Combination of Chemo- and Biocatalysis: Conversion of Biomethane to Methanol and Formic Acid

Abstract

In the present day, methanol is mainly produced from methane via reforming processes, but research focuses on alternative production routes. Herein, we present a chemo-/biocatalytic oxidation cascade as a novel process to currently available methods. Starting from synthetic biogas, in the first step methane was oxidized to formaldehyde over a mesoporous VO_x/SBA-15 catalyst. In the second step, the produced formaldehyde was disproportionated enzymatically towards methanol and formic acid in equimolar ratio by formaldehyde dismutase (FDM) obtained from Pseudomonas putida. Two processing routes were demonstrated: (a) batch wise operation using free formaldehyde dismutase after accumulating formaldehyde from the first step and (b) continuous operation with immobilized enzymes. Remarkably, the chemo-/biocatalytic oxidation cascades generate methanol in much higher productivity compared to methane monooxygenase (MMO) which, however, directly converts methane. Moreover, production steps for the generation of formic acid were reduced from four to two stages.

1. Introduction

Nowadays, methane is mainly used as a source of energy [1] and its direct chemical utilization is restricted to the formation of (i) syngas [2,3] or (ii) hydrogen cyanide via the Andrussow or Degussa process [4] as well as (iii) for the production of methyl halides or carbon disulfide [5].

Oxygenates from methane, like methanol, formaldehyde or formic acid, are conventionally being produced via the syngas route because the direct production routes are rather limited. Homogeneous, heterogeneous and enzymatic direct catalytic conversions of methane primarily focus on the target molecule methanol. However, all of these approaches have certain restrictions like batchwise production, low productivity or selectivity owing to the nature of the applied chemistry [6].

The direct chemocatalytic synthesis of methanol from methane can be divided into homogeneous and heterogeneous methods. Among homogenous catalytic approaches the Periana system, [(bpym)PtCl₂] (bpym: bipyrimidine), converts methane in oleum directly to methyl bisulfate as a methanol derivative with a productivity of 0.115 kg_CH3OH∙L⁻¹∙h⁻¹ [7]. The major drawback of this reaction is the need for sulfur trioxide as transfer oxidant which makes it difficult to run the process in larger scale. Schüth and coworkers have found that K₂[PtCl₄] exhibits a higher productivity (0.48 kg_CH3OH∙L⁻¹∙h⁻¹), but SO₃ concentration has to be even increased [8]. Milder oxidants, like oxygen or hydrogen peroxide have been investigated, too, but to the detriment of very low productivities [9,10,11]. In order to overcome limitations in productivity when using milder oxidants, the groups of Stahl and Strassner have recently shown that the concept of redox mediation could be a future way in activating alkanes [12,13].

Heterogeneous catalytic approaches work with immobilized Pt(II) species on N-rich covalent triazine frameworks (CTF), but this may also include oleum as solvent/oxidant [14]. Alternatively, milder oxidants have been used by the group of Hutchings which applied gold-palladium alloy nanoparticles for the oxidation with H₂O₂ [15] or aqueous Au-Pd colloids for the oxidation with O₂ in the presence of H₂O₂ [16]. Obviously, the oxidation of methane with oxygen is still challenging. Several efforts have been made utilizing Cu sites incorporated in zeolites to activate methane and form methoxide species [17]. However, a subsequent hydrolyzing step with water is required in order to close the reaction cycle [18]. Recently and perhaps revolutionarily, Sushkevich et al. opened the possibility to transform methane to methanol by an anaerobic oxidation with water over copper-exchanged mordenite zeolite [19].

Biocatalysis shows that the activation of gas phase oxygen and its further reaction with methane is principally feasible. Enzymes classified as methane monooxygenases (MMO) [20,21,22,23], found in methanotrophic bacteria, do not only produce methanol selectively, but they also work under mild conditions of bacterial living. In soluble MMO (sMMO) di-iron clusters with bis-μ-oxo diamond core structure are discussed as active sites [21] whereas in particulate MMO (pMMO) a tri-copper center catalyzes the reaction [24]. Synthetic approaches mimicking the active site of the latter have already been successfully demonstrated [25,26,27]. Although the direct conversion of methane to methanol is possible using such natural or biomimetic structures, conversion rates and turn over numbers, respectively, are rather low and chemocatalytic activation of methane remains preferred.

With regards to thermodynamic data [28] the fundamental problem of chemocatalytic approaches can be expressed as sensitivity of oxygenates against over-oxidation (see Equations (1)–(4)).

CH₄ + 0.5 O₂ → CH₃OH ΔG° = −27.7 kcal∙mol⁻¹

(1)

CH₄ + O₂ → CH₂O + H₂O ΔG° = −69.1 kcal∙mol⁻¹

(2)

CH₄ + 1.5 O₂ → CO +2 H₂O ΔG° = −134.0 kcal∙mol⁻¹

(3)

CH₄ + 2 O₂ → CO₂ + 2 H₂O ΔG° = −195.5 kcal∙mol⁻¹

(4)

However, the direct formation of formaldehyde from methane and oxygen has been known for years and satisfying selectivities can be achieved at low conversion (X_CH4 < 10%) [6]. At very low residence times the reaction can be controlled partially. During recent years, researchers focused their studies on methods to increase the productivity of this reaction [29,30,31] and found that a CH₂O space-time yield above 5 kg_CH2O∙kg_cat⁻¹∙h⁻¹ is possible by using an optimized mesoporous silica supported VO_x catalyst [32,33].

Ohkubo and Hirose transformed methane directly to methanol and formic acid via chlorine dioxide radicals under photoirradiation [34]. Moreover, self-disproportionation of formaldehyde to methanol and formic acid under hydrothermal or supercritical conditions occurs spontaneously [35,36]. However, both the products can be obtained under much milder conditions applied in an enzymatic conversion of formaldehyde. Formaldehyde dismutase (FDM; EC 1.2.99.4) is an enzyme which converts formaldehyde into methanol and formic acid [37,38]. By a so-called flip-flop mechanism, formaldehyde is disproportionated to formic acid and methanol. Initially, hydrated formaldehyde is oxidized to formic acid. Thereafter the reduction equivalents are transferred to the enzyme bound cofactor NAD⁺ which is regenerated by the reduction of another formaldehyde molecule to methanol. As described by Rodewyk [39], the enzyme is independent of further cofactors and very robust. The associated and self-regenerating cofactor makes the FDM interesting to use as a technical enzyme because neither an additional cofactor has to be added nor an alternative regeneration path has to be developed. Most importantly, the specific activity of FDM can be up to 6000 times higher compared to MMO (138,000 ± 16,000 nmol_CH3OH∙mg⁻¹∙min⁻¹ vs. 22.9 ± 6.1 nmol_CH3OH∙mg⁻¹∙min⁻¹ over FDM [38] and pMMO [24], respectively). This led us to the idea to combine highly productive chemocatalytic methane oxidation with fast biocatalytic conversion of the formaldehyde intermediate to generate methanol and formic acid in equimolar amounts.

Scheme 1 compares the different routes for methanol production: (i) in industrial scale including formic acid synthesis, (ii) via direct oxidation routes and (iii) the here presented chemo-/biocatalytic oxidation cascade.

The latter, our approach, uses molecular oxygen in the chemocatalytic methane activation step followed by a relatively fast (selective) biocatalytic methanol production, additionally yielding formic acid in equimolar amounts. Moreover, methane from mimicked biogas (CH₄:CO₂ = 1:1) was utilized in this work to further explore this renewable feedstock, which is produced at a scale of Gm³∙a⁻¹ [41]. The following results should be also set into context of presently discussed methanol production routes from biogas which follow the conventional syngas strategy [42] or might use the hydrogenation of carbon dioxide [43,44,45,46]. Clearly, the following results represent insights of a new, alternative route to already existing methods from biomethane to methanol.

2. Materials and Methods

2.1. Chemocatalysis

Plain SBA-15 was synthesized as described by Iglesias et al. [47]. In brief, 8 g Pluronic P123 (Aldrich, St. Louis, MO, USA, 1.4 mmol) was diluted in 240 g of 2 M HCl (Fisher Scientific, Pittsburgh, PA, USA) at 40 °C. Afterwards 17.35 g tetraethyl orthosilicate (Aldrich, 98%, 83.3 mmol) was added dropwise to the solution, which was further stirred for 24 h at 40 °C. Subsequently the resulting suspension was treated hydrothermally in an autoclave with a PTFE inlet at 100 °C. After 24 h the white solid was isolated and washed with deionized water until the filtrate was free of Cl⁻. The solid residue was dried overnight at 70 °C and calcined for 16 h at 625 °C (heating rate 1 K∙min⁻¹) in air.

The functionalization of SBA-15 with VO_x species was prepared as described by Baltes et al. [48]. 2.7 g VO(acac)₂ (Strem Chemicals, 98%, Newburyport, MA, USA, 10.2 mmol) was dissolved in 450 mL of dry toluene in argon atmosphere. To this solution 4.5 g SBA-15 which was dried overnight at 120 °C was added and stirred for 5 h at room temperature. Afterwards, the solid was isolated and washed four times with toluene, followed by drying at 70 °C overnight and calcination at 625 °C for 16 h in air (heating rate 1 K∙min⁻¹) yielding 4.4 g of a bright yellow solid. The resulting catalyst is denoted as V/SBA-15.

Catalytic oxidation of methane was carried out in a horizontal fixed bed plug flow reactor consisting of a quartz tube with an inner diameter of 8 mm. Experiments were performed with 25 mg of the catalyst placed in the center of the quartz tube, fixed by quartz wool. Behind the catalyst bed the inner diameter of the reactor decreases to 4 mm. The oven temperature was controlled by a thermocouple in the middle of the catalyst bed. The catalytic performance was measured in a temperature range of 575–650 °C at atmospheric pressure for 30 min at each testing point after reaching steady state. The premixed reactant gas contained CH₄, CO₂ and O₂ in a ratio of 9:9:1 (Air Liquide, N45, Düsseldorf, Germany) and the gas flow passing the reactor was controlled by a mass flow controller (MKS-Instruments, Andover, MA, US). The on-line gas analysis was carried out with a FT-IR spectrometer (Bruker Matrix-MG01, Ettlingen, Germany) and the Bruker software Opus GA which uses reference spectra of the single compounds and a nonlinear fitting algorithm.

Methane conversion (X_CH4), formaldehyde yield (Y_CH2O), formaldehyde selectivity (S_CH2O) and space-time yield (STY_CH2O) were calculated according to Equations (5) to (8):

$$X_{C{H}_4}=\frac{c_{C{H}_4}^f-c_{C{H}_4}^p}{c_{C{H}_4}^f}·100\%$$

(5)

$$Y_{C{H}_{2}O}=\frac{c_{C{H}_{2}O}^p-c_{C{H}_{2}O}^f}{c_{C{H}_4}^f}·100\%$$

(6)

$$S_{C{H}_{2}O}=\frac{Y_{C{H}_{2}O}^{}}{X_{C{H}_4}^{}}·100\%$$

(7)

$$ST{Y}_{C{H}_{2}O}=\frac{{\dot{m}}_{C{H}_{2}O}^{}}{m_{cat}}$$

(8)

where $c_i^f$ and $c_i^p$ denote the concentration of component i in the feed gas f and product gas p, ${\dot{m}}_{C{H}_{2}O}$ the mass flow of formaldehyde and $m_{cat}$ the mass of the catalyst.

2.2. Biocatalysis

The heterologous expression of FDM from the Pseudomonas putida J3 strain by E. coli and its purification was described earlier [38]. Pseudomonas putida was isolated from sewage sludge at Fraunhofer-Institute for Interfacial Engineering and Biotechnology IGB Stuttgart and cultivated in a broth consisting of peptone (10 g∙L⁻¹), meat extract (10 g∙L⁻¹) and NaCl (5 g∙L⁻¹) at a pH value between 7.0 and 7.2. After chemo-mechanical extraction of the genomic DNA the heterologous expression of the FDM J3 gene was carried out by the E. coli strain BW3110 with the expression plasmid pJOE5940.1 in the rhamnose promotor system. E. coli was further cultivated on LB medium consisting of bacto peptone (10 g∙L⁻¹), yeast extract (5 g·L⁻¹) and NaCl (10 g∙L⁻¹) at a pH value of 7.0. For cell disruption cells were centrifuged, washed with aqueous NaCl solution (0.9%), suspended in aqueous Na₂HPO₄ (100 mM) at a pH value of 7, treated by ultrasound and again centrifuged. The resulting supernatant protein extract was used for further experiments.

Particles of the company Align Chemicals (Amersham, UK) were used for the immobilization of enzymes. Each immobilization was done in a column with a volume of 1 mL. To prepare for immobilization, the empty column weight was first recorded, the particles were then packed into columns, the dry weight of the packed column was determined, the package was rinsed with two column volumes of 2-propanol and weighed again after 10 min. The pore volume can be calculated from the wet weight of the column and the density of the 2-propanol. The column was then rinsed with five times the column volume of phosphate buffer (0.5 M, pH 6.8). For immobilization, 3 mL of centrifuged raw extract was used and flowed through the column using a syringe. The column was than incubated for 12 h at 4 °C with the remaining raw extract. After immobilization, an air-filled syringe was used to press the residual raw extract solution out of the column and the protein concentration was measured. The amount of immobilized protein can be calculated from the initial and residual protein concentrations. As reference, unloaded particles treated in the same way as described above were used but without protein loading.

The activity of the immobilized proteins in the respective columns was determined by rinsing the columns with 3 mL of a 1 g∙L⁻¹ formaldehyde solution, squeezing out the liquid after 30 min with an air-filled syringe and measuring the formaldehyde concentration. The enzyme activity was derived from the difference of the formaldehyde concentrations between protein loaded and reference column.

Biocatalytic reactions were performed in a pressure reactor (Büchi AG, Uster, Switzerland) at 37 °C on 500 mL scale. The formaldehyde and methanol concentrations were determined by quadrupole mass spectroscopy with liquid inlet (Variolytics, Stuttgart, Germany). Calibration of the mass spectrometer was done with methanol/formaldehyde solutions of known concentration.

2.3. Chemo-/Biocatalytic Oxidation Cascade

CH₄, CO₂ and O₂ in a ratio of 9:9:1 was fed into the chemocatalytic reactor and was partly converted to formaldehyde. A 1 L bioreactor was used to produce methanol and formic acid from the gas obtained downstream the chemocatalytic reactor.

In the first chemocatalytic step the feed was passed through the homemade plug flow quartz reactor described above which was equipped with 25 mg of the silica supported VO_x catalyst, V/SBA-15. A tube furnace was employed to reach the desired reaction temperature of 575 °C. The gas flow was adjusted to 300 mL∙min⁻¹ via a mass flow controller (MKS-Instruments, Andover, MA, USA).

The formaldehyde-containing product gas from chemocatalysis was passed through a 150 °C heated stainless-steel capillary (inner diameter: 3 mm) and fed into the buffer solution to run the biocatalytic conversion. Using a stirrer, the gas bubbles were shredded at a speed of 1500 rpm for better mass transfer from the gas phase into the liquid phase. In the biocatalytic reactor, free FDM or FDM immobilized on particles was introduced and the temperature of the buffer solution was kept at 37 °C by means of a thermostat. In detail, 500 mL of the 0.5 M phosphate buffer and 176 mg free FDM in raw extract or 176 mg FDM in raw extract immobilized onto 2 g IB-150A were placed in the stirrer tank for the biocatalytic disproportionation of formaldehyde at atmospheric pressure. The reaction was stopped when a pH value of 6.3 was reached. Online monitoring of the formaldehyde and methanol concentration was realized by a quadrupole mass spectrometer with electron ionization and a membrane inlet for liquids enabling the detection of volatile components from the solution (200 amu, Variolytics, Stuttgart, Germany).

2.4. Product Separation

The selectivity of polydimethylsiloxane (PDMS) membranes (PolyAn, Berlin, Germany) was determined by the use of the aforementioned mass spectrometer with membrane inlet using a PTFE membrane for reference measurements. The pressure in the membrane module was 0.16 mbar and the temperature of the solution 25 °C. Membranes were allowed to swell in the respective solution for at least 12 h to become fully functional. The volume of the solution was 500 mL and the flow at the membrane inlet was 10 mL∙min.

The depletion of methanol over time was investigated with the M12 membrane. The used membrane area was 0.005 m² and the volume of the solution was 500 mL. To improve mass transfer each solution was stirred. Methanol was collected using a cooling trap and the pressure at the permeate side was 20 mbar. Selectivity for methanol was calculated on the base of its concentrations c at the permeate and feed side according to Equation (9):

$$S= \frac{c_{permeate}}{c_{feed}}$$

(9)

For ion exchange experiments Purolite A111 (Purolite, Bala Cynwyd, PA, USA) was used for the separation of formic acid. The material was activated by washing first with 4 wt% NaOH for 1 h and afterwards with double-distilled water. Then, 51.5 g of Purolite A111 was added to 200 mL of an 80 mM solution of formic acid. The adsorption was done under stirring at room temperature and the concentration of formic acid was determined by mass spectrometry.

Extraction experiments of formic acid were done using a solution with 0.45 g∙L⁻¹ formic acid. Following this, 15 mL of formic acid solution and 15 mL organic phase were mixed. For an optimal segregation the mixtures were centrifuged for 30 min and were allowed to separate for 2 h afterwards. The concentration of formic acid in each phase was determined by HPLC-UV.

3. Results and Discussion

3.1. Chemocatalytic Methane Activation—Formaldehyde Production

Methane from synthetic biogas was transformed in the presence of equimolar amounts of carbon dioxide to formaldehyde. As catalyst VO_x species supported on SBA-15 (V/SBA-15) were prepared by a grafting procedure since this method provides highly dispersed vanadia species [48] which are selective for formaldehyde. The investigated catalyst possesses a vanadia content of 2.2 wt% as determined by ICP-OES. In the X-ray diffraction pattern (Figures S1 and S2) reflections for the (100), (110) and (200) planes were observed confirming the hexagonal structure of the pure support and the derived catalyst. Nitrogen sorption experiments (Figures S3 and S4) revealed a specific surface area of 581 m²∙g⁻¹ and a mean pore diameter of 5.1 nm of the fresh catalyst. Spent V/SBA-15 slightly loses surface area (of 507 m²∙g⁻¹) whereas pore diameter is nearly not affected. Raman spectroscopic investigations on the V/SBA-15 catalyst as well as the pure support (Figure S5) revealed that in V/SBA-15 a band at 1035 cm⁻¹ appears, which can be assigned to the V=O stretching mode, while the signals from the support at 970 cm⁻¹ (ν_s(Si-OH)) and 800 cm⁻¹ (ν_s(Si-O-Si)) decreased, respectively [31,49,50,51,52,53,54], indicating the molecular attachment of VO_x species on the silica surface. Furthermore, the infrared spectrum of the catalyst (Figure S6) shows, beside the typical ν_s(SiO-H) band, a small signal at 3660 cm⁻¹ which has been assigned to the ν_s(VO-H) mode [55]. According to literature the latter species were responsible for a high selectivity towards formaldehyde [50]. The UV-vis DRS spectra (Figure S7) of fresh and spent V/SBA-15 indicate a high amount of monomeric and low oligomeric VO₄ species which are stable during reaction. Both the samples possess absorption maxima at 250 nm and an UV-vis edge energy (E_g) of 3.5 eV. The transparency above 400 nm is a measure for the absence of polymeric or particulate VO_x species [51,52,56].

Before integrating into the cascade, the performance of the chemocatalytic first stage was tested under different conditions to evaluate the optimal set of reaction parameters. For this purpose, the reaction temperature and the continuous total flow were varied from 575–650 °C and from 100–300 mL∙min⁻¹, respectively. The latter correspond to gas hourly space velocities (GHSVs) from 114,000–342,000 L_CH4∙kg_cat⁻¹∙h⁻¹. The resulting selectivity vs. conversion dependency is illustrated in Figure 1a and shows a typical behavior for this selective oxidation of methane which was normally carried out in the absence of carbon dioxide [6].

The observed methane conversion (X_CH4) ranges from 1.0 to 6.8% corresponding to a formaldehyde selectivity (S_CH2O) between 50.3 and 12.5%, respectively. Side products were mainly CO and 5–10% CO₂. These values are comparable to those reported earlier by our group [31,33] for SBA-15 supported VO_x catalysts obtained from carbon dioxide free reactant gases. Figure 1b depicts the formaldehyde space-time yields (STY) obtained at different conditions. In general, the formaldehyde space-time yield grows continuously with an increasing gas flow. The behavior was observed by increasing the reaction temperature except at very low flow rates caused by overoxidation of produced formaldehyde. The resulting CH₂O space-time yield ranges from 1.0 to 2.7 kg_CH2O∙kg_cat⁻¹∙h⁻¹. These values are lower compared to previous studies [32] because of the lower methane concentration in biogas instead of using pure methane/oxygen mixtures.

In order to minimize the content of side products (CO and CO₂) the reaction conditions of the highest CH₂O selectivity were chosen (575 °C, total flow: 300 mL∙min⁻¹) for the experiments which combine chemo- and biocatalysis. Perspectively, the yield of the chemocatalytic conversion unit could be improved by a gas recycling. In this way, a total methane conversion of 89% with a formaldehyde yield of up to 50% was already reported by Bafas et al. [57] who separated the formaldehyde from the gas phase via a liquid trap after each cycle.

3.2. Biocatalytic Disproportionation of Formaldehyde

The produced formaldehyde can be transformed into methanol and formic acid by FDM. The immobilization of FDM on specific carriers offers long term and continuous processing modes as well as simple product recovery. Hence, we first tested several carriers with different functional groups. The surface specific activity allows a comparison of the different carriers.

Table 1. Activity of the FDM on different carriers.

Column	Bond Type	Matrix	Functional Group	Surface Based Enzyme Activity [U∙mm−2]	Number of Measurements	Measurement Period/Half-Life/Residual Activity [Days]/[Days]/[%]
IB-150A	covalent	polyacrylic	epoxy, nonpolar	53.3	29	440/291/38.8
IB-150P	covalent	polyacrylic	epoxy, polar	29.5	16	145/129/19.3
IB-D152	cationic	polyacrylic	carboxylic group	32.0	14	90/129/37.5
IB-C435	cationic	polyacrylic	carboxylic group	25.7	16	348/329/32.0
IB-A161	anionic, strong	polystyrene	quaternary ammonium	51.8	9	302/96/34.0
IB-A171	anionic, strong	polystyrene	quaternary ammonium	62.5	9	96/28/33.5
IB-A369	anionic, weak	polystyrene	quaternary ammonium	61.5	2	<20 (no activity)
IB-EC1	non-ionic bond	polyacrylic	carboxyl group	18.6	2	<20 (activity 6.4 U/mm2)
IB-S861	non-ionic bond	polystyrene	aromatic	26.0	7	57/31/30.6

shows the calculated enzyme activity related to the surfaces of the carrier particles. As can be seen, the carriers with a quaternary ammonium compound as functional group showed the highest area-specific activity. With regards to long-term stability, the best results were achieved with the columns IB-150A and IB-C435. After 440 days and 29 measurements the residual activity of the enzyme on a polyacrylic carrier IB-150A with an epoxy group was still 40%. The carrier IB-C435 with carboxyl groups showed a half-life of 329 days after 348 days and 16 measurements. The carrier IB-150A had a lower half-life of 291 days compared to the carrier IB-C435 but a higher activity. We have chosen carrier IB-150A for further experiments since it offers sufficient initial activity at high long-term stability.

The activity of the immobilized FDM was tested in batch and compared to that obtained over free FDM. The reaction catalyzed by the free FDM was completed seven- to eightfold faster compared to the immobilized FDM what might be caused by mass transport limitations and damages or hindrance of the enzymes caused by the immobilization process. The maximal reaction rate v_max of the free FDM reached a value of 44.7 mmol·min⁻¹ which was much higher than the v_max of the immobilized FDM with 1.6 mmol·min⁻¹. The corresponding Michaelis constants K_m, the substrate concentration with half maximum reaction rate, were determined to a value of 50.9 mM and 41.2 mM, respectively.

However, increasing concentrations of methanol and formic acid rapidly inhibit product formation (see K_i in Figure 2). In order to illustrate this serious problem, activity tests with the free enzyme at different starting concentrations of either methanol or formic acid, respectively, are shown in Figure 3. In the experiments with formic acid, a buffer was added to exclude inhibition by a lower pH value. The temporal decrease of the formaldehyde concentration with ongoing reaction time at different product concentrations is obvious. For both products, an inhibition effect was observed. While methanol completely inhibits the reaction at a concentration of 1.5 M an addition of formic acid already stops the reaction at a concentration of 0.7 M.

3.3. Cascade Combining Chemo- and Biocatalysis

After the optimization of the separate processes both were combined in the chemo-/biocatalytic oxidation cascade. The used set-up is presented in Scheme 2. The feed was composed of CH₄, CO₂ and O₂ in the ratio of 9:9:1.

In the first mode, formaldehyde was first introduced into the buffer solution by absorption in the absence of free FDM. The formaldehyde feed (chemocatalytic product gas stream) was switched off after 160 min and 176 mg free FDM containing raw extract dissolved in 10 mL buffer was added to the stirred vessel. The enrichment of the formaldehyde and the formation of methanol after addition of the FDM are shown in Figure 4a. After 30 min the reaction was completed. Nearly no formaldehyde could be detected, and the methanol concentration was about 0.2 g∙L⁻¹. The concentration of formic acid increases in the same way and the initial enzyme activity in this experiment was 17,900 nmol∙mg⁻¹∙min⁻¹ for both components.

In the second mode, the cascade was operated continuously since the use of immobilized enzymes enables a reaction during absorption. Moreover, product removal is simplified reducing the problems caused by product inhibition, which might be interesting for possible applications. Figure 4b shows the enrichment of formaldehyde in the tank as consequence of ongoing feeding from chemocatalytic methane conversion. The constant gassing caused a saturation of the buffer solution with carbon dioxide and methane, so that bubbles entered the membrane inlet of the mass spectrometer and led to the noise of the measurement signal. In parallel, the formaldehyde in the buffer solution within the stirrer tank is simultaneously converted by the immobilized enzyme to methanol and formic acid. After an induction period of about 30 min, the onset of the reaction was observed. The methanol concentration increased linearly up to 20 mg∙L⁻¹ and the formaldehyde concentration reached a maximum of 0.3 g∙L⁻¹ at the end of the experiment. The overall productivity was 30 nmol∙mg⁻¹∙min⁻¹ for both components.

Two points speak clearly against an exact implementation: (i) the lower productivity compared to industrial methanol production and (ii) ongoing product enrichment. Hence, it can be stated that the aforementioned approach only demonstrates the concept and a direct transfer to industry would be impractical. We suggest that productivity could be improved by site-selective immobilization. Moreover, in order to retain the protein structure and high activity of the FDM immobilization in cavities using porous polymers or glasses could be utilized. Furthermore, the inhibition of FDM by formed products could be solved by in situ product removal as it will be shown on idealized mixtures in the following.

3.4. Product Separation

As shown in Figure 3, product inhibition is one of the limiting factors for the presented oxidation cascade necessitating an in situ product removal. Accordingly, we investigated first principles on the separation and enrichment of methanol by pervaporation and the separation of formic acid by ion exchange and extraction to evaluate possible routes for further process development.

3.4.1. Pervaporation of Methanol

Due to the low methanol concentration in the solution a direct distillation is rather uneconomic. Therefore, pervaporation was investigated since it offers recovery of FDM and enrichment of methanol in one step which can be complemented by further distillation. In this regard we tested different commercially available polydimethylsiloxane membranes. M1 and M2 are different ultrafiltration membranes which are further optimized for either high selectivity (M11, M21) or high flux (M12, M22). The selectivity obtained at different concentrations of methanol is given in Figure 5a. All membranes show a high selectivity at low methanol concentrations. By increasing methanol concentration, selectivity drops. Membrane M12 performed best over the whole concentration range and was therefore chosen for further experiments.

Figure 5b illustrates the methanol concentration in the retentate using membrane M12. After 22 h the methanol concentration decreased from 5 to 2.1 g∙L⁻¹ at 25 °C and 20 mbar at the permeate side and reached a minimum concentration of 0.1 g∙L⁻¹ at the end of the experiment.

3.4.2. Ion Exchange and Extraction of Formic Acid

The separation of formic acid is even more crucial since a decrease of the pH affects FDM activity (Figure S8). The high boiling point of formic acid and the fact that it is dissociated at the present low concentrations make distillation nearly impractical. Hence, we investigated possibilities of ion exchange (as used in industrial formic acid production) and extraction of formic acid.

When ion exchange is used a buffer solution can no longer be utilized. To maintain a neutral pH value under these conditions, formate (and carbonate) species have to be directly separated. The formate adsorption was investigated using an ion exchanger (Purolite A111, Figure S9) and an 80 mM solution of formic acid. The loading over time and formic acid concentration are illustrated in Figure 6a.

After 8 min half of the maximum loading (0.31 mmol∙g⁻¹ as given by the manufacturer) was reached followed by a decrease in the further loading rate. For regeneration of the ion exchanger a basic treatment is required. However, FDM is not stable under basic conditions. In order to overcome this problem commercially available mixed bed filters can be used which allow a separation of different support materials.

Besides ion exchange liquid/liquid extraction represents another option for the separation of formic acid which is also used in industry. For this purpose, different mixtures of tributylphosphate (TBP) or trioctylamine (TOA) in diethylcarbonate were investigated. The partition coefficients K_D were determined by extraction of 10 mM formic acid solutions. Prior to these experiments a possible inhibition of FDM by these extraction agents was tested revealing no effect on enzymatic activity. The K_D values for the different extraction mixtures are shown in Figure 6b. The K_D values increase with increasing concentration TBP from 0.23 (0% TBP) to 1.75 (100% TBP) whereas TOA mixtures perform best at low concentration (K_D 1.24 at 5%). Higher concentrations of TOA decrease the extraction efficiency up to 0.13 for 100% TOA.

4. Conclusions

We presented a successful cascadic combination of (i) chemocatalytic direct methane conversion to formaldehyde and (ii) its biocatalytic disproportionation to methanol and formic acid. Moreover, we illustrate possible pathways for the separation of the two final products.

The cascade is characterized by the following facets:

(1)

Productivity. It was possible to produce methanol with an initial FDM activity up to 17,900 nmol∙mg⁻¹∙min⁻¹, which is significantly higher compared to those obtained over MMO, which use methane as substrate.

(2)

Production stages. Nowadays, formic acid is prepared in four production stages (methane → syngas → methanol → methyl formate → formic acid) whereas the herein presented route requires two steps, only. Moreover, methanol is being produced conventionally in a two-step process including the high endothermic syngas production. In this regard our method is clearly favored due to the exothermic nature of the chemocatalytic oxidation process.

(3)

Biomass utilization. It was demonstrated that the production of methanol and formic acid is possible using biogas as methane source. It could be one option for present and future fermentation plants to broaden their product portfolio beyond the generation of electrical energy and biomethane.

With this approach, a novel synthesis route towards both methanol and formic acid starting from methane is now available.