Absorption of Carbon Dioxide in Aqueous

Piperazine/Methyldiethanolamine
Sanjay Bishnoi and Gary T. Rochelle
Dept. of Chemical Engineering, The University of Texas at Austin, Austin, TX 78712

Carbon dioxide absorption in 0.6 M piperazine (PZ)r4 M methyldiethanolamine

(MDEA) was measured in a wetted wall contactor. The data were simulated using a
model that accounts for chemical reactions and transport effects with the eddy diffusi®-
ity theory. PZrMDEA blends absorb CO2 faster than monoethanolamine (MEA) or
diethanolamine (DEA) blends with MDEA at similar concentrations. The reaction of
PZ to form a monocarbamate is dominant at low loading (-0.14). The reaction of the
monocarbamate to form dicarbamate is dominant at high loading. The absorption rate
did not follow pseudo first-order beha®ior except at ®ery low loading. All carbamate and
dicarbamate formation reactions approached instantaneous beha®ior at high loading.
The series resistance due to pseudo first-order reaction and instantaneous carbamate
formation matched data throughout the entire loading range. A typical ammonia plant
absorber was simulated at isothermal conditions using the rigorous model. Gas film
limitations were dominant ()50%) only at the top of the absorber.

Introduction
The BASF corporation has commercialized activated low loading and concentrations determined from NMR data
methyldiethanolamine ŽMDEA. solvents with up to 0.8 M at high loading. The VLE measurements of Xu et al. Ž1998.
piperazine in 1.5 to 4.5 M MDEA ŽAppl et al., 1982.. These and Liu et al. Ž1999. were evaluated by Bishnoi and Rochelle
solvents are widely used in the removal of carbon dioxide Ž2002.. Unfortunately, these data have not been acquired in a
from synthesis gas in ammonia plants and hydrogenrcarbon region where there is any significant deviation from the be-
monoxide plants ŽWammes et al., 1994.. The popularity of havior of MDEA.
these solvents appears to be due to the very fast kinetics Bishnoi and Rochelle Ž2000. studied the aqueous PZ sys-
ŽBishnoi and Rochelle, 2000. which enables PZ to be added tem and showed that the rate constant of PZ with CO 2 is an
in very low concentrations. However, fundamental data on order of magnitude higher than that of conventional carba-
the PZrMDEArH 2 OrCO 2 system are very limited. mate formers such as monoethanolamine ŽMEA.. Their stud-
Bishnoi and Rochelle Ž2002. present data and a model of ies on equilibrium partial pressure and NMR have shown that
CO 2 solubility and piperazine speciation with MDEArPZ the second nitrogen on PZ is reactive to form protonated PZ
blends. Data have been obtained at CO 2 loading Ža measure- carbamate and PZ dicarbamate. The pH of these systems,
ment of CO 2 concentration in the liquid defined as moles of however, is never low enough to observe di-protonated PZ.
CO 2 per mole of total amine. from 0.005 to 0.2 in solutions Xu et al. Ž1992. studied the absorption of CO 2 from gas at
of 0.6M PZr4M MDEA at 313 to 343 K. In this loading re- ambient pressures into aqueous blends of PZrMDEA. The
gion, PZ species make a significant difference on the partial work of Xu et al. was performed at conditions where PZ was
pressure and information can be obtained about speciation at almost completely depleted at the gasrliquid interface and,
low loading. The electrolyte NRTL model was adjusted to therefore, no information can be gained about the kinetics of
match partial pressure data in the blended amine system at CO 2 with PZ ŽBishnoi and Rochelle, 2000..
We now combine knowledge of the single amine systems
ŽPZrH 2 OrCO 2 . and ŽMDEArH 2 OrCO 2 . with knowledge of
Correspondence concerning this article should be addressed to G. T. Rochelle. the thermodynamics of the blended amine system to study
Current address of S. Bishnoi: The El Paso Corp., 1001 Louisiana St., Houston,
TX 77002. the rate of absorption of CO 2 into PZrMDEA blends. We

2788 December 2002 Vol. 48, No. 12 AIChE Journal

specifically target experiments where the total CO 2 concen- 2 ⑀
tration is less than the total PZ concentration and driving
forces where PZ is not depleted at the interface. The appar-
rs
␲
tany1
ž( /
x
D
Ž3.

ent second-order rate constant of PZrCO 2 is seen to double

This dimensionless transformation is helpful in the numerical
in the presence of 4M MDEA, leading us to believe that
solution since it bounds the space variable and results in a
MDEA participates in the reaction of PZ with CO 2 . This is
linear concentration gradient for physical mass transfer.
consistent with the zwitterion mechanism proposed for DEA
We divide the r space from 0 to 1 into 38 nodes Žstarting at
and other secondary amines ŽCaplow, 1968.. Measurements
the gas liquid interface and moving into the liquid phase: 1
are also made at loading, partial pressure, and temperature
node at r s 0, 10 nodes with grid spacing of 1.0 E-4, 9 nodes
representative of industrial absorption.
with grid spacing of 1.0 E-3, 9 nodes with grid spacing of 1.0
A new rigorous model for calculating the concentration
E-2, 9 nodes with grid spacing of 0.1.. Three point forward
gradients of all species throughout the boundary layer and
differencing was used at the interface to describe all deriva-
the enhancement factor for CO 2 absorption is presented. The
tives
new model is based largely on Glasscock Ž1990. who showed
that the eddy diffusivity theory is equivalent to the penetra-
df 1
tion theory or surface renewal theory. The resulting rate-
based model is combined with the thermodynamics model of ž /dr i
s
2⌬ r
Ž y3 f i q4 f iq1 y f iq2 . Ž4.
Bishnoi and Rochelle Ž2002. to make predictions of the rate
of absorption, enhancement factor, and height and volume of Three point central difference formulas were used where
the transfer unit at industrially significant conditions. These there were equal grid sizes on each side of the node of inter-
results are presented and compared to other promoted est
MDEA systems.
df 1
Model Description ž / dr i
s
2⌬ r
Ž y f iy1 q f iq1 . Ž5.
Eddy diffusivity theory is used in this work. Glasscock and
d2 f 1
Rochelle Ž1989. demonstrated that eddy diffusivity theory was
effective for modeling acid gas absorption by amines. They ž /dr 2
i
s
Ž ⌬ r .2
Ž f iy1 y2 f i q f iq1 . Ž6.
showed that the predicted absorption rate never deviates
more than 3% from surface renewal theory predictions at The finite difference scheme for nonuniform grid size of Liu
conditions indicative of CO 2 absorption with amines. et al. Ž1994. was used at nodes where the grid size changed.
For absorption of CO 2 into a nonreactive solvent, Glass- The material balance equations are derived and presented
cock and Rochelle used the following expression to represent in Appendix E of Bishnoi Ž2000.. In x space we can express
the material balance of CO 2 in the liquid where x represents the material balance in the following form
the liquid depth away from the gas-liquid interface

ⵜ Ž Ž Diq ⑀ x 2 . ⵜ w i x . sy Ri Ž7.
d d
dx
Ž DCO q ⑀ x . dx wCO2 x
2
m
s0 Ž1.
where i is a component to be included in the model, ⑀ is the
mass transfer coefficient parameter, Ri is the overall rate of
Close to the interface Ž x s 0., mass transfer is dominated by the reaction of species i, and ⵜ is the derivative operator
the diffusion coefficient of CO 2 . As x increases Žtoward the with respect to the space variable x.
bulk solution., the effect of eddies becomes more important. We simplify this further as
This expression excludes the effect of electrical potential
on ion diffusion, which Glasscock and Rochelle show to be
ⵜ 2 w i x sy Ri Ž8.
insignificant for promoted amine systems. By making this
simplifying assumption, we must assume all ion diffusion co-
efficients to be equal. We use a value of 2 for m. This value where ⵜ 2 can be considered the Eddy Diffusivity operator.
is suggested by Prasher and Fricke Ž1974. and is consistent A proper understanding of the model used in this work
with the work of Glasscock. The mass-transfer coefficient then begins with a discussion of all species and reactions used.
becomes The following six reactions are assumed to be kinetically con-
trolled and reversible
2
o
k l,CO 2
s
␲
'⑀ D CO 2 Ž2.
MDEA:H 2 OqCO 2
Littel et al. Ž1991 .
MDEAHqqHCO3y
R
R

Ž9.
The integration is performed to infinite value of x. In or-
der to deal with the integration numerically, we must per- Pinsent et al. Ž1956 .
CO 2 qOHy HCO 3y Ž 10.
R

form a variable transformation such that the upper bound is

R

finite. This work uses the space variable transformation used Bishnoi and Rochelle Ž2000 .
PZ:H 2 OqCO 2 PZCOOyqH 3 Oq
R

in the work of Glasscock, originally proposed by Versteeg

R

Ž1987. Ž 11.

AIChE Journal December 2002 Vol. 48, No. 12 2789

 k 25 C s 1.46E4 m 6rkmol 2 ⴢ s, ⌬ H a s 8.43E4 kJrkmol
MDEAqPZqCO2 PZCOOyqMDEAHq Ž 12.

R
R
k 25 C s 4.70E4 m 3rkmol ⴢ s, ⌬ H a s 3.36E4 kJrkmol
PZCOOy:H 2 OqCO 2 PZ Ž COOy . 2qH 3 Oq Ž 13.

R
R
k 25 C s 1.27E4 m rkmol ⴢ s, ⌬ H a s 8.43E4 kJrkmol
6 2

MDEAqPZOOyqCO 2 PZ Ž COOy . 2qMDEAHq Ž 14.

R
R
For example, the rate of MDEA reaction with CO 2 is given mate does not go through a rapid proton exchange, its flux is
by considered to be zero at the interface. Electroneutrality is
also assumed at the interface. Table 1 also documents the
w MDEAHq x HCO3y boundary conditions used in this work.
ž
R 9 s k MDEA w MDEAx w CO 2 x y
K9 /
Ž 15. Table 1. Model Equations and Boundary Conditions
Conservation Equations at Each Node
where k MDEA is given by O®erall Species Material Balance
ⵜ 2 wPZxqⵜ 2 wPZCOOy xqⵜ 2 wHq PZCOOy xqⵜ 2 wPZHq x
⌬ Ha 1 1 qⵜ 2 wPZŽCOOy . 2 x s 0
k 2 s k 25⬚C ⴢ exp y ž y / Ž 16.
R T 298.15 ⵜ 2 wMDEAxqⵜ 2 wMDEAHq x s 0
ⵜ 2 wCO 2 xqⵜ 2 wHCO 3y xqⵜ 2 wCO 3s xqⵜ 2 wPZCOOy x
and the equilibrium constant qⵜ 2 wHq PZCOOy xq2ⵜ 2 wPZŽCOOy . 2 x s 0
Equilibrium Relationships
w MDEAHq x HCO3y w MDEAH q xw OH y x
K9s K 18 s
w MDEAx w CO 2 x w MDEAx
w PZH q xw OH y x
K 19 s
and is calculated by the ratio of the species in the bulk solu- w PZx
tion. w H q PZCOO y xw OH y x
K 20 s
Note that water is left out of the kinetic and equilibrium w PZCOO y x
expressions since it is considered to be constant across the w CO 3 s x
boundary layer and, therefore, can be lumped with the ap- K 17 s
w HCO 3 y xw OH y x
parent rate and equilibrium constants.
Reactions involving only a proton transfer are always con- Material Balance for Molecular CO2
sidered to be in equilibrium. All equilibrium constants are ⵜ 2 wCO 2 xyŽR 9 qR 10 qR 11 qR 12 qR 13 qR 14 . s 0
calculated directly from the ratio of products and reactants in Carbamate Buffer System Balance
the bulk solution. Since the bulk solution is speciated using ⵜ 2 wPZCOOy xqⵜ 2 wHq PZCOOy xqR 11 qR 12 yŽR 13 qR 14 . s 0
the equilibrium model of Bishnoi and Rochelle Ž2002., the Dicarbamate Material Balance
equilibrium constants used in this work are consistent with ⵜ 2 wPZŽCOOy . 2 xqR 13 qR 14 s 0
that work. These reactions are Electroneutrality
wMDEAHq xqwPZHq x s wHCO 3y xq2wCO 3s xqwOHy x
HCO 3yqOHy CO 3s qH 2 O Ž 17.
R

qwPZCOOy xq2wPZŽCOOy . 2 x
R

MDEA:H 2 O MDEAH qOH q y

Ž 18. Boundary Conditions
R
R

At x s 0
PZ:H 2 O PZHqqOHy Ž 19.
R

wCO 2 x s wCO 2 xI
R

y
PZCOO :H 2 O q
H PZCOO qOH y y
Ž 20. D PZⵜwPZxq D PZHqⵜwPZHq x s 0
R
R

K 19 wPZxywPZHq xwOHy x s 0
The equations to be solved at each node are presented in D MD EAⵜwMDEAxq D MDEAHqⵜwMDEAHq x s 0
Table 1. Along with these 11 equations at each node, we de- K 18 wMDEAxywMDEAHq xwOHy x s 0
fine boundary conditions at the interface and in the bulk so- D PZ COOyⵜwPZCOOy xq D HqPZCOOyⵜwHq PZCOOy x s 0
lution. We use the condition that all concentrations are equal K 20 wPZCOOy xywHq PZCOOy xwOHy x s 0
to the equilibrium concentrations as liquid depth approaches D HC O 3y ⵜwHCO 3y xq DCO 3 sⵜwCO 3s x s 0
infinity Ž x s⬁, r s1.. We also assume phase equilibrium of K 17 wHCO 3 y xwOHy xywCO 3 s x s 0
CO 2 at the interface leading to a known concentration of ⵜwPZŽCOOy . 2 x s 0
CO 2 at the interface for a given interfacial partial pressure of wMDEAHq xqwPZHq x s wHCO 3y xq2wCO 3s xqwOHy x
CO 2 . The concentration of species which undergo proton ex- qwPZCOOy xq2wPZŽCOOy . 2 x
change are defined by the combined buffer system flux being
At x s⬁
zero at the interface and chemical equilibrium between the
w i x s w i x o For all species i in solution
two species involved in the proton exchange. Since dicarba-

2790 December 2002 Vol. 48, No. 12 AIChE Journal

 Table 2. Absorption of CO 2 into 0.6 M PZ, 4 M MDEA at Low Loading
Loading Flux=10 8 Žmolrcm2 ⴢ s.
T Bulk Gas k g =10 5 k 1 =10 5 CO 2rmol Predicted
ŽK. PCO 2 ŽPa. Žmolratm ⴢ cm2 ⴢ s. Žmrs. Amine. Measured ŽRigorous.
295 151 2.84 1.83 0.0015 1.32"0.11 1.30
295 207 2.87 1.82 0.0021 1.87"0.15 1.78
295 261 2.90 1.82 0.0018 2.44"0.20 2.27
295 290 2.89 1.83 0.0013 2.77"0.22 2.51
313 15 2.86 2.36 0.0056 0.15"0.01 0.16
313 169 2.89 2.06 0.0039 2.05"0.16 2.10
313 212 2.90 2.47 0.0050 2.47"0.20 2.64
313 252 2.91 2.06 0.0044 2.95"0.24 3.15
313 276 2.95 2.07 0.0044 3.31"0.26 3.45
343 127 3.02 5.38 0.0011 2.15"0.17 2.17
343 188 3.06 5.30 0.0020 3.10"0.25 2.89
343 224 3.14 5.42 0.0024 3.49"0.28 3.39
343 243 3.17 5.32 0.0016 4.02"0.32 3.88

G ( 5 SLPM, L ( 2 ccrs. See Bishnoi Ž2000 ..

Physical Properties Here, the solution viscosity is calculated using the correlation
The Henry’s law constant for CO 2 ŽH CO 2 . is obtained using of Glasscock Ž1990..
the N2 O analogy with the solubility of N2 O measured in We assume that the physical properties do not vary with
MDEA and correlated by Al-Ghawas et al. Ž1989. solution loading.

HCO 2 , H 2 O Experimental Methods

HCO 2 s Ž 21. The absorption rate of carbon dioxide was determined in a
H N2 O, H 2 O
wetted wall column. The use of this apparatus to provide data
for CO 2 absorption in amine solutions has been described in
The diffusion coefficient of MDEA is obtained from corre-
detail by Bishnoi and Rochelle Ž2002. and Bishnoi Ž2000.. So-
lation of the Rowley Ž1999. data reported in Bishnoi Ž2000..
lution flowed down the outside of a stainless steel tube that
Diffusion coefficients for PZ are calculated using the diffu-
was 9.1 cm long with a contact area of 38.52 cm2. The column
sion coefficient of MDEA corrected for molecular weight by
operated at 1 to 8 atm with gas rates of 4 to 6 Lrmin. Loaded
multiplying by a factor of 1.2. Diffusion coefficients of all ions
amine solution was recirculated at 2 cm3rs from reservoirs of
are arbitrarily set at the same value as PZ.
0.4 to 2.4 L volume.
The diffusion coefficient of CO 2 in concentrated MDEA
The absorption rate was determined from continuous in-
solutions is calculated using the N2 O analogy with values of
frared analysis of the gas leaving the system. The reported
D N2 O measured by Versteeg Ž1987.. Pacheco Ž1998. presents
CO 2 partial pressure is the log mean average of the gas en-
a correlation for the diffusion coefficient of N2 O in amine
tering and leaving the contactor.
solution that takes the form of the Stokes-Einstein equation
The liquid film mass-transfer coefficient was determined
by CO 2 desorption from water and ethylene glycol mixtures.
T
D N2 O Ž cm2rs . s 5.33 Ey8 Ž 22. The gas film coefficient was determined by the absorption of
␮ 0.545
sulfur dioxide into sodium hydroxide solutions.

Table 3. Absorption of CO 2 into Partially Loaded 0.6 M PZ, 4 M MDEA Solutions at Low Driving Force
Bulk Gas k g =10 5 Loading Flux=10 7 Žmolrcm2 ⴢ s.
T PCO 2 U
PCO ŽmolrŽatm ⴢ k 1 =10 5 Žmol CO 2r Predicted
2
ŽK. ŽkPa. ŽkPa. cm2 ⴢ s.. Žmrs. mol Amine. Measured ŽRigorous.
313 5.25 0.36 1.43 2.87 0.095 3.61"0.29 3.09
313 3.42 0.54 2.36 3.00 0.111 2.61"0.21 2.07
313 3.47 0.76 2.36 2.87 0.125 2.43"0.19 1.83
313 3.53 1.05 2.36 2.89 0.140 2.19"0.18 1.52
313 4.29 1.31 2.88 2.89 0.151 0.96"0.08 1.72
313 9.04 2.40 2.85 2.87 0.185 2.60"0.21 2.85
313 14.4 4.26 2.29 2.84 0.225 3.33"0.27 3.09
313 20.5 7.91 1.62 2.84 0.281 3.21"0.26 2.61
313 27.3 10.33 1.19 2.87 0.311 4.02"0.32 2.87
343 1.08 0.01 3.42 6.31 0.004 1.83"0.15 1.41
343 2.32 0.09 3.42 6.19 0.013 3.47"0.28 2.75
343 3.53 0.42 3.42 6.00 0.027 4.43"0.35 3.10
343 4.44 0.81 3.44 6.14 0.038 5.02"0.40 3.10

G ( 5 SLPM, L ( 2 cm 3. ŽSee Bishnoi Ž2000 ...

AIChE Journal December 2002 Vol. 48, No. 12 2791

 Table 4. Absorption of CO 2 into Partially Loaded 0.6 M PZr
r4 M MDEA Solutions at High Driving Force and
313 KU
k g =10 6 Loading Flux=10 6 Žmolrcm2 ⴢ s.
Bulk Gas U
PCO ŽmolrŽatm ⴢ k 1 =10 5 Žmol CO 2r Predicted
2
PCO 2 ŽkPa. ŽkPa. cm2 ⴢ s.. Žmrs. mol Amine. Measured ŽRigorous.
97 5.37 9.35 3.44 0.244 1.93"0.97 1.23
100 7.68 9.23 3.43 0.278 1.70"0.85 1.10
99 10.15 9.39 3.46 0.309 1.59"0.80 1.00
103 12.48 9.13 3.49 0.335 1.43"0.72 0.94
95 18.25 9.30 3.53 0.391 1.20"0.60 0.74
100 17.47 9.04 3.46 0.384 1.07"0.54 0.77
150 27.86 3.08 3.66 0.466 1.42"0.71 0.76
156 31.99 3.03 3.49 0.493 1.27"0.64 0.70
155 33.14 3.07 3.62 0.500 1.25"0.63 0.70
165 40.14 3.04 3.63 0.539 0.95"0.48 0.64
166 44.02 3.04 3.64 0.558 0.94"0.47 0.61
167 49.46 3.03 3.67 0.582 0.91"0.46 0.55
175 61.10 3.01 3.65 0.625 0.64"0.32 0.46
U
See Kaganoi Ž1997 ..

Experimental Data Figure 2 shows the apparent second-order reaction of PZ

Experimental data obtained in this work are presented in with CO 2 for PZ alone and in the presence of 4 M MDEA.
Tables 2, 3 and 4 along with model predictions. A subset of Each point represents several experimental partial pressures
the Xu data is shown in Figure 4 Ždetails are given in Bish- and the procedure of extracting the apparent second-order
noi, 2000.. Model predictions of all the Xu data are shown in rate constant from the data shown in Table 2 has been de-
Figure 1. The model does a good job of fitting data over a scribed by Bishnoi and Rochelle Ž2000.. The rate constant for
wide range of measured fluxes as shown by the parity plot in reaction 11 was documented in that article as well. MDEA
Figure 1. The data of Xu et al. Ž1992. are divided into two appears to participate in the reaction of CO 2 and PZ to form
subsets: accepted and rejected. The details of this will be dis- carbamate. This observation is consistent with the zwitterion
cussed in a later section. mechanism, originally reported by Caplow Ž1968.. The zwitte-
rion mechanism has been observed in several other sec-
ondary amines ŽDanckwerts, 1979; Littel, 1991..
Interpretation of data taken at low loading Glasscock et al. Ž1991. showed that this effect can be mod-
Table 2 shows data for the absorption of CO 2 into 0.6 M eled by including the participation of MDEA in carbamate
PZ, 4 M MDEA at low loading. The bulk phase partial pres- formation Žreaction 12.. The third-order rate constant for this
sure of CO 2 in this table is the log mean average of inlet and reaction was regressed to fit the absorption data at low load-
outlet pressures. At these conditions, the concentration of all ing and found to follow the Arrhenius expression presented
ionic species is very low making all reactions except 11 and in the second section. It should be noted that modeling the
12 unimportant. Since the concentration of all reaction prod- zwitterion mechanism as two parallel reactions Žone with wa-
ucts is low, we only need to consider the forward reactions at ter deprotonation and one with MDEA deprotonation. as-
these conditions. As shown by Table 2, the model matches sumes that the deprotonation by PZ is negligible. This ap-
absorption data for CO 2 into unloaded solutions well. pears to be true, based on the data presented by Bishnoi and

Figure 1. Fit of the rigorous model to all PZr

rMDEA ad- Figure 2. Comparison of CO 2 absorption into aqueous
sorption data. PZ, PZr
rMDEA blend, and MEA.

2792 December 2002 Vol. 48, No. 12 AIChE Journal

 stant for reactions 13, however, is low requiring us to adjust it
manually. The rate constant for reaction 14 is set by assum-
ing the same ratio for the rate constant of 14 to 13 as 12 to
11. This assumes that the effect of MDEA on the reaction of
PZ with CO 2 is the same as the effect of MDEA on the
reaction of PZCOOy with CO 2 .
Figure 3 compares measured enhancement factor and pre-
dictions from several simple models. The solid points repre-
sent experimental data, while the open points represent the
model predictions.
The pseudo first-order model ŽDanckwerts, 1979; shown in
Eq. 23. is a good assumption for MDEA solutions since the
reaction of MDEA is slow enough that no significant deple-
tion of MDEA occurs at the interface. The pseudo first-order
curve for the blend matches the data only at low loading.
However, it is not a good assumption for PZrMDEA blends
Figure 3. Enhancement factors for 0.6 M PZr
r4 M MDEA
at the practical conditions of the data taken in loaded solu-
at 40⬚C and k1o s3.3E-5mr
rs.
tions in this work

Ž Ýk Am w Am x . DCO
Rochelle Ž2000. where there was no effective change in the
second-order rate constant of PZ in 0.2 M solutions and 0.6
M solutions. We would expect the deprotonation by PZ to be
E PFO s ( k l2
2
Ž 23.

as fast as the deprotonation by MDEA, however, we can ig- The enhancement factor predicted by the simple model of
nore the PZ reaction because its low concentration never al- instantaneous reactions and small driving force Žderived in
lows it to dominate. In the aqueous system, the deprotona- Bishnoi, 2000; E GLBL,INST . is also not a good approximation
tion of the zwitterion by water to form hydronium ion will of the data
dominate.
k o l, PROD w Am x T HCO 2
E GLBL,INST s U Ž 24.
Data taken in loaded solutions o
k l,CO 2
⭸ PCO 2
r⭸␣
Data Taken in this Work. Table 3 shows the absorption
measurements taken in this work and model predictions in A good approximation is obtained at high loading if we
loaded solutions of PZrMDEA. We have included a finite only consider the PZ reactions to be instantaneous by in-
rate for the formation of PZ dicarbamate from PZ mono- creasing all carbamate and dicarbamate formation rate con-
carbamate Žreactions 13 and 14.. Without the inclusion of re- stants to very large numbers Ž E PZ,INST .. The dashed line in
actions 13 and 14, the data taken in loaded solutions was Figure 3 is the combined enhancement factor for pseudo first
always underpredicted Ž40% lower than current model pre- order with instantaneous PZ reactions defined by Eq. 25. A
dictions.. The sensitivity of the predicted flux to the rate con- derivation of Eqs. 24 and 25 can be found in Bishnoi Ž2000.

1
Es
 1
E PZ,INST
q
E PFO
1 0 Ž 25.

In the rigorous model, the second-order rate constant for

piperazine carbamate and CO 2 to form piperazine dicarba-
mate Žreaction 13. was adjusted and found to be 90,000
Lrmol ⴢ s at 40⬚C. This is lower than the rate constant of
piperazine to form piperazine carbamate Žreaction 11., which
was found by Bishnoi and Rochelle Ž2000. to be 102,000
Lrmol ⴢ s. This is consistent with the Bronsted theory, which
suggests a decrease in carbamate formation rate with a de-
crease in amine pK a. We believe we have found a higher
value for the reaction 13 rate constant than the Bronsted the-
ory would suggest Žaround 60,000 Lrmol ⴢ s. since it is the
only parameter adjusted to match the high loading data. It,
therefore, encompasses many other charges as the solution is
loaded with CO 2 .
The gas-phase partial pressure of CO 2 in the experimental
Figure 4. Enhancement factors for 0.1 M PZrr4.21 M data taken here are 5 to 6 times above the value in equilib-
MDEA at 40⬚C and k1o s 2E-5 mr
rs. rium with the liquid phase. The presence of 0.6 M PZ en-

AIChE Journal December 2002 Vol. 48, No. 12 2793

hances the rate of removal above that of 4M MDEA by al- Once again, the assumption of instantaneous reactions with
most two orders of magnitude at moderate loading Ž-0.15. small driving forces or pseudo first-order behavior would re-
and up to an order of magnitude at high loading Ž-0.3.. sult in a very large error of the predicted enhancement fac-
Data Taken in Loaded Solutions with High Dri®ing Force. tor. The combination of instantaneous PZ reactions with
We analyze CO 2 absorption rates in 0.6M PZr4M MDEA at pseudo first order does an adequate job of modeling the data
40⬚C with high driving force Ždata of Kaganoi Ž1997... Data Ž1 E PZ,INST q1rE PFO .. The rigorous model does an adequate
and model predictions are shown in Table 4. The gas side job of predicting the measured flux.
bulk phase partial pressures of CO 2 studied in this section Data of Xu et al. (1992). Xu et al. Ž1992. measured absorp-
are up to 2 atm. tion of pure CO 2 Ž1 atm. in MDEA solutions with 0.04 to 0.2
The model presented in this work consistently underpre- M PZ. Unfortunately, the measured enhancement was not
dicts Kaganoi’s data by 30᎐40%. No further parameters were significantly different from enhancement in MDEA solutions
adjusted, however, to match this data. This is because no pa- making it impossible to gain any information about the effect
rameters were found to make a significant difference in the of PZ. This is shown in Figure 4, which examines a subset of
model fit of the data, within reasonable changes of the pa- the Xu et al. Ž1992. data at 40⬚C and 0.1 M PZ, 4.21 M
rameters. The sensitivity of the model prediction of flux to MDEA. This shows that the data of Xu et al. is best de-
the important parameters is discussed in Figure 6. There are scribed as pseudo-first-order absorption into unpromoted
two explanations for the discrepancy of the model predic- MDEA. They have measured the flux by taking the differ-
tions and the measured values. ence of outlet and inlet gas-flow rates. Since absorption is
small, their flux measurements may be prone to large errors.
Ž1. There is a possibility that the discrepancy between the
Details of their data and our model predictions of their data
model prediction and the data could be an experimental er-
are shown in Appendix D of Bishnoi Ž2000..
ror. For this subset of data, the percent removal of CO 2 across
Figure 4 also demonstrates the experimental error of the
the wetted wall contactor is only about 15᎐20%. This means
Xu et al. data. It is unreasonable to expect that PZ has a
that the flux is calculated by taking the difference of two large
negative effect on the enhancement factor, yet this is what
numbers and can lead to large errors. A combined error of
the data above a loading of 0.3 seems to suggest. Further-
5% in reading the partial pressure in and out of the contac-
more, one of the published points predicts an enhancement
tor can result in a 25% error in the flux. Also, error in the
factor of 0.3, which is unreasonable. The data set of Xu et
driving force due to uncertainty in the equilibrium partial
al., therefore, is divided into two sets: accepted points Žthose
pressure could result in an error of up to 50% in the en-
that lie above the MDEA curve. and rejected points Žthose
hancement factor at Kaganoi’s experimental conditions. All
that lie below..
of this data was taken in two experiments where it appears
Model Predictions for the Single Amine Systems. The data
that the amine solution and analyzer calibrations were not
presented by Bishnoi and Rochelle Ž2000. for CO 2 absorp-
changed from one run to the next, leading us to believe the
tion into aqueous PZ have been reinterpreted using the rig-
error could be systematic.
orous model ŽTable 5.. As can be seen, the inclusion of the
Ž2. The model does not account for the direct reaction of PZ carbamate rate constant has improved the model predic-
MDEA with carbamate species. With Table 6, we show that tions to within 10%, showing that the model can be used in
carbamate reversion to CO 2 and the subsequent reaction of the limit of no MDEA.
the freed CO 2 with MDEA occurs, but is never significant. Model Predictions. The model developed in this work has
There is possibly a direct reaction of carbamate with MDEA been used to generate insight as to how PZ activated MDEA
which makes absorption faster than predicted in this work. blends derive their effectiveness. We present a comparison of
We recommend experimental investigation of this phe- PZrMDEA blends to other promoted systems ŽMEArMDEA
nomenon by keeping the loading of the solution constant and and DEArMDEA.. The relative sensitivity of model parame-
varying the driving force. If there is a direct interaction of the ters is also shown along with predictions that show what the
carbamate with MDEA, the model prediction of the flux will important reactions are at different conditions. The specific
increasingly underpredict the measured flux with increasing case of ammonia plants is also discussed.
driving force. Comparison with other Promoters. Figure 5 compares the
The high driving force data is also shown in Figure 3. Even effectiveness of PZrMDEA blends with DEArMDEA and
at these conditions, it appears as though 0.6 M PZ results in MEArMDEA blends. Results are compared to pseudo-first-
a significant enhancement above the performance of 4M order predictions for MDEA. The MEA and DEA blend cal-
MDEA. culations are from the work of Glasscock et al. Ž1991.. We

Table 5. Model Predictions of CO 2 Absorption into 0.6 M PZ at 313 K (data of Bishnoi and Rochelle, 2002)
Loading Flux=10 7 Žmolrcm2 ⴢ s.
Bulk Gas k g =10 5
k 1 =10 4 Žmol CO 2r Previous Predicted
PCO 2 ŽPa. ŽmolrŽatm ⴢ cm2 ⴢ s.. Žmrs. mol Amine. Measured PredictionU ŽRigorous.
2,208 2.11 1.11 0.29 2.39"0.19 2.19 2.18
2,207 2.02 1.10 0.47 2.14"0.17 1.60 1.81
2,470 2.18 1.12 0.56 1.88"0.15 1.52 1.79
2,600 2.11 1.11 0.67 1.47"0.12 1.02 1.29
U
Predictions of Bishoni and Rochelle Ž2000. which exclude PZCOOy kinetics.

2794 December 2002 Vol. 48, No. 12 AIChE Journal

Figure 5. Comparison of PZr
rMDEA blends to conven- Figure 6. Model sensitivity to rate and mass-transfer
tional blends. parameters.
i U Predictions in 5r45 wt. % PZrMDEA at 40⬚C and k 1o s1.0
Predictions at 40⬚C, k 1o s1.0E-4 mrs. P rP s1.05 unless
specified. E-4 mrs.

see that the PZ blend enhances the performance of MDEA

by a factor of 50 at low loading. Even at high loading Ž0.5 mol total concentration of PZ Žloading - 0.13., the rate constant
U
CO 2rmol amine; PCO 2
s 33 kPa., the PZ blend can be up to of PZ is by far the most sensitive parameter showing that the
a factor of 5 more effective than 50% MDEA. dominant phenomenon is still the direct reaction of CO 2 with
PZ is found to be a more effective promoter than MEA or PZ. In the second region, where the total concentration of
DEA, especially at low loading. The high driving force is not CO 2 is lower than two times the concentration of PZ Žload-
seen to have a major effect on the predicted enhancement ing - 0.26., the rate constant of PZ carbamate becomes the
factor except at high loading. The decrease in enhancement most sensitive parameter Žregion 2.. In the third region, where
is due to the depletion of PZ and PZ carbamate at the inter- the total concentration of CO 2 is higher than two times the
face. total concentration of PZ, the most sensitive parameter is the
Sensiti®ity to Model Parameters. Figure 6 shows the sensitiv- diffusion coefficient of ions showing that the removal of reac-
ity of the calculated flux to several parameters using 5 wt. % tion products has become the dominant phenomenon. An-
PZr45 wt. % MDEA at 313 K. The rate constant for PZ and other interesting result is that the rate constant of MDEA is
MDEA catalyzed PZ Žreactions 11 and 12. are lumped into not more sensitive than the rate constant of PZ or PZ carba-
one parameter, as are the rate constants for PZCOOy and mate until a loading of around 0.4.
PZCOOy catalyzed with MDEA Žreactions 13 and 14.. The Reaction Zones and De®iation from Approximate Solutions.
diffusion coefficients of MDEA, PZ, and ions are also stud- Table 6 shows model predictions of the enhancement factor
ied. of CO 2 absorption into 0.6 Mr4 M blend of PZrMDEA at
The results can be divided into three regions. In the first 313 K and a k 1o s1.0 E-4 mrs. A series of driving force con-
region, where the concentration of CO 2 is lower than the ditions varying from a low driving force Ž1.05U PCO
U
2
. to a high

Table 6. Enhancement Factors for 0.6 M PZ, 4 M MDEA at 313 K and Various Driving Force
U
PCO 2 U U U U
Loading ŽPa. E P.F.O. E 1.05 P E Series E2 P E5P E 10 P E PZ,INST E GLPL,INST
0.01 5 123.9 121.9 121.8 121.8 121.7 26,600
0.02 17 118.4 115.8 115.7 115.6 115.4
0.05 82 110.6 106.9 106.9 106.7 106.2 105.4 3214 7,840
0.07 167 103.3 98.5 98.2 97.2 95.6 4,810
0.10 404 92.4 85.5 85.8 84.8 82.8 79.7 1194 2,480
0.15 1,272 75.7 64.8 65.2 63.4 59.3 53.5 469 1,120
0.20 2,991 62.7 48.4 49.5 46.3 40.7 34.0 235 647
0.25 5,685 53.1 36.5 37.5 34.0 28.1 22.0 128 451
0.30 9,309 46.0 28.0 29.3 25.5 20.0 15.0 81 356
0.40 19,099 36.0 17.8 18.9 15.5 11.4 8.2 40 252
0.50 25,400 29.0 11.8 12.8 10.0 7.2 5.2 23 176
0.60 32,800 23.3 7.8 6.6 4.7 3.5 109
0.70 89,700 18.2 5.0 4.2 3.1 2.4

1 E series s Ž1rE PFO q1rE PZ,INST .y1

AIChE Journal December 2002 Vol. 48, No. 12 2795

 Table 7. Concentrations and Approach to Equilibrium for CO 2 Absorption into 5r rMDEA at 313 KU
r45 wt. % PZr
0.07 0.2 0.65
mol CO 2rmol Amine mol CO 2rmol Amine mol CO 2rmol Amine
Parameter Interface Bulk Interface Bulk Interface Bulk
wPZx 0.25 0.28 0.02 0.05 2.9E-4 1.8E-3
wPZCOOy x 0.15 0.14 0.098 0.13 0.02 0.04
wPZŽCOOy . 2 x 0.028 0.020 0.22 0.11 0.31 0.14
wMDEAx 3.67 3.69 2.92 3.18 1.00 1.42
Carbamate Approach to 13 100 28 100 46 100
Equil. at Interface Ž%.
Dicarbamate Approach to 14 100 40 100 86 100
Equil. at Interface Ž%.
U I U
All Concentrations presented in molrL. PCO 2
rPCO 2
s10, k 1o s1.0E-4 mrs.

driving force Ž10.0U PCOU

2
. are considered. Also included are interface Žabout 60% depletion.. Interestingly, the produc-
predictions of the pseudo-first-order enhancement factor, the tion of PZ carbamate from PZ under these conditions helps
enhancement factor predicted assuming global instantaneous the depletion of carbamate to form dicarbamate. As a result,
reactions Ž E GLBL,INST, Eq. 24., the enhancement factor pre- there is almost no depletion of PZ carbamate at the inter-
dicted assuming all carbamate formation approaches instan- face. At higher loading in region 3, we finally see some deple-
taneous behavior Ž E PZ,INST ., and the combined pseudo-first- tion of the MDEA. PZ is severely depleted, as well as PZ
orderrPZ instantaneous enhancement factor Ž1rE PFO carbamate which has changed from being a reaction product
q1rE PZ,INST .. E PZ,INST is calculated by increasing all rate in the region 1 to a reactant in region 3. There is also a steep
constants for carbamate formation until the enhancement gradient in dicarbamate at these conditions showing that
factor approaches a steady value and will increase no more much of the CO 2 is moving across the boundary layer in the
with increasing rate constant. form of dicarbamate.
Table 6 demonstrates that, although the reaction of PZ with We have also analyzed the concentration of mono and di-
CO 2 is fast, it is not fast enough to be considered instanta- carbamate as a percentage of their equilibrium value. The
neous. The behavior does, however, approach an instanta- carbamate formation reactions approach equilibrium at high
neous reaction with respect to the carbamate formation at loading, consistent with our observation that these reactions
high loading. Pseudo first order is also not a good assumption approach instantaneous behavior.
at higher loading. A simple addition of the pseudo first-order It is also useful to analyze the total amount of CO 2 being
resistance and the PZ instantaneous resistance leads to a very transferred across the boundary layer by different mecha-
accurate prediction of the enhancement factor with low driv- nisms. We define the total amount of carbamate and carbon-
ing force Ž1rE PFO q1rE PZ,INST .. ate species respectively as
Table 7 shows interfacial and bulk phase concentrations of
PZ species at various loadings. At low loading Ž ␣ s 0.07., it is w CARBAMATEx s w PZCOOy x q w Hq PZCOOy x
easy to see that pseudo first order is a good approximation.
q2 PZ Ž COOy P2 .
There is not much depletion of PZ at the interface and there
is still enough PZ around in the bulk solution. Examining the w CARBONATEx s w HCO3y x q w CO3s x
concentration of PZ carbamate, it is diffusing away from the
interface, which makes it appear as a reaction product. The At low loading, almost all the gradient in the total CO 2
moderate loading Ž ␣ s 0.20. demonstrates that there is be- species is in the form of carbamate species Ž99%.. The same
ginning to be a significant amount of depletion of PZ at the can be said about moderate loading where 94% of total CO 2

Table 8. Conditions for Ammonia Plant Example

Parameter Value Notes
Temperature 313 K Hefner et al. Ž1992.
Pressure 300 psi Hefner et al. Ž1992.
Treated Gas CO 2 Conc. 0.05 vol. % Typical Ammonia Plant
Inlet Gas CO 2 Conc. 18 vol. % Typical Ammonia Plant
Lean Loading Žmol CO 2rmol Amine. 0.095 Calculated using VLE code factor of 3 from pinch
Rich Loading Žmol CO 2rmol Amine. 0.75 Calculated using VLE code factor of 3 from pinch
Ave. Molecular Wt.-Gas 16.11 grmol Hefner et al. Ž1992.
Mass Velocity ŽGas. 2.55 kgrm2 ⴢ s Hefner et al. Ž1992.
Mass Velocity ŽLiquid. 10.3 kgrm2 ⴢ s Calculated from solvent rate
Packing Specific Area 170 my1 IMTP噛40 Wagner et al. Ž1997.
Calculated Wetted Area 132 my1 Calculated per Onda et al. Ž1968.
Packing Particle Diameter 40 mm Nominal Packing Diameter
k g ŽmolrŽcm2 ⴢ atm ⴢ s.. 1.19E-5 Calculated from Onda et al. Ž1968.
k 1 Žcmrs. 3.04E-3 Calculated from Onda et al. Ž1968.

2796 December 2002 Vol. 48, No. 12 AIChE Journal

Figure 7. McCabe Theiele diagram and enhancement Figure 8. Change in liquid and gas-phase compositions
factor prediction for ammonia plant case us- along column length for ammonia plant exam-
ing rigorous model. ple using the rigorous model.
40⬚C, P s 300 psig, k 1o s3.0E-5 mrs.
40⬚C, P s 300 psig, k 1o s 3.0E-5 mrs.

is being transferred as carbamate species. At high loading, ical phenomenon at the top of the column. The total bed
however, only 66% of the total CO 2 gradient is due to the depth calculated was 65 ft, which is larger than expected yet
carbamate species while 24% is due to molecular CO 2 and reasonable. This is probably due to the fact that most plants
the balance is due to the carbonate species. would use a higher solvent rate to account for the tempera-
ture bulge and to reduce the rich loading. Figure 8 shows the
loading and partial pressure as a function of bed depth. The
Ammonia Plantsr
rHydrogen Plants mass transfer will be in a region of low loading Ž - 0.14. for
We consider a typical ammonia or hydrogen plant scrubber approximately the first 15% of the bed height. Moderate
and make model predictions at industrial conditions. Table 8 loading Ž0.14 - ␣ - 0.28. is passed through quickly Žat 30%
shows the conditions used in developing the example. Calcu- of bed depth.. The bulk of the column is at high loading.
lated parameters such as lean amine loading, mass-transfer Figure 9 shows an overall gas phase transfer unit height and
coefficient calculated using the model of Onda et al. Ž1968. gas film resistance Ž%. as a function of bed depth. We define
and other performance parameters are shown in Table 8. We gas film control Ž%. as
have used a factor of three away from equilibrium at the lean
and rich ends in order to define the solvent rate. Mathemati- 1rk g
cally Resistance Ž % . s ⴢ 100 Ž 27.
1rK G

U
PCO 2
PCO @Top,Bottom s @Top,Bottom Ž 26.
2
3

Although we do not account for a temperature bulge in

order to keep the example simple, the chosen solvent rate
would still lie above the equilibrium curve for a bulge of 15⬚C
at the bottom of the column. Figure 7 shows the operating
line for this example and the corresponding isothermal equi-
librium curve. The operating line may be as much as a factor
of 30 deviated from the equilibrium curve at a loading of
0.20. Also shown on Figure 7 are the predicted and approxi-
mate enhancement factors according to Eq. 25 assuming in-
stantaneous reactions for all carbamate formation. The ap-
proximate enhancement factors are quite accurate at the top
and bottom, but have a significant deviation from the rigor-
ous enhancement factors in the middle.
Using the enhancement factors generated above, we can
calculate the height of a transfer unit and estimate the bed
height for the isothermal case being studied. Although this is Figure 9. Mass-transfer performance parameters for
a simplification, the top of the column will approach isother- ammonia plant example using rigorous model.
mal behavior and we can gain insight into the dominant phys- 40⬚C, P s 300 psig, k 1o s3.0E-5 mrs.

AIChE Journal December 2002 Vol. 48, No. 12 2797

 The top of the column is 50% controlled by gas film resis- CO 3ss carbonate, CO 3s
tance. The gas film resistance has reduced to 20% at a di- H 3 Oqs hydronium ion, H 3 Oq
OHys hydroxide ion, OHy
mensionless bed depth of 0.2. The transfer unit curve dis- H 2 Os water, H 2 O
tinctly shows the three regions. At dimensionless bed depths
of less than 20%, the HTU curve has one slope that corre- Symbols
sponds to low loading Ž-0.14.. At bed depth above a dimen-
Disdiffusion coefficient of species i
sionless depth of 0.35 is a second slope that corresponds to Es true enhancement factor Žboth predicted and ob-
high loading Ž)0.28. and is analogous to absorption into served.
MDEA. E GL BL,INST senhancement factor assuming all reactions are in-
stantaneous
E PF O s pseudo first-order enhancement factor
Conclusions E PZ ,INST s E. factor assuming only PZ and PZCOOy r = ns to
form carbamate are instantaneous
PZr
rMDEA blend performance H CO 2 ,H 2 O s Henry’s constant for CO 2 in H 2 O
5 wt. % PZr45 wt. % MDEA provides almost two orders H N 2 O,H 2 O s Henry’s constant for N2 O in H 2 O
o
k l,CO s liquid film mass-transfer coefficient for CO 2
of magnitude and more enhancement than 50 wt. % MDEA 2
k gs gas film mass-transfer coefficient
at low loading and one order of magnitude enhancement at K G s overall mass-transfer coefficient, expressed using gas
moderate loading. PZrM DEA solvents outperform side
MEArMDEA and DEArMDEA blends at the same concen- k 25 s rate constant at 25⬚C
tration. k xs rate constant for reaction x
mseddy diffusivity theory mass-transfer parameterᎏset
to 2 in this work
U
Approximate solutions PCO 2sequilibrium partial pressure of CO 2
PCO 2s partial pressure of CO 2
Pseudo first order is a good approximation to adsorption Rs universal gas constant
into PZrMDEA blends at low loading, but not at moderate rs liquid depth, dimensionless
to high loading. At high loading, instantaneous carbamate Ris overall rate of reaction for species i
Ts temperature
formation is approached. A simplified model that combines Xs liquid depth measured from gas liquid interface
the resistance of pseudo first-order reactions in series with y ⌬ Hasactivation energy
instantaneous carbamate formation that matches the experi- ⑀seddy diffusivity theory mass-transfer parameter
mental data and rigorous model well. ⵜsderivative operator
␮sliquid solution viscosity

Important phenomena at different absorption conditions

Literature Cited
The reaction of PZ to form monocarbamate is the domi-
Al-Ghawas, H. A., D. P. Hagewiesche, G. Ruiz-Ibanez, and O. C.
nant effect at low loading. At moderate loading, the domi- Sandall, ‘‘Physicochemical Properties Important for Carbon Diox-
nant reaction is the reaction of monocarbamate to form di- ide Absorption in Aqueous Methyldiethanolamine,’’ J. Chem. Eng.
carbamate. At high loading, the solution behaves like MDEA Data, 34, 385 Ž1989..
with instantaneous carbamate reactions. Appl, M., U. Wagner, H. J. Henrici, K. Kuessner, F. Volkamer, and
N. Ernst Neust, ‘‘Removal of CO 2 andror H 2 S andror COS From
Mass-transfer limitations due to gas-phase limitations are Gases Containing These Constituents,’’ U.S. Patent No. 4336233
important at the top of an absorber in an ammonia plant. In Ž1982..
some natural gas and flue-gas applications, gas-phase limita- Bishnoi, S., and G. T. Rochelle, ‘‘Absorption of Carbon Dioxide into
tions would not be large due to slower liquid side enhance- Aqueous Piperazine: Reaction Kinetics, Solubility and Mass
Transfer,’’ Chem. Eng. Sci., 55, 5531 Ž2000..
ment caused by a higher lean loading and higher gas film
Bishnoi, S., and G. T. Rochelle, ‘‘Thermodynamics of Pipera-
mass-transfer coefficients due to lower pressure. zinerMethyldiethanolaminerWaterrCarbon Dioxide,’’ Ind. Eng.
Chem. Res., 41, 604 Ž2002..
Bishnoi, S., ‘‘Carbon Dioxide Absorption and Solution Equilibrium
Acknowledgments in Piperazine Activated Methyldiethanolamine,’’ PhD Diss., The
This work was supported by the Separations Research Program at University of Texas at Austin Ž2000..
the University of Texas and other industrial sponsors. Caplow, M., ‘‘Kinetics of Carbamate Formation and Breakdown,’’ J.
Amer. Chem. Soc., 90, 6795 Ž1968..
Chakravarty, T., U. K. Phukan, and R. H. Weiland, ‘‘Reaction of
Notation Acid Gas with Mixtures of Amines,’’Chem. Eng. Prog., 91, 32 Ž1995..
Critchfield, J. E., ‘‘CO2 AbsorptionrDesorption in Methyldi-
ethanolamine Solutions Promoted With Monoethanolamine and
Chemical Species Diethanolamine: Mass Transfer and Reaction Kinetics,’’ PhD Diss.,
The University of Texas at Austin Ž1988..
Symbol, Name, Formula Danckwerts, P. V., ‘‘The Reaction of CO 2 with Ethanolamines,’’
PZs piperazine, C 4 H 10 N2 Chem. Eng. Sci., 34, 443 Ž1979..
PZHqs protonated piperazine, C 4 H 11 N2 Glasscock, D. A., and G. T. Rochelle, ‘‘Numerical Simulation of
PZCOOys piperazine carbamate, C 4 H 10 N2 COOy Theories for Gas Absorption with Chemical Reaction,’’ AIChE J.,
PZŽCOOy . 2 s piperazine dicarbamate, C 4 H 9 N2 ŽCOOy . 2 35, 8, 1271 Ž1989..
Hq PZCOOys protonated piperazine carbamate, C 4 H 12 N2 COO Glasscock, D. A., ‘‘Modelling and Experimental Study of Carbon
MDEAsmethyldiethanolamine, C 5 H 13 O 2 N Dioxide Absorption into Aqueous Alkanolamines,’’ PhD Diss., The
MDEAHqs protonated MDEA, C 5 H 14 O 2 N University of Texas, Austin Ž1990..
CO 2s carbon dioxide, CO 2 Glasscock, D. A., J. E. Critchfield, and G. T. Rochelle, ‘‘CO 2 Ab-
HCO 3 ys bicarbonate, HCO 3y sorptionrDesorption in Mixtures of Methyldiethanolamine with

2798 December 2002 Vol. 48, No. 12 AIChE Journal

 Monethanolamine and Diethanolamine,’’ Chem. Eng. Sci., 46, 2829 Prasher, B. D., and A. L. Fricke, ‘‘Mass Transfer at a Free Gas Liq-
Ž1991.. uid Interface in Turbulent Thin Films,’’ Ind. Eng. Chem. Process
Hefner, W., H. Meissner, and J. Davis, ‘‘Revamping of CO 2 Re- Des. De®., 13, 4, 336 Ž1974..
moval Systems by Activated MDEA Process,’’ AIChE Symp., San Pinsent, B. R., L. Pearson, and F. J. W. Roughton, ‘‘The Kinetics of
Antonio, TX ŽSept. 27᎐30, 1992.. Combination of Carbon Dioxide with Hydroxide Ions,’’ Trans.
Kaganoi, S., ‘‘Carbon Dioxide Absorption in Methydiethanolamine Faraday Soc., 52, 1512 Ž1956..
with Piperazine or Diethanolamine: Thermodynamics and Rate Rowley, R. L., ‘‘Diffusion Coefficients in Aqueous Alkanolamines,’’
Measurements,’’ MS Thesis, The University of Texas at Austin GRIrGPA Research Report RR-163, GPA Project 911, GRI Con-
Ž1997.. tract No. 5092-260-2356 Ž1999..
Lewis, W. K., and W. G. Whitman, ‘‘Principles of Gas Absorption,’’ Toman, J. J., Draft PhD Diss., The University of Texas at Austin
Ind. Eng. Chem., 16, 12 Ž1924.. Ž1990..
Littel, R. J., ‘‘Selective Carbonyl Sulfide Removal in Acid Gas Treat- Versteeg, G. F., ‘‘Mass Transfer and Chemical Reaction Kinetics in
ing Processes,’’ PhD Diss., Twente University, The Netherlands Acid Gas Treating Processes,’’ PhD Diss., Twente University, The
Ž1991.. Netherlands Ž1987..
Littel, R. J., G. F. Versteeg, and W. P. M. Van Swaaij, ‘‘Kinetics of Wagner, I., J. Stichlmair, and J. R. Fair, ‘‘Mass Transfer in Beds of
CO 2 with Primary and Secondary Amines in Aqueous Solutions: Modern, High Efficiency Random Packings,’’ Ind. Eng. Chem. Res.,
II. Influence of Temperature on Zwitterion Formation and Depro- 36, 227 Ž1997..
tonation Rates,’’ Chem. Eng. Sci., 47, 2037 Ž1992.. Wammes, W., H. Meissner, and W. Hefner, ‘‘Activated MDEA Pro-
Liu, J., G. Pope, and K. Sepehrnoori, ‘‘A High Resolution Finite cess: A Flexible Process for Acid Gas Removal from Natural Gas,’’
Difference Scheme for Non-uniform Grids,’’ Appl. Math. Mod- Conference Preprint, Annual GPA Convention, New Orleans, LA
elling, 19, 162 Ž1995.. ŽMar. 7᎐9, 1994..
Liu, H.-B., C.-F. Zhang, and G.-W. Xu, ‘‘A Study on Equilibrium Xu, G.-W., C.-F. Zhang, S.-J. Qin, and Y.-W. Wang, ‘‘Kinetics Study
Solubility for Carbon Dioxide in Methyldiethanolamine-Pipera- on Absorption of Carbon Dioxide into Solutions of Activated
zine-Water Solution,’’ Ind. Eng. Chem. Res., 38, 4032 Ž1999.. Methyldiethanolamine,’’ Ind. Eng. Chem. Res., 31, 921 Ž1992..
Onda, K., H. Takeuchi, and Y. Okumoto, ‘‘Mass Transfer Coeffi- Xu, G.-W., C.-F. Zhang, S.-J. Qin, W.-H. Gao, and H.-B. Liu, ‘‘Gas-
cients between Gas and Liquid Phases in Packed Columns,’’ J. of Liquid Equilibrium in a CO 2-MDEA-H 2 O System and the Effect
Chem. Eng. of Japan, 56, 1 Ž1968.. of Piperazine on It,’’ Ind. Eng. Chem. Res., 37, 1473 Ž1998..
Pacheco, M. A., ‘‘Mass Transfer Kinetics and Rate-Based Modeling
of Reactive Absorption,’’ PhD Diss., The University of Texas at
Austin Ž1998.. Manuscript recei®ed Aug. 22, 2001, and re®ision recei®ed Apr. 10, 2002.

AIChE Journal December 2002 Vol. 48, No. 12 2799