Long-Chain Aliphatic Polymers To Bridge the Gap between

Publication Date (Web): March 29, 2016 .... Yet, several thousand tons end up in the synthesis of nylon-6.10, which shows good chemical resistance and...
1 downloads 23 Views 16MB Size
This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Review pubs.acs.org/CR

Long-Chain Aliphatic Polymers To Bridge the Gap between Semicrystalline Polyolefins and Traditional Polycondensates Florian Stempfle, Patrick Ortmann, and Stefan Mecking* Chair of Chemical Materials Science, Department of Chemistry, University of Konstanz, Universitätsstrasse 10, D-78457 Konstanz, Germany ABSTRACT: Other than their established short-chain congeners, polycondensates based on long-chain difunctional monomers are often dominated by the long methylene sequences of the repeat units in their solid-state structures and properties. This places them between traditional polycondensates and polyethylenes. The availability of longchain monomers as a key prerequisite has benefited much from advances in the catalytic conversion of plant oils, via biotechnological and purely chemical approaches, likewise. This has promoted studies of, among others, applications-relevant properties. A comprehensive account is given of long-chain monomer syntheses and the preparation and physical properties, morphologies, mechanical behavior, and degradability of longchain polyester, polyamides, polyurethanes, polyureas, polyacetals, and polycarbonates.

CONTENTS 1. Introduction 2. Sources of monomers 2.1. Classical organic synthesis 2.2. Plant-oil fatty acids as a source of midchain monomers 2.3. Biotechnological ω-oxidation 2.4. Olefin metathesis 2.4.1. Self-metathesis reactions 2.4.2. Cross-metathesis reactions 2.5. Isomerization functionalization 2.6. Combination of isomerization functionalization and metathesis transformations 3. Long-chain aliphatic polymers 3.1. Synthesis of long-chain aliphatic polyesters 3.2. Typical properties of long-chain aliphatic polyesters 3.3. Synthesis of long-chain aliphatic polyamides 3.4. Typical properties of long-chain aliphatic polyamides 3.5. Further long-chain polycondensates 3.5.1. Long-chain aliphatic polyurethanes and polyureas 3.5.2. Long-chain aliphatic polyacetals and polycarbonates 4. Conclusion Author Information Corresponding Author Notes Biographies Abbreviations References © 2016 American Chemical Society

1. INTRODUCTION For the largest part, synthetic polymers possess an aliphatic backbone structure. In terms of scale, the most prominent examples are polyolefins. However, polycondensates like polyesters and polyamides also are largely aliphatic. In technically relevant polyesters and polyamides, the aliphatic segments are relatively short, typically amounting to six atoms or shorter linear carbon chains, −(CH2)n−. Thus, the physical and applications properties are often dominated by the polycondensates’ functional groups. For example, the advantageously high modulus and heat-distortion temperature of polyamide-6,6 arise from hydrogen bonds between the amide groups. However, an intermediate situation in which both a polycondensate nature and an aliphatic chain nature contribute substantially is of interest in its own right. For example, other than low-density polyethylene (LDPE), polyesters can allow for hydrolytic degradability and enable applications like disposable bags for compostable waste. To achieve sufficient melting and crystallization points, however, aromatic repeat units are required and the commercial products are mixed aromatic− aliphatic polyesters. In principle, such higher melting points can also be achieved in all-aliphatic polyesters by virtue of longerchain aliphatic repeat units that crystallize via van der Waals interactions between the hydrocarbon segments, akin to polyethylene. Polyesters and polyamides with longer-chain linear aliphatic repeat units were studied early on, already in the pioneering work of Carothers.1 The relationship between monomer chain length and melting points was studied in much detail for

4597 4598 4598 4599 4601 4602 4602 4604 4605 4608 4609 4609 4613 4617 4618 4623 4623 4628 4631 4631 4631 4631 4631 4631 4632

Received: December 9, 2015 Published: March 29, 2016 4597

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

polyamides and also polyesters. Other than this, longer-chain difunctional monomers and their corresponding polymers have received comparatively little attention. This can be traced to the lack of viable routes to access these monomers in a more practical fashion than the multistep traditional organic syntheses employed for the preparation of these model polymers. In principle, fatty acids are an attractive substrate to consider for this purpose in that they contain longer aliphatic segments −(CH2)n− as a structural feature along with a terminal functional group. To illustrate the problem of their conversion to polycondensation monomers, the existing application of fatty acid derivatives as cross-linkers in polyurethanes is instructive. In this case, a multiple midchain functionalization, e.g., via epoxidation of the double bonds and subsequent hydrolysis, affords multifunctional molecules.2 As another example, branched “dimer fatty acids” as difunctional monomers are produced by a midchain dimerization of two molecules of unsaturated fatty acids.3−5 However, to obtain crystallizable linear α,ω-difunctional monomers, other approaches like a selective terminal functionalization of the fatty acid chain are required. Recent advances in catalytic conversions of plant oils provide solutions to this problem. This encourages bridging the gap between semicrystalline polyolefins and traditional polycondensates. Concerning the scope of this Review, long-chain monomers are considered to consist of a chain of 14 carbon atoms or longer. As a background, there is no consistent and generally accepted precise definition of the term “long-chain”,6−10 but there is a rather general agreement that 10−15 carbon atoms represents the lower limit of “long-chain” compounds. Our definition is based on the analogy to myristic acid as a wellavailable representative of the lower end of the spectrum of different carbon number fatty acids.

Scheme 1. Structure of the Major Repeat Units of Suberin and Cutin11

By a repeated sequence of monobromination of the C22 diol and coupling to double the original chain length, an extension up to 44 carbon atoms was possible, yielding tetratetracontane1,44-diol.13−15 Even longer α,ω-functionalized compounds have been obtained by a multistep procedure developed for the synthesis of very long chain alkanes16 and functionalized derivatives thereof.17 On the basis of an iterative sequence involving a coupling step through Wittig olefination of an aldehyde, straight-chain aliphatic dicarboxylic acids containing from 48 up to 192 methylene groups were generated (Scheme 3). Another useful coupling protocol to linear long-chain α,ωfunctionalized compounds is Kolbe electrolysis.18 Anodic decarboxylation of dicarboxylic acid (half-)esters gives access to radicals for dimerization and coupling, leading to the higher homologues of the dicarboxylic acids (Brown-Walker coupling).19−21 Starting from sodium ethyl sebacate, the linear long-chain aliphatic C34 diester can be obtained via two subsequent electrolysis steps (Scheme 4). As a side note, this synthetic method has also been applied to the methyl ester of the C36 dimer fatty acid (isomeric mixture), which was coupled to yield a mixture of branched C70 diesters.22 Further approaches have been developed to extend the chain length of α,ω-functionalized compounds. Starting from aliphatic α,ω-dihalides, long-chain dicarboxylic acids with up to 22 carbon atoms have been prepared by metal-catalyzed coupling reactions with short-chain α,ω-ester acid chlorides.23 In this way, for example, 1,10-dibromodecane can be extended to yield docosane-1,22-dioic acid (Scheme 5). Even longerchain dicarboxylic acid esters, namely, dimethyl tetracosane1,24-dioate and diethyl tetratriacontane-1,34-dioate, can be prepared by coupling with iodo-substituted carboxylic acid esters of the appropriate chain lengths.24 An alternative route to docosane-1,22-dioic acid is a chain extension of decane-1,10-dioic acid by 12 carbon atoms via C− C scission of a cyclic malonic ester as the key step (Scheme 6).25 This chain-extension approach has been applied to different α,ω-dicarboxylic acids.26 Moreover, by both increasing the ring size of the enamine and multiple application of this reaction sequence, chains of considerable length can be constructed.27−29 In this way Hünig and Buysch, for example, synthesized linear α,ω-dicarboxylic acids with up to 56 carbon atoms. They already pointed out the decreasing solubility of the products as a limiting factor for this synthetic pathway.29 Nevertheless, Wakselman could even achieve heptacontane1,70-dioic acid by a combination of these chain-extension reactions.30

2. SOURCES OF MONOMERS For the synthesis of long-chain aliphatic polyesters and polyamides, polycondensation monomers with appropriate amounts of methylene segments are required. Traditionally, these linear long-chain difunctional compounds are prepared via sequential build-up starting from shorter-chain building blocks. An interesting alternative to these multistep syntheses is a selective terminal functionalization of fatty acid derivatives, which already contain linear, long-chain crystallizable segments. A number of such straight-chain compounds actually also occur naturally. Aliphatic long-chain dicarboxylic acids as well as ωhydroxycarboxylic acids, for example, are building blocks of naturally occurring polyesters like cutin and suberin in cork (Scheme 1).11,12 Nevertheless, these natural resources normally are not used to recover these compounds, as their purity is often low and removal of other contaminants is extremely difficult. Hence, these polycondensation monomers have often been prepared via classical organic synthesis. 2.1. Classical organic synthesis

A range of laboratory-scale multistep synthesis schemes starting from smaller building blocks, often applying elaborated protection and deprotection steps, have been developed in the past. Syntheses of long-chain α,ω-functionalized compounds, for instance, can be achieved by coupling of two shorter-chain fragments; e.g., docosane-1,22-diol can be synthesized in three steps from commercially available 11bromoundecan-1-ol (Scheme 2).13 4598

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Scheme 2. Multistep Synthesis of Docosane-1,22-diol and Tetratetracontane-1,44-diol13

Scheme 3. Multistep Synthesisa of Pentacontane-1,50-dioic Acid (n = 2) and Tetranonacontahectane-1,194-dioic Acid (n = 4)17

Scheme 5. Chain Extension of 1,10-Dibromodecane via Grignard Coupling with Short-Chain α,ω-Ester Acid Chlorides23

inefficient as a source of monomers for purposes others than model polymers on a small scale. 2.2. Plant-oil fatty acids as a source of midchain monomers

These classical organic syntheses still remain tedious and inefficient. Purification and isolation of the long-chain compounds become more and more difficult as the chain length increases. Differences in physical properties between starting material, desired product, and possible side products vanish, eventually making a separation difficult. Moreover, although conversions of the single reaction steps might be convincing, the overall yield of the desired final product relative to the starting material utilized is often limited. Alternative approaches have emerged utilizing plant-oil-derived fatty acids,

a The first reaction sequence starts with x = 10, whereas n is the number of cycles applied.

Many further examples of such multistep approaches to α,ωlong-chain compounds have been reported.31−41 While they are elegant in providing even components with a very large number of carbon atoms precisely, they are very tedious and rather

Scheme 4. Electrolytic Synthesis of Diethyl Tetratriacontane-1,34-dioate (Two Cycles with x = 4)20

4599

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Scheme 6. Chain Extension of Decane-1,10-dioic Acid to Docosane-1,22-dioic Acid25

Scheme 7. Mid-Chain Length Monomers and Polyamides Based on Ricinoleic Acid

capacity of about 600 000 tons.44 Most of the oil is used for motor lubrication or in the cosmetics industry. However, a substantial part is allocated to the production of sebacic acid, affording ca. 100 000 tons of this compound. Large parts thereof are used in plasticizers, lubricants, or cosmetics. Yet, several thousand tons end up in the synthesis of nylon-6.10, which shows good chemical resistance and low water absorption characteristics and thus is often used in applications under wet conditions such as filter clothes or toothbrush filaments. The production capacity of the second castor-oilderived polyamide described above, nylon-11 (marketed by Arkema as Rilsan), is in the same order of magnitude. Due to its excellent physical, chemical, and mechanical properties, including high-temperature service, long-term aging, electrical resistance, and high chemical and hydrocarbon resistance, it is often used in demanding applications such as automotive fuel lines, pneumatic airbrake tubing, electrical cable sheathing, or flexible oil and gas pipes.

as these substrates already contain linear long-chain hydrocarbon segments. In this context, it is instructive to briefly review existing routes to shorter, medium-chain monomers from fatty acids. Some pathways to such polycondensation monomers (with numbers of carbon atoms ranging up to 13) from unsaturated fatty acids are well-established industrially. As the most prominent starting material, ricinoleic acid, the major component of castor oil, has been used for many years. 10Undecenoic acid, for example, is obtained via thermal rearrangement with chain cleavage.42 Further addition of hydrobromic acid and reaction with ammonia yields 11aminoundecanoic acid, the starting material for nylon-11. Likewise, sebacic acid, which is used for nylon-6.10 synthesis, can be generated by cleavage under strongly basic conditions (Scheme 7).43 Both of these routes require a hydroxy-substituted unsaturated fatty acid, of which ricinoleic acid is the only practically available example. Castor oil, consisting mainly of ricinoleic acid (85−95%), is currently produced with an annual 4600

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Scheme 9. Enzymatic ω-Oxidation of Fatty Acids

However, castor oil is significantly more costly (ca. double the price) of other plant oils like soybean, palm, or rapeseed oil. Moreover, in the transformations described above, only one side of the fatty acid chain with respect to the double bond is incorporated and stoichiometric amounts of less-valuable byproducts are formed. These arguments also apply to ozonolysis, which converts monounsaturated fatty acids to α,ω-diacids. In this way, several thousand tons of the mediumchain length diacids azelaic acid (C9) and brassylic acid (C13) are produced industrially by oxidative cleavage of oleic acid and erucic acid, respectively, affording pelargonic acid as a byproduct (Scheme 8).45,46

Scheme 10. Mechanism of Terminal Oxidation of Fatty Acids51

Scheme 8. Synthesis of Azelaic and Brassylic Acid by Ozonolysis of Oleic Acid and Erucic Acid, Respectively, Both Giving Pelargonic Acid as a Byproduct45

By comparison, a total of ca. 20 million tons of plant oil are used as a feedstock for chemical production annually.47 For the largest part, this oleochemistry deals with conversions of the fatty acids’ carboxyl group. In addition, as a recent development, more than 30 million tons are used for the production of “biodiesel”,48 although the benefit of this biofuel is questioned increasingly. To employ the potential of the linear long-chain hydrocarbon segments and also to utilize the feedstock most efficiently, a full incorporation of the entire fatty acid chain into linear long-chain α,ω-functionalized compounds is desirable. Biotechnological transformations provide a possible approach to this challenge.

enrichment of the α,ω-functionalized target compound is only possible by blocking these degradation pathways, either by deletion of genes involved in this metabolic oxidation53,54 or by manipulation of transport processes within the cell.55 As a trade-off, an additional, often costly carbon source like glucose is necessary to maintain the energy supply. Further optimization of the biocatalyst was achieved via overexpression of enzymes. Amplification of the Cyt-P450 monooxygenase (CYP) and NADPH:Cyt oxidoreductase (CPR) genes in Candida tropicalis involved in the ω-oxidation pathway leads to a 30% increase in productivity compared to the β-oxidation blocked wild type (Table 1, entries 1 and 2).56,57 The productivity of these biotechnological transformations strongly depends on a variety of different parameters (e.g., medium composition, availability of oxygen, pH value, temperature, and emulsification of hydrophobic substrates) and has to be optimized during bioprocess engineering. Key features of selected optimized fermentation approaches for the microbial transformation of long-chain dicarboxylic acids with different chain lengths are summarized in Table 1. Notably, substrates with shorter chain length can be converted with higher efficiency when compared to long-chain substrates (Table 1, entries 3−5). Moreover it could be demonstrated that the position of the double bond in unsaturated fatty acids is

2.3. Biotechnological ω-oxidation

Certain yeasts strains, e.g., Candida tropicalis, Candida maltosa, and Yarrowia lipolytica, are able to oxidize terminal aliphatic carbons to carboxylic acids. This ω-oxidation enables the conversion of fatty acids and their derivatives to long-chain dicarboxylic acids (Scheme 9).49−51 Linear aliphatic diacids with the same number of carbon atoms as the fatty acid starting material, that is, an even number usually in the range of 14−22, can be obtained. The first step of this biotechnological transformation is catalyzed by a hydroxylase complex and involves a terminal oxidation of the fatty acid to a primary alcohol (Scheme 10). In a second step the alcohol is oxidized by a fatty alcohol oxidase to the corresponding aldehyde, which is subsequently converted to the carboxyl group of the corresponding diacid.52 Given that the fatty acids as well as dicarboxylic acids can be metabolized further via the β-oxidation pathway to produce energy, an 4601

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

products are required for their application as polycondensation monomers, the latter point is especially important. Because of the classification of Candida tropicalis as a pathogenic microorganism,62 current research efforts are also focused on the utilization and optimization of nonpathogenic microorganisms such as Yarrowia lipolytica or Saccharomyces cerevisiae.51 Currently, several companies are active in the microbial production of medium- and long-chain dicarboxylic acids from fatty acids. This technology has been pioneered among others by Cognis,63,64 but most producing companies using this technology are located in China (e.g., Cathay Industrial Biotech and Shandong Hilead Biotechnology).

Table 1. Maximum Product Concentration (CMax) and Maximum Production Rate of Dicarboxylic Acid (RMax) in ωOxidation entry

strain

substrate

CMax

1

C. tropicalis (AR40) C. tropicalis (H5343) C. tropicalis (AR40) C. tropicalis (AR40) C. tropicalis (AR40) C. tropicalis (ATCC20962) C. tropicalis (ATCC20962)

95% methyl myristate (14:0) 95% methyl myristate (14:0) 95% methyl myristate (14:0) 95% methyl palmitate (16:0) 95% methyl stearate (18:0) oleic acid (18:1)

150 g L−1 in 92 h 115 g L−1 in 92 h 145 g L−1 in 118 h 94 g L−1 in 118 h 71 g L−1 in 118 h 18 g L−1 in 72 h 14 g L−1 in 72 h

2 3 4 5 6 7

erucic acid (22:1)

RMax 1.63 g L−1 1.25 g L−1 1.47 g L−1 0.80 g L−1 0.60 g L−1 0.25 g L−1 0.20 g L−1

ref h−1 h−1 h−1 h−1 h−1 h−1 h−1

56 56 56 56 56

2.4. Olefin metathesis 58

An entirely chemical-synthetic approach to linear long-chain α,ω-functionalized compounds is provided by olefin metathesis. In the last two decades major improvements have been achieved in metathesis catalyst performance, and olefin metathesis has evolved as a tool for oleochemistry.65−70 Metathesis products like unsaturated diacids, diesters, or mixed α,ω-functionalized compounds can be used for polycondensation reactions after hydrogenation of the carbon−carbon double bonds. 2.4.1. Self-metathesis reactions. An early example of selfmetathesis of unsaturated fatty esters into monounsaturated hydrocarbons and α,ω-diesters was reported by Van Dam et al. in 1974.71 By using WCl6/SnMe4 as a catalyst system, methyl oleate (and methyl elaidate) were converted to dimethyl octadec-9-ene-1,18-dioate and octadec-9-ene, reaching equilibrium conversion within a few hours (Scheme 12). Subsequently, several other classical in situ catalyst systems and also heterogeneous catalysts were found to convert unsaturated fatty acids and oil substrates.72,73 However, their performance in these reactions is limited due to their (partially) insufficient tolerance toward the substrates’ carboxylic acid or ester groups.67 This issue was advanced with the development of more functional group tolerant, defined metal alkylidene metathesis catalysts precursors.74,75 In particular, ruthenium alkylidenes, most prominently [(PCy3)2Cl2RuCHPh] (Grubbs first-generation catalyst) and [(PCy3)(η-C-C3H4N2Mes2)Cl2RuCHPh] (Grubbs second-generation catalyst), have been found to be precursors to very active catalysts. High productivities of up to several 105 turnovers have been reported for self-metathesis of methyl oleate, applying these catalyst precursors.76 Nevertheless, molar conversions are limited (at its best ∼50% in homogeneous solution), because such reactions typically are subjected to thermodynamic control. This limitation can be overcome by applying solvent-free self-metathesis of monounsaturated fatty acids. Under these bulk conditions, the diacid products formed during the reaction are not soluble in the reaction medium and precipitate from the mixture. This removal of a product shifts the equilibrium mixture. Thus, conversion of the starting material and yields of the diacid products increase. In this way even carbon-numbered, monounsaturated dicarboxylic acids

58

maintained (Table 1, entries 6 and 7). Thus, also unsaturated α,ω-functionalized diacids can be prepared.58 A related transformation of fatty acid substrates is ωhydroxylation, introducing a terminal hydroxy group at the unsubstituted hydrocarbon chain end. By applying modified yeast strains, 14-hydroxytetradecanoic acid can be obtained with yields of 174 g L−1 with 99% by crystallization from the reaction solvents), enabling utilization for polycondensation reactions.145 These long-chain α,ω-functionalized diesters can also be obtained in a one-pot procedure from different plant oils,146 including high oleic sunflower oil.147 Yields correlate with the oleate content of the starting material, but otherwise the catalyst performance appears not to vary dramatically between pure oleate (99%) starting material and technicalgrade methyl oleate (92.5%) or plant oil (triglyceride). Catalyst performance can be enhanced by utilization of [(dtbpx)Pd(OTf)2] as a defined catalyst precursor, which eliminates the need for using an excess of the diphosphine ligand.148 Thus, in a standard 1 L pressure reactor, 100 g batches of >99% pure dimethyl nonadecane-1,19-dioate (Figure 3) or dimethyl tricosane-1,23-dioate can be prepared routinely from high oleic sunflower oil methyl ester and technical-grade methyl erucate, respectively, with ca. 90% conversion and 90% selectivity with respect to the contained methyl oleate (Scheme 21).149 Utilizing long-chain α,ω-diesters as platform chemicals, the generation of further monomers, like diols or diamines, for the preparation of long-chain polycondensates becomes possible.145,148 Algae oils (extracted from the microalgae Phaeodactylum tricornutum) with their unique fatty acid spectrum are also suited as a feedstock.150 Isomerizing methoxycarbonylation of the crude algae oil multicomponent mixture, containing among others derivatives of palmitoleic acid (16:1), oleic acid (18:1), and eicosapentaenoic acid (20:5), yielded a mixture of dimethyl heptadecane-1,17-dioate and dimethyl nonadecane-1,19-dioate for polyester synthesis. 4606

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

this dormant species can interconvert into the productive catalytic cycle (left cycle in Scheme 22). Under pressure reactor conditions (90 °C, 20 bar of CO), all conceivable branched alkyls are formed in very small amounts. By the utilization of an adamantyl-substituted diphosphine ligand, which specifically destabilizes the transition states leading to branched products, an even higher selectivity for the linear α,ω-diester product of 96% (at 95% conversion) was achieved.154 Notably, multiple unsaturated analogues, commonly occurring in monounsaturated fatty acids from natural plant oils, can in principle be converted to the same α,ω-diacid esters (as formed as the main product from the monounsaturated fatty acid starting material) by a sequence of isomerizing carbonylation and catalytic hydrogenation (Scheme 23).155 The selectivity in isomerizing methoxycarbonylation of methyl linoleate to the desired unsaturated α,ω-diester is reduced considerably compared to monounsaturated substrates. Since the formation of Pd-allyl intermediates increases the energy barrier for the isomerization (and further) steps, the reaction becomes slower.152 This favors other reaction pathways, and besides the unsaturated α,ω-diester as the main product, among others, ketone- and methoxy-functionalized monoesters and the (branched) triester are formed.156 For comparison, a sequence of self-metathesis and subsequent double-bond hydrogenation would convert the multiple unsaturated fatty acids to additional α,ω-diacid esters of different chain lengths. This difference between isomerizing carbonylation and olefin metathesis can be related to fundamental characteristics of these reactions: isomerizing carbonylation is strictly kinetically controlled and yields a product not representing the thermodynamically favored outcome. On the other hand, in olefin metathesis often equilibria are obtained because there is no extreme kinetic preference for a particular product. Isomerizing alkoxycarbonylation is also emerging as a route to unsymmetrical α,ω-difunctionalized products.157 Considering a preparation of unsymmetric α,ω-diesters, a suppression of transesterification with the alcohol employed as a reactant (and solvent) is crucial. Under appropriate acid-free conditions, transesterification can be suppressed completely (200 °C at 9 Torr31). Thus, already in the initial reaction mixture, an accurate stoichiometric balance between the diol and the diacid (or the diacid derivative) is required. This obviously does not apply to an AB-type approach, applying ω-hydroxy acids (Scheme 28b).130,168 However, such unsymmetric monomers are less well accessible in general. An alternative route to long-chain aliphatic polyesters is ringopening polymerization (ROP) of cyclic monomers (Scheme 28c). As a chain-growth reaction ROP is not subjected to the aforementioned restrictions of step-growth polycondensations. However, besides ω-pentadecalactone, which can be directly derived from Angelica archangelica L. root oil,169 large ring

Figure 4. Dependence of standard Gibbs energy (ΔGop, ▽), enthalpy (ΔHop, ○), and entropy (ΔSop, ●) of lactones polymerization on their ring sizes (n). The temperature was dependent on the lactone and ranged from 350 to 430 K, for monomer and polymer liquid. Reprinted with permission from ref 170. Copyright 2009 John Wiley and Sons.

First investigations of metal-catalyzed anionic ROP of large ring lactones were reported by Endo and co-workers, applying alkali alkoxides.172 Using potassium alkoxides for ROP of pentadecalactone, molecular weights Mn up to 92 000 g mol−1 were possible (determined by gel permeation chromatography (GPC) vs polystyrene standards).173 Recently, Duchateau and co-workers reported the synthesis of high-molecular-weight (Mn > 150 000 g mol−1, determined by GPC vs polyethylene standards) poly(pentadecalactone), applying aluminum−salen complexes (Scheme 29).174−176 In the last few years, numerous examples for catalyzed ROP of various lactones have been reported, screening activity and molecular weights of the resulting polymers.177−184 Nakayama et al. reported neodymium tetrahydroborates, a very efficient catalyst, achieving high conversions already after 1 min of reaction time. 185 Also organic catalysts like 1,5,7triazabicyclo[4.4.0]dec-5-ene (TBD) have been explored, however yielding relatively low molecular weights.186,187 4610

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Scheme 29. Aluminum−Salen-Catalyzed Ring-Opening Polymerization of Pentadecalactone174

Scheme 30. ADMET Polymerization of Undec-10-en-1-yl Undec-10-enoate Followed by Exhaustive Hydrogenation83

Although ROP is generally designated as more controlled than polycondensation reactions, for polyesters from macrolactones generally broader molecular-weight distributions are observed compared to polyesters from lactones with small ring sizes. Studies on the ROP of ambettolide (as a 17-membered, unsaturated lactone) demonstrated that during the whole polymerization relevant amounts of cyclic oligomers are present, proving that there is no clear preference between ROP and transesterification.188 Therefore, a living character and consequently narrow molecular-weight distributions are hardly achievable in ROP of macrolactones. A further access to linear polyesters is provided by olefin metathesis polymerizations,189 namely, acyclic diene metathesis (ADMET) polymerization of ester-functionalized α,ω-diene monomers and ring-opening metathesis polymerization (ROMP) of unsaturated lactones, both yielding unsaturated polyesters. While only a few examples for ROMP have been reported,190−192 ADMET polymerization was established as an efficient method to yield long-chain model polyesters (Scheme 28d).193,194 As a polycondensation reaction (releasing ethylene as the byproduct), ADMET polymerization underlies the typical limitations of step-growth reactions. However, the formed ethylene can be efficiently removed from the reaction equilibrium, and molecular weights on the order of (several) 104 g mol−1 can be achieved. Upscaling of these reactions appears problematic, but they are useful for the preparation of model polycondensates in gram-scale quantities for studies of their physical properties. Exhaustive ADMET polymerization of undec-10-en-1-yl undec-10-enoate, followed by postpolymerization hydrogenation, yields a saturated polyester (Scheme 30), containing the same density of ester groups compared to polyester-20.20 (PE-20.20) generated by polycondensation of eicosane-1,20-dioic acid with eicosane-1,20-diol.83 However, this ADMET-generated polyester exhibits an irregular structure, where the ester groups are randomly oriented along the polymer chain (−O(CO)− vs −C(O)O−) and the distances between the ester groups differ (18, 19, or 20 carbon atoms). Compared to its regularly oriented PE-20.20 congener, random ester group orientation results in an observable reduction of the melting point. As a further example, a randomly oriented, ADMET-derived polyester corresponding to regular PE-19.19 was reported by Watson and Wagener.195 Such irregularities can be avoided by applying symmetric diene monomers (Scheme 31). Several examples of different long-chain unsaturated polyesters generated in this way have been reported, although hydrogenation of the remaining carbon−carbon double bonds was not always performed.97,110,196 Double bonds within the hydrocarbon polymer backbone have a drastic influence on polymer chain crystallization, resulting in reduced crystallinity and melting points (a Tm difference of 43 °C has been reported

Scheme 31. ADMET Polymerization of Symmetric Monomers to Unsaturated Polyesters41

for the saturated and unsaturated polyesters from Scheme 30).83 Saturated PE-38.23 and PE-44.23 containing very long hydrocarbon segments have been prepared by this approach.41 Yet, irregular structures of ADMET polymers also can be imparted by isomerization of the double bond as an undesired side reaction.88 As a consequence, chain lengths of the methylene segments between these ester groups vary. This effect has been quantified by Meier and co-workers by ADMET polymerization of dianhydro-D-glucityl diundec-10-enoate (applying different olefin metathesis catalysts) and subsequent degradation of the polymer via transesterification with methanol (Scheme 32).197 GC-MS analysis of the resulting diesters, which represent the repeat units of the previously prepared polyesters, clearly show that well-defined polymer architectures can be obtained by applying Grubbs first-generation catalyst, whereas ADMET polymerizations with Grubbs second-generation catalyst shows a temperature-dependent isomerization tendency, resulting in less-defined polymeric architectures (Figure 5). This irregularity in the polymers’ structure also affects thermal properties. In general, the melting points of the polyesters from methyl 10-undecanoate and isosorbide decrease with increasing degree of isomerization. Hence, the unsaturated polyester with the highest degree of isomerization exhibits a melting point of only 17 °C, whereas an analogue prepared by ADMET polymerization applying Grubbs firstgeneration catalyst melts at 56 °C. Nevertheless, an improved control over the polymer microstructure by applying second4611

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Scheme 32. Approach for the Quantification of Isomerization during ADMET Polymerizations197

Scheme 33. Catalytic Dehydrogenation Polymerization of Different α,ω-Diols199

show no evident limitations for a further increase of the carbon spacer of the diol component. For diols with fewer carbons, however, cyclization to the corresponding lactones is observed. Besides (organo-)metal-catalyzed polymerizations, enzymatic polyesterification has been studied intensively.203−205 The first reports on the enzymatic production of oligomeric esters from dicarboxylic acids and diols via A2 + B2 polycondensation were reported in the 1980s.206 Later on, higher-molecular-weight polyesters were also obtained from ω-hydroxy acids. Eventually both routes were also applied to linear long-chain monomers emploing different enzymes as catalysts.58,109,207 Significant differences in the behavior toward monomers of variable chain lengths were observed for different enzymes.208−211 For example, systematic studies using ω-hydroxy acids of various chain lengths showed that Candida antarctica Lipase B (Novozyme-435) was able to polymerize 10-hydroxydecanoic acid, 12-hydroxydodecanoic acid, and 16-hydroxyhexadecanoic acid, while Humicola insolens cutinase immobilized on Amberzyme oxiranes was just active toward ω-hydroxy acids with 12 and 16 carbons.210 Initial reports on polyester synthesis by enzymatic ringopening polymerization appeared in the 1990s.212,213 Since then, various cyclic compounds have been polymerized by enzymatic catalysis, targeting different applications.214−223 Among others, several unsubstituted lactones of different ring sizes, like pentadecalactone (PDL)214,224−226 and hexadecalactone (HDL),224,226,227 have been polymerized to yield the corresponding long-chain aliphatic polyesters with molecular weights up to 105 g mol−1 in solution as well as under bulk reaction conditions via reactive extrusion techniques.219,228 By contrast to metal-mediated ROP, where a decrease of reactivity with increasing ring size of the lactone is observed, the situation in the enzymatic ROP of macrolactones appears to be reversed.229 Enzymes show an unusually high reactivity in the ROP of macrolactones.224 Reactivity is no longer dominated by the ring strain, but rather by monomer recognition of the enzyme, which is enhanced for macrolactones as their hydrophobic nature and molecular shape resemble aliphatic fatty oilsthe lipase’s natural substrate. Nevertheless, restricted physiological conditions (temperature, pH value, etc.) under which enzymes work as well as difficulties in isolation and purification of the resulting polymers from the fermentation media still limit the application of enzyme catalysis.230 Especially the preparation of long-chain aliphatic polyesters in bulk is difficult due to the high melting points of the corresponding monomers.58 Yet, such enzymatic approaches broaden the scope of polymer syntheses. Especially polymers with additional reactive functional groups, which are unsuitable for classical high-temperature polycondensations with chemical catalysts, have become accessible. In this way, for example, long-chain unsaturated polyesters58 and long-chain

Figure 5. GC-MS traces of the transesterification products formed according to Scheme 32. Adapted with permission from ref 197. Copyright 2009 American Chemical Society.

generation ruthenium metathesis catalysts can be achieved by choosing appropriate reaction conditions (catalyst loading and temperature) and by addition of 1,4-benzoquinone,198 which is known to effectively suppress double-bond isomerization.93 Finally, an unconventional approach to aliphatic polyesters of short- and medium-chain length by in vacuo dehydrogenation polymerization of different α,ω-diols also appears attractive for long-chain aliphatic polyesters (Scheme 33).199 The commercially available Milstein catalyst,200−202 which is known to be highly effective for the conversion of alcohols to esters, can be used to generate polyesters. Because the gaseous hydrogen byproduct can be easily removed, driving the equilibrium toward the polymer product’s side, high-molecular-weight polyesters (Mn up to 145 000 g mol−1, determined by GPC vs polystyrene standards) can be obtained. Polymerizations of 1,6-hexanediol and 1,10-decanediol under reduced pressure using catalyst loadings of 0.2 mol % are highly effective and 4612

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Table 2. Melting Points of Different All-Aliphatic Linear Polyesters of the PE-X.Y and PE-X Typea

Other long-chain aliphatic polyesters: PE-25.25, 108 °C;150 PE-26.26, 114 °C;41 PE-30.30, 113 °C;167 PE-38.23, 109 °C;41 PE-38.26, 111 °C;255 PE-44.5, 101.5 °C;15 PE-44.23, 111 °C;41 PE-44.26, 110 °C.255 Polyesters based on mixed components: PE-17/19.17/19, 99 °C;150 PE-32/33/ 34.32/33/34, 109 °C.87 a

poly(butylene adipate) at Tm of ca. 55 °C, c.f. Table 2). This lower crystallinity also goes along with a decreasing stability toward hydrolytic degradation, which may be undesirable for a given application. These potential drawbacks of short- and midchain all-aliphatic polyesters can be overcome, however, by introduction of longer, crystallizable methylene segments. Although general trends such as the increase of the melting point with increasing hydrocarbon chain length of the diol or the diacid component become evident, gaps in the data, especially for long-chain monomers, and the great variation that is often found between the values given by different authors complicate the derivation of a fully comprehensive picture. Korshak and Vinogradova prepared a number of aliphatic polyesters based on eicosane-1,20-diol.264 They clearly demonstrated that the melting points of the polyesters obtained depend not only on the number of carbon atoms in the diacid monomer but also on whether this number is even or odd (Figure 6).

polyesters with epoxidized58,231,232 or hydroxy-functionalized231 repeat units can be prepared without applying tedious protection group strategies necessary for conventional chemical approaches. In this sense, these biocatalytic routes can complement the traditional polymerization approaches. 3.2. Typical properties of long-chain aliphatic polyesters

Today’s applications of polyesters are dominated by materials based on aromatic diacids, most prominently polycondensates of terephthalic acid with C2 to C4 linear diols. Applications of entirely aliphatic polyesters, in contrast, are limited to smallerscale specialties. One reason is certainly given by their low melting points, making thermoplastic processing problematic due to low crystallization temperatures (apart from very shortchain aliphatic polyesters like polylactic acid or polybutylene succinate).233−235 Also an undesired softening in applications, e.g., at elevated ambient temperature, could be problematic (e.g., poly(ε-caprolactone) melts at Tm of ca. 60 °C and 4613

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

number of methylene groups in a link but also on their distribution within the polymer chain (e.g., Tm(PE-6.8) = 65− 68 °C; Tm(PE-10.4) = 70−74 °C).234 By increasing the amount of long, crystallizable methylene segments within both the diacid and the diol components, aliphatic polyesters with melting points > 100 °C are obtained (c.f. Table 2). With further increasing chain lengths of the aliphatic repeat units, melting points should finally converge toward linear polyethylene. To quantify this effect over a broad range of different ester group densities within the polymer chain, classical polycondensation approaches would require the synthesis of numerous monomers, each one tedious in itself. To overcome this issue, linear long-spaced aliphatic model polymers with adjustable degrees of ester functionalization were synthesized by random ADMET copolymerization of undeca-1,10-diene and undec-10-en-1-yl undec-10-enoate, followed by exhaustive hydrogenation (Scheme 34).269 In this way, polyesters very close to linear polyethylene become accessible.

Figure 6. Melting points of polyesters-20.X with α,ω-diacid repeat units of different chain lengths (adopted from ref 264).

In general, polyesters derived from dicarboxylic acids containing even numbers of carbon atoms melt at higher temperatures than those derived from diacids containing odd numbers of carbon atoms. This observation was ascribed to the fact that the crystalline phase of an aliphatic polyester with an even number of carbon atoms between ester groups contains close layers of dipoles of opposite direction (effectively canceling the local polarization), whereas close carbonyl layers of identical direction are present in the crystal of polyesters derived from dicarboxylic acids having odd numbers of carbon atoms (Figure 7).265−267

Scheme 34. Synthesis of Long-Spaced Aliphatic Model Polyester via ADMET Copolymerization of Undeca-1,10diene and Undec-10-en-1-yl Undec-10-enoate and Subsequent Exhaustive Post-polymerization Hydrogenation269

Figure 7. Schematic illustration of the arrangement of the polar layers in aliphatic polyesters with even number of carbons between the ester groups (left) and odd number of carbons between ester groups (right), to account for the observed trends in the melting behavior. The arrows indicate the directions of polarizations.234

A related approach was reported by Duchateau and coworkers, applying ring-opening metathesis copolymerization (ROMP) of ambrettolide and cis-cyclooctene to synthesize related long-spaced polyesters (after hydrogenation of the remaining carbon−carbon double bonds, Scheme 35).270 Recently, a third method for the preparation of random polyesters was reported by Maeda et al., namely, crossmetathesis of 1,4-polybutadiene as an unsaturated hydrocarbon polymer with an unsaturated polyester, yielding copolymers by main chain exchange reactions.271 Double-bond hydrogenation yielded saturated compounds, designated as polyethylene/ polyester copolymers. Long-spaced aliphatic model polyesters from both ringopening and acyclic diene metathesis copolymerization approaches coincide well in their melting behavior (Figure 8).269,270 A linear increase of the melting points toward the melting temperature of defect-free, linear polyethylene is observed (Tm = 134 °C)272 if Tm is plotted against the mole fraction of ester groups (regarding the polymers as random

Parallel to the odd−even effect in the melting behavior, also variations in the crystalline structures are observed for these polyesters. While an orthorhombic unit cell is generally found for polyesters with odd numbers of carbon atoms in the (shortchain) diacid repeat unit, monoclinic structures are found for even-numbered congeners. Obviously, for these latter polyesters, distorted chain conformations (compared to the all-trans orthorhombic structure) increase the overall crystalline stability.268 With increasing numbers of methylene groups in the dicarboxylic acid, the extent of this odd−even effect decreases and the orthorhombic crystal structure becomes dominant for all long-chain polyesters (vide infra). However, the melting point of an aliphatic polyester depends not only on the total 4614

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Detailed studies on the crystalline properties of long-spaced polyesters showed that the incorporation of ester groups in the hydrocarbon chain crystals also affects the lamellar thickness.270 However, the melting-point depression compared to linear polyethylene is mainly related to the disturbing effect of the ester groups within the crystalline phases and only to a minor extent related to a reduction of the overall lamellar thickness. By comparison with regularly spaced long-chain polyesters from A2 + B2 polyesterification, the randomly spaced model copolyesters melt at slightly lower temperatures (Figure 8). This effect can be ascribed to the hindered formation of polar ester layers within the crystal lattices due to inhomogeneous spacing of the ester groups in such random copolyesters. In general, the solid-state structures of long-chain aliphatic polyesters are dominated by van der Waals interactions between adjacent stretched methylene sequences, that is, by their hydrocarbon nature. For linear, long-chain A2 + B2 polyesters (where both diacid and diol monomer compounds have long-chain character), generally orthorhombic, polyethylene-like crystal structures are found (Figure 9),145,148,269,270,274 because the ester groups (acting as structural

Scheme 35. Synthesis of Long-Spaced Aliphatic Model Polyester via Ring-Opening Copolymerization of Ambrettolide and cis-Cyclooctene, Followed by Exhaustive Hydrogenation270

copolymers consisting of methylene and ester units as disturbing units). This behavior is in accordance with the inclusion model described by Sanchez and Eby,273 which provides the relationship ⎛ ε 2σ ⎞ ⎟ Tm = Tm0⎜1 − − X E ΔHm0 ΔHm0l ⎠ ⎝

(1)

T0m

were Tm is the melting temperature of the copolyester, is the equilibrium melting temperature, ε is the energy penalty created by incorporation of ester groups into the crystal lattice, ΔH0m is the heat of fusion for linear polyethylene, XE is the molar fraction of ester units (−C(O)O−) in the copolyester, σ is the surface free energy of the crystal surface, and l is the lamellar thickness. The melting behavior of long-spaced polyesters containing a random distribution of ester groups along the polymer chain follows Tm = (133 − 1033 × XE) °C.

Figure 9. Wide-angle X-ray diffraction (WAXD) patterns of PE-23.23 poly[1,23-tricosadiyl-1,23-tricosanedioate] (red) and linear polyethylene (black).

Figure 8. Peak melting points (Tm) of random, long-spaced copolyesters from acyclic diene269 (■) and ring-opening metathesis copolymerizations270 (▲), as well as symmetrical, regularly spaced long-chain polyesters from A2 + B2 polyesterification (●) for comparison, vs number of ester groups per 1000 CH2 and mole fraction XE. 4615

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

the orthorhombic unit cell with dimensions comparable to polyethylene is observed.186,278 Dynamic mechanical analysis (DMA) on poly(pentadecalactone) with a number-average molecular weight of 6.5 × 104 g mol−1 (determined by GPC vs polystyrene standards) revealed a glass transition at −27 °C and a high storage modulus, attributed to high crystallinity as determined by DSC (64%, compared to linear polyethylene) and WAXD (54%).279 Mechanical analyses on poly(pentadecalactone) samples with different molecular weights exhibit a strong dependence of the mechanical properties on the molecular weight in the regime studied (Table 3).263 An incorporation of ε-caprolactone-based repeat units in random copolymers with ω-pentadecalactone did not impact crystallinities and lamella thicknesses significantly. However, the yield stress is lowered compared to either homopolymer. This was traced to a more facile defect propagation as a result of a less regular stacking of ester groups in the lamellae in the copolymer.280 Gross and co-workers observed a brittle-to-ductile transformation for low-molecular-weight samples with Mw values between 4.5 × 104 and 8.1 × 104 g mol−1 (Table 3, entries 1 and 2). Moreover, with elongations at break and stresses at break reaching asymptotic values of 650% and 16 MPa, respectively, these high-molecular-weight samples also exhibit mechanical properties comparable to those of high-density polyethylene. Similar trends were observed for poly(ωhydroxytetradecanoate) with its slightly shorter C14 repeat units.168 High-molecular-weight poly(pentadecalactone) was also melt-processed into fibers, which were further elongated to ∼10 times their original length.219 Preliminary tensile tests revealed fiber tensile strengths up to 0.74 GPa depending on the crystal orientation and thus already compare to common polyethylene fibers, which typically exhibit tensile strength between 1.0 and 3.0 GPa.281 Narine and co-workers recently reported tensile properties of PE-9, PE-13, and PE-18 over a range of molecular weights, generated from the AB polycondensation of the corresponding ω-hydroxy esters, as well as related molecular weights, tensile strength, and crystallinity to each other.260 Concerning the mechanical properties of linear long-chain A2 + B2 type polyesters, Cho and Lee investigated PE-30.30.167 However, an elongation of break of only 5% was found, indicating that the molecular weight of the material was likely insufficient. By contrast, tensile testing of PE-19.19 and PE23.23 on samples generated by injection molding show Young’s moduli around 400 MPa and elongations at break >600% (Table 4 and Figure 11).149 To probe a possible application in packaging, films of PE19.19 and PE-23.23 were generated using a twin-screw miniextruder. A sufficient combination of melt stability and solidification behavior was observed, and films with thicknesses of ca. 60 μm were obtained without breaking during extrusion (Figure 12, top). By electrospinning of PE-19.19 from solution, fibers with an average diameter of several μm were obtained (Figure 12, bottom). Additionally, both long-chain polyesters were melt-compounded with talcum as a well-established inorganic filler of engineering thermoplastics, especially in automotive applications.282,283 The results clearly display an improvement in the mechanical properties, increasing the storage moduli, e.g., 2-fold with 25 wt % of talcum filler compared to the pure bulk polyesters. All-aliphatic thermoplastic polyester elastomers (TPEs) can be generated by copolymerization of dimethyl tricosane-1,23-

defects along the crystalline polymer chains) have no fundamental influence on the crystallization of hydrocarbon segments in the all-trans zigzag conformation. Typically, narrow melting transitions and distinct melting points, together with crystallinities up to 80%, are observed (compared to fully extended chain crystals of linear polyethylene as a reference for 100% crystalline material).41,145,148 Consequently, these novel materials are frequently discussed as “polyethylene mimics” or “polyethylene-like”.174,250,275,276 By means of solid-state NMR and small-angle X-ray scattering (SAXS) data, Schmidt-Rohr and co-workers elucidated the solid-state structure of PE-22.4 in detail (Figure 10).277

Figure 10. Solid-state structure of the aliphatic polyester PE-22.4 according to detailed NMR and SAXS studies. Reprinted with permission from ref 277. Copyright 2007 American Chemical Society.

Their results show that the ester functionalities exert a detectable influence on the crystalline chain arrangements as well as the crystal thickness (while still allowing for crystallization in a polyethylene-like crystal structure). In PE22.4 typically three diester layers can be found in the crystal: two of them at the surface and one in the center (Figure 10). Related observations were made for PE-22.5 and PE-44.5.15,274 In contrast to their branched analogues PE-22.5-Prop and PE44.5-Prop (Scheme 36), which reject the propyl side chains outside the crystalline phase, these two linear examples also feature an inclusion of polar ester groups into the crystal lattice. Also for AB-type polyesters like ROP-generated poly(pentadecalactone) (and its copolymers with ε-caprolactone), Scheme 36. Chemical Structures of the Polyesters Investigated by Penelle and Co-workers274

4616

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Table 3. Young’s Modulus, Stress at Yielding, Strain at Yielding, Elongation at Break, and Stress at Break for Poly(pentadecalactone) Samples of Different Molecular Weights263

a

entry

Mwa [g mol−1]

1 2 3 4 5

4.5 × 104 8.1 × 104 18.9 × 104 28.0 × 104 48.1 × 104

Young’s modulus [MPa] 690 620 450 290 390

± ± ± ± ±

stress at yield [MPa]

40 50 20 30 10

24.1 17.2 13.3 18.2

± ± ± ±

2.0 0.3 1.3 0.3

strain at yield [%] 12.0 13.0 20.6 17.4

± ± ± ±

1.0 1.0 1.3 0.4

stress at break [MPa] 20.4 17.7 16.2 13.7 17.3

± ± ± ± ±

3.2 1.3 0.5 1.0 0.9

(21.3 (42.0 (60.8 (55.0 (58.8

± ± ± ± ±

3.3)b 3.0)b 2.0)b 4.0)b 3.1)b

elongation at break [%] 4.5 ± 0.8 137 ± 25 650 ± 28 703 ± 72 580 ± 30

Determined by GPC in chloroform against polystyrene standards. bTrue stress at break as calculated after cross-sectional area correction.

Table 4. Young’s Modulus, Stress at Yielding, Strain at Yielding, Elongation at Break, and Stress at Break for PE19.19 and PE-23.23149

[g mol ]

Young’s modulus [MPa]

stress at yield [MPa]

strain at yield [%]

stress at break [MPa]

elongation at break [%]

3.0 × 104

408

15.3

18.3

15.9

619

3.9 × 10

436

16.7

19.7

17.0

678

Mna −1

PE19.19 PE23.23

4

a Determined by GPC at 160 °C in 1,2,4-trichlorobenzene vs polyethylene standards.

Figure 12. (Top) Melt-extruded films of PE-19.19 (left) and PE-23.23 (right). (Bottom) Scanning electron microscopy (SEM) image of fibers of PE-19.19 prepared by electrospinning. Reprinted with permission from ref 149. Copyright 2014 The Royal Society of Chemistry.

Figure 11. Stress−strain curves of PE-19.19. Reprinted with permission from ref 149. Copyright 2014 The Royal Society of Chemistry.

caprolactone) (which is utilized, e.g., for drug-delivery applications), long-chain polyesters did not show relevant hydrolytic or enzymatic degradation, which was related to the increased crystallinity and hydrophobicity, hindering water from penetration into the materials to hydrolyze ester groups. Hydrolytic degradation experiments of poly(pentadecalactone) fibers in phosphate-buffered saline (PBS) solution (pH = 7.4) displayed no changes in molecular weight and crystallinity over a period of two years.

dioate with tricosane-1,23-diol as long-chain aliphatic hard segments and short-chain polyethers like poly(tetramethylene glycol) (PTMG, with Mn of 1000 and 2000 g mol−1) or poly(trimethylene glycol) (PPDO, with Mn of 1000 and 2000 g mol−1) as soft segments.284 At an appropriate composition, e.g., 65 wt % of polyether phase, these polymers (Mn = 6 × 104 g mol−1, determined by GPC vs polystyrene standards) display elastomeric behavior with a high recovery (Figure 13, left). The thermal behavior is dominated by the aliphatic hard phase (Figure 13, right) at this composition, and melting points are well above 80 °C. This is the most notable difference toward shorter-chained analogous C12, melting considerably lower by ΔTm ≈ 20 °C. The biocompatibility and nontoxicity of long-chain polyesters like poly(pentadecalactone), poly(hexadecalactone), and unsaturated congeners were confirmed by Heise and coworkers.217 In contrast to short-chain polyesters like poly(ε-

3.3. Synthesis of long-chain aliphatic polyamides

Although preparation from various reactive derivatives is possible, aliphatic polyamides are most commonly prepared directly from dicarboxylic acids and diamines, ω-amino acids, or (short-ring) lactams (Scheme 37). On a commercial scale, most aliphatic polyamides are synthesized by melt-phase polycondensation processes. Typical molecular weights of commercial materials are on the order of Mn = 104 g mol−1.164,285 By 4617

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Figure 13. (Left) Stress−strain curves from cyclic tensile testing with a constant strain of 100% for TPEs based on dimethyl tricosane-1,23-dioate and a 1/1 molar ratio of tricosane-1,23-diol and PTMG (Mn = 2000 g mol−1, red), and a 1/1 molar ratio of tricosane-1,23-diol and PPDO (Mn = 2000 g mol−1, green). (Right) Differential scanning calorimetry (DSC) melting traces of TPEs based on dimethyl tricosane-1,23-dioate and different molar ratios of tricosane-1,23-diol and PTMG (Mn = 2000 g mol−1) of 3/1 (black), 2/1 (red), 1/1 (green), and 0/1 (blue).284

Scheme 37. Polyamide Synthesis from (a) A2 + B2 Polycondensation of a Diamine (CX) and a Diacid (CY) Yielding Polyamides of the PA-X.Y Type and (b) AB Polycondensation of an ω-Amino Acid, Which Together with (c) ROP of a Lactame Yields Polyamides of the PA-X Type

symmetrical α,ω-dienes containing amide functionalities using Grubbs- and Hoveyda−Grubbs-type second-generation catalysts (Scheme 38), yielding molecular weights Mn on the order

comparison to polyester synthesis, polyamidation can often be carried out without adding any catalyst. However, different kinds of catalysts, such as strong acids or metal oxides, also have been reported.285 A nontraditional approach for polyamide synthesis was presented by Zeng and Guan in 2011286 and subsequently also by Milstein and colleagues.287 Both groups independently reported the preparation of various polyamides via direct polyamidation by catalytic dehydrogenation of different nonactivated diols and diamines, applying the commercially available Milstein catalyst (cf. Scheme 33 for the related approach for the generation of linear polyesters). The long-chain aliphatic polyamides reported to date were predominately formed by melt A2 + B2 polycondensations at high temperatures of carboxylate/ammonium salts or by the reaction of the respective long-chain diacid derivatives with diamines,148,288−293 or by AB polycondensation of ω-amino acids or esters.87,294,295 Although ROP of strained cyclic amides (such as ε-caprolactam) is well-established, the polymerization of other lactams is less studied. To our knowledge ROP of macrocyclic lactams larger than ω-laurolactam (C12)296,297 to yield linear, long-chain aliphatic polyamides has not been reported so far. Moreover, unsaturated long-chain polyamides PA-X.20 (X = 2, 4, 6, 8) were prepared by ADMET polymerization of

Scheme 38. Synthesis of Unsaturated PA-X.20 (X = 2, 4, 6, 8) by ADMET Polymerization298

of 104 g mol−1 (as determind by GPC vs poly(methyl methacrylate) (PMMA) standards).298 However, hydrogenation of the remaining double bonds was not reported for these compounds so far. 3.4. Typical properties of long-chain aliphatic polyamides

Linear aliphatic polyamides find widespread use as fibers and engineering thermoplastics. In contrast to polyesters, the crystalline structures of polyamides are dominated by 4618

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Table 5. Melting Points of Different All-Aliphatic Linear Polyamides of the PA-X.Y and PA-X Typea

Other long-chain aliphatic polyamides: PA-2.34, 190 °C;291 PA-4.32, 189 °C;291 PA-6.34, 177 °C;291,331 PA-8.34, 174 °C;291 PA-10.34, 169 °C;291 PA-12.34, 166 °C.291 a

intermolecular hydrogen bonds between amide units, while van der Waals interactions between hydrocarbon segments play a minor role. Consequently, melting and crystallization temperatures are significantly enhanced for polyamides (c.f. Table 5).299 As an example, poly(ε-caprolactam) exhibits a melting point of ∼223 °C, whereas the melting temperature of the corresponding poly(ε-caprolactone) is only ∼60 °C. The transition in the physical properties is illustrated by polyesteramides as intermediates to homopolymers, generated by random ring-opening copolymerization of lactam/lactone mixtures.300−302 With decreasing amide group content in the polymer chains (going in hand with an overall reduction of the hydrogen-bond density), melting points generally decrease for linear polyamides. Related to aliphatic polyesters, variations in the melting points between even-numbered and odd-numbered polyamides are observed (Figure 14). This phenomenon can account for the different overall symmetry and the resulting different possibilities for hydrogenbond formation. In PA-7, for example, assuming an all-trans

Figure 14. Melting points of n-nylons with n as a function of chain length displaying pronounced odd−even effects.332

conformation, more hydrogen bonds can be formed than in PA-6 (Figure 15a and b).338 Analogously, the higher melting 4619

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Figure 15. Effects of all-trans conformation on hydrogen-bond formation as a model to explain odd−even effects in the melting behavior.338

points of aliphatic polyamides based on even-carbon-numbered diacid and diamine monomers in comparison to compounds based on odd-numbered compounds can be related to conformational effects (c.f. Figure 15c−f). Note that this picture is a simplified one. The assumption of all-trans conformations (related to long-chain polyesters and polyethylene) in crystalline polyamide chains is not fulfilled, as the vast majority of amide groups is actually found to be arranged in hydrogen bonds (up to 100%).339−341 To generate these high degrees of hydrogen bonding, slightly twisted chain arrangements are adopted, if all-trans conformations hinder hydrogen-bond formation to a relevant extent. Accordingly, the observed odd−even oscillations are a direct consequence of different crystal modifications.332 Hence, polyamide chains crystallize from the melt such that as many hydrogen bonds as possible form (also in the melt significant numbers of hydrogen bonds remain present). Early on, Kinoshita reported an approach to classify crystalline phases of polyamides containing odd, even, or mixed alternating numbers of methylene units, described as α-, β-, or γ-forms.342 Polyamides of the PA-X.Ytype (X = number of carbon atoms in the diamine compound, Y = number of carbon atoms in the diacid compound), where both the diamine and the diacid component contribute an even number of main-chain carbon atoms, mainly crystallize in the αform, where stacking of chain-folded, hydrogen-bonded sheets occurs in a progressive way (the angles between the amides plains and direction of extended polymer chains is Tm (polyesters) > Tm (polycarbonates) as a result of the decreased polarity of the carbonyl moieties with their increasing electron-withdrawing substituents in this series.395 Conformational reasons account for the drastically reduced melting points of polyacetals compared to polycarbonates. As the bond conformations in carbonyl functionalities (like ketones, esters, and carbonates) agree well with the all-trans zigzag conformation of crystalline hydrocarbon chains, orthorhombic crystal structures generally are found for regular and random long-chain polycarbonates, polyesters, and polyketones.195,393,395,396 However, this polyethylene-related crystal structure is not observed for long-chain polyacetals. For regular PAc-18, PAc-19, and PAc-23 and most random ADMET-derived polyacetals, WAXD patterns with various reflexes and shifting intensities were found (Figure 30). A transition to the orthorhombic crystal structure is observed only for polyacetals with less than 20 acetal groups per 1000 methylene units. This was related to the preferred gauche conformation in acetal units (due to the anomeric effect), already hindering polymer chains with intermediate degrees of acetal functionalization (referring to the regime of generally low functional group densities considered) from uniform crystallization in the all-trans conformation.393 According to the Sanchez−Eby inclusion model, Tm = (133−1033 × XC) °C (for polycarbonates) and Tm = (133−1412 × XAc) °C (for polyacetals) were found (with XC and XAc as the mole fraction of carbonates and acetals, respectively). Studies of the hydrolytic degradability for mid- and longchain polyacetals were performed in acidic media in order to

bottom). While condensation of linear diols with dimethyl carbonate yields biscarbonates, which can be isolated, purified, and then converted to the corresponding polycarbonates,394 also the direct polycondensation in a one-pot reaction is possible.393 Comparable to linear aliphatic polyesters, melting points generally increase for polyacetals and polycarbonates with increasing hydrocarbon chain lengths originating from the monomer diol component (Figure 28). Odd−even effects are

Figure 28. Melting points of mid- and long-chain polyacetals PAc-X (blue) and polycarbonates PC-X (red). Numerical values for polyacetals: PAc-5 (38 °C), PAc-6 (28 °C), PAc-7 (46 °C), PAc-8 (47 °C), PAc-9 (55 °C), PAc-10 (59 °C), PAc-11 (57 °C), PAc-12 (64/68 °C), PAc-18 (82 °C), PAc-19 (81/83 °C), PAc-23 (87/88 °C).391−393 Numerical values for polycarbonates: PC-5 (40 °C), PC-6 (54 °C), PC-7 (45 °C), PC-8 (55 °C), PC-9 (53 °C), PC-10 (59 °C), PC-12 (68 °C), PC-18 (89 °C), PC-19 (89 °C), PC-23 (97 °C).393,394

observed for both polyacetals and polycarbonates containing short hydrocarbon segments. While for polyacetals evennumbered compounds melt at lower temperatures than their odd-numbered congeners with related aliphatic spacer lengths, for polycarbonates higher melting points are found for the even-numbered compounds. The highest melting points have been reported so far for PAc-23 (88 °C) and PC-23 (97 °C) based on tricosane-1,23-diol, remaining still considerably below the Tm of linear polyethylene. To fill the gap between these long-chain compounds and polyethylene, random ADMET 4629

DOI: 10.1021/acs.chemrev.5b00705 Chem. Rev. 2016, 116, 4597−4641

Chemical Reviews

Review

Scheme 45. Synthesis of Random Polyacetals (Left) And Random, Unsaturated Polycarbonates (Right) by ADMET Copolymerizations and Post-polymerization Hydrogenation393

enable an observation and comparison of degradation rates in relative short-term experiments (weeks to months). The incorporation of longer aliphatic chain units between the acetal functions, going in hand with an enhanced hydrophobicity, increases the degradation time. While for PAc-12 (in the form of polymer pellets) a rapid degradation in a THF/conc. aq. HCl = 9/1 mixture with a weight loss of 50% occurred within 15 min, for long-chain PAc-19 and PAc-23 degradation times were extended to >100 h (at 40 °C). In aqueous systems (without further addition of organic solvents) a relevant degradation occurred only in strongly acidic and basic media. In 3 M aq. HCl and concentrated aq. HCl, weight losses