A Career in Catalysis: Max McDaniel - ACS Catalysis (ACS Publications)

Dec 19, 2017 - Still going strong in his seventies, some of the work of Dr. Max McDaniel is briefly described ... Heinz, Lutz, Simmons, Miller, Ewing,...
0 downloads 0 Views 6MB Size
Account Cite This: ACS Catal. 2018, 8, 602−614

pubs.acs.org/acscatalysis

A Career in Catalysis: Max McDaniel Kumudini C. Jayaratne,† Ted H. Cymbaluk,‡ and Michael D. Jensen*,§ †

Borealis, Innovation Center, Porvoo 06100, Finland Chevron Phillips Chemical Company, Pasadena, Texas 77506, United States § Formosa Plastics Corporation, Polyolefins Technology Division, Point Comfort, Texas 77978, United States ‡

ABSTRACT: Still going strong in his seventies, some of the work of Dr. Max McDaniel is briefly described in this account. Regarded as a titan of polyolefin catalysis by his industrial and academic colleagues, he arguably stands among the topmost important scientists to ever make contributions to the polyolefin industry. Several of the highlights and insights he has contributed to the field over his long and prolific career are summarized.

KEYWORDS: Phillips catalysts, Ziegler catalysts, metallocenes, long-chain branching, polyethylene, ethylene polymerization



INTRODUCTION Commercial ethylene polymerization by transition-metal catalysis began over 60 years ago at Phillips Petroleum with the discovery of Cr/silica catalyst by Hogan and Banks in 1951,1−3 and the discovery by Ziegler in 1953 of TiCl3−AlR3.4 Today the “Phillips” and “Ziegler” catalysts are used to produce the world’s most widely used plastic: linear polyethylene (HDPE and LLDPE) with an annual worldwide production of about 130 billion pounds.5 During this creative 60 years, both catalyst types have gone though many major advancements, and for over 40 of those years, Max McDaniel, a protégé of Paul Hogan, has led a small group of scientists at Phillips (now Chevron-Phillips) in the development of new catalysts and polymers. This perspective provides a brief review of some of his insights and contributions to the field. McDaniel began his career in chromium catalysis at Northwestern University where he received his Ph.D. in 1974 under Robert L. Burwell, writing his dissertation on the redox behavior of chromia.6,7 After a year as a postdoctoral fellow at the Institut Catalyse in Lyon, France, McDaniel accepted a position with Paul Hogan at Phillips Petroleum Co. in Bartlesville, Oklahoma. Authoring over 120 journal publications on polymerization catalysts, and almost 400 U.S. patents, McDaniel and his group have changed the PE industry with inventions that benefit mankind. McDaniel’s career spanned the period from the tail end of the post-Sputnik boom, when central R&D groups were the norm for industry, and long-term projects were on the severalyear, not the several quarter, scale; where management was willing to invest in high risk/high reward projects, to the current day, where R&D is often considered as just another cost-center, and technical expertise as fungible; and indeed, © XXXX American Chemical Society

where the Bunsen burner has largely been replaced by the heat gun, the latter being deemed safer by those with little experience using either. Through these headwinds, McDaniel continued to be an innovation machine, responsible for the discovery and development of nearly all the catalysts, from chromium and Ziegler to metallocene, used to make commercial resins by Chevron Phillips and its licensees around the world. These inventions have contributed greatly to society by, for example, providing stronger and lighter weight substitutes for conventional materials, and replacing materials derived from environmentally less friendly, or more energy intensive, processes. And though in his 70s, he does not appear to be slowing down. Max is still as prolific as ever, the last Mohican of the Hogan era, so to say, still creating at the Bartlesville Research Center. On the personal side, when not pursuing answers to chemistry’s persistent questions, he peppers conversation with quotes from Moby Dick, and Greek and Roman classics. And every weekend before July fourth, he can be found somewhere making his own fireworks. Nevertheless, he is truly modest, almost embarrassed, when the subject comes up about the number of patents or journal publications he has authored, the large body of which reflect a quest for knowledge, rather than an appeal to the bean-counting gods. McDaniel is perhaps best known for his “industrial” approach to Cr/silica science, deriving insights into polymerization catalyst behavior by observing the character of the polymer produced, something that is frequently overlooked by Received: October 5, 2017 Revised: November 20, 2017

602

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis academicians in the field. As he once wrote, “to understand the catalyst, you have to make polymer”. In other words, only through knowing the polymer can one begin to understand the nature of the catalyst. Particularly for Phillips catalysts, this approach forces one to selectively observe the behavior of only the small number of active sites (of total Cr on SiO2), a failing of most types of spectroscopic studies. It also allowed McDaniel to draw on the wealth of commercial experience accumulated over 60 years at Phillips and its many worldwide licensees, information that is usually not published except through patents, a jungle through which few academicians venture. That said, Max keenly follows academic contributions to the field, which are frequent topics at his group meetings. Two themes consistently stand out in all of McDaniel’s work. First, he observed the role that Lewis acidity, or electron deficiency, plays in determining the activity of the catalyst, and more importantly, in determining the type of polymer it makes. Controlling the character of the polymer made is nearly always the first goal in industrial research, not determining, for example, the active Cr valence, which dominates the scientific literature. Indeed, polymer properties always take precedence over catalyst activity, and the molding behavior usually dominates both of these considerations. Second, McDaniel observed how the physical structure of the catalyst support (silicas, aluminas, aluminophosphates, etc.) can independently play a major role in influencing not just the activity but more importantly the molding behavior and physical properties of the polymer made. In the following pages, we trace these two themes through some of his prolific career.

extensional viscosity, die swell in blow molding, warpage in injection and rotational molding, and bubble stability in film. To be most effective, an industrial catalyst scientist must be aware of all these polymer characteristics and how they relate back to the catalyst. Preferably, in order to tune it in a rational manner, he or she also wants to know what mechanistic considerations are driving LCB formation. Despite its central importance to the Phillips catalyst and to the PE industry in general, LCB from Cr/silica has received little, if any, attention in the scientific literature other than that contributed by the McDaniel group.8−13 Hogan was aware that Phillips polymers could contain LCB,14 but its importance to PE molding, its origin, and its connection with the catalyst were all unknown. In fact, LCB was thought to be somehow derived from terminal vinyl groups during pelletization. Because rheology was still in its infancy, and LCB could not (and still cannot) be measured by NMR or IR spectroscopy, Hogan had no way to even measure LCB in his polymers. Today we understand that vinyl end-groups on terminated chains may become reincorporated as branches into other chains, as depicted in Scheme 1. To be rheologically significant, such Scheme 1. Formation of Long-Chain Branching via the Intra-Molecular Mechanism



branches must be longer than the critical entanglement length, or about 150 carbons for PE.15 Thus, the numerous publications claiming to measure LCB by NMR are actually measuring oligomer-derived branches of possibly no further significance than to suppress polymer density.16,17 In several of his papers, McDaniel shows how the formation of LCB is affected by nearly all of the catalyst variables,8−13 including the Cr loading, the calcination temperature, the presence of titania or other metal oxides, the method of reduction, and some reactor variables such as reaction temperature, ethylene concentration, and the presence of poisons or cocatalysts. Trying to grasp a common underlying reason driving these diverse catalyst and reaction interactions, he concluded that the amount of LCB produced is a strong function of the degree of Lewis acidity of the active Cr sites. Applying these same ideas to metallocene chemistry does provide some parallels. For example, tightly bridged metallocenes generally produce higher LCB amounts than similar unbridged ones. Presumably tightly bridged species inhibit ligand π orbital overlap and donation, leaving the Zr cation in the active site more electropositive relative to their unbridged counterparts, and therefore, from a strictly electronic argument, they increase the bond strength of coordinated α-olefin comonomers, including terminated chains (i.e., macromers), which are better electron donors than ethylene. Thus, the more Lewis acidic catalyst sites, whether Cr or Zr, have an increased affinity for macromer. Of course, this argument ignores the steric contribution to LCB formation, which the McDaniel group has also studied but is beyond the scope of this paper. However, McDaniel often turns the conventional views on their head. Such is the case with the view that soluble catalysts, containing pendant olefins in their structure, can undergo “self-heterogenization”, whereby the pendant olefin is inserted into a growing polymer chain, as

LONG-CHAIN BRANCHING To understand the research contribution of McDaniel and his various co-workers, it is first necessary to appreciate the dominant role that long-chain branching (LCB) plays in governing all types of PE molding behavior. This includes (1) extrusion ease or pressure relative to the polymer MW; (2) melt strength or elasticity, which determines resistance to sag (e.g., in pipe and milk bottle manufacturing); (3) chain relaxation time, which influences polymer chain orientation in the final part and thereby its impact and tear resistance; and (4) melt fracture (Figure 1), which governs the degree of surface roughness on finished parts and also clarity in film. Generally, LCB is the predominant factor affecting melt behavior in the processing of manufactured parts, such as shear-thinning,

Figure 1. Melt fracture can be one of the many manifestations of longchain branching (LCB) in PE. Depending on the level of LCB, polymers can exhibit any of the above states during extrusion in a particular application. 603

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis popularized by Alt et al.18 However, McDaniel’s group19 found that the actual function of the olefin pendant on a bridged metallocene is to drive down to near zero the otherwise high LCB content that one would normally expect. An especially vivid example is shown in Scheme 2, where two analogous

encapsulated or entrained, but notice that in Figure 2C it quickly desorbed when a competing α-olefin electron donor is added (e.g. a few mL of 1-hexene). This conclusion is in sharp contrast to those in a 2005 Perspective in Dalton Transactions21 and picture that adorned the cover of another issue.18 Thus, the intramolecular mechanism helps to explain much of the otherwise confusing observations pertaining to LCB formation.



Scheme 2. Two Metallocenes Differing Only in the Bridge Substituent

SUPPORT ACIDITY The majority of PE manufacturing is done utilizing heterogeneous catalysts, principally SiO2 or MgCl2-based, even in solution reactors where the polymerization is done at high temperatures to keep the polymer soluble in the reactor. The electron density of the Cr active site is most often a function of the acidity of the support. Although silica is not noted for being particularly acidic, the addition of other metal oxides creates acidity. One very useful example is the effect of titania, which can fit into the tetrahedral silica lattice22−24 creating acidic hydroxyls to which Cr then attaches. Cr/silica− titania was first reported by Hogan25 in the 1960s, but it was then modified and embellished by others, including McDaniel’s group, to produce some very useful and now widespread commercial catalysts.26−35 McDaniel noted that the increase in acidity can be correlated with (1) increased activity, (2) lower MW polymer, and (3) increased LCB formation. Enhanced activity is the result of a new population of sites, possibly more active and probably more of them. These Ti-associated Cr sites are distinguished by producing considerably lower MW polymer, as shown in Figure 3. Electron deficiency of the Cr increases its tendency for

metallocenes are compared, one with a butyl substituent off the methine bridge, and the other with a butenyl substituent. The dramatic 2 orders of magnitude drop in LCB suggests that the process of LCB incorporation is not random. Instead, the terminated chain must have a tendency to remain coordinated to the active site for a time, even as another chain grows on the same site. McDaniel coined this the “intra-molecular mechanism of LCB formation”, which is shown in Scheme 1. The pendant butenyl group must compete for macromer coordination to the Zr, effectively shutting down LCB formation. Many other similar metallocenes have since been synthesized to successfully test the principle, which has been used commercially for over two decades.20 This principle is visually illustrated by the following experiment. Metallocene A in Scheme 2, with the saturated, butyl substituent, was activated by dissolving it into a toluene MAO solution, resulting in the beautiful magenta color shown in Figure 2A. Then ethylene was added to the solution at 25 °C to produce polyethylene, which, being insoluble, quickly sank to the bottom of the bottle, as shown in Figure 2B. However, notice that it (solid polymer) took the magenta colored species (presumably the cationic site) with it, consistent with macromer remaining coordinated to the Zr. One might argue that the metallocene was merely physically adsorbed,

Figure 3. Acidity in the support, from the combination of silica and titania, increases activity and lowers polymer MW.

agostic coordination of β-hydride, which is a precursor to H elimination and chain termination. It also increases the tendency of the Cr to retain the terminated macromer by strengthening coordination, leading to LCB formation as in Scheme 1. This is observed with silica-coated titania, as well as titania-coated silica. It is interesting that these effects are seen only with the combination of silica and titania, and not with titania alone, which yields little activity and very high MW PE. Noting the connection, McDaniel went on to explore many other acidic supports. Addition of species containing Zr, Al, Sn, PO4, and SO4 to Cr/silica produced very similar changes to that from Ti addition, that is, more activity, lower MW, and higher LCB.13 Cr/AlPO4 was an especially useful variant commercially

Figure 2. Metallocene precipitation with PE, then liberation by 1hexene (reproduced from ref 19). 604

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis

be extruded. Therefore, the grade is sometimes called the “extrudable UHMW-PE”. Other dopants were also investigated in attempts to increase the acidity of Cr/alumina catalysts. The addition of silica48−50 or boria43 increased activity up to 10-fold. In fact, in recent years novel materials have been synthesized that hold the record in activity for all known Phillips-type catalysts, well outdistancing even the most active Cr/silica−titania catalysts. Unlike silica, however, other dopants such as zinc or titania, when applied to Cr/alumina, had little effect on activity.

and has been used to produce many millions of pounds of specialty PE resins.36−42 Although AlPO4 is isoelectronic and isostructural with SiO2, it has a population of surface hydroxyl groups to which Cr can bind that is more varied in acidic strength than silica. This again broadens the MW distribution of the PE product. However, the increase in breadth is much greater than that derived from Cr/silica−titania. An example is depicted in Figure 4 which shows the MW distribution of PE from a Cr/AlPO4 catalyst. Note that there



METALLOCENE ACTIVATION BY SOLID ACIDS Aware of these phenomena, McDaniel postulated in the early 1990s that such materials might also be capable of ionizing metallocenes. Scheme 3 shows how such “solid acids” could be Scheme 3. Metallocene Activation by Solid Acids

capable of replacing the conventional and much more costly MAO and perfluorophenylborate activators. Note that this involves Lewis acidity, not the Bronsted sites to which Cr attaches. Nevertheless, he reasoned, and subsequent measurements have confirmed, that doping silica or alumina or similar supports to increase the Bronsted acidity also usually enhances Lewis sites as well. In fact, many formidable metallocene activators were then synthesized,51−53 and some have been commercialized to produce (in combination with the discoveries noted in Scheme 2) billions of pounds of metallocene-based PE. The technology was also licensed to other PE producers and its benefits realized beyond ChevronPhillips Chemical Co. There are some exceptions to the Bronsted/Lewis acidity generalization noted above. For example, chlorided zinc/ alumina is a strong metallocene activator but produces a poor support for chromium, yielding almost no activity. This is because it has Lewis acid sites but is essentially lacking Bronsted acidity. In contrast, when silica is treated with sulfuric or phosphoric acids, such materials perform well as Phillips catalyst supports, yielding the expected advantages of high acidity, but they are often useless as metallocene activators, because these solid acids generate no Lewis sites. Another interesting consequence of metallocene activation by solid acids results from the fact that such materials nearly always contain a distribution of Lewis sites varying in acidity. Generally, there is evidence that a given metallocene activated by the most acidic sites is much more active than the same metallocene activated by the weakest acid sites. This emphasizes how remarkably active a fully ionized metallocene can be. This is demonstrated in Figure 5, where the average metallocene activity is plotted against its loading on the solid acid. This assumes that the most acidic sites are occupied by metallocene first. In addition, the activation of metallocenes by solid acids as depicted in Scheme 3 should not place surface ligands into the inner coordination sphere of the zirconium, and therefore, the polymer MW and MW distribution should be dominated only by the steric environment around the Zr. This has indeed been observed, as the polymer retains a single-

Figure 4. MW distribution and 29Si NMR of Cr/aluminophosphates.

are now two distinct peaks in this bimodal distribution, corresponding to the two types of surface hydroxyls present, P−OH and Al−OH. Consistent with results from Cr/SiO2− TiO2 catalysts, the more acidic P−OH sites produce the lowMW peak, whereas Al−OH produce the high-MW peak. Also consistent with trends noted above, the more acidic the support (i.e., the more phosphate it contained), the higher the LCB content in the polymer. Treating these same surfaces with (CH3)3SiCl vapor affords the solid-state 29Si NMR spectra shown in Figure 4.36,42 Note that two broad peaks were obtained, corresponding to P−O−Si(CH3)3 and Al−O− Si(CH3)3, whose areas varied with the composition of the support. It was found that aluminas could also be doped with acidic components to achieve some spectacular increases in activity. For example, Cr/alumina catalysts are barely active on a commercial scale, but the addition of phosphate, fluoride, boria, silica, or sulfate greatly increases the acidity of the support and also raises its activity by up to 6-fold, well into the industrially useful range.13,43,44 Polymer MW also tends to decrease, although this is complicated by the already high MW alumina produces, which necessitates using H2 with these catalysts to regulate MW. The LCB content is localized in the mid/low MW range, which is consistent with being generated by the most acidic sites. The short-chain branch profile was absolutely flat, completely unprecedented from other Phillips catalysts. This was exploited to make the world’s first pipe under single reactor conditions and a single catalyst that met with PE-100 qualification (indicating high burst resistance).45−47 Because of the high MW, this PE grade is extraordinarily tough, being as abrasion resistant as the ultrahigh MW PE grades, which cannot 605

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis

Figure 5. Metallocene and solid acid activity versus loading.

site MW distribution and the MW does not change as the activator is varied. However, the LCB content does vary with the choice of solid acid, implying that (consistent with other observations noted above) the acidity of the support may influence the active Zr to retain macromer more or less strongly. The general observation noted in the last paragraph, that the solid acid does not influence the MW distribution, applies primarily to full-sandwich metallocenes that have sufficient organic substituents to prevent coordination of the Zr by the solid acid. In contrast, tetra-alkyl Zr compounds, such as those Ballard described,54 are also very active on solid acids, and produce a very broad MW distribution, suggesting major or maximum interaction with the support. In between these two extremes are sparsely substituted metallocenes and halfsandwich complexes, which are able to coordinate with the solid acid enough to broaden the MW distribution beyond that of a single-site.51 Another convenient consequence of metallocene activation by solid acids is the ability to incorporate other polymerizationactive metals directly into the support. For example, the doping of nonmetallocene Zr or Ti species onto a halided support results in Ziegler catalyst sites operating simultaneously with whatever metallocene is also present, each producing its own characteristic polymer MW and degree of branching. An example is shown in the upper part of Figure 6, where the MW distribution of the polymer obtained is the result of two catalyst types each operating independently on the same support to yield intimately mixed PE chains. The main central peak is due to metallocene and the long, high-MW tail is the result of using a fluorided silica−titania as the solid acid, which activated the Ti sites. In fact, it was the same silica−titania used as a support for chromium catalysts discussed above. The high MW tail tends to tie the growing catalyst/polymer particle together, thus resisting breakage, and inhibiting polymer swelling. Major increases in PE bulk density have been observed by this small modification. Chromium can also be incorporated onto the solid acid to produce a Phillips catalyst operating in combination with metallocene sites. Such Ziegler or Cr-based resin components, intimately mixed on the molecular scale in situ, thus avoids the necessity of postreactor mixing and extrusion. This can be done to lend melt strength and other properties to the polymer during molding (e.g., film blowing).

Figure 6. (Top) Addition of a high-MW tail by the solid acid. (Bottom) Bimodal polymer from two metallocenes adsorbed on the same solid acid.

Alternatively, two metallocenes can be fed simultaneously to be adsorbed only angstroms apart on the same solid acid surface. Each metallocene produces a single-site distribution of its own characteristic MW and branching content to make bimodal PE, or in principle, any combination of polymer types desired. An example of a bimodal polymer obtained in this way is shown in the lower part of Figure 6. This had been the “Holy-Grail” of the PE industry for decades. This method of making bimodal polymers is superior to the usual arrangement of a Ziegler catalyst traveling through multiple reactors because (1) the use of two single-site components provides less overlap between the peaks (greater short chain branch segregation, i.e., between the high density and low density components), which has produced some incomparable polymer properties; (2) the two peak heights can be varied to produce any ratio desired, unlike conventional multiple reactors in series with fixed volumes; and (3) lower density polymers can be made as easily as homopolymers by varying the metallocene structures, whereas multiple reactors must make at least one of the polymer extremes (very low MW or very low density) separately in one reactor and (4) it obviates the need for higher energy intensive shearing during extrusion to blend the polymers made in separate reactors. This technology is quite unique in the industry, inexpensive, and it allows for the catalyst to be made in situ at the PE manufacturing plant by simply pumping the three ingredients (solid acid, metallocenes, and aluminum alkyl cocatalyst) simultaneously into the reactor. Adjustments in the recipe can be made on the fly by control of the feed ratios. The world’s largest loop reactor was recently built on the U.S. Gulf Coast to utilize this technology, producing about 120 000 lbs PE/h. 606

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis



IN SITU 1-HEXENE GENERATION Phillips-type catalysts can also be made using organochromium compounds, such as bis(η 6 −aryl)Cr(0), dialkylCr(II), tetraalkylCr(IV), η3-allylCr(II or III), η5-cyclopentadienyl Cr(II), and the η 5 -pentadienyl analogues, among others.42,55−62 To make such organo-Cr catalysts, the usual process is reversed. That is, the support is first calcined alone, and then afterward it is treated with the organo-Cr compound. The chromium attaches by reacting with one or more surface hydroxyl groups, losing at least one organic ligand in the process, as illustrated in Scheme 4A. McDaniel observed that Scheme 4. Formation of Coexisting Mono- and Di-Attached Cr Species by Different Routesa

a

Figure 7. MW distributions from organo-Cr compounds on three different supports calcined at varying temperatures.

4A by addition of organo-Cr compound to partially dehydroxylated silica, 4B by addition of organo-Cr compound to calcined Cr/silica after reduction by ethylene, 4C by addition of an organo-boron compound to CO-reduced Cr/silica catalyst.

storage, purification, and plant feeding equipment for the comonomer. Another advantage is improved comonomer incorporation efficiency. When the α-olefin generating and consuming sites exist side-by-side on the catalyst surface, the consuming sites experience a local comonomer concentration that is higher than exists in the reactor as a whole. Consequently, a lower average concentration of comonomer is required in the reactor to produce a certain polymer density (crystallinity). This in turn also has benefits. α-Olefin comonomer is a better solvent for PE than isobutane, and therefore, the in situ generation of 1-hexene resulted in less polymer swelling, which can increase the production rate. Unfortunately, to produce mostly polymer with some liquids, as would be needed commercially, the organo-Cr catalysts often do not have sufficient activity or MW-capability at the required loading and silica calcination temperatures. A way around this limitation was found when it was observed that the normal Cr oxide catalysts could be treated with small amounts of organoCr compounds in the reactor.64−66 For example, when Cr/silica is calcined at the usual commercial temperatures to produce the traditional Cr(VI) oxide catalyst, it is reduced by ethylene upon introduction into the reactor to a lower-valent active species. If simultaneously the catalyst is exposed to ppm levels of an organo-Cr compound in the reactor, it reacts with remaining isolated hydroxyl groups to afford type I sites that exist alongside doubly anchored Cr oxide species. This approach is illustrated in Scheme 4B, where the redox byproduct from ethylene is shown as methyl formate,67 although other similar species have also been detected.68 Whatever the mechanism of reduction, the resulting catalyst generates α-olefin in situ (especially 1-hexene) which is then consumed by the Cr oxide

when the silica support had been calcined at approximately 600 °C or higher, only isolated hydroxyls remain, giving rise to singly attached surface Cr species I depicted in Scheme 4A. However, when the silica was calcined at lower temperatures, at least some of the organo-Cr compound reacts with hydroxyl pairs, giving the doubly anchored Cr species II in Scheme 4A. This was established by OH measurements.63 Testing each kind of catalyst, he observed that when exposed to ethylene these two species yielded completely different products. Species II produced high-MW (solid) polymer, similar to Cr-oxide catalysts when made by calcination of Cr/ silica. This would perhaps be expected given that the latter (Croxide) route yields a similar doubly attached species II. However, catalysts containing only the singly attached species I produced liquid α-olefin oligomers, especially 1-hexene. The upper plot in Figure 7, utilizing the open-ring chromocene bis(2,4-dimethylpentadienyl)chromium, (DMPD)2Cr, shows the MW distribution generated by the two species. The extreme difference is remarkable. The amount of species I and II could be manipulated by means of the silica calcination temperature or the loading of the organo-Cr compound on the silica. The α-olefins generated by species I became incorporated as branches into the high-MW polymer generated by species II. Therefore, it is possible to produce lower density (i.e., highly branched) PE copolymers from ethylene addition alone. This offered certain commercial advantages: 1-Hexene is more costly than the equivalent weight in ethylene. Producing 1-hexene in situ in the reactor also avoids the expense of a second monomer supply. That is, it avoids the purchase, 607

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis type II-derived sites to produce the desired polymer grades, which is otherwise unaffected. Consequently, an advantage of this process, sometimes called the “two-valent” approach, is that the polymer-producing catalyst need not be changed, ensuring that the existing polymer character remains intact. Later the process was simplified further when it was realized that Cr(II)/silica, which is generated by CO reduction of Cr(VI)/silica, could be modified by metal alkyls to produce species I. Scheme 4C illustrates how this can be accomplished. Fortunately, the conversion of Cr(VI)/silica to Cr(II)/silica has only minor effects on the polymer character, and therefore, 1hexene can be produced in situ from such catalysts without a major change in the character of the polymer. Figure 8 shows

other candidates failing to attain the same selectivity and activity required. Supports other than silica were also explored with these organo-Cr catalysts. Surprisingly, the presence of most any other oxide, such as alumina or titania, did not yield the same result as shown in the upper plot of Figure 7. Instead, as shown in the lower parts of Figure 7, a higher-MW polymer is mostly produced, even at the highest calcination temperatures. This suggests that the other oxides could either serve as ligands, or that the organo-Cr compound could possibly insert itself into Si−O−Ti or Si−O−Al bonds to yield doubly anchored chromium (species II in Scheme 4C). Nevertheless, there is always production of at least some oligomers or 1-hexene, which indicates the coexistence of some of the monoattached species I. Even though such oxides were thus unsuited for the selectivity required in a trimerization plant, this made it possible to use Cr/silica−titania to produce high-MW copolymers from ethylene feed only, via in situ 1-hexene formation. Organo-Cr compounds deposited on acidic supports were nearly always significantly more active than when deposited on silica alone. In some cases, this could be attributed to enhanced hydrolysis of the organic ligand by the surface hydroxyl group. For example, one could see from the color change that bis(η6aryl)Cr(0) compounds and sterically encumbered tetrakis(trimethylsilylmethyl)Cr(4) were more reactive with silica− alumina than with silica.58 However, the higher activity of these and other compounds on acidic supports is also no doubt a more general phenomenon, attributable to reduced electron density on the chromium. The polymer MW is significantly affected by acidity on the support surface, as was noted above in Figure 4, where the more acidic phosphate sites produced lower MW polymer. The effect of fluoride, sulfate, and other treated metal oxides is often clearly visible in the MW distribution of the polymer, as well as in the activity of the catalyst. And he showed that poisons added to the reactor could sometimes be seen to interact selectively with some sites.13

Figure 8. Polymer density and branch content from in situ generated comonomer as a function of the triethylboron to chromium molar ratio in the reactor.



how the branch concentration and the density of the polymer (i.e., the crystallinity) can be easily manipulated in the absence of any added comonomer, simply by adjusting the amount metal alkyl in the reactor. In this example triethylborane was used to produce almost entirely 1-hexene as the comonomer, but other metal alkyls can also be added instead to vary the olefin composition. Silicon hydrides are especially potent. The process was quickly commercialized as a way to make LLDPE in a slurry process, which was formerly not possible due to polymer swelling in the presence of externally added 1-hexene. This in situ 1-hexene process has since accounted for about 4 billion lbs of LLDPE production, winning two of R&D Magazine’s “Best 100 New Products” award and an ACS “Hero of Chemistry” award for contributions to the environment by a new geomembrane polymer.69−72 Because of the broader MW distribution and consequent higher melt-strength (compared to Ziegler-based LLDPE), the polymer is easily blown into film at high MW, giving it unusual toughness. As part of these studies, McDaniel’s group also discovered that when treated with ethylene, chromium pyrrole compounds produced selectively 1-hexene as the only product.73 No highMW PE is obtained. This discovery was then developed into the first commercial process to selectively produce 1-hexene from ethylene. Subsequently, the search by many academic and industrial groups for other highly selective ethylene trimerization catalysts ensued. However, today the two large ethylene trimerization plants that are operating, one in Qatar and another in Houston, utilize the Cr-pyrrole-based catalysts, with

CALCINATION In addition to the chemical modifications above, McDaniel has extensively explored other methods of decreasing electron density on the chromium sites. These were specifically developed by manipulation of the calcination process. Cr/silica catalysts must be calcined (i.e., “activated”) to develop any polymerization activity. At 250−400 °C, Cr(III) is oxidized to Cr(VI), which then becomes attached to the silica by esterification with surface OH groups. Still higher temperatures, up to about 900 °C, increase the acidity of Cr active sites by removal of OH ligands, and probably by introducing strain into the Si−O−Cr−O-Si bonds.74−83 Consequently, with rising calcination temperature comes (1) higher polymerization activity, from an increase in the number and acidity of active Cr sites; (2) lower polymer MW, because decreased electron density favors agostic β-hydride coordination, which is necessary for chain termination; (3) narrowing of the MW distribution, as ligands are removed; (4) increased comonomer incorporation, because α-olefins are better electron donors than ethylene to the now more electron-deficient Cr; and (5) higher levels of long-chain branching, because the affinity of the Cr to macromer is also enhanced. Therefore, calcination not only “activates” Cr/silica catalyst but also controls all the polymer properties as well. These trends are summarized in Figure 9, where catalyst activity, polymer MW, MW breadth (MW/MN), LCB content, and melt index (a melt viscosity measurement 608

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis

reactivity toward ethylene. This was found to be true even when some nonincorporating secondary or cyclic olefins are added to the reactor. It was proposed that the coordination of electron-donating olefins to highly acidic metal centers facilitates the incorporation of coordinated ethylene. 19 Consequently, the more “naked” Cr(II)/silica, or Ziegler catalysts containing less coordinated “donor”, or the more tightly bridged metallocenes, all seem to display an enhanced “comonomer effect”. Another way the McDaniel group found of enhancing acidity on Cr/silica catalysts is through chemical dehydroxylation of the support,13,74,87 perhaps more accurately described as an extension of the trends shown in Figure 9. For instance, making use of the water−gas shift reaction, if the catalyst is first calcined at 800−900 °C in CO instead of air, this results in enhanced dehydroxylation of the silica surface but without the sintering that would normally accompany thermal treatment above 900 °C. This is illustrated in Scheme 5. As a final step in

Figure 9. Influence of calcination temperature on activity, polymer character.

Scheme 5. Dehydroxylation via the Water−Gas Shift Reaction and via Similar Reactions with Sulfur Compounds and Halides

that is inversely proportional to MW raised to the 3.4 power) are plotted as a function of the catalyst calcination temperature. Importantly, McDaniel found that deconvolution of the branch profiles and MW distributions of copolymers indicated that the type of sites involved do not change with calcination temperature. Instead there is a redistribution in the population of those site types.84 In contrast, the addition of titania, which also increases the acidity, does change the essential character of the Cr sites. Both ways of decreasing electron density on the Cr (temperature and TiO2) increased the frequency of long-chain branching in the polymer. Cr(VI)/silica is usually reduced by ethylene in the reactor, which leaves redox ligands still attached to the site.67,68 Alternatively, the Cr(VI) catalyst can be reduced by CO at 350 °C at the conclusion of the calcination process, yielding only CO2 as the oxidative byproduct that is immediately swept away, leaving a more “naked” Cr(II) species. As might be expected, this version of the catalyst is more active, and he found its branching profiles are also affected.13 The decreased ligation (enhanced Cr acidity) results in increased comonomer incorporation, especially in the low-MW portions of the MW distribution. McDaniel theorizes that because lower electron density on the Cr accelerates chain termination, the MW distribution tends to reflect the distribution of acidity on the catalyst. That is, short chains tend to come from the more acidic sites, whereas longer chains tend to come from less acidic, and less reactive, sites. An example of this was shown in Figure 3, where the addition of titania to a Cr/silica catalyst, which generates acidity, also produces a large new population of low-MW chains in the MW distribution.13,22 The Cr(VI) can also be partially reduced by exposure to CO at lower temperatures, followed by poisoning of the Cr(II) sites through CO retention. This provides another means of identifying site diversity.85 The most electron-deficient Cr(VI) sites that produce low-MW polymer were found to be the first to reduce, and the resultant MW distribution shows the loss of the low-MW component. The remaining sites, being less reactive with CO (less electron deficient), produce the longest chains. Acidity of the active site is probably also the root of the “comonomer effect” which has been observed on many different catalyst types, including Phillips, Ziegler, and metallocene catalysts.19,86 The addition of very small increments of α-olefin to the reactor causes a large boost in the catalyst’s

the procedure, the catalyst is treated with dry air, to convert the Cr into Cr(VI) and attach it to the surface. This procedure results in polymer having much lower MW, i.e. higher melt index (a long-standing goal in industrial research), narrower MW distribution (another industrial goal), and higher longchain branching. These trends are shown in Figure 10. He demonstrated that the effect is even more significant when the calcination is done in the presence of small amounts of sulfur containing species, such as COS or CS2, shown in

Figure 10. (Dark symbols) Silanol population remaining on the silica surface after calcination in three atmospheres. (Open symbols) Melt index of the polymer obtained varies inversely with the OH population. 609

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis

a function of time and temperature. Therefore, commercial calcination provides a unique set of challenges, namely maximizing production while minimizing exposure of the catalyst to moisture at high temperatures, which can decompose Cr(VI) species to inactive α-Cr2O3. McDaniel, using TGA and Cr(VI) measurements in the laboratory, carefully determined the time/temperature activation parameters. Cr(VI)/silica was found to be relatively insensitive to moisture at 600 °C or lower. However, at 800− 900 °C even traces of moisture have a devastating effect on catalyst quality. The lab data collected allowed McDaniel to produce a model of catalyst calcination. Accurate predictions could then be made for the catalyst quality resulting from any calciner scale or procedure, and from these predictions, more efficient recipes were soon developed to improve catalyst quality and calcination efficiency. This involved trading time for temperature, increasing gas flow during selected parts of the calcination, and using a “bent” temperature ramp instead of the traditional linear profile. This information became instrumental in designing and building new continuous commercial activators, which are much smaller than previous batch calciners, and result in better catalyst quality, lower capital costs, higher output, and better energy efficiency.92

Figure 10. Other methods of dehydroxylation, as shown in Scheme 5, yielded similar results, including the addition of halides to displace OH groups, followed by burning off the halide in air at 800−900 °C.88,89 Interestingly, the use of iodine for this purpose resulted in an unusual paramagnetic surface iodate, which was stable in air at 900 °C but easily reduced in the presence of organics even at 25 °C. About a decade ago, McDaniel showed that hexavalent Cr is extremely mobile during the calcination step on silica, alumina and other surfaces.90 Remarkably, in a fluidized bed, full equilibration between particles was found to occur within an hour through interparticle contact. In fact, he demonstrated that migration of Cr(VI) between particles could even be seen in a static bed. Cr(VI) migration was found to be caused by redox cycling between attached Cr(VI) and unattached Cr(III), sometimes facilitated by hydrolysis from atmospheric moisture. Removal of oxygen from the calcination atmosphere stopped migration, yielding mostly static Cr(III). Likewise, the addition of moisture to the calcination atmosphere tended to produce more Cr(III) due to hydrolysis of Cr(VI). The migration of Cr(VI) between silica particles was observed in the lab and also in the large commercial fluidized-bed calciners. It became the basis of a commercial way to make catalysts in situ by mixing silica with Cr/silica, or simply with Cr2O3 powder itself. During calcination, the Cr transfers onto the silica to form an active catalyst. It is a way to reduce catalyst cost, and storage and shipping of Cr/silica. The approach also permits prior calcination of the silica at a high temperature, and then addition of the Cr species at lower temperatures, to achieve certain polymer characteristics. Cr(III) acetylacetonate is particularly useful for this approach because it sublimes at only 100−150 °C onto the silica. If done in dry air, a Cr(VI)/silica catalyst is formed. Surprisingly, if the sublimation is done in N2, instead of the normal yellow/ orange catalyst, a green, very active Cr(III)/silica catalyst is produced that affords some very unusual polymer properties.91 In further studies of the calcination process, McDaniel found that the phenomena depicted in Figure 9 also have a timedependence, which indicates an annealing process driven by surface mobility at higher temperatures.91 Consequently, polymer properties were shown to be a function of the time the catalyst annealed during calcination. The annealing process is almost static at 200−300 °C, becomes significant at 500−600 °C, and is so fast at 800−900 °C that it appears complete (i.e., equilibrated) within a couple of hours. From this observation, it was concluded that time and temperature are interchangeable. That is, catalyst calcined for 20 h at 600 °C is equivalent in activity and polymer character to catalyst calcined at 750 °C for only 15 min. Trading time for temperature became an important way of increasing the output of commercial activators, which are sometimes the bottleneck of the entire plant. Another important observation to come from these studies was the sensitivity of Cr(VI) to moisture, which McDaniel carefully mapped out for the first time. Moisture in the fluidization air was known to reduce conversion of Cr(III) to Cr(VI) during calcination. But he found that moisture impurities in the gas stream were insignificant compared to moisture generated in situ by the dehydroxylation process itself. This explains why large commercial calcinations typically exhibit less Cr(VI) content than the small, more efficient laboratory activators. Water generated in situ by dehydroxylation causes the moisture concentration within the bed to be



PHYSICAL STRUCTURE OF THE CATALYST Much of McDaniel’s work has concentrated on the fragmentation process of Cr/silica catalysts during polymerization, and his papers are some of the earliest on the subject.93−100 He is unique among other researchers, especially the particle-growth modeling community, in espousing the following views: (1) Polymerization-grade silicas fragment until all of the BET surface area contributes to the observed activity; (2) Some of this surface is still inside pores within fragments; and (3) Consequently, the structure of silica plays an important role in determining the character of the polymer produced. Papers on the fragmentation of Cr/silica have been sparse in the literature, and these ideas have not yet been widely recognized. The first conclusion (no. 1 above) comes from a comparison of the activities of various Cr/silicas per square meter of BET surface. Nonporous silicas, which do not fragment, yielded equal or similar activities to Cr on higher porosity silica gels (so-called “polymerization grade” silicas). This can only mean that all of the surface does indeed participate in the polymerization. However, this is by no means true of all silica gels. Many, usually of lower porosity, do yield considerably lower activity per square meter of BET surface. At lower porosity, the silicas are stronger and resist fragmentation more effectively. In fact, many Cr/silica samples are completely dead because they do not fragment. However, “polymerization grade” silicas do fragment until all the surface contributes. The second and third conclusions (nos. 2 and 3 above) come from years of commercial experience in which chemically identical catalysts, which have different physical structures, can produce vastly different kinds of polyethylene. Molecular weight varies widely with the silica structure, as well as longchain branch content, and even comonomer incorporation. After making and studying many different silica structures, he connected the polymer traits with the overall structure of the silica, not just the porosity. For example, silicas subjected to Ostwald ripening during their preparation, or thermal sintering, tend to produce high amounts of LCB compared to other silicas having identical surface area and pore volume, but not 610

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis Table 1. Variation in Pore Volume Caused by Different Methods of Drying Silica Gel pore liquida

pore volume

surface area

pore diameter

catalyst activity

molecular weight, MW

melt index

(azeotroped, then dried)

(mL/g)

(m2/g)

(Å)

(kgPE/gCat-h)

(kg/mol)

(dg/min)

water toluene toluene +5% i-pentanol toluene +10% i-pentanol butyl acetate BuOAc + 10% i-pentanol BuOAc + 25% i-pentanol ethyl acetate i-Pentanol

0.7 1.0 1.3 1.7 1.8 2.0 2.2 2.6 3.3

407 409 408 411 409 413 417 426 408

56 80 104 136 144 160 176 208 264

2.6 3.5 3.9 5.1 6.0 6.1 6.9 7.4 7.5

231 212 185 133 116 102 90 82 70

0.12 0.32 0.64 1.40 1.72 2.02 2.31 2.53 2.88

a

The pore water in chromia−silica−titania hydrogel was removed by azeotropic distillation while refluxing in various organic solvents. Initial water content was 7.5 mL/g.

subjected to the same coalescence process. Such aging or sintering treatments fuse the primary silica particles together, making the structure more resistant to fragmentation, but not enough to lower the activity. Therefore, a higher portion of the polymerization probably occurs within pores of fragments as opposed to being made on the exterior of fragments. Polymer made inside pores seems to produce more long-chain branching and higher MW. In another example, samples of a common silica−titania hydrogel having a water content of 7.5 mL/g, were dried by different methods, giving different degrees of shrinkage. On one extreme, the hydrogel was simply placed in an oven which subjects the pore walls to the high surface tension of water. This yields a pore volume of only 0.5−0.7 mL/g. Alternatively, the hydrogel pore water can be removed by azeotropic distillation with a 5- or 6-carbon alcohol. This subjects the structure to a very low oil/water interfacial tension, followed by an oil/air interface of only slightly higher surface tension. The result is a catalyst having a pore volume of about 3.3−3.5 mL/g. In between these extremes the hydrogel can be dried by washing or azeotropic distillation with other organic liquids, as listed in Table 1. The surface area remained constant in all these samples, indicating no significant Oswald ripening. In this series, catalyst activity is found to vary with pore volume, which is understandable since the strength of the silica matrix also varies with pore volume. However, what is less intuitive is the fact that the MW of the polymer varied over the entire series, from high MW at low PV to low MW at high PV. This is consistent with his hypothesis that some of the polymer is made inside pores, at least down to some shallow depth within the fragments. Had all the polymer been made on the exterior of the fragments, as is usually proposed by the modeling community, one would expect the polymer MW to have been constant. This response is summarized in Table 1. As control experiments, McDaniel mechanically compressed the high PV sample from Table 1 under different pressures up to 70 000 psig. The pore volume dropped with the force applied, back down to 0.5 mL/g. These samples, when allowed to polymerize ethylene, produced an almost identical relationship to that in Table 1, indicating that it makes no difference whether the force used to shrink the matrix is applied from without or from within. Many other structures were also synthesized or otherwise obtained and investigated, including several colloidal silicas, in which primary silica particles of different sizes are kept in aqueous suspension by repulsion of their negatively charged surfaces at certain pH values. When the pH is adjusted, these

particles coalesce into a gel, which can then be dried in different ways. McDaniel concluded that the size of the primary particles, which varied in surface area from 30 up to 300 m2/g, made no difference to the polymer properties. However, the way these silica gels were dried had a large effect, because that determined the “coordination number” of each particle (i.e., the number of neighboring particles to which each one is connected), which in turn determines the strength of the matrix, and its resistance to fragmentation. He infers that a stronger matrix probably yields larger fragments, and therefore a larger percentage of the polymer is made inside pores. For further comparison, he conducted experiments with pyrogenic silicas of varying surface areas, which are nonporous. And in still other studies, gelled silicas were contrasted with precipitated silicas. The former type is usually made by the addition of sodium silicate to excess acid so that pH neutrality is approached from the acid side.101 The result is a gelled network of uniform primary particle size, giving a narrow pore size distribution. In contrast, precipitated silicas are usually approached from the basic side. Often the primary particles agglomerate into tight, larger aggregates which then coalesce into a larger framework. The result is a bimodal pore size distribution of very small and very large diameters. Such precipitated silicas produced some extraordinary behavior when converted into Cr/silica catalysts. The activity can be quite respectable, indicating that a significant fraction, or in some cases all, of the surface is active, even that inside small pores. In fact, most of the surface is inside the small pores. This affords extreme levels of LCB, and relatively high MW polymer, both of which suggest that the small pore activity dominates the polymer character. In contrast, a gelled silica with such small pores is not active, i.e. it is too strong to allow fragmentation. The high activity of some precipitated silicas suggests that significant fragmentation still occurs, perhaps along the larger pores, to provide access to a large portion of the surface in the small pores. This is consistent with the aggregates being small enough, and the fragmentation complete enough, to permit significant egress of the polymer. Therefore, according to his view, some (usually most) of the polymer is made on the exterior of the fragments and has a relatively low MW and LCB content; however, some polymer is also made within (possibly shallow) pores inside of fragments, the amount varying with the size of the fragments made and also with a pore diameter that allows polymer egressthe polymer made inside pores tending to have higher MW and LCB content. 611

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis

Ziegler, Ballard, and metallocene catalysts. Indeed, he has found that comparison between these catalyst types is often a rich source of insight into mechanism and polymer behavior, and proposes no reaction for one without also considering the others. Acidity has been a common underlying theme, governing not only the polymerization activity but many polymer properties and reaction behavior. In addition, he has made, and continues to make, seminal contributions to our understanding of how the extended structure of the catalyst (e.g., the silica itself) can also be a strong factor influencing catalyst reactivity and polymer character.

In fact, polymers usually have a bimodal MW distribution reflecting this dual source of polymer formation. He illustrates an example in Figure 11. In this series, a silica was given various



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Michael D. Jensen: 0000-0001-6957-8642 Notes

The authors declare no competing financial interest.



Figure 11. Change in MW distribution as the Cr/silica structure coalesces. The bimodal shape suggests that two different site types exist, such as those on the exterior of fragments vs those still in pores of the fragments.

REFERENCES

(1) Hogan, J. P.; Banks, R. L. U.S. Patent 2,825,721 to Phillips Petroleum Company, filed Aug.1954, Issued Mar. 1958. (2) Hogan, J. P.; Banks, R. L. Belgian Patent 530,617, filed Jan, 1955; filed in the U.S. Jan 1953 S.N. 333,576; also issued as Austrian Pat App. 864/54 filed Aug 6, 1954. (3) Sailors, H. R.; Hogan, J. P. A History of Polyolefins. Polymer News 1981, 7, 152−167. (4) Ziegler, K.; Breil, H.; Martin, H.; Holzkamp, E. German Patent 973,626, Filed Nov 17, 1953, Issued April 14, 1960. See also U.S. Patent 3, 257,332, Filed Nov 15, 1954, Issued June 21, 1966. (5) Zletz, A. U.S. Patent 2,692,257 (filed April 28, 1951, issued Oct. 19, 1954) to Standard Oil (later Amoco) describes polymerizing ethylene using molybdenum oxide on alumina as a catalyst. However, management, perhaps influenced by consultants, did not understand the potential importance of the polymer. For historical detail see Hutley, T. J.; Ouderni, M. In Polyolefin Compounds and Materials: Fundamentals and Industrial Applications; AlMa’adeed, M. A., Krupa, I., Eds.; Springer: Berlin, 2015; Chapter 2, pp 13−50. (6) McDaniel, M. P.; Burwell, R. L. J. Catal. 1975, 36, 394−403. (7) McDaniel, M. P.; Burwell, R. L. J. Catal. 1975, 36, 404−412. (8) McDaniel, M. P.; Rohlfing, D. C.; Benham, E. A. Polym. React. Eng. 2003, 11, 101−132. (9) Schwerdtfeger, E. D.; Jensen, M. D.; Yang, Q.; McDaniel, M. P. Current Topics in Catalysis 2016, 12, 1−27. (10) Yu, Y.; Schwerdtfeger, E.; McDaniel, M. P. J. Polym. Sci., Part A: Polym. Chem. 2012, 50, 1166−1173. (11) DesLauriers, P. J.; McDaniel, M. P.; Rohlfing, D. C.; Krishnaswamy, R. K.; Secora, S. J.; Benham, E. A.; Maeger, P. L.; Wolfe, A. R.; Sukhadia, A. M.; Beaulieu, W. B. Polym. Eng. Sci. 2005, 45 (9), 1203−1213. (12) DesLauriers, P. J.; Tso, C.; Yu, Y.; Rohlfing, D. L.; McDaniel, M. P. Appl. Catal., A 2010, 388, 102−112. (13) McDaniel, M. P. Advances in Catalysis; Gates, B. C., Knoezinger, H., Jentoft, F. C., Eds.; Academic Press, Elsevier: Amsterdam, 2010; Vol 53, Chapter 3, pp 123−606. (14) Hogan, J. P.; Levett, C. T.; Werkman, R. T. SPE Journal 1967, 23, 87−90. (15) Janzen, J.; Colby, R. H. J. Mol. Struct. 1999, 485−486, 569. (16) Jung, M.; Lee, Y.; Kwak, S.; Park, H.; Kim, B.; Kim, S.; Lee, K. H.; Cho, H. S.; Hwang, K. Y. Anal. Chem. 2016, 88, 1516−1520. (17) Liu, P.; Liu, W.; Wang, W.; Li, B.; Zhu, S. Macromol. React. Eng. 2017, 11, 1600012. (18) Alt, H. G. J. Chem. Soc., Dalton Trans. 1999, 1703−1709. (19) Yang, Q.; Jensen, M. D.; McDaniel, M. P. Macromolecules 2010, 43, 8836−8852.

degrees of coalescence through an alkaline aging process. Notice that the untreated silica produces a single, lower-MW peak, but as the fusion between subparticles becomes ever more severe, a shoulder develops on the high-MW side of the MW distribution that finally turns into a second peak of similar size. He postulates that the two peaks come from polymerization in or out of pores. And some polymerization grade silicas resist fragmentation more than others, depending on their structure, which can result in relatively large or small fragments, with narrow or wide pores. Therefore, this determines the amount of each type of polymer that is produced. Indeed, this is the mechanism by which silica structure can have such a large influence on PE processing and manufactured-part properties. Exactly why polymer made inside pores should be different is not entirely clear. However, he notes that many of the polymer chains are thousands of times longer than the width of the pores they are made in. And he envisions that they are “extruded” out of the pores under considerable stress and pressure. Such conditions may restrict conformational freedom of the growing chain and thus inhibit agostic β-H coordination necessary for chain termination, thus raising the nascent MW. Crowding might also make it more difficult to dissociate terminated macromer, resulting in higher LCB. Alternatively, perhaps ethylene concentration is lower in crowded pores, which is also known to increase LCB. In these and many other experiments, he finds the polymer MW seems to always correlate with catalyst pore diameter, while LCB content seems to be determined by the matrix strength, thus, answering his own long-standing question: “How does the Cr know what size pore (or structure) it is in?” Despite our current lack of full understanding of these phenomena, these relationships are still used daily to manufacture billions of pounds of polymers for a diverse assortment of commercial applications.



CONCLUSIONS For four decades, in order to develop insight into their nature, McDaniel probed Phillips catalysts, many of which he invented, and studied the polymers that they produce. Many of the lessons learned, he extended to other systems as well, such as to 612

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis (20) Fahey, D. R.; Lauffer, D. E.; Dokter, D. W.; Das, P. K.; Boudreaux, E.; Whitte, W. M. MetCon International Conference, Houston, Texas, June 9−10, 1999. (21) Alt, H. G. Dalton Transactions 2005, 20, 3271−3276. (22) McDaniel, M. P.; Welch, M. B.; Dreiling, M. J. J. Catal. 1983, 82, 118−126. (23) Bouh, A. O.; Rice, G. L.; Scott, S. L. J. Am. Chem. Soc. 1999, 121, 7201−7210. (24) Liu, B.; Terano, M. Advances in Polyolefins, Sonoma Valley, California, September 25−28, 2005. (25) Hogan, J. P.; Witt, D. R. U.S. Patent 3,622,521, assigned to Phillips Petroleum Company. Filed August 1967, issued Nov 1971. (26) Pullukat, T. J.; Hoff, R. E.; Shida, M. J. Polym. Sci., Polym. Chem. Ed. 1980, 18, 2857−2866. (27) Deitz, R. E. U.S. Patent 3,887,494, issued June 3, 1975, assigned to Phillips Petroleum Co. (28) Conway, S. J.; Falconer, J. W.; Rochester, C. H.; Downs, G. W. J. Chem. Soc., Faraday Trans. 1 1989, 85, 1841−1851. (29) McDaniel, M. P. U.S. Patent 4,402,864, September 6, 1983. (30) McDaniel, M. P. U.S. Patent 4,405,768, September 20, 1983. (31) McDaniel, M. P. U.S. Patent 4,294,724, October 13, 1981. (32) McDaniel, M. P. U.S. Patent 4,368,303, January 11, 1983. (33) McDaniel, M.P. U.S. Patent 4,382,022, May 3, 1983. (34) McDaniel, M. P. U.S. Patent 4,424,320, January 3, 1984. (35) Van Loon, J. D. Maack PE 2004 Conference on Polyolefins, Zurich, Switzerland, February, 2004. (36) Wharry, S. M.; Martin, S. J.; McDaniel, M. P. J. Catal. 1989, 115, 463−472. (37) McDaniel, M. P.; Johnson, M. M. U.S. Patent 4,444,962, April 24, 1984. (38) McDaniel, M. P.; Johnson, M. M. U.S. Patent4,364,854, December 21, 1982. (39) McDaniel, M. P.; Johnson, M. M. J. Catal. 1986, 101, 446−457. (40) McDaniel, M. P.; Johnson, M. M. Macromolecules 1987, 20, 773−778. (41) Cheung, T. T. P.; Willcox, K. W.; McDaniel, M. P.; Johnson, M. M. J. Catal. 1986, 102, 10−20. (42) McDaniel, M. P. Ind. Eng. Chem. Res. 1988, 27, 1559−1564. (43) McDaniel, M. P.; Collins, K. S. U.S. Patent 7,638,456, issued Dec. 29, 2009 to Chevron Phillips Chemical Co. (44) McDaniel, M. P.; Collins, K. S.; Benham, E. A.; DesLauriers, P. J. U.S. Patent 7,214,642, issued May 8, 2007 to Chevron Phillips Chemical Co. (45) Deslauriers, P. J.; McDaniel, M. P.; Rohlfing, D. C.; Krishnaswamy, R. K.; Secora, S. J.; Benham, E. A.; Maeger, P. L.; Wolfe, A. R.; Sukhadia, A. M.; Beaulieu, W. B. Polym. Eng. Sci. 2005, 45, 1203−1213. (46) DesLauriers, P.J.; McDaniel, M.P.; Rohlfing, D.C.; Krishnaswamy, R.J.; Secora, S.J.; Maeger, P.L.; Benham, E.A.; Wolfe, A. R.; Sukhadia, A.M.; Beaulieu, W.M. Novel Phillips loop-slurry based polyethylene resins using modified chromium oxide catalysts for high performance pipe applications. Proceedings of the International Conference on Polyolefins, Houston, TX, Feb 22−25, 2004; Society of Plastics Engineers, 2004; pp 343−357. (47) McDaniel, M. P.; Benham, E. A.; Wolf, A. L. U.S. Patent 6,495,638, December 17, 2002. (48) McDaniel, M. P.; Smith, P. D.; Norwood, D. D. U.S. Patent 4,806,513, February 21, 1989. (49) McDaniel, M. P.; Smith, P. D.; Norwood, D. D. U.S. Patent 5,037,911, August 6, 1991. (50) McDaniel, M. P.; Smith, P. D.; Norwood, D. D. U.S. Patent 5,401,820, March 28, 1995. (51) McDaniel, M. P.; Jensen, M. D.; Jayaratne, K.; Collins, K. S.; Benham, E. A.; McDaniel, N. D.; Das, P. K.; Martin, J. L.; Yang, Q.; Thorn, M. G.; Masino, A. P. In Tailor-Made Polymers: Via Immobilization of Alpha-Olefin Polymerization Catalysts; Severn, J. R., Chadwick, J. C., Eds.; Wiley-VCH: Weinheim, 2008, XVI, Chapter 7, pp 171−210.

(52) McDaniel, M. P.; Collins, K. S.; Eaton, A. P.; Benham, E. A.; Jensen, M. D.; Martin, J. L.; Hawley, G. R. U.S. Patent 6,613,852, September 2, 2003. (53) McDaniel, M. P.; Benham, E. A.; Martin, S. J.; Collins, K. S.; Smith, J. L.; Hawley, G. R.; Wittner, C. E.; Jensen, M. D. U.S. Patent 6,300,271, October 9, 2001. (54) Ballard, D. G. H.; Jones, E.; Medinger, T.; Pioli, A. J. P. Makromol. Chem. 1971, 148, 175−194. (55) McDaniel, M. P.; Johnson, M. M. U.S. Patent 4,364,841, December 21, 1982. (56) McDaniel, M. P.; Johnson, M. M. U.S. Patent 4,444,968, April 24 1984. (57) Smith, P. D.; McDaniel, M. P. J. Polym. Sci., Part A: Polym. Chem. 1989, 27, 2695−2710. (58) Smith, P. D.; McDaniel, M. P. J. Polym. Sci., Part A: Polym. Chem. 1990, 28, 3587−3601. (59) Walker, D. L.; Czenkusch, E. L. U.S. Patent 3,157,712, November 1967. (60) Ajjou, J. A. N.; Scott, S. L. Organometallics 1997, 16, 86−92. (61) Ajjou, J. A. N.; Scott, S. L.; Paquet, V. J. Am. Chem. Soc. 1998, 120, 415−416. (62) Karol, F. J.; Karapinka, G. L.; Wu, C.; Dow, A. W.; Johnson, R. N.; Carrick, W. L. J. Polym. Sci., Part A-1: Polym. Chem. 1972, 10, 2621−2637. (63) McDaniel, M. P.; Leigh, C. H.; Wharry, S. M. J. Catal. 1989, 120, 170−181. (64) Smith, P. D. U.S. Patent 4,668,808, May 26, 1987. (65) Smith, P. D.; Hsieh, E. T. U.S. Patent 4,665,263, May 12, 1987. (66) Smith, P. D.; Hsieh, E. T., U.S. Patent 4,587,227 issued May 6 1986, to Phillips Petroleum Co.. (67) Barzan, C.; Piovano, A.; Braglia, L.; Martino, G. A.; Lamberti, C.; Bordiga, S.; Groppo, E. Conference on Phillips Catalysis, Lyon, June, 2016; published August 21, 2017 on-line as J. Am. Chem. Soc. Article ASAP, DOI: 10.1021/jacs.7b07437. (68) Potter, K. C.; Beckerle, C. W.; Jentoft, F. C.; Schwerdtfeger, E. D.; McDaniel, M. P. J. Catal. 2016, 344, 657−668. (69) Chemical Week, May 5, 1993; pp 12. (70) R&D Magazine, September 1996, 38, 46. (71) McDaniel, M. P.; Benham, E. A. U.S.. Patent 5,274,056, December 28, 1993. (72) Benham, E. A.; Smith, P. D.; McDaniel, M. P. Polym. Eng. Sci. 1988, 28, 1469−1472. (73) Reagen, W. K.; McDaniel, M. P. U.S. Patent 5,382,738, issued January 17, 1995 to Phillips Petroleum Co. (74) McDaniel, M. P. Appl. Catal., A 2017, 542, 392−410. (75) McDaniel, M. P.; Welch, M. B. J. Catal. 1983, 82, 98−109. (76) McDaniel, M. P. J. Catal. 1982, 76, 29−36. (77) McDaniel, M. P. J. Catal. 1982, 76, 37−47. (78) McDaniel, M. P. in Transition Metal Catalyzed Polymerizations, Alkenes and Dienes; MMI Press, Ed. Quirk, Roderic P. 1983, Vol.4, Part B, p713−735. (79) Demmelmaier, C. A.; White, R. E.; Van Bokhoven, J. A.; Scott, S. L. J. Phys. Chem. C 2008, 112, 6439−6449. (80) Demmelmaier, C. A.; White, R. E.; Van Bokhoven, J. A.; Scott, S. L. J. Catal. 2009, 262, 44−56. (81) Liu, B.; Fang, Y.; Terano, M. J. Mol. Catal. A: Chem. 2004, 219, 165−173. (82) McDaniel, M. P.; Collins, K. S.; Benham, E. A.; Cymbaluk, T. H. Appl. Catal., A 2008, 335, 252−261. (83) McDaniel, M. P.; Collins, K. S.; Benham, E. A.; Cymbaluk, T. H. Appl. Catal., A 2008, 335, 180−186. (84) DesLauriers, P. J.; McDaniel, M. P. J. Polym. Sci., Part A: Polym. Chem. 2007, 45, 3135−3149. (85) McDaniel, M. P.; Martin, S. J. J. Phys. Chem. 1991, 95, 3289− 3293. (86) Yu, Y.; Schwerdtfeger, E. D.; McDaniel, M. P. J. Polym. Sci., Part A: Polym. Chem. 2012, 50, 1166−1173. (87) Welch, M. B.; McDaniel, M. P. J. Catal. 1983, 82, 110−117. (88) McDaniel, M. P. J. Phys. Chem. 1981, 85, 532−537. 613

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614

Account

ACS Catalysis (89) McDaniel, M. P. J. Phys. Chem. 1981, 85, 537−541. (90) McDaniel, M. P.; Collins, K. S.; Benham, E. A. J. Catal. 2007, 252, 281−295. (91) McDaniel, M. P.; Clear, K. S. Appl. Catal., A 2016, 527, 116− 126. (92) Benham, E. A.; McDaniel, M. P.; Cymbaluk, T. H.; Newsome, C. K.; Nease, C. R.; Staffin, H. K.; Parr, T. R., U.S. Patent 8,349,264, January 8, 2013. (93) McDaniel, M. P. J. Polym. Sci., Polym. Chem. Ed. 1981, 19, 1967−1976. (94) Webb, S. W.; Conner, W. C.; Laurence, R. L. Macromolecules 1989, 22, 2885−2894. (95) Whitaker, H. L.; Wills, G. B. J. Appl. Polym. Sci. 1969, 13, 1921− 1927. (96) McDaniel, M. P. ACS Catal. 2011, 1, 1394−1407. (97) McDaniel, M. P.; Collins, K. S. J. Catal. 2009, 261, 34−49. (98) McDaniel, M. P.; Collins, K. S. J. Polym. Sci., Part A: Polym. Chem. 2009, 47, 845−865. (99) Conner, W. C.; Weist, E. L.; Ali, A. H.; Chiovetta, M.; Laurence, R. L. Transition Metal Catalyzed Polymerization, Proceeding of the International Symposium, 2nd, Akron, OH 1986, Quirk, R.P.,Ed.; Cambridge Univ. Press: Cambridge, 1988 pp 417−427. (100) Niegisch, W. D.; Crisafulli, S. T.; Nagel, T. S.; Wagner, B. E. Macromolecules 1992, 25, 3910−3916. (101) Iler, R. K. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties, and Biochemistry; John Wiley: New York, 1979.

614

DOI: 10.1021/acscatal.7b03393 ACS Catal. 2018, 8, 602−614