Manufacture of n-Propyl Propionate Using Ion Exchange Resins: Reaction Kinetics and Feasibility of Reactive Distillation

Reaction kinetics of esterification of propionic acid with 1-propanol to produce n-propyl propionate using Amber-lyst-15 as a catalyst was investigated in this work. Effect of various parameters was studied. Initially, the efficacy of different catalysts was investigated which revealed that Amberlyst-15 can be suitably used for the reaction. This was followed by mass transfer exclusion which included effect of agitation speed, catalyst size and calculation of Weisz-Prater Criterion. Parameters selected for kinetic study were mole ratio of propionic acid to 1-propanol, temperature, catalyst loading and initial water concentration. Catalyst reusability was also verified. Aspen Custom Modeler (ACM) was used to regress kinetic parameters and a good match was obtained. Reactive Distilla-tion (RD) runs in the batch mode were performed which showed that RD is a better option for process intensification (PI) for this reaction.


Introduction
There is an increasing demand of chemical industries towards new processes or techniques that can meet requirements like generation of nearly zero waste, using less energy, while producing chemicals for variety of applications. The flavour and fragrance industry mostly uses small, volatile organic compounds to create pleasant odours or tastes, many of which are mixtures of naturally-occurring substances.
foul smell, but when these react with an alcohol, these can produce wide-ranging flavours and fragrances. Esterification of carboxylic acids with alcohols is catalyzed by either homogeneous or heterogeneous catalysts [1]. Sulphuric acid (H2SO4) is the most common homogeneous catalyst because of its effectiveness [2], but a small increase in concentration or temperature can cause dehydration of alcohol to ethers or olefins. Acidic catalysts like ion exchange resins (IERs) can also be used as an alternative [3]. IERs possess good catalytic activity and generally cause less side reactions than H2SO4. IERs have received better attention than H2SO4 because of several advantages: recovery of the catalyst by filtration [4][5]; possible continuous operation; higher purity products due to higher selectivity [4]; reduced waste disposal problems and possibility of isolation of reaction intermediates [5]. IERs like Amberlyst-15 (A-15) have played an important role in organic synthesis and its applications include: esterification, transesterification, addition, cyclization, alkylation, acylation, etc. [6][7][8][9][10][11].
Dakshinamurty et al. studied the esterification of 1-propanol (ProOH) with propionic acid (ProAc) using Dowex-50W (Scheme 1) [12]. Lilja [4]. Buchaly et al. studied modelling, simulation and process analysis of synthesis of n-propyl propionate (ProPro) by using RD with membrane separation in a hybrid process. Catalyst used in reactive zone was Amberlyst-46 (A-46). Membrane was located after distillate to remove reaction water without using entrainer [14]. Altman et al. studied synthesis of ProPro using reactive distillation (RD) with decanter separator for reactant recovery using A-46 [15]. Keller et al. studied ProPro synthesis in a RD column (RDC) coupled with a liquid-liquid phase separator using A-46 [16]. Cruz-Diaz et al. studied synthesis of ProPro in a pilot-plant RDC using A-46 [17].
Bhandare et al. synthesized isoamyl propionate by esterification of ProAc with isoamyl alcohol in batch as well as packed bed reactor using A-15 [10]. Conversions obtained were 33 and 53 %, respectively, at 343 K and 1:3 mole ratio [10]. Tekale and Yadav carried out esterification of ProAc in a batch reactor with 1,2-propanediol using a clay supported heteropolyacid Cs2.5H0.5PW12O40/K-10 as the solid acid catalyst [18]. Conversion of ProAc was ~75 % at 180 0 C [18]. Leyva [17] tion of ProAc with isoamyl alcohol in a batch reactor with Amberlyst-70 catalyst and obtained a conversion of 69 % [19]. Table 1 gives the instances of different IERs for the given reaction (ProAc reacting with ProOH to give Pro-Pro).

Analysis
Samples taken out at regular time intervals were analyzed by titration as well as by gas chromatography. Samples were titrated with NaOH solution, standardized with 0.1 N oxalic acid solution. Liquid samples were weighed and analyzed with a gas chromatograph (NUCON) using thermal conductivity detector. Porapack-Q packed column (length 1.82 m, O.D. 1/8 inch, made up of S.S.) was used for separation. For preparation of the GC samples, mixture of collected samples with known amount of ethyl alcohol (EtOH, used as internal standard) was used for analysis. From the prepared sample, 0.5 μL was injected into the column through a syringe. Oven temperature was 150 °C constant for 1 min with a ramp of 5 °C/min up to 230 °C. Injector and detector temperatures were maintained at 230 °C. Using standard calibration curves prepared for all components, the integrated areas were converted to weight % of each component in the sample. Back calculation of samples showed that the results obtained by GC were about 98 % accurate and those for titration were ~ 99 % accu-rate. Amount of sample withdrawn was very small so that it did not affect the material balance. Conversion of ProAc was calculated from Equation (1): (1)

Kinetic Experiments
When heterogeneous catalysts are used, a rather standardized methodology is adopted. First, the better catalyst among those available is selected. Secondly, experiments are conducted to ensure that mass transfer resistances are absent (for this, generally in order to assess the inter-particle and intra-particle diffusion, two experimental sets are evaluated: namely, effect of particle size and speed of agitation). Investigation of effect of various parameters on kinetics is evaluated after this (temperature, catalyst loading, mole ratio and initial water concentration). This is generally followed by catalyst reusability and reproducibility of results. An error bar analysis is also sometimes employed and reported. The procedure employed in the present study is exactly the same.  Laboratory batch reactor with 4 necks, 500 mL volume (~ 330 mL of reaction mixture) with condenser, oil bath (accuracy ±1°C) with stirrer and sampler was used. ProOH was taken in reactor and was heated till desired temperature was obtained. After the desired temperature was attained, separately preheated required quantities of ProAc and catalyst maintained at same temperature were added to the batch reactor and this was taken as "zero" time. Liquid sample was pipetted out from the reactor at regular intervals for analysis. All experiments were performed at least twice to ensure authenticity of the results.

Results and Discussion
Effect of different parameters including catalyst selection, molar ratio, temperature, catalyst loading, water concentration, particle size, catalyst reusability and stirrer speed on the reaction rate was studied.

Effect of Speed of Agitation
Effect of mass transfer on the kinetics was studied by performing the reaction at various agitation speeds. Figure 2(a) shows that speed of agitation does not affect conversion. Hence, all the experiments were conducted at 900 rpm. Figure 2(b) shows the effect of catalyst particle size on conversion. There is not much change in conversion of ProAc. Also, calculation of Weisz-Prater coefficient (CWP) was performed [24] which represents the ratio of the intrinsic reaction rate to intra-particle diffusion rate. This can be evaluated from the observed rate of reaction (-r'A(obs)), the average particle radius (RP), the effective diffusivity of the limiting reactant (De) and concentration of the reactant at the external surface of the particle (CAS):

Effect of Catalyst Particle Size
If CWP <<1 then there are no diffusion limitations and consequently no concentration gradients exist within the pellet. In the present case, the value of CWP was calculated to be 0.00231, which is much less than 1, therefore the reaction is seen to be intrinsically kinetically controlled.

Effect of Temperature
The reaction was conducted at four different temperatures (70-100 0 C) with mole ratio 1:1 (acid to alcohol) and catalyst 5% of limiting reactant. From Figure 3(a), the conversion of ProAC is observed to increase with temperature. 80 0 C was the chosen temperature at which ~ 58 % conversion was obtained.

Effect of Molar Ratio of Reactants
An increase in alcohol concentration leads to an increase in conversion as seen in Figure 3 (b). Use of excess alcohol is typical in order to shift equilibrium. It is seen that the effect is not very prominent after mole ratio of 1:2, hence the mole ratio of 1:2 was chosen for future runs.

Effect of Catalyst Loading
Reaction was carried out by varying the catalyst amount at 80 °C and mole ratio 1:1 (acid to alcohol). It is seen from Figure 3(c) that as the catalyst loading is increased, conversion of ProAc increases. The 5 % catalyst loading was chosen for further runs.

Effect of Water Concentration
Reaction was carried out with initial addition of water (Figure 3(d)). As the initial concentration of water was increased, the % conversion of ProAc decreased.

Catalyst Reusability
There is no significant decrease in catalyst activity when the same catalyst is reused even up to 3 times as seen from Figure 4. This proves the capability of IERs and also of A-15 to be reused several times. Figure 5 shows the results of the run (80 0 C, Acid to alcohol mole ratio 1:1, cat. 5 % of ProAc, 900 rpm) when conducted three times. It is seen that almost unchanged results were obtained.

Regression of Kinetic Parameters
Heterogeneously catalyzed reactions (and especially those where IERs are used), are generally represented in terms of either of the following: (a) Pseudohomogeneous (PH) (b) Eley Rideal (ER) and (c) Langmuir-Hinshelwood-Hougen-Watson (LHHW) models [25,26]. For most esterification reactions catalyzed by A-15, the LHHW type model has been found applicable. Concentration based or activity based models can be used. The present reaction scheme is a typical esterification system in which A-15 is employed. Different forms of ER and LHHW were tested and the LHHW type equation (Equation 3) using concentration was seen to provide best fit.
For regression, i.e. estimation of kinetic parameters, the following are needed: kinetic data in prescribed form (time vs. concentration of all the species involved), range of parameters covered (temperature, catalyst amount, moles, water concentration), objective function and initial guesses. There are various methods for solving estimation problems, but two are commonly employed are: weighted least square method and maximum log likelihood method, the former being generally employed. For a chosen model, parameter estimation is done based on several criteria like accuracy of the fit, parameter accuracy, etc. The given model is either accepted as such, is altered or is replaced by another model until the chosen criteria are fulfilled. Optimization methods are used to move the model parameters from initial guesses to their optimal destination by way of minimizing the objective function. The accuracy is measured using standard deviation of estimation parameter with the help of a confidence interval and certainty values.
Batch kinetics was utilized for the regression of parameters of kinetics. Following Equation (3) was utilized for this purpose: where, C is total concentration, kf is forward rate constant, kb is backward rate constant, kp is adsorption constant of ProAc, kw is adsorption constant of water and x1, x2, x3, x4 are mole fractions of ProAc, ProOH, ProPro and water, respectively. ACM was used for regression. Data related to mole ratio, temperature, catalyst loading and initial water concentration was used for regression. Table 3 shows the kinetic parameters obtained from regression. Figure 6 (a) to (d) shows the predicted concentration of ProAc compared with the experimentally observed concentration of ProAc. A good fit is indicated by R 2 value which is close to unity (Figure 7).

Reactive Distillation
RD is a combination of reaction and distillation in a single vessel which can offer many ad-   vantages over conventional method of reaction followed by separation. RD offers advantages like increased conversion, improved selectivity, better heat control and effective utilization of reaction heat [27]. Although invented in 1921, the industrial application of RD mostly took place after the 1980s [7,[27][28][29]. RD is especially useful for equilibrium limited reactions such as esterification and hydrolysis. Due to continuous removal of reaction products from the reaction zone, conversion increases beyond the equilibrium value in RD. Thus RD helps for lower consumption of resources thereby reducing capital and investment cost. Use of RD may not be feasible for every reaction, as the suitability of RD for a particular reaction depends on various factors such as volatilities of reactants and products, reaction and distillation temperatures, etc. RD can be useful for process intensification (PI). The present work has aimed at establishing the feasibility of use of RD using A-15 for this important manufacturing processes.

Laboratory reactive distillation set up
Esterification of ProAc with ProOH was studied in batch RD setup shown in Figure 8, which consists of a reaction vessel (2000 mL capacity) fitted with stop cock and thermocouple for temperature measurement. The setup consists of two glass columns of lengths 49 cm (0.49 m) and 51 cm (0.51 m) and each of diameter 51 mm (0.051 m). Teflon disks were used to facilitate sample collection, to input the reactants and for temperature indication. Catalytic reactive packings were used (Katapack, Sulzer, 30) which contained the catalysts. Dean and Stark assembly was fitted at the top along with condenser as shown ( Figure 8).

Batch reactive distillation
In a batch RD run, initially 700 mL of ProOH (1:1 mole ratio) was taken in reboiler and 500 mL ProOH in Dean and Stark assembly and then heating was started. Catalyst packings were fitted in the column, with screened A-15 catalyst (110 gm, washed and dried) with a particle size > 600 μm filled in the packing. When the reboiler temperature reached the boiling temperature, vapor rose through column. Vapors condensed in Dean and Stark assembly, which supplied the reflux to the column. ProAc was added through sample point S4 using peristaltic pump. Total 1200 mL of ProAc was added to the reactor. Column was operated with total reflux. Temperatures at intermediate stages and reboiler were noted. Samples were withdrawn from sample points S1 to S5 and were analyzed. Temperature and composition profile thus obtained was plotted as shown in Figure 9 (a) and (b), respectively. In about 5 h, 72 % conversion was achieved. In batch runs under similar conditions, conversion was about 58% in 5 h. Obviously, this is a better way of carrying out the reaction. Probably due to insufficient separation stages, sharp separation was not achieved. But it is seen that RD helps to enhance the conversion. Figures 9 (a) and (b) show the temperature and composition profile along the height of column at 5 h. It is clear that as ProAc flows towards the bottom, the temperature of top portion decreases and the bottom temperature increases. It is thus seen that use of RD helps as follows: (a). the conversion was increased in lesser time; (b). the process can be converted into a continuous one, thereby helping for PI; (c). about 70 % pure product was obtained from the bottom with ~ 24 % of acid and less alcohol and water; (d). this can be improved in a continuous RD column.
A genuine design of RD column can help to increase conversion and also separation and purification of products. Continuous RD runs are planned. A 51 mm (0.051 m) diameter, 3 meter high RD column is presently being commissioned. Detailed experimental data will be obtained and reported. Vapor liquid equilibrium (VLE) data for the different binaries in the system is being compiled and modeled and the design of continuous RD sequence will be done as a natural course of action using modelling and simulation. Experimental validation of the predictions from simulations will be done and will be communicated for consequent publication.

Conclusions
Esterification of ProAc with ProOH was carried out in this work in a batch reactor and in batch reactive distillation column. Kinetic data was obtained in batch reactor by variation of parameters like temperature, mole ratio, catalyst loading. A kinetic model was proposed and was validated through regression. Experimentation using batch reactive distillation shows its applicability. The conversions obtained were: ~ 58 % in batch reactor and ~ 72 % in RD. Genuine design and operation of RD column can help in process intensification.

CRediT Author Statement
Yogesh S. Mahajan: Conceptualization, methodology, investigation, resources, data curation, writing, review and editing, supervision; Reshma R. Devale: Partly helped in experimental investigation and in regression. She is also responsible for writing, draft preparation, Review and Editing etc.; Bipin S. Deogharkar: Experimental and simulation work, validation and initial draft of the article. All authors have read and agreed to the published version of the manuscript.