Introduction

The addition of lithium halides has previously been applied to polymer electrolytes in polymer solar cells. Watanabe et al. [1] demonstrated how the glass transition temperature, Tg, of PEO-based complexes increases with the addition of various salts, including lithium halides such as LiCl, LiBr, and LiI, and that there exists a correlation between the ionic conductivities and the polymer chain mobilities in PEO–salt complex membranes. Yang et al. [2] investigated the ion-conduction behavior of LiI in PEO-PVDF polymer blends and identified the optimal salt concentration in the systems based on the mobilities of I and I3.

Lithium halides have been widely used as additives for polymer modification, such as the functionalization of porous membranes for filtration [3, 4]. Zhang et al. [3] fabricated a novel porous polyethersulfone (PES) membrane with high vapor permeation by coating it with LiCl-doped polyvinyl alcohol (PVA). Idris et al. [4] investigated the influence of LiF, LiCl, and LiBr on the water vapor permeability of PES membranes and concluded that membranes consisting of 2 wt% LiBr exhibited excellent separation and high permeation rates. Furthermore, the doping of lithium halides has been used as a method to control the crystalline structure of semicrystalline polymers, such as PVA [5, 6] and polyamides [7,8,9]. For example, Jiang et al. [5] found that the addition of LiCl and other chloride salts acts as a plasticizer to starch/PVA blends and that these salts act as good compatibilizers and result in high thermal stability. They concluded that the strong coordination effects with the hydroxyl groups in the polymers reduced the crystallinity of PVA, leading to good compatibility.

Saari et al. [6] investigated the effects of the addition of LiCl, LiBr, and LiI on the rheological properties of aqueous solutions of PVA, focusing on the Hofmeister series (HS) [10]. They described the trend of the ability of these halides to precipitate proteins in an aqueous solution. They found that the plateau moduli of the dynamic mechanical spectra of PVA with these salts decreased in the order of anion species in the HS [6]. They also studied the effects of the addition of inorganic salts, such as LiClO4, LiI, LiBr, LiNO3, and LiCl, to PVA films and revealed that the addition of the salts gives rise to a lower Tg of PVA according to HS, owing to the decrease in the crystallinity of PVA caused by intermolecular interactions of the salts with hydrogen bonds in PVA [11]. They concluded that these salts act more strongly as water–structure breakers according to the order of the HS [11]. Sato et al. [12] investigated the optical and mechanical properties of melt-mixing polyamide 6 (PA6) films containing LiBr with different salt concentrations.

Recently, the salt-based modification of glassy polymers has been reported [13,14,15,16,17]. Although these glassy polymers have polar groups and are not as hydrophilic as PVA and PA6, their additional salt effects are different from those of such crystalline polymeric materials. Miyagawa et al. [13] found that the addition of a generic lithium salt, lithium trifluoro sulfonate (LiCF3SO3), leads to an increase in the glass transition temperature of poly(methyl methacrylate) (PMMA), owing to the ion–dipole interaction between lithium cations and carbonyl groups in PMMA. We previously investigated the effects of the addition of various generic salts, for example, LiCF3SO3, on PMMA and found that in such generic lithium salts, the aggregates of the salts interact with the PMMA matrix [16,17,18].

According to previous studies [17, 18], these halide ions can be molecularly dispersed in various polymers containing polar groups. The purpose of this work is to clarify the effects of anion size on the static and dynamic mechanical properties of PMMA. For this purpose, we selected a series of lithium halides as additive salts and investigated the tensile fracture behavior of comp-molded sheets of PMMA samples doped with LiCl, LiBr, and LiI in relation to their rheological properties at the compression molding temperature [16, 17].

Experimental procedure

Materials and sample preparation

PMMA pellets with an average molecular weight Mw = 1.0 × 105 and molecular weight distribution Mw/Mn = 1.9 (calibrated with PMMA standard) were used in this study. Lithium salts, such as lithium chloride (LiCl, purity ≥99.0%; Kanto Chemical Co., Ltd.), lithium bromide (LiBr, purity >99.0%; Tokyo Chemical Industry Co. Ltd. (TCI)), and lithium iodide (LiI, purity: 99.9%; Sigma‒Aldrich), were used without any additional purification. The PMMA pellets and lithium salts were added with a salt concentration of 0.03 molar ratio to the carbonyl groups in PMMA ([Li]/[C=O] ratio of 0.03) to a solvent mixture of dichloromethane and methanol in a 9:1 weight ratio and stirred for 1 h. The cast films of the PMMA/LiX (X = Cl, Br, and I) blends were then prepared by air-drying at room temperature for 24 h and dried by vacuum heating in an oven at 135 °C for 30 h to remove the residual solvents. The cast films were pulverized and compression-molded at 200 °C and 20 MPa for 5 min using a hot press machine (Techno Supply Corp., Japan) after preheating at 200 °C for 5 min. After a rapid quenching process at 25 °C for 5 min, we obtained sample sheets with a thickness of ~200 µm. We confirmed that the absence of crystallization in the added salts suggests that the salts were uniformly dispersed in the PMMA matrix, as shown in a previous paper [18]. For PMMA/LiX (X = Cl, Br and I), the IR peak of the carbonyl group at 1724 cm−1 was broadened to lower wavenumbers, i.e., the lower energy region, at room temperature [14, 18]. These results suggest ion–dipole interactions between the lithium cation and the carbonyl groups in PMMA/LiX.

Measurements

Differential scanning calorimetry (DSC) using a Diamond Differential Scanning Calorimeter (Perkin Elmer, USA) was performed to estimate the Tg of the sample sheets. The first run was conducted with a scanning speed of 10 °C/min in a temperature range of 25–220 °C. The temperature was maintained at 220 °C for 5 min between the first and second run. The Tg values of PMMA in the doped samples were evaluated after the second run of DSC measurements.

The dynamic mechanical properties of the cast films were evaluated using a viscoelastic spectrometer (DVE-V4, UBM Co., Ltd., Japan). The temperature dependence of the dynamic mechanical properties was obtained in the temperature range of −150 to 200 °C, increasing at a rate of 2 °C/min and a frequency of 10 Hz. The distance between the chucking apparatuses was set to 20 mm.

Rheological characteristics were investigated using a rotational rheometer (Discovery HR-2, TA Instruments, USA) with a parallel plate with a diameter of 8 mm under nitrogen flow. The initial gap distance was 1000 µm, and the angular frequency ranged from 0.1 to 100 rad/s within a temperature range of 120–210 °C for PMMA, 130–240 °C for PMMA/LiCl, 130–250 °C for PMMA/LiBr, and 130–290 °C for PMMA/LiI in 10 °C increments.

Tensile tests were performed using a miniature tensile testing machine (TC 05-010, Abe Seisakusho, Japan). Thin rectangular specimens (gauge size 5 × 10 mm) were cut from the sheets using an ultrasonic cutter. The tensile tests were performed at room temperature at an elongation speed of 10 mm/min.

Results and discussion

DSC curves of the glass transition temperature (Tg) of the PMMA/LiX samples are shown in Fig. 1. The glass transition shifts to higher temperatures and becomes broader in the order of PMMA/LiI > PMMA/LiBr > PMMA/LiCl; thus, PMMA doped with a larger ionic size shows a higher Tg.

Fig. 1
figure 1

Differential scanning calorimetry (DSC) heating curves obtained at 10 °C/min for pure PMMA and PMMA/LiX (X = Cl, Br and I, [Li]/[C = O] = 0.03) sheets: (black) pure PMMA, (blue) PMMA/LiCl, (green) PMMA/LiBr, and (red) PMMA/LiI

Figure 2 shows the temperature dependence of the dynamic mechanical spectra of the sample sheets in the solid state. Two typical peaks of α at ~120–130 °C and a broad β peak around room temperature were observed. The Tg value estimated from the DSC curves at 10 °C/min is identical to the peak temperature of α relaxation in the E” curves at 10 Hz. The β relaxation peak is ascribed to the rotational motion of the methoxy groups in PMMA. The Tg value increased with the ionic radius of the halide [19], as shown in Fig. 3. In addition, we found that the relaxation intensities of the β relaxation for the salt-doped sheets are lower than those of neat PMMA. It is likely that the addition of lithium halides suppressed the overall molecular mobility of the PMMA side chains, leading to higher temperature shifts of α relaxation (Tg).

Fig. 2
figure 2

Temperature dependence of dynamic tensile moduli at 10 Hz of PMMA/LiX (X = Cl, Br and I, [Li]/[C=O] = 0.03) sheets: (black) pure PMMA, (blue) PMMA/LiCl, (green) PMMA/LiBr, and (red) PMMA/LiI

Fig. 3
figure 3

Glass-transition temperature of PMMA/LiX (X = Cl, Br and I, [Li]/[C=O] = 0.03) plotted against the ionic radii r of Cl, Br, and I

Figure 4 shows the master curves of the rheological spectra, including the glassy, rubbery, and flow zones, in the order of descending frequency for the neat PMMA and PMMA/LiX samples. The master curves were obtained by shifting horizontally to the curve at 200 °C as the reference temperature (Tr), which corresponds to the compression-molding temperature of the samples. The glass–rubber (Tg) transition region became obscured, and the flow or terminal relaxation region was prolonged to lower frequencies as the anionic radius of the halide increased. These results suggest that the molecular interaction between lithium salts and PMMA chains affects the segmental motion of the PMMA chains and suppresses the fluidity of the PMMA chains. The frequency dependence of tanδ is shown in Fig. S1.

Fig. 4
figure 4

Dynamic viscoelastic spectra of a PMMA, b PMMA/LiCl, c PMMA/LiBr, and d PMMA/LiI sheets at a molar salt concentration of 0.03

The master curves, representing data at different temperatures, were shifted along the log ω axis for superposition. We define the shift distance required to superpose data at temperature T with those at reference temperature Tr as aT. A plot was constructed of –(TTr)/ln aT vs. T–Tr for all samples, where Tr = 200 °C. As shown in Fig. S2, almost all data fall on a straight line, indicating that the Williams–Landel–Ferry (WLF) equation fits the temperature dependence of ln aT. The slope and intersection of the straight line give rise to the following WLF equation in (1). As shown in Fig. S2, except in the terminal flow regions of PMMA/LiX, the WLF formula was followed. This is due to the pinning effect caused by the interaction between the carbonyl groups in PMMA and the halide ions, as described later.

$$\log a_T - \frac{{C_1\left( {T - T_r} \right)}}{{C_2 + T - T_r}}$$
(1)

The values of C1 and C2 are summarized in Table 1.

Table 1 Values of slope and intercept evaluated from WLF analysis for PMMA and PMMA/LiX (X = Cl, Br, and I) at a molar salt concentration of 0.03

There were no significant differences between the PMMA and the salt-doped PMMA within the experimental error. These values for PMMA/LiX were almost identical to those of neat PMMA. Thus, the overall molecular mobility, except for the terminal region, was independent of the addition of the salts, which is significantly different from the additional effects of generic salts, highlighting the aggregation of the salts in the PMMA matrix. These results imply that halide ions were molecularly dispersed in the PMMA matrix.

Here, we used the relaxation indices in three relaxation regions (glassy, rubbery, and terminal flow regions) in the master curves at the compression-molding temperature (200 °C) for our investigation, namely, the ratio of the relaxation time of the glass transition in the glassy region (<τG>/<τG0>), the ratio of the entanglement density, i.e., “specific entanglement density” (<νe>/<νe0>) in the rubbery plateau, and the ratio of the relaxation time in the flow region (<τF>/<τF0>), which are the ratios to the respective values of the neat PMMA [16]. The present work follows the assumption that the molecular motional states in the molten state at the molding temperature are frozen under rapid cooling [16, 17].

Figure 5a, c indicates that the ratio of the relaxation time in the glassy region, <τG>/<τG0>, as well as the ratio of the relaxation time in the flow region, <τF>/<τF0>, increase linearly with an increase in the ionic radii, r, of the anion [19]. These results indicate that the larger ionic radii of the lithium salts suppress not only the micro- but also the macro-Brownian motions of the PMMA chains. The former result is consistent with the results in Figs. 1 and 2, where the Tg of PMMA increased with the addition of lithium salts with larger anion radii. These results suggest that the addition of lithium salts with larger anionic radii has a greater pinning effect on PMMA chains by physical cross-linking.

Fig. 5
figure 5

Dependence of ionic radii r of anions within the ratios of a relaxation times in the glassy region, b entanglement density, and c relaxation times in the flow region to that of neat PMMA for PMMA/LiX (X = Cl, Br and I, [Li]/[C=O] = 0.03)

The entanglement density νe was estimated from the plateau modulus GN° in the rubber region of the viscoelastic spectra (Fig. 4) as follows:

$$v_e = \frac{c}{{M_e}} = \frac{{G_N^ \circ }}{{RT}}$$
(2)

where c is the polymer concentration (g/m3) and Me is the molecular weight (Da) between entanglements. For GN°, we adopted the value of G’ when the slope of G” in the rubbery-plateau region had the minimum value. The ratio of the entanglement densities, defined as the specific entanglement density <νe>/<νe0>, is plotted against the anion radii r in Fig. 5b.

As shown in Fig. 5b, the specific entanglement density of PMMA/LiCl increased noticeably with an increase in the size of the halide salts. In particular, the addition of Li salts with larger sizes of Br- and I- showed higher <νe>/<νe0> values greater than unity, indicating that the addition of LiBr or LiI enhances the entanglement density compared to that of neat PMMA, which is significantly different from the additional effects of the generic salts.

Figure 6 shows the strain–stress curves obtained from the tensile tests of the samples. The results reveal that the salt-doped samples exhibit lower fracture stress (strength) and strain at break than neat PMMA, indicating that the addition of salts embrittled the PMMA. In Fig. S3, we summarize the mechanical data of Young’s modulus, fracture stress, and toughness, which is the area under the stress–strain curve up to the break point, plotted against the ionic radii r of the anions [19].

Fig. 6
figure 6

Tensile stress–strain curves of PMMA/LiX (X = Cl, Br and I, [Li]/[C=O] = 0.03) sheets: (black) pure PMMA, (blue) PMMA/LiCl, (green) PMMA/LiBr, and (red) PMMA/LiI

We also compared the tensile toughness, which was estimated from the areas under the stress–strain curves up to breakage at room temperature, for the PMMA/Li-salt sheets. As shown in Fig. 7, the toughness corresponding to the fracture energy decreases monotonically with increasing <τF>/<τF0>. These results lead us to conclude that PMMA is embrittled by doping with salts because ionic atoms cause defects, such as microcracks and voids, as shown in Fig. 7.

Fig. 7
figure 7

Toughness of PMMA/LiX (X = Cl, Br and I, [Li]/[C=O] = 0.03) plotted against <τF>/<τF0> [16]

According to the Griffith theory, when a brittle material is stressed in the uniaxial direction, intrinsic voids and cracks grow in the direction perpendicular to the tensile direction. Taking the crack length as 2a, the change in energy (per unit thickness) resulting from the release of stored elastic energy due to crack length growth excluding the newly generated incremental surface energy can be expressed based on the assumption that the stress concentration occurs at the edge of the crack, and the crack propagates in-plane to maintain crack equilibrium under perpendicular stress [20]. The concept of this theory assumes that the materials originally possess microscopic cracks of their own. Then, the critical stress or strength can be written as follows:

$$\sigma _c = \sqrt {\frac{{2E\gamma }}{{\pi a}}} $$
(3)

where γ is the surface energy of the material, E is Young’s modulus, and 2a is the crack length. Polymeric materials such as PMMA undergo extensive plastic deformation before forming a new surface, a phenomenon called crazing and shear yielding. When an amorphous polymer is in a glassy state, the molecular chains are essentially frozen, and the free movement of the chains is considerably hindered. From the standpoint of fracture toughness, the occurrence of a craze near the crack tip is interpreted in the Griffith equation as plastic work, where γ is the plastic work rather than the surface energy. The fracture data of cracked PMMA confirmed that the critical stress value is proportional to \(\sqrt a \), as predicted by Griffith [21].

Figure 8 shows the fracture stress of the salt-doped samples plotted against \(\sqrt {E/r_{total}} \), where r is the total radius of Li+ [22] and the halide anion [19], that is, Cl, Br, and I.

Fig. 8
figure 8

Fracture stress plotted against \(\sqrt {E/r_{total}} \), where E is Young’s modulus and rtotal is the total ion radii of cations and anions in salts. The sample is PMMA/LiX (X = Cl, Br and I, [Li]/[C=O] = 0.03)

The fracture stress was found to be proportional to \(\sqrt {E/r_{total}} \) according to the Griffith equation. Thus, Fig. 8 strongly suggests that the fracture stress corresponds to the critical stress for crack propagation in brittle materials, and the radii of the ions are proportional to the size of the defects and cracks. Therefore, the points at which the salt was present in the salt-doped samples acted as the origin of crack initiation, and the critical stress for crack propagation decreased as the anion size increased; thus, the salt-doped sample became brittle with increasing anion size. It is likely that the halide ions are molecularly dispersed in the PMMA matrix. This is consistent with the pinning effects caused by halide doping in the flow region in the terminal flow zone. It was possible to estimate the intrinsic crack size within the PMMA samples from their tensile fracture stress values using this empirical result.

Conclusion

In this study, we investigated the effects of doping with lithium halides, LiCl, LiBr, and LiI, which have different anion sizes, on the rheological and tensile properties of PMMA. The results of DSC, dynamic mechanical properties, and rheological properties suggest that the addition of lithium halides causes pinning effects on PMMA chains by chain association with the salts. The entanglement densities increased in the Br- and I- systems, where the anions were larger. This may be because lithium halides are less likely to aggregate with each other and interact with PMMA at the molecular level, as reported previously [17, 18]. Although the shape of the master curves in the lithium halide-doped system are different from that of the original PMMA, it was indicated that the three parameters for glass transition, rubbery, and terminal zones, calculated from the rheological spectra of each sample, dominate the mechanical properties of the frozen solid state. This implies that the terminal relaxation time of the viscoelastic properties at the compression-molding temperature dominates the properties in the solid state. The results of uniaxial stretching tests of solids indicate that the breaking stress, which corresponds to the critical stress for crack propagation in brittle materials, decreases with increasing anion size. Therefore, the lithium halide-doped samples became brittle with increasing salt anion size because crack initiation points occurred where the salts were present.