REPORTS nofhelarn mac 2.24 e o tep,the tw couple their pe nent shape to fit aseq ed.In the 50 Fig.1.and enthalpie gon 130全 by an external s as a tem 570 atureTof the polymer movie S 20 open circles. 10 he thermally i 101 90 All of t allic alloys(13):and c able inp logical environm ts and man 2 Cyclic therm lity or compliance i 3,0 (37)with 2.5 de are sh 0 is due toa m designed 4 to the b e of leve 1,5 the Ab 00 of th 20 -10 15 100 (% g 20 an 30 % (/5).which ar 40 esent work as pu 0.8 ()w 0.6 gher than in )w inea for our 50 Dradatlpenio9al 20 300 pic properties by variation of mole m In the first ster of the re cd山 on of CAM e the precurso for the sw ching se DX 1674 31 MAY 2002 VOL 296 SCIENCE www.sciencemag.orgly be introduced into the body in a compressed temporary shape by means of minimally inva￾sive surgery and then be expanded on demand to their permanent shape to fit as required. In the same way, a complex mechanical deformation could be performed automatically instead of manually by the surgeon. The transition from the temporary to the permanent shape could be initiated by an external stimulus such as a tem￾perature increase above the switching transition temperature Ttrans of the polymer [movie S1 (19)]. The thermally induced shape-memory ef￾fect has been described for different material classes: polymers (2, 3), such as polyure￾thanes (4–7), poly(styrene-block-butadiene) (8) and polynorbornene (9, 10); hydrogels (11, 12); metallic alloys (13); and ceramics (14). All of these materials are nondegrad￾able in physiological environments and many lack either biocompatibility or compliance in mechanical properties. In metallic alloys, the shape-memory effect is due to a martensitic phase transition (13). In contrast, the polymers designed to exhibit a thermally induced shape-memory effect require two components on the molecular level: cross￾links to determine the permanent shape and switching segments with Ttrans to fix the tempo￾rary shape. Above Ttrans, the permanent shape can be deformed by application of an external stress. After cooling below Ttrans and subsequent release of the external stress, the temporary shape is obtained. The sample recovers its per￾manent shape upon heating to T . Ttrans. Cross-links can be either covalent bonds or physical interactions. Recently, we have report￾ed on shape-memory polymers (15), which are covalently cross-linked polymer networks con￾taining hydrolyzable switching segments. Em￾phasis in the present work was put on the de￾velopment of a group of polymers that contain physical cross-links. These thermoplastics are easily processed from solution or melt and are substantially tougher than polymer networks. In particular, they are degradable, showing linear mass loss during hydrolytic degradation. We selected linear, phase-segregated multiblockcopolymers as the structural con￾cept for our polymer system, because this polymer architecture allows tailoring of mac￾roscopic properties by variation of molecular parameters. In the first step of the polymer synthesis, macrodiols with different thermal character￾istics are synthesized through ring-opening polymerization of cyclic diesters or lactones, with a low-molecular-weight diol as initia￾tor, and purified (16). In the current study, oligo(ε-caprolactone)diol (OCL) was cho￾sen as the precursor for the switching seg￾ments having a melting transition tempera￾ture (Tm). Crystallizable oligo(p-dioxanone)diol (ODX), with a higher Tm than OCL was chosen as a hard segment to provide the physical cross-links (17). The melting transi￾tion of the latter macrodiols is determined by the average chain length, which can be tai￾lored by the monomer/initiator ratio (16, 17). In the second step, the two macrodiols are coupled with 2,2(4),4-trimethylhexanediisocya￾Fig. 1. Tm and enthalpies DHm of multiblockcopoly￾mers (36). Tm (OCL), solid squares; DHm (OCL), solid circles; Tm (ODX ), open squares; DHm (ODX ), open circles. Fig. 2. Cyclic thermome￾chanical experiment of PDC35 (37) with Ttrans 5 40°C. Results of the first cycle are shown. Step 1 of the experiment was strain￾controlled; steps 2 through 4 to the beginning of next cycle were stress-con￾trolled. Fig. 3. Hydrolytic deg￾radation of thermoplas￾tic shape-memory elas￾tomers in aqueous buffer solution (pH 7) at 37°C. The relative mass loss for multiblockco￾polymers differing in their hard segment con￾tent is shown (PDC10, circles; PDC17, squares; PDC31, upward-pointing triangles; PDC42, down￾ward-pointing triangles). m(t), Sample mass after a degradation period t; m(t0), original sample mass. Fig. 4. Results of CAM tests of PDC38 (sam￾ple length: left, ;0.3 cm; right, ;0.5 cm). For a positive control sample, see (25). R EPORTS 1674 31 MAY 2002 VOL 296 SCIENCE www.sciencemag.org on March 31, 2007 www.sciencemag.org Downloaded from