Deprotonation of coordinated ethylene may start Phillips catalysis

The metal-catalyzed polymerization of simple olefins, such as ethylene and propylene, has been one of the transformational discoveries in chemistry of the 20th century. It is difficult to imagine today’s civilization without ubiquitous plastics, such as polyethylene and related polymers. Although often attributed to Karl Ziegler and Giulio Natta, the winners of the 1963 Nobel Prize in Chemistry for the discovery of titanium-based catalysts and their use in propylene polymerization, much (∼40–50%) of the polyethylene produced worldwide is actually made with catalysts containing chromium as the catalytic metal. Discovered in 1951 by Paul Hogan and Robert Banks at the Phillips Petroleum Co. in Bartlesville, Oklahoma, the so-called “Phillips catalyst” remains a major workhorse of the polymerization industry to this very day (1). Despite its significant commercial impact, this apparently “simple” heterogeneous catalyst has been the focus of longstanding scientific arguments regarding its mechanism of operation. Most enigmatic among the questions posed is the issue of its initiation. In other words, because the formation of the polymer chains presumably proceeds by repeated insertion of olefin monomer into the metal–carbon bond of an ever-growing chromium alkyl (2), what sequence of chemical events transforms the inorganic catalyst precursor into an organometallic compound—i.e., a material containing a covalent chromium–carbon bond—that catalyzes the polymerization? Delley et al. address this important question, among others, in their elegant study published in PNAS (3).

To appreciate the conundrum posed by the Phillips catalyst, it is necessary to review its preparation and the conditions of its use. At the most basic level, the catalyst consists of chromium atoms bound to the surface of an oxide support (typically silica; i.e., SiO₂). It is prepared by contacting the support particles with an aqueous solution of a chromium compound, followed by calcination in oxygen at high temperatures (600–900 °C). Although hexavalent chromium compounds (e.g., CrO₃) were originally used as chromium source, concerns about the carcinogenicity of the former eventually led to the substitution of a variety of simple Cr(III) salts as precursors. Either way, reaction with oxygen at high temperature oxidizes the surface-bound chromium to Cr(VI). Another important effect of the heat treatment is the condensation of silanol groups [(-O)₃Si-OH], which terminate the surface of silica under ambient conditions. Extrusion of water leaves behind a surface populated by hexavalent chromium, likely in the form of chromate esters, Si–O–Si linkages, and the occasional isolated silanol group. A schematic representation of this transformation is shown in Fig. 1A.

Fig. 1.

(A) Treatment of silica particles with a chromium salt solution, followed by calcination in oxygen at 600–900 °C, yields the Phillips catalyst precursor. It consists of hexavalent chromium sites on a dehydroxylated silica surface. (B) As proposed by Delley et al. (3), reduced chromium (III) sites bind ethylene; heterolytic cleavage of a C–H bond of the latter generates an organometallic vinyl chromium complex. (C) Rapid, repetitive insertion of ethylene into the chromium-carbon σ-bond leads to growth of the polymer chain.

The hexavalent catalyst precursor can be directly introduced into the polymerization reactor. However, this process results in an induction period, with full catalytic activity developing over the course of an hour or so. During this time, the chromium is likely reduced and transformed into an organometallic derivative (i.e., it acquires a ligand—such as alkyl group or a hydride—that can start a polymer chain). In this context it is worth noting that the Phillips catalyst does not require any cocatalyst, which might serve as the source of the organic ligand. Typical additives for Ziegler/Natta catalysts are aluminum alkyls (AlR₃), which are thought to transfer their alkyl groups to the titanium. Because no such compounds are needed for the development and maintenance of the catalytic activity of the chromium catalyst, it must somehow acquire its alkyl group from the substrate (ethylene, C₂H₄).

The chemistry of chromium derives much of its interest from the multitude of possible oxidation states exhibited by this metal, ranging from 0 to +VI. The decades since the discovery of the Phillips catalyst have seen a lively discussion regarding the oxidation state of the active catalyst, with almost all possibilities having been proposed at some time. Spectroscopic studies of the actual catalyst have been hindered by the fact that only a small fraction of the surface chromium is catalytically active. A recent review of much of this work concluded that “the still open questions remain [regarding] the structures of the active sites and the exact initiation/polymerization mechanism” (4) (i.e., much that is of interest). Under these circumstances, the study of well-characterized homogeneous model compounds has proven helpful. The preponderance of the evidence from this work suggests that coordinatively unsaturated or substitutionally labile Cr(III) alkyls consistently polymerize ethylene (5, 6). Combined with the well-established fact that Cr(III) is by far the most stable/common oxidation state of this element, Occam’s razor would suggest that the “active” Phillips catalyst (i.e., the surface species that actually inserts ethylene into the growing polymer chain) contains trivalent chromium [Cr(III)]. Indeed, previous experiments from the Copéret laboratory showed that dinuclear Cr(III) complexes grafted onto silica are catalytically active, whereas related Cr(II) sites showed little activity (7). In the present contribution (3), this observation is extended to mononuclear Cr(III) sites, which are considered more realistic models of the Phillips catalyst. The crucial point here is that it has long been known that the hexavalent Phillips catalyst precursor can be reduced all of the way to Cr(II) (for example with CO), and that reduction of the precursor eliminates the induction period (8). What has been missing is a convincing pathway for a reaction of Cr(II) with ethylene to generate a chromium alkyl. However, Cr(II) is easily oxidized to Cr(III) and the present paper (3) provides evidence for a straightforward transformation of a Cr(III) ethylene complex into a Cr(III) vinyl species by a proton transfer (Fig. 1B). By this pathway, easily accessible Cr(III) can enter a catalytic cycle (Fig. 1C).

Copéret et al. deliver a powerful combination of careful experimental “surface organometallic chemistry” (9) and theoretical modeling (with density functional theory calculations). The experimental work involves the grafting of a molecular Cr(III) siloxide

The paper by Delley et al. sets a high bar for those who would cling to more esoteric schemes for the formation of one of the most important industrial catalysts.

onto dehydroxylated silica, followed by extensive characterization of the resultingmaterial by a wide variety of spectroscopic methods, thus establishing the formation of two different Cr(III) binding sites. These coordinatively unsaturated sites bind CO, and one of them catalyzes the polymerization of ethylene yielding a polymer that resembles the material produced by the Phillips catalyst. A revealing experiment—regarding the crucial initiation step—is the observation of Si–OD groups in the vibrational spectrum of catalyst treated with deuterium labeled ethylene (C₂D₄), whereas at the same time X-ray near-edge absorption spectroscopy shows that the oxidation state of chromium remains +III during this transformation. This result serves as direct evidence for the postulated heterolytic activation of C–H(D) bonds of the monomer. The spectroscopy is augmented with computational experiments on chromium siloxide clusters as models for the two surface sites. These calculations are validated by their agreement with experimental data (bond distances, CO stretching frequencies, and so forth), and they allow the estimation of thermochemical parameters for the individual steps in the catalytic cycle, including their activation barriers. Although the proton transfer step faces an appreciable activation energy, it appears to be the most accessible pathway to an active catalyst.

Although it seems unlikely that the report by Delley et al. (3) will end the long-standing discussion about the intricacies of the Phillips catalyst, it provides a wealth of evidence for a proposal that is both appealingly simple as well as compatible with the results of molecular organometallic chemistry. The paper by Delley et al. sets a high bar for those who would cling to more esoteric schemes for the formation of one of the most important industrial catalysts around.