Materials Transactions, Vol. 46, No. 6 (2005) pp.

1348 to 1353
#2005 The Japan Institute of Metals

Vapor Pressure Measurements for Metal Chloride Systems

by the Knudsen Eusion Method
Yanling Zhang, Etsuro Shibata, Eiki Kasai and Takashi Nakamura
Institute of Multidisciplinary Research for Advanced Materials (IMRAM), Tohoku University, Sendai 980-8577, Japan
Thermodynamic knowledge of metal chlorides, such as vapor pressure and activity in salt/slag, are of remarkable industrial interest,
particularly in elds related to the recovery/recycle/reuse of metal materials. In this study, a new apparatus for measuring the vapor pressure of
metal chlorides from molten salt/slag was designed and tested with reference compounds. The results are in a good agreement with the available
literature data. Using the developed apparatus, vapor pressure measurements for a KClNaClCaCl2 system have been performed. Using the
measured data the activities of components were obtained. Simultaneously, the activities of components were calculated using a thermodynamic
code, Factsage 5.2 wherein the Modied Quasi-chemical Model was employed to obtain the necessary thermodynamic parameters. The both
data agree reasonably well with each other. The results suggested that the KClNaCl system exhibits similar behavior with ideal solution; while
in the KClCaCl2 and NaClCaCl2 systems there are possible interactions between CaCl2 and KCl/NaCl. For the ternary system KClNaCl
CaCl2 , in some concentration range (mole fraction of NaCl 0.250.50) NaCl behaves as in an ideal solution, suggesting a larger anity between
KCl and CaCl2 than that between NaCl and CaCl2 .
(Received March 1, 2005; Accepted April 21, 2005; Published June 15, 2005)
Keywords: vapor pressure measurement, metal chlorides, Knudsen eusion method, Modied Quasi-chemical Model

1.

Introduction

Molecular effusion
d

The thermodynamic behaviors of metal chlorides are of

considerable industrial interest, particularly in elds related
to the recovery/recycle/reuse of metal materials. For
example, some basic knowledge, such as vapor pressures
and activities of metal chlorides in salt/slag, are very useful
and urgently needed during the recovery/recycling of metals
from wastes by high temperature processes; and the recovery
process of metals from their ores by chlorination. However,
such basic information, particularly experimental data, is
rather limited, primarily because of operational diculties.
There are several methods of determining the vapor
pressure, e.g., static, boiling point, and transpiration. The
Knudsen method is a direct way and is particularly
appropriate in the case of pressures lower than 100 Pa.1,2)
Although it has still not been used for metal chlorides, it has
long been employed for the vapor pressure measurements of
compounds.37)
In the present study, an apparatus was designed for the
measurement of the vapor pressure of metal chlorides from
salt/slag systems by the Knudsen eusion method; it was
then tested with reference compounds. The vapor pressure of
a KClNaClCaCl2 system was measured, thus obtaining the
activities of its components. Simultaneously, the activities of
components were calculated using FactSage 5.2,8) wherein
the Modied Quasi-chemical Model912) was employed to
obtain the necessary thermodynamic parameters. The calculated activities of components agree reasonably well with the
experimental data.
2.

Principle of the Knudsen Eusion Method

The theoretical basis of the Knudsen eusion method is the

kinetic theory of gas. In the experiment, the sample is
contained in an inert cell provided with a small, thin orice
on the center of the lib, as shown in Fig. 1. The cell is placed
in a high vacuum system at a constant temperature. If the

L
H
D

Fig. 1

Scheme of Knudsen eusion cell.

orice is suciently small, an equilibrium condition is

maintained inside the cell. Hence, the amount of vapor
eusing through the orice in a certain period of will be
determined by the vapor pressure, temperature, molecular
weight of vapor species, and the dimensions of the orice.
Equation (1) describes this relationship, which was derived
by Knudsen:13)
r
1

W
2
RT


1
PK


Kc A0
t
M
where PK (Pa) is the vapor pressure near the orice, A0 (m2 )
is the area of the orice, M (kg/mol) is the molecular mass of
the eusing vapor, t (s) is the experimental time,
W (kg) is
the mass loss of the sample, T (K) is the temperature, R
(8.314 J/mol
K) is the gas constant, and Kc is a coecient
related with the geometrical condition of the orice,
determined by its thickness L (m) and diameter d (m).2)
Equation (1) holds when there are no collisions between
the molecules either in the cell or near the orice and when
the escaped molecules do not disturb the equilibrium between
the vapor and the condensed phases. These conditions are
established when the mean free path of the molecular ,
which is determined by the vapor pressure and temperature,
is considerably larger than the diameter of the orice d; and

Vapor Pressure Measurements for Metal Chloride Systems by the Knudsen Eusion Method

when the surface area of the condensed phase As is

suciently larger than the orice area A0 . It was reported1,2)
that eq. (1) is accurate when =d > 10 and A0 =As < 100.
The saturated vapor pressure of the sample Peq would be
higher than PK since the Knudsen cell is not actually an
equilibrium system. While both Whitman14) and Motzfeldt15)
arrived at the conclusion that in a Knudsen cell of typical
dimensions (cylindrical shape wherein height equals diameter), if A0 =As < 100, the saturated vapor pressure of the
sample is very close to that near the orice: Peq  PK .
Moreover some researches3) indicated that in such geometric
shapes of the Knudsen cell the experimental error is great
enough to mask the dierence between Peq and PK . Therefore, this study used eq. (1) to evaluate the vapor pressure of
the sample.
3.

Experimental Condition and Theoretical Basis of

Calculation

turbo molecular pump and a rotary pump, and the temperature of the sample was controlled by an electric furnace.
Weight loss of the sample was continuously recorded by the
data system.
3.3 Analysis method
The vapor pressure of pure chlorides can be obtained
directly using eq. (1). In the case of KClNaCl system, it was
assumed that the mixed condensed and vapor phases exhibit
the same behavior with ideal solution and ideal gas,
respectively, since they have similar molecule structures
and vapor pressure values. Therefore, the composition and
molecular weight of the vapor could be estimated by eqs. (2)
and (3):
v
c
0
NKCl
NKCl
PKCl

v
0
c
NNaCl
NNaCl
PNaCl

Mv
3.1 Samples
KCl (Wako Pure Chem. Ltd., mass fraction 99.5%), NaCl
(Wako Pure Chem. Ltd., mass fraction 99.5%) and CaCl2
(Wako Pure Chem. Ltd., mass fraction 99.5%) were used as
reference compounds to test the apparatus. Dierent mixtures
of these were used to measure the vapor pressures of a KCl
NaClCaCl2 system. All samples were in powdered state.
3.2 Apparatus and procedure
The schematic diagram of the apparatus is shown in Fig. 2.
A cylindrical cell made of pure platinum with equal height
and diameter (D H 10 mm) was used. The cell consists
of two parts, i.e., upper and lower pans. The upper pan has an
orice and its diameter d is 0.07, 0.13, 0.20, or 0.30 mm. The
sample powder was charged and compressed in a thin layer
on the bottom of the lower pan. Then, the lips of the two pans
were press-sealed and the cell was hung with the hook of the
balance by a platinum wire, which was set at the center of the
reaction tube. The balance gives an accuracy of 0.01 mg. A
high vacuum condition ( 102 Pa) was maintained by a
1

11

4
5

10

9
3

Fig. 2

5.Thermal couple
6. Electric furnace
7. Turbo molecular pump
8. Rotary pump

9. Temperature controller
10. Data system
11. Reaction tube

Apparatus schematic for Knudsen eusion method.

c
0
c
0
MKCl
NKCl

PKCl
MNaCl
NNaCl

PNaCl
3
0
0
c
c
NKCl

PKCl
NNaCl

PNaCl

c
c
v
v
where NKCl
, NNaCl
, NKCl
, NNaCl
are the mole fractions of KCl
and NaCl in condensed and vapor phases, respectively, and
0
0
PKCl
, PNaCl
are the vapor pressures of pure KCl and NaCl,
respectively. Mv is the molecular mass of the vapor; MKCl ,
MNaCl are the molecular mass of KCl and NaCl, respectively.
After the experiment, the total vapor pressure of the sample
was obtained by using its weight loss at a controlled
temperature and combining eqs. (1) and (3).
For KClCaCl2 , NaClCaCl2 , and KClNaClCaCl2
systems, after the experiment, the compositions of K and
Na were analyzed by ICP (inductively coupled plasma),
based on which the weight loss of KCl and NaCl were
calculated. Then, their partial vapor pressures and activities
in the system could be obtained by using eq. (1). However in
the case of CaCl2 , the weight loss of CaCl2 could not be
detected because of its extremely low vapor pressure (0.05
0.18 Pa) at the experimental temperature (10731103 K) and
the ICP analysis error. Therefore, it is very dicult to obtain
the partial vapor pressure and activity of CaCl2 in the present
study.

3.4 Thermodynamic model for solutions

In addition to experimental measurements, the activities of
components in the chloride systems of KClCaCl2 , NaCl
CaCl2 , and KClNaClCaCl2 were calculated using FactSage 5.2,8) wherein the Modied Quasi-chemical Model was
employed to obtain the necessary thermodynamic parameters. The theory of the Modied Quasi-chemical Model for
solution was proposed by Pelton et al. and described in detail
in the literatures.912) In this paper, a brief introduction of the
model is presented as follows. The parameters of the model
are the Gibbs energy change
gAB/Cl of the pair change for
the following pair-exchange reactions:
(A-Cl-A)pair (B-Cl-B)pair 2(A-Cl-B)pair

1. Balance
2. High vacuum room
3. Water cooling
4. Knudsen cell

1349

As
gAB/Cl becomes more negative, reaction (4) is shifted

to the right, (A-Cl-B) pairs dominate, and the solution
becomes progressively more ordered. The
gAB/Cl approximately corresponds to the excess free energy of the solution.
For example, XKCa is the mole fraction of the second-

1350

Y. Zhang, E. Shibata, E. Kasai and T. Nakamura

Table 1

Cation-cation coordination numbers of the solution.

Ziij

Zijj

Na

Na

Ca

Ca

further measurements of the KClNaClCaCl2 system was

selected in a similar manner.
The comparisons between the experimental and reference
data for the vapor pressures of KCl, NaCl, and CaCl2 are
shown in Figs. 3, 4, and 5, respectively. There are reasonable
agreements between the data for KCl and NaCl (shown in
Figs. 3 and 4). Although most of the experimental data fall
within the range of reference values, some deviation is
observed in the case of CaCl2 . This might be due to the

nearest-neighbor (KClCa) pairs, and the model also

requires denitions of the cation-cation coordination numbers Ziij for all binary subsystems. The coordination numbers
chosen in this study based on the literature16) are listed in
K
Ca
Table 1. For example, the choice of ZKCa
1=2ZKCa
ensures
that the composition of the maximum short-range ordering
will approximate the K2 CaCl4 composition, wherein the
molar enthalpy and entropy of mixing for the system assume
the lowest values. The mole fractions of KCl and CaCl2 in the
binary solution can be derived using the coordination
numbers XKCa , XKK , and XCaCa .912) As described in detail
in the literatures,912) it is necessary to geometrically dene
all ternary systems as either symmetric or asymmetric.
In the present case, the systems with one alkali-earth and two
alkali chlorides are asymmetric, with the alkali-earth chloride
as the asymmetric component.
Through optimization with available experimental data,
the parameters
gAB/Cl of reaction (4) are expanded for each
pair (KClNa, KClCa and NaClCa) as empirical
polynomials in molar fractions Xi j .16) The equations of

gAB/Cl for each pair are shown in Table 2. Ternary excess
parameters for the solution were not required for the ternary
system of KClNaClCaCl2 . Table 3 lists all thermodynamic
data (H  298.15 K , S 298.15 K , and C p ) for the pure liquids of the
KClNaClCaCl2 system.17,18)
4.

7
6
5
ln (p/Pa)

4
3
2
1

by Kubaschewski 19)

this study

-1
-2
0.8

1.1

1.2

1.2

1.3

1000 (K/T)

Fig. 3

Vapor pressures of KCl.

5
4

ln (p/Pa)

Results and Discussion

2
1
0
-1

by I.Barin20)

-2

4.1 Reliability of the apparatus

The vapor pressures of the reference compounds KCl,
NaCl, and CaCl2 were measured at 8731023 K, 9131023 K
and 12731353 K, respectively. This is because under such
temperature conditions, their vapor pressure values fall in the
range of 0.150 Pa, in which the Knuden eusion method is
more accurate. The experimental temperature range for the
Table 2

0.9

this study

-3
-4
0.9

1.1
1000 (K/T)

Fig. 4

Vapor pressures of NaCl.

Gibbs energy change of the pair change for the pair-exchange reactions.

gAB/Cl (J/mol)

Binary System (A-Cl-B)

gNaK/Cl 695:5  67:0XNaNa =XKK XNaK XKK

NaClKCl
NaClCaCl2

gNaCa/Cl 4710:2 0:7805T 1347:8 0:5726TXNaNa

1580:0 0:1473TXCaCa

KClCaCl2

gKCa/Cl 12281:0 3:9706T  852:56XKK  5930:6XCaCa

Table 3 Thermodynamic properties of pure liquids.16;17

Temperature range
(T/K)

H  298.15 K
(J/mol)

S 298.15 K
(J/mol K)

Cp
(J/mol K)

NaCl17

298.15 to 1500
1500 to 2000

394956:0
390090:1

76.0761
84.5055

77:7638  0:0075312T
66.9440

KCl17

298.15 to 2500

421824:9

86.5225

73.5966

660 to 2500

606887:4

117.2971

92.0480

CaCl2 18

Vapor Pressure Measurements for Metal Chloride Systems by the Knudsen Eusion Method

performed from 963 to 1003 K. Figure 6 shows the measured

results, wherein the line is the relation between the vapor
pressures of pure KCl and NaCl. As evident, at lower
temperatures the points representing the total vapor pressure
of the KClNaCl system mostly lie on the line, which
suggests that the total vapor pressure could be estimated by
eq. (5):

4
by Kubaschewski 19)

ln (p/Pa)

this study

0
c
0
c

NKCl
PNaCl

NNaCl
Pt PKCl

0
0.7

0.75

0.8

0.85

1000 (K/T)

Fig. 5

Vapor pressures of CaCl2 .

30
1003K
Vapor pressures of the KCl-NaCl
system (p/Pa)

1351

25
993K

20

983K

15

973K
10

963K

5
0
0

0.2

0.4

0.6

0.8

1.0

Mole fraction of KCl in the KCl-NaCl system

Fig. 6

Vapor pressures of the KClNaCl system.

handling diculty caused by the strong deliquescent property of CaCl2 , despite its preparation in a glove box lled
with Ar. However, the results reveal that this apparatus is
suitable for measuring the saturated vapor pressures of
substances at high temperatures.
4.2 Measurement for the KClNaClCaCl2 system
4.2.1 KClNaCl
Samples of KCl:NaCl = 0.2:0.8, 0.5:0.5, and 0.8:0.2 (in
molar fractions) were prepared, and the measurements were
Table 4

where Pt (Pa) is the total pressure of the KClNaCl system.

Further, it proves that the condensed and vapor phases of the
KClNaCl system exhibit same behavior with ideal solution
and ideal gas, respectively. The deviation from the line
increases with increasing temperature. This could be explained by the fact that the experiment was conducted from
low to high temperatures, during which the composition of
the sample changed continuously. The concentration of KCl
decreased becauae its vapor pressure was slightly higher than
that of NaCl. This led to a decrease in the total vapor pressure
of the sample.
4.2.2 KClCaCl2
The experimental situation and the measured results of the
partial vapor pressure and activity of KCl in the KClCaCl2
system are shown in Table 4, along with the calculated
activity of KCl obtained using Factsage 5.2, wherein the
necessary thermodynamic parameter was obtained from the
above mentioned Modied Quasi chemical Model. Figure 7
shows the comparison at 1073 K. As evident, the data
measured using this apparatus agree reasonably well with the
calculation result. The activities of KCl showed a negative
deviation from Raoults Law, which suggests a possible
interaction between KCl and CaCl2 .
4.2.3 NaClCaCl2
Table 5 showes the experimental condition and the
measured results of the partial vapor pressure and activity
of NaCl in the NaClCaCl2 system, along with the caculated
activity of NaCl using Factsage 5.2, in a manner similar to the
KClCaCl2 system. Figure 8 shows the comparison at
1073 K. There is a reasonable agreement between the
experimental and calculated results. The activities of NaCl
showed a negative deviation from Raoults Law, which also
suggests a possible interaction between NaCl and CaCl2 .

The experimental and calculated data on partial vapor pressure and activity of KCl in the KClCaCl2 system.
Caculation
data

Experimental results
Sample compositon,
mole fraction

0.25KCl+0.75CaCl2

0.50KCl+0.50CaCl2

0.75KCl+0.25CaCl2

Temperature
(T/K)

Partial vapor
pressure of
KCl (p/Pa)

Activity of
KCl

Activity of
KCl

1073

13.07

0.14

0.05

1103

16.98

0.11

1123

43.06

0.19

0.06

1073

25.45

0.27

0.23

1103

37.98

0.24

1123

83.61

0.37

0.24
0.61

1073

62.10

0.67

1103

113.71

0.71

1123

138.70

0.61

0.62

1352

Y. Zhang, E. Shibata, E. Kasai and T. Nakamura

1.0

1.0

0.9

0.9

0.8
Calculation result

0.6

0.7

Activity of NaCl

0.7
Activity of KCl

Experimental data

0.8
Experimental data

0.5
0.4
0.3

Calculation result

0.6
0.5
0.4
0.3

0.2

0.2

0.1

0.1

0
0

0.2

0.4

0.6

0.8

1.0

Mole fraction of KCl in KCl-CaCl2 system

Fig. 7

Activity of KCl in the KClCaCl2 system at 1073 K.

Table 5

0.2

0.4

0.6

0.8

1.0

Mole fraction of NaCl in the NaCl-CaCl 2 system

Fig. 8

Activity of NaCl in the NaClCaCl2 system at 1073 K.

The experimental and calculated data on partial vapor pressure and activity of NaCl in a NaClCaCl2 system.
Caculation
data

Experimental results
Sample compositon,
mole fraction

0.25NaCl+0.75CaCl2

0.50NaCl+0.50CaCl2

0.75NaCl+0.25CaCl2

Temperature
(T/K)

Partial vapor
pressure of
NaCl (p/Pa)

Activity of
NaCl

Activity of
NaCl
0.12

1073

7.32

0.11

1103

20.01

0.18

1123

28.53

0.18

0.12

1073

24.45

0.36

0.34

1103

55.24

0.48

1123

66.22

0.42

0.35

1073

45.01

0.66

0.67

1103

81.15

0.71

1123

109.76

0.70

0.67

Table 6 The experimental and calculated data on partial vapor pressures and activities of KCl and NaCl in the KClNaClCaCl2 system.
Caculation
data

Experimental result
Sample composition,
mole fraction

Temperature
(T/K)

Partial vapor
pressure (p/Pa)
KCl

0.50CaCl2 +0.25NaCl+0.25KCl

0.50KCl+0.25CaCl2 +0.25NaCl

0.50NaCl+0.25CaCl2 +0.25KCl

NaCl

Activity
KCl

Activity
NaCl

KCl

NaCl
0.20

1073

12.92

13.12

0.14

0.19

0.10

1103

31.10

19.41

0.09

0.13

1123

34.62

27.75

0.15

0.18

0.10

0.20
0.25

1073

41.44

18.57

0.44

0.27

0.35

1103

64.75

17.55

0.40

0.15

1123

87.93

41.42

0.39

0.26

0.36

0.25
0.47

1073

9.35

33.47

0.10

0.49

0.16

1103

20.52

51.39

0.13

0.45

1123

37.08

60.89

0.16

0.39

0.16

0.47

4.2.4 KClNaClCaCl2
Table 6 shows the measured data of partial vapor pressures
and activities of KCl and NaCl in the KClNaClCaCl2
system, along with the calculated results obtained using
Factsage 5.2. Figure 9 shows the comparison between the
experimental and calculated data on the activities of KCl and
NaCl in a 0.25CaCl2 NaClKCl system at 1073 K. As

evident, they agreed well with each other. As compared to

NaCl, KCl shows a larger negative deviation from Raoults
Law, and in some concentration scopes (mole fraction 0.25
0.50), the activity coecient of NaCl is approximately l.0,
which suggests that NaCl behaves as in an ideal solution.
This suggests that there is a stronger anity between KCl and
CaCl2 than that between NaCl and CaCl2 .

Vapor Pressure Measurements for Metal Chloride Systems by the Knudsen Eusion Method

1353

suggests a greater anity between KCl and CaCl2 than

between NaCl and CaCl2 .

0.75
0.6

REFERENCES

Activity

0.45
NaCl
KCl

0.3

0.15

0
0

0.15

0.3

0.45

NaCl

0.6

0.75
KCl

Mole fraction

Fig. 9 Activities of KCl and NaCl in the 0.25CaCl2 KClNaCl system at

1073 K.

5.

Conclusion

The testing results indicated that the present apparatus

is suitable for detecting the vapor pressures of metal
chlorides. The vapor pressures of the KClNaClCaCl2
systems were measured, and the results agree reasonably
well with the calculation data obtained using Factsage 5.2.
For the KClNaCl system, the results suggested that its
condensed and vapor phases exihibit similar behaviors
with ideal solution and ideal gas, respectively. In the KCl
CaCl2 and NaClCaCl2 systems, the activitis of KCl/NaCl
showed negative deviations from Raoults Law, which
suggested possible interactions between KCl and NaCl with
CaCl2 . In the ternary system, KClNaClCaCl2 , the activity
of KCl showed a larger negative deviation from Raoults
Law than that of NaCl, and in some concentration range
(mole fraction 0.250.50), NaCl behaves as in an ideal
solution (the activity coecient is approximately 1.0), which

1) K. D. Carlson: The Characterization of High-Temperature Vapors,

ed. by J. L. Margrave (John Wiley and Sons, Inc., New York, 1967)
Chap. 5, pp. 115129.
2) E. D. Cater: Physicochemical Measurements in Metals Research, Part
1, ed. by R. A. Rapp (John Willey and Sons, Inc., New York, 1970)
Chap. 2A, pp. 2290.
3) X. Li, E. Shibata, E. Kasai and T. Nakamura: Mater. Trans. 43 (2002)
29032907.
4) A. Boehncke, K. Martion, M. G. Multer and H. K. Cammenga:
J. Chem. Eng. Data 41 (1996) 543545.
5) V. Oja and E. M. Sunnberg: J. Chem. Eng. Data 43 (1998) 486492.
6) M. J. S. Monte and D. M. Hillesheim: J. Chem. Thermodyn. 33 (2001)
849857.
7) C. G. De Kruif: J. Chem. Thermodyn. 12 (1980) 243248.
8) A. D. Pelton, S. A. Degterov, G. Eriksson, C. Robelin and Y.
Dessureault: Metall. Mater. Trans. B 31 (2000) 651659.
9) C. W. Bale, P. Chartrand, S. A. Degterov, G. Eriksson, K. Hack, R. Ben
Mahfoud, J. Melancon, A. D. Pelton and S. Petersen: Calphad 26
(2002) 189228.
10) A. D. Pelton and P. Chartrand: Metall. Mater. Trans. A 32 (2001) 1355
1360.
11) A. D. Pelton and P. Chartrand: Metall. Mater. Trans. A 32 (2001) 1397
1407.
12) A. D. Pelton, P. Chartrand and G. Eriksson: Metall. Mater. Trans. A 32
(2001) 14091416.
13) M. Knudsen: Ann. Phys. 34 (1911) 593656.
14) C. I. Whitman: J. Chem. Phys. 20 (1952) 161164.
15) K. Motzfeldt: J. Phys. Chem. 59 (1955) 139147.
16) P. Chartrand and A. D. Pelton: Metall. Mater. Trans. A 32 (2001) 1361
1383.
17) I. Barin, O. Knacke and O. Kubaschewski: Thermochemical Properties
of Inorganic Substances, (Springer-Verlag, Berlin, 1977).
18) D. R. Stull and H. Prophet: JANAF Thermochemical Tables, (Unite
States Department of Commerce, Washington D.C., 1985).
19) O. Kubaschewski: Metallurgical Thermochemistry, (Pergamon Press
Ltd., 1979).
20) I. Barin: Thermochemical Data of Pure substances, (Weinheim, New
York, VCH, 1989).