Studies in Surface Science and Catalysis, Vol. 139 J.J. Spivey, G.W. Roberts and B.H. Davis (Editors) 9 2001 Elsevier Science B.V. All rights reserved.
R e c e n t D e v e l o p m e n t s in the C a t a l y s t R e g e n e r a t i o n for the B e c k m a n n R e a r r a n g e m e n t of C y c l o h e x a n o n e O x i m e to s - C a p r o l a c t a m G. Dahlhoff, W. Eickelberg and W. F. H61derich* Department of Chemical Technology and Heterogeneous Catalysis, University of Technology RWTH Aachen, Worringerweg 1, 52074 Aachen, Germany
A fluidised bed reactor with continuous regeneration has shown to be the most suitable reactor for the large scale heterogeneously catalyzed Beckmann rearrangement of cyclohexanone oxime to e-caprolactam. For the operation of a fluidised bed plant with the catalyst being circulated between a reactor and a regenerator a close examination of the regeneration behaviour of the catalyst and an estimation of the necessary regeneration time is needed. The necessary regeneration time for modified [B]-MFI in a fluidised bed reactor using an empirical activation energy approach was determined. Dependent on the reaction conditions the amount of the deactivating compounds on the outer surface and also in the pores of the catalyst can be high, causing hot spots close to and on the catalyst surface during the regeneration with air which can destroy the active sites. Oxygen could be replaced by nitrogen as regeneration gas in a new reactor setup. 1
An important precursor for the synthesis of polyamide is e-caprolactam with capacities over 4.4 million tons per year world-wide, estimated to increase with 4-5 % per year in the next decade. The main disadvantage of the classic procedure starting from benzene is the production of 2.0 - 4.5 t (NH4)2804 per ton of product in the last two main steps of the manufacture, i.e. the oximation of cyclohexanone with hydroxylamine, and the Beckmannrearrangement of cyclohexanone oxime to e-caprolactam using concentrated sulfuric acid. An alternative for the oximation is the liquid-phase oxidation with hydrogenperoxide and ammonia over TS-1 [ 1]. For the last step, the heterogeneously catalysed gasphase Beckmannrearrangement, a catalyst with a balanced activity, a long life and a good regeneration behaviour is necessary. Thus far, results on the use silica-alumina , boronoxide on various carriers [3, 4, 5, 6], tantalumoxide  and several different zeolites including [8, 9, 10, 11, 12], ZSM-5 , Beta  and MCM-22  have been published. H61derich [16, 17, 18, 19] presented several surveys on the heterogeneously catalysed Beckmann-rearrangement. The same research group  recently published a review of the up to date production
* Corresponding Author, Tel.: + 49 / (0)241 / 80 65 60, Fax: + 49 / (0)241 / 88 88 291, E-Mail: [email protected]
336 processes including not only the heterogeneously catalysed processes, but also the butadiene based route and recycling routes as well as the activities of the industry in this field [27, 30]. For a heterogeneously catalysed Beckmann-rearrangement it is necessary to find a suitable reactor system for the catalysts, in this case a B-MFI catalyst. After investigating several alternatives a fluidised bed with continuous regeneration was found to be the best [27, 30]. In this reactor setup the catalyst is circulated from the reaction zone through a specially designed solid valve  into the regenerator. The objective of the presented work is to analyse the regeneration behaviour of the catalyst and to develop an empirical estimation method for the necessary regeneration time. Furthermore some investigations concerning the regeneration gas are presented. 2
2.1 Catalyst preparation B-MFI was prepared as described elsewhere [21 ]. The catalyst was extruded, crushed and sieved in fractions of 0.5-1.0 mm for fixed bed experiments and 63-250 gm for the fluidised bed. The yielded material was calcined in air with two steps. The B-MFI was first dried 4 h by 120~ to remove the physisorbed water from the surface and after this calcined by 550~ for 8 h. The catalyst was characterised with XRD (SIEMENS D5000), ICP AES (Spectroflame D) and FT-IR (PROTEGE460).
2.2 Fixed bed experiments Screening experiments were carried out in a tubular fixed bed reactor with an inner diameter of 10 mm and a vertical reaction zone of 40 mm. The liquid reactant was pumped in the evaporation zone by a membrane pump. The reaction pressures were varied from 10 kPa to 0.1 MPa with a membrane vacuum pump.
2.3 Fluidised bed experiments The fluidised bed reactor consisted of several seamless welded steel tubes (1.4571) with an inner diameter of the reaction zone of 51.2 mm, an inner diameter of the freeboard of 100 mm and a total height of 800 mm. The gas distributor grid was a sintered metal plate with a medium pore size of 50 gm. The whole reactor was heated using several heating coils. To preheat the entering carrier gas and reactants an evaporator was situated in front of the reactor. For the regeneration an identical reactor was connected to the first reactor. To circulate the solid catalyst from the reactor to the regenerator a solid circuit was set up consisting out of a specially designed solid valve (Figure 1) , a short air lift and a redistributor. Average reaction temperatures in the reactor were 280 to 400~ and pressures between 10 kPa and 0.1 MPa. To prevent temperature effects the gas entering the regenerator was preheated in a special turbulence zone.
2.4 Thermogravimetric analysis The regeneration behaviour of the catalyst was investigated with thermogravimetric analysis (NETZSCH209/2/E) under flowing air and nitrogen.
~~ DL04[~ N2011 ~
~.q ......... ...~a.=.;yst...i
b"t" .... ,
.c.~.st............ l ~~
I-%-_ ~ o
DL021~ Figure 1" Fluidised double bed with the solid valve.
RESULTS AND DISCUSSION
There are several papers on the kinetic analysis using thermogravimetric data [28, 29], which primarily deal with the thermal cracking of polymers used for the isolation of electric conductors. With the help of these investigations some reliable information about thermal resistance could be obtained in a short time thereby avoiding time consuming experiments. Femandez and Araujo  used thermogravimetric analysis for the description of the regeneration of deactivated catalysts in air, examining deactivated HY from the gas phase alkylation of benzene with dodecene. The development of a kinetic equation for this system was first described by Doyle  and was completed by several other authors. This thermogravimetric approach circumvents the usual problems appearing when defining a reaction set (like the order of the reaction). However, the use of the system is restricted to simple reactions with straight thermogravimetric plots. The model itself is based on a simple solid to gas oxidation A(s) --~ B(c), the Arrhenius equation and a very general reaction rate (1): dC dt
= k. f (C)
338 To solve this system of equations it is necessary to remove the time dependency from the basic kinetic equation, by introducing a linear heating rate, which is also time related. Now the function only depends on the remaining coke concentration and the heating temperature. This results in an equation for the activation energy of the regeneration process. Doyle  and Toop  developed a mathematical relationship between the activation energy and thermal durability of polymers. Here, this equation is used for an estimation of the regeneration time of a completely deactivated catalyst, i.e. the time necessary for the complete combustion of the residuals from the catalyst's surface.
logti = log
+ - ~ . T . l o g e + l o g p R-T,.
In this equation (2) ti is the time spent necessary for the isothermal combustion of the deactivating compounds, b the heating rate, T the absolute temperature, R the ideal gas constant, p the pressure and E the activation energy for an oxidative regeneration. The temperature Ti is the temperature at which a defined percentage of the residuals is burned off, and can be determined with the help of thermogravimetric analysis.
99%,, 95%\\. 90%\"\
GH 145 b = 10 K/min E* = 54 kJ/mol
7 5 %\""\" \ ~ "' 50%
823 Temperature [K]
Figure 2 Time of regeneration as a function of the temperature for different rates of regeneration. In Figure 2 the regeneration time as a function of different degrees of regeneration is shown. For the calculation a heating rate of 10~ was used from which an activation energy of 54 kJ/mol was determined. The regeneration time decreased with increasing regeneration temperature. The upper temperature limit was set by 850~ due to limitations of
339 the reactor design and to prevent sintering, which irreversibly destroys the active sites of the catalyst. The calculated regeneration times (figure 2) are in good agreement with the experimental data. The regeneration temperatures did not exceed 830 ~ because of the inherent instability of the linearisation. For this the resulting regeneration times can oscillate between a maximum and a minimum. However, the calculated regeneration times of ca. 7 h at 500 ~ seem to be realistic. The deactivating residuals on the outer surface of B-MFI are not of a aromatic nature. Albers et al.  showed that on the outer surface of the catalysts after the use in the Beckmann-rearrangement mainly long chain aliphatic compounds are formed. In the pores C-N compounds were found with concentrations much higher than that of the parafinic carbons on the outer surface. The reason for the much lower calculated activation energies than those found by Fernandez et al. . These observations were the basis for the next part of our experiments. The regeneration of the catalysts is done at temperatures around 500~ With the oxidation of hydrocarbons in air being an exothermic reaction the local temperatures on the surface of a strongly coked catalyst (on an average of ca. 3.9 wt % coke, Figure 3) are
(.9 }'-- 96 [1 ] -3.90%~ 3
Temperature [~ Figure 3" Thermogravimetric analysis of coked B-MFI in nitrogen atmosphere. much higher than the 500~ in the reactor. These hot spots can destroy the active sites of the catalyst after long time on stream. Strongly coked B-MFI from fixed-bed-reactions after long time on stream showed a slow but irreversible deactivation under certain circumstances. This irreversible deactivation was not found when using nitrogen instead of air as regeneration gas.
340 In table 1 the conversion and the selectivity of the Beckmalm rearrangement of cyclohexanone oxime to e-caprolactam over a regenerated B-MFI are shown. The catalyst was only weakly coked. Thus the temperature on the surface is not rising up to the critical limit, at which the active sites would be destroyed during the regeneration with air. As can be seen, there is no essential difference with regard to the performance between the use of nitrogen or oxygen as regeneration gas. Regeneration gas
Selectivity [%] 92.8
Table 1: Comparison of the performance of B-MFI after one regeneration with nitrogen and oxygen. Therefore nitrogen, which was also the carrier gas of the reaction, was tested for the regeneration in the fluidised bed. The loss of mass found with TG under nitrogen was identical to the loss of mass under oxidative conditions. In figure 3 is shown the diagram of the nitrogen TG. The first the step at ca. 120~ is the removal of water and some organic
Figure 4: Flow of the catalyst for regeneration with two steps
341 compounds (1.8 wt %) and the second loss of mass (350 - 750~ is the removal of coke (3.9 wt %). A TG under air of an material regenerated under nitrogen showed that all coke was removed from the catalyst. After the regeneration under nitrogen the performance of the catalyst was the same as that of a fresh catalyst. However, with the conditions in the mini plant being easier to control than in a large scale fluidised bed plant, we expect the sole use of nitrogen might be not sufficient for a complete regeneration. For an industrial scale plant we therefore propose a two step regeneration (figure 4): First under nitrogen to remove the main part of the coke, and a second under air to remove any remaining residues. Thus, hot spots that might cause an irreversible deactivation can be prevented, increasing the catalyst life time. 4
The regeneration of B-MFI as the heterogeneous catalyst for the gas phase Beckmannrearrangement of cyclohexanone oxime to ~-caprolactam in a continuous fluidised bed was investigated. A model of the necessary residence time of the deactivated catalyst in the regenerator of a continuous plant was created. This allows the design of a plant without unnecessary reaction volume containing only the necessary amount of catalyst to perform the reaction. Replacement of oxygen by nitrogen as the regeneration gas prevents hot spots on the catalyst surface during the regeneration. That stops a gradually increasing irreversible deactivation of the catalyst, which could very well be a significant problem in a non-ideal technical scale fluidised bed reactor. 5
The authors appreciate the steady interest of SUMITOMO CHEMICAL COMP. in their research and are grateful in particular to the research group of Hiroshi Ichihashi.
[ 1]. Roffia, P., Padovan, M., Moretti, E.,Alberti, G., EP 267.362, Montedipe S.p.A. . Immel, O., Schwarz, H.H., Starke, K. and Swodenk, W., Chem.-Ing.-Tech., 56(8) (1984), 612 . Immel, O., Schwarz, H.H., and Schnell, H., DP 1.670.816, BAYERAG (03.03.1967). . Irnich, R.,DP 1.227.028, BASF AG (25.07.1962). . Sato, S., Urabe, K., and Izumi, Y., J. CataL, 102 (1986), 99 . Curtin, T., McMonagle, J.B., Ruwett, M., and Hodnett, B.K., J. CataL, 142 (1993), 172 . Ushikubo, T., and Wada, K., J. CataL, 148 (1994), 138 . Landis, P.S.; Venuto, P.B.; J. Catal., 6 (1966), 245 . Sato, H., Hirose, K., Kitamura, M., and Nakamura, Y., Stud. Surf Sci. Catal., 49 (1989), 1213 . Sato, H., Hirose, K., Chem. Let., (1993), 1765 [ 11]. Ichihashi, H., Suzuki, T., Yago, M., JP 1105 7483, (1999) . H61derich, W.F., Heitmann, G.P., Dahlhoff, G., Ichihashi, H., DE 10010189 A1, (2000) . Takahashi, T., JP 09012539 A2, (1997) [ 14]. Dai, L.-X., Hayasaka, R., Iwaki, Y., Koyano, A.and Tatsumi, T., Chem. Commun., (1996), 1071 . H61derich, W.F., Dahlhoff, G., Barsnick, U.,AppL CataL A, 210 (2001), 83 . H61derich, W.F., Stud. Surf Sci. Catal., 46, (1989), 193 [ 17]. H6lderich, W.F., in "Comprehensive Supramolecular Chemistry", (G. Alberti and T. Bein, Eds.) Vol. 7, Pergamon, New York (1996), 671 . H6lderich, W.F., van Bekkum, H., Stud. Surf Sci. Catal., 58 (1991), 631 . H~ilderich, W.F., R~iseler, J., Heitmann, G., Liebens, A.T., Catal. Today, 37 (1997), 353 . Dahhoff, G., Niederer, J., H61derich W.F., CataL Rev. accepted for publication [21 ]. H61derich, W.F., R6seler, J., Arntz, D., DP 19608660, DEGUSSAAG . Doyle, C.D., J. Appl. Polym. Sci., 15 (1961), 285 . Toop, D.J.,IEEE Trans. Electr. InsuL EI-6 (1) (1971), 2 . Fernandez Jr., V. J.; Araujo, A. S.; Thermochim. Acta, 255 (1995), 273 . Albers, P.; Seibold, K.; Haas, T.; Pretscher, G.; H6lderich W. F., J. CataL, 176 (1998), 561 . H6lderich, W.F., Dahlhoff, G., Weckes, P.,DE 199 39 830 A1, (22.02.2001) . H6lderich, W.F., Dahlhoff, G., Chem. Inovation, 2 (2001), 29 . Coats, A. W.; Redfern, J. P.,Polym. Lett., 3 (1965), 917 . Flynn, J. H.; Wall, L. A.; Poym Lett., 4 (1966), 232 . Dahlhoff, G., H~lderich W.F., Proceedings of the 219th National Meeting of the American Chemical Society, San Francisco (2000)