Molecular dynamics in polyurethane foams chemically reinforced with POSS

The polyurethane foams were obtained in a three-step process, in detail described in our previous work [8, 9, 19]. Briefly, dispersions of different POSS in THF were introduced to polyol under ultrasonic conditions, and afterwards, the catalyst, water, and surfactant were added in order to prepare the polyol premix (component A). In the next step, n-pentane as a physical blowing agent was added to component A. Then, the component A and B [polymeric 4,4′-diphenylmethane diisocyanate (PMDI)] were mixed together and stirred for 15 s. Finally, the prepared mixtures were dropped into a Teflon open mould. All the experiments were performed at ambient temperature (ca. 20 °C).

DSC curves were recorded with a PerkinElmer Pyris 6 calorimeter, cooled with liquid nitrogen, and purged with nitrogen gas. In order to increase the mass of the samples that could be fitted inside standard aluminium crucibles, foam samples were immersed in liquid nitrogen and crushed to powder. Samples of 4–7 mg were used for the measurement. A first heating run at 10 K/min to 250 °C was used to remove water and erase the thermal history. During this run, a water evaporation peak was recorded, but no peaks related to microphase mixing. A cooling run down to − 100 °C at 10 K/min followed, and a final heating at 10 K/min up to 250 °C was used for evaluation. Glass transition temperatures were calculated as midpoint values.

DMA thermograms were recorded with a Netzsch 242C dynamic mechanical analyser in compression mode. The compression was applied along the foaming direction on samples of 5 mm thickness and 10 × 10 mm² cross section. The amplitude of the oscillation was 25 μm. Three rather low frequencies were used (1, 2, and 10 Hz) because during test measurements we observed that higher frequencies tended to fracture the rigid foams. Temperature range was − 105 to 105 °C at a rate of 2 K/min. Please note though that in the region around 50 °C, the samples collapsed even at the very low stresses we applied.

The main event in the DSC curves is a very broad glass transition step around 140 °C (Fig. 1). As we mentioned also in the introduction, several authors have detected a glass transition step at the same region in rigid PU foams [11, 20]. Small loadings (5 wt%) of both POSS species enhance the midpoint T_g (Table 1) indicating a deceleration of the segmental dynamics, similar to what has been observed in elastomeric systems with the same particles [21‐23]. Here, we may attribute this effect to immobilization of PU chain segments on the bulky POSS particles. On the contrary, upon further loading, the T_g decreases for both types of silsesquioxane additives (Table 1), denoting an acceleration of dynamics. The same trend has been also observed in very similar systems but with disilanolisobutyl POSS incorporated along the chain contour, being actually part of the main chain [9].

There are four possible reasons for that, explaining also the counterintuitive observation that a high amount of a crosslinking moiety can accelerate dynamics: (1) The PU matrix itself is highly crosslinked and an excess of POSS particles may actually disrupt this crosslinking, even if it is multifunctional [22]. (2) Possible aggregation of POSS at high loadings hinders the interaction with the matrix. (3) It has been observed that during the synthesis of the foams the viscosity of the polyol component increased with addition of the POSS diols [8, 19], thus a reduction in the efficiency of the reaction may have occurred, and thus a looser, more mobile network may have formed. (iv) The organic groups of the POSS moieties may be more flexible than the hard macromolecular segments of the PU-foamed matrix.

We now turn our attention to secondary features of the DSC curves. For the OCTA-POSS hybrids, a wide endothermic event starts at 0 °C and is superimposed as a shoulder on the low-temperature side of the glass transition step. We may assume that it is related to the relaxation of POSS-crosslinked regions inside the foam, thus supporting the assumption that POSS actually have faster dynamics than the polymeric matrix.

Interestingly, although OCTA-POSS crystals were detected by wide-angle X-ray diffraction already at 10 wt% (WAXD) [8], only a very weak narrow endothermic peak is observed for 15 wt% loading and around 150 °C in the studied region, presumably because of the lower sensitivity of DSC with respect to crystallinity. On the other hand, melting peaks of PHI-POSS are detected at 153 °C for all the hybrids which contain it. This is in agreement with WAXD studies that indicate more intense crystallinity for this kind of silsesquioxane [8].

The tan δ profiles, for all PU/POSS materials, show three distinct regions in the studied temperature range (Figs. 2, 3). A weak peak at temperatures below −30 °C, a prominent peak in the vicinity of 80 °C, and a shoulder on the latter around 0 °C for the OCTA-POSS hybrids. The storage modulus E′ (Fig. 1), initially decreases in a continuous manner with temperature, but, contrary to the expected behaviour increases again after approximately 50 °C. This increase is accompanied by a significant decrease in the thickness of the sample, indicative of a mechanical collapse of the foam. Indeed, upon measurement, the sample was found irreversibly collapsed. The collapse starts at approximately the same temperature with the water evaporation peak at the first run of DSC (not shown here). It is probable that the loss of water makes the material harder and more brittle. One more factor contributing to the increase in the mechanical modulus at above ambient temperatures is the increase in gas pressure in the cells. The strong tan δ peak around 80 °C is clearly related to this phenomenon, and does not seem related to a molecular relaxation as no dependence of its peak temperature is observed with changing frequency (Fig. 2). As observed in the tan δ profiles of the matrix and the composites (Fig. 3), this high-temperature peak consists of at least two components, reflecting that more than one phenomena should be taken into account. The temperature where this macromorphological change happens, as quantified by the peak temperature in the tan δ profiles, does not change considerably with POSS type or content.

The low-temperature peak shifts to higher temperatures with increasing frequency (Fig. 2) and thus may be attributed to a molecular relaxation. At 5 wt% PHI-POSS (Fig. 3; Table 1), the peak migrates to higher temperatures, i.e. the relaxation becomes much slower. With further increasing PHI-POSS content, it regains its mobility and becomes faster, even compared to the reference material. On the contrary, at 5 wt% OCTA-POSS the relaxation is accelerated, as is peak migrates to lower temperatures, but again with increasing POSS content, the peak approaches that of the pristine material. Hence, we can observe that POSS does have an effect on the relaxation, but in a non-monotonic manner dependent on the architecture of the polymeric chains.

There is no consensus on the origin of this low-temperature peak in the literature. As mentioned in the introduction, in this area Hatakeyama et al. [11] found a step in DSC which they attributed to a low-temperature glass transition. Piszczyk et al. [24] observed in the same area a tan δ peak in flexible foams, which they attributed to a glass transition. On the contrary, Narine et al. [18] attributed a similar peak to the β relaxation. This is consistent with the observation that polyurethanes exhibit in this area a characteristic relaxation related to the fluctuation of the urethane bond which is typically coined as β relaxation [21]. In the next paragraph, we will try to contribute to the clarification of the origin, based on our observations.

As we pointed out earlier, the peak temperature of the low-temperature relaxation varies considerably with POSS content in a non-monotonic manner (Fig. 3; Table 1). Sensitivity to the filler would be expected by a segmental relaxation like a glass transition. Changes in local relaxations are hardly expected and definitely not observed in PU elastomers bearing the same POSS moieties [21, 25]. On the other hand, a glass transition would have a footprint on the DSC curves, so assignment of the peak to a glass transition would not be justified. In addition, Narine et al. [18] observed a shift of the peak with different polyols. Moreover, the topology of the PU chains seems to influence the dynamics of β relaxation of polyurethane, as proven by studies of hyperbranched polyurethanes [26‐29].

In order to shed more light on this complex situation, we construct the Arrhenius map of Fig. 4, including the data from our study, traces of β relaxations of hyperbranched PUs and elastomers, as well as the area where Piszczyk et al. [24] detected a relaxation in flexible foams. Indeed, the dynamics of β relaxation is strongly affected by the geometry/topology of the PU chain, and it may occur at very different regions. Calculation of the activation energy by three points is not reliable; however, we can see that the slope of the traces of our materials is very similar to the one of hyperbranched systems as well as to elastomers, which is another clue that the peak should be attributed to the β relaxation. Albeit this relaxation in elastomers is not sensitive to filler content, we should keep in mind that in the foam, the environment faced by the carbonyl groups is much different than the more compact environment in elastomers, and POSS may affect it by direct interaction or indirectly, by changing the foam micromorphology. This is consistent with the much opposite effects of the crosslinking moiety (OCTA-POSS), which accelerated the relaxation, and the pendent moiety (PHI-POSS), which decelerated it. This change in micromorphology has been observed in our earlier work [8]. Finally, the dynamics of the relaxation is known to be affected by water molecules [30]. Thus, we may attribute the shift to the lower hydration level of the highly POSS-loaded hybrids. Nevertheless, a more detailed investigation would be needed for the clarification of the origin of the low-temperature relaxation.

It is not clear if the shoulder starting at 0 °C for the OCTA-POSS hybrid shifts with frequency, thus it is not easy to attribute it to a structural change or a molecular relaxation. However, bearing in mind that it has a counterpart in the DSC curves of the same materials, we believe it is related to a transition in the POSS-crosslinked regions.