Heat insulators are key materials for efficient energy use and reduction of CO2 emissions. Heat-resistant insulation materials are indispensable for applications under harsh conditions, including those for automobile and aircraft engines and space technology. Insulation materials are classified into organic and inorganic materials. Organic heat insulators such as plastic and nonwoven fabrics have the advantages of high processability and low weight; however, they are usually flammable and undergo thermal decomposition at high temperatures, hindering their use under harsh conditions [1, 2]. On the other hand, inorganic heat insulators such as glass wool and rock wool usually have high thermal and mechanical stability but are saddled with disadvantages such as low processability and high weight compared with organic heat insulators [3, 4].

Polysilsesquioxane (PSQ) has received much attention as a typical organic–inorganic hybrid material that possesses advantageous properties derived from both organic and inorganic components [5,6,7,8]. PSQ can be prepared by the hydrolysis/condensation polymerization of trifunctional silanes such as organotriethoxysilane (RSi(OEt)3), forming thermally stable siloxane (Si–O–Si) bonds. The generated siloxane-based highly branched network structure provides high heat resistance to PSQ, and the organic group (R) on the silicon atom improves its solubility in organic solvents, resulting in high processability. For example, PSQ-modified cellulose materials have been reported as heat insulation materials [9, 10]. We have studied the development of PSQ-based heat insulation materials with high heat resistance. PSQ-based heat insulation films can be readily produced by the sol–gel process as follows: the PSQ polymer sol is prepared by partial polymerization and calcined after a sol film is formed to further facilitate condensation; this process produces a gel film [11]. An example using triethoxymethylsilane as the monomer is presented in Scheme 1a. An issue in the gelation process involves the condensation of silanols; here, significant film shrinkage and cracking would occur, and the process is difficult to control. To overcome this problem, we introduced a new method that involves the hydrosilylation of Si–H- and vinyl-terminated methyl-PSQs (MSQ-SiH and MSQ-SiVinyl) in a 1:1 ratio in the presence of a catalytic amount of a platinum catalyst (MSQ-ethylene in Scheme 1b) [12]. When a small amount of MSQ-SiVinyl was used, the remaining Si–H groups underwent platinum-catalyzed oxidative coupling, resulting in siloxane linkages that promoted gelation. We also showed that void spaces formed around the resultant ethylene units and enhanced the heat insulation properties. Further studies on the introduction of cross-linking units by hydrosilylation indicated that the rigid and bulky cross-linking units could enhance void space formation in the films [13,14,15]. For example, a PSQ film was prepared by hydrosilylation of divinyl-substituted double-decker silsesquioxane cage compounds (DDSQ in Fig. 1) with MSQ-SiH in a vinyl:Si–H ratio of 25% (MSQ-DDSQ25) followed by oxidative coupling of Si–H groups; this film had high heat insulation properties with a low thermal diffusivity of 1.02 × 10−7 m2 s−1 [14]. The MSQ-DDSQ25-derived film also had high heat resistance; moreover, based on its thermogravimetric analysis (TGA), a high temperature for 5% weight loss (Td5) of 438 °C in air was observed. Despite these efforts, systematic studies aimed at elucidating the relationship between the cross-linking units and the thermal properties of PSQ materials are needed to improve the molecular design. In this study, we prepared PSQs with different cross-linking units to determine the effects of the cross-linking units on the heat resistance and heat insulation properties. PSQ linked by diethynylbenzene (DEB) provided sufficiently low thermal diffusivity with moderately high heat resistance.

Scheme 1
scheme 1

Preparation of PSQ

Fig. 1
figure 1

Structures of the cross-linking reagents

PSQs with various cross-linking units were prepared by the hydrosilylation of MSQ-SiH and aromatic diynes and a triyne. The cross-linking reagents were selected on the basis of the notion that the steric rigidity of the units would enhance void space formation through the pillar effect (Fig. 1). The progress of hydrosilylation was monitored by 1H nuclear magnetic resonance (NMR) measurements, as illustrated by the reaction of MSQ-SiH with 1,4-DEB with an ethynyl:SiH ratio of 5% in the presence of Karstedt’s catalyst in Fig. 2 (Figs. S1S8). The reaction proceeded smoothly, and the spectrum indicated that the ethynyl units were completely consumed after 12 h. The IR spectrum also revealed that no ethynyl units remained at this stage. After solvent evaporation, the resulting polymer liquid was poured into a Teflon vial. The vial was heated stepwise from 60 to 200 °C in the air to promote the platinum-catalyzed oxidative coupling of hydrosilyl units, forming a self-standing colorless gel film of MSQ-DEB5. Under the same conditions, films with higher DEB contents (MSQ-DEB15 and MSQ-DBE25) were also obtained using the hydrosilylation reaction with ethynyl:SiH ratios of 15% and 25%, followed by gelation, respectively.

Fig. 2
figure 2

1H NMR spectra of MSQ-SiH, DEB, and MSQ-DEB polymer and the vertical magnification (from bottom to top) and a photograph of the film obtained by calcination. The MSQ-DEB sample contained residual EtO units. The signals of the reaction solvent of toluene are also observed

The gel film formation was examined using other cross-linking reagents, as shown in Fig. 1. Some of the films were yellow-colored but remained transparent, with the exception of the MSQ-2,6-DEN film; this film became reddish brown and partly opaque during the gelation process, as shown in Fig. S9. This likely occurred because a larger amount of the platinum catalyst (H2PtCl6 6H2O) was used for the hydrosilylation of MSQ-SiH with 2,6-DEN. The hydrosilylation reaction by Karstedt’s catalyst proceeded very slowly with 2,6-DEN. For comparison, hydrosilylation reactions of MSQ-SiH with 5% monofunctional ethynylbenzene (EB) and less rigid divinylbenzene (DVB) were also carried out. Subsequent gelation in air provided the MSQ-EB5 and MSQ-DVB5 films.

The thermal stability of the cross-linked PSQ films was evaluated via TGA in air as well as in a nitrogen atmosphere (Fig. S9). The thermal diffusivities of the films were also measured to estimate their heat insulation properties. Table 1 summarizes the thermal properties and densities of the cross-linked PSQ-based films. All films showed high thermal stability, with Td5 values higher than 410 °C in air and 480 °C in nitrogen. A comparison of the thermal stability of the films with 5% cross-linking units revealed that the MSQ-1,4-DEN5 film had the highest thermal stability under a nitrogen atmosphere with a Td5 value of 666 °C according to TGA. The thiophene-containing films MSQ-2,5-DET5 and MSQ-3,4-DET5 also exhibited high thermal stability in nitrogen. This was likely caused by the high thermal stability of these aromatic units and the tendency for π‒π interactions to occur in the films. The MSQ-2,6-DEN5 film containing 2,6-naphthylene units that were sterically less favorable for π‒π interactions than 1,5-naphthylene units showed a lower Td5 value in air and nitrogen. The MSQ-TEB5 film cross-linked by trisubstituted benzene units presented a higher Td5 value than MSQ-DEB5. With respect to TGA in air, the MSQ-3,4-DET5 film had the highest Td5 of 512 °C. All films had high char yields in a nitrogen atmosphere, indicating the formation of carbon-containing residue after heating at high temperatures; these results were supported by the formation of the black residues under a nitrogen atmosphere. In contrast, heating the films in the air produced white or pale gray powder, indicating the formation of silica. The calculated char yields assuming complete conversion of silicon in the film to SiO2 were in good agreement with the experimental data (Table 1). Similar formations of carbon-containing residues such as SiOC ceramics and silica from PSQ by heating in nitrogen and air, respectively, have been reported previously [16].

Table 1 Thermal properties and densities of the PSQ-based films

As shown in Table 1, the thermal diffusivities ranged from 1.09 to 1.40 × 10−7 m2 s−1, depending on the type of cross-linking unit and its content in the film. For the MSQ-DEB films, increasing the DEB content from 5% to 15% resulted in a decrease in the thermal diffusivity. However, a further increase in the DEB content to 25% resulted in a slight increase in the thermal diffusivity. The thermal diffusivity of MSQ-DEB15 was slightly greater than that of MSQ-DDSQ25 [14] but lower than that of MSQ-ethylene [12], as reported previously. Among the films containing 5% cross-linking units, the MSQ-2,6-DEN5 film had the lowest thermal diffusivity. However, increasing the 2,6-DEN content in the film was difficult because of the low reactivity of 2,6-DEN. As expected, a strong positive correlation was observed between the film density and the thermal diffusivity, as shown in Fig. 3, with the exception of the MSQ-TEB5 and MSQ-3,4-DET5 films; their plots showed significant deviation from the linear approximation of other plots. These results were probably caused by the larger number of conduction paths in these films through the trisubstituted benzene cross-linking units and the π‒π interactions that were sterically preferred between the 3,4-thienylene units. The thermal conduction paths through the π‒π interaction could also slightly influence the MSQ-1,4-DEN5 film, which showed higher thermal diffusivity than the MSQ-2,6-DEN5 film, even though the densities of these two films were nearly the same. The MSQ-EB5 film derived from monosubstituted benzene had much greater thermal diffusivity than the MSQ-DEB5 film, indicating that the pillar effect plays a crucial role in enhancing the heat insulation properties of the film. Compared with the MSQ-DEB5 film, the MSQ-DVB5 film containing more flexible cross-linking units had slightly greater thermal diffusivity. These flexible cross-linking units likely enhanced molecular mobility to facilitate the π‒π interactions and to suppress the pillar effect. However, the sterically bulky ethylene units could increase the free volume across the cross-linking units more than the less bulky ethenylene units. These factors could balance each other, resulting in a slightly higher thermal diffusivity of the MSQ-DVB5 film.

Fig. 3
figure 3

Density–thermal diffusivity plots for the PSQ-based films with different cross-linking units. The broken line shows the linear approximation for the blue plots. The red plots are not included in the linear approximation

In conclusion, we prepared MSQ materials containing a variety of cross-linking units and demonstrated their structure–thermal property relationships. Although other influencing factors could exist, the rigidity, π–π interactions, and bulkiness of the cross-linking units were found to predominantly influence the thermal properties. These findings will lead to new molecular designs for PSQ-based heat-resistant heat insulators with high performance. Studies to develop new heat insulation materials are currently being performed, and the results will be reported in future publications.