The Fouling Mechanism of Ceramic Membranes Used for Recovering TS-1 Catalysts*

ZHONG Zhaoxiang (仲兆祥), LI Dongyan (李冬燕), LIU Xin (劉馨), XING Weihong (邢衛(wèi)紅),** and XU Nanping (徐南平)

?

The Fouling Mechanism of Ceramic Membranes Used for Recovering TS-1 Catalysts*

ZHONG Zhaoxiang (仲兆祥)1, LI Dongyan (李冬燕)2, LIU Xin (劉馨)1, XING Weihong (邢衛(wèi)紅)1,** and XU Nanping (徐南平)1

1State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemistry and Chemical Engineering, Nanjing University of Technology, Nanjing 210009, China2Chemical Engineering Department, Nanjing College of Chemical Technology, Nanjing 210048, China

Ceramic ultrafiltration membranes were used to separate titanium silicalite-1 (TS-1) catalysts from the slurry of catalytic ammoximation of cyclohexanone to oxime. Silica was shown to have a great effect on membrane fouling in the alkaline environment of this system. In the ammoximation system, there are three main silica sources, which are residual silica on the catalyst particles surface during preparation, silica dissolved from TS-1 catalyst particles by ammonia solvent, and silica sol added into the reaction slurry to inhibit the dissolution erosion of the TS-1 catalyst. The silica dissolved by ammonia has been proved to influence membrane fouling most among the three silica sources. This was because the amount of silica dissolved by ammonia was the largest, and the polymerization of silica monomers at high concentration caused colloid particles formation, which led to a dense cake layer depositing on the membrane surface. Meanwhile, the size reduction of catalyst particles caused by alkaline dissolution also increased specific resistances of cake layers.

TS-1 catalyst, ceramic membrane, membrane fouling

1 Introduction

Cyclohexanone oxime is a primary intermediate in producing-caprolactam，which is an important chemical [1]. In 1983, Taramasso. [2] first synthesized titanium silicates successfully, and then, the Italian company Enichem developed a new method of catalytic direct ammoximation of cyclohexanone with NH3/H2O2to the oxime on titanium silicalite-1 (TS-1) catalysts [3]. However, the activity of TS-1 catalyst decreased much after a long period of operation [4]. Petrini. [5-7] proved that the dissolution erosion of silicon and titanium from the TS-1 catalyst caused by ammonia solvent was the primary factor leading to deactivation of the catalyst. So in production, besides the reactants and chemical solvents, a certain content of silica sol additive was also added into the reaction system to inhibit the dissolution of TS-1 molecular sieve in the alkaline environment [8].

A major difficulty in the separation of TS-1 particles from reaction slurry arises because TS-1 particles are too fine to be removed by gravity settling and porous tube filtration, which are often used to separate catalysts in industrial application [9]. Microfiltration (MF) and ultrafiltration (UF) have emerged as useful processes for separation of fine particles, microorganisms, and emulsion droplets. Ceramic ultrafiltration membrane was used to recover the ultra-fine catalyst particles [10]. However, the membranes tend to be fouled during the separation of reaction slurry. For the system of cyclohexanone ammoximation, the fouling materials have been determined to include organic matter, fine particles, impurities, and the added silica sol mentioned above. And the mechanism of membrane fouling could be the formation of dense cake on the membrane surface, adsorption of silica additive, pore plugging,[10]. The silica has been found to have a great effect on membrane fouling in water treatment processes [11-13]. The type and extent of silica fouling depends on the conditions of system, such as concentration, pH of feed solution, temperature, and presence of some divalent captions,[14-17].

Our previous works have suggested that silica played an important role in membrane fouling in the production of cyclohexanone oxime [10]. However, what and how silica acts in this system are still not very clear. In this study, the effects of different silica sources and solution environment on the membrane fouling were investigated in detail.

2 Experimental

2.1 Materials

The tested stock solutions contained water, ammonia (Shanghai Chemical Reagents Co. Ltd., China), and silica sol additive (Yueyang Wenli Industrial Co. Ltd., China). The concentration of TS-1 catalysts (Research Institute of Petroleum Processing, China) was 30 g·L－1.

The ultrafiltration membrane used in the experiment was a multi-channel tubular ceramic membrane (0.5 m long, 19 channels of 4.0 mm inner diameter) supplied by Nanjing Jiusi High-Tech Co. Ltd., China. Its nominal pore size was 0.05 μm and the filtration area was 0.12 m2. The membrane was composed of a top layer of ZrO2and a support layer of α-Al2O3.

2.2 Experimental methods and procedures

The experimental setup was constructed by stainless steel and comprised a recirculation loop (Fig. 1). The recirculation loop was composed of a 20 L reaction tank (and jacketed for temperature control), two flowmeters, a centrifugal pump, a membrane module, and the accessories of pressure gauges, valves, and pipe.

Figure 1 Schematic diagram of the crossflow filtration equipment

1—feed tank; 2—sampling port; 3—centrifugal pump; 4,6—rotameter; 5—membrane module; P1, P2—pressure gauges; V1-V6—valves

The crossflow filtration was run at a constant temperature of 353 K, crossflow velocity of 3 m·s－1, and trans-membrane pressure of 0.1 MPa. The feed suspension was maintained at a constant volume by recycling the permeation back into the reaction tank. After each run, the tank was emptied, the system was thoroughly rinsed with deionized water to remove residual process solution, the membrane was cleaned by circulating 1% (mass concentration) NaOH (Shanghai Chemical Reagents Co. Ltd, China) solution and 1% (by volume) nitric acid (Shanghai Chemical Reagents Co. Ltd, China) solution at 353 K for several hours with the permeate line open. The apparatus was then rinsed with deionized water until the pH returned to 7. In order to ensure that the experiments had good reproducibility, the pure water flux (PWF) was measured using deionized water after every cleaning operation. The PWF measurement showed that the membrane could be fully restored.

Adsorption test was conducted by placing the TS-1 particles in silica sol solution. The initial silica concentration was 50 mg·L－1. The TS-1 particles concentration added to the solution was 30 g·L－1[10]. The solution was stirred to prevent precipitation and the temperature was kept 353 K. The solution samples were centrifuged to remove TS-1 particles, and then, analyzed by inductively coupled plasma (ICP) (Optima2000 DV, Perkin Elmer, USA) to determine the concentration of silica.

3 Results and Discussion

3.1 Effect of silica on membrane fouling

Adsorption measurements showed that silica concentration decreased quickly because of the adsorption of silica to TS-1 particles (Fig. 2). It is well known that molecular sieves are often used as adsorbents in a variety of industrial applications because of their large surface area [19], meanwhile silica sol is a polymer in the colloidal state and has high surface energy and bonding strength [20]. So, when the silica sol was added into the TS-1 suspension, the colloidal silica would be readily adsorbed on the surfaces of TS-1 particles present in suspension or cake layer. For the membrane filtration of the silica-contained solution, therefore, the pore size of the cake layer decrease gradually because of the adsorbed silica on the TS-1 particles embedded in the cake layer, which essentially lead to a low permeability. Fig. 3 shows SEM pictures taken from the fresh membrane and fouled membrane. It seems that a dense cake layer is formed on the fouled membrane surface.

Figure 2 Adsorption of silica sol by TS-1. time

3.2 Effect of silica sources on the filtration

Figure 3 SEM pictures of membrane surface

Besides the addition of silica sol added into the reaction slurry to inhibit the dissolution erosion of the TS-1 catalyst, there are two other silica sources in this system: residual silica on the fresh catalyst particles surface during preparation [21], and silica dissolved from TS-1 catalyst particles by ammonia solvents. The residual silica on the catalyst particles surface can be removed by nitric acid. The filtration experiments of catalyst particles with or without pre-treatment by nitric acid have been performed, and the results were shown in Fig. 4. As shown in Fig. 4, the flux of filtrating solution containing fresh catalysts was lower than that of the solution containing pretreated catalysts. During the filtration, the residual silica on the catalyst particles surface became the binder of particles and formed dense cake layer [22, 23]. Therefore, it is necessary to remove residual silica on the particles before being applied for production.

Figure 4 Flux. operation time with or without treatment by nitric acid [3% (by mass), 0.1 MPa, 3 m·s－1, 80°C]

○?TS-1 treated by nitric acid;□?TS-1 untreated by nitric acid

Three different suspensions of pre-treated TS-1 particles in water, ammonia solution, and silica sol solution were filtrated respectively to investigate the effect of different silica sources on membrane fouling. The plots of flux ratio. operation time under different solution conditions are shown in Fig. 5. The results showed that the membrane flux of TS-1 only declined relatively slowly compared with the other two cases. The stable flux was found to be approximately 91% of the initial flux. And the membrane fouling in ammonia and silica sol solutions was more serious than that in water. For example, the membrane flux under ammonia condition after filtration for 8 h was only approximately 60% of the initial flux.

Figure 5 Variation of flux of different solution conditions [3% (by mass), 0.1 MPa, 3 m·s－1, 80°C]

○?TS-1 only;□?TS-1+silica sol;△?TS-1+ammonia

A resistances analysis of these three filtration processes was calculated according to the article [24], and the results were shown in Table 1. The results showed that cake layer resistances were the main resistance for three solution conditions, especially for ammonia condition, which implies that the densest cake layer is formed in ammonia system.

Table 1 Ratio of filtration resistance in different solution conditions

Note:t, total resistance of fouled membrane;c, resistance of cake layer;p, resistance of pore blockage;m, resistance of new membrane.

3.3 Effect of silica forms on the filtration

Figure 6 The schematic representation of the three states of silica as monomer, polymer and colloid

The concentrations of silica in different solutions were shown in Fig. 7. It showed that the silica sol solution contained mainly silica colloids. The content of total silica dissolved in water was too low to polymerize, so the concentration of monomer and total silica of water were close. By contrast, the total silica under ammonia condition was very high, which reached to approximately 730 mg·L－1in 8 h. Because of high concentration, silica molecules had much more chance to meet each other, which promoted the monomer silica to polymerize [26].

Figure 7 Total and monomer silica concentration of different solution

The polymerization of silica correlates significantly to the size of colloids. The size of silica colloids were shown in Fig. 8. It showed that colloids in ammonia solution had a wide size distribution. This could ascribe to the fact that the high pH value of 12 and high concentration of silica were unfavorable to form colloids particles with uniform size [27, 28]. It seemed that mean size of particles was approximately 260 nm after 8 h. By contrast, the silica sol added into the slurry had a narrow size distribution with the mean size of only approximately 30 nm. Bringing together the consideration of Fig. 5, it might imply that the large-size silica colloids have strong influence on membrane flux. According to some reports, organic matter with high molecular weight (w) also influences membrane fouling greater than low one [29, 30]. Based on the experimental results above, it can be explained that silica sol with big size will constrict pores of cake layer and membrane in a great extent, which leads to a denser cake layer and more serious membrane pore plugging as compared with the silica sol with small size.

Figure 8 The size distribution of silica in different solutions

□?silica sol solution;○?ammonia solution

3.4 Effect of the dissolution on the size of TS-1 particles

As we know, the serious dissolution of silica from the TS-1 catalyst might influence the size of particles. In this study, TS-1 particles were soaked in simulated industrial feed for seven days. The samples were measured to investigate the change of particle size. As shown in Fig. 9, after soaking for seven days, the mean size decreased from 0.47 μm to 0.37 μm. While in industrial production, TS-1 catalyst will be used for a long period and be gradually dissolved by ammonia to a greater extent in a real industrial feed. Based on the analysis with K-C equation [31], it can be concluded that the resistance of cake layer formed with small particles is high. Furthermore, the small particles could more easily enter the membrane pores and lead to pore plugging compared with big particles.

Figure 9 Mean particle size of TS-1 particleoperating time

4 Conclusions

Silica played an important role in membrane fouling in the system of catalytic ammoximation of cyclohexanone to oxime. In this system, silica from different sources had different concentration and forms of composition. The content of silica dissolved in the ammonia environment was the highest, and the great extent of polymerization of silica lead to colloid particles formation, which had a great influence on membrane fouling. The size decrease of catalyst particles for alkaline dissolution also increased specific resistances of cake layers. Because deactivated TS-1 catalysts are often replaced by adding new TS-1 catalysts in industrial application, residual silica on the catalyst will accumulate in the system, leading to a significant flux decline after a long-time filtration. Therefore, TS-1 catalysts should be pre-treated by nitric acid to reduce membrane fouling caused by silica adsorption on catalyst surfaces. It is necessary to develop new anti-dissolution additives to the ammoximation system instead of silica sol.

1 Cesana, A., Mantegazza, M.A., Pastori, M., “A study of the organic by-products in the cyclohexanone ammoximation”,..., 117, 367-373 (1997)

2 Taramasso, M., Perego, G., Notari, B., “Preparation of porous crystalline synthetic material comprised of silicon and titanium oxides”, U.S. Pat., 4410501 (1993).

3 Rpffia, P., Padovan, M., Moretti, E., De Albetti, G., “Catalytic process for preparing cyclohexanone-oxime”, EP Pat., 0208311 (1987).

4 Zhang, X.J., Wang, Y., Xin, F., “Coke deposition and characterization on titanium silicalite-1 catalyst in cyclohexanone ammoximation”,..:., 307 (2/3), 222-230 (2006).

5 Petrini, G., Cesana, A., Alberti, G.D., Genoni, F., Leofanti, G., Padovan, M., Paparatto, G., Roffia, P., “Deactivation phenomena on Ti-silicalite”,...., 61, 761-766 (1991).

6 Liu, N., Guo, H.C., Wang, X.S., Chen, L.X., Chen, Y.Y., “Hydrothermostability of titanium silicate TS-1 zeolite in environment of cyclohexanone ammoxidation”,..., 24 (6), 441-446 (2003).

7 Sun, B., “Study on dissolution erosion of titanium silicalite zeolite in cyclohehanone ammoximation”,..., 36 (11), 54-58 (2005). (in Chinese)

8 Wu, W., Sun, B., Li, Y.X., Chen, S.B., Wang, E.Q. Zhang, S.Z., “Process for ammoximation of carbonyl compounds”, U.S. Pat., 20050215810 (2005).

9 Fu, S.B., Wang, H.B., Xu, F.H, Zhu, Z.H., “Cyclic separation of titanium silicalite-1 catalysts in their catalytic reactions”, CN Pat., 00113447.7 (2003).

10 Zhong, Z.X., Xing, W.H., Liu, X., Jin, W.Q., Xu, N.P., “Fouling and regeneration of ceramic membranes used in recovering titanium silicalite-1 catalysts”,..., 301 (1-2), 67-75 (2007).

11 Sahachaiyunta, P., Koo, T., Sheikholeslami, R., “Effect of several inorganic species on silica fouling in RO membranes”,, 144, 373-378 (2002).

12 Bremere, I., Kennedy, M., Mhiyo, S., Jaljuli, A., Witkamp, G.J., Schippers, J., “Prevention of silica scale in membrane systems: removal of monomer and polymer silica”,, 132, 89-100 (2000).

13 Sheikholeslami, R., Tan, S., “Effects of water duality on silica fouling of desalination plants”,, 126, 267-280 (1999).

14 Sch?fer, A.I., Schwicker, U., Fischer, M.M., Fane, A.G., Waite, T.D., “Microfiltration of colloids and natural organic matter”,..., 171 (2), 151-172 (2000).

15 Amy, G., Cho, J., “Interactions between natural organic matter (nom) and membranes: Rejection and fouling”,..., 40 (9), 131-139 (1999).

16 Braghetta, A., DiGiano, F.A., Ball, W.P., “NOM accumulation at NF membrane surface: Impact of chemistry and shear”,..., 124, 1087-1098 (1998).

17 Jucker, M.M., Clark, J., “Adsorption of aquatic humic substances on hydrophobic ultrafiltration membranes”,..., 97, 37-52 (1994).

18 Semiat, R., Sutzkover, I., Hasson, D., “Scaling of RO membranes from silica supersaturated solutions”,, 157 (1-3), 169-191 (2003).

19 Lee, G.D., Jung, S.K., Jeong, Y.J., Park, J.H., Lim, K.T., Ahn, B.H., Hong, S.S., “Photocatalytic decomposition of 4-nitrophenol over titanium silicalite (TS-1) catalysts”,..,, 239 (1/2), 197-208 (2003).

20 Bergna, H.E., Roberts, W.O., Colloidal Silica: Fundamentals and Applications, CRC Taylor & Francis, New York (2006).

21 Wittmann, G., Demeestere, K., Dombi, A., “Preparation, structural characterization and photocatalytic activity of mesoporous Ti-silicates”,.., 61 (1/2), 47-57 (2005).

22 Zhang, M.M., Li, C., Benjamin, M.M., Chang, Y.J., “Fouling and natural organic matter removal in adsorbent/membrane systems for drinking water treatment”,..., 37 (8), 1663-1669 (2003).

23 Lee, S.A., Choo, K.H., Lee, C.H., Lee, H.I., Hyeon, T., Choi, W., Kwon, H.H., “Use of ultrafiltration membranes for the separation of TiO2photocatalysts in drinking water treatment”,...., 40 (7), 1712-1719 (2001).

24 Ousman, M., Bennasar, M., “Determination of various hydraulic resistances during cross-flow filtration of a starch grain suspension through inorganic membranes”,..., 105 (1), 1-21 (1995).

25 Sj?berg, S., “Silica in aqueous environments”,.-., 196, 51-57 (1996).

26 Chang, S.M., Lee, M., Kim, W.S., “Preparation of large monodispersed spherical silica particles using seed particle growth”,...., 286 (2), 536-542 (2005).

27 Adamczyk, Z., Jachimska, B., Kolasińska, M., “Structure of colloid silica determined by viscosity measurements”,...., 273 (2), 668-674 (2004)

28 Bengoa, J.F., Gallegos, N.G., Marchetti, S.G., Alvarez, A.M., Cagnoli, M.V., Yeramián, A.A., “Influence of TS-1 structural properties and operation conditions on benzene catalytic oxidation with H2O2”,., 24 (4-6) ,163-172 (1998).

29 Lee, N.H., Amy, G., Lozier, J., “Understanding natural organic matter fouling in low-pressure membrane filtration”,, 178 (1-3), 85-93 (2005).

30 Son, H.J., Son, Y.D., Roh, J.S., Jik, W., Sin, P.S., Jung, C.W., Kang, L.S., “Application of MIEX?pre-treatment for ultrafiltration membrane process for NOM removal and fouling reduction”,.., 5 (5), 15-24 (2005).

31 Altmann, J., Ripperger, S., “Particle deposition and layer formation at the crossflow microfiltration”,..., 124 (1), 119-128 (1997).

2008-09-03,

2008-10-27.

the National Basic Research Program of China (2009CB623406), the National Natural Science Foundation of China (20806038), the Natural Science Foundation of Jiangsu Province (BK2008504), the National Science Foundation for Postdoctoral Scientists of China (20070421005) and Jiangsu Planned Projects for Postdoctoral Research Funds (0702020B).

** To whom correspondence should be addressed. E-mail: xingwh@njut.edu.cn