Introduction

Lower critical solution temperature (LCST)-type phase behavior in polymer solutions has been attracting a great deal of attention over the past decade. Polymers that exhibit LCST are a class of stimuli-sensitive polymers that are soluble at temperatures below the LCST and become insoluble above the LCST. Many studies have focused on LCST-type phase behavior in aqueous polymer solutions, and many theories and applications have been developed.1, 2, 3 Today, LCST-type phase behavior of polymers in nonaqueous media is also attracting interest as an opportunity for research and development of smart materials. For example, LCST polymer solutions of thermoresponsive polymer/ionic liquid systems have been reported.4, 5, 6, 7

Conventional LCST polymer solutions in organic media have critical temperatures that are higher than the boiling points (BPs) of the solvents.8, 9, 10, 11, 12, 13, 14 These reported systems can exhibit LCST-type phase behavior under some extreme conditions, for example, in sealed high-pressure cells above the BP of the solvent or using considerably high-molecular-weight polymers. Some theories have been proposed to explain or predict this type of LCST-type phase behavior.15, 16, 17 In free-volume theories, the difference between the densities or the expansion coefficients of the polymer and the solvent above the solvent's BP is proposed to cause phase separation at elevated temperatures.17 Because these conditions are inconvenient to study, a novel polymer/organic media solution that can present LCST under mild conditions is expected to be discovered. Recently, a few of these novel polymer solution systems in organic media have been reported.18, 19, 20 For example, polymers with certain unique structures or chemical compositions can exhibit LCST-type phase behavior. Fluoroalkyl-end-capped polymers present LCST behavior that may be primarily related to the balance between the oleophilic character of the adamantyl segments and the oleophobic character of the fluoroalkyl groups.18 Thermally reversible, heat-set gel-like networks in organic media with LCST characteristics were developed by combining elements of one-dimensional metal complexes and mesoscopic supramolecular assemblies.19 Poly(vinyl ether) with ionic liquid pendants of imidazolium or pyridinium salt can also undergo sensitive LCST-type phase separation in organic media.20

Herein, we report a novel polymer solution in common organic media that can present LCST behavior under mild conditions. As has been reported in our recent communication,21 the alternating copolymer of 2-chloroethyl vinyl ether (CVE) and maleic anhydride (MA) known as poly(2-chloroethyl vinyl ether-alt-maleic anhydride), henceforth P(CVE-MA), can exhibit LCST-type phase separation in n-butyl acetate (BuAc) with a critical temperature sufficiently lower than the BP of BuAc. In this paper, the detailed phase behavior of BuAc solutions of P(CVE-MA) with various molecular weights will be described, and the effects of various cosolvents on LCST behavior will also be presented.

Experimental procedure

Materials

Benzene (Nacalai tesque, Kyoto, Japan; 99.5%), toluene (Nacalai tesque; 99.5%), cyclohexane (Nacalai tesque; 99.5%), dimethylsulfoxide (DMSO; Kishida Chemical; 99%), 2-hexanone (TCI, Tokyo, Japan; >98.0%), sodium hydroxide (NaOH; Nacalai tesque; 97%), anhydrous sodium sulfate (Na2SO4; Nacalai tesque; 97%), calcium hydride (CaH2; Nacalai tesque; EP) and 2,2′-azobisisobutyronitrile (Wako, Osaka, Japan; >98%) were used as received. CVE (TCI; >97.0%) was washed three times with an equal volume of water made slightly alkaline with NaOH, dried with Na2SO4, refluxed in the presence of CaH2 for 2 h and distilled under reduced pressure. MA (TCI; >99.0%) was recrystallized in benzene. Methyl ethyl ketone (TCI; >99.0%) was refluxed in the presence of CaH2 for 2 h, distilled under reduced pressure and kept in the presence of a molecular sieve. Propyl acetate (TCI; >98.0%), BuAc (Aldrich, St Louis, MO, USA; ACS reagent; ⩾99.5%), 1,2-dichloroethane (Wako; >99.5%) and tetrahydrofuran (Wako; >99.5%) were distilled before use.

Synthesis of copolymer

MA and CVE were copolymerized by free-radical solution polymerization. Typically, specific amounts of MA, CVE, 2,2′-azobisisobutyronitrile and solvent were added to a 50 ml three-neck flask equipped with a condenser and a nitrogen gas inlet. The reaction solution was stirred for a specified time at a given temperature. The resultant polymer was precipitated in diethyl ether and dried in vacuum at 100 °C for at least 8 h.

Gel permeation chromatograph

The molecular weight of the prepared copolymer was determined by gel permeation chromatograph using two columns of TSKgel, α-3000 and α-5000, connected to an SD-8022 system (Tosoh, Tokyo, Japan) with a 1.0 ml min−1 N,N-dimethylformamide/lithium bromide flow rate (the concentration of lithium bromide was 15 mg l−1). Polystyrene standards were used to calibrate the molecular weight during measurement.

Preparation of polymer solutions

Polymer solutions in BuAc were prepared by direct dissolution of specified amounts of polymer in BuAc. The bottle containing the polymer and BuAc was placed in an oven at 60 °C for about 8 h to dissolve the polymer. The bottle containing the polymer solution was then kept in a refrigerator at about 3 °C until measurements were made.

Polymer solutions in BuAc/cosolvent were prepared as follows: a certain amount of the polymer was added to BuAc, the bottle was kept in an oven at 60 °C for about 8 h, and then a specified amount of the cosolvent was added to the prepared polymer solution. The bottle containing the polymer solution in BuAc/cosolvent was kept in a refrigerator at about 3 °C until measurements were performed.

The polymers were measured immediately after polymerization to prevent degradation, oxidation or water absorption.

Light scattering

The thermoresponsive phase behavior of the polymer solution was characterized using a light scattering technique. The scattered light intensity (Is) at a fixed angle of 90° and at various temperatures was measured using a laboratory-made apparatus equipped with an ALV/SO-SIPD detector (ALV, Langen, Germany) and a He–Ne laser (wavelength λo=633 nm).22 The sample solutions were optically purified by a Millipore filter (Millipore, Billerica, MA, USA) with a nominal pore size of 1.0 μm and then transferred into the optical tube. The heating and cooling rates of the system were both about 0.2 °C min−1. The Is of the polymer solution was measured immediately when the temperature of the system reached the desired temperature.

Results and Discussion

Synthesis of copolymers

CVE has an electron-releasing alkoxyl group and is usually used in cationic polymerization,23, 24, 25, 26, 27 whereas MA contains an electron-accepting carboxylic anhydride group. These two groups can form charge transfer complexes of donor–acceptor monomer systems in which MA and CVE act as the accepter monomer and the donor monomer, respectively.28 It is well known that alternating copolymers of vinyl ether and MA can be formed in free-radical polymerization.29, 30, 31, 32 Our previous report21 used 1H nuclear magnetic resonance to show that the molar composition of MA and CVE of the formed copolymer was 1:1, and the 13C nuclear magnetic resonance spectrum suggested that these monomers were alternately polymerized.

Copolymers with different weight-averaged molecular weights (Mw) were synthesized. The reaction conditions and the Mw values measured through gel permeation chromatograph are listed in Table 1. The various copolymers are designated P(CVE-MA)-1, -2, -3, -4 and -5. Higher molecular weight polymers (Mw=36 400, 77 200 and 92 200 g mol−1) were synthesized in methyl ethyl ketone, and polymers with lower Mw (7700 and 17 500 g mol−1) were synthesized in tetrahydrofuran with much more initiator added.

Table 1 Polymerization conditions of P(CVE-MA) with different molecular weights
Full size table

LCST-type phase behavior of polymer solution

The temperature-sensitive behavior of the polymer solution can be studied through various methods, such as turbidity, differential scanning calorimetry, light scattering measurements, viscometry and fluorescence.3 In our research, the temperature-sensitive behavior was characterized using the scattered light intensity Is at a fixed 90° angle. Figure 1 shows that the Is of the polymer solution increased abruptly as the temperature increased above a certain temperature. Above a certain polymer concentration (>0.1 wt%), the polymer solution became visually turbid, meaning that phase separation occurred above a certain temperature. This point is called the cloud point temperature (Tcp). In other words, the polymer is soluble below Tcp and insoluble above Tcp, which is characteristic of LCST-type phase behavior in a polymer solution. In our experiment, Tcp is defined from the temperature dependence of Is in Figure 1, that is, Tcp is the temperature at which the tangents to the lines in the higher and lower temperature regimes intersect (the points A, B, C and D for different polymer solutions in Figure 1).

Figure 1
figure 1

Temperature dependence of the scattered light intensity (Is) of polymer/n-butyl acetate solutions with different weight-averaged molecular weights. (◊) 0.05 wt% solution of P(CVE-MA)-2 (Mw=92 200), (▿) 0.05 wt% solution of P(CVE-MA)-1 (Mw=77 200), (▵) 0.05 wt% solution of P(CVE-MA)-3 (Mw=36 400), (○) 1 wt% solution of P(CVE-MA)-4 (Mw=17 500) and (□) 10 wt% solution of P(CVE-MA)-5 (Mw=7700). The points A, B, C and D indicate the Tcp of each solution, and the broken lines are tangent lines for evaluating Tcp.

Full size image

The value of Tcp decreased as the Mw of the polymer increased. The higher molecular weight polymers (P(CVE-MA)-1, 2 and 3) exhibited LCST-type phase behavior at a relative low Tcp (<90 °C) and low polymer concentrations. In contrast, the Tcp of the polymer P(CVE-MA)-4, with a lower molecular weight, was about 97 °C even at a relatively high polymer concentration of 1 wt%. The lowest molecular weight polymer, P(CVE-MA)-5, did not exhibit thermoresponsive phase behavior until 99 °C (the upper limit of the apparatus) even at a concentration of 10 wt%. To evaluate the Tcp of P(CVE-MA)-5 in pure BuAc, the non-solvent n-octane was added to the BuAc-polymer solution. As shown in the latter part of this report, the addition of a cosolvent can influence Tcp in a manner dependent on the solvent quality of the cosolvent. Figure 2 indicates that a small amount of n-octane (5 wt%) added to the BuAc-polymer solution did not induce thermoresponsive behavior below 100 °C. As the amount of n-octane in the mixed solvent increased over 10 wt%, LCST-type phase behavior was recognized, and further increasing the n-octane content of the mixed solvent caused the Is ratio to increase and Tcp to be reduced. In Figure 3, the Tcp values of 1 wt% polymer solutions in BuAc/n-octane are plotted against the n-octane content in the mixed solution. Tcp decreased linearly with increasing n-octane content, and the extrapolated line intersects the vertical axis at about 132 °C, which is higher than the BP of BuAc. This value could provide information about the Tcp of P(CVE-MA)-5 in pure BuAc with a polymer concentration of 1 wt%.

Figure 2
figure 2

Temperature dependence of scattered light intensity (Is) of P(CVE-MA)-5 solutions with 1 wt% polymer concentration in a solvent mixture of n-butyl acetate and n-octane. The weight percentage of n-octane is indicated in the graph.

Full size image
Figure 3
figure 3

Octane weight percentage dependence of the Tcp of the P(CVE-MA)-5 solution.

Full size image

Plots of Tcp versus polymer concentration of the BuAc-polymer solutions are shown in Figure 4. The value of Tcp increased as the Mw of P(CVE-MA) decreased. Similar molecular weight dependence of Tcp or LCST behavior has frequently been observed in aqueous polymer solutions.33, 34, 35 As mentioned above, the lowest molecular weight P(CVE-MA)-5 did not exhibit thermoresponsive phase behavior even in a 10 wt% solution, and the value of evaluated Tcp indicated in Figure 4 was derived from extrapolating the measurements shown in Figure 3. The Tcp of P(CVE-MA)-4/BuAc was much higher than that of the higher molecular weight polymers P(CVE-MA)-1, -2, and -3 and varies over a wide range of polymer concentrations from 1 to 10 wt%. The molecular weight dependence of Tcp became weaker as the Mw of the polymer increased to higher values. For the higher molecular weight polymers, Tcp initially decreased significantly as the polymer concentration increased and then converged to an almost constant value above a certain polymer concentration (>1 wt%). Similar dependences of Tcp on polymer concentration have been reported in other water-soluble polymers.36, 37, 38, 39

Figure 4
figure 4

Tcp of polymer/n-butyl acetate solutions at different polymer concentrations. Mw is (○) 7700, (□) 17 500, (▵) 36 400, (▿) 77 200 and (⋄) 92 200 g mol−1. Binodal curves for these polymers were calculated by using Equations (5) and (6) and are presented as solid (P=38), broken (P=86), dashed (P=178), thin solid (P=378) and thin broken (P=451) lines, respectively.

Full size image

It should be noted that the Tcp of P(CVE-MA)/BuAc at higher Mw is sufficiently below the BP of BuAc (126 °C) such that this LCST phase behavior is not caused by the free-volume difference between the polymer and solvent but instead probably originates from some specific interaction between the polymer and the solvent. This kind of phase behavior in polymer solutions can be conventionally described by the Flory–Huggins (F–H) lattice model based on mean-field theory in which the chemical potential of the solvent (1) and the polymer (2) are expressed by Equations (1) and (2), respectively. The molecular weights of the polymers are considered to be monodisperse.

In these equations, R is the gas constant, T is the absolute temperature, P is the number of segments per polymer chain and φ is the volume fraction of the polymer in solution. χ is the thermodynamic segmental interaction parameter between the polymer and the solvent, and its temperature dependence is conventionally assumed to follow the Shultz–Flory equation40 shown in Equation (3).

In Equation (3), Θ is the theta temperature and ψ is the entropy parameter. The lower molecular weight polymers revealed higher critical temperatures; this finding means that the apparent χ value becomes larger as the temperature increases, so the sign of the parameter ψ must be negative. According to Equation (3), Tcp at the critical point Tc can be calculated from Equation (4).40

Because the minimum value of Tcp for each polymer in Figure 4 can be regarded as its critical point, the inverse of the minimum Tcp was plotted against and displayed in Figure 5. P was evaluated as Mw/M0, where M0=204.6 is the molecular weight of the repeating unit of P(CVE-MA). The derived values of the intercept and slope of the straight line in Figure 5 can be related to the parameters in Equations (3) and (4) and give Θ=333 K and ψ=−1.62. The value of Θ=333 K means that BuAc is a good solvent for P(CVE-MA) below 60 °C (χ<1/2). Above 60 °C (χ>1/2), BuAc becomes a poor solvent, and its solvent quality for this polymer becomes poorer and poorer as the temperature increases, resulting in phase separation at a certain temperature.

Figure 5
figure 5

Plot of 1/Tcp versus ().

Full size image

Because the χ parameter varies with temperature, the chemical potentials in Equations (1) and (2) can be represented as functions of P, T and φ, that is, Δμi (φ, T, P). In the liquid–liquid phase-separated state, the chemical potentials of the solvent (i=1) and the polymer (i=2) should satisfy Equations (5) and (6), respectively.

Here, φ′ and φ″ are the volume fractions of the polymer in the phase-separated dilute and concentrated solutions, respectively. For a solution of polymer with a degree of polymerization of P, χ can be calculated at a given temperature T from Equation (3), and by solving Equations (5) and (6), the values of φ′ and φ″ for the separated phases can be obtained at that temperature. Through this procedure, binodal curves for P(CVE-MA)/BuAc of different molecular weights can be computed and are shown as lines in Figure 4. In these calculations, the P values for P(CVE-MA)-1 (Mw=77 200), -2 (92 200), -3 (36 400), -4 (17 500) and -5 (7700) were assumed to be 378, 451, 178, 86 and 38, respectively. The calculated curves reasonable reproduced the experimental phase diagram, that is, (i) the binodal lines exhibit LCST-type behavior, (ii) the lines shift to lower temperatures as the molecular weight of the polymer increases, and (iii) when Mw is high, the lines exhibit sudden decreases in the lower concentration region and level off in the higher concentration region. The value of Tcp (=132 °C) for P(CVE-MA)-5 deduced from the mixed BuAc/octane solvent at 1 wt% is also located near the calculated line. Of course, the present calculation is based on the condition that the polymer is monodisperse, although the polymers used in the experiments have a certain molecular weight distribution. The Mw distribution may be the reason for some observed quantitative discrepancies between the experimental and calculated Tcp. For example, the concentration dependence of Tcp for P(CVE-MA)-3 was broader than the calculated binodal line, and the calculated line for P(CVE-MA)-1 is located in a lower temperature regime than the experimental results. Despite these discrepancies, the overall trend of Tcp variation with polymer concentration and molecular weight can be reproduced with the F–H calculation, which suggests that the LCST-type phase diagram of P(CVE-MA)/BuAc can be described within the framework of basic F–H theory. However, the negative temperature dependence of the χ parameter cannot be explained by the original concept for χ, which was based on a van Laar-type segmental interaction between the polymer and the solvent, suggesting that some other specific interactions may exist.

Effect of addition of cosolvent on the Tcp of P(CVE-MA)/BuAc

The effect of adding various cosolvents on the Tcp of the P(CVE-MA)/BuAc solution was investigated using P(CVE-MA)-2 at a fixed polymer concentration of 0.1 wt%. The temperature dependences of Is in different solvent mixtures are shown in Figure 6. As observed in pure BuAc, the Is of the polymer solution increased sharply at a certain Tcp. The cosolvents benzene, toluene, cyclohexane and 1,2-dichloroethane all caused Tcp to decrease as the amount of cosolvent added to BuAc increased, as shown in Figures 6a–d, respectively. This phenomenon was also observed in the P(CVE-MA)-5/BuAc/n-octane solutions of Figure 2 and arises because these solvents make the solvent mixture more poor than pure BuAc. In other words, these solvents as well as n-octane should be non-solvent for P(CVE-MA).

Figure 6
figure 6

Temperature dependence of Is of polymer solutions with different amounts of various cosolvents added. (a) Benzene, (b) toluene, (c) 1,2-dichloroethane, (d) cyclohexane, (e) 2-hexanone, (f) propyl acetate and (g) dimethylsulfoxide.

Full size image

The relationship between Tcp and the added amounts of these solvents is shown in Figure 7. There exists an overall tendency that the Tcp of the polymer solution decreased as the amount of cosolvent increased. In the cases of benzene, toluene and 1,2-dichloroethane, the slopes of the Tcp versus added solvent lines were similar, and the effectiveness of these solvents at reducing Tcp is ranked as follows: 1,2-dichloroethane>benzene>toluene. Cyclohexane exhibits the highest effectiveness at reducing Tcp, meaning that aliphatic cyclohexane is the least powerful solvent of P(CVE-MA) among the solvents in Figure 7.

Figure 7
figure 7

Tcp of polymer solutions in n-butyl acetate with different amounts of the indicated cosolvents added.

Full size image

By contrast, in Figures 6e–g, 2-hexanone, propyl acetate and DMSO were the cosolvents added to the P(CVE-MA)/BuAc solutions. As 2-hexanone was added to the polymer/BuAc solution, Tcp shifted to higher temperatures until the amount of added 2-hexanone reached 15 wt%, at which no thermoresponsive behavior was observed below 100 °C (Figure 6e). Tcp shifted to higher temperature slowly as the amount of added propyl acetate to BuAc increased (Figure 6f). When DMSO was added, shown in Figure 6g, the solution remained transparent until 100 °C without any thermoresponsive behavior, even when the amount of added DMSO reached 2 wt%.

Figure 8 shows the Tcp of the polymer solution as a function of the amount of other added solvents including 2-hexanone and propyl acetate. It is clear that 2-hexanone increased the Tcp of the polymer/BuAc solution more effectively than propyl acetate. The phenomenon of Tcp increasing as larger amounts of good solvents are added has also been observed in aqueous poly(vinyl methyl ether) solutions and may arise because the interactions between the good solvents and the polymers are heavily favored over the interaction between the polymer and the original solvent.41 The Tcp in Figure 8 may increase because these cosolvents interact with the polymer more strongly than BuAc, and the phenomena in Figures 7 and 8 present clues about the ability of each solvent to dissolve this polymer. Our experiments showed that P(CVE-MA) can be dissolved in 2-hexanone and propyl acetate but cannot be dissolved in the solvents shown in Figure 7. Figure 6g shows that DMSO, which produced a pink solution after dissolving the polymer, should be the best solvent of P(CVE-MA) used in this work.

Figure 8
figure 8

Tcp of polymer solutions in n-butyl acetate with different amounts of the indicated cosolvents added.

Full size image

The slope of the curves in Figures 7 and 8 should represent the ability of each solvent to dissolve P(CVE-MA). The values of this slope for each of the tested cosolvents are listed in Table 2. The dielectric constants (ɛ) and solubility parameters (δ) of these solvents are also indicated; the former value relates to the polarity of the solvent, and the latter value is conventionally used to indicate the solubility of the polymer solution.42 These lists indicate that the δ values do not exhibit a linear relationship with the solubility of P(CVE-MA). In contrast, the value of the slope seems to increase with ɛ, with the exception of 1,2-dichloroethane. In Table 2, the polar contribution to the solubility parameter proposed by Hansen,43 δp, is also listed. δp and ɛ exhibit similar relationships with the slope of the curve. In Figure 9, the values of the slope are plotted against δp, and a good linear relationship can be seen, with the exception of 1,2-dichloroethane. Therefore, except for 1,2-dichloroethane, we can mention that a solvent's ability to dissolve P(CVE-MA) is determined by the polarity of the solvent. As reported in our recent studies,21, 44 tetrahydrofuran is a good solvent of P(CVE-MA), whereas n-hexane is a non-solvent. Some polar interaction between P(CVE-MA) and the solvent is likely to be an important factor for solubilization. This hypothesis seems reasonable because P(CVE-MA) contains polar ether and carboxylic anhydride groups and the good solvents also contain oxygen atoms in their chemical structures, whereas the non-solvents are oxygen-free non-polar solvents. The LCST behavior of P(CVE-MA)/BuAc suggests that the polar interaction between the polymer and the solvent becomes weak as the temperature increases, which results in the segregation and precipitation of P(CVE-MA). Because of chemical structure similarities, 1,2-dichloroethane would prefer to interact with the 2-chloroethyl group in P(CVE-MA) and prevent the interaction between the polymer and BuAc, which might be the reason that 1,2-dichloroethane did not follow the linear relationship in Figure 9.

Table 2 Various physical properties of the tested cosolvents
Full size table
Figure 9
figure 9

Relationship between the slope of the curve of Tcp versus the added cosolvent amounts evaluated from Figures 7 and 8 and the polar contribution to the solubility parameter (δp) of the cosolvent. (•) Cyclohexane, (○) benzene, (▪) toluene, (□) propyl acetate, (▴) 2-hexanone and (Δ) 1,2-dichloroethane. The straight line is a guide for the eye.

Full size image

Conclusion

Novel LCST-type phase behavior of polymer solutions in common organic media under mild conditions has been investigated. Alternating copolymers of CVE and MA produced an LCST-type solution when dissolved in BuAc, and the Tcp values of the higher molecular weight polymers were much lower than the boiling temperature of the solvent. As the Mw of the polymers increased, the Tcp of the polymer solution decreased. The experimental LCST-type phase diagram of this polymer solution can be reasonably described by F–H theory. The Tcp of the polymer solution in BuAc can also be tuned by adding cosolvents. Adding non-solvent decreases the Tcp, whereas adding proper good solvents increases the Tcp of the polymer/BuAc solution. This increase can be ascribed to the polar interaction between P(CVE-MA) and the solvent. Effects that strengthen the polar interaction between the polymer and the solvents are inclined to increase the Tcp of the polymer solution.

To understand the LCST behavior of P(MA-CVE)/BuAc systems, dilute polymer solution properties such as the second virial coefficient A2 and its temperature dependence measured through precise light scattering or viscosity measurements around the evaluated Θ temperature (=60 °C) will offer useful information. Further investigations of these novel polymer solutions with LCST-type phase behavior will be reported in future studies.

Finally, we should mention that this copolymer can be conveniently chemically modified because it has functional alkyl chloride and carboxylic anhydride groups. Thus, research and development can be expected to discover applications of the tunable LCST thermoresponsive behavior under mild conditions of this novel polymer solution in organic solvent.