You are on page 1of 9

Chemical Engineering Journal 169 (2011) 290298

Contents lists available at ScienceDirect

Chemical Engineering Journal


journal homepage: www.elsevier.com/locate/cej

Kinetics research on fast exothermic reaction between cyclohexanecarboxylic acid and oleum in microreactor
K. Wang , Y.C. Lu, Y. Xia, H.W. Shao, G.S. Luo
The State Key Lab of Chemical Engineering, Department of Chemical Engineering, Tsinghua University, Beijing 100084, China

a r t i c l e

i n f o

a b s t r a c t
Microreactors are effective tools for the intensication of fast exothermic chemical reactions. In this work, we focus on the kinetics study of a microreacting process to provide a deeper understanding of the transport and reaction performance within microreactors. An experimental setup incorporating an online kinetic measurement method was developed based on the temperatureconversion relationship in the cyclohexanecarboxylic acidoleum reaction a crucial reaction for the preparation of -caprolactam. The reactant conversion was successfully recorded in a reaction time of less than 1.0 s, and it was determined that the reaction rate was mainly controlled by the mixing of reactants. The mass transfer coefcient in the microreactor reached 104 m/s, and the observed selectivity of the main-product was higher than 97%. Based on the experimental results, a single-droplet model was developed to establish a better understanding of the temperature and concentration distributions in the reacting system as well as analyze the effect of drop size on main-product selectivity. Crown Copyright 2011 Published by Elsevier B.V. All rights reserved.

Article history: Received 2 December 2010 Received in revised form 18 February 2011 Accepted 24 February 2011 Keywords: Microreactor Fast exothermic reaction Kinetics Liquidliquid multiphase CFD simulation

1. Introduction Fast exothermic reactions play essential roles in the chemical engineering industry for their wide applications. Although these types of reactions have appeared hundreds of years until now, people still nd they are hard to operate. Many fast exothermal reactions take place in batch reactors with low reaction efciencies, low yield and low safety. In recent years, however, the introduction of micro-structured chemical system has brought about many new advantages for those reactions, including higher product yield, higher space rate, lower energy consumption and safer operation [13]. Fast mixing can be obtained in microreactors with residence times on the millisecond level [4] while high volumetric heat transfer coefcients 10 times larger than those observed in common heat exchangers can be obtained in microcontactors [5]. Because of their excellent mixing and transport performance, many different types of chemical reactions have been intensied using microreactors, such as fast precipitation reaction for nanoparticles [6], strong exothermic polymerization for polymer materials [7], and dangerous organic synthesis with reactive reactants [8].

Corresponding author at: The State Key Lab of Chemical Engineering, Department of Chemical Engineering, Tsinghua University, Gongwu Building 477, Beijing 100084, China. Tel.: +86 10 62783870; fax: +86 10 62783870. Corresponding author. Tel.: +86 10 62783870; fax: +86 10 62783870. E-mail addresses: kaiwang@tsinghua.edu.cn (K. Wang), gsluo@tsinghua.edu.cn (G.S. Luo).

Microreactors are effective tools for improving fast exothermic reaction processes, especially for solving low selectivity problems. The efcient mixing of reactants, fast heat removal and controllable reaction time allow the overall reaction process to be controlled more effectively, resulting in a higher selectivity of the mainproducts. The FriedelCrafts aminoalkylation reaction is a classic example of a reaction system to which a microreactor can be applied. Nagaki et al. reported 92% main-product yields by using micromixing technology [9]. In a different study, Park and Kim carried out an oxidative Heck reaction in a dual-channel microreactor [10]. Their results showed signicant improvements in yield, selectivity, and reaction time in microreactors over traditional batch reactors. Jovanovic et al. reported a selective alkylation reaction of phenylacetonitrile in a 250- m internal diameter microchannel reactor [11]. They found both the conversion and selectivity in their microreactor increased signicantly compared with a stirred reactor. In our previous work, we also studied the enhancement of selectivity in fast exothermic consecutive reactions using the reaction between cyclohexanecarboxylic acid and oleum, a crucial process for the preparation of -caprolactam. Microreactors generating microdroplets in a liquidliquid multiphase reaction process were used to enhance reactant mixing and main-product selectivity which reached levels (>97%) much higher than those in common batch reactors [12,13]. Except for the advantage research of microreactors, deeply understanding the transport and reaction characteristics of microreacting process is very important for further development of microreaction technology. In this work we investigate the appar-

1385-8947/$ see front matter. Crown Copyright 2011 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.cej.2011.02.072

K. Wang et al. / Chemical Engineering Journal 169 (2011) 290298

291

Nomenclature AH+ AS by-P CCA Ci Cpalkyl CpCCA Cpoleum D Din Dout d dav d32 Ha k kC kC a kin kout Kb Km MCCA MH2 SO4 MSO3 nCCA Qoil Qoleum q qr R s T in TC in TD uT uC V xCCA x CCA xSO3 Cm r HM
r HP

protonated cyclohexanecarboxylic acid mixed anhydride by-products cyclohexanecarboxylic acid concentration, i refers to CCA, H2 SO4 , HSO4 , SO3 , AS, AH+ , by-P (mol/m3 ) heat capacity of alkyl-hydrocarbons, alkyl refers to C6 H14 , C7 H16 , C8 H18 [kJ/(kg C)] heat capacity of CCA [kJ/(kg C)] heat capacity of oleum [kJ/(kg C)] diffusion coefcient (m2 /s) diffusion coefcient in the droplet (m2 /s) diffusion coefcient out of the droplet (m2 /s) inner diameter of the reaction pipe (m) average droplet diameter (m) Sauter-mean droplet diameter (m) Hatta number (Ha =
0 m CSO Din /kC )
3

Fig. 1. Sketch view of the working system.

oleum m s
C oil oleum

thermal conductivity [W/(m C)] mass transfer coefcient of CCA (m/s) volumetric mass transfer coefcient of CCA (1/s) thermal conductivity in the droplet [W/(m C)] thermal conductivity out of the droplet [W/(m C)] equilibrium constant of proton exchange reaction equilibrium constant of mixed anhydride formation reaction molecular weight of CCA (g/mol) molecular weight of H2 SO4 (g/mol) molecular weight of SO3 (g/mol) molar transport rate of CCA (mol/s) volume feeding rate of oil phase (m3 /s) volume feeding rate of oleum (m3 /s) heat source in model equation (W/m3 ) released heat ux of the reacting system (W) reaction source in model equation [mol/(m3 s)] cross-sectional area of the mixing channel (m) temperature ( C, K) inlet temperature of continuous phase ( C) inlet temperature of dispersed phase ( C) average velocity in microreactor [uT = 4(Qoil + Qoleum )/ d2 , m/s] average velocity of continuous phase in micromixer (uC = Qoil /s, m/s) inner volume of the microreactor (m3 ) mass concentration of CCA in the feeding oil (wt%) mass concentration of CCA in the hydrolyzed oil phase (wt%) mass concentration of SO3 in the feeding oleum (wt%) concentration driving force (mol/m3 ) reaction enthalpy of mixed anhydride formation reaction (kJ/mol SO3 ) reaction enthalpy of proton exchange reaction (kJ/mol H2 SO4 ) conversion of oleum kinetic constant of the main reactions [m3 /(s mol)] kinetic constant the of side reactions (1/s) interfacial tension (N/m) viscosity of continuous phase (Pa s) density of oil phase (kg/m3 ) density of oleum (kg/m3 ) phase ratio of dispersed phase reaction time (s)

ent kinetics of cyclohexanecarboxylic acidoleum reaction in a microreactor, a topic which, up to now, has not been frequently discussed in the literature before. An online kinetic measurement method was developed to measure the reactant conversions at different reaction times. The mixing performance in the microreactor was investigated experimentally, and a single droplet model was established to show the nature of the temperature and concentration distributions in the micro-scale reacting system. 2. Experiment and simulation 2.1. Working system The reaction process between cyclohexanecarboxylic acid (CCA) and oleum, also referred to as the Premixing Reaction in the SINA process for the preparation of caprolactam [14], was selected as a representative fast exothermic reaction system. Fig. 1 shows this multiphase reaction system with a CCA dissolved alkylhydrocarbon solution, as the organic feed solution, and oleum as the sulfuric phase solution (Fig. A1). Throughout the process, the phase ratio between the oil and sulfuric acid phase was larger than 6, with the oil phase acting as the continuous phase. CCA rst diffused from the continuous phase to the dispersed phase, followed by reactions between CCA and oleum in the droplet. Two main reactions take place in this working system [13]. One is the proton exchange reaction between CCA and sulfuric acid:

(I) The other is the mixed anhydride formation reaction between CCA and sulfur trioxide:

(II) The mixed anhydride is the main-product, since it can react with nitrosylsulfuric acid to form caprolactam in the following process [15]. These two main reactions are ultra-fast reactions [16,17]. They take place almost instantaneously when the reactants meet and stop quickly when the feed stream is terminated [18]. The reactions are always shown as reversible in the literatures [19,20]. We measured the chemical equilibrium of the proton exchange reaction and determined the equilibrium constant, K b , shown in Eq. (1). According to previous studies, the equilibrium constant of the mixed anhydride formation reaction (K m ) is 104 times greater than the proton exchange reaction [12,19]. Thus, both reactions can be seemed as irreversible processes for simplication with stoichiometric excess of CCA in all operations.
Kb = CAH+ CHSO4 CCCA CH2 SO4 = 0.06 exp 13, 900 RT = 19.48.4 (20 70 C)

(1)

where AH+ refers to the protonated cyclohexanecarboxylic acid. Both reactions are highly exothermal. Maggiorotti measured their

292

K. Wang et al. / Chemical Engineering Journal 169 (2011) 290298 Table 1 Locations of the temperature sensors. T1 Distances from the micro-sieve pores (cm) 10 T2 22 T3 28 T4 36 T5 46 T6 60 T7 75 T8 90

reaction enthalpies at r HP = 13.86 kJ/mol H2 SO4 for the proton exchange reaction and r HM = 50.4 kJ/mol SO3 for the mixed anhydride formation reaction [18]. Throughout the reaction process, irreversible side reactions involving mixed anhydride begin, forming undesirable sulphonic acids such as sulphocyclohexanecarboxilic acid and benzensulphonic acid [18]. To simplify the reaction scheme we refer to all products of side reactions as by-products in this article.

(III) The kinetic rates of the side reactions are slower than those of the main reactions. Maggiorotti measured the yields of by-products according to the reaction time in a batch reactor with strong stirring (600 rpm stirring rate) [18]. Based on those results, we determined that the side reactions can be considered as rst-order reaction processes with an average reaction rate described by (Fig. A2): dCby-P d = s CAS , s = 1.8 109 exp 8.9 104 RT (2)

(30 C < T < 90 C)

where is the reaction time, by-P refers to the by-products and AS refers to the mixed anhydride. The reaction process of the working system can be quenched by adding water into the system. The protonated CCA and mixed anhydride quickly hydrolyze and revert to CCA as shown in the following schemes. The by-products, however, cannot be hydrolyzed, resulting in the partial consumption of CCA in the experiments. Using these relations, the yield of by-products can be analyzed and the main-product selectivity can be calculated. (IV)

in height. The volume of distribution region is approximately 80 L. The mixing channel and the distribution region are connected via the micro-sieve pores, which are 0.4 mm in diameter and 0.5 mm in depth. A reaction pipe with a 1 mm inner diameter and 0.3 mm wall thickness is placed at the outlet of the mixing channel to complete the reaction process. Thus, the micromixer and reaction pipe combine to form a complete microreactor. For the online measurement of reactant conversion, eight mini-sensors are placed on the reaction pipe to collect temperature data. The pipe temperature represents the uid temperature during the stable reaction process. The microreactor is completely insulated with NBRPVC rubber foam before and during the experiment ensuring that the temperature is only inuenced by the reactant conversion. This type of microreactor has already been shown to exhibit plug ow performance [23], and the reaction time can therefore be calculated by comparing the sensor location to the average velocity of the reacting uids. The locations of the mini-sensors are given in Table 1. At the end of the reaction pipe a hydrolyzer is introduced to quench the reacting process for the analysis of main-product selectivity. A microscope is also used to obtain images of the microdroplets produced in the microreactor. A polymethyl methacrylate chamber is introduced to replace part of the reaction pipe for observation. 3.1. Operation and analysis The CCA solution and oleum were delivered to the microreactor via metered pumps (Beijing Satellite Co. Ltd.). In each trial, the oil was injected into the microreactor rst followed by the oleum. The hydrolysis temperature was maintained below 45 C. During the experiment, temperatures were measured after the reaction had stabilized and samples of hydrolyzed products were collected at the outlet. Oleum conversion and mixed anhydride selectivity were the main parameters considered in this experiment. The oleum conversion was calculated by monitoring the rise in reactor temperature due to the release of heat as shown in the following equations:
T

(V) 3. Methods and setup For this microreactor kinetics study, the measurement method is very important. Due to the small size of microreactors, accurate measurement of reactant conversion with very short reaction time is very difcult. Because the sampling time may be several times longer than the reaction time, an online measurement method is most appropriate for a kinetics study. Several online methods have been proposed in the literature. In 2003, Song and Ismagilov used uorescence signals to analyze the turnover kinetics of Ribonuclease A in their microuidic chips [21]. In 2009, Han et al. measured the decomposition kinetics of H2 O2 with microelectrodes in time-controlled microchannels [22]. In this work, the relationship between reactant conversion and temperature was used. The experimental setup is given in Fig. 2. Fig. 2a is a sketch map of the experimental devices and Fig. 2b provides an illustration. The reactants come from the right side of the devices with the feed temperatures controlled by coiled pipes in a water bath. Mini-sensors (6 mm 3 mm 3 mm metal shell with PT100 thermal resistance wire in it) placed on the inlet pipes of the micromixer are connected to a data recording system (Beijing Riubohua Co. Ltd.). The micromixer is a micro-sieve dispersion device, which produces microdroplets. It consists of a mixing channel (yellowred color in Fig. 2a), a distribution region (blue color in Fig. 2a) and three microsieve pores (For interpretation of the references to color in this sentence, the reader is referred to the web version of the article.). The mixing channel is 10 mm in length, 1 mm in width and 0.5 mm

qr =

Qoil
T in
C

oil

xCCA CpCCA + (1 xCCA )Cpalkyl dT

+
T in
D

Qoleum

oleum Cpoleum dT

(3)

oleum = Qoleum
oleum

qr xSO3
r HM /MSO3

+ (1 xSO3 )

100%
r HP /MH2 SO4

(4)

Here, qr is the released heat ux; oleum is the oleum conversion; T in is the measured temperature; TC is the inlet temperature of continin is the inlet temperature of dispersed phase; Q uous phase; TD oil is the volumetric feed rate of oil (continuous phase); Qoleum is the volumetric feed rate of oleum; oil is the oil density; oleum is the oleum density; xCCA is the mass concentration of CCA in the oil feed; xSO3 is the mass concentration of SO3 in the oleum feed; CpCCA is the heat capacity of CCA; Cpalkyl is the heat capacity of alkyl-hydrocarbons; Cpoleum is the heat capacity of oleum; MSO3 is the molecular weight of SO3 and MH2 SO4 is the molecular weight of H2 SO4 . We use this denition for oleum conversion not only because of the need for online kinetic measurement, but also due to the present inability to accurately quantify oleum, mixed anhydride and protonated CCA

K. Wang et al. / Chemical Engineering Journal 169 (2011) 290298 Table 2 The heat capacities of the working system [2426]. CpCCA (20100 C) CpC6 (2060 C)a CpC7 (2080 C)a CpC8 (20100 C)a Cpoleum (20100 C)
a

293

1.67 + 0.0069T [J/(g C)] 2.16 + 0.0043T [J/(g C)] 2.14 + 0.0041T [J/(g C)] 2.16 + 0.0026T [J/(g C)] 1.82 + 0.0025T [J/(g C)]

C6, C7 and C8 refer to hexane, heptane and octane individually.

in the working system. Although this conversion is not based on the molar quantity, it still reects the actual change in reactant concentration. Heat loss from the experimental setup was tested by owing hot n-hexane through the microreactor, and the results showed that heat dissipation was less than 3% of the total enthalpy ux (the enthalpy at room temperature was set to zero). Thus heat loss was neglected in the conversion calculation. To measure the selectivity, the concentrations of CCA in the oil feed and the oil phase output with hydrolyzed products were measured using standard potentiometric titration. An automatic titrator (Shanghai Leici Co. Ltd.) with a relative error of less than 0.5% was used. A 70% (v/v) acetone/water solution was introduced as the titration solution, and CCAs equivalence point lies between pH 10 and pH 11. In hydrolyzed solutions, CCA is water-insoluble and is concentrated in the oil phase, but the sulphonic by-products are highly water-soluble substances. Thus, the molar quantity of CCA converted to by-products can be calculated by the concentration difference between the oil feed and the hydrolyzed oil output. At 100% oleum conversion, the main-product selectivity is shown as: SM = Qoil
oil

xCCA (1 xCCA )/(1 x CCA )x CCA /MCCA Qoleum


oleum xSO3 /MSO3

100% (5)
Fig. 3. The measured temperatures at different pipe positions and the oleum conversions according to the residence times.

where x CCA is the mass concentration of CCA in the hydrolyzed oil phase, and MCCA is the molecular weight of CCA. The physical properties of the working system are given in Table 2 [2426]. 3.2. Model simulation

4. Results and discussion 4.1. Measured kinetics in the microreactor A mathematical model detailing temperature and concentration elds was developed to gain a deeper understanding of the reaction process, mass transport and heat dispersion characteristics within the microreactor. The model contained combined mass and heat transport equations solved using computational uid dynamics (CFD) software COMSOL 3.4 with a nite element solver. The running time for each simulation was approximately 5 min on a personal computer with an Intel Core II CPU and 4 GB of RAM. Using the experimental setup with online temperature sensors, the reaction kinetics were measured for three working systems oleum/CCA hexane, oleum/CCA heptane and oleum/CCA octane. The nal temperature of each system did not exceed the boiling point of the solvent. The measured temperatures are given in Fig. 3. As shown on the left side of the picture, the temperature increases near the beginning of the reaction pipe but stabilizes towards the

Fig. 2. (a) The experimental setup and process; (b) picture of the microreactor.

294

K. Wang et al. / Chemical Engineering Journal 169 (2011) 290298

Fig. 6. The main-product selectivity at different nal temperatures. Fig. 4. (a) The droplet sizes and their distributions. (b) Microscope picture of the microdroplets.

by the reacting system as shown in the following equation: Qoleum nCCA = T1 oleum
oleum

end of the pipe, suggesting the reaction has reached completion inside the microreactor. In the gure, uT is the average velocity of the reacting uids in the reaction pipe. The right side of Fig. 3 shows the conversion of oleum with reaction time. The graph shows that the reaction progresses quickly requiring 1 s or less for the complete conversion of oleum. Fig. 3 also illustrates the effect of different ow conditions on the reaction rate. The reaction rate increases with the increase of ow velocity for all experimental systems and operating conditions, and the measured kinetic rate, strongly affected by the mixing process in the microreactor, is therefore the apparent kinetic rate. 4.2. Mixing performance in the microreactor For a liquidliquid multiphase process, ow velocity is a character of mixing performance in the microreactor. Since this is a mixing controlled reaction, it is important to understand exactly how mixing inuences the overall microreaction process. Liquidliquid two-phase mixing is primarily determined by droplet size and the mass transfer coefcient of the working system. In this study, we used an online CCD camera to record the owing droplets in the reaction pipe. The sulfuric acid/CCA alkyl-solutions were chosen as cold test systems to give investigations on the droplet size. Fig. 4 shows the average diameters of droplets and their standard deviations, which were obtained by counting at least 300 droplets in the recorded images with image-analysis software. Nearly all the average droplet diameters were 60 m, with almost no change resulting from the variation of ow velocity. Fig. 4 shows ow velocity has little effect on the droplet size and only affects the mass transfer coefcient of the working system. By using the measured average droplet sizes we can estimate the CCA mass transfer coefcient. Since the main reactions are very fast, almost all the transported CCA is quickly consumed by the oleum at the beginning of the process, taking place in the reaction region between the micromixer and Sensor T1 . Assuming the two main reactions have equal kinetic constants, the transported molar quantity of CCA can be calculated from the amount of heat released

xSO3 /MSO3 + (1 xSO3 )/MH2 SO4

(6)

where nCCA is the molar transport rate of CCA in the reaction region is the oleum between the micromixer and Sensor T1 , and T1 oleum conversion at the location of Sensor T1 . Since CCA is almost consumed at low oleum conversions, the CCA concentration in the droplets can be assumed to be zero. Thus, the mass transport driving forces can be calculated by the following equation: Cm =
in T1 CCCA CCCA in T1 ln(CCCA /CCCA )

(7)

in where CCCA is the incoming concentration of CCA in the oil phase, T1 and CCCA is the CCA concentration in the oil at Sensor T1 . Assuming the average droplet sizes in the cold test systems represent the average droplet sizes in the reacting system, the mass transfer coefcient (kC ) and volumetric mass transfer coefcient (kC a) of CCA can be estimated as: nCCA (8) kC 6V/d32 Cm

and kC a = nCCA V Cm (9)

where is the phase ratio of the dispersed phase, d32 is the Sautermean diameter of the droplets, and V is the inner volume of the microreactor. The calculated results given in Fig. 5 shown the mass transfer and volumetric mass transfer coefcients increasing with the increasing velocity of the uids. The mass transfer coefcient of CCA in the microreactor is in the range of 104 m/s, much higher than those in common liquidliquid processes [27]. The volumetric mass transfer coefcient, in the range of 1/s, also increased with the increase in ow velocity. This enhanced mass transport performance in the microreactor allows the reactions to reach completion quickly. 4.3. Main-product selectivity High reaction selectivity can be obtained by quickly quenching the fast reaction process. Considering that the side reactions are temperature sensitive, the advantage of using a microreactor device operating at high temperatures becomes more apparent. In our previous work we demonstrated that main-product selectivity higher than 97% can be obtained when the nal reactor temperature ranged from 45 to 65 C [12]. In this study, we show that this high selectivity is maintained even at temperatures near 90 C (Fig. 6). Obtaining high selectivity at high temperatures is desirable especially because the reactions immediately following these reactions in the SINA caprolactam production process also require high temperatures. Compared to the results obtained in a common batch reactor [18], the main-product selectivity in the microreactor is much higher.

Fig. 5. (a) The mass transfer coefcients of CCA; (b) the volumetric mass transfer coefcients of CCA.

K. Wang et al. / Chemical Engineering Journal 169 (2011) 290298

295

RAS = m CCCA CSO3 s CAS Rby-P = s CAS q = m CCCA CSO3


r HM

(15) (16)

m CCCA CH2 SO4

r HP

(17)

The kinetic equations of the mixed anhydride formation reaction and the proton exchange reaction are very important for characterizing the rate of mass generation due to reaction. However, there has been little or no information on the kinetics of these reactions available in the literature. Here in the present model, considering both the mixed anhydride formation and proton exchange reactions are ultra-fast processes, we assume them to be secondorder reactions with equal kinetic coefcients, m , as seen in Eqs. (12)(14). The value of m was estimated using the Hatta number (Ha), whose value should be higher than 3 for a mixing controlled reaction. The Ha number is dened as: Ha =
0 m CSO Din /kC
3

(18)

Fig. 7. The single droplet reaction model.

4.4. Modeling of the reaction process Knowing the reaction and mass transport details in the reacting system is crucial for the in-depth understanding of the overall microreaction process. However, due to the small sizes, measuring the temperature and concentration elds in microdroplets is very difcult, and in this work they are instead expressed with a mathematical model. For the purpose of understanding the characterized information of the reacting process in detail, a single droplet model was adopted in this work. Considering the co-directional ow of the two phases in the reaction pipe, the continuous phase can be considered static from the inertial reference frame of the droplet and vice versa. As shown in Fig. 7, the reacting ow consists of many different uid elements. Neglecting droplet coalescence, we can use a single droplet model to describe the reaction process. The model is based on one reaction element as shown in Fig. 7b. The element has an axisymmetric structure. Diffusion and heat conduction equations with sources are used to describe the mass transport and heat transport processes in the model. Ci + (D Ci ) = Ri T + (k T ) = q Cp (10) (11)

0 where Din is the diffusion coefcient in the droplet and CSO is the 3 feeding concentration of SO3 in oleum. The nal value of m chosen was 0.2 m3 /(mol s) (Ha = 3), and the calculation results changed little by increasing m beyond this value. The kinetic equation for the side reactions is given in Eq. (2), shown above. The parameters for the model equations are given in Table 3 [24,26]. Because of its small size, uid movement within the droplet is neglected and the molecular diffusion coefcient is used as shown in the table. On the outside of the droplet, the diffusion coefcient is given as the effective diffusion coefcient. An enhancement factor is introduced to represent the effect of ow on the mass transfer coefcient. Similar assumptions are made corresponding to the heat conductivities. The effect of CCA mass transport on droplet size is neglected, since the diameter expansion ratio was less than 1.2 post-reaction. However, the effect of CCA mass transport on the variation of heat capacity is considered in the model parameters. The initial and boundary conditions of the model equations are given as: For the time = 0,

CCCA = CAS = Cby-P = 0

(in droplet) (in oil phase)

(19) (20)

CSO3 = CH2 SO3 = CAS = Cby-P = 0 On the surface of the droplet,

CSO3 = CH2 SO3 = CAS = Cby-P = 0


Din CCCA = Dout CCCA kin T = kout T On the external boundary of continuous phase,

(21) (22) (23)

where i refers to CCA, H2 SO4 , SO3 , AS, AH+ and by-P, D is the diffusion coefcient, and k is the thermal conductivity. R and q are the rates of mass and heat generation represented as: RCCA = m (CCCA CSO3 + CCCA CH2 SO4 ) RSO3 = m CCCA CSO3 RH2 SO4 = m CCCA CH2 SO4
Table 3 The parameters in the model equations (2060 C) [24,26]. Diffusion coefcient D (m2 /s) Inside of droplet Din = 2.0 109 Outside of droplet Deff = DCCA = 0.5 109 a
a

CCCA = 0 T = 0

(24) (25)

(12) (13) (14)

Since oleum droplet formation in the micromixer only takes several milliseconds much shorter than the reaction time we assume no product exists initially. The partition coefcient of CCA between the two phases is near unity for the operating conditions specied (Fig. A3). Thus, the inuence of CCA concentration on the

Thermal conductivity k [w/(m C)] kin = 0.31 keff = koil = 0.17

Heat capacity Cp [kJ/(m3 C)]


oleum Cpoleum

+ (CCCA + CAS + Cby )CpCCA MCCA

xC6 CpC6

oil

+ CCCA CpCCA MCCA

DCCA = 0.5 109 m2 /s measured with metallic diaphragm cell method.

296

K. Wang et al. / Chemical Engineering Journal 169 (2011) 290298

Fig. 8. The variations of temperature and concentration eld according to the reaction time.

Fig. 9. (a) The calculation results of oleum conversion; (b) the intensied factors at different average ow velocities.

Fig. 10. (a) The variation of reaction time according to the droplet size; (b) the variation of reaction selectivity according to the droplet size.

two-phase interface is negligible and not considered. The droplet region covered by this model is shown in Fig. 7c. It is an axisymmetric structure, scattered with triangle elements. The model was solved using COMSOL 3.4 CFD software and the results are given in Fig. 8. Fig. 8 shows the variation of the temperature and CCA/SO3 concentration elds with reaction time. The temperature gradient from the inside to the outside of the reacting droplet is weak with nearly no temperature difference between the droplet and its surrounding solution for all simulation cases in this study. Thus heat generated by the reaction can be removed quickly from the reacting droplet. The CCA concentration eld shows a very large concentration gradient near the droplet surface. Taken together with the SO3 concentration eld, it becomes apparent that the concentration of CCA in the droplet is nearly zero as SO3 has not been completely consumed. Thus, the reactions mainly taken place near the droplet surface. The enhancement factor in the model (Fig. 9) was obtained using the experimental results in Fig. 3. Fig. 9a shows good agreement in a comparison between the experimental and simulation results. The enhancement factors, given in Fig. 9b, increase with the acceleration of the reacting uids. These results are in accord with the variation of the mass transfer coefcient. According to the values of the enhancement factor, mass transport processes can be intensied several times over by using microreactor ows to increase the effective diffusion coefcient.

We also analyzed the effect of droplet size on main-product selectivity using this model. Fig. 10 displays the total reaction time and main product selectivity (calculated using the by-product concentrations) when the amount of transported CCA had reached 95%. The calculated results reect the variation trend in reaction time and selectivity. The reaction time increases rapidly with the increase of droplet size, while the main-product selectivity decreases. Interestingly, the selectivity decreased slowly on the micrometer scale, and much more quickly on the millimeter scale. This phenomenon accords with the achievement of high selectivity using microdroplets with size distributions. Thus, the micro-scale mixing of the liquidliquid multiphase system is crucial for fast exothermic reactions. 5. Conclusion In this study, we designed an online measurement method and a mathematic model to study the apparent kinetics of fast exothermic multiphase reactions in a microreactor. The online measurement was based on the relationship between temperature and reactant conversion in the working system an ultra-fast, highly exothermic, liquidliquid multiphase reaction process between cyclohexanecarboxylic acid and oleum. The entire reaction process was completed in less than 1 s and the reaction rate was dominated by the mixing of the reactants. A small average droplet diameter

K. Wang et al. / Chemical Engineering Journal 169 (2011) 290298

297

(60 m), reective of the mixing scale of the system, combined with high ow velocities in the microreactor enhanced the CCA mass transport between phases. The CCA mass transfer coefcients were in the range of 104 m/s, much higher than those observed in common liquidliquid processes. Enhanced mixing performance and precise control of the reaction time yielded main product selectivity as high as 97% at nal temperatures ranging from 40 to 90 C. To gain a deeper understanding of the transport and reaction processes occurring on the droplet scale within the microreactor, a single droplet model was developed to calculate the temperature and concentration distributions in the working system. The model showed that heat generated by the reaction is removed quickly from the reacting droplet, and CCA transport from the oil phase to the sulfuric acid phase is the rate controlling step of the reaction process. The reactions mainly take place near the droplet surface due to the speed with which this fast, multiphase reaction process occurs. Using this model as an analysis tool, we studied the effects droplet size on reaction time and selectivity and determined that the micrometer scale is crucial for the enhancement of selectivity in fast exothermic multiphase reactions. Acknowledgements We would like to acknowledge the support of the National Natural Science Foundation of China (21036002, 20876084) and the National Basic Research Program of China (2007CB714302) for this work. Thanks to Chris P. Tostado, Ph.D. candidate in our group, for the English revision of this paper. Appendix A. Appendixes Considering the physical similarity of oleum and sulfuric acid, H2 SO4 CCAhexane was chosen as a cold test system in the present work. Fig. A1 gives its triangle phase diagram. The miscibility of the two phases is weak with CCA concentration lower than 70 wt%. For the kinetic study of side reactions, the experimental results of Maggiorotti are retreated in the present work [18]. It is found the side reactions can be seem as rst-order processes and their average kinetic contant with the variation of temperature is given in Fig. A2. Correlated with those kinetic contants the effect of temperature on the reaction rate can be discribed by Eq. (2) in the text. To build the reaction model, the partition of CCA between two phases is an important parameter. Using the phase diagram of H2 SO4 CCAhexane and the equilibrium coefcient of proton exchange reaction (Eq. (1)), the partition coefcients of CCA between sulfuric acid phase and oil phase can be calculated. Fig. A3

Fig. A2. The average kinetic constants of the side reactions.

Fig. A3. The partition coefcients of CCA between sulfuric acid phase and oil phase.

gives the calculation results. For the situation of CCA concentrations between 1 mol/L and 5 mol/L in oil, the operating concentrations in the experiment, the partition coefcient nearly equals to 1.0. References
[1] T. Illg, P. Lob, V. Hessel, Flow chemistry using milli- and microstructured reactors from conventional to novel process windows, Bioorg. Med. Chem. 18 (2010) 37073719. [2] E.E. Coyle, M. Oelgemoller, Micro-photochemistry: photochemistry in microstructured reactors. The new photochemistry of the future? Photochem. Photobiol. Sci. 7 (2008) 13131322. [3] K. Jahnisch, V. Hessel, H. Lowe, M. Baerns, Chemistry in microstructured reactors, Angew. Chem. Int. Ed. 43 (2004) 406446. [4] S. Verguet, C.H. Duan, A. Liau, V. Berk, J.H.D. Cate, A. Majumdar, A.J. Szeri, Mechanics of liquidliquid interfaces and mixing enhancement in microscale ows, J. Fluid Mech. 652 (2010) 207240. [5] K. Wang, Y.C. Lu, H.W. Shao, G.S. Luo, Heat-transfer performance of a liquidliquid microdispersed system, Ind. Eng. Chem. Res. 47 (2008) 97549758. [6] K. Wang, Y.J. Wang, G.G. Chen, G.S. Luo, J.D. Wang, Enhancement of mixing and mass transfer performance with a microstructure minireactor for controllable preparation of CaCO3 nanoparticles, Ind. Eng. Chem. Res. 46 (2007) 60926098. [7] T. Iwasaki, J. Yoshida, Free radical polymerization in microreactors. Signicant improvement in molecular weight distribution control, Macromolecules 38 (2005) 11591163. [8] N. Kockmann, M. Gottsponer, B. Zimmermann, D.M. Roberge, Enabling continuous-ow chemistry in microstructured devices for pharmaceutical and ne-chemical production, Chem.-Eur. J. 14 (2008) 74707477. [9] A. Nagaki, M. Togai, S. Suga, N. Aoki, K. Mae, J. Yoshida, Control of extremely fast competitive consecutive reactions using micromixing. Selective FriedelCrafts aminoalkylation, J. Am. Chem. Soc. 127 (2005) 1166611675. [10] C.P. Park, D.P. Kim, Dual-channel microreactor for gasliquid syntheses, J. Am. Chem. Soc. 132 (2010) 1010210106. [11] J. Jovanovic, E.V. Rebrov, T.A. Nijhuis, V. Hessel, J.C. Schouten, Phase-transfer catalysis in segmented ow in a microchannel: uidic control of selectivity and productivity, Ind. Eng. Chem. Res. 49 (2010) 26812687. [12] K. Wang, Y.C. Lu, H.W. Shao, G.S. Luo, Improving selectivity of temperaturesensitive exothermal reactions with microreactor, Ind. Eng. Chem. Res. 47 (2008) 46834688. [13] K. Wang, Y.C. Lu, H.W. Shao, G.S. Luo, Measuring enthalpy of fast exothermal reaction with micro-reactor-based capillary calorimeter, AIChE J. 56 (2010) 10451052. [14] L. Giuffre, E. Tempesti, M. Fornarol, G. Sioli, R. Mattone, G. Airoldi, New caprolactam process, Hydrocarbon Process. 52 (1973) 199204.

Fig. A1. The phase diagram of H2 SO4 CCAhexane system.

298

K. Wang et al. / Chemical Engineering Journal 169 (2011) 290298 [21] H. Song, R.F. Ismagilov, Millisecond kinetics on a microuidic chip using nanoliters of reagents, J. Am. Chem. Soc. 125 (2003) 1461314619. [22] Z.Y. Han, W.T. Li, Y.Y. Huang, B. Zheng, Measuring rapid enzymatic kinetics by electrochemical method in droplet-based microuidic devices with pneumatic valves, Anal. Chem. 81 (2009) 58405845. [23] G.G. Chen, G.S. Luo, S.W. Li, J.H. Xu, J.D. Wang, Experimental approaches for understanding mixing performance of a minireactor, AIChE J. 51 (2005) 29232929. [24] Company-Technic-Group, Physical Data of SINA Viscosa Process, rst ed., Shijiazhuang Chemical Fiber Co. Ltd., Shijiazhuang, 1995. [25] G.Q. Liu, L.X. Ma, J. Liu, Handbook of Physical Properties for Chemistry and Chemical Engineering Process, rst ed., Chemical Industry Press, Beijing, 2002. [26] S.W. Liu, Y. Qi, D. Liu, Y.P. Liu, Service Manual of Sulfuric Acid, rst ed., Southeast University Press, Nanjing, 2001. [27] A. Kumar, S. Hartland, Correlations for prediction of mass transfer coefcients in single drop systems and liquidliquid extraction columns, Chem. Eng. Res. Des. 77 (1999) 372384.

[15] L. Giuffre, E. Tempesti, G. Sioli, M. Fornarol, G. Airoldi, Nitrosation of pentamethyleneketene with nitrosylsulphuric acid in aqueous sulphuric acid, Chem. Ind. (1971) 10981099. [16] E. Tempesti, L. Giuffre, G. Sioli, M. Fornarol, G. Airoldi, Mixed anhydrides of cyclohexanecarboxylic acid with sulfonic acids, J. Chem. Soc., Perkin Trans. (1974) 771773. [17] E. Tempesti, L. Giuffre, G. Buzziferraris, G. Sioli, F. Serena, E. Montoneri, Investigation into kinetics of reaction between cyclohexanecarboxylic acid and oleum, Chem. Ind. 58 (1976) 247251. [18] P. Maggiorotti, The application of the reaction calorimetry to investigate reactions involving unstable compounds, J. Therm. Anal. Calorim. 38 (1992) 27492758. [19] L. Giuffre, G. Sioli, Valutazione della composizione di soluzioni concentrate di acido cicloesancarbossilico ed oleum, Chim. Ind. 51 (1969) 787794. [20] L. Giuffre, G. Sioli, E. Losio, G. Airoldi, Comportamento dellacido cicloesancarbossiliconel solvente solforico Nota II, Chim. Ind. 51 (1969) 33 37.

You might also like