compounds; he had just published a paper of a chiral ansa bistetrahydroindenyl-titanium dichloride.35 I contacted him and asked for some milligrams of such an ansa metallocene, if possible with zirconium as metal, since Cp2Zr(CH3)2 had been much more active than Cp2Ti(CH3)2. Fortunately, he had a student who was nearly ready with the synthesis of the ansa zirconocene. In spring 1984, my Ph.D. student Klaus Kuelper traveled from Hamburg to Konstanz to pick up the zirconocene. Back in Hamburg, we carried out the first polymerization runs using this complex. As anticipated, isotactic polypropylene was formed. This could clearly be seen by the formation of a suspension of insoluble polymer in the autoclave while all our previous experiments had given a clear solution of atactic polypropylene. The ansa zirconocene [Et(THind)2]ZrCl2 exists in three structures (Fig. 3). The rotation of the indenyl rings is hindered by the CH2-CH2-bridge. Besides the R and the S forms, a meso form is possible. In the case of [Et(THind)2]ZrCl2 only traces of the meso form are obtained, which can be eliminated by re-crystallization of the complexes. The meso form has no asymmetric symmetry and produces only atactic polypropylene similar to the unbridged Cp2ZrCl2/MAO catalyst. The [Et(THInd)2]ZrCl2/MAO catalyst used produced some stereo errors. The errors measured by 13C-NMR spectroscopy are randomly distributed along the polymer chain; while in polypropylene made with Ziegler–Natta catalysts, the errors are concentrated at chain ends and in

HIGHLIGHT

Figure 3. Structures of the R, S, and meso form of [Et(Thind)2]ZrCl2.

oligomers. The former is called isoblock-polypropylene and is characterized by a lower melting point but higher film transparency. We investigated the influence of temperature and catalyst concentration on molecular weight and isotacticity and wrote together with the research group of Hans Brintzinger a publication.36 During this, we read that John Ewen,37 working for Exxon, had carried out similar experiments with the ansa-[Et(Ind)2TiCl2]/MAO catalyst and had obtained a mixture of isotactic and atactic polypropylene due to the fact that the complex had not been clean and had contained a high amount of meso complex. The metallocene had shown a low activity and had been destabilized by the addition of MAO and at higher temperatures. The discovery of the production of isotactic polypropylene by a highly active homogeneous metallocene catalyst increased the interest of research groups working worldwide in the field of olefin polymerization. In the following years, the number of patents increased from 1 in 1980 to 300 in 1990. The new catalyst provided the opportunity to tailor not only the polymer structure but also the tacticity of polyolefin resins in combination with an extremely high activity. Researchers of the Hoechst company optimized later the ansa zirconocene complexes by using different bridges and substituents at the indenyl rings.38 They were able to obtain isotactic polypropylene with an ac-

Figure 4. Structure of (neomenthyl-C5H4)2ZrCl2.

3915

tivity of 900,000 kg PP/molZr䡠h, a molecular weight of 700,000 g/mol, an isotacticity of 99%, and a melting point of 160 °C. Another step forward was the synthesis of a Cs-symmetric zirconocene complex with a bridged cyclopentadienyl and a fluorenyl ring in 1987 by Ewen, Jones, and Razavi,39 working for Fina Corp. This metallocene offers two different bonding positions for the inserting propylene. As a consequence of the Brintzinger, Corradini, Cossee mechanism,40 the chain migration allows this metallocene to produce syndiotactic polypropylene with high activity and again with a narrow molecular weight distribution of Mw/Mn ⫽ 2. One year before, in 1986, I had discussed with my coworkers the question, what would happen if the rotation and change of place of the aromatic rings in unbridged zirconocenes were hindered by bulky substitutions. Maria Buschermoehle synthesized a metallocene in which a hydrogen atom of the cyclopentadienyl ring was substituted by a neomenthyl group (Fig. 4).41 Such a bulky ligand could stabilize an asymmetric (chiral) active site for a short time. During this time, some propylene units could be inserted with the same stereospecificity forming an isotactic block, followed by a change in conformation. As the movement is slower at low polymerization temperature, the isotactic block length must increase with decreased temperature. Indeed, we obtained an elastomeric polypropylene with a stereoblock microstructure and an increasing isotacticity with decreasing polymerization temperatures. Using different ligand substituents, it was now possible to obtain isotactic, isoblock, stereoblock, syndiotactic, and atactic polypropylenes in high purity (Fig. 5). All these polypropylenes with different microstructures show special physical and mechanical properties. Encouraged by this success, we used the [Et(THInd)2]ZrCl2/MAO catalyst to polymerize different types of olefins such as 1-butene, 1-pentene, 1-hexene, and cyclic olefins.42

Figure 5. Microstructures of polypropylenes produced by different metallocene catalysts.

3916

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 42 (2004)

Table 2. Homopolymerization of Cycloalkenes with [Et(Ind)2]ZrCl2 (I), [Me2Si(Ind)2]-ZrCl2 (II), and MAO as Cocatalyst

Monomer

Catalyst

Polymerization Temp. (°C)

Activity (kg poly/mol Zr 䡠 h)

Melting Point of the Polymer (°C)

Cyclobutene Cyclopentene Cyclopentene Norbornene

I I II I

0 30 30 20

149 480 810 150

⬎400 (decomposition) 395 395 ⬎400 (decomposition)

Surprisingly, it was possible to homopolymerize cyclic olefins such as cyclobutene, cyclopentene, and norbornene to partially crystalline materials with high melting points (see Table 2).43 While it is very difficult to polymerize strained cyclic olefins with heterogeneous Ziegler–Natta catalysts without ring opening, the zirconocene aluminoxane catalyst system polymerizes these olefins reacting only with the double bond. The unusual formation of 100% of 1,3enchainments for poly(cyclopentene) was later shown by Collins.44 The rate of isomerization is enhanced relative to the rate of insertion of the olefin into the metal carbon bond, which is reduced through non bonding repulsion of the ligands around the Zr center.

COPOLYMERS Since metallocene catalysts were found to be highly active polymerization catalysts not only for propylene and ethylene but also for longer chain 1-olefins, they were suitable for different copolymerizations. Between 1980 and 1987 we synthesized the following copolymers for the first time with metallocene catalysts using the complexes Cp2Ti(CH3)2, (CH3C5H4)2Ti(CH3)2, Cp2Zr(CH3)2, Cp2ZrCl2, Cp2*ZrCl2, Cp2TiCl2, [Et(THInd)2]ZrCl2, [Et(Ind)2]ZrCl2: ethylene ethylene ethylene ethylene ethylene ethylene ethylene ethylene ethylene ethylene

– – – – – – – – – –

propylene (EP)45 propylene, diene (EPDM)46 1-butene (LLDPE)47 1-hexene (LLDPE)46 1,5-hexadiene (elastomer)46 cyclopentene (COC)48 cycloheptene (COC)48 cyclooctene (COC)48 1,3-butadiene (elastomer)47 isoprene (elastomer)47

While the comonomers are distributed randomly in the polymer chain, only low amounts are needed to decrease the density and the melting point. The low

amount of oligomers compared to copolymers produced with Ziegler–Natta catalysts is responsible for a high tensile strength and other mechanical properties of the obtained LLDPE. Our main interest, however, was focused on the copolymerization of ethylene with cyclic olefins. The homo poly(cycloolefins) are not processible due to their high melting points and their insolubility in common organic solvents. Therefore, we copolymerized cyclic olefins with ethylene or other ␣-olefins and obtained cycloolefin copolymers (COC), representing a new class of thermoplastic, amorphous materials.43 The metallocene catalysts are much more active than previously used vanadium-based catalysts, and the comonomer distribution can be varied from statistical to alternating. Statistical copolymers are amorphous if more than 10 –15 mol % of cycloolefins are incorporated in the polymer chain. The glass transition temperature can be varied over a wide range by selection of norbornene as cycloolefin and variation of the amount of norbornene incorporated into the polymer chain.49 Cycloolefin copolymers are characterized by excellent transparency, high glass transition temperatures of up to 200 °C, and excellent long-life service temperatures. They are resistant to polar solvents and chemicals and can be meltprocessed. Due to their high carbon/hydrogen ratio, these polymers have a high refractive index (1.53 for an ethylene/norbornene copolymer at 50 mol % incorporation). Their stability against hydrolysis and chemical degradation, in combination with their stiffness, makes them interesting materials for optical applications, for example, in compact discs, lenses, optical fibers, or films.50 Meanwhile, ethylene-norbornene COC-material is commercially available under the name of TOPAS (Ticona, Celanese). Investigating the kinetics of the different copolymerization reactions, we observed that in many cases the polymerization rate was higher than that of ethylene homopolymerization. For example, the highest ethylene1-hexene copolymerization rate was reached at 50 mol % of hexene in the starting mixture. It was twice as high as the rate for the homopolymerization of ethylene. A ratio

HIGHLIGHT

3917

of ethylene/norbornene ⫽ 0:4 gives the highest activity of 3780 kg polymer/molZr 䡠 h, which is about 5 to 6 times higher than that of the homopolymerization of ethylene. This comonomer effect was later also observed by other research groups. A convincing explanation has still not been given; the different theories—a higher solubility of ethylene in the comonomer, a faster diffusion or shifting of the equilibrium of an active/sleeping site— cannot explain such a high effect.

CHIRAL POLYMERS/OLIGOMERS The soluble chiral metallocene catalyst systems offer another new possibility. By insertion of propylene or higher 1-olefins, a new stereocenter is formed in the growing polymer chain. In heterogeneous Ziegler–Natta catalysts, the amount of stereospecific centers with Rand S-configuration is equal, thus producing a mixture of both structures: CH3 CH3 P P Ti OCH [ OCHOCH OCHO ]n 2 2 Ti OCH [ ]n 2OCHOCH2OCHO P P CH3 CH I thought it must be possible to obtain only one of these configurations by using only one enantiomer of the racemic mixture of the ansa zirconocene shown in Figure 3. Hans-Herbert Brintzinger succeeded in separating Rand S-ansa zirconocene complexes by preparation of diastereomers with optically active compounds like (R) binaphthol or O-acetyl-(R)-mandelate.51 We used the mandelate complex since its catalytic activity is similar

Figure 6. Structure of (S)-[1,1⬘-ethylene bis(4,5,6,7tetrahydro-1-indenyl)]zirconiumbis(o-acetyl-(R)-mandelate and methylaluminoxane.

Figure 7. Reaction scheme for the oligomerization of olefins by chiral zirconocene/MAO catalysts: a) initiation and propagation, and b) chain termination.

to that of the corresponding dichloride and it can be obtained in higher purity than the binaphtholate (Fig. 6). Only the S-ansa zirconocene forms a complex, while the R-form does not react with the R-mandelate. The complex can be separated by crystallization from the unreacted form and obtained in a purity of 99%. The polymerization starts when an olefin undergoes insertion into the zirconium carbon bond formed by methylation with MAO (Fig. 7). Subsequent insertions lead to long alkyl chains. Chain growth termination takes place by ␤-hydrogen transfer to the transition metal center or a metal-bound olefin, resulting in formation of a zirconocene hydride or alkyl and an olefin terminated polymer or oligomer chain. We hoped to receive a polypropylene crystallizing only in a right or left handed helix that would be optically active by this procedure. We never obtained such an optically active polypropylene; crystallization of polymer chains initiates from the middle of the chain, forming left and right handed helices at the same time. I then focused on the formation of oligomers. Here, the situation is different. If our hypothesis about the production of only one of the stereo configurations with the S-ansa zirconocene was right, then the trimeric oligopropylene or higher oligomers would have to have one or more chiral carbon atoms and would have to be optically active.52 The conditions to obtain oligomers were extreme since the catalyst normally produces high molecular weight polymer chains. High S-zirconocene and very low propylene concentration, fed continually, were chosen for this oligomerization at different reaction temperatures. The yield of trimers was up to 14%. They were separated by distillation from the other oligomers. But also the trimers were a mixture of different isomers. At 50 °C oligomerization temperature and a propylene feed of 10 mL/min, 88% of the timeric fraction consists of 2,4-dimethyl-1-heptene. Other isomers were formed by double bond migration, 2,1- and 1,3-insertion, and were a first evidence for the mechanisms of chain termi-

3918

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 42 (2004)

Figure 8. Asymmetric oligomerization of propene; gaschromatographic separation of 2,4-dimethyl-2-heptene (trimer) using an octakis(6-o-methyl-2,3-di-o-pentyl)cyclodextrine phase.

nation in the olefin polymerization by metallocene catalysts. It was luck that I know Wilfried Koenig, a colleague in the organic institute at Hamburg University who is an expert in gaschromatographic separation of racemic mixtures. He was able to separate our trimeric oligopropylene into the pure enantiomers by a cyclodextrine phase (Fig. 8).53 It can be seen that we obtained a trimer with 95,3 enantiomeric excess (ee) at 20 °C oligomerization temperature. The tetrameric oligopropylene shows a higher optical rotation than the trimeric. This was the direct proof that a chiral (pure enantiomer) soluble metallocene/MAO catalyst produces optically active oligomers and isotactic polymers. Similar results had been found just before by Pino54 using hydrooligomerization to obtain saturated branched alkanes.

HETEROGENIZATION OF METALLOCENES There was another phenomenon that held a large fascination for me. Compared to the traditional heterogeneous Ziegler–Natta catalyst, we had a new very active soluble catalyst in our hands. This catalyst could be adsorbed on surfaces of particles and fibers, covering them uniformly with active sites. As MAO is relatively stable against OH-groups covering inorganic, metal, or biomass surfaces, the aluminoxane can be fixed without losing its activity. After treatment with metallocenes and addition of olefins, the particles are covered with a thin polyolefin film. The polymer thickness depends on the polymerization time. With this procedure, new materials can be obtained combining the properties of polyolefin resins with those of fillers such as starch grains, cellulose fibers, carbon fibers, metal powders, silica monospheres, or nano-particles. Figure 9 shows polyethylene on the surface of starch.33 Every starch grain is covered with polyethylene particles of similar size. Polyethylene can

be filled with up to 85 vol % aluminum powder without significant loss of electrical resistance. This can be explained by a closed shell of polymeric material around each aluminum particle, which prevents the metal particles from forming electrically conductive paths. On the other hand, an exponential increase with growing filler content can be observed for the thermal conductivity. These early experiments were the start for both the synthesis of polyolefin composites and the supporting of metallocene catalysts on silica or alumina to use them as heterogeneous catalysts in existing technical processes (drop-in technology).55

ACTIVATION MECHANISM The start for the investigation and discovery of the metallocene/MAO catalysts was the interest in aging reactions of Ziegler–Natta catalysts. It is clear that we looked at the active site in metallocene catalysts shortly after their discovery. These investigations were complicated by the fact that the structure of MAO is a complicated mixture of basic units. Hansjo¨rg Sinn, back to Hamburg University in 1984, brought some light into the structure of the amorphous white powder. By cryoscopic measurements in benzene or liquid trimethylaluminum, decomposition reactions with water, hydrochloric acid, alcohols, and NMR-measurements, he came to the conclusion that MAO consists of basic units of Al4O3(CH3)6.56 These units can combine to form cage structures of preferentially 4 units, which could be the active form. Rytter is convinced that three units are sufficient because bridging methyl groups are already existing.57 By IRand UV-investigations we found that the metallocene and MAO rapidly form a complex (Fig. 10). Alkylation can be seen by IR- and NMR-experiments. Shilov and Dyachkovsky,58 Eisch,59 Jordan,60 and Bochmann61 showed that the activity of metallocene catalysts depends on the formation of cationic species.

HIGHLIGHT

3919

Figure 9. Starch grains covered with polyethylene; catalyst Cp2ZrCl2/MAO.

Today, most research groups agree with this statement. Not totally clarified is the function of MAO during the polymerization and why we need such a high excess. One explanation is the fact that MAO acts as a scavenger

on impurities. Another explanation goes back to our starting investigations. ␣-Hydrogen transfer reactions from a zirconium methyl bond to a methyl group of another zirconium complex or MAO can occur in the

Figure 10. Reactions of zirconocenes with MAO; formation of active sites and deactivation.

3920

J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 42 (2004)

system. The fast formation of methane was measured to reach one mole CH4 per mole of zirconocene in 10 min. To reactivate the inactive Zr-CH2-Al-structure formed, a high excess of MAO is necessary. This and other equilibriums exist between active and inactive complexes.62 Knowing the function of MAO for the activation of metallocene catalysts, other bulky and weakly coordinating cocatalysts such as perfluorophenyl-borate anions, boranes, and aluminum fluorides were introduced by Marks and Hlatky63,64 and others.

PRESENT AND FUTURE ACTIVITIES The discovery of metallocene/MAO catalysts has opened up the door to a much better understanding of the mechanism of olefin polymerization and to the creation of new types of single site catalysts. Metallocene-based catalyst systems are dramatically different from Ziegler–Natta catalysts as their homogeneous nature leads to lower polydispersities and more uniform incorporation of comonomers. The resulting improvements in the properties and manufacturing of polyolefin resins by metallocene catalysts represent a revolution in the polymer industry. Polymerization of olefin monomers with single site catalysts allows the synthesis of polymers with a highly defined structure and superior properties such as film clarity, tensile strength, and lower extractables. For chemists investigating organometallic compounds, a wide research field was opened that shows useful applications and sometimes surprising properties of many of their synthesized complexes. Investigations have not ended up to now in this fascinating research field. Another important step were the new catalyst developments by Brookhart65 and Gibson,66 who used diimine nickel, palladium, iron, and cobalt complexes in combination with methyl aluminoxane or other bulky cocatalysts. These catalysts allow the preparation of highly branched polyolefin resins. Such late transition metal catalysts show less oxophilicity than titanium or zirconium complexes and are less poisoned by functionalized polar vinyl monomers such as methyl(meth)acrylate, (meth)acrylic acid, and vinyl acetate. Copolymers containing these monomers, up to now prepared by freeradical processes, can be obtained by single site catalysts in the future. Some new titanium complexes show living character and give the possibility to produce polyolefin resins with extremely high molecular weight or block copolymers.67 In addition, copolymerizations of ethylene with cyclic olefins and styrene open the door to new commodity polymers.

All these different kinds of catalyst systems offer designed polyolefin resins with improved strength, toughness, enhanced optical and sealing properties, and increased elasticity and cling performance. These polymers have many applications such as film packaging, flexible films, adhesives, foam products, cable insulation, and materials for lenses. Flexible polyolefin resins can also replace plasticized PVC in many applications without requiring the addition of environmentally damaging plasticizers and stabilizers. The metallocene technology has matured significantly in the last years, starting from small production with higher prices to an increasing factor in the polymer industries.68 New estimated fields for polypropylene are nanocomposites and block copolymers. The detailed understanding of the mechanisms in metallocenes and other single site catalysts by structures and modes that have an analogy to enzymes will push the whole field of catalysis significantly.

REFERENCES AND NOTES 1. Sinn, H.; Kaminsky, W.; Vollmer, H. J; Woldt, R. DE Patent 3007725, 29, Feb. 1980. 2. Sinn, H.; Kaminsky, W., Vollmer, H. J. Angew Chem Int Ed Engl 1980, 19, 390. 3. Sinn, H.; Kaminsky, W.; Adv. Organomet Chem 1980, 18, 99. 4. Brintzinger, H. H.; Fischer, D.; Mu¨lhaupt, R.; Rieger, B.; Waymouth, R. Angew Chem Int Ed Engl 1995, 107, 1255. 5. Kaminsky, W.; Macromol Chem Phys 1996, 197, 3907. 6. Chen, E. Y.-X.; Marks, T. J Rev 2000, 100, 1391. 7. Scheirs, J.; Kaminsky, W. Metallocene Based Polyolefin; Wiley: Chester, 2000; Vols. Ia, II. 8. Terano, M.; Shiono, T.; Eds. Future Technology for Polyolefin and Olefin Polymerization Catalysis; Technology Publishers: Tokyo, 2002. 9. Sinclair, K. B. Macromol Symp 2001, 173, 237. 10. Natta, G.; Pino, P.; Mazzanti, G.; Giannini, U. J Am Chem Soc 1957, 79, 2975. 11. Chien, J. C. W. J Am Chem Soc 1959, 81, 86. 12. Henrici-Olive´, G.; Olive´, S. J Organomet Chem 1969, 16, 339. 13. Sinn, H.; Patat, F. Angew Chem 1963, 75, 805. 14. Fink, G.; Rottler, R.; Schnell, D.; Zoller, W. J Appl Polym Sci 1976, 20, 2779. 15. Keii, T. Kinetics of Ziegler–Natta Polymerization; Kodansha Ltd.: Tokyo, 1972. 16. Breslow, D. S.; Newburg, N. R. J Am Chem Soc 1957, 79, 5072. 17. Kaminsky, W.; Vollmer, H.-J.; Heins, E.; Sinn, H. Makromol Chem 1974, 175, 443. 18. Kaminsky, W.; Sinn, H. Justus Liebigs Ann Chem 1975, p. 424.

HIGHLIGHT 19. Kaminsky, W.; Kopf, J.; Thirase, G. Justus Liebigs Ann Chem 1974, p. 1531. 20. Mottweiler, R. Thesis, University of Hamburg, 1975. 21. Reichert, K. H.; Meyer, K. R. Makromol Chem 1973, 169, 163. 22. Long, W. P.; Breslow, D. S. Justus Liebigs Ann Chem 1975, p. 463. 23. Andresen, A.; Cordes, H. G.; Herwig, J.; Kaminsky, W.; Merck, A., Mottweiler, R.; Pein, J.; Sinn, H.; Vollmer, H. J. Angew Chem Int Ed Engl 1976, 15, 630. 24. Kaminsky, W.; Miri, M.; Sinn, H.; Woldt, R. Makromol Chem Rapid Commun 2000, 21, 1333. 25. Herwig, J. Thesis, University of Hamburg, 1979. 26. Pino, P.; Muelhaupt, R. Angew Chem 1980, 92, 869. 27. Kaminsky, W.; Sinn, H.; Woldt, R. IUPAC Internat Symp on Macromolecules, 7–12 Sept., 1980, Firenze, Abstract Vol. 2, p. 59. 28. Kaminsky, W. 11th Midland Macromolecular Meeting, 17–21 Aug. 1981, Midland, Proceedings, Transition Metal Catalyzed Polymerizations, Quirk, R. P.; Ed.; MMI Press: New York, 1983; Part A, p. 225. 29. Kaminsky, W.; Sinn, H. IUPAC 28th Macromolecular Symposium, 12 July 1982, Amherst, USA, Proceedings, p. 247. 30. Tait, P. In Transition Metals and Organometallics as Catalysts for Olefin Polymerization; Kaminsky, W.; Sinn, H.; Eds.; Springer: Berlin, 1988. 31. Chien, J. C. W.; Wang, B. P. J Polym Sci 1989, Part A 27, 139. 32. Kaminsky, W.; Haehnsen, H.; Kuelper, K.; Woldt, R. (to Hoechst AG). Patent DE 3127133, 9th July, 1981. 33. Kaminsky, W. Nachr Chem Tech Lab 1981, 29, 373i. 34. Kuelper, K. Diploma thesis, University of Hamburg, 1981. 35. Wild, F. R.; Zsolnai, L.; Huttner, G.; Brintzinger, H. H. J Organomet Chem 1982, 232, 233. 36. Kaminsky, W.; Kuelper, K.; Brintzinger, H. H.; Wild, F. R. Angew Chem Int Ed Engl 1985, 24, 507. 37. Ewen, J. A. J Am. Chem Soc 1984, 106, 6355. 38. Spaleck, W.; Aulbach, M.; Bachmann, B.; Kueber, F.; Winter, A. Macromol Symp 1995, 89, 237. 39. Ewen, J. A.; Jones, R. L.; Razavi, A.; Ferrara, J. P. J Am Chem Soc 1988, 110, 6255. 40. Grubbs, R. H.; Coates, G. W. Accounts of Chem Research 1996, 29, 85. 41. Kaminsky, W.; Buschermoehle, M. In Recent Adv in Mechanistic and Synthetic Aspects of Polymerization; Fontanille, M.; Guyot, A.; Eds.; Reidel Publ. Dordrecht, 1987; p. 503. 42. Kaminsky, W. Angew Makromol Chem 1986, 145– 146, 149.

3921

43. Kaminsky, W.; Spiehl, R. Makromol Chem 1989, 190, 515. 44. Collins, S.; Kelly, W. M. Macromolecules 1992, 25, 233. 45. Herwig, J.; Kaminsky, W. Polym Bull 1983, 9, 464. 46. Kaminsky, W.; Miri, M. J Polym Sci 1985, 23, 2151. 47. Schlobohm, M. Thesis, University of Hamburg, 1985. 48. Spiehl, R. Thesis, University of Hamburg, 1987. 49. Kaminsky, W.; Bark, A.; Arndt, M. Makromol Chem Macromol Symp 1991, 47, 83. 50. Cherdron, H.; Brekner, M.-J.; Osan, F. Angew Makromol Chem 1994, 223, 121. 51. Schaefer, A.; Karl, E.; Zsolnai, L.; Huttner, G.; Brintzinger, H. H. J Organomet Chem 1987, 328, 87. 52. Kaminsky, W.; Ahlers, A.; Moeller-Lindenhof, N. Angew Chem Int Ed Engl 1989, 28, 1216. 53. Kaminsky, W.; Ahlers, A.; Rabe, O.; Koenig, W. In Organic Synthesis via Organometallics; Enders, D.; Gais, H.-J.; Keim, W. Eds.; Vieweg: Braunschweig, 1992; p. 151. 54. Pino, P.; Cioni, P.; Wei, J. Am Chem Soc 1987, 109, 6189. 55. Dutschke, J.; Kaminsky, W.; Lueker, H. In Polymer Reaction Engineering; Reichert, K. H.; Geiseler, W.; Eds.; Hanser Publ.: Munich, 1983; p. 207. 56. Sinn, H. Macromol Symp 1995, 97, 27. 57. Ystenes, M.; Eilertsen, J. L.; Liu, J.; Ott, M.; Rytter, E.; Storneng, J. A. J Polym Sci Part A Polym Chem 2000, 38, 3106. 58. Dyachkovskii, F. S.; Shilova, A. K.; Shilov, A. E. J Polym Sci 1967, Part C, 16, 2333. 59. Eisch, J. J.; Pombrick, S. I.; Zheng, G. X. Organometallics 1993, 12, 3856. 60. Jordon, R. F.; Dasher, W. E.; Echols, S. F. J Am Chem Soc 1986, 108, 1718. 61. Bochmann, M.; Wilson, L. M. J Chem Soc Chem Commun 1986, 1610. 62. Kaminsky, W.; Steiger, R. Polyhedron 1988, 7, 2375. 63. Hlatky, G. G.; Upton, D. J.; Turner, H. W. US Pat Appl. 1990 459921; Chem Abstr 1991, 115, 256897 v. 64. Yang, X.; Stern, C. L.; Marks, T. J Organometallics 1991, 10, 840. 65. Ittel, S. D.; Johnson, L. K.; Brookhart, M. Chem Rev 2000, 100, 1169. 66. Britovsek, G. J. P.; Bruce, M.; Gibson, V. C.; Kimberley, B. S.; Massox, P. J.; Mastroianni, S.; McTavish, S. J.; Redshaw, C.; Solan, G. A.; Stromberg, S.; White, A. J. P.; Williams, D. J. J Am Chem Soc 1999, 121, 8728. 67. Saito, J.; Mitani, M.; Matsui, S.; Kashiwa, N.; Fujita, T. Macromol Rapid Commun 2000, 21, 1333. 68. Stevens, JC. Stud Surf Sci a Catal 1996, 101, 11.