Separation and Purification Technology 69 (2009) 48–56
Contents lists available at ScienceDirect
Separation and Purification Technology
journa l homepage: www.e lsev ier .com/ locate /seppur
henol recovery with tributyl phosphate in a hollow fiber membrane contactor:xperimental and model analysis
.F. Shen a, K.H. Smith a, S. Cook a, S.E. Kentish a, J.M. Perera a, T. Bowser b, G.W. Stevens a,∗
The Particulate Fluids Processing Centre, Department of Chemical and Biomolecular Engineering, The University of Melbourne, Parkville, Victoria 3010, AustraliaGlaxoSmithKline, Princes Highway, Port Fairy 3284, Australia
r t i c l e i n f o
rticle history:eceived 11 March 2009eceived in revised form 25 June 2009ccepted 25 June 2009
a b s t r a c t
The extraction and stripping of phenol using a solution of tributyl phosphate in kerosene in a hydrophobicpolypropylene hollow fiber membrane contactor has been studied. The effect of the aqueous and theorganic phase flow rates on the overall mass transfer coefficient for both extraction and stripping steps wasinvestigated. Experimental values of the overall mass transfer coefficient were determined and compared
eywords:embrane contractorollow fiber membraneass transfer coefficient
henolributyl phosphate
with predicted values from the resistance in series model. Results showed that the overall mass transfercoefficients for extraction were about one order of magnitude greater than those measured during thestripping process. The experimental values were in good agreement with the predicted values for theextraction module. However, the predicted values were slightly overestimated for the stripping module.The individual mass transfer resistances were analyzed and the rate-controlling steps of mass transferwere also identified in both extraction and stripping modules. The major resistance in extraction andstripping was in the aqueous phase and in the membrane phase, respectively.
. Introduction
Phenol and its derivatives are often present in the wastewaterischarged of manufacturing processes such as in the petro-hemical and agrochemical industries and in coal gasificationastewaters. Due to their potential harm to human health, most
re listed as priority pollutants by the US Environmental Protectiongency [1]. Therefore, treatment processes must often be imple-ented before the waste streams can be safely discharged.
One of the most widely used methods of removing phenol isiquid–liquid extraction using esters, alcohols [1,2], alkylamines [3]nd trialkylphosphine oxides [4] as extractant solvents. The use ofributyl phosphate (TBP) as an extractant is the focus of the presentork. Compared to other solvents, TBP shows good extraction per-
ormance, has low solubility in water, a high flash point whichowers the flammability potential and is relatively inexpensive [5,6].henol forms a complex with TBP as a solvating reagent as follows6]:
H OH + TBP = C H OH·TBP (1)
6 5 6 5
Solvent extraction is conventionally conducted in devices suchs mixer-settlers and pulsed plate columns. However, these devicesxhibit important drawbacks associated with the dispersion of the
two phases, such as emulsion formation and organic phase carry-over. As a result, the wastewater streams can become contaminatedby these organic chemicals, which are often equally as polluting.These problems can be substantially eliminated using membranecontactors [7].
The use of hollow fiber membrane contactors for non-dispersivesolvent extraction has been extensively studied for many applica-tions [7–11]. In this case, the interface between the aqueous andthe organic phases is stabilized within the pores of the polymermembrane. There are many advantages associated with this non-dispersive configuration: they are highly compact devices withlarge specific surface area; the dispersion-free operation eliminatesemulsion formation, phase entrainment and downstream phaseseparation; there is no limitation on the phase flow rate ratio; noneed for density differences between the extractant and strippingphases; and there is greater flexibility in equipment configuration.Compared to dispersive extraction devices, the main drawback isthe additional membrane resistance which lowers the overall masstransfer rate. However, this can be successfully offset by the greaterinterfacial mass transfer area found within a hollow fiber mem-brane unit [7].
In the literature, many researchers have studied solvent-based
phenol recovery within hollow fiber membrane contactors [11–21].However, the use of tributyl phosphate (TBP) as a phenol carrier insuch a configuration has not been investigated. Further, there areconflicting reports on the use of empirical correlations for estimat-ing the shell side mass transfer coefficient within such a hollow
S.F. Shen et al. / Separation and Purification Technology 69 (2009) 48–56 49
Nomenclature
A membrane area of mass transfer (m2)c phenol concentration (g L−1)D distribution ratiodin inner diameter of hollow fiber membrane (m)dh the hydraulic diameter of shell side (m)dlm logarithmic mean diameter of hollow fiber (m)dout outer diameter of hollow fiber membrane (m)Daq diffusivity of phenol in the aqueous phase (m2 s−1)Dorg diffusivity of the organic phase (m2 s−1)Fi fractional resistance (%)Jsol phenol mass transfer flux (g s−1)k local mass transfer coefficient (m s−1)K overall mass transfer coefficient (m s−1)Keq the equilibrium constant of chemical reaction (Eq.
(6))l length of hollow fiber (m)M molar concentration (kmol m−3)Q defined in Eq. (5)Qaq flow rate of the aqueous phase (m3 s−1)Qorg flow rate of the organic phase (m3 s−1)Re Reynolds numberSc Schmidt numberSh Sherwood numbert time (s)V defined in Eq. (4)Vaq volume of the aqueous phase (m3)Vorg volume of the organic phase (m3)
Greek lettersı membrane thickness (m)ε membrane porosity� viscosity (kg m−1 s−1)� density (kg m−3)� tortuosity of membrane pores� interfacial tension (N m−1)� the contact angle˚ the packing fraction of membrane contactorˇ the constant, defined in Eq. (15)� defined in Eq. (2)
Superscripts* the equilibrium concentration between two phases
fitamt1ioso
part of the stripping curves multiplied by the volume of the organic
0 the initial concentration
ber contactor [11,16,18,20,21]. The aim of this study is to identifyhe rate-limiting step for phenol extraction with TBP and to developreliable model for predicting the extraction and stripping perfor-ance in a hydrophobic hollow fiber membrane contactor. In order
o do this, distribution ratios were determined for a system with0% (v/v) TBP/Shellsol2046 as the extractant and sodium hydrox-
de (NaOH) as the stripping agent. The effect of phase flow ratesn the overall mass transfer coefficient for both the extraction andtripping steps were studied and compared with predicted valuesbtained from the mass transfer model.
Fig. 1. Schematic diagram of experimental set-up for phenol extraction and strip-ping.
2. Theory
2.1. Overall mass transfer coefficient calculation
The transfer of phenol from the aqueous phase to the organicphase during the extraction process can be described in terms ofthe overall mass transfer coefficient, Kaq, which is defined by
Jsol = KaqA(caq − c∗aq) (2)
where Jsol is the phenol flux, A is the effective interfacial area, caq isthe concentration of phenol in the aqueous solution at time t, andc∗
aq is the equilibrium concentration of phenol in the aqueous phase.By making an unsteady state mass balance and assuming a con-
stant distribution ratio, Kaq can be determined from concentrationof phenol in the aqueous phase with time [12,22,23]. The equationsfor recirculating co-current flow as shown in Fig. 1 are
caq = Vc0aq
1 + V+ c0
aq
1 + Vexp(−�t) (3)
Kaq = −Qaq
(1 + Q )Amln
[1 − �
(Vaq
Qaq
)(1 + Q
1 + V
)](4)
V = Vaq
Vorg × D(5)
Q = Qaq
Qorg × D(6)
c0aq is the initial concentration in the aqueous feed solution. Qaq
and Qorg are the aqueous and organic phase flow rates, respectively.Vaq and Vorg are the volumes of the aqueous and organic phases.Am is the effective interfacial area, � is the fitting coefficient, that isa function of flowrate, volume of solvent and mass transfer coeffi-cient, and D is the distribution ratio.
In the stripping process, the transfer of phenol from the organicphase back to the aqueous phase involves the following chemicalreaction [16,19]
C6H5OH + NaOH → C6H5ONa + H2O (7)
The phenol flux for stripping can be expressed based on thepermeation model of such a stripping process as follows [8,11,16]:
Jsol = KorgA(corg − c∗org) (8)
where Jsol, mol s−1, can be calculated from the slope of the linear
phase. Korg is the overall mass transfer coefficient. A is the effec-tive area of mass transfer based on the inside diameter of hollowfiber membrane and corg is the bulk concentration of phenol in theorganic phase at time t and c∗
org is the concentration of phenol in
50 S.F. Shen et al. / Separation and Purification Technology 69 (2009) 48–56
F nd stre ores;t ial chb
ti
2c
t
slfsittcdfi
isare
bpt
wbmrotd
F
ig. 2. Schematic diagram of the phenol mass transfer process in the extraction aquilibrium distribution between the two phases; (3) diffusion in the membrane phe organic boundary layer; (6) solute diffusion in the membrane pores; (7) interfacoundary layer of the tube side.
he organic phase in equilibrium with the aqueous phase at thenterphase at the same time.
.2. Mass transfer models and estimation of overall mass transferoefficients
A schematic of the proposed permeation model for phenol massransfer in the extraction and stripping modules is shown in Fig. 2.
In the extraction process, the phenol transfer involves fourequential steps: phenol diffusion through the aqueous boundaryayer; equilibrium distribution at the aqueous–membrane inter-ace; diffusion through the membrane pores wetted by the organicolvent and diffusion through the organic boundary layer. Four sim-lar steps were assumed for the stripping process: diffusion throughhe organic phase from the bulk phase through the organic layer;ransfer through the membrane pores filled by the organics; thehemical reaction (Eq. (7)) to form the phenolate at the interface;iffusion of phenolate through the aqueous boundary layer in theber.
In this study, it is assumed that the extraction equilibrium at thenterface and the interfacial chemical reaction between phenol andodium hydroxide are both very fast [11,16,19]. Therefore, the over-ll mass transfer coefficient can be estimated using a conventionalesistance in series model in order to make a comparison with thexperimental values.
If the assumption is made that the organic phase fills the mem-rane pores, the overall mass transfer coefficient based on aqueoushase in the tube side is predicted by the local mass transfer resis-ances as follows [7,8]:
1Kaq
= 1kt
+ 1D × km
× din × ln(dout/din)dout − din
+ 1D × ks
× din
dout(9)
here kt, km and ks are the local mass transfer coefficient for theoundary layer resistance in the tube side, phenol diffusion in theembrane pores and the boundary layer resistance in the shell side,
espectively. The corresponding three items on the right hand side
f Eq. (9) are the local resistance of the tube side, membrane andhe shell side, respectively. din and dout are the inner and outsideiameters of the hollow fiber, respectively.
The fractional resistance of each step in this process, Ft, Fm ands, can be calculated. For example, Ft is defined by the following
ipping modules. (1) Diffusion in the aqueous boundary layer of the tube side; (2)(4) diffusion in the organic boundary layer of the shell side; (5) solute diffusion inemical reaction with sodium hydroxide; (8) phenolate ion diffusion in the aqueous
equation:
Ft (%) = 1/kt
1/Kaq× 100 (10)
When the aqueous phase is in the membrane pores, Eq. (11) canbe used:
1Kaq
= 1kt
× dout
din+ 1
km× dout × ln(dout/din)
dout − din+ 1
D × ks(11)
For phenol stripping, the overall mass transfer coefficient is esti-mated from the following equation
1Korg
= D
kt+ 1
km× din × ln(dout/din)
dout − din+ 1
ks× din
dout(12)
D = 1Keq[NaOH]
(13)
where Keq is the equilibrium constant of chemical reaction (Eq. (7))and D is a function of the NaOH concentration. In this study it isassumed that the value of D can be estimated by that at the initialNaOH concentration which is considered constant along the lengthof fibers.
For the laminar flow in the tube side of hollow fibers, the tubeside individual mass transfer coefficient was estimated from theLévêque correlation, a limiting case of solutions applicable to a lam-inar flow in a tube when the Graetz number (Gz) is large (Gz > 4)[7,24,25]:
Sht = ktdin
Daq= 1.62 Sc1/3 Re1/3
(din
l
)1/3
(14)
where Sh, Re and Sc are the Sherwood, Reynolds and Schmidt num-bers, respectively. Daq is the diffusivity of phenol in the aqueousphase, and l is the length of fibers. In the experimental conditionsin this study, Gz ranges between 5.0 and 11.0.
For pores filled with organic phase, the mass transfer coefficientthrough the membrane itself can be determined by the membrane
parameters of porosity (ε), tortuosity (�) and thickness (dout − din)as well as the phenol diffusivity in the organic phase (Dorg) [7]:
km = 2εDorg
�(dout − din)(15)
rification Technology 69 (2009) 48–56 51
tdm(Pt
S
wt
S
wm
3
3
TSeo
aa(
w2sNmw
masc(e
3
etTatadp
3fi
itA
Table 1Characteristics of the hollow fiber membrane.
Membranecharacteristics
Measured/determineda Supplied bymanufacturerb
Fiber/pottingmaterial
Polypropylene/polyethylene
Fiber insidediameter, Din
(�m)
310 ± 10 300
Fiber outsidediameter, Dout
(�m)
560 ± 10 670
Average porediameter (�m)
0.2 0.2
Porosity, ε 0.65 0.5
4.1. Equilibrium distribution of phenol
The partition behavior of phenol in the mixed solvent, 10% (v/v)TBP/Shellsol 2046, was investigated under a range of pH conditions
Table 2Physical properties, equilibrium parameters and module characteristics used forextraction and stripping experiments.
Module characteristics Extraction Stripping
Glass inside diameter (m) 0.0075 0.0075Number of fibers, N 60 60Effective length of fiber, L (m) 0.24 0.22Packing fraction, ˚ 0.33 0.33Interfacial area, Ain (m2) 0.01402 0.01286Distribution ratio, D 33.50 ± 2.0 0.064 ± 0.001(0.2 M NaOH)Volume of aqueous phase (L) 0.2 0.1Volume of organic phase (L) 0.1 0.1Viscosity of aqueous phasea, � (cP) 0.98 1.02
S.F. Shen et al. / Separation and Pu
For the shell side, there are many published empirical correla-ions which relate to the type of flow within the modules and theyiffer widely [7–9,21,26]. Reviews [21,26] show that the estimatedass transfer coefficients from the Costello et al. correlation (Eq.
16)) [27] are the highest while the lowest are obtained from therasad and Sirkar correlation (Eq. (17)) [9] for shell side flow parallelo the hollow fibers
hs = ksdh
Dorg= (0.53 − 0.58ϕ)Sc0.33 Re0.53 (16)
here ˚ is the packing fraction and dh is the hydraulic diameter ofhe shell side; in the range of 21 < Re < 324 and 0.32 < ϕ < 0.76
hs = ksdh
Dorg= ˇ(1 − ϕ)
dh
lSc0.33 Re0.6 (17)
here ˇ is a constant, 5.85 and 6.1 for hydrophobic and hydrophilicembranes, respectively; 0 < Re < 500 and 0.04 < ϕ < 0.4.
. Experimental
.1. Reagents and analytical method
The organic phase was prepared by diluting the extractantBP (Consolidated Chemical Company, Australia) to 10% (v/v) withhellsol 2046 kerosene (Shell Chemical Ltd., Australia) and pre-quilibrating with water. Phenol, sodium hydroxide (NaOH) andther chemicals were of analytical reagent grade.
The aqueous feed solution was prepared with deionised waternd phenol as stock solution, with the concentration of solutepproximately 1.0 kg m−3. Solutions of 0.1–1.0 M sodium hydroxideNaOH) were used as the stripping phase.
The concentration of phenol in the feed solution was determinedith a Cary 100 UV–vis spectrophotometer (Varian, Australia) at
70 nm under acidic medium. The phenol concentration in thetripping phase was measured at 287 nm after dilution with 0.1 MaOH. The concentration in the organic phase was calculated by theass balance. The errors associated with experimental data wereithin 1.0%.
The viscosity of the aqueous and the organic phases was deter-ined using a Cannon-Fenske Routine viscometer (tube size 75)
Equilibrium experiments were performed by vigorously mixingqual volumes (25 mL) of aqueous and organic phases in separa-ory funnels on a shaking device at room temperature (22 ± 1 ◦C).he pH of the aqueous solution was 6.00 ± 0.02 for extraction testsnd 12.25 ± 0.02 for stripping tests. After equilibrium between thewo phases had been reached, the aqueous phase was separatednd analyzed for phenol concentration. The distribution ratio (D) isefined as the ratio of the total phenol concentration in the organichase to that in the aqueous phase.
.3. Liquid–liquid extraction and stripping of phenol with hollowber membrane contactor
The membrane contactors used in this study were preparedn-house. Polypropylene hollow fiber membranes (Memtec, Aus-ralia) were potted into glass tubing using epoxy resins (Selleysraldite). Physical properties as well as characteristics of the mem-
Tortuosity, � 2.5 3.0
a Used in this study.b Data from Ref. [25].
brane material and the two contactors used in these experimentsare presented in Tables 1 and 2.
A schematic diagram of the experimental set-up in a co-currentrecirculation mode is shown in Fig. 1. The aqueous phase waspumped through the tube side of the hollow fiber membranes andcontinuously returned to the feed reservoir, while the organic phasewas circulated through the shell side in a co-current arrangement.Both reservoirs were stirred vigorously. Since the membrane ishydrophobic, a slight overpressure on the tube side was maintainedto stabilize the interface within the membrane pores [7].
The influence of both aqueous and organic phase flow rates onthe overall mass transfer coefficient was investigated for both phe-nol extraction and stripping. The experiments were carried out byvarying the flow rates of one phase with a specific flow rate for theother phase. In this study, the tube side Reynolds numbers rangedfrom 3.76 to 8.32 and the shell side Reynolds numbers rangedfrom 1.01 to 2.55, indicating laminar hydrodynamic conditions. Thechange of phenol concentration in the aqueous phase (feed or strip-ping reservoir) was measured with time (at 10–30 min intervals)and analyzed using a UV–vis spectrophotometer.
In the stripping experiments, the effect of NaOH concentrationwas investigated in a once-through co-current operation mode inorder to make a clear comparison. Once a steady state had beenachieved in each run, three samples were taken at 5 min internals.
4. Results and discussion
Viscosity of organic phasea, � (cP) 1.75 1.75Diffusivity of phenolb (m2 s−1) 1.051 × 10−9 1.010 × 10−9
Diffusivity of TBP–phenolb (m2 s−1) 3.492 × 10−10
a Determined experimentally at 22 ± 1 ◦C.b Diffusivity of the solutes was estimated using the Wilke–Chang equation [28].
52 S.F. Shen et al. / Separation and Purification Technology 69 (2009) 48–56
FT
abia
Iicow(snFcsfotm
4e
s
was estimated by the following equation
ig. 3. Effect of pH on phenol recovery and distribution ratio using 10% (v/v)BP/Shellsol 2046.
t room temperature as shown in Fig. 3. It was found that at pHelow 8.0 the phenol recovery was almost independent of the acid-
ty of the aqueous phase while at strongly alkaline conditions (pHbove 12.0) most of the phenol partitioned into the aqueous phase.
An extraction equilibrium isotherm at pH 6.0 is shown in Fig. 4.t can be seen that the distribution ratio is constant (D = 33.50 ± 2.0)n the concentration range studied. Therefore, Eqs. (3)–(6) are appli-able and this distribution ratio can be used directly in the analysisf the membrane contactor performance. This value is consistentith comparable studies using other extractants such as decanol
D = 24.4 ± 2.0) [11,12] and Cyanex 923 (D = 42.0) [16,17]. For phenoltripping, the distribution ratio (D) was also constant with phe-ol concentration but varied with the concentration of NaOH (seeig. 5). With decreasing concentration of NaOH, D increased signifi-antly. In this study, 0.2 M NaOH was used as the stripping solutiono the value of D used for estimation of the overall mass trans-er coefficient was 0.064 ± 0.001. This relatively low concentrationf NaOH was selected for membrane studies as higher concentra-ions can lead to higher solution viscosities. Further, a lower pH will
inimize any damage to the membrane module itself.
.2. Influence of hydrodynamics on mass transfer of phenol
xtraction with membrane contactor
Experiments were carried out in the membrane contactor totudy the influence of both phase flow rates on the overall mass
Fig. 4. Equilibrium distribution curve for phenol extraction at pH 6.0.
Fig. 5. Effect of NaOH concentration on distribution ratio in phenol stripping pro-cess.
transfer coefficient. In order to do this, the flow rate of one phasewas varied while the other one was kept constant. The aqueous andorganic phase volumes were 0.2 and 0.1 L, respectively. Extractionprofiles of phenol concentration are shown in Figs. 6 and 7.
As illustrated in Fig. 6, at a constant organic flow rate(5.16 mL min−1), the extraction rate gradually improved as theaqueous phase flow rate increased from 3.49 to 6.32 mL min−1.However, the extraction profile did not show any further changewhen the aqueous phase flow rate was increased to 7.27 mL min−1.Conversely, the extraction rate did not change with the organicphase flow rate in the wide range from 3.48 to 8.70 mL min−1, whenthe aqueous phase flow rate was kept constant, as depicted in Fig. 7.
It can be seen from these results that the aqueous phase hydro-dynamic conditions have a significant effect on mass transfer ofphenol during the extraction step. This suggests that the aqueousphase is the controlling resistance for mass transfer at the loweraqueous flow rates. This is expected for hydrophobic membraneswetted by the extractant solvent with a large distribution ratio.
The aqueous phase breakthrough pressure (P) of membrane
P = − 4� cos �
d(18)
Fig. 6. Effect of the aqueous phase flow rate on phenol mass transfer in the extractionstep. Solid lines represent concentration profiles with time by fitting the experimen-tal data to Eq. (3).
S.F. Shen et al. / Separation and Purification Technology 69 (2009) 48–56 53
Fst
wpbocmdtteimsiabiid
4
eacsametttltst
S
Ppt
ig. 7. Effect of the organic phase flow rate on phenol mass transfer in the extractiontep. Solid lines represent concentration profiles with time by fitting the experimen-al data to Eq. (3).
here � is the interfacial tension, � is the contact angle and d is theore diameter of membrane. In this work, the interfacial tensionetween the organic phase (10% TBP/Shellsol 2046) and the aque-us phenol solution was determined to be 14.0 ± 0.3 N m−1 and theontact angle between the aqueous phase and the polypropyleneembrane was determined to be 105–110◦. While the nominal pore
iameter provided by the manufacturer is 0.2 �m, SEM pictures ofhis membrane have shown that the pores have a wide distribu-ion of diameters with some around 1.0–2.0 �m [25]. Therefore, thestimated breakthrough pressure would be around 15 kPa when ds assumed to be 1.0 �m or 75 kPa if d is 0.2 �m. During the experi-
ents, about 3.0 kPa overpressure was maintained on the aqueouside at the outlet of the module. The transmembrane pressure at thenlet increased gradually with the aqueous flow rate, from 5.0 kPat 3.49 mL/min to 9.0 kPa at 7.27 mL/min. Given these transmem-rane pressures, it is expected that the organic/aqueous interface
s likely to remain at the aqueous side of the membrane pore thats the pores will be filled with the organic phase, as assumed in theevelopment of Eq. (9).
.3. Model validation and analysis in the extraction process
The overall mass transfer coefficients (Kaq) determined from thexperimental extraction profiles (Eq. (4)) are shown in Fig. 8 asfunction of the Reynolds number in both phases. These coeffi-
ients ranged from 6.0 × 10−6 to 8.0 × 10−6 m s−1, which are of theame order of magnitude as recent studies using decanol [11,12]nd Cyanex 923 [16,17] as extractant. The corresponding overallass transfer coefficients predicted by Eq. (9) are also shown. The
xperimental data are located between the predicted values usinghe Prasad and Sirkar correlation for the shell side coefficient andhe Costello et al. correlation. Indeed, the predicted overall massransfer coefficients from the Prasad and Sirkar correlation are farower than the experimental data. To obtain a better fit to the data,he parameter ˇ in the Prasad and Sirkar correlation for the shellide mass transfer coefficient (Eq. (17)) was increased from 5.85o 43.5:
hs = ksdh
Dorg= 43.5(1 − ϕ)
dh
lSc0.33 Re0.6 (19)
This result is consistent with other workers who find that therasad and Sirkar correlation (Eq. (17) with ˇ = 5.85) tends to under-redict the shell side coefficient and who also suggest increasinghe ˇ parameter [21,29]. The resulting predicted coefficients using
Fig. 8. Experimental and predicted overall mass transfer coefficients as a functionof hydrodynamics of the aqueous and organic phases in the extraction step.
the modified correlation (Eq. (19)) fit the experimental values well.The overall mass transfer coefficient increases gradually with anincrease in the aqueous Reynolds number (Reaq), however no sig-nificant changes are seen within the range of organic Reynoldsnumbers (Reorg) studied.
When the Reaq increased to around 8.0, the experimental valueof the overall mass transfer coefficient decreased significantly, sug-gesting that the aqueous phase may have penetrated the pores dueto the higher overpressure. However, if the membrane coefficient iscalculated using the aqueous phase diffusion coefficient (Eq. (11)),then the predicted value falls much lower than the experimentaldata points (Fig. 8a). These results suggest that at these higher flowrates the interface between the two phases moves from the mouthof the membrane pore due to the higher transmembrane pressure.However, the penetration into the pore itself is not deep, the mem-brane mass transfer coefficient is still better predicted by an organicphase diffusivity than an aqueous one.
As shown in Fig. 9, the fractional resistance to mass transfer fromthe tube side boundary layer was found to be in the range of 64–70%of the total resistance. This indicates that the mass transfer on thetube side is the main rate-limiting step. The fractional resistance on
the shell side is very low, less than 15%, due to the large distributionratio. As expected, the contribution of the membrane resistanceis almost independent of the hydrodynamic conditions of eitherphase. However, it contributes to about 20% of the overall resistancedue to the relatively large thickness of the membrane wall.
54 S.F. Shen et al. / Separation and Purification Technology 69 (2009) 48–56
Fo
4s
mapie
r5nwfihmact
potowa
chemical reaction rate in this process which was linearly depen-dent on the concentration of NaOH. The chemical reaction rate isnot very fast compared to mass transfer rates in the other steps.Therefore, the chemical reaction resistance at the interface con-
Fig. 11. Effect of NaOH (0.2 M) flow rate on mass transfer of phenol stripping.
ig. 9. Individual fractional resistances against the overall resistance as a functionf hydrodynamics of the aqueous and organic phases in the extraction step.
.4. Influence of hydrodynamics on mass transfer of phenoltripping with membrane contactor
Stripping experiments were carried out separately in anotherodule with the phenol-loaded organic solution on the shell side
nd 0.2 M NaOH solution on the tube side. The organic and strippinghase volumes were both 0.1 L. Stripping performance was studied
n a similar recirculating co-current manner to that used with thextraction module.
The stripping rate increased slightly with organic phase flowate when the aqueous phase flow rate was kept constant at.26 mL min−1 as shown in Fig. 10. It can be seen that the phe-ol concentration in the aqueous phase had a near-linear increaseith time during the first 150 min. Conversely, the stripping pro-les did not change as a function of the flow rate of 0.2 M sodiumydroxide (Reaq = 3.76–6.69) when the organic phase flow rate wasaintained at 8.70 mL min−1 (Fig. 11). The rate of phenol stripping
lso increased with base concentration, with the total phenol con-entration 30% higher after stripping for 150 min at 1.0 M NaOHhan at 0.2 M NaOH.
The effect of the concentration of sodium hydroxide in the strip-ing phase was also investigated in a once-through co-currentperation mode. As can be seen from Fig. 12, the phenol concen-
ration in the organic phase decreased when the concentrationf NaOH on the tube side was increased while other conditionsere kept constant. The overall mass transfer coefficient improved
bout 30% when the concentration of NaOH is increased from 0.2
Fig. 10. Effect of the organic phase flow rate on mass transfer of phenol stripping.
to 1.0 M. This observation is likely to be related to the effect of
Fig. 12. Effect of NaOH concentration on phenol stripping in a once-through co-current mode.
rification Technology 69 (2009) 48–56 55
ts
4
ipr
dtpaiSdt
stnop
ss
Fo
S.F. Shen et al. / Separation and Pu
ributes significantly to the overall mass transfer resistance duringtripping.
.5. Model validation and analysis in the stripping process
The effect of the Reynolds number of both phases on the exper-mental and predicted overall mass transfer coefficients (Korg) isresented in Fig. 13. The relative contribution of local mass transferesistance to the overall mass transfer is illustrated in Fig. 14.
As can be seen from Fig. 13, the overall mass transfer coefficientsuring the stripping step are almost one order of magnitude lowerhan those in the extraction step. The use of Eq. (11) with either theredictions from the Prasad and Sirkar correlation or the Costello etl. correlation used for the shell side coefficient did not fit the exper-mental data. The predicted values from the modified Prasad andirkar correlation (Eq. (19)) also overestimated the experimentalata by about 20%. This difference is probably due to the contribu-ion of chemical reaction resistance as discussed previously.
The predicted overall mass transfer coefficients improvedlightly with increasing Reynolds number of the organic phase onhe shell side (Fig. 13a). However, the experimental values showedo comparable trend with all data points to within ±5% of each
ther. There was no effect from the hydrodynamics of the aqueoushase on the overall mass transfer coefficient (Fig. 13b).
The contribution of individual mass transfer resistances duringtripping is shown in Fig. 14. The major resistance is due to the diffu-ion in the membrane phase and represents 60–72% of the overall
ig. 13. Experimental and predicted overall mass transfer coefficients as a functionf hydrodynamics of the aqueous and organic phases in the phenol stripping step.
Fig. 14. Individual fractional resistances against the overall resistance as a functionof hydrodynamics of the aqueous and organic phases in the phenol stripping step.
resistance. The fractional resistance in the shell side varied fromabout 40% down to about 28%. It can be clearly seen that the effectof the hydrodynamics of the aqueous phase on the tube side wasvery small. This clearly indicates that the rate-controlling step is inthe membrane phase and the organic shell side. Therefore, one ofthe effective approaches is to improve the mass transfer rate in themembrane pores. From Eq. (15), it can be deduced that increasingthe diffusion coefficient in the membrane pores or reducing the dif-fusion length (dout − din) will improve the mass transfer coefficientin these pores. The use of a hydrophilic membrane in stripping,would allow the aqueous phase to fill the membrane pores. As canbe seen in Table 2, the phenol diffusivity in the aqueous phase isabout three times faster than the complex diffusivity in the organicphase. Thus, it could be expected that the resistance of the mem-brane phase would decrease greatly if the hydrophobic membranewas replaced with a hydrophilic membrane [30]. Alternatively theuse of a thinner commercial membrane (30–50 �m membranethickness) could improve the results. The membrane used in thisstudy had a thickness greater than 120 �m.
5. Conclusions
The extraction and stripping of phenol was carried out in twoseparate hollow fiber membrane modules in co-current mode using
10% (v/v) TBP/Shellsol2046. Extraction and stripping equilibriumdistributions were determined. Experiments were performed toassess the influence of hydrodynamic conditions in the aqueoustube side and the organic shell side on the overall mass transfer
5 rificat
cm
irctassccTtcTpn
rrns
A
(iR
R
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
6 S.F. Shen et al. / Separation and Pu
oefficient for both extraction and stripping. A resistance in seriesodel for predicting performance was also outlined.
The overall mass transfer coefficients were determined exper-mentally and then compared with the predictions from theesistance in series model. The shell side mass transfer coeffi-ients were evaluated with a modified correlation. Results showedhat the overall mass transfer coefficients during extraction werebout one order of magnitude greater than those obtained duringtripping. The hydrodynamics of the aqueous phase on the tubeide had a significant effect on the overall mass transfer coeffi-ients in the extraction module. However, stripping profiles did nothange significantly under the hydrodynamic conditions studied.he mass transfer rate was enhanced when using a more concen-rated NaOH stripping solution, which indicated that the interfacialhemical reaction was influencing the overall stripping kinetics.his was consistent with the finding that the experimental strip-ing data were about 20% lower than model predictions, which didot include an interfacial reaction resistance.
The individual mass transfer resistances were evaluated and theate-controlling steps were identified. The analysis of fractionalesistance against the overall resistance indicated that the domi-ant resistance in extraction and stripping was on the aqueous tubeide and in the membrane phase, respectively.
cknowledgements
The authors would like to thank the Australian Research CouncilARC), GlaxoSmithKline Australia Ltd. (GSK) and the Particulate Flu-ds Processing Centre, a Special Research Centre of the Australianesearch Council, for financial support.
eferences
[1] C. Yang, Y. Qian, L. Zhang, J. Feng, Solvent extraction process development andon-site trial-plant for phenol removal from industrial coal-gasification wastew-ater, Chem. Eng. J. 117 (2006) 179–185.
[2] H. Jiang, Y. Fang, Y. Fu, Q.X. Guo, Studies on the extraction of phenol in wastew-ater, J. Hazard. Mater. B101 (2003) 179–190.
[3] T. Dobre, A. Guzun-Stoica, O. Floarea, Reactive extraction using sulphuric acidsalts of trioctylamine, Chem. Eng. Sci. 54 (1999) 1559–1563.
[4] P.F.M.M. Correia, J.M.R. Carvalho, Salt effects on the recovery of phenolby liquid–liquid extraction with Cyanex 923, Sep. Sci. Technol. 40 (2005)3365–3380.
[5] A. Stella, K.H. Mensforth, T. Bowser, G.W. Stevens, H.R.C. Pratt, Mass transferperformance in Karr Reciprocating plate extraction columns, Ind. Eng. Chem.
Res. 47 (2008) 3996–4007.
[6] K. Igarashi, Y. Yamada, K. Kurumada, Removal of phenol from an aqueous solu-tion using hydrogel incorporated with extractant tributyl phosphate, J. Chem.Eng. Japan 37 (2004) 1279–1283.
[7] A. Gableman, S.-T. Hwang, Hollow fiber membrane contactors, J. Membr. Sci.159 (1999) 61–106.
[
[
ion Technology 69 (2009) 48–56
[8] R. Prasad, K.K. Sirkar, Membrane-based solvent extraction, in: W.S. WinstonHo, K.K. Sirkar (Eds.), Membrane Handbook, Van Nostrand Reinhold, New York,1992, pp. 727–763.
[9] R. Prasad, K.K. Sirkar, Dispersion-free solvent extraction with microporoushollow-fiber modules, AIChE J. 34 (2) (1988) 177–188.
10] R. Prasad, K.K. Sirkar, Hollow fiber solvent extraction: performances and design,J. Membr. Sci. 50 (1990) 153–175.
[11] Z. Lazarova, S. Boyadzhieva, Treatment of phenol-containing aqueous solutionsby membrane-based solvent extraction in coupled ultrafiltration modules,Chem. Eng. J. 100 (2004) 129–138.
12] M.J. González-Munoz, S. Luque, J.R. Álvarez, J. Coca, Recovery of phenol fromaqueous solutions using hollow fiber contactors, J. Membr. Sci. 213 (2003)181–193.
13] W. Cichy, J. Szymanowski, Recovery of phenol from aqueous streams in hollowfiber modules, Environ. Sci. Technol. 213 (2002) 2088–2093.
14] W. Cichy, S. Schlosser, J. Szymanowski, Recovery of phenol with Cyanex 923in membrane extraction-stripping systems, Solv. Ext. Ion Exch. 19 (5) (2001)905–923.
15] R. Basu, R. Prasad, K.K. Sirkar, Nondispersive membrane solvent back extractionof phenol, AIChE J. 36 (3) (1990) 450–460.
16] M. Teresa, A. Reisa, O.M.F. de Freitasa, M. Rosinda, C. Ismaela, J.M.R. Carvalho,Recovery of phenol from aqueous solutions using liquid membranes withCyanex 923, J. Membr. Sci. 305 (2007) 313–324.
[17] W. Cichy, S. Schlosser, J. Szymanowski, Extraction and pertraction of phe-nol through bulk liquid membranes, J. Chem. Technol. Biotechnol. 80 (2005)189–197.
18] A. Nanoti, S.K. Ganguly, A.N. Goswani, B.S. Rawat, Removal of phenols fromwastewater using liquid membranes in a microporous hollow-fiber membraneextractor, Ind. Eng. Chem. Res. 36 (1997) 4369–4373.
19] K. Trivunac, S. Stevanovic, M. Mitrovic, Pertraction of phenol in hollow-fibermembrane contactor, Desalination 162 (2004) 93–101.
20] M.J. González-Munoz, S. Luque, J.R. Álvarez, J. Coca, Simulation of integratedextraction and stripping processes using membrane contactors, Desalination163 (2004) 1–12.
22] R.M.C. Viegas, M. Rodrııguez, S. Luque, J.R. Alvarez, I.M. Coelhoso, J.P.S.G. Cre-spo, Mass transfer correlations in membrane extraction: analysis of Wilson-plotmethodology, J. Membr. Sci. 145 (1998) 129–142.
23] Y.M. Lee, J.S. Kang, S.Y. Nam, C.H. Choi, Removal of acetic acid with amineextractants from fermentation broth using hydrophobic hollow-fiber mem-brane contactor, Sep. Sci. Technol. 36 (2001) 457–471.
24] F.X. Pierre, I. Souchon, M. Marin, Recovery of sulfur aroma compoundsusing membrane-based solvent extraction, J. Membr. Sci. 187 (2001) 239–253.
25] P.J. Harrington, The removal and recycling of heavy metals using hollow fiberliquid membranes, Ph.D. Thesis, The University of Melbourne, 1999.
26] F. Lipnizki, R.W. Field, Mass transfer performance for hollow fiber modules withshell-side axial flow: using an engineering approach to develop a framework,J. Membr. Sci. 193 (2001) 195–208.
27] M.J. Costello, A.G. Fane, P.A. Hogan, R.W. Schofield, The effect of shell side hydro-dynamics on the performance of axial hollow fiber modules, J. Membr. Sci. 80(1993) 1–11.
28] C.R. Wilke, P. Chang, Correlation of diffusion coefficients in dilute solutions,AIChE J. 1 (2) (1955) 264–270.
29] C.H. Yun, R. Prasad, A.K. Guha, K.K. Sirkar, Hollow fiber solvent extractionremoval of toxic heavy metals from aqueous waste streams, Ind. Eng. Chem.Res. 32 (1993) 1186–1195.
30] P.B. Kosaraju, K.K. Sirkar, Novel solvent-resistant hydrophilic hollow fiber mem-branes for efficient membrane solvent back extraction, J. Membr. Sci. 288 (2007)41–50.
