Polymerization of Alkoxides to Polyethers
While ethylene oxide is an important building block for synthesis of polyurethane polyether soft segments, there are only a few exceptions where polyethylene glycol (PEG) is utilized as the sole soft
FIGURE 2.8 Production of polypropylene oxide via base catalyzed ring opening polymerization.
segment. This is because of the crystallinity of PEG and because of its affinity for water. The affinity for water limits PEGs use since the properties of the polyurethane can be strongly dependent on the humidity of the environment. Instead, for polyurethanes, ethylene oxide is usually used as a coreactant with other alkylene oxides, usually propylene oxides, to create block or random copolymers.
Propylene oxide is polymerized to polypropylene oxide by two industrially dominant processes. Both processes are defined by the catalyst that is used. The dominant method employs an initiator or starter molecule with active hydrogens such as water, ethylene glycol, glycerine, ethylene diamine, or sorbitol and a base catalyst like potassium hydroxide . This approach is usually referred to as the base-catalyzed process. The other process to polypropylene oxide uses a class of heterogeneous metal-based catalysts termed "double-metal cyanide" catalysts (aka DMC catalysts) . Each process has unique attributes including differences in cost, process flexibility, and even structure of the final product. The base-catalyzed process follows the simplified path given in Figure 2.8.
The functionality of the final product is determined by the functionality of the initiating alcohol (see Table 2.2) and the final molecular weight influenced by the ratio of initiating alcohol to oxirane monomer. The base resides in a molecularly dispersed state with the oxirane and is termed a homogeneous catalyst. In actuality, the process is more complicated than expressed in Figure 2.8. The reaction mechanism of Figure 2.8 suggests that the result of the reaction is all secondary hydroxyl end groups. However, this is not in fact observed. In the synthesis of polypropylene oxide, there are also observed a population of primary hydroxyl end groups and a population with vinyl termination. The vinyl termination is a particular concern since it is not reactive with the other components of polyurethane chemistry and so degrades the polymer network structure. The actual chemistry that occurs within the reactor is explicitly given in Figure 2.9. The reaction resulting in secondary hydroxyls predominates, but the other structures do form and the process and specific product influence the product distribution.
The primary product in base-catalyzed (or explicit in these reactions—base initiated) polymerization is the product predominated by secondary hydroxyls due to the relative ease of nucleophilic attack on the less hindered carbon of the oxirane. Alkoxide attack on any of the methyl hydrogens results in unsaturation given
TABLE 2.2 Common polymerization starters or initiators for polyether polymerization
mechanistically in Figure 2.10 for the formation of allyl alcohol, which can then serve as the initiating alcohol for monohydroxy polyether formation referred to as "monol" .
The amount of monol formed in the polymerization is a function of several process details increasing primarily with temperature. An empirical equation for estimation of polyol functionality taking into account the effect of monol is given by Equation 2.1 :
Alternatively, the functionality can be calculated from the measured OH number (ASTM D4274)  and the ASTM measured allyl concentration using Equation 2.2:
The amount of unsaturation is measured by a standard method such as ASTM D4671 . Unsaturation is quantified in units of "milhequivalents unsaturation per gram of polyol," and the diol content is obtained from either chemical analysis or by calculation. Figure 2.11 shows the potential loss of network connectivity resulting from formation of unsaturation of a nominally 5000 molecular weight triol. Similarly, a 4000 molecular weight diol can have a true functionality of 1.7 due to monol formation.
Ethylene oxide cannot form unsaturation in the manner of propylene oxide since there is no beta carbon from which to abstract a hydrogen. Butylene oxide is capable of forming an unsaturated monol initiator in a manner analogous to propylene oxide.
The problems associated with monol formation have stimulated research for alternative polymerization catalysts less prone to formation of unsaturation. One approach has been to vary the complexing counterion. It has been shown that the tendency for the counterion to promote formation of allyl alcohol decreases in order down the periodic table from Li+>Na+>K+>Rb+>Cs+. It has been reported that the use of CsOH instead of KOH can reduce the formation of unsaturated end groups by 50% [30, 31]. Alkaline earth salts bearing Ba+2 and Sr1"2 have also been reported to result in significantly reduced formation of unsaturation in polyether synthesis .
As with all homogeneous catalysts, the base must either be extracted or neutralized after the polymerization step. If the polyol is sufficiently immiscible with water, the catalyst can be extracted with water. The polyol will coalesce as a separate phase and be easily removed and, if necessary, the catalyst recovered. In general, the rate of base-catalyzed reaction is considered to be first order in catalyst and in oxirane:
FIGURE 2.9 Possible pathways PO polymerization can traverse to products. The route to secondary hydroxyls is the most prevalent.
FIGURE 2.10 Mechanism for formation of unsaturation in polyether polyols. PEGs cannot do this due to the absence of pendant carbons for attack.
FIGURE 2.11 The calculated actual functionality for a for a 5000 molecular weight (1666 equivalent weight) triol reflecting the effects of end-group unsaturation developed during polymerization.
where K2 is the second-order rate constant. However, since the catalyst is not consumed, the rate can be further simplified to Equation 2.4:
where Kj is a pseudo first-order rate constant equaling the product of [catalyst] xK2.
Separate from the homogeneous base initiation of oxirane polymerization, a heterogeneous catalyst system has been developed. The most successful approach has been the development of heterogeneous catalysts described collectively as "DMC" catalysts . DMC describes any of the class of catalysts of general structure
where M refers to a metal, CN refers to the CN ligand, and A is an additive.
The effectiveness of DMC catalysis in polymerization of epoxies was first noticed in the early 1960s and rediscovered for the purpose of polypropylene oxide production by Shell, Arco, Asahi Glass, Union Carbide, and others in the 1980s. Considering the amount of time and effort that has been put into this very enticing class of catalysts, it remains safe to say that there are many open questions about how they operate. While DMC catalysts show a high level of reproducibility and reliability in their function, this technology has largely advanced through empirical observation and what can only be termed as "art."
The most efficient DMC catalysts for polymerization of polypropylene oxide contains Zn3[Co(CN)6]2 scaffolding in its structure. However, it is well established that this structure alone is inactive for polymerization of propylene oxide and that additional additives must be incorporated for desired results. DMC additives typically include metal salts such as Zn(Cl)2 and low-molecular-weight solvents such as tert-butanol . Preparation of the optimized active catalyst is an involved and precise procedure. The mechanism of the catalyst is highly uncertain. The catalyst is sometimes referred to as a "precatalyst" since there is usually an induction time on the order of 30min to several hours between when reaction conditions are reached and when molecular weight build due to polymerization commences. Several theories exist for this behavior; however, given the fundamental limitations of precisely determining the surface conditions during reaction, full understanding remains an area open for research.
One mechanism suggested for the action of DMC catalysts involves the initial substitution of the CI ligand of ZnCl2 with the alcohol (i.e., tert-butanol) stated in the previous text to be necessary for catalyst activity . Another substitution of CI on the Zn metal center coordinates the reactive species to a single site in a geometry potentially effective for an addition reaction to occur. Proton transfer from the alcohol coordination site to the Zn-PO bond ring opening would be facilitated. This model is consistent with the observation that too much alcohol (R'OH in Fig. 2.12) depresses the activity of the catalyst. In this scheme, too much R'OH would possibly compete with the alkoxide for coordination to Zn . It is not immediately clear how this model accounts for the induction period, though it is possible to rationalize that
FIGURE 2.12 Proposed coordination-insertion mechanism for activity of DMC catalysts polymerizing propylene oxide.
attaining the condition of cocoordinate alcohol and alkoxide may take time and may depend on such experimental details as mixing, temperature, the population of active catalyst sites on the surface, etc. Similarly, it is not immediately clear why the ratio of Zn to Co in the catalyst package has such a significant effect on catalyst activity.
An alternative (but somewhat related) mechanism also relies on coordination of reactants to the surface but takes into account the totality of the catalyst morphology including its layered structure and the coordination geometry. As mentioned previously, when the entire catalyst package including added metal salts such as ZnCl2 is taken into account, the total formula for the catalyst can be generalized from Zn3[Co(CN)6]2 to
For a Zn/Co ratio of 1.5, each Zn is coordinated to four cyanides and lacks a coordination site (assuming 4-coordinate Zn). In fact, it is established that DMC catalyst activity for this series of catalysts requires a Zn/Co ratio greater than 1.5. For a Zn/Co of 2.0, zinc obtains a population of coordination conditions with an average
FIGURE 2.13 Proposed acid-base coordination mechanism for activity of DMC catalysts polymerizing polypropylene oxide.
of three cyanides and one chloride for every two zinc ions. By this model, a surface fragment of DMC catalyst may have two Zn sites: one acid capable of binding to an alcohol function and, the other, a basic site capable of binding to an alkoxide. The binding of the alcohol to the acid can activate the proton as suggested in the model described by Figure 2.12. Figure 2.13 describes this model for the acid-base-regulated coordination chemistry.
Regardless of the details of the mechanism of DMC polymerization of polyols, there are several advantages associated with this process. These include higher activity at any given temperature (allowing low catalyst loadings), narrow molecular weight distributions, a highly atactic product (low tendency to crystallize), and very low levels of unsaturation, well under O.Olmeq/g (see Fig. 2.14 for comparison to KOH). The meaning of the lower unsaturation to the polymer structure can be seen in Figure 2.14. Since the number of end groups decreases rapidly with increasing molecular weight, and the formation of unsaturation is dependent on reaction conditions but are otherwise fixed, the effects of unsaturation on polyurethane network connectivity will intensify as the molecular weight increases. The differences between KOH- and DMC-catalyzed polymerizations become more noticeable at equivalent weights greater than 1000g/eq .
The obvious advantages associated with DMC catalysis in the final product come with a price. A well-known process caution is associated with the induction period. Given the very high activity of DMC-catalyzed ring opening and the high
FIGURE 2.14 Difference of measured polyol functionality attained by KOH and DMC catalysis for a triol of variable equivalent weight. The difference becomes appreciable at higher equivalent weights as the number of end-groups decreases. Reprinted with permission from Ref. . © John Wiley & Sons, Inc. (See insert for color representation of the Figure .)
exotherm of the reaction (~25 kcal/mol), the induction results in a very significant and potentially unpredictable exotherm produced during the polymerization. Process optimization has dealt with this on an industrial scale by standardization of materials and process .
A larger and more intractable problem is the inability to develop block copolymer structures of ethylene and propylene oxide. As will be discussed later, there is a significant reactivity difference between the secondary hydroxyls from polypropylene oxide and the primary hydroxyls of polyethylene oxide. One processing compromise to obtain primary hydroxyls without the drawbacks of all PEG structures is to polymerize PO to polypropylene oxide to a given molecular weight and subsequently add EO. When handled appropriately, homopolymerization of the EO can be minimized, and ethylene oxide blocks can grow from the PPO polymer ends. With KOH catalysis, this is not a problem since PO and EO can be equally polymerized via the mechanism of Figure 2.9. A similar process has not been achieved with DMC catalysis. The reason for this has not been well established. One hypothesis has been that the growing PEG block has a much higher affinity for the DMC catalyst surface causing the initially formed end blocks to grow inhomogeneously. The ever-growing PEG blocks become ever less miscible with the PPO bulk, and so polymer chain exchange off the surface to the bulk becomes increasingly less favorable. Instead, obtaining EO-capped PPO polyols prepared by DMC catalysis has required deactivation of the DMC catalyst followed by EO end capping using conventional KOH catalysis. A process in which EO mixed with the PO in the reacting mixture has been developed and proven effective at producing random PO/EO polymer chains .
A product-related problem with DMC catalysis is ironically caused by the high level of patent activity that has been pursued by a few competitors. The result has
FIGURE 2.15 Measured unsaturation levels by titration of three polyols for a 2000 molecular weight diol. The lower the value the less vinyl end groups found in the system .
been that only a few industrial concerns have developed or licensed patent coverage for making DMC-derived polyols. At the same time, there have been advances in traditional base catalysis using varied counterions (Fig. 2.15). The motive for further innovation in heterogeneous coordination catalysis has also been dampened by the observation that polyurethane foam properties, accounting for a large volume of polyurethane usage, are not noticeably affected by the undeniable polyol structural improvements derived by DMC catalysis. This has relegated DMC polyols to nonfoam applications—those applications where the low reactivity of secondary hydroxyls is irrelevant. DMC-catalyzed polyols are also preferred for higher equivalent weight polyols since at lower equivalent weights the structural differences are minimized as the fixed number of vinyl chain ends is less important due to the large number of total chain ends.
Today, the vast majority of industrially produced polyols are prepared by KOH-catalyzed polymerization of PO or PO/EO monomers. The process in concept is remarkably simple, but the details associated with producing a commoditized product can create significant amounts of redundancy, controls, and safety additions. To make polyols efficiently and competitively, the production volumes must be very large and the accompanying costs very large as well. Plant capacities range from about 50 thousand metric tons to about 500 thousand metric tons. For the purpose of scale, a 50 thousand metric tons/year plant can produce about 12 thousand lbs/h of polyol. Smaller quantities are sometimes produced by enterprises that use the entirety of their product captively. Figure 2.16 provides a very simplified view of a polyol plant process to nominally produce a three-functional (vis-a-vis glycerin) 12.5% ethylene oxide-capped PO or a random PO/EO polyether polyol by a batch process . Also provided are the approximate unit ratios required to make that polyol 1000 g/OH eq. An important and unappreciated component of the process is the
FIGURE 2.16 Simplified block diagram for a small polyol production of a glycerine initiated (tnol) KOH catalyzed PO/EO block co-polymer (12.5% EO) with total molecular weight 3000 (equivalent weight 1000 g/eq).
treatment stage at which unreacted monomer may be driven off and the KOH removed by filtering with magnesium silicate, or neutralized with acidified water, and the polyol coalesced as a separate phase from the water and decanted. These neutralization and filtering steps are important aspects of the process and an important component of process research when introducing a new product.
The polyether polyols from THF (referred to commonly as polyfctramethylene ether glycol (PTMEG)) are treated separately here and in many other places for several reasons. One reason is because, although a relatively high volume polymer (~500 million pounds/year) , PTMEG occupies a specialty position within polyurethane chemistry as a whole. As a specialty, it is produced at smaller volumes, by a significantly different process, and at much higher costs to manufacture and prices to the consumer. Growth rates are in-line with most other polyols (~4% per annum) reflecting growth in the overall global economy. The major uses by volume for PTMEG are polyurethane elastomers, sealants, adhesives, polyesters, and spandex fibers.
The production of PTMEG is achieved by two routes . The most common way is via acid-catalyzed ring-opening polymerization (Fig. 2.17). This path has the advantages of simplicity. The most common acid is fluorosulfonic acid, although other variations are practiced as well. Disadvantages of this process include inability to recover the acid, the potential for the acid to degrade the product, and difficulties in molecular weight control due to the limited but still finite ability of terminal hydroxyl groups to be deprotonated and initiate polymerization.
FIGURE 2.17 Acid catalyzed polymerization of THF to make polytetramethylene ether glycol (PTMEG).
FIGURE 2.18 Process for preparation of PTMEG by reaction of THF with acetic anhydride to form the stable ester.
An alternative process has been established producing the final polyether product by preparation of an intermediate but stable diacetate ester, which can subsequently be converted to diol by alcoholysis (Fig. 2.18) . The greater stability of the ester end group in this environment allows for better molecular weight control.
While many older plants utilize the acid-catalyzed polymerization route, there has been significant optimization of the acetic anhydride polymerization process now used by many newly constructed plants. Industrial production always entails separation of unwanted side products by either volatilization or filtration and recyclization of unused reagents back to earlier stages of the reaction. Figure 2.19 is a highly simplified version of what would be included in a real plant but does provide detail of the primary unit operations.
Because of the similarity of polymerization processes, monomers of polyether polyols synthesis are very amenable to copolymer design. Virtually any imaginable
FIGURE 2.19 Simplified block diagram for production of PTMEG using acetic anhydride/ alcoholysis process.
FIGURE 2.20 Cartoon representations of the types of co-monomer organization available to produce polyalkylene glycol soft segment structures. Reprinted with permission from Ref. . © John Wiley & Sons, Inc.
design and combination of oxirane monomers have been combined to examine properties and look for advantageous structures. Figure 2.20 illustrates the various combinations that are imaginable with these building blocks.
The most common type of block copolymer structure is the internal block polyol in which a polypropylene oxide block is reacted at either end with ethylene oxide to
Figure 2.21 Fractional consumption of polyester polyols in 2012.
balance the low T, low tendency to crystallize with the relatively high reactivity of the primary hydroxyl end group.