Hierarchical Fe-ZSM-5/SiC foam catalyst as the foam bed catalytic reactor (FBCR) for catalytic wet peroxide oxidation (CWPO)

Hierarchical Fe-ZSM-5/SiC catalyst as structured reactors for catalytic wet peroxide oxidation, showing good longevity and performance under flow conditions. A R T I C L E I N F O

Structured Fe-ZSM-5 zeolite supported on silicon carbide (SiC) foam catalyst was develop and applied as the packed foam bed catalytic reactor (FBCR) to the liquid phase catalytic wet peroxide oxidation (CWPO) of phenol under flow conditions. Fe catalysts were uniformly incorporated into ZSM-5 coatings using chemical vapour deposition (CVD). Fluid flow analysis of FBCR was performed by varying liquid flow rate, showing the transition of reactor type from the plug flow reactor to the continuous stirred tank reactor at about 1 ml min −1 . Systematic investigation of CWPO within Fe-ZSM-5/SiC FBCRs identified the optimum condition of operating the FBCR (80 mm bed length) as at 60°C and 1 ml min −1 with the turnover frequency (TOF) of 95 h −1 for phenol degradation (apparent activation energy = 3.4 kJ mol −1 ) and total organic carbon (TOC) conversion of 45.5%. It

Introduction
The increasing industrial activity of modern society has raised issues concerning environmental pollution. Many processes produce organic waste ending up on the atmosphere, soil and waterways, endangering all forms of life. One class of organic pollutants that is vastly produced by various industrial processes, such as oil refining, petrochemical and polyolefines processing, textile and leather processing industries [1], referring to phenolic compounds. Quantitatively, 70,190 tons per annum of phenol are produced in European territory (exposure limit: < 20 mg m −3 ), which has a relatively low degradation rate constant of 0.05 d −1 in surface waters. Low doses of 140-290 mg phenol kg −1 through the oral route, or dermal exposure could be lethal for humans due to its fast adsorption through tissues [2].
Destructive methods for phenol treatment include biological (e.g. biodegradation which is lengthy), thermal (e.g. incineration which is energy intensive) and chemical treatment (e.g. advanced oxidation processes, AOP), among which AOPs using ozone (O 3 ), hydrogen peroxide (H 2 O 2 ) and/or UV light stand out as a continuous progression, showing high efficiency in mineralisation of organic pollutants [3,4]. Catalytic wet peroxide oxidation (CWPO) processes employ ecofriendly non-toxic H 2 O 2 and hold good promise for potential applications in practical settings [5][6][7][8][9] since gas-to-liquid mass transfer (for O 3 ) and special reactor configuration (for UV) can be avoided. In CWPO, excessive H 2 O 2 is usually used to ensure the complete mineralisation of organics into CO 2 and H 2 O under mild conditions [10], for example, 14 mol of H 2 O 2 are needed to mineralise one-mole phenol fully. H 2 O 2 is catalytically decomposed to produce hydroxyl radicals, which has a very high oxidation potential (+2.8 V at pH 0) and high O-H bond energy (109 kcal mol −1 ) [10], virtually being able to oxidise any organic pollutants. Heterogeneous CWPOs with iron-based catalysts have the potential benefits of taking advantage of the Fenton process (of fast production of hydroxyl radicals [11]) and overcoming the problem of recovery, regeneration and reuse of catalysts [5,12].
However, the pressure drop issue of conventional packed bed configurations makes their practical applications challenging, especially handling liquid phase reactions at scales. To prepare a packed bed, pelletisation is needed to transfer powder catalysts into granules, spheres and extrudates [27], in which the use of inert binders, as well as the excessive shaping pressure, may result in the compromised catalytic activity. In addition, the maldistribution of reacting flows through packed bed reactors may also decrease the catalyst efficiency [28].
Solid open-cell cellular foams (based on base materials of polyurethane, carbon [37,38] and silicon carbide, SiC [39]) were proposed to develop flow-through reactors for environmental catalytic applications, especially photocatalytic treatment of chemical and biological pollutants in gas or liquid phase. Benefiting from features of the cellular structure, such as static mixer effect, high surface-to-volume ratio, low-pressure drop and high transparency to light [39], as well as properties of the base SiC (e.g. the semiconductive property [40]), SiC foams could be promising candidates for performing continuous-flow environmental catalysis. More importantly, SiC foam has gained the success at scales and demonstrated the possibility of industrial use [41]. For example, SiC foams have been used as distillation trays at pilot scale and outperformed the conventional trays regarding all key performance indicators [42].
Impregnation is commonly used to deposit catalysts on structured supports [6,21,51]. However, impregnation is prone to promote the inhomogeneous special distribution of active species across supports and the variation of catalyst sizes [52], which may abate the catalytic performance of the resulting structured catalyst. Recently, chemical vapour deposition (CVD) is shown to disperse catalytically active components on structured supports (e.g. LTA zeolite/stainless steel fibres [53]) homogeneously and facilely without requirements of drying and reduction.
In this study, the development of structured catalysts of Fe supported on ZSM-5 coated SiC foam (Fe-ZSM-5/SiC) is presented. Uniform incorporation of Fe catalysts in ZSM-5/SiC supports is achieved by CVD. Fe-ZSM-5/SiC catalyst is used as the foam bed catalytic reactors (FBCRs) and assessed by CWPO with phenol as the model compound under laminar flow conditions. The reaction temperature and flow rate are investigated systematically to understand their effects on the residence time distribution and catalytic performance of FBCRs. The longevity of the developed catalytic system is evaluated as well to assess the catalytic FBCRs for potential practical applications.

Preparation of ZSM-5 zeolite supported on SiC foam supports (ZSM-5/ SiC)
SiC foam supports with an average cell diameter of ca. 0.13 mm and porosity (ε) of 70% (10 × 10 × 80 mm) were used in this work. ZSM-5/ SiC was prepared by a dip-coating method (Fig. S1), as previously described in [33] and the average ZSM-5 coating on SiC foams was about 5 wt% (estimated by the weight gain of the support after the dip coating).

Synthesis of Fe-ZSM-5/SiC by chemical vapour deposition (CVD)
Fe-ZSM-5/SiC catalysts were prepared by a CVD method using iron (III) acetylacetonate (98%, Aladdin Industrial Corporation) as the precursor in a tube furnace (Hefei Kejing Materials Technology Co., Ltd.). Prior to CVD, the iron precursor and ZSM-5/SiC were mixed in a crucible and placed in the tube furnace under nitrogen (N 2 , the vacuum was applied to the system to extract air before introducing N 2 ). The evaporation of the precursor was carried out at 120°C for 1 h. The system was then heated up to 310°C (with a ramp rate of 5°C min −1 , held at 310°C for 2 h) for the deposition of iron species on ZSM-5/SiC to prepare Fe-ZSM-5/SiC catalysts. After CVD, the resulting Fe-ZSM-5/SiC catalyst was calcined at 550°C for 6 h (in a muffle furnace Nabertherm LE6/11 with P300 controller, ramp rate = 1°C min −1 ) prior to the catalytic test.
The quantity of Fe species in Fe-ZSM-5/SiC catalysts was determined by atomic absorption spectroscopy (AAS, Shimadzu AA-6800). Fe-ZSM-5/SiC samples were immersed in 3 wt% HCl solution overnight then solutions were analysed by AAS for the quantitative determination of Fe species. It was estimated that about 0.2 wt% iron species were present in Fe-ZSM-5/SiC catalysts (based on the total weight of the whole structured catalyst of 7.6-8.0 g).

Characterisation of materials
X-ray diffraction (XRD) patterns of materials were obtained using a Philips X'Pert X-ray diffractometer (CuKα 1 radiation, λ = 1.5406 Å, 40 kV, 40 mA, 5°< 2θ < 65°, 0.0167°step size). X-ray photoelectron spectroscopy (XPS) analysis was carried out on a Kratos AXIS Ultra DLD apparatus equipped with monochromated Al Kα X-ray source, a charge neutraliser and a hemispherical electron energy analyser. During data acquisition, the chamber pressure was kept below 10 −9 mbar and a pass energy of 40 eV. The spectra were analysed using CasaXPS software pack and corrected for charging using C 1s binding energy (BE) as the reference at 284.8 eV. The morphology and quantitative composition analysis of materials were acquired by a Philips XL30 FEG scanning electron microscope (SEM, beam acceleration voltage = 15 kV) equipped with an energy-dispersive (EDS) detector. Nitrogen (N 2 ) sorption analysis at −196.15°C was performed using a Quantachrome surface area analyser (Nova 2200). Before N 2 sorption analysis, samples were outgassed at 300°C overnight. The specific surface area of materials was determined using the Brunauer-Emmett-Teller (BET) method. Pore size analysis was performed using the Barrett-Joyner-Halenda (BJH) method and the desorption branch of isotherms. The meso-/macro-pore sizes of the samples were also analysed using mercury (Hg) intrusion porosimetry on Micromeritics' AutoPore IV 9510 (pressure range: 0.0007-413.6854 MPa).

Catalytic wet peroxide oxidation (CWPO) of phenol in foam bed catalytic reactors (FBCRs)
CWPO of phenol was carried out in the foam bed catalytic reactor (FBCR loaded with a single Fe-ZSM-5/SiC catalyst of 80 mm length and ca. 8.0 g weight, as seen the inset in Fig. 2) under continuous flow conditions and the experimental set-up is shown in Fig. 2. The synthetic analogue of wastewater containing phenol was prepared to have a concentration of 100 mg l −1 phenol (C phenol , or part per million, ppm). The system temperature was maintained by circulating isothermal water through the jacket and measured by a thermal couple. To start CWPO, a mixture of phenol solution and H 2 O 2 (molar ratio of H 2 O 2 to phenol = 14) was fed into the upflow FBCR via an HPLC pump (KNAUER P.4.1S). The mixture was preheated (by going through a tube which was immersed in an isothermal water bath) to the required temperature before going into FBCR. The effluent from CWPO was cooled down by an ice bath (to avoid evaporation) then collected for various analyses.
Different temperatures (T) of 25, 40 and 60°C (measured by a thermocouple at the outlet of FBCRs) and flow rates (F) of 0.5, 1 and 2 ml min −1 were investigated to study the effect of kinetics and hydrodynamics on CWPO. All experiments were carried out under the continuous flow condition for 6 h with internally collected samples for determining concentrations of H 2 O 2 , phenol, total organic carbon (TOC) and Fe. pH values of effluents were monitored as well using a pH were performed to understand the contribution of adsorption and liquid phase reaction to the measured catalytic performance. High-performance liquid chromatography (HPLC) was used to analyse concentrations of phenol (Fig. S2) and degradation products in the effluent. HPLC analysis was performed using an Agilent 1220 Infinity LC system equipped with a Diode Array Detector. An Agilent HC-C18(2) column was used for the detection of organics. Relevant HPLC conditions are: wavelength = 210 nm, oven temperature = 30°C, and sample injection volume = 20 μL, flow rate = 1 ml min −1 , mobile phase = ultrapure water and HPLC grade methanol with the volume ratio of 60:40. Calibration of HPLC analysis for phenol and calculation of conversions (phenol, TOC and H 2 O 2 ) are presented in the Supplementary Information (SI). TOC analysis of the effluent was performed on a TOC analyser (TOC-V CSH , Shimadzu). H 2 O 2 concentration was determined by the iodometric titration method. Iron leaching from the catalyst during continuous flow operation was measured by inductively coupled plasma optical emission spectrometry (ICP-OES, Thermo iCAP 6000 SERIES, 259.940 nm wavelength). Repeated reactions were performed with catalyst samples from the same synthesis protocols, showing that the conversion values were reproducible to better than ± 10%.

Pulse response experiments
Residence time distribution (RTD) measurements were performed using the pulse response experiment on FBCR. Deionised water (DI) was continuously flowed through FBCR to achieve steady state, then 2 ml tracer (4 wt% NaCl solution) was injected into the feed stream via a tee junction (to ensure that the tracer was inserted as a whole and avoid dispersion in the inlet flow, as indicated in Fig. 2). For each flow rate (i.e. F = 0.5, 1 and 2 ml min −1 ), at least three individual RTD measurements were performed to verify the results. The tracer was injected instantaneously (< 1 s), simulating a Dirac delta function of the concentration. The detection system was composed of a solids detector dipped inside 200 ml of DI, where the effluent stream was discharged.
The detected signal (at every five seconds on stream) was quantified at the ppm scale. The analysis of concentration profiles and RTDs were based on the RTD function (E(t)), cumulative distribution function (F (t)), average residence time (t m [min]), theoretical space-time (τ [min]) and variance (σ t 2 [min 2 ]) [54], as defined in SI.

Characterisation of materials
CVD was effective to deposit iron into ZSM-5/SiC support, as shown in Fig. 1, illustrating the colour change of ZSM-5/SiC from chalk white to dark brown after CVD. SEM-EDS elemental mapping of one crosssection of the strut of Fe-ZSM-5/SiC reveals a uniform distribution of Fe across the ZSM-5 coating (Fig. 3). In addition, EDS point analysis at different depths across Fe-ZSM-5 layer (thickness = ca. 30 μm) shows comparable Fe concentrations of 7.2, 7.1, 6.6 and 7.2 wt% (local mass fractions of Fe element in the sample), respectively, demonstrating the effectiveness of CVD for loading Fe within the ZSM-5 framework. XRD pattern of Fe-ZSM-5/SiC (Fig. S3) is similar to that of ZSM-5/SiC. Fe species are not identified by XRD analysis which may be due to the overlapped peak positions of iron oxide (2θ = 33.0°and 35.6° [21]) with the SiC phase. N 2 sorption analysis and Hg intrusion porosimetry (Fig. S4) were performed to understand the textural property of Fe-ZSM-5/SiC catalyst. BET surface area drops from 23.4 m 2 g −1 (ZSM-5/SiC) to 19.8 m 2 g −1 (Fe-ZSM-5/SiC) after CVD, indicating possible pore blockage by the deposition of Fe species. Meso-/macro-pore size distributions are presented in Fig. 4a with 4 nm mesopores probed by N 2 sorption and ca. 95 nm macropores by Hg intrusion. The hierarchical porous system of Fe-ZSM-5/SiC catalysts can be attributed to the intercrystal/particle pores of the ZSM-5 coating. The presence of meso-and macro-pores in the developed structured catalysts may be beneficial to CWPO in the liquid phase, in which large pores can contribute to mass transfer significantly since the diffusivity of species in the liquid phase is generally five orders of magnitude lower than that in the gas phase [55].
XPS analysis was carried out to understand the chemical state of Fe species in Fe-ZSM-5/SiC catalysts. Fe 2p XPS spectra are complicated to be fitted due to the complex multiplet splitting observed for atoms which contain unpaired electrons. Fe 2p 3/2 portion of the spectrum was   fitted for the fresh and used Fe-ZSM-5/SiC catalyst (Fig. 4b) according to the fitting parameters reported by Beisinger et al [56]. Fig. 4b shows a good fit for both the fresh and used catalyst when the spectral fitting parameters for Fe 2 O 3 were used (Table S1), suggesting that Fe is in the 3 + oxidation state in the two catalysts. Additionally, neither significant changes in Fe peaks nor Fe(0) (an absence of any peaks at 706.6 eV) are observed in either sample. In Fig. 4c, O 1s XPS profile shows three peaks in both the fresh and used catalyst. The peak at ∼530.0 eV can be assigned to the lattice oxygen in the metallic Fe oxide structures, confirming the main presence of Fe 2 O 3 , the peaks at ∼531.3 and ∼533.0 eV could be assigned as surface hydroxyl or oxygen species, unsurprisingly, small changes in these peaks are observed in the used catalyst in comparison to the fresh one.

Effect of temperature on CWPO in FBCRs
At F = 1 ml min −1 , the effect of reaction temperature on the performance of Fe-ZSM-5/SiC FBCR in CWPO was investigated with findings shown in Fig. 5 [10,59]. The apparent activation energies (E a ) of Fe-ZSM-5/SiC catalyst for H 2 O 2 decomposition and phenol oxidation were calculated according to Arrhenius equation (i.e. a plot of ln(TOF) versus T −1 giving a straight line, whose gradient can be used to determine E a ), as shown in Fig. 5c, the E a value of Fe-ZSM-5/SiC for phenol degradation is ca. 3.7 kJ mol −1 , lower than that reported for packed beds [60] and slurry reactors [24], also suggesting an improved catalytic system by the developed structured catalyst. Blank experiments were carried out to evaluate (i) the adsorption of phenol on Fe-ZSM-5/SiC under flow conditions (without the oxidising agent of H 2 O 2 ) and (ii) the phenol conversion caused by the non-catalytic oxidation of phenol by H 2 O 2 under flow conditions (without the catalyst). Fig. 6a shows that phenol has a strong affinity towards Fe-ZSM-5/SiC catalyst, e.g. a maximum disappearance rate of about 55% was measured at 60°C. As no H 2 O 2 was present in blank experiments, the measured phenol disappearance was solely due to the adsorption of phenol on FBCR. Adsorption was favoured at the relatively low temperature of 25°C, as expected. Negative values of phenol disappearance rate at 40°C (in the saturated region of > 4 h) might be attributed to the desorption of adsorbed phenol molecules back to the flow, causing the excess mass balance between the inlet and outlet of FBCR. Blank flow-through experiments with an inert FBCR (with a SiC foam insert) and H 2 O 2 /phenol mixture (molar ratio = 14) were also carried out, showing that phenol degradation by H 2 O 2 without catalyst was insignificant with an average phenol conversion of 16.5 ± 3.1% at 60°C under the steady state (Fig. 6b). Based on the findings from blank experiments, Langmuir-Hinshelwood mechanism was very likely accountable for CWPO through FBCRs, in which oxidation reactions on the surface of Fe-ZSM-5/SiC catalysts prevailed.
Steady-state TOC conversions are presented as the inset in Fig. 5b, showing lower conversions of TOC than the corresponding phenol conversions due to the production of the first and secondary intermediates (e.g. acids such as maleic acid, oxalic acid, acetic acid and etc.) in CWPO of phenol [61]. For Fe-ZSM-5/SiC catalysts, under the condition used by this work, catechol and p-benzoquinone were detected as the main first generation intermediates at temperatures of 25 and 40°C, contributing to TOCs of the effluent (Fig. S5), as well as the toxicity (half maximal effective concentration, EC 50 , phenol (16.7) > catechol (8.32) > p-benzoquinone (0.1) [62]). For FBCR at 60°C (Fig. S5), toxic intermediates were further decomposed into less toxic secondary acidic intermediates (e.g. EC 50, maleic acid ≈ 247, EC 50, oxalic acid > 450 and EC 50, acetic acid ≈ 130 [62]), decreasing the pH of the catalytic system (pH = 3.5-4, under steady-state conditions, Fig. S6) and causing the leaching of Fe from FBCR (Table S2). Although Fe species in the aqueous solution may cause homogeneous Fenton-like reaction, a number of previous work have demonstrated that heterogeneous catalysis played the major role in mineralising phenol compared to the homogeneous catalysis (e.g. higher H 2 O 2 degradation and TOC conversion can be observed through heterogeneous catalysis) [7,63].

Effect of macromixing in FBCR on CWPO
The cellular structure of Fe-ZSM-5/SiC catalyst is expected to play a role in enhancing hydrodynamics (i.e. improved transport phenomena), and hence the apparent rate of phenol degradation. In comparison to packed bed configurations (Fe 2 O 3 supported on MCM-41 [6] or SBA-15 [7]), Fe-ZSM-5/SiC catalyst achieved better TOFs than packed beds, e.g. 95.0 h −1 for Fe-ZSM-5/SiC versus 28.6 h −1 for Fe 2 O 3 @SBA-15 at ca. 60°C and 1 ml min −1 . To investigate the effect of hydrodynamics of FBCR on catalytic reacting flows, CWPO of phenol aqueous solutions over Fe-ZSM-5/SiC FBCR was performed at 60°C at flow rates of 0.5, 1 and 2 ml min −1 , respectively. It was found that the decomposition of H 2 O 2 was favoured by low flow rates. Under steady state conditions as in Fig. 7a, the average conversion of H 2 O 2 was measured to increase from about 59% to 88% by decreasing the flow rate from 2 to 0.5 ml min −1 . The pore-scale Reynolds numbers of the conditions used were estimated to be < 40 (Darcy flows), where the viscous force was prevailing [64,65]. Previously, we have shown that, for Darcy flows, foam supports were able to produce complex flow patterns at the millisecond scale and led to the axial and radial dispersion of species in and after cellular structures [66], as well as outperforming the conventional monoliths [35]. Therefore, the enhanced H 2 O 2 conversions at low flow rates (i.e. by ca. 49% from 2 to 0.5 ml min −1 ) could be explained by the improved liquid-to-solid transport by extending the residence time. The effect of flow rate on the conversion of phenol was minor as shown in Fig. 7b, showing that the first step of phenol degradation could be easily achieved by FBCR with > 59% H 2 O 2 conversions (at 60°C) under flow conditions.
Interestingly, the highest TOC conversion of 45.5% was achieved at 1 ml min −1 . A decrease in TOC conversion was measured by either increasing or decreasing of the flow rate, as the inset in Fig. 7b. In order to understand this phenomenon, pulse response experiments were carried out to characterise FBCR, gaining insight into the macromixing of FBCR on the continuous CWPO. Cumulative distribution functions (F (t) curves), concentration-time curves and residence time distribution (RTD) curves (E(t) curves) of FBCRs are shown in Fig. 8 (concentration profiles were smoothed using a fast Fourier transform filter to exclude noises). Other relevant parameters of RTD are summarised in Table 1.
Comparing normalised responses to pulse inputs, F(t) curves at the inlet and outlet of FBCRs show that the tracer needed much more time to exit FBCR than that needed to enter the reactor ( Fig. 8a and 8b). At 2 ml min −1 , it was observed that 80% F(t) of tracer molecules spent 16 min or less in FBCR, while 27 min and 50 min were estimated for flow rates of 1 and 0.5 ml min −1 , respectively.
Calculated RTD curves at the outlet of FBCR as a function of time are shown in Fig. 8d for three flow rates. At 2 ml min −1 , E(t) curve was narrow with reduced tailing, suggesting that FBCR behaved more like a non-ideal plug flow reactor (FB-PFR). At lower flow rates of 1 and 0.5 ml min −1 , E(t) curves developed long tails with relatively long residence times. Especially, the RTD at 0.5 ml min −1 is very close to continuous stirred tank reactors (FB-CSTR), implying the possible existence of stagnant zones and/or recirculating zones in FBCRs under these flow rates. Theoretical space time (τ, or ideal residence time for a perfect plug flow reactor, PFR) of three cases is shown in Table 1, also as the vertical solid lines in Fig. 8d. It was found that all E(t) curves arrived later than the expected τ, excluding the presence of stagnant fluids within FBCR. Therefore, the recirculation of fluid within the open-cell structure of FBCRs is likely to be responsible for the macromixing phenomena measured at 1 and 0.5 ml min −1 . Values of mean residence time (t m ) and variance (σ t 2 ) were deduced from E(t) for FBCR (as shown in Table 1). The value of σ t 2 at 0.5 ml min −1 is one order of magnitude higher than those at 1 and 2 ml min −1 , suggesting a greater RTD's spread of fluid elements in the FBCR at 0.5 ml min −1 (i.e. similar to a not well mixed CSTR). Based on the RTD characterisation of FBCR, ineffective phenol mineralisation at F = 0.5 ml min −1 (i.e. low TOC conversion of 33.1%)  could be explained due to the poor macromixing in FBCR, which could lead to the undesirable consumption of H 2 O 2 in CWPO such as (i) the conversion of H 2 O 2 into the less reactive hydroperoxy radical [10,12] and (ii) the consumption of hydroxyl radicals by the auto-scavenging reactions (with H 2 O 2 ) [10]. For FB-PFR at 2 ml min −1 , short-circuiting fluid and channelling might exist, resulting in the short residence time, and hence low conversions of H 2 O 2 and TOC (in comparison to FB-CSTR at 1 ml min −1 ).

Catalyst longevity and repeatability of FBCRs in CWPO
The longevity and repeatability of Fe-ZSM-5/SiC catalysts under flow conditions were evaluated over six-hour cycles at 60°C and 1 ml min −1 as shown in Fig. 9. For CWPO catalysed by Fe-ZSM-5/SiC FBCR, conversions of H 2 O 2 and phenol stabilised at between 15 and 60 min after the start-up of the system and remained on same levels for the continuous run. The catalyst used in the longevity test was not fresh and bed washing using DI was performed between cycles. Catalyst activity at phenol conversions < 100% (ca. 93%, C phenol(0) = 1000 mg l −1 ) was also checked showing no deactivation under the flow condition (Fig. S7). The catalyst longevity demonstrated by Fe-ZSM-5/SiC FBCR may have implications for relevant applications in practical settings where catalyst stability is often of paramount importance. Fe leaching as a function of the time-on-stream during the stability test was shown in Fig. S8, showing that the Fe leaching approached to a plateau at ca. 4.14 mg l −1 in the last two cycles. Postreaction XPS analysis of the used Fe-ZSM-5/SiC catalyst (Fig. 4b and c) showed that the chemical state of iron species in the catalyst remained the same as that found for the fresh catalyst. Interestingly, the leaching of Fe species did not result in the significant drop in the catalytic activity of the developed FBCR, i.e. the conversions at the same time were shown to be relatively stable (Fig. 9). The origin of this phenomenon can be attributed to the leaching of inactive part of Fe or formation of the active clusters during the reaction [22], which requires further investigation.

Conclusions
Fe (0.2 wt%) supported on ZSM-5/SiC catalysts were prepared by chemical vapour deposition (CVD) with good dispersion of Fe 2 O 3 in the ZSM-5 coating. Catalytic wet peroxide oxidation (CWPO) of phenol was carried out in Fe-ZSM-5/SiC based foam bed catalytic reactors (FBCRs) and showed good steady-state conversions of phenol (99%) and TOC (45.5%), as well as the TOF value of 95 h −1 , at the optimum condition of 60°C and 1 ml min −1 (bed length = 80 mm).   The characteristics of FBCRs were studied and research findings suggested that both hydrodynamics and residence time affected the decomposition of phenol significantly in FBCRs. RTD characterisation of the FBCR revealed that the reactor behaved differently before and after the transition flow rate of 1 ml min −1 , e.g. as a plug flow reactor at 2 ml min −1 and as a continuous stirred tank reactor at 0.5 ml min −1 . An optimum flow rate is necessary to operate the FBCR for continuous flow CWPO reactions, in which foam supports are able to maximise the macromixing as static mixers without compromising the residence time. In this work, it was anticipated that (i) short-circuiting and channelling at the high flow rate (2 ml min −1 ) and (ii) recirculating zones at the low flow rate (0.5 ml min −1 ) were plausible explanations for the reduced effective reactor volume, compromising the effective conversion of H 2 O 2 , and leading to less mineralisation of phenol in FBCRs. By comparing the effect of temperature and flow-rate on CWPO through FBCR, it was found that the effect of the system temperature on H 2 O 2 decomposition was more profound than that by the flow-rate.·H 2 O 2 conversion was improved by ca. 180% by increasing the reaction temperature from 25°C to 60°C at 1 ml min −1 . Conversely, it was decreased by ca. 38% by increasing the flow-rate from 0.5 to 2 ml min −1 at 60°C.
A longevity test of a Fe-ZSM-5/SiC based FBCR was carried out, showing a stable performance in terms of the phenol conversion. However, the Fe leaching from the Fe-ZSM-5/SiC catalyst (prepared by CVD) was observed, which is the concern for the further exploitation of the FBCR technology in practical settings. Further developments in the catalyst design are underway to address this issue.