You are on page 1of 23

Fluid Phase Equilibria 208 (2003) 199221

Critical curves of liquidliquid equilibria in ternary systems Description by a regular-solution model


Josef P. Novk, Karel Rehk , Petr Vo ka, Jaroslav Matou n
Department of Physical Chemistry, Prague Institute of Chemical Technology, Technick 5, CZ-166 28 Prague 6, Czech Republic Received 21 June 2002; received in revised form 4 February 2003; accepted 4 February 2003

Abstract In this paper relations between the shapes of the liquidliquid critical curves in ternary systems and the temperature characters of the heterogeneous regions are discussed. The regular-solution model with temperature-dependent parameters was employed to describe critical curves of liquidliquid equilibria in ternary systems. Relations for the limiting values of the slopes of critical curves were derived. The corresponding type of the critical curve and consequently the temperature character of the liquidliquid equilibrium can be deduced from these values. Model calculations were performed for ternary systems characterised by two binary subsystems with either an upper critical solution temperature or with a lower critical solution temperature, and the third binary subsystem which is homogeneous in the considered temperature range. In the calculations the following approach was used: the behaviour of two binary systems was xed by the temperature dependencies of their parameters and ternary critical curves of LLE were then calculated for variants differing in non-ideality (deviations from the Raoults law) of the third homogeneous binary subsystem. It was found that the level of non-ideality of the homogeneous system can inuence the shape of the critical curve signicantly. 2003 Elsevier Science B.V. All rights reserved.
Keywords: Liquidliquid equilibria; Critical state; Mixtures; Ternary system

1. Introduction Two essential problems are encountered on studying liquidliquid phase equilibria in ternary systems. First, the experimental study of liquidliquid equilibria in a wider temperature range is seldom carried out, because such measurements are usually very demanding. This fact is corroborated in the compilations [1,2] in which more than 80% of experimental data on LLE in ternary systems are given for the temperature range from 0 to 35 C.
Corresponding author. a E-mail address: karel.rehak@vscht.cz (K. Reh k). 0378-3812/03/$ see front matter 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0378-3812(03)00037-2

200

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

Second, thermodynamic descriptions or predictions of the liquidliquid equilibrium often fail both qualitatively and quantitatively. This happens due to the fact that a very precise description of component activities is required [3,4]. The aim of the present paper is the study of liquidliquid equilibrium in ternary systems using critical curves. The critical curve is understood as a set of liquidliquid equilibrium critical points [5]. Our discussion is restrained to ternary systems with one homogenous and two heterogeneous binary subsystems, the heterogeneous subsystems being characterised by an upper or a lower critical solution temperature (i.e. UCST or LCST) and by a critical composition. It is also restrained to low pressures and consequently the inuence of pressure on critical point compositions is omitted. In addition, it is assumed that the binary systems create a ternary system with the two-phase regions only. An appearance of the three-phase regions (see [6,7] for examples) is not discussed in this work. To describe the thermodynamic behaviour of heterogeneous systems the regular-solution model, i.e. the simplest model for the excess Gibbs energy, is used in this study. Recently, the similar approach for binary systems was applied by Rebelo [8]. In the theory of phase equilibria the equations of state are often used as well [911]. Their application for description of the LLE in comparison to the VLE is more complicated and number of possible ternary diagrams classes becomes enormous [11]. The different basic types of LLE are schematically outlined in Figs. 14. Their parts (a) show the liquidliquid equilibria in the systems [1 + 2] and [2 + 3]; parts (b) demonstrate isotherms in ternary LLE diagrams; and parts (c) sketch shapes of the corresponding critical curves. The shaded areas represent two-phase regions in these gures. Examples of ternary systems which qualitatively follow the described behaviour are summarised in Table 1. Fig. 1 shows the case that the critical curve connecting binary critical points is monotonous. At a temperature lower than the critical temperatures of both binary subsystems, the heterogeneous region is band-shaped with no critical point. With increasing temperature, the heterogeneous region rst detaches from the axis that represents the binary subsystem characterised by lower critical temperature (the [2 + 3]

Fig. 1. The liquidliquid equilibrium in a ternary system: (a) LLE in binary subsystems [1 + 2] and [2 + 3]; (b) LLE in a ternary system at selected temperatures; (c) the composition dependence of the critical LLE temperature in a ternary system. The curve is monotonous.

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

201

Fig. 2. The liquidliquid equilibrium in a ternary system: (a) LLE in binary subsystems [1 + 2] and [2 + 3]; (b) LLE in a ternary system at selected temperatures; (c) the composition dependence of the critical LLE temperature in a ternary system. The curve shows a minimum.

system in Fig. 1). Further temperature increase makes the heterogeneous region, which now has one critical point, shrink. Fig. 2 illustrates the LLE behaviour related to a ternary critical curve with a temperature minimum. The heterogeneous region is band-like with no critical point at a temperature below Tmin . The two-phase region narrows with increasing temperature and splits into two separate regions at Tmin . This behaviour is also called the cosolvency effect [25].

Fig. 3. The liquidliquid equilibrium in a ternary system: (a) LLE in binary subsystem [2 +3] (the system [1+2] is homogeneous with high positive deviations from ideality); (b) LLE in a ternary system at selected temperatures; (c) the composition dependence of the critical LLE temperature in a ternary system. The curve shows a maximum.

202

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

Fig. 4. The liquidliquid equilibrium in a ternary system: (a) LLE in binary subsystems (the subsystem [1 + 2] with an UCST and the subsystem [2 + 3] with a LCST); (b) LLE in a ternary system at selected temperatures; (c) the composition dependence of the critical LLE temperature in a ternary system. The curve shows two branches.

A ternary critical curve showing a temperature maximum (Tmax ) is presented in Fig. 3. A closed heterogeneous region with two critical points appears at temperatures above TUC,23 . This region shrinks on further temperature increase and disappears at Tmax . An example of a ternary system with one binary subsystem showing an upper and the other a lower critical temperature is shown in Fig. 4. In these cases, the ternary critical curve shows two branches. Figs. 14 demonstrate only few basic LLE critical curves shapes and their relations to LLE temperature dependency in ternary systems. There are also many other combinations of the above-mentioned behaviours. Several questions concerning the topic can be asked and their resolution can lead not only to a better thermodynamic description but also to a better prediction of the behaviour of heterogeneous ternary systems.
Table 1 Examples of systems with different temperature characters of the LLE System Aniline (1) + water (2) + phenol (3) Benzylalcohol (1) + water (2) + 4-hydroxy-4-methyl-2-pentanone (3) Cyclohexane (1) + methanol (2) + hexane (3) Nitromethane (1) + water (2) + 1-butanol (3) 1-Butanol (1) + water (2) + furfural (3) Nitromethane (1) + 1-hexanol (2) + octanoic acid (3) Acetone (1) + water (2) + phenol (3) Acetic acid (1) + cyclohexene (2) + dimethylformamide (3) Au (1) + Ni (2) + Cu (3) Phenol (1) + water (2) + triethylamine (3) Phenol (1) + water (2) + tetrahydrofuran (3) Figure 1 1 2 2 2 2 3 3 3 4 4 Reference [12,13] [14] [15] [16] [17] [18] [19,20] [21] [22] [23] [24]

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

203

(a) Under which conditions is the ternary critical curve monotonous (see Fig. 1)? (b) What conditions must be fullled in case that the concentration dependence of the critical temperature shows an extreme (see Figs. 2 and 3)? (c) Are other shapes (e.g. showing both maximum and minimum) possible? (d) Does a critical curve lying between two upper critical solution temperatures differ from the one lying between two lower critical solution temperatures? (e) Is it possible that a ternary system characterised by two heterogeneous binary subsystems (one with an upper and the other with a lower critical solution temperature), where TLC,23 > TUC,12 would be (in contrast to the system phenol (1) + water (2) + tetrahydrofuran (3)) homogeneous within the temperature interval T (TUC,12 , TLC,23 )? From this point of view it is obvious that critical curves can serve as a valuable tool to the study of the liquidliquid equilibria. In the present paper, the regular-solution model was used to resolve these questions. 2. Theoretical The regular-solution model was chosen to describe the thermodynamic properties of studied systems primarily for its simplicity. 2.1. Describing binary systems In the regular-solution model the excess molar Gibbs energy in a binary subsystem (i) + (j) is given by [3] GE = xi xj bij (T ) RT (1)

where bij (T) is a temperature-dependent parameter of the corresponding binary system. At the critical temperature bij (T = Tc,ij ) = 2. Considering a system characterised by an upper critical solution temperature (UCST), bij > 2 at temperatures below the critical one. The parameter bij (T) can, at these temperatures, be evaluated from the mutual solubility xi (with regard to symmetry x1 = x2 = 1 x1 ) using the formula [3,26] bij = ln[(1 xi )/xi ] 1 2xi (2)

In this work, the following temperature dependence of the parameter bij = bij (T ) is considered ij bij = ij + T

(3)

where ij and ij are constants usually determined from a selected critical temperature of the binary system (i) + (j ) and a chosen solubility at another temperature. Variants of parameters ij and ij were obtained (see Table 2) and were used to carry out model calculations. The employed temperature dependence of the parameters determines the behaviour of the modelled system. Eq. (3) can be used both for systems characterised by an upper critical solution temperature

204

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

Table 2 Variants (IVI) of parameters ij and ij used to carry out model calculations UCST I Tc (K) T (K) bij (T) ij ij (K) xi (T ) HE (xi = 0.5) (J mol1 ) 300 250 2.5 0.5 750 0.1448 1558 II 350 300 2.5 1 1050 0.1448 2182
E

LCST III 250 350 0.40 6.4 2100 4364 IV 290 350 2.5858 5.4 986 0.1272 2049 V 350 400 2.5 6 1400 0.1448 2910 VI 300 350 2.5 5.5 1050 0.1448 2182

Values of Tc , T and bij were chosen, and ij , ij , xi and H were then calculated.

(ij > 0) and for those characterised by a lower critical solution temperature (ij < 0). Another consequence of the relation is the fact that the temperature dependence of the mutual solubility of the components is monotonous. Binary systems presenting a closed region of limited miscibility can however not be modelled by Eq. (3). This equation must be extended to be able to describe those systems. Appendix A summarises some circumstances concerning these cases. The conception of the regular-solution model based on Eq. (3) differs from the theory of the strictly regular solution employed by Prigogine [27], Prigogine and Defay [28] or Meijering [29,30]. These authors did not consider the absolute term in Eq. (3), i.e. they considered ij = 0. The advantage of their approach lies in the fact that only one parameter in necessary to describe a binary system. On the other hand, the description of systems characterised by a lower critical solution temperature is impossible using the strictly regular-solution model. A simple GE model used for generating all basic types of binary liquidliquid equilibria and their pressure dependence was also developed by Rebelo [8].

3. Describing ternary systems The regular-solution model expresses the molar excess Gibbs energy in ternary systems in the following way [3,28]: GE = x1 x2 b12 + x1 x3 b13 + x2 x3 b23 RT (4)

Relations for activity coefcients and the second to the fourth derivatives of the Gibbs energy with respect to x1 and x2 derived from this equation are summarised in [3]. It should be emphasised that in this work the Gibbs energy is considered as a function of temperature and composition only. The pressure dependence of GE is disregarded. The shape of the critical curve as a function of pressure at a constant temperature will therefore not be discussed, although it exhibits similar features under these conditions [7,25,31,32].

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

205

Eq. (4) can be also extended by a ternary term x1 x2 x3 C with a ternary parameter C. In this case the equation no longer describes a regular solution. For that reason the effect of this ternary parameter on the liquidliquid equilibrium character will not be discussed in the paper. Possibilities to use Eq. (4) in describing liquidliquid equilibria were partly discussed previously [3,28,33]. Nevertheless, cited works were restrained to ternary systems at constant temperatures described by parameters b12 = b23 = b and b13 . The temperature dependence of these parameters was not analysed in those papers. 4. Critical curve The molar Gibbs energy is given by the relation
3 3

Gm =
i=1

xi G m,i

+ RT
i=1

xi ln xi + GE

(5)

where G is the molar Gibbs energy of the pure component in the liquid state at the temperature and m,i pressure of the system, xi the mole fraction of the ith component and GE is the excess molar Gibbs energy. Let us consider temperature and mole fractions x1 and x2 as independent variables. The critical point of a ternary system at a constant temperature is given by [3,34,35] D = G11 G22 G2 = 0 12 D = D G22 x1 D G12 = 0 x2 D x2 = G112 G22 + G11 G222 2G12 G122 (6) (7)

Derivatives of the determinant D are given by D = G111 G22 + G11 G122 2G12 G112 , x1

(8)

where Gij = 2 (Gm /RT)/xi xj and Giij = 3 (Gm /RT)/xi2 xj are the second and the third derivatives of the dimensionless Gibbs energy with respect to mole fractions (Modell and Reid [35] use similar symbolisation, but in another context). The critical composition at a given temperature and pressure is determined by solving a set of two non-linear Eqs. (6) and (7). The calculation is described elsewhere [3]. Quantities D and D are functions of T, x1 and x2 in ternary systems, and their differentials are expressed as D D D dx1 + dx2 + dT , x1 x2 T D D D dD = dx1 + dx2 + dT x1 x2 T dD =

(9)

Eqs. (6) and (7) are valid for the critical curve and it also holds for this curve that dD = dD = 0. The dependence of the critical temperature and of x2 on x1 is expressed by means of Eq. (9) D dT D dT D dx2 D D dx2 D , (10) + = + = T dx1 x2 dx1 x1 T dx1 x2 dx1 x1

206

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

i.e. (D /x1 )(D/x2 ) (D /x2 )(D/x1 ) dT = , dx1 (D /x2 )(D/T ) (D /T )(D/x2 ) dx2 (D /T )(D/x1 ) (D /x1 )(D/T ) = dx1 (D /x2 )(D/T ) (D /T )(D/x2 ) Detailed expressions for D and D , and their derivatives are given in Appendix B. 4.1. Limiting values of slopes (dTc /dxi ) The shape of the critical curve is affected by limiting values of slopes, limx1 0 (dTc /dx1 ) and limx3 0 (dTc /dx3 ), i.e. slopes in critical points in binary subsystems [2 + 3] and [1 + 2]. The value of the slope, limx3 0 (dTc /dx3 ), is used to be interpreted as a quantity characterising the effect of the third component on the mutual solubility of the rst and the second component. Prigogine and Defay [28] state that when the solubility increases as a consequence of adding third component to the mixture this leads to a decrease in the critical temperature, while as the solubility decreases this leads to an increase of the critical temperature. So, the mutual solubility of hexane and methanol increases on adding a small amount of cyclohexane to the system. In a similar manner, a small amount of hexane increases the miscibility of the system cyclohexane + methanol (see the cosolvency effect [25]). Addition of acetone in the system water + phenol causes an increase of the critical temperature and the decrease of the mutual solubility of phenol and water. Prigogine [27], Prigogine and Defay [28] and before them Wagner [36] have studied this behaviour from a thermodynamic standpoint. In his work Prigogine [27] does not use the denition of the critical point based on Eqs. (6) and (7), but he employs equivalent functions (see [28]) 1 = x1 12 13 + x2 12 23 + x3 13 23 = 0 and 2 = 2 2 n2 2 =0
T ,p,1 ,n3

(11)

(12)

(13)

where ij is the derivative of the chemical potential of the ith component with respect to the number of moles of the jth component ij = ji = i nj (14)
T ,p,nk=j

Using these equations, Prigogine has shown that


x3 0

lim

dTc limx3 0 (1 /x3 )T ,x2 = dx3 limx3 0 (1 /T )x2 ,x3

(15)

Prigogines approach was applied to the case of the regular-solution model considered in this paper and given by Eqs. (3) and (4). The following relations were derived for 12 , 13 , 23 (n = n1 + n2 + n3

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

207

is the total number of moles):


2 n12 = 1 + 2x2 (1 x2 )b12 2x3 b13 + x3 (b12 + b13 b23 ) + 2x2 x3 (b23 b13 b12 ), 2 n13 = 1 + 2x3 (1 x3 )b13 2x2 b12 + x2 (b12 + b13 b23 ) + 2x2 x3 (b23 b13 b12 ), n23 = 1 + 2x2 (1 x2 )b12 2x3 (1 x3 )b13 + (1 x2 x3 + 2x2 x3 )(b23 b12 b13 )

(16)

The equation expressing the limiting value of dTc /dx3 (at x1 = x2 = 0.5, x3 = 0) is obtained from the corresponding relations for 1 (Eqs. (12) and (15)) with the temperature dependence of bij (T) being expressed by Eq. (3). dTc (b23 + b12 b13 )(b23 b13 b12 ) = x3 0 dx3 212 /T 2 lim (17)

On cyclic substitution of indexes, an analogous relation for the limiting slope of the critical curve in the binary system [2 + 3] can be obtained dTc (b13 + b23 b12 )(b13 b12 b23 ) = x1 0 dx1 223 /T 2 lim (18)

One should note an important fact: the limiting value of the composition change of the critical temperature depends only on values of b12 , b13 and b23 and on the derivative dbij /dT = ij /T 2 in the corresponding binary system ([1 + 2] or [2 + 3]). Prior to the analysis of Eqs. (17) and (18), the behaviour of the considered binary systems must be further specied. Positive deviations from the Raoults law in binary systems [1 + 2] or [2 + 3] and a critical solution temperature in the studied temperature interval for at least one of them will be supposed. In such case b12 > 0, b23 > 0. Moreover, if the subsystems [1 + 2] and [2 + 3] separate into two liquid phases at the given temperature, then b12 > 2 and b23 > 2. The temperature dependence of bij is given by Eq. (3). In case that the system shows an upper critical solution temperature ij > 0. For a system presenting a lower critical solution temperature ij < 0. The binary system [1 + 3] will be considered as homogeneous and showing either slightly positive or negative deviations from the Raoults law, i.e. b13 < 2. With regard to these restrictions, the following conclusions result from Eq. (17). The sign of the denominator is given by the value of 12 , i.e. systems with an upper critical solution temperature will be characterised by positive values of the parameter 12 and systems with a lower critical solution temperature will show negative values of the parameter 12 . Let us rst discuss the value of the rst term in the numerator of the Eq. (17), i.e. the value of (b23 + b12 b13 ). Then in the considered case, i.e. at the critical temperature of the binary system [1+2], b12 = 2 and also, according to the assumption of positive deviations from the Raoults law, b23 > 0. In agreement with our assumptions (i.e. b13 < 2), the expression (b23 + b12 b13 ) is always positive. The second term (b23 b13 b12 ) in the numerator of the Eq. (17) or, as the case may be the term (b13 + b23 b12 ) in Eq. (18), are more interesting. They can, under the above-described conditions, be positive or negative according to the value of b13 . The values of b13 that sets these terms to zero will be 0 0 referred to as b13(12) and b13(23) , i.e.
0 b13(12) = b23 2, 0 b13(23) = b12 2

(19)

208

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

At this critical value of b13 , (dTc /dx3 ) = 0. The derivative for a denite system [1 + 2] is affected by the value of the parameter b13 in the following way: dTc 0 >0 (20) b13 < b13(12) lim x3 0 dx3
0 b13 > b13(12) lim

dTc <0 x3 0 dx3

(21)

In the former case the critical temperature increases on addition of a third component to the system and it decreases in the latter case. The conditions given by Eq. (19) remain valid for binary systems [1 + 2] showing lower critical solution temperatures with the sole difference that the effect of the third component (i.e. the signs of Eqs. (20) and (21)) is opposite. Eq. (18) can be analysed in an analogous way. All the conclusions are summarised in Table 3. 4.2. Results of model calculations The preceding analysis concerned slopes of the critical curve at the concentration limits, i.e. in the vicinity of binary subsystems [1+2] and [2+3]. The shape of the critical curve at a specied concentration and temperature dependence of the Gibbs energy inside the ternary system can only be determined solving Eqs. (6) and (7). The following procedure has been adopted on modelling. Temperature dependencies of bij (T) (see Table 2) were selected for the systems [1 + 2] and [2 + 3]. For the model calculations of critical curves
Table 3 Relations for limiting values of critical curve slopes in critical points (I) At the critical point of systems [1 + 2], i.e. x3 = 0 and b12 = 2 (a) Systems [1 + 2] with UCST 0 b13(12) = b23 2 b13 < b13 >
0 b13(12) 0 b13(12)

limx3 0 (dTc /dx3 ) = 0 limx3 0 (dTc /dx3 ) > 0 limx3 0 (dTc /dx3 ) < 0 limx3 0 (dTc /dx3 ) = 0 limx3 0 (dTc /dx3 ) < 0 limx3 0 (dTc /dx3 ) > 0

(b) Systems [1 + 2] with LCST 0 b13(12) = b23 2 b13 < b13 >
0 b13(12) 0 b13(12)

(II) At the critical point of systems [2 + 3], i.e. x1 = 0 and b23 = 2 (a) Systems [2 + 3] with UCST 0 b13(23) = b12 2
0 b13 < b13(23) 0 b13 > b13(23)

limx1 0 (dTc /dx1 ) = 0 limx1 0 (dTc /dx1 ) > 0 limx1 0 (dTc /dx1 ) < 0 limx1 0 (dTc /dx1 ) = 0 limx1 0 (dTc /dx1 ) < 0 limx1 0 (dTc /dx1 ) > 0

(b) Systems [2 + 3] with LCST 0 b13(23) = b12 2 b13 < b13 >
0 b13(23) 0 b13(23)

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

209

Fig. 5. Shapes of the modelled critical curves in a ternary system in which are both binary subsystems characterised by an UCST.

in ternary systems, different values of the parameter b13 describing the subsystem [1 + 3] were chosen. For a better transparency of the results, the parameter b13 was considered as temperature-independent. By virtue of selected parameters bij , values T (=Tc ) and x2 (=x2c ) have been calculated from Eqs. (6) and (7) for specied values of x1 (=x1c 0, 0.5 ). It follows from the effectuated calculations that x2 differs only insignicantly from 0.5 (the regular solution is considered) and hence only diagrams T = T (x1 ) will further be discussed. 4.3. Binary systems [1 + 2] and [2 + 3] with upper critical solution temperatures The results of the rst series of the model calculations are shown in Fig. 5. The parameters in the binary subsystem [1 + 2] were described by a temperature dependence corresponding to variant II, variant I was employed for the subsystem [2 + 3] (see Table 2 for the variants). The acquired dependencies T = T (x1 ) for different values of the parameter b13 = 13 are shown in Fig. 5. The dependence of the critical temperature on the mole fraction of the rst component is almost linear in the case that b13 = 0 (the function is, in effect, slightly concave) and a monotonous critical curve is obtained. On increasing positive values of b13 the curve acquires a convex character. The critical curve 0 0 attains a zero value of the limiting slope (at x1 = 0) for b13 = b13(23) (b13(23) = 0.5 in the considered case) 0 and for b13 > b13(23) the curve shows a minimum deepening with increasing values of b13 . The concave character of the critical curve accentuates itself at negative values of b13 and the zero value of the slope of 0 the critical curve for b13(12) = b12 2 = 1.6428 2 = 0.3572 is attained at x3 = 0 or at x1 = x2 = 0.5. As the parameter b13 < 0.3572 then (limx1 0 (dT /dx1 ) > 0), (limx1 0.5 (dT /dx1 ) < 0) and there is a weak maximum on the critical curve. The co-ordinates of this maximum (or minimum in the preceding case), i.e. values of x1 , x2 , T for a specied set of parameters dening the Gibbs energy course, can be obtained by solving a set of equations (see Eqs. (6) and (7), and dT /dx1 = 0, Eq. (11)): D = 0, D = 0, D x1 D x2 D x2 D x1 =0 (22)

210

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

Fig. 6. Shapes of the modelled critical curves in a ternary system in which are both binary subsystems characterised by a LCST.

4.4. Binary systems [1 + 2] and [2 + 3] with lower critical solution temperatures Results of the second series of calculations are shown in Fig. 6. The considered binary systems had the same values of critical temperatures as those shown in Fig. 5. The difference is that binary systems in Fig. 5 were characterised by upper critical solution temperatures, while the subsystems [1 + 2] and [2 + 3] considered in this paragraph show lower critical solution temperatures. Parameters corresponding to variant V in Table 2 were employed for the system [1 + 2], variant VI was used for the system [2 + 3]. 0 It follows from the obtained results that the critical curve is monotonous at b13(23) = 2/3 < b13 < 0 0.5 = b13(12) (see Table 3). A minimum occurs on the critical curve at b13 < 2/3 (in the vicinity of the [2 + 3] system), and the curve shows a maximum at b13 > 0.5 (near to the subsystem [1 + 2]). Figs. 5 and 6 differ qualitatively in the fact that the heterogeneous region lies under the critical curve in Fig. 5, while it nds itself above the critical curve in Fig. 6. The critical curve with a maximum can be observed in the system 2-butanone (1) + water (2) + nicotine (3) [37]. The curve connects the hypothetical LCST in the subsystem 2-butanone (1) + water (2) and the LCST in the water (2) + nicotine (3) system. 4.5. Binary system [1 + 2] with an upper critical solution temperature, binary system [2 + 3] with a lower critical solution temperature A variant of a system whose binary heterogeneous subsystems [1 + 2] and [2 + 3] show different types of critical points is given in Fig. 7. Variant I (see Table 2) was employed in the binary subsystem [1 + 2]. Lower critical solution temperature and parameters corresponding to variant V (Table 2) were considered in the system [2 + 3]. Sassen et al. studied a ternary system [7] with similar behaviour type but as function of pressure. The lower critical solution temperature in the binary subsystem [2 + 3] shown in Fig. 7 is 50 K higher than the upper critical solution temperature in the binary subsystem [1 + 2]. The parameter 0 0 b13 attains the following limiting values: b13(12) = 2/3, b13(23) = 0.3572 and b13S = 1.0239 (see further).

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

211

Fig. 7. Shapes of the modelled critical curves in a ternary system in which the subsystem [1 + 2] is characterised by an UCST, and the subsystem [2 + 3] by a LCST. It holds TLC > TUC : b13 = 0.5, curves 1; b13 = 0.3572, curves 2; b13 = 0.5, curves 3; b13 = 2/3, curves 4; b13 = 0.9, curves 5; b13 = 1.0237, curves 6; b13 = 1.5, curves 7.

According to the value of b13 the following shapes of critical curves and corresponding heterogeneous regions can be acquired:
0 (a) b13 > b13(23) (see curves 1 in Fig. 7): both of the branches of the critical curve are monotonous. Heterogeneous regions are adjacent to binary systems [2+3] and [1+2] at corresponding temperatures (T > TLC,23 for the system [2 + 3] and T < TUC,12 for the system [1 + 2]). The ternary system is homogeneous in the temperature interval T (TUC,12 , TLC,23 ). The ternary system ethanol (2) + water (2) + triethylamine (3) can be cited as an example if a hypothetical low value of the upper critical solution temperature is attributed for the system ethanol (1) + water (2). 0 (b) b13 = b13(23) (see curves 2 in Fig. 7): the critical curve starting in the binary system [2 + 3] shows a slope equal to zero. 0 0 (c) b13(12) < b13 < b13(23) (see curves 3 in Fig. 7): the critical curve starting in the binary system [2 + 3] presents a weak minimum. This leads to the consequence that the heterogeneous region adjoins the binary system [2 + 3] at temperatures T > TLC,23 . The ternary system forms a closed binodal curve at temperatures Tmin < T < TLC,23 , where Tmin is the temperature corresponding to the minimum on the critical curve. 0 (d) b13 = b13(23) (see curves 4 in Fig. 7): the critical curve starting in the binary system [1 + 2] shows a slope equal to zero. 0 (e) b13(12) > b13 < b13S (see curves 5 in Fig. 7), where b13S is the value of b13 given by relations derived in Appendix C at which the two branches of the critical curve meet. Both branches of the critical curve show extremes. Temperatures corresponding to the minimum (on the branch starting in the binary system [2 + 3]) and to the maximum (on the branch starting in the binary system [1 + 2]) will be referred to as Tmin and Tmax , respectively. Ternary heterogeneous regions schematically shown in Fig. 8 correspond to values of b13 belonging to the considered interval. Such behaviour has, so far, not been observed in any system. (f) b13 = b13S : both critical curves touch asymptotically (see curves 6 in Fig. 7). (g) b13 < b13S (curves 7 in Fig. 7): an interconnection of both the heterogeneous regions is observed and a closed region of limited miscibility is formed.

212

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

0 Fig. 8. Heterogeneous regions in a ternary system satisfying the condition b13(12) > b13 > b13S at different temperatures T: (A) T > TLC,23 ; (B) Tmin < T < TLC,23 ; (C) Tmin < T < Tmax ; (D) Tmax < T < TUC,12 ; (E) T < TUC,12 .

A similar inuence of the parameter b13 can be seen in Fig. 9. The temperature dependence of the parameter b12 (T) is the same as for the subsystem in Fig. 7: variant I in Table 2 was employed; variant IV 0 in Table 2 was used for the subsystem [2 + 3]. It holds for the considered binary systems that b13(12) = 0 0.1133, b13(23) = 0.0862, b13S = 0.1995. With decreasing value of the parameter b13 the following shapes of the critical curve and of the heterogeneous region can be obtained: (a) b13 > b13S (curves 1 and 2 in Fig. 9): both branches of the critical curve show a monotonous course. Ternary heterogeneous regions are adjacent to corresponding binary systems. (b) b13 = b13S (curves 3 in Fig. 9): both branches of the critical curve meet and are asymptotic to preceding (see (a)) and following (see (b)) curves. 0 (c) b13(12) < b13 < b13S : both of the branches of the critical curve show less accentuated extremes. The branch starting in the [1 + 2] system shows a minimum, the branch corresponding to the [2 + 3] system shows a maximum. The heterogeneous region is band-shaped at temperatures lying between the minimum and the maximum. One heterogeneous region adjacent to the corresponding binary systems occurs in the ternary system out of this temperature interval. 0 (d) b13 = b13(12) (curves 4 in Fig. 9): the branch of curve 5 originating in the system [1 + 2] starts with a zero slope and then becomes monotonous. 0 0 (e) b13(12) < b13 < b13(23) : the critical curve starting in the binary system [1 + 2] is monotonous and the curve starting in the system [2 + 3] keeps its maximum. 0 (f) b13 = b13(23) : the critical curve starting in the system [2 + 3] looses its maximum.

Fig. 9. Shapes of the modelled critical curves in a ternary system in which the subsystem [1 + 2] is characterised by an UCST, and the subsystem [2 + 3] by a LCST. It holds TLC < TUC : b13 = 0.5, curves 1; b13 = 0.3, curves 2; b13 = 0.1995, curves 3; b13 = 0.1133, curves 4; b13 = 0, curves 5; b13 = 0.25, curves 6.

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

213

Fig. 10. Ternary system with an S-shaped LLE critical curve. Both binary subsystems are characterised by an UCST.
0 (g) b13 < b13(23) (see curves 5 and 6 in Fig. 9): both of the branches of the critical curve are monotonous. The ternary heterogeneous region is adjacent to the binary system [1 + 2] at temperatures T < TLC,23 and it adjoins the binary system [2 + 3] at T > TUC,12 . The heterogeneous region is band-shaped with no critical point in between these two extreme temperatures. Such behaviour can be observed in systems phenol (1) + water (2) + triethylamine (3) and diethylether (1) + water (2) + triethylamine (3) [23] (the existence of an upper critical solution temperature in the system diethylether (1) + water (2) is assumed).

4.6. Binary systems [1 + 2] and [2 + 3] with upper critical solution temperatures, ternary system with an S-shaped critical curve Fig. 10 represents a ternary system in which the critical curve is S-shaped. This was achieved using the following temperature dependencies of parameters bij (T): variant II in the binary system [1 + 2], variant I in the system [2 + 3] and variant III in the system [1 + 3] (for the individual variants see Table 2). 0 The considered dependence b13 (T) satises previously mentioned conditions: b13(12) = 0.3572 > 0 b13 (T = TUC,12 ) = 0.4 and b13(23) = 0.5 < b13 (T = TUC,23 ) = 0.6 and, in agreement with before mentioned ndings, leads to the S-shaped course of the critical curve (see Fig. 10). 5. Discussion The questions asked in the introduction can be now discussed. It follows from the calculations and from Eqs. (17) and (18) that the character of the shape of the critical curve connecting the critical points of binary subsystems [1 + 2] and [2 + 3] does not depend on whether the systems show an upper or a lower critical solution temperature (compare Figs. 5 and 6). In both cases the major inuence on the shape of the curve is depending on the third binary subsystem 0 0 [1 + 3]. If the parameter b13 does not exceed limits given by values of b13(12) and b13(23) , the critical curve shows a monotonous course. 0 A more or less accentuated minimum occurs on the critical curve at b13 > b13(12) (TUC,12 < TUC,23 ) or 0 at b13 > b13(23) (TUC,12 > TUC,23 ) in case that systems [1 + 2] and [2 + 3] show an upper critical solution

214

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

0 temperature. The critical curve can, on the other hand, show a maximum at b13 < b13(12) (TUC,12 > TUC,23 ) 0 or at b13 < b13(23) (TUC,12 < TUC,23 ). These conclusions are in agreement with the behaviour of previously mentioned systems. Upper critical temperatures of both of the heterogeneous binary subsystems differ signicantly in the system aniline (1) + water (2) + phenol (3) [12,13] (TUC,12 = 441 K, TUC,23 = 337.05 K). The binary system aniline (1) + phenol (2) shows negative deviations from Raoults law with b13 0.35 (this value was determined using the azeotropic point), deviations being relatively small to lead to an appearance of a maximum on a critical curve. Upper critical temperatures of heterogeneous binary subsystems in the system cyclohexane (1) + methanol (2) + hexane (3) are very close to one another (TUC,12 = 318.93 K, TUC,23 = 306.5 K). Hence, even very small positive deviations from ideal behaviour in the binary system cyclohexane (1) + hexane (3) [15,38] are sufcient to lead to a creation of a minimum on the critical curve. Hradetzky and Lempe [15] have published data which lead to an approximate value of b13 0.1. The binary subsystem acetone (1) + water (2) in the system acetone (1) + water (2) + phenol (3) is homogeneous at all temperatures [19,20]. However, the system shows large positive deviations from the Raoults law (T = 300 K, b12 1.75) and can be looked upon as lying on the brink of immiscibility. The system acetone (1) + phenol (3) shows large negative deviations (b13 < 1) [19,20,38] which are obviously sufcient for the creation of a maximum on the critical curve. On studying ternary systems characterised by two heterogeneous binary subsystems [1 + 2] and [2 + 3] showing different types of critical solution temperatures (upper versus lower) it was found that even in cases where it holds that TLC,23 < TUC,12 (see Fig. 9) it was possible that the heterogeneous regions would not touch. The two regions merge at b13 = b13S . Then, in case that for the binary system [1 + 3] b13 = 13 = f (T ), the value of b13S can be determined using the relation 0 0 b13S = b13(12) + b13(23)

(23)

It holds, for example, for the system in Fig. 7 that b13S = 0.3572 + (0.6666) = 1.0238. Heterogeneous regions adjacent to corresponding binary subsystems remain separate at b13 > b13S . The regions merge at b13 < b13S and form a closed binodal curve in case that TLC,12 > TUC,23 . The ternary system tetrahydrofuran (1) + water (2) + phenol (3) shows such behaviour. As was already said, the system is heterogeneous even at temperatures T (TUC,23 , TLC,12 ). According to the regular-solution model applied in this paper, such behaviour is only possible if the system tetrahydrofuran (1) + phenol (3) showed large negative deviations from the Raoults law, which in this particular case corresponds to reality [24,38]. Another example can also demonstrate the acquired ndings. The system acetic acid (1) + cyclohexene (2) + dimethylformamide (3) form a closed region of limited miscibility. No heterogeneous region characterised by an upper critical solution temperature was observed in the binary subsystems [21], but a similar system acetic acid + cyclohexane shows a TUC 273 K. Its thermodynamic behaviour (H E > 0), makes us suppose that the upper critical temperature in the system acetic acid (1) + cyclohexene (2) would be lower than the eutectic temperature in corresponding binary subsystems. The system acetic acid (1) + dimethylformamide (3) however shows large negative deviations from ideal behaviour, which causes that the heterogeneous region appears at much higher temperatures than hypothetical upper critical solution temperatures in both of the binary subsystems [1 + 2] and [2 + 3]. The experimentally found temperature dependence of the miscibility is in agreement with these facts: the heterogeneous region gradually diminishes on increasing temperature to disappear totally in the end [21].

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

215

6. Conclusion Making general conclusions regarding results of effectuated descriptions of the critical curve by means of the regular-solution model, the following facts may be stated. (a) A critical curve connecting critical points of two binary subsystems [1+2] and [2+3], characterised by the same type of the critical solution temperature (lower or upper), is monotonous if the binary system [1 + 3] shows relatively small deviations from the ideal behaviour (limits for the regular-solution model are specied in Table 3). (d) If the systems [1 + 2] and [2 + 3] show upper critical solution temperatures, then sufciently high positive deviations from ideality in the binary system [1 + 3] cause the creation of a minimum on the critical curve and the ternary heterogeneous region splits at a temperature lower than the lower of the two critical temperatures. Conversely, sufciently large negative deviations from the Raoults law result in the creation of a maximum on the critical curve and in the formation of a closed limited miscibility region at temperatures higher than the critical temperatures of both of the heterogeneous binary subsystems. In case that systems [1 + 2] and [2 + 3] show lower critical solution temperatures, opposite effects can be observed. (c) In case that binary systems [1+2] and [2+3] show different types of the critical solution temperature, two branches of the critical curve are obtained. Each of the branches originates in the critical point of the corresponding binary system. These branches may be monotonous, but may also show minima or maxima depending on the non-ideality of the binary subsystem [1 + 3]. Equations derived for the limiting slopes in critical points are valid in this case as well. A relation which, when fullled, leads to the junction of both of the heterogeneous regions adjacent at sufciently high values of b13 to heterogeneous binary subsystems, is derived in Appendix C. (d) A closed region of limited miscibility found in some ternary systems is obviously a consequence of large positive or large negative deviations from ideality in the binary system [1 + 3], which lead to the appearance of the critical curve at temperatures outside of the temperature interval determined by critical temperatures of the two heterogeneous binary subsystems. (e) A closed limited miscibility region can also occur in case that binary systems [1 + 2] and [2 + 3] are homogeneous in the whole concentration range (see Appendix A). List of symbols b13S bij 0 0 b13(12) , b13(23) D D (G11 )x0 Gij Gij k Gm GE G m,i

value of the parameter b13 at which the two branches of the critical curve meet parameter of the regular-solution model (Eq. (1)) for the system i + j special value of the parameters b13 (see Eq. (19)) determinant dened by Eq. (6) determinant dened by Eq. (7) the minimum value of G11 the second derivative of the molar Gibbs energy with respect to mole fractions xi and xj the third derivative of the molar Gibbs energy with respect to mole fractions xi , xj and xk molar Gibbs energy molar excess Gibbs energy molar Gibbs energy of the pure component i at the temperature and pressure of the system

216

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

ni T Tc TLC,12 , TLC,23 Tmax Tmin TUC,12 , TUC,23 x1c xi xi , xi Greek letters ij , ij , ij i ij 1, 2

number of moles of ith component temperature critical temperature lower critical solution temperature for binary subsystem [1 + 2] and [2 + 3], respectively critical curve maximum temperature critical curve minimum temperature upper critical solution temperature for binary subsystem [1 + 2] and [2 + 3], respectively critical mole fraction of the component 1 mole fraction of the ith component mole fraction of the ith component in the conjugated phases

constants in the equations for the temperature dependence of the b parameter (see Eqs. (3) and (4)) chemical potential of the ith component derivative of i with respect to the mole number of the jth component (see Eq. (14)) functions dened by Eqs. (12) and (13) [28]

Acknowledgements The authors gratefully acknowledge nancial support form the Ministry of Education of the Czech Republic under Grant No. CB MSM 2234 00008. Appendix A. Modelling with an extended temperature dependence of the model parameters Binary systems presenting a closed region of limited miscibility cannot be modelled by Eq. (3). ConE sidering a non-zero and constant Cp , the relation must be extended as follows: ij bij = ij + + ij ln T (A.1) T Parameters of Eq. (A.1) can be evaluated using LCST and UCST (bij = 2) and a solubility at a selected temperature. The lower critical solution temperature TLC = 345 K, the upper critical solution temperature TUC = 411 K [39] and b (T = 378 K) = 2.1 (resulting from x1 = 0.3146) permitted in the system tetrahydrofuran + water the following relation to be obtained: bij = 183.716 9856.18 26.2079 ln T T (A.2)

2 Eq. (A.1) is also to be applied on systems where the temperature dependence of [ 2 (G/RT)/x1 ] shows a positive minimum. The binary systems 3-methylpyridine + water, 4-methylpyridyne + water and acetone + water belong to such systems. The excess enthalpy is negative at lower temperatures, positive at higher ones and the minimum of the second derivative of the Gibbs energy with respect to composition

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

217

Fig. 11. Modelled LLE critical curves in ternary systems: curve 1, methanol (1) + water (2) + tetrahydrofuran (3) (b12 = 0.5, b13 = 0.5); curve 2, ethanol (1) + water (2)+tetrahydrofuran (3) (b12 = 1.2, b13 = 0.5); curve 3, 1-propanol (1) + water (2) + tetrahydrofuran (3) (b12 = 1.8, b13 = 0.3); curve 4, 1-butanol (1) + water (2) + tetrahydrofuran (3) (b12 = 0.095+834.3/T , b13 = 0.2); curve 5, dimethylsulfoxide (1) + tetrahydrofuran (2) + water (3) (b12 = 1.2, b13 = 2.5); curve 6, dimethylsulfoxide (1) + tetrahydrofuran (2) + water (3) (b12 = 1.2, b13 = 3).

occurs at approximately 35 C ((G11 )x0 0.61) in the system acetone + water [3]. The system acetone + water can therefore be considered as homogeneous in the whole temperature and concentration ranges. A.1. Binary system [1 + 2] homogeneous or with an upper critical solution temperature and binary system [2 + 3] with a closed region of limited miscibility The binary system tetrahydrofuran (2) + water (3) forms a closed region of limited miscibility and offers another possibilities of shapes of critical curves sketched in Fig. 11. Ternary systems in which the third component is methanol, ethanol, 1-propanol, 1-butanol and dimethylsulfoxide are discussed. Values of parameters used in calculations characterise the behaviour of corresponding binary systems only roughly and are given in the legend to Fig. 11. Eq. (A.2) was employed for the system tetrahydrofuran (2) + water (3). It follows from the effectuated calculations that the heterogeneous region is relatively small in case of methanol, ethanol and 1-propanol and that the critical curve is closed-shaped (see curves 1, 2 and 3 in Fig. 11). Similar critical line was found by Kumar et al. in the system 3-methylpyridine + water +heavy water [40]. The critical curve presents two branches in the system containing 1-butanol (see curve 4). A less accentuated minimum is assumed to occur on the upper branch. A closed region of limited miscibility at 20 C occurs in the system dimethylsulfoxide (1) + tetrahydrofuran (2) + water (3) [41]. The binary system dimethylsulfoxide (1) + tetrahydrofuran (2) presents positive deviations from the Raoults law [38] and the parameter b12 = 1.2 was employed to characterise the system. On the other hand, the system dimethylsulfoxide (1) + water (3) shows large negative deviations from the Raoults law [38] and two values of the parameter b13 were employed for its description: b13 = 2.5 and b13 = 3. Calculated critical curves are represented in Fig. 11 and are referred to as curves 5 and 6. It follows from the calculations that the occurrence of a closed heterogeneous region in the ternary system dimethylsulfoxide (1) + tetrahydrofuran (2) + water (3) at 20 C, a temperature lower than the lower critical solution

218

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

Fig. 12. Shape of the modelled critical curve in a ternary system in which all binary subsystems are homogeneous.

temperature in the system tetrahydrofuran (2) + water (3), is caused by large negative deviations from the Raoults law in the system dimethylsulfoxide (1) + water (3). A.2. All binary systems homogeneous and ternary system with a closed region of limited miscibility The extreme inuence on the occurrence of the heterogeneous region can be demonstrated by a system whose critical curve is sketched in Fig. 12. In this case, the following dependence of the parameter bij was employed for the binary subsystems [1 + 2] and [2 + 3]: bij = 568.144 25539.9 84.349 ln T T (A.3)

which was obtained using parameters bij = 1.5, 1.9 and 1.6 at temperatures 275, 300 and 330 K, respectively. The dependence shows a maximum at bij = 1.9063 and T = 302.79 K. Binary systems are thus homogeneous in the whole concentration and temperature range. Nevertheless, in accordance with the previously acquired ndings [28,33], if the b13 is sufciently low a heterogeneous region appears in the ternary system. A curve for b13 = 3 is shown in Fig. 12. Appendix B. Expressions for D and D , and their derivatives D= 1 2b12 2b13 2b23 2 2 2 b12 b13 b23 + 2b12 b13 + 2b12 b23 + 2b13 b23 x1 x2 x3 x3 x2 x1
2 3x1 1 b12 [x3 x2 (1 + 2x1 )] b13 [x2 x3 (1 + 2x1 )] 4b23 + + 2 2 2 2 2 2 2 2 x1 x 2 x3 x1 x2 x3 x1 x2 x3 x1

(A.4)

D =

b23 [x1 (1 x1 ) + 2x2 x3 (2 x1 )] 2b12 (b12 + b23 b13 ) 2b13 (b13 + b23 b12 ) + + 2 2 2 2 2 x1 x 2 x 3 x3 x2 (A.5)

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

219

D x1 D x2 D T

x2 ,T

2 2 x1 x3 + 2b23 x2 x3 2b12 x1 x2 2 2 x 1 x2 x 3 2 2 x2 x3 + 2b13 x1 x3 2b12 x1 x2 2 2 x1 x 2 x 3

(A.6)

x1 ,T

(A.7)

x1 ,x2

2 T2

12 13 23 + 13 b12 + 12 b13 + 23 b12 + 12 b23 x3 x2 x1 (A.8)

+23 b13 + 13 b23 12 b12 13 b13 23 b23

D x1

x2 ,T

3x1 (2x1 x3 ) + 2(x3 x1 ) b12 [x3 (2x3 1) + 2x1 x2 (1 + 2x1 )] 3 2 3 2 2 3 x1 x2 x3 x1 x 2 x 3 b13 [x3 (x2 x3 ) + x1 x3 (1 + 2x1 ) 2x1 x2 ] 4b12 (b12 + b23 b13 ) + 3 2 2 3 x1 x2 x3 x3 +
2 2 2 b23 [2x2 x3 (x1 4) + x1 x3 (1 + 4x2 2x1 x2 ) + 2x1 (x1 1)] 8b23 + 3 3 2 3 x 1 x2 x 3 x1

(A.9)

D x2

x1 ,T

2 2(3x1 1)(x2 x3 ) 2b12 [x3 (x1 x2 x3 + x2 ) x2 (1 + 2x1 )] + 3 3 2 3 3 x 1 x2 x3 x1 x2 x3 2 2b13 [x3 (1 + 2x1 ) + x2 (x2 x3 x1 x3 )] 4b12 (b12 b13 + b23 ) + 3 3 3 x1 x 2 x 3 x3 2b23 [x1 x3 (x1 x2 x3 1) + x2 x3 (x1 x2 2x2 + 2x3 ) + x1 x2 (1 x1 )] 2 3 3 x1 x 2 x 3 4b13 (b12 b13 b23 ) + 3 x2

(A.10)

D T

x1 ,x2

1 12 [x3 x2 (1 + 2x1 )] 13 [x2 x3 (1 + 2x1 )] + 2 2 2 2 T2 x1 x 2 x 3 x1 x2 x3 23 [2x2 x3 (2 x1 ) x1 (1 x1 )] 212 (b12 + b23 b13 ) + + 2 2 2 2 x1 x2 x3 x3 213 (b13 + b23 b12 ) 8b23 23 2b12 (12 + 23 13 ) + + 2 2 2 x2 x1 x3 2b13 (13 + 23 12 ) + 2 x2

(A.11)

220

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

Fig. 13. Schematic dependence of the temperature on b13 at a constant mole fraction x1 : (A) regular points only; (B) one double singular point S; (C) two singular points S1 and S2 .

Appendix C. Conditions for splitting or conjunction of two critical curves Two conditions (see Eqs. (6) and (7)) are satised in the critical point D(x1 , x2 , T , b13 ) = 0, D (x1 , x2 , T , b13 ) = 0 (A.12)

Let x1 and b13 be independent variables. The point (x10 , x20 , T0 , b130 ) satisfying Eq. (A.12) is a regular point if only one pair of explicit functions T = T (x1 , b13 ) and x2 = x2 (x1 , b13 ) satises Eq. (A.12) in a neighbourhood of the point (x10 , b130 ). In an opposite case, the point (x10 , x20 ,T0 , b130 ) is a singular point. It follows from the theorem of implicit functions that the following determinant E is equal to zero in the singular point. D x2 D x2 D T D T

E=

=0

(A.13)

Singular points S1 and S2 (see Fig. 13) satisfy set of three equations D = 0, D = 0, E = 0 for four unknowns x1 , x2 , T and b13 . Set of three functions x1 = x1 (b13 ), x2 = x2 (b13 ) and T = T (b13 ) is not unambiguously determined in the double singular point S and therefore it holds D x2 E1 = D x2 E x2 D T D T E T D x1 D x1 E x1 =0 (A.14)

in the point S. Values of x1c , x2c , Tc and b13S are obtained by solving the set of Eqs. (A.12)(A.14). References
[1] J.M. Sorensen, W. Arlt, LiquidLiquid Equilibrium Data Collection, Ternary and Quaternary Systems, Chemistry Data Series, vol. V, DECHEMA, Frankfurt am Main, Germany, 1980.

J.P. Nov k et al. / Fluid Phase Equilibria 208 (2003) 199221 a

221

[2] E.A. Macedo, P. Rasmussen, LiquidLiquid Equilibrium Data Collection, Chemistry Data Series, vol. V (Suppl. 1), DECHEMA, Frankfurt am Main, Germany, 1987. [3] J.P. Novk, J. Matou, J. Pick, LiquidLiquid Equilibria, Elsevier, Amsterdam, 1987. [4] A. Arce, A. Blanco, J. Martnez-Ageitos, I. Vidal, Fluid Phase Equilib. 109 (1995) 291297. [5] H.W.B. Roozeboom, F.A.H. Schreinemakers, Heterogenen Gleichgewichte. III. Heft, I, II, Vieweg u.Sohn, Braunschweig, 1911. [6] T. Adrian, S. Oprescu, G. Maurer, Fluid Phase Equilib. 132 (1997) 187203. [7] C.L. Sassen, A.G. Casielles, Th.W. de Loos, J. de Swaan Arons, Fluid Phase Equilib. 72 (1992) 173187. [8] L.P.N. Rebelo, Phys. Chem. Chem. Phys. 1 (1999) 42774286. [9] P.H. van Konynenburg, R.L. Scott, Philos. Trans. R. Soc. London 298 (1980) 495540. [10] Y.S. Wei, R.J. Sadus, Phys. Chem. Chem. Phys. 1 (1999) 43294336. [11] M. Bluma, U.K. Deiters, Phys. Chem. Chem. Phys. 1 (1999) 43074313. [12] F.A.H. Schreinemakers, Zeitschr. Phys. Chem. 29 (1899) 577602. [13] F.A.H. Schreinemakers, Zeitschr. Phys. Chem. 30 (1899) 460480. [14] H.N. Slimo, M.B.G. de Doz, Fluid Phase Equilib. 107 (1995) 213227. [15] G. Hradetzky, D.A. Lempe, Fluid Phase Equilib. 69 (1991) 285301. [16] A.C.G. Marigliano, M.B.G. de Doz, H.N. Slimo, Fluid Phase Equilib. 149 (1998) 309322. [17] A.C.G. Marigliano, M.B.G. de Doz, H.N. Slimo, Fluid Phase Equilib. 153 (1998) 279292. [18] V.P. Sazonov, N.I. Lisov, N.V. Sazonov, J. Chem. Eng. Data 47 (2002) 599602. [19] F.A.H. Schreinemakers, Z. Phys. Chem. 33 (1900) 7898. [20] F.A.H. Schreinemakers, Z. Phys. Chem. 39 (1902) 485510. [21] F. Becker, P. Richter, Fluid Phase Equilib. 49 (1989) 157167. [22] E. Raub, A. Engel, Z. Metallkd. 38 (1947) 1116. [23] P.A. Meerburg, Zeitschr. Phys. Chem. 40 (1902) 641688. [24] K. Rehk, J. Matou, J.P. Novk, Fluid Phase Equilib. 109 (1995) 113129. [25] A. Kordikowski, G.M. Schneider, Fluid Phase Equilib. 90 (1993) 149162. [26] E.A. Guggenheim, Mixtures, Clarendon Press, Oxford, 1952. [27] I. Prigogine, Bull. Soc. Chim. Belg. 52 (1943) 115123. [28] I. Prigogine, R. Defay, Chemical Thermodynamics, Longmans, London, 1954. [29] J.L. Meijering, Philips Res. Rep. 5 (1950) 333356. [30] J.L. Meijering, Philips Res. Rep. 6 (1951) 183210. [31] I.F. Hlscher, M. Spee, G.M. Schneider, Fluid Phase Equilib. 49 (1989) 103113. [32] A. Kordikowski, G.M. Schneider, Fluid Phase Equilib. 105 (1995) 129139. [33] J.P. Novk, P. Vo ka, J. Matou, J. Pick, Coll. Czech. Chem. Commun. 44 (1979) 34693489. n [34] R. Haase, Thermodynamik der Mischphasen, Springer, Berlin, 1956. [35] M. Modell, R.C. Reid, Thermodynamics and its Applications, Prentice-Hall, Englewood Cliffs, NJ, 1974, pp. 188230. [36] C. Wagner, Z. Phys. Chem. 132 (1928) 273294. [37] A.N. Campbell, E.M. Kartzmark, W.E. Falconer, Can. J. Chem. 36 (1958) 14751486. [38] J. Gmehling, U. Onken, W. Arlt, VaporLiquid Equilibrium Data Collection, Chemistry Data Series, vol. I (Suppl. 1), DECHEMA, Frankfurt am Main, 1981. [39] J. Matou, J. Hrnk, J.P. Novk, J. obr, Coll. Czech. Chem. Commun. 35 (1970) 19041905. cr [40] A. Kumar, S. Guha, E.S.R. Gopal, Phys. Lett. A 123 (1987) 489493. [41] T. Wolski, Rocz. Chem. 44 (1970) 22372242.

You might also like