Journal of Chromatography A, 1127 (2006) 237–245

The phase behaviour of poly(styrene-co-methacrylic acid)/poly(2,6-dimethyl-1,4-phenylene oxide) by inverse gas chromatography Z. Benabdelghani a , A. Etxeberria b,∗ , S. Djadoun a , J.J. Iruin b , C. Uriarte b a

Faculty of Chemistry, University of Sciences and Technology Houari Boumediane, BP 32, El Alia, Algiers 16111, Algeria b Departamento de Ciencia y Tecnologia de Polimeros and Instituto de Materiales Polimericos (POLYMAT), Universidad del Pais Vasco UPV/EHU, P.O. Box 1072, 20080 San Sebastian, Spain Received 21 March 2006; received in revised form 23 May 2006; accepted 30 May 2006 Available online 7 July 2006

Abstract The miscibility behaviour of blends of poly(styrene-co-methacrylic acide) (PSMA-12) containing 12% of methacrylic acid with poly(2,6dimethyl-1,4-phenylene oxide) (PPO) at five different compositions has been studied by inverse gas chromatography (IGC). The adequacy of two types of solvents (solvents and precipitants) has been tested in order to detect the glass transition temperature, Tg . A single Tg has been observed in PSMA-12/PPO blends containing less than 67 wt% of PPO. Two glass transition temperatures appeared, however, in blends containing higher PPO content. The polymer–polymer interaction parameters have been calculated in the molten state (260–280 ◦ C), their values being in good agreement with the observed phase behaviour. Moreover, the methods proposed by Farooque–Deshpande and Huang for obtaining the true polymer–polymer interaction parameters have been compared. Finally, the ability of IGC in order to determine Tg s has been extended to a ternary system, PSMA-12/PPO/EMV4P-8 poly(ethylmethacrylate-co-4-vinylpyridine) which, depending on the composition, gives a single, two and even three Tg s. © 2006 Elsevier B.V. All rights reserved. Keywords: Inverse gas chromatography; Glass transition temperature; Interaction parameter; Poly(styrene-co-methacrylic acid)/poly(2,6-dimethyl-1,4-phenylene oxide)

1. Introduction In most cases, polymers give immiscible blends [1]. This fact is, in general, undesirable in many industrial applications due to their poor mechanical properties. In spite of the extensive research carried out in polymer synthesis, it could be advantageous to obtain polymeric materials with appropriate combinations of properties from miscible polymer blends [2]. In this sense, it is known that poly(2,6-dimethyl-1,4-phenylene oxide) (PPO) is miscible in all proportions with polystyrene (PS) [3]. On the contrary, it has been shown that both the blends of a homopolymer with a copolymer or of two copolymers may be even miscible in the absence of specific interactions, provided that sufficient repulsions between the comonomers are present

∗

Corresponding author. Tel.: +34 943 01 53 49; fax: +34 943 01 52 70. E-mail address: [email protected] (A. Etxeberria).

0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2006.05.093

[4–6]. This fact has promoted the study of the miscibility of mixtures based on PPO with styrene copolymers [7–11]. In this sense, the effect of incorporating carboxylic groups into polystyrene chains on the well-known miscible PPO/PS blends has been already reported showing that PSMA/PPO are totally or partially miscible with PPO depending on the methacrylic acid (MA) content in the copolymer [9,12,13]. In previous studies, Benabdelghani et al. [12] have found that PSMA and PPO are totally miscible up to 7.8% mol fraction of MA. Akiba et al. [13] have also reported similar conclusions. Inverse gas chromatography (IGC) has been used in order to determine different magnitudes of pure polymers, solutions and mixtures [14]. Among others, glass transition temperature (Tg ) [15] and the polymer–polymer interaction parameter have been, probably, the main topic of the IGC studies [16–18]. Using the Tg detection by IGC, in this work we would like to show the ability of this technique in order to study the miscibil-

238

Z. Benabdelghani et al. / J. Chromatogr. A 1127 (2006) 237–245

ity of polymers blends following the well-known macroscopic criteria according to that miscible blends have a unique Tg while immiscible blends have two or more Tg s. At this point, it is important to remind a fact that has received much attention in IGC studies for polymer–polymer blends. As it has been cited above, IGC has been largely applied to determine the polymer–polymer interaction parameter, χ23 . However, it has been shown that the nature of the solvent used as a probe has a large influence in the obtained χ23 value. This influence has been ascribed to the so-called χ effect, related to the non-random partitioning of the probe [19–22]. Most of these last cited papers have tried to find the conditions that a solvent [23,24] should fulfil in order to obtain the true χ23 value. Therefore, it could be possible to wonder if the nature of the used solvent could play a role in order to detect the Tg . Therefore, the first aim of this work will be to answer this question and to find the conditions that a solvent should fulfil in order to measure the Tg . As the second aim, we have measured the Tg of the above-mentioned systems (PSMA-12/PPO and PSMA12/PPO/PEM4VP-8) by IGC in order to check this technique ability for establishing the polymer miscibility. Finally, we will apply two simple and phenomenological methods [25,26] to obtain the true polymer–polymer interaction parameter in the PSMA-12/PPO system. 2. Experimental 2.1. Materials PPO was supplied by Aldrich. Before use, it was purified, by dissolution in chloroform and precipitation in methanol. After this purification, the viscosimetric molecular weight has been calculated (Mv = 2 × 104 ). Both poly(styrene-co-methacrylic acid) and poly(ethylmethacrylate-co-4-vinylpyridine) copolymers were prepared by free radical polymerization at 60 ◦ C using azo-bis-isobutyronitrile (AIBN) as initiator. The styrene content in the copolymers was determined by UV spectroscopy [27]. DSC measurements have been carried out on the same conditions that those explained in a previous publication [12]. 2.2. Procedures and equipment The chromatographic support was Chromosorb W (acid washed and treated with dimethyldichlorosilane AW-DMSC) 80–100 mesh. Packed columns were prepared from a 150 cm long and 0.635 cm outer diameter stainless-steel tube. The tubes were first rinsed with acetone, after passing dry nitrogen gas through them, and dried under vacuum for 5 h at 60 ◦ C. The polymer and their mixtures were coated on the solid support by a soaking method [28]. Columns were conditioned at 250 ◦ C under a fast carrier gas flow rate for 12 h prior to use. The total percent loading of polymer over the support is almost 7% in weight in all the cases. The loading was measured by calcinations [20] (1 h at 450 ◦ C). Characteristics of the columns used in this work are given in Table 1.

Table 1 Column loading data Column

Support (g)

Polymer (g)

PSMA-12 PPO ME4VP-8 PSMA-12/PPO 2/1 PSMA-12/PPO 1/1 PSMA-12/PPO 1/2 PSMA-12/PPO 1/3 PSMA-12/PPO 1/4 PSMA-12/PPO/PEM4VP-8 33/33/33 PSMA-12/PPO/PEM4VP-8 39/39/22 PSMA-12/PPO/PEM4VP-8 58/20/22 PSMA-12/PPO/PEM4VP-8 80/10/10

1.270 1.564 1.569 1.570 1.564 1.560 1.739 1.575 1.546 1.554 1.535 1.490

0.127 0.156 0.155 0.157 0.156 0.155 0.150 0.158 0.155 0.155 0.154 0.149

The blends composition is given by weight fraction.

Measurements were carried out on a Varian gas chromatograph (model 3700) equipped with a flame ionization detector. Nitrogen was used as the carrier gas. Flow rate (≈7 ml/min) was measured with a soap-bubble flowmeter at the room temperature. Three or more injections of an infinitesimal quantity, using a 1.0 ␮l Hamilton syringe, were carried out for each probe. The net retention time was taken as the difference of the retention times of the probe and the methane peaks. The required thermodynamic data of solvents have been taken from usual compilations [29,30]. The densities of the PSMA-12 and PPO were taken from the literature [31]. 2.3. Data reduction The specific retention volume, Vg0 , were calculated from the expression: Vg0 = t

F 3 273.15 (Pi /Po )2 − 1 ω 2 Tr (Pi /Po )3 − 1

(1)

where t is the net retention time; ω the mass of the polymer in the column, Pi and Po the inlet and outlet pressures and F is the carrier gas flow at room temperature, Tr . The values obtained for Vg0 using the above given procedure were found to agree within 5 × 10−3 for each measurement. The solvent–polymer interaction parameter, χ1i , can be calculated from the Vg0 value using the following equation: χ1i = ln

273.15Rvi B11 − V1 0 −1− P1 0 0 RT Vg,i V1 P1

(2)

0 are, respectively, the specific volume and spewhere vi and Vg,i cific retention volume of the polymer i; Vi , P10 and B11 are, respectively, the molar volume, the vapour pressure and the second virial coefficient of the solvent. When the stationary phase is a polymer blend, Eq. (2) allows the determination of the ternary solvent (1)-polymer (2)-polymer (3) interaction parameter, χ1(23) , assuming an additive specific volume for the polymer blend, vb = w2 v2 + w3 v3 where wi is the weight fraction of polymer i in the blend. On the contrary, assuming the ScottTompa approximation [32], which describes a ternary system as

Z. Benabdelghani et al. / J. Chromatogr. A 1127 (2006) 237–245

a simple balance of the corresponding binary systems, it is possible to calculate the polymer–polymer interaction parameter, χ23 , by: χ1(23) = χ12 φ2 + χ13 φ3 −

χ23 V1 φ2 φ3 V2

(3)

where φi is the volume fraction of polymer i and V2 the molar volume of polymer 2. In the framework of the Flory–Huggins theory a reference volume must be defined in order to calculate the interaction parameter, thus, the polymer–polymer interaction  is introduced as: parameter related to the solvent volume, χ23  χ23 = χ23

V1 V2

(4)

However, Al-Saigh and Munk [28] have demonstrated that if IGC measurements are carried out in identical experimental conditions of flow, temperature, and inlet and outlet pressures, the polymer–polymer interaction parameter determination is greatly simplified and only the specific volumes and specific retention volumes of the pure polymers and the blends are required to  values: obtain χ23  χ23

=

0 /v ) − φ ln(V 0 /v ) − φ ln(V 0 /v )] [ln(Vg,b 2 3 b g,2 2 g,3 3

φ2 φ 3

239

specific retention volumes by:  0     0   0   Vg,b Vg,2 Vg,3 1 1 = φ2 ln + φ3 ln ln V1 vb V1 v2 v3 +

χ23 φ2 φ3 V2

(9)

Therefore, a plot of the left side term as a function of the expression between brackets of the right side term of Eq. (9) allows us to calculate from the intercept the trueχ23 /V2 . In both methods the plots could be performed using the interaction parameters, χ1i , i.e. Eqs. (6) and (8), or using specific reten0 , i.e. Eqs. (7) and (9). In this paper, we have tion volumes, Vg,i applied the latter ones because they are experimental magnitudes directly extracted from IGC data. In this way the error introduced by the other needed magnitudes (V1 , P10 and B11 ) for the calculation of χ1i , Eq. (2), would disappear. Moreover, due to that, in both methods, the intercepts give directly the χ23 /V2 magnitude. In the following pages, we will use this magnitude instead of the polymer–polymer interaction parameters. 3. Results and discussion

(5)

As it has been mentioned above, the polymer–polymer interaction parameter determined by IGC shows a clear dependence on the solvent used as a probe. In order to solve this problem, different methods have been proposed [18,22,23]. Between others, the simple method of Farooque and Deshpande gives a reliable true polymer–polymer interaction parameter [25,33–35] after a rearrangement of Eq. (3):
   χ123 − χ13 χ12 − χ13 χ23 − φ2 φ3 = φ2 (6) V1 V1 V2 Moreover, if the conditions given by Al-Saigh and Munk [28] are still obeyed, the Eq. (6) could rearrange in a similar way:       0 0 v3 Vg,b v3 Vg,2 1 χ23 1 = φ2 + ln ln φ2 φ3 0 0 V1 V1 V2 vb Vg,3 v2 Vg,3 (7) Thus, using Eq. (6) or, alternatively, Eq. (7), a plot of the left side term versus the expression between brackets of the right side term allows us to calculate from the intercept the true polymer–polymer interaction parameter. More recently, Huang [26,36] have proposed an alternative rearrangement of Eq. (3) obtaining this expression: 
χ1(23) φ2 φ3 χ23 φ2 χ12 + φ3 χ13 − = (8) V1 V1 V2 Following the same assumptions that transform Eq. (6) into Eq. (7), we propose the determination of the true polymer–polymer interaction parameter χ23 using only experimental measured magnitudes, such as specific volumes and

3.1. PMA-12/PPO system. determination of glass transition temperatures From IGC measurements, phase transition can be detected by plotting the logarithm of the specific retention volume versus the reciprocal of temperature. At any physical state a straight line is observed. However, when a phase transition occurs, the physico-chemical properties change from the original state to the second one, thus, the retention diagram gives a curve showing a minimum at lower temperatures and then a maximum at higher temperatures, after which a new straight line starts (see Fig. 1). In this way, both glass transition (Tg ) and melting temperatures can be determined [14]. The shape of the retention diagram was explained on the basis that, at the glassy state, the probe only interacts with the polymer surface due to the fact that the diffusion into the polymer is too slow to allow the bulk interaction. At temperatures higher than Tg , the retention volume is a measure of the interaction of the probe with the bulk polymer in liquid state. Close to the Tg , both factors contribute to the retention volume, which suffers an increase as the solvent penetrability increases. In relation to the straight-line regions, the negative slop is due to the increase of vapour pressure with temperature [37,38]. In order to obtain Tg s from retention diagrams different protocols can be applied: very start detection, onset, midpoint (the most commonly accepted way), endpoint temperatures or very end detection, as also are defined for DSC [39]. Even though in some cases onset or midpoint definitions could be applied, in others the retention diagram does not show a clear curve, as described above. This could be due to the fact that the two Tg s can overlap over an extended temperature region. Other reason could be that the amount of one of the phases can be so small that the transition region cannot be clearly detected. Thus, we have

240

Z. Benabdelghani et al. / J. Chromatogr. A 1127 (2006) 237–245

Fig. 1. Retention diagrams of PSMA-12 using the seven probes.

preferred to define a common criterion for obtaining all the Tg s. Actually, we have measured the Tg as the temperature where the straight line obtained in the glassy state starts to deviate from linearity, so, its slope changes, as it has been marked in some figures. This definition corresponds to the very start detection of

the glass transition event [39]. From the measurements carried out in this work, we estimate that the error involved in the very start detection of the Tg by IGC is about of 1 ◦ C. As it has been commented in the introduction, the Tg transition region depends on the nature of the probes. Thus, in this work we have used two types of probes (solvents and precipitants) in order to find the best solvents for the Tg detection in the PSMA-12/PPO system. In Fig. 1, the retention diagrams of seven solvents in a poly(styrene-methacrylic acid) copolymer with 12% mol fraction in methacrylic acid (PSMA-12), are shown. First at all the solvents allow an adequate determination of Tg of this copolymer but the transition region (see the rectangle areas in Fig. 1) is observed in a different way for poor solvents (precipitants) and good solvents. In the case of poor solvents, the retention diagram shows a curve as expected in the Tg region, so, the Tg can be determined without any trouble. However, in the case of good solvents, the curve associated to glass transition is not as clear as for precipitants. The transition occurs in a narrower temperature interval and, principally, the increase of the specific retention volume (Vg0 ) is lower. This behaviour renders more difficult the Tg determination, especially in the halogenated solvents. We believe that there are two factors contributing to the different behaviour between good and poor solvents. First, the fact that the good solvents have stronger interactions with the polymer would provide a weak diffusion of the solute at the glassy state that does not occur in the precipitant cases. Then, a minor temperature increase is enough for the solvent to interact with the bulk polymer. This fact explains the narrower transition region observed for good solvents compared to those of the precipitants. The second fact is the vapour pressure. As it can be seen in Fig. 1 for both groups of solvents, the higher is the vapour pressure (see Table 2) the clearer is the Tg curve region. At the glassy state, the solvents with lower vapour pressure have larger Vg0 . At liquid state, the solvent solubility is governed by pressure, thus, the higher is the vapour pressure the higher is Vg0 . 0 0 − Vg,glass Resuming, the partition coefficient (Vg0 = Vg,liquid ) increases strongly for solvents that have higher vapour pressure [40,41]. In this sense, n-decane and benzene have the largest transition region between precipitants and good solvents, respectively. Taking into account the retention diagrams of Fig. 1 and the similar ones obtained for PPO, EMV4P-8 and blends, it can

Table 2 Glass transition temperatures determined for the PSMA-12/PPO system and the vapour pressures of the solvents at 120 ◦ C, in mmHg % of PPO

n-C10a

n-C11a

n-C12a

Benzene

Toluene

Cl-Bena

DSC [12]

0 100 33.3 50.0 66.7 75.0 80.0 P10

121 210 130 133 143, 205 145, 207 139, 210 151.30

120 210 132 133 143, 205 – 140, 210 74.34

120 212 132 133 144, 205 – 140, 209 36.90

120 212 130 133 143, 205 147, 207 143, 209 2249.16

120 213 130 135 144, 205 147, 207 143, 209 984.70

120 213 130 – 144, 205 147, 207 144, 209 547.7

125 214 1393 147 146, 201 150, 208 151, 209 275.1

a

n-C10, n-C11, n-C12 and Cl-Ben correspond to n-decane, n-undecane and chlorobenzene, respectively.

Z. Benabdelghani et al. / J. Chromatogr. A 1127 (2006) 237–245

241

Fig. 3. Tg -composition of PSMA-12/PPO as determined by IGC using n-decane as probe.

Fig. 2. Retention diagrams of PSMA-12, PPO and their 2/1,1/2 ratio blends using n-decane as probe.

be conclude that the molecular size and the chemical nature of the probe are important factors in determining of the Tg region. Consequently, for our system, n-decane could be considered the best probe in order to detect the glass transition temperature by IGC. Fig. 2 shows the n-decane retention diagrams corresponding to the 2/1 and 1/2 compositions in weight fraction in PSMA-12. In the first case, the Tg of the pure polymers and the blend are well delimited, being the blend Tg intermediate to pure polymer ones. In the case of the PSMA-12/PPO 1/2 blend two glass transitions are observed by IGC. So, the immiscibility at this composition can be concluded, in good agreement to the DSC data [12]. Furthermore, the first Tg , corresponding to a phase richer in PSMA-12, is also well delimited but the other one occurs in a very small transition region. Even for the diagram shown in Fig. 2, corresponding to the n-decane, which gives the best defined Tg region, the Tg detection is not possible using the midpoint or onset methods. This fact has definitely forced us to

determine the Tg using the very start detection method, as it has been mentioned above. Fig. 3 displays the variation of the glass transition temperature (Tg ) with the weight fraction of PPO in the blends, as determined by IGC using n-decane as probe. As it can be seen, the miscibility behaviour is in excellent agreement with previous results obtained by DSC. At less than 67 wt% PPO, only a single glass transition temperature can be observed, indicating that these blends are miscible. Moreover, for immiscible blends, the Tg corresponding to the phase rich in PSMA-12 takes a similar value to that corresponding to a 1/1, miscible blend while the Tg of the phase rich in PPO is slightly lower than the pure PPO. Tomita and Register [42] have reported similar observations. From this fact, it can be concluded that PSMA-12 achieves its PPO saturation concentration near to 50%, the remaining PPO forming the second phase. The glass transition temperatures obtained by IGC are summarized in Table 2. The agreement between Tg data of different solvents is excellent and similar to Tg s determined by DSC. The differences could be attached to the fact that different detection methods have been used in IGC and DSC. In this sense, the very start detection method gives Tg values lower than those obtained by the midpoint method. Also, it must be taken into account the dynamic basis of DSC would provide higher Tg values than the IGC case, where thermal equilibrium is attained at every temperature. 3.2. Determination of polymer–polymer interaction parameters As it has been pointed out in the introduction, the IGC measurements in the liquid state allow the determination of thermodynamic magnitudes [16], specially the interaction parameter that plays an important role as indicative of the blend misci0 take bility. In relation to the previous experimental data, Vg,i similar values to reported by Su and Fried [31] for PPO, PS and the 1/1 ratio blend. Firstly, we have examined, qualitatively, the phase behaviour of the PSMA-12/PPO system measured in the

242

Z. Benabdelghani et al. / J. Chromatogr. A 1127 (2006) 237–245

Fig. 4. ln Vg —composition of PSMA-12/PPO using the seven probes at 260 ◦ C.

molten state, using the following expression [43]: 0 0 0 Vg,b = w2 Vg,2 + w3 Vg,3

(10)

0 ) is plotted versus the blend composition In Fig. 4, ln (Vg,i for all probes. Up to 50 wt% PPO, a negative deviation of ln 0 ) from the additive ones, Eq. (10), can be observed for all (Vg,b the probes. At 67 wt% PPO and above, the probe curves show a positive deviation from linearity. Thus, all the probes, good solvents as well as poor ones, have a S-shaped form respect to the additive rule.

As shown by DiPaoli-Baranyi and Degree [44], negative devi0 means that probe has more difficulties to interact ation of Vg,b with polymers due to their own interaction, thus, partition coefficient is displaced to gas phase and this fact is resumed in negative  ), as it can be seen polymer–polymer interaction parameters (χ23  in Table 3, and, of course, positive χ23 values are obtained for 0 . positive deviation of Vg,b  absolute values in an Of course, all the solvents give χ23 excellent agreement to the blend composition effect on misci values show a slight dispersion. But, in bility. As usual, the χ23 IGC studies is usual to obtain, depending of the injected probe,  for the same blend, so, hinpositive and negative values of χ23 dering to classify it as miscible or immiscible, but fortunately this fact does not occur in our system. Anyway, between the methods proposed [18,22,23,25,26] in order to obtain a true polymer–polymer interaction parameter, the simple and phenomenological method of Farooque– Deshpande [25] as well as the recent method proposed by Huang [26] have been shown as reliable alternatives. In both cases, the true polymer–polymer interaction parameter is obtained from the intercept of a plot derived from an arrangement of Eq. (3), actually Eqs. (7) and (9). Strictly, Eqs. (3)–(9) are only valid for miscible blends, even though this fact has been usually overlooked in the literature [45]. However, the next objective of this paper is to check the ability of Farooque–Deshpande and Huang methods in order to obtain the true polymer–polymer interaction parameter and therefore, in our opinion, those methods would be reliable to classify the blends as miscible or immiscible by means of the true χ23 /V2 value. Thus, we will apply also Eqs. (7) and (9) to immiscible compositions of PSMA-12/PPO system. In Fig. 5, the fittings provided by the Farooque–Deshpande method, using Eq. (7), and the method of Huang, using Eq. (9), for the blend PSMA-12/PPO 2/1 at 260 ◦ C are shown. Some comments are appropriate. First, both methods give very similar intercept values, so the true polymer–polymer interaction parameter, χ23 /V2 , determined by both methods, is practically similar and this result would be taken as a proof of their reliability. Second, the correlation of the linear regression is excellent for the Huang method while the Farooque–Deshpande method gives poorer results, even though its appearance is good. Third, although the correlation of the Huang method is nearly to one, the error calculated for the intercept value is twice the one determined by the Farooque–Deshpande method (2.38 × 10−4 and 1.13 × 10−4 , respectively). This fact could be due to the experimental points fitted by Eq. (9) are so far from the intercept

Table 3  calculated by Eq. (5), for the PSMA-12/PPO system at 260 ◦ C Polymer–polymer interaction parameters, χ23 % of PPO

n-C10

n-C11

n-C12

Benzene

Toluene

Cl-Ben

Br-Bena

33.3 50.0 66.7 75.0 80.0

−1.03 −0.13 1.02 1.37 1.98

−1.32 −0.88 1.11 – 2.08

−1.59 −1.04 1.71 – 2.99

−1.09 −0.34 1.25 1.44 1.92

−0.89 −0.33 1.39 1.87 2.24

−0.69 −0.33 1.29 1.62 2.10

−0.66 −0.22 1.26 1.88 2.19

a

Br-Ben corresponds to bromobenzene.

Z. Benabdelghani et al. / J. Chromatogr. A 1127 (2006) 237–245

243

Table 5 χ23 /V2 values, in mol/cc × 103 , determined by the Farooque–Deshpande method (FD) and by the Huang method (H) for the miscible PSMA-12/PPO blends (2/1 and 1/1 weight ratios) T (◦ C)

260 265 270 275 280

Fig. 5. Application of Eq. (7) (
) and Eq. (9) (䊉) in order to obtain the true polymer–polymer interaction parameter for the PSMA-12/PPO 1/1 at 260 ◦ C. The straight lines are the calculated linear regressions, whose parameters are given.

and its determination is subjected to a larger error. Whereas in the case of Eq. (7), the fitted data are closer to the intercept and, moreover, there are points at both sides of it. In Table 4 we have resumed the results obtained by Eqs. (7) and (9) for all the blends at 270 ◦ C. It can be observed that similar errors are involved in the calculation of the intercept in the rest of the compositions. In general the Eq. (9), fits the experimental 0 data considerably better than Eq. (7). In this context, the Vg,i correlations of Eq. (9) are, in all cases, superior to 0.99 while whose corresponding to the Eq. (7) varies from 0.80 to 0.98. The same trend has been reported for the Farooque– Deshpande method, where the intercept was involved in errors superior to 100%, even though their corresponding fits were close to one [33–35]. This error could be partially diminished using a previous solvent selection procedure [35]. Possibly, applying this idea to the Huang method framework, the error would be reduced. But, as it has pointed out by Huang [26] the requirements on the selection procedure should be different to those given for the Farooque–Deshpande method. This point would be an attractive aim of future works. Concerning to the χ23 /V2 values shown in Table 4, both methods give similar absolute values as well as a good agreement to Table 4 Intercept values, φ2 φ3 χ23 /V2 in mol/cc, and their error obtained by the Farooque–Deshpande, Eq. (7), and Huang, Eq. (10), methods for the PSMA12/PPO at 260 ◦ C % of PPO

33.3 50.0 66.7 75.0 80.0

Farooque–Deshpande

Huang

Intercept × 103

Error × 104

Intercept × 103

Error × 104

−1.18 −0.62 1.24 1.59 1.51

1.25 1.11 1.06 2.36 0.99

−1.25 −0.59 0.92 1.43 1.11

2.36 2.38 3.95 1.58 3.70

2/1

1/1

FD

H

FD

H

−5.2 −5.0 −4.7 −4.5 −4.6

−5.5 −5.3 −5.2 −5.0 −4.8

−2.5 −2.3 −2.1 −1.9 −1.8

−2.4 −2.4 −2.2 −1.7 −1.2

the phase behaviour. Thus, this result corroborates that these methods could be applied to obtain reliable values for miscible blends as well as to certify that positive values are calculated for immiscible blends. However, the large negative values of χ23 /V2 for miscible blends would reflect the existence of strong interactions between PSMA-12 and PPO through the methacrylic acid and the ether groups, respectively. So, it would be expected that as PPO concentration increases, this intermolecular interaction would be facilitated while MA dimerization would be hindered. Finally, in Table 5 the χ23 /V2 behaviour of miscible blends with the temperature is resumed. Again, both methods give similar χ23 /V2 absolute values and, as it is usual in polymer blends [46], the interactions are weakened increasing the temperature, thus, χ23 /V2 becomes more positive at higher temperatures. In conclusion, both the Farooque–Deshpande and the Huang methods give a reasonably reliable true polymer–polymer interaction parameter. The large error on its calculation is a reflect that χ23 /V2 is determined from a sensitive balance of binary system in a ternary one [22]. 3.3. Miscibility of the ternary system PSMA-12/PPO/PEM4VP-8 On the basis of the results obtained above with the binary system SMA-12/PPO, we also have investigated by ICG some ternary SMA-12/PPO/EMV4P-8 blends. We have reported in our previous work that SMA copolymer was found miscible in all proportions with poly(ethyl methacrylate-co-4-vinylpyridine) (EM4VP) while PPO/EM4VP has been found immiscible [47]. Furthermore, as it has been pointed out in the introduction, it is known that when a third component is added to an immiscible polymer blend, is possible to achieve a miscible ternary blend [4–6]. In our case, PSMA-12 was used as a compatibilizer of binary pair PPO/PEM4VP-8. Actually, the following glass transition temperatures characterized the phase behaviour. The lowest one (Tg,1 ≈ 103–109 ◦ C) is intermediate between pure PSMA-12 and PEM4VP-8 Tg s, that can be attributed to binary PSMA-12/PEM4VP-8 miscible blend. Whereas the highest one (Tg,2 ≈ 205 ◦ C) would be given by a phase formed mostly by PPO with a small amounts of PSMA-12 (see Fig. 3). Even, a third phase (Tg,3 ≈ 152 ◦ C) was observed and should be attributed to the miscible phase of binary pair SMA-12/PPO. On the contrary, it should be important to cite that the dissolution of PPO in the SMA-12 decreased

244

Z. Benabdelghani et al. / J. Chromatogr. A 1127 (2006) 237–245

Table 6 Glass transition temperatures determined for the PSMA-12/PPO/PEM4VP-8 system Column

n-Decane

n-Undecane

DSC

PEM4VP-8 PSMA-12/PPO/PEM4VP-8 33/33/33 PSMA-12/PPO/PEM4VP-8 39/39/22 PSMA-12/PPO/PEM4VP-8 58/20/22 PSMA-12/PPO/PEM4VP-8 80/10/10

75 95, 149, 180 101, 150, 205 95, 194 110

75 98, 153, 185 103, 155, 205 98.5, 195.5 113.5

77.5 103, 152, 205 102, 153, 205 109, 152 116

in the presence of EMV4P-8 as confirmed by formation of such a high phase, observed only with binary SMA-12/PPO blends containing high fraction of PPO. It is interesting to notice that, in all cases of these ternary blends, specific interactions occurred preferentially between EMV4P-8 and PSMA-12 and reduced the dissolution of PPO in such phase. With this aim, four different compositions of ternary blend PSMA-12/PPO/PEM4VP-8, 80/10/10, 58/20/22, 39/39/22 and 34/33/33 in weight, respectively, have been studied by IGC. These compositions have been chosen in such a way that the PSMA-12/PPO binary ratio correspond to binary miscible blends. Also, we have chosen substantially low PPO and PEM4VP-8 compositions in the ternary blends in order to get miscible ternary systems. In Fig. 6, retention diagrams of ternary system corresponding to three blends are shown. Fortunately, the PSMA-12/PPO/PEM4VP-8 80/10/10, in weight fraction, gives only a Tg with a value slightly lower than pure PSMA-12 and so different to the Tg s corresponding to phases detected previously by DSC [12], therefore, it can be concluded that at this composition the blend is miscible as it has been also confirmed by DSC (see Table 6). However, in other three ternary compositions two or three Tg s have been found, thus, they must be considered as immiscible. In the case of the blend PSMA-12/PPO/PEM4VP-8 39/39/22, two Tg s, which correspond to a phase of PSMA-12/EM4VP and other rich in PPO with a small amount of PSMA-12 (Tg,1 and Tg,2 , respectively), are notorious in the retention diagram. But, as it can be seen in Fig. 6, the retention diagram maximum corresponding to Tg,1 occurs in a larger Tg interval than the preceding ones. We have concluded that such a width can be only due to the existence of a third Tg (Tg,3 ) which is formed by PSMA-12/PPO approximately at 50%, near to the retention diagram maximum. Moreover, the obtained Tg values are in good agreement to the values determined by DSC (see Table 6). Finally, in the PSMA-12/PPO/PEM4VP-8 58/20/22 and 34/33/33 compositions the Tg,3 does not appear and the maximum is not so wide to suppose that it would include Tg,3 , even though it had been detected by DSC. However, there is another difference between DSC and IGC data. As it is shown in Table 6, DSC is not able to detect the Tg,2 while IGC does. In our opinion, the quantity of PPO in this composition would be in the limit to give detectable amount of the phases denoted as 2 and 3. Therefore, DSC would find difficulties in order to detect the Tg,2 while IGC has troubles with the Tg,3 . In conclusion, the use of the most appropriates probes, the n-decane and n-undecane due to their precipitant character and higher vapour pressures, allows to reproduce such complicate

Fig. 6. Retention diagrams of the PSMA-12/PPO/PEM4VP-8 system using ndecane as probe.

Z. Benabdelghani et al. / J. Chromatogr. A 1127 (2006) 237–245

PSMA-12/PPO/PEM4VP phase behaviour, thus, the ability of IGC in order to detect the glass transition temperatures has been confirmed. 4. Conclusions In order to detect the glass transition temperatures in polymer blends by IGC, we have shown that not only the ideal probes are non-solvents, but rather the size of probe would be the smallest possible. In general, good solvents give small changes in the retention diagram in the Tg region. Moreover, between the alkanes the principle property to choose the best solvent is the vapour pressure, which should be the highest one. Similar trends would be applied to other solvent families. Furthermore, the good agreement with DSC measurement found for both the PSMA-12/PPO and the PSMA-12/PPO/ PEM4VP-8 blends, which give two or three Tg s in the immiscible compositions, allows us to conclude that IGC is an adequate technique to study the miscibility of polymer blends through the glass transition temperatures. The phase behaviour has been corroborated by the polymer–  , which takes negative valpolymer interaction parameter, χ23 ues in the miscible region and positive ones for the immiscible compositions. Moreover, two methods proposed to obtain the true polymer–polymer interaction parameter have been compared, i.e., the Farooque–Deshpande and the Huang methods. The results show that both methods give similar χ23 /V2 values, so, in our opinion, both methods can be considered as reliable methods in order to calculate the true polymer–polymer interaction parameter. Moreover, the determined χ23 /V2 values are also subject to an important error which is, on the other hand, intrinsic to the IGC technique due to the fact that χ23 /V2 is determined from a delicate balance between a ternary system (χ1(23) ) and two binary systems (χ12 and χ13 ). Acknowledgements A.E., C.U. and J.J.I. would like to acknowledge the support by the project of UPV/EHU 9/UPV 00203.215-13519/2001. Z.B. and S.D. would like to acknowledge the support of Minister of Sciences and Technology of Algeria (MESRS/E 1602/71/05). References [1] D.R. Paul, S. Newman, Polymer Blends, vol. 1, Academic Press, New York, 1978. [2] D.R. Paul, J.W. Barlow, Polymer 25 (1984) 487. [3] J. Stoelting, F.E. Karasz, W.J. MacKnight, Polym. Eng. Sci. 10 (1970) 133. [4] G.T. Brinke, F.E. Karasz, W.J. MacKnight, Macromolecules 16 (1983) 1827.

245

[5] J. Kressler, H.W. Kammer, K. Klostermann, Polym. Bull. 15 (1986) 113. [6] J.M.G. Cowie, G. Li, R. Ferguson, I.J. McEwen, J. Polym. Sci. B 30 (1992) 1351. [7] H. Tomita, A. Register, Macromolecules 26 (1993) 2796. [8] C. Auschra, R. Stadler, Macromolecules 26 (1993) 6364. [9] C.G. Bazuin, L. Rancourt, S. Villeneuve, A. Soldera, J. Polym. Sci. B 31 (1993) 1431. [10] A. Gottschalk, K. Muehlbach, F. Seitz, R. Stadler, C. Auschra, Macromol. Symp. 83 (1994) 127. [11] R. Vucovic, G. Bogdanic, F.E. Karasz, W.J. MacKnight, J. Phys. Chem. Ref. Data 28 (1999) 851. [12] Z. Benabdelghani, R. Belkhiri, S. Djadoun, Polym. Bull. 35 (1995) 329. [13] I. Akiba, Y. Ohba, Y. Kobori, S. Akiyama, Polym. J. 32 (2000) 402. [14] A. Etxeberria, J. Alfageme, C. Uriarte, J.J. Iruin, J. Chromatogr. 607 (1992) 227. [15] O. Smidrød, J.E. Guillet, Macromolecules 2 (1969) 272. [16] D.D. Deshpande, D. Patterson, H.P. Schreiber, C.S. Su, Macromolecules 7 (1974) 530. [17] B. Mandal, C. Bhattacharya, S.N. Bhattacharya, J. Macromol. Sci. Chem. A26 (1989) 175. [18] P. Munk, P. Hattam, A.A. Abdel-Azim, Q. Du, Macromol. Symp. 38 (1990) 205. [19] D. Robard, D. Patterson, Macromolecules 10 (1977) 1021. [20] H.P. Schreiber, Y.B. Tewari, D. Patterson, J. Polym. Sci. B 11 (1973) 15. [21] D. Walsh, J.G. McKeown, Polymer 21 (1980) 1335. [22] M.G. Prolongo, R.M. Masegosa, A. Horta, Macromolecules 22 (1989) 4346. [23] M.J. El-Hibri, W. Cheng, P. Punk, Macromolecules 21 (1988) 3458. [24] C.S. Su, D. Patterson, Macromolecules 10 (1977) 708. [25] A.M. Farooque, D.D. Deshpande, Polymer 33 (1992) 5005. [26] J.C. Huang, J. Appl. Polym. Sci. 90 (2003) 671. [27] S. Djadoun, F.E. Karasz, F. Metref, Macromol. Symp. 78 (1994) 155. [28] Z.Y. Al-Saigh, P. Munk, Macromolecules 17 (1984) 803. [29] C. Reid, J.M. Prausnitz, T.K. Sherwood, The Properties of Gases and Liquids, third ed., McGraw-Hill, New York, 1977. [30] H. Romdhane, R.P. Danner, J. Chem. Eng. Data 36 (1991) 15. [31] A.C. Su, J.R. Fried, ACS Symp. Ser. 391 (1989) 155. [32] H. Tompa, Polymer Solutions, Butterworths, London, 1956. [33] A. Etxeberria, C. Uriarte, M.J. Fernandez-Berridi, Macromolecules 27 (1994) 1245. [34] E.G. Lezcano, D. Ramirez de Arellano, M.G. Prolongo, C. Salom Coll, Polymer 39 (1998) 1583. [35] A. Etxeberria, C. Etxabarren, J.J. Iruin, Macromolecules 33 (2000) 9115. [36] J.C. Huang, J. Appl. Polym. Sci. 89 (2003) 1242. [37] A. Lavoie, J.E. Guillet, Macromolecules 2 (1969) 443. [38] M. Braun, A. Lavoie, J.E. Guillet, Macromolecules 8 (1975) 311. [39] B. Wunderlich, in: E.A. Turi (Ed.), Thermal Characterization pf Polymeric Materials, Academic Press, San Diego, 1997, Chapter 2. [40] Z.Y. Al-Saigh, Polym. Int. 40 (1996) 25. [41] Z.Y. Al-Saigh, Polym. News 9 (1996) 269. [42] H. Tomita, R.A. Register, Macromolecules 26 (1993) 2796. [43] C. Zhikuan, D.J. Walsh, Eur. Polym. J. 19 (1983) 519. [44] G. DiPaola-Baranyi, P. Degree, Macromolecules 14 (1981) 1456. [45] Q. Du, W. Chen, P. Munk, Macromolecules 32 (1999) 1514. [46] M.M. Coleman, P.C. Painter, Specific Interactions and the Miscibility of Polymer Blends, Technomic Publishing, Lancaster, 1991. [47] Z. Benabdelghani, These de Magister, USTHB, Algiers, Algeria, 1997.