ARTICLE IN PRESS
Chemical Engineering Science
(
)
– www.elsevier.com/locate/ces
Performance of catalytic membrane reactor in multiphase reactions Matevž Vospernika , Albin Pintara,∗ , Gorazd Berˇciˇca , Janez Leveca , John Walmsleyb , Henrik Ræderc , Eduard Emil Iojoiud , Sylvain Miachond , Jean-Alain Dalmond a Laboratory for Catalysis and Chemical Reaction Engineering, National Institute of Chemistry, P.O. Box 660, SI-1001 Ljubljana, Slovenia b SINTEF Materials and Chemistry, Richard Birkelands vei 2B, NO-7465 Trondheim, Norway c SINTEF Materials and Chemistry, P.O. Box 124 Blindern, NO-0314 Oslo, Norway d Institut de Recherches sur la Catalyse/CNRS, 2 Avenue Albert Einstein, FR-69626 Villeurbanne Cedex, France
Received 24 February 2004; received in revised form 23 April 2004; accepted 2 July 2004
Abstract Single-channel catalytic membranes were prepared using an evaporation-crystallization Pt deposition method and characterized by employing SEM, EDX and EPMA techniques. Their activity was tested by conducting liquid-phase formic acid oxidation, and effects of trans-membrane pressure difference, catalyst loading and re-circulation rate on their performance is reported. The results obtained have revealed that the measured conversions are preferentially determined by diffusion of formic acid through the top and intermediate layers to the reaction zone on one hand, and by concentration gradient of gaseous reactant on the other hand. Which effect prevails, depends on the position of gas–liquid interface and the instantaneous molar ratio of reactants. Finally, thickness and reactants’ concentrations in the reaction zone established within the membrane wall were calculated. 䉷 2004 Elsevier Ltd. All rights reserved. Keywords: Catalytic membrane reactor; Formic acid; Laminar flow; Membrane; Multiphase reactor; Oxidation; Static mixer
1. Introduction Catalytic membrane reactor (CMR) is one of many different configurations of membrane reactors developed in recent years (Tsotsis et al., 1993). According to classification proposed by Mota et al. (2001), only the contactor-mode CMR necessitates a catalytically active membrane. A membrane can either be catalytic by itself or act as a support for catalytically active-phase deposited on its surface. By employing this reactor type for gas–liquid applications one can benefit from the unique structure of the catalytic membrane that provides a well-defined contact region between the gas- and liquid-phase flowing from the opposite sides of the membrane. As gas–liquid interface is established within the porous membrane structure, gas-phase can be supplied ∗ Corresponding author. Tel.: +386-1-47-60-282; fax: +386-1-47-60-300. E-mail address:
[email protected] (A. Pintar).
0009-2509/$ - see front matter 䉷 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2004.07.092
directly to the catalytic region (Cini and Harold, 1991), which consequently increases the concentration of gaseous reactant and enhances the overall reaction rate. This represents a key improvement with respect to conventional reactors, as gaseous reactant does not have to diffuse through the liquid film to reach the catalyst. Besides providing reduced transport limitations, this reactor type offers better temperature control, enables independent variation of gas and liquid flow rates, reactant concentration and pressure in a wide range of operating conditions. The above-mentioned advantages would make contactormode CMR ideal for conducting for instance catalytic wetair-oxidation (CWAO), which involves the oxidation of organic compounds in water. Wet-air-oxidation (WAO) is a liquid-phase process that takes place at elevated temperatures (473–593 K) and pressures (2–20 MPa) (Mishra et al., 1995). These severe conditions are somewhat reduced by the use of heterogeneous catalyst in CWAO (T = 393–473 K, P = 0.5–9 MPa). Nevertheless, the amount of energy needed
ARTICLE IN PRESS 2
M. Vospernik et al. / Chemical Engineering Science
is still quite high, and under these conditions, material selection becomes very critical as corrosion problems become substantially greater. Scrutiny of the recent literature concerning this type of membrane reactor reveals that it has been successfully implemented on a laboratory scale mainly in hydrogenation reactions. Such a system has been studied thoroughly through the model reaction of nitrobenzene hydrogenation (Peureux et al., 1993) and in particularly through the catalytic liquidphase denitrification (Daub et al., 1999; Ilinitch et al., 2000; Dittmeyer et al., 2001; Vospernik et al., 2003a). Recently, this type of reactor has been applied for WAO of formic acid (Miachon et al., 2003). These authors observed 3–6 times higher initial activity over the one obtained in conventional slurry batch reactor, and the fast deactivation of the catalytic membrane. In this paper, we focused on studying the effects of catalyst loading, gas–liquid interface position, and operating conditions on the performance of CMR. 2. Methods and materials 2.1. Ceramic membranes The oxide membrane supports used in this work, provided by Pall Exekia (Bazet, France), have tubular geometry (OD 10 mm, ID 7 mm) with a total length of 250 mm. They consisted of coarse-porous support (average pore diameter: 10 m) and two intermediate layers (all made of -Al2 O3 coated with TiO2 ), showing an average pore size decreasing from external to internal side of the tube, and a final mesoporous top layer (20 nm), made of ZrO2 , located on the inner side of ceramic membranes. The endings of all tested membranes were enamelled to insure a smoother surface on which the O-ring seals were placed. 2.1.1. Catalytic membrane preparation The evaporation-crystallization technique used for the platinum deposition in the porous framework of the ceramic membranes is described in detail elsewhere (Iojoiu et al., 2003) and will thus only be briefly summarized here. The platinum precursor used to prepare the catalytic membranes was H2 PtCl6 (Strem Chemicals, 39.85% Pt). Before impregnation, all the tubes were dried in air at 443 K for 24 h. The samples were then soaked overnight, in a vertical
(
)
–
position, within an H2 PtCl6 precursor solution. In order to provide a more homogenous contact of the hexachloroplatinic acid with the support, a mechanical stirrer (60 rpm) has been used. The samples were then kept in horizontal position at room temperature under air and rotated (60 rpm) in order to allow the solvent evaporation and a uniform distribution of the precursor. The impregnated membranes have then been dried in nitrogen flow (60 ml min−1 ) at 373 K for 1 h and calcined overnight at 473 K (heating rate 1 K min−1 ) in order to decompose the platinum precursor. The gas flux was then switched to hydrogen (60 ml min−1 ) for 6 h, in order to reduce the Pt species to metal nanoparticles. The impregnated membranes were referred to as CAT-1–CAT-4 samples (Table 1). 2.1.2. Membrane characterization Bubble-point pressure, gas permeability and pore volume of employed single-channel membranes were determined. The latter was carried out by boiling a weighed membrane sample immersed in distilled water. Nitrogen permeation measurements were performed before and after deposition and measured by employing a homemade apparatus (a detailed description of both the apparatus and measuring procedure is provided by Uzio et al., 1993), with an overpressure on the tube side and the atmospheric pressure on the shell side. No sweep gas was used and the flux was measured at the inlet of the device. In order to obtain the information about Pt distribution in prepared catalytic membranes, Electron Probe Microanalysis (EPMA) and Scanning Electron Microscope (SEM) examinations were conducted on several fractured pieces of CAT-4 membrane. Initially, the membrane was cut to lengths of ∼3 cm and then sliced longitudinally using a handsaw. For EPMA and SEM examination, sections of tube were cut and mounted in resin in cross-section orientation. In order to confirm that no artifacts are produced during sample preparation, a fractured cross-section through the tube wall was also examined in the SEM analysis. SEM examination was performed on the inner and outer tube surfaces in plan-view orientation. Before analysis, the samples were coated in a thin layer of carbon to prevent charging. EPMA was performed using a JEOL JXA-8900 instrument operating at 15 kV. SEM was performed using a Hitachi S-4300SE Field Emission Gun instrument and the accelerating voltage
Table 1 Characteristics of Pt-impregnated (-Al2 O3 –TiO2 )/ZrO2 (20 nm) ceramic membrane contactors: bubble-point pressure (BP) values and nitrogen gas permeation data Sample
Amount of Pt active phase (mg)
BP value (kPa)
Permeancea (mol m−2 s−1 Pa−1 )
CAT-1 CAT-2 CAT-3 CAT-4
3 21 31 39
330 340 220 40
29 31 24 24
a Measured after impregnation.
ARTICLE IN PRESS M. Vospernik et al. / Chemical Engineering Science
was varied between 5 and 20 kV. Higher accelerating voltages were used in order to excite the PtLa1 line at 9.44 keV, avoiding the overlap between PtMa1 line at 2.050 keV and the Zr La1 line at 2.042 keV. H2 pulse chemisorption measurements of a piece of crushed -Al2 O3 /ZrO2 (20 nm) catalytic membrane (CAT4) used in the present study were performed by means of an automated catalyst characterization system (Micromeritics, model AutoChem II 2920). The sample weight was 1.4 g. Prior to measurements the crushed membrane samples were activated at 623 K for 2 h using a flow of pure hydrogen at atmospheric pressure. Then, the samples were degassed at 623 K for 2 h under flowing Ar and cooled to 308 K, at which pulses of H2 (5 vol%)/Ar were injected into a stream of Ar flowing through the sample bed.
(
)
–
3
O-rings, which assure tight sealing and prevent gas bypass. To minimize diffusion resistances within the membrane structure, the gas-phase was supplied from the outer (shell) side, while the liquid-phase containing dissolved reactant (formic acid) was fed through the membrane channel. The catalyst was deposited primarily on the membrane filtration layers. The performance of impregnated ceramic membranes was examined by conducting catalytic liquid-phase oxidation of aqueous solutions of formic acid. During the experimentation, the liquid-phase was contained in a stirred and heated vessel and re-circulated through the reactor by a gear pump. The temperature of the liquid-phase was successfully controlled within ±0.2 K of the set value by employing the PID control. An electronic mass flow controller (Brooks, model 5850E) was used for adjusting the gas flow. Pressure was measured by fine pressure transducer (Baratron MKS). Trans-membrane pressure difference (TMP) was controlled by a fuzzy logic controller, which regulates back pressure according to a signal from the pressure transducer by adjusting the current through the outlet solenoid valve (Aalborg, model PSV2). Formic acid (2.0 g l−1 ) was used as a model solution in this study. Experiments were conducted at T = 308 K, and
2.2. Experimental setup The catalytic properties of prepared membranes were tested in a batch-recycle laboratory-scale membrane reactor illustrated in Fig. 1. The reactor setup consists of a glass housing in which a ceramic membrane contactor is placed, liquid and gas delivery systems, and a data acquisition system. The tubular ceramic membrane was sealed by Viton
2
8 P
N2 3
6
4
7
3
15
1
16 2
14
5
6
10
T 13
O2
3
P
12 9
11
DAS A /D 17 2
H2
18 1
Fig. 1. Schematic drawing of a membrane reactor system. 1, gas cylinder; 2, gas regulator; 3, multi-port valve; 4, flow controller; 5, membrane reactor; 6, pressure sensor; 7, back-pressure regulator; 8, valve; 9, continuously stirred tank (V = 450 ml); 10, venting line; 11, pump; 12, heat exchanger; 13, temperature sensor; 14, sampling; 15, graduated pipette; 16, feed tank (V = 250 ml); 17, data acquisition system; 18, PC computer.
ARTICLE IN PRESS 4
M. Vospernik et al. / Chemical Engineering Science
TMP set to 35, 100, 200 or 300 kPa. The liquid-phase was pumped at atmospheric pressure from a reservoir of 450 ml through the reactor at different flow rates ranging from 180 to 1200 ml min−1 (Reynolds numbers between 660 and 4860). Several runs were also performed in the presence of Kenics-type static mixer (OD 6.35 mm) inserted in the inner tube of a membrane reactor and used for inducing additional turbulence (Khinast et al., 2003). Until the desired operating conditions, i.e., pressure and temperature were reached, the system operated under nitrogen. The run was started by replacing nitrogen flow with the oxygen flow. Conversion of formic acid was followed by repetitive sampling and total organic carbon (TOC) analysis (Rosemount/Dohrmann TOC analyzer, model DC-190). In parallel to formic acid oxidation, some experiments of nitrite reduction were performed in the presence of Pd-containing -Al2 O3 /ZrO2 (20 nm) membrane contactor.
3. Results and discussion 3.1. Membrane characterization Results of bubble-point pressure values (BP) along with results of permeation measurements of ceramic membranes
(
)
–
used in this study are reported in Table 1. During permeance measurements, no significant difference has been detected for membrane samples before and after impregnation, so it can be assumed that the structure of the membrane was not modified during the impregnation. This was verified by liquid-displacement tests (Vospernik et al., 2003a). Pore volume of all tested membranes was found to be approximately 4.0 ml (overall porosity ∼0.4). The results of H2 pulse chemisorption analysis show that the dispersion of Pt ensembles in tested CAT-4 membrane equals to 5.4%, with the average active particle diameter of about 21 nm (assuming hemispherical shaped Pt clusters), and the metallic surface area of 0.018 m2 g−1 sample . Low values of Pt dispersion might be attributed either to low BET surface area of employed ceramic membranes (being in the range of 1 m2 g−1 ) or to the applied preparation technique of catalytic membrane contactors. The mapping and composition line-scans show that Pt is concentrated mostly within the nanoporous zirconia inner layer (Fig. 2a). EPMA measured a normalized Pt level of between 4 and 8 at% in the zirconia layer. It should be noted that the layer is quite thin (approx. 3 m) compared with the EPMA spot size (∼1 m), which may lead to some variability in the maximum value measured in the profiles. Contrast in X-ray maps was uniform suggesting that
Fig. 2. SEM-BSE cross-section images of the fractured Pt-impregnated CAT-4 ceramic membrane showing: (a) all four layers and average EPMA composition of three line-scans; (b) intermediate -Al2 O3 /TiO2 layers and top ZrO2 layer.
ARTICLE IN PRESS M. Vospernik et al. / Chemical Engineering Science
qualitatively the distribution of Pt within the zirconia layer was uniform around the circumference of the layer. This was also confirmed by AES analysis. Further, the X-ray maps suggest a distribution of Pt within the first TiO2 modified alumina layer. The Backscatter Electron (BSE) images, which are highly sensitive to atomic number, show this in more detail (Fig. 2b). Areas of light contrast in the BSE images showed the presence of Pt when analyzed by EDS. The particle size in these areas is quite variable and particles that can be resolved individually have a size from tens of nm to 100 nm or so. A low concentration of isolated, larger Pt particles was present in the second, coarser, TiO2 modified alumina layer. The density of these particles is not high enough to show contrast in X-ray maps but the individual particles are seen clearly in BSE images (Fig. 2b). The overall level of Pt in the two TiO2 modified alumina layers was lower than in the inner ZrO2 layer and higher in the first layer than the second. The bulk of the alumina substrate was found to contain occasional large Pt-containing particles when observed in cross-section orientation. At the outside of the tube, the Pt level was found to increase. Examination of a plan-view orientation of the sample in BSE mode showed a coarse distribution of Pt along grain boundary grooves and triple points. Some loose debris, probably from the cutting process, is also present at the surface.
3.2. Catalytic membrane reactor performance The positive effect of elevated TMP on the performance of ceramic membrane reactors has already been demonstrated (Vospernik et al., 2003a). Results obtained in denitrification runs showed that the position of gas–liquid interface within the membrane wall is the most important parameter concerning the employability of ceramic membrane reactors. The nitrite disappearance rate is governed by external mass-transfer limitations and diffusion of nitrite ions from the bulk liquid-phase through the top (filtration) layer into the intermediate layers, where reaction takes place. To obtain further insight in the behavior of ceramic membrane reactors, additional experiments of liquid-phase oxidation of aqueous HCOOH were conducted. The results obtained during the catalytic oxidation of this compound over different Pt/ceramic membranes are reported in Fig. 3 and Table 2. Typical reaction profiles of temporal course of formic acid concentration obtained during oxidation runs conducted at T = 308 K in laminar flow regime (vol.,L = 360 ml min−1 , Re = 1320) with TMP set to 35 kPa are illustrated in Fig. 3a. This TMP was chosen according to cognition obtained in our previous work (Vospernik et al., 2003a), where it is suggested as optimal for the applied ceramic membranes (nevertheless, one must be aware of the fact that with such a low TMP applied in the experiment, the gas–liquid interface can only be shifted within the coarseporous support of the membrane, and TMPs higher than 100 kPa would be needed to shift the gas–liquid interface
(
)
–
5
to the intermediate layers). The obtained results can be well represented by means of apparent pseudo-first-order kinetics with respect to formic acid concentration and show that the amount of Pt loading affects the observed reaction rate. As expected, catalytic oxidation of formic acid conducted by employing a membrane with the lowest amount of Pt active sites (CAT-1) is the slowest, while for membranes with higher Pt loadings (CAT-2–CAT-4) the results show interestingly that the quantity of active phase deposited on ceramic membrane is only slightly affecting the observed reaction rate. This might be attributed to similar concentration of discrete Pt particles in the coarse-porous support, over which reaction takes place. Considering the fact that at the given operating conditions (i.e., low TMP) the gas–liquid interface and the reaction zone were located within the coarse support (resulting in longer diffusion path of the liquidphase reactant to active sites), the Pt-impregnated ceramic membranes exhibit a promising activity for efficient treatment of model HCOOH solutions, as conversions in the range from 31% to 45% were attained after 480 min of operation. In Figs. 3b–d the experimental results measured at higher TMPs are shown. Accordingly to liquid displacement measurements (Vospernik et al., 2003a), at such TMPs the liquid is already completely displaced from the coarse-porous support and the gas–liquid interface shifts towards intermediate layers. As observed, the catalytic activity increases with the Pt loading and TMP, i.e., oxygen partial pressure. Membranes CAT-3 and CAT-4 exhibit low bubble-point pressure values, so they were not used in the whole range of TMPs applied between the gas and liquid side of membrane. It should also be stressed that stable operation of the reactor unit (i.e., volume of displaced liquid) was obtained and no gas leaks were observed during the runs conducted at elevated TMPs. Finally, the results depicted in Fig. 3 are not influenced by catalyst deactivation, as identical concentration-time profiles were observed in repetitive oxidation runs. The effect of TMP on the membrane reactor performance is for membrane CAT-2 shown in Fig. 4a. The results presented demonstrate a positive effect of TMP on the activity of the membrane reactor. The global reaction rate constants (k) were calculated for experiments carried out at different TMPs by fitting experimental data by the first-order decay model. The obtained values of apparent reaction rate constants are summarized in Fig. 4a. Fig. 4b presents an influence of oxygen overpressure on calculated rate constants. It can be seen that an almost linear dependency was obtained, since the global reaction rate constant increases linearly (within the experimental error) with the applied oxygen overpressure, indicating that mass-transfer of oxygen within the porous membrane structure is a major factor determining the overall reaction rate. A deviation of experimental point tot. from the best-fit curve at Poxygen = 1.35 kPa is attributed to incomplete displacement of liquid from the coarse membrane support, and correspondingly longer diffusion path of formic acid (approx. 320 m).
ARTICLE IN PRESS 6
M. Vospernik et al. / Chemical Engineering Science
1.0
(
)
–
1.0 TMP: 100 kPa
0.8
c(HCOOH)/c(HCOOH)0, /
c(HCOOH)/c(HCOOH)0, /
TMP: 35 kPa
0.6
0.4 CAT-1 CAT-2 CAT-3 CAT-4 first-order fit
0.2
0.0 (a)
0
60
120
180
240
300
360
420
480
0.8
0.6
0.4
0.0 (b)
1.0
CAT-1 CAT-2 CAT-3 first-order fit
0.2
0
60
120
180
240
300
480
TMP: 300 kPa
0.8
c(HCOOH)/c(HCOOH)0, /
c(HCOOH)/c(HCOOH)0, /
420
1.0 TMP: 200 kPa
0.6
0.4 CAT-1 CAT-2 CAT-3 first-order fit
0.2
0.0
360
0.8
0.6
0.4
0.0
0
60
120
(c)
180 240 time, min
300
360
420
CAT-1 CAT-2 first-order fit
0.2
0
60
(d)
120
180 240 time, min
300
360
420
Fig. 3. Relative formic acid concentration as a function of time obtained with different catalytic membranes at T = 308 K, vol.,L = 360 ml min−1 and different trans-membrane pressure differences (TMP): (a) 35 kPa, (b) 100 kPa, (c) 200 kPa, and (d) 300 kPa.
Table 2 Global reaction rate constants (k) and initial reaction rates (−rHCOOH )0 obtained during the oxidation of aqueous formic acid solutions carried out with different catalytic membranes in the CMR reactor at T = 308 K and various trans-membrane pressure differences (TMP) Sample
TMP (kPa)
103 × k(min−1 )
k/mPt (min−1 g −1 Pt )
(−rHCOOH )0 , mmol s−1 g −1 Pt
CAT-1
35 100 200 300 35 100 200 300 35 100 200 35
0.74 1.4 1.9 2.0 1.2 2.4 4.1 4.7 1.1 3.2 5.7 1.1
0.25 0.47 0.63 0.67 0.06 0.11 0.19 0.22 0.04 0.10 0.18 0.03
0.08 0.15 0.21 0.22 0.02 0.04 0.06 0.07 0.01 0.03 0.06 0.009
CAT-2
CAT-3
CAT-4
ARTICLE IN PRESS M. Vospernik et al. / Chemical Engineering Science
0.8
–
7
liquid completely displaced from the coarse-porous support (diffusion path: ~50 µm)
0.0045
0.6
k, min-1
c(HCOOH)/c(HCOOH)0, / (a)
)
0.0060
1.0
0.4 TMP [kPa]: 35 100
0.2
0.0
(
-5%
0.0030
-1
k [min ]: -3 1.2 × 10
0.0015
-3
2.4 × 10
-3
200
4.1 × 10
300
4.7 × 10
60
120
exp. best fit
-3
0.0000
0
+5%
180 240 time, min
300
360
420
0
50
(b)
100 150 200 250 tot. Poxygen, kPa
300 350
400
Fig. 4. (a) Relative formic acid concentration vs. time dependencies obtained in the presence of CAT-2 membrane at different TMP values; (b) global reaction rate constant as a function of oxygen overpressure. T = 308 K, vol.,L = 360 ml min−1 .
φvol.,L (mL min-1)
1.0
c(NO2)/c(NO2)0, /
0.8
-
TMP (kPa)
Vdispl.(mL)
static mixer
35 35 35 45 45 45
3.3 3.3 3.2 3.3 3.4 3.4
+ +
90 180 360 360 360 360
0.6
first-order fit
-
A summary of the reactivity of all tested catalytic membranes is shown in Table 2. A comparison of k values obtained with different catalytic membranes at the same TMPs shows that higher Pt loadings contribute to higher overall reaction rates of HCOOH oxidation, which means that besides mass-transfer limitations global reaction rate constant is influenced also by chemical resistances. However, it can be seen from the listed k/mPt ratios and (−rHCOOH )0 values that this increase is not proportional, which suggests that only part of the total deposited Pt efficiently participated in the performed oxidation runs. Furthermore, based on the fact that the values of k/mPt ratios in the case of CAT2 and CAT-3 membrane contactors (higher catalyst loadings of 21 and 31 mg, respectively) are almost identical, one might tentatively speculate that in the transition region from the coarse-porous support to the first intermediate layer (Fig. 2a), i.e., where gas–liquid interface and reaction zone are located at TMP 100 kPa, the applied Pt deposition technique results in the proportional increase of the concentration of Pt particles. To conclude, in the given range of operating conditions, in which mass-transfer limitations prevail and the reaction zone is very narrow (see the discussion below), it is advantageous to use catalytic membrane contactors with lower catalyst loading providing that the deposition of active phase is limited to intermediate and top layers. Nitrite hydrogenation runs carried out in the presence of Pd-doped -Al2 O3 /ZrO2 (20 nm) catalytic membrane contactor at various re-circulation rates of liquid-phase and the use of a static mixer inserted in the inner membrane tube, have revealed that external mass-transfer limitations might play a significant role concerning the practical use of this reactor type (see Fig. 5). Correspondingly, experiments were conducted in order to ascertain possible the influence of external mass-transfer limitations on liquid-phase HCOOH oxidation. On the contrary to expectations, results presented in
0.4
0.2
0.0
0
60
120
180
240
300
360
time, min
Fig. 5. Nitrite concentration vs. time curves obtained in the single-channel Pd(0.1 wt%)-doped catalytic membrane contactor during the course of liquid-phase nitrite reduction conducted at different trans-membrane pressure differences. T = 303 K, pH = 5.8 (const.).
Fig. 6 show that the observed formic acid disappearance rate is not affected by the re-circulation rate (Re = 660–4860), or by insertion of a static mixer in the inner tube, which implies that in the given range of operating conditions liquidto-solid mass-transfer resistance does not influence overall reaction rate. This might be attributed to the fact that the concentration of formic acid is in great excess in comparison to the concentration of oxygen, while in the case of catalytic nitrite reduction the concentrations of reactants were comparable. 3.3. Determination of the thickness of reaction zone To obtain an insight into the situation in the membrane wall during the reaction, the simplified approach based on
ARTICLE IN PRESS 8
M. Vospernik et al. / Chemical Engineering Science
c(HCOOH)/c(HCOOH)0 , /
T: 308 K TMP: 200 kPa
–
xg = (exp(−171.2542 + 8391.24/T pO2 + 23.24323 ln(T ))) · . (6) 1bar By inserting diffusion path of formic acid, LHCOOH , reduced for the calculated thickness of the reaction zone, into modified Eq. (4), instantaneous concentrations of formic acid in the reaction zone can be calculated. dCHCOOH,bulk dt DHCOOH CHCOOH,bulk − CHCOOH,reaction zone = εA . (7) LHCOOH − l
0.8
0.6
0.4 -1
φ vol.,L: 180 mL min
-1
φ vol.,L: 1200 mL min
0.2
-1
φ vol.,L: 1200 mL min + static mixer first-order fit
0
100
200 time, min
300
400
Fig. 6. Influence of the re-circulation rate and the use of a static mixer on formic acid oxidation in the presence of CAT-2 membrane.
mass fluxes was used for the estimation of reaction zone thickness as well as the diffusion path of formic acid. Oxidation of formic acid can be represented by the following reaction: HCOOH + 21 O2
Pt/ceramic membrane
−→
CO2 + H2 O.
(1)
By implementing fluxes of reactants, one obtains: N(HCOOH) = 21 N(O2 ),
(2)
where each of fluxes is given by the modified Fick’s law: N=
)
Fogg and Ferrard (1990):
1.0
0.0
(
C D εA . l
(3)
Combining Eqs. (2) and (3) gives: CO∗ − CO2,reaction DO2 dCHCOOH,bulk = εA 2 dt 2 l
zone
.
(4)
By employing Eq. (4), thickness of the reaction zone (l) can be calculated. Calculations are based on previously shown concentration-time profiles (Fig. 3) presenting disappearance of formic acid as a function of time, and under the following assumptions: (i) oxygen is the rate-limiting reactant; (ii) reaction on the catalyst surface is proportional to oxygen flux. Tortuosity factor () and porosity (ε) of the employed membranes were obtained experimentally and equal to 2.5 and 0.4, respectively (Vospernik et al., 2003b). Diffusion coefficient of oxygen was calculated by means of Wilke–Chang correlation (Reid et al., 1987): DAB =
7.4 × 10−8 (B · MB )0.5 T , B VA0.6
(5)
whereas equilibrium concentration of oxygen in the liquidphase was calculated using a correlation proposed by
Diffusion coefficient of formic acid was estimated by means of Hayduk–Minhas equation (Reid et al., 1987):
DAB = 1.25 × 10−8 (VA−0.19 − 0.292)T 1.52 B , 9.58 − 1.12. = VA
(8)
Molar volume (VA ) was estimated using a correlation proposed by Tyn and Calus (Reid et al., 1987): VA = 0.285Vc1.048 .
(9)
Calculations were performed for the CAT-1 membrane operating at different trans-membrane pressure differences. Results presented in Fig. 7 confirm that the disappearance rate of formic acid is influenced by a combined mass-transfer of both reactants (formic acid and oxygen) to the reaction zone. For example, Fig. 7a shows the calculated thickness of the reaction zone as a function of formic acid conversion for experiments conducted at different TMPs. One can see that the thickness of the reaction zone is in the range of some microns, indicating that it is located very close to the gas–liquid interface within the membrane wall. Calculations further show that there exists a sharp oxygen concentration gradient, which means that O2 is completely depleted, while the concentration of formic acid in the reaction zone decreases and drops to about 77% (at TMP = 100 kPa) or 72% (at TMP = 200 kPa) with respect to its concentration in bulk liquid-phase (Fig. 7b). When conducting oxidation at TMP set to 35 kPa (which means that lower amount of liquid is displaced from the membrane wall), concentration of formic acid in the reaction zone considerably decreases and approaches to 0 (Fig. 7b). 4. Conclusions The results indicate that external and/or internal masstransfer resistances considerably influence membrane reactor performance. For the reaction studied, in which concentration of reactant in the liquid-phase is appreciably higher with respect to equilibrium concentration of gaseous reactant, the reaction rate is predominantly determined by internal mass-transfer resistances, i.e., diffusion of formic acid
ARTICLE IN PRESS M. Vospernik et al. / Chemical Engineering Science 16
cHCOOH,reaction zone / cHCOOH,bulk, /
∆l, µm 8 TMP [kPa]: 35 100 200
(a)
)
–
9
0.8
12
4 0.0
(
0.1
0.2
0.3
0.4
0.5
0.4
TMP [kPa]: 35 100 200
0.2
0.0 0.0
0.6
XHCOOH , /
0.6
0.1
(b)
0.2
0.3 XHCOOH , /
0.4
0.5
0.6
Fig. 7. Calculated thickness of the reaction zone as a function of formic acid conversion (a) and formic acid concentration in the reaction zone as a function of formic acid conversion (b) for CAT-1 membrane at T = 308 K and different TMP values: TMP = 35 kPa, k = 7.4 × 10−4 min−1 , diffusion path of the liquid-phase reactant ∼320 m; TMP = 100 kPa, k = 1.4 × 10−3 min−1 , diffusion path of the liquid-phase reactant ∼50 m; TMP = 200 kPa, k = 1.9 × 10−3 min−1 , diffusion path of the liquid-phase reactant ∼50 m.
through the top and intermediate layers to the reaction zone, and by the amount of oxygen present in the reaction zone. The experiments performed on catalytic oxidation of aqueous solutions of formic acid and subsequent calculations concerning the thickness of reaction zone (being in the range of some microns) show that only part of deposited Pt takes place in the reaction. With this respect, the amount of Pt active phase deposited in the intermediate and top layers is still to be optimized.
Subscripts
Notation
Acknowledgements
A C D l N VA VC xg
membrane surface area, cm2 concentration in liquid-phase, mol l−1 diffusion coefficient, m2 s−1 thickness of the reaction zone, m flux, mol s−1 molar volume at normal boiling temperature, m3 mol−1 critical volume, m3 mol−1 mole fraction solubility, dimensionless
Greek letters ε
porosity, dimensionless factor, dimensionless viscosity, mPas density, kg m−3 torturosity factor, dimensionless association factor (2.26 for water), dimensionless
A B
solute solvent
Superscript ∗
equilibrium
The authors express their acknowledgement to the European Commission, who founded this work through the WATERCATOX Project, contract No. EVK1-CT-2000-00073, and the Slovenian Ministry of Education, Science and Sport (program No. PO-0521-0104). The authors are grateful to Pall Exekia (Bazet, France), an industrial partner involved in the WATERCATOX Project, for supplying ceramic membranes used in the present study. References Cini, P., Harold, M.P., 1991. Experimental study of the tubular multiphase catalyst. A.I.Ch.E. Journal 37 (7), 997–1008. Daub, K., Emig, G., Chollier, M.-J., Callant, M., Dittmeyer, R., 1999. Studies on the use of catalytic membranes for reduction of nitrate in drinking water. Chemical Engineering Science 54, 1577–1582. Dittmeyer, R., Höllein, V., Daub, K., 2001. Membrane reactors for hydrogenation and dehydrogenation processes based on supported palladium. Journal of Molecular Catalysis A: Chemical 173, 135–184. Fogg, P.G.T., Ferrard, W., 1990. Solubility of Gases in Liquids. Wiley, Chichester, pp. 293–294. Ilinitch, O.M., Cuperus, F.P., Nosova, L.V., Gribov, E.N., 2000. Catalytic membrane in reduction of aqueous nitrates: operational principles and catalytic performance. Catalysis Today 56, 137–145.
ARTICLE IN PRESS 10
M. Vospernik et al. / Chemical Engineering Science
Iojoiu, E.E., Walmsley, J., Raeder, H., Bredesen, R., Miachon, S., Dalmon, J.-A., 2003. Comparison of different support types for preparation of nanostructured catalytic membranes. Reviews on Advanced Materials Science 5, 160–165. Khinast, G.J., Bauer, A., Bolz, D., Panarello, A., 2003. Mass-transfer enhancement by static mixers in a wall-coated catalytic reactor. Chemical Engineering Science 58, 1063–1070. Miachon, S., Perez, V., Crehan, G., Torp, E., Raeder, H., Bredesen, R., Dalmon, J.-A., 2003. Comparison of a contactor catalytic membrane reactor with a conventional reactor: example of wet air oxidation. Catalysis Today 82, 75–81. Mishra, V.S., Mahajani, V.V., Joshi, J.B., 1995. Wet air oxidation. Industrial and Engineering Chemistry Research 34, 2–48. Mota, S., Miachon, S., Volta, J.-C., Dalmon, J.-A., 2001. Membrane reactor for selective oxidation of butane to maleic anhydride. Catalysis Today 67, 169–176. Peureux, J., Torres, A., Mozzanega, H., Giroir-Fendler, A., Dalmon, J.-A., 1993. Nitrobenzene liquid-phase hydrogenation in a membrane reactor. Catalysis Today 25, 409–415.
(
)
–
Reid, R.C., Prausnitz, J.M., Poling, B.E., 1987. The Properties of Gases and Liquids, fourth ed. McGraw-Hill, New York, pp. 53–54. Tsotsis, T.T., Minet, R.G., Champagnie, A.M., Liu, P.K.T., 1993. Catalytic membrane reactors. In: Becker, E.R., Pereira, C.J. (Eds.), ComputerAided Design of Catalysts. Marcel Dekker, New York, pp. 471–551. Uzio, D., Peureux, J., Torres, A., Giroir-Fendler, A., Ramsay, J., Dalmon, J.-A., 1993. Platinum/-Al2 O3 catalytic membrane—preparation, morphological and catalytical characterisation. Applied Catalysis 96, 83–93. Vospernik, M., Pintar, A., Berˇciˇc, G., Levec, J., 2003a. Experimental verification of ceramic membrane potentials for supporting three-phase catalytic reactions. Journal of Membrane Science 223, 157–169. Vospernik, M., Pintar, A., Berˇciˇc, G., Levec, J., 2003b. Mass transfer studies in gas–liquid–solid membrane contactors. Catalysis Today 79–80, 169–179.
