You are on page 1of 8

Chemical Engineering Science 64 (2009) 4016 -- 4023

Contents lists available at ScienceDirect

Chemical Engineering Science


journal homepage: w w w . e l s e v i e r . c o m / l o c a t e / c e s

Effect of amine degradation products on the membrane gas absorption process


Julianna A. Franco a , David deMontigny b , Sandra E. Kentish a , Jilska M. Perera a , Geoffrey W. Stevens a, ∗
a
The Cooperative Research Centre for Greenhouse Gas Technologies; Department of Chemical and Biomolecular Engineering, The University of Melbourne, Victoria 3010, Australia
b
The International Test Centre for Carbon Dioxide Capture; Faculty of Engineering, University of Regina, Saskatchewan, Canada S4S 0A2

A R T I C L E I N F O A B S T R A C T

Article history: The effect of the presence of monoethanolamine (MEA) degradation products on membrane hollow fibers
Received 24 December 2008 was investigated using untreated polypropylene (PP) as a model material. Common amine oxidative
Received in revised form 26 May 2009 degradation products were added to MEA to simulate a degraded solution. The effect of these degradation
Accepted 2 June 2009
products on the membrane gas absorption process using PP hollow fiber membrane was quantified.
Available online 10 June 2009
When PP membrane which has been exposed to amine degradation products is used in a membrane gas
Keywords:
absorption contactor, the mass transfer rate of CO2 is reduced relative to the use of unexposed PP. It was
Absorption found that the presence of oxalic acid reduced the mass transfer rate of CO2 in MEA most significantly
Amine followed by formic acid and then acetic acid. These acids are believed to adsorb into the PP, altering the
Degradation surface properties and reducing the hydrophobicity of the membrane. This in turn increases the degree
Gases of wetting of the membrane pores. The membrane was characterized before and after use in a membrane
Membranes gas absorption contactor containing degraded MEA solvent and studies showed that membrane pore
Process control wetting increased by 22–31% after 69 h of use. SEM images and XPS spectra of exposed PP membrane
indicate that wetting may be due to both morphological and chemical changes in the membrane due to
contact with the solvent. This study highlights the need to consider reductions in the mass transfer rate
of membrane gas absorption processes associated with inevitable changes in the solvent composition that
comes with prolonged use.
Crown Copyright 䉷 2009 Published by Elsevier Ltd. All rights reserved.

1. Introduction Polypropylene (PP) and polyethylene (PE) membranes are cheap


and readily available and thus would be useful in an MGA service.
Absorption in a packed column using monoethanolamine (MEA) However, several studies that have been carried out to determine
or an amine equivalent is currently the preferred technology for the resistance of PP and PE fibers to water and amine solutions have
CO2 separation from flue gas streams. This approach gives high reported an increase in the wettability of the membrane surfaces
rates of absorption, high capacities for CO2 absorption at low con- after exposure (Kamo et al., 1992; Barbe et al., 2000; Wang et al.,
centrations and low raw material costs (Kohl and Nielsen, 1997). 2004). On exposure to organic solvents, the pore size of PP and PE
However, column separation is costly and it is likely that only incre- fibers has been found to increase by an amount that is a function of
mental improvements in the mature process will be possible. Mem- the surface tension of the solvent; solvents with higher surface ten-
brane gas absorption (MGA) is an emerging technology that will sion affect membrane morphology to a greater degree (Kamo et al.,
potentially reduce the cost of CO2 separation. CO2 is transferred from 1992). A mechanism for this change has been suggested; the solvent
the flue gas stream through a microporous hydrophobic membrane forms a microscopic liquid bridge between fiber filaments and sub-
into a chemical solvent. Membrane contactors allow a reduction by sequent drying of the fibers results in these filaments being drawn
4–5 times in equipment size compared with conventional packed closer together such that van der Waals forces act to maintain a more
columns (Falk-Pedersen et al., 2005). To optimise performance of open pore structure. Wang et al. (2004) exposed PP membranes to
membrane contactors, the solvent should not wet the pores of the DEA solution and observed that the pores became more elliptical in
membrane. This condition ensures that the membrane resistance shape and some pore sizes increased while others reduced, increas-
remains negligible relative to that of the gas and liquid phases. ing the pore size distribution of the membranes. Barbe et al. (2000)
also observed a change in the membrane morphology when exposed
to water, but attributed this change to the non-wetting intrusion
of the water meniscus into some pores which acts to enlarge some
∗ Corresponding author. of the pore entrances. The chemical composition of the surface of
E-mail address: gstevens@unimelb.edu.au (G.W. Stevens). PP membranes has also been studied after exposure to DEA (Wang

0009-2509/$ - see front matter Crown Copyright 䉷 2009 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2009.06.012
J.A. Franco et al. / Chemical Engineering Science 64 (2009) 4016 -- 4023 4017

et al., 2004). Amine exposure introduced a C–N bond into the mem- 2. Theory
brane composition. This work confirmed that amines can react with
PP which may be another cause for their increase in wettability 2.1. Mass transfer of CO2
following amine solvent exposure.
One of the major problems associated with the use of amine The mass transfer rate of CO2 is hindered by three resistances in
solvents is that they undergo irreversible degradation in the presence series which include the gas, membrane and liquid phases (Franco
of CO2 and other flue gas components (Kohl and Nielsen, 1997). et al, 2008a):
Three types of degradation can occur when MEA is contacted with
1 1 1 1
flue gas streams: carbamate polymerisation, thermal degradation = + + (1)
K
 kg km mEkl
and oxidative degradation (Strazisar et al., 2003).     
Overall Resistance Membrane Resistance Liquid Resistance
Carbamate polymerisation requires high temperatures and the Gas Resistance

presence of CO2 and is therefore likely to occur in the stripping


where K is the overall mass transfer coefficient, kg , km and kl are the
process. It results in the production of high molecular weight degra-
gas, membrane and liquid side mass transfer coefficients, respec-
dation products. Thermal degradation requires temperatures in
tively, m is the partition coefficient and E is the enhancement factor
excess of 205 ◦ C and is therefore less of a concern; although heat
due to chemical reaction. The CO2 flux (N) through the membrane
stable salts can be formed via thermal degradation in the regen-
contactor is given by:
erator bottoms (Kohl and Nielsen, 1997). Oxidative degradation
requires O2 which is present in abundance in flue gas streams and (Y0 − YX )G
requires the replacement of about 2.2 kg MEA per tonne of CO2 N= (2)
A
captured (Kohl and Nielsen, 1997). It is likely to occur under normal
absorption conditions. Oxidised fragments of the solvent such as where Y is the mole ratio of CO2 in the gas phase and G is the inert gas
organic acids and ammonia (and gaseous products) are produced flow rate. The overall mass transfer coefficient (K) in the contactor
which results in the most significant loss of free amine. Organic is calculated using a similar method to that used for absorption in a
acids such as formic acid and acetic acid can react to form heat packed column system (deMontigny et al., 2001).
stable salts such as acetates, formates, glycolates and oxalates that
G dY
may bind to free amine (Rooney et al., 1998; Chi and Rochelle, 2002; Kav = (3)
APT (y − y∗ ) dx
Dawodu and Meisen, 1996; Supap et al., 2006; Bello and Idem,
2005; Lawal and Idem, 2005). These salts are difficult to regenerate, where av is the specific surface area, PT is the total pressure, y is the
have no capacity for CO2 absorption, lead to increased solvent vis- mole fraction of CO2 in the gas phase and x is the contactor length.
cosity, a loss of free amine, solvent foaming and fouling and cause For chemical absorption, the equilibrium mole fraction of CO2 in the
corrosion to equipment. The currently preferred method of dealing gas phase (y∗ ) is assumed to be zero.
with this loss is to separate the active amine from its degradation In order to minimize the membrane resistance (Eq. (1)), it is re-
products using ion exchange, vacuum distillation or electrodialysis quired that conventional liquid absorbents will not wet the mem-
(Kohl and Nielsen, 1997). However, degradation products such as brane's pores. It has been shown that the membrane mass transfer
heat stable salts are always present in ppm concentrations in com- coefficient is appreciably higher for the case of non-wetted pores
mercial CO2 separation processes employing amine based solvent relative to that of wetted pores (Rangwala, 1996) due to the four-
absorption. fold increase in the mass transfer coefficient through gas-filled pores
MGA may minimize the solvent oxidation rate because the gas relative to liquid-filled pores. The contact angle  can be used to de-
and liquid phases are not in direct contact. However, it is also scribe the hydrophobicity of a surface and thus its resistance to pore
possible that the inevitable presence of degradation products will wetting. For low energy surfaces such as Teflon and PP, the contact
have an adverse effect on the membrane gas absorption process. angle generally depends only on the surface tension of the liquid lg
Solvent degradation products may lead to a higher pressure drop because a low degree of interaction (only van der Waals dispersion
across the membrane length (when a hollow fiber membrane con- forces) exists between the solid and liquid. By relating the contact
figuration is used) and cause damage to the membrane because angle of a solution on a low energy surface to the solution's surface
the solvent is in direct contact with the membrane cartridge. If tension, the wettability of a system can be assessed:
the gas-side pressure remains constant, a higher pressure drop in
the liquid will bring the transmembrane pressure difference (the lg cos  = −lg + c (4)
pressure of the liquid over that of the gas) closer to the pressure
at which liquid will penetrate the membrane, particularly at the This equation may not be valid for all systems involving low sur-
fiber inlet. Further, any membrane damage may reduce the hy- face tension alkanolamines because they can adsorb into polymeric
drophobicity of the membrane and thereby reduce the pressure materials and cause a deviation from linearity. Even so, Eq. (4) serves
at which liquid will penetrate the membrane pores. In combina- to establish the high dependency of membrane wettability on liquid
tion, these two effects may lead to (or increase) membrane pore surface tension.
wetting.
In this work we consider the effect that three common MEA 2.2. Model development
degradation products have on the performance of a polypropy-
lene membrane contactor. Oxalic, formic and acetic acid have A numerical model has been developed for the prediction of CO2
been chosen because they are produced when MEA degrades mass transfer in a hollow fiber MGA system. Flow of the liquid is
(Kohl and Nielsen, 1997). The characterisation of PP membrane parallel and counter-current to the gas flow with gas flow through
after exposure to model degraded aqueous MEA solutions is the shell side of the contactor. Basic assumptions are made in de-
performed. The CO2 mass transfer rate is measured using hol- veloping the mathematical models which are similar to those used
low fiber PP membrane contactors and different degraded MEA by other researchers (Malek et al., 1997, Karoor and Sirkar, 1993,
solutions. Finally, these hollow fiber experiments are mod- Kreulen et al., 1993):
elled in order to predict the degree of membrane wetting that
occurs. (1) The operation is steady state and isothermal.
4018 J.A. Franco et al. / Chemical Engineering Science 64 (2009) 4016 -- 4023

(2) The velocity profile down the length of the fiber (z-direction) is *CMEA

(0, z) = 0, Boundary Condition (1) (10b)
parabolic in shape and depends upon the position in the radial *r
direction (r-direction).
(3) There is symmetry in the radial direction of the hollow fibers.
*CMEA

(ri , z) = 0, Boundary Condition (2) (10c)
(4) Negligible axial diffusion in the tube side occurs which is reason- *r
able to assume because the concentration gradient in the axial These equations can be made dimensionless according to the follow-
direction is much smaller than that in the radial direction. ing definitions:
(5) Gas behaviour is ideal, the solvent is non-volatile and Henry's

CMEA
law applies. z r Cl
Z= , = , X1 = , X2 =  (11)
(6) All mass transfer occurs due to chemical reaction and interface L ri mCg,0 CMEA,0
reactions are fast which is valid for the low CO2 partial pressures
(14 kPa) used. where X1 is the dimensionless free CO2 liquid phase concentration
and X2 is the dimensionless free MEA concentration.
An analysis of the system based on a differential mass balance is The following empirical correlation developed for gas shell-side
required. The following partial differential equations and boundary flow parallel to hollow fibers for gas absorption and stripping (Yang
conditions written in cylindrical coordinates can be used to describe and Cussler, 1986) has been used to estimate the gas resistance:
the mass transfer of CO2 :  
de 0.93 1/3
 Sh = 1.25 Reg Scg (12)
 2
2 L
r * Cl * Cl 1 * Cl 
2UL 1− = Dl + − k2 Cl CMEA (5)
ri *z *r2 r *r where de is the equivalent shell side diameter, L is the length of
the hollow fiber module. Other correlations give similar predictions
where l denotes liquid flow, Cl is the free CO2 concentration in the (Prasad and Sirkar, 1988; Dahuron and Cussler, 1988).

liquid phase, CMEA 
is the concentration of unreacted MEA, k2 Cl CMEA An implicit numerical method was used to solve the model using
is the rate of reaction of CO2 in the liquid phase, Dl is the diffusivity Matlab 6.5䉷 modelling software. The calculated concentration pro-
of CO2 in the liquid, UL is the average liquid velocity and k2 is the files of free CO2 and unreacted MEA could be integrated across the
2nd order reaction rate constant. The relevant initial and boundary fiber radius to obtain average concentrations at the fiber exit. By de-
conditions include: termining the average number of moles of reacted MEA exiting the
contactor, the average bound CO2 concentration could be calculated.
Cl (r, 0) = 0, Initial Condition (6a) Thus, the average CO2 flux or overall mass transfer coefficient could
be calculated as required. A good agreement between the theoretical
*Cl
(0, z) = 0, Boundary Condition (1) (6b) model and the experimental data under different experimental con-
*r ditions was obtained by determining the proportion of pore wetting,
* Cl w that minimized the sum of the squares of the errors.
Dl (ri , z) = kex (C̄g − Cg,int ), Boundary Condition (2) (6c)
*r
3. Material and methods
In this case, the external resistance (1/kex ) includes both the mem-
brane and gas phase resistances: Laboratory grade MEA (Fisher Scientific, NJ, USA) was used for
all experiments. Oxalic acid (Sigma-Aldrich Inc., > 98% purity, MO,
1 1 1 USA), formic acid (Sigma-Aldrich Inc., > 88% purity, MO, USA) and
= + (7)
kex kg km acetic acid (Sigma-Aldrich Inc., > 99.7% purity, MO, USA) were added
in parts per million concentrations to aqueous MEA solution in or-
The driving force for mass transfer across these phases is equivalent
der to imitate degraded MEA solutions. CO2 absorption experiments
to the difference between the CO2 concentration in the gas bulk
were conducted using hollow fiber membrane cartridges and af-
and that at the gas–liquid interface Cg,int . The CO2 concentration in
ter exposure of PP membrane to these solutions, various membrane
the gas bulk C̄g has been assumed to be constant and equal to the
characterisation methods were used.
inlet concentration Cg,0 , while the CO2 gas phase concentration at
Flat sheet PP was sourced from Membrana Accurel (Germany)
the gas–liquid interface Cg,int can be related to the liquid phase by
with an average thickness of 92.5 m, a pore size of 0.1 m and
assuming applicability of Henry's law. Membrane wetting can be
a porosity of 61%. Hollow fiber PP membranes were sourced from
estimated using:
Memtec (Australia) with an inner/outer diameter of 0.3/0.67 mm, an
1 (1 − w) mw average pore size of 0.2 m and a porosity of 50%.
= + (8)
km Dg  Dl 
3.1. Membrane and solvent characterisation apparatus
where w denotes the fraction of wetted membrane pores,  is the
membrane tortuosity,  is the membrane thickness,  is the mem- Since the wetting properties of a solvent are a function of its
brane porosity and Dl and Dg are the diffusivities of CO2 in the liquid surface tension, the surface tension of aqueous MEA solutions was
and gas phases respectively. measured using the pendant drop method. A modified pendant drop
The MEA concentration profile must also be considered. It is given method was used. A clear cuvette was filled with aqueous MEA so-
by the differential mass balance: lution in which an upturned metal needle with a horizontal flat tip
   ⎛ ⎞ was immersed. A bubble was formed at the end of the needle fed
2 
r 2 *CMEA

* CMEA 1 *CMEA

⎠−2k2 Cl C  by a syringe, which was controlled by a syringe pump. This equip-
2UL 1− = DMEA + MEA (9)
ri *z *r 2 r *r ment was interfaced to FTÅ200 software (First Ten Ångstroms Inc.,
Software Version 2.0 Build Number 187, UK) which allows careful
The relevant initial and boundary conditions include: control of the pump such that the bubble remains attached to the
needle. From a series of still shots of the bubble, the FTÅ200 software
 
CMEA (r, 0) = CMEA,0 , Initial Condition (10a) calculates the liquid surface tension, lg .
J.A. Franco et al. / Chemical Engineering Science 64 (2009) 4016 -- 4023 4019

Liquid Out

T5 Gas In

T4

Hollow G1
Fibres
G2
T3

Gas Out

T2
Liquid In
T1
G3

Fig. 1. Membrane contactor with liquid flow through fiber lumen (a) schematic diagram and (b) photograph of middle piece of contactor. G1–G3 = gas sample points,
T1–T5 = thermocouples.

Contact angle of water and solvent solutions on PP membrane Table 1


Properties of hollow fiber membrane modules used for MEA degradation product
was measured using the sessile drop technique on a contact angle
experiments.
goniometer which directly measures the profile of a liquid drop on a
solid surface. The contact angle goniometer was interfaced to FTÅ200 MEA impurity Base case Oxalic acid Formic acid Acetic acid
software (First Ten Ångstroms Inc., Software Version 2.0 Build Num- Number of fibers 275 275 275 275
ber 187, UK), which used “drop shape” analysis to analyse drop shape
and size and to calculate the contact angle. “Drop shape” analysis
Contactor length (cm) 13.3 12.5 12.7 12.9
calculates contact angle by measuring the tangent to the liquid pro-
file at the point of liquid contact with the solid surface by finding
the drop's baseline and curvature profile. Mass transfer area based on
ID (cm2 ) 334 324 329 334
XPS was performed using an Axis Ultra spectrometer (Kratos An- OD (cm2 ) 770 724 735 747
alytical, UK) equipped with a monochromatised X-ray source (Al
K, h = 1486.6 eV) operating at 150 W (15 kV and 10 mA) in order
Membrane contactor void fraction (%) 84 84 84 84
to determine the chemical composition of the flat sheet membrane
surfaces. The spectrometer energy scale was calibrated using the
Au 4f7/2 photoelectron peak at a binding energy of 83.98 eV. Sur- Cross-sectional area for flow
Through lumen (cm2 ) 0.194 0.194 0.194 0.194
vey scan spectra were acquired from 0 to 1100 eV binding energy
Through shell (cm2 ) 5.19 5.19 5.19 5.19
and a pass energy of 160 eV and region spectra were scanned at a
pass energy of 20 eV to obtain higher resolution spectra. The anal-
Volume of lumen (cm3 ) 0.0026 0.0024 0.0025 0.0025
ysis area was 700 m × 300 m. The peaks were fitted with syn-
thetic Gaussian–Lorentzian components and were quantified using
the sensitivity factors for the Kratos instrument. The spectra were Volume of shell (cm3 ) 0.069 0.065 0.066 0.067
calibrated to the C 1 s peak at 285.0 eV and peaks were identified
according to their binding energies (Moulder et al., 1992, Beamson
and Briggs, 1992). Polymer samples required no pre-treatment, but rent flow membrane module (Fig. 1). Membrane cartridges were
were mounted in a copper holder to prevent them from charging. constructed in house and their specifications are listed in Table 1.
Scanning electron microscopy (SEM) was conducted using a JEOL A fresh PP membrane cartridge was used for each of the four
JSM-5600 SEM (Akishima-Shi, Japan) with a tungsten hairpin fila- cases: (1) the base case that employed uncontaminated aqueous
ment and a spot size of 36, power of 10 kV, working distance of MEA solution; the cases where aqueous MEA solution was contam-
approximately 20 mm and magnification ranging from 150–2000×. inated by the impurities (2) oxalic acid (1000 ppm), (3) formic acid
Samples were attached to aluminum stubs using carbon tabs and (1000 ppm) and (4) acetic acid (100 ppm). Properties of the four
sputtered coated with a conducting material to prevent the samples membrane cartridges appear in Table 1.
from charging. Experiments were conducted for 69 h continuously in each
membrane module with 20 wt% MEA flowing through the fiber
lumen.
3.2. Hollow fiber membrane absorption apparatus Industrial grade CO2 (Praxair, > 99.9% purity, Canada) was mixed
with air in the proportion 14:86 by volume to simulate a flue gas
A series of hollow fiber membrane cartridges, which were con- stream. Gas flow controllers (Aalborg, Model GFC-17A, NY, USA) were
structed in the laboratory, were separately tested in a counter cur- used to regulate the gas flows before they entered a pipe mixer
4020 J.A. Franco et al. / Chemical Engineering Science 64 (2009) 4016 -- 4023

and filter. Analytical grade MEA (Fisher Scientific, USA) was diluted different periods of time at room temperature. Contact angles were
with deionised water to 10–30 wt% and preloaded to 0.27–0.30 mol measured on a fresh membrane and then on the same membrane
CO2 /mol MEA by bubbling CO2 through the solution using a sintered after it had been exposed to MEA for 2 and 25 days (see Fig. 2).
glass sparger before use to simulate use of a regenerated solution. As shown in Fig. 2, the PP membrane experiences a drop in contact
A magnetic drive gear pump (Cole Parmer, IL, USA) pumped MEA angle of over 10 degrees after 2 days. However, beyond 2 days, the
solution through a liquid filter and rotameter before being passed contact angle for the PP membrane remains stable which suggests
through the membrane contactor. The liquid passed through a liq- that the PP membrane is degraded by the solvent very quickly and
uid trap before entering the outlet tank to create a positive liquid that further degradation is either very slow or does not occur. The
head and prevent gas entrainment. Analogue pressure readings were surface morphology of the PP membrane was observed to determine
available for the differential pressure across both the gas and liquid the reason for the contact angle change.
phases and also for the liquid inlet and outlet (Ashcroft, CT, USA). The PP membrane suffers extensive degradation on contact with
The CO2 gas phase concentration was measured using an on-line 20 wt% MEA for a period of 25 days. The pores appear larger and
infra-red gas analyser from Nova Analytical Systems Inc. (Hamilton, disrupted and a change in porosity is also apparent. Similar changes
Model 302WP, ON, Canada). are also evident from just 2 days of MEA contact. Various mechanisms
The CO2 gas phase concentrations at equilibrium at the contactor for the degradation of the PP membranes are possible. Studies have
inlet, outlet and mid-point were measured and used to calculate found that solvents can act to increase the pore size distribution and
the overall mass transfer coefficient. Liquid CO2 loading was also reduce the porosity of membranes by either intruding into the larger
measured to complete a mass balance for CO2 absorption across the pores and displacing the pore walls or by causing the pore structure
hollow fiber unit to verify the quality of each experiment. to contract as the wet membrane fibers dry (Kosaraju et al., 2004;
Kumar et al., 2002). This mechanism alone is unlikely to be causing
4. Results and discussion the substantial morphological changes evident in Fig. 3. The MEA
could be swelling or adsorbing into the PP or alternatively, it could
4.1. Membrane and solvent characterisation be reacting with the PP.
The surface tension and pH of 20 wt% MEA does not change sig-
In order to confirm that PP membranes will become degraded and nificantly when the aqueous solvent is contaminated by oxalic acid
lose hydrophobicity during MEA absorption using un-degraded sol- (1000 ppm), acetic acid (100 ppm) or formic acid (1000 ppm) (see
vent, PP membrane was left to float on a 20 wt% solution of MEA for Table 2). Moreover, the contact angle of each of these solutions with
untreated PP membrane is similar, with the exception of formic acid,
125 which affects a slight drop in the contact angle relative to uncon-
taminated 20 wt% MEA solution.
120 It is known that MEA contact causes PP degradation. However, it
is unknown whether the presence of MEA degradation products acts
Contact Angle, θ (Degrees)

115 to increase this degradation. In order to determine whether this is


the case over short periods of time, the contact angle of water on
110 PP membrane that had been exposed to 20 wt% MEA containing the
three common MEA degradation products was measured (Fig. 4).
105 The PP exposed to small concentrations of oxalic, acetic or formic
acid each experience a lower water contact angle relative to PP
100 that has been exposed to uncontaminated aqueous MEA solution.
In order to determine whether chemical changes to the PP are re-
95 sponsible for this change in hydrophobicity, an XPS analysis was
conducted.
90
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 An XPS analysis was performed on the inner surface of PP hollow
fiber membrane that was fresh (no MEA exposure) or that had previ-
Exposure to 20 wt% MEA (days)
ously been exposed to either uncontaminated 20 wt% MEA solution
Fig. 2. Relationship of contact angle on PP membrane with exposure time to 20 wt% or MEA contaminated by either oxalic, acetic or formic acid for
MEA. 69 h for CO2 absorption experiments with solvent flow through the

Fig. 3. SEM images depicting the change in surface morphology of PP membrane between (a) fresh PP membrane and (b) PP membrane that has been exposed to 20 wt%
MEA for 25 days at a magnification of 5000×.
J.A. Franco et al. / Chemical Engineering Science 64 (2009) 4016 -- 4023 4021

Table 2
Properties of pure water, 20 wt% MEA and 20 wt% MEA contaminated with oxalic, acetic or formic acid.

Solution Surface tension (mN/m) Contact angle on PP pH

Distilled water 73.5 ± 0.7 128 ± 0.7 9.76


20 wt% MEA 67.2 ± 0.8 120 ± 2.2 12.63

20 wt% MEA+impurity
Oxalic acid (1000 ppm) 67.0 ± 0.5 117 ± 1.9 12.03
Acetic acid (100 ppm) 67.8 ± 0.4 120 ± 0.9 12.61
Formic acid (1000 ppm) 66.0 ± 0.4 115 ± 1.8 12.36

Properties were measured at 20 ◦ C.

126
Static
Contact Angle with Water (degrees)

123
Advancing
120 Receding
117

114

111

108

105

102

99

96
20 wt% MEA 20 wt% MEA + Oxalic 20 wt% MEA + Acetic 20 wt% MEA + Formic
Acid 1000ppm Acid 100ppm Acid 1000ppm
Solution Impurity
Contact Solution

Fig. 4. Water static, advancing and receding contact angles measured on PP membrane. Each PP membrane has been exposed to 20 wt% MEA or 20 wt% MEA contaminated
with 1000 ppm oxalic acid, 100 ppm acetic acid or 1000 ppm formic acid for 1 day prior to the contact angle measurement.

Table 3
XPS analysis of inside of PP hollow fiber before and after exposure to both fresh and degraded 20 wt% MEA solution inside the fiber lumen.

PP membrane exposed to Amount of contaminant (ppm) Mass concentration (%)

Oxygen Nitrogen Carbon

Unexposed – 0.6 ± 0.1 0 99 ± 10

Fresh MEA – 1.5 ± 0.2 0 99 ± 10

Degraded MEA containing


Oxalic acid 1000 2.7 ± 0.3 0.67 ± 0.07 97 ± 10
Acetic acid 100 1.7 ± 0.2 1.0 ± 0.1 97 ± 10
Formic acid 1000 11 ± 1.0 1.8 ± 0.2 87 ± 9

fiber lumen. Table 3 shows that the mass concentration of oxygen in a MGA contactor for CO2 absorption, the performance of PP mem-
increases from 1.5% for contact with uncontaminated MEA solution brane is likely to drop at a faster rate for degraded MEA solution.
to 11.2% for contact with MEA contaminated with formic acid.
This oxygen increase could be due to oxidation of the PP mem- 4.2. Membrane gas absorption tests
brane due to direct contact with a CO2 /air stream or chemical re-
action between the membrane and MEA and the MEA degradation It has been established that the presence of MEA degradation
products. However, the fact that the oxygen concentration varies in products acts to reduce the hydrophobicity of PP membrane because
all four cases along with the nitrogen concentration suggests that the these reagents chemically react with the PP. Here, the absorption
MEA and its degradation products are interacting with the PP. It is performance of PP is examined when 20 wt% MEA solvent contains
likely that the hydrophobicity of the PP hollow fiber membrane con- either 1000 ppm oxalic, 100 ppm acetic or 1000 ppm formic acid. In
tacted with uncontaminated MEA is the highest (followed by MEA addition, the change in absorption performance over a period of 69 h
containing acetic acid, MEA containing oxalic acid and then MEA has been considered. Fig. 5 shows the difference in the CO2 flux for
containing formic acid) due to these chemical reactions. When used both pure and degraded MEA solvent.
4022 J.A. Franco et al. / Chemical Engineering Science 64 (2009) 4016 -- 4023

As expected from the membrane characterization study con- in the elemental mass concentration of oxygen on the surface of PP
ducted in Section 4.1, the presence of each degradation product (by 1 and 9%, respectively), while the presence of acetic acid does
causes the mass transfer rate to be reduced relative to the case of not increase the amount of oxygen relative to when PP is exposed
pure 20 wt% MEA. Since each solvent mixture contains the same to pure 20 W% MEA solution. Experiments show that the that the
number of moles of MEA to react with CO2 , it is likely that the PP fibers used with pure 20 W% MEA show a 22% reduction in mass
drop in mass transfer performance is caused by an increase in the transfer rate after 69 h, while those used with the contaminated MEA
membrane mass transfer resistance due to the PP membrane being solutions show a 22–31% reduction in mass transfer rate after 69 h
degraded or fouled by the acids. However, in the concentrations (Fig. 6).
commonly occurring in degraded MEA solvent (Kamo et al., 1992),
PP appears to be more susceptible to degradation by oxalic and 5. Conclusions
formic acid than by acetic acid. This is supported by XPS data (Table
3) which shows that oxalic and formic acid both cause an increase Aqueous MEA solvent was contacted with PP membrane to con-
firm that destructive morphological changes occur to the PP mem-
brane surface, which reduces the membrane's hydrophobicity (a
4.5 drop in contact angle of approximately 10◦ was measured). Com-
Base mon amine oxidative degradation products (oxalic, formic and acetic
AA acid) were then added to MEA to simulate a degraded solution. The
4.0 FA effect of these degradation products on the membrane gas absorp-
OA tion process using PP hollow fiber membrane was quantified.
3.5 Contact of the three common MEA degradation products was con-
firmed to change the chemical composition of the membrane sur-
face, presumably through chemical reactions with the PP. The mass
CO2 Flux (mol/m2s) ×104

3.0 concentration of oxygen on the surface increased by approximately


10% after exposure to MEA contaminated with formic acid. A reduced
contact angle was measured on PP membrane of up to 5◦ using the
2.5
degraded aqueous MEA solutions relative to un-degraded MEA. Ex-
posure of PP membrane to degraded MEA solutions also caused a
2.0 reduction in contact angle relative to the case where PP membrane
was exposed to un-degraded MEA.
The degradation products were also found to reduce the CO2 ab-
1.5 sorption rate over time below that of a system operating with uncon-
taminated MEA solvent, presumably due to the degradation products
1.0 reacting with the PP surface and changing the chemical composi-
tion of the PP membrane surface, which in turn reduces membrane
hydrophobicity. The results indicate that the PP fibers used with un-
0.5 degraded 20 wt% MEA show a 22% reduction in mass transfer rate
after 69 h, while those used with model degraded MEA solutions
show a reduction in mass transfer rate of 22–31% in the same time
0.0
period. Preliminary work on improving the hydrophobicity of flat
0 22 44 66
sheet PP membrane is promising (Franco et al., 2008b) but is yet to
Hours of Absorption
be adapted for hollow fibers.
It is important to consider the effect that non-ideal solvent so-
Fig. 5. Change in CO2 flux for a solvent flow rate of 17 mL/min. Solvent flow is
through the fiber lumen. The four solvent cases including uncontaminated 20 wt% lutions will have on the CO2 absorption efficiency through different
MEA solution (base) and 20 wt% MEA contaminated with 1000 ppm formic acid (FA), membrane surfaces. Furthermore, the effect of solvent impurities
100 ppm acetic acid (AA) and 1000 ppm oxalic acid (OA) are shown. should be considered in the design of MGA systems.

Base Case FA AA OA
0.8
0.75
KGav (kmol/m3 kPa hr)

0.7
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.3
0 100 200 300 0 100 200 300 0 100 200 300 0 100 200 300
Liquid Flow Rate (m3/m2hr)

Fig. 6. Change in the product KG aV with liquid flow rate. Experiments were conducted with solvent flowing through the fiber lumen for fresh 275 fiber PP cartridges (䊏)
and then after 69 h of absorption (×). Four cases are shown: uncontaminated 20 wt% MEA solution (base) and 20 wt% MEA contaminated with 1000 ppm formic acid (FA),
100 ppm acetic acid (AC) and 1000 ppm oxalic acid (OA).
J.A. Franco et al. / Chemical Engineering Science 64 (2009) 4016 -- 4023 4023

Notation Centre for CO2 Capture, Canada. The authors would also like to ac-
knowledge Paul Pigram and Penelope Hale for assistance provided
av specific surface area (mass transfer area per unit vol with XPS analysis.
contactor), m2 /m3
A cross-sectional area for fluid flow, m2 References
C concentration of CO2 , mol/m3
Barbe, A.M., Hogan, P.A., Johnson, R.A., 2000. Surface morphology changes during
C̄g CO2 concentration in the gas bulk, mol/m3 initial usage of hydrophobic, microporous polypropylene membranes. Journal of
CMEA concentration of MEA, mol/m3 Membrane Science 172, 149–156.

CMEA free concentration of MEA, mol/m3 Beamson, G., Briggs, D., 1992. High Resolution XPS of Organic Polymers, The Scienta
ESCA300 Database. Wiley, Chichester.
de equivalent shell side diameter, m Bello, A., Idem, R., 2005. Pathways for the formation of products of the oxidative
Dg diffusivity of CO2 in gas, m2 /s degradation of CO2 -loaded concentrated aqueous monoethanolamine solutions
Dl diffusivity of CO2 in aqueous MEA solution, m2 /s during CO2 absorption from flue gases. Industrial & Engineering Chemistry
Research 44, 945–969.
E enhancement due to reaction, dimensionless Chi, S., Rochelle, G., 2002. Oxidative degradation of monoethanolamine. Industrial
G inert gas flow rate, mol/s Engineering Chemistry Research 41, 4178–4186.
k individual phase mass transfer coefficient, m/s Dahuron, L., Cussler, E., 1988. Protein extractions with hollow fibers. A.I.Ch.E. Journal
34, 130.
k2 second order reaction rate constant, m3 /mols Dawodu, O., Meisen, A., 1996. Degradation of alkanolamine blends by carbon dioxide.
kex external mass transfer coefficient, m/s The Canadian Journal of Chemical Engineering 74, 960–966.
K overall mass transfer coefficient, m/s deMontigny, D., Tontiwachwuthikul, P., Chakma, A., 2001. Parametric studies
of carbon dioxide absorption into highly concentrated monoethanolamine
L length of contactor, m
solutions. The Canadian Journal of Chemical Engineering 79, 137.
m partition coefficient, dimensionless Falk-Pedersen, O., et al., 2005. CO2 capture with membrane contactors. International
N CO2 flux, mol/m2 s Journal of Green Energy 2, 157–165.
PT total pressure, Pa Franco, J.A., deMontigny, D., Kentish, S.E., Perera, J.M., Stevens, G.W., 2008a. A
study of the mass transfer of CO2 through different membrane materials in
r radial coordinate, m the membrane gas absorption process. Separation Science and Technology 43,
ri inside radius of hollow fiber, m 225–244.
Re Reynolds number, dimensionless Franco, J.A., Kentish, S.E., Perera, J.M., Stevens, G.W., 2008b. Fabrication of a
superhydrophobic polypropylene membrane by deposition of a porous crystalline
Sc Schmidt number , dimensionless polypropylene coating. Journal of Membrane Science 318, 107–113.
Sh Sherwood number, dimensionless Kamo, J., Hirai, T., Kamada, K., 1992. Solvent-induced morphological change
UL Average liquid velocity, m/s of microporous hollow fiber membranes. Journal of Membrane Science 70,
217–224.
w fraction of membrane pores wetted by solvent, dimen- Karoor, S., Sirkar, K.K., 1993. Gas absorption studies in microporous hollow fiber
sionless membrane modules. Industrial & Engineering Chemistry Research 32, 674.
x length of contactor, m Kohl, A., Nielsen, R., 1997. Gas Purification. fifth ed. Gulf Publishing Company,
Houston. pp. 41–186.
X dimensionless concentration, dimensionless Kosaraju, P., Kovvali, A., Korikov, A., Sirkar, K., 2004. Hollow fiber membrane
y mole fraction of CO2 in the gas phase, dimensionless contactor based CO2 absorption-stripping using novel solvents and membranes.
y∗ equilibrium mole fraction of CO2 in the gas phase, di- Industrial & Engineering Chemistry Research 44, 1250–1258.
Kreulen, H., Smolders, C., Versteeg, G., van Swaaij, W., 1993. Microporous hollow
mensionless
fiber membrane modules as gas liquid contactors. Part 2. Mass transfer with
Y mole ratio of CO2 in the gas, dimensionless chemical reaction. Journal of Membrane Science 78, 217.
Z axial coordinate, dimensionless Kumar, P., Gogendoorn, J., Feron, P., Versteeg, G., 2002. New absorption liquids
for the removal of CO2 from dilute gas streams using membrane contactors.
Chemical Engineering Science 57, 1639–1651.
Lawal, A., Idem, R., 2005. Effects of operating variables on the product distribution
Greek letters
and reaction pathways in the oxidative degradation of CO2 -loaded aqueous
lg surface tension of interface between liquid and gas, MEA-MDEA blends during CO2 absorption from flue gas streams. Industrial &
mN/m Engineering Chemistry Research 44, 986–1003.
 tortuosity of membrane, dimensionless Malek, A., Li, K., Teo, K., 1997. Modeling of microporous hollow fiber membrane
modules operated under partially wetted conditions. Industrial & Engineering
 thickness of membrane or membrane wall, m Chemistry Research 36, 784.
 porosity of membrane, dimensionless Moulder, J.F., Stickle, W.F., Sobol, P.E., Bomben, K., 1992. Handbook of X-ray
 contact angle, ◦ Photoelectron Spectroscopy. second ed. Perkin Elmer Corporation (Physical
Electronics), Minnesota.
Prasad, R., Sirkar, K., 1988. Dispersion-free solvent extraction with microporous
hollow-fiber modules. A.I.Ch.E. Journal 34, 177.
Subscripts Rangwala, H., 1996. Absorption of carbon dioxide into aqueous solutions using
g gas hollow fiber membrane contactors. Journal of Membrane Science 112, 229–240.
m membrane Rooney, P., DuPart, M., Bacon, T., 1998. Oxygen's role in alkanolamine degradation.
Hydrocarbon Processing 77, 109–113.
l liquid Strazisar, B., Anderson, R., White, C., 2003. Degradation pathways for
int interface monoethanolamine in a CO2 capture facility. Energy and Fuels 17, 1034–1039.
0 entering membrane contactor Supap, T., Idem, R., Tontiwachwuthikul, P., Saiwan, C., 2006. Analysis of
monoethanolamine and its oxidative degradation products during CO2 absorption
X exiting membrane contactor from flue gases: a comparative study of GC-MS, HPLC-RID, and CE-DAD analytical
techniques and possible optimum combinations. Industrial & Engineering
Chemistry Research 45, 2437–2451.
Wang, R., Li, D.F., Zhou, C., Liang, D.T., 2004. Impact of DEA solutions with and
Acknowledgements without CO2 loading on porous polypropylene membranes intended for use as
contactors. Journal of Membrane Science 229, 147–157.
Yang, M.C., Cussler, E.L., 1986. Designing hollow-fiber contactors. A.I.Ch.E. Journal
The authors would like to thank the Cooperative Research Cen-
32, 1910.
tre for Greenhouse Gas Technologies (CO2CRC), Particulate Fluids
Processing Centre (PFPC) and Australian Research Council (ARC) for
financial assistance. Also, the receipt of a Melbourne University Post-
graduate Overseas Research Scholarship (PORES) and help from Har-
ald Berwald and the workshop at the University of Regina allowed
absorption experiments to be conducted at the International Test

You might also like