Professional Documents
Culture Documents
The Isothermal Van't Hoff Equation For Phase Equilibria-A Forgotten Relation?
The Isothermal Van't Hoff Equation For Phase Equilibria-A Forgotten Relation?
The isothermal van’t Hoff equation for phase equilibria—A forgotten relation?
Ulrich K. Deiters ∗
Institute of Physical Chemistry, University of Cologne, Luxemburger Str. 116, D-50939 Köln, Germany
a r t i c l e i n f o a b s t r a c t
Article history: The extended van’t Hoff equations for dilute mixtures relate the differences of initial slopes of phase
Received 22 June 2012 boundary curves to the phase transition enthalpy or volume of the solvent. While the isobaric van’t
Received in revised form 22 August 2012 Hoff equation is frequently used in mixture thermodynamics, its isothermal counterpart has received
Accepted 27 August 2012
much less attention. We show that the isothermal van’t Hoff equation can be used to interpolate dew
Available online 11 September 2012
point curves in isothermal phase diagrams or to create a model-free consistency test for experimental
phase equilibrium data. Furthermore, an expression for the initial curvature of phase boundary curves is
Keywords:
derived.
Dilute mixtures
Van’t Hoff equation © 2012 Elsevier B.V. All rights reserved.
Consistency test
Phase equilibria
Limiting azeotropy
1. Introduction ∂ ln K Vm
=− , (3)
∂p RT
T
The physical–chemical laws of colligative properties date back
to the end of the 19th Century. In his famous article of 1887 [1] van’t where Hm and Vm are the enthalpy and volume change of
Hoff not only presented the theory of the osmotic pressure in dilute the process under consideration, and K is an equilibrium constant.
solutions, but also linked this phenomenon to the melting-point In the case of a solid–liquid phase equilibrium with a vanishing
depression of solutions and derived the equation amount of component 1, the first one of these equations reduces to
van’t Hoff’s equation, Eq. (1).
m Hm,2 1 1
x1 = − , (1) In his textbook of 1935 [6], van Laar explicitly describes the
R T T2sl calculation of the initial slopes of phase boundary curves with equa-
tions derived from these two:
where x1 is the (very small) mole fraction of a solute (1), and
m Hm,2 and T2sl are the enthalpy of melting and the melting tem- dx1 dx1 Hm,2
perature of the solvent (2); the solvent is assumed to form pure lim − =− (4)
x1 →0 dT dT RT 2
crystals upon freezing.
Previously, van’t Hoff had worked on the temperature and dx1 dx1 Vm,2
pressure dependence of chemical equilibrium constants [2]. His lim − =+ (5)
x1 →0 dp dp RT
theoretical explanation originally referred to reacting ideal gases
only, but Planck pointed out that this restriction was not neces- These equations might be called extended van’t Hoff’s equations,
sary [3].By 1894 it had become clear that melting-point depression, as they are not restricted to phase equilibria in which one phase is
boiling-point elevation, and the shifting of chemical equilibria were pure. Here Hm,2 and Vm,2 are the enthalpy and volume change,
manifestations of some universal laws. In his impressive review, respectively, of the phase transition of pure component 2. The
van Laar gave an axiomatic derivation of these laws together with derivatives are taken along the phase equilibrium curves.
a multitude of applications[4,5]: Eq. (4) appears in many textbooks thereafter, e.g., in the influ-
ential textbook of Lewis and Randall [7], and is frequently used in
∂ ln K Hm
=+ (2) metallurgy or mineralogy to check the validity of phase diagrams.
∂T p RT 2 Eq. (5) is mentioned in the textbook of Lewis and Randall,
too, but merely in the context of assessing the influence of pres-
sure on melting equilibria involving pure solid phases, i.e., with
∗ Tel.: +49 221 470 4543; fax: +49 221 470 4900. dx1 /dp = 0. Here, however, the pressure effect is small, and the
E-mail address: ulrich.deiters@uni-koeln.de textbook conveys the impression that this equation is of little
0378-3812/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.fluid.2012.08.028
U.K. Deiters / Fluid Phase Equilibria 336 (2012) 22–27 23
practical importance. From then on, the equation seems to have Table 1
Parameters of the Peng–Robinson of state used for the computation of the figures. T*:
disappeared from the textbooks.
characteristic temperature, v∗ : characteristic volume, ω: acentric factor (commonly
We hope to show in this work that this equation can be rather used parameters of this equation of state are obtained as ac = 8RT ∗ v∗ , b = v∗ ).
useful.
Substance T* (K) v∗ (cm3 mol−1 ) ω
We will now consider two-phase equilibria of binary mixtures Division of Eq. (5) by (dx1 /dp)∞ leads to
where both phases are not pure. Then Eq. (5) can be quickly derived
dx1 Vm,2 dp
from Eq. (3) by applying the latter to the enhancement factor (“K lim −1= . (12)
factor”) of component 2, x1 →0 dx1 RT dx1
∞
Here the Vi˛ (˛= , ) denote partial molar volumes of the coex- where p2 is the vapour or sublimation pressure of component 2,
isting phases, and the ˛ the chemical potentials of component and then Eq. (14) reduces to
1
1. Readers interested in the derivations or applications of these
1 dp
equations are referred to textbooks, e.g., [9] (Section 4.6) or [10] K1,∞ =1+ . (16)
p2 dx1
(Section 5.5). In the limit of x1 → 0, the V2˛ approach Vm,2 ˛
; fur- ∞
thermore, the derivatives of the chemical potentials diverge as Switching the indices in Eqs. (14) and (16) gives the analogous
(∂˛ 1
/∂x1 ) (x1˛ )−1 . Hence Eqs. (8) and (9) reduce to [10] relations for K2,∞ .
dp RT (K1,∞ − 1)
lim = . (10) 2.2. Application: interpolation of phase diagrams
x1 →0 dx1 Vm,2 − Vm,2
A recurrent problem in mixture thermodynamics is having to
and
work with phase diagrams which are given as a table of data points
dp RT (1 − 1/K1,∞ ) only, so that interpolation becomes necessary. Then, however, the
lim = . (11)
x1 →0 dx1 Vm,2 − Vm,2 “density” of the data points becomes an issue.
For the following discussion of interpolation errors of phase
K1,∞ represents the enhancement factor of component 1 at infi- boundary curves it is evidently important to know the correct
nite dilution. Taking the difference of the reciprocals of these two results. We therefore work here with a computed phase dia-
equations then yields Eq. (5). gram of {methane + propane}, which has been obtained with the
The important feature of Eqs. (4) and (5) is that they are exact Peng–Robinson equation of state [11] (using the mixing rules pro-
and do not depend on equations of state, activity coefficients, or posed by its authors) by means of the ThermoC program [12]; the
other models. model parameters are given in Table 1. In this diagram, six equilib-
rium states on the propane side have been marked; the task is now
to generate the “true” phase boundary curves from these selected
1
D.P. Konovalov published his first article in a German journal, transliterating his states. This is easy for the bubble point curve, where the selected
name as “D. Konowalow” [8]. Therefore this spelling is also found in the literature. data points on the bubble point curve are closely spaced at 0.01
24 U.K. Deiters / Fluid Phase Equilibria 336 (2012) 22–27
10 100
methane + propane 50
8 40
250 K 30
20
6
p/MPa
K1
10
4 l lg
5
4
3
2
2
g
0 1
0 0.2 0.4 0.6 0.8 1 0.1 0.2 0.3 0.4 1 2 3 4 5 10
x1 p/MPa
Fig. 1. Isothermal phase diagram of the {methane + propane} system at 250 K. —: Fig. 3. K1 as a function pressure for the system {methane + propane} at 250 K (dou-
calculated with the Peng–Robinson equation of state; ◦: data points used for inter- ble logarithmic). —: true curve (computed from an equation of state), · · ·: spline
polation (see Section 2.2). interpolation of the dew point data points indicated in Fig. 1 with x1 > 0, - - -: spline
interpolation including the K1,∞ value according to Eq. (14). The insert is scaled up
by a factor of 10.
mole fraction units. The distances between the corresponding data
points on the dew point curve, however, can exceed 0.3 mole frac-
tion units. Consequently, an interpolation of the dew point curve is In a case like this it is better to interpolate the function
difficult (Fig. 1). lg p → lg K1 , as shown in Fig. 3. This diagram contains
In such a situation it helps to know the initial slope of the dew
point curve. Fig. 2 shows the error of interpolation attempts with (a) the “true” curve, i.e., the curve obtained by solving the phase
cubic spline functions: Starting from x1 = 0, the natural spline func- equilibrium conditions,
tion through the indicated data points quickly reaches a pressure (b) a natural cubic spline interpolation of the dew point data indi-
error of 1.5 kPa, which is equivalent to approximately 0.5%, whereas cated in Fig. 1 with the exception of x1 = 0, extrapolated to lg p2 ,
the clamped spline function reaches 0.5 kPa only. At larger values the vapour pressure of propane,
of x1 , however, both spline functions show the typical oscillations (c) and a natural cubic spline interpolation of the dew point data
that occur when spline interpolation is attempted at a too low data as before plus the K1,∞ value calculated from Eq. (14).
density.
For this example, (dp/dx1 )∞ had been determined from a cubic Evidently, curve (c) is practically on top of the exact data,
spline interpolation of the bubble point curve. whereas the curve (b) deviates significantly.
At this point the phase equilibrium calculation method of
Hu et al. [13] should be mentioned, which is also based on the
interpolation of experimental bubble point data. Their method
4 is model-free and can be applied over the whole concentration
range, but is restricted to vapour–liquid equilibria at low pressures,
whereas the interpolation scheme presented here is for dilute mix-
tures only, but can in principle applied to other kinds of equilibria
2
and at arbitrary pressures.
δp/kPa
2 3.1. Theory
Fig. 4. Isothermal phase diagram of the {carbon dioxide + propane} system at and rearranging leads to
277.59 K. ◦: experimental data [14], - - -: tangents to the dew point curve as predicted
3
by Eq. (14). 3 d2 p d2 p 1 dVm,2 dVm,2 dp
K1,∞ = − −
dx1 2 dx1 2 RT dp dp dx1
∞ ∞ ∞
if the data points are sparse or the dew point curve is strongly 3
curved at its end. In such cases it is advantageous to extrapolate
2
d p 1
dp
= + Vm,2 T,2 − Vm,2 T,2 , (20)
the experimental ln p → ln Ki data towards the vapour pressure dx1 2 RT dx1
∞ ∞
of the other compound, p3−i . This extrapolation can be per- ˛
where the T,2 denote the isothermal compressibilities of compo-
formed by spline interpolation, Lagrange interpolation, or fitting
nent 2 and K1,∞ can be taken from Eq. (14).
a low-order polynomial. The best method depends on the quality
For low-pressure vapour–liquid equilibria, this equation
of the experimental data, and it is advisable to vary the method
reduces to
or the number of data used for this extrapolation in order to 3
establish the error margins of the extrapolated value. 3 d2 p d2 p 1 dp
K1,∞ = + , (21)
dx1 2 dx1 2 p2 2 dx1
∞ ∞ ∞
As an example, we present an experimentally obtained isother-
mal phase diagram of the {carbon dioxide + propane} system at or, after substitution of Eq. (16), to
277.59 K (Fig. 4) [14]. The calculated slopes and Ki,∞ values are
d2 p 1 d2 p
shown in Table 2. Evidently, the K1,∞ (at x1 → 0) predicted with Eq. = + (K1,∞ − 1)3 p2 . (22)
(14) agree reasonably well with the extrapolated values. The K2,∞ dx1 2 ∞
3
K1,∞ dx1 2 ∞
values (at x1 → 1), however, significantly disagree. Hence we must The initial curvature of the dew point curve therefore depends
conclude that one of the experimental data points in this region is on the enhancement factor and the curvature of the bubble point
in error. curve: If the latter is zero, K1,∞ > 1 (i.e., component 2 is less volatile
Here a word of caution is in order: According to Eq. (14), than component 1) implies a positive initial curvature; if the bub-
a very low enhancement factor Ki,∞ 1 value can result only ble point curvature dominates, the curvature of the dew point curve
if (dp/dxi )∞ Vm,3−i /(RT ) ≈ −1. In such a case, however, Ki,∞ is equals that of the bubble point curve, multiplied with K1,∞ −3
, i.e., a
obtained as a difference of numbers of almost equal size, and then large enhancement factor causes a low curvature.
the extrapolation or the consistency test may be unreliable.
3.2. Application: limiting azeotropy
2.4. Application: binary mixture at a pure-fluid critical point
A limiting azeotrope is an azeotrope located at the mole fraction
For completeness’ sake we briefly consider a special case which xiaz = 1; it is also called “pure azeotrope” or “border azeotrope”. In
can either be described as an isothermal vapour–liquid equilibrium phase diagrams, limiting azeotropes are the starting or end points
at infinite dilution of component 1 at the critical temperature of of azeotropic curves.
component 2 or as a binary critical point at infinite dilution of com- At a limiting azeotrope, K1,∞ = 1 and (dp/dx1 )∞ = 0; hence Eq.
ponent 1. At this point, the bubble point curve and dew point curve (20) predicts that not only the slopes, but also the curvatures of the
detach from the x2 = 1 axis of the corresponding px phase diagram. bubble point curve and the dew point curve must be equal:
This case had been studied by the school of van der Waals at the az az
begin of the 20th Century and later by Krichevsky; for a historical d2 p d2 p
= (23)
overview see [15]. It turns out that, in the instance of detaching, the dx1 2 ∞
dx1 2 ∞
bubble point curve retains a finite slope,
Consequently, the bubble point curve and the dew point curve meet
2
in a very sharp cusp.
dp ∂ Am
=− , (17) This is illustrated in Figs. 5 and 6 for a binary van der Waals
dx1 ∂Vm ∂x1
∞ T,∞ mixture, i.e., a mixture described by the van der Waals equation
26 U.K. Deiters / Fluid Phase Equilibria 336 (2012) 22–27
Table 2
Enhancement factors at infinite dilution, Ki,∞ , of the {carbon dioxide + propane} system at 277.59 K [14]: comparison of the prediction with Eq. (14) and extrapolations of the
experimental data by cubic spline or Lagrange extrapolation.
5 Table 3
Parameters of the van der Waals equation of state used for the computation of
Figs. 5–8. T*: characteristic temperature, v∗ : characteristic volume [the commonly
used “van der Waals parameters” are obtained by a = 2 RT ∗ v∗ , b = v∗ ].
The critical point of substance A is at 5 MPa and 200 K, of B and C at 6.25 MPa and
250 K.
2
As the Taylor series of the phase boundary curves at the limiting
azeotrope is
dp 1 d2 p
1 p = p1 + x2˛ + (x2˛ )2 + · · ·, (24)
dx2˛ 2 d(x2˛ )2
∞
∞
=0
5
2
0
4
–2
ln((pσ -p )/MPa)
3
–4
p/MPa
–6 2
–8
1
–10
–12 0
–6 –5 –4 –3 –2 –1 0 0 0.2 0.4 0.6 0.8 1
ln x2 x1
Fig. 6. Double-logarithmic representation, ln x2˛ → ln(p1 − p) of the phase diagram, Fig. 7. Isothermal phase diagram of a van der Waals fluid exhibiting normal positive
Fig. 5. —: bubble point curve, - - -: dew point curve. azeotropy at x1az = 0.75: system {A + C} (see Table 3 for the parameters) at 180 K.
U.K. Deiters / Fluid Phase Equilibria 336 (2012) 22–27 27
chemical potential
–4
Subscripts
–6 c critical point
i belonging to component i
m molar property
–8 ∞ infinite dilution
Superscripts
–10
az azeotropic
sl solid–liquid equilibrium
–12 ˛ phase indicator
–6 –5 –4 –3 –2 –1 0 saturation property
ln |Δx1| , phase indicators
A Helmholtz energy
H enthalpy