Synthesis and catalytic advantage of a hierarchical ordered macroporous KIT-6 silica

Ordered mesoporous silicas are important technological materials in catalysis, sorption and separation science, however new architectures are desired to improve in-pore accessibility. Here we report the ﬁ rst synthesis of an ordered macroporous KIT-6, obtained by optimizing the ratios of Pluronic P123: sodium dodecyl sulfate cosurfactants


Introduction
Ordered mesoporous silicas (OMS), employing surfactants including, block copolymers, as structure-directing agents (SDAs) or templates, find broad applications in adsorption [1], catalysis [2], separation science [3], drug delivery [4] and biomaterials engineering [5]. Academic and industrial interest in OMS materials reflects their desirable textural properties, with BET surface areas reaching 1500 m 2 /g, uniform pores tunable from~1.5 to 40 nm diameter and diverse pore connectivity, and their ease of chemical functionalization through one-pot (co-condensation) or post-synthesis modification [6e8]. Regarding the latter, organo-acid, amine, alkyl, aromatic and thiol groups have been covalently bound to surface silanols [9,10], with alkene and aromatic groups incorporated into the silica framework to form periodic mesoporous organosilicas [11e14].
A significant limitation of OMS is their pore diameter, and for the commonly used MCM-41 and SBA-15 silicas, their twodimensional pore connectivity, both of which hinder access of bulky molecules to the internal surface of the mesopores. OMS pore dimensions and connectivity are largely dictated by the choice of soft template (typically a surfactant which may be neutral, anionic or cationic and with a tunable headgroup [15] and molecular shape), around which a silica sol-gel is condensed [16], and the subsequent solvothermal processing used to transform the sol-gel into a covalently bonded silica network [17]. Soft templating of OMS materials, and the role of swelling agents (poragens) to increase mesopore diameter, has been comprehensively reviewed. Regarding pore connectivity, this is predominantly dictated by the structure of the organic mesophase (surfactant template and additives) formed prior to genesis of the silica sol-gel. Surfactants such as cetyltrimethylammonium bromide and Pluronic 123 (P123) triblock copolymer act as SDAs, enabling the synthesis of hexagonal arrangements of one-dimensional cylindrical silica mesopores possessing thin (Mobil Composition of Matter No. 41, MCM-41) [18] and thick (Santa Barbara Amorphous-15, SBA-15) [19] walls respectively. OMS with a three-dimensional cubic Ia3d mesostructure comprising two interpenetrating continuous networks of chiral pore channels can also be synthesized using cetyltrimethylammonium chloride [20], cetyltrimethylammonium tosylate [18], cetyltrimethylammonium bromide [21] or metal acetates [22] (MCM-48), or with butanol [23,24] or sodium dodecyl sulfate (SDS) [25,26] as cosurfactants with P123 (KIT-6), as SDAs. Chen et al. report the synthesis of KIT-6 using mixtures of nonionic P123 and anionic sodium dodecyl sulfate (SDS) surfactants [25], with an optimal SDS:P123 M ratio between 2.1 and 2.5.
The triblock copolymer P123, (PEO) 20 -(PPO) 70 -(PEO) 20 , comprises hydrophilic polyethylene oxide (PEO) and more hydrophobic polypropylene oxide (PPO) units. At temperatures >15 C the PPO block is insoluble in water, inducing spontaneous self-assembly of P123 into micelles [27,28] with a hydrophobic PPO core of 4.8 nm radius and a 4.6 nm thick hydrophilic PEO corona [2]. At critical volume fractions of 20e40 wt% in water mixture and temperatures <50 C, P123 cylindrical micelles form an isotropic, cubic closepacked mesophase [29], however it has not proven possible to template KIT-6 from this without cosurfactants. Higher critical volume fractions of 40e80 wt% and temperatures of 20e80 C favor an isotropic, hexagonal close-packed mesophase which soft templates SBA-15. Soft templating of OMS may proceed by the addition of a silicic precursor to a pre-formed isotropic mesophase of the appropriate surfactant followed by (typically) acid hydrolysis to form a silica replica, denoted true liquid crystal templating (TLCT), or via cooperative self-assembly of the surfactant and silicic precursor mixture followed by (typically) acid hydrolysis to form a silica replica [30]. The TLCT process requires higher surfactant concentrations, but yields highly ordered OMS due to confined condensation of the silicic precursor around the organic mesophase, sometimes described as nanocasting. The TLCT synthesis of SBA-15 is reported [31], but not to our knowledge the analogous synthesis of KIT-6 for which only cooperative self-assembly is reported in the presence of butanol [23] or SDS [25] cosurfactants. Such cosurfactants preferentially bind to PPO units in P123, thereby swelling the hydrophobic volume, reducing the interfacial curvature of the micelles, and promoting transition of the organic mesophase from hexagonal (which templates SBA-15) to cubic (which templates   [23,25,32]. Functionalized OMS possessing three-dimensional pore networks offer many advantages over their one-dimensional counterparts in catalysis. Pirez et al. demonstrated increased turnover frequencies for C 3 eC 16 fatty acid esterification with methanol over propylsulfonic acid grafted KIT-6 (PrSO 3 H-KIT-6) vs. an analogous grafted SBA-15 (PrSO 3 H-SBA-15) due to improved reactant in-pore diffusion and active site accessibility [33]. Gas phase selective oxidation of diphenylmethane is faster over V-doped KIT-6, and more selective to the desired benzophenone product, than a corresponding V-doped SBA-15, attributed to a higher concentration of V 5þ species on the external walls of the KIT-6 which also suppressed coking [34]. Parlett et al. also observed that KIT-6 increased the dispersion of Ni nanoparticles active for the structure sensitive, high temperature steam reforming of ethanol to H 2 relative to SBA-15 (which was also more prone to mesopore blockage by carbon deposition) [35]. Nickel nanoparticles dispersed throughout KIT-6 exhibit superior on-stream stability for the dry reforming of methane than those dispersed in SBA-15 [36], attributed to greater sintering and coking resistance afforded by the Ia3-d mesopore network. KIT-6 also provides improved reactant accessibility and metal dispersion vs. SBA-15 to in-pore PdO x [37] or PtO x [38] active sites for the aerobic selective oxidation of allylic alcohols to aldehydes. Vapor phase ethanol dehydrogenation to acetaldehyde is faster and more selective over Cu nanoparticles within KIT-6 than SBA-15, in part reflecting a higher metal dispersion and genesis of (Lewis acidic) Cu þ active sites [39]. Palladium nanoparticles prepared by impregnation of a nitrate precursor exhibit a slightly higher dispersion over KIT-6 compared with SBA-15, which Venkateswarlu and co-workers ascribe responsible for the former's enhanced activity in base-and ligand-free Suzuki-Miyaura crosscouplings [40]. The catalytic advantage of three-dimensional vs. one-dimensional OMS is sometimes unclear due to difficulties in preparing frameworks with the common pore diameters [41,42], or attempted benchmarking at high reactant conversion wherein mass transport limitations are problematic [43]. Nevertheless, where fair comparisons are possible, OMS possessing threedimensional pore networks permit faster in-pore molecular diffusion and higher active site dispersions, and are mechanically stronger than their one-dimensional counterparts [44].
As humanity transitions from a fossil carbon dependence to renewable carbon feedstocks (notably biomass), catalytic science faces numerous challenges to producing identical, analogous or new organic molecules and materials to those which underpin modern society. One such challenge is the requirement for new solid catalysts suitable for transforming bulky (and highly functional) bio-derived reactants, often in the liquid phase at mild temperatures. Micro-and mesoporous catalysts such as zeolites, which underpinned last century's petrochemical industry, are illsuited for these transformations due to limited accessibility of inpore active sites. The introduction of macropores (with diameters >50 nm [45]) into smaller pore frameworks can dramatically enhance molecular transport without compromising surface area or active site density [46]. The catalytic advantage of the resulting hierarchical porous catalysts has been demonstrated for SBA-15; macropores increase per site activities for biodiesel production from free fatty acids (FFAs) and triacyl glycerides over PrSO 3 H groups [47], and for the selective oxidation of sesquiterpenoids over PdO nanoparticles [48]. However, to our knowledge the analogous macroporous KIT-6 framework has never been synthesized, despite its potentially superior mass transport, active site dispersion and resistance to coking. Here we report the first synthesis and resulting striking catalytic advantage of an ordered macroporous KIT-6 sulfonic acid silica for the esterification of hexanoic acid with methanol.

Polystyrene nanosphere synthesis
Monodispersed polystyrene nanospheres were prepared according to the literature [49]. Styrene (105 cm 3 , Sigma 99%) was washed five times with aqueous NaOH (0.1 M, 100% v/v) and five times with deionized water (100% v/v) to remove the 4-tertbutylcatechol polymerization inhibitor. Purified styrene was then added to N 2 degassed (50 cm 3 $min -1 for 1 h) deionized water (1000 cm 3 ) at 80 C, to which aqueous K 2 S 2 O 8 (0.024 M, 50 cm 3 , Sigma !99%) was added dropwise over 15 min, and the resulting mixture then stirred at 300 rpm under flowing N 2 for 22 h. The resulting white emulsion was cooled to room temperature and the solid product collected by centrifugation at 8000 rpm for 3 h. The colloidal nanosphere aggregates thus obtained were ground in a mortar and pestle to a fine, free-flowing powder containing monodispersed particles of 420 ± 5 nm before use.

Hierarchical porous silicas
Macroporous-mesoporous silicas were prepared by dual hardsoft templating, employing polystyrene nanospheres as hard macropore templates and P123 as a surfactant mesopore template [50]. 10 g (1.7 mmol) P123 (Sigma, 30 wt%, average Mn~5800) and 1.15 g (4 mmol) sodium dodecyl sulfate (SDS, Sigma, anhydrous !99%) were dissolved in 360 cm 3 deionized water and 20 cm 3 HCl (Sigma, 12 M) in a polypropylene bottle at 30 C under stirring at 400 rpm. To this mixture, 5 g of 420 nm polystyrene nanospheres and 23 cm 3 tetraethyl orthosilicate (TEOS, 98%) were added under vigorous stirring (800 rpm) for 5 min to form a homogeneous liquid. Stirring was maintained at the same temperature for 20 h before transferring to an oven for hydrothermal treatment at 100 C (static) for 3 days. The resulting solid was filtered and air dried at room temperature, prior to calcination at 550 C (ramp rate 1 C.min -1 ) for 5 h in a muffle furnace under air to remove organic templates. The preceding synthesis was also performed for SDS:P123 M ratios spanning 1.8 to 3.5, with materials designated MM-SiO 2 -1.8, MM-SiO 2 -2.3 and MM-SiO 2 -3.5. Note that shorter hydrothermal treatments (1 day) yielded less ordered materials and there was no benefit to longer hydrothermal (6 days) or additional recrystallization treatments ( Fig. S1 and Tables S1eS2).

Acid functionalization
The preceding hierarchical porous silicas were functionalized with propylsulfonic acid by an alkoxide grafting method [49]. 600 mg MM-SiO 2 was dispersed in 30 cm 3 toluene under flowing N 2 at 120 C and 700 rpm stirring for 15 min. 1 cm 3 3mercaptopropyl)trimethoxysilane (MPTMS) was subsequently added, and the reaction mixture stirred at 600 rpm at 120 C for 24 h. The resulting solid was filtered, washed with ethanol to remove residual silane, and dried at 80 C overnight. Surface thiols (-SH) were oxidized to sulfonic acids (-SO 3 H) by treatment of the dried functionalized silica powder with 20 mL of 30 vol% H 2 O 2 with 400 rpm stirring overnight. The final materials were washed with deionized water and ethanol, and dried overnight at 80 C. The overall catalyst synthesis is illustrated in Fig. 1.

Materials characterization
Structural order was evaluated by powder X-ray diffraction (XRD) using a Bruker D8 Advance Diffractometer fitted with a LynxEye highspeed strip detector and Cu K a source (1.54 Å) with a Ni filter and calibrated against a quartz standard. Low angle patterns were measured for 2q ¼ 0.4e8.0 with a step size of 0.02 and a scan speed of 0.014 s. Textural properties were determined by N 2 porosimetry on Quantachrome Nova 2000 and 4000 porosimeters and processed using Novawin v2.2 analysis software. Samples were degassed at 120 C for 2 h before recording N 2 adsorption/ desorption isotherms at À196 C. Specific surface areas were calculated using the Brunauer-Emmett-Teller (BET) method over relative pressures P/P 0 ¼ 0.05e0.2, where a linear relationship was maintained. Microporosity was evaluated using the t-plot method, over a relative pressure of 0.2e0.5 which presented a linear correlation. Mesopore size distributions and mesopore diameters were calculated applying the Barrett-Joyner-Halenda (BJH) model to the desorption isotherm. Transmission electron microscopes (TEM) images were recorded on a JEOL 2100F FEG scanning transmission electron microscope (STEM) operating at 200 keV and equipped with a spherical aberration probe corrector (CEOS GmbH). Samples were prepared by dispersion in methanol and dropped onto a copper grid coated with a holey carbon film. Scanning electron microscopy (SEM) was performed on a FEI Verios 460L XHR microscope at 700 V, equipped with an Oxford XMax30 energy dispersive spectroscopy (EDS) detector. Samples were ultrasonicated in 2-propanol then drop-cast on an aluminum stub for imaging, or carbon tape and coated with 3 nm carbon for EDS. Thermogravimetric analysis (TGA) was performed under flowing N 2 (20 cm 3 .min -1 ) from 40 to 800 C (ramp rate of 10 C.min -1 ) on a Metter Toledo TGA/DSC 2 Star System equipped with a Pfeiffer Vacuum ThermoStar MS GSD 301 T3 mass spectrometer. Approximately 10 mg of sample was treated with 0.5 cm 3 propylamine and left for 2 h. Physisorbed propylamine was subsequently removed by in vacuo drying at 40 C overnight. Quantification of sulfonic acid loadings was performed by temperature programmed decomposition of the remaining chemisorbed propylamine in the TGA; reactively-formed propene and ammonia being liberated at Brønsted acid sites [51]. Ammonia pulse chemisorption was also performed on pre-dried samples using a Quantachrome ChemBET Pulsar chemisorption analyser at 100 C. Magic-angle spinning solid-state nuclear magnetic resonance (MAS-NMR) spectra were recorded on an Agilent DD2 500 MHz spectrometer equipped with a 4 mm MAS solid-state triple-resonance probe. The samples were packed into a 4.0 mm zirconia rotor with 10 kHz sample spin rate. 1

Results and discussion
Propylsulfonic acid functionalized MM-SiO 2 all exhibited type IV adsorption isotherms with H1 hysteresis loops characteristic of mesoporous materials with open cylindrical pores (Fig. 2a). BET surface areas and mesopore volumes decreased significantly with increasing SDS:P123 M ratio (Table 1), reflecting differences in the textural properties of the parent supports and the introduction of surface acidity (which also lowers the porosity of PrSO 3 H-SBA-15 and PrSO 3 H-KIT-6 prepared using a butanol cosurfactant [33]). Low angle XRD patterns evidence the impact of SDS addition on the ordering of silica mesopores (Fig. 2b). At SDS:P123 ratios <2.2, three distinct reflections are observed consistent with the (100), (110) and (200) reflections of SBA-15, wherein two-dimensional, hexagonal mesopore channels close pack to adopt a p6mm symmetry. For 2.1 SDS:P123 ratios 2.8 a completely different set of reflections is apparent consistent with the (211) and (220) reflections of KIT-6, wherein three-dimensional, gyroidal mesopore channels intertwine to adopt a Ia3d symmetry [23]. The unit cell size of 19.4 nm is also consistent with that previously reported for conventional KIT-6 [25]. Higher SDS concentrations appear to disrupt micelle formation and/or self-assembly, resulting in an ill-defined material with a single, broad reflection~1.15 which cannot be indexed to any crystalline phase. Differential analysis of the low angle XRD patterns (Fig. 2c) confirms the formation of SBA-15 and KIT-6 and phases for SDS:P123 ratios of 1.8 and 2.3 respectively. Peaks and troughs in the second derivative of the latter data align perfectly with those for a reference mesoporous KIT-6 prepared with butanol but without SDS, while those for the former are consistent with those for a reference mesoporous SBA-15 with an expanded unit cell (shifting reflections in Fig. 2b and peaks/troughs in Fig. 2c to lower angle). Note that the weaker intensities of higher order reflections for the macroporous KIT-6 and SBA-15 relative to the mesoporous reference silicas is expected due to the smaller mesopore domains (<200 nm) in the hierarchical porous materials.
Mesoporous silicas prepared using co-block polymer templates such as Pluronic P123 typically exhibit microporosity [52,53], however the magnitude strongly depends on the hydrothermal synthesis temperature. At 80 C, approximately half of the total surface area of cooperative self assembled SBA-15 is attributed to complementary microporosity [31] resulting from interactions between hydrophilic PEO chains of micelles sharing their hydration sphere during silica condensation. Microporosity is commonly identified through t-plot analysis of N 2 adsorption isotherms, which assumes that the surface of the porous material is chemically identical to a non-porous analogue (e.g. Aerosil silica) used to determine the reference isotherm. Geometric models which make no assumptions regarding surface properties demonstrate that SBA-15 microporosity approaches zero for hydrothermal temperatures of 130 C [52]. Introduction of a sodium dodecylsulfonate (SDS) co-surfactant during SBA-15 and SBA-16 syntheses also eliminates microporosity [54]. Our use of high temperature (100 C) hydrothermal aging and a SDS co-surfactant to synthesize  (Tables S2 and S4).
Transmission electron microscopy (Fig. 3) of silicas prior to acid functionalization supports these phase assignments and provides additional information on the size and ordering of the templated macropores. A face-centered, cubic close-packed arrangement of macropore voids, templated by polystyrene nanospheres, is observed for SDS:P123 ratios <2.5, with some loss of macropore ordering apparent when additional SDS was added in the synthesis. The mean macropore diameter slightly decreased with SDS concentration from 370 nm (SDS:P123 ¼ 1.8) to 360 nm (SDS:P123 ¼ 3.5). For SDS:P123 ratios <2.2, domains of hexagonally close-packed mesopores are observed between macropores (Fig. 3a), characteristic of hierarchical ordered macroporous SBA-15 [47,48,50]. Ordered mesopore domains are also observed for 2.1 SDS:P123 ratios 2.8, but with pores adopting a square arrangement (Fig. 3b(iii)) consistent with a cubic phase. No ordered mesopores are visible for higher SDS:P123 ratios (Fig. 3c).
Chemical properties of the  (Fig. 5a). The 7.2 ppm peak is assigned to sulfonic acid protons, while those at 4.5 and 1.4 ppm are assigned to protons in methyl and CH 2 groups of the propyl backbone [57]. The  corresponding 29 Si NMR spectrum exhibits peaks at À112.28 ppm, À104.59 ppm and À95.24 ppm (Fig. 5b) assigned to Q 4 , Q 3 and Q 2 siloxane structural units in the silica support [6]. Due to the low propylsulfonic acid loading, peaks associated with SieC bonds (expected at À67 ppm and À58 ppm) could not be observed. Surface chemistry of the PrSO 3 H/MM-SiO 2 -2.3 silica was further studied by DRIFTS to confirm organic functionalization and the absence of residual thiol precursor (Fig. 6a). Strong bands spanning 700e1300 cm and 3000e3800 cm are associated with the silica framework and surface silanols/water respectively; the sharp band at 3740 cm arising from isolated SieOH groups, with the broader features at 3550 cm and 3200 cm arising from H-bonded silanols in vicinal chains and residual physisorbed water respectively [58].
Weaker bands due to n(CH 2 ) and scissor d(CH 2 ) modes of the propyl backbone appear at 2890e2934 cm and 1400 cm respectively. Although S]O stretches for sulfonic acid are expected between 950 and 1150 cm, these are obscured by strong overlapping bands from the silica support, however the absence of a 2700 cm band confirms complete oxidation of the MPTMS precursor (confirmed by sulfur 2p XPS, Fig. S2). TGA of PrSO 3 H/MM-SiO 2 -2.3 revealed the loss of physisorbed water <100 C (Fig. 6b) and subsequent decomposition of hydroxyl (200e400 C) and propylsulfonate groups from~450 to 550 C (consistent with PrSO 3 H/KIT-6 [33], PrSO 3 H/SBA-15 and PrSO 3 H/MM-SBA-15 [47]). Corresponding measurements from a chemisorbed n-propylamine adlayer evidenced a greater low temperature mass loss (150e450 C) due to the desorption of reactively-formed propene and NH 3 (Hoffmann elimination) [51]. The propene peak maximum desorption temperature of 350 C  indicated mid-strength Brønsted acid sites similar to those in WO x / ZrO x functionalized periodic mesoporous organosilicas [59] and H-ZSM-5 [51]. Quantification of the mass losses for propylsulfonic acid functionalized silicas yielded acid site loadings between 0.15 and 0.21 mmol$g -1 , consistent with elemental S loadings (0.57e0.68 wt%) from EDS, and comparable to our previous reports on PrSO 3 H/SBA-15 prepared without SDS [47,49]. The sulfonic acid function is only stable to 450 C, precluding acid site analysis by TPD for stronger bases, however NH 3 pulse chemisorption confirmed an acid loading of 1.5 mmol$g -1 .
The preceding characterization demonstrates that SDS promotes the transition of P123 organic mesophases from hexagonal to cubic close-packed arrangements, even in the presence of PS nanospheres, enabling the templating of (hierarchical) macroporous-mesoporous KIT-6 (MM-KIT-6). If the SDS concentration is too low then the hexagonal close-packed arrangement of P123 micelles remains more stable, and (hierarchical) macroporous-mesoporous SBA-15 is obtained (MM-SBA-15); if the SDS concentration is too high then micelle self-assembly is disrupted and neither macropore nor mesopore ordering occurs.
The optimum PS loading was 5 g, with lower macropore template loadings favoring SBA-15 over KIT-6 and a loss of macropore ordering (Figs. S3eS4 and Tables S3eS4) and higher template loadings diminishing both mesopore and macropore ordering. Attempts to adapt the same synthesis using~250 nm diameter PMMA beads as the macropore template were unsuccessful, with only mesoporous SBA-15 or disordered silica phases formed with relatively poor macropore ordering (Figs. S5eS7 and Tables S5eS8).
Having successfully synthesized a macroporous PrSO 3 H/MM-KIT-6, the question arises whether its three-dimensional network of interconnected mesopores is beneficial for mass transport vs. the one-dimensional network of mesopores in macroporous PrSO 3 H/ MM-SBA-15.
The catalytic performance of these hierarchical porous solid acids was therefore examined for hexanoic and palmitic acid esterification in methanol and benchmarked against conventional (mesoporous only) PrSO 3 H/SBA-15 and PrSO 3 H/KIT-6 counterparts and commercial Amberlyst-15 sulfonic acid resin (Fig. 7). Esterification of such fatty acids is an important step in the pretreatment of low-grade oil feedstocks for biodiesel production. Turnover frequencies (TOFs) based on initial rates normalized per acid site (for conversion <25%) are poor for both acids over Amberlyst-15 (acid loading~1.7 mmol$g -1 ) [33], likely reflecting poor swelling of the resin in methanol and hence limited active site accessibility [60].
Sulfonated SBA-15 is much more active than Amberlyst-15, but significantly outperformed by sulfonated KIT-6, attributed to improved mass-transport in the three-dimensional OMS [33]. Hexanoic acid esterification over Amberlyst-15 and mesoporous sulfonic acid catalysts is approximately twice as fast as that for the larger palmitic acid, attributed to diffusion limitations in combination with polar and steric effects for the long chain fatty acid [61e64]. Macropore incorporation was beneficial, irrespective of the mesopore arrangement, and essentially eliminated differences in esterification rates between hexanoic and palmitic acids. Since all the silica catalysts contain identical sulfonic acid species, rate enhancements for hierarchical catalysts must be a consequence of reduced pore tortuosity and faster diffusion throughout silica particles to in-pore active sites [49]. Nevertheless, the presence of three-dimensional interconnected mesopores was essential to  maximize the macropore enhancement, with sulfonic acid active sites within PrSO 3 H/MM-KIT-6 producing methyl hexanoate and methyl palmitate 39% and 33% faster respectively than those in PrSO 3 H/MM-SBA-15 (which was comparable to the disordered PrSO 3 H/MM-SiO 2 -3.5). This may be readily understood as the three-dimensional, interconnected pore channels present in KIT-6 provide more mesopore entrances and shorter pathways from the bulk media to in-pore active sites. In contrast, diffusion through the ordered, but long and poorly interconnected, pore channels in SBA-15 offers no significant advantage over that through disordered mesopore channels.
Note the three-fold rate enhancement for palmitic acid esterification over macroporous vs. conventional PrSO 3 H/SBA-15 is similar to that previously reported [47], demonstrating that the SDS cosurfactant had minimal impact on the silica surface chemistry or sulfonic acid active site. Palmitic acid esterification is five times faster over PrSO 3 H/MM-KIT-6 than a conventional PrSO 3 H/KIT-6, highlighting the impact of macroporosity on mass transport. The PrSO 3 H/ MM-KIT-6 catalyst maintained excellent stability during the 6 h reaction, evidenced by constant rates of palmitic and hexanoic acid conversion (Fig. S8), consistent with literature reports for propylsulfonic acid silicas under identical reaction conditions [14,61,65].
There are few examples of macroporous sulfonic acids for catalytic esterification that report kinetic turnover frequencies (TOFs). Macroporous sulfonic acid resin beads prepared using paraffin liquid and dibutyl phthalate as poragens exhibit TOFs of only 3 h -1 for rapeseed oil esterification with methanol at 70 C [66]; unfortunately, the macroporosity was not characterized. Sulfonated macroporous carbon spheres, templated by ordered arrays of colloidal silica beads, delivered a slightly improved performance (TOF ¼ 2 h -1 ) for acetic acid esterification with ethanol at 70 C [67]. A hierarchical, interconnected micro-meso-macroporous solid acid constructed from zeolite nanocrystals only achieved a TOF of 7 h -1 for palmitic acid esterification with methanol, despite forcing conditions of 150 C [54]. Our PrSO 3 H/MM-KIT-6 (TOF of~100 h -1 for palmitic acid esterification at 60 C) therefore significantly outperforms literature macroporous catalysts.

Conclusions
Here we adapt a bottom-up (dual hard-soft templating) route, using polystyrene nanosphere templates in conjunction with P123 and SDS cosurfactants, to introduce ordered macropores into a mesoporous KIT-6 silica (MM-KIT-6). The resulting hierarchical porous silica possesses a framework of three-dimensional interconnected mesopore channels with cubic Ia3d symmetry interspersed by an ordered array of 370 nm macropores. Macropore ordering and formation of the Ia3d mesopore structure were sensitive to the SDS:P123 M ratio (2.1 optimum 2.8) and amount of macropore template (optimum mass ratio of PS:TEOS:P123:SDS of 4.3 : 18.7: 8.7 : 1). Sulfonic acid functionalization yields a solid acid catalyst (PrSO 3 H/MM-KIT-6) with significantly enhanced per site activity for the esterification of (bulky) fatty acids to biodiesel. The combination of macropores and three-dimensional interconnected mesopores accelerates in-pore molecular diffusion and reactant accessibility to active sites, delivering a five-fold enhancement in turnover frequency for palmitic acid esterification in methanol over PrSO 3 H/MM-KIT-6 vs. a conventional mesoporous PrSO 3 H/KIT-6. This study highlights the impact of pore architecture on the catalytic performance of heterogeneous catalysts for liquid phase reactions. Future work will explore molecular simulations and diffusion NMR to further optimize pore connectivity in hierarchical porous silicas.

Declaration of competing interest
None.

Data availability
Data will be made available on request.