A Simplified Overall Kinetic Model for Cyclohexanone Oximation by Hydroxylamine Salt David Lorenzo,* Arturo Romero, and Aurora Santos Departamento de Ingeniería Química, Facultad de Químicas, Universidad Complutense de Madrid, Ciudad Universitaria, 28040 Madrid, Spain *S Supporting Information ABSTRACT: Cyclohexanone oxime (ONEOX) is the precursor of ε-caprolactam, a monomer used within the nylon-6 industry. This work studies the production of ONEOX from cyclohexanone (ONE) oximation with hydroxylammonium sulfate (HAS). The reaction involves two liquid phases: HAS is present in the aqueous phase, and ONEOX and ONE are present in the organic phase. The influence of pH, the interfacial area between phases, and the concentration of reagents on the oximation rate have been studied, taking into account the effect of these variables on the proposed model for the overall rate. Oximation runs have been carried out in a batch reactor at low agitation speed to control the interfacial area between phases. The pH ranged from 3 to 5.5, and the temperature was set at 353 or 358 K. The contact model proposed for the two liquid phases was based on the two-film theory, and it was assumed that the oximation reaction takes place in the aqueous film. Because of the low solubility of organic species in the aqueous phase (salting-out effect), the concentration of salts was considered negligible in the organic phase and the concentration of hydroxylamine was assumed to be constant in the aqueous film. The rate expression obtained was first order for ONE and 0.5 for HAS, using a pseudokinetic constant that included the transport coefficient of ONE in the aqueous film, and the pH effect. It was found that this apparent kinetic constant shows a slight increase when the pH is increased within the studied range. The kinetic model proposed predicts well the overall oximation rate under the experimental conditions studied. This model includes quantitatively the influence of the main variables: temperature (353−358 K), pH (3−5), and concentration of reagents. 1. INTRODUCTION Cyclohexanone oxime (ONEOX) is a compound used as an intermediate in the manufacture of ε-caprolactam, which is polymerized to obtain nylon-6.1−4 The principal precursor of ONEOX is pure cyclohexanone (ONE), which reacts in the oximation process. This reaction is produced in different ways. The commercial processes and conditions are summarized in Table 1. The named processes can be categorized into two groups. The first one corresponds to the most commonly employed process, involving the reaction of pure ONE with hydroxylamine (HA) to produce ONEOX.1,5,6 Due to the high volatility of HA, this species is supplied in the form of either sulfate (HAS) or phosphate (HAP). The second group uses a more recent technology, where ONE reacts with ammonium and a peroxide compound over a heterogeneous catalyst to produce ONEOX. This process has been carried out on an industrial scale known as ammoximation.7−9 BASF and Inventa produce ONEOX in two stages. The first stage is the preparation of HAS solution by nitric oxide hydrogenation, using platinum as a heterogeneous catalyst. The HAS produced is then used to carry out the cyclohexanone oximation reaction at 80 °C in a weak acidic solution. Sulfuric acid is produced as a byproduct in both HAS and in the cyclohexanone oximation reaction and is neutralized with ammonia to form ammonium sulfate.10,11 The main drawback of this process is the high amount of ammonium sulfate produced. For this reason, BASF6 and DSM10,12 have developed different technologies in order to reduce the amount of this byproduct that is formed. The BASF process is based on oximation without neutraliza- tion, also known as acidic oximation, in which ammonium hydrogen sulfate is recycled to the HA production process.6,10 The DSM process uses HA as a phosphorus salt produced by reduction of the phosphoric acid and ammonium nitrate solution with hydrogen over palladium on graphite as a heterogeneous catalyst, as can be seen in Table 1. The HAP salt then reacts with cyclohexanone at pH 2 to produce ONEOX.10,12,13 Enichem patented a process for producing ONEOX using ONE and a mixture of ammonium and hydrogen peroxide as reactants.7−9 The reaction is carried out over a heterogeneous catalyst of SiO2−TiO2. The reaction medium can be an organic solvent, such as tert-butanol,8,14 or a three-phase reactor (organic−aqueous−solid) in which the catalyst can be more Received: February 29, 2016 Revised: May 20, 2016 Accepted: May 24, 2016 Published: May 24, 2016 Article pubs.acs.org/IECR © 2016 American Chemical Society 6586 DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 D ow nl oa de d vi a U N IV C O M PL U T E N SE D E M A D R ID o n Ja nu ar y 10 , 2 02 4 at 1 5: 29 :5 5 (U T C ). Se e ht tp s: //p ub s. ac s. or g/ sh ar in gg ui de lin es f or o pt io ns o n ho w to le gi tim at el y sh ar e pu bl is he d ar tic le s. pubs.acs.org/IECR http://dx.doi.org/10.1021/acs.iecr.6b00813 easily removed.8 The ammoximation process can be developed by two different routes: (a) Iminic route:15 According to Reddy et al.,16 the reaction mechanism can be explained in two steps. First, the Table 1. Commercial Methods for Cyclohexanone Oxime Production Industrial & Engineering Chemistry Research Article DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 6587 http://dx.doi.org/10.1021/acs.iecr.6b00813 cyclohexanone reacts to produce the corresponding imine. Second, it is oxidized by hydrogen peroxide over TS-1 to form cyclohexanone oxime,16 achieving high conversion of cyclohexanone and high selectivity. This mechanism can be seen in Table 1. (b) Hydroxylamine route:15 TS-1 catalyzes the formation of hydroxylamine and this compound reacts with the pure ONE. Moreover, hardly any nondesirable reactions are produced.8 In order to improve this production method, the Sumitomo Chemical Co. recently patented the ammoximation process using an organic peroxide instead of hydrogen peroxide.17 This study is focused on the production of ONEOX using pure ONE and HAS. As can be seen in Scheme 1, this reaction takes place in two stages. In the first of these, the hydroxylamine sulfate is instantaneously and completely dissociated in the aqueous phase, producing a hydroxylammonium cation (HA+), which is hydrolyzed to produce HA. This is also an instantaneous reaction, the equilibrium constant being Ka. The second stage is the oximation reaction. ONE reacts with the HA to yield ONEOX. As can be deduced from Scheme 1, the pH affects the equilibrium between free HA and HA+: the higher the pH, the higher the hydrolysis of HA+ to HA.10,11 In order to control the pH with the reaction progress, the acid produced has to be neutralized. To do this, ammonium is continuously added to promote the formation of ammonium sulfate (AS).6 The global reaction is summarized in Scheme 2, where the acidity constant of HA+ dissociation can be expressed using eq 1: = + + K C C Ca H HA HA (1) The ONE oximation reaction involves two liquid phases. Hydroxylamine sulfate is very soluble in an aqueous solution, whereas cyclohexanone and cyclohexanone oxime form the organic phase;10,11 therefore, the chemical reaction and mass transfer occur simultaneously.18,19 An approach to this system, which can be found in the literature, is the work of Sharma et al. where a two phase reaction is proposed. It is a fast reaction, where an extraction is coupled with a second order reaction in a spray column.19 On the other hand, Egberink and Vanheerden (1980) had proposed a oximation reaction kinetic in an aqueous buffer solution to control the pH. ONE was diluted in the buffer aqueous phase and the HA was added as hydroxylammonium chloride. In this work, the temperature was studied within the range from −11 to 70 °C and the pH influence was also taken into account They studied the hydrolysis of ONEOX as well in order to test the reversibility of the oximation reaction.20 A model based on the two-film theory can be proposed in order to obtain a global rate expression, considering that the solubility of inorganic salts in the organic phase is negligible. On the other hand, the solubility of the organic compounds in the aqueous phase can be dramatically decreased by a high salt content in the aqueous phase (salting-out effect).21 The aim of this study is to examine the oximation reaction of pure ONE with HAS as a source of HA under real industrial conditions.22 The effects of pH, temperature, and reagent concentrations are analyzed and a kinetic model is proposed, taking the liquid phase contact into account. 2. EXPERIMENTAL SECTION 2.1. Chemicals. The following reactants or standards were used: cyclohexanone (Fluka, catalog no. 29135); cyclohexanone oxime (Aldrich, catalog no. 102202); ammonium sulfate (Aldrich, catalog no. A4418); hydroxylamine sulfate (Aldrich, catalog no. 210250); hydroxylamine solution 50 wt % in H2O (Aldrich, catalog no. 467804); 1-butanol (Aldrich, catalog no. 360465); 1,4-benzodioxane (Aldrich, catalog no. 179000); N- ethylaniline (Aldrich, catalog no. 426385). The chemicals used are summarized in Table S1 provided as Supporting Information. 2.2. Experimental Setup and Procedure. Oximation runs were carried out in an isothermal batch reactor, a chart of which is shown in Figure 1. A jacket and a pumping bath of silicon were used to achieve a constant temperature in the reactor, and the temperature was fixed using a PID controller. The melting point of cyclohexanone oxime is 74.6 °C (content 3 wt % water);10 therefore, in order to ensure the organic phases were in the liquid phase, the reactor temperature was fixed to 353 or 358 K. Besides, these values have been selected to simulate real operation conditions because these are near industrial conditions used for oximation and ammoximation of ONE.10,23 The upper temper- ature used was selected to avoid HA losses and secondary reactions. The reactor was initially loaded with around 130 g of an aqueous solution of 45 wt % HAS and heated until the reaction temperature was reached.22 The pH of this aqueous solution was about 3.2, so ammonium or sulfuric acid was added to reach the desired pH at which the reaction was carried out. The reaction started when the organic phase, which was heated separately, was added to the reactor. The runs carried out and the corresponding experimental conditions are summarized in Table 2. The stoichiometric ratio for HAS, indicated in Table 2 (CHAS,0/ CHAS,0 stc ), is calculated as the ratio between the initial concentration of HAS used and the concentration corresponding to the amount of HAS required to ensure total oximation of the ONE in the organic phase. To keep the pH constant over reaction time, the free acid formed was neutralized by feeding in an ammonium solution (25 wt %). The ammonium dosage was Scheme 1. Steps in the Oximation Reaction To Produce Cyclohexanone Oxime (ONEOX) from Hydroxylamine Sulfate (HAS) and Pure Cyclohexanone (ONE)a aKa is the acid disassociation constant of HA+ expressed in eq 1. Scheme 2. Global Reaction between Pure Cyclohexanone (ONE) and Hydroxylamine Sulfate (HAS) To Produce Cyclohexanone Oxime (ONEOX) Industrial & Engineering Chemistry Research Article DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 6588 http://pubs.acs.org/doi/suppl/10.1021/acs.iecr.6b00813/suppl_file/ie6b00813_si_001.pdf http://dx.doi.org/10.1021/acs.iecr.6b00813 controlled by an 800 Dosino supplied by Metrohm and a pH glass electrode with temperature compensation that was introduced into the reaction medium. The liquid phases in the reactor were mixed using an IKA laboratory agitator (RW 20 digital). However, due to the low agitation speed, two phases with a defined interface correspond- ing to the transversal section of the batch reactor (S) were obtained, as shown in Figure 2. The S value is also shown in Table 2. This experimental procedure yields a known value of S in rate expression development. Reaction samples were taken at different reaction times. About 1 g from the organic phase and 1 g from the aqueous phase were taken separately. The organic phase was diluted in 4 mL of BuOH and the aqueous phase was diluted in 4 mL of pure water in order to avoid salt precipitation at room temperature. The organic phase was filtered through a 40 μm filter and analyzed. Triplicate tests were performed on four runs (1−4), and the average values were used for figures with an experimental error lower than 5%. The oximation reaction is supposed to take place in the aqueous phase, so the ONE profile could be developed in the aqueous phase bulk. That could affect the model settled. To work it out, the solubility of ONE in an aqueous medium, which contains AS and HAS salts, was studied. In order to determine the solubility of ONE at reaction conditions, an aqueous dissolution of 45 wt % AS was prepared (HAS was not used in order to avoid the oximation of ONE), andmixed with an organic phase formed by ONE. The ratio of the phases used (AP to OP) was about 2. The aqueous mixture was heated to 353 K and the pH was fitted adding sulfuric acid or ammonium (pH 3.4 and 5 were tested). These conditions were settled in order to reproduce a realistic reaction system, and the pH values are within the industrial reaction pH range. Both phases were blended at 2000 rpm for 120 min. The agitation was then stopped and the mixture decanted. After 300 min three samples were taken from each phase and analyzed in order to quantify the amount of ONE in the aqueous phase and AS in the organic phase. 2.3. Analytical Methods. The cyclohexanone and cyclo- hexanone oxime concentrations in the organic phase were determined by gas chromatography with flame ionizatin detector (GC/FID; HP 6890) to obtain data using N-ethylaniline as ISTD for calibration. The organic impurities were also identified and quantified using a capillary INNOWAX column (60 m × 0.32 mm inside diameter × 0.25 μm) in a gas chromatograph− mass spectrometer (GC/MS; HP5975B). In this case, 1,4- benzodioxan was used as ISTD for calibration. Samples were analyzed in triplicate, and the error was lower than 5%. Concentrations of free sulfuric acid, SA, and HAS in the aqueous phase were determined by titration with a solution of NaOH with a known concentration, using a Metrohm pH glass electrode. Figure 1. Schematic setup to develop the oximation reaction. Table 2. Operation Conditions Used in the Oximation Runsa run T, K pH CONE,0, mmol·(kgorg) −1 CONEOX,0, mmol·(kgorg) −1 CHAS,0/CHAS,0 stc 1 353 3.50 2034 6937 4.51 2 353 4.50 2063 7049 4.42 3 353 5.51 2047 7062 4.47 4 358 3.50 2189 6951 4.20 5 358 4.50 2029 7090 4.53 6 358 5.50 2028 7091 4.52 7 353 4.50 10204 0 0.83 8 358 4.50 10204 0 0.83 aAw = 100 rpm. Waq = 0.125 kg. Worg = 0.075 kg. S = 30.9 cm2. CHAS,0 = 2471 mmol·(kgaq) −1. Figure 2. Images of phase mixture in oximation runs in Table 2. Industrial & Engineering Chemistry Research Article DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 6589 http://dx.doi.org/10.1021/acs.iecr.6b00813 3. RESULTS AND DISCUSSION 3.1. Liquid−Liquid Contact and Reagent Solubility. According to the two-film theory, if the reaction takes place in the aqueous phase, ONE is transported from the organic phase to the aqueous phase. On the other hand, the higher the cyclohexanone solubility, the higher the oximation rate in the aqueous phase. Cyclohexanone solubility in pure water reaches 8.5 wt % at 353 K.24 However, the presence of sulfate salts (HAS and AS) can affect the solubility of ONE in the aqueous phase because of salting-out effects. It was observed from the solubility runs that the solubility of ONE in the aqueous phase was lower than 400 mg/kg (0.04%) and slightly dependent on pH (the higher the pH, the lower the solubility of ONE in water), therefore proving the important effect of salting-out on the solubility of organic compounds in the aqueous phase, in accordance with the literature.21 Also, an almost negligible concentration of AS was found in the organic phase. At the agitation speed used (100 rpm), it was evaluated if there was a concentration profile of ONE and HAS along either the organic or aqueous phase. To do this, samples from organic and aqueous phases were taken in run 1 at three different heights of each phase (top, middle, and 0.2 cm from the interface for the organic phase and 0.2 cm from the interface, middle, and bottom for the aqueous phase) and analyzed. The concentration of organic species (ONE and ONEOX) in the organic phase and the concentration of HAS in the aqueous phase were measured, and the values obtained were almost independent of the distance to the interface. Therefore, in runs 1−8, while each phase is considered to be homogenized, the phases are not mixed and present a clear interface, S, as shown in Figure 2. 3.2. Influence of pH, Temperature, and Concentration on theOximation Rate.As can be seen in Scheme 1, hydrolysis of the hydroxylamine cations (HA+) to hydroxylamine (HA) depends on the pH. SinceHA reacts with cyclohexanone (ONE), the pH should affect the overall reaction rate. In order to study the effect of pH, runs 1−6 (summarized in Table 2) were carried out at pH values from 3 to 5.5. As plotted in Figure 3, the higher the pH, the higher the cyclohexanone conversion. This trend was observed for both temperatures used. It was also observed that temperature has a positive effect on the conversion of ONE. As Scheme 1 shows the reaction between ONE and HA takes place in two stages: (a) HA is produced by dissociation of HA+; (b) then, HA reacts with ONE in the oximation reaction. According to Scheme 1 and the Ka definition, the first reaction is affected by pH. If pH increases one unit, an increase of 1 order of magnitude in the concentration of HA, which is obtained by HA+ dissociation, is expected. However, analyzing Figure 3, a lower effect is experimentally observed than the expected influence. In order to explain the low pH influence on the global oximation rate, the mechanism in Scheme 3 can be considered.13,25 That conclusion was also observed by Egberink and Vanheerden, in their work. The pH effect was explained by considering that the protonated compound, which is represented in Scheme 3b, is formed.20 The oximation reaction mechanism is based on consecutive steps13,25 as shown in Scheme 3. First, a nucleophilic attack of HA to ONE (Scheme 3a) occurs, and then a dehydration step takes place (Scheme 3b). Nucleophilic attack is favored by higher pH (more free HA in the reaction medium), but the higher dehydration-step rate, the lower pH. These opposite effects of pH on the global oximation rate could explain the weaker effect of the pH observed in Figure 3 The effect of the reagent concentration was studied with the stoichiometric ratioCHAS,0/CHAS,0 stc changed for both temperatures (runs 7 and 8). Results are shown in Figure 4. As expected, the higher the stoichiometric ratio used, the lower the time required to complete cyclohexanone oximation. 3.3. Kinetic Model. The kinetic and transport models proposed were coupled in order to obtain an overall oximation rate to explain the behavior of the different species in the reaction media. As mentioned previously, the two-film theory was used,26 assuming the following simplifications: • The oximation reaction was considered to have taken place in the aqueous phase, because the HA is not experimentally found in the organic phase. Figure 3. pH influence on the remaining ONE (1 − XONE) profile in runs 1−6 at CHAS,0/CHAS,0 stc ≈ 4.5 ratio AP to OP about 2: (a) T = 353 K; (b) T = 358 K. Symbols depict experimental results, and lines depict predicted values by kinetic model using eqs 19−29. Scheme 3. Full Mechanism for ONEOximation with HA13,25a a(a) Nucleophilic attack of HA to ONE. (b) Dehydration or reaction intermediates affected by pH. Figure 4. CHAS,0/CHAS,0 stc influence on the remaining ONE (1 − XONE): (a) runs 2 and 7 at 353 K; (b) runs 5 and 8. Operation conditions in Table 3. Symbols depict experimental results, and lines depict predicted values by kinetic model using eqs 19−29). Industrial & Engineering Chemistry Research Article DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 6590 http://dx.doi.org/10.1021/acs.iecr.6b00813 • Therefore, ONEmust transfer from the organic phase to the aqueous phase.19 • Solubility of the organic compounds in the aqueous phase proved to be dramatically reduced by the high salt content in the aqueous phase due to the salting-out effect.21 Therefore, the concentration of ONE in the bulk of the aqueous phase is considered negligible in the presence of HAS, and the concentration of HAS in the film is constant and equal to the concentration value of HAS in the bulk of the aqueous phase • HAS is not soluble in the organic phase, and the ONE concentration in the bulk of the organic phase can be considered constant with distance to the interface. Based on the above assumptions, the profiles of reagents ONE and HAS are represented in the schematic chart in Figure 5. The model for the global reaction rate deduced explains well the experimental data, which means to validate the model proposed and the named assumptions. The concentration of ONE on the aqueous side of liquid film can be calculated by using a partition coefficient: α = C CONE ONE,i org ONE,i aq (2) where CONE,i aq and CONE,i org are the ONE concentrations on both sides of the interface. Moreover, in the bulk of the organic phase CONE,i org and the ONE concentration are considered equal. The reactions that take place are summarized in eqs 3−6, based on Scheme 1: r1: → +− +HAS (SO ) 2HA4 2 (3) r2: ⇄ ++ +HA HA H (4) r3: + → +HA ONE ONEOX H O2 (5) r4: + + →− +2NH (SO ) 2H AS3 4 2 (6) where r1 and r4 are irreversible and instantaneous reactions, r2 is reversible and instantaneous, and the oximation reaction r3 can be considered irreversible under the conditions used here. The equilibrium constantKa defined in eq 1 is used to calculate dissociated HA: = + + C K C CHA a HA H (7) where CHA and CHA + are expressed in mmol·(kgaq) −1. In order to relate CH + with the pH, CH + is defined in mol·(Laq) −1. The Ka values reported in the literature are quite low, about 10−6 mol· (Laq) −1.27 Therefore, under reaction test conditions, the HA concentration is much lower than the HA+ concentration and the hydroxylamine cation concentration can then be calculated by eq 8: ≈+C C2HA HAS (8) The reaction and diffusion of ONE in the aqueous film (RONE) are equal and are expressed by eq 9: − = =R r D C x d dONE 3 ONE 2 ONE aq 2 (9) where RONE and r3 are given as mmol·min−1·cm−2, DONE is the diffusion coefficient of ONE in the aqueous phase expressed as cm2·min−1, and x is the distance in the aqueous film (from the interphase to δD). The kinetic equation for r3 is obtained by considering the elementary reaction in the aqueous film: − = =R r k C CONE 3 s ONE aq HA aq (10) As the HA concentration is assumed to be constant in the aqueous film, a pseudoconstant ks′ is defined as ′ =k k Cs s HA aq (11) According to Figure 5, the boundary conditions, used to solve eq 9, are formulated as follows: = ∴ = =x C C C0 ONE aq ONE,i aq ONE org (12) δ= ∴ =x C 0D ONE aq (13) By solving the differential eq 9, the rate of mass transfer of ONE at the interface is obtained as δ − =R Ha Ha D C( ) tanhONE i ONE D ONE,i (14) Ha is the Hatta number, expressed in eq 15. δ= ′⎛ ⎝⎜ ⎞ ⎠⎟Ha k DD s ONE 0.5 (15) Because of the fast reaction rate, the Hatta number is supposed to be high and tanhHa trends toward unity. Therefore, eq 14 can be written as eq 16. − = ′R k D C( ) ( )ONE i s ONE 0.5 ONE,i aq (16) By substituting eqs 1, 2, 7, 8, and 11, and including eq 16, the ONE flux through the interface is expressed as α− = + ⎛ ⎝⎜ ⎞ ⎠⎟R k D K C C C( ) 2ONE i s ONE a HAS aq H 0.5 ONE ONE org (17) As explained previously, the lower the pH the higher the dehydration rate in Scheme 3. Therefore, ks′ must be also expressed as a function of pH: = +k k C( )n s s,0 H (18) Taking eq 18 into account, eq 17 gives Figure 5. Schematic profiles for reagents based on a simplified two-film theory. Industrial & Engineering Chemistry Research Article DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 6591 http://dx.doi.org/10.1021/acs.iecr.6b00813 − =R k C C( ) ( )ONE i ap ONE org HAS aq 0.5 (19) with an apparent constant kap defined as follows: = =+ +k k D K C a C2 ( ) ( )b b ap s,0 ONE a H H (20) kap is expressed in kg1.5·mmol−0.5·min−1·cm−2. 3.4. Mass Balances in the Batch Reactor. The mass balance for ONE in the batch reactor is expressed as − =R S w M t ( ) d( ) dONE i ONE ONE (21) where RONE,i is the flux rate of ONE through the interface S (mmol·cm−2·min−1), wONE is the total mass of ONE in the reactor, andMONE is the molecular weight of ONE. The interface S is expressed in cm2. By substituting eq 19 in eq 21 the ONE mass balance can be written as = − w M t k C C S d( ) d ( )ONE ONE ap ONE org HAS aq 0.5 (22) The HAS mass balance can be related to the ONE mass balance by stoichiometry = − ⎡ ⎣ ⎢⎢ ⎛ ⎝⎜ ⎞ ⎠⎟ ⎤ ⎦ ⎥⎥w w w M X M 1 2HAS HAS,0 ONE,0 ONE ONE HAS (23) wHAS is the mass of HAS in kilograms at each time, wHAS,0 and wONE,0 are the initial masses of HAS and ONE, respectively, and MHAS is the molecular weight of HAS. XONE is the ONE conversion, calculated with eq 24. = − X w w wONE ONE,0 ONE ONE,0 (24) The concentration of each compound at each time is calculated as the ratio of the millimoles of the j species in the mass (kg) of the corresponding phase: = =C n W p aq, orgj p j p (25) Themass of the organic and aqueous phases changes with time as a result of both the reaction and the addition of NH3 to keep the pH constant. The mass balances for both phases are shown in sections 3.4.1 and 3.4.2. 3.4.1. Organic Phase. The following equations give the mass balance for the organic phase. = +W w worg ONE ONEOX (26) = ⎛ ⎝⎜ ⎞ ⎠⎟w W C 98 10ONE org ONE 6 (27) = +⎜ ⎟⎛ ⎝ ⎞ ⎠ ⎛ ⎝⎜ ⎞ ⎠⎟w w X W C 113 98 113 10ONEOX ONE,0 ONE 0 org ONEOX,0 6 (28) 3.4.2. Aqueous Phase. The mass of the aqueous phase changes with time, due to the reaction and the NH3 added (as an aqueous solution 25% in weight) to neutralize the acid produced by the reacted ONE. The mass of the aqueous phase can be calculated at any time by the following equation: = + − + ⎜ ⎟⎛ ⎝ ⎞ ⎠ W W W W w X M M ( ) 1 0.25 aq 0 aq 0 org org ONE,0 ONE NH ONE 3 (29) The experimental ONE conversion values with time in runs 1−8 were fitted to eq 14. The experimental data were obtained from three different tests and the average values were used for fitting the kinetics parameters; standard deviations lower than errors lower than 5% were obtained. In Figure 3 and Figure 4, the error bars are provided. Nonlinear regression (Marquardt algorithm) coupled with the Runge−Kutta algorithm were used in the fitting procedure. The parameter estimation method uses least-squares as an objective function. The residual sum of squares was calculated by comparing data from the experiment with data predicted by the kinetic model. The kinetic parameters obtained by data fitting are summarized in Table 3. It also shows standard deviation errors and the percentage of explained values. Predicted values of the ONE conversion (1 − XONE) in runs 1−8, using eqs 19−29, and the kinetic parameters provided in Table 3 are plotted as lines in Figure 3 and Figure 4. As can be seen, good agreement between experimental and predicted values was found, confirming the validity of the proposed kinetic model from the mechanism assumptions. As can be seen in Figure 3 and 4, main deviations between experimental and predicted values are obtained at lower ONE conversion. This can be probably due to the higher relative error in the experimental ONE conversion at this region. The runs were carried out at an agitation speed of 100 rpm. If the reaction takes place at higher speed but within the narrow range of temperature and salt concentration in the aqueous phase used, a significant change in the value of the a parameter in Table 3 is not expected, but in the interfacial surface. 4. CONCLUSIONS A kinetic model capable of predicting the global oximation rate of ONE with aqueous solution of HAS was proposed and validated by data fitting. To do this, transport and chemical phenomena were taken into account. A kinetic approach based on the two liquid films was proposed, with the following simplifications: Solubility of HAS in the organic phase can be neglected, so the reaction takes place in the aqueous phase. The high salt content in the aqueous phase yields a low solubility of ONE in the aqueous phase (salting-out effect). Because of this, the concentration of ONE in the bulk of the aqueous phase can be considered zero. The concentration of HAS in the film can be considered equal to the value of the concentration of HAS in the bulk of the aqueous phase. With these hypotheses, the ONE profile in the aqueous film was obtained and used to calculate the flux of ONE at the interface and therefore the global disappearance rate of ONE in Table 3. Kinetic Parameter Values Estimated by Fitting of Experimental Data XONE vs t Obtained in Runs 1−8 to eqs 20 and 21 T, K (a ± CI) × 1011, kg1.5·mmol−0.5·min−1·cm−2·(mol·L−1)0.1 b ± CI SQR % VE 353 0.945 ± 0.088a −0.098 ± 0.02 0.04 98.0 358 1.060 ± 0.102a −0.105 ± 0.02 0.04 98.0 aCI, 95% confidence interval. Industrial & Engineering Chemistry Research Article DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 6592 http://dx.doi.org/10.1021/acs.iecr.6b00813 the reaction medium. The pseudoconstant, kap in eq 20, includes the kinetic constant of the oximation rate, the diffusion coefficient D, and the pH influence. The reaction orders obtained for ONE and HAS, using the model above, are consistent with the experimental data and the values deducted by the model proposed. In the pH range studied (3−5.5), it was observed that the overall oximation rate was slightly influenced by pH. This can be explained by the two stages considered in the oximation mechanism in Scheme 1 and Scheme 3. The reaction order found for the H+ concentration was about −0.1. Therefore, an increase in pH would produce a small increase in the apparent kinetic constant. However, pH values higher than 5.5 are not recommended, while the low boiling point of this species may give rise to losses of free HA. The reaction rate expression obtained is given as the flux of ONE through the interphase S, this being a crucial parameter for obtaining the ONE conversion. Low agitation was used in the runs carried out in order to control the interphase value and obtain reliable kinetic parameters. ■ ASSOCIATED CONTENT *S Supporting Information The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.iecr.6b00813. Table S1, chemicals used in this work (PDF) ■ AUTHOR INFORMATION Corresponding Author *E-mail: dlorenzo@ucm.es. Notes The authors declare no competing financial interest. ■ ACKNOWLEDGMENTS FPU grant with reference AP2012-0250 from the Spanish Ministry of Education, Culture and Sports. ■ NOTATION % VE = percentage of explained values with the kinetic model a = kinetic parameter, kg1.5·mmol−0.5·min−1·cm−2·(mol·L−1)0.1 AP = aqueous phase AS = ammonium sulfate Aw = agitation speed, rpm b = kinetic parameter Cj p = molar concentration of compound j within phase p in mmol·kgp −1 CI = 95% confidence interval Dj = coefficient of mass transfer of compound j, cm2·min−1 HA = hydroxylamine HA+ = hydroxylammonium Ha = Hatta number HAP = hydroxylammonium phosphate HAS = hydroxylammonium sulfate ISTD = internal standard compound ks = reaction rate constant for oximation reaction referred to superficial area, kg mmol−1·min−1·cm−2 ks′ = reaction rate pseudoconstant for oximation reaction referred to superficial area, min−1·cm−2 kap = apparent reaction rate constant for oximation reaction, kg1.5·mmol−0.5·min−1·cm−2 Ka = acid dissociation constant Mj = molecular weight mol·g−1 nj = moles of compound j, mmol ONE = cyclohexanone ONEOX = cyclohexanone oxime OP = organic phase rk = reaction rate for k reaction, mmol·cm−2·min−1 Rj = molar flux rate of j compound, mmol·cm−2·min−1 S = interface area of contact of liquid phases, cm2 SQR = residual sum of squares, ∑(Xi,experimental − Xi,simulated) 2 T = temperature, K wj = mass of compound j, kg Wp = total mass of phase p, kg Xj = conversion of compound j Subscripts i = interface j = compounds k = reaction number 0 = initial f = final Superscripts aq = aqueous org = organic stc = stoichiometric p = phase ■ REFERENCES (1) Belden, R. H.; Roth, J. D. W. H. (Allied Chem., USA). Production of cyclohexanone oxime. U.S. Patent US3338965, 1967. (2) Jodra, L. G.; Romero, A.; Garciaochoa, F.; Aracil, J. Impurity Content and Quality Definition of Commercial Epsilon-Caprolactam. Ind. Eng. Chem. Prod. Res. Dev. 1981, 20, 562−566. (3) Heitmann, G. P.; Dahlhoff, G.; Hölderich, W. F. Catalytically Active Sites for the Beckmann Rearrangement of Cyclohexanone Oxime to ε-Caprolactam. J. Catal. 1999, 186, 12−19. (4) Zhang, J. S.; Riaud, A.; Wang, K.; Lu, Y. C.; Luo, G. S. Beckmann Rearrangement of Cyclohexanone Oxime to epsilon-Caprolactam in a Modified Catalytic System of Trifluoroacetic Acid. Catal. Lett. 2014, 144, 151−157. (5) Kuma, K. (Mitsubishi Kasei Corp., Japan). Process for preparing oximes. Eur. Patent EP0346674 A1, 1989. (6) Rapp, G.; Fuchs, H.; Thomas, E. (Basf Aktiengesellschaft, Germany). Manufacture of cyclohexanone oxime. U.S. Patent US4031139, 1977. (7) Mantegazza, M. A.; Padovan, M.; Petrini, G.; Roffia, P. (Enichem Anic S.R.L., Italy). Direct catalytic process for the production of hydroxylamine. U.S. Patent US5320819, 1994. (8) Liu, G. Q.; Wu, J.; Luo, H. A. Ammoximation of Cyclohexanone to Cyclohexanone Oxime Catalyzed by Titanium Silicalite-1 Zeolite in Three-phase System. Chin. J. Chem. Eng. 2012, 20, 889−894. (9) Mantegazza, M. A.; Cesana, A.; Pastori, M. Ammoximation of ketones on titanium silicalite. A study of the reaction byproducts. Top. Catal. 1996, 3, 327−335. (10) Ritz, J.; Fuchs, H.; Kieczka, H.; Moran, W. C. Caprolactam. In Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-VCH Verlag GmbH & Co. KGaA: 2000. (11) Ritz, J.; Fuchs, H.; Perryman, H. G. Hydroxylamine. In Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-VCH Verlag GmbH & Co. KGaA: 2000. (12) Damme, J.; Van Goolen, J. T.; Derooij, A. H. Cyclohexanone oxime made without by-product (NH4)2SO4. Chem. Eng. 1972, 79, 54. (13) Weissermel, K.; Arpe, H. J. Industrial Organic Chemistry; Wiley: 2008. (14) Roffia, P.; Padovan, M.; Moretti, E.; De Alberti, G. Catalytic process for preparing cyclohexanone-oxime. U.S. Patent US4745221, 1988. (15) Dal Pozzo, L.; Fornasari, G.; Monti, T. TS-1, catalytic mechanism in cyclohexanone oxime production. Catal. Commun. 2002, 3, 369−375. Industrial & Engineering Chemistry Research Article DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 6593 http://pubs.acs.org http://pubs.acs.org/doi/abs/10.1021/acs.iecr.6b00813 http://pubs.acs.org/doi/suppl/10.1021/acs.iecr.6b00813/suppl_file/ie6b00813_si_001.pdf mailto:dlorenzo@ucm.es http://dx.doi.org/10.1021/acs.iecr.6b00813 (16) Reddy, J. S.; Sivasanker, S.; Ratnasamy, P. Ammoximation of cyclohexanone over a titanium silicate molecular-sieve, ts-2. J. Mol. Catal. 1991, 69, 383−392. (17) Tsujiuchi, S. (Sumitomo Chemical Co., Japan). Method for producing oxime. U.S. Patent US20120016161, 2012. (18) Zaldivar, J. M.; Molga, E.; Aloś, M. A.; Hernańdez, H.;Westerterp, K. R. Aromatic nitrations by mixed acid. Slow liquid-liquid reaction regime. Chem. Eng. Process. 1995, 34, 543−559. (19) Sharma, M. M.; Nanda, A. K. Extraction with second order reaction. Trans. Inst. Chem. Eng. 1968, 46, T44−T52. (20) Egberink, H.; Vanheerden, C. The mechanism of the formation and hydrolysis of cyclohexanone oxime in aqueous-solutions. Anal. Chim. Acta 1980, 118, 359−368. (21) Xie, W. H.; Shiu, W. Y.; Mackay, D. A review of the effect of salts on the solubility of organic compounds in seawater. Mar. Environ. Res. 1997, 44, 429−444. (22) Romero, A.; Santos, A.; Yustos, P. Effect of Methyl-δ- valerolactams on the Quality of ε-Caprolactam. Ind. Eng. Chem. Res. 2004, 43, 1557−1560. (23) Cavani, F. Liquid Phase Oxidation via Heterogeneous Catalysis. Organic Synthesis and Industrial Applications. Edited by Mario G. Clerici and Oxana A. Kholdeeva. Angew. Chem., Int. Ed. 2014, 53, 7707. (24) Pei, Y. C.; Wang, Q. B.; Gong, X.; Lei, F. Q.; Shen, B. W. Distribution of cyclohexanol and cyclohexanone between water and cyclohexane. Fluid Phase Equilib. 2015, 394, 129−139. (25) Klein, D. R.Organic Chemistry, 2nd ed.; JohnWiley & Sons: 2013. (26) Levenspiel, O. Chemical Reaction Engineering; Wiley: 1972. (27) Christian, G. D.; Dasgupta, P.; Schug, K. Analytical Chemistry, 7th ed.; Wiley Global Education: 2013. Industrial & Engineering Chemistry Research Article DOI: 10.1021/acs.iecr.6b00813 Ind. Eng. Chem. Res. 2016, 55, 6586−6594 6594 http://dx.doi.org/10.1021/acs.iecr.6b00813