Introduction

Because of well-developed synthetic methods, the properties of poly(monosubstituted ethylene)s (vinyl polymers) have been widely investigated. Controlling the chemical structure of side groups enables us to fine tune physical properties. For example, the glass transition temperature (Tg) of poly(alkyl methacrylate)s can be controlled from –70 °C to 111 °C simply by varying the length of alkyl side chains [1]. The manipulation of the glass transition is equivalent to the control of the α-relaxation process [2]. The effectiveness of random copolymerization to control the physical properties has been reported [3].

In contrast to vinyl polymers, research regarding poly(1,2-disubstituted ethylene)s, that is, poly(substituted methylene)s, is limited [4, 5]. The extremely low propagation kinetics originating from steric hindrance prevents the polymerization in many cases. An exception is the radical polymerization of some dialkyl fumarates [6,7,8]. When the ester alkyl group of the fumarate is bulky, a high-molecular-weight poly(dialkyl fumarate) can be synthesized [4, 9]. Polymers are produced when not only the propagation rate but also the termination rate becomes reasonably slow. A fundamental difference between poly(monosubstituted ethylene)s and poly(1,2-disubstituted ethylene)s is whether or not a methylene spacer (–CH2–) in the main chain is present. Due to the absence of a methylene spacer, poly-(dialkyl fumarate) forms a rigid chain structure with high rotational energy [10,11,12,13].

It is well known that typical amorphous polymers exhibit a main α-relaxation and a weaker β-relaxation process [14]. These relaxation processes can be observed by dielectric spectroscopy (DS) [15], mechanical analysis [16], thermal analysis [17], solid-state nuclear magnetic resonance (NMR) spectroscopy [18], friction force microscopy [19], and fluorescence microscopy [20]. It is noteworthy that the detailed and fundamental mechanisms of both the α- and β-relaxations are not perfectly understood despite a large amount of knowledge about these relaxations [21]. A simple view of α- and β-relaxations for amorphous polymers correlates these dynamics to the segmental dynamics of the main chain and the local motion of the side group, respectively [14]. However, this simple picture does not always apply, especially for rigid molecules [21]. It was revealed that α- and β-relaxations are not fully independent [18]. Some β-relaxations are considered to be a local step of the α-segmental relaxation [21]. In the case of poly(methyl methacrylate) (PMMA), a detailed analysis by NMR spectroscopy showed that the side-group flip correlates with the main chain rearrangement [22]. Based on the results of the dynamic mechanical analysis (DMA) of poly(diisopropyl fumarate) (PDiPF) film, Yamada et al. reported a change in the storage modulus from ~1 GPa to ~0.1 GPa ca. 100 °C, which corresponded to the β-relaxation [23]. They also pointed out that the observed intensity of the α-relaxation was much smaller than that of the β-relaxation. Kurosu et al. investigated the chain dynamics of PDiPF in solid-state 13C cross-polarized magic angle spinning and 1H pulse NMR spectroscopic experiments [24]. They revealed that the reorientation frequency of the main chain of PDiPF was 42.6 kHz at 120 °C and that the molecular motion was highly restricted by an increase in the bulkiness of the ester alkyl group.

Previous research has clarified unique properties of PDiPF, including gas permeation [25], surface properties [26], solution properties [10,11,12,13], solid-state chain dynamics [23, 24], liquid crystals [27,28,29], and electro-optical properties [30,31,32]. At least some of these unique properties are intimately related to relaxation processes [25]. Thus, it is of great importance to control and understand the relaxation behavior of PDiPF. In this study, we synthesized random copolymers of diisopropyl fumarate (DiPF) with two different alkyl acrylates, that is, 1-adamantyl acrylate (AdA) and n-butyl acrylate (nBA). While the Tg of poly(n-butyl acrylate) (PnBA) is –53 °C, that of poly(1-adamantyl acrylate) (PAdA) is 156 °C [33]. We discuss the shift in relaxation temperatures and the merging of the α- and β-relaxations as a function of the copolymer composition.

Experimental methods

Materials

DiPF (Fujifilm Wako Pure Chemical Corporation, Tokyo, Japan) and nBA (Tokyo Chemical Industry Corporation, Ltd, Tokyo, Japan) were distilled under reduced pressure prior to use. Dimethyl 2,2′-azobis(2-methylpropionate) (MAIB, Fujifilm Wako Pure Chemical Corporation, Tokyo, Japan) was purified via recrystallization in n-hexane before use. AdA was provided by Osaka Organic Chemical Industry Ltd, Osaka, Japan and was purified via recrystallization in n-hexane before use.

Synthesis of random copolymers

Here, the synthetic methods are briefly summarized. The detailed synthetic methods of the homopolymers and random copolymers of DiPF are reported elsewhere [33, 34]. DiPF, an acrylate (nBA or AdA), and MAIB were mixed in a polymerization tube. Freeze–pump–thaw cycles were repeated three times for degassing. Under a nitrogen atmosphere, the polymerization was conducted at 60 °C for nBA or 80 °C for AdA. The obtained polymer was dissolved in chloroform and subsequently precipitated in an excess amount of a methanol/water mixture (9/1 vol/vol). The polymer was washed with the methanol/water mixture and collected by decantation or percolation. The polymer was dried under vacuum at 60 °C. The precipitation process was repeated twice for purification.

Measurements

1H NMR spectra were recorded using ECS-400 and ECX-400 spectrometers (JEOL, Tokyo, Japan). Size exclusion chromatography (SEC) was employed to elucidate the number-average (Mn) and weight-average (Mw) molecular weights. Tetrahydrofuran (flow rate 0.8 mL/min) was used as the eluent at 40 °C. The SEC system was developed with a PU-2080-Plus (pump), a DG-2080-53 (degasser), an RI-2031-Plus (RI detector), and a UV-2031-Plus (UV detector, 300 nm) (JASCO, Tokyo, Japan). TSK gel GMHHR-N (Tosoh Corporation Ltd, Tokyo, Japan) was used as the column. The molecular weights were calibrated against standard polystyrenes (Tosho Corporation Ltd, Tokyo, Japan). Thermogravimetric (TG) analysis was conducted using a DTG-60 (Shimadzu Corporation Ltd, Kyoto, Japan) with a scanning rate of 10 °C/min under a nitrogen atmosphere. Differential scanning calorimetry (DSC) was measured using a DSC-60 (Shimadzu Corporation Ltd, Kyoto, Japan) under nitrogen flow with a scanning rate of 10 °C/min.

Samples for DMA were prepared by thermal welding. Approximately 1.5 g of the polymer was inserted in a rectangular mold (5 cm × 1 cm). After the polymer was annealed for 30 min at 150 °C, pressure (approximately 1 MPa) was applied for 1 h at the same temperature. Typically, a rectangular sample with a thickness of 2 mm was obtained. A DMS 6100 (Seiko Instruments Inc., Tokyo, Japan) was utilized for the DMA measurement. The instrument was operated in dual cantilever mode with a heating rate of 2 °C/min. Sinusoidal strains with an amplitude of 10 μm at 2 Hz was applied. DMA data for some polymers was collected at five different frequencies in a range of 0.5‒10 Hz.

Results and discussion

In this study, three kinds of homopolymers, PDiPF, PnBA, and PAdA, as well as the random copolymers of DiPF with nBA [P(DiPF-co-nBA)] and of DiPF with AdA [P(DiPF-co-AdA)], were synthesized based on the procedure reported previously [33, 34]. The schematic poly(1,2-disubstituted ethylene) [i.e., poly(substituted methylene)] and poly-(monosubstituted ethylene), as well as the structures of PDiPF, PnBA, and PAdA, are displayed in Fig. 1. PDiPF was selected as a model poly(1,2-disubstituted ethylene) because of the relatively high polymerization reactivity of the corresponding monomer. To obtain high-molecular-weight polyfumarates, the ester alkyl groups need to be bulky because such monomers have reasonably slow termination rates that compensate for the slow propagation rate [35,36,37]. For example, the reaction of dimethyl fumarate only produces oligomers because the methyl group is not bulky enough. In contrast, previous studies showed that the polymerization of DiPF proceeds to give high-molecular-weight polymers via bulk polymerization or in solution. Among the polyfumarates, the physical properties of PDiPF have been well investigated [10,11,12,13, 27,28,29,30,31,32]. AdA and nBA were selected as acrylates in this study. Adamantane has a rigid symmetrical structure with a high melting point (270 °C). Typically, the insertion of adamantyl groups in the polymer main and side chains increases the Tg of the polymers [33, 34, 38,39,40,41,42,43,44,45]. Due to the side-chain structure, the Tg values of PnBA and PAdA are –53 °C and 156 °C, respectively [33].

Fig. 1
figure 1

Schematics of a poly(1,2-disubstituted ethylene) and b poly(monosubstituted ethylene). Structures of c poly(diisopropyl fumarate) (PdiPF), d poly(n-butyl acrylate) (PnBA), and e poly(1-adamantyl acrylate) (AdA) are provided

Representative 1H NMR spectra of P(DiPF-co-nBA) and P(DiPF-co-AdA) are shown in Fig. 2. The signal from the proton denoted as b in the schematic as well as the proton denoted as f in the case of P(DiPF-co-nBA) or other protons in the case of P(DiPF-co-AdA) were used to determine the copolymer composition. The characteristics of the copolymers prepared in this study are summarized in Table 1. The higher the acrylate composition is, the broader the dispersity (Mw/Mn) becomes. This is due to the reactivity differences between DiPF and acrylates. In our previous study, monomer reactivity ratios and the number-average sequence lengths for DiPF and AdA were calculated [34]. Five different copolymers whose DiPF composition spanned from 3 to 59 mol% were analyzed. The number-average sequence lengths of DiPF and AdA ranged from 1.01 to 2.44 and 1.69 to 30.7, respectively. This result suggests that the DiPF sequence is randomly well distributed for all the analyzed copolymers. In other words, the probability of a long DiPF sequence in the copolymer was low for the copolymers produced under the conditions used in the previous and present studies. Td5 was determined from the 5% weight loss in the TG curves. The decomposition of an isopropyl ester group in PDiPF occurs at the lowest temperature [34]. As the DiPF content in copolymer decreased, Td5 increased.

Fig. 2
figure 2

1H nuclear magnetic resonance (NMR) spectra of a P(DiPF-co-nBA) and b P(DiPF-co-AdA). Based on the ratio between the signal from the proton denoted as b and other protons, the DiPF content was calculated to be 64 mol% and 76 mol% for a, b, respectively

Table 1 List of homopolymers and random copolymers synthesized in this study

Figure 3 shows the DMA results for the homopolymers. Figure 3a displays the storage and loss moduli of PAdA observed at 2 Hz as a function of temperature. Because of the adamantyl moiety in the side chain, the Tg of PAdA is higher than typical polyacrylates [2]. The storage modulus was almost the same below the onset of the α-relaxation process at approximately 140 °C. Then, the storage modulus suddenly decreased from ~1 GPa to ~10 kPa. At temperatures above the α-relaxation, the polymer was rubbery. The peak temperature of the loss modulus for PAdA shown in Fig. 3a was 151.4 °C and is a good indication of the segmental motion. The slight decrease in the storage modulus ca. ‒100 °C indicates the existence of the β-process (see Fig. S1 in the Supporting Information). In any case, the intensity of the β-process was much smaller than that of the α-process. This behavior was qualitatively similar to other vinyl polymers such as PMMA [46]. Compared to the data of PAdA, the DMA curve of PDiPF was fundamentally different (Fig. 3c). Two relaxation processes were observed at 73.4 °C and ca. 215 °C as the peaks of the loss moduli (Fig. 3c). The processes were also confirmed from the peaks of the tan δ (Fig. 3d). During the lower temperature relaxation, the storage modulus decreased from 1.3 GPa at 50 °C to 0.2 GPa at 100 °C. Although the relaxation process is clear, the absolute value of storage modulus (0.2 GPa) indicates that the sample is still glassy, that is, a nonfluid solid. The storage modulus in a range between the low- and high-temperature relaxations was more or less constant. The intensity of the higher temperature process was weak. In this study, the measurement was stopped at 230 °C to avoid the decomposition of the sample. Consistent with previous reports [23], it is reasonable to assign the high- and low-temperature processes to the α- and β-relaxations, respectively. The α-relaxation of PDiPF was observed ca. 220 °C, which was close to the onset temperature of thermal decomposition. Therefore, we checked the relaxation processes for another fumarate polymer. The relaxation processes of poly(diethyl fumarate) (PDEF) were detected at 18.9 °C and 159.4 °C, which were assigned to the β- and α-relaxations, respectively (Fig. 3e, f). Both relaxation temperatures for PDEF were lower than those for PDiPF because the steric bulkiness of the ester in the side chain was small. As discussed later, the low-temperature relaxation process of the polyfumarates possesses unusual characteristics.

Fig. 3
figure 3

Dynamic mechanical analysis (DMA) of poly(1-adamantyl acrylate) (PAdA) (a and b), poly(diisopropyl fumarate) (PDiPF) (c and d), and poly(diethyl fumarate) (PDEF) (e and f). E, E″ (a, c, and e) and tan δ (b, d, and f) are plotted as a function of time. The α-process of PDiPF is indicated with a gray arrow in d

Selected DMA curves of PDiPF and P(DiPF-co-AdA)s with different compositions are displayed in Fig. 4a. As discussed earlier, the storage modulus of PDiPF stayed approximately 0.2 GPa above the β-relaxation temperature at 73.4 °C. As the composition of the DiPF units decreased, the storage modulus after the β-relaxation decreased more significantly. In the case of P(DiPF-co-AdA)-46, the storage modulus at 180 °C was approximately 0.6 MPa. The shift in relaxation temperatures is discussed with respect to the tan δ presented in Fig. 4b. The α- and β-relaxation of PDiPF are indicated in Fig. 4. Comparing the results from PDiPF, P(DiPF-co-AdA)-89, and P(DiPF-co-AdA)-76, the β-relaxation temperature increased as the composition of DiPF decreased. In many cases, β-relaxation is a very local motion of the side groups [21]. The fact that the β-relaxation temperature shifted along with the copolymer composition in the present study implies the local dynamics of the side chains are influenced by the sequence structure of the main chain. In contrast to an increase in the β-relaxation temperature, the α-relaxation temperature decreased as the composition of DiPF decreased. The broad α-relaxation may reflect the statistical distribution of the successive DiPF segments. Interestingly, these two processes seemed to merge in the case of P(DiPF-co-AdA)-46, resulting in the observation of an apparent single relaxation process with a high intensity of the tan δ peak.

Fig. 4
figure 4

Selected dynamic mechanical analysis (DMA) data from poly(diisopropyl fumarate) (PDiPF) and P(DiPF-co-AdA)s with different copolymer compositions. a Storage modulus (gray) and loss modulus (red) are plotted as a function of temperature. b Tan δ is plotted as a function of temperature. The number after the name of copolymer indicates the PDiPF content in the copolymer

Similarly, the DMA results of P(DiPF-co-nBA)s with different compositions are shown in Fig. 5a. Because of the flexible alkyl side group, the Tg of PnBA was –53 °C. Again, as the acrylate composition increased, the storage modulus above the β-relaxation temperature decreased. While the storage modulus was almost constant in the case of PDiPF between the α- and β-relaxation temperatures, the storage modulus slightly decreased as the fraction of DiPF units decreased. For PDiPF, P(DiPF-co-nBA)-95, and P(DiPF-co-nBA)-78, almost constant values were observed for the α- and β-relaxation temperatures (Fig. 5b). For these three samples, the α-relaxation process was very weak. The data from P(DiPF-co-nBA)-64 were fundamentally different. From the tan δ, there were two strong relaxation processes with peak temperatures at 84 °C and 30 °C.

Fig. 5
figure 5

Selected dynamic mechanical analysis (DMA) data from poly(diisopropyl fumarate) (PDiPF) and P(DiPF-co-nBA)s with different copolymer compositions. a Storage modulus (gray) and loss modulus (red) are plotted as a function of temperature. b Tan δ is plotted as a function of temperature. The number after the name of copolymer indicates the PDiPF content in the copolymer

To further clarify the features of the chain dynamics of the DiPF copolymers, their thermal properties were recorded. Figure 6 depicts traces for the DSC measurements of PDiPF and of the copolymers. DSC captured the change in the heat capacity as a step in the heat flow curve. When the relaxation process caused a detectable change in heat capacity, this event was recorded by DSC. Typically, only a single glass transition, which is associated with the α-relaxation process, is detected for common polymers including polyacrylates and polymethacrylates [2]. Although less common, some studies have detected β-relaxation from calorimetry [47]. Here, we exclusively used TDSC to represent relaxations observed by DSC because it was difficult to distinguish the mode of the chain dynamics based only on the calorimetric experimental results. Interestingly, PDiPF showed a clear step in the heat flow curve, which corresponded to the β relaxation temperature observed by DMA. This is a unique feature of PDiPF. This implies that the strong β-relaxation of DiPF segment causes the change in heat capacity. It was difficult to detect any α relaxation process of PDiPF by DSC under the present conditions. The thermal decomposition of the polymer actually occurred at a high temperature. To protect the instruments, other scans were made for the copolymers in a temperature range clearly below the decomposition temperature. Figures 6a and 6b present the results from P(DiPF-co-AdA)s and P(DiPF-co-nBA)s, respectively. As the PDiPF content decreased, the TDSC of P(DiPF-co-AdA)s increased. For example, the samples with 100%, 65%, and 46% PDiPF content showed a TDSC of 66 °C, 105 °C, and 114 °C, respectively. In contrast, the TDSC of P(DiPF-co-nBA)s decreased as the PDiPF content decreased. The TDSC values of P(DiPF-co-nBA)-64, P(DiPF-co-nBA)-35, and P(DiPF-co-nBA)-19 were 58 °C, −17 °C, and −53 °C, respectively.

Fig. 6
figure 6

Differential scanning calorimetry (DSC) curves of selected P(DiPF-co-AdA)s and P(DiPF-co-nBA)s with different copolymer compositions. The data were recorded upon heating with a rate of 10 °C/min

The relaxation temperatures obtained from DMA and DSC are summarized in Table 1 and Fig. 7. In the case of P(DiPF-co-AdA)s, the TDSC increased almost linearly as the PDiPF content decreased. According to the Fox equation [48], the Tg of a binary random copolymer can be estimated by the following equation:

$$1/T_{\mathrm{g}}=x_1/T_{{\mathrm{g}},1} + (1 - x_2)/T_{{\mathrm{g}},2}.$$
(1)
Fig. 7
figure 7

Summary of the relaxation temperatures of a P(DiPF-co-AdA)s and b P(DiPF-co-nBA)s as a function of copolymer compositions. The filled symbols are from the dynamic mechanical analysis (DMA) data. The empty circles are from the differential scanning calorimetry (DSC) data. The black lines are predicted by the Fox equation. Red lines and dotted gray lines are visual aids

Here, x indicates the composition of the two polymers denoted as 1 and 2. The calculated Tg from the Fox equation was plotted as a black line in Fig. 7. It is emphasized that for the samples containing more than 76% DiPF content, the relaxation detected by DSC coincides with the β-relaxation measured by DMA. In contrast to the β-relaxation, the weak α-relaxation temperature decreased as the DiPF content decreased. Eventually, these two relaxation processes merged when the DiPF content was 46%. Figure 6b shows the relaxation temperatures for P(DiPF-co-nBA)s. Again, the β-relaxation process from PDiPF and the α-relaxation process followed the Fox equation. As the DiPF content decreased, both the α-relaxation and β-relaxation temperatures decreased. The α-relaxation temperature decreased significantly when the DiPF content decreased from 78 to 64%. This behavior suggests the merging of the α- and β-relaxation, and further investigation of the molecular dynamics observed during the α- and β-relaxation processes are now discussed.

Figure 8 displays selected DMA data of P(DiPF-co-AdA)-89 and P(DiPF-co-nBA)-89 measured at five different frequencies in a range from 0.5 to 10 Hz. The tan δ of the β-relaxation processes shows a clear maximum, while that of the α-relaxation processes was broad. Assuming the Arrhenius dependence of the β-relaxation processes, apparent activation energies were obtained (Table 1). It should be noted that an α-relaxation time (τ) as a function of temperature (T) generally follows the Vogel–Fulcher–Tammann equation [2]:

$$\tau = \tau _0{\mathrm{exp}}\left( {\frac{B}{{T - T_0}}} \right),$$
(2)

where τ0 is the relaxation at the high temperature limit, B is a positive constant, and T0 is the ideal glass transition temperature. The characteristic time approaches infinity at T0. The current frequency range (0.5‒10 Hz) was too narrow to accurately fit the data (see Fig. S2 in the Supporting Information). The β-relaxation activation energy was 270 and 280‒310 kJ/mol for PDiPF and the DiPF copolymers, respectively. The standard deviation of the fit was approximately ±10 kJ/mol. The literature activation energy values are 58‒125 kJ/mol for the β-relaxation and 325‒585 kJ/mol for the α-relaxation of PMMA [2, 16, 49,50,51,52,53,54,55,56,57,58,59], 59‒63 kJ/mol for the β-relaxation and 180‒238 kJ/mol for the α-relaxation of poly(methyl acrylate) (PMA) [54], and 113‒152 kJ/mol for the β-relaxation and 210‒294 kJ/mol for the α relaxation of PAdA and its block copolymers [33], as summarized in Table 2. Thus, the β-relaxation activation energy of PDiPF is much larger than that of conventional polyacrylates and polymethacrylates, but the frequency range investigated in this study was limited. A detailed analysis of the β-relaxation containing the fumarate segments is ongoing [60, 61].

Fig. 8
figure 8

Selected dynamic mechanical analysis (DMA) data from P(DiPF-co-AdA)-89 and P(DiPF-co-nBA)89 measured at 0.5, 1, 2, 5, and 10 Hz. Storage modulus (dotted line) and tan δ (line) as a function of temperature are plotted

Table 2 List of activation energies estimated for some acrylic polymers in the literature

Our results indicate the unique properties of β-relaxation arising from DiPF segments. This process is categorized as the relaxation due to side-chain motion. On the other hand, the fact that clear transition behavior was detected by DSC implies that this dynamic has some similarity with common α-relaxations for any other conventional vinyl polymers. These unique properties should be associated with hindered rotational motion due to the absence of a methylene group. As shown in the case of PMMA [18], the β-relaxation of PDiPF may be a local step for the segmental motion associated with the α-relaxation.

Conclusion

We synthesized the DiPF homopolymer and two series of random copolymers containing DiPF and acrylates with several compositions. As counterpart acrylates, nBA and AdA whose glass transition temperatures are distinctively different were selected. The DMA measurement of PDiPF as the homopolymer showed a strong β-relaxation. The storage modulus decreased from ~1 GPa to ~0.1 GPa during the β-relaxation process. In both cases of copolymers of DiPF with AdA and nBA, we found that the β-relaxation from the DiPF segments and the α-relaxation from the polyacrylates segments were correlated with each other and merged according to the copolymer compositions. It was demonstrated that the random copolymerization of DiPF and acrylates enabled the control of both α- and β-relaxation processes for flexible and nonflexible polymer chains. Because the control of the dynamics of PDiPF by random copolymerization may open new application areas using side-chain dynamics, further characterization of the β-relaxation of PDiPF as well as some polymers containing DiPF repeating units is ongoing [60, 61].