Environ. Sci. Technol. 2001, 35, 2352-2357
Kinetics of Removal of Carbonyl Sulfide by Aqueous Monoethanolamine SANG CHEOL LEE,* MICHAEL J. SNODGRASS,† MOON KI PARK,‡ AND ORVILLE C. SANDALL† Department of Chemical Engineering, Kunsan National University, Miryong Dong, Kunsan, Chonbuk 573-701, South Korea, Deparment of Chemical Engineering, University of California, Santa Barbara, California 93106-5080, and Department of Environmental Science and Engineering, Kyungsan University, Kyungsan, Kyungpook 712-715, South Korea
The kinetics of the reaction between carbonyl sulfide and aqueous monoethanolamine (MEA) were studied over a range in temperature (298-348 K) and amine concentrations (5-20 wt %) using a wetted-sphere absorber. The experimental data were interpreted using a zwitterion mechanism. The key physicochemical properties needed to interpret the data are the solubility and diffusivity of COS in the aqueous amine solution. These properties were estimated using the N2O analogy method. Experimental values of N2O solubility were correlated using an extended scaled-particle model, and the measured N2O diffusion coefficients were correlated using a modified StokesEinstein equation. Solution densities and viscosities were also measured and correlated in this work. On the basis of the zwitterion mechanism whose rate-limiting step was the deprotonation of a zwitterion, the Arrhenius relationship between the third-order rate constant and the temperature was well correlated with an absolute mean deviation of 0.3%. It could be thus concluded that the overall reaction rate was first-order in the COS concentration and second-order in the MEA concentration.
transfer. COS has much slower kinetics with alkanolamines. Sharma (1) measured the chemical kinetics for the liquidphase reaction between CO2 and COS with several primary and secondary amines and found that the reactions with COS are approximately 100 times slower than with CO2. To properly design and develop an acid gas treating system, it is necessary to know the kinetics of the reaction and the key physicochemical properties which include the gas solubility and diffusivity in the aqueous amine solution. The purpose of the research described here was to measure the chemical kinetics of the reaction between COS and MEA in aqueous solution and to determine the physical solubility and diffusion coefficient for COS in aqueous MEA over a range in temperature and concentration. The nitrous oxide analogy method was used to determine COS solubility and diffusivity in the aqueous MEA solutions. This method, originally proposed by Clarke (2) to estimate properties for CO2 in aqueous amine solutions, was shown by Al-Ghawas et al. (3) and Littel et al. (4) to be also valid for COS. The density and viscosity of aqueous MEA solutions were also measured in this work since these properties were needed to interpret the absorption data. These are useful physical properties for the design of industrial absorbers. Reaction Mechanism. Sharma (1) studied the kinetics of COS in primary and secondary amines at 298 K and proposed a reaction mechanism involving the formation of an intermediate zwitterion. The zwitterion mechanism for the reaction of COS with the primary amine is given by: ka,k-a
COS + RNH2 798 RNH2+COSkb
RNH2+COS- + RNH2 98 RNHCOS- + RNH3+
* To whom correspondence should be addressed, Kunsan National University. Phone: +82-63-469-4775; fax: +82-63-469-4779; e-mail:
[email protected]. † University of California. ‡ Kyungsan University. 2352
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 11, 2001
(2)
where ka, k-a, and kb are rate constants for zwitterion formation, zwitterion reversion and zwitterion deprotonation, respectively. Littel et al. (5) also found that the reaction between COS and primary or secondary amines could be described by the above zwitterion mechanism. The rate of consumption of COS per unit volume (rcos) can be derived as
rcos )
Introduction In the processing of natural gas, it is usually necessary to remove acid gas compounds such as H2S, CO2, and COS. Sulfur species such as H2S and COS are toxic and corrosive and the removal of these components down to very low concentrations is often required because of severe environmental regulations. The most common means of removal of acid gas compounds in natural gas is absorption using aqueous solutions of various alkanolamines. Some typical alkanolamines used are the primary amine, monoethanolamine (MEA), the secondary amine, diethanolamine (DEA), and the tertiary amine, methyldiethanolamine (MDEA). H2S reacts extremely fast with all amines since in aqueous solution the reaction involves only a proton transfer and the kinetics may be considered to be instantaneous with respect to mass
(1)
ka[COS][RNH2] 1 1+ kb [RNH2] k-a
( )
(3)
Equation 3 assumes that the zwitterion is a reactive intermediate that rapidly establishes its steady-state composition.
Experimental Section MEA was supplied by Fisher Scientific with a minimum purity of 99% and was used without further purification. The aqueous MEA solutions were prepared by weight with deionized water. The nitrous oxide was medical grade with a stated purity of 99.99%. Carbonyl sulfide was obtained from Aldrich Chemical Co. with a minimum purity of 96%. The densities of aqueous MEA solutions were measured over the concentration range 10-40 wt % and over the temperature range 298-348 K. The density measurements were performed by using a 25-mL Gay-Lussac pycnometer which was submerged in a thermostated water bath. The bath temperature was controlled within ( 0.1 K of the test temperature. The viscosity measurements were made using a Cannon-Fenske type viscometer (B58, size 100) for 10-40 10.1021/es0017312 CCC: $20.00
2001 American Chemical Society Published on Web 04/21/2001
TABLE 1. Fitted Parameters for Density and Viscosity Correlation T (K )
A0 (cm3/mol)
A1 (cm3/mol)
B0 (cP)
B1 (cP)
B2 (cP)
298 313 333 348
-3.529 -2.588 -2.987 -2.712
-1.941 -0.3542 -1.085 - 0.6697
36.86 23.17 9.384 6.610
82.12 48.45 17.28 13.33
35.91 21.49 6.960 6.310
wt % MEA solutions over the temperature range 298-348 K. In this work, we used a modified Zipperclave Reactor to measure the solubility of N2O in aqueous MEA solutions. The apparatus and experimental procedure are described in detail by Rinker and Sandall (6). The diffusivity experiments were carried out in a wetted-sphere absorber. The experimental setup and procedure are described by Tamimi et al. (7). The chemical kinetics of the aqueous reaction between COS and MEA were measured over the MEA concentration range 5-20 wt % and over the temperature range 298-348 K. Absorption data were also obtained using the wettedsphere absorber. The experimental procedure is described by Rinker et al. (8).
Results and Discussion Density of Aqueous MEA. The density data were correlated using a Redlich-Kister type correlation (9) for the excess specific volume (VE, cm3/mol) as follows: n
VE ) xMEAxH2O
∑A (x k
MEA
k)0
- xH2O)k
(4)
where xMEA and xH2O are the mole fractions of MEA and H2O, respectively, and Ak is a correlation coefficient for VE. The specific volume of the MEA solution (Vsol) is determined from
F)
xMEAMMEA + xH2OMH2O Vsol
(5)
where MMEA and MH2O are the molecular weights of MEA and H2O, respectively, and F is the density (g/cm3) of the MEA solution. The mole fraction (xi) is calculated from the weight fraction (wi):
xi )
(wi/Mi)
(6)
∑(w /M ) j
j
FIGURE 1. Density of aqueous MEA solutions compared to correlation [(b) 10 wt %, (9) 20 wt %, (2) 30 wt %, (1) 40 wt %, this work; (O) 10 wt %, (0) 20 wt %, (4) 30 wt %, (3) 40 wt %, Weiland et al. (14); (-) correlation]. aqueous MEA density data are fitted in this way, it is found that only the first two terms are necessary to give an excellent fit to the data. Table 1 gives the A0 and A1 coefficients found as a function of temperature and Figure 1 shows the density data compared to the correlation. The correlation fits the data with an absolute mean deviation of 0.02%. The densities of aqueous MEA at 298 K reported by Weiland et al. (14) are also plotted in Figure 1 for comparison. Our experimental measurements agree with their reported values with a mean deviation of 0.18%. Viscosity of Aqueous MEA Solutions. The viscosity data were correlated in a similar manner as the density data:
ηsol ) xMEAηMEA + xH2OηH2O + ηE
(10)
where ηsol is the viscosity (cP) of the MEA solution. The excess viscosity (ηE) is correlated as n
ηE ) xMEAxH2O
∑B (x k
k)0
MEA
- xH2O)k
(11)
where Bk is the correlation coefficient for ηE. The pure water viscosity (ηH2O) data were obtained from the literature (3, 13, 15) and were correlated to give
j
The specific volume of the aqueous MEA solution is calculated from
Vsol ) xMEAVMEA + xH2OVH2O + VE
(7)
The specific volume of pure water (VH2O) was determined from an equation given by Ashour et al. (10):
VH2O ) MH2O/(0.7536 + 1.8775 × 10-3T 3.5640 × 10-6T2) (8) where T is the temperature in K. An expression for specific volume of pure MEA was developed using the MEA density data of DiGuillo et al. (11), Li and Shen (12), and Lee and Lin (13) over the temperature range 294-373 K and is given by
VMEA ) MMEA/(1.2551 - 8.1249 × 10-4T)
(9)
where VMEA is the specific volume of pure MEA. When the
ln ηH2O ) -1.4345 - 1237.4/T + 486763/T2
(12)
The pure MEA viscosity (ηMEA) data of DiGuillo et al. (11) and Lee and Lin (13) were correlated to give
ln ηMEA ) - 1.0878 - 1600.0/T + 833553/T2
(13)
The aqueous MEA solutions viscosity data obtained in this work were correlated according to eq 10 using eqs 11-13 to give the B parameters listed in Table 1. We found that the use of three parameters (B0, B1, and B2) gave an excellent fit to the data with an absolute mean deviation of 0.17%. The experimental data are shown compared to the correlation in Figure 2, which also shows that our predictions are in good accordance with the solution viscosity data of Lee and Lin (13) and Snijder et al. (16). Solubility of N2O in Aqueous MEA. A key property needed in the evaluation of the chemical kinetics is the physical solubility of COS in aqueous amine solution. As explained VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2353
TABLE 3. Values of HCOS/HN2O and DCOS/DN2O in Pure Water
a
temp (K)
HCOSa/HN2Oc
DCOSb/DN2Oc
298 313 333 348
1.038 1.041 1.157 1.419
0.879 0.829 0.888 0.946
Calculated from eq 16.
b
Calculated from eq 23. c This work.
method to correlate our solubility data for MEA. The method is described as follows:
ln(Hi) ) FIGURE 2. Viscosity of aqueous MEA solutions compared to correlation [(b) 298 K, (9) 313 K, (2) 333 K, (1) 348 K, this work; (0) 313 K, Lee and Lin (13); (O) 298 K, (4) 333 K, Snijder et al. (16); (-) correlation].
TABLE 2. Energy Parameter Estimates from N2O Solubility Data Fitting E12/k (K) system N2O-H2O
N2O-MEA
temp (K)
this work
Li and Mather
298 313 333 348 298 313 333 348
233.17 243.27 251.18 257.79 243.28 256.39 261.70 265.40
235.94 244.13 251.81 257.25 246.01 251.88 257.10 260.65
( ) ( )
kT ∂Ni
+
(17)
3y2σ1 3y1σ12 9y22σ12 + + 1 - y3 1 - y3 2(1 - y )2 3
y 2 σ1 y3
2
ln(1 - y3) +
( )[ y2σ1 y3
3
2ln(1 - y3) +
y3 y32 1 - y3 2(1 - y )2 3
]
y3(2 - y3) 1 - y3
] (18)
where
(15)
where pi is the partial pressure (MPa) of the gas component i. Another Henry’s constant H′i can be also defined in units of atmosphere‚liters/mole (atm‚L/mol) when pi is in atmospheres (atm) and Ci is the liquid-phase concentration (mol/ L) of the gas component i. The Henry’s constant for COS in water [(HCOS)H2O] as a function of temperature was determined by Al-Ghawas et al. (3):
)
(16)
In this work, we measured HN2O values for pure water and aqueous MEA solutions; thus with these experimental measurements the ratio HCOS/HN2O in water is determined (see Table 3) and eq 14 allows prediction of HCOS for aqueous MEA solutions (see Table 4). Li and Mather (17) correlated the solubility of N2O in several aqueous amine solutions using an extended scaledparticle theory proposed by Hu et al. (18). We also used this 9
j
j
]
( )[
(14) H2O
pi ) Hixi ) H′iCi
2354
∑n )
[
Hi is the Henry constant (MPa) in Henry’s law for the physical solubility:
(
+ ln(kT
3y1y2 3y23 y3y22 y0 + + σ3 1 - y3 (1 - y )2 (1 - y )3 (1 - y )3 1 3 3 3
-
1045.3 -log10(HCOS)H2O ) - 4.158 T
kT ∂Ni
1 ∂Ah ) -ln(1 - y3) + kT ∂Ni
in the Introduction, we used the N2O analogy method to estimate the COS solubility according to
HCOS ) HN2O MEA
1 ∂As
+
where Ah is the change in Helmholtz energy (J) in the transition to a hard-sphere mixture from an ideal gas mixture, As is the change in Helmholtz energy (J) from the hard-sphere mixture to a real liquid mixture, k is the Boltzmann constant, N is the number of molecules, and n is the number density (1/m3) of molecules. The Boublik (19)-Mansoori-CarnahanStarling-Leland (20) equation is used to estimate Ah. The first term in eq 17 is expressed as
+3
HCOS HN2O
1 ∂Ah
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 11, 2001
yn )
π
∑n σ 6
n
(19)
j j
j
The second term in eq 17 can be expressed by the following (17):
1 ∂As kT ∂Ni
) -5.14025
π kT
∑n jσ j j 1
1
3
(20)
j
where
1 σ1j ) (σ1 + σj) 2
(21)
In the above equations, j represents the solvent species. σ1 is the hard sphere diameter (m) of N2O and σj is the hardsphere diameter (m) of the solvent species (H2O and MEA). The hard-sphere diameters for N2O, H2O, and MEA are reported by Li and Mather. They presented a method to calculate the energy parameters, 1j, which are determined from actual experimental solubility data. Table 2 shows a comparison between the energy parameters calculated for this work with those calculated by Li and Mather for the N2O-H2O-MEA system. Figure 3 shows a plot for both
TABLE 4. Absorption Data for COS in Aqueous MEA temp (K)
[MEA] (mol/L)
H′COS (atm L/mol)
DCOS × 105 (cm2/s)
PCOS (atm)
liquid rate (cm3/s)
rCOS × 105 (mol/s)
k2,app (L mol/s)
298 298 298 298 313 313 313 313 333 333 333 333 348 348 348 348
0.817 1.638 2.462 3.293 0.814 1.631 2.451 3.277 0.807 1.615 2.428 3.261 0.800 1.601 2.405 3.211
43.07 44.42 45.77 45.85 64.03 64.71 64.90 64.91 102.16 101.41 100.26 100.44 144.67 149.92 136.64 135.13
1.626 1.477 1.362 1.231 2.122 1.840 1.791 1.560 2.779 2.602 2.282 2.042 3.330 3.046 2.933 2.611
0.967 0.969 0.969 0.969 0.926 0.928 0.931 0.933 0.804 0.808 0.809 0.815 0.624 0.630 0.639 0.645
1.23 1.22 1.11 1.40 0.58 0.43 0.80 0.60 0.88 1.66 0.83 0.95 1.07 0.97 0.35 0.43
1.330 1.627 2.115 2.822 1.335 2.301 3.089 3.809 1.635 2.336 3.426 4.495 1.449 2.053 4.044 5.016
28.06 24.92 33.53 54.00 43.13 77.75 96.01 122.98 171.18 180.97 289.31 396.54 383.35 358.53 845.21 967.18
FIGURE 4. Diffusivity of nitrous oxide in aqueous MEA solutions compared to correlation [(b) 298 K, (9) 313 K, (2) 333 K, (() 348 K, this work; (O) 298 K, Sada et al. (22); (-) correlation]. FIGURE 3. Solubility of nitrous oxide in aqueous MEA solutions compared to correlation [(b) 298 K, (9) 313 K, (2) 333 K, (1) 348 K, this work; (4) 333 K, (3) 348 K, Littel et al. (4); (-) correlation]. experimental and predicted values of N2O solubility in MEA. The absolute average deviation between N2O solubility in MEA predicted by their model and experimental N2O solubility in MEA from this work is 1.9%. Also, the solubility data of Littel et al. (4) are included in Figure 3 for comparison. Our experimental solubility data and predictions show reasonable agreement with the authors. Diffusivity of N2O in Aqueous MEA. Since COS reacts with MEA, the diffusion coefficient for COS (DCOS, cm2/s) in aqueous MEA cannot be measured directly. The nitrous oxide analogy was also used to determine the diffusivity of COS in MEA.
( ) ( ) DCOS DN2O
)
MEA
DCOS DN2O
(22)
H2O
The diffusion coefficient for COS in water [(DCOS)H2O] as a function of temperature and viscosity is given by Al-Ghawas (21) as follows:
(DCOS)H2O ) 5.54 × 10-8T/η0.682
(23)
where T is in Kelvin (K) and η is in centipoise (cP). We measured the diffusivity of N2O (DN2O) in water. Table 3 gives the ratio DCOS/DN2O in water.
The experimental diffusivity values of N2O in aqueous MEA solutions were correlated using a modified StokesEinstein equation
DN2O ) RηβT
(24)
The best fit values of R and β are 6.19 × 10-8 and -0.581, respectively. The data fit eq 24 with an absolute mean deviation of 2.9%. Figure 4 shows the diffusivity data of Sada et al. (22) and the present work compared to eq 24. The predictions of DN2O from eq 24 agree to within 9.3% with their data at 298 K. Chemical Kinetics of the COS/MEA Reaction. Absorption data were interpreted using the theoretical results of Ashour and Sandall (23) for absorption into a liquid flowing over a sphere with a second-order irreversible chemical reaction. Their model includes simultaneous diffusion and reaction in the liquid film and is described only by two partial differential equations no matter what process in the film is rate controlling. The estimated diffusivities and solubilities of COS in aqueous MEA solutions, and the experimentally measured kinetic data are given in Table 4. These values were used to solve the equations whose solutions give the relation between the rate of absorption and the apparent second-order rate constant (k2,app). Ultimately, the value of the apparent second order rate constant was optimized so that the theoretically predicted rate of absorption might agree with the experimentally measured rate of absorption (rCOS) in Table 4. The values of k2,app are also listed in Table 4. The VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2355
FIGURE 5. Apparent second-order rate constant for COS/MEA reaction as a function of temperature. apparent second-order rate constant is related to the kinetic rate constants of the zwitterion mechanism
k2,app )
1 1 1 + ka kakb [MEA] k-a
(25)
Figure 5 shows the k2,app plots as a function of MEA concentration for the temperature 298, 313, 333, and 348 K. It is seen in these plots that the k2,app data may be represented by a straight line through the origin. This indicates that the term, 1/ka, is small compared to the other term in the denominator of eq 25 and the rate-limiting step is the deprotonation of the zwitterion, eq 2. The kinetics for the reaction between COS and aqueous MEA is thus expressed by
rCOS ) k3[COS][MEA]
2
(26)
where the third-order rate constant is given by
k3 )
kakb k-a
(27)
Figure 6 shows an Arrhenius plot of the third-order kinetic rate constants. A fit of these data gives
k3 ) 1.32 × 1010 exp
(-6136 T )
(28)
Sharma (1) presented kinetics data in the form of a pseudofirst-order rate constant at 298 K and at a single MEA concentration of 1 mol/L. His reported value is 16 s-1. Equation 28 gives 15.1 s-1 for the apparent pseudo-first order rate constant at [MEA] ) 1 mol/L and 298 K. Thus, the kinetics data found in this work are in relatively close agreement with the single measurement of Sharma. Littel et al. (5) report a pseudo-first-order rate constant for the COS/MEA reaction of 18.2 s-1 at a temperature of 303 K and [MEA] ) 1 mol/L. 2356
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 11, 2001
FIGURE 6. Arrhenius plot of third-order rate constants for COS/ MEA reaction. Equation 28 gives a pseudo-first-order constant of 21.2 s-1 for these conditions. This is also fairly close agreement. However, Littel et al. considered that the contribution of H2O as well as MEA to the zwitterion deprotonation rate was important, but their zwitterion mechanism could not predict our kinetic data as satisfactorily as our mechanism did. We therefore concluded that only MEA contributed to the deprotonation of the zwitterion and the reaction orders in COS and MEA were 1 and 2, respectively.
Acknowledgments This work was funded by the Gas Processors Association.
Literature Cited (1) Sharma, M. M. Trans. Faraday Soc. 1965, 61, 681-688. (2) Clarke, J. K. A. Ind. Eng. Chem. Fundam. 1964, 3, 239-245. (3) Al-Ghawas, H. A.; Ruiz-Ibanez, G.; Sandall, O. C. Chem. Eng. Sci. 1989, 44, 631-639. (4) Littel, R. J.; Versteeg, G. F.; van Swaaij, W. P. M. J. Chem. Eng. Data 1992, 37 (7), 49-55.
(5) Littel, R. J.; Versteeg, G. F.; van Swaaij, W. P. M. AIChE J. 1992, 38, 244-250. (6) Rinker, E. B.; Sandall, O. C. Chem. Eng. Commun. 1996, 144, 85-94. (7) Tamimi, A.; Rinker, E. B.; Sandall, O. C. J. Chem. Eng. Data 1994, 39, 330-332. (8) Rinker, E. B.; Ashour, S. S.; Sandall, O. C. Chem. Eng. Sci. 1995, 50, 755-768. (9) Prausnitz, J. M.; Lichtenthaler, R. N.; de Azevedo, E. G. In Molecular Thermodynamics of Fluid-phase Equilibria; Prentice Hall: New Jersey, 1986; 2nd ed., Chapter 6. (10) Ashour, S. S.; Rinker, E. B.; Sandall, O. C. Chem. Eng. Commun. 1997, 161, 15-24. (11) DiGuillo, R. M.; Lee, R.; Schaeffer, S. T.; Brasher, L. L.; Teja, A. S. J. Chem. Eng. Data 1992, 37, 239-242. (12) Li, M.; Shen, K. J. Chem. Eng. Data 1992, 37, 288-290. (13) Lee, M.; Lin, T. J. Chem. Eng. Data 1995, 40, 336-339. (14) Weiland, R. H.; Dingman, J. C.; Cronin, D. B.; Browning, G. J. J. Chem. Eng. Data 1998, 43, 378-382. (15) Dizechi, M.; Marschall, E. J. Chem. Eng. Data 1982, 27, 358363.
(16) Snijder, E. D.; te Riele, J. M.; Versteeg, G. F.; van Swaaij, W. P. M. J. Chem. Eng. Data 1993, 38, 475-480. (17) Li, Y. G.; Mather, A. E. Fluid Phase Equilib. 1994, 96, 119-142. (18) Hu, Y.; Xu, Y. N.; Prausnitz, J. Fluid Phase Equilib. 1985, 25, 15-40. (19) Boublik, T. J. Chem. Phys. 1970, 53, 471-472. (20) Mansoori, G. A.; Carnahan, N. F.; Starling, K. E.; Leland, T. W. J. Chem. Phys. 1971, 54, 1523-1525. (21) Al-Ghawas, H. A. Ph.D Dissertation, University of California at Santa Barbara, 1988. (22) Sada, E.; Kumazawa, H.; Butt, M. A. J. Chem. Eng. Data 1978, 23, 3(2), 161-163. (23) Ashour, S. S.; Sandall O. C. Chem. Eng. Commun. 1997, 160, 199-210.
Received for review October 3, 2000. Revised manuscript received December 22, 2001. Accepted February 8, 2001. ES0017312
VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2357
