Introduction

Cellulose, a linear polysaccharide with a β-(1 → 4)-linked d-glucose unit and the most abundant natural polymer on earth from green plants and some bacteria (Wang et al. 2012), has been widely applied in packaging, durable goods, electronics, biomedical uses, etc. (Gericke et al. 2013; Reddy et al. 2013) for its renewability, biocompatibility and outstanding mechanical properties (Diddens et al. 2008). Due to the numerous inter- and intramolecular hydrogen bonds (H-bonds) (Pinkert et al. 2009) and amphiphilicity of repeat units (Lindman et al. 2010; Medronho et al. 2012), cellulose can neither dissolve in common solvents nor melt at elevated temperature even above the decomposition temperature (Calahorra et al. 1989), which makes this fantastic material difficult to process. Although chemical modifications such as esterification and etherification provide processable derivatives (Klemm et al. 2005), direct utilization of natural cellulose remains a big challenge. The development of efficient solvent systems has led to breakthrough progress in wet processing (Johnson 1969; Dupont 2003; Cai and Zhang 2005; Swatloski et al. 2002), but thermal processing methods, which can produce bulky materials with various shapes conveniently and efficiently, are preferred in the modern plastics industry. Unfortunately, only limited progress has been achieved. Schroeter and Felix (2005) fabricated a transparent cellulose disk with a diameter of 3 mm and thickness of 20 μm via strong mechanical shear, high pressure and large-dose laser radiation. By means of back pressure-equal channel angular press (BP-ECAP) technology, Zhang et al. (2012) successfully prepared bulky plastics with a storage modulus up to 1.0 GPa from cellulose powders under an ultrahigh pressure of above 100 MPa. However, both technologies need specific instruments and are highly inefficient for industrial applications.

According to the reptation theory proposed by Edwards and de Gennes, the motion of a whole macromolecular chain leading to flowing comprises the segmental motion, which is directly related to the glass transition behavior. Since the cellulose starts to oxidize and decompose at ~473 K (Mamleev et al. 2007), and there are small variations in the heat capacity during the glass transition of cellulose (Szcześniak et al. 2008), it is not a trivial task to measure the glass transition temperature (T g) of dry cellulose directly. Recently, the experimental T g values of dehydrated cellulose and ball-milled dry cellulose were reported to be 357 and 433 K, respectively, by differential scanning calorimetry (DSC) and dynamic mechanical analysis (DMA) (Roig et al. 2011; Paes et al. 2010). To filter out the thermal decomposition, Kim et al. (2013) used a step scan mode DSC to measure the T g of 573 K for cellulose films . Obviously, the above values are discrepant and difficult to understand. From the predictive relationship between T g and solvent content in solvent-containing samples, T g of pure cellulose has been extrapolated to be 493–523 K (Szcześniak et al. 2008; Bizot et al. 1997), which is consistent with the simulation result through specific volume (Chen et al. 2004). Setting the T g of pure cellulose as a generally accepted value of ~500 K, the relationship between T g and water content has been successfully described by the Gordon-Taylor/Kelley-Bueche (G-T/K-B) equation (Hancock and Zografi 1994), Kaelbe’s approach (Szcześniak et al. 2008) and Couchman’s equation (Paes et al. 2010). However, the assumption of “no specific interactions between components” in the G-T/K-B equation based on free volume could not be satisfied because of the existence of powerful H-bonding networks (Mazeau and Heux 2003). It is well known that strong H-bonds exist in cellulose, and their confinement on molecular chain motion leads to a high T g. Though the intermolecular H-bonds in hydrated cellulose can be interrupted by water, the effect is limited, especially at low water content. Experimental T g of cellulose with 5.3 and 7 % water content shows a discrepancy from the prediction of Kaelbe’s approach, which is attributed to the strong H-bonding interactions between the chains (Szcześniak et al. 2008); that is, the effect of on T g is considerable, but it cannot be reflected in the free volume-based equations. Therefore, to differentiate the effects of H-bonding and free volume on the T g of cellulose is urgently needed for understanding the basics of glass transition behavior.

Ionic liquids have been found to be effective plasticizers for polymers such as poly(methyl methacrylate) (Scott et al. 2002) and poly(vinyl chloride) (Choi et al. 2011, 2014). Their influences on the thermal processability and the relationship of dissolution and gelatinization of starch were also reported (Sankri et al. 2010; Mateyawa et al. 2013). In this work, an ionic liquid of 1-butyl-3-methylimidazolium chloride (BmimCl) was used as a plasticizer to prepare ionic liquid plasticized cellulose (IPC) materials. Due to the aggressive ability to break H-bonds and crystal structures of cellulose, BmimCl is better than water to improve the mobility of cellulose chains. Furthermore, its nonvolatility broadens the processing temperature window, which is helpful to the thermal processability of cellulose. IPC materials with 25–70 wt% BmimCl were prepared using direct thermal processing, and their glass transition mechanism was investigated via a linear form of the G-T/K-B equation and a prediction about the percolation transition of H-bonding networks. Finally, the contributions of free volume and H-bonding networks to the glass transition of cellulose were discussed.

Experimental section

Materials and sample preparation

Microcrystalline cellulose (MCC) was purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China) with a crystallinity of 66.9 %. About 1.5 wt% moisture was confirmed by thermogravimetry analysis and removed at 125 °C under vacuum for 48 h. 1-Butyl-3-methylimidazolium chloride (BmimCl, 99 %) was purchased from Dibo regent Co., Ltd. (Shanghai, China) and used as received.

BmimCl plasticized cellulose (IPC) materials with 25–70 wt% BmimCl were first hot compounded on a ZG-76 two-roll mill (Zhenggong Mechanical and Electrical Equipment Technology Co., Ltd., Dongguan, China) at 130 °C for 15 min with the two roll speeds set to 8.6 and 10 rpm. Then, they were calendared into film samples with a thickness of about 0.25 mm for FT-IR, DSC and XRD measurements. The thick film samples (~1 mm) with 30–50 wt% BmimCl for mechanical tests and fracture morphology measurement were compression-molded on a PL-6170C plate vulcanizing machine (Bolon Precision Testing Machines Co., Ltd., Dongguan, China) under 130 °C and 10 MPa for 5 min, with a 3-min preheating period. For evaluating the thermal processability, the IPC materials with 40 wt% of BmimCl were also injection-molded using a TY-200 vertical injection machine (Tayu Machinery Co., Ltd., Hangzhou, China). The barrel-zone temperatures of the injection-molding machine were set at 160, 155, 155, 150 and 150 °C, the injection pressure and rate were 13 MPa and 10 cm3/s, and the cooling time was set at 60 s for sufficient solidification of samples. To check the thermal reprocessability of the IPC materials, the sample with 40 wt% of BmimCl was prepared for five cycles of repeated thermoprocessing on the two-roll mill and cooled to room temperature.

All samples were sealed immediately and naturally cooled to room temperature after processing and kept in a drier with argon before use.

Measurements

Fourier transform infrared (FT-IR) analysis was conducted on a spectrometer (Bruker V70, Germany) with a reflectance accessory. A scan range from 4,000 to 600 cm−1 was chosen with a resolution of 4 cm−1, and 16 scans were performed for each sample. Tensile tests of IPC with a dimension of about 10 × 25 × 1 mm3 were conducted on a CMT4104 universal testing machine (SANS, Shenzhen, China) with 25 mm gauge length and a cross-head rate of 10 mm/min. At least six samples of each IPC were tested, from which the mean values and standard deviations were calculated. Field emission scanning electron microscopy (SEM, Sirion 200, The Netherlands) was used to determine the morphology of the fracture surface of IPC materials. A thin layer of platinum was sputtered onto the samples after breaking off in liquid nitrogen prior to SEM imaging. Wide-angle X-ray diffraction (XRD) analysis was carried out on an X-ray diffractometer (X’Pert PRO, The Netherlands) by a reflection method using a Cu Kα target at 40 kV, and the diffraction angle (2θ) ranged from 5° to 60°. Crystallinity was calculated through the ratio of the area of crystalline peaks to total area. The thermal analyses of IPC materials were conducted with a differential scanning calorimeter (DSC, Q2000, TA Instruments) at a rate of 20 K/min in nitrogen. About 10 mg of IPC materials was subjected to an initial temperature ramp to 433 K and kept for 15 min to completely remove any residual absorbed water and then ramped down to 193 K. Eventually, the IPC materials were heated to 393 K. To clarify the transition terrace, the temperature ranges for IPC materials and pure BmimCl were given in the DSC curves from 223 to 333 and 213 to 333 K, respectively. The T g of IPC materials was determined by the deflection point of the second heat flow curves. The T g for pure BmimCl was 225 K according to the reference (Yamamuro et al. 2006) and also confirmed in this work.

Results and discussion

Thermal processability, mechanical properties, morphology and formation mechanism of IPC materials

In general, the processability of IPC materials increases with BmimCl content. IPC materials with BmimCl content <25 wt% were too brittle to process, while IPC materials with BmimCl content more than 70 wt% were not geometrically stable. Therefore, IPC materials with 25–70 wt% BmimCl were successfully fabricated through calendaring, and thick film samples (~1 mm) with 30–50 wt% BmimCl were compression molded. Figure 1a and b shows typical photos of the calendared film and the hot-compressed plate of IPC with 40 wt% of BmimCl. It can be seen that the calendared film with 0.25 mm thickness is uniform and partially transparent, and the hot-compressed plate with dimensions of 70 × 50 × 1 mm3 is successfully fabricated. Further, typical impact (80 × 10 × 4 mm3) and tensile (75 × 10 × 2 mm3) samples without surface defects can be injection molded, as shown in Fig. 1c. These indicate that the IPC materials have sufficient processability for direct thermal processing.

Fig. 1
figure 1

Photographs of IPC with 40 wt% BmimCl processed by calendaring (a), compression molding (b) and injection molding (c)

Full size image

FT-IR spectra of BmimCl, MCC and IPC with 30 wt% BmimCl are shown in Fig. 2. For BmimCl, absorption peaks at 3,095, 2,865, 1,563 and 1,463 cm−1 are attributed to stretching vibrations of unsaturated C–H, saturated C–H, C–N and C=C in the imidazole cation, respectively. For MCC, the broad absorption peak from 3,540 to 3,180 cm−1 is attributed to O–H stretching vibrations of various forms of free and hydrogen-bonded hydroxyl groups. The absorption peak ranging from 1,160 to 1,030 cm−1 is attributed to C–OH stretching vibrations of primary and secondary hydroxyl, and peaks at 1,430 and 1,315 cm−1 are contributed by scissor bending and twisting of –CH2 in the main chain. Compared with MCC and pure BmimCl, IPC with 30 wt% BmimCl shows no new absorption peaks. This is because BmimCl is a nonderivative solvent for cellulose (Swatloski et al. 2002), there is no chemical reaction between cellulose and the ionic liquid, BmimCl only acts as a plasticizer to improve the processability of MCC, and the IPC materials are stable during thermal processing.

Fig. 2
figure 2

FT-IR spectra of BmimCl, MCC and IPC with 30 wt% BmimCl

Full size image

The mechanical properties of IPC materials with 30–50 wt% BmimCl are listed in Table 1. Due to the plasticization effect of BminCl, the tensile strength and Young’s modulus decrease almost one order of magnitude as the BmimCl content increases from 30 to 50 wt%, i.e., from 200.7 and 8.3 MPa to 32.0 and 1.0 MPa, respectively. Although these data are far from those for native cellulose (138 and 13–17 GPa for cellulose I, 88 and 9 GPa for cellulose II, 10 GPa and 800 MPa for amorphous cellulose) (Huber et al. 2011; Nishino et al. 1995), these materials show a potential for making bulk materials, especially with a low BmimCl content. Meanwhile, their elongation at break increases from 6.4 to 14.2 %, which is indicative of a little enhancement of toughness. It is worth noted that the IPC with 40 wt% BmimCl had the highest elongation at break. With the addition of BmimCl, the plasticization effect increases the free volume and weakens the H-bonds in the IPC materials, leading to an increase in elongation at break. However, excess plasticizer acts as an activation cell for crack formation (Jacobsen and Fritz 1999), which makes the IPC easier to break and results in a decrease in elongation.

Table 1 Mechanical properties of IPC materials with different contents of BmimCl and IPC with 40 wt% of BmimCl after five reprocessed times (denoted as 40 %-5R)
Full size table

The fracture surface morphologies of IPC materials are shown in Fig. 3. For the IPC with 30 wt% BmimCl as shown in Fig. 3a and d, the boundaries of the spliced MCC domains are sharp and clear. When the BmimCl content is increased to 40 wt%, the plasticized MCC domains with obtuse and fuzzy boundaries can be observed (Fig. 3b and e). For IPC with 50 wt% BmimCl (Fig. 3c, f), the MCC domains are merged together by the swollen MCC boundaries, which form the continuous matrix. Thus, a formation mechanism of IPC during thermal compression can be illustrated in Scheme 1. As a good solvent for cellulose, BmimCl disaggregates MCC particles from the outside, forming MCC domains with swollen amorphous boundaries. The chain mobility in the outside swollen boundaries is enhanced by the interruption of H-bonds and increase in free volume. The disentanglement and sliding of the cellulose chains in the swollen boundaries occur under heating and shearing. As a result, the effective interchain penetration of the cellulose chains in the amorphous boundaries merges individual MCC domains. With increasing BmimCl content, the mobility of cellulose chains is further enhanced, leading to improved processibility.

Fig. 3
figure 3

SEM images of cryofractured IPC materials with BmimCl content of 30 wt% (a, d); 40 wt% (b, e); 50 wt% (c, f)

Full size image
Scheme 1
scheme 1

Formation mechanism of IPC materials during thermal processing

Full size image

Based on the plasticization effect of ILs, this is the first report on direct thermal processing of natural cellulose using industrial equipment including a two-roll mill, plate vulcanizing machine and injection machine. More interestingly, the IPC materials can be processed repeatedly. Due to the oxidation and decomposition of cellulose, the hue of IPC materials became a bit darker with increased reprocessed times. However, the mechanical properties of IPC materials remained good after reprocessing five times. Hence, from Table 1, it can be seen that the IPC with 40 wt% BmimCl retains a Young’s modulus of 68.1 MPa, tensile strength of 3.1 MPa and elongation of 13.6 %. Recently, Wang et al. (2013) prepared a hydrogel with NaOH/urea aqueous solution as solvent and fabricated the cellulose plastic with excellent mechanical performance under 90–190 °C and pressure of 60 MPa by thermal compression accompanied by removal of water. It is clear that this cellulose plastic cannot be reprocessed, which is different from our IPC materials.

Crystalline structure of IPC materials

The normalized XRD spectra of MCC and IPC materials with different BmimCl contents are shown in Fig. 4. All IPC materials have the characteristic peaks of cellulose I crystal in natural cellulose at 14.8°, 16.2° and 22.7° for (\(1\bar{1}0\)) (110) and (200) planes, respectively (Nishiyama et al. 2002; French 2014). Also, a clear peak at 20.6° for (012) and (102) planes in cellulose Iα is observed in MCC and IPC materials with low BmimCl content, and no characteristic peaks of cellulose II crystal exist in regenerated cellulose (Duchemin et al. 2009; Han and Yan 2010). Although having a melting point of 68 °C, BmimCl is a liquid once heated at high temperature and cooled to room temperature due to its strong tendency for supercooling (Galiński et al. 2006). Also, the XRD curves of IPC materials only show the characteristic peaks of cellulose I even at high BmimCl contents up to 70 wt%. Thus, the crystalline peaks of the XRD curves are attributed to the cellulose crystalline phases in the IPC materials. As shown in the inset of Fig. 4, the amorphous and crystalline phases in MCC and IPC materials are distinguished quantitatively via the peak fitting technique, and the crystallinity of cellulose (φ) is achieved. Figure 5 shows the crystallinity of MCC and IPC materials with different BmimCl contents. The crystallinity decreases almost linearly with increasing BmimCl content. With a fixed intercept of 66.9 % (crystallinity of MCC), the BmimCl content dependence of crystallinity can be fitted by the equation below:

$$\varphi = - 0.557c + 0.669$$
(1)
Fig. 4
figure 4

Normalized XRD spectra of MCC (a) and IPC materials with BmimCl content of 30 wt% (b), 40 wt% (c), 40 wt% reprocessed five times (d), 50 wt% (e), 60 wt% (f) and 70 % (g), respectively. Inset is the peak fitting of MCC

Full size image
Fig. 5
figure 5

Crystallinity of IPC materials as a function of BmimCl content

Full size image

Even at high BmimCl content up to 70 wt%, the IPC still retains a crystallinity of 23.5 %. It is noted that the dependence of BmimCl on the crystallinity of cellulose is different from the independence of water (Hancock and Zografi 1994; Szcześniak et al. 2008), indicating the stronger plasticization effect of ionic liquid on the processability of cellulose. Furthermore, the regeneration of cellulose II did not occur in the IPC materials during thermal processing, while the recrystallization to cellulose I from amorphous phases might happen as reported by Schall (Samayam et al. 2011). Additionally, the crystallinities of IPC with 40 wt% BmimCl are 44.6 and 38.8 % before and after reprocessing five times. Like in the discussion of the hue change during reprocessing, this decrease is attributed to the oxidation and decomposition of the cellulose.

Glass transition behavior of IPC materials

The DSC curves of MCC and IPC materials with different BmimCl contents are shown in Fig. 6. For MCC, no thermal transition signal can be observed in the experimental temperature range. For pure BmimCl, a T g of 227.0 K is in good agreement with the 225.0 K °C reported by Yamamuro et al. (2006). The IPC with 30 wt% BmimCl shows a tiny glass transition around 281.7 K. With increasing BmimCl content, the transition shifts to lower temperature, suggesting that the cellulose segments move more easily because of the plasticization effect of BmimCl. The change of heat capacity of IPC materials during glass transition also becomes greater, which makes the transition terrace clearer. All IPC materials show only one transition terrace, which indicates that MCC and BmimCl are compatible (Hancock and Zografi 1994). Interestingly, the transitions are also observed in cooling runs and subsequent cycles. The reversible glass transition behavior in these IPC materials is different from that in wet-processed samples (Shi et al. 2011).

Fig. 6
figure 6

DSC curves of MCC (a), IPC materials with BmimCl content of 30 wt% (b), 40 wt% (c), 50 wt% (d), 60 wt% (e), 70 wt% (f) and pure BmimCl (g)

Full size image

The G-T/K-B equation (Gordon and Taylor 1952) (Eq. 2) simplified by the Samba-Boyer rule (Eq. 3) is usually used to describe the glass transition temperature (T g,m) in compatible amorphous polymer blends or plasticized systems (Hancock and Zografi 1994; Schneider 1989).

$$T_{\text{g,m}} = \frac{{w_{\text{p}} \cdot T_{\text{g,p}} + K \cdot w_{\text{c}} \cdot T_{\text{g,c}} }}{{w_{\text{p}} + K \cdot w_{\text{c}} }};\quad K = \frac{{\rho_{\text{p}} \cdot\Delta \alpha_{\text{c}} }}{{\rho_{\text{c}} \cdot\Delta \alpha_{\text{p}} }}$$
(2)
$$\Delta \alpha \cdot T_{\text{g}} \approx \, 0.113; \, K = \frac{{\rho_{\text{p}} \cdot T{\text{g,p}}}}{{\rho_{\text{c}} \cdot T{\text{g,c}}}}$$
(3)

where T g,p and T g,c are the glass transition temperatures of the plasticizer and pure polymer, i.e., cellulose in this work; w p and w c represent the mass fraction of each component. K is a constant decided by the density (ρ) and the change of thermal expansivity (Δα) of components at T g. Herein, ρ p and ρ c are 1.08 and 1.54 g/cm3, respectively.

However, the glass transition behavior of a semicrystalline polymer originates from the amorphous region in which the plasticizer is located (Zografi et al. 1984). Therefore, w p should be recalculated with the apparent mass fraction of the plasticizer (c), the crystallinity φ of cellulose, which can be written as

$$w_{\text{p}} = \frac{{m_{\text{p}} }}{{m_{\text{p}} + m_{\text{c}} \cdot (1 - \varphi )}} = \frac{c}{1 - \varphi + c\varphi }$$
(4)

where m p and m c represent the masses of plasticizer and cellulose, respectively.

In addition, the strong intermolecular H-bonding interaction between MCC and BmimCl gradually destroyed the crystalline regions of cellulose with increasing BmimCl content, so the crystallinity of cellulose in IPC materials should be taken as a variable parameter.

As mentioned above, the G-T/K-B equation is based on the free volume assumption, and the existence of H-bonds leads to a high T g. Compared to water, BmimCl has a greater ability to interrupt H-bonds in cellulose and a higher actual content in the plasticized system; thus, a weaker H-bonding level gives rise to a lower T g in IPC materials. Since the strength of the H-bonds depends on the BmimCl content, the continuous H-bonding networks will be broken at a critical BmimCl content. That is, there is a percolation transition for the continuous H-bonding networks. Therefore, it is not reasonable to take all data equally in a single fitting line; otherwise, the result would be inaccurate with systematic error caused by the H-bonding interactions. To differentiate the effects of free volume and H-bonding interactions on T g, a linear form of the G–T/K–B equation is proposed for further investigation. Rather than an intangibe fitting, which considers every datum equally, a linear form is a more reasonable method to determine whether the experimental data are consistent with the physical meaning or not. Consequently, the influence of H-bonding interactions on T g can be visualized.

Through a mathematical transformation of the classic G-T/K-B equation, a linear correction form with the apparent content and crystallinity can be written as:

$$\frac{{T_{\text{g,p}} }}{{T_{\text{g,m}} }} = b \cdot \left[ {\frac{{\rho_{\text{p}} \cdot (T_{\text{g,p}} - T_{\text{g,c}} )}}{{T_{\text{g,c}} }} \cdot \frac{1}{{\rho_{\text{p}} + z}} + 1} \right]$$
(5)
$$\, z = \frac{{w_{\text{p}} \cdot \rho_{\text{c}} }}{{1 - w_{\text{p}} }} = \frac{{c \cdot \rho_{\text{c}} }}{(1 - c) \cdot (1 - \varphi )}$$
(6)

where b is a correction factor with respect to the deviation from the ideal G-T/K-B correlation. The closer the intercept b is to 1, the better the experimental data obey the theoretical equation.

Experimental data points are plotted in the new coordinate system of T g,p/T g,m versus 1/(ρ p + z) in Fig. 7. Obviously, the experimental data with 1/(ρ p + z) from 0.174 to 0.414 (w p from 74.9 to 55.2 wt%; c from 70 to 30 wt%) in the left part of the plot have a good linear relationship. However, the experimental data for the IPC materials with 1/(ρ p + z) more than 0.414 (w p < 55.2 wt%; c < 30 wt%) in the bottom right part of Fig. 7 clearly deviate negatively from the fitting line. This deviation is attributed to the interactions of continuous H-bonding networks in IPC materials. The onset of deviation, 0.414 of 1/(ρ p + z) (w p is 55.2 wt%; c is 30 wt%), is a critical content where the transition of free volume and percolation of H-bonding networks happen simultaneously. Below the critical BmimCl content, the H-bonding interaction in IPC materials is still strong enough to form continuous networks and restricts the segmental motion of cellulose even if the free volume provides enough space for the segmental relaxation. Actually, T g,m is the dissociation temperature of the H-bonding networks. By contrast, above the critical BmimCl content, the temperature for destroying the continuous H-bonding networks is lower than that for the segmental relaxation from the free volume contribution. Thus, T g,m matches the prediction of the G-T/K-B equation well.

Fig. 7
figure 7

Relationship between T g,p/T g,m and 1/(ρ p + z) for the linear correction G-T/K-B equation

Full size image

The left part of the plot in Fig. 7 can be expressed as

$$\frac{{T_{\text{g,p}} }}{{T_{\text{g,m}} }} = 0.963 \times \left( { - 0.385 \times \frac{1}{{\rho_{\text{p}} + z}} + 1} \right)$$
(7)

As expected, the intercept b of 0.963 is very close to unity. From the slope of the straight line, a glass transition temperature of pure cellulose, T g,c, can be estimated as 353 K, which is ascribed to the contribution of free volume. This value is close to 357 K of pure dehydrated cellulose (Roig et al. 2011).

For comparison, a direct fitting curve of all IPC materials (dashed curve) and the curve fitted by our method using partial IPC materials with w p from 74.9 to 55.2 wt% (c from 70 to 30 wt%, solid curve) are shown in Fig. 8. Without the interruption of data points with continuous H-bonding networks, the solid curve fits the data points with high BmimCl contents better than the dashed curve, which is supported by the higher correlation coefficient R. That is, R 2 = 0.96 versus R 2 = 0.93.

Fig. 8
figure 8

Direct fitting curve of samples with all BmimCl contents (dashed line) and the curve fitted using samples with BmimCl from 30 to 70 wt% contents (solid line)

Full size image

Up to now, the quantitative relationship between the dissociation temperature of H-bonding networks (T H) and the H-bond density is still unclear. Recently, Brovchenko et al. (2011) found that the dissociation temperature of H-bonding networks depended linearly on the average number of H-bonds by computer simulation. Therefore, the relationship between T H and w p can be ratiocinated to be linear. Since IPC materials with c < 25 wt% (w p < 41.4 wt%) are difficult to prepare, we only achieved three points below the critical value to roughly forecast the dependence of w p on T H. The linear fitting result is shown in Fig. 9, which can be expressed as:

$$T_{\text{H}} = - 373.7w_{p} + 455.8$$
(8)

The intercept of about 456 K is considered to be T H of pure cellulose, i.e., the actual T g of cellulose.

Fig. 9
figure 9

Relationship between the dissociating temperature of continuous H-bonding networks and w p

Full size image

With the two functional relations in Eqs. 7 and 8, a phase diagram of apparent T g,m as a function of w p is clearly defined in Fig. 10a. For better perception and easier application, the phase diagram is transformed with the x-axis of c, as shown in Fig. 10b. According to the basics of segmental mobility, the apparent T g,m of IPC materials is decided by the dominant contribution of either the free volume or the specific interaction of continuous H-bonding networks. Based on the contribution of the two factors on glass transition, the phase diagram can be divided into four regions: (I) the rubber state with sufficient free volume and without H-bonding networks, (II) glass state without sufficient free volume and H-bonding networks, (III) glass state with insufficient free volume and continuous H-bonding networks, and (IV) glass state with sufficient free volume and continuous H-bonding networks. For pure cellulose, the free volume provides enough space for segmental relaxation at 353 K, while the continuous H-bonding networks are destructed at 456 K. By adding BmimCl, both transition temperatures decrease, but the latter one decreases faster. The apparent T g is gradually close to the predicted value from the G-T/K-B equation, and finally they coincide. The intersection at w p of 46.2 wt% (c of 30 wt%) with the temperature of about 281 K for both transitions indicates a mechanism transform from the H-bonding network control to the free volume control in the glass transition behavior. Further, region (I) is the thermal processing window in which the high temperature and sufficient plasticizer break the continuous H-bonding networks and provide sufficient free volume, so that the enhanced segmental mobility is useful for processing the IPC materials. Besides, the cellulose chains are restricted the most in region (III) by the continuous H-bonding networks and insufficient free volume, thereby imparting the materials with excellent mechanical properties and stability. From the phase diagram, we can optimize the thermal processing and modulate the properties of the cellulose materials by adjusting the temperature and plasticizer content synergistically.

Fig. 10
figure 10

Phase diagrams of apparent T g,m as a function of w p (a) and c (b). The four regions in the diagrams represent: (I) rubber state with sufficient free volume and without H-bonding networks, (II) glass state without sufficient free volume and H-bonding networks, (III) glass state with insufficient free volume and continuous H-bonding networks, and (IV) glass state with sufficient free volume and continuous H-bonding networks

Full size image

Conclusion

IPC materials with 25–70 wt% BmimCl have been successfully prepared by thermal processing. There is no chemical reaction between cellulose and the ionic liquid, BmimCl only acts as a plasticizer to improve the thermal processability of MCC, and the IPC materials can be processed repeatedly. Due to the plasticization effect of BminCl, the tensile strength and Young’s modulus of IPC materials decrease with increasing BminCl content, and the IPC with 40 wt% BmimCl has the highest elongation at break. Additionally, the crystallinity of cellulose in IPC materials decreases linearly with BmimCl content. Even at high BmimCl contents up to 70 wt%, the IPC still retains a crystallinity of 23.5 % compared with 66.9 % of MCC. Interestingly, all IPC materials show only one glass transition, and their T g,m values decrease with BmimCl content.

A linearized form of the G-T/K-B equation containing apparent BmimCl content (c) and crystallinity of cellulose (φ) has differentiated the effects of free volume and H-bonding interactions on T g. The proposed 353 and 456 K are related to the transitions of free volume and percolation of H-bonding networks in cellulose, respectively. Subsequently, the phase diagram with four regions has been plotted for the IPC materials, which is useful for the optimization of thermal processing and modulation of the properties of the cellulose materials. Further, the phenomenological method used in this work is expected to apply to other polymer systems with H-bonding interactions.