Effect of mixed solvent on solution properties and gelation behavior of poly(vinyl alcohol)

European Polymer Journal 45 (2009) 1158–1168

Contents lists available at ScienceDirect

European Polymer Journal journal homepage: www.elsevier.com/locate/europolj

Effect of mixed solvent on solution properties and gelation behavior of poly(vinyl alcohol) Shi-Jie Hong a, Po-Da Hong a,*, Jyh-Chien Chen a, Kan-Shan Shih b a b

Department of Polymer Engineering, National Taiwan University of Science and Technology, Taipei 10607, Taiwan School of Dentistry, National Defense Medical Center, Taipei 114, Taiwan

a r t i c l e

i n f o

Article history: Received 26 February 2008 Received in revised form 17 December 2008 Accepted 5 January 2009 Available online 13 January 2009

Keywords: Poly(vinyl alcohol) Solution properties Aging effect Phase separation Gelation kinetics

a b s t r a c t In this work, the static and dynamic light scattering measurements were used to investigate the solution properties and the aging effects on PVA/DMSO/water ternary system in dilute region at 25 °C. It was found that the phase separation and aggregate behavior occurs rapidly and obviously when DMSO mole fraction (X1) in the solvent mixture is between 0.2 and 0.33, especially at 0.25. In this solvent composition range, a broad peak which indicates phase separation and chain aggregation can be observed from static light scattering measurement. However, when DMSO mole fraction is increased to 0.37, no such peak is present. For this ternary system, the gelation mechanism and the relationship between the phase separation behavior and the gelation of the formed physical gels were also investigated through the gelation kinetic analyses in the dilute and semi-dilute region. It is concluded that the cononsolvency effect in the dilute solution is not the sole origin that affects the phase separation, aggregation, and gelation behavior for the ternary system in a higher polymer concentration range. The hydrodynamic factors such as the higher viscosity and slower polymer chain diffusion that are resulted from higher polymer concentration should be also considered. Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction Poly(vinyl alcohol) (PVA) exhibits important commercial applications due to its water-soluble and biocompatible properties and nowadays it is widely applied to the hydrogels as biomaterials. Mixtures of DMSO and water have also been used in biological applications ranging from antibacterial activity to membrane permeability [1]. On the other hand, PVA gels are also well known in preparing high-modulus PVA fibers by the drawing of gel films from the crystallization of semidilute solutions [2]. It is generally recognized that the degree of polymer–solvent interaction would affect significantly the physical and chemical properties of polymer solutions. It is also accepted that the formation of co-solvent complex could play an important role on the preferential adsorption phenom-

* Corresponding author. Tel.: +886-2-27376539; fax: +886 2 27376544. E-mail address: [email protected] (P.-D. Hong). 0014-3057/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.eurpolymj.2009.01.005

enon and the thermodynamic properties of polymer solutions. In order to study the solution properties of a ternary system PVA/DMSO/water, the phenomenon of cononsolvency which is related to the molecular interaction is needed to be addressed. Tacx et al. [3] gave a possible explanation for the different affinity between the two solvents and PVA. It was concluded that PVA could be molecularly dissolved in DMSO and the formed solutions would not age. However, water could only be a moderate good solvent for PVA, the formation of aggregates would be easy to occur. From the viewpoint of chemical structures, the dipole moments and the donor-acceptor properties of each component could clarify these molecular interactions which would change with the compositions of the solvent mixture and the formation of a third component i.e., cosolvent complex. In our previous study [4], it was concluded that the formation of a third component would give rise to various phenomena of preferential adsorption coefficient for dilute PVA/DMSO/water ternary solution. It is

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

therefore important to further discuss on the solution properties, the phase separation, coil aggregate and gelation behaviors in the semi-dilute region with different concentration and composition of co-solvent mixture. Takahashi et al. [5] reported that the fastest gelation and phase separation rate for a PVA/DMSO/water ternary system would occur at the volume fraction of DMSO UDMSO = 0.6. It was ascribed to the cononsolvency effect due to the formation of stable DMSO/(water)2 complexes at UDMSO = 0.66. They also pointed out that spinodal decomposition type of phase separation would occur before gelation for the PVA solution with UDMSO = 0.6, resulting in polymer-rich and polymer-poor phases. The formation of junction points (PVA crystallites) of gel network would occur mainly in the polymer-rich phase, purposing that overall gelation rate should be affected by the phase separation process. Kaji et al. [5–11] have reported a serious of studies on PVA gels formed in DMSO/ water mixtures by various scattering techniques in a broad composition range (UDMSO = 0.3–0.8). It was found that the mixed solvent would exhibit the poorest affinity to PVA at DMSO, UDMSO = 0.6. Takeshita et al. [9] also confirmed the spinodal decomposition type of phase separation actually would occur in the formation process of the opaque gel using time-resolved light scattering measurements. Wolf et al. [12] suggested that cononsolvency would usually occur either when the two solvents were close to demixing or when the complex was formed. On the other hand, Schild et al. [13] reported that perturbation of solvent–solvent interaction parameter would not be the origin of cononsolvency and the gel collapse transition in a ternary system. It is believed that the phase separation, the aggregation behavior and the gelation kinetics of PVA solutions might be strongly dependent on the composition of co-solvent mixture and the PVA concentration. However, the composition usually studied at UDMSO = 0.6 is actually nearer the 1/3 ratio, corresponding to XDMSO = 0.25. The aim of this study was to investigate if the cononsolvency would be the origin of phase separation and gelation in the compositions near the formation of cononsolvency.

1159

Millipore filter to remove dusts before using. The as-received PVA powder was purified by the following method before using. First, a 10 gL1 PVA aqueous solution was prepared at 95 °C, and then the homogeneous PVA solution was filtered through a 0.45 lm Millipore filter to remove the dusts. Finally, the purified PVA sample was obtained from the precipitation of the solution by adding dust-free acetone. The precipitated PVA was collected and dried in vacuum oven at 70 °C for 3 days. Mixed solvents with various compositions of DMSO/water mixture, 0/10, 1/9, 2/8, 3/7, 4/6, 5/5, 6/4, 6.6/3.4, 7/3, 8/2, 9/1, 10/0 (V/V) were used to prepare PVA solutions. Four dilute solutions for every studied composition of co-solvent mixture were prepared at least. Homogeneous PVA solutions were obtained by heating at 95 °C for 5–6 h, and then sealed in the Pyrex tube. The solutions were heated to become homogeneous again, and then quenched rapidly to 25 °C for dynamic light scattering measurements. The dynamic light scattering was conducted on a Malvern series 4700 apparatus equipped with the 7132 multiple-s autocorrelator recording on 128-channels. The light source was an Argon ion laser, operating at power of 20– 50 mW with a wavelength of 514.5 nm. The light was vertically polarized and focused on the sample cell through a temperature-controlled chamber filled with dust-free distilled water. The measurements were carried out at the scattering angle h, from 30° to 120°. The data was collected from the intensity autocorrelation function. Given that the polymer chain obeys Gaussian statistics, the measured normalized intensity correlation function g(2)(t) is related to the normalized electric correlation function g(1)(t) through Siegert relation [12–15]:

g ð2Þ ðtÞ ¼ 1 þ bjg ð1Þ ðtÞj2

ð1Þ

where b is an optical factor related to the coherence area viewed by the detector. Some useful parameters related to the solvent quality such as hydrodynamic radius Rh, diffusion coefficient D, and second virial coefficient Kd can be determined by this operation. 2.3. Kinetics of gelation

2. Experimental 2.1. Materials The PVA powder (Mw = 124,000–186,000, from Aldrich) with a high degree of hydrolysis (about 99.9%) was used in this work. The solvent, dimethylsulfoxide DMSO, was purchased from Aldrich and used as received. The PVA/DMSO/ water ternary solutions with various compositions were prepared by dissolving PVA in mixed DMSO/water solvent. The combined mixtures were heated at 95 °C then cooled to 25 °C. The PVA concentration was controlled in both dilute and semi-dilute region. The DMSO compositions (V/V) in the mixed solvent ranged from 10/0 to 0/10. 2.2. Dynamic light scattering For dynamic light scattering experiment, the solvent, DMSO and water were repeatedly filtered using a 0.02 lm

The homogeneous PVA solutions with different polymer concentrations and compositions of co-solvent mixture were prepared in the sealed test tubes which were then kept in an oven at 95 °C for 2 h to make the solutions homogeneous again. The hot solutions were rapidly transferred into a water bath at 25 °C. The quenched temperature (25 °C) was always kept constant for different polymer concentration and compositions of co-solvent mixtures to prevent temperature effect which could weaken the DMSO/(water)n hydrate obviously. The test tube tilting method was used to determine the gelation time, tgel which was defined as the time (minutes) required for the cessation of the solution flow inside the test tube when it was tilted. The reciprocal of gelation time of the solution is referred as the apparent gelation rate t1 gel which can be expressed as a function of polymer concentration and temperature: [8,17,18]

t 1 gel / f ðCÞf ðTÞ

ð2Þ

1160

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

that the water and DMSO molecules adsorbed on the PVA chains might tend to leave PVA coils to form stable DMSO/(water)2 complexes(hydrate), which would increase the molar size of the co-solvent. The decreased number of adsorbed solvent molecules in the polymer coils directly cause a more contracted conformation, which may be assigned to the reduction in the excluded volume. When a small amount of DMSO(X1 0.07) was present in the mixed solvent, the dimension of the PVA coil in this solution was slightly larger than that in water alone. It was suggested that the tetrahedral structures of water molecules would be easily disrupted at the position close to the polar sites as indicated in the previous study [20] (i.e., the S=O group of DMSO). As a result, more free water molecules were released to slightly increase PVA-solvent affinity. In order to clarify the change in molecular interactions in the PVA/DMSO/water ternary system, Read’s equation [21] of the theoretical preferential adsorption was modified to include the effects of not only the molar volumes of solvent, but also the interaction among the complex [20], the polymer chain, and the solvents. It is clear that the water molecules are preferentially adsorbed by PVA chains at lower X1, while DMSO molecules at higher X1. An inversion of theoretical preferential adsorption phenomenon appears at X1 0.33, implying that the PVA chains have no preferential adsorption of solvents at this co-solvent composition. In other words, the solvent desorption and the interaction of solvent–solvent attraction of this ternary system appears strongest. In order to clarify the previous study on theoretical preferential adsorption and intrinsic viscosity, we performed the dynamic light scattering experiment (DLS). The DLS results are helpful to understand a wide range of dynamic processes; it can also reflect conformational relaxation and viscoelastic properties. However, it is more sensitive to both diffusive and non-diffusive process than that studied by dynamic mechanical methods or dielectric measurement. Attention was primarily focused on several compositions of solvent mixtures (that is, X1 = 0.2–0.37) which are corresponding to the ratios of co-solvent complexes in 1:4, 1:3, and 1:2, respectively. The averaged relaxation rates at several angles for monitoring fluctuations were measured with different momentum vector, q.

At a specific temperature, a more general relationship is given as follows:

t1 gel /

" #n C  C gel C gel

ð3Þ

where C gel is the critical gelation concentration and n is an exponent value depending on the various gelation mechanisms. 2.4. Static light scattering Light scattering measurements were conducted on a Malvern series 4700 apparatus with an Argon ionic laser (k = 514.5 nm, 20–50 mw) as the light source. The scattering vector, Q ¼ 4pk n sin 2h is in the range of 0.3 to 3.33103 Å1, where n is the reflective index. 3. Results and discussion Fig. 1 describes the schematic representation of the most probable forms for DMSO/(water)n (n = 2–3) complexes. It is well-known that DMSO and water molecules tend to form complex by hydrogen bond in the mole ratio of 1/2 (DMSO mole fraction, X1 = 0.33) or 1/3 (X1 = 0.25). It was also reported that the structure of the complex formed by perturbing the pattern of hydrogen bonds in water is dependent on the size, charge, shape, and chemical character of the solvent [19]. In our previous study [16], it was found that the intrinsic viscosities [g] for the PVA/water and PVA/DMSO solutions are about 0.93 and 3.25 dLg1, respectively. It suggested that DMSO is a better solvent for PVA than that for water. The formation of strong PVA–DMSO hydrogen bonds and the large molar volume of DMSO (V1 ffi 71.313 cm3 mol1) could induce a higher excludedvolume effect which would extend the PVA coil in solution state. On the other hand, the PVA chains exhibit a contracted coil in the aqueous PVA solution due to less affinity to PVA and small molar volume (V2 ffi 18.083 cm3 mol1) of water. The study was also focused on two special compositions at X1 0.07 and 0.28. It was found that the intrinsic viscosity would be the minimum at X1 = 0.28. It is accepted

H

CH 3 O

S

H

O

O

CH3 H

H H

CH3 O

or

S

O

CH3

H

H CH3 O

CH3

S CH3

O

O

S

CH3 O

H

H

H

II DMSO/(water)n , n=2

DMSO/(water)n , n=3

The most probable forms of co-solvent complexes Fig. 1. Schematic representation of the most probable forms for DMSO/(water)n (n = 2–3) complexes.

1161

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

ses of scattered light autocorrelation functions. The results of the sA(s) distribution revealed obviously that the PVA dilution solutions, in which the concentration is far below the concentration of chain overlapping, simply exhibited a single relaxation mode related to the translational diffusion of individual PVA coil. The decay rate for diffusion, C = 1/s (inset of Fig. 2), for all of the relaxation modes shows a linear dependence on the square of the scattering vector, q2 at qRg << 1. In other words, the dynamic structure factor of a diffusing coil relaxes to zero showing a mono-exponential decay and this dynamics follows a diffusion equation. For other compositions, the similar results and the D0 values of PVA at specific concentration in dilute region have also been derived. The mutual diffusion coefficients of PVA (i.e., the unit for the diffusivities ffi1011 m2/s) as a function of concentration for all the studied PVA/DMSO/water solutions at 25 °C are shown in Fig. 3(a) and (b), respectively. The results reveal that the mutual diffusion coefficients of coil in solutions vary almost linearly with the polymer concentration in dilute region. It is believed that the change in slope can be related to the thermodynamic driving force, hydrodynamic driving force and solvent quality. Usually, the mutual diffusion coefficient at a finite concentration can be represented as follows [16,22–27]:

The average diffusion coefficients were then calculated in infinite dilution to exclude inter-chain aggregation or association. The aggregates that formed in the dilution solution may result in an increase in characteristic domain size and the scattering intensity. The aggregates would also drive the system to be opaque with time, which would hinder DLS measurement. Thus, the time-trace experiment (i.e., Scattering Intensity vs. time plot) was conducted to monitor the presence of the aggregates so that more reliable g(1)(t), A(s) and C, D0, Rh values can be obtained [16,22]. These values reveal the information of polymer–solvent interaction, coil size and the effect of solvent complexes on dilute regime. Note that DMSO could either be a strong electron donor or hydrogen acceptor that would enhance the affinity between PVA and DMSO. The hydrogen bonding between DMSO and PVA is helpful toward the formation of the homogeneous solution, which shows no aggregation, no aging, and a lower scattering intensity. It takes more time to collect the enough count number for conforming the statistic regular. Once the solvent complex is formed, it would reduce the dipole–dipole interaction and result in poorer affinity and stronger scattering intensity [23]. Fig. 2 shows the intensity time correlation functions, g(2)(t)1, for the PVA dilute solutions in X1 = 0.2, 0.28, 0.33, 0.37 at 25 °C together with the corresponding relaxation time distributions obtained by using cumulant analy-

D ¼ ðC=q2 Þq¼0 ¼ D0 ð1 þ K d C þ . . .Þ;

0.4

0.4

(a)X1= 0.20 -1

Γ × 10

0.3

6 4 2

0.2

0

0

2

4

6

2

-10

q × 10

0.1

0

2

4

8

6 4 2

0.2

0

10

2

4

6

8

10

0.1

6

0.0

8

0

2

4

6

8

log τ/μs

0.4

0.4

(c) X1= 0.33

0 0

2

4 2

(s )

-1

6 -10

q × 10

8

10

-2

−3

6 4 2

0.2

(2)

2

g (t)-1

Γ × 10

4

single coil diffusive motion

8

0

(cm )

0

2

4

6 2

8 -10

q × 10

0.1

10

12

τ A(τ)

(2)

0.3

6

τ A(τ)

0.2

10

(d) X1=0.37

Single coil diffusive motion

Γ × 10

8

-1

(s )

10

−3

0.3

g (t) -1

0

-2

(cm )

log τ/μs

0.0 0

single coil diffusive motion

8

g (2)(t)-1

(s )

8

10

τ A(τ)

(2)

single coil diffusive motion

τ A(τ)

g (t)-1

(b) X1=0.28

10

−3

0.3

0.0

ð4Þ

-2

(cm )

0.1

2

4

log τ/μs

6

8

0.0

0

2

4

6

8

log τ/μs

Fig. 2. The normalized intensity-intensity autocorrelation function, g(2)(t)1, for dilute solutions of PVA/DMSO/water at: (a)X1 = 0.20, [g]C = 0.21, and (b)X1 = 0.28, [g]C = 0.074, (c)X1 = 0.33, [g]C = 0.09, (d) X1 = 0.37, [g]C = 0.12 at 25 °C and a scattering angle h = 90°, along with their corresponding distributions of relaxation time. Inset: q2 dependence of the corresponding decay rate, C.

1162

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168 2.0

a

1.2

11

2

-1

D× 10 (m s )

1.6

X1=0 X1=0.03 X1=0.06 X1=0.10 X1=0.14 X1=0.20

0.8

0.4

0.0

0.0 2.0

1.2

0.4

0.6

0.8

1.0

b

X1=1 X1=0.7 X1=0.50 X1=0.37 X1=0.33 X1=0.28

11

2

-1

D× 10 (m s )

1.6

0.2

0.8

0.4

0.0

0.0

0.2

0.4

0.6

0.8

1.0

concertration (g/dL) Fig. 3. The diffusion coefficients D as a function of concentration of the water-rich fractions and the DMSO-rich fractions at 25 °C.

where mutual diffusion coefficients D is determined by measuring autocorrelation functions at several angles above 30°, D0 is the mutual diffusion coefficient at infinite dilution, and Kd is the second virial coefficient composed of thermodynamic and frictional parameters, representing the intermolecular equilibrium interactions and intermolecular hydrodynamic interactions, respectively[22]. The hydrodynamic radius, Rh, at the studied compositions can be also evaluated from the diffusion coefficient at infinite dilution, D0, using Stokes–Einstein relation:

Rh ¼ kB T=6pgs D0

ð5Þ

where kB is the Boltzmann constant, and gs is the solvent viscosity. These calculated parameters are summarized in Table 1. From Fig. 4(a) and Table 1, it is observed that the value of Kd is negative at X1 = 0, (i.e., PVA dissolved in pure water). It indicates that the hydrodynamic interaction will

begin to intensely influence the dynamic behavior of polymer chains in solutions. The value of D0 obtained from the intercept is almost four times larger in PVA/water than that in PVA/DMSO, implying that PVA coils in DMSO are not so easy to aggregate than in water. When the DMSO mole fraction in co-solvent mixture was increased, a slightly positive slope appeared at X1 = 0.05–0.10 (i.e., the volume ratio is near 2/83/7). This result indicates that the affinity is improved among these compositions, and thus PVA chains may become more extended. The results are also consistent with the changes in Rh and Kd values in Fig. 4(a) and Table 1, respectively. The results are in good agreement with our previous studies of the theoretical preferential adsorption and intrinsic viscosity at X1 0.07 [20]. It suggested that the preferential adsorption on water molecules would break the H-bonded tetrahedral structure of water, thus more free water molecules could be created to improve the affinity of PVA. When the DMSO concentration was increased (X1 0.07), DMSO might act as a structure-breakers of water–water network owing to polar interaction. However, when the DMSO concentration was below this fraction, only a few DMSO molecules might act as the structure-makers for stabilizing the H-bonded water–water network structure [28]. Generally, for an aqueous ternary system with water and a low polar organic solvent, the tetrahedral structure of water could not be broken. The organic solvent with weak-polarity could always serve as a structure-maker for water [29,30]. When the DMSO concentration was increased to X1 = 0.1–0.2, the tendency of the slope was steady, indicating that the solvent power might be between h and marginal solvent. Therefore, the change in coil size and dynamic behavior might be similar to those in pure water. Furthermore, the curve shows a negative slope and a steep drop with concentration in dilute region at X1 0.28. It suggested that the hydrodynamic interaction might dominate the dynamic behavior of the ternary system at this fraction. There is no doubt that this composition is the poorest solvent and its Rh and Kd show the minimum of about 13.7 nm and 1.0, respectively in Fig. 4(a) and Table 1. When the DMSO concentration was further increased to X1 0.33 (V/V = 6.6/3.4), the smallest D0 value and the largest solvent viscosity gs as shown in Fig. 4(b) suggested that no preferential adsorption would occur at this fraction [20]. Therefore it reconfirms the most stable and the

Table 1 Characteristic parameters of dilution solution for PVA in various solvent compositions at 25 °C. V/V (volume ratio)

X1 (mole fraction)

D0  1011 (m2/s)

Kd (dLg1)

Rh (Å)

Solvent viscosity (cp)

0/100 10/90 20/80 30/70 40/60 50/50 60/40 66/34 70/30 80/20 90/10 100/0

0 0.0274 0.0596 0.0981 0.1446 0.2013 0.2756 0.33 0.3718 0.5037 0.6953 1

1.7416 1.3170 0.9344 0.7215 0.5977 0.4852 0.4521 0.3609 0.4000 0.4135 0.4236 0.4697

0.2788 0.1063 0.4248 0.3505 0.0786 0.1628 1.0063 0.2830 0.1750 0.8610 2.0910 3.0360

140 151 169 162 166 155 137 163 152 170 204 237

0.898 1.11 1.38 1.87 2.20 2.91 3.52 3.70 3.60 3.11 2.53 2.003

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

240

4

(a) X1≅ 0.07

O

200 0

Kd

Rh (A )

2

160 -2

X1≅ 0.28 120 0.0 1.8

-4

0.2

0.4

0.6

0.8

1.0 4

(b)

D 0× 10

s

1.2

2

1

X1≅ 0.33

0.6

0.0

η (cp)

-11

2

(m /s)

3

0.2

0.4

0.6

0.8

0 1.0

X1 (mol%) Fig. 4. The hydrodynamic radius Rh, the dynamic second virial coefficient Kd, the diffusion coefficient at infinite dilute D0 and the solvent viscosity g0 as a function of X1 at 25 °C.

largest amount of DMSO/(water)2 complexes could form at this fraction without showing the poorest affinity between co-solvent and polymer. We suggest that when DMSO concentration is at X1 0.28 (near the composition X1 = 0.25 which corresponds to the formation of DMSO/(water)3 complexes, though this complexes cannot exist steadily owing to the higher free energy of mixing than DMSO/ (water)2 complexes), the large-scale of concentration fluctuation due to the formation and elimination of this unstable complexes can result in intrachain contraction or interchain association. When PVA concentration is increased further at this solvent composition, the resulted ternary system might tend to undergo liquid–liquid phase separation and gelation. When DMSO concentration is at X1 = 0.33, it is the most stable co-solvent composition of the thermodynamic equilibrium structure and has the lowest free energy of mixing without the presence of solute or when the amount of solute is small enough to be neglected. Hence, gs and D0 show the maximum and the minimum, respectively. Also, the preferential adsorption coefficient predicts aa0 on this fraction. Based on the considerations of hydrodynamic interaction, Rh value derived from the Stokes–Einstein relation is dependent on the solvent viscosity gs and the average diffusion coefficient D0 in this ternary system. The formation of DMSO/(water)2 complexes would make the effect of solvent viscosity more important than that of the solute diffusion coefficient. Thus, it does not show the minimum for Rh and Kd values because polymer coils and co-solvents are difficult to move at this fraction. When the DMSO fractions are even higher at X1 > 0.33–1, the movement of DMSO molecules to the di-

1163

luted phase (polymer-poor phase) becomes easy because the redundancy of the free DMSO molecules. The PVA coils might extend even more than those at X1 = 0.07 because of the increased excluded volume and the donor-accepter interaction between PVA and DMSO. Furthermore, the thermodynamic driving force would start to dominate the dynamic behavior in PVA dilution solution of these DMSO fractions. The thermodynamic driving force as shown in Fig. 3(a) and (b) would become more profound when the DMSO mole fraction was increased above 0.33. Different solvent quality could affect the molecular interactions of this ternary system, the relaxation time of the specific molecular motion, and thus result in the change of the coil size and dynamic second virial coefficient in the dilute region. A further investigation was made to clarify whether the behavior of preferential adsorption could affect the polymer–solvent interaction in semi-dilute region. It was found that it would be difficult to investigate the dynamics of this ternary system by dynamic light scattering in the semidilute region because of the strong aggregation and the opaque surface. Moreover, the excluded volume that would swell the polymer chains at low concentration might screened at higher concentrations. Therefore, efforts were focused on investigating the relationship between the phase separation and the structural characteristics of polymer gels through gelation kinetic analyses instead. It is well-known that PVA solutions can form physical gels from various kinds of solvents or co-solvents. When it comes to thermoreversible gelation, a fluid solution is converted into a rigid medium of infinite viscosity, and vice versa. Many methods, such as test tube, ball-dropping and gelation timer, have been used in determining the gelation process and mechanism of polymer gel. However, these methods cannot provide the detail information at the gelation threshold. In general, the gelation of polymer solution from a poor solvent must occur after the spinodal decomposition because the liquid–liquid phase separation is easy to achieve for the thermodynamic driving force, @ 2 f =@C 2 < 0. As a matter of fact, the compositions higher than 10 gdL1 in this examination were so viscous as to be regarded as gel-like solutions. Gelation occurred gradually with the evolution of the gel-like property in sol state without a definite gelation concentration. This gel-like heterogeneity is related to the liquid–liquid phase separation [31]. Matsuo and co-workers [32–34] have studied the phase separation behavior of PVA/DMSO/water ternary solutions. They pointed out that the phase separation would occur rapidly as the DMSO content in the mixed solvent was about 0.5–0.7 volume fraction, and the process of gel formation was faster for 0.5–0.6 volume fraction than that for 0.7 volume fraction. It was also found that the linear relationship of ln(I) vs. time did not reflect the initial state of pure spinodal decomposition. It indicated that the chain mobility was slowed down at this moment and gelation would start to take place. Therefore, the simultaneous progress of gelation and spinodal decomposition would be suggested. It is evident that the PVA solutions in X1 = 0.28 have a UCST type phase diagram with a critical temperature about 70 °C. Thus, the experiment temperature at 25 °C is much deeper into the SD-region [6]. Fig. 5

1164

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

shows the t1 gel as a function of the [g]C value for various solvent compositions at 25 °C. The t1 gel would increase gradually with concentrations at all studied compositions. The calculated exponent values n were not equal to 0.45, indicating that the gelation process would not be a percolation type association [17,18,35,36]. All the curves in Fig. 5 were extrapolated to zero gelation rates to determine the critical concentration of gelation C gel . It is generally accepted that analysis of the concentration effect can be used to study the macroscopic mechanism of gelation. As a result, the C gel values obtained at X1 = 0.20, 0.28, 0.33, 0.37, which gelation can occur are about 1.8, 0.9, 1.8, and 1.95 gdL1, respectively. Note that the co-solvent composition at X1 = 0.25 corresponding to the formation of DMSO/ (water)3, PVA coils are the most difficult to be dissolved in the co-solvent above the semi-dilute solution region. Therefore, in the poorest solvent composition for X1 = 0.25–0.28, the C gel values are the smallest; the crystallite formation and the connection of chains are also easier to occur than any other fractions. However, the C* (=1/[g]) values as derived in the preceding section disagree with the experimental C gel values, suggesting that occurrence of the spinodal decomposition may strongly affect the gelation behavior. Under this circumstance, the chain overlap concept cannot be applied. Thus, according to the report by Frisch and Simha [37], it was suggested that the dynamic behavior of polymer solution could be classified into several regions through the semi-empirical rule based on the interaction between polymer chains. It is obvious that only the composition at X1 = 0.28, the gel could still

Infinite dilution limit

form obviously at several specific infinite dilution region of [g]C < 1, suggesting that the poorest affinity at this fraction as discussed previously could lead the PVA coils with stronger capacity to aggregate. The considerable amount of the intermolecular association between these chains could become possible. Furthermore, for X1 = 0.2, 0.33, 0.37, the macroscopic gel could form at the hydrodynamic screening limit, for 1 < [g]C < 4. It suggested that the polymer–polymer intermolecular interactions might start to affect the aggregation behavior of polymer chains in solution. However, the gelation does not occur fastest at X1 = 0.33 which the largest amount of DMSO/(water)2 complexes could form. When gelation occurs at the hydrodynamic screening limit at X1 = 0.28 as shown in Fig. 5, the gelation rate is so fast as to interfere with spinodal decomposition process corresponding to ½gC trans ffi 1:287, which was obtained from the intersect of two straight lines as shows in Fig. 6. When it comes to X1 = 0.20 and 0.33, corresponding to ½gC trans ffi 3:5 and 2.29, respectively, the apparent gelation rates rise rapidly also at the 1 < [g]C < 4 region. At X1 = 0.37 corresponding to ½gC trans ffi 4:42, apparent rate rises rapidly and gelation starts to dominate in the semi-dilute region of [g]C > 4. The influence of phase separation at this fraction would be weakened gradually due to the effect of chain overlapping. Consequently, the gelation occurs at the region of 1 < [g]C < 4 and [g]C > 4 for X1 = 0.20, 0.33 and 0.37. Fig. 6 shows the double logarithmic plots of t1 gel as a function of reduced concentration at all studied fractions at 25 °C. Two different rate-dependent regions were ob-

Chain overlapped concentration

0.12

Coil-coil interpentration

hydrodynamic screening region

-1

-1

tgel (min )

0.08

0.04

0.00

0

2

4

6

8

10

12

[ C

Fig. 5. Apparent gelation rate, t1 gel , as a function of [g]C for PVA/DMSO/water solutions at 25 °C at various solvent composition: (d) X1 0.20; (N) X1  0.28; (5) X1 0.33; (}) X1  0.37.

1165

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

0

0

(b) X1~0.28

n2=2.4

-1

n2=2.68

-1

-1

log tgel (min )

-2

n1=1.17

-1

-1

-1

log tgel (min )

(a) X1~0.20

[η]Ctran*=3.5

-3

-4 -0.8

-0.4

0.0

0.4

0.8

-2

n1=0.81 -3

[η]Ctran*=1.287

-4 -0.8

1.2

-0.4

-1

-1

1.2

n2=2.29

-1

-1

log tgel (min )

-1

n2=1.97

n1=0.91

-3

-4 -0.8

0.8

(d) X1~0.37

(c) X1~0.33

-2

0.4

0

0

log tgel -1(min )

0.0

log[(c-cgel*)/cgel*]

log[(c-cgel*)/cgel*]

0.0

0.4

n1=0.80 -3

[η]Ctran*=2.29

-0.4

-2

[η]Ctran*=4.42 0.8

-4 -0.8

1.2

-0.4

0.0

0.4

0.8

log[(c-cgel*)/cgel*]

log[(c-cgel*)/cgel*]

  Fig. 6. The log t 1 gel vs. log½ðC  C gel Þ=C gel  plot of PVA/DMSO/water gel at X1 = 0.2–0.37 at 25 °C.

served at low and high concentration region for every studied DMSO fraction. It was found that the ‘‘n” values obtained from the linear fit of each concentration region were about 1 or 2 [8,17,18,38–40]. This divergence may be attributed to the various kinetic processes involved during gelation in PVA/DMSO/water ternary system. The exponent value ‘‘2” could be regarded as the binary association of segments in the cross-linking junction, and ‘‘1” as the diffusion of polymer chains in the phase separation process. The kinetic transition concentration C trans can be determined from the intersection of two straight lines. These characteristic values of C gel , C trans and n for the kinetic processes of gelation at the four studied DMSO fractions are summarized in Table 2. In order to investigate the relation between the kinetic exponent ‘‘n” and the corresponding gelation mechanisms, it could be said that the gelation process would exist in two competitive mecha-

nisms. For n ffi 1 at lower concentration, the gelation is related to be diffusion controlling mode. In other words, the desorption of DMSO and water, the strong association of DMSO/(water)n solvent complexes and the SD type phase separation do occur before the macroscopic gelation, then induce and dominant the gelation process. It is obvious that the phase separation rate is faster than the aggregation rate. Takeshita and co-workers have confirmed that the SD type phase separation would occur in the formation of the opaque gel by using TRLS measurements [9]. For n ffi 2 at higher concentration, the gelation mechanism is known to be crystalline nucleation control or reaction control. It is assumed that the formation of nucleus is the ratedetermining step which induces and dominates the gelation process even in the phase separation region. All the studied results suggest that the influence of phase separation on gelation process may be weakened when the

Table 2 The critical gelation concentration C gel , transition concentration C trans and exponent n values for PVA/DMSO/water gels between X1 = 0.2–0.37 at 25 °C. X1 (mole %)

V1 (volume %)

C gel (gdL1)a

C trans (gdL1)b

[g] (dLg1)

n1

n2

0.20 0.28 0.33 0.37

0.5 0.6 0.66 0.7

1.8 0.9 1.8 1.95

4.65 3.48 4.09 4.42

0.75 0.37 0.56 1

1.17 0.81 0.91 0.8

2.4 2.68 1.97 2.29

a b

Obtained in this work by extrapolation to t1 gel ¼ 0.   The transition concentration was determined from the intersection of two lines from the plots of log t 1 gel vs. log½ðC  C gel Þ=C gel  in Fig. 7.

1166

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

concentration is higher than the kinetic transition concentration, C trans . It should be mentioned that Ohkura [10] showed the kinetic characteristic exponent of the same ternary system was about 2 on their macroscopic kinetic analysis. That is because that the concentration region that they studied is deep enough into the coexistence curve of this ternary system at X1 = 0.28; thus the crystalline nucleation process can always be the rate determine step on gelation. As comparing with other compositions, it is obvious that gelation occurs easiest at X1=0.28 and the lowest concentration C gel =0.9 gdL1. It suggested that the poorest affinity at this fraction might greatly facilitate the coil aggregation and the gel formation, but the whole gelation process would proceed for a longer time than other fractions due to the long-range heterogeneity caused by phase separation. In addition, the n values obtained at X1 0.28 is

somewhat higher than those at other fractions. It suggested that phase separation might accelerate gelation rate by itself. Although this method may only offer the apparent gelation rates, it can still provide the qualitative analysis and the useful information on gelation mechanisms. In order to further investigate the relationship between the polymer-solvent interaction and the gelation behavior of this ternary system, static light scattering were used to study the aging effect of this ternary system. Fig. 7 shows the plots of intensity vs. scattering vector for 1 gdL1 of PVA solution with different aging time at several solvent compositions (i.e., at X1 = 0.2, 0.25, 0.28, 0.33, 0.37) at 25 °C. All the studied mole fractions except at X1 = 0.37 obviously exhibited a broad peak at around 1.5–2  103 Å1 and the scattering maximum would change with aging time. At X1 = 0.37, the redundant free DMSO molecules

20

20

X1~0.20

-3

10

A

0 -1

8 4 0

8 4

1

2 3

0 -1

0

4

3

Q 10 (A )

20

2

1

0 -1

30 hr 20 hr 10 hr 5 hr

16 12

3

3

12

4

3

X1~0.33 I(Q) 10 (a.u.)

16

3

Q 10 (A )

20 20hr 10hr 5hr

X 1 ~0.28 I(Q) 10 (a.u.)

12

3

1.6

16

I(Q) 10 (a.u.)

12

3

I(Q) 10 (a.u.)

16

20 hr 10 hr 5 hr

X 1 ~0.25

20 hr 10 hr 5 hr

8

4

4

1

2 3

0 -1

3

0

4

2

1

1.6

3

0 -1

3

Q 10 (A )

Q 10 (A )

X1~0.37

30 hr 20 hr 10 hr 5 hr

1.2

3

I(Q) 10 (a.u.)

0

8

0.8

0.4

0.0

1

2 3

0 -1

3

4

Q 10 (A ) Fig. 7. Plots of intensity vs. scattering vector Q for 1 gdL1 PVA solution with various aging times at 25 °C.

4

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

would increase the polymer-co solvent interactions and hinder the aggregation, thus the broad peak cannot be observed. Takeshita and co-workers [9] performed this time evolution of aging effect at the same temperature, 25 °C. The appearance of scattering maximum in this system is attributed to the phase separation and the significant degree of the chain aggregation. Indeed, when DMSO fractions were at X1 = 0.25, and 0.28, especially at X1 = 0.25, the results showed the fastest phase separation and aggregation. This behavior was considered as the many-body effect of short-period forms of the DMSO/(water)3 complexes [41,42]. Larger amount of DMSO/(water)2 complexes and other forms would induce the stronger and faster concentration fluctuation and aggregation. Thus, it was suggested that the coil concentration might follow the solvent adsorptions, the initial stage of phase separation, coil aggregate; and then gel formation would occur gradually at this temperature at lower concentration which could form the physical gel. Further investigations on how these processes affect the final results will be conducted through computer simulation and spectrum analysis in the future. When the concentration and/or the DMSO fractions were increased, the effect of crystallization would outshine that of phase separation on gelation. Thus, different initial conditions might result in the different gelation mechanism. The shift of the scattering peak and no value in low scattering vector might imply the saturation of scattering intensity and/or the form of the characteristic size of this physical gel. In addition, from the result of solvent quality, it seems reasonable to conclude that the heterogeneous, opaque crosslink network could form when DMSO fractions were at X1 = 0.25–0.33 and the homogeneous, transparent network could form when DMSO fractions were higher than 0.37 if gelation could occur.

4. Conclusions In this study, the effects of solvent quality on the dynamic behavior of PVA co-solvent solution in dilute region have been clarified. It was found that the thermodynamic driving force might dominate the dynamic behavior of PVA dissolved in pure DMSO and DMSO-rich solvent mixture in dilute solution, while the hydrodynamic interaction might dominate in pure water and water-rich solvent mixture. Furthermore, The reasons that the smallest values of Rh, Kd and [g] appeared at X1 0.28 but not at X1 = 0.33 might be related to the formation of the largest amount of stable DMSO/(water)2 complexes, which may increase the co-solvent viscosity, slow the molecular mobility and influence the accompany kinetics and mechanisms. As a result, the phase separation, intrachain contraction and interchain aggregation must occur at X1 = 0.25, corresponding to the existence of a large amount of DMSO/(water)3 complexes. Also, all the formed gels are opaque at 25 °C and for DMSO fraction near X1 = 0.28 at which the gelation rate for this ternary system is the fastest than other studied fractions. Analysis of the kinetic data exhibits that the gelation process in this ternary system obeys the kinetic mechanisms

1167

of a diffusion control and a nucleation control determined by different concentrations of the solution and the co-solvent compositions. The result is similar to our previous study on PVDF/TG system. However, this result has slightly deviation from the result suggested by Takeshita who thinks that the composition of cononsolvency should occur at X1 = 0.33 corresponding to the most stable DMSO/(water)2 complexes. Combine the course among the molecular interactions in dilute solution and the phase separation, aggregate, and gelation behavior in higher concentration by tracing the aging effect on this studied composition. Our results would conclude that the driving force for the spinodal decomposition demixing and clusters aggregate in higher concentration is the poor affinity resulted from the long-range heterogeneity and the formation of strong DMSO/(water)n complexes due to the thermodynamic and the hydrodynamic factor. Moreover, the smallest characteristic size obtained at X1 = 0.25 might imply that the junction zone of this PVA physical gel was constructed by the denser aggregation of chain coils. References [1] Jacobs SW. In: Rosenbaum EE, Wood DC, editors. Dimethyl sulfoxide. New York: Marcel Dekker; 1971. [2] Hyon SH, Chu WI, Ikada Y. Rep Poval Committee 1986;89:1. [3] Tacx JCJF, Schoffeleers HM, Brands AGM, Teuwen L. Polymer 2000;41:947. [4] Hong PD, Huang HT. Polymer 2000;41:6195. [5] Takahashi N, Kanaya T, Nishida K, Kaji K. Polymer 2003;44:4075. [6] Takeshita H, Kanaya T, Nishida K, Kaji K. Macromolecules 2001;34:7894. [7] Ohkura M, Kanaya T, Kaji K. Polymer 1992;33:3686. [8] Ohkura M, Kanaya T, Kaji K. Polymer 1992;33:5044. [9] Takeshita H, Kanaya T, Nishida K, Kaji K. Macromolecules 1999;32: 7815. [10] Kanaya T, Ohkura M, Takeshita H, Kaji K. Macromolecules 1995;28:3168. [11] Kanaya T, Ohkura M, Kaji K. Macromolecules 1994;27:5609. [12] Wolf BA, Willms MM. Makromol Chem 1978;179:2265. [13] Schild HG, Muthukumar M, Tirrell DA. Macromolecules 1991;24: 948. [14] Sigert AJF. Massachusetts Institute of Technology, Radiation Laboratory Report No. 465, 1943. [15] Chu B. Laser light scattering. 2nd edn. New York: Academic Press; 1991. [16] Brown W. Dynamic light scattering: the method and some applications. New York: Oxford Press; 1993. Chapter 6. [17] Mal S, Nandi AK. Polymer 1998;39:6301. [18] Dikshit AK, Nandi AK. Macromolecules 1998;31:8886. [19] Rahman A, Stillinger FH. J Am Chem Soc 1973;95:7943. [20] Hong SJ, Huang HT, Hong PD. J Appl Polym Sci 2004;92:3211. [21] Read BE. Trans Faraday Soc 1960;56:382. [22] Brown W. Light scattering: principles and development. New York: Clarendon Press; 1996. Chapter 7. [23] Vavra J, Antalik J. Polymer 1997;38:6281. [24] Cotts PM, Selser JC. Macromolecules 1990;23:2050. [25] Han CC, Ziya Akcasu A. Polymer 1981;22:1165. [26] Ziya AA. Polymer 1981;22:1169. [27] Hong PD, Chou CM, He CH. Polymer 2001;42:6105. [28] Costa Ricardo OR, Freitas Roberto FS. Polymer 2002;43:5879. [29] Zhang G, Wu C. Phys Rev Lett 2001;86:822. [30] Zhang G, Wu C. J Am Chem Soc 2001;123:1376. [31] Choi JH, Ko SW, Kim BC, Blackwell J, Lyoo WS. Macromolecules 2001;34:2964. [32] Matsuo M, Kawase M, Sugiura Y, Takematsu S, Hara C. Macromolecules 1993;26:4461. [33] Hara C, Matsuo M. Polymer 1995;36:603. [34] Matsuo M, Sugiura Y, Takematsu S. Polymer 1997;38:5953. [35] Mal S, Maiti P, Nandi AK. Macromolecules 1995;28:2371. [36] Sudip M, Tushar J, Nandi K. Macromolecules 2001;34:275.

1168

S.-J. Hong et al. / European Polymer Journal 45 (2009) 1158–1168

[37] Frish HL, Simha R. In: Eirich FR, editor. Reology: theory and applications. New York: Academic Press; 1956. [38] Hong PD, Chou CM. Polymer 2000;41:8311. [39] Hong PD, Chou CM. Macromolecules 2000;33:9673.

[40] Guenet J. Thermoreversible gelation of polymers biopolymers. London: Academic Press; 1992. [41] Rasmussen DH. Nature 1968;220:1315. [42] Barbara K, Markus R. J Am Chem Soc 2002;124:6206.

and