Microreactor technology for on-site production of methyl chloride

2.1 Microreactor setup

The two microreactors used were stainless steel gas-phase microreactor units (GPMR-mix) produced and purchased from the Institut für Mikrotechnik Mainz (IMM, Mainz, Germany) (Figure 1).

Figure 1

Gas-phase microreactor (GPMR)-mix from the Institut für Mikrotechnik Mainz (IMM).

The microreactor consisted of a mixing and a catalytic zone, each equipped with 10 stainless steel microstructured platelets. The mixing plates had curved channels leading the reactants from the inlets to the mixing chamber, where they are contacted in a diffusion tunnel prior to entering the catalytic zone. The catalyst platelets had nine straight channels each, 90–100 μm deep, 460 μm wide and 9.5 mm long. The platelets were coated with a neat γ-alumina (UOP Versal VGL-25, UOP, Des Plaines, USA) catalyst using a slurry coating method prior to use. The coating method was described in a previous publication [12]. The thickness of the catalyst layer in the microchannels was 15±3 μm and the mass of catalyst in a single microreactor, i.e., on 10 coated catalyst platelets, was 3.4 mg. The reactor was heated with two heating cartridges which can be inserted in the reactor housing. The temperature in the catalyst zone was controlled by a thermocouple directly in contact with the catalytic zone. The first microreactor in series corresponded to the above description, while the second microreactor was slightly modified. As only one inlet was needed, one of the inlets was closed with a stainless steel cylinder. After this modification, it was not possible to insert the 10 mixing platelets. Furthermore, the partially reacted mixture that entered the second microreactor did not need additional mixing. Consequently, the mixing zone of the second microreactor was a chamber with a volume of 296 mm³. In some experiments, the mixing chamber was filled with quartz wool. A schematic view of the reaction setup is displayed in Figure 2.

Figure 2

Schematic representation of the experimental setup. MR1, MR2 microreactors: (A) neutralization bottle; (B) MeCl absorption.

Two microreactors were coupled in series with the possibility to bypass the second microreactor and obtain data for a single microreactor only. If not in use, the bypass line or the second microreactor was kept under a steady nitrogen flow. Helium and a mixture of 20% HCl in helium (AGA, München, Germany, 20% HCl in He) were fed from gas cylinders. The flow was controlled by mass flow controllers (Bronkhorst, Ruurlo, The Netherlands, EL-flow). Methanol [J.T. Baker, Center Valley, USA, high performance liquid chromatography (HPLC) gradient grade] was fed from a tank containing liquid methanol using an HPLC pump (Shimadzu, Kyoto, Japan, LC-20AD). All the lines and equipment from the methanol inlet to the gas chromatography (GC) were isolated and heated by electrical heating wires to 90°C.

After the microreactors, a neutralization bottle entirely filled with calcium oxide lumps (approximately 80 g, Fischer, Waltham, USA, general purpose grade) was installed to remove HCl and water from the product gas. In this way, the corrosion problem was minimized and HCl and water injections to the GC were prevented. The temperature of the neutralization bottle was adjusted to 100°C to ensure reaction with HCl and CaO to CaCl₂ and prevent condensation of methanol in the bottle. Depending on the HCl content of the reaction mixture after the reactors the bottle had to be replaced every 1–10 h. At the reactor outlet, a washing bottle filled with a volume based 2:6:2 mixture of water/methanol/ethanol amine (Sigma Aldrich, Seelze, Germany, ≥98%) at room temperature was placed to strip the methyl chloride from the product gas. Swagelok stainless steel lines and valves were used in the equipment (Swagelock, Solon, OH, USA). Samples were withdrawn with a gas-tight syringe through a septum in the sampling section. The syringe and an isolated transportation box were heated to 100°C prior to sampling in order to prevent condensation of methanol. The gas samples were manually injected to a gas chromatograph.

When investigating the separation of gaseous products from the flow, the neutralization bottle was replaced by a glass-made condenser. Three different condensers were compared: an Allihn condenser with a cooling surface of 61 cm², a coil condenser with 86 cm² of cooling surface and a Dimroth condenser with 210 cm² of cooling surface. An aqueous glycol solution was used as the cooling medium and the temperature varied between +1°C and -15°C. For the experiments conducted with the condensers, the mixing chamber was filled with quartz wool.

2.2 Analytical methods

The gaseous products were analyzed by gas chromatography using an Agilent 6890N GC equipped with a flame ionization (FI) detector and an HP-Plot/U Column [30 m, inner diameter (ID) 0.530 mm, film thickness: 20 μm, T=110°C]. The GC was calibrated for methyl chloride, DME and methanol by using mixtures of known concentrations. As no products were present in the feed, the concentrations of HCl and water, which could not be determined by GC were calculated from the known concentrations as follows:

n(H2O)=n(MeCl)+n(DME) and n(HCl)=n0(HCl)-n(MeCl).

The methanol content of the liquid phase obtained from the condenser was determined using the abovementioned GC, using ethanol as an internal standard. Water/methanol/ethanol mixtures of known composition were used for calibration. The water content was determined by volumetric Karl-Fischer titration (Hydranal, Fluka, Seelze, Germany) and the content of hydrochloric acid was determined by titration with NaOH (Sigma-Aldrich, Seelze, Germany).

3.1 Conversion and selectivity

The conversion that can be obtained with the microreactors is limited by the lowest flow rate that can be sustained by the equipment. If not otherwise stated, the microreactors were operated at a flow rate of 3.8 ml_n/min (ml_n at 0°C and 101.325 kPa) He/HCl, 4.0 ml_n/min He and 0.0013 ml_n/min methanol, which corresponds to a methanol:HCl ratio of 1:1 and a residence time in the catalytic zone of 0.174 s at 340°C and 0.186 s at 300°C. This corresponds to the lowest flow rates that can be sustained by the pump and the mass flow controllers available.

The methanol conversion and methyl chloride selectivity after two microreactors were determined in the temperature range of 300–340°C. The results and the calculated thermodynamic equilibria for selectivity and conversion (HSC7 software) are displayed in Figure 3.

Figure 3

Influence of temperature on () methanol conversion and () methyl chloride selectivity. The dashed lines represent the calculated thermodynamic equilibrium () conversion and () selectivity.

The conversion increased steeply from 300°C to 330°C, where it approached the thermodynamic equilibrium (97% conversion and 96% selectivity). Upon further increase of the temperature to 340°C, the conversion increased slightly and was marginally higher than the calculated thermodynamic limit. The fact that the experimental results slightly exceed the theoretical thermodynamic equilibrium can be attributed to imprecision of the gas chromatographic analysis. As the conversion increases to very high values, the methanol peak area decreases strongly and thus the relative error of the analysis increases. The selectivity increased steadily from 300°C to 340°C. The selectivity of this reaction increased with the reaction time due to the slow formation of MeCl from previously formed DME (reaction III) [3]. It can be concluded that the reaction equilibrium can be approached with two microreactors coupled in series at a temperature of 330–340°C.

3.2 Coupling of two microreactors: reaction engineering aspects

In a previously published work, the reaction kinetics of methyl chloride formation on γ-alumina in one microreactor was studied in detail [3]. Thus, the methanol conversion obtained using two microreactors coupled in series was compared to the conversion determined at the corresponding residence time from the obtained kinetic model. The measured and calculated values are presented in Table 1. It is inevitable to notice that the conversion obtained in two microreactors was considerably higher than that predicted by the kinetic model. At identical residence times (0.0678 s, 340°C, not at maximum conversion), the methanol conversion in two microreactors was 91.5% which was significantly higher than in one microreactor, in which 76.4% was reached.

There are two possible explanations for this observation; external mass transfer limitations due to the lower flow rate at identical residence time in one microreactor, or a blank activity, i.e., activity without a catalyst, of the second microreactor. Although external mass transfer limitations are unlikely to appear in the microreactor due to the short diffusion distances, experiments were carried out to distinguish between the two effects.

In the first microreactor, no considerable blank activity existed. Although the microreactor showed a slight blank activity with uncoated platelets, the microreactor with coated platelets showed the same conversion as an inert quartz-made tubular reactor at an identical catalyst mass-to-residence time ratio. This suggests, however, that the stainless steel surface at the reaction temperatures shows a small activity in the methyl chloride synthesis. Stainless steel subjected to the reaction atmosphere (HCl and methanol) can form a small amount of metal chlorides on its surface, which are active in methyl chloride synthesis [6]. In the first microreactor, the reaction mixture was mainly in contact with the catalyst at the reaction temperature, while in the second microreactor, the reactants were in contact in the empty mixing chamber, too. The chamber was at the reaction temperature and had a high volume compared to the catalytic bed. Consequently, the residence time was much higher in the mixing chamber than in the actual catalytic zone. Assuming a plug flow, the residence time in the mixing chamber would be 0.9 s instead of 0.17 s on the two catalytic beds.

To investigate the effects of the free volume in the mixing chamber, it was filled with quartz wool to reduce the residence time. The results were compared to experiments carried out without quartz wool and with the results of a single microreactor (Figure 4).

Figure 4

Influence of the dead volume on (A) the methanol conversion and (B) methyl chloride selectivity. () one microreactor () two microreactors () two microreactors no quartz wool. T=340°C. The red line represents the thermodynamic equilibrium.

Upon introduction of quartz wool to the mixing chamber, the methanol conversion in the two microreactors setup was significantly lowered from 91.5% to 83.2%, while in one microreactor, the conversion at an identical residence time was 76.3%. Furthermore, at shorter residence times, e.g., higher flow rates, the conversion in the two microreactors (with quartz wool) setup was constantly about 9% higher than in one microreactor. Similar observations were made for the selectivity towards methyl chloride. Generally, the selectivity increased with the residence time due to the reaction of DME with HCl (III). As the conversion was very similar for one microreactor and two microreactors with a quartz wool filling, the selectivities obtained from both setups were close. However, when the quartz wool was removed, the selectivity increased strongly and remained above the values obtained with the quartz wool filling, although the conversion reached at high residence times was similar. The reason is that the formation of methyl chloride from DME (reaction III) is very slow compared to the formation of methyl chloride and DME from methanol (reactions I and II).

External mass transfer limitations thus can be excluded and the additional activity is attributed to the reaction inside the mixing chamber on the stainless steel surface.

The two microreactors setup does not correspond to a simple doubling of the residence time. Instead, it introduces a dead volume that significantly contributes to the productivity of the reactors. This phenomenon has to be taken into account if the reactor is modeled mathematically. A schematic representation of the reaction space in the reactor is shown in Figure 5. The catalytic zones 1 and 2 can be described with the kinetic model published [3], while a new model is necessary to describe the reaction inside the mixing chamber. The kinetic modeling of the two microreactors setup will be addressed in a future work.

Figure 5

Schematic representation of the reaction space inside the microreactors.

The use of a larger single microreactor would be beneficial for the production of methyl chloride, but was not available in this work. However, the blind activity in the mixing chamber can be suppressed by improving the design of the steel cylinder, which was used to block the second inlet of the second microreactor, so that the mixing platelets can be inserted to the reactor. In this case, the setup consisting of two serially coupled reactors, including a valve system to bypass the second reactor, is an ideal tool for kinetic studies, as the available flow rates are doubled. Furthermore, the effects of external diffusion limitations can be evaluated, as a given residence time can be achieved with either one or two microreactors by adjusting the flow rate.

3.3 Catalyst long-term stability

The catalyst-coated platelets were used at a temperature of 340°C for 2750 min without observing a decrease of performance. Corrosion of the stainless steel reactor and lines is a risk in the methyl chloride synthesis, due to the formation of water as a reaction coproduct. Corrosion of the microreactor housing was not observed. However, a slight yellow color of the catalyst layer after reaction for 2750 min indicates that corrosion takes place on the stainless steel surface below the catalyst coating. Neither the performance nor the catalyst layer stability was affected. Considerable corrosion of the lines was observed in places where the lines had a temperature below 100°C and a corrosive mixture of water and HCl could condense. It is crucial that the equipment that is in contact with the reacted mixture is either kept well heated or is made of a corrosion resistant material such as Hastelloy, tantalum or glass.

3.4 Product separation

The product stream leaving the microreactor setup consists mainly of methyl chloride and water, with traces of DME, in addition to unreacted methanol and hydrochloric acid. In industrial scale, methyl chloride is separated from DME by extractive distillation. Prior to the distillation, water and unreacted HCl have to be removed [13]. Furthermore, methyl chloride can be purified by a series of scrubbers consisting of aqueous, caustic and sulfuric acid scrubbers. The first two scrubbers remove methanol, water and HCl from methyl chloride, while DME is trapped in the sulfuric acid scrubber [14].

In this work, the separation of methyl chloride and DME from water, HCl and methanol with a condenser was investigated. The use of a condenser avoids the production of large amounts of scrubber solutions and is less complex than extractive distillation. For a process that does not require high purity methyl chloride, this separation method can be attractive.

Three condensers with different internal surfaces were compared. The smallest condenser was an Allihn condenser followed by coil and Dimroth condensers.

The influence of temperature and condenser design on the separation efficiency was studied. The composition of the gas phase was investigated with the microreactors operated at the highest obtained conversion (97.6%), while the liquid phase analysis was performed at a lower conversion (83.3%), since no considerable amounts of condensate could be obtained at higher conversion due to the low flow rate.

Using the smallest condenser, the methanol content detected in the gas phase decreased upon lowering the temperature of the cooling fluid from 0.6 wt% at +1°C to 0.1 wt% at -10°C. When a larger condenser was used, or the temperature of the cooling fluid was further lowered to -15°C, the methanol content of the gas phase could not be further reduced. The gas-phase composition prior to and after the different condensers is given in Table 2. Furthermore, the gas phase contains traces of HCl, which could not be determined quantitatively. However, the presence of HCl was confirmed by leading the gas over an amine solution: a white smoke consisting of ammonium chloride evolved.

The composition of the liquid phase collected from the condenser suggested, though, that most of the unreacted HCl remained in the condensate. The composition of the fluid phase for the coil and the Dimroth condenser and the theoretical composition for complete condensation of water, methanol and HCl are given in Table 3. The composition of the gas phase leaving the condenser is specified in brackets. Both condensers gave a comparable separation of the liquid products. Only trace amounts of MeCl and DME were present in the condensate. The content of water increased compared to the theoretical composition (assuming complete condensation), while methanol was lower and the HCl content was close to the theoretical value. This shows that there is a considerable methanol slip and a minor HCl slip.

3.5 Production capacity of the microreactor setup

The present setup made from two serially coupled GPMR microreactors by IMM produces approximately 800 g of methyl chloride/year at the highest conversion and selectivity (97.5% at 340°C). However, due to corrosion problems the reaction mixture consisted of 82% helium. If a concentrated reaction mixture can be used in a corrosion-resistant microreactor, the system would produce approximately 5 kg of MeCl/year or 720 kg/year/g of catalyst. The production capacity of the two coupled microreactors is low. For actual production, the microreactors should be slightly larger, so that higher flow rates, i.e., production rates, can be sustained. The production capacity is increased by numbering up. Even though the production capacity of one microreactor is low, the catalyst efficiency is significantly enhanced. In a 15 μm thick catalyst layer, the effectiveness factor for methanol was shown to be 0.91, while in a 2 mm thick catalyst layer which corresponds to an industrial fixed bed, the effectiveness factor dropped to 0.03 [3]. The use of thin catalyst layers in microreactors thus considerably diminishes the equipment size compared to conventional fixed beds.

3.6 Summary

Methyl chloride was produced by catalytic hydrochlorination in two serially coupled microreactors achieving equilibrium conversion and selectivity. The highest conversion and selectivity were obtained at 340°C. Lowering the temperature to 330°C is possible, resulting in a slightly lower conversion and selectivity. The reaction was carried out in the catalyst durability test for 2750 min at 340°C without any drop of performance.

It was observed that the reaction takes place, to a considerable extent, in the dead volume between the microreactors. This effect was accounted for by the reaction of HCl with the stainless steel surface leading to the formation of metal chlorides, which are active in the methyl chloride synthesis. The gaseous products contained in the reaction mixture leaving the reactors were separated from the liquid components using glassware condensers. The maximum performance was reached at a cooling medium temperature of -10°C, independent of the condenser design. The gas phase contained 99 wt% of methyl chloride, 0.5–0.9 wt% DME, 0.1 wt% of methanol and traces of HCl. The condensate contained water and most of the unreacted HCl and methanol. Only traces (<0.01 wt%) of MeCl and DME could be detected in the condensate.

In conclusion, a setup consisting of two microreactors, or ideally one larger microreactor, equipped with a condenser, can produce methyl chloride in high yield and purity.