Materials

4,4′-Thio-bis(6-tert-butyl-m-methyl phenol) (AO300, T_g=302 K, n=0.43, dipole moment μ=2.5 D), 4,4′-thiodiphenol (TDP, T_g ~210 K, μ=3.0 D), 1,3,5-trimethyl-2,4,6-tris(3,5-di-tert-butyl-4-hydroxybenzyl) benzene (AO330, T_g=368 K, n=0.46, μ=1.8 D), and PMMA (M_w=90 kg/mol, PDI=1.90, T_g=381 K, μ=1.7 D) were provided by Aladdin. 1,1,3-Tris(2-methyl-4-hydroxy-5-tert-butylphenyl) butane (CA, T_g=371 K, n=0.53, μ=2.0 D) was provided by Tianjin Lisheng Chemical. PBMA (M_w=330 kg/mol, PDI=2.27, T_g=298 K, μ=1.8 D) was synthesized via solution polymerization under a nitrogen atmosphere by using azobis(isobutyronitrile) (AIBN, ≥99.5%, Aldrich) as an initiator. All reagent molecules were used as received, and their chemical structures are plotted in Fig. 1.

1Full-screen View

Molecular structures of (a) PMMA, (b) PBMA, (c) TDP, (d) AO300, (e) CA and (f) AO330.

Sample Preparation

The polymer was dissolved in ethyl acetate (EAc) solution at a concentration of 20 wt% with continuous stirring. Small molecules were then added and stirred until completely dissolved to prepare hybrids with different additive weight concentrations. The solution was cast into the glass mold, kept at room temperature for 48 h, and further dried under vacuum at 120 °C for 2 h to ensure a complete evaporation of solvents.

Broad Band Dielectric Spectrometer (BDS)

The dielectric spectra were collected with a Novocontrol Concept 40 broad band dielectric spectrometer in the frequency domain of 0.01 Hz to 10 MHz. Samples with a diameter of 20 mm and a thickness of 250 μm were prepared via hot pressing above the T_m of small molecules, followed by a quenching process. The temperatures were controlled within 0.1 K with a Novocontrol Quatro Cryosystem, and liquid N₂ was used to cool the samples. The dielectric loss spectra were fitted using a single or multiple Havriliak-Negami (HN) Eq. (2):^[60,61]

2 ${\varepsilon^{{\tiny{*}}} }\left( \omega \right) = {\varepsilon _\infty } + \sum\limits_{{i}} {\frac{{{\varepsilon _{{\rm{0i}}}} - {\varepsilon _{\infty {{i}}}}}}{{{{\left[ {1 + {{\left( {j\omega {\tau _{{\rm{HN}}}}_{\rm{i}}} \right)}^{{a_{{i}}}}}} \right]}^{{b_{{i}}}}}}}} - j{\left( {\frac{\sigma }{{{e_{\rm{0}}}\omega }}} \right)^s}$

where ε^* represents the complex dielectric permittivity. Δε = ε₀ − ε_∞ is the dielectric strength, where ε₀ and ε_∞ are the relaxed and unrelaxed dielectric constants at extremely low and high frequencies, respectively. ai and bi (0<ai, bi≤1) are shape parameters of relaxation process i (primary or secondary) that represent symmetric and asymmetric broadening, respectively. σ is the ionic conductivity, e₀ (8.854 pF/m) is the free space dielectric permittivity, and s is a parameter very close to the unit.^[60] τ_HN is the relaxation time by HN equation and is used to obtain the peak location τ_max:^[62]

3 ${\tau _{{\rm{max}}}} = {\tau _{{\rm{HN}}}}{\left( {{\rm{sin}}\frac{{ab{\text{π}} }}{{2\left( {b + 1} \right)}}} \right)^{1/a}}{\left( {{\rm{sin}}\frac{{a{\text{π}} }}{{2\left( {b + 1} \right)}}} \right)^{ - 1/a}}$

In the PBMA hybrids, we set b<1 for asymmetric primary α-peaks, and b=1 for symmetric secondary β_JG-peaks. The PMMA hybrids have an unusually asymmetric secondary β_JG-peak with b<1.^[63,64] The HN fitting procedures for PBMA or PMMA mixtures are clearly shown in the Supporting Information (Figs. S1−S5, Tables S1 and S2 in the electronic supplementary information, ESI). The increase of temperature results in an accelerated β_JG-process (as manifested by a higher peak frequency) with enhanced peak intensity. The increase of temperature also accelerates α-process, but it reduces the intensity of α-process due to the weakening α-cooperativity.

Effect of AO300 Concentration on the β_JG-Relaxation

Fig. 2(a) shows the dielectric loss ε″ near T_g of PMMA containing different amounts of AO300. It can be observed that neat PMMA displays a prominent and well-resolved β_JG-peak at the high-frequency flank of α-peak whose strength is nearly twice as that of α-relaxation as previous reports.^[63,64] The addition of AO300 substantially enhances the α-peak intensity and simultaneously reduces the β_JG-intensity, thereby leading to a diminishing β_JG-peak that merges gradually with α-peak. HN fitting curves are given for better identifying the respective peak times. Fig. 2(a) also depicts the temperature dependence of the α- and β_JG-relaxation time for these PMMA/AO300 hybrids. We can see that the addition of AO300 accelerates α-relaxation of PMMA (decreased τα) at a given temperature. Meanwhile, the β_JG-motion of PMMA is apparently restricted, as manifested by an increased τ_JG, after hydroxyl groups of AO300 are hydrogen bonded to the carbonyl moiety of the PMMA side group as detected by FTIR analysis.^[52] These results demonstrate that AO300 acts typically as an anti-plasticizer^[65-67] in the PMMA matrix.

2Full-screen View

Dielectric loss ε″ at indicated temperatures near T_g and temperature dependence of the α- and β_JG-relaxation time for (a) PMMA and (b) PBMA hybrids containing different AO300 loadings. The dashed red, purple and green curves represent the HN fittings of α-, β_JG- and β_fast-peaks, respectively.

Neat PBMA exhibits a prominent β_JG-process whose intensity is close to that of the α-relaxation (Fig. 2b). Compared to rigid PMMA, the flexible PBMA exhibits a distinguishable β_fast-peak (the dependence of β_fast-peak time and intensity on temperature is shown in Fig. S2 in ESI) at a frequency near to 10⁶ Hz, and a small α, β_JG separation due to the reduction in coupling factor n caused by the inner plasticizing effect of flexible butyl group.^[39] Addition of AO300 results in the slowing down of both α- and β_JG-relaxations, as well as a narrowed α, β_JG separation. As will be further discussed below, we have confirmed that both the β_fast-peak and the suppressed β_JG-peak in Fig. 2(b) are resulted from the cooperative impact of the polymer chain and the small molecule, rather than from the extra relaxation modes of the additives. The addition of AO300 in PBMA leads to a weak and slow β_JG-relaxation as PMMA, clearly demonstrating that phenolic molecules act like the chemical modification of side group and effectively restrict the β_JG-relaxation of poly(n-alkyl methacrylate)s when they are hydrogen bonded to the carbonyl moiety of acrylic pendent groups.

Separation of the β_JG-peak from the α-peak of PBMA and PBMA/phenol mixtures is particularly difficult by HN fittings due to their overlapping near T_g. We recognize that the multiple HN fittings must create large uncertainties in determining the dispersion of α- and β_JG-peaks,^[68] despite that the uncertainties of the most probable intensity and peak time can be effectively minimized if the HN fitting is carefully performed under procedures given in Fig. S1 (in ESI).^[68,69] Therefore, the dashed fitting curves for the PBMA systems in this work are just for guide of eye to identify the position of various peaks. As shown in Fig. 2(a), with the acceleration of the α-relaxation and the slowing down of the diminishing β_JG-relaxation, the α- and β_JG-traces of the PMMA/AO300 hybrids gradually move closer to each other, and the β_JG-peak cannot be resolved from the α-peak when the AO300 content reaches 75 wt% (Fig. S6 in ESI). The PBMA/AO300 hybrids also display a reduced α, β_JG separation (Fig. 2b and Table 1) due to the fact that AO300 molecules more significantly slow down the β_JG-relaxation than the α-relaxation. Notably, when the loading of AO300 exceeded 17 wt%, β_JG-relaxation of PBMA is encroached by α-relaxation and cannot be resolved well from α-relaxation. At 50 wt%, β_JG-relaxation is suppressed completely (Fig. S7 in ESI).

Effect of Molecule Size on the β_JG-Relaxation

The triphenyl-ring CA (V_w=373 cm³/mol) molecule has a similar steric hindrance as AO300 (V_w=229 cm³/mol) on hydroxyl groups (Fig. 1); thus, despite of bulkier van der Waals volume (V_w) and capability of forming triple inter-HBs per molecule, it owns an inter-HB strength Δυ_i (158 cm⁻¹, determined by FTIR analysis as shown in Figs. S8 and S9 in ESI) very close to AO300 (Δυ_i=167 cm⁻¹) at ambient temperatures. Fig. 3(a) shows dielectric loss ε″ curves and temperature dependence of the α- and β_JG-relaxation time for PMMA containing 0 wt%, 17 wt% and 60 wt% CA. The addition of CA substantially enhances the α-intensity and simultaneously reduces the β_JG-intensity, which is in similar tendency as AO300-loaded PMMA mixtures. However, different from the AO300 system, the CA molecule with a comparable T_g to that of PMMA matrix trivially affects the τ_α of PMMA. Furthermore, an unrestricted β_JG-motion, as manifested by little change in τ_JG, can be observed when bulkier CA molecules are added. We observe a slight acceleration of τ_JG with increasing CA loading (Fig. 3a), a trend which is different from the τ_JG change in the case of adding AO300.

3Full-screen View

Dielectric loss ε″ at indicated temperatures near T_g and temperature dependence of the α- and β_JG-relaxation time for (a) PMMA and (b) PBMA hybrids containing different amounts of CA (wt%). The dashed red, purple and green curves represent the HN fittings of α-, β_JG- and β_fast-peaks, respectively.

When the bulky CA mixes with the flexible PBMA matrix, a remarkable increase in τ_JG (Fig. 3b) is observed, suggesting a substantially constrained β_JG-motion of PBMA. Addition of 17 wt% CA leads to an increase in β_JG-peak temperatures at τ_max=1 s (T_JG) of PBMA about 9 K (Table 1), which is higher than the increment in T_JG (6 K) of the PBMA/AO300 hybrids at the same content range. Due to the large mobility difference between CA and PBMA, bulky CA molecules more significantly slow down the α-relaxation than the β_JG-relaxation of PBMA (Fig. 3b). Consequently, the α- and β_JG-traces of the PBMA/CA hybrids separate from each other with increasing CA content. We noticed again a decrease in τ_JG of the PBMA/CA hybrids as the CA loading reaches 60 wt%. This phenomenon arises from a weakened Δυ_i at the increased T_g, which will be further discussed in the next section.

Effect of Inter-HB Strength on the β_JG-Relaxation

A robust inter-HB is crucial for coupling the motion between the guest molecules and the acrylic pendent groups. To elucidate the effect of the inter-HB strength Δυ_i on the β_JG-properties, various phenolic molecules with adjusted van der Waals volume (V_w) and steric hindrance on their hydroxyls were mixed with the PBMA matrix. Among them, AO330 (V_w=515 cm³/mol, Δυ_i=103 cm⁻¹) has a larger bulkiness but a weaker Δυ_i than CA (V_w=373 cm³/mol, Δυ_i=158 cm⁻¹), whereas TDP (V_w=113 cm³/mol, Δυ_i=313 cm⁻¹) has a smaller size but a stronger Δυ_i than AO300 (V_w=229 cm³/mol, Δυ_i=167 cm⁻¹). Note that the Δυ_i mentioned here is determined at ambient temperatures, while in Table 1 it is at T_g.

Compared to the β_JG-traces of PBMA with adding CA, Fig. 4 shows that adding bulky AO330 with a weak Δυ_i has little effect on change in τ_JG of PBMA, though intensity of the β_JG-peak is suppressed remarkably. By contrast, the rigid AO330 significantly restricts the segmental relaxation of PBMA, thereby leading to an apparent increase in τα or Tα of the PBMA/AO330 hybrids. As a result, the α- and β_JG-traces of the PBMA/AO330 hybrids gradually separated from each other with increasing AO330 content, and the significant increase in τα or Tα is responsible for the increased α, β_JG separations. On the contrary, addition of small TDP molecules with a stronger Δυ_i effectively restricts the β_JG-motion of PBMA and leads to a remarkable increase in τ_JG of PBMA as shown in Figs. S10 and S11 (in ESI). Compared with the β_JG-traces in PBMA/AO300 hybrids, the β_JG-peak of PBMA rapidly converges with α-peak after adding TDP, and the β_JG-peak becomes un-resolvable when TDP loading exceeds 17 wt%. We noted that the TDP mixtures exhibit more pronounced α-, β_JG- and β_fast-peaks than AO300, CA and AO330 mixtures (Fig. S11 in ESI). This may be rationalized by the fact that TDP (μ=3.0 D) has the largest dipole moment among these additives. However, variations of intensity, relaxation time and activation energy of the secondary relaxations with the TDP loading exhibit a similar tendency as that of AO300 mixtures (Fig. 5 and Fig. S12 in ESI), further verifying that the β_JG-peak stems from the hydrogen bonding-induced cooperative impact of PBMA and TDP, rather than from the independent relaxation of TDP.

4Full-screen View

Dielectric loss ε″ at indicated temperatures near T_g and temperature dependence of α- and β_JG-relaxation time for PBMA hybrids containing 0 wt%, 17 wt%, and 50 wt% AO330. The dashed red, purple and green curves represent the HN fittings of α-, β_JG- and β_fast-peaks, respectively.

5Full-screen View

Temperature dependence of β_JG-relaxation time for PBMA hybrids mixed with different phenols at a given loading of 17 wt%.

Fig. 5 shows a comparison of τ_JG at different temperatures for PBMA hybrids containing different types of small molecules. The loading is given at 17 wt%. As is seen, the temperature dependence of τ_JG in all PBMA hybrids follows Arrhenius fashion with slightly varied activation energies from 77 kJ/mol to 84 kJ/mol. At a given temperature (for example, at 280 K denoted by the dashed arrow in Fig. 5), the τ_JG is found to increase with increasing Δυ_i in the order AO330<AO300<TDP. Clearly, a stronger Δυ_i between the small molecule and the polymer chain leads to a more restricted β_JG-motion and a longer τ_JG. Apart from the significant effect of Δυ_i, we also note a longer τ_JG in the CA mixture than that in the AO300 system, and the τ_JG in the CA mixture is slightly longer than that in the high-Δυ_i TDP mixture. These results highlight a strong impact of the additive size and demonstrate that a bulkier molecule is also beneficial for a more restricted β_JG-motion.

Generally, Δυ_i decreases with increasing temperatures and its temperature dependence shall change a little with the type of small molecules (Figs. S8 and S9 in ESI). These variations may provoke slight changes in the activation energy (E_JG) of τ_JG. We observe in Fig. 4 that the τ_JG of the PBMA/AO330 at high-temperature range is longer than that of neat PBMA whereas at low-temperature range it becomes shorter. The variation of Δυ_i with temperatures can also account for the decrease of τ_JG in PBMA/CA hybrids when a high CA content is loaded (Fig. 3b): at a loading of 60% CA, T_g of PBMA increases significantly from 298 K to 351 K, leading to the apparent reduction of Δυ_i from 158 cm⁻¹ to 147 cm⁻¹ (Table 1). If the PBMA matrix is replaced by high-T_g PMMA, a reduction of Δυ_i can also be observed near the T_g of the mixtures (Table 1 lists T_g and the corresponding Δυ_i for various mixtures). This fact can explain why its τ_JG decreases slightly with increasing CA loadings (Fig. 3a).

Effect of Small Molecule-bridged HBs on α, β_JG Separation

Fig. 6(a) depicts the effect of small molecule loading on the α, β_JG separation, which is denoted by the time span logτα–logτ_JG at τα=10 s. As seen, the logτα–logτ_JG of PMMA is apparently reduced with the addition of AO300, while it decreases slightly by adding bulky CA. The α- and β_JG-peaks of PBMA converge when TDP and AO300 are added. According to Eq. (1), the decreased logτα–logτ_JG strongly demonstrates the fact that the inter-chain cooperativity/constraints must be reduced, that is, a decreased coupling factor n by introducing small molecule-bridged inter-HBs despite of the HB-strengthened intermolecular interactions.

6Full-screen View

(a) Effect of small molecule loading on α, β_JG separation (logτα–logτ_JG) at τα=10 s for PMMA and PBMA mixed with various hindered phenols. (b) Effect of small molecule size (V_w) on logτα–logτ_JG at τα=10 s for PBMA hybrids containing different phenols at a loading of 17 wt%. Here, the van der Waals volume (V_w) of small molecules is normalized by the V_w of PBMA monomer unit (V_w(BMA)=87 cm³/mol).

However, we observed a gradual separation between α- and β_JG-peaks of PBMA with increasing loadings of CA and AO330, a trend completely different from other systems shown in Fig. 6(a). Furthermore, we found an enlarged separation with the increasing bulkiness of small molecules. Fig. 6(b) shows the normalized size (V_w/V_w(BMA)) dependence of the α, β_JG separation in PBMA hybrids at a given loading of 17 wt%. Clearly, the logτα–logτ_JG monotonically increases with the V_w of hindered phenols. At first glance, this trend appears reasonable because the attachment of bulky phenolic molecules to acrylic ester groups via inter-HBs can be regarded as enlarging the acrylic pendent groups and may impose greater steric hindrance effects on the segmental relaxation of flexible PBMA. However, if this was true, then addition of CA and AO330 in PBMA should strengthen topological constraints on segmental cooperative rearrangement during glass transition and leads to an increased coupling factor n.

To elucidate the changes in topological constraints, we checked the effect of CA and AO330 loadings on the α-dispersion of PBMA because a broad α-dispersion corresponds to an increased n. Due to the uncertainties of α-dispersion by HN fittings in poly(n-alkyl methacrylates),^[68] the normalized α-dispersion curves are explored in the following analysis. As shown in Figs. 7(a) and 7(b), the narrowing trend in α-dispersion of PBMA/CA and PBMA/AO330 hybrids is distinguishable with increasing small molecule loadings, though it is interfered by the β_JG-peak at the high-frequency flank of α-peak. When the loading exceeds 17 wt%, the narrowing effect becomes clearer as the β_JG-peak is dramatically suppressed and becomes resolvable from the giant α-peak. To remove the influence from the β_JG-peak, we change the matrix to be poly(butyl acrylate) (PBA), a homologue of PBMA with a much lower T_g and independent α-loss peaks. Although PBA has a T_g that is, respectively, 74 and 137 K lower than AO300 and CA, both of PBA/CA and PBA/AO300 mixtures exhibit a definitely narrowing α-dispersion (Figs. 7c and 7d) with increasing additive content, demonstrating the promoted dynamic homogeneity or reduced inter-chain constraints of PBA by adding CA and AO300.

7Full-screen View

Normalized α-loss spectra near T_g of (a) PBMA/CA, (b) PBMA/AO330, (c) PBA/CA and (d) PBA/AO300 hybrids containing different amounts of small molecules.

Dynamic fragility m quantifies the variation steepness of τα with T_g-scaled temperature at T=T_g.^[70] It measures how cooperative the α-relaxation is, and a decrease in m is generally a consequence of reduced α-cooperativity.^[36,37] As shown in Fig. 8, although a higher Δυ_i can more positively promote T_g (AO330<CA<AO300), the HB-driven hybrids exhibit an apparently lower m than both of its two constituting components at additive contents lower than 29 wt%, and a stronger Δυ_i between phenolic hydroxyl and acrylic carbonyl groups leads to a more substantial decrease in m. Accordingly, the inter-chain constraints must be alleviated by small molecule-bridged inter-HB networks, further supporting that the coupling factor n of PBMA is reduced by adding CA and AO330.

8Full-screen View

Concentration dependence of T_g, m/m_(P) of (a) P(BMA-co-St), (b) PBMA/AO330, (c) PBMA/CA, and (d) PBMA/AO300. T_g and m were determined at τ_max=1 s, and m is normalized by the neat PBMA’s own m (m_(P)). For comparison, concentration dependence of T_g, m/m_(P) of the P(BMA-co-St) is given in (a) based on data from work of Kahle et al.^[30] Here, P(BMA-co-St) is a copolymer of butyl methacrylate (BMA) and styrene (St).

Universality of Crossover Time t_c

The question is why the α, β_JG separations increase with decreasing coupling factor n for CA and AO330-loaded PBMA hybrids. It is essential to re-identify whether the sub-T_g secondary process is the β_JG-relaxation as it may be encroached by other relaxation modes of small molecules or stem from the α-relaxation due to phase separation of small molecules. In the present work, the narrowing dielectric α-dispersion (Fig. 7) together with an independent heat capacity (C_p) jump in differential scanning calorimetry (DSC) (Fig. S15 in ESI) confirm that small molecules are uniformly dispersed in polyacrylate matrix. Moreover, the continuous changes in dielectric spectra (Fig. S13 in ESI) of PBMA hybrids with increasing temperature or small molecule content can also exclude the untenable conjectures. We confirmed that small molecules used in this study display no β_JG-relaxations (Fig. S13d in ESI) and their β_fast-relaxations are well separated from the β_JG-relaxations of PBMA (Fig. S14 in ESI). Furthermore, phase separation gives rise to α-relaxation time obeying the VFT fashion, which is in contrary to the Arrhenius type found in all secondary traces of PBMA hybrids. More importantly, if phase separation took place, an enhancement in peak intensity should be observed with the increasing small molecule’s loading. The suppressed β_JG-peak observed in the present work clarifies that it must not be the extra relaxation mode of additives phase separated from the polymer matrix. In addition, the activation energy of the secondary process (E_JG, listed in Table 1) calculated from the Arrhenius plots of relaxation time approaches to the prediction by E_JG=26RT_g^[71] as the loading of small molecules increases to be higher than 17 wt% (the predicted E_JG is in a range of 65−76 kJ/mol). All of these evidences assert that the secondary peaks around the α-peak flank must originate from the β_JG-relaxation of PBMA that is restricted by bulky phenols via inter-HB interactions, rather than from the extra relaxation modes of additives.

The Coupling Model proposed by Ngai provides a connection between the cooperative τ_α and primitive τ_JG via coupling factor n through the following Eq. (4):

4 ${\rm{log}}{\tau _{{\alpha }}} - {\rm{log}}{\tau _{{\rm{JG}}}} = n\left( {{\rm{log}} {\tau _{{\alpha }}} - {\rm{log}} {t_{\rm{c}}}} \right)$

where t_c is time at the crossover from exponential to non-exponential decay, the magnitude of which is entirely determined by the interaction potential. Given that t_c = 2 ps, Eq. (4) can be simplified to Eq. (1). The prediction of the α, β_JG separations by Eq. (4) works well for weakly-interacted carbon-based small molecular/ionic glass-formers and polymers, because crossover time t_c in these systems keeps nearly constant at about 1−2 ps and is insensitive to temperature or pressure.^[72]

The counterintuitive phenomenon that separation of the β_JG-peak from the α-peak by reducing n in this study can be rationalized only when one assumes that the crossover time t_c is not a universal quantity and decreases with increasing Δυ_i. For example, the logτ_α–logτ_JG of PBMA is 2.98 (see Fig. 6a), which corresponds to a value of n=0.23 according to Eq. (1). If the addition of 17 wt% TDP or AO300 leads to a reduction of n from 0.23 to 0.18 (the same decrement of n in PEA/TDP hybrids as previously reported)^[56] and a decrease in t_c from 2 ps to 0.02 ps, then logτ_α–logτ_JG decreases from 2.98 to 2.65 as expected. Since adding bulky molecules like CA and AO330 must alleviate the reduction of n, given that n=0.22 and t_c=0.02 ps, then logτα-logτ_JG increases to 3.23. That is to say, if t_c minimizes by introducing small molecule-bridged HBs between polymer chains, one can observe the anomalous dynamics that the β_JG-peak separates from the α-peak despite of reducing its inter-chain cooperativity.

Actually, t_c decreases with increasing Δυ_i, which has been verified by molecular dynamics simulations of imidazolium-based ionic liquids.^[73] In that case, the formation of HB interactions engenders severe anharmonic chaos on the rotation motion of ions, which significantly reduces the t_c value by nearly 2 orders of magnitude (from 1 ps to ~0.02 ps).