Electrocarboxylation in supercritical CO
2
and CO
2
-expanded liquids
S. Chanfreau, P. Cognet
∗, S. Camy, J.-S. Condoret
Laboratoire de G´enie Chimique, Site Basso-Cambo, 5 rue Paulin Talabot, BP 1301, 31106 Toulouse cedex 1, France
Keywords: High pressure Supercritical CO2 Expanded liquids Electrocarboxylation Benzyl chloride Sacrificial anode
a b s t r a c t
In this study, the electrocarboxylation of benzyl chloride in pressurized CO2, or pressurized mixtures of dimethylformamide (DMF) and CO2, was investigated in order to synthesize phenylacetic acid. A stainless steel cathode was used as the working electrode, whereas a sacrificial massive magnesium rod or a platinized platinum grid was used as the anode, tetrabutylammonium perchlorate (TBAP) or tetrakis(decyl)ammonium tetraphenylborate (TDATPhB) being the supporting electrolyte. The electrocar-boxylation was carried out at 40◦C, at operating pressures of 1, 6, 7, 8, 9 and 12 MPa, using current densities ranging from 0.1 to 150 mA cm−2. It was found that a small amount of DMF was necessary to ensure the solubility of the supporting electrolyte, to obtain sufficient electrical conductivity of the medium. The best results were obtained using the magnesium sacrificial anode, at 6 MPa. After consumption of the the-oretical amount of electrical current (2F mol−1), 65.7% benzyl chloride conversion was reached, together with an 82.4% phenylacetic acid selectivity and a 54.2% faradaic yield. Detected by-products were toluene, bibenzyle, benzyl alcohol and benzaldehyde.
1. Introduction
Carboxylation reactions involve reactions between an organic molecule and carbon dioxide. They are generally achieved by means of a Grignard reaction, which consists in making an aliphatic or aro-matic halide to react with magnesium to form a Grignard reagent. The latter can then further react with carbon dioxide to form the desired carboxylate. Several drawbacks are associated to this well-known chemical pathway: (i) the Grignard reaction is often carried out in diethylether, a highly flammable solvent, (ii) water traces have to be avoided because they would lead to the destruc-tion of the Grignard compound and (iii) CO2 bubbling has to be
performed very rapidly after the first step, because of the weak sta-bility of the Grignard intermediate. In this last step, mass transfer limitations in this gas–liquid system, constitute an engineering dif-ficulty, especially when one intends to perform the reaction at large scale.
In this context, it is of interest to develop alternative carboxy-lation routes, preferably based on Green Chemistry principles. Indeed, since the 1980s, numerous attempts have been made to use the electron as a non-polluting and cheap reagent, and electro-chemical routes were proposed to replace conventional dangerous and polluting synthesis processes[1–3]. Among the investigated
∗ Corresponding author. Tel.: +33 5 34 61 52 60; fax: +33 5 34 6152 53. E-mail address:[email protected](P. Cognet).
reactions, promising results were obtained concerning the electro-carboxylation as an alternative to classical chemical electro-carboxylation route. Electrocarboxylation involves the electroreduction of an organic halide into a carbanion, which further reacts with car-bon dioxide to form the carboxylate. Both reactions are performed simultaneously, in an electrochemical device where gaseous CO2
is bubbling in a solvent, containing the reactant and a supporting electrolyte. One crucial issue is then the control of the oxidation reaction at the counter electrode. One solution consists in using a sacrificial anode, as, for instance, magnesium, a strongly electropos-itive metal[4,5]. In this case, magnesium is oxidized at the anode into Mg2+ions, thus preventing from anodic degradation reactions.
A magnesium rod as the sacrificial anode is often chosen as the more suitable material electrode for operation in aprotic media[5]. In the case of aromatic halides (RCl), the reactions taking place at the electrodes are:
At the anode: At the cathode:
Mg→ Mg2++ 2e− RCl + CO2+ 2e−→ RCO2−+ Cl−
So the overall reaction can be written as follow: Mg+ RCl + CO2→ RCO2−+ Cl−+ Mg2+
The metallic sacrificial anode is consumed and conjugated car-boxylates are obtained under the form of magnesium salts. Sacrificial magnesium anodes were shown to be suitable to elec-trocarboxylate mono-halogenated aromatic compounds[6,7]and poly-substituted aromatic chlorides[8]. In such processes, CO2
saturation tank. Increase of CO2 solubility was also obtained by
increasing CO2pressure[5].
Using this technique, a non-compartmented cell can be used. Such configuration also allows minimizing the amount of salt required to ensure the conductivity of the medium since, as elec-trolysis proceeds, magnesium salts are formed. With this technique, good results were obtained at laboratory scale[9,10], and led to the development of a new type of pilot scale reactor: the “V elec-trochemical reactor”[11]. The main characteristic of this device is to keep a constant spacing between both electrodes, in spite of the magnesium consumption. Another technical advantage was the possibility to work under increased CO2pressure, up to 5 bar,
allowing increasing the CO2concentration in the medium. Using
dimethylformamide (DMF) as the solvent, such a process was envisaged for pharmaceutical applications, and was successfully tested for the synthesis of anti-inflammatory compounds such as Naproxen, Fenopren or Ibuprofen [12]. Unfortunately industrial production was not developed because of difficulties in the purifi-cation of the product, arising from to the presence of impurities, generated by the degradation of the solvent.
Indeed, the main obstacle towards industrialisation of the elec-trocarboxylation process is the use of DMF as the solvent. Firstly, it is toxic and very hygroscopic. Secondly, from an economical point of view, it is expensive and requires purification steps before use. Finally, its degradation during electrolysis generates impurities. So, it is of interest to avoid, or at least reduce the use of such a solvent in the perspective of sustainable engineering develop-ment. Liquid or supercritical CO2is today considered as a promising
alternative to the use of toxic and expensive conventional organic solvents. As high pressure technology is today better and better mastered, the use of such rich CO2 media could be envisaged to
design green processes. In addition, CO2 is cheap, non-toxic and
has a relatively low critical temperature and pressure (TC= 31◦C;
PC= 7.383 MPa). A few attempts to carry out electrochemical
reac-tions in liquid or supercritical CO2 have already been reported
[13,14]. Especially, a few results were obtained for
electrocarboxy-lation in liquid rich CO2media[15,16]. In the present work, results
obtained for electrocarboxylations carried out under high CO2
pres-sure are described, and, in this case, CO2is used as a reagent and
a solvent. Benzyl chloride (PhCH2Cl) was chosen as a model
sub-strate. The major problem encountered with electrochemistry in scCO2 is the solubility of electrolytes, due to its very low
per-mittivity (ε < 2). So, a small amount of a polar co-solvent, such as DMF, is always required. Nevertheless, one of the objectives of this work was to indeed reduce the amount of this organic sol-vent, by combining it with pressurized carbon dioxide. Recently, such configurations, termed CO2-expanded liquids (CXLs), were
also proposed to perform catalytic reaction[17–19]. Indeed, CXLs are organic solvents, where dense CO2 is dissolved at rather mild
pressure (a few tens of bar). In this case a tunable property medium (gas or catalyst solubility, reduced viscosity. . .) can be obtained. Eventually, this results in the use of smaller quantities of problem-atic compounds.
In our previous study[20], the phase diagram of CO2–DMF
mix-ture under high pressure was presented, and these results were used here for the choice of conditions for the electrocarboxylation of PhCH2Cl in CO2-expanded DMF liquid. Tetrabutylammonium
perchlorate (TBAP) was chosen as the supporting electrolyte
[21]. As proposed by Abbott and Harper [22], some experi-ments were also run using a very specific hydrophobic salt, tetrakis(decyl)ammonium tetraphenylborate (TDATPhB). Finally, some experiments were run using CO2-expanded PhCH2Cl liquid,
without any use of DMF. In this case, in order to evaluate PhCH2Cl
and CO2mutual solubility, high pressure CO2–PhCH2Cl phase
equi-libria must be estimated. In this study, in addition to pressure,
support electrolyte and solvent modifications, current density and electrodes materials were studied.
2. Experimental 2.1. Reagents
Benzyl chloride (Sigma–Aldrich, 99% purity), tetrabutylammo-nium perchlorate (TBAP) (Fluka, electrochemical grade 99% purity), sodium tetraphenylborate (Fluka, ACS reagent,≥99.5% purity) and tetrakis(decyl)ammonium bromide (Fluka, 99.0% purity) were used as received. DMF (Sigma–Aldrich, 99.9% purity) was dried over molecular sieves before use. CO2 cylinder containing a dip tube
was provided by Air Liquide.
Tetrakis(decyl)ammonium tetraphenylborate (TDATPhB) was synthesized by previously described methods [22]. In order to remove residual traces of the solvent, used in the salt synthesis (acetone, in this case), the produced salt was contacted with a con-tinuous flux of supercritical CO2 (200 bar, 40◦C) during 1 h, in a
conventional supercritical extraction device (not described here). 2.2. high-pressure electrochemical cell
The high-pressure electrochemical cell, a 250 mL thermostated jacketed vessel made of stainless steel (Parr Instrument Com-pany, USA), and electrochemical analytic electrode system were described in our previous works[20,21]. Two sapphire port-holes, facing each other, allowed observation inside the reactor. High pressure CO2was supplied by a pneumatic pump (Top Industrie,
France), equipped with a pressure regulation.
2.3. Electrodes used for the electroanalytical study of the carboxylation of PhCH2Cl
Such study is done using a micro electrode. A three-electrode system was chosen to carry out experiments in a single cell reac-tor. A 100m diameter platinum wire (Goodfellow) sealed in a 3 mm diameter glass tube was used as the working micro elec-trode. Before every experiment, the tip of this electrode was polished with a 0.3m roughness abrasive band. The counter-electrode was a platinized titanium grid (2 cm× 2 cm approx.). A silver wire pseudo-reference electrode (Goodfellow, 99.99% purity 1 mm diameter wire) was fixed near the working electrode glass tube in order to minimize the distance between both electrodes. PTFE electric wire feedthrough made it possible to ensure the seal-ing of the reactor.
Voltamperometric measurements were run on a Voltalab 32 potentiostat (Radiometer, France). This DEA-I-type digital electro-chemical analyzer is divided in two parts, an IMT 102 interface and a 33V/1A DEA 332 potentiostat. Data acquisition is ensured by Voltamaster 2 electrochemistry software.
2.4. Electrodes used for the reaction of PhCH2Cl
A silver wire pseudo-reference electrode (Goodfellow, 99.99% purity 1 mm diameter) was also used (Ag ref) in this case. A coax-ial configuration of the anode and the cathode was chosen, held up by a home-made Teflon electrode support. The central anode was either a magnesium massive rod (Prolabo Rhˆone-Poulenc Ltd., 99.8% purity 10 mm diameter), or a home-made platinized titanium grid tube-like electrode (METAL DEPLOYE S.A., 1 mm thickness, 0.17 cm2cm−2 frontal transparency). 17 mm diameter
cathodes were chosen in order to minimize the inter-electrode spacing and a home made 0.35 cm2cm−2 frontal transparency
or an expanded grid tube-like electrode (METAL DEPLOYE S.A., 55± 0.8 mm height, 0.5 cm2cm−2frontal transparency) were
cho-sen as the 316 L stainless steel electrode. 2.5. Procedure for the synthesis
A typical solution was composed of 25 mL of DMF, at 0.1 M in PhCH2Cl and TBAP. Same TBAP support electrolyte concentration
was employed when PhCH2Cl was used as the solvent. When TBAP
was substituted by the hydrophobic salt, 19 mM of TDATPhB were added to the solvent.
All freshly prepared solutions were degassed under CO2gas flow
before experiments. Pressurization of the apparatus was run as previously described[20,21].
2.6. Analytic procedure
After electrolysis completion, resulting solutions were acidified with 37% (w/w) aqueous hydrochloric acid, and extracted two times with diethyl ether. Samples of the obtained organic phase were ana-lyzed by Gas Chromatography (GC) where biphenyl was chosen as the internal standard. Gas Chromatography was performed using a capillary column (Hewlett Packard HP-5, cross linked 5% Ph Me Sil-icone, 25 m× 0.2 mm × 0.33 m) operated in split mode, and a FID detector was used. Injector temperature was set to 270◦C. Oven temperature was set to 70◦C during 5 min. Then it was increased (40◦C/min) up to 290◦C (5 min plateau). Prior to experiments, response coefficients of all products, with respect to biphenyl, were determined.
2.7. Criteria for electrosynthesis performance
As conventionally proposed[21], conversion rate (C), selectivity (S) and faradaic yield (F) are defined as follows:
C(%) =n i reagent− nfreagent ni reagent ; S(%) = n f product ni reagent− nfreagent ; F(%)= 2F · nf product Q whereni
reagent is the initial quantity of reagent (mol), (nireagent−
nf
reagent) is the quantity of consumed reagent (mol), nproductis the
quantity of the desired product (mol), F is the Faraday’s constant (C mol−1) and Q is the transferred quantity of electricity (C).
The quantity of electricity for the desired product synthesis is represented byF. It is obtained by multiplying C (%) by S (%) when
2F mol−1are transferred.
3. Results and discussion
3.1. Electroanalytical study of the carboxylation of PhCH2Cl
For the reaction at the cathode, the following mechanism is conventionally admitted for the electrochemical carboxylation of aromatic halides (ArX)[22]:
ArX−→ArX+e− −• (E) ArX−•Rapid−→ X−+ Ar• (C)
Ar• +e−→Ar− − (E)
Fig. 1. Cyclic voltammetry at different scan rates of a N2degassed DMF/0.1 M
TBAP solution containing PhCH2Cl. [PhCH2Cl] = 0.1 M, T = 24◦C. Working
elec-trode: 100m diameter platinum microelectrode, counter-electrode: 2 cm × 2 cm (approx.) platinized titanium grid, reference: silver wire pseudo-reference electrode.
The Ar− reduced species are obtained by one reaction step, fol-lowing an Electrochemical–Chemical–Electrochemical (ECE) three step mechanism, the reaction being limited by the electrochemi-cal steps. The chemielectrochemi-cal step is the fastest but it can be slowered depending on the presence of electron acceptor groups on the aro-matic ring[22]. The reaction is ended by the carboxylation chemical step.
The electrochemical reduction of aromatic halides is irre-versible, according to the overall reaction. It was confirmed by cyclic voltammetry as shown inFig. 1.
Benzyl chloride (PhCH2Cl) was chosen as the studied compound.
Peak-shaped voltammograms were obtained at different scan rates. No backward oxidation peak was observed and so the electroreduc-tion of PhCH2Cl was shown to be non-reversible (Fig. 1).
According toFig. 2, the electrochemical reduction of CO2was
determined at potential values lower than−3000 mV/Ag reference. A wave, at potentials close to−2000 mV, was also observed, cer-tainly due to the reduction of water traces. Reduction potential of PhCH2Cl was found to be close to−2600 mV. As PhCH2Cl is reduced
before CO2, working potential should be chosen close to−2600 mV,
so that only PhCH2−is generated, without reducing CO2, thus
avoid-ing the generation of by-products comavoid-ing from CO2reduction.
Fig. 2. Linear voltammetry at different scan rates of a CO2saturated DMF/0.1 M
TBAP solution containing PhCH2Cl. [PhCH2Cl] = 3.25× 10−3M, T = 20◦C. Work
elec-trode: 100m diameter platinum microelectrode, counter-electrode: 2 cm × 2 cm (approx.) platinized titanium grid, reference electrode: silver wire pseudo-reference electrode.
3.2. Results for electrolytic synthesis 3.2.1. Choice of operating conditions
With a view of possible industrial development, constant inten-sity operation was chosen, rather than using constant potential operation. Indeed, for constant current operation, potential of the working electrode may vary during electrolysis, consequently, selectivity is less controlled. The influence of different parameters, such as CO2pressure, electrode material, reagent and DMF
concen-trations, current density (see experimental conditions inTable 1), was investigated.
In most cases, 40◦C and 6 MPa were chosen, respectively, as the operating temperature and pressure (Table 1, entries 2 and 5–8). Such temperature and pressure conditions were shown to be suit-able from our previous study[20]upon electrochemical analysis of ferrocene under similar conditions. In this study, it was shown that ferrocene diffusion coefficient value was 5 times higher at 60 bar than at 10 bar. The same effect on benzyl chloride diffusion and on mass transfer may be expected.
From these previous calculations[20], we were able to estimate the respective proportions and compositions of vapour and liquid phases in the reactor, as shown inTable 2.
From this table, it can be seen that, when increasing the pressure, proportions of the DMF rich liquid phase increases in the reactor, but concentration of CO2also increases in this phase, leading to a
decrease of its density.
Influence of current density and electrode nature were also studied, together with the influence of the supporting electrolyte. In particular, TDATPhB was used as a hydrophobic support elec-trolyte to run experiments with PhCH2Cl being simultaneously the
co-solvent and the reagent. Experimental results are presented in
Table 1, and commented in the following paragraph.
3.2.2. Influence of the pressure
Experiments at 1, 6, 7 and 8 MPa were carried out under the same experimental parameters (Table 1, entries 1–4).
At the end of the experiments, it was visually observed that only the part of the magnesium rod located in the liquid phase was consumed, revealing that the reaction only took place in the liquid phase. This was explained by very low TBAP solubility in the CO2-rich phase. Due to the already mentioned increase of the
liquid phase volume, associated with the pressure increase, there was a dilution of the electrolytic species in the liquid phase, and their effective concentrations were reduced. From our computa-tions of the volume increase, we estimated from that concentration dropped from 0.1 M, at ambient pressure (entry 1) to 0.028 M at 6 MPa (entry 2).
Table 1 showed that, although conversion is quite low (less
than 40% in entries 1–4), phenylacetic acid was found to be the main synthesized product, and few by-products, namely toluene and bibenzyle, were obtained. Considering the poor conversion and selectivity obtained above 6 MPa, an optimum in pressure seems then to exist, around 6 MPa, for this electrocarboxylation reaction. The performances of the reaction seem to be improved when increasing CO2 pressure, providing the reacting mixture is
rich enough in DMF to insure TBAP solubilization. Indeed, when the whole reacting mixture becomes a single-phase system, then DMF concentration is too low, and this leads to precipitation of the salt, and therefore to a sharp decrease of the conductivity of the mixture. In this case, very low faradaic yield were obtained (entries 11–14).
From these considerations, 6 MPa was chosen as the pres-sure for the following experiments, where it enpres-sures a two-phase system, in which reaction can occur in the CO2-expanded DMF
phase. Table 1 Experimental par ame ters and results of PhCH 2 Cl electr ocarbo xy lation Entr y Pr essur e (MP a) Electr ode mat erials Initial liq uid phase Curr ent density i(mA cm − 2) Electricity Q (F mol − 1) Con v ersion C (%) Selecti vity S (%) T oluene (%) Bib enzy le (%) Benzaldeh yde (%) Benzy l alcohol (%) Far adaic yield F (%) Anode Cathode Co-sol v ent v olume Support electr ol yt e concentr ation [PhCH 2 Cl] (M) 1 1 Mg SSDT b 25 mL-DMF 0. 1 M-TB AP 0.1 − 30 2.2 3 5.4 95.8 0.8 0.7 – – 3 3.9 2 6 Mg SSDT 25 mL-DMF 0. 1 M-TB AP 0 1 − 30 1.9 7 39.2 99 0.26 0.1 4 – – 38.8 3 7 Mg SSDT 25 mL-DMF 0. 1 M-TB AP 0.1 − 30 1.9 7 1 2.8 9 7.8 0 0.28 – – 1 2.6 4 8 Mg SSDT 25 mL-DMF 0. 1 M-TB AP 0.1 − 30 1.9 7 0.5 55.4 0.0 6 0.1 6 – – 0.3 5 6 Mg ESSG c 25 mL-DMF 0. 1 M-TB AP 0.1 − 1 0 1.95 28.2 96.8 0.89 0 – – 2 7.3 6 6 Mg ESSG 25 mL-DMF 0. 1 M-TB AP 0 1 − 30 2 4 9.5 95 1.4 9 1 – – 4 7 7 6 Ti–Pt a ESSG 25 mL-DMF 0. 1 M-TB AP 0.1 − 1 0 1.95 6 6.8 4 7.8 0.32 0.2 4 3 4.29 – 3 1.9 8 6 Mg SSDT 25 mL-DMF 1 9 mM-TD A TPhB 0.1 − 11.2 1.95 65.7 82.4 0.26 0.1 6 11.1 7 5 4.2 9 8 Mg SSDT 25 mL-DMF 1 9 mM-TD A TPliB 0.1 − 0.1 0.58 46.2 4 7.9 0.28 0.2 1 4.1 5 9.4 1 22.1 1 0 8 M g SSDT 2 5 m L-PhCH 2 Cl 1 9 mM-TD A TPliB 8 − 1 4.6 0.2 4 0.6 6 6.6 2.0 7 0.4 8 11.5 1 9.3 5 0.43 11 9 M g SSDT 2 5 m L-PhCH 2 Cl 1 8.6 mM-TD A TPhB 8 − 1 4.6 6 0.6 7 0.8 9 0.7 6 0 8.89 81.36 0.08 1 2 9 M g SSDT 2 5 m L-PhCH 2 Cl 1 8.6 mM-TD A TPhB 8 − 75.22 2 0.4 7 3 0 1.4 7 23.1 2 2.4 1 0.32 1 3 9 M g SSDT 2 5 m L-PhCH 2 Cl 1 8.6 mM-TD A TPhB 8 − 1 50 1.9 8 1.8 7 0.1 9 1.26 4 4 8 87.08 0.1 2 1 4 1 2 Mg SSDT 2 5 m L-PhCH 2 Cl 1 8.6 mM-TD A TPhB 8 − 11 5 1.9 8 1.6 9.4 0.1 6 0.0 4 5.86 84.5 0.1 5 T =4 0 ◦C. a Platinize d titanium Grid. b S tainless st eel drille d tub e. c Expande d stainless st eel grid.
Table 2
Calculated CO2–DMF proportions, according to the pressure
P (MPa) Vapour phase Liquid phase
V (mL) (kg m−3) m (g) mCO2(g) mDMF(g) V (mL) (kg m−3) m (g) mCO2(g) mDMF(g) 0.1 218.35 1.71 0.38 0.37 0.01 25.72 924.16 23.77 0.16 23.60 1 214.72 17.69 3.80 3.79 0.01 29.34 867.36 25.45 1.85 23.60 3 203.37 58.83 11.97 11.95 0.02 40.71 754.85 30.73 7.14 23.60 5 183.41 112.55 20.64 20.61 0.03 60.65 659.36 39.99 16.41 23.58 7 130.22 196.21 25.55 25.49 0.06 113.85 568.29 64.70 41.15 23.55
T = 40◦C, VinitialDMF = 25 mL, total volume: 244 mL.
Experiments were also carried out in order to study the influence of other parameters such as current density, electrodes materials or supporting electrolyte.
3.2.3. Influence of current density and electrode nature
Syntheses were carried out at 6 MPa and 40◦C, at two differ-ent currdiffer-ent densities, i.e.,−30 and −10 mA cm−2(Table 1, entries
5 and 6). Phenylacetic acid was obtained as the main product in both cases. By-product ratio was increased with current density, but better conversion was reached (Table 1, entries 5 and 6). Despite a slight decrease in selectivity, faradaic yield was considerably raised. An Expanded Stainless Steel Grid (ESSG) was employed as the cathode during these experiments. Its frontal transparency was given at 50%, while it was estimated at 35% for the Stainless Steel Drilled Tube (SSDT). The use of ESSG was more efficient than SSDT, as shown inTable 1(entries 6 and 2, respectively), where faradaic yield was increased in the order of 10%, despite a longer exper-imental time. Indeed, circulation of the electrolytical solution in the inter-electrodes annular spacing was improved by the higher frontal transparency electrode, which favoured convection, leading to better mass transfer of PhCH2Cl from the solution to the vicinity
of the cathode.
The good course of the electrosynthesis was also favoured by the use of magnesium massive rod sacrificial anodes. Neverthe-less, the use of a non-consumable anode was also considered in the perspective of a continuous process (Table 1, entries 5 and 7). Therefore magnesium massive rod was substituted by a platinized titanium grid tube (Table 1, entry 7). In this case, small toluene and bibenzyl quantities were synthesized as by-products, but benzalde-hyde was obtained in great quantities (Table 1, entry 7). Another by-product, benzyl alcohol, may be also synthesized by a possible benzyl chloride hydrolysis and benzaldehyde production may have resulted from benzyl alcohol anodic oxidation. Faradaic yield was lowered by a decrease in selectivity, although a better conversion for PhCH2Cl was observed. Contrary to the sacrificial anode process,
anodic reaction was not imposed, and benzaldehyde synthesis was favoured. Thus, best results for phenylacetic acid synthesis were obtained when Mg massive rod and ESSG were used as the anode and the cathode, respectively. Nevertheless, the reaction was lim-ited by TBAP solubility in CO2rich solutions. Therefore, change in
the nature of the supporting electrolyte was envisaged. 3.2.4. Change of the supporting electrolyte
For these experiments, TBAP was replaced by TDATPhB (entries 8–14). It was demonstrated that this hydrophobic salt is soluble in non-polar media such scCO2 [22]. Experiment was run under
the best operating conditions found for phenylacetic acid synthesis using TBAP as the supporting electrolyte (Table 1, entry 8). With TDATPhB, a better faradaic yield was obtained (Table 1, entry 8) but selectivity was lowered. A significant amount of benzyl alcohol was synthesized while a high PhCH2Cl conversion was achieved.
Nev-ertheless, phenylacetic acid was obtained as the main product and benzyl alcohol production can be attributed to PhCH2Cl hydrolysis.
TDATPhB was also used to run experiment under single-phase CO2-expanded DMF mixture conditions. However, no
electrochem-ical operation was possible under these operating conditions, because it outranged the cell voltage that could be applied by the potentiostat, due to a too low electrical conductivity of the medium. As a consequence, a−2 mA intensity value was imposed as the highest current value accepted by the electrochemical apparatus. For this very low current density, the total theoretical charge was not applied, because it would have led to prohibitive experimen-tal time (Table 1, entry 9). Then, analyses were carried out after passing 30% of the total theoretical electric charge (Table 1, entry 9). Phenylacetic acid was found to be the main product, and small quantities of toluene and bibenzyle were also synthesized. Never-theless, significant quantities of benzyl alcohol and benzaldehyde were analyzed. Indeed, PhCH2Cl hydrolysis may have occurred, and
the produced benzyl alcohol was partially oxidized. Nevertheless, using TDATPhB, it was possible to operate experiment under single-phase CO2-expanded DMF mixture conditions.
3.3. Experiments without DMF
Attempts have been made to perform solvent-free PhCH2Cl
elec-trocarboxylation with CO2. As for the first investigations done with
DMF as the solvent, it is necessary to have an estimation of the ther-modynamic behaviour of CO2–PhCH2Cl mixtures, in order to know
physical state of the reacting mixture.
3.3.1. Computation of phase diagram for CO2–PhCH2Cl mixtures Calculation for the phase diagram of binary CO2–PhCH2Cl
mixture was done using a specific fluid-phase equilibrium ther-modynamic model. This type of calculation is usually preferably based on experimental fluid-phase equilibrium data, but in the case of CO2–PhCH2Cl mixture, such experimental results were
not found in the literature. In such case, predictive thermody-namic modelling has to be done. In this context, fluid phase equilibria of CO2–PhCH2Cl mixture were computed using the
Soave–Redlich–Kwong (SRK) equation of state[23], and, in order to account for interactions taking place between components in the mixture, we used the mixing rules developed by Huron and Vidal[24]and modified by Michelsen[25](MHV2 mixing rules). The UNIFAC modified by Larsen et al.[26]predictive activity coef-ficient model was chosen to determine the value of the free excess Gibbs energy needed in the calculation of the mixture parameters. In this case, the model is totally predictive, i.e., it does not need experimental binary interaction coefficients. Such a computational procedure was shown to be adequate for mixtures at high pressure composed of constituents with different type of polarity[27]. Cal-culations have been carried out using a comercial software, Prophy PlusTM(Prosim S.A., France).
Using this method, phase diagrams at four different tempera-tures have been established for the binary mixture CO2–PhCH2Cl
Fig. 3. Calculated P(x,y) phase equilibrium CO2–PhCH2Cl mixtures at different
tem-peratures.
For each temperature, we can evaluate values of bubble and dew pressures for the entire range of compositions of the mixture. These two curves define a zone of two-phase mixture. For each case, bubble pressure curve and dew pressure curve join each other at the critical point of the mixture, at the considered temperature. At 35.5◦C, i.e., in the vicinity of the critical temperature of CO2,
criti-cal pressure is close to that of pure CO2, corresponding toxCO2very close to 1. As far as temperature increases, critical point moves away from thexCO2= 1 axis. We encountered some numerical difficulties in the precise determination of position of the critical points, due to the very particular properties of the mixture in this zone. Nev-ertheless, the accuracy is sufficient to give a good indication upon the physical state of the mixture for a given composition, at known pressure and temperature.
From this figure, we can also observe that at a given temperature, the rich PhCH2Cl liquid phase is enriched in CO2as the pressure is
increased, indicating that CO2is very soluble in the PhCH2Cl liquid
phase. Conversely, we observe that the vapour phase is very poor in PhCH2Cl (dew curves very close toyCO2= 1 axis), this fraction being increased with temperature.
3.3.2. Results
Phenylacetic acid synthesis was carried out in CO2-expanded
PhCH2Cl liquid and in single-phase near the critical point of the
CO2–PhCH2Cl mixture, TDATPhB being used as the supporting
elec-trolyte (Table 1, entries 10–14). First experiment (entry 10) was run at 8 MPa and 40◦C, the initial mixture composition beingxCO2 = 9. According to bubble and dew pressure curves at 40◦C, the mixture is likely to be a two-phase system in the reactor. In this case, it was assumed that the reaction only took place in the liquid phase, i.e., in the CO2–PhCH2Cl expanded liquid. As can be observed from
experi-mental results (Table 1, entry 10) phenylacetic acid was obtained as the main product, but conversion was very low. Benzaldehyde and benzyl alcohol were mainly synthesized as by-products. This was also observed at 9 MPa, where the mixture was also presumably a two-phase system (Table 1, entries 11 and 13). Benzyl alcohol was obtained as the main product, whatever the applied current den-sity value. Finally, no major change was observed at higher pressure (12 MPa,Table 1, entry 14), where the mixture was at its supercrit-ical state. Synthesis performances were comparable to entry 13 for which current density was applied in the same order of magni-tude. Benzyl alcohol was also obtained as the main product. No conclusion may be drawn from the increase in pressure or from the biphasic to supercritical state change while small amounts of phenylacetic acid were achieved.
As a conclusion on these results, poor PhCH2Cl conversion can be
explained by the high CO2concentration, which hinders the
forma-tion of electrogenerated PhCH2−species. Moreover, it may also be
due to cathode geometry, for which larger surface was preferred to a larger frontal transparency. So, in this configuration, mass trans-fer across the electrode spacing was not favoured. At moderate current density, phenylacetic acid was obtained as the main prod-uct, despite a very low conversion (Table 1, entry 12). Such current density was more suitable but low faradaic yield was reached. 4. Conclusions
Consumable magnesium anode process was found to be selec-tive for phenylacetic acid production in CO2–DMF expanded liquid.
Indeed, 6 MPa pressure was found to be the optimal operat-ing pressure, when two-phase CO2–DMF binary mixtures were
employed. Few results could be obtained under single-phase condi-tions. However, process performances were increased when a very hydrophobic salt was used as the supporting electrolyte. Benzalde-hyde and benzyl alcohol were obtained as the main by-products. This phenomenon was increased when CO2-expanded PhCH2Cl
liq-uid was used, where, in this case, PhCH2Cl acts as a reagent and a
co-solvent. In some cases, phenylacetic acid was not synthesized as the main product. PhCH2− anions were less electrogenerated
because of high CO2concentrations and lowered mass transport at
the electrodes.
Operating parameters such as electrodes and reactor geometries can be improved. Moreover, the observed low electric conductivity of the reaction mixture, when high CO2concentrations are used, as
in single-phase experiments, proved to be crucial to operate elec-trochemical reactions. This is due to the low conductivity of CO2and
poor solubility of conventional salts, like TBAP. Electrical conductiv-ity of the reactional medium must be increased in order to decrease cell electrical resistance. Performances were improved by the use of TDATPhB, but it was still not satisfactory. Finally, until very efficient CO2soluble salts are proposed, single-phase electrochemical
reac-tion will remain problematic, and operareac-tion in CO2-expanded DMF
liquid might then be the sole solution to minimize the solvent quan-tities in such process. Another interesting alternative could also be to substitute the aprotic organic co-solvent and support elec-trolyte by a hydrophobic room temperature ionic liquid. Nowadays, such commercial CO2miscible compounds can be found[28,29]and
green consumable anode process might then be designed. In addi-tion, the use of catalysts[30–34]or operation of a modified cathode, grafted with onium salts to ensure the conductivity in the vicinity of the electrode, may also be envisaged, so that the use of sup-porting electrolyte could be suppressed. As a final remark, PhCH2Cl
was chosen here as a model molecule, but such an approach may be extended towards other aromatic halides, provided that their reduction potential is higher than CO2reduction potential.
References
[1] A. Savall, Recent developments in organic electrosynthesis, J. Phys. IV 4 (C1) (1994) 163–174.
[2] M.A. Matthews, Green electrochemistry. Examples and challenges, Pure Appl. Chem. 73 (8) (2001) 1305–1308.
[3] T. Fushigami, T. Tajima, Development of new methodologies toward green sus-tainable organic electrode processes, Electrochemistry 47 (8) (2006) 585–589. [4] G. Silvestri, S. Gambino, G. Filardo, Use of sacrificial anodes in synthetic elec-trochemistry. Processes involving carbon dioxide, Acta Chem. Scand. 45 (10) (1991) 987–992.
[5] J. Chaussard, M. Troupel, Y. Robin, G. Jacob, J.P. Juhasz, Scale-up of electrocar-boxylation reactions with a consumable anode, J. Appl. Electrochem. 19 (1989) 345–348.
[6] O. Sock, M. Troupel, J. P ´erichon, Electrosynthesis of carboxylic acids from organic halides and carbon dioxide, Tetrahedron Lett. 26 (12) (1985) 1509– 1512.
[7] M. Heintz, O. Sock, C. Saboureau, J. P ´erichon, Electrosynthesis of aryl-carboxylic acids from chlorobenzene derivatives and carbon dioxide, Tetrahedron 44 (6) (1988) 1631–1636.
[8] J.F. Fauvarque, C. Chevrot, A. Jutand, M. Franc¸ois, J. Perichon, Catalytic elec-trosynthesis of para-substituted benzoic acids from the corresponding halogen derivatives and carbon dioxide, J. Organometallic Chem. 264 (1–2) (1984) 273–281.
[9] C.L. Scortichini, S.J. Babinec, The effect of cathode composition on the effi-ciency of electrocarboxylation of aromatic halides, J. Electroanal. Chem. 379 (1–2) (1994) 111–120.
[10] O. Scialdone, G. Filardo, A. Galia, D. Mantione, G. Silvestri, Performances of homogeneous charge transfer catalysts in the electrocarboxylation of benzyl halides, Acta Chem. Scand. 53 (10) (1999) 800–806.
[11] J. P ´erichon, O. Sock, M. Troupel, Patent n◦FR2566434-A1 (1985).
[12] F. Maspero, O. Piccolo, U. Romano, S. Gambino, Office europ ´een des brevets, brevet n◦EP0286944, 1988.
[13] V.A. Grinberg, V.M. Mazin, Electrochemical processes in liquid and supercritical dioxide, Russ. J. Electrochem. 34 (3) (1998) 223–229.
[14] A.P. Abbott, J.C. Harper, Electrochemical investigations in supercritical carbon dioxide, J. Chem. Soc. Faraday Trans. 92 (20) (1996) 3895–3898.
[15] V.A. Grinberg, T.A. Koch, V.M. Mazin, E.I. Mysov, S.P. Sterlin, Electrocarboxylation of 1,4-dibromobut-2-ene in a CO2–DMF liquid mixture, Russ. Chem. Bull. 46 (9)
(1997) 1560–1564.
[16] A. Sasaki, H.O. Kudoh, H. Senboku, M. Tokuda, Electrochemical carboxylation of several organic halides in supercritical carbon dioxide, in: Novel Trends in Electroorganic Synthesis (Papers Presented at the International Symposium on Electroorganic Synthesis), vol. 3, Kurashiki, Japan, Sept 24–27, 1997, 1998, pp. 245–246.
[17] B. Rajagopalan, M. Wei, G.T. Musie, B. Subramaniam, D.H. Busch, Homogeneous catalytic epoxidation of organic substrates in CO2-expanded solvents in the presence of water-soluble oxidants and catalysts, Ind. Eng. Chem. Res. 42 (2003) 6505.
[18] M. Wei, G.T. Musie, D.H. Busch, B. Subramaniam, CO2-expanded solvents:
unique and versatile media for performing homogeneous catalytic oxidations, J. Am. Chem. Soc. 124 (11) (2002) 2513–2517.
[19] P.G. Jessop, Homogeneously-catalyzed syntheses in supercritical fluids, Top. Catal. 5 (1998) 95–103.
[20] S. Chanfreau, P. Cognet, S. Camy, J.-S. Condoret, Electrochemical determina-tion of ferrocene diffusion coefficient in liquid media under high CO2pressure:
application to DMF–CO2mixtures, J. Electroanal. Chem. 604 (1) (2007) 33–40.
[21] S. Chanfreau, P. Cognet, J.-S. Condoret, Electrocarboxylation sous fortes pres-sions de CO2, R ´ecents Progr `es en G ´enie des Proc ´ed ´es 92 (C19) (2005) 1–8.
[22] A.P. Abbott, J.C. Harper, Electrochemical investigations in supercritical carbon dioxide, J. Chem. Soc., Faraday Trans. 92 (20) (1996) 3895–3898.
[23] G. Soave, Equilibrium constants from a modified Redlich–Kwong equation of state, Chem. Eng. Sci. 27 (1972) 1197–1203.
[24] M.-J. Huron, J. Vidal, New mixing rules in simple equations of state for rep-resenting vapour-liquid equilibria of strongly non-ideal mixtures, Fluid Phase Equilib. 3 (4) (1979) 255–271.
[25] M.L. Michelsen, A modified Huron–Vidal mixing rule for cubic equations of state, Fluid Phase Equilib. 60 (1990) 213–219.
[26] B.L. Larsen, P. Rasmussen, A. Fredenslund, A modified UNIFAC group-contribution model for prediction of phase equilibria and heats of mixing, Ind. Eng. Chem. Res. 26 (1987) 2274–2286.
[27] S. Dahl, A. Dunalewicz, A. Fredenslund, P. Rasmussen, The MHV2 model: Predic-tion of phase equilibria at sub- and supercritical condiPredic-tions, J. Supercrit. Fluids 5 (1) (1992) 42–47.
[28] A. Shariati, C.J. Peters, High-pressure phase behavior of systems with ionic liquids: Part III. The binary system carbon dioxide + 1-hexyl-3-methylimidazolium hexafluorophosphate, J. Supercrit. Fluids 30 (2) (2004) 139–144.
[29] P.G. Jessop, R.R. Stanley, R.A. Brown, C.A. Eckert, C.L. Liotta, T.T. Ngo, Pollet, Neo-teric solvents for asymmetric hydrogenation: supercritical fluids, ionic liquids, and expanded ionic liquids, P. Green Chem. 5 (2003) 123.
[30] O. Scialdone, A. Galia, G. Silvestri, C. Amatore, L. Thoulin, J.-N. Verpeaux, Elec-trocaboxylation of benzyl halides through redox catalysis on the preparative scale, Chem. Eur. J. 12 (28) (2006) 7433–7447.
[31] J. Damodar, M. Krishna, R. Srinivasulu, R. Jayarama, R. Srinivasulu, Synthesis of 2-arylpropionic acids by electrocarboxylation of benzyl chlo-rides catalysed by PdCl2(PPh3)2, Electrochem. Commun. 3 (12) (2001),
792–766.
[32] A. Gennaro, A.A. Isse, F. Maran, Nickel(I)(salen)-electrocatalyzed reduction of benzyl chlorides in the presence of carbon dioxide, J. Electroanal. Chem. 507 (1–2) (2001) 124–134.
[33] C. Amatore, A. Jutand, Activation of carbon dioxide by electron transfer and tran-sition metals. Mechanism of nickel-catalyzed electrocarboxylation of aromatic halides, J. Am. Chem. Soc. 113 (8) (1991) 2819–2825.
[34] G.D. Zheng, Q. Ding, Q.D. An, X.Z. Cao, Electrocarboxylation of organic com-pounds with carbon dioxide catalyzed by metalloporphyrins. IV. Effect of various factors on electrocarboxylation, Chin. Chem. Lett. 3 (5) (1992) 357– 358.
