Integrated Extrinsic and Intrinsic Self‐Healing of Polysiloxane Materials by Cleavable Molecular Cages Encapsulating Fluoride Ions

Abstract Self‐healing ability is crucial to increasing the lifetime and reliability of materials. In this study, spatiotemporal control of the healing of a polysiloxane material is achieved using a cleavable cage compound encapsulating a fluoride ion (F−), which triggeres the dynamic rearrangement of the siloxane (Si–O–Si) networks. A self‐healing siloxane‐based elastomer is prepared by cross‐linking polydimethylsiloxane (PDMS) with a F−‐encapsulating cage‐type germoxane (Ge–O–Ge) compound. This material can self‐heal repeatedly under humid conditions. The F− released by hydrolytic cleavage of the cage framework contributes to rejoining of the cut pieces by promoting the local rearrangement of the siloxane networks. The use of a molecular cage encapsulating a catalyst for dynamic bond rearrangement provides a new concept for designing self‐healing polysiloxane materials based on integrated extrinsic and intrinsic mechanisms.

germoxane (Ge-O-Ge) [12] cage compounds can encapsulate F − .By incorporation of the molecular capsules of F − into siloxane networks, the rearrangement of the Si-O-Si bonds for selfhealing can be triggered by external stimuli to release F − .This concept is similar to that of extrinsic self-healing systems [13] in which healing agents in microcapsules are dispersed in a matrix and released at the damaged area.The essential difference lies in the capsule sizes and release mechanisms.The D4R-type siloxane and germoxane compounds are quite small (< 1 nm) so that transparent materials with no light scattering can be obtained.The encapsulated F − can be released not by mechanical rupture, but by hydrolytic cleavage of the cage frameworks.The germoxane cage is superior to the siloxane cage because Ge-O-Ge bonds are more susceptible to hydrolysis. [14]When the material is damaged, the D4R cages near the cut or cracked surfaces are exposed to moisture in the air and release F − to induce local rearrangement of the siloxane networks.This can be regarded as a combined extrinsic and intrinsic healing system.
To the best of our knowledge, silicone-based self-healing materials containing F − have not been reported, probably because of the difficulty in uniformly incorporating F − into relatively hydrophobic and nonpolar networks.The D4R-type germoxane cage can act not only as a capsule of F − but also as crosslinkers of PDMS because of the eight functional groups radially arranged from the cage.We recently reported the functionalization of the corner Ge-OH groups of F − -encapsulating germoxane cages with dimethylvinylsilyl groups. [15]This allowed postchemical modification through hydrosilylation without deteriorating the cage structure. [15,16]These features of F --encapsulating germoxane cages guarantee their uniform molecular-level incorporation into silicone networks.
In this study, a self-healing siloxane-based material was designed by using F − -encapsulating germoxane cages as crosslinkers of PDMS.The hydrosilylation reaction between F −encapsulating D4R-type germoxane cages functionalized with dimethylvinylsilyl groups (GeD4R-Vi) and H-terminated PDMS (PDMS-H) produced a transparent elastomer (PDMS-GeD4R) (Figure 1a).The cut elastomers healed under humid conditions, which was triggered by the cleavage of the germoxane cages to release F − (Figure 1b).For investigation of the healing mechanism, we prepared two types of F − -free samples where PDMS chains are cross-linked with either D4R siloxane cages or GeO 4 units.A comparison of the healing efficiencies of these samples with that of PDMS-GeD4R indicated that the rearrangement of the Si-O-Si networks catalyzed by F -played a crucial role in rejoining of the cut surfaces.

Results and Discussion
GeD4R-Vi, containing a fluoride anion inside and a tetraethylammonium cation outside the cage, was synthesized according to our previous report [15] (for details, see Procedures S1 and S2  and Figures S1 and S2, Supporting Information).GeD4R-Vi and PDMS-H (average molecular weight = 17200) were dissolved in toluene and reacted in the presence of Karstedt's catalyst.After stirring at room temperature for 10 min, the reaction mixture was poured into a cylindrical perfluoroalkoxy alkane vial and allowed to stand at 60 °C for 2 d in an open system under nitrogen atmosphere.The removal of the remaining solvent under re-duced pressure produced a clear and colorless elastomer (PDMS-GeD4R, Figure 1a, see also Procedure S3, Supporting Information).
The solid-state 29 Si magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectrum of PDMS-GeD4R is shown in Figure 1c.The absence of the signal for the terminal SiMe 2 H groups of PDMS (≈−7 ppm) and the appearance of the signals for (SiO)SiMe 2 CH 2 -(C:  = 8.6 ppm) and (GeO)SiMe 2 CH 2 -(D:  = 16.3 ppm) [16] indicated that hydrosilylation reaction proceeded.Signals for the residual vinylsilyl groups [(GeO)SiMe The solid-state 19 F MAS NMR spectrum of PDMS-GeD4R (Figure 1d) showed a signal at  = −14.5 ppm, which was assigned to F − inside the germoxane cage. [12,15,16]Other signals at  = −131.2and −163.9 ppm were assignable to the Si-F and Ge-F species, respectively, based on the 19 F NMR spectra of F − -treated siloxane and germoxane compounds (for details, see Figures S4-S6, Supporting Information).These species indicated that F − was partially released from the cage during the synthesis of PDMS-GeD4R.The integral ratio of the 19 F signal for encapsulated F − to that for Si-F/Ge-F was 0.83:0.17,indicating that 83% of F − were still inside the germoxane cages.The partial release of F − was probably due to the cleavage of the germoxane cage by a small amount of water in the solvent or by the moisture in the air.The scanning electron microscopy (SEM)-energydispersive X-ray spectroscopy (EDS) mapping images of PDMS-GeD4R (Figures 1e,f,g) revealed uniform elemental distributions of Ge, Si, and F, indicating that the germoxane cages and encapsulated F − were well-dispersed in the siloxane networks.
An elastomer with a thickness of ≈1.5 mm was cut into a rectangular shape and then cut in half using a razor blade to evaluate its self-healing ability.The cut pieces were placed together and treated under humid conditions to cleave the germoxane cages near the cut surfaces.After 3 d at 25 °C and 80% RH, the cut surfaces appeared to rejoin and did not separate when pulled apart with tweezers (Figure 2A(a)).The healed elastomer was cut vertically across the rejoined cut surfaces, and the cross-section was observed by SEM.No gap was observed between the rejoined cut surfaces from the surface to the interior (Figure 2B(a)).When the cut elastomer was treated at a higher temperature with the same amount of water vapor (60 °C and 15% RH), healing occurred within a shorter time (1 d) than that required at 25 °C (3 d) (Figure 2A(b) and 2B(b)).The cut PDMS-GeD4R was not self-healed under a low humidity condition using a drying oven even if heated at 80 °C (data not shown).
Tensile tests were performed to evaluate the healing efficiency of PDMS-GeD4R.The stress-strain curves of dumbbell-shaped pieces of the as-prepared and healed elastomers (at 25 °C and 80% RH and at 60 °C and 15% RH) are shown in Figure 2C.The stiffness of the elastomer, as indicated by the slope of the stress−strain curve, was unchanged before and after healing, suggesting that the elastomer was not deteriorated by the treatment under humid conditions.The maximum stress, maximum strain, and average healing efficiency values calculated from the maximum stresses before and after self-healing are listed in Table 1.Comparison of the maximum stress values before and after healing revealed that the average healing efficiencies at 25 °C and 80% RH and at 60 °C and 15% RH were ≈72% and 77%, respectively.The results suggest that the cut surfaces are not merely attached by noncovalent interactions such as van der Waals force but are rejoined at the molecular level by the rearrangement of the siloxane networks.It should be noted here that the healing efficiency varied from ≈50% to ≈90% depending on each sample piece tested (see Table S1, Supporting Information).The low efficiency values (≈50%) should be due to the misalignment between the attached cut surfaces, causing (i) decrease in the crosssectional area of the sample piece at the healed region and (ii) tearing from the steps on the elastomer surface by the stress concentration upon tensile test.
Tracking of the healing process showed that the healing efficiencies were 37%, 57%, 77%, and 90% after treatment at 60 °C and 15% RH for 1, 3, 6, and 24 h, respectively, and the crosssectional SEM images showed no gaps between the attached cut surfaces (Figure S7, Supporting Information).Thus, the cut surfaces are healed in the early stages, and the healing efficiency increased with time.Furthermore, a similarly high healing efficiency (84% on average for three samples, Table S2, Supporting Information) was obtained when the healed sample was cut again at a different position (≈300 μm away from the first cut) (Figure S8, Supporting Information) and treated at 60 °C and 15% RH for 1 d, suggesting that the healing of PDMS-GeD4R was repeatable.
The 19 F MAS NMR spectrum of PDMS-GeD4R treated at 60 °C and 15% RH for 1 d (see Figure S9, Supporting Information) showed signals assigned to encapsulated F -, Ge-F, and Si-F.The integral ratio indicated that the remaining encapsulated F -was 39%, which was lower than that before the treatment (83%).Wa-   b) Based on a previous report, [17] the healing efficiency was calculated as the ratio of the maximum stress of the healed samples to that of the as-prepared samples.
ter vapor was assumed to diffuse from the surface of the cut gel, promoting the cleavage of the cages and release of F -.The released F -reacted not only with the Ge species, but also with the PDMS chains to induce the rearrangement of the siloxane chains.
Because the amount of germoxane cage contained in PDMS-GeD4R was too small to spectroscopically investigate the cleavage behavior of the cage under humid conditions, we prepared a similar elastomer by the reaction of GeD4R-Vi with H-terminated PDMS with a lower molecular weight (consisting mainly of linear hexasiloxane) to significantly increase the germoxane content (see Procedure S4, Supporting Information).The resulting clear solid named D6-GeD4R (Figure S10A, Supporting Information) was treated at 60 °C and 80% RH for 1 d.Note that the humidity was set higher than that for the self-healing experiment to accelerate the reaction with moisture.
The 29 Si MAS NMR spectra of D6-GeD4R before and after the treatment at 60 °C and 80% RH (Figure S10B, Supporting Information) were almost the same, indicating that the main siloxane networks were retained.The 19 F MAS NMR spectrum (Figure S10C) revealed a decrease in encapsulated F − and increase in Si-F and Ge-F species.After the treatment, the FT-IR spectrum (Figure S10D, Supporting Information) showed that the Ge-O-Ge stretching band of D4R germoxane units (≈910 cm −1 ) [15] significantly decreased.Another Ge-O-Ge stretching-derived band at lower wavenumbers (≈868 cm −1 ) assignable to cleaved germoxane cages [15] increased.The decrease in the intensity of the O-Ge-O bending of D4R units (≈496 cm −1 ) [18] was also observed.In the Raman spectrum (Figure S10E, Supporting Information), the intensity of the peak assigned to the four-membered germoxane ring [19] (≈420 cm −1 ) decreased after the treatment.These spectroscopic results indicate that the germoxane cages are cleaved under humid conditions, and the released F − reacts with Si and Ge species.
For detailed investigation of the healing mechanism, we prepared a F − -free siloxane-based elastomer (PDMS-SiD4R) using a F − -free siloxane cage (SiD4R-Vi) instead of GeD4R as a crosslinker (see Procedure S5 and Figure S11A, Supporting Information).The healing efficiency of PDMS-SiD4R after treatment at 60 °C and 15% RH for 1 d was as low as 30.9% (Figure S11B, Supporting Information).The cut surfaces were still observable in the cross-sectional SEM image, and the sample easily separated when pulled apart with tweezers (Figure S11C and D, Supporting Information).It was assumed that the cut surfaces were physically adhered and did not heal at the molecular level.When a tetrahydrofuran (THF) solution of tetrabutylammonium fluoride (TBAF) was dropped onto the cut surface, PDMS-SiD4R was repaired (Figure S11E, Supporting Information).These results in-dicate that the released F -from the germoxane cage significantly contributes to self-healing.
Although TBAF was effective in healing cut elastomers as described above, incorporating TBAF into the elastomer to impart self-healing ability was difficult.For example, when PDMS-SiD4R was swollen with a THF solution of TBAF and subsequently dried (Procedure S6, Supporting Information), no selfhealing behavior was observed after cutting (Figure S12A, Supporting Information).The hydrophobic PDMS networks likely hindered the impregnation of TBAF, as suggested by the SEM-EDS analysis (Figure S12B, Supporting Information).We also attempted to introduce TBAF during the cross-linking reactions between SiD4R-Vi and PDMS-H (Procedure S6, Supporting Information).8e] Furthermore, cut pieces were not rejoined, possibly due to phase separation of hydrated TBAF in the elastomer (Figure S12C and D (c), Supporting Information).These results highlight the effectiveness of our concept based on encapsulation and uniform distribution of F -using the molecular cages.
The dehydration condensation of Ge-OH and Si-OH groups generated by the cleavage of the Ge-O-Ge and Ge-O-Si bonds might be involved in rejoining the cut surfaces.To evaluate the contribution of the Ge-OH groups to the healing efficiency, we prepared a F − -free elastomer containing Ge-O-Si bonds (PDMS-GeO 4 ) via the dealcoholization reaction between Si-OH terminated PDMS and Ge(OEt) 4 (Procedure S7 and Figure S13, Supporting Information).The Ge-O-Si bond is known to be hydrolytically unstable as is the Ge-O-Ge bond. [14]Cut pieces of PDMS-GeO 4 were placed together and treated at 60 °C and 15% RH for 1 d.However, the cut pieces were not rejoined but easily separated when pulled apart with tweezers.Thus, the contribution of the Ge-O-Si bonds to healing was negligible.
Figure 3 shows the proposed reaction mechanisms during healing based on the rearrangement of the Si-O-Si networks.First, the germoxane cages cleave upon hydrolysis of the Ge-O-Ge bonds to release F − .Second, the following two types of cleavage of the siloxane bonds in PDMS occur: i) the coordination of F − to silicon [10a,10j,20] promotes nucleophilic attack by H 2 O, resulting in the formation of Si-OH end groups (Figure 3a), and ii) the nucleophilic attack of F − on the silicon atoms directly forms Si-O − and Si-F end groups (Figure 3b).Si-OH groups can be subsequently formed by the attack of the Si-O − groups on H 2 O (Figure 3b). [21]he next step is the reformation of the siloxane bonds.The nucleophilic attack of the Si-OH groups on the F − -coordinated sili- con atoms in other PDMS chains [10a,10j,20] results in the exchange of Si-O-Si bonds (Figure 3c).10a] A problem in the applications of silicone elastomers is the volatilization of low-molecular-weight oligomers, which, for example, causes electrical contact failure. [23]8d] A similarly small weight loss was observed for the sample treated at 60 °C and 15% RH for 1 d followed by air-drying at 105 °C.We also confirmed that PDMS-GeD4R showed self-healing ability even after heating at 185 °C for 1 d (Figure S14, Supporting Information).8c] Such a difference can be attributed to the higher thermal stability of the F − -encapsulating cage structures and Si-F bonds over silanolates.We have previously synthesized F −encapsulating cage germoxane-based cross-linked networks [16a] and ionic liquids [16b] and investigated their thermal stability, confirming that they were thermally stable without decomposition up to ≈200 °C.The thermal and chemical stability of Si-F compounds such as SiF 4 was also revealed by computational simulations. [24]These results show that PDMS-GeD4R has a significant advantage for practical applications.

Conclusion
We prepared a self-healing siloxane-based elastomer (PDMS-GeD4R) by cross-linking PDMS with F − -encapsulating cage-type germoxane.The germoxane cages acted not only as capsules of fluoride ions, but also as cross-linking sites.Humid conditions promoted the cleavage of Ge-O-Ge bonds and release of fluoride ions.The released fluoride ions triggered the rearrangement of the siloxane networks, resulting in the self-healing of the cut sample.PDMS-GeD4R remained self-healing even after repeated cutting of the elastomer.We also confirmed that the self-healing elastomer containing the F − -encapsulating germoxane cage had higher thermal stability than previously reported PDMS-based self-healing elastomers.Comparison with fluoride-free samples revealed that the F − in the elastomer dominantly contributed to self-healing.Improvement of the self-healing ability and mechanical properties of this material will be possible by changing the cage-type germoxane content and the cross-linking density.Thus, a new concept of healing mechanism in response to damage can be realized through the introduction of well-dispersed cleavable molecular capsules into polymer materials.

Figure 2 .
Figure 2. A) Product appearances and B) cross-sectional SEM images of PDMS-GeD4R after cutting and healing at a) 25 °C and 80% RH for 3 d and b) 60 °C and 15% RH for 1 d.The yellow dashed frames in the SEM images mark the self-healed cut surfaces.C) Stress−strain curves of (a) as-prepared PDMS-GeD4R and self-healed PDMS-GeD4R treated at (b) 25 °C and 80% RH for 3 d and c) 60 °C and 15% RH for 1 d and a dumbbell-shaped piece of PDMS-GeD4R used for the tensile test (inset).

Figure 3 .
Figure 3. F − -catalyzed cleavage and reformation of Si-O-Si bonds for self-healing.