Oxidized Dihydrocarvone as a Renewable Multifunctional Monomer

Biomacromolecules 2009, 10, 2003–2008

2003

Oxidized Dihydrocarvone as a Renewable Multifunctional Monomer for the Synthesis of Shape Memory Polyesters Jennifer R. Lowe, William B. Tolman,* and Marc A. Hillmyer* Department of Chemistry, University of Minnesota, 207 Pleasant Street SE, Minneapolis, Minnesota 55455-0431 Received April 24, 2009; Revised Manuscript Received May 19, 2009

The natural product dihydrocarvone, found in caraway oil, was oxidized to an epoxylactone on a multigram scale. The resulting epoxylactone was used as a multifunctional monomer and cross-linker in ring-opening polymerizations. Homopolymerization using diethylzinc and tin(II) 2-ethylhexanoate gave only low molecular weight oligomers (apparent Mn less than 2.5 kg/mol). Copolymerizations of ε-caprolactone and 0.3 to 50% of the epoxylactone gave flexible cross-linked materials on a multigram scale in a one-step synthesis. The gel fraction of these copolymers was determined. These copolymers showed near perfect shape memory properties even after repeated bending.

Introduction The preparation of polymers from renewable resources is a burgeoning topic, with the desire to reduce dependency on fossil fuels driving interest in alternatives to traditional petroleum feedstocks.1-4 In addition, many renewable resource polymers are degradable under compost conditions, thus alleviating waste disposal problems inherent to petroleum-based plastics. Yet applications of renewable polymers (e.g., polylactide5) are not as widespread as those of traditional oil-derived materials, in part because of limitations in the monomer pool and the derived polymer property profiles. Thus, the development of new renewable monomers for the synthesis of sustainable polymeric materials with useful properties is an important research area. Motivated by our previous work on the polymerization of menthide,6,7 a seven-membered lactone derived from menthol, we have turned our attention to dihydrocarvone, which is found as a component of caraway oil.8 It can be produced by the hydrogenation of carvone, a component of caraway and spearmint oils, or by the oxidation of limonene, found in citrus peels (Scheme 1).2 Both carvone and limonene are produced on the scale of tens of thousands of tons annually.9 Because dihydrocarvone contains both cyclic ketone and olefinic functional groups, it has been used to test the selectivity of oxidation systems toward either epoxidation or Baeyer-Villiger oxidation, although the resulting products have not been widely studied and/or characterized fully.10-13 The presence of two different functionalities enables the potential use of dihydrocarvone as a monomer (or monomer precursor) for the generation of branched or cross-linked polymers. An intriguing application of cross-linked polymers is their use as shape memory devices, wherein the material can be deformed upon heating and then retains its deformed shape once cooled; upon reheating, it can then revert to its original shape. This behavior is similar to that seen in shape memory metal alloys, but such alloys are generally limited to recovery deformations of less than 8%.14 Polymeric materials can not only be produced inexpensively but also have greater recovery after deformation of over 100%, especially if the polymers are * To whom correspondence should be addressed. E-mail: hillmyer@ umn.edu (M.A.H.); [email protected] (W.B.T.).

Scheme 1

chemically rather than physically cross-linked.15-17 Important tunable properties of shape memory polymers include tensile strength and impact resistance, as well as the softening temperature needed to cause a reversion back to the original shape (typically, the melting temperature of the polymer crystallites in a semicrystalline cross-linked network). For example, a biodegradable shape memory polymer that has a softening temperature approximately that of body temperature could be used as a self-tightening suture or as a shape-changing stent that allows for easier implantation.18-20 Low levels of creep from the deformed shape back to its original shape while at the usage temperature (below Tm) is an important attribute of shape memory polymers. Shape memory polymers have been produced from biodegradable ε-caprolactone (CL)16,17,19,21 in physically cross-linked multiblock copolymers incorporating urethanes18,22-24 or in homopolymers with chain ends modified to chemically crosslink with acrylates or isocyanates in subsequent steps.16,17,21,25 High loadings of cross-linking agents are commonly required. Herein, we describe the use of renewable dihydrocarvone to produce 7-methyl-4-(2-methyloxiran-2-yl)oxepan-2-one (1) on scales sufficient to examine its utility in the preparation of chemically cross-linked statistical copolymers with CL in a facile

10.1021/bm900471a CCC: $40.75  2009 American Chemical Society Published on Web 06/08/2009

2004

Biomacromolecules, Vol. 10, No. 7, 2009

one-step process. We find that only small amounts (less than 1 mol %) of 1 are needed to effect this cross-linking reaction at moderate temperatures. The resulting materials show impressive shape memory properties with essentially complete strain recovery and are able to maintain some mechanical strength as solvent-swollen gels expanded to over ten times their original size.

Experimental Section Materials and Methods. (+)-Dihydrocarvone (98%), diethyl zinc (1.0 M in hexanes), tin(II) 2-ethylhexanoate (98%), diethylaluminum ethoxide (1.6 M in hexanes), and benzyl alcohol (98%) were used as received from Aldrich. m-Chloroperbenzoic acid (m-CPBA, 70%) was used as received from Fluka. ε-Caprolactone (CL, 97%, Aldrich) and δ-valerolactone (technical grade, Aldrich) were dried over CaH2, distilled, and stored under nitrogen. All polymerizations were carried out in a nitrogen-atmosphere drybox unless otherwise specified. Toluene used in polymerizations was purified using an in-house solvent purification system with an alumina column and copper catalyst and was stored over sodium. Dichloromethane used in polymerizations was dried on MBraun solvent columns. NMR spectroscopy was performed on a Varian Inova VI-500 spectrometer with CDCl3 as the solvent and TMS as an internal standard. High resolution ESI-TOF mass spectrometry was performed on a Bruker BioTOF II using reflectron mode and an internal PEG standard. Size exclusion chromatography was carried out in CHCl3 using an Agilent 1100 chromatograph containing a Varian PLgel 5 guard column followed by a series of three Varian PLgel Mixed C columns. A Hewlett-Packard 1047A refractive index detector was used. All samples were run at 35 °C and compared to polystyrene standards. Differential scanning calorimetry was performed on a Texas Instrument QA1000 DSC (calibrated with an indium standard) using hermetically sealed aluminum pans and sample sizes of about 3 mg. The heating cycle consisted of equilibration at -80 °C, heating at 10 °C/min to 100 °C, cooling to -80 at 10 °C/min, then heating again to 100 at 10 °C/min. Glass transition, melting temperatures, and ∆Hm were measured on the second heating cycle. 7-Methyl-4-(2-methyloxiran-2-yl)oxepan-2-one (1). A round-bottom flask was charged with (+)-dihydrocarvone (9.95 mL, 60.6 mmol), CH2Cl2 (100 mL), and sodium acetate (15.12 g, 185 mmol). The flask was cooled to 0 °C in an ice bath, after which 70% m-CPBA (46.04 g, 187 mmol) was added in three portions over 90 min. The mixture was then allowed to warm to room temperature and stirred for 2 d, at which time the mixture was filtered to remove the solid portions. The yellow filtrate was chilled in dry ice, the resulting solid was removed by filtration and was washed with 40 mL of cold CH2Cl2, and the filtrate was concentrated under vacuum until solid began to precipitate. This filtration and chilling procedure was repeated twice. When no additional solid precipitated upon cooling in dry ice, the yellow solution (approximately 50 mL) was then washed 10 times with 50 mL of saturated sodium bicarbonate solution and two times with 50 mL of sodium chloride solution. The aqueous layers were back extracted at each step with 20 mL of CH2Cl2. The organic layers were combined, dried over magnesium sulfate, and then concentrated under reduced pressure to give a viscous, yellow oil. By 1H and 13C NMR spectroscopy (Figure S1), this oil contained two regioisomeric epoxylactones, 1 and 2 (Scheme 1) in approximately a 3:1 ratio. The oil was then vacuum distilled at 93 °C (approximately 100 mtorr) to give the product as a colorless, transparent oil (2.73 g, 24.5%) containing 90% 1 and the 10% 2. This 9:1 mixture was used in all polymerizations and is referred to by its major component. Density: 1.16 g/cm3. Data for 1: 1H NMR (CDCl3, 500 MHz) 4.43 (m, 2H), 2.95 (m, 1H), 2.79 (d, 2H), 2.70-2.55 (m, 7H), 2.07 (m, 1H), 1.99-1.90 (m, 2H), 1.75-1.46 (m, 5H), 1.36 (d, 6H), 1.25 (s, 6H) ppm; 13C NMR (CDCl3, 125 MHz) 174.0 (OCOCH2), 77.1/76.1 (CH3CHO), 59.6/59.3 (CH3CO), 54.2/54.1 (CCH2O), 41.6/41.3 (CH2CHCH2), 38.0/37.5 (COCH2CH), 35.5 (OCHCH2), 31.3/30.7 (CH2CHCH2CH2), 22.7 (CHCH3), 17.4/17.2

Lowe et al. (CCH3) ppm. ESI-TOF MS: 207.1000 (M + Na+); theoretical mass, 207.0997. NMR data for 2 is provided in the Supporting Information. Homopolymerizations of 1. (a) SnOct2. In a typical polymerization, a 15 mL pressure vessel was charged with 1 (159 mg, 0.863 mmol) and toluene (300 µL). SnOct2 (14.2 µL, 4.38 × 10-2 mmol) was injected into the pressure vessel, which was capped and immersed in a 60 °C oil bath. The reaction mixture stopped stirring within a few hours. After 72 h, the pressure vessel was opened to air, was allowed to cool to room temperature, and 1 mL of CH2Cl2 was added to dissolve any soluble fractions. The CH2Cl2 soluble fraction as well as the remaining solid were precipitated into 7 mL of pentane and allowed to stir overnight. The pentane was decanted off and the solid was dried under vacuum overnight at room temperature to give an insoluble hard yellow solid (101 mg, 63.7%). (b) ZnEt2. In a typical polymerization, a 15 mL pressure vessel was charged with 1 (152 mg, 0.826 mmol), toluene (300 µL), and ZnEt2 (44 µL, 4.4 × 10-2 mmol). The vessel was capped and immersed in a 60 °C oil bath. The reaction turned vivid yellow and stopped stirring within a few hours. After 70 h, the pressure vessel was opened to air and allowed to cool, and the product isolated as above to give a soluble, transparent, yellow solid (128 mg, 84.0%). Copolymerizations of 1 with CL. (a) SnOct2. In a typical polymerization, a 15 mL pressure vessel was charged with 1 (146 mg, 0.793 mmol) and CL (1.316 g, 11.53 mmol). The pressure vessel was then opened to air and SnOct2 (0.9 µL, 2.8 × 10-3 mmol) was injected. The vessel was resealed and heated to 120 °C in an oil bath for 23 h. After cooling to room temperature, the vessel was opened and the reaction was quenched with 0.15 M aqueous HCl (50 µL). The crude product was dissolved in CH2Cl2 (∼2 mL) and precipitated into 40 mL of methanol. The resulting polymer was collected and dried under vacuum, giving an opaque off-white solid (1.35 g, 92.3%). (b) ZnEt2 (Small Scale; Solution Method). In a typical polymerization, a 15 mL pressure vessel was charged with 1 (75.4 mg, 0.41 mmol), CL (45.0 µL, 0.41 mmol), and toluene (250 µL). ZnEt2 (46 µL, 4.6 × 10-2 mmol) was injected into the pressure vessel, which was capped and immersed in a 60 °C oil bath. The reaction mixture stopped stirring within a few hours. After 72 h, the pressure vessel was opened to air and 1 mL of CH2Cl2 was added to dissolve any soluble fractions. The CH2Cl2 soluble fraction as well as the remaining solid were precipitated into 10 mL of pentane and allowed to stir overnight. The pentane was decanted off and the solid was dried in vacuo overnight to give a hard, insoluble yellow solid (115 mg, 94.4%) that could not be dissolved for NMR spectroscopy or SEC analysis. (c) ZnEt2 (Large Scale; Bulk Method). In a typical polymerization, 1 (3.69 g, 0.0200 mol), CL (8.05 mL, 0.0726 mmol), and ZnEt2 (2.0 mL, 4.6 × 10-5 mol) were premixed in a scintillation vial at room temperature for 30 min. The hexanes from the ZnEt2 solution were removed under vacuum and the remaining solution was poured into a PTFE-coated aluminum pan and placed inside a desiccator. The desiccator was placed under reduced pressure (approximately 1 Torr) and then heated to 60 °C in a sand bath. After 72 h, the desiccator was opened to air and the polymer disk (6 cm diameter, ∼3.5 mm thickness) was allowed to cool before it was removed from the aluminum pan (11.5 g, 96.1%). The resulting disk was an opaque, hard solid that was insoluble and was characterized by solvent extraction and in a shape memory angle test. Copolymer Property Evaluation. (a) SolVent Extraction and Gel Swelling. In a typical experiment, a rectangle of 10 × 8 × 3.5 mm3 of dried polymer was weighed and its dimensions measured with digital calipers. The dried polymer was placed in a 30 mL jar and CHCl3 (14.5 mL) was added. After 48 h, the swollen polymer gel was removed and its dimensions measured with digital calipers. The swollen gel was also weighed to determine the amount of CHCl3 absorbed. The polymer gel was then dried under reduced pressure to determine the weight fraction of the original polymer that was part of the cross-linked network.

Synthesis of Shape Memory Polyesters

Biomacromolecules, Vol. 10, No. 7, 2009

2005

Scheme 2

(b) Water Uptake. In a typical experiment, the tare weight of a 2 dram vial was recorded. Dried polymer (40.0 mg) was added to the vial, followed by 15 mL of distilled water. After 36 h, the polymer solid was removed, the exterior dabbed dry, and the sample weighed to determine water uptake. The polymer solid was then dried under reduced pressure to determine the weight lost during the experiment. (c) Shape Memory Effect Angle Test. The experiment was based upon the procedure used by Lin, et al.26 A 40 × 4 × 3 mm3 sample bar cut from cross-linked P(1-co-CL) copolymer containing 2.1% 1 was heated in a 52 °C water bath for 5 min. The sample was bent around a 6 mm diameter glass stir rod to a 60° angle. The sample was removed from the hot water bath and plunged into an ice bath for 90 s while still bent. The internal angle of the bent sample was measured using a protractor with 1° divisions. The sample was left at room temperature for 5 min before the internal angle of the sample was measured again. The sample was placed in a 52 °C water bath for 5 min during which time the sample flattened. The sample was removed from the hot water bath and plunged into an ice bath for 90 s. The internal angle of the sample was measured a final time. This bending cycle was repeated nine times on each sample bar. To determine the creep, one modified cycle was run where the bent sample was left at room temperature for 24 h before its internal angle was measured to determine if there was any long-term creep. After measuring, the sample was treated as in the shorter 5 min cycles and was submerged in a 52 °C water bath to flatten.

Results and Discussion Synthesis and Characterization of Monomer 1. While 1 was reported previously as a product of (+)-dihydrocarvone oxidation, no characterization data, purification procedure, or indication of the presence of regioisomer 2 had been noted.11 We prepared 1 on a multigram scale from commercially available (+)-dihydrocarvone using m-chloroperbenzoic acid (mCPBA) and a sodium acetate buffer (Scheme 1). Conversion of starting material to a 3:1 mixture of regiosiomers 1 and 2 (each as a pair of diastereomers) was complete after two days, with no remaining unsaturated ketone, unsaturated lactone, or ketone-epoxide evident by GC-MS. The two regioisomers were identified by NMR spectroscopy, with assignments determined using 2D and other methods (DEPT, HMQC, COSY; Figures S2-S4). The primary isomer 1 was concentrated by vacuum distillation, resulting in a mixture with approximately 90% 1 and 10% regioisomer 2; this mixture was used in all polymerizations reported below.

Homopolymerization of 1. Solution and bulk polymerizations of 1 using diethylzinc (ZnEt2) and tin(II) 2-ethylhexanoate (SnOct2) catalysts, both with and without benzyl alcohol as an initiator, were carried out at temperatures ranging from 20 to 120 °C (Table S2). In all cases, 1H NMR spectroscopy revealed that both the epoxide and lactone rings had opened to give low molecular weight branched oligomers (Figures S7-S8). Apparent number average molecular weights of 1.7-2.6 kg/mol and PDI values ranging from 1.4 to 1.6 were determined by SEC (versus polystyrene standards). The oligomers were hard, vibrantly yellow solids27 with less than 5 wt % gel fractions, indicating a lack of any significant cross-linking. Polymerization through both the epoxide and lactone rings is expected, as many similar or identical catalysts are used in the ring-opening polymerization of both ring types.28 The opening of the lactone ring is easily tracked by 1H NMR spectroscopy (Figure S7), with the chemical shift of the methine proton adjacent to the acyl oxygen atom shifting from 4.4 ppm in the monomer to 4.9 ppm in the polymer. The baseline resolution of these peaks allows for the calculation of the percent of lactone rings opened in the polymer, which averaged about 63%. To incorporate the remaining 37% of lactone rings that are unopened into the polymer, they must be pendent from opened epoxide rings. While the exact percentage of opened epoxide rings could not be determined from the 1H NMR spectrum because of overlapping resonances, 13C NMR spectroscopy analysis confirmed that the epoxide rings are opening (Figure S8). The decreased intensity of the resonances at 59 and 54 ppm, corresponding to the epoxide carbon atoms in the monomer, and the concomitant growth of peaks at 76 and 68 ppm, corresponding to a linear structure in the polymer, constitutes evidence of epoxide ring-opening. The opening of both the epoxide and lactone rings allows for a branched structure of the polymer. Copolymerization of 1 with CL. In initial attempts at copolymerizing 1 and CL, we used SnOct2 as catalyst without solvent at 120 °C for 24 h (1 g scale, Scheme 2). The polymers produced (Table S3) contained between 10 and 21 mol % 1, but in all cases the amount of 1 incorporated into the polymer was less than the percentage of 1 in the feed. In addition, while the polymers contained both opened epoxide and lactone rings (as determined by NMR spectroscopy, see Supporting Information), cross-linking was not apparent by solvent extraction, even

2006

Biomacromolecules, Vol. 10, No. 7, 2009

Lowe et al.

Table 1. Properties of Statistical Copolymers of 1 and CL 1 in the feed (mol %)

synthesis method

gel fractionb (wt %)

0.8 6.0 12.0 28.0 51.0 75.0 90.0 0.0 0.3 1.0 2.1 4.6 7.1 10.0 21.6

solution solution solution solution solution solution solution bulk bulk bulk bulk bulk bulk bulk bulk

>99 >99 >99 >99 >99 8.2 6.7 0.0 41.8 94.2 76.4 91.9 89.0 84.2 36.6

a

Mxc (kg/mol)

13.1 17.5 6.4 5.1 7.1 7.2 14.9

CHCl3 absorbedd

0.0 45.1 24.4 16.8 12.1 15.3 14.9 56.3

Tge (°C)

Tme (°C)

crystallinitye,f (%)

-54 -45 -49 -49 -30 -15 -3

47, 51 56 32, 41 28 25

41 47 35 7 2