Electrochemical Synthesis of d, l-Homocysteine Thiolactone

A simple reactor mathematical model was used to assess the effect of operative parameters (the initial substrate concentration, the reaction temperatu...
1 downloads 0 Views 139KB Size
2360

Ind. Eng. Chem. Res. 2007, 46, 2360-2366

APPLIED CHEMISTRY Electrochemical Synthesis of D,L-Homocysteine Thiolactone Hydrochloride in a Batch Continuous Recirculation Reactor Equipped with Carbon Felt Cathodes: A Study for the Optimization of the Process Alessandro Galia* Dipartimento di Ingegneria Chimica dei Processi e dei Materiali, UniVersita` di Palermo, Viale delle Scienze, 90128 Palermo, Italy

The electrochemical reduction of D,L-homocystine hydrochloride to the corresponding cysteine was performed in a batch continuous recirculation reaction system that was equipped with a FM01-LC electrochemical reactor. Electrolyses were performed under amperostatic conditions, using carbon felt cathodes and DSA anodes divided by Nafion 324 cationic membranes. A simple reactor mathematical model was used to assess the effect of operative parameters (the initial substrate concentration, the reaction temperature, the current density, and the linear velocity of the catholyte) on the performances of the process, with the objective of determining the optimized conditions to perform the electrosyntheses at an industrial scale with limited batch reaction time and costs for the isolation of the product. Thanks to the adoption of the three-dimensional electrode, this process can be considered to be a typical case study in which the utilization of electrochemical technology can afford concrete advantages, with respect to conventional chemical routes. Introduction One applicative scenario in which organic electrosyntheses can compete with conventional manufacturing technologies is when the economics of the process are mainly affected by the cost of separation procedures, rather than by the energy consumption. The former is particularly relevant when a highpurity product must be obtained, as usually occurs in the pharmaceutical industry, where product specifications are stringent and imposed by national regulations and, quite often, the chemical must be synthesized with high yields, because the starting material is expensive. In similar scenarios, electrochemical syntheses can become the best choice to accomplish this goal, especially if a mathematical model of the process is available such that the conceptual design and the optimization of the process can be performed with limited risks for the investor. The electrochemical cleavage of disulfide to afford the corresponding thiol is a synthetic route that has been widely adopted both in the laboratory and at the industrial scale to prepare several compounds of pharmaceutical interest, such as L-cysteine, S-carboxymethyl-L-cysteine, N-acetyl-L-cysteine and D,L-homocysteine.1-5 The electrolyses are usually performed in divided plate and frame electrochemical reactors operating in the batch recycle mode and using an aqueous concentrated HCl solution (usually 2 M) as the catholyte. As a counter reaction, chlorine or oxygen evolution can be used, with the latter being preferred if in situ utilization of the produced Cl2 is not possible. The electrolysis is commonly continued up to very high values of reactant conversion, to avoid the necessity of separating small amounts of the unconverted disulfide from the product, whose final specifications are stringent. This strategy leads to relatively * Tel./Fax: 39 0916567258. E-mail address: [email protected].

low global values of current efficiency; however, this figure of merit is not relevant in the case of processes where the cost of energy is of minor concern, so that excess molar energy consumption can be tolerated, provided that at the end of the electrolysis quantitative conversion of the substrate occurs. The electrochemical reduction of L-cystine has been studied in a small divided laboratory reactor of parallel-plate design on different cathode materials. Because, under the adopted configuration, the fractional conversion per pass through the electrolytic cell is small, the reactor was modeled as a simple batch.6,7 This simple approach to describe the reactor performances is highly valuable, because it offers a rationale to define the operative conditions that can lead to quantitative conversion of the substrate in short batch time and with acceptable electric energy consumption. In this article, a similar approach was adopted to study the optimization of the electrochemical cleavage of D,L-homocystine (Figure 1). In this case, the target product is the thiolactone derivative that is in equilibrium with the linear thiol and is a key intermediate for the production of Citiolone, which is a powerful mucolytic agent. Despite its high applicative interest, this electrosynthesis has been the object of less investigation, compared to that of the inferior homologue. In this study, the electrochemical reduction of D,L-homocystine was performed in a commercially available FM01-LC parallelplate electrochemical reactor. Constant current electrolyses were performed by dissolving the disulfide in aqueous 4.5 M HCl as the catholyte and using a carbon felt cathode. Under these conditions, both the reactant and the products (linear and cyclic form) are in the hydrochloride form. An aqueous H2SO4 solution with a concentration in the range of 2-4 M was used as the anolyte at a METCOTE ES1 301161 low-oxygen-overpotential electrode. A Nafion 324 cationic membrane was adopted to prevent the anodic oxidation of the homocysteine. During electrolysis, periodic samples of catholyte were taken and

10.1021/ie061340h CCC: $37.00 © 2007 American Chemical Society Published on Web 03/20/2007

Ind. Eng. Chem. Res., Vol. 46, No. 8, 2007 2361

Figure 1. Reaction scheme.

Figure 2. Schematic representation of the experimental apparatus adopted to perform the electrochemical cleavage of D,L-homocystine. Legend: 1, jacketed reservoirs; 2, centrifugal pumps; 3, drain valves; 4, catholyte rotameter; 5, FM01-LC reactor; and 6, current feeder.

analyzed via high-performance liquid chromatography (HPLC), to monitor the substrate concentration. Using this approach, it was possible to obtain the experimental conversion data that could be used to test the simple batch model. This was done with the specific task of defining the best process strategy to obtain quantitative conversion of the disulfide with acceptable values of global current efficiency and limited batch reaction time. Experimental Section Materials. Water that was used as the solvent was HPLC grade and supplied by Fluka. Both H2SO4 and HCl (37% w/w aqueous solutions) were analytical grade and obtained from Acros. The substrate D,L-homocystine (99%+ purity) was purchased from Fluka. Nitrogen that was used to purge the reaction system was obtained from Rivoira (99.9990% pure). Reaction System. The experimental apparatus used to perform the electrolysis of D,L-homocystine was home-built, using commercial elements, and it is schematically depicted in Figure 2. It is composed of a FM01-LC commercial laboratory filter-press electrolyzer (with a projected electrode area of 64 cm2 and electrolyte channels 16 cm long in the direction of the electrolyte flow) equipped with a GF-S6 carbon felt cathode and a METCOTE ES1 301161 anode for oxygen evolution. The carbon felt was glued onto a graphite plate, using conductive silver epoxy. The electrodes are divided by a Nafion 324 perfluorinated cationic membrane in acidic form. The membrane was prepared fresh for each electrolysis; it was expanded by soaking it in hot water for at least 60 min, and then it was allowed to cool while still immersed and

installed wet. Under the adopted configuration, the carbon felt does not completely fill the cathodic compartment and a gap of ∼0.5 cm is maintained between the membrane and the surface of the three-dimensional (3D) cathode. This gap is filled with a polytetrafluoroethylene (PTFE) mesh, which is used as a turbulence promoter. The electrolytic solutions are recirculated by Iwaki magnetically driven pumps. All experiments with the 3D electrode were performed using the same carbon felt. Electrolytic solutions were introduced to the reactor from two jacketed reservoirs, each of which had a capacity of 0.75 L and was equipped with thermometers. They also served as the heatexchange section of the system, as electrolyses were performed at 45 and 60 °C. The temperature of the heating fluid (water) was regulated using a Ultratemp 2000 Julabo F30 thermostat. The fluidic circuit was constituted, in part, by poly(vinyl chloride) (PVC) tubes (with an inner diameter of 14 mm) and partially by PTFE tubes (with an inner diameter of 8 mm). The recirculating flow rate in the cathodic circuit was measured by a rotameter that was calibrated for aqueous solutions of HCl at a concentration of 33% w/w. During the electrolyses, gas evolution occurs at both electrodes (hydrogen at the cathode and oxygen at the anode) and gaseous nitrogen was bubbled through both reservoirs, to remove these gases and to prevent the dissolution of atmospheric oxygen in the catholyte. At the end of each electrolysis and after the withdrawal of the electrolytic solutions, the cathodic circuit was carefully washed with 2 M HCl, followed by deionized water. The anodic compartment was carefully cleaned with distilled water only. Experiments were performed under constant current operation using an AMEL 555C potentiostat. The circulated charge was computed by measuring the electrolysis time, and the cell voltage was measured by a Simpson 360 electronic voltmeter. Analysis of D,L-Homocystine by High-Performance Liquid Chromatography (HPLC). At regular time intervals, weighed samples (500 mg) of catholyte were withdrawn from the reservoir and analyzed by HPLC. Measurements were performed with a Perkin-Elmer HPLC 410 LC liquid chromatograph that was equipped with an ultraviolet-visible light (UV-vis) detector LC-95 using a wavelength of λ ) 220 nm. The column was provided by Supelchem (model LC-18). The mobile phase was an aqueous buffer solution (5 mM KH2PO4, 1 mM triethylamine, pH adjusted to 2 with H3PO4) and acetonitrile in a ratio of 99:1 that was pumped at a flow rate of 0.8 mL/min. Quantitative determination of the reactant concentration in the catholyte was made using a calibration curve that was determined by analyzing disulfide solutions at known concentrations. Samples and standards were prepared for injection by dilution with the aqueous eluent. A typical plot of the substrate concentration, as a function of the time of electrolysis, is reported in Figure 3.

2362

Ind. Eng. Chem. Res., Vol. 46, No. 8, 2007

Figure 3. Typical substrate concentration profile obtained during an exhaustive electrolysis.

Reactor Model. A simple batch reactor model was adopted to fit the experimental data of substrate conversion, the detailed description of which can be found elsewhere.8 Two zones of behavior can be predicted by the model, depending on the rate-determining step at the working electrode. Initially, the reactant concentration is high and the kinetics of the process is controlled by the rate of the electron transfer at the interphase. Under amperostatic electrolysis, assuming a constant value for the intrinsic current efficiency of the process (φ0), a linear trend of the conversion X, as a function of time, is expected, according to eq 1:

X)

( )

jAφ0 t ) φ0Θ nFVC0b

(1)

Here, j is the applied nominal current density, A the surface area of the electrode, n the number of electrons exchanged per molecule of reduced substrate, F the faraday constant, V the volume of the electrolytic solution, C 0b the initial bulk concentration of the substrate. In eq 1, the adimensional time Θ is given by t/τ ) Q/Qt, where t and Q are, respectively, the actual electrolysis time and passed charge and τ and Qt are, respectively, the electrolysis time and the stoichiometric electric charge required to convert the entire amount of the substrate initially loaded in the reaction system under the hypothesis of 100% faradic yield. According to the model, as the electrolysis proceed, a critical time t* is reached where the substrate concentration has fallen to a threshold value Cb(t*), where the electrochemical reaction starts to be controlled by mass transport as the partial current density for the electroreduction becomes equal to the limiting current density:

φ0j ) jlim(t*) ) nFkmCb(t*)

(2)

From this simple equation, one can estimate the lower value of the substrate concentration or conversion (X*) beyond which the rate expression given by eq 1 cannot be used anymore. For substrate concentrations lower than the aforementioned threshold value, the kinetic expression for the electrochemical process becomes of first order and the substrate concentration decreases exponentially with time with a time constant given by V/(Akm), where km is the average mass-transfer coefficient.

Figure 4. Conversion profiles computed on the basis of the model, at different γ values and assuming an intrinsic faradic efficiency of 70%.

In this case, the conversion also can be expressed in adimensional form, in terms of X, Θ, and γ (γ ) j0lim/j ) nFkmC 0b/j), by eq 3:

ln(1 - X) ) -γΘ +

[

]

γ γ - 1 - ln 0 φ0 φ

(3)

According to this equation, a plot of ln(1 - X) versus Θ should have a linear trend whose slope can allow the operator to estimate the average mass-transfer coefficient km. Several conversion profiles, computed with previously reported equations, are depicted in Figure 4. Calculations were performed assuming an intrinsic faradic efficiency of 70% with different values of the γ parameter. It can be easily understood that the linear portion represents the maximum rate of change of conversion, coupled with the better selectivity for electron utilization, and the higher the value of γ, the wider the portion of time where the performances of the electrochemical reactor are at their best. From a physical viewpoint, the most effective and versatile parameter that can be manipulated to have high values of γ, with limited time of electrolysis, is the mass-transfer coefficient, km. High values of km, coupled with high initial disulfide loading of the catholyte, can also lead to high γ-values with high applied nominal current density (j). Electrosynthesis of D,L-Homocysteine Thiolactone Hydrochloride at Carbon Cathodes: Effect of the Electrode Morphology A set of experiments was performed at 60 °C, under constant current conditions, using both planar and three-dimensional carbon felt electrodes (see Table 1). Samples of the catholyte were taken at different electrolysis times and analyzed by HPLC to monitor the conversion of the substrate. In all experiments, a linear plot of the homocystine conversion, as a function of the adimensional passed charge, was obtained, thus showing that, under the adopted operative conditions, the process was always under charge-transfer kinetic control. From the slope of the plotted straight line, it was possible to compute the current efficiency (using eq 1). In the case of massive carbon plates, two different nominal current densities (30 and 75 mA/cm2) were chosen. Unacceptably low faradic yields were obtained on the plate electrodes also when the current density was decreased from 75 mA/cm2 to 30 mA/cm2. It can be concluded that the adoption of carbon

Ind. Eng. Chem. Res., Vol. 46, No. 8, 2007 2363 Table 1. Electrochemical Synthesis of D,L-Homocysteine: Comparison between Planar and Three-Dimensional (3D) Carbon Electrodesa

run

electrode

initial molar concentration of D,L-homocystine, C0 (M)

1 2 3

planar planar GF-S6

0.80 0.69 0.74

nominal current density, j (mA/cm2)

cell voltage, Ecell (V)

average linear catholyte velocity, V (m/s)

adimensional passed charge, Θb

intrinsic current efficiency, φ (%)

conversion, X (%)

30 75 250

3.2 3.6 4.0-4.5

0.03 0.03 0.04

0.31 0.50 0.57

32 29 74

10 15 42

a Test conditions: catholyte, 0.7 L of 4.5 M HCl; anolyte, 0.6 L of 4 M H SO ; T ) 60 °C. b Θ ) Q/Q ; evaluated with respect to the stoichiometric 2 4 t amount for a two-electron electroreduction (Qt ) 2FC0V, where V is the volume of catholyte).

Figure 5. Scanning electron microscopy (SEM) micrograph of the carbon felt electrode.

planar electrodes does not permit sufficiently high production rates, when coupled with sustainable molar energy consumption. Electrolyses with three-dimensional electrodes were performed using a GF-S6 felt with an initial thickness of ∼1 cm. The electrode was glued onto a massive carbon support and mounted in the cathodic compartment, pressed against the turbulence-promoting net. The turbulence promoter was used to improve the mass-transfer coefficients, because it has been reported that the space-averaged mass-transfer coefficient in an FM-01 electrolyzer improved by a factor of 1.7-3.8 at the electrolyte linear velocities adopted in this study.9 A micrograph of the electrode is reported in Figure 5. The electrode is composed of randomly dispersed fibers that are separated by large voids, creating an interconnected continuous space that is homogeneously distributed in the structure. In Figure 5, it is possible to observe that the single fiber has a cylindrical geometry and seems to be constituted by several thinner fibers that have been melted together longitudinally. From this visual inspection, it is evident that the material is characterized by an high specific surface area (expressed in units of m2/m3 of material), and, indeed, a specific surface area of 105 m2/m3 was estimated using the Ergun method for similar felt electrodes.10 With the porous electrocatalyst, it was possible to operate at a much higher nominal current density (j ) 250 mA/cm2), corresponding to a higher cell production rate, with a significant improvement in the current efficiency, which increased from ∼30% to 74%. It is relevant to emphasize that such an interesting result was obtained without changing the electrode material, thus avoiding the necessity of using lead, which could pose severe constraints for the final purification of the target product. Based on these results, all further experiments were performed using the carbon felt cathode. Electrosynthesis of D,L-Homocysteine Thiolactone Hydrochloride at Carbon Felt Cathodes From the applicative viewpoint, the main goal in the development of an industrial process for the electrochemical

cleavage of disulfide is to determine the better operating conditions to reach quantitative conversion of the substrate. Because the electrolysis is conducted under constant current conditions in batch systems, this result is unavoidably accompanied by a reduction in the current efficiency of the process, which leads to an increase of the molar energy consumption and selectivity problems. In the process under consideration, the only parasitic process that supplies current when the substrate concentration is depleted is the hydrogen evolution from the reduction of water molecules, and, for this reason, no problem of product purification exists. Thus, the only inconvenience to consider is the higher energetic cost of the process. Nevertheless, in the case of electro-organic syntheses intended for the production of fine chemicals, the economics of the manufacturing route can be mainly dependent on the costs for the purification of the product, especially when it must be addressed to a pharmaceutical application. This is exactly the case of the process under study. It would be optimal if the electrolysis could be performed up to almost quantitative conversion of the substrate that is the only potential impurity in the final product. If this process strategy could be coupled with sufficiently low values of the batch electrolysis time, and without decreasing the final value of the faradic yield too much, the increased energetic cost would be overcompensated by the higher space time yield of the electrocatalyst and by the reduction of the cost of product purification, which, in principle, could be limited just to solvent evaporation and drying of the product. This goal can be achieved quite easily and with a limited number of experiments, provided that a reliable mathematical model of the process is available. Indeed, the model offers a powerful tool for the conceptual design of experiments, because it clearly defines the key parameters of the process and the maximum performances that can be obtained, thus strongly decreasing the experimental effort for optimizing the synthetic route. Given the aforementioned considerations, we have tested the possibility of using the simple batch reactor model previously described to determine the optimized conditions for the electrochemical cleavage of the disulfide performed using 3D carbon felt electrodes. This model was already tested by Walsh et al. in regard to the electrochemical cleavage of the inferior homologue Lcystine6 on planar massive electrodes, and a good fit of experimental data was obtained at high hydrogen overpotential cathodes, such as mercury-plated copper and lead. These electrodes were selected to kinetically hinder the hydrogen evolution reaction, thus keeping the faradic efficiency of the cleavage reaction within acceptable values. In the case of the electrochemical synthesis of homocysteine thiolactone (HCT), the use of heavy-metal cathodes should be avoided, considering the pharmaceutical utilization of the product. For this reason, in the attempt to improve faradic yields by only changing the morphology of the carbon electrode, we have tested the applicability of the model to a 3D carbon felt cathode.

2364

Ind. Eng. Chem. Res., Vol. 46, No. 8, 2007

Table 2. Electrochemical Synthesis of D,L-Homocysteine on Carbon Felt Cathodesa,b run

initial molar concentration of D,L-homocystine, C0 (M)

temperature, T (°C)

nominal current density, j (mA/cm2)

φ0 (%)

Θmin

average linear catholyte velocity, V (m/s)

Θ*

X*

1 2 3 4 5 6

0.65 0.68 0.78 0.69 0.74 0.43

45 45 45 60 60 60

206 200 150 200 250 250

72 68 65 66 79 74

1.39 1.47 1.54 1.51 1.27 1.35

0.04 0.07 0.06 0.07 0.04 0.08

0.23 0.95 1.06 1.21 0.60 0.96

0.16 0.65 0.69 0.80 0.48 0.70

a Conditions: Catholyte, 0.7-0.8 L of 4.5 M HCl; anolyte, 0.6 L of 4 M H SO . b Figures of merit were computed from the model in the activation 2 4 control kinetic regime. φ0 ) intrinsic faradic efficiency estimated from the model for the electrosynthesis of D,L-homocysteine thiolactone hydrochloride; Θmin ) minimum adimensional charge for complete substrate conversion; Θ* ) estimated adimensional passed charge at the onset of pure mass-transfercontrolled kinetics; and X* ) substrate conversion at the onset of pure mass-transfer-controlled kinetics (see the model section).

easy to understand that the maximum global faradic yield, which corresponds to the intrinsic value φ0, could be better approached if γ f ∞, that is to say, performing the electrolysis with a high initial concentration of the substrate and high values of the masstransfer coefficient. The latter can be enhanced using a higher linear velocity of the catholyte, in the presence of turbulence promoters. This prediction is confirmed by the experimental results, because higher values of γ allowed the operator to reach almost quantitative conversions with acceptable faradic yields. In regard to the operating temperature, better performances were obtained at 60 °C, because of the cumulative effect of a higher intrinsic faradic efficiency and an improved mass-transfer rate, which is due to the decreased viscosity of the electrolytic solution. From the applicative viewpoint, it is interesting to make an estimation of the molar energy consumption (EC) for the process. This can be done from the equation11 Figure 6. Plot of ln(1 - X) (where X is the conversion) at high values of adimensional passed charge (Θ) for the experiments reported in Table 3: (O) run 1, (0) run 2, (∆) run 3, (b) run 4, (9) run 5, and (2) run 6.

A set of exploratory syntheses was performed at two different operating temperatures (45 and 60 °C), while changing the initial substrate concentration, the current density, and the linear velocity of the catholyte. According to the model, conversion profiles in the activation control region should be linear, relative to the adimensional passed charge Θ, and from their analysis, it is possible to define the limiting performances of the process. The maximum faradic yields obtainable with the adopted electrocatalyst can be estimated from the slope of the profile, and the minimum electrolysis time Θmin can be obtained from the extrapolation of the line at unit conversion. In all the experiments performed in this study, experimental conversions measured from HPLC analyses for adimensional times below the value of Θ* ) t*/τ could be fitted with good agreement by straight lines, the slopes of which give estimated current efficiencies in the range of 65%-72% at 45 °C and 66%-79% at 60 °C. These figures of merit are reported in Table 2, together with the computed values of Θmin. They are obviously limiting factors, because the onset of the mass-transfer-controlled regime cannot be avoided in a batch system. The nominal critical values of the adimensional passed charge (Θ*) and the conversion (X*) for the onset of the mass-transfer-controlled regime are listed in the same table. Figure 6 shows a plot of the experimental values of ln(1 X) as a function of Θ, computed at Θ > Θ*, where the process should be under purely mass-transfer control. Experimental points can be fitted with good accuracy by straight lines and from their slope, according to the model, it is possible to estimate the space-average mass-transfer coefficient km, the values of which are summarized in Table 3. From the model, it is quite

EC )

-nFEcell φ

During the syntheses, the cell voltage varies gradually between the values reported in Table 3. A value of 4.5 V was the maximum measured under the adopted operative conditions, and it was used, together with a global faradic efficiency of φ ) 40%, to make an overestimation of the parameter EC. The calculation gives a value of ∼0.3 kWh/mol, which can be considered to be acceptable for industrial applications, because it is quite close to that obtained in the Monsanto undivided cell process for the electrohydrodimerization of acrylonitrile,12 which, to date, is the largest-scale organic electrosynthesis that has been applied industrially. It seems relevant to emphasize once again that this result was obtained without the use of lead or mercury in the cathode: the morphology of the electrode was simply changed, going from a massive electrode to a 3D electrode. A comparison between the calculated trends of the conversion and the experimentally determined values are reported in Figure 7 for two of the electrolyses reported in Table 3, a similar agreement between the computed and experimental data was obtained for all other runs performed in this study. Some discrepancy can be observed in the intermediate region of the adimensional passed charge Θ. The model predicts a sharp transition from charge-transfer control to mass-transport control at the point (Θ*, X*); however, as also reported by Ralph et al.,6 a significant region of mixed control can exist. Considering that experiments are conducted under constant current conditions, the transition from one regime to the other can be very rapid only if the electrode potential shifts to values significantly different from the initial value to drive the secondary reaction rapidly to supply the applied current intensity. Under this condition, the electrode potential enters the mass-transfer-

Ind. Eng. Chem. Res., Vol. 46, No. 8, 2007 2365 Table 3. Electrochemical Synthesis of D,L-Homocysteine on Carbon Felt Cathodesa,b

run

initial molar concentration of D,L-homocystine, C0 (M)

temperature, T (°C)

nominal current density, j (mA/cm2)

cell voltage, Ecell (V)

average linear catholyte velocity, V (m/s)

γ

km (× 103 cm/s)

Θf

φf (%)

Xf

1 2 3 4 5 6

0.65 0.68 0.78 0.69 0.74 0.43

45 45 45 60 60 60

206 200 150 200 250 250

3.6-4.2 3.8-4.3 3.7-4.3 3.6-4.0 4.0-4.5 3.7-4.0

0.04 0.07 0.06 0.07 0.04 0.08

0.85 1.92 2.07 3.33 1.51 2.46

1.4 2.9 2.1 5.0 2.6 7.4

3.3 2.4 2.1 2.1 2.4 2.1

29 41 46 47 40 47

0.95 0.98 0.96 0.99 0.96 0.99

a Conditions: Catholyte, 0.7-0.8 L of 4.5 M HCl; anolyte, 0.6 L of 4 M H SO . b Figures of merit were computed from the model in the mass-transfer2 4 controlled kinetic regime. γ ) slope of the function ln(1 - X) vs Θ (a straight line in the pure mass control regime); km ) space average mass-transfer coefficient estimated from γ; Θf ) total adimensional passed charge evaluated with respect to the stoichiometric amount for a two-electron electroreduction (Qt ) 2FC0V, where V is the volume of the catholyte); φf ) global faradic efficiency for the electrosynthesis of D,L-homocysteine thiolactone hydrochloride; Xf ) final conversion of the disulfide.

(given as a percentage), Ast the mean peak area of HCT in the chromatograms of the standard solution, and Ws the weight of the sample (also given in milligrams). The solid was determined to be constituted by the thiolactone hydrochloride, with an assay of 98.2% (corresponding to a gravimetric yield of 93%), which makes it already commercial quality.13 The only detectable impurity was the disulfide. It must be emphasized that the conversion from a linear thiol to a cyclic thiol is a slow process under the adopted conditions, and both forms are present at the end of the electrolyses. It was observed that water removal led to quantitative conversion of the linear form to the cyclic form that is the target product of the manufacturing route. This result confirms that, by pushing the conversion to values of g0.99, it is possible to synthesize the target product at high purity just by removing the solventsupporting electrolyte system (a concentrated aqueous solution of HCl).

Figure 7. Experimental and calculated conversion profiles for selected experiments presented in Table 3.

controlled plateau region of the current-potential curve for the target process. This is the reason why, in the case of the electrochemical cleavage of L-cystine hydrochloride on bulk planar electrodes, the model satisfactorily fits the experimental data only when high hydrogen overpotential cathodes, such as mercury-plated copper and lead, are used. On these materials, a shift of approximately -200 mV of the cathode potential was measured during the exhaustive electrolysis of the disulfide at 200 mA/cm2. The potential decrease is probably not as significant in the case of the carbon felt cathode, thus leading to a more-extended mixed-control region. Despite this observation, considering the high simplicity of the model, which made it easily applicable, the accuracy of the prediction that can be obtained under the operative conditions used in this study are good enough for the purpose of conceptual design of a synthetic strategy at an industrial scale. It seems interesting to report that the catholyte that corresponds to run 4 in Table 3 (X ) 0.99) at the end of the electrolysis was evaporated under vacuum to dryness at 50 °C, obtaining a white solid. A weighed sample of it (with a mass of Ws) was analyzed by HPLC versus a standard to estimate its assay, using the simple equation

assay (%) )

AsWstS AstWs

where As is the mean peak area of HCT in the chromatograms of the sample solution, Wst the weight of the reference standard (given in milligrams), S the strength of the reference standard

Conclusions D,L-Homocysteine hydrochloride was synthesized using a three-dimensional (3D) carbon felt cathode in a divided parallelplate electrochemical reactor that was operating at constant current in the batch recycle mode. A simple mathematical model was adopted to optimize the process. The model suggests that, to obtain the thiol with quantitative yields, acceptable molar energy consumption, and limited batch reaction time, it is necessary to adopt operative conditions that lead to high values of γ (the ratio between the initial limiting current density j0lim and the applied current density j; γ ) j0lim/j). Experimental results confirm this prediction, even if some discrepancy with the model has been detected at intermediate values of the adimensional passed charge Θ, presumably as a consequence of the occurrence of a mixed-control region during the operation of the reactor. From the applicative viewpoint, industrial electrosyntheses must be conducted with a high initial concentration of the substrate and high catholyte flow rates, using a turbulence promoter to improve the mass-transfer coefficient. Also the operating temperature can have a significant role in enhancing the mass-transfer kinetics inside the reactor. Planar carbon electrodes lead to poor figures of merit; however, quite interestingly, the aforementioned good reactor performances were obtained simply by changing the morphology of the cathode material passing through a 3D carbon felt electrode without the necessity of using heavy toxic metals to slow the hydrogen evolution from the background electrolyte. When γ values of g2.5 were achieved, substrate conversions of 99% with global faradic yields close to 50% were obtained, leading to a product that, after solvent-supporting electrolyte evaporation, was constituted by D,L-homocysteine thiolactone hydro-

2366

Ind. Eng. Chem. Res., Vol. 46, No. 8, 2007

chloride with an assay of >98%. This product can be classified as a fine chemical and, thus, be characterized by high added value and limited volume of production; therefore, the excellent figures of merit obtained strongly suggest that the electrochemical route can become the election choice for its preparation at the industrial scale. Literature Cited (1) Ralph, T. R.; Hitchman, M. L.; Millington, J. P.; Walsh, F. C. The electrochemistry of L-cystine and L-cysteine. Part 1: Thermodynamic and kinetic studies. J. Electroanal. Chem. 1994, 375, 1. (2) Ralph, T. R.; Hitchman, M. L.; Millington, J. P.; Walsh, F. C. The electrochemistry of L-cystine and L-cysteine. Part 2: Electrosynthesis of L-cysteine at solid electrodes. J. Electroanal. Chem. 1994, 375, 17. (3) Sa´nchez-Cano, G.; Perez, J. R.; Montiel, V.; Aldaz, A. Anal. Quim. 1989, 85, 526. (4) Sa´nchez-Cano, G.; Montiel Leguay, V.; Elias Vilarelle, E.; Ginebreda Martin, A.; Aldaz Riera, A. Eur. Pat. Appl. 0618312A1, 1994. (5) Aldaz, A.; Montiel, V.; Gonza´les-Garcı´a, J.; Garcı´a-Garcı´a, V. Electrochemical synthesis of L-cysteine derivatives. Presented at the 11th International Forum on Electrolysis in the Chemical Industry, Clearwater, FL, November 2-5, 1997. (6) Ralph, T. R.; Hitchman, M. L.; Millington, J. P.; Walsh, F. C. Evaluation of a reactor model and cathode materials for batch electrolysis of L-cystine hydrochloride. J. Electroanal. Chem. 1999, 462, 97.

(7) Ralph, T. R.; Hitchman, M. L.; Millington, J. P.; Walsh, F. C. The importance of batch electrolysis conditions during the reduction of L-cystine hydrochloride. J. Electrochem. Soc. 2005, 152, D54. (8) Danly, D. E. Emerging Opportunities for Electroorganic Processes; Marcel Dekker: New York, 1984; pp 116-120. (9) Brown, C. J.; Pletcher, D.; Walsh, F. C.; Hammond, J. K.; Robinson, D. Studies of space-averaged mass transport in the FM01-LC laboratory electrolyser. J. Appl. Electrochem. 1993, 23, 38. (10) Gonza´les-Garcı´a, J.; Bonete, P.; Expo´sito, E.; Montiel, V.; Aldaz, A.; Torregrosa-Macia´, R. Characterization of a carbon felt electrode: structural and physical properties. J. Mater. Chem. 1999, 9, 419. (11) Pletcher, D.; Walsh, F. C. Industrial Electrochemistry; Blackie Academic & Professional: Glasgow, U.K., 1990; pp 74-75. (12) Danly, D. E. Adiponitrile via improved EHD. Hydrocarbon Process. 1981, 60, 161. (13) AK Scientific, Inc. Technical information regarding D,L-Homocysteine Thiolactone Hydrochloride. http://www.aksci.com/ search2.asp?message2)568682.

ReceiVed for reView October 18, 2006 ReVised manuscript receiVed February 9, 2007 Accepted February 10, 2007 IE061340H