The design and development of new catalysts and new ways to use catalysts are a continuing effort in many laboratories. It is an effort that has led to major advances in the synthesis of new polymers, in the more efficient synthesis of new fine chemicals, and in the development of new ways to convert a variety of feedstocks into the materials that form the basis of a 21st century economy. The catalysts used for these purposes are often subdivided into the categories of heterogeneous and homogeneous catalysts. Heterogeneous catalysis typically involves some sort of surface reaction, most often in a gas/solid or liquid/solid mixed-phase system. Typical catalysts are insoluble metal or metal oxide particles—materials that are usually easily separated from products. In contrast, homogeneous transition-metal catalysts are soluble and more difficult to separate from products at the end of a reaction. Homogeneous catalysts have advantages, however, because they offer chemists exquisite control of catalyst reactivity and selectivity based on the structure and organization of ligands around the transition-metal catalyst. In the ensuing review, I have summarized some of our work that uses polymer supports to deal with the problem of homogeneous catalyst separation and recovery. This review also includes limited examples of other work with soluble polymers that either was a precedent for our work or that has been contemporaneous with our studies. This informal review does not attempt to discuss homogeneous catalysts in detail but instead briefly describes the advantages and problems of using homogeneous catalysts in the context of the problems subsequently discussed. The objective of this review is to describe the history and ideas underlying our successful design of a number of different polymer supports that facilitate the use, separation, and reuse of homogeneous transition-metal catalysts.
Polymer chemistry has extensively influenced and has been broadly influenced by developments in both heterogeneous and homogeneous catalysts. For example, polymers like polyolefins are prepared using either heterogeneous or homogeneous catalysts. Indeed, the discovery of the heterogeneous Ziegler–Natta catalysts was the Nobel prize-winning discovery that ushered in the era of polyolefins. More recently, there has been a great deal of interest in new homogeneous catalysts for polymer synthesis. For example, new homogeneous metallocene catalysts control polyolefin microstructure and tacticity;1 metathesis catalysts lead to new sorts of polymers;2 catalysts for atom transfer radical polymerization enable ‘living’ radical polymerization;3 catalysts for biaryl couplings afford specialty conjugated polymers of interest in electronic and optical applications;4 and new catalysts are being successfully developed for the synthesis of biodegradable polymers.5 The impact of such catalysts on the area of polymer chemistry can be judged from the impact of just one example of such catalysis, atom transfer radical polymerization, which has rapidly developed in only a few years since its original introduction by three different groups.6 It is thus not surprising that many sorts of homogeneous catalysts are now being studied and discovered for polymer synthesis. More conventional organic synthesis too often uses homogeneous catalysts, and there is an even greater diversity of homogeneous catalysts being discovered and developed for fine chemical synthesis.7, 8 Such catalysts provide synthetic chemists both with more efficient ways to form simple bonds and, in an increasing number of cases, provide chemists with the same sort of exquisite control of stereochemistry that nature achieves with enzymatic catalysis.8 It is not surprising that homogeneous catalysts are becoming increasingly more important in all sorts of syntheses.
However, although it is evident that what homogeneous catalysts do is important and valuable, there are problems in using homogeneous catalysts. One specific issue is the separation of product from catalysts and ligands.9–11 This is a continuing problem in homogeneous catalysis. Such separations are often necessary based on catalyst or ligand cost. In some cases, for example, in polyolefin synthesis, catalyst efficiency is extraordinarily high, and the product polyolefin utility is essentially unaffected by trace catalyst residues. In those cases, separation and catalyst recovery is neither economic nor necessary. However, product purity of fine chemical or pharmaceutical products is very important. In these cases, the potential toxicity of even trace amounts of ligand or of a heavy metal serves as a justification for catalyst separation even when very efficient catalysts are used.
Our earliest work using polymers to recycle and reuse catalysts began in the early 1980s—at the end of a period of intense research both in academia and in industry on so-called “heterogeneized” catalysts. That work mainly used crosslinked polymers like divinylbenzene crosslinked polystyrene as a support for homogeneous catalysts. That work had begun in several laboratories in the 1960s and was work aimed at making heterogeneous (and hence recoverable) analogues of existing homogeneous catalysts.12
All of this catalysis work was precedented by Merrifield's13 pioneering work on solid-phase peptide synthesis. It was contemporaneous with other work that led to stoichiometric polymer-bound reagents useful in organic synthesis. As a result, there is now a very large body of work on insoluble polymer supports describing their use in catalysis and synthesis. This work has also become much more important recently. In the past 5–10 years there has been an upsurge of interest in polymer-bound reagents and supports resulting from developments in combinatorial chemistry and high throughput synthesis.14, 15 These developments have led to the successful launch of several companies providing such supports. Because the bulk of our work has involved the use of soluble polymers and because there are a number of comprehensive reviews on solid-phase synthesis and solid-phase polymer supports, the use of crosslinked polymers is not further discussed.
Although insoluble polymer supports are the dominant sort of support used in synthesis and catalysis, our own work has focused on soluble polymers. This was a result of our other interests in organometallic chemistry and polymer chemistry. However, it also reflected the fact that we were aware of some of the deficiencies of these “heterogeneized” catalysts. Specifically, catalysts on insoluble polymers were difficult to characterize fully; they had different (often lower) activity than their homogeneous analogues and often altered selectivity, and neither they nor their homogeneous counterparts were always as durable as a heterogeneous catalyst in relation to heat and adventitious impurities (e.g., oxygen) in a reaction mixture.
LIQUID-PHASE PEPTIDE SYNTHESIS
Our interest in developing soluble polymers as alternatives to insoluble polymers as supports for catalysts has precedent in earlier developments of peptide synthesis. Soluble polymers were recognized as potential alternatives to the insoluble Merrifield-type supports early on by Bayer's group at Tübingen.16, 17 What Bayer's group in Tübingen developed was an alternative to Merrifield's solid-phase synthetic approach to peptides called liquid-phase peptide synthesis. This work has been reviewed and was a harbinger of our work.17 The most common polymer used in this work was poly(ethylene oxide) (PEO), and the general idea was to use the terminal group of PEO as a site to attach a reagent, catalyst, or substrate. PEO would then provide solubility control for this terminal group. In appropriate solvents, PEO would be soluble, and it would be possible to do homogeneous chemistry and to readily characterize the end group by spectroscopy. The solubility of the PEO support ensures that the end group reacts like their low molecular weight analogues, avoiding many of the reactivity problems seen with crosslinked supports. When the synthesis is complete, the PEO support and the polymer-bound synthesis product can be isolated by solvent precipitation. PEO is insoluble in solvents like isopropanol or ether, and the polymer and its attached group could thus be isolated and ‘purified’ by the simple expedient of adding a solution of the PEO in a ‘good’ solvent to an excess of a poor solvent. The PEO support is then recovered by filtration (Fig. 1).
This polymer-based separation scheme can be put to practical use in the case of a peptide synthesis as is shown in the simplified example in Scheme 1. In this case, the chemistry is very much like the chemistry used on insoluble resins. The principle difference is that the reactions and the characterization of products can be carried out under homogeneous conditions. Moreover, the carbonyl substitution reactions on PEO have kinetics that are comparable to those of their low molecular weight analogues.18 Problems with solubility can develop as the end group's size increases if the end group becomes a significant part of the support. The relative size of the polymer support and attached group is also of interest in connection with loadings of functionality on these PEO supports. A hydroxyl-terminated PEO polymer has at most two OH groups/molecule, and if the polymer has a high molecular weight, the loadings are at best modest. Thus, PEO oligomers described as poly(ethylene glycol)s are often used in place of a high molecular weight polymer. Regardless, the scheme involved in this approach to peptide synthesis and the separation concept illustrated in Scheme 1 and Figure 1 has wide generality.
ETHYLENE OLIGOMER SUPPORTS
Our efforts to use soluble polymers as supports in synthesis began as a result of other work in our group in the early 1980s. At that time, we had the advantage of knowing about the liquid-phase peptide synthesis approach of Bayer and of others' work with soluble polymers.17, 19 At that time, my group had begun a new and still continuing effort in studying polymer-surface-modification chemistry.20 That work began slowly given our modest experience and knowledge of surface chemistry and of polymers generally at the beginning of these studies. Although I had a background of success in physical organic and organometallic chemistry, our proposals for funding of this polymer chemistry were not always well received by the polymer community, usually for good reason. Nonetheless, my group persisted in its efforts, and in the course of some of this early work to obtain preliminary results to support our hypotheses for studies in polymer-surface chemistry, we prepared spin-labeled polyethylene oligomers.21 These oligomers were prepared using first anionic oligomerization of butadiene22 and later anionic oligomerization of ethylene.23 We found these highly linear ethylene oligomers prepared from ethylene were completely insoluble in any solvent at room temperature. For example, a student working on the surface-modification project prepared a spin-labeled oligomer like 1. He found 1 was insoluble at room temperature in toluene, but that 1 visually dissolved when the toluene was heated to reflux. When the hot homogeneous solution of spin-labeled oligomer 1 was cooled to room temperature, the polyethylene oligomer reprecipitated and was quantitatively recovered based on gravimetric analysis after a simple filtration. Moreover, when the toluene filtrate was analyzed by electron spin resonance (ESR) spectroscopy, no detectable spin label was seen. Because ESR spectroscopy was capable of detecting even trace quantities of spin label, this experiment suggested that precipitation of these linear oligomers was quantitative. This high-temperature solubility/low-temperature insolubility of these oligomers immediately suggested a role for these polyethylene oligomers as supports for recoverable catalysts.
Our first idea was to use a polymer's temperature-dependent solubility to prepare a polymer-bound catalyst that is soluble under the reaction conditions and insoluble for recovery and workup. This concept is illustrated in Figure 2. The ideas for dissolution/precipitation of a polymeric catalyst described in this figure are really just an extension of the experiments with spin-labeled polymer 1. They are generally applicable to any sort of catalyst attached to the terminus of a linear polyethylene oligomer, and this concept has been used by several groups with a variety of catalysts. Table I includes a list of catalysts that others and we have appended to polyethylene oligomers. Because the solubility of the catalyst in each case is determined by the ethylene oligomer, each catalyst is essentially used in the same way—the polymer-bound catalyst is active and homogeneous when in hot solvent but inactive, heterogeneous, and recoverable in room temperature solvent.
Table I. Catalysts or Reagents Successfully Attached to Polyethylene Oligomers
Several aspects of polymer chemistry affect the utility of these polyethylene-bound catalysts. The most important feature is that the catalyst solubility is dictated by the polyethylene oligomer. Because the polymer makes up the bulk of the ligand–catalyst complex and the actual ligating group and catalyst are only a small part of the final molecule, we get predictable polyethylene-like solubility at high temperature in solvents like toluene, chlorinated aromatics, or dibutyl ether. However, although this solubility is useful, polyethylene-like solubility can be a problem in certain cases. For example, when a chiral ligand was appended to the ethylene oligomer for a catalytic asymmetric reaction, better stereoselectivity was expected at a lower temperature. This was a problem in Rh(II)-catalyzed asymmetric carbene insertion chemistry we studied with the Doyle group. In these cases, we could not run the reaction below 70 °C because the oligomer completely precipitated below about 70 °C, leading to an inactive catalyst.28 Tests of solubility based on a fluorescence analysis of a solution containing a polyethylene-Eu(+3) carboxylate show the temperature-dependent solubility behavior of the polyethylene oligomers increases markedly above 70–80 °C (Fig. 3).39
These terminally functionalized polymer-bound catalysts have other important features that are not always seen with insoluble polymer-supported catalysts. Specifically, they have reactivity that is more predictable and more like their low molecular weight analogs.31 This similarity in reactivity is expected based on the general observation that the reactivity of end groups on polymers with low molecular weight substrates (e.g., monomers in propagation steps) is typically not dependent on polymer molecular weight. These soluble polyethylene oligomer-bound catalysts, the ligands, and the intermediates in their synthesis are also more easily characterized than insoluble polymer-bound species. Specifically, we can use techniques like fluorescence, UV–visible spectroscopy, and solution-state NMR spectroscopy. 1H NMR spectroscopy in particular is very useful in following the course of synthesis with these oligomers because the signals as a result of the protons of the oligomer are all in a narrow window (δ 1.0–1.5), affecting only other signals that are in a relatively uninformative region for 1H NMR spectroscopy.23
The use of these soluble polymers also had some unexpected effects. Early on in our work I had prepared a (PEOlig-PPh2)4Pd catalyst from a stoichiometric amount of PEOlig-PPh2 and (PPh3)4Pd. In this synthesis, I noted that mixing a hot solution of the polyethylene-bound phosphine with the yellow-orange solution of the (PPh3)4Pd did not lead to any immediate color change but that the filtrate after cooling was colorless. This suggested a high yield in this ligand exchange reaction, a result we casually attributed to the presence of a more basic alkyldiphenylphosphine on the ethylene oligomer. However, when we later prepared an oxidation catalyst from a stoichiometric amount of PEOlig-CO2H and the deep purple ruthenium cluster [(CH3CH2CO2)Ru3O(H20)3]+ CH3CH2CO,34 we noted a similar effect. In an early draft of an article about these latter catalysts, we noted this supposed unusual effect. A referee pointed out that this was actually a well known simple entropic effect. This effect arises because these polyvalent metal complexes can have a mixture of polymeric and low molecular weight ligands. However, the isoenthalpic equilibria associated with the competition of electronically and sterically equivalent polymeric and low molecular weight ligands for a metal center entropically disfavor metal complexes that have only low molecular weight or only high molecular weight ligands attached to the metal. Because any complex with even one polyethylene ligand is insoluble at room temperature, precipitation and recovery of the metal or catalyst are favored even when some low molecular weight species are present. We subsequently verified this in studies with a fluorescent europium carboxylate and used this effect to advantage in asymmetric catalysis and metal sequestration where we were able to deliberately add a low molecular weight ligand without affecting catalyst or metal recovery.28, 39, 40
SYNTHESIS ON SOLUBLE SUPPORTS
Our work with polyethylene oligomers as soluble polymers containing terminal functional groups as catalyst ligands complements both the original studies of the Tübingen group and studies by a variety of other groups on syntheses with soluble polymers. Soluble polymers have experienced renewed interest in recent years in synthesis. This renewal of interest is reflected both by a number of recent reviews and by some innovative chemistry.41 In many of these studies, PEO continues to be the material that has attracted the most interest, However, soluble polymers with pendant groups are also useful. Selected examples of this chemistry taken from recent literature are provided and illustrate the useful properties of soluble polymers in synthesis. The examples shown here include the synthesis of solubilizing/protecting groups, the synthesis of protein tyrosine phosphatase inhibitors, and the synthesis of chiral catalysts. In each of these cases, the soluble polymer afforded advantages in purification without compromising reagent, substrate, or catalyst reactivity—often a critical issue in cases where chirality is involved. In each case, polymer recovery was effected by solvent precipitation, an approach used successfully in the liquid-phase peptide synthesis described previously.
POLY(ETHYLENE OXIDE) SUPPORTS IN SYNTHESIS AND CATALYSIS
Janda's group has been among the most active in using soluble polymers as reagent and catalyst supports. Although they have used other polymers, much of their work has focused on PEO or poly(ethylene oxide)-like polymers. They have described the innovative synthesis of block and star polymers like 2 and 3 to control solubility and increase loading, respectively.42, 43
The focus of a number of the contributions from Janda's group has been the use of polymer supports including soluble PEO supports in asymmetric synthesis. One of the more significant results from this research was their development of PEO-supported Sharpless dihydroxylation catalysts.44 This olefin dihydroxylation reaction is one of the most successful polymer-supported asymmetric catalytic reactions yet developed.8 Janda's contribution was to show that the soluble polymer-bound ligand 4 was especially effective in this chemistry. The soluble polymer-bound catalyst Structure 4 could be recovered by solvent precipitation but had reactivity and selectivity that mirrored its low molecular weight analogue. This catalyst was also effective in the oxidation of a polystyrene-bound substrate—chemistry that is of interest in the context of solid-phase synthesis.45 These dihydoxylation catalysts are also of interest because they provide a good example where multiple groups have discussed the relative merits of soluble versus insoluble polymers and organic supports versus inorganic supports for catalyst immobilization.46
SOLUBLE LINEAR POLYSTYRENE SUPPORTS IN SYNTHESIS
Linear polymers like polystyrene with pendant functionality have some advantages over PEO in cases where low-temperature solubility or acid stability is important.47, 48 Janda has described the use of such polymers in prostaglandin synthesis—syntheses that required reactions conditions (low-temperature chemistry in tetrahydrofuran) that were incompatible with the solubility of a PEO support. Taylor described the use of the more acid-stable polystyrene as a support in a recent article detailing the synthesis of α,α-difluoromethylenephosphonic acids (Scheme 2).48 In this chemistry, a hydroxyl-terminated tether was first bound to a linear chloromethylated polystyrene. Then a bromophenyldifluoromethyl phosphonic acid was bound to these pendant hydroxyl groups using Mitsunobu chemistry. The reactions could be followed by both 1H and 19F NMR spectroscopies, and the polymer was recovered and separated from other soluble byproducts by precipitation with 80% MeOH. A Suzuki coupling with arylboronic acids then completed the synthesis. After isolation of the product via solvent precipitation, trimethylsilyl iodide (TMSI) cleavage of the biaryl-α,α-difluoromethylenephosphonic acid from the polymer yielded the desired products in high yield. In this example, the soluble polymer facilitated separation yet allowed analysis of the intermediates by conventional NMR spectroscopy. This work like Janda's prostaglandin synthesis demonstrates the feasibility of multistep syntheses with soluble polymer supports.
SOLUBLE POLYSTYRENE OLIGOMERS IN SYNTHESIS
Although the aforementioned examples utilize linear high molecular weight polystyrene with pendant functionality as a support, the possible application of shorter linear polystyrene oligomers in synthesis has also been noted. For example, Quiclet-Sire et al.49 recently noted that short oligomers (15-mers) of polystyrene can easily be prepared using readily available xanthate esters as initiators for a controlled radical polymerization. They describe the use of such chemistry to prepare polystyrene derivatives like 5–7 from xanthate esters (cf. eq 1).
This chemistry may have applications in parallel syntheses because the polymer groups in effect behave like a large bulky protecting group that dictates the solubility of the oligomeric adduct in much the same way the polyethylene oligomers we previously used modified the solubility of a catalyst. However, unlike our chemistry, a polystyrene oligomer that serves as a temporary protecting/solubility control agent needs to be easily separated from the supported organic molecule of interest. Fortunately, the initial report by Quiclet-Sire's group shows that the oligomers were readily removed by base hydrolysis. An important implication of this work and of the polyethylene work previously discussed is that polymers that have little utility as materials (the polystyrene and polyethylene oligomers molecular weights are too low to have useful materials properties) can still have utility as supports in synthesis and catalysis (5 and 6).
POLYMERS WITH INVERSE TEMPERATURE-DEPENDENT SOLUBILITY
Our work with polyethylene-bound catalysts along with previous work with other soluble polymers has established the utility of soluble polymers as supports for catalysts and synthesis. However, although we had a lot of success using polyethylene oligomers as ligands, the 100 °C temperature required for ethylene oligomer dissolution and the limited solubility of polyethylene in many solvents precluded our using these catalysts in some reactions and was probably a deterrent to others using these catalysts. Around this time, I was attending a series of Gordon Research Conferences as a way to ‘educate’ myself in the precepts of polymer chemistry. At one of these conferences, I remember meeting a very well-known surface chemist and discussing our new soluble polymeric catalysts. I am sure I was very enthusiastic about how these ethylene oligomers went into and out of solution on heating and cooling. I am less sure that the professor from Lehigh University was that impressed. What I do remember is his telling me that polymers do not always dissolve on heating and that some precipitate on heating. I believe he used as an example the phase separation of PEO from water at 95–100 °C. That story stuck with me for a number of years and eventually led us to a whole new project and a new way to use polymers to recover and reuse catalysts.
The idea to use polymers to prepare ‘smart’ catalysts actually originated from a combination of several events. Specifically, I was faced with the all too frequent periodic necessity of writing another proposal in my near future. I also was lucky enough to get the opportunity to go to Japan for a catalysis conference. The necessity for some innovative ideas in the proposal, the long flight, the more temporally distant advice by that professor about inverse temperature-dependent solubility, and the disappointment from some colleagues about the ‘inactivity’ of our polyethylene oligomer-bound catalysts below 60 °C fortuitously led me to the realization that a catalyst on a polymer that precipitated on heating would both provide a way to isolate or separate a catalyst from products or reagents and a way to turn a reaction off. I also realized that a catalyst on such a polymer that redissolves reasonably rapidly on cooling would be a ‘smart’ catalyst—a catalyst that would control an exothermic reaction by turning a hot reaction off until it cooled. I remember distinctly refining these initial ideas in a discussion with a former student, now a professor in Japan, while he drove me through the Japanese countryside from Narita airport into the Tokyo gigopolis.
On returning to the U.S. I proceeded to look to see if this seemingly obvious idea had already been described. Somewhat to my surprise, I found that others had not used this idea to control a catalytic reaction. I rationalized this as being the result of most professors having the sense to have a lack of interest in developing new ways to make reactions fail. However, I nevertheless set out to find or make some of the sorts of polymers the professor from Lehigh University had described. Fortunately, our initial synthetic task was simplified when some of my industrial contacts led me to contact Wyandotte Chemicals, a division of BASF. They provided our laboratory with a collection of oligomeric triblock surfactants containing ethylene oxide/propylene oxide/ethylene oxide blocks.
The triblock copolymers available through BASF under the tradename Pluronics have a lower critical-solution temperature (LCST, the temperature at which the polymer phase separates from solution on heating) that varies depending on the sizes of the hydrophilic PEO and hydrophobic poly(propylene oxide) blocks. We initially chose a substrate that had an LCST near room temperature and carried out the series of synthetic operations as shown in Scheme 3.
The results for these experiments were the successful synthesis of several hydrogenation catalysts—materials that did in fact turn reactions off and on in an anti-Arrhenius manner by heating and cooling, respectively (Fig. 4).50
Although these materials worked as ‘smart’ catalysts, we soon realized that materials that phase separate as oils were less desirable than materials that would separate from solution as a solid. This is because a ‘smart’ catalyst that separates from solution as an oily phase may actually accelerate a reaction if the substrate partitions into the oily catalyst phase above the LCST. Moreover, at that time we also wanted ways to recover and separate catalysts from products and reagents using a solid/liquid separation. Thus, we turned our attention to polymers that had LCST properties but whose molecular weight or structure led to their separating from solution as solids.
In looking for alternatives to these poly(alkene oxide)s we were specifically attracted to polyacrylamides like poly(N-isopropylacrylamide) (PNIPAM). We already knew at this point that addition polymers with pendant functionality can be useful in synthesis, and we had the good fortune to be able to build on work by others who had earlier described the synthesis of functional PNIPAM polymers and PNIPAM derivatives whose LCST temperature could be tuned by adjusting the hydrophobicity of the pendant N-alkyl groups of the acrylamide.51–53
PNIPAM chemistry proved to be a very useful polymer support in our studies for several reasons. First, it was easy to make high molecular weight polymer by radical polymerization. High molecular weight polymers are better for our purposes because the insolubility with temperature is most simply seen with high molecular weight polymers. High molecular weight also enhances the phase separation of polymers in mixed liquid-phase systems—an important consideration in later work. Second, we could use the synthetic chemistry schematically shown in eq 2 to prepare a wide variety of polymer-bound ligands, catalysts, and substrates. Third, it was relatively simple to adapt the synthetic reaction shown in eq 2 so that a dye molecule was included in the polymer in place of a ligand or substrate. Incorporating a dye molecule facilitates initial quantitative studies to establish that polymer separation (and hence ligand, catalyst, or substrate separation) is quantitative.
is that it is very versatile. We have successfully used PNIPAM derivatives where the loading of the active ester is as high as 5–10 mol %. The product polymer can also be readily analyzed by 1H or 13C NMR spectroscopy.54 Although the polymer backbone has broad peaks, the use of a short tether for a ligand or substrate leads to sharp peaks for the ligand or substrate protons. In cases where the solubility of the product is affected by the hydrophilicity or hydrophobicity of the ligand, we can also modify the product polymer through synthesis of a terpolymer. For example, use of a substoichiometric amount of the desired amine-functionalized ligand or substrate followed by the addition of a more hydrophilic amine (e.g., NH3) or a more hydrophobic amine (e.g., decylamine) leads to terpolymers whose LCST is similar to PNIPAM. For example, in the specific ensuing example where the ligand was the hydrophobic diarylphoshphine NH2CH2CH2CH2PPh2, addition of ammonia produced a product polymer that was soluble in cold water and insoluble in water at room temperature.
We have described a number of ligands and catalysts attached to PNIPAM. Examples include the compounds 9–12 in addition to the simple phosphine ligands shown in eq 3.
These catalysts have been used in a variety of reactions including Heck catalysis,55 simple hydrogenation,54 asymmetric hydrogenation, allylic substitution of acetates by amines,56 alkyne–aryl halide couplings,57 aryl amination, and Suzuki reactions.57 Our results suggest >99% recovery of catalysts in each case. A problem we are now dealing with is the instability of phosphine ligands to adventitious oxidation—a problem that is not present in the more stable terdentate system 12.
AMPHOTERIC POLYMER SUPPORTS
Another approach to engineer solubility into a polymer is to take advantage of soluble polymers that have amphoteric behavior. Both my group and others have described such chemistry.58, 59 In our case, we prepared the polymeric ligand 13 using a commercially available maleic anhydride-methyl vinyl ether copolymer as a starting material. Only a portion of the anhydride groups were used to bind a phosphine ligand via amide bonds. The balance of the anhydrides was converted to succinic acid groups. Solubilization of the product polymer and polymer-bound catalyst in water was then effected by merely increasing the pH to form carboxylate groups. Decreasing the pH to make the solution acidic regenerated the CO2H groups and precipitated the polymer (see eq 4).
FLUOROUS PHASE-SOLUBLE POLYMER SUPPORTS
An entirely different approach to catalyst recovery that did not require a polymeric support was described by Horvath.60(c) This fluorous-phase chemistry was first discussed by Horvath as a nonaqueous analog to aqueous biphasic catalysis already used in industry. This fluorous-phase chemistry offered chemists an alternative to soluble or insoluble supports for catalyst or reagent recovery and reuse based on the immiscibility of fluorocarbons and organic solvents. This liquid/liquid biphasic separation scheme is modeled after aqueous biphasic separations already used in catalysis. However, to be selectively soluble in fluorocarbon phases, catalysts need to have ligands with “Teflon” ponytails. Such perfluoro groups need to be electronically isolated from the actual ligating group to preserve the original ligand character.
We readily recognized that our soluble polymer approach was potentially applicable to the problems of preparing a generic fluorous-phase soluble support. The idea in all of the aforementioned chemistries is that polymers can be designed so that they impart a desired solubility to a bound substrate without profound modification of the substrate's (ligand's) reactivity. In our initial experiment, we tested this using a fluoroacrylate bound azo dye. The necessary polymer was prepared from a fluoroacrylate/N-acryloxysuccinimide copolymer and an amine-derivative of p-methyl red according to eq 5.61
This dye cleanly separated into a fluorocarbon phase from fluorocarbon/organic emulsion. It also cleanly, rapidly, and reversibly reacted with organic-phase soluble acids and bases, showing that a fluoracrylate-bound species can retain good reactivity.
Extending this chemistry to catalysis involved similar chemistry (eq 6).62, 63
The product Rh(I) catalyst was unfortunately not easily characterized by 31P NMR spectroscopy, which only showed a broad singlet at δ 32.8. However, elemental analysis showed that the polymer contained 0.013 mmol of Rh and 0.043 mmol of phosphine/g of polymer. These catalysts were indeed active in hydrogenation chemistry with turnover numbers of 21,700 mol of alkene reduced/mole of Rh. Rates for alkene hydrogenation varied with the alkene and were typically in the range of 65–203 mmol of H2/mmol of Rh. Separations involved a resting, biphasic stage and a mixing emulsion stage for the actual reaction (Fig. 5).
Although the fluorous systems offer many advantages, I felt that the problems of using a costly fluorocarbon solvent could pose problems with this system. This concern may not be well founded because a number of groups are making considerable strides in fluorous biphasic catalysis and in using fluorinated tags in synthesis.60, 64 Nonetheless, we turned our attention to a minor feature of Horvath's original Science article—his observation that some fluorous biphasic systems become monophasic on mild heating and revert to the original biphasic condition on cooling. Brief further investigation including some empirical experiments with common solvents showed quickly that such phase behavior is quite common. Thus, we conceived yet another approach to separation in catalysis where a polymer's solubility would simplify catalyst/product separation. This idea is illustrated in Figure 5. In the thermomorphic system shown in Figure 6, we rely on a polymer's large distribution coefficient—its preference for one of two phases in the cool resting state–to ensure a simple liquid/liquid separation. We reasoned that if the polymer we chose maintained its solubility in the miscible mixture, we could carry out true homogeneous catalysis and still use this liquid/liquid-separation process.
Our expectations have generally proven to be correct.60, 64 Although our efforts in this area are ongoing, it is clear that the use of solvent mixtures whose miscibility is temperature dependent in conjunction with soluble polymeric ligands is a viable method to engineer catalyst recovery in homogeneous catalysis. We feel this scheme is particularly advantageous because it combines a simple separation with more predictable catalyst reactivity. Not only does the polymer support's solubility facilitate catalyst characterization but we can still use techniques like membrane filtration to recover the polymer-bound catalyst at the end of the reaction.
TRACE METAL SEQUESTRATION WITH SOLUBLE POLYMERS
Finally, the complete control of solubility of metal complexes through the use of a polymer-bound ligand suggested yet another application of these materials—sequestration of trace metals.40 In work that is still ongoing, we have been able to show that a soluble polymer ligand can be used to design materials that both sense and sequester a trace metal. Using hydroxamic acid ligands, hydroxypyridinone ligands, and imidazole ligands on PNIPAM and fluorous polymers (14–16), we have been able to design soluble systems that completely remove sample trace metals from aqueous or organic solutions.
Soluble (and insoluble) polymers have demonstrated success as supports for catalysts, reagents, and sequestering agents. They also have a very promising future as supports in catalysis and synthesis. Dendrimers, new co- and terpolymers, new insoluble supports, and polymer-enclosed vesicles, are among the sorts of soluble or homogeneous macromolecular assemblies that will likely have increased use in the future. In all of these cases, it is likely that polymers can be designed that affect the physical handling or separation of a catalyst without impugning or altering the catalyst's selectivity or reactivity. Similar consideration should apply to the design of new polymeric supports for synthesis. In many cases, concepts or ideas that are well known to the polymer community may be ‘novel’ to the organic community and can be developed into useful chemistry for synthetic applications. Thus, there is considerable opportunity for chemists with knowledge of both areas to develop creative approaches that address problems in synthesis and catalysis.
However, although there are many opportunities in studying soluble supports in synthesis and catalysis, there are also many unanswered questions and problems associated with using polymers as supports for catalysts or reagents. For example, the use of soluble polymers in synthesis is not nearly as extensive as is the use of insoluble crosslinked polymers. This partly reflects the fact that the technology for combinatorial chemistry and high throughput synthesis is based on the more standard insoluble resins. Nonetheless, it is plausible that schemes can be devised whereby many of the advantages of soluble polymers too can be used to simplify the separation processes in such chemistry. The dynamics of complexation of metals by a single polymer or multiple polymers has not been extensively studied. There should also be ways to use soluble polymers effectively in combinatorial development of new catalysts.
Support of our work on soluble polymer supports and the development of new media and new processes by both the National Science Foundation and the Robert A. Welch Foundation is gratefully acknowledged. Finally, although the ideas the author has discussed may have occasionally been developed when he was at a meeting or conference, the graduate students, undergraduate students, and postdoctoral students whose names are listed in the cited references really deserve the credit because it is only through their hard work and creativity that any of this work reached fruition.
David Bergbreiter was born in 1948 in Chicago, IL. Through college, he was the product of public education, matriculating from high school in the Chicago public schools and receiving a B.S. in Chemistry from Michigan State University in 1970. He received his Ph.D. in 1974 from the Massachusetts Institute of Technology (MIT) where he worked under the tutelage of Prof. G. Whitesides. At MIT he was exposed to a variety of intellectual experiences in Whitesides' group, working on projects in polymer chemistry, organometallic chemistry, and organic synthesis. Since 1974, he has been a faculty member of Texas A&M University. In Texas, his research has spanned several areas of chemistry including topics like asymmetric synthesis, catalysis, organometallic chemistry, surface chemistry, and physical organic chemistry. After over 26 years, he considers himself an immigrant Texan because he and his wife Lynne own a few acres of land, have a tiny share of an oil well, and have two “native-Texan” daughters, Sarah and Amy.