Enthalpies of absorption and solubility of CO2 in aqueous solutions of methyldiethanolamine

ELSEVIER

Fluid Phase Equilibria

140 (1997) 171-182

Enthalpies of absorption and solubility of CO,in aqueous solutions of methyldiethanolamine C. Mathonat

a,‘, V. Majer

a. *, A.E. Mather

b.‘, J.-P. E. Grolier

”

a Laboratoire de Thermodynamique et G&ie Chimique, Uniuersite’ Blake Paxal / CNRS, 63177 Aubi;re. France h Department of Chemical Engineering,

University of Alberta, Edmonton, Alberta, Canadu, T6G 2Gh

Received 29 August 1996; accepted 18 June 1997

Abstract A flow mixing unit using a SETARAM C-80 calorimeter, developed for measuring the enthalpy of solution of two fluids, has been used to measure enthalpies of absorption of carbon dioxide in a 30 wt.% aqueous solution of methyldiethanolamine (MDEA) at three temperatures 313.15, 353.15, 393.15 K and three pressures 2.0, 5.0, 10.0 MPa. We have established that the effect measured by calorimetry corresponds not only to the absorption of CO, in the aqueous solution but also to the vaporisation of water into the carbon dioxide depending on the temperature and the pressure of the experiment. The enthalpies measured by calorimetry were compared with those calculated from solubility measurements and a reasonable agreement within the accuracy of measurement and calculation was found. 0 1997 Elsevier Science B.V. Keywords:

Experiments; Data; Enthalpy; Carbon Dioxide; Alkanolamine

1. Introduction Some industrial processes such as natural gas purification require the removal of acidic impurities, carbon dioxide and hydrogen sulphide. They are often absorbed in chemical solvents, which are generally aqueous solutions of alkanolamines. The industrial removal process needs two important thermodynamic values which are the limit of the solubility and the enthalpy of absorption of the gas in the solvent. At Brigham Young University some measurements of the latter have been made. Christensen et al. [l], Merkley et ai [2] and Oscar-son et al. [3] reported results for absorption of CO, (MDEA) and Diethanolamine (DEA) respectively in Diglycolamine (DGA), Methyldiethanolamine solutions and Oscarson et al. [4] for H,S in DEA solutions.

* Corresponding author. ’Present address: SETARAM, 7 rue de I’Oratoire, 69300 Caluire, France. * Also corresponding

author.

037%3812/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved. PI/ SO378-3812(97)00182-9

C. Mathonat et al. /Fluid Phase Equilibria 140 (1997) I71 - 182

172

Absorption of CO, in aqueous solutions of tertiary alkanolamines is controlled by the chemical reaction which takes place in the solution [5]. The tertiary alkanolamine cannot react directly with CO, because of the lack of an hydrogen attached to the nitrogen. Barth et al. [6] discussed the possibility of forming an intermediary zwitterion (by analogy with the formation of a carbamate in the case of primary and secondary alkanolamines) and the formation of alkylcarbonates. CO, + RR’R’N = RR’R”N+COORR’R”N+COO-

(1)

+ H,O -+ RR’R”NH+

The overall reaction can be considered

+ HCO,

(2)

as a base catalysis of the direct reaction of CO, with water:

CO, + RR’R’N + H,O + RR’R”NH+

+ HCO,

(3)

Reactions are stoichiometrically limited to 1 mole of CO, per mole of amine. Tertiary amines are used preferentially in industry to remove H,S selectively. Reactions between CO, and tertiary alkanolamine solutions occur with a slow rate compared with reaction between H,S and tertiary alkanolamines. Absorption of CO, to a loading exceeding unity is possible if physical absorption is considered which explains the important absorption obtained at the temperature and pressure of the experiments. When the temperature is increased, the basicity of alkanolamine is decreased and hence the absorption of CO, by alkanolamine is decreased.

2. Experimental The chemicals employed, CO, (Alphagaz, N48, 99.998 mol% pure) and MDEA (Fluka, 98% pure), were used without any further purification. Water was deionized and degassed before use. The mixing unit has been recently developed in the laboratory of Clermont-Ferrand [7] and will be described briefly. The calorimetric system uses a commercial SETARAM C-80 differential heat-flux apparatus based on the Calvet principle. The reactants are supplied to the system from two metering pumps; a Gilson Master-305 pump (operating range between 20 and 5000 ,~l min- 1 was mostly used for the liquid while an ISCO-3 14 pump (operating range starting from 80 ~1 min- ’> supplied the gas. The flow circuit consists of stainless steel capillary tubing (o.d. 1.6 mm, wall thickness 0.3 mm). The adjustment of the required pressure and maintenance of its constancy to better than 0.2 bar is assured by means of a back pressure regulator. A Grove-Mity-Mite-Mode1 91 pressure controller with an operating range up to 20 MPa connected at the end of the flow system was used. The fluids pumped into the system pass first through a heated counter-current heat exchanger allowing their thermostatting to a temperature close to that of the calorimetric block. The temperature of the exchanger is regulated by means of a Leeds & Northrup (Electromax V) PID controller, using an iron/constantan thermocouple measuring the difference of temperature between the calorimetric block and the tubes carrying the fluids entering the cell. The exchanger is located outside the active part of the calorimeter and is buried in tightly packed thermal insulation material (glass wool). The temperature of the fluids is further regulated during their passage through the temperature controlled head. The head is located close above the calorimetric block. The core of the head is a temperature controlled ‘tee’-union-piece (Valco Instrument) and serves for the final thermostatting of the fluids before mixing and for minimizing the heat flow between the cell and the surroundings along the inlet

r

and outlet tubes. The temperature of the ‘tee’-union-piece is adjusted with a precision of f O.Ol”C. A Wheatstone bridge is used to control the current supplied, by means of a power pack, to two cartridge heaters (Acim-Jouanin) supported by two metallic extension pieces placed on each side of the ‘tee’ union piece each. A 100 R platinum resistance thermometer is used as a temperature detector located inside one of the extension pieces. The output of the Wheatstone bridge is connected to a Leeds & Northrup (Electromax III) analog regulator. The mixing cell consists of a stainless steel tube (outer diameter I .6 mm) tightly coiled in about 45 loops inside a metallic confinement cylinder which fits into the well surrounded by the thermopile detector in the calorimetric block. The fluids are introduced towards the lowest point of the cell through two vertical and concentric tubes. The actual mixing (dissolution, reaction) starts at the point where the thinner tube ends; the heat accompanying this process is exchanged between the coil and the block across the wall of the confinement cylinder and the thermopile at near isothermal conditions. During an experiment, the baseline was realized with the aqueous solution of alkanolamine because when the baseline is done with the carbon dioxide, the introduction of the aqueous solution ot alkanolamine in the gaseous flow involved strong perturbation of the mixing and the calorimetric signal is stable only after a long period of time. The gas has to be absorbed in the solvent. With a proper adjustment of the head and of the preheater the baseline signal is always small (0 to 30 PV). For medium concentrations it represents in most cases less than 0.3% of the overall recorded mixing effect. An error resulting from a possible thermal perturbation after switching from one to two fluids is not very important considering the magnitude of the measured heat effects (typically 1000 to 22 000 PV). When the base line signal is well-stabilized it is recorded over IS to 30 min: subsequently the second pump is started. Before starting the second pump, it is important to be sure that the CO, pressure is slightly less than the pressure in the system to avoid perturbation in the system at the beginning of the experiment. The pressure in the pump which contains CO, is indicated by a digital manometer (Druck, DPI 600 model) with an uncertainty of 0.1% of the full scale and serves also tin precise pressure indication during an experiment. A second manometer (O-10 MPa gauge, Besanc;on Instrument, precision 0.5% of full scale) is located between the calorimeter and the back pressure regulator. The use of two manometers ensures that there is no overpressure when connecting the pump with CO,. The ‘plateau’ signal corresponding to the actual mixing effect is achieved about 20 min after starting the mixing process and the signal is recorded for about I h. After stopping introduction of one of the fluids the signal comes back to the baseline value and normally does not differ significantly from that observed before the experiment. Altogether about 2 h are necessary fog obtaining one data point for one concentration. The signal is registered on a recorder to follow the evolution of the thermopile signal in real time. The exact determination of the difference between the plateau signal and the baseline is obtained from the average values determined from the readings on a multimeter, Metrix ITT Instruments, indicating the signal with a precision of 0.0 1c/c and recorded with a computer. The actual amount of heat corresponding to the mixing process is obtained after conversion of the difference in the thermopile signals (recorded in PV) to energy units. The ISCO pump has to be calibrated with water at each experimental pressure to determine the volumetric flow rates which allow the calculation of the molar fractions of resulting mixtures (provided the densities of CO, are known). The Gilson pump is calibrated directly with aqueous solution of alkanolamine at each experimental pressure to determine the mass tlow rate\.

174

C. Mathonat et al. / Fluid Phase Equilibria 140 (1997) 171-182

The precision of the calorimeter has been shown to be better than + 1.5% for heat of mixing runs for a liquid-liquid or liquid-gas system. Because of the difficulties in mixing CO, with aqueous solutions of alkanolamine and determining the gas flow rate, the accuracy of the calorimetric measurements is estimated to be + 4% at 2.0 and 5.0 MPa and f 7% at 10.0 MPa.

3. Results and discussion Enthalpies of absorption of carbon dioxide (CO,) in aqueous solution of methyldiethanolamine were measured at the conditions listed in Table 1. It is a convenient way to present the results of the measured enthalpy AH, for CO, in MDEA-water solutions as shown in Fig. 1 for 3 13.15 K, 2.0 MPa. AH, expressed as of kJ/mol MDEA is plotted against loading (Y (mol CO,/mol MDEA). The AH, increases with the CO, loading first as a straight line which corresponds to complete absorption of the CO, in the aqueous solution of methyldiethanolamine, and then AH, grows slower to become quite constant after the saturation point or loading point as the aqueous solution of alkanolamine is saturated and no additional CO, is absorbed. The limit between ranges where the CO, is completely absorbed and where the aqueous solution of alkanolamine is saturated is marked by a saturation loading point which is dependent on the temperature and amine solution composition. The saturation point can be determined easily from the AH,,, (kJ/mol MDEA) versus (Y curves. The results from Brigham Young University are presented as enthalpies of absorption AH,,, in kJ/mol CO, as a function of cr. Merkley et al. [2,8] assumed that AH, is independent of the amount of CO, absorbed below the saturation loading point, and can be fitted by a straight line which is obtained from the slope of the first straight line describing AH, (kJ/mol MDEA) versus (Y below the saturation point. The slope was found by a modified least-squares linear regression which forces the line describing the experimental data to pass through the origin. In fact we observed that AH, is not exactly constant under the saturation point, decreasing slowly when approaching the loading point. It seems that the determination of the enthalpy of absorption below the loading point using the regression method depends on the number of points used for the calculation. We prefer to say that the enthalpy measured becomes constant when (Y tends to zero. In order to calculate the value of the enthalpy measured which is constant when cr is going to zero, we plotted AH, (kJ/mol CO,) versus the number of moles of aqueous solution (number of moles of water (n,> + number of moles of alkanolamine (n,)) per mole of CO, (n,). The results are presented in Fig. 2a,b,c for each temperature and the various pressures. In a first region which corresponds to the saturation of the alkanolamine solution, AH, can be described by a straight line which passes through the origin. For all temperatures and pressures except 2.0 MPa and 393.15 K, all the straight lines can be superposed. The corresponding slope is approximately 3 kJ/mol solution within 10%

Table 1 Experimental

conditions

at which enthalpies

Temperature (K) Concentration (wt.% MDEA) Pressure (MPa)

of absorption

(AH,,,) were measured 313.15, 353.15,393.15 30 2.0, 5.0, 10.0

C. Mathonat

0 -“~---0.00

et al./

Fluid

I

Phase Equilibria

-1

140 (1997)

17/LIX2

~~

2.00

4.00

ci (mol CO*/ mol MEA)

Fig. I. Measured enthalpies AH, (kJ/mol MDEA) versus CO, loading in an aqueous solution 30 wt.% of MDEA at 313.15 K and 2.0 MPa.

LY(mol CO? /mol

MDEA) for absorption

of CO,

depending on temperature and pressure. In a second region where CO, is completely absorbed (solution is not saturated), the data are much more scattered and AH, seems to be nearly constant when the number of moles of CO, is tending to zero. In this region, at a given temperature, AH,,, seems to be independent of pressure and an average value was calculated. Similar plots with the data from the Brigham Young University for 20 and 40 wt.% MDEA solutions at different temperatures, were made. Both, our and Brigham Young University results are presented in Table 2 with the corresponding standard deviation. We observed that A H,, increases with temperature within the error bars; the results are presented in Fig. 3. During the plotting of the results from Brigham Young University, we observed some changes in the slopes of AH,,, versus ~1,+ n,/n2 in the region where the aqueous solution is saturated and corresponding lines do not pass through the origin. We also observed a change of slope during our experiment at the highest temperature and at the lowest pressure. Measurements [9] of the solubility of water in CO, at these conditions show that the vapour phase is 10.2 mol% H,O. Hence during our calorimetric experiment, the enthalpy measured corresponds not only to the absorption of CO, in the aqueous solution of alkanolamine but also to the vaporisation of the water into the CO,. At the experimental conditions from Brigham Young University work, the vaporisation of water is much more important as the solubility of the water increases as the temperature increases and as the pressure decreases. Measurements are much more complicated at the highest pressure where the flow rates of CO, are small (typically between 8 and 20 ~1 min -I) and are very sensitive to the variation of the CO, pressure in the pump due to possible changes of the room temperature. In this case the pressure of COZ is more difficult to control, and AH,,, values, expressed in kJ/mol CO,, are more scattered. Measurements were restarted at different periods of time and the agreement between the data is about

176

C. Mathonat et al./Fluid

O

--~‘-

0

‘--I 20

Phase Equilibria 140 (1997) 171-182

I

I

40

60

T-

i’e-m-~m 80

100

(na+nw)/nCO,

Fig. 2. (a) Measured enthalpies AH,,, (kJ/ mol CO,) versus number of mole of solution (water + MDEA) per mole of CO* for absorption of CO, in an aqueous solution 30 wt.% MDEA solution at 313.15 K. 0 2.0 MPa, A 5.0 MPa, 0 10.0 MPa. (b) Measured enthalpies AH,,, (kJ/mol CO,) versus number of mole of solution (water + MDEA) per mole of CO, for absorption of CO, in an aqueous solution 30 wt.% MDEA solution at 353.15 K. 0 2.0 MPa, A 5.0 MPa, 0 10.0 MPa. (c) Measured enthalpies AH, (kJ/mol COz) versus number of mole of solution (water+MDEA) per mole of CO, for absorption of CO, in an aqueous solution 30 wt.% MDEA solution at 393.15 K. ? ? 2.0 MPa, A 5.0 MPa, 0 10.0 MPa.

C. Mutlwnut

et al. /Fluid

Phme Equilihrirr

140 CIYY71 I?/-

80

I82

100

Fig. 2. continued

& 1.5%. In the present study, data representing certain CO, loadings were taken at several different volumetric flow rates to see if the AH,,, is affected when flow rates increases. We have observed no flow rate dependence in the range of our experimental flow rates. Thus we have assumed that equilibrium is reached under the various conditions of flow rates.

Table 2 Enthalpies of absorption of CO,

in aqueous solution of MDEA

below the saturation loading point when cy ten&

T(K)

p (MPa)

- LI H,,, kJ/mol

30

313.15

2.0, 5.0, 10.0

4Y

30

353.15

2.0, 5.0, 10.0

55

3

30

393.15

2.0, 5.0, 10.0

5x

5

20

288.75

0.156, 1.121

47

5

20

333.15

0.156,

54

2

20

388.15

1.121

59

3

20

422.05

I.466

64

2

40

288.75

0.156,

I.121

4Y

2

40 40

333.15 388.15

0.156, 1.121

I.121

57 62

3 3

40 __-

422.05

1.466

64

2

wt.9

CO,

u kJ/mol CO,

This wvrk

Merkle~

4

ct al. 181

I.121

to /cm

C. Mathonat et al./Fluid

178

Phase Equilibria 140 (1997) 171-182

1, I 1, ,’

,’

A

280

320

360

400

440

W)

Fig. 3. Measured enthalpies AH,,, (kJ/mol CO,) MDEA, ? ?Merkley et al. [8] for 20 wt.% MDEA,

when

(Y tends to zero versus temperature.

0

this work for 30 wt.%

0 Merkley et al. [8] for 40 wt.% MDEA.

3.1. Limit of solubility The calorimetric data can serve for indirect determination of the saturation loading point of CO, in the solution. From Fig. 1, it is apparent that AH,,, is independent of the CO, loading after the saturation loading point. The point where the enthalpy begins to deviate from the straight line describing the data for the completely saturated solution is the saturation loading point. It was determined as the intersection between a polynomial function describing data just before the saturation point (identified by a dashed line) and the straight line after this point. More than four parameters are usually necessary for good description of the dashed curve.

Table 3 Limit of solubility of CO, in aqueous solution of MDEA (30 wt.%) obtained from calorimetric with the values from the literature T (K)

p (MPa)

cr (mol CO, /mol

313.15 313.15 313.15 353.15 353.15 353.15 393.15 393.15 393.15

2.0 5.0 10.0 2.0 5.0 10.0 2.0 5.0 10.0

1.17 1.22 1.26 0.87 1.20 1.35 0.51 0.76 0.90

“Literature values obtained by interpolation

MDEA); this work

a’ (mol CO, /mol 1.21 1.33 1.42 0.93 1.18 1.37 0.56 0.87 1.09

of the experimental

data.

measurements

compared

MDEA); Jou et al. [IO]

C. Mathonat

et cd./ Fluid Phase Equilihriu

17’)

140 (19971 171-182

The saturation loading points determined by means of calorimetric measurements are reported in Table 3. Values of limit of solubility of CO, obtained by direct measurements in aqueous solutions of MDEA [lo] are also reported in Table 3. A comparison of the experimental loading points and the loading points determined by direct determination of the solubility show a good agreement (within 8%). At the highest pressure the agreement is not as good than at the lowest pressures (about 12%). The saturation loading point decreases when the temperature increases and increases with the pressure. 3.2. Comparison of enthalpies of absorption obtained from calorimetric obtained .from solubility measurements

measurements

with enthalpies

Few values of AH,,, for aqueous MDEA solutions are available for comparison. Jou et al. [ 101 measured the solubility of CO, in aqueous solutions of MDEA over a wide range of conditions of temperature, pressure, and concentration. From those measurements, the enthalpy of solution of carbon dioxide in the amine solutions was obtained by application of the Gibbs-Helmholtz equation to the solubility data: [tlln

P/d(l/T)I,=AHm/R

(4)

In order to compare our values with the values in the literature, we assumed a relationship between this differential enthalpy obtained from the solubility measurements and the integral enthalpy obtained from the calorimetric experiments. In the case of a mixing including alkanolamine/water + carbon dioxide, the enthalpy of solution may be written:

where ~1, is the number of moles of a pseudo compound, which is equal to n, + n ~, II,, consisting of the moles number of alkanolamine, and ~1,. moles of water; n2 is the number of moles of CO,. H,. and H,, are the enthalpies of the ‘pure’ solvent and solute, respectively. The-integral enthalpy per mole of CO, is written:

(6)

A Hi", = and: AHi,, = 2A,,

+ A&

(7)

n,

where A EZ = A Hdiff is the differential enthalpy. This equation corresponds to a straight line which is tangent to the curve representing enthalpy A Hint versus n,/n2. The slope can be written as: AH,=

aA 1

an,/%

which gives: A Hint

Hint

____

[

=

2 n2

the integral

~

I

(8)

T.,’

‘A

h/n,

Hint +

I T,p

a

Hdiff

(9)

C. Mathonat et al. / Fluid Phase Equilibria

180

We fitted our experimental A Hi,, (kJ/molCO,)

A I&, (kJ/mol

140 (I 997) 17/-1X2

CO,) versus IZ,/Q

using the following

equation:

=

where aj are fitted parameters. The number of parameters was varied between 1 and 3. Then we calculated the differential enthalpy by combining Eqs. (9) and ( 10). The results of AHdiff (kJ/mol CO,) given by Jou et al. correspond to the equilibrium between the liquid and gaseous phases, i.e. at the loading point. Hence, we calculated AI&, for each temperature and pressure studied during our work at the corresponding loading point. Thus we can compare our experimental values with the calculated data obtained from the solubility measurements obtained by Jou et al. Results are reported in Table 4 and in Fig. 4. We reported our values for each temperature and pressure. Values for Jou et al. are given for temperatures between 298.15 K and 393.15 K and pressures included between 0.1 and 20.0 MPa. Enthalpies from direct solubility measurements are given with an accuracy of + 20% which is represented in Fig. 4 by dashed lines. The calculation does not allow any temperature dependence to be seen as the procedure involves the differentiation of the pressure versus the temperature over a narrow range of temperature. However the calculation indicates that enthalpy is quite constant when a tends to zero.

Table 4 Comparison between measurements LY(mol CO, /mol

the differential

MDEA)

enthalpies -

calculated

A Hdiff&J/mol

from the direct solubilities

1.o I.1 1.17 1.2

1.22 1.26 1.35

62 62 62 61 60 58 55 51 47 43 _ 39 33 2x _ 24 _

and the calorimetric

CO,)

Jou et al. [IO] 0.01 0.05 0.1 0.2 0.3 0.4 0.5 0.5 I 0.6 0.7 0.76 0.8 0.87 0.9

measurements

This work

_ _ _ _ 30 _ -

393. I5 K, 2.0 MPa

32 _

393. I5 K, 5.0 MPa

33 25 _ _

353.15 K, 2.0 MPa 393.15 K, 10.0 MPa

21 22 18 15 13

313.15 353.15 313.15 313.15 353.15

K, K, K, K, K,

2.0 MPa 5.0 MPa 5.0 MPa 10.0 MPa 10.0 MPa

C. Mathmat

et al. / Fluid

0.40

Phuse

Eyuilihrirr

140 f/997)

0.80 c\

1 20

I7lLiX2

1 60

(mol CO,/mol MDEA)

Fig. 4. Differential enthalpies (kJ/mol CO,) versus CO, loading (Y (mol CO, /mol MDEA) at the loading point. 0 Jou et al. [IO]-error band; + this work 313.15 K and 2.0 MPa, W 313.15 K and 5.0 MPa, 0 313.15 K and 10.0 MPa. A 353.15 K and 2.0 MPa, v 353. I5 K and 5.0 MPa. b 353. I.5 K and 10.0 MPa, 4 393. IS K and 2.0 MPa, + 393. I5 K and 5.0 MPa, 4 393.15 K and 10.0 Ml%.

The comparison of the enthalpies obtained from the two methods shows a relatively good agreement within the accuracy of the calculation. The enthalpies obtained from the calorimetric measurements at 393.15 K and 2.0, 5.0, 10.0 MPa are in disagreement. At this condition of temperature the enthalpy measured corresponds also to the vaporisation of the water into the CO,. especially at 2.0 MPa.

4. Conclusion A new mixing flow apparatus using a differential calorimeter allowed the determination of enthalpies of absorption of CO, in an aqueous solution of alkanolamine with an accuracy of +4% at 2.0 and 5.0 MPa and f 7% at 10.0 MPa. Enthalpies have been measured in a range of temperatures and pressures which correspond to strong exothermic effects and where in part H,O is vaporised into CO, at the highest temperature and lowest pressure. AH, (kJ/mol CO,), when Q tends to zero, is temperature dependent and independent of the pressure within the experimental errors. The enthalpy increases slightly with temperature. Study of the curve shows that AH, expressed in kJ/mol CO, is dependent on the loading below the saturation point and tends to be constant when (Y is going to zero. Comparison of the enthalpies obtained from the calorimetric measurements and those obtained from solubilities shows a reasonable agreement.

182

C. Mathonat et al./Fluid

Phase Equilibria 140 (1997) 171-182

5. List of Symbols

X

Enthalpy Number of moles Pressure Gas constant Absolute temperature Weight percent Mole fraction

Greek letters CY CT

CO, loading Standard deviation

H n P R T wt.%

Subscripts 1 2 diff int m

Water + alkanolamine

co2 Differential Integral Measured

Superscripts

??

Pure component Partial property

Solvent abbreviation DEA DGA MDEA

Diethanolamine Diglycolamine Methyldiethanolamine

References [l] [2] [3] [4] [5] [6] [7]

[8] [9] [lo]

S.P. Christensen, J.J. Christensen, R.M. Izatt, Thermochim. Acta 106 (1986) 241-251. K.E. Merkley, J.J. Christensen, R.M. Izatt, Thermochim. Acta 121 (1987) 437-446. J.L. Oscarson, R.H. Van Dam, J.J. Christensen, R.M. Izatt, Thermochim. Acta 146 (1989) 107-l 14. J.L. Oscarson, R.H. Van Dam, J.J. Christensen, R.M. Izatt, Thermochim. Acta 154 (1989) 89-95. D.A. Glasscock, J.E. Critchfield, G.T. Rochelle, Chem. Eng. Sci. 46 (1991) 2829-2845. D. Barth, C. Tondre, G.T. Lappai, J-J. Delpuech, J. Phys. Chem. 85 (1981) 3660-3667. C. Mathonat, V. Hynek, V. Majer, J-P.E. Grolier, J. Sol. Chem. 11 (1994) 1161-l 182. K.E. Merkley, J.J. Christensen, R.M. Izatt, Gas Processors Association Research Report RR-102, September G. Mliller, E. Bender, G. Maurer, Ber. Bunsenges Phys. Chem. 92 (1988) 148-160. F-Y. Jou, F.D. Otto, A.E. Mather, Ind. Eng. Chem. Res. 33 (1994) 2002-2005.

1986.