4.1.1 The Van Der Waals Equation of State
The van der Waals1theory assumes a molecular structure of matter, where matter means gases or liquids. The interaction between molecules requires modification of the ideal gas law:
P
P+a(n/V)2
V
V−nb
=n RT,
i.e.
P= n RT V −nb−a
n V
2
. (4.1)
Molecules, or atoms in the case of noble gases, at close proximity tend to repel.
The attending volume reduction is−nb, wherebis a parameter accounting for the exclusion of (molar) volume that a particle imposes on the other particles inV. At large distances particles attract, which in turn reduces the pressure. The particular form of this pressure correction, i.e. a(n/V)2, may be motivated as follows. The number of particle pairs in a system consisting ofNparticles isN(N−1)/2≈N2/2.
Expressed in moles this leads to the factorn2. The attraction is limited to “not too large” particle-to-particle separation. We assume that two particles feel attracted if they are in the same volume element V. The probability that two particular particles are found withinV simultaneously(V/V)2. Assuming this to be true for all possible pairs leads to an overall number of attracted molecules proportional to(n/V)2. The resulting Eq. (4.1) is the van der Waals equation of state for gases and
1Johannes Diderik van der Waals, Nobel prize in physics for his work on the phase behavior of gases and liquids, 1910.
R. Hentschke,Thermodynamics, Undergraduate Lecture Notes in Physics, 125 DOI: 10.1007/978-3-642-36711-3_4, © Springer-Verlag Berlin Heidelberg 2014
liquids. The (positive) parametersaandbare characteristic for the specific material.
The van der Waals equation of state is by no means accurate, but its combination of simplicity and utility is outstanding.
The parametersaandbmay be estimated by measuring the pressure as function of temperature at low densities. The result may then be approximated using the following low density expansion of Eq. (4.1):
P =RT ∞ i=1
Bi(T;a,b)n V
i
, (4.2)
where
B1(T;a,b)=1 (4.3)
B2(T;a,b)=b− a
RT (4.4)
B3(T;a,b)=b2 (4.5)
...
are so called virial coefficients. In practice one determinesB2(ex p)(T)by fitting low order polynomials to experimental pressure isotherms at low densities. The resulting B2(ex p)(T)is then plotted versus temperature. Nowaandbmay be obtained by fitting Eq. (4.4) to these data points.
If we introduce the following reduced quantities, p,t, andv, via
P=Pcp (4.6)
T =Tct (4.7)
V =Vcv, (4.8)
where
Pc = 1 27
a
b2 (4.9)
RTc= 8 27
a
b (4.10)
Vc=3nb, (4.11)
we may rewrite the van der Waals Eq. (4.1) into p = 8t
3v−1− 3
v2. (4.12)
This is the so called universal van der Waals equation. It is universal in the sense that it does no longer depend on the material parametersa andb. Notice that the ideal
gas law in these units is
pi d.gas= 8t
3v. (4.13)
4.1.2 Gas-liquid Phase Transition
The upper portion of Fig.4.1shows plots of the universal van der Waals equation for three different values oft(solid lines). Of course Eq. (4.12) always deviates from the ideal gas law at lowv. In fact we did not plot the pressure forv-values below the singularity atv=1/3, because there the molecules overlap. But we also notice that the universal van der Waals equation exhibits strange behavior ift <1.0. There is av-range in which the pressure rises even though the volume increases. Here we find an isothermal compressibilityκT < 0—in clear violation of the mechanical stability condition in Eq. (3.16)! Had we plotted Eq. (4.12) for even smallert-values, we would have obtained negative pressures in addition. All in all, for certainvandt, the van der Waals equation does describe states which cannot be equilibrium states.
It turns out however that we can fix this problem, and at the same time we may describe a new phenomenon—the phase transformation between gas and liquid.
To understand how the model may be fixed we look at the free energy obtained via integration of the pressure
F(V)=Fo− V
Vo
d V P, (4.14)
cf. (2.109). The bottom part of Fig.4.1shows the result obtained using the universal van der Waals equation (with Fo = 3) for the same three temperatures as above.
Thet =0.9-curve, which violates mechanical stability according to the attendant pressure isotherm, is sketched in somewhat exaggerated fashion in Fig.4.2.
We notice that the system represented by the filled black circle may lower its free energy by decomposing into regions in which the free energy is flor fg. In between the free energy is
f= flx+ fg(1−x) with x= v−vg
vl−vg . (4.15) Notice that this is the lowest free energy the system can achieve via decomposition into regions with high density, denotedvl, and regions with low density, denoted vg. The respective volume fractions of the two different regions are assigned by the parameter x(note:v =xvl +(1−x)vg) according to the value ofv. In other words, for volumesvl < v < vgthe homogeneous system is unstable relative to the decomposed or inhomogeneous system.
Imagine we move along an isothermt <1 starting from a large volumev > vg. We are in a homogeneous so called gas phase. Upon decreasingvwe are entering the rangevg > v > vl. Here, depending on the value ofv, we observe a “mixture”
0.5 1.0 1.5 2.0 2.5 3.0v
2 1 0 1 2
p
t 0.9 t 1.0 t 1.1
f
t 0.9 t 1.0
t 1.1
vg vl
-
-
= = =
=
=
=
Fig. 4.1 Van der Waals pressure and free energy vs volume at three different temperatures
of regions having a homogeneous density n/vg or n/vl. As we approach vl the volume fraction of the latter regions increases to unity. Ifvl ≥vwe are again inside a homogeneous system—the liquid phase. This augmented van der Waals theory therefore predicts a phase change from gas to liquid and vice versa.
Notice also that forvg ≥v≥vl
p= − ∂f
∂v
T
= −fg− fl
vg−vl =constant
(the straight line in Fig. (4.2) is the common tangent to f atvl andvg) and fg− fl = pvl−pvg or fg+pvg
=nμg
= fl+pvl
=nμl
,
Fig. 4.2 Reduction of the van der Waals free energy via phase separation below t = 1
vl vg
v fl
fg f
f f'
which means that mechanical and chemical stability are satisfied.
In turn we may calculate vl andvg via the conditions p(t, vl) = p(t, vg)and μ(t, vl)=μ(t, vg)based on the universal van der Waals equation itself, i.e.
p(t, vl)= p(t, vg) (4.16) and
− vl
vo
dvp(t, v)+p(t, vl)vl = − vg
vo
dvp(t, v)+p(t, vg)vg (4.17) usingnμ= f+pv. The numerically obtained valuesvl(t)andvg(t)are shown as a dashed line, the binodal line, in the upper part of Fig.4.1. The area beneath the binodal line is the gas-liquid coexistence region. Notice that no solutions exist ift >1, i.e.
no gas-liquid phase transition is encountered abovet=1. In addition to the binodal line there is a dotted line, the spinodal line, which indicates the (mechanical) stability limit. This means that the isothermal compressibility,κT, is negative below this line.
We remark that, among other methods, the simultaneous numerical solution of Eqs. (4.16) and (4.17) may be programmed as “graphical” search for the intersection of pressure and chemical potential in a plot of μ(v)versus p(v). An example is shown in Fig.4.3fort =0.9.
The largest t-value for which a solution is obtained ist = 1. Here one finds vl =vg. The corresponding values of the unreduced pressure, temperature, and vol- ume arePc,Tc, andVcgiven by the Eqs. (4.6–4.8). We can find this so called critical point directly by simultaneous solution ofPc = P(Tc,Vc),d P/d V|Tc,Vc =0, and d2P/d V2|Tc,Vc =0 (the second and third equations are due to the constant pressure in the coexistence region).
We may rewrite Eq. (4.17) using p(t)= p(t, vl)=p(t, vg)as
4.20 4.15 4.10 4.05 4.00 3.95 v 0.60
0.62 0.64 0.66 0.68 0.70 0.72
p (v)
- - - - - -
( )
Fig. 4.3 Pressure vs chemical potential below t = 1 vg
vl
dvp(t, v)−p(t)(vg−vl)=0. (4.18) The left side of this equation is the sum of the two shaded areas, one positive and one negative, in the upper graph in Fig.4.1. The equation states that the two areas between the van der Waals pressure p(t, v), and the constant pressure,p(t), which replaces it betweenvg andvl, are equal. Therefore vg andvl may also be found graphically via this equal area or Maxwell construction . We remark that betweenvg
andvlthe van der Waals pressure isotherms are said to exhibit a van der Waals loop.
Figure4.4shows the possibly simplest of all phase diagrams in thet-v- and in thet-p-plane. The solid line in the upper diagram is the same as the dashed line, i.e.
the phase coexistence curve, in Fig.4.1. The dotted line is the spinodal. The lower graph shows the phase boundary between gas and liquid in the pressure-temperature plane. Notice that here no coexistence region appears because the pressure is constant throughout this region (at constantt). The crosses are vapor pressure data for water taken from HCP.
The next figure, Fig.4.5, shows three isobars above, at, and below the critical pressure.
Figure4.6shows the van der Waals chemical potential along three isobars close to the critical point (top) and over an expanded temperature range (bottom), i.e. we computeμalong three horizontal lines in the lower panel of Fig.4.4just below, at, and just above the critical pressure. Forp=0.96 we cut across the gas-liquid phase transition. Notice that the dashed lines indicate the continuation of the liquid (lowt) and gas (hight) chemical potentials into their respective metastable region, i.e. the region between spinodal and binodal line. This justifies our sketches of the chemical potential in Fig. (3.20) and in principle also in Fig. (3.21). At p = 1 the chemical potential still exhibits a kink, whereas above its slope changes smoothly.
0.5 1.0 5.0 10.0 50.0 100.0v 0.4
0.5 0.6 0.7 0.8 0.9 1.0 t
liquid gas
critical point
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.0 t 0.2 0.4 0.6 0.8 1.0 p
liquid
gas critical
point
Fig. 4.4 Phase diagrams in the t-v- and in the t-p-plane
0.6 0.8 1.0 1.2 1.4 t
0.5 1.0 1.5 2.0 2.5 3.0
1/v
p 1.5 p 1.0
p 0.5
=
=
=
Fig. 4.5 Isobars in the vicinity of the critical pressure
0.985 0.990 0.995 1.000 1.005 1.010t 3.90
3.88 3.86 3.84 3.82 3.80
p 0.96
p 1.0 p 1.04
0.90 0.92 0.94 0.96 0.98 1.00
t 3.90
3.88 3.86 3.84 3.82 3.80
p 0.96 p 1.0 p 1.04
Fig. 4.6 Van der Waals chemical potential along three isobars
The general quality of the van der Waals equation is nicely demonstrated in Fig.4.7.2The figure shows coexistence data for seven different substances plotted in units of their critical parameters. The data, in almost all cases, indeed fall onto a universal curve. This behavior is calledlaw of corresponding states. The universal van der Waals equation certainly is not an exact description, but considering its simplicity the agreement with the experimental data is quite remarkable!
We briefly return to the second virial coefficient, B2(T;a,b), in Eq. (4.4).
Figure4.8illustrates the comparison between the van der Waals prediction (solid line), i.e.
n B2
Vc =1 3
1−27
8t
, (4.19)
and experimental data from HTTD (Appendix C). The agreement with the experi- mental data is qualitatively correct. We note however, that the form ofB3in Eq. (4.5)
2The data shown here are taken from the book by Stanley (1971); the original source is Guggenheim (1945).
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.6
0.7 0.8 0.9 1.0
TTc
gas liquid
gas liquid coexistence Ne N2
Xe CO CH4
O2 Kr
ρ ρC
Fig. 4.7 Law of corresponding states demonstrated for various compounds in comparison to the van der Waals prediction
1 2 3 4 5 6 t
3.0 2.5 2.0 1.5 1.0 0.5 nB2Vc
CH4
Ar O2
Fig. 4.8 Reduced second virial coefficient according to the van der Waals equation in comparison to experimental data
is an oversimplification. The third virial coefficient, B3, is not independent of tem- perature as this equation suggests. Notice thatB2(T)=0 defines the so called Boyle temperature,TBol yle. According to the van der Waals theory
TBoyle=27
8 Tc. (4.20)
Another quantity of interest is the compressibility factor at criticality, for which the van der Waals theory predicts that it is a simple universal number:
PcVc
n RTc = 3
8 =0.375. (4.21)
In the cases of argon, methane, and oxygen the experimental values are close to 0.29.
4.1.3 Other Results of the Van Der Waals Theory
The gas-liquid phase behavior described by the van der Waals theory is considered
“simple”. In this sense it is a reference distinguishing “simple” from “complex”. We emphasize that this refers to the qualitative description rather than to the quantitative prediction of fluid properties. Other phenomenological equations of state may be better in this respect, but it is the physical insight which here is important to us.
Because of this we compute a number of other thermophysical quantities in terms of t,p, and/orv.
Isobaric Thermal Expansion Coefficient
Figure4.9shows the temperature dependence of the isobaric thermal expansion coef- ficient,αp, below, at, and above the critical pressure. The dashed line is the ideal gas result. Below the critical pressure a jump occurs when the gas-liquid saturation line (cf. Fig.4.4; right panel) is crossed. At the critical point we observe a diver- gence. Above the critical point a maximum marks the smooth “continuation” of the gas-liquid saturation line, which sometimes is called Widom line.
Isothermal Compressibility
Figure4.10 shows the volume dependence of the isothermal compressibility, κT, below, at, and above the critical temperature. The dashed lines are the continuation ofκT into the metastable region. Notice thatκT diverges when∂p/∂v|T =0. This condition defines the stability limit,κT ≥0, i.e. inside the gap between the dashed lines the van der Waals equation gives negativeκT. Notice thatκT also diverges at the critical temperature. Above the critical temperatureκT exhibits a maximum near
0.6 0.8 1.0 1.2 1.4
t 2
4 6 8 10 p
p 1.5 p 1.0 p 0.5
α
Fig. 4.9 Temperature dependence of the isobaric thermal expansion coefficient for different pres- sures
0 1 2 3 4 v 0
2 4 6 8 10 T
t 1.4 t 1.02
t 1.0
t 0.98
Fig. 4.10 Volume dependence of the isothermal compressibility in the vicinity of the critical tem- perature
the critical volume, which diminishes as the temperature increases. In general the compressibility increases asvincreases—the gas is less dense and easier to compress.
Isochoric Heat Capacity
Having discussedαP andκT the next obvious function to look at is the isochoric heat capacity,CV. It turns out that within the van der Waals theory all we obtain is
CvVd W =CVvd W(T), (4.22) i.e.CVvd W is a function of temperature only.
We show this via
∂
∂T
∂F
∂V
T
V
=−∂∂PT
V
= ∂
∂V
∂F
∂T
V
T
=−∂∂VS
T
.
Therefore
∂2P
∂T2
V = ∂
∂T
∂S
∂V
T
V = ∂
∂V
∂S
∂T
V
T
(2.=146) ∂
∂V CV
T
T = 1 T
∂CV
∂V
T. This proves Eq. (4.22), because according to the van der Waals equation
∂2P/∂T2|V =0.
Remark:The above equation, i.e.
∂CV
∂V
T =T∂2P
∂T2
V, (4.23)
may be integrated to yield
CV −CV,i deal ≈ −n R
2T∂B2(T)
∂T +T2∂2B2(T)
∂T2 n
V, (4.24)
which is a low density approximation toCV. Of course this correction vanishes if the second virial coefficient,B2(T), of the van der Waals theory, cf. (4.4) is used.
Inversion Temperature
On p. 51 we had discussed the Joule-Thomson coefficient. Based on the universal van der Waals equation (4.12) we want to calculate the inversion line in thet-p-plane.
Equation (2.94) is inconvenient, because we have to expressvin terms oft andp.
However, using Eq. (A.2) we may write
∂v
∂t
p= −∂v
∂p
t
∂p
∂t
v= −
∂p
∂t
v
∂p
∂v
t
.
The inversion line now is the solution of 0=t∂p
∂t
v
∂p
∂v
t+v, which we may find analytically:
p=12(√ t−
3/4)(
27/4−√
t). (4.25)
The inversion temperatures at p = 0 therefore are tmi n =3/4 andtmax = 27/4.
Equation (4.25) is shown in Fig.4.11. The curve encloses the area in thet-p-plane, where the Joule-Thomson coefficient,μJ T, is positive (cooling). Outside this area μJ T is negative corresponding to heating. Experimental data for methane, oxygen, and argon from Fig.4.12in Hendricks et al. (1972) are included for comparison.
Qualitatively the van der Waals predictions are correct. But the quantitative quality is quite poor fort-p-conditions far from the critical point (the gas-liquid saturation line (cf. Fig.4.4; right panel) terminating in the critical point (circle) is included), which we use to tie our theory to reality.
It is interesting to work out ∂v/∂t|p based on the virial expansion Eq. (4.2), because this allows a better understanding of the Joule-Thomson effect on the basis of molecular interaction. To leading order we find
∂v
∂t
p= v
t +t∂(b2/t)
∂t , (4.26)
0 2 4 6 8 10 12 p
CH4
Ar O2
JT cooling
JT heating
0 1 2 3 4 5 6 7
0
0
t Fig. 4.11 Inversion temperature according to Eq. (4.25) including experimental data
v gas p
liquid a
c c t=1 a
v gas t
liquid
c c
b
t gas liquid
p
a a b
a a
coex coex
Fig. 4.12 Different paths approaching the gas-liquid critical point
whereb2=n B2(T)/Vc. Consequently the Joule-Thomson coefficient in this approx- imation is
μJ T = Vc
CP
t2∂(b2/t)
∂t
. (4.27)
This equation is general, i.e. we have not yet used the van der Waals equation of state. If we do this, i.e. we insert Eq. (4.4), the result is
Vc−1CPμJ T =1 3
27 4t −1
. (4.28)
We recognize that the equation describes the Joule-Thomson coefficient at P =0 (or smallP) near the upper inversion temperature. In particular we verify thatμJ T >
0 fort<tmax andμJ T <0 fort>tmax.
Van Der Waals Critical Exponents
Close to the gas-liquid critical point one can show that the quantities±δp,±δv, and
±δt, which are small deviations from the critical point in terms of the variablesp, v, andt, are simply related to each other as well as to thermodynamic functions like the isobaric thermal expansion coefficient,αp, and the isothermal compressibility, κt(as well as all others!).3
Again we focus on the universal van der Waals equation, i.e. Eq. (4.12). Along the critical isotherm, path (a) in Fig.4.12, we setp=1+δpandv=1−δv. Inserting this into Eq. (4.12) we obtain
δp= −3
2δv3+O(δv4).
Ignoring constant factors and additional terms, like the corrections to the leading behavior, we write instead
δp∼ ±δv3 where ± :v=1∓δv. (4.29) Approaching the critical point from above,t = 1+δt, along the critical isochor, v=1, i.e. path (b) in Fig.4.12, we find
δp∼δt. (4.30)
Another special line is the coexistence curve, shown as the dashed line in Fig.4.1.
On the sketches in Fig.4.12the path along the coexistence curve is labeled (c). We insertt = 1−δt andv = 1±δvinto the universal van der Waals equation and expand the result in powers ofδv. Finally we assumeδv≡δv±=c±δtβ.4Note that δt andδvare positive. The result is
p±=1−4δt∓3
2c3±δt3β±6c±δt1+β+. . . .
Here. . .stands for higher order termsO(δt4β)andO(δt2+β), and+and−stand for the gas and the liquid side of the coexistence curve respectively. The stability condition Eq. (4.16) requires p−= p+. Settingc+ = −c−fulfills this equality but does not yield the desired result becausev−=v+. The only other solution requires 3β=1+βand∓32c3±±6c±=0. Consequently we obtain
β = 1
2 and c±=2, (4.31)
3The small quantitiesδp,δv, andδtare all positive.
4This is the leading term in a power series expansion ofδvinδt. Notice that the coexistence curve is not symmetric with respect to reflection across the critical isochore—except very close to the critical point (cf. below).
i.e. the leading relation betweenδvandδtalong the coexistence curve is
δv∼δt1/2. (4.32)
We may use this to work out the dependence of the isothermal compressibility, κt = −(1/v)∂v/∂p|t, near the critical point and along the coexistence curve. Again we insertt =1−δtandv =1±δvinto the universal van der Waals equation and expand the result in powers ofδv. We then work out the derivative∂p/∂v|tand insert the above result Eq. (4.32). This yields
κt−1∼δt. (4.33)
Using the general thermodynamic relation ∂v∂t
p= −∂∂pt
v/∂∂vp
t, cf. (A.3) we obtain theδt-dependence of the isobaric expansion coefficient,αp =(1/v)∂v/∂t|p, near the critical temperature and also along the coexistence curve. Because∂p/∂t
vto leading order contributes a constant only, we obtain, as above forκt,
αp∼δt−1. (4.34)
We return briefly to the critical isochore,v =1, and compute theδt-dependence of κt−1, when we approach the critical point via this path. Working out∂p/∂v|t and settingv=1 as well ast=1+δtyields
κt−1∼δt (4.35)
as before. Only the prefactors (called scaling amplitudes) are different, i.e. κt−1
≈12δton the path along the coexistence curve andκt−1 ≈6δt along the isochore.
Again we find the relation Eq. (4.34) for the same reason as before, i.e.∂p/∂t
vto leading order contributes a constant only.
The reader may want to work out the divergences of αp in Fig.4.9andκt in Fig.4.10to leading order inδtandδv, respectively. The result isαp∼δt−2/3along the critical isobar andκt ∼δv−2along the critical isotherm.
The exponents in the above power laws are called critical exponents. Table4.1 compiles a selected number of them together with their definition, thermodynamic conditions, and van der Waals values. Hereρl−ρg is the density difference across the coexistence curve.5This quantity is called order parameter. Notice that by con- struction the order parameter vanishes aboveTc. In addition,δT = |T −Tc|and δP = |P− Pc|. Notice also that we have not yet talked about the heat capacity exponentα.6The prime indicates the same critical exponent belowTc. The van der Waals theory yields the same values for the two exponents listed here, i.e.α =α
5What is the relation betweenρl−ρgand±δv? Settingρl =ρc+δρlandρg=ρc+δρgand using|δρ/ρ| = |δv/v|yieldsρl−ρg∝δv−+δv+.
6The present critical exponent notation is standard. We want to adhere to it, even though certain letters are used for other quantities also.
Table 4.1 Selected critical exponents and their vdW-values
Exponent Definition Conditions vdW-value
α CV ∼δTα V =Vc(T<Tc) 0
α CV ∼δTα V =Vc(T>Tc) 0
β ρl−ρg∼δTβ coex curve 12
γ κT−1∼δTγ coex curve (T <Tc) 1
γ κT−1∼δTγ V =Vc(T>Tc) 1
δ δP∼ ±(ρl−ρg)δ T=Tc+:ρ > ρc
-:ρ < ρc
3
andγ =γ. But the van der Waals values are not correct! Even though the correct exponent values turn out to be nevertheless the same below and aboveTc, we again adhere to the (safe) standard notation, which distinguishes the two conditions.αP
does not appear in this list, because, as we have seen, its exponents are alsoγ and γat the indicated conditions.