Reductive amination of cyclopentanone

Reductive amination of cyclopentanone

Applied Catalysis A: General 286 (2005) 202–210 www.elsevier.com/locate/apcata Reductive amination of cyclopentanone P. Dolezˇal a,*, O. Machalicky´ ...

414KB Sizes 22 Downloads 95 Views

Applied Catalysis A: General 286 (2005) 202–210 www.elsevier.com/locate/apcata

Reductive amination of cyclopentanone P. Dolezˇal a,*, O. Machalicky´ b, M. Pavelek a, P. Kubec a, K. Hra´dkova´ a, R. Hrdina b, R. Sˇula´kova´ b b

a BorsodChem MCHZ, s.r.o., Chemicka´ 1/2039, 709 03 Ostrava–Maria´nske´ Hory, Czech Republic Department of Organic Technology, University of Pardubice, Studentska´ 95, 532 10 Pardubice, Czech Republic

Received 23 December 2003; received in revised form 3 March 2005; accepted 7 March 2005 Available online 27 April 2005

Abstract The kinetics of heterogeneous catalytic reductive amination of cyclopentanone has been studied on a pilot plant PARR autoclave. Ncyclopentyliminocyclopentane was detected as the main intermediate in a reaction mixture. It was found that, at the given conditions, the main intermediate does not form the undesirable N,N-dicyclopentylamine but undergoes slow hydrolysis, and the desired product, cyclopentylamine, results in a good yield. Slight amounts of by-products such as cyclopentanole and N,N-dicyclopentylamine were obtained. The experimental data were confronted with the suggested kinetic model. # 2005 Elsevier B.V. All rights reserved. Keywords: Cyclopentanone; Cyclopentylamine; N-cyclopentyliminocyclopentane; Heterogeneous catalytic reductive amination; Kinetics

1. Introduction Cyclopentylamine is an important industrial intermediate, e.g. in production of the fungicides based on N-benzylcycloalkylamines [1]. The respective industrial manufacturing process (used in BorsodChem s.r.o., Ostrava, The Czech Republic) is based on reductive amination of cyclopentanone catalysed by powdered Raney nickel (Ra–Ni) and carried out in an autoclave. This reaction belongs among the group of reductive alkylations of amines [2]. It is known that hydrogenation of the imine (prepared in situ by condensation of a ketone with ammonia) occurs at room temperature and at atmospheric pressure in ethanol in the presence of catalyst (5% palladium on charcoal). Raney nickel in ethanol at 326–363 kPa is an alternative catalyst, where reductive amination of ketone preferably occurs and parallel hydrogenation of ketone also takes place [3]. Since the process consists in bimolecular reactions on the catalyst surface, the yield of cyclopentylamine strongly depends on the reaction conditions. Moreover, the amine primarily produced competes with ammonia in the carbonyl * Corresponding author. Tel.: +420 596643451; fax: +420 596642647. E-mail address: [email protected] (P. Dolezˇal). 0926-860X/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2005.03.007

condensation step. A proper choice of concentrations of the reagents can lead to the desired product (primary amine) in a high yield [2]. The use of a large excess of ammonia gives the primary amine as the predominant product. An excess of carbonyl compound gives the symmetrical secondary or tertiary amines. The preparation of cyclopentylamine by reductive amination of cyclopentanone over Raney nickel catalyst was described by Corrigan et al. [4], Plate et al. [5] and Xing et al. [6]. The procedure described in the patent [6] uses ammonia/ cyclopentanone in the molar ratio 4/1; the catalyst used is a mixture of Ra–Ni and MgCl2; the reaction pressure is 5– 11 MPa, the reaction temperature is 120–180 8C. The declared conversion of the ketone is 99%, selectivity 96% and the yield of amine is 89%. Other ways of reductive amination have also been developed, e.g. by Panfilov et al. [7] and Johansson et al. [8], where sodium borohydride and borane-dimethylsulphide complex were used as a reducing agents. However, these procedures are only convenient for laboratory purposes because of the very expensive reducing agents. The aim of this present work is a kinetic study of the reductive amination of cyclopentanone in a laboratory

P. Dolezˇal et al. / Applied Catalysis A: General 286 (2005) 202–210

PARR autoclave (BorsodChem) which should provide a kinetic model of the process as well as kinetic data useful for industrial purposes.

2. Experimental 2.1. Apparatus The reductive amination was carried out in a PARR 4522 autoclave from Parr Instrument Company. The reactor of total volume 2 dm3 was equipped with the mechanic stirrer, inlet of gases, outlet and the sample port (Fig. 1). 2.2. Process of reductive amination The reactor was filled with 0.95 kg cyclopentanone (11.29 mol, commercial sample from Rhodia Organique), and 0.95 g powdered Ra–Ni (commercial sample form Endelhard De Meern B.V.) was added. Then, the reaction vessel was closed and loaded with 0.37 kg liquid ammonia (21.73 mol); the starting temperature was 20 8C and the initial pressure was 1 MPa. The starting liquid reactants (mixture of ammonia and cyclopentanone) occupied ca. 80.3% of total reactor volume. Subsequently, the reactor was heated to the experimental temperature (80–150 8C) and hydrogen was introduced to reach and maintain the desired constant experimental pressure (4–9 MPa). At definite time intervals, reaction samples (1.5 cm3) were taken and cooled to the laboratory temperature (22 8C), their residual free ammonia was volatilised, the catalyst was filtered off, and the samples were analysed by gas chromatography.

203

2.3. Gas chromatography Gas chromatography (GC) was performed on a Hewlett Packard 6890 apparatus equipped with a 25 m capillary column HPI-dimethylsilicon with the inner diameter of 0.2 mm and the film width of 0.25  106 m. A temperature gradient of 30 8C min1 from 60 to 250 8C was used for the analysis. Helium 30 mL min1 was used as a carrier gas. A volume of 0.2  106 L sample was injected without dilution at the temperature of 60 8C. Chromatographic responses of analysed compounds were detected at 250 8C with FID Hewlett Packard detector. Concentrations of compounds were determined by comparing with chemical standards of cyclopentanone, cyclopentylamine and cyclopentanole supplied from Aldrich (99%). The other components (N-cyclopentyliminocyclopentane, N,N-dicyclopentylamine and 2-cyclopentylcyclopentanone) were identified by GC–MS (for the spectra, see Chapter 3) before they were isolated from the reaction mixtures by means of fraction distillation. Chromatographic responses were converted to mass%. 2.4. GC–MS The chromatographic apparatus consisted of a Hewlett Packard HC 5890 pump, a 30 m capillary column HPIdimethylsilicon with an inner diameter of 0.25 mm and the film width of 0.25  106 m. A temperature gradient of 30 8C min1 from 60 to 250 8C was used for the analysis. Helium 0.5 mL min1 was used as a carrier gas. The chromatographic responses of analysed compounds were detected by means of a 5971A MS detector. The quadrupole mass spectrometer with EI probe was used for all the measurements. The EI ion source temperature was set at 250 8C. 2.5. Calculations All the calculations such as linear regression and nonlinear fits of measured data were performed on PC using commercial software Microsoft Excel 97 SR-1.

3. Results and discussion Reductive amination of cyclopentanone to cyclopentylamine catalysed by powdered Raney nickel (Ra–Ni) was carried out in a laboratory autoclave (Fig. 1) and can be described by the total chemical equation (Fig. 2). All the experiments described in this work were carried out at a constant concentration of catalyst Ra–Ni (1 g per Fig. 1. Scheme of the PARR reactor used. (1) Pressure gage; (2) stirrer drive system; (3) gas inlet valve; (4) liquid sampling valve; (5) thermocouple; (6) water cooling channel; (7) dip tube; (8) gas release valve; (9) safety rupture disc; (10) stirring shaft; (11) lower guide bearing.

Fig. 2. Synthesis of cyclopentylamine (C5H11N) from cyclopentanone (C5H8O).

P. Dolezˇal et al. / Applied Catalysis A: General 286 (2005) 202–210

204

Fig. 3. GC chromatogram of typical final reaction mixture. Reaction conditions: 1 g Ra–Ni per 1 kg cyclopentanone, 1.92 mol NH3 per 1 mol cyclopentanone; reaction temperature: 80–150 8C; reaction pressure: 4– 9 MPa.

1 kg cyclopentanone) and at a constant charge of ammonia (1.92 mol NH3 per 1 mol cyclopentanone). The reaction temperature was 80–150 8C, and reaction pressure was between 4 and 9 MPa. At the above-described experimental conditions, there are two phases in the reactor: liquid (mostly consisting of starting cyclopentanone, its reaction products, and liquid ammonia) and gaseous (a mixture of hydrogen and ammonia). The reductive amination of cyclopentanone is a multistep process; therefore, the reaction mixture was analysed by gas chromatography (GC) and mass spectrometry (MS). An example of a typical GC chromatogram of final reaction mixture obtained is shown in Fig. 3 and Table 1. The mass spectra of N-cyclopentyliminocyclopentane, N,N-dicyclopentylamine and 2-cyclopentylcyclopentanone are depicted in Fig. 4. The unidentified signals of substances A and B in Table 1 belong to more complex poly-cyclopentyl derivatives with relative molecular masses of 220 and 204. Each kinetic run at given reaction conditions was repeated twice. An example of kinetic data collected in a typical experiment is shown in Table 2 and Fig. 5. The reaction conditions used and the kinetic data obtained led us to the presumption that the reductive amination proceeds in the liquid phase, where the concentration of

Fig. 4. EI mass spectra of (a) N-cyclopentyliminocyclopentane (C10H17N, Mr = 151), (b) N,N-dicyclopentylamine (C10H19N, Mr = 153) and (c) 2cyclopentylcyclopentanone (C10H16O, Mr = 152).

Table 1 GC data of components detected in typical final reaction mixture Formula

Compound

Retention time (min)

Concentration (mass%)

C5H8O C5H11N C10H17N C5H10O C10H16O C10H19N A B

Cyclopentanone Cyclopentylamine N-cyclopentyliminocyclopentane Cyclopentanol 2-Cyclopentylcyclopentanone N,N-dicyclopentylamine Unidentified signal 1 Unidentified signal 2 Unidentified noise

3.34 3.16 8.06 3.41 7.94 8.09 10.1–10.3 10.6–10.8

0.00 85.30 0.61 6.07 2.50 0.93 2.31 0.26 2.03

Reaction conditions: 1 g Ra–Ni per 1 kg cyclopentanone; 1.92 mol NH3 per 1 mol cyclopentanone; reaction temperature: 80–150 8C; reaction pressure: 4–9 MPa.

Fig. 5. The relationship between actual concentrations of components and reaction time (the kinetic data correspond to those in Table 2). (1) cyclopentanone (*); (2) N-cyclopentyliminocyclopentane (); (3) cyclopentylamine (&).

P. Dolezˇal et al. / Applied Catalysis A: General 286 (2005) 202–210

205

Table 2 Experimental data obtained in typical experiment Time (h)

Actual concentrations of compounds (mol kg1) Computed from material balancesa

Measured by means of GC

0 1.50 2.50 6.33 9.33 21.00

[C5H8O]

[C10H17N]

[C5H11N]

[NH3]

[C5H9N]b

[H2O]

8.464 1.086 0.603 0.121 0 0

0 1.899 1.461 0.464 0.162 0.041

0 2.874 4.018 6.121 6.668 6.623

16.93 11.03 10.10 8.573 8.116 7.898

0 0.283 0.493 0.818 0.962 1.152

0 6.955 7.433 7.867 7.954 7.857

Reaction conditions: 1 g Ra–Ni per 1 kg cyclopentanone; 1.92 mol NH3 per 1 mol cyclopentanone; reaction temperature: 80–150 8C; reaction pressure: 4– 9 MPa. a See below Eqs. (8a), (9) and (10). b Iminocyclopentane.

dissolved hydrogen was low and constant (fast mass transfer from gaseous phase). Thus, the following set of chemical and kinetic equations can be written (Scheme 1). The symbol kN represents the rate constant of reaction N, and [X] represents the actual concentration of compound X at the reaction time t. Since the actual concentrations of by-components C5H10O, C10H16O, C10H19N, A and B in the reaction mixtures were relatively low (approximately 6 times lower than the concentration of the main product C5H11N), the reactions (r4)–(r6) can be neglected, and the following set of differential Eqs. (1)–(6) can be written as:

the symbol [X]0 represents the starting concentration of compound X (e.g. at the reaction time t = 0).

d½C5 H8 O ¼ k2 ½C5 H9 N½H2 O  k1 ½C5 H8 O½NH3  dt þ k4 ½C10 H17 N½H2 O  k3 ½C5 H8 O½C5 H11 N

½NH3  ¼ ½C5 H8 O þ ½NH3 0  ½C10 H17 N0  ½C5 H8 O0 (8a)

(1) d½NH3  ¼ k2 ½C5 H9 N½H2 O  k1 ½C5 H8 O½NH3  dt

(7a)

d½C5 H9 N d½NH3  d½C5 H11 N d½C10 H17 N þ þ þ ¼0 dt dt dt dt

(7b)

d½H2 O d½NH3  d½C10 H17 N þ  ¼0 dt dt dt

(7c)

½C5 H9 N ¼ ½NH3   ½C10 H17 N  ½C5 H11 N þ ½NH3 0 þ ½C10 H17 N0 þ ½C6 H11 N0 þ ½C5 H9 N0

(2)

d½C5 H9 N ¼ k2 ½C5 H9 N½H2 O þ k1 ½C5 H8 O½NH3  dt  kexp ½C5 H9 N

d½C5 H8 O d½NH3  d½C10 H17 N  þ ¼0 dt dt dt

(8b) ½H2 O ¼ ½NH3  þ ½C10 H17 N þ ½NH3 0  ½C10 H17 N0

(3)

d½H2 O ¼ k2 ½C5 H9 N½H2 O þ k1 ½C5 H8 O½H2 O dt  k4 ½C10 H17 N½H2 O þ k3 ½C5 H8 O½C5 H11 N (4) d½C5 H11 N ¼ k4 ½C10 H17 N½H2 O  k3 ½C5 H8 O½C5 H11 N dt (5) þ kexp ½C5 H9 N d½C10 H17 N ¼ k4 ½C10 H17 N½H2 O þ k3 ½C5 H8 O½C5 H11 N dt (6) The integration of differential form of material balance expressed by Eqs. (7a)–(7c) leads to Eqs. (8a)–(8c), where

þ ½H2 O0

(8c)

The set of Eqs. (8a)–(8c) makes it possible to calculate the concentrations of the compounds (ammonia, iminocyclopentane, water) that were not directly measured in reaction mixture. The concentrations of ammonia were computed easily from Eq. (8a), those of iminocyclopentane and water follow from Eqs. (9) and (10). ½C5 H9 N ¼ ½C5 H8 O  2½C10 H17 N  ½C5 H11 N þ ½C5 H8 O0 þ 2½C10 H17 N0 þ ½C5 H11 N0 þ ½C5 H9 N0 ½H2 O ¼ ½C5 H8 O þ ½C5 H8 O0 þ ½H2 O0

(9) (10)

From the results (see Table 2) it can be seen that the actual concentrations of cyclopentanone and ammonia rapidly decrease at the beginning of reaction. The actual concen-

P. Dolezˇal et al. / Applied Catalysis A: General 286 (2005) 202–210

206

Scheme 1. Model of reductive amination of cyclopentanone.

tration of the intermediate N-cyclopentyliminocyclopentane increases from zero to the maximum and then falls to a trace concentration. The actual concentrations of products, cyclopentylamine and water, always increase during the reaction. However, the actual concentration of the predicted intermediate iminocyclopentane is unexpectedly fluctuating (decreasing/increasing) in the course of the reaction time. Mathematical integration of the set of Eqs. (1)–(6) is impossible. For this reason the concentration fractions in Eqs. (11) and (12) were tested to simplify the above-defined kinetic model. K1 ¼

½C5 H9 N½H2 O ½C5 H8 O½NH3 

(11)

K2 ¼

½C10 H17 N½H2 O ½C5 H8 O½C5 H11 N

(12)

It was found that the concentration fraction K1 was time dependent as is illustrated in Fig. 6. On the other hand, the concentration fraction K2 was time independent. Therefore, it can be considered that K2 represents the equilibrium constant. Its value was estimated from the linear fit of experimental data (Fig. 7). The concentration product [C5H8O][NH3] was strongly time-dependent at the beginning of the process, which can be interpreted as an effect of fast iminocyclopentane formation (reaction rate r1 in Scheme 1). We suppose, that the actual concentrations of cyclopentanone, NH3 and H2O shift immediately from starting values (in the next text

P. Dolezˇal et al. / Applied Catalysis A: General 286 (2005) 202–210

207

½NH3 S ¼ ½C5 H8 OS  ½C5 H8 O0 þ ½NH3 0

(13)

½C5 H9 NS ¼ ½C5 H8 OS þ ½C5 H8 O0

(14)

½H2 OS ¼ ½C5 H8 OS þ ½C5 H8 O0

(15)

Combination of formulas in Eqs. (13)–(15) with formula in Eq. (11) gives expression as in Eq. (16), allowing determination of the equilibrium concentration of ammonia [NH3]S in the liquid phase, where parameters are c = [NH3]02, b = K1[C5H8O]0  K1[NH3]0 + 2[NH3]0 and a = K1  1. The calculated equilibrium concentrations (for K1 = 1.92, see bellow) are summarised in Table 3. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi b þ b2 þ 4ac (16) ½NH3 S ¼ 2a

denoted as [C5H8O]0, [NH3]0 and [H2O]0) to the equilibrium ones ([C5H8O]S, [NH3]S and [H2O]S). Relationships among them can be written in the form of formulas in Eqs. (13)–(15).

A comparison of actual concentrations of ketone, ammonia and water in Table 2 with those in Table 3 shows that the actual concentrations of compounds at the time = 1.5 h (Table 2) are quite near to those of the equilibrium ones (in Table 3). We can also see (Table 2) that the final concentration of amine is lower than that of water, which means that cyclopentanone was converted to cyclopentylimine, but this imine was not completely converted to the desired amine. Since it can be supposed that the equilibrium between concentrations of cyclopentanone and iminocyclopentane is practically reached immediately at the beginning of the process, relationships (1), (2), (4) and (6) can be put equal to zero. Subsequently, Eqs. (3) and (5) can be written in the

Fig. 7. The relationship between the concentration products [C5H8O][C5H11N] and [C10H17N][H2O]. Linear fit: y = K2x, where x = [C5H8O][C5H11N] and y = [C10H17N][H2O], gives the slope K2 = 4.4 (S.D. 0.2 and correlation coefficient 0.996).

Fig. 8. Linearisation of Eq. (19) using starting concentration of cyclopentanone [C5H8O]0 = 8.46 mol kg1. Linear fit: y = ax + b gives the slope a = 0.7 (standard deviation 0.1) and the intercept b ¼ lnðkexp K1a Þ ¼ 0:3 (standard deviation 0.1).

Fig. 6. The time dependence of the concentration fraction K1.

P. Dolezˇal et al. / Applied Catalysis A: General 286 (2005) 202–210

208

Table 3 Calculated steady-state concentrations of reactants in the liquid phase in the beginning of reaction Steady-state concentrations (mol kg1)

Reaction time (h)

t = tS ! 0

[C5H8O]S

[C10H17N]S

[C5H11N]S

[NH3]S

[C5H9N]S

[H2O]S

2.0

0.0

0.0

10.5

6.4

6.4

form of Eq. (17), where kexp is an experimental (apparent) reaction constant and a is the reaction order. d½C5 H9 N d½C5 H11 N ¼ ¼ kexp ½C5 H9 Na dt dt

(17)

The relationship between actual concentrations of iminocyclopentane and cyclopentanone can be written as follows: ½C5 H9 N ¼

K1 ½C5 H8 O½NH3  ½C5 H8 O0  ½C5 H8 O

(18)

The reaction order a can be determined from Eq. (19), as the result of connection of Eqs. (17) and (18). In our experiments the reaction orders obtained ranged between 0.7 and 1. The plot of real data in accordance with Eq. (19) is shown in Fig. 8.     d½C5 H11 N ½C5 H8 O½NH3  ln ¼ a ln dt ½C5 H8 O0  ½C5 H8 O þ

lnðkexp K1a Þ

(19)

Since the concentration of ammonia was practically timedependent only within a short period at the beginning of the reaction (see Table 2), it can be supposed that the term K1[NH3] in the Eq. (18) converges to the constant K1A. The derivation of Eq. (18) leads to relationship as in Eq. (20) between reaction rates of formation of iminocyclopentane and cyclopentanone, and subsequently Eq. (21) can be written.

(20) ða1Þ

kexp K1A d½C5 H8 O ¼ ½C5 H8 Oa dt ½C5 H8 O0 (21)

Eq. (21) can be simplified to give Eq. (22) if a = 1.   1 1 þ d½C5 H8 O ½C5 H8 O ½C5 H8 O0  ½C5 H8 O ¼ kexp dt

½C5 H8 O ¼

½C5 H8 O0 g ekexp t 1 þ g ekexp t

(24)

Rate Eq. (25) of formation of cyclopentylamine is obtained by the substitution of Eq. (24) into Eqs. (17) and (18).   d½C5 H11 N ½C5 H8 O½NH3  ¼ kexp K1 dt ½C5 H8 O0  ½C5 H8 O kexp K1A g ekexp t ¼ kexp ½C5 H9 NS ekexp t

(25)

The integration of differential Eq. (25) gives the kinetic Eq. (26) for cyclopentylamine. ½C5 H11 N ¼ ½C5 H9 NS ð1  ekexp t Þ þ ½C5 H11 NS

(26)

Kinetic Eq. (27) of formation of Schiff’s base C10H17N is obtained by the substitution of Eqs. (24) and (26) into Eq. (12). ½C10 H17 N ¼ K2

d½C5 H9 N ½C5 H8 O0 d½C5 H8 O ¼ K1A dt dt ð½C5 H8 O0  ½C5 H8 OÞ2

 ð½C5 H8 O0  ½C5 H8 OÞð2aÞ

Experimental data and their linear fit based on Eq. (23) are illustrated in Fig. 9. From the parameters a = 0.7, lnðkexp K1a Þ ¼ 0:3 and kexp = 0.47 h1, the equilibrium constant K1 = 1.92 (S.D. 0.01) was calculated (compare with the value 0.09 found at atmospheric pressure and at an ambient temperature in literature [9]). The explicit form of the kinetic curve of cyclopentanone is described by Eq. (24).

½C5 H8 O½C5 H11 N ½C5 H8 O0  ½C5 H8 O

¼ K2 g ekexp t ð½C5 H9 NS ð1  ekexp t Þ þ ½C5 H11 NS Þ

(27)

Eqs. (24), (26) and (27) represent simplified integral kinetic curves of the model designed in Scheme 1. Table 4 summarises the parameters and experimental forms of kinetic curves used in the representative Fig. 10. Table 4 Parameters and experimental kinetic curves of cyclopentanone, cyclopentylamine and N-cyclopentyliminocyclopentane

(22)

The integration of differential Eq. (22) leads to integral Eq. (23), where g = [C5H8O]S/[H2O]S.   ½C5 H8 O ln (23) ¼ kexp t þ lnðgÞ ½C5 H8 O0  ½C5 H8 O

List of parameters [C5H8O]0 = 8.5 mol kg1 g 0.300 kexp. = 0.47 h1 [C5H9N]S = 6.44 mol kg1 [C5H11N]S = 0.0 mol kg1 K2 = 4.4

Experimental forms of kinetic curves 0:47t

2:54e ½C5 H8 O ¼ 1þ0:30e 0:47t

½C5 H11 N ¼ 6:44  ð1  e0:47t Þ ½C10 H17 N ¼ 8:40ð1  e0:47t Þ  e0:47t

P. Dolezˇal et al. / Applied Catalysis A: General 286 (2005) 202–210

Fig. 9. Linear fit of Eq. (23) using starting concentration of cyclopentanone [C5H8O]0 = 8.46 mol kg1 gives y = 0.468t  1.296 with a correlation coefficient of 0.997. The slope kexp. = 0.47 h1 represents the experimental rate constant. Standard deviations of slope and intercept are 0.04 and 0.15, respectively.

Fig. 11. Analytical curve of concentration-time dependence of iminocyclopentane during reductive amination using Eq. (28) and optimised parameters in Table 4.

½C5 H9 N ¼  Finally, the kinetic equation for iminocyclopentane can be expressed using Eqs. (24), (26), (27) and (8a), (8b), and regarding [C10H17N]S = [C5H11N]S = 0, as follows:

Fig. 10. Reductive amination of cyclopentanone. Experimental data (Table 2) and their fit with optimised kinetic curves (Table 4). (1) Cyclopentanone (*); (2) N-cyclopentyliminocyclopentane (); (3) cyclopentylamine (&).

209

½C5 H8 O0 g ekexp t  ½C5 H9 NS ð1  ekexp t Þ 1 þ g ekexp t  ð1 þ 2K2 g ekexp t Þ þ ½C5 H8 O0 (28)

Eq. (28), describing the relationship between the actual concentration of iminocyclopentane and reaction time, exhibits a minimum (shown in Fig. 11). The intercept equals to

Fig. 12. Dependences of the equilibrium concentration [C5H8O]S (curve 1) and the yield of cyclopentylamine (curve 2) on the starting ammonia concentration.

P. Dolezˇal et al. / Applied Catalysis A: General 286 (2005) 202–210

210

[C5H9N]S and an asymptote equals to the equilibrium concentration [C5H8O]S. As far as we accept this principle, we can conclude that the lower the equilibrium concentration [C5H8O]S, the better the yield of cyclopentylamine. We have suggested relationships as in Eqs. (13)–(16) between starting and equilibrium concentrations in the previous text. If we express equilibrium concentration [C5H8O]s of cyclopentanone as a function of the starting concentration [NH3]0 of ammonia (using Eqs. (13) and (16)), then we can also define the dependence of the yield h of cyclopentylamine on the starting ammonia concentration (Eq. (29)). The dependences of both concentration [C5H8O]S and the yield h of cyclopentylamine on the starting ammonia concentration are shown in Fig. 12. h¼

½C5 H8 O0  ½C5 H8 OS  100 ½C5 H8 O0

(29)

It can be concluded, that the presented kinetic model is in good accordance with experimental results acquired on pilot plant PARR autoclave. Subsequent tests on industrial BorsodChem autoclave exhibited similar results.

yield of cyclopentylamine and, consequently, increases the production of by-products. The starting ratio of ammonia/ cyclopentanone is a key parameter for good yield of the desired cyclopentylamine. We suppose that the ratio ammonia/cyclopentanone 4.7 is needed to reach 90% yield. A mass balance of the suggested reaction scheme gives a somewhat surprising result for supposed intermediate iminocyclopentane. Its maximum concentration at the beginning of the process decreases to the minimum and then asymptotically increases. Theoretical final residual concentration of iminocyclopentane is inversely proportional to the starting molar ratio of ammonia/cyclopentanone. The cyclopentylamine resulting from the reduction of iminocyclopentane immediately forms N-cyclopentyliminocyclopentane by the reaction with cyclopentanone, where the equilibrium constant of that process was found to be K2 = 4.4 (S.D. 0.2). The concentration of N-cyclopentyliminocyclopentane formed decreases in the final part of the process due to the reversible hydrolysis.

4. Conclusions The results of the study of the kinetics of the reductive amination of cyclopentanone in a pilot plant PARR autoclave suggest the following conclusions valid for the reaction temperature 80–150 8C and reaction pressure 4– 9 MPa. Desired cyclopentylamine is formed by the hydrogenation of iminocyclopentane which results from the reaction of cyclopentanone and ammonia at the beginning of the process. The starting formation of iminocyclopentane is an equilibrium process, where the equilibrium constant is K1 = 1.92 (S.D. 0.01).

The equilibrium concentration of iminocyclopentane is probably the most important variable in the process. This concentration is given by the charged molar ratio ammonia/ cyclopentanone. Lower concentration of ammonia at the beginning of the process decreases the reaction rate and the

Acknowledgement The authors wish to thank BorsodChem MCHZ, s.r.o. for financial support and for permission to publish the above results. References [1] S. Junichi, T. Tatsuo, Jpn. Patent 53,149,949 (1976). [2] W.S. Emerson, Org. React. IV (1948) 174. [3] R.L. Augustine, Catalytic Hydrogenation, Marcel Dekker Inc., New York, 1965, pp. 103. [4] Corrigan, et al. J. Am. Chem. Soc. 75 (1953) 6258. [5] Plate, et al. Zh. Obsch. Chim. 26 (1956) 684–687. [6] X. Xing et al., CN 1,063,680 (1992); through C.A. 118, 254444h (1993). [7] A.V. Panfilov, Yu.D. Markovich, A.A. Zhirov, I.P. Ivashev, A.T. Kirsanov, V.B. Kondratev, Pharm. Chem. J. 34 (7) (2000) 371–373. [8] A. Johansson, E.-L. Lindstedt, T. Olson, Acta Chem. Scand. 51 (1997) 351–353. [9] M. Brˇezina, P. Zuman, Chemicke´ listy 46 (1952) 599.