P R E P R I N T ICPWS XV

Berlin, September 811, 2008

Equations for the Thermodynamic Properties at the Saturation Line in the

Supercooled Water Region
Jana Kalovaa and Radim Maresb
a

Institute of Technology and Business in Ceske Budejovice, Czech Republic

b
University of West Bohemia in Pilsen, Czech Republic
Email: kalova@volny.cz

Vapour-pressure equations are often used in atmospheric applications. Also properties of water
at temperatures below the triple point are needed for calculating saturation pressure in clouds,
because clouds are often composed of deeply supercooled water. There are many equations used
in meteorology that describe the saturation pressure of H2O also below the triple point. Most of
them are based on the Clapeyron equation, because experimental data are not of high quality.
Measurements are restricted to temperatures above approximately 235 K, which is the limit of
homogenous nucleation. We have compared calculated values with the Wagner and Pru
equation, which also appears to be a very good approximation in the supercooled water range.
We have used this equation as a standard equation for vapour pressure and added only
a correction term for temperatures from 130 K to 255 K.

Introduction
The vapour pressure of supercooled water
plays an important role in cirrus clouds, polar
stratospheric clouds, and in a large volume of the
atmosphere, where water is cooler than 0 C.
Meteorology is interested in the properties of water
at temperatures above 160 K, but there are only
experimental data in the region above 233 K, and
existing relations are valid mainly in the region
above 0 C. Extrapolation of existing vapourpressure equations is mainly used for temperatures
below 0 C, but there are large differences between
them. A variety of vapour-pressure equations are
listed in [1].
Murphy and Koop [2] carried out an evaluation
of many vapour-pressure equations. They used
the knowledge of cp above 233 K and the
asymptotic behaviour of cp at the temperature limit
of amorphous ice (130 K - 150 K) to develop an
equation of the vapour pressure for supercooled
liquid water. They made detailed comparisons of
the new equation with other vapour-pressure
equations and with existing experimental data. The
crucial area is the range of temperatures between
150 K and 232 K, referred to as a no-mans land,
where experimental data are missing. There are
three theories to explain the properties of water in
this temperature range: the theory of spinodal curve,
the theory of a second critical point and a
singularity-free hypothesis. Recent molecular
simulations and some experiments support the
theory of a second critical point, the critical point of
liquid liquid phase transition. Water exhibits

a well-known (first) critical point at 647 K, the

predicted second critical point in supercooled water
describes the pressure and temperature, where two
distinct liquid phases the low-density liquid and
high-density liquid become identical. If there is a
second critical point, it is possible to describe the
thermodynamic properties near the critical point
with a scaled parametric equation, which can
explain the divergent behaviour of many
thermodynamic properties. Fuentevilla and
Anisimov proposed a scaled equation based on only
a few parameters [3]. We have used their equation
to predict the behaviour of cp in the range of 150 K
to 233 K and from a knowledge of cp we have
started our calculation of the vapour pressure in the
range 130 K - 273 K.

Thermodynamic basis of the vapour pressure of

liquid water
Our derivation of the vapour-pressure relation
is based on the procedure from [2] and will be
derived from the Clausius-Clapeyron equation
extrapolated to the ideal-gas limit:
d ln p Lliq (T )
(1)
=
,
dT
RT 2
where Lliq(T) is the latent heat of vaporization as
a function of temperature and R is the molar gas
constant (8.314472 J/mol/K).
The integration of Eq. (1) can start at the
triple-point (pressure pt = 611.657 0.01 Pa and
temperature Tt = 273.16 K [4]).

often called amorphous ice. Amorphous ice

crystallizes upon warming to about 155 K. It is
possible to obtain some thermodynamic data
concerning the properties of amorphous ice and two
states of amorphous ice were discovered so called
high-density amorphous ice and low-density
amorphous ice. These discoveries supported the
theory named liquid-liquid phase transition.
The isobaric heat capacity cp is extrapolated
with a smooth function. It is assumed that cp has
a maximum near 235 K, and then decreases until,
close to 150 K it is near the heat capacity of ice.
Murphy and Koop [2] obtained the equation for
cp,ice (in J/mol/K):

To integrate Eq. (1) it is necessary to know the

temperature dependence of Lliq(T). An equation for
Lliq(T) was published in [2,5]:
T

Liq (T ) = Lliq ,t + c p (T , )dT , +

Tt

(2)

,
, ( v v vl )
dT ,
(vv vl ) T
,
T
p

where Lliq,t is the latent heat of vaporization at the

triple-point temperature Tt, cp is the difference in
molar heat capacities, vv and vl are the molar
volumes of vapour and liquid, resp. The second
integral is very small compared to the first. For
example if vv >> vl and the behaviour of the
vapour is approximated by an ideal-gas
RT
, it is possible to calculate the
equation v =
p
second integral:
T
,
dp
, ( v v vl )
dT , (vv vl ) T T , dT

p
T

dp
+ ,
T dT
t

c p ,ice = 2.0572 + 0.14644 T +

+ 0.06163 T exp((T / 125.1) 2 ) .

In the temperature range between 130 K and 150 K

it is assumed [2], that cp,liquid is equal to cp,ice of
hexagonal ice plus 2 JK-1mol-1. Hence in this
temperature interval (in J/mol/K):
c p ,liquid = 0.0572 + 0.14644 T +

dp

Tt

+ 0.06163 T exp((T / 125.1) 2 ) .

,
,
v
dT , vv T T , dT

dp

= 0.

Tt

The latent heat of vaporization at the triple point is

Lliq,t =45 054.7 J/mol (IAPWS95). Because the
second integral of Eq. (2) is very small, it can be
written (in J/mol):
T

Lliq (T ) = 45054.7 + c p (T , ) dT , ,

(3)

Tt

where cp=cp,liquid cp,vapour.

From [2] one can obtain a relation for cp,vapour
(in J/mol/K):
c p,vapour = 33.2618 + 0.00187 T

0.06165 T exp((T /129.85) 2 ) +

(6)

Hruby et al. [11] have developed equations for

the properties of water at ambient pressure for
temperatures between 253 K and 383 K. The
properties are based on IAPWS-95. The cp,liquid from
this work can be used in this temperature interval.
Recently, Fuentevilla and Anisimov developed
a scaled equation based on the Landau theory of
phase transitions describing the behaviour of
properties in the vicinity of the liquid-liquid critical
point [3]. To obtain values far from the critical
point, it is necessary to use some background
functions. The description of some properties far
from the critical point consists of two parts. The
first part describes the critical behaviour and the
second is a background function which balances the
values far from the critical point. For the critical
part of cp,crit they obtained [3] and [12]:

dT , [vv vv ]dT

(5)

(4)

c p ,crit = R

+ 0.06163 T exp((T /125.1) ) .

2

T 2
(a 2 1 + 2a 2 b1 12 + b12 2 ) ,
Tc

(7)

where a and b are known constants and where 1 ,

12 , 2

are functions of p, T gained from the

solution of a cubic equation. Because the cubic
equation cannot be solved analytically, it is
necessary to calculate cp,crit for each p and T
individually. These values represent only a singular
part of cp,liquid, the background functions are
necessary for a complete description of cp,liquid.
Above we have described the approximation of
cp,liquid for temperatures 130 K - 150 K, the relation
for critical part of cp,liquid, and the equation for
temperatures between 253 K and 383 K. We have
also used the following steps to get cp,liquid in the
entire temperature interval between 130 K and
383 K, including the no-mans land region:

Relation for cp,liquid (T)

Murphy and Koop [2] used cp,liquid(T) at

atmospheric pressure to calculate the vapour
pressure. Experimental data are used at
temperatures above 233 K [6 - 8]. It is assumed that
at temperatures near 155 K the heat capacity of
liquid is close to the heat capacity of ice [9]. The
range between 155 K and 233 K is not open to
experiments on supercooled water [10], because
liquid water rapidly crystallizes. The crystallization
of ice can be avoided only by hyperquenching
liquid water at extremely high cooling rates
exceeding 105 K/s to temperatures below 100 K.
The viscosity of water increases rapidly and this
procedure results in an amorphous form of water,
2

A) Generation of data for cp,liquid in interval from

130 K to 150 K using Eq. (6).
B) Generation of data for cp,liquid from an equation
of Hruby et al. [11] in the interval 253 K - 383 K.
C) For both temperature intervals (130 K - 150 K
and 253 K - 383 K) the critical part of cp,liquid is
calculated from Eq. (7).
D) The critical part cp,crit is subtracted from the
values gained in steps A and B. As a result the
cp,backround values are gained from these temperature
intervals. It is possible to fit the values with some
analytical function the least-squares method was
used.
E) The cp,backround is fitted with some analytical
functions and we suppose that the background is
valid in all temperature intervals between 130 K
and 383 K. The critical part cp,crit is possible to
calculate (see (7)) and then the
cp,liquid = cp,crit + cp,backround.
In units of J/mol/K the background of the molar
heat capacity can be fitted to data from step D) by:

Latent heat as function of temperature J/mol

48500
48000
47500
47000
46500
46000
45500
45000
44500
120

1/ 2

240

260

280

Now we can use the Clausius-Clapeyron

equation (1) and when the equation is integrated
(numerically), we obtain values for the vapour
pressure ln p(T) displayed in Fig. 3:
Vapour pressure - ln(p)
10
5
0

100

-15
-20
130

(8)

150

170

190

210

230

250

270

T[K]

Figure 3: Vapour pressure

There is a very accurate equation for the

vapour pressure, namely, the Wagner and Pru
equation [13]:
p

ln
=
7
2
.
2064
10

C P J/mol/K
110

T
= c
T

90
70
50
30

7.85951783 + 1.84408259 1.5

11.7866497 3 + 22.6807411 3.5

4
7 .5

15.9618719 + 1.80122502

where = 1

10
150

220

Figure 2: Latent heat as a function of temperature

When the data for c p ,crit are added (see step E),
we obtain for cp,liquid values displayed in Fig. 1:

-10130

200

-10

T
+ 5274.9173

100

T
409.64849 426.00397 e
10
+ 64.65854 e

180

-5

400 T

100

160

T[K]

c p ,background = 5455.2493
T
10133.798

100

140

170

190

210

230

250

270

(9)

T
, Tc = 647.096 .
Tc

This equation is valid in the range

273.16 K T 647 K. The equation can be
extrapolated below the triple point. In Fig. 4 the
difference between calculated vapour-pressure data
from Eq. (1) and Eq. (9) is displayed.
The Wagner-Pru equation also describes the
calculated vapour-pressure data very well in the
region below the triple point, to temperatures near
-30 C. We have also added to the Wagner-Pru
equation correction terms, which are significant
only in the region 130 K - 250 K.

T[K]

Figure 1: cp in the interval 130 K 270 K

This behaviour is numerically consistent with

IAPWS-95 in the region above 230 K, but it also
takes temperatures around 150 K into account.
cp,liquid is now calculated not only in the No-mans
land region, but in the entire range from
130 K - 383 K.

Calculation of vapour pressure

The values of cp,liquid and cp,vapour (4) can be

substituted in Eq. (3) for a calculation of Lliq . We
have performed a numerical integration and we
obtained values displayed in Fig. 2:

The resulting equation is:

1.5
3
T 7.85951783 + 1.84408259 11.7866497 +
ln( p ) = ln(22064000) + c

T + 22.6807411 3.5 15.9618719 4 + 1.80122502 7.5

0.004496417 e

where = 1

T 182
0.18

25

(10)

+ 8.6073267 1 4 17.40625115 + 9.0495481 1 6 ,

T 120
T
, Tc = 647.096 , 1 =
.
Tc
10

Murphy a Koop [2] have derived:

Acknowledgements

6763.22
4.210 ln(T )
T
(11)
+ 0.000367 T + tanh{0.0415 (T 218.8)}
1331.22
(53.878
9.44523 ln(T ) + 0.014025 T ),
T
ln ( p ) = 54.842763

This work was supported by the Ministry of

Education, Youth and Sports of the Czech Republic,
project No. 1P04LA241 and by the Grant Agency
of the Czech Republic, project No. 101/05/2214.

which is valid in the temperature interval

123 K < T < 332 K.
A comparison of values calculated from Eq. (10)
with Eqs. (9) and (11) is shown below:

Literature

K-W

Difference of ln(p ) - Wager-Pru (K-W) and

Murphy-Koop (K-K)

K-K

[1] http://cires.colorado.edu/~voemel/vp.html
[2] D. M. Murphy, T. Koop: Review of the vapor
pressures of ice and supercooled water for
atmospheric applications, Q. J. R. Meteorol.
Soc. 131: 1539-1565 (2005)
[3] D. A. Fuentevilla, M. A. Anisimov: Scaled
Equation of State for Supercooled Water Near
the Liquid-Liquid Critical Point, Phys. Rev.
Lett. 97: 195702 (2006)
[4] L. A. Guildner, D. P. Johnson, F. E. Jones:
Vapor pressure of water at its triple point,
J. Res. Natl. Bur. Stand. 80A: 505-521 (1976)
[5] H. R. Pruppacher, J. D. Klett: Microphysics
of clouds and precipitation, 2nd ed., Kluwer,
Dordrecht, The Netherlands (1997)
[6] C. A. Angell, M. Oguni, W. J. Sichina: Heat
capacity of water at extremes of supercooling
and superheating, J. Phys. Chem. 86: 998-1002
(1982)
[7] E. Tombari, C. Ferrari, G. Salvetti: Heat
capacity anomaly in a large sample of
supercooled water, Chem. Phys. Lett. 300:
749-751 (1999)
[8] D. G. Archer, R. W. Carter: Thermodynamic
properties of the NaCl + H2O system. 4: Heat
capacities of H2O and NaCl(aq) in cold-stable
and supercooled states. J. Phys. Chem. 104:
8563-8584 (2000)
[9] F. W. Starr, C. A. Angell, H. E. Stanley:
Prediction of entropy and dynamic properties
of water below the homogeneous nucleation
temperature, Physica A 323: 51-66 (2003)
[10] O. Mishima, H. E. Stanley: The relation
between liquid, supercooled and glassy water,
Nature, 396: 329-335 (1998)
[11] J. Hrub, J. Ptek, J. Klomfar, M. Soukov,
A. H. Harvey: J. Phys. Chem. Ref. Data, in
preparation

0,4
0,3
0,2
0,1
0
130
-0,1

150

170

190

210

230

250

270

Figure 4: Differences of vapour-pressure

equations

In this picture K-W shows the difference between

the calculated values ln( p) with Eq. (10) and
values from Wagner-Pru equation (9), K-K shows
the difference between calculated values and the
Murphy-Koop equation (11).
It is necessary to mention again that the
Wagner and Pru equation is valid at temperatures
273 K - 647 K, and Murphy and Koop in the
interval from 123 K - 332 K.

Conclusion

In this article we have shown that it is possible

to use the Wagner-Pru equation down to 30 C.
The difference in vapour pressure was observed for
temperatures below than 30 C, which is caused
by the anomalous behaviour of cp, described in the
scaled equation of Fuentevilla and Anisimov [12].
The behaviour of the vapour pressure around 150 K
is influenced by the accuracy of cp when the region
of amorphous ice is approached. Correction terms
to the Wagner-Pru equation are proposed to
correct differences in the region below 30 C.

[12] D. A. Fuentevilla: Scaled Parametric Equation

of State for the Liquid-Liquid Critical Point in
Supercooled Water, Thesis submitted to the
Faculty of the Graduate School of the
University of Maryland (2007)
[13] W. Wagner and A.Pru: International
equations for the saturation properties of
ordinary water substance: Revised according
to the international temperature scale (1990).
Addendum to J. Phys. Chem. Ref. Data 16:
893 (1987), J. Phys. Chem. Ref. Data 22:
783-787
[14] W.Wagner and A.Pru: The IAPWS
Formulation 1995 for the thermodynamic
properties of ordinary water substance for
general and scientific use, J. Phys. Chem. Ref.
Data 31: 387-535 (2002)