Towards oxidative denitrogenation of fuel oils: Vanadium oxide-catalysed oxidation of quinoline and adsorptive removal of quinoline-N-oxide using 2,6-pyridine-polybenzimidazole nanofibers

2.3.1 Synthesis of 2,6-pyridine-polybenzimidazole

A general procedure for the synthesis of pyridine-PBIs (PPBIs) is described as follows (Sannigrahi et al., 2010): Pyridine dicarboxylic acid (3.9 g, 0.023 mol) and TAB (5 g, 0.023 mol) were added to a three-neck reaction flask in a nitrogen atmosphere, followed by 70 g of polyphosphoric acid (Scheme 1). The reaction mixture was stirred under nitrogen environment and the reaction temperature was ramped to 200 °C and stirring continued for 24 h. The reaction mixture became more viscous and developed a dark brown colour. A small amount of the reaction mixture was poured into water to isolate a brown mass. The mass was pulverized, neutralized with ammonium hydroxide, washed thoroughly with water, and dried in a vacuum oven for 24 h at 100 °C to obtain the 2,6-pyridine-polybenzimidazole.

Close

Scheme 1 Synthesis of 2,6-pyridine-polybenzimidazole (2.6-PyPBI).

Export to PPT

2.3.2 Electrospinning of imprinted 2,6-pyridine-polybenzimidazole solutions

2,6-Pyridine-polybenzimidazole solution was prepared by dissolving 2.0 g in 10 mL N,N-dimethylacetamide. Quinoline-N-oxide (0.001 M) in 1 mL of acetonitrile containing 40 µL Triton X-114 (surfactant agent) was added to the 2,6-pyridine-polybenzimidazole in N,N-dimethylacetamide. The dissolution of the mixture was carried out under reflux condition in a nitrogen atmosphere for 4 h. The resulting homogeneous viscous solution was cooled, and filtered to remove any particulates. The polymer solution was transferred into a 25 mL syringe and electrospun using the following optimized electrospinning conditions: A voltage of 25 kV at a flow rate of 0.3 mL/h, with a distance between the needle tip and collector plate maintained at 15 cm. The resulting nanofibers were subjected to an extensive washing process using a mixture of acetonitrile and methanol (1:1) via Soxhlet extraction to remove residual solvent (N,N-dimethylacetamide). The washed nanofibers were later dried overnight in an oven at 60 °C.

2.4.1 Metal content determination and leaching studies

Vanadium loading on fumed silica was determined by weighing out 0.0025 g into a vial, and 5 mL of TraceSelect HNO₃ (69%) was added. This mixture was heated at 50 °C for 48 h to leach out the vanadium. The acid-leached solution was then diluted with deionized-distilled water to 100 mL, filtered with 0.45 µM filters and analysed using ICP-OES. Prior to the vanadium loading analysis, five standard samples of known vanadium concentrations were run.

2.5.1 Investigation of OH^• radical generation during oxidation

Electron paramagnetic resonance (EPR) was employed for the detection of generated radicals during the catalytic oxidation process (Eaton et al., 2005). For this purpose, the following stock solutions were prepared: 1 mM V₂O₅, 10 mM t-BuOOH (in water) and 1 M 5,5-dimethyl-1-pyrroline N-oxide (DMPO). 150 μL of distilled water was added to an Eppendorf tube followed by the addition of 20 μL DMPO and 20 μL V₂O₅ of the stock 1 mM solution. The reaction was initiated by adding 20 μL of t-BuOOH in water. On addition of the oxidant, the solution mixture was quickly transferred into an EPR flat cell in order to acquire the spectrum. A control experiment was also carried out with the exclusion of V₂O₅. Another EPR study was also carried out in the presence of quinoline (0.1 mM) by the addition of 40 μL of 1 mM V₂O₅ solution.

2.5.2 Batch oxidation

The catalytic oxidation of a known quantity of quinoline (2.58 mL, 0.02 mol) was carried out using vanadium(V) oxide supported on silica as a catalyst (0.05 g, 0.000167 mol of vanadium) and t-BuOOH as an oxidant. In a 25 mL round bottom flask containing quinoline, 15 mL of ethanol was added and the solution was heated at 70 °C under continuous stirring at 300 rpm. The oxidation reaction was considered to begin after the addition of a required volume of t-BuOOH and the catalyst. An oxidant-to-substrate ratio of 7:1 was employed for the oxidation reaction.

2.5.3 Flow micro-reactor oxidation

Two glass syringes (1 and 2) (1 mL each), containing (1) reactant solutions, (quinoline, 0.02 mol) and (2) (t-BuOOH, 0.14 mol) were fed into a micro-reactor (Chemtrix catalyst reactor 3026) (which consisted of a catalyst chamber (0.05 g V₂O₅/SiO₂) and T-mixer (width 300 µm, depth 60 µm and volume 5.0 µL)) through PTFE tubing at a total flow rate of 25 μL/min. The delivery of the reactants was enabled by Chemyx Fusion 100 syringe pumps. The temperature was set at 70 °C and controlled using a Peltier heater placed directly under the reactor (Fig. 2). Electron paramagnetic resonance (EPR) was employed for the detection of generated radicals during the catalytic oxidation process (Eaton et al., 2005).

Close

Figure 2 A simple schematic set-up of the flow micro-reactor employed in the oxidation reaction. Two (1 mL) glass syringes containing, quinoline (0.02 mol) and t-BuOOH (0.14 mol) were fed into a micro-reactor which consisted of a T-mixer (width 300 µm, depth 60 µm and Volume 5.0 µL) at a total flow rate of 25 μL/min followed by a catalyst chamber containing V₂O₅/SiO₂ catalyst (0.05 g), at 70 °C.

Export to PPT

Please see Sections S1.1 and S1.3 in the supplementary data. The chemistry of the catalysed oxidation of quinoline is discussed in detail in the supplementary section (Sections S1.1–S1.4).

2.5.4 Calorimetric studies (oxidation reaction)

Micro-calorimeter was employed to quantify the amount and rate of heat release in terms of heat flow, heat and heat capacity. A sample mixture containing quinoline (0.02 mol), V₂O₅ (0.000167 mol of vanadium) and t-BuOOH (0.14 mol) was analysed using the isothermal micro-calorimeter operated at 25 ± 1 × 10⁻⁵ °C. The heat flow for the oxidation reaction ampoule is measured relative to the reference cell. A 30-min thermal equilibrium period was allowed before data collection began. The calorimetry study was carried out in the presence and absence of a catalyst. Calorimetric experimental runs lasted typically for 5 days before the experiment were terminated.

2.6.1 Oxidized quinoline (quinoline-N-oxide) adsorption

A solution of quinoline-N-oxide in ethanol (2 mL) was passed through a syringe packed with 50 mg of 2.6-PyPBI nanofibers. A flow rate of 0.1 mL h⁻¹ was employed for the adsorption process. Prior to adsorption, 2,6-pyridine-polybenzimidazole nanofibers were conditioned by wetting the adsorbents with methanol. The final concentration after adsorption was quantified using GC-FID.

2.6.2 Isothermal titration calorimetry (ITC) experiments

Isothermal titration calorimetry (ITC) which is based on the detection of the heat released or consumed upon titration of an analyte with (sub) microlitre aliquots of a titrant, was employed for thermodynamic characterization of interactions in solution (Wiles and Watts, 2009). ITC experiments were carried out at 25 °C by employing 10 mg of 2,6-PyPBI dissolved in 10 mL N-acetamide, and 20 μmol/L quinoline-N-oxide. 2,6-PyPBI vs quinoline-N-oxide titration was performed on a modular titration nanocalorimeter TAM III with an injection volume of 5 μL, a time spacing of 5 min between injections and a stirrer speed of 40 rpm. The reference cell was kept empty.

2.7.1 Interaction studies

The atomic level interaction of the quinoline-N-oxide with poly(pyridylbenzimidazoles) was predicted by molecular interaction studies using B3LYP functional with a SDD basis set. The enthalpies of formation, Gibb’s free energies and entropy of formation were calculated by using Eqs. (1) and (2), respectively.

ΔΔG, T and ΔΔS are the Gibbs free energy for the adduct formation, temperature (298 K) and entropy for adduct formation at standard conditions (i.e. 1 M concentration for solvents and 1 atm pressure) respectively. Isothermal titration calorimetry (ITC) which is based on the detection of the heat released or consumed upon titration of an analyte with (sub) microlitre aliquots of a titrant, was employed for thermodynamic characterization of interactions in solution quinoline-N-oxide vs. 2,6-PyPBI.

3.1.1 FTIR spectroscopy

The V₂O₅ synthesized at 600 °C showed peaks for V⚌O and V—O—V stretching vibrations at 1010 cm⁻¹ and 809 cm⁻¹ respectively, and the V—O stretching frequency was also observed at 413 cm⁻¹ (Fig. 3a). The absorption band at approximately 450 cm⁻¹ was assigned to the rocking motion of oxygen atoms bridging silicon atoms in siloxane bonds (Si—O—Si) (Fig. 3b). The most intense frequency observed in a silica spectrum is present around 1100 cm⁻¹ and is dominated by anti-symmetric motion of silicon atoms in siloxane bonds (Si—O—Si) Burcham et al., 2000. This broad peak consists of three components: (i) a broad shoulder centred at 1115–1130 cm⁻¹ assigned to longitudinal optical mode (Si—O—Si), (ii) a strong peak centred at 1035–1045 cm⁻¹ (transverse optical mode Si—O—Si), and (iii) a weak absorption at 950 cm⁻¹ due to silanol (Si—OH) stretching vibrations. The O—H stretching in the region 3300–3500 cm⁻¹ is due to H-bonded H₂O, hydroxyl terminals, H-bonded OH vibrations and H-bonded Si—OH in the chain (Fig. 3b). The absorption band at approximately 971 cm⁻¹ on the vanadium impregnated silica is attributed to Si—O—V linkages (Burcham et al., 2000; Kera and Hirota, 1969).

Close

Figure 3 (A) FT-IR spectrum of V₂O₅ produced after the calcination of NH₄VO₃ at 600 °C and (B) FT-IR spectra of silica and vanadium impregnated on silica (V₂O₅/SiO₂).

Export to PPT

3.1.2 X-ray diffraction (XRD) of synthesized V₂O₅, SiO₂ and V₂O₅/SiO₂

Room temperature XRD patterns of synthesized V₂O₅, SiO₂ and V₂O₅/SiO₂ catalysts are shown in Fig. 4 respectively. Topas® (V4.1) software was used to do full pattern Rietveld refinement to fit the diffraction pattern to the known reference pattern published by Filonenko et al. (2004) and Bruker AXS (2008) for the orthorhombic phase which was obtained from the Topas® Structural Database (Filonenko et al., 2004; Bruker, 2008). Amorphous SiO₂ gives a broad peak at around 2θ = 22, while V₂O₅ and V₂O₅/SiO₂ showed similar crystalline diffraction patterns that corresponded to the orthorhombic phase of V₂O₅. A similar diffraction pattern was observed for the V₂O₅/SiO₂ catalyst, with the crystalline peaks appearing on top of a broad diffraction halo which was due to the amorphous silica. The amorphous halo was refined by inserting a single peak fit at the maximum definable halo of the diffraction pattern. The peak position of the amorphous halo was subsequently fixed and the area of the peak was refined, depending on the size of the contribution of the amorphous halo. Within the Topas® software, the degree of crystallinity was calculated as the ratio between the total contributions of the crystalline material (V₂O₅) to that of the amorphous content (SiO₂). X-ray diffraction patterns obtained of material that was made at 500 °C showed additional small peaks at 2θ = 27.5° and 29°, in addition to the diffraction pattern of V₂O₅ (Fig. S1). The additional peaks were identified to be due to VO₂ and upon full pattern Rietveld quantification using Topas, the amounts were about 1.29% (Fig. S1). The presence of VO₂ can be attributed to the reduction in V₂O₅ by ammonium ions. The various diffractograms for the different NH₄VO₃ masses used for loading on SiO₂ (12.5, 25 and 37.5 wt%) at 600 °C are presented in Figs. S2–S4.

Close

Figure 4 XRD pattern of (A) V₂O₅, (B) SiO₂ and (C) V₂O₅/SiO₂ after calcination at 600 °C (Blue colour = experimental XRD pattern, red colour = Topas® refined theoretical V₂O₅ match based on reference pattern [ref], grey colour = observed difference plot between the measured and calculated patterns). The Space Group of the V₂O₅ was Pmnms with the refined lattice parameters a = 11.504 Å, b = 4.379 Å and c = 3.562 Å. Millers indices of V₂O₅ are presented in the supplementary information (Fig. S4b).

Export to PPT

The XRD patterns of V₂O₅/SiO₂ revealed that vanadyl phosphate phase was absent, (Wada et al., 1998; Makgwane et al., 2010, 2011) and this was necessary to confirm since phosphoric acid was used for the preparation. The catalyst that was made from 50 wt% (NH₄VO₃:SiO₂ mass ratio) was employed for the oxidation reaction and its vanadium content (as V not V₂O₅) was analysed by ICP-OES and was found to be 17 ± 1 wt%. The use of the catalyst, V₂O₅/SiO₂ @ 600 °C is because XRD confirmed V₂O₅ phase as the only phase obtained on the catalyst surface.

The degree of crystallinity (Table 1), i.e. the amount of amorphous SiO₂ to crystalline material (V₂O₅) within various calcined NH₄VO₃:SiO₂ wt% ratios, was calculated using Topas. The difference in the Topas wt% measurement was attributed to (i) the formation of V₂O₅ from NH₄VO₃ upon calcination and (ii) structural effect of the amorphous material (SiO₂) onto the structure of the crystalline materials (V₂O₅).

3.1.3 Scanning Electron Micrograph (SEM) and Electron Dispersive Spectroscopy (EDS) of V₂O₅/SiO₂

The SEM micrographs of the V₂O₅/SiO₂ showed no significant change w.r.t the fumed silica (SiO₂). Aggregated particles with uneven surface morphology in both fumed silica after functionalizing and V₂O₅/SiO₂, were observed. Hence, confirming the amorphous nature of the silica was preserved even after the immobilization of V₂O₅ at 600 °C. In addition, SEM showed that the shapes of V₂O₅/SiO₂ gave a higher diameter distribution range of 10–150 μm (Fig. S5a) as compared to the pristine fumed silica (0.2–0.3 μm). The use of EDS is reportedly sufficient for the identification of elements present in a material and not reliable for quantitative analysis. In the current work, EDS indicates the presence of vanadium on the catalyst support (SiO₂) (Fig. S5a).

However, the total vanadium content of V₂O₅/SiO₂ calcined at 600 °C was quantified using ICP-OES as presented in the experimental section and found to be 17 ± 1 wt%. SEM micrographs of V₂O₅/SiO₂ and SiO₂ are presented in Fig. S5b.

3.1.4 BET surface area

A Barrett–Emmett–Teller (BET) model was used to calculate the specific surface area and a Barrett–Joyner–Halenda (BJH) model was used to calculate the pore volume distribution and the average pore size. The BET surface area and average pore volume of fumed SiO₂ were found to be 301.5 m² g⁻¹ and 0.0183 cm³ g⁻¹, respectively, while that of V₂O₅/SiO₂ was found to be 67.4 m² g⁻¹ and 0.0036 cm³ g⁻¹ respectively. It was observed that the surface area of fumed SiO₂ reduced slightly upon functionalization and formation of V₂O₅/SiO₂. The drop in surface area going from fumed SiO₂ to V₂O₅/SiO₂ may be due to filling of the pores by vanadium, as evidenced by the slight drop in pore volume.

3.2.1 OH^• radical generation during oxidation

The reaction between V₂O₅ and t-BuOOH resulted in the generation of hydroperoxyl (OH^•) radicals (Scheme 1). The presence of the OH^• radicals was investigated by EPR spectroscopy using a spin-trap compound, 5,5-dimethyl-1-pyrroline N-oxide (DMPO), which binds with free radicals in solution to yield stable products known as spin adducts (Fig. S6) (Eaton et al., 2005). The hyperfine couplings of this radical indicated a DMPO-OH radical with observed A and g values of 12.1 cm⁻¹ and 2.4980 respectively. The production of hydroxyl radicals by oxidants in the presence of metal ions follows the mechanism reported by Walling (1975) and Hutzinger (1980).

3.2.2 Proposed catalytic oxidation route

V₂O₅ was generally perceived to be EPR inactive because it gives rise to a broad Zeeman line. This is due to the clustering of V⁴⁺ owing to significant dipolar interactions (Typek et al., 2010). The isolated V⁴⁺ (VO²⁺) species in V₂O₅ exhibit an eightfold hyperfine splitting derived from the interaction of free electrons (3d¹) with the magnetic nuclear moment of ⁵¹V (I = 7/2) (Typek et al., 2010).

In the synthesized catalyst, V₂O₅/SiO₂, the hyperfine structure was visible (Fig. 5A), as the interaction within V₂O₅/SiO₂ was attributed to the hopping of a mobile electron along V⁴⁺—O—V⁵⁺ bonds (Typek et al., 2010; Gupta et al., 1995). The presence of V⁴⁺, through the alleged super-exchange of electrons in V⁴⁺—O—V⁵⁺ bonds, from V₂O₅/SiO₂ calcined at 600 °C was confirmed using the EPR (Fig. 5A). This phenomenon was also reported by both Typek et al. (2010) and Gupta et al. (1995). The hyperfine interaction lines predominantly ascribed to V⁴⁺ of (V₂O₅/SiO₂) (Typek et al., 2010; Gupta et al., 1995; Shul'pin and Kozlov, 2003) disappeared upon the addition of t-BuOOH (Fig. 5B), indicating the oxidation of the V⁴⁺ to V⁵⁺. Addition of quinoline to this solution resulted in reduction of the metal centre to V⁴⁺ species, but with the hyperfine splitting not well resolved albeit with a clear Zeeman line (Fig. 5C). This phenomenon indicated that quinoline may have been chemisorbed onto the catalyst surface for the electron transfer processes to occur. The relative role of this liquid mass transfer mechanism is important for the reaction, as chemical species move from an area of high chemical potential on the catalyst surface to area of low chemical potential via diffusion.

Close

Figure 5 First derivative EPR spectra of suspended V₂O₅/SiO₂ catalyst (0.000167 mol, vanadium) (A) in EtOH/H₂O; (B) after addition of 10 equivalent of t-BuOOH to the solution of (A); (C) after addition of quinoline (0.02 mol) to the solution of (B); and (D) solution of (C) after 24 h.

Export to PPT

From the batch reaction process, we propose the oxidation mechanism as shown in Scheme 2. The reaction between t-BuOOH and V₂O₅/SiO₂ resulted in the generation of OH^• radicals (I) (Shul'pin and Kozlov, 2003). The generated OH^• abstracts a proton from quinoline to form quinoline radical (II). Reactive oxygen species (generated from the decomposition of t-BuOO^•) (Chuang et al., 1983) attack the quinoline radical to form reactive oxoquinoline and dioxo-quinoline radical species (III), hence leading to ring opening and further oxidation. Several products were formed through the intermediate III as continuous attack of the ring structure could lead to the formation of other products (see Section 3.2.4) (Thomsen, 1998). Wang et al. (2015) proposed the formation of the V-O-penolate surface species (complex) in the metal-based mechanism of V₂O₅ catalysed oxidation of chlorobenzene.

Close

Scheme 2 Summary of the proposed catalytic oxidation mechanism leading to the formation of quinoline radical species (Batch oxidation process).

Export to PPT

Evidence of possible radicals [such as t-BuO^•, t-BuOO^•, HO^• and R^• (quinoline-based radicals)] were noticed after the addition of quinoline to the mixture of t-BuOOH and V₂O₅/SiO₂ using the EPR, nonetheless, the EPR signal interpretation was difficult (Fig. S7). Interestingly, the formation of acetone as one of the reaction product within the matrix confirmed the decomposition product of t-BuO radical (Chuang et al., 1983). The recombination of t-BuO radical also lead to the formation of di-tert-butyl peroxide (Fig. S8). After a 24 h period, the catalyst recovers the EPR spectrum observed for V⁴⁺ as shown in Fig. 5D.

3.2.3 Quinoline oxidation reaction (batch process)

The oxidation reaction progressed steadily on addition of the oxidant, t-BuOOH, to quinoline in the presence of catalyst (V₂O₅/SiO₂) at 70 °C. When an oxidant:quinoline ratio 7:1 was used, there were no oxidation product(s) observed after 30 min although the EPR experiment showed that hydrogen abstraction had already occurred. However, it was very difficult to establish the hydrogen abstraction from GC–MS data, due to the highly reactive nature of the radical species (Fig. S9). After 3 h of oxidation reaction, several products were observed, such as 7-methylfuro[3,4-b]pyridin-5(7H)-one, nicotinaldehyde (positions 2 and 3), and nicotinic acid and pyridine-2-ol (Fig. S10). Further oxidation for another 2 h resulted in the formation of more products, amongst which is 7-methylfuro[3,4-b]pyridin-5(7H)-one (Fig. S11). More new products were observed after 12 h of oxidation and these are 1-(pyridin-2-yl)ethanone, pyridine-2,3-dicarboxylic acid and quinolin-5-ol (Fig. S12). Acetone was also observed as a new reaction product, hence confirming the decomposition of t-BuO radical formed via the recombination of t-BuOO radicals (Chuang et al., 1983). A total oxidation of 78% was observed after 12 h.

The quantification was based on the amount of quinoline left after quinoline oxidation. Some quantified yields of the various products formed after 12 h of oxidation are nicotinic acid (9%), 3-nicotinaldehyde (28%), 2-nicotinaldehyde (10%), 7-methylfuro[3,4-b]pyridin-5(7H)-one (4%), 1-(pyridin-2-yl)ethanone (5%), pyridine-2,3-dicarboxylic acid (2%), and quinolin-5-ol (3%). Mechanistic steps leading to the formation of various products are discussed in the next section. The oxidation trend for the formation of nicotinic acid, 3-nicotinaldehyde, 2-nicotinaldehyde, 7-methylfuro[3,4-b]pyridin-5(7H)-one and quinolin-5-ol over the 12 h span is presented in Fig. 6. A reduction in the amount of 7-methylfuro[3,4-b]pyridin-5(7H)-one after 7 h was as a result of further oxidation leading to the formation of 3-nicotinaldehyde.

Close

Figure 6 Formation of nicotinic acid (● NCA), 3-nicotinaldehyde (○ 3-NCH), 2-nicotinaldehyde (▾ 2-NCH), 7-methylfuro[3,4-b]pyridin-5(7H)-one (■ MFP) and quinolin-5-ol (Δ QUN) with time. (A mixture containing quinoline (2.58 mL, 0.02 mol), V₂O₅/SiO₂ catalyst (0.05 g, 0.000167 mol of vanadium) and t-BuOOH (7 mol equiv.) at 70 °C, stirring speed, 300 rpm).

Export to PPT

The oxidation of quinoline in the absence of a catalyst resulted in an overall oxidation yield of 9% after 12 h, with quinolin-5-ol accounting for about 96% of the total yield. The mechanistic route for the oxidation of quinoline in the absence of V₂O₅/SiO₂ could not be established. The generation of radicals (tert-butoxy radical and hydroxyl radical) was established for the decomposition of tert-butyl hydroperoxide by Morse (1957) and Swern (1970). According to the two studies, the generation of radicals in the absence of a catalyst is possible at relatively higher temperature condition as shown in equation below.

The oxygen molecule generated via the reaction of two molecules of t-BuOOH binds to the quinoline radical (Scheme 2). Lower oxidant-to-quinoline ratios (3:1) and (5:1) presented a total conversion of 56% and 69% respectively after oxidation for 12 h (Fig. 7). It is worth noting that the low amount of the oxidant was completely decomposed to O₂ thereby generating limited OH^• radicals necessary for H-abstraction of quinoline. Percentage conversions and turnover frequency (TOF) of the various oxidations are provided in Table 2.

Close

Figure 7 % Conversion on varying the ratio of quinoline-to-t-BuOOH at 70 °C. (A mixture containing quinoline (2.58 mL, 0.02 mol), V₂O₅/SiO₂ catalyst (0.05 g, 0.000167 mol of vanadium) and t-BuOOH (1, 3 and 7 mol equiv.) at 70 °C, stirring speed, 300 rpm).

Export to PPT

3.2.4 Mechanism of quinoline oxidation

The formation of OH^• radical responsible for the abstraction of hydrogen atom (H) from the available substrate (quinoline), is in accordance with the findings of Thomsen (1998) and Thomsen and Kilen (1998). The reactive oxygen molecule obtained from t-BuOOH decomposition binds to the quinoline radical to produce a reactive dioxo-quinoline (2) radical which could also carry out hydrogen abstraction of other quinoline molecules within the reaction mixture to form 5-hydroperoxyquinoline (3). Further protonation of molecule (4) could lead to the formation of quinolin-5-ol (4A) and/or quinolinone (Scheme 3, detected via GC–MS, Fig. S13).

Close

Scheme 3 H-abstraction from quinoline and subsequent formation of quinolin-5-ol (RH represents quinoline and R • represents quinoline radical).

Export to PPT

The oxidation of quinolin-5-ol produces quinolin-5-ol-1-oxide present in Scheme 3 (GC–MS chromatogram in Fig. S14). From the literature (Chuang et al., 1983; Swern, 1970), hydrogen abstraction on quinoline via OH^• takes place at positions 5 and 6, 5 and 7, with positions 5 to 8 resulting to the formation of similar end products. The reaction mechanism for the abstraction of hydrogen in the 5 and 6 position of quinoline followed by subsequent oxidation on those positions is shown in Scheme 4. Oxyquinoline radical is further oxidized with O₂ on the position 8 of the quinoline radical resulting in the release of OH^• radical, hence forming quinoline-5,8-dione ring. Further H abstraction and oxidation of quinoline-5,8-dione ring resulted in the formation of 2,2′-(pyridine-2,3-diyl)bis(2-oxoacetic acid) (Scheme S1).

Close

Scheme 4 Mechanism on the oxidation of quinoline intermediates.

Export to PPT

With the formation of 2,2′-(pyridine-2,3-diyl)bis(2-oxoacetic acid), the reaction environment becomes acidic which induces decarboxylation of the compound to form pyridine-2,3-dicarbaldehyde (Scheme S2). Oxidation and further decarboxylation lead to the formation of nicotinaldehyde (Scheme S2). Quinoline-5,6-dione obtained through the abstraction of hydrogen and oxidation (Scheme 4) also resulted in ring opening, similar mechanistic pathway as described for quinoline-5,8-dione was also obtained for quinoline-5,6-dione with the formation of more products (Scheme 4). Of importance is the observation of 7-methylfuro[3,4-b]pyridin-5(7H)-one (11) on the GC–MS chromatogram and this was attributed to the deprotonation of 2-(2-carboxyvinyl)nicotinic acid to form 2-(5-oxo-5,7-dihydrofuro[3,4-b]pyridin-7-yl)acetic acid, followed by decarboxylation. Furo[3,4-b]pyridin-5(7H)-one (12) occurred as a result of the oxidation of 7-methylfuro[3,4-b]pyridin-5(7H)-one radical generated by H-abstraction of molecule 11. It is proposed that the radical, 7-methylfuro[3,4-b]pyridin-5(7H)-one (11) –O₂ formed releases OH^• radicals which was then hydrolysed H₂O(H⁺/OH⁻) to generate furo[3,4-b]pyridin-5(7H)-one (12) and formic acid.

The mechanistic steps A {responsible for the conversion of 2-(2-carboxyvinyl)nicotinic acid to 2-(2-carboxyacetyl) nicotinic acid} and B {responsible for the conversion of 2-(5-oxo-5,7-dihydrofuro[3,4-b]pyridin-7-yl)acetic acid to acetic acid and pyridine-2,3-dicarboxylic acid} are presented in Schemes S2 and S3. GC–MS chromatograms of some of the identified products are available in the supplementary information (Figs. S13–S21).

3.2.5 Theoretical (DFT) study on the mechanism of some oxidation intermediates and products

The intrinsic reaction coordinate (IRC) of the reacting species in Schemes 3 and 4 was modelled in an attempt to follow some intermediates of interest in the oxidation pathway. It is to be noted that the energies contributed by the catalyst were not taken into account. The generation of OH^• responsible for the abstraction of hydrogen to form quinoline radical (1) was slightly endergonic (by +1 cal/mol). The reactive oxygen (O⁻) released during the decomposition of oxidant (t-BuOOH) reacts with quinoline radical to form a quinoline-oxygen radical adduct (2) which corresponds to the transition state (TS 2) with an energy of +40 cal/mol. This process is shown to require free energy to proceed (Fig. 7). The formation of 5-hydroperoxyquinoline (3) was endergonic as energy is released (+14 cal/mol). A further loss of OH^• radicals from (3) resulted in a quinoline-oxygen radical adduct, with an imaginary frequency of 1 (TS 4). The loss of OH^• was observed to be endergonic (+36 cal/mol). H-transfer from quinoline to (TS 4) resulted in the formation of quinolin-5-ol (4A), and this process resulted in the release of energy (+6 cal/mol) (Fig. 8).

Close

Figure 8 Energy profiles of the oxidation mechanism (see Schemes 2 and 3).

Export to PPT

Based on the oxidation mechanism shown in Scheme 3, the reaction of molecule 4 with oxygen molecule [O₂] to form molecule 5 was endergonic (+5 cal/mol, Fig. 8). An endergonic reaction (by +24 cal/mol) resulted in the formation of molecule 6, a transition state (TS 6). Molecule 6 undergoes ring opening to form molecule 7 with the change in free energy ΔG of −11 cal/mol. Somewhat release of free energy (+4 and +0.6 cal/mol, respectively) occurred when molecule 8 and 9 were formed through the oxidation and subsequent release of CO₂ from molecule 7. Molecule 7 undergoes alternative reactions (route 2) to produce molecules 10, 13 and 14 through free energies +26 cal/mol, −2 cal/mol and +12 cal/mol, respectively. Molecule 10 also undergoes other reactions via bond cleavage leading to ring opening to produce molecules 11 and 12 (+6 cal/mol and +1 cal/mol, respectively).

3.2.6 Calorimetric studies (oxidation reaction)

Calorimetric oxidation studies for both the catalysed and the uncatalysed systems showed exothermic heat flow (Fig. 9). The result clearly shows that the catalysed quinoline oxidation reaction exhibited a much higher exothermic heat flow of 1481 J (74.1 kJ/mol of quinoline) as compared to the non-catalysed systems, 6 J (0.3 kJ/mol of quinoline). The high heat flow reported from the catalysed system was mainly attributed to the generation of reactive oxygen molecule as well as other radicals from the reaction of catalyst and oxidant, and slightly contributed by the various reactions of quinoline oxidation intermediates generated hence leading to the oxidation product.

Close

Figure 9 Exothermic heat flow time curves for catalysed and uncatalysed oxidation of quinoline at 25 °C. (A mixture containing quinoline (0.02 mol), V₂O₅ (0.000167 mol of vanadium) and t-BuOOH (0.14 mol) were analysed using the isothermal micro-calorimeter operated at 25 ± 1 × 10⁻⁵ °C).

Export to PPT

3.2.7 Quinoline oxidation reaction (micro-reactor)

A micro-reactor was employed for the oxidation reaction as it provides ideal conditions such as higher heat and mass transfer rates, which enable fast and highly exothermic reactions to be performed under isothermal conditions. Thus, better product yield and selectivity are achieved as compared to the conventional reactors. The micro-reactor also allows a better process control, especially when dealing with toxic or hazardous substances (Watts et al., 2012; Wiles and Watts, 2011, 2009, 2007; Guangwen et al., 2008). Micro-reactor system was adopted in order to eliminate the formation of several oxidized quinoline products. The oxidation reaction presented a high degree of selectivity for quinoline-N-oxide with an overall oxidation yield of 67%. Herein, we propose that the reactive oxygen radicals generated from the reaction of t-BuOOH and V₂O₅/SiO₂ reacted with quinoline to form quinoline-N-oxide. The selectivity towards one product (quinoline-N-oxide) is due to a better controlled reaction thermodynamics offered by the micro-reactor as compared to the conventional batch reactor (Watts et al., 2012; Wiles and Watts, 2011).

The product (quinoline-N-oxide) was isolated and analysed by GC-FID (Fig. S22), GC–MS (Fig. S23) and NMR (Fig. 10). Quinoline-N-oxide: ¹H NMR (400 MHz, CDCl₃) δ 8.81 (s, 1H), 8.12–7.98 (m, 2H), 7.72 (d, J = 8.0 Hz, 1H), 7.62 (t, J = 7.2 Hz, 1H), 7.45 (t, J = 7.1 Hz, 1H), and 7.33–7.26 (m, 1H). ¹³C NMR (400 MHz, CDCl₃) δ 150.06, 147.84, 136.36, 130.07, 129.01, 128.50, 128.26, 127.77, 126.61, 121.04. The NMR data obtained are in agreement with those reported by Ding et al. (2011). Quinoline: ¹H NMR (400 MHz, CDCl₃) δ 8.54 (s, 1H), 7.84 (d, J = 8.4 Hz, 1H), 7.59 (d, J = 8.3 Hz, 1H), 7.29 (t, J = 9.6 Hz, 2H), 7.07 (t, J = 7.4 Hz, 1H), and 6.88–6.80 (m, 1H). ¹³C NMR (400 MHz, CDCl₃) δ 150.19, 150.02, 149.75, 148.00, 135.54, 135.25, 129.10, 128.80, 127.71, 126.31.

Close

Figure 10 Overlaid ¹H NMR spectra of quinoline and quinoline-N-oxide (Inset: ¹H NMR of quinoline-N-oxide).

Export to PPT

3.3.1 FT-IR of 2,6-PyPBI nanofibers

The brownish imprinted 2,6-pyridine-polybenzimidazole (2,6-PyPBI) nanofiber obtained after electrospinning and washing, displayed similar IR spectrum when compared to the synthesized 2,6-pyridine-polybenzimidazole, hence indicating that polymer properties were retained. The FT-IR spectra of imprinted nanofibers are presented in Fig. S23. FT-IR (ν, cm⁻¹) synthesized 2,6-PyPBI: 3301 (N—H), 1621, 1487 (C⚌N), 1324 (C⚌C) and FT-IR (ν, cm⁻¹) 2,6-PyPBI nanofibers: 3164 (N—H), 1611 (C⚌N), 1436 (C⚌C), 1278 (C = C—C aromatic), 900–680 (C—H aromatic).

3.3.2 SEM and BET surface area of 2,6-PyPBI nanofibers

SEM images of 2,6-pyridine-polybenzimidazole nanofibers are presented in Fig. 11. The electrospun nanofibers displayed uniform morphology without the formation of beads. A diameter range of 0.34–2.21 µm was presented by the nanofibers. The BET surface area and average pore diameter of 2,6-PyPBI nanofibers were observed to be 6.65 m² g⁻¹ and 54.8 Å, respectively.

Close

Figure 11 SEM micrograph of 2,6-pyridine-polybenzimidazole nanofibers.

Export to PPT

3.4.1 Computational studies

Quinoline-N-oxide was adsorbed using pyridine-polybenzimidazole nanofibers. Prior to material synthesis, four sets of pyridine-polybenzimidazoles, previously synthesized by Sannigrahi et al. (2010), were modelled to evaluate the polymer with the best electronic properties suitable for quinoline-N-oxide adsorption. Computational analysis was calculated with SDD basis set using Gaussian 09 (Frisch et al., 2009).

The highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energies for the various pyridine-polybenzimidazoles are presented in Fig. S24. Hard molecules/adducts have a large HOMO–LUMO gap, while a small HOMO–LUMO gap depicts high polarizability, a characteristic of soft molecules/adducts, therefore a small amount of energy is required to get the pyridine-polybenzimidazole with small HOMO–LUMO gap to the excited states (Hizaddin et al., 2013). From the modelled pyridine-polybenzimidazoles as well as its interactions with quinoline-N-oxide, 2,6-pyridine-polybenzimidazole presented the lowest energy gap (Fig. S24) and was employed as the suitable polymer for the adsorption studies. Molecular modelling further confirmed π-π stacking interactions between quinoline-N-oxide and pyridine-polybenzimidazole. NBO further confirms the changes in the electronic properties of the polymers before and after interaction (Fig. S25).

3.4.2 Oxidized quinoline (quinoline-N-oxide) adsorption

A solution of quinoline-N-oxide in ethanol (2 mL) was passed through a syringe packed with 50 mg of 2.6-PyPBI nanofibers. A flow rate of 0.1 mL h⁻¹ was employed for the adsorption process. Prior to adsorption, 2,6-pyridine-polybenzimidazole nanofibers was conditioned by wetting the adsorbents with methanol. The final concentration after adsorption was quantified using GC-FID. A maximum adsorption of 86% quinoline-N-oxide was observed with an adsorption capacity (q_e) of 4.8 mg/g.

3.4.3 Isothermal titration calorimetry (ITC) studies

Isothermal titration calorimetry (ITC) which is based on the detection of the heat released or consumed upon titration of an analyte with (sub) microlitre aliquots of a titrant, was employed for thermodynamic characterization of interactions in solution (Ogunlaja et al., 2016). ITC experiments were carried out at 25 °C by employing 10 mg of 2,6-PyPBI dissolved in 10 mL N-acetamide, and 20 μmol/L quinoline-N-oxide. 2,6-PyPBI vs quinoline-N-oxide titration was performed on a modular titration nanocalorimeter TAM III with an injection volume of 5 μL, a time spacing of 5 min between injections and a stirrer speed of 40 rpm. The reference cell was kept empty.

The ITC interaction titrations for 2,6-pyridine polybenzimidazole (2,6-PyPBI) vs quinoline-N-oxide are presented in Fig. 12. The free energy (ΔG_b), enthalpy (ΔH_b) and entropy (ΔS_b) of interaction are presented in Table 3. A decrease in free energy of the system (ΔG_b < 0) confirmed that the adsorption process was spontaneous. A decrease in the randomness of the interacting molecules was observed, hence giving rise to negative entropies (ΔS_b < 0). The negative enthalpy (ΔH_b) value for the interaction contributed to an exothermic process. This confirmed a favourable interaction between 2,6-pyridine-polybenzimidazole and quinoline-N-oxide.

Close

Figure 12 Isothermal titration calorimetric analysis of quinoline-N-oxide with 2,6-pyridine polybenzimidazole (2,6-PyPBI). (Ampoule containing 700 μL of 10 mg of 2,6-PyPBI dissolved in 10 mL N-acetamide was titrated against 200 μL quinoline-N-oxide (20 μmol/dm³) with an injection volume of 5 μL, a time spacing of 5 mins between injections and a stirrer speed of 40 rpm).

Export to PPT

It was observed via DFT studies that quinoline-N-oxide interacts with 2,6-PyPBI through hydrogen bonding (the pyrolle-type amine group of 2,6-PyPBI forms a hydrogen bond with one of the oxygen atoms of the quinoline-N-oxide), pi-pi stacking and extensive number of van der Waals interactions (Ogunlaja et al., 2014). The binding modes as well as the HOMO and LUMO positions of adduct are shown in Fig. 13. The value obtained for the enthalpy of formation (ΔΔH < 0), Gibb’s free energies (ΔΔG < 0) and entropy of formation (ΔΔS < 0) agrees with the ITC data (Table 3).

Close

Figure 13 (A) HOMO and (B) LUMO positions of 2,6-PyPBI vs quinoline-N-oxide.

Export to PPT

Computational interaction studies using (A) 2,5-pyridine-polybenzimidazole (2,5-PyPBI), (B) 2,4-pyridine-polybenzimidazole (2,4-PyPBI), and (C) 3,5-pyridine-polybenzimidazole (3,5-PyPBI) presented result with +VE Gibb’s free energies, hence confirming a non-spontaneous adsorption process (Table S1).