Engineering the site-accessibility for robust, continuous flow-through hydrogenation of styrene

Abstract

The evolution of micro-scale chemical factories operating under flow conditions has fueled growing interest in designing new catalytic materials capable of sustaining multiphasic reactions with enhanced kinetics and selectivity over extended periods. Unlike batch-mode, the flow-through configuration demands continuous catalytic interfacial engineering stretching from the micro-scale of the active site to the macro-scale of the reactor geometry. Here, we address this multi-scale, multiphasic challenge by demonstrating a macro-scale, shape-tunable carbon monolith, chemically decorated at the microscale with well-defined nickel (Ni) nanoparticles. The resulting Ni-monolith is both processable and mouldable into desired geometries, offering precise control, high selectivity, and excellent yield for the catalytic mono-hydrogenation of styrene under continuous flow conditions over prolonged durations. Through kinetic analysis, we reveal that the outstanding performance of the granulated Ni-on-monolith (NOM-G) catalyst is enabled by a transition in the rate-determining step from half-hydrogenation of styrene to complete hydrogenation forming ethylbenzene. Additionally, the engineered turbulent flow channels within the NOM-G monolith significantly enhance the interfacial accessibility of reactants to active sites, sustaining high yields over extended timescales. As a result, the system delivers one of the highest space-time yields (231 g L⁻¹ h⁻¹) reported thus far and establishes the viability of multiphasic reactions through flow.

---

1 Introduction

Energy-efficient and selective catalytic processes enabled by advanced materials are gaining increasing attention as a strategic pathway toward sustainable chemical manufacturing. Herein, heterogeneous catalysis has witnessed the largest industrial scale adoption due to the scalability, recyclability and the degree of controllability it provides over key process parameters.^1–3 In fact, the twelve principles of green and sustainable chemistry have transformed into important guidelines that emphasize the significance of catalysis as a pivotal element for synthetic transformations.⁴ Heterogeneous catalytic hydrogenation is one such process that assumes significance in both lab-scale and industrial settings. The hydrogenation of organic compounds involves the addition of molecular hydrogen in the presence of a suitable catalyst, typically a precious metal. Metal-catalytic systems are ideal candidates based on Sabatier's principle, which suggests that the best catalyst binds molecules just right—strong enough to adsorb and activate reactants while binding weak enough to enable product formation followed by desorption – best represented by the volcano plot.^5,6 In this regard, batch-mode reactions have traditionally been preferred for chemical transformations due to their operational flexibility and suitability for multi-step synthesis.⁷ They allow for easy modifications in reaction parameters, making them ideal for pharmaceuticals, specialty chemicals, and research applications.⁸ Additionally, batch reactors can accommodate a wide range of reaction types, including those requiring long residence times or complex reagent additions. However, one of the main challenges in scaling up of a batch-scale process arises from intrinsic heat-transfer limitations leading to large thermal non-uniformity in the reaction mixture. Such poor control over the distribution of heat across the batch fuels high energy demand for the process, leading to poor energy efficiencies. Further limitation of batch-scale reactions includes significant downtime between filling, reacting, and emptying cycles, which reduces the overall throughput. Scalability is also a challenge, as increasing reactor size can lead to inconsistent mixing and difficulties in maintaining uniform reaction conditions.⁹ Moreover, batch processes often have higher operational costs and labour requirements, since manual intervention is frequently needed for monitoring and handling.¹⁰

Continuous flow hydrogenation protocols in contrast to such batch-mode operations, offer an attractive alternative that enables high space-time yield (STY), precise control over process parameters, and stoichiometry that leads to improved product selectivity and enhanced cyclability, thereby promoting atom-economy.¹¹ A typical continuous flow system would contain the catalysts loaded in the flow channels across which the chemical conversion is enabled by the passage of the reactants. While greater control of the process parameters is enabled in such a configuration, the residence time and space-time yield emerge as critical parameters that dictate the outcome of the reaction, its energy implications and overall sustainability of the process.^12,13 The ability to continuously introduce reactants and immediately remove products enhances the overall throughput, making it possible to achieve higher production rates without the need for larger reactor volumes.¹⁴ Moreover, precise application of external reaction-driving parameters such as heat, electric field, magnetic field, light and pressure has been demonstrated in flow-mode reactors, thereby increasing their versatility.¹¹ Importantly, the heat and mass-transfer limitations in a conventional batch process can be easily overcome through the flow-geometry. The continuous nature of flow chemistry also allows for easier product isolation since reaction products can be collected directly from the outlet without extensive filtration or separation steps that are common in batch processes. Furthermore, scale-up in continuous flow systems is straightforward; it simply involves adjusting flow rates rather than redesigning reactor systems as required in batch processes. This flexibility allows for rapid adaptation to different reaction conditions or scales.⁷ By enabling better mixing, controlled reaction times, and real-time monitoring, flow reactors reduce energy consumption, minimize waste, and enhance safety, making continuous flow systems a promising solution for modern chemical manufacturing.

Organic hydrogenation reactions often necessitate the mixing of liquid-phase olefins with gas-phase hydrogen, followed by their interfacing at the active sites of the solid catalyst.^15–17 Given the multiphasic nature of this reaction, maintaining the control over the stoichiometry of the feedstock and simultaneously ensuring its availability at the catalytic sites in the continuous flow system is challenging.¹⁸ Furthermore, the intrinsic flow pathways offered by the catalytic column should facilitate extensive interfacing between the reactant and the catalyst, while simultaneously minimizing the pressure drop across the reactor column. In addition, strategies for reactivation of the catalysts are desirable, particularly after increased reaction timescales and throughput.^18–20 In this context, a catalytic flow-through process achieving rapid, selective hydrogenation of olefins is highly desirable and poorly investigated, particularly due to the challenging multiphasic solid–liquid–gas interphase nature of the reaction. The design of a robust catalyst that can be machined into desired arbitrary geometries, while achieving high selectivity, lifetime and high space-time yield is highly desirable in the context of green and sustainable chemistry.

Here, we report a materials-based approach to design advanced catalysts with porosities that are tunable at both the macro-scale and micro-scale. Towards this, we leverage the chemical anchoring of well-defined nickel nanoparticles (Ni-NPs) in precisely chosen chemical environments on a carbon-based support. Such Ni-decorated carbon catalysts are further processed and shaped into monolithic structures that simultaneously provide extensive interfacial access to the flow of the reactants to deliver high space-time yields and reaction kinetics, with excellent selectivity and control.

2 Experimental

2.1. Materials

For the carbon honeycomb monolith based catalysis, run-of-mine Loy Yang (LY) brown coal (<3 mm), KOH pellets (Merck®, Darmstadt, Germany), Methocel® (proprietary mixture of methylcellulose and hydroxypropyl methylcellulose from Dow®, Michigan, United States), paraffin (Merck®, Darmstadt, Germany), glycerol (Merck®, Darmstadt, Germany), nickel acetate tetrahydrate solution (Ni(CH₃COO)₂·4H₂O, Sigma-Aldrich (Australia)) were procured and used. All precursors were used directly without additional purification.

2.2. Methods

2.2.1. Synthesis of Ni-based CHM-supported catalysts. The porous carbon honeycomb monoliths were synthesized from Victorian brown coal (VBC) through an extrusion process that utilizes VBC as the primary carbon precursor, incorporating various additives according to an established protocol.²¹ Step-by-step synthesis procedure is detailed in the SI (Fig. S1). Such a structured honeycomb carbon monolith demonstrates a high surface area (1160 m² g⁻¹) and pore diameters of 3–5 nm. It features an interconnected network of macropore channels and a sponge-like structure with cavities providing large accessible surface area (Fig. S2 and S3).²² Employing their unique morphological and physical attributes, carbon honeycomb monoliths (CHMs) were utilized as the catalyst support and Ni nanoparticles were incorporated following two different approaches.
Approach 1: Ni on monolith-granular (NOM-G). In this approach, the as-prepared monolith block was first crushed and sieved to form granulated monoliths in the size range of 180–250 μm. 1 g of such granular monoliths was added to 10 ml of a 6 wt% nickel acetate tetrahydrate solution (Ni(CH₃COO)₂·4H₂O), and the mixture was stirred and sonicated for about 2 h. During the experiment, samples of the solution were collected at 10 min intervals to check the metal loading on the monolith particles using spectroscopic measurements (UV-vis) until equilibrium was achieved. The Ni-loaded monolithic particles were subsequently washed repeatedly with distilled water, centrifuged, and then freeze-dried overnight. To convert the loaded Ni ions to Ni metal, the dried particles were further reduced under ∼99% pure hydrogen in a custom-built flow-through setup (refer to Fig. S6) at 300 °C for 2 h, forming the NOM-G catalyst (Fig. S4).
Approach 2: Ni in monolith (NIM). For the NIM catalyst, the nickel precursor was added during the kneading process. The optimized formulation was identified based on the plasticine-like consistency of the dough and is detailed in Table S5. The kneaded dough was subsequently extruded and air dried to form sturdy blocks of the NIM catalyst. Subsequently, the monolith blocks were carbonized at different temperatures (600 °C, 700 °C, 800 °C and 900 °C forming NIM-600, NIM-700, NIM-800, and NIM-900 respectively) to identify the best carbonization temperature for forming a uniform metal distribution. Post carbonization, the monolith blocks were passivated with air before recovery from the carbonization furnace. This was essential to avoid the development of a rapid exotherm as the Ni surface is oxidized by air, which would then cause the carbon to combust. The steps of carbonization including passivation with an air–N₂ mixture are described in the SI (Fig. S5). After carbonization, the NIM-900 catalyst was processed in two different ways to study the effect of the monolith structure on the catalytic activity.
(i) Ni in monolith block (NIMB-900). After carbonization, the CHM was machined to a reduced dimension (length ≃ 3.5 cm, diameter ≃ 1.5 cm) so that it could be fitted into the reactor column (Fig. S7). The NIM-900 catalyst was found to be sturdy and could be machined successfully to the desired dimensions. This small block of monolith was used directly for testing the activity of the NIMB-900 catalyst for the styrene hydrogenation reaction in the flow-through configuration.
(ii) Granulated Ni in monolith (NIM-G). Herein, the block of the NIM-900 catalyst was crushed into the size range of 180–250 μm to form the granulated NIM (NIMG-900) catalyst which was further packed into the reactor column to test its catalytic activity (Fig. S7).

2.2.2. Catalyst characterization. The morphology of the prepared catalysts was confirmed by scanning electron microscopy (SEM: ZEISS Ultra 55 Field Emission Scanning Electron Microscope, 5 keV) and transmission electron microscopy (TEM: FEI Tecnai G2, F30, 300 kV). XRD patterns were obtained using a Bruker D8 Advance diffractometer with a Cu radiation source (Kα = 1.54 Å) at 20 kV and 40 mA using a LynxEye detector. The samples were scanned from 15° to 100° at a step size of 0.05° and an acquisition time of 2 s per step. X-ray photoelectron spectroscopy (XPS; Axis Supra model, SHIMADZU group) was employed to investigate the surface characteristics of the catalysts. In this study, peak fitting of the Ni 2p spectra focused exclusively on the Ni 2p_3/2 region; in line with the standard practice reported in several literature sources.^23–27 The binding energies and full widths at half maximum (FWHMs) for the deconvoluted peaks corresponding to Ni metal, NiO, and Ni(OH)₂ were selected based on values reported by Davydova et al. (refer to the SI).²⁴ Both CO₂ (Dubinin–Radushkevich) and BET (N₂) surface area were measured using a TriStar II 3020 (Micromeritics, USA). Prior to analysis, each sample was de-gassed at 110 °C for ≃18 h, and back-filled with N₂ on a Micromeritics VacPrep 061. The specific surface area normalized with the weight of the catalysts was determined by analysing their N₂ and CO₂ adsorption isotherms through the Brunauer–Emmett–Teller (BET) and Dubinin–Radushkevich (D–R) equations, respectively.^28–30 Additionally, the pore size and pore volume distribution were determined using the BJH adsorption method. The metal concentration of the catalyst samples was determined using a PerkinElmer Avio 200 ICP-OES spectrometer. About 10 mg of solid catalysts were ashed in a muffle furnace to remove the carbon phase, which is known to protect some metal NPs from acid dissolution. After ashing, the residue was treated with acid (EMPARTA® ACS, Merck, HNO₃ ∼ 69%) under continuous agitation using a 5-roller tube mixer for 48 hours. The catalyst particles were separated, and the process was repeated to ensure complete etching of the metals into the acid solution. A 0.1 ml aliquot was taken from the combined centrifuged digestate and further diluted with 2 wt% HNO₃ to a total volume of 10 ml. The diluted solution was then passed through a 0.2 μm syringe filter before analysis and used for final measurements. NEXAFS measurements were undertaken at the Australian Synchrotron (AS) on the Soft X-ray (SXR) beamline.³¹ Both carbon and oxygen K-edge NEXAFS spectra were acquired on all the Ni-CHM catalysts in total electron yield (TEY), partial electron yield (PEY) and total fluorescence yield (TFY) modes. The C K-edge NEXAFS spectra were collected over a photon energy range of 270–340 eV. The energy step used was 0.1 eV and the dwell time of each reading was 0.8 s. Data processing and fitting were subsequently performed. The method employed involved using QANT to normalise the spectra with an I₀ signal as well as normalising the pre-edge to zero and post edge to 1.³² For C K-edge scans we also used a photodiode signal collected to ‘double normalise’ the spectra in order to remove any carbon contamination from the beamline prior to normalising pre and post edges to 0 and 1 respectively.³³ The liquid products were analysed using GC-MS: ISQ 7000 Single Quadrupole GC-MS system, EI/CI source, Thermo Fisher Scientific equipped with an FID detector and a TC-WAX capillary column. The amount of H₂ desorbed was determined from the H₂-TPD study using a BELCAT-II Catalyst Analyzer unit (Microtrac BEL, Corp.) equipped with a thermal conductivity detector (TCD). It was assumed that the amount of H₂ desorbed was equal to the amount of H₂ chemisorbed. During the H₂-TPD studies 30 mg of each catalyst was initially reduced in situ in flowing H₂ at 300 °C for 1 h, followed by cooling to 50 °C in the same environment at a rate of 30 ml min⁻¹. The sample was held at 50 °C under flowing Ar to remove weakly bound physiosorbed species. With a flow of Ar the temperature of the sample was ramped to 900 °C at a rate of 10 °C min⁻¹ and maintained at hold for 0.5 h under flowing Ar. The H₂-TPD profile was then integrated and the number of moles of H₂ desorbed was determined by using MFC calibration of H₂ in Ar. The area under the H₂-TPD profile was used to determine the % Ni dispersion using the following equation:³⁴where mol of H₂ = number of moles of H₂ desorbed, calculated from the H₂-TPD curve (mol H₂ per g_cat), SF = stoichiometric factor (1.0), and MW of Ni = molecular weight of atomic Ni (58.69 g mol⁻¹).

2.2.3. Hydrogenation set-up. The catalyst activity tests with CHM-supported catalysts were conducted in a Phoenix™ Flow Reactor equipped with a digital temperature controller for the furnace, HPLC pump, gas module and pressure module (Fig. S8). In a standard reaction, a known quantity of catalyst (granulated or block), typically ≃1 g, was loaded into the reactor column sandwiched between small beds of ZrO₂ beads as an inert support. 1% styrene in isopropanol solution was used as the reagent and reactions were carried out under various conditions. A fixed substrate-to-catalyst ratio (mol_styrene : mol_{active metal} = 18 : 1) was maintained for all reactions. Post run, the reactor path was washed with pure isopropanol to ensure no product remained from the previous run. Subsequently, the products were collected and further diluted to analyse via gas chromatography-mass spectrometry (GC-MS). The liquid products were analysed using GC-MS (Agilent GCMS 5977B, HES EI source, TC-WAX capillary column). After the reaction, the spent catalysts were washed with pure isopropanol and further analysed using various characterization techniques. During the reaction the styrene conversion and ethylbenzene selectivity were calculated using eqn (2.2) and (2.3).

The rate of the reaction, TOF and space-time yield (STY) for the flow-through system were calculated using the following formula:

3 Results & discussion

3.1. Design principles for synthesis of Ni-based monolith-supported catalysts

The unique catalyst design strategy described here leverages the surface textural properties of CHMs to design two well-defined and distinctly different chemical environments around the metal (nickel) nanoparticles, while retaining other parameters such as dimensions, weight fraction and crystallinity (Fig. 1). Leveraging these advantages, we developed complementary strategies that result in two different catalysts possessing significantly differing spatial access of the reactants to the catalytic sites, while exhibiting identical chemical constitution, as discussed below (full details in Section 2.2.1).

Fig. 1 Schematic representation for the design of Ni on monolith (NOM) and Ni in monolith (NIM) catalysts, highlighting their impact on reaction facilitation and overall mechanism.

The first strategy involves utilizing the high surface area of the pre-formed CHMs (790 ± 10 m² g⁻¹), along with the additional terminal carboxylic functionalities (–COOH) as anchoring points, to facilitate the nucleation of Ni²⁺ ions.²² This is followed by gas-phase reduction of the chelated Ni²⁺ under a H₂ stream (100 sccm, 300 °C, 2 h) leading to the formation of Ni nanoparticles (Ni-NPs) on the exposed surface of CHMs. HR-TEM images of such CHM@Ni-NPs show a uniform distribution of distinct nanoparticles with a mean diameter of 5.2 ± 1.1 nm (Fig. 2a and b), which are predominantly located on the surface of the CHM matrix. Furthermore, such CHM@Ni-NPs were granulated using a ball mill to yield granules of size 180–250 μ (Fig. S4), and are therefore referred to as granulated Ni-on-CHMs (NOM-G). Granulated NOM-G is utilized to enhance the accessible metal surface area, thereby promoting interaction between the Ni-NPs and the reactant. This approach is chosen to investigate its effect on reaction facilitation and influence on the overall reaction mechanism.

Fig. 2 Characterization of the NOM-G catalyst: (a) TEM images with the HRTEM image in the inset, (b) particle size distribution, (c) X-ray diffractogram, (d) Ni 2p XPS, and (e and f) NEXAFS analysis of the Ni L-edge and carbon K-edge, respectively.

The second approach entails pre-mixing the Ni-precursor with brown coal, followed by kneading, extrusion, and air drying to obtain monoliths with desired configuration (approximately 3.5 cm in length and 1.5 cm in diameter). Subsequent carbonization results in the formation of Ni-NPs uniformly embedded within and on the surface of the CHM. The monoliths are then washed to remove physiosorbed salts (primarily excess of Ni and acetate ions), resulting in 3 wt% of chemisorbed Ni within the CHM. The innate –COOH groups of the CHM are also important for ion exchange and help evenly disperse Ni²⁺ during the kneading process.²² Importantly, HR-TEM images confirm that the dimensions of the Ni-NPs formed through this approach (6.8 ± 1.5 nm, Fig. 3) are similar to those estimated for NOM-G. Considering the uniformly distributed Ni-NPs on and within the CHM, this sample is termed Ni-in-CHM (NIM). The temperature of the carbonization is varied from 600 °C to 900 °C (N₂, 100 sccm, 1 h, see SI – Experimental details) to understand the thermal effect on the nucleation and reduction of the chemisorbed Ni²⁺ ions. The four samples thus prepared, are termed NIM-600, NIM-700, NIM-800, and NIM-900, with the numbers denoting the carbonization temperature (in °C). Importantly, the weight fractions of Ni (≃3 wt%, Table S6) in all these four variants are maintained similar to each other and to NOM-G.

Fig. 3 Characterization of the NIM catalyst: (a and b) TEM images with the HRTEM image in the inset for NIM-600, (c and d) TEM images with the HRTEM image in the inset for NIM-900, (e) X-ray diffractogram, (f) NEXAFS analysis of the Ni L-edge, and (g) Ni 2p XPS of NIM-600 and NIM-900 catalysts.

These two design principles result in two distinct catalysts (NOM and NIM) that exhibit near identical chemical and thermal characteristics, while providing a stark difference in the diffusional accessibility of active sites to the reactants (0.1 M styrene solution in isopropanol) and products (ethylbenzene: semi-hydrogenation product, ethyl cyclohexane – complete hydrogenation product) during the course of catalysis (refer to Fig. 1). Thus, the active content, dimensions, and uniformity in distribution of Ni-NPs in both the NIM and NOM class of catalysts are identical, with a wide variation in their accessibility and spatial localization. This set of spatially diverse and yet chemically identical catalysts provided a model system for deeper understanding of the various critical parameters of heterogeneous catalysis such as the role of the support, the support–catalyst interaction, and the accessibility of catalytic sites.

3.2. Characterization of Ni-monolith catalysts

NOM-G exhibits uniform distribution of Ni-NPs on the surface of CHM, with crystalline and spherical Ni nanoparticles deposited on the CHM surface. The average d-spacing as estimated from the HR-TEM and pXRD is 0.2 ± 0.1 nm, corresponding to the Ni (111) planes. Furthermore, the pXRD indexing of peaks at 45°, 52°, 76° and 93° corresponds to the (111), (200), (220), and (311) facets of Ni (refer to Fig. 2c),³⁵ confirming the phase purity of the as-prepared NOM-G catalysts. The crystallite sizes are estimated from the Debye–Scherrer equation to be in the range of 7–9 nm, which is in good agreement with the HR-TEM patterns (refer to Table S6 and Fig. 2). The chemical environment of the Ni 2p region from the NOM-G, as evidenced from XPS (Fig. 2d), exhibits a distinct peak for Ni(0) at 852.8 eV, confirming the presence of surface Ni(0) in NOM-G.^23,36 Additionally, the spectra show two characteristic peaks at 854.2 eV and 856.2 eV corresponding to 2p_3/2 of NiO and Ni(OH)₂, respectively. Furthermore, surface oxygen vacancies at the nickel active sites, are indicated by the presence of a satellite feature at 861.4 eV for Ni²⁺ 2p_3/2. The ratio of Ni(0) to Ni²⁺ is found to be 0.07 for the NOM-G catalyst (Table S7). More detailed characterization of the oxidation states of Ni is carried out through NEXAFS Ni L-edge spectra showing two distinct features corresponding to the L₃-edge at around 853 eV (peaks ‘a’ and ‘b’) and the L₂-edge at around 870 eV, resulting from the Russell–Saunders spin–orbit coupling (Fig. 2e).³⁷ The Ni L₃-edge peak reflects the Ni 2p → Ni 3d transition and splits into peaks ‘a’ and ‘b’, due to the crystal field effect sensitivity towards sub-optimally oxidized Ni.³⁸ Furthermore, the C K-edge shows a distinct peak at 288.4 eV corresponding to the π–π* transition of C=O functionality (Fig. 2f).³⁹ In summary, the surface spectra of NOM-G indicate the predominance of Ni(0) and the favourable anchoring of Ni-NPs to the surface of CHMs.

A bottom-up approach is adopted for synthesizing the NIM catalysts, wherein the metal precursor (Ni²⁺ in the form of Ni(Ac)₂) is blended with the VBC, followed by kneading and extrusion. This ensures a uniform chemical composition of the admixture, leading to the formation of the NIM class of catalysts. The diffraction peaks of this admixture are essentially featureless, indicating the absence of any crystalline Ni particles or nanoparticles (Fig. 3e). Importantly, sharp, and distinct diffraction peaks, at 45°, 52°, 76° and 93° due to (111), (200), (220), and (311) planes, emerge from the same samples after carbonization. Thus, the reduction and growth of NPs from the Ni²⁺ precursors is driven by the thermal energy of the carbonization step. Accordingly, the NIM samples are subjected to varying carbonization temperatures and denoted as NIM-600, NIM-700, NIM-800, and NIM-900, with the number signifying the temperature of carbonization. A systematic sharpening of these diffraction peaks is observed with increase in carbonization temperature (Fig. S9), providing confirmation of the thermally-driven Ostwald ripening, leading to the formation of Ni-NPs. This is further supported by the monotonic dependence of the crystallite size on the carbonization temperature. Higher temperature drives greater sintering and growth of larger crystallites, as seen from the average crystallite size of 5–6 nm and 12–14 nm for NIM-600 and NIM-900, respectively (Fig. 3a–d and Table S6).⁴⁰ These overall dimensions of the Ni NPs, also increase with carbonization temperature, while retaining the uniformity of particle dispersion.

XPS analysis of the carbonized CHMs reveals a distinct Ni(0) peak at 853.2 eV, confirming the presence of metallic nickel on the catalyst surface. Additional peaks corresponding to NiO and Ni(OH)₂ are also detected, similar to those observed in NOM-G. Notably, the Ni(0)/Ni²⁺ ratio is higher for NIM-600 (0.3) than for NIM-900 (0.1), indicating a greater surface concentration of metallic Ni in the NIM-600 catalyst (Table S8).

Additionally, NEXAFS spectra at the C K-edge and Ni L-edge were acquired for the NIM catalyst variants. A prominent peak (‘a’) at 288.4 eV, assigned to the C 1s π → π* transition of C=O in carbonyl species, is observed (Fig. S10). The intensity of this feature increases with carbonization temperature, suggesting an elevated concentration of surface oxygen functionalities at higher temperatures. This observation is in agreement with the XPS analysis, which similarly indicates increased surface oxygen content. The NEXAFS Ni L-edge spectra (Fig. 3f) of the NIM catalysts show the distinct features of the L₃-edge at around 853 eV (peaks ‘a’ and ‘b’) and the L₂-edge at 870 eV (peaks ‘c’ and ‘d’), resulting from the core-hole spin–orbital coupling split.³⁷ However, no significant difference is observed in the Ni L-edge of the different NIM catalysts, showing they have similar chemical environments for Ni species.

Heterogeneous catalysis has conventionally looked at increasing the accessible surface area to accelerate the kinetics of the reaction. Among several aspects such as crystallinity, surface energy of exposed facets and the role of support, the information drawn from gas-adsorption isotherms has been the major focus to qualitatively assess the accessibility of the active sites. Such an analysis reveals close similarity in the specific surface area, pore size and pore volume between the NOM (790 m² g⁻¹, 4.5 nm, 0.03 cm³ g⁻¹) and NIM class (800 m² g⁻¹, 6.0 nm, 0.13 cm³ g⁻¹) of catalysts (Table S6). An interesting trend emerging from this study is the systematic increase in CO₂-accessible surface area with carbonization temperature, for the NIM-class of catalysts (Table S6). This can be attributed to the increased availability of pores within the carbon matrix with temperature. The thermal carbonization is also observed to increase the size of the Ni-NPS from 6–8 nm in the case of NIM-600 to 12–14 nm for NIM-900, while preserving the uniformity in dispersion (Fig. 3). Furthermore, the mechanical strength also increases with carbonization temperature, allowing the NIM-900 to be machined into two distinct forms, namely,

(1) As a cylindrical monolith block (NIM-900B).

(2) As small granules (NIM-900G).

3.3. Heterogeneous catalytic activity of Ni-monolith catalysts

The catalytic activities of NOM and NIM catalysts are evaluated for the catalytic potential, with the mono-hydrogenation of styrene serving as a model reaction. Product selectivity, yield and the key process parameters are evaluated based on ethylbenzene (EB) formation, with H₂ being available in large excess (pH₂: 5–30 bar). Importantly, the catalytic forms of both NIM and NOM enable the entire catalytic study to be done in a flow-through geometry (Fig. S8), with stringent controls implemented to eliminate the possibility of surface-induced catalysis originating from the reactor setup.

Considering the paucity of studies done in flow-through geometry, a detailed parametric assessment is required for deeper understanding of the catalysis. Accordingly, it is observed that lower flowrate (0.3 ml min⁻¹) results in greater conversion of styrene to ethylbenzene (≃99%) owing to the increase in residence time (Fig. 4a). Importantly, the selectivity of ethylbenzene remains consistently high (≃99%) across the entire range of flowrates tested (0.3–3 ml min⁻¹), indicating a well-balanced interplay between the kinetics and selectivity of the NOM-G catalyst. Similarly, higher pressures also promote the kinetics of the reaction, while retaining the selectivity towards formation of ethylbenzene (Fig. 4b). The overall reaction order, estimated from these studies, is 0.4 (Fig. 4c). In the context of the Horiuti–Polanyi mechanism, this implies that the semi-hydrogenation step proceeds at a slower rate and is the rate determining step, compared to the complete hydrogenation to EB.^41–44 Finally, the catalysis is carried out at different temperatures (100–150 °C), to arrive at the Arrhenius plot and thereby estimate the apparent activation energy over the NOM-G catalyst to be 9.07 kJ mol⁻¹ g_met⁻¹, which is comparable to that reported for a Ni–Ir catalyst (Fig. 4d and e).⁴¹ On a comparative basis, the NIM class presents two variants of NIM-900G and NIM-900B that differ exclusively in the packing structure of the catalytic bed in the flow-through geometry. This difference is realized by utilizing (a) as prepared Ni metal incorporated monolith, termed NIM-900B and (b) Ni-loaded catalyst that is subsequently granulated to form particles with the average particle size of 180–250 μ (details in the SI), termed NIM-900G. We note that the total weight of the catalyst, the active Ni-loading, and the dimensions of packing (length ≃ 3.5 cm, diameter ≃ 1.5 cm) are kept constant for both NIM-900G and NIM-900B, to afford a direct comparison of their activities. Thus, NIM-900B and NIM-900G represent catalysts in identical chemical environments but carrying significant differences in the tortuosity of channels in the packed, flow-through reactor bed.

Fig. 4 Catalytic activity of the NOM-G catalyst: (a) effect of flowrate at constant pressure and temperature (T = 100 °C, P = 7 bar), (b and c) effect of pressure at constant temperature and flowrate (T = 100 °C, v = 3 ml min⁻¹) (bar diagram represents conversion/selectivity (%) (left axis) and the straight line represents the rate of reaction (right axis)), (d) effect of temperature on catalytic performance at constant pressure and flowrate, and (e) Arrhenius plot (reaction conditions: P = 10 bar, v = 3 ml min⁻¹) (bar diagram represents conversion/selectivity (%) (left axis) and the straight line represents the rate of reaction (right axis)).

Under optimized reaction conditions (180 °C and 30 bar H₂ pressure), the conversion of styrene to EB is always found to be higher for NIM-900G compared to NIM-900B, irrespective of the flow rate of the reactant mixture (Fig. 5a). Moreover, this conversion decreased with increase in flow rate for both NIM-900G and NIM-900B, albeit at different rates. While NIM-900G exhibits similar conversion for both 0.3 ml min⁻¹ and 1 ml min⁻¹ (≃96%), NIM-900B exhibits significant lowering of conversion from ≃53% at a flow rate of 0.3 ml min⁻¹ to ≃12% for a flow rate of 1 ml min⁻¹. This observation brings out the importance of enhancing the tortuosity of the flow-channels that is effectively realized in NIM-900G, compared to straight and relatively unobstructed flow channels for NIM-900B. Quantitatively, we correlated this behaviour to the Reynolds number (R_e), estimated for non-circular geometry as:^45,46

where ρ represents the density (kg m⁻³), v is the flow speed (m s⁻¹), D is hydraulic diameter (0.8 mm) and μ is dynamic viscosity (kg m⁻¹ s⁻¹). Accordingly, the R_e for NIM-900B with 1% styrene in IPA is estimated to be ≃6, indicating a laminar flow of styrene through the oriented channels of NIM-900B.⁴⁷ Such laminar flow is expected to set up interfacial domains between the side-wall of the channel and the flowing styrene, where the flow velocity is significantly lower than that of the bulk. Besides causing mass-transfer limitations of the reacted product, such a parabolic velocity profile of fully developed, steady, incompressible laminar flow (Hagen–Poiseuille flow) also impedes the diffusional access of the reactants to the surface of the NIM-900B catalysts, accounting for their poor conversion kinetics even at low flow rates. Converting such a laminar flow with a low Reynolds number into turbulent behaviour, without modifying the geometry, would require an extremely high flow rate (≃5 L min⁻¹), which is beyond the capabilities of the current setup and therefore could not be tested. In contrast, the NIM-900G presents a highly disordered flow-channel with a significantly complex flow path that increases the turbulence in the flow of both the reactant (styrene–isopropanol mixture) and product (EB), thereby minimizing the diffusional restrictions. Such turbulent flow regimes enhance the interfacial contact between styrene and NIM-900G, leading to high conversion and selectivity achieved over a range of flow rates (0.3–1.0 ml min⁻¹). Finally, the selectivity remains consistently high for both NIM-900G and NIM-900B over the entire range of flow rates tested, confirming the thermodynamically driven nature of mono-hydrogenation of styrene to EB.

Fig. 5 Catalytic activity of NIM variants: (a) effect of flow rate on performance of NIM-900G and NIM-900B catalysts at P = 30 bar, T = 180 °C (bar diagram –% conversion, scattered points – % EB selectivity), (b and c) probable flow patterns through the bed of NIM-900G and NIM-900B catalysts, respectively, effect of pressure on the activity of (d) NIM-900G, (e) NIM-900B catalysts at constant temperature and flow rate (T = 150 °C, v = 0.3 ml min⁻¹) (bar diagram represents conversion/selectivity (%) (left axis) and the straight line represents the rate of reaction (right axis)), and (f) order of reaction from the slope of the graph.

As with the NOM-G catalysts, the presence of H₂ enhances both the conversion and reaction rate for NIM-900G and NIM-900B. Interestingly, the conversion and EB selectivity are consistently high and less dependent on pH₂ for NIM-900G. In comparison, the conversion reaches a maximum of ≃46% for NIM-900B at pH₂ of 30 bar, which is significantly lower than the conversion observed with NIM-G at the same pressure (≃95%). The near-independency of selectivity on pH₂ for both NIM-900G and NIM-900B, along with pressure-independent conversion observed for NIM-900G alone, provides vital clues for the underlying reaction mechanism. NIM-900G with multiple exposed catalytic sites and longer residence times of styrene presents a thermodynamic control over the catalysis leading to stable EB formation, without any further hydrogenation. In contrast, the NIM-900B providing a laminar flow passage leads to relatively low mass-transfer from the active catalyst–substrate interface, leading to longer dwell-times and lower pressure-dependent conversion.

To further elucidate the mechanistic differences between the two geometries of catalysts, we compare the reaction kinetics through a log–log plot of reaction rate and pH₂. A noticeable difference between NIM-900G and NIM-900B emerges in terms of the reaction order (Fig. 5f), with NIM-900G exhibiting a reaction order of <1, whereas the NIM-900B catalyst displays a reaction order of ≃1. Accordingly, it implies that the initial hydrogenation of adsorbed olefins on the metal surface is the slowest step for NIM-900 G, while the second hydrogenation of the semi-hydrogenated olefin species on active metals is the slowest or rate-determining step for NIM-900B (Fig. 7a).^47–49 Additionally, the apparent activation energy over the NIM-900B catalyst (E_a = 14.02 kJ mol⁻¹ g_met⁻¹) is found to be higher than that of the NIM-900G (E_a = 5.92 kJ mol⁻¹ g_met⁻¹) catalyst, confirming that the NIM-900G catalyst is more reactive towards the styrene hydrogenation reaction (Fig. S13). While the block structure exhibits lower catalytic efficiency compared to granular catalysts in this lab scale evaluation, in industrial catalysis, the cylindrical monolith block structure may offer several advantages. Among these, a significant benefit is the reduction of pressure-drop across the industrial scale catalyst bed, which might make it a preferred option. Therefore, this study sheds light on the performance of NIM catalysts (both NIM-900G and NIM-900B), highlighting their structure-specific characteristics, thus offering flexibility in choosing the appropriate catalyst based on specific applications.

Furthermore, the stability of the NOM-G and NIM variants of catalysts is evaluated through long-duration reactions, and the activity of the spent catalysts is analysed via detailed characterization. For the NOM-G catalyst, catalytic activity remains stable after approximately 22 h of reaction, showing negligible reduction in styrene conversion and ethylbenzene selectivity (Fig. 6a). TEM images of the spent catalyst reveal uniformly distributed Ni-NPs (fresh catalyst: 5–7 nm; spent catalyst: 9–12 nm) on the carbon matrix with minimal agglomeration or sintering over time, indicating long-term stability (Table S9). However, XPS analysis of the spent catalyst shows a slight decrease in the surface concentration of metallic Ni (Ni⁰ – fresh catalyst: 3.5%; spent catalyst: 2.7%), suggesting a minor reduction in available active site concentration during or after extended use. An increase in the surface concentration of Ni(OH)₂ is observed, indicating surface oxidation of the catalyst either during the extended run or as a consequence of oxidation after removal from the reactor and storage (Table S9). Similarly, the activity of the NIM variants remains stable after approximately 23 h of reaction, with negligible changes in styrene conversion and ethylbenzene selectivity, as shown in Fig. 6e and i. TEM images of the spent NIM-900G and NIM-900B catalysts (Fig. 6f and j) display uniform distribution of Ni nanoparticles with minimal aggregation or sintering. The average particle size of Ni nanoparticles in spent NIM-900G is around 15 nm, closely matching that of the fresh catalyst (12–13 nm). However, XPS analysis again shows a slight reduction in the concentration of metallic Ni, which is more pronounced in the NIM-900G (fresh: 4.8%, spent: 3.9%) catalyst compared to NIM-900B (fresh: 4.8%, spent: 4.1%) (Table S9). This may be due to the fact that Ni-NPs in NIM-900B are embedded within the carbon matrix, reducing their tendency to be oxidized. Thus, based on its higher catalytic activity, lower activation energy, and long-term stability, the NIM-900G catalyst is presented as a favourable option for hydrogenation applications under the tested conditions.

Fig. 6 Study of the long-run reaction for NOM-G, NIM-900G and NIM-900B catalysts, respectively: (a, e and i) catalytic activity*, (b, f and j) TEM images, (c, g and k) size distribution and (d, h and l) XPS analysis pre- and post-reaction (*reaction conditions – NOM-G: P = 1 bar, T = 30 °C, v = 0.3 ml min⁻¹, NIM-G and NIM-B: P = 10 bar, T = 150 °C, v = 0.3 ml min⁻¹).

To draw an overall comparison between the performance of Ni-based monolith catalysts (Table S10), specifically focusing on two main categories: Ni deposited on the surface of CHM (NOM) and within the bulk of CHM (NIM), we selected the NOM-G and NIM-900G catalysts, as both are utilized in granular form with identical weight percentages (≃3 wt%). Upon comparing their catalytic activity, a significant difference emerged. The NOM-G catalyst exhibits ≃99% styrene conversion and ≃99% EB selectivity at relatively lower pressure and temperature (T = 100 °C, P = 7 bar), with a flow rate of 0.3 ml min⁻¹. Even with a higher flow rate of 3 ml min⁻¹, it attains superior catalytic activity (at T = 150 °C, P = 15 bar), confirming the suitability of the catalyst for high throughput. Conversely, the NIM-900G catalyst requires relatively higher pressure and temperature (T = 180 °C, P = 30 bar) to achieve comparable activity and could only do so with a lower flow rate of 0.3 ml min⁻¹, necessitating more stringent reaction conditions at higher flow rates.

Additionally, H₂-pulse chemisorption measurements reveal the number of active sites available for catalysis represented by their corresponding dispersion numbers.⁴¹ The results show that the NIM-900G catalyst has a lower apparent metal dispersion compared to the NOM-G catalyst (NOM-G = 3.5%, NIM-900G = 1.38%). This may be attributed to the fact that in the NIM-900G catalyst, Ni-NPs with a comparatively larger crystallite size (13–15 nm) are distributed throughout the bulk of the carbon matrix, resulting in lower metal dispersion. Consequently, higher pressure is required to overcome the diffusional barrier, leading to mass transfer limitations. This explains why the NIM-900G catalyst requires critical reaction conditions to achieve the same activity level as the NOM-G catalyst, where active sites are readily accessible during catalysis. Furthermore, Fig. 7b and c provide an overall comparison of the NOM-G and NIM-G catalysts with state-of-the-art catalysts, demonstrating that NOM-G achieves the highest turnover frequency (TOF) (≃28 000 h⁻¹) compared to other catalysts reported in the literature.^41,50–52

Fig. 7 (a) Mechanistic differences between NOM-G, NIM-900G and NIM-900B catalysts, and (b and c) comparative analysis of NOM-G and NIM-G catalysts with state-of-the-art catalysts.

4 Conclusions

This study presents the fundamental design principles for chemical and morphological tailoring of heterogeneous catalysts, which are leveraged to unravel critical mechanistic insights of an olefin hydrogenation reaction. In doing so, we demonstrate that the size-engineered nickel nanoparticles anchored on granulated carbon monolithic supports are viable and sustainable catalysts for selective conversion of styrene to ethyl benzene. Furthermore, the versatility of the NOM-G enables direct conversion of this conventional batch-mode reaction to a flow-through configuration, while achieving high conversion (≃99%), STY (231 g L⁻¹ h⁻¹) and selectivity (≃99.9%) over continuously extended timescales (≃22 h). Such high-performance metrics overcome fundamental limitations of the flow-through process by adoption of a rationale materials-based approach that configures the catalyst–reactant interface for enhanced kinetics without compromising on the stability. Therefore, this study would lead to new directions that couple catalyst discovery with materials design, for driving transition towards flow-based process strategies.

Author contributions

Gitika Rani Saha: conceptualisation, investigation, methodology, data curation, writing – original draft. Marc Marshall: formal analysis, writing – review & editing. Yi Fei: methodology (flow reactor). Fatima Shahid: investigation, data curation (electron microscopy). Mamun Mollah: methodology (gas adsorption). Daryl J. H. Lee: methodology (XRD, chemisorption), formal analysis, visualisation. Lars Thomsen: methodology (NEXAFS), writing – review & editing. Chandramouli Subramaniam: conceptualization, resources, supervision, writing – review & editing. Alan L. Chaffee: conceptualization, resources, supervision, writing – review & editing.

Conflicts of interest

The authors declare that they have no financial or personal conflicts of interest that may have influenced the research reported in this paper.

Data availability

The data supporting this article have been included as part of the SI. Supplementary information is available. See DOI: https://doi.org/10.1039/d5ta04645a.

Acknowledgements

The authors gratefully acknowledge the financial support received from the IITB-Monash Research Academy, India, and BASF Chemicals India Private Limited for this research. We extend our sincere thanks to the Sophisticated Analytical Instrumentation Facility (SAIF-IIT Bombay), the Central Surface Analytical Facility (ESCA Lab), the Industrial Research & Consultancy Centre (IRCC), and the Central Facility of the Department of Chemistry, IIT Bombay, for access to their instrumentation. At Monash University, the authors acknowledge use of facilities within the Monash X-ray Platform and Monash Centre of Electron Microscopy. The authors also wish to thank Dr Boujemaa Moubaraki for assistance with GC-MS analyses. Part of this research was undertaken on the Soft X-ray beamline at the Australian Synchrotron, part of the ANSTO. Chandramouli Subramaniam thanks the Department of Science and Technology, Government of India, for generous support through grants Swarna Jayanti Fellowship, Government of India (DST/SB/SJF/2021-22/07-G) (dated January 23, 2020), and the HSBC-IITB Green Hydrogen program through grant DO/2023-HSBC001-002.

References

1. N. Mizuno and M. Misono, Chem. Rev., 1998, 98, 199–217.
2. T. W. Deelen, C. H. Mejía and K. P. Jong, Nat. Catal., 2019, 2, 955–970.
3. L. Liu and A. Corma, Chem. Rev., 2018, 118, 4981–5079.
4. R. Noyori, Nat. Chem., 2009, 1, 5–6.
5. H. Toulhoat and P. Raybaud, J. Catal., 2003, 216, 63–72.
6. A. J. Medford, A. Vojvodic, J. S. Hummelshøj, J. Voss, F. Abild-pedersen, F. Studt, T. Bligaard, A. Nilsson and J. K. Nørskov, J. Catal., 2015, 328, 36–42.
7. G. L. Foutch and A. H. Johannes, in Encyclopedia of Physical Science and Technology, 2003, pp. 23–43.
8. S. Liu, in Bioprocess Engineering, 2017, pp. 139–178.
9. M. Elhassan, R. Abdullah, M. R. R. Kooh and Y. F. Chou Chau, Bioresour. Technol. Rep., 2023, 21, 101314.
10. M. V. Bukhtiyarova, A. L. Nuzhdin and G. A. Bukhtiyarova, Int. J. Mol. Sci., 2023, 24, 1–23.
11. C. Wiles and P. Watts, Green Chem., 2012, 14, 38–54.
12. H. Jung, C. Kim, H. W. Yoo, J. You, J. S. Kim, A. Jamal, I. Gereige, J. W. Ager and H. T. Jung, Energy Environ. Sci., 2023, 16, 2869–2878.
13. X. Jing, Z. Li, W. Geng, H. Lv, Y. Chi and C. Hu, J. Mater. Chem. A, 2021, 9, 8480–8488.
14. B. G. Sumpter, K. Hong, R. K. Vasudevan, I. Ivanov and R. Advincula, Carbon Trends, 2023, 10, 100234.
15. M. Ibrahim, M. M. Wei, E. Deydier, E. Manoury, R. Poli, P. Lecante and K. Philippot, Dalton Trans., 2019, 48, 6777–6786.
16. M. Turáková, T. Salmi, K. Eränen, J. Wärnå, D. Y. Murzin and M. Králik, Appl. Catal., A, 2015, 499, 66–76.
17. M. C. M. Pérez, C. S. M. De Lecea and A. L. Solano, Appl. Catal., A, 1997, 151, 461–475.
18. T. Yu, J. Jiao, P. Song, W. Nie, C. Yi, Q. Zhang and P. Li, ChemSusChem, 2020, 13, 2876–2893.
19. K. Chai, R. Shen, T. Qi, J. Chen, W. Su and A. Su, Processes, 2023, 11(1074), 1–14.
20. Royal Society of Chemistry, Org. Biomol. Chem., 2015, 13, 7119–7130.
21. M. R. Parsa and A. L. Chaffee, Carbon monoliths and a process for producing same, WO Pat., WO 2018/213890 Al, 2018, pp. 1–30.
22. M. Parsa, Y. Qi, J. J. Di Nuzzo, Y. Moussakhani, A. Tirto and A. L. Chaffee, Chem. Eng. J., 2023, 471, 1–13.
23. A. P. Grosvenor, M. C. Biesinger, R. S. C. Smart and N. S. McIntyre, Surf. Sci., 2006, 600, 1771–1779.
24. E. S. Davydova, M. Manikandan, D. R. Dekel and S. Sunde, ACS Appl. Energy Mater., 2021, 4, 3404–3423.
25. L. Zhu, S. Shan, V. Petkov, W. Hu, A. Kroner, J. Zheng, C. Yu, N. Zhang, Y. Li, R. Luque, C. J. Zhong, H. Ye, Z. Yang and B. H. Chen, J. Mater. Chem. A, 2017, 5, 7869–7875.
26. P. Dubey, N. Kaurav, R. S. Devan, G. S. Okram and Y. K. Kuo, RSC Adv., 2018, 8, 5882–5890.
27. A. M. Hengne, A. K. Samal, L. R. Enakonda, M. Harb, L. E. Gevers, D. H. Anjum, M. N. Hedhili, Y. Saih, K. W. Huang and J. M. Basset, ACS Omega, 2018, 3, 3688–3701.
28. L. Liu, X. B. Luo, L. Ding and S. L. Luo, in Nanomaterials for the Removal of Pollutants and Resource Reutilization, Elsevier Inc., 2019, pp. 83–147.
29. M. M. Baig, I. H. Gul, S. M. Baig and F. Shahzad, J. Energy Storage, 2021, 44, 103370.
30. S. Kusuma, D. B. Bawiskar, C. Singh, P. Panneerselvam, P. Sinha, A. K. Samal and A. H. Jadhav, RSC Adv., 2023, 13, 32110–32125.
31. B. C. Cowie, A. Tadich and L. Thomsen, AIP Conf. Proc., 2010, 307, 307–310.
32. E. Gann, C. R. McNeill, A. Tadich, B. C. C. Cowie and L. Thomsen, J. Synchrotron Radiat., 2016, 23, 374–380.
33. B. Watts, L. Thomsen and P. C. Dastoor, J. Electron Spectrosc. Relat. Phenom., 2006, 151, 105–120.
34. T. Witoon, N. Kachaban, W. Donphai, P. Kidkhunthod, K. Faungnawakij, M. Chareonpanich and J. Limtrakul, Energy Convers. Manag., 2016, 118, 21–31.
35. X. Kou and J. Li, Bull. Mater. Sci., 2008, 31, 97–100.
36. L. Fan, P. F. Liu, X. Yan, L. Gu, Z. Z. Yang, H. G. Yang, S. Qiu and X. Yao, Nat. Commun., 2016, 7, 1–7.
37. D. Drevon, M. Görlin, P. Chernev, L. Xi, H. Dau and K. M. Lange, Sci. Rep., 2019, 9, 1–11.
38. H. Wang, D. S. Patil, W. Gu, L. Jacquamet, S. Friedrich, T. Funk and S. P. Cramer, J. Electron Spectrosc. Relat. Phenom., 2001, 855–863.
39. A. Dey, S. Krishnamurthy, J. Bowen, D. Nordlund, M. Meyyappan and R. P. Gandhiraman, ACS Nano, 2018, 12, 5473–5481.
40. S. Das, A. Jangam, Y. Du, K. Hidajat and S. Kawi, Chem. Commun., 2019, 55, 6074–6077.
41. J. Q. Bai, M. Tamura, A. Nakayama, Y. Nakagawa and K. Tomishige, ACS Catal., 2021, 11, 3293–3309.
42. C. Hu, D. Creaser, H. Grönbeck, H. Ojagh and M. Skoglundh, Catal. Sci. Technol., 2015, 5, 1716–1730.
43. A. C. Pauls, E. W. Comings and J. M. Smith, AIChE J., 1959, 5, 453–457.
44. C. J. Heard, C. Hu, M. Skoglundh, D. Creaser and H. Grönbeck, ACS Catal., 2016, 6, 3277–3286.
45. E. S. Menon, in Transmission Pipeline Calculations and Simulations Manual, 2015, pp. 149–234.
46. B. Rehm, D. Consultant, A. Haghshenas, A. S. Paknejad and J. Schubert, in Managed Pressure Drilling, Gulf Publishing Company, 2009, pp. 39–80.
47. J. Y. Lee, T. Y. Yung and L. K. Liu, Nanoscale Res. Lett., 2013, 8, 1–6.
48. J. Q. Bai, M. Tamura, A. Nakayama, Y. Nakagawa and K. Tomishige, ACS Catal., 2021, 11, 3293–3309.
49. J. A. Anderson, L. Daza, S. Damyanova, J. L. G. Fierro and M. T. Rodrigo, Appl. Catal., A, 1994, 113, 75–88.
50. Y. Ma, Z. Su, N. Tang, S. Chen, W. Wang and J. Yuan, Chem. Phys. Lett., 2021, 775, 138604.
51. Y. F. Jiang, C. Z. Yuan, X. Xie, X. Zhou, N. Jiang, X. Wang, M. Imran and A. W. Xu, ACS Appl. Mater. Interfaces, 2017, 9, 9756–9762.
52. F. Corvaisier, Y. Schuurman, A. Fecant, C. Thomazeau, P. Raybaud, H. Toulhoat and D. Farrusseng, J. Catal., 2013, 307, 352–361.

---