Deposition of poly(furfuryl alcohol) in mesoporous silica template controlled by solvent polarity: A cornerstone of facile and versatile synthesis of high-quality CMK-type carbon replicas. Nanocasting of SBA-15, SBA-16, and KIT-6 *

Six CMK-n replicas ( n ¼ 3, 5, 6, 7, 8, and 9) were synthesized using a new, facile and versatile method based on a surface-selective deposition of poly(furfuryl alcohol) onto the silica matrix. The concept of the synthesis relies on the in situ polycondensation of the carbon precursor in a toluene suspension of the hard template. Preferential pre-adsorption of the monomer onto silica provides even coverage of its surface with polymer, which is a prerequisite to the formation of highly ordered carbon negative structures (either hollow-or rod-type ). The procedure is operated under ambient pressure and at moderate temperature, and a preliminary modi ﬁ cation of silica and vacuo carbonization are no longer needed. A systematic study on materials (SEM, TEM, XRD, N 2 , and Ar adsorption) evidenced that the proposed pathway offers the preparation of carbon mesostructures featuring excellent structural and textural parameters while maintaining the morphology of the mother silica. CO 2 adsorption revealed inhomogeneous porosity of the carbonaceous material itself. The rod-type replicas showed a core-shell structure, in which each carbon rod was built of the outer nonporous shell enveloping the microporous core . In contrast, the hollow-type carbons were constituted of non-microporous turbostratic carbon. © 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
The history of carbon replicas has begun at the turn of the 20th and 21st centuries, when a group of researchers from the Korea Advanced Institute of Science and Technology published the first successful synthesis of a negative carbon mesostructure cast from ordered mesoporous silica [1].The carbon material termed CMK-1 was an inverse structure of MCM-48 silica synthesized by its socalled nanoreplication.The idea of the synthesis, somewhat similar to mold casting, was pretty simple.It involved filling the pores of the silica matrix (mold) with a carbon precursor (therein sucrose), followed by carbonization of the composite and leaching of the silica template to obtain a carbonaceous cast.The curious features of the synthesized carbon inspired the researchers to attempt nanoreplication of other silicas.A year later, the same group reported on another mesostructured carbon, i.e.CMK-3, which was templated using SBA-15 silica [2].In this way, the authors became the fathers of a new class of porous materials, that quickly gained widespread interest from the scientific community worldwide.
Since the publication of these pioneering works, the synthesis routes of nine types of carbon replicas have been reported (CMK-n, n ¼ 1e9).These materials were the negative structural analogues of various siliceous templates (i.e.SBA, M41S, KIT, and FDU families), which featured a long-range ordered framework [1e15].The resulting replicas exhibit a combination of the surface chemistry typical of activated carbons and the pore system array, as well as textural characteristics inherited from the silica' matrices.Such a propitious combination of properties created an excellent opportunity for the utilization of carbon replicas for a variety of practical purposes.Heretofore, numerous applications of carbon replicas were investigated, ranging from adsorption and catalysis [16e26] through medicine and drug development [27e32] up to hydrogen storage [33e36], as well as biosensor construction and electrochemical engineering [37e44].Furthermore, ordered mesoporous carbons pose ideal model materials for the study of adsorption and mass transfer within their highly-ordered porous structures [34,45e49], as well as arrays convenient for the theoretical considerations of the small-angle X-ray scattering and/or diffraction on amorphous, but periodically ordered solids [45,47,50,51].Besides, carbon nanoreplicas may serve as structural matrices for synthesis of ordered mesoporous metal oxides (secondary replication) [52,53], as well as an elegant tool for non-direct investigation of the parent silica structures [2,13e15,54e56].
Nevertheless, the fundamental inconvenience impeding the application of carbon replicas on a technical scale is a laborious, multistage synthesis pathway, that includes time-and energyconsuming operations, most of which require the use of expensive chemicals and an advanced laboratory setup.
The original nanoreplication route to the synthesis of ordered carbon mesostructures, which is the most common so far, is called the hard templating pathway and involves the following steps: (i) preparation of the silica matrix (hard template), (ii) deposition of the carbon precursor inside the template channel system, (iii) carbonization, and (iv) removal of the silica matrix [11,16,17,56e59].Among these stages, the deposition of the carbon source and carbonization are the crucial ones governing the quality of expected structure of the final replica.Before the incorporation of the carbon precursor, the silica matrix is typically modified by the introduction of a suitable polymerization catalyst (e.g.Al 3þ , p-toluenesulfonic or oxalic acid) [4,11,13,16,20,40e42,45,47].Subsequently, the organic material is introduced either by the incipient wetness technique (often repeated impregnation is required), or by chemical vapor deposition (CVD) (of e.g.acetylene, ethylene, propylene, styrene, or naphthalene) [4,9,11,16].The silica-organic composite is then subjected to carbonization.In the case of some carbon precursors (e.g.sucrose), pre-polymerization in a multistep temperature regime is necessary prior to carbonization [1,2,5,11,49].Finally, the silica template is removed from the carbonized composite by treating it with alkali or hydrofluoric acid.Such a sophisticated route is costineffective and does not guarantee obtaining high-quality, reproducible carbon structures.
Depending on whether the carbon precursor infiltrates pores of a matrix completely (volume-templating) or clads an inner surface with a thin layer (surface-templating), two types of carbon replica structure can be distinguished, i.e. rod-, and hollow-type, respectively.Both CMK-1 and CMK-3 replicas originally reported by Ryoo et al. belong to the first type [1,2].
CMK-3 is the best known and the most often studied carbon replica.More recently, several papers have also been published on the synthesis and interesting applications of CMK-5 (the hollowtype analogue of SBA-15) [40e42] and CMK-1 [19,31,34].Nonetheless, other carbon replicas, especially these of the hollow-type, have been reported scarcely, except for their early synthesis methods [5,11].Apparently, this is due to the particularly tedious synthesis routes necessary to obtain such subtle structures.Indeed, besides the aforementioned multistage preparative path, most authors have reported additionally carrying out the carbonization in vacuo as the factor that enables the formation of highly ordered hollow-type replicas [1,5,8,9,11e14,17,27,60].
In the present work, we discuss a new, facile and versatile method of synthesis of carbon replicas based on a surface-selective deposition of poly(furfuryl alcohol) onto a silica matrix.The developed procedure relies on a simple one-pot incorporation of carbon precursor in the silica matrix suspended in a nonpolar liquid medium [61].The deposition step is carried out under ambient pressure at a moderate temperature; thus, it does not require employing an advanced synthetic apparatus.Furthermore, a preliminary modification of the silica matrix and carbonization in vacuo are no longer needed.The intentional carbon structures can be readily obtained by changing just two synthesis parameters, i.e. choosing a suitable silica matrix and adjusting the amount of monomer used.
The feasibility and versatility of the proposed procedure were proven by the successful synthesis of six CMK-n replicas derived from three silica matrices.Namely, three hollow-type carbons (n ¼ 5, 7, and 9) and their rod-type counterparts (that is, n ¼ 3, 6, and 8), were cast from SBA-15, SBA-16, and KIT-6, respectively (see Scheme 1).
A thorough study on morphology, structural and textural characteristics of the resulting carbons evidenced that the proposed synthesis pathway provides obtaining carbon mesostructures with excellent long-range ordering and textural parameters while maintaining the morphology of the template.Such perfectly ordered carbon mesostructures (particularly these hollow ones) have been rarely or not at all reported hitherto.Moreover, the in-depth investigation on the internal porosity of the carbon material itself disclosed the core-shell structure of the rod-type mesostructures, in which the outer nonporous shell envelops the microporous core.In contrast, the hollow carbons are entirely constituted of nonmicroporous turbostratic carbon.To the best of our knowledge, herein we report on this for the first time.

SBA-15, SBA-16, and KIT-6
Mesoporous SBA-15, SBA-16 and KIT-6 matrices were synthesized under the acidic conditions, according to the relevant procedures reported elsewhere [14,56,62].Details of the synthesis are provided in the Supplementary Information section.The calcined SBA-15, SBA-16, and KIT-6 silicas were labeled as S15, S16, and K6, respectively.Additionally, small portions of the as-made materials were annealed under a static air atmosphere using the same thermal programme as for the carbonization of poly(furfuryl alcohol)/ silica composites (850 C for 4 h with a heating rate of 1 C min À1 ).These references were marked as S15@850, S16@850, and K6@850.

Carbon replicas
Six carbon replicas were cast from the silica matrices by the surface-selective acid-catalyzed precipitation polycondensation of furfuryl alcohol in a toluene suspension of the hard templates.More specifically, both hollow-and rod-type carbon analogues were obtained from each silica, as shown in Table 1.
The deposition of poly(furfuryl alcohol) (PFA) into the silica matrix pore system was carried out in a 100 cm 3 round-bottom Scheme 1. Pictorial representation of the pore architectures of the studied silica matrices (top), and their hollow-(middle) and rod-type analogues (bottom).

Table 1
Intended loading of silica matrices with carbon precursor (expressed as a mass ratio) corresponding to the given carbon replica structures and their space groups.flask immersed in an oil bath, placed on a magnetic stirrer and equipped with a reflux condenser.An amount of 1.50 g of freshly calcined silica was added at room temperature (21 C) under vigorous stirring to a mixture containing toluene (reaction medium), anhydrous sodium sulphate (desiccant), furfuryl alcohol (carbon source), and tartaric acid (polyreaction catalyst).Noteworthy, the use of fresh (or freshly distilled) monomer devoid of the FA oligomers and freshly calcined silica are essential to the successful synthesis.The intended PFA/SiO 2 mass ratios (cf.Table 1) were adjusted by the amount of monomer used.These ratios were individually optimized beforehand towards obtaining the desired replica structures.We selected these ratios, which guaranteed the highest possible XRD intensity while maintaining the mutual proportions of the intensity of individual reflections, as well as the highest textural parameters (specific surface area and total pore volume).The cumulative mass of toluene and FA was kept constant at 50.00 g.The syntheses were carried out at constant molar ratios of TA/FA ¼ 0.50, and Na 2 SO 4 /FA ¼ 0.15.After the silica matrix was put into the reaction system, the mixture was stirred (600 rpm) at room temperature for 30 min, and then the heating was turned on.Since the temperature reached 100 C, the suspension was isothermally held for the next 24 h under vigorous stirring (800 rpm).The resulting brown PFA/SiO 2 composite was then isolated using a fluted filter, washed with a small portion of toluene (approx.20 cm 3 ), and dried at 105 C overnight.Finally, both TA and Na 2 SO 4 were removed from the composite by washing with 200 cm 3 of hot distilled water (~70 C) and dried again at 105 C for 24 h.This prevents the high-temperature oxidation of carbonaceous material during carbonization (i.e. its reaction with Na 2 SO 4 [61]).The final composites were marked as PFAeYex, where Y refers to the silica matrix and x stands for the real PFA/SiO 2 mass ratio (determined from thermogravimetric measurements).The composites were carbonized in a tubular furnace under an argon atmosphere (40 cm 3 min À1 ) at 850 C for 4 h with a heating ramp of 1 C min À1 .This temperature regime was found to be optimal for obtaining PFA-derived carbon frameworks with rigidity sufficient to be stable upon silica removal [16e19, 56,57,59,61].Eventually, the silica matrix was removed by double etching with 5 wt% HF at room temperature for 90 min (1.00 g of a carbonizate was immersed in 30.0 cm 3 of HF and the suspension was shaken ever and again).The carbonized composites were marked as CeYex, while the relative CMK-n replicas were denoted according to the generally adopted nomenclature (Table 1).

Methods
Textural parameters of the studied materials were determined by means of low-temperature adsorption-desorption of nitrogen and argon; both at À196 C. The isotherms were recorded using an ASAP 2020 sorptometer (Micromeritics) operated in a conventional volumetric mode.Prior to the measurements, the samples were outgassed at 250 C for 6 h under vacuum.The calculation of the textural parameters was carried out following our previous works with respect to suitable non-porous reference materials [17,56,63e68].The CO 2 adsorption isotherms were acquired for the carbon replicas using the same apparatus operated at 0 C. Beforehand, the samples were evacuated under identical conditions as for other adsorptives.The detailed calculation of the textural parameters can be found in the Supplementary Information section [67,69,70].
Structural parameters were examined by means of low-angle Xray powder diffraction (XRD) using a D2 Phaser instrument (Bruker) equipped with a LYNXEYE detector.XRD patterns were collected with Cu Ka radiation (l ¼ 1.54184 Å) in the angular range of 2q ¼ 0.80e3.15with a step of 0.02 .
Transmission electron microscopy (TEM) images were taken on an FEI Tecnai TF20 X-TWIN (FEG) microscope operated at an accelerating voltage of 200 kV.Prior to the analyses, the samples were dispersed in isopropanol and sonicated for 10 min, and then deposited onto carbon-coated copper grids by the drop-casting technique.
Scanning electron microscope (SEM) images were acquired using a HITACHI S-4700 field emission scanning electron microscope (FESEM) working at an accelerating voltage of 20 kV.Samples were mounted on carbon adhesive discs and coated with gold.
High-resolution thermogravimetric (TG) measurements were performed using an SDT Q600 analyzer (TA Instruments).An amount of ca.0.01 g of a sample placed in a corundum crucible was heated from 30 to 1000 C with a heating rate of 20 C min À1 under an air atmosphere (100 cm 3 min À1 ).The true amounts of the carbon source deposited onto the silica matrices (i.e. the real PFA/silica mass ratios in the as-synthesized composites) were assessed on the basis of the mass loss related to the burning-off of the organic component.Then, the silica's pore filling degree was calculated as a ratio of the volume of PFA deposited (the density of bulky polymer d PFA ¼ 1.55 g cm À3 [17,56,57,61]) with respect to the total pore volume of the relevant silica matrix.Thermo-oxidative measurements for the ultimate carbon replicas were performed using the same apparatus and settings.Additionally, based on the latter analyses, the content of silica residues in the replicas was calculated.
Elemental analysis (EA) measurements of the carbon replicas were performed using an EA3100 (EuroVector) CHNS-O analyzer.Each analysis was repeated three times.

Effectiveness of carbon source deposition
The efficiency of loading the carbon source into the silica matrix mesochannels was investigated by means of thermogravimetric analysis under the oxidative atmosphere.The real contents of the deposited polymer, expressed additionally with regard to the textural parameters of the silicas (cf.Table 2), are depicted in Fig. 1.
It is reasonable to presume that the real PFA/SiO 2 ratio should be proportional to the amount of monomer available in the reaction medium.In fact, the higher the intended PFA/SiO 2 ratio, the higher the real loading can be found.Nevertheless, the deposition effectiveness calculated as the quotient of the genuine and intended polymer/silica mass ratio shows the opposite trend.This is most likely due to the facilitated deposition of PFA directly onto the silica surface, which is favored by the pre-adsorption of monomer molecules thereon.The detailed discussion of this issue is provided in our recent work [61].The incorporation of further amounts of polymer proceeds without this effect; additionally, the gradual filling of the pores entails a hindered monomer transport due to diffusion limitations.It should be noted that the highest deposition effectiveness was observed for the S15 matrix with relatively broad, double-sided open mesopores (cf.Scheme 1 and Fig. 1).The lower effectiveness of the S16 decoration can be attributed to the impeded transfer of monomer through the narrow constrictions interconnecting the spherical cages.In the case of the K6 template, the likely factor that limits monomer diffusion is the complex geometry of its pore network (a similar effect was reported for PFA deposition in MCM-48 silica with the same pore architecture, although narrower pores [72]).The filling of silica templates with polymer with respect to the total pore volume thereof spans the range of 53e64%, and 68e99% for the low-and high-loaded composites, respectively.The latter suggesting the incomplete pore filling may be somewhat puzzling taking into account the rod-type structure of the ultimate replicas derived from these composites.This becomes understandable given the high-temperature shrinkage of porous silicas during composite carbonization (the decrease in V t attains 26, 46, and 42% of the original values for S15, S16, and K6, respectively; see Table 2).The surface content of PFA ranges from 0.8 to 2.0 mg of polymer per square meter (S BET ) of silica, which corresponds to the even covering the surface with the PFA layer of a roughly nanometer-scale thickness.

Silica matrices
The textural and structural parameters of the studied materials were investigated by low-temperature adsorption-desorption of nitrogen and low-angle X-ray diffraction.The recorded isotherms and pore size distributions (PSDs) are shown in Fig. 2, while the XRD patterns are presented in Fig. 3.The isotherms for pristine S15, S16, and K6 silicas exhibit essentially similar behavior of the adsorption branches with distinct capillary condensation steps, characteristic of type IV(a) according to the IUPAC classification [67].The S15 and K6 materials (Fig. 2A, C) reveal textbook examples of the H1 hysteresis loop with almost parallel adsorption and desorption branches ascribed to materials featuring mesopores highly uniform in diameter.Consequently, the NLDFT curves for both silicas (cf.Fig. 2A 0 , C 0 ) exhibit sharp pore size distributions with the maxima at 7.6 and 8.5 nm, respectively, accompanied by smaller fractions of narrow mesochannels of the main size centered at ca. 2.6 nm (interconnecting pores).The corresponding XRD patterns (Fig. 3A, C) reveal the distinct Bragg's reflections typical of the respective long-range ordering of the mesoporous frameworks.This confirms the successful synthesis of the desired structures, i.e. 2D hexagonally arranged parallel mesopores in the case of S15 (p6mm), and interwoven bicontinuous mesochannels cubic framework in K6 (Ia3d) (Scheme 1).These findings are also supported by TEM imaging (see Morphology and phase purity of carbon replicas section).
The S16 silica exhibits a model H2(a) hysteresis loop with a very steep desorption branch and a closure point at p=p 0 ¼ 0.43 (Fig. 2B).Such a nature of the isotherm is indicative of the cage-like pore structure, in which adjoining cages are interconnected by narrower apertures (below 5 nm).The corresponding PSD (Fig. 2B 0 ) shows a broad peak centered at ca. 2.6 nm, which may be roughly assigned to the window diameter (Scheme 1).However, it should be kept in mind that the cavitation-controlled evaporation of the adsorptive (for N 2 it is below the critical relative pressure of ca.0.43) precludes quantitative estimation of the actual pore size [67].Therefore, the reliability of this value is limited.
b Minimal distance between the surface of adjacent spherical cages (i.e.length of the cage-to-cage interconnecting window) calculated as: c Dw ¼ a 0 3:0919 Fig. 1.Effectiveness of PFA deposition in the S15, S16, and K6 pore systems.The areas marked in cyan and yellow pertain to the real PFA/SiO 2 mass ratios required for the synthesis of hollow-and rod-type carbon replicas, respectively.the typical ranges reported elsewhere for such materials.Interestingly enough, the pore wall thickness of S15 and the minimal distance between the surfaces of adjacent cages in S16 (i.e. the linear size of the cage-to-cage interconnecting neck), as well as the share of microporosity, are slightly lower compared to the typical values reported previously [2,4,5,8e12,16,17,55e57,61].In fact, this is not surprising in view of the applied synthesis conditions (i.e. higher temperature and prolonged aging time as compared to the typical procedure).The XRD pattern collected for S16 (Fig. 3) exhibits a typical shape reported for such arrays [10,12,62].Combining nitrogen adsorption and XRD data for this silica template, one may infer that the structure consisting of a body-centered cubic array of mesopore cages (Im3m symmetry) interconnected with eight neighbouring cavities through narrower apertures has been successfully synthesized (Scheme 1).This long-range ordering can be clearly seen in the TEM images (Fig. 8).

CeYex carbonizates
The nitrogen isotherms recorded for the carbonized PFA/silica composites are depicted together with the corresponding PSDs in Fig. 2. It can be seen at a glance that the deposition of a carbon source brought substantial changes in the textural parameters of the resulting materials.For the CeS15 composite with partially filled pores, the isotherm type remained unaltered (type IV(a)), although the shape of the hysteresis loop changed to H2(a) [67], and the nitrogen uptake drastically dropped compared to the bare silica (Fig. 2A).Additionally, the loop shifted towards lower relative pressures.Consequently, the specific surface area decreased by ca.68%, while the total pore volume dropped by roughly 74% compared to the pure silica.The maximum of pore size distribution shows a shift to 4.1 nm (Fig. 2A').This pores narrowing arises from the two coinciding effects, namely the formation of a homogeneous PFA film inside the matrix mesochannels, and the thermal shrinkage of the silica framework itself during carbonization (cf.Fig. 2A, sample S15@850).The latter finds reflection in a decrease in the lattice parameters of the carbonizates (Table 2).Expectedly, the carbonizate with fully filled pores reveals a further drop in the textural parameters (S BET decreases by 93%, whereas V t lowers by 95%).This is seen in the isotherm, which shows the nitrogen uptake close to nil, and featureless PSD (Fig. 2A and A', respectively).
The isotherms of the CeK6 composites exhibit a similar behavior except for the microporous region (Fig. 2C).Namely, the steep nitrogen uptake accompanied by sharp knees observed at p= p 0 < 0.02 for both materials points to the formation of a certain share of micropores.Therefore, the micropore volume increases by 300e350% with regard to the silica matrix (cf.Table 2), giving rise to the development of the micropore surface that reaches 112e153 m 2 g À1 .This in turn causes slighter drops in the S BET after polymer deposition (decreased by 53 and 64% for partially and completely filled matrix, respectively).The appearance of microporosity may arise from the detachment of carbon from the silica pore walls during carbonization.This should not be surprising given the relatively broad pores of K6 (8.5 nm; Table 2) and the discrepancies in the scale of high-temperature shrinkage of both materials [16,17,19,57,61,72].A slightly different scenario is observed for the CeS16 carbonizates (Fig. 2B).In this case, both the specific surface area and the total pore volumes decreased substantially (S BET by 97 and 98%, respectively; V t for both carbonizates is close to nil; Table 2), notwithstanding the pore filling degree.The hysteresis loop for the partially filled composite reveals an alteration into a H4 type (Fig. 2B).Nevertheless, the negligible nitrogen uptake for both samples results in the featureless shape of the respective PSDs (Fig. 2B').The low-angle XRD patterns acquired for all carbonizates evidence that the long-range ordering of materials throughout PFA deposition and carbonization is preserved (Fig. 3).The positions of the characteristic reflections are in line with those of the silica references calcined at 850 C. The lower intensities thereof are stem from the shielding effect of the carbon layer onto the silica surface against the X-rays.

Carbon replicas
The low-temperature nitrogen adsorption isotherms and corresponding PSDs of the ultimate replicas are shown in Fig. 2, and the respective low-angle XRD patterns are presented in Fig. 3.The relevant textural and structural parameters are gathered in Table 3.
Essentially, the behavior of the nitrogen isotherms for the synthesized replicas corresponds to the types of isotherms for the respective matrices (Fig. 2A, B, C).Namely, all the isotherms recorded for carbons are of a IV(a) type, characteristic of mesoporous materials [67].The isotherms of replicas cast from S15 and K6 (Fig. 2A, C) are accompanied by H1 hysteresis loops, typical for materials with mesopores uniform in a size.Both hollow-type replicas (i.e.CMK-5, and CMK-9) feature prominent two inflections in the adsorption branches at the relative pressures of p= p 0 ¼ 0.30e0.50,and 0.55e0.70.These abrupt adsorptive uptakes correspond to the delayed capillary condensation in two independent mesopore systems, i.e. the intertubular pores, originating from the leached silica walls, and the intratubular ones, corresponding to voids unfilled with the carbon precursor (see Scheme 1).Combining the PSDs of these replicas with the respective carbonizates (Fig. 2A', C'; Table 3), one can infer that in both cases the narrower sizes (2.9 and 3.0 nm, for CMK-5 and CMK-9, respectively) correspond to the intertubular channels, while the wider ones (4.1 and 4.4 nm, respectively) are the intratubular pores inherited from the carbonizate.It is worth noting that the adsorption and desorption branches of the hysteresis loops at p=p 0 ¼ 0.55e0.70 are almost parallel (especially for CMK-9).This is indicative of excellent uniformity of the diameter of the intratubular pores.Naturally, such a widely developed porosity influenced the textural parameters; S BET of 2208 and 2113 m 2 g À1 , while V t of 2.12 and 2.08 cm 3 g À1 were calculated for CMK-5 and CMK-9, respectively (Table 3).The carbon wall thickness of the hollow-type replicas (Table 3) of approximately 1.2e1.4nm is consistent with the previously reported critical minimum wall thickness (i.e.1.0e1.8nm), which allows maintaining a stable and rigid structure of the inverse replica after removal of the silica matrix [1,3,6,9,11,45,50].To the best of our knowledge, the synthesis of hollow-type replicas of so high-quality has not been reported heretofore.
The complete pore filling of the S15 and K6 matrices with PFA led to the formation of rod-type CMK-3 and CMK-8 replicas, respectively (Scheme 1).As a consequence, the step of capillary condensation in the intratubular pores has vanished (cf.Fig. 2A, C), and the materials feature a purely monomodal porosity (cf.Fig. 2A',  C').Extinction of intratubular porosity brings a roughly 50% drop in both apparent surface area and total pore volume as compared to the respective hollow-type counterparts (see Table 3).These values are consistent with those published elsewhere [2,4,5,7,11,13e19,25,33e35,51,54e58].
The isotherms of the replicas cast from S16 (i.e.CMK-6, and CMK-7) show the H2(a) hysteresis loop (Fig. 2B) typical for materials with cage-like pores interconnected by narrow apertures (Scheme 1).Regardless of the mutual similarity of the carbonizates from which the CMK-6 and CMK-7 materials were derived (Table 2), the difference between the isotherms for the final replicas is evident (Fig. 2B).This is likely caused by the relaxation of the carbon framework upon silica leaching.Apart from substantially higher adsorption uptake throughout the isotherm, CMK-7 carbon shows a small inflection in the adsorption branch at p= p 0 ¼ 0.50e0.65.This is related to the condensation of adsorbate in the voids that emerged in place of the cages of the silica matrix covered with the PFA film.This is better seen in Fig. 2B', which shows a bimodal porosity with average pore size of 3.0 and 4.2 nm.Keeping in mind the lower total pore volume of S16 compared to other templates, a relatively small share of the intratubular pore volume in this case is understandable (Table 2).Nevertheless, the CMK-7 replica shows a highly developed apparent surface area of 2141 m 2 g À1 and a total pore volume of 1.72 cm 3 g À1 , which are coherent with the parameters reported elsewhere [11,12].Expectedly, complete pore filling of the S16 matrix with carbon precursor led to the formation of CMK-6 carbon with monomodal porosity centered at 3.2 nm.Extinction of intra-cage porosity caused a 36% and 32% drop in the S BET and V t , respectively [10e12].
Based on the suitable experimental data collected for the broader series of the synthesized materials (which are not reported here), we defined the critical ranges of PFA loading required to obtain given replica structures (see Fig. 1, the cyan-and yellowcoloured regions).Namely, in the case of insufficient loading of SBA-15 with PFA, the final carbon structure collapses, resulting in the formation of an amorphous material.The moderate loadings yield the hollow-type replicas, while for the rod-type materials we have indicated the minimal PFA loading threshold (note: by crossing this threshold, the desired rod-type structures are obtained all the time, but when a large excess of monomer is used, there is a threat of deterioration of the quality of the final material by the formation of disordered amorphous carbon agglomerates).
The low-angle X-ray patterns (Fig. 3) confirm the excellent fidelity of the rod-type carbon structures with respect to the corresponding matrices.According to the Babinet's principle, CMK-3 and CMK-8 show identical angular positions and the same relative intensities of characteristic reflections as for the respective silicas calcined at 850 C. With this, one can infer that the original silica symmetries were retained (hexagonal p6mm and cubic Ia3d space groups, respectively).Meanwhile, the CMK-6 material features the XRD pattern similar to the S16 matrix (cubic Im3m phase), however, in this case, the relative intensities of the corresponding maxima were not preserved.This effect, also observed by others, has not been fully explained yet [45,47,50,51].It can be hypothesized that the ratio of these intensities depends on the electron density of the carbon material, which may differ on the outer surface and in its bulk.This is in turn influenced by the internal porosity of the carbon material itself.Nonetheless, the position of the reflections also confirms that the S16 matrix architecture was maintained in the CMK-6 material [10e12].
The positions of the XRD reflections recorded for the hollow-type replicas are roughly coherent with those for their rod-type analogues, but their relative intensities change substantially (Fig. 3).This stems from the discrepancies in the electron density between rod-and hollow-type replicas and the interferences of X-rays (either constructive or destructive) that may occur on the carbon walls of these latter ones.It is well documented that X-ray patterns can differ strongly even for the same hollow-type replicas depending on their wall thicknesses [45,50,51].With this, we may conjecture that the respective structural arrays were preserved also in the case of the subtle arrays of CMK-5, CMK-7, and CMK-9.

In-depth study of carbon replicas porosity
The adsorption of argon at À196 C provides a deeper insight into the nature of the pore structures of the carbon replicas [12,55,67,68].Because of its lower desorption closure point than nitrogen at the same temperature, argon as the adsorptive probe offers a lower minimum limit of the analyzeable diameters of constrictions/plugs, which can be particularly informative.The minimal limit of desorption for argon equals p=p 0 z 0.27e0.38,while for nitrogen reaches 0.40e0.50.These ranges correspond to the evaporation of adsorptive from pores 3.4e4.0nm and 4.5e5.0nm in diameter, respectively [55].The argon adsorptiondesorption isotherms of the studied carbon replicas are shown in Fig. 4 (N 2 adsorption isotherms are added to facilitate direct comparison).
When considering the shape of the argon isotherms for the CMK-5 and CMK-9 carbons, the conspicuous difference can be seen, although the respective nitrogen isotherms look basically identical.Namely, in the case of the CMK-5 material, the argon isotherm hysteresis loop may be classified in between H2(a)/H2(b) type characteristic of pore systems with constrictions (but relatively broad) [67].Interestingly, unlike the nitrogen isotherm, the argon one does not show two deflections in the desorption branch.The single steepness of the desorption branch is indicative of the impeded adsorbate evacuation from certain regions of the pore structure.Likely, this occurs in the interstitial space, i.e. this one located in between the adjacent carbon nanopipes (Scheme 1).Assuming the size of these slits to be approximately 1.5 nm (which corresponds to the thickness of the walls of the parent silica matrix; Table 2), the foregoing reasoning appears to be sound.
Another scenario is observed for CMK-9 replica.In this case, the character of the hysteresis loops of both argon and nitrogen isotherms remains the same (i.e.H1 type) and features two pronounced steps in the desorption branch.This is not totally unexpected considering the structure of the CMK-9 replica, which does not possess these slit constrictions as in the case of CMK-5 (viz.Scheme 1).An analogous conjuncture can be observed for the respective rod-type carbon replicas, i.e.CMK-3 and CMK-8.The former one shows a small tail in the desorption branch of the argon isotherm, which allows us to classify this hysteresis loop as the H5 type, while the CMK-8 carbon is characterized by the H1 type hysteresis loop.
A profound difference in the nitrogen and argon isotherms can be seen for the CMK-7 replica (viz.Fig. 4).The argon isotherm shows the H5 type with a slightly tailing desorption branch, which probably corresponds to the evaporation of the adsorbate through the narrow necks interconnecting the carbon spherical cages (Scheme 1).Expectedly, this tail disappears for CMK-6 and its argon isotherm exhibits the H2(a) hysteresis.
The basic textural parameters (i.e.specific surface areas and total pore volumes) of carbon replicas calculated from the argon adsorption isotherms are gathered in Table S2.All these values, although slightly lower, are roughly in line with those calculated from nitrogen adsorption (cf.Table 3).To summarize this part, it can be stated that the isotherms of argon adsorption at À196 C can be successfully used to achieve complementary information to nitrogen.
The microporous structure of carbon replicas was investigated in more detail by means of carbon dioxide adsorption at 0 C. The collected isotherms together with the respective PSD curves are presented in Fig. 5.
The differences between the isotherms for hollow-and rod-type replicas can be seen at first glance.Namely, the latter show higher CO 2 adsorption throughout the curves and steep adsorptive uptakes at the lowest relative pressures (i.e. up to about 10 À3 ).This suggests the presence of a certain fraction of ultramicropores in these materials (D p < 0.7 nm [67]).Indeed, the respective PSDs (Fig. 5, insets) disclose the presence of ultramicropores of two sizes, namely around 0.4 and 0.6 nm.Meanwhile, the hollow-type carbon replicas show slightly wider micropores centered at 0.6 and 0.7 nm.This is reflected in the micropore volumes calculated according to the Dubinin-Radushkevich model [67] (see Fig. S3 and Table 3).The micropore volume of the rod-type replicas contributes to the total pore volume about 11e17%, whereas the hollow-type carbons reveal only 1e5% of the micropore volume share.As such, the CMK-5, CMK-7, and CMK-9 replicas can be considered to be practically non-microporous [69e71].
The above discussion provides particularly interesting  information.Namely, it is evident that the porosity of the final carbonaceous material itself in the rod-type replicas is inhomogeneous.Let us consider holistically the following findings: (i) hollow-type carbons and the crossbars merging the whole structure are almost entirely mesoporous; (ii) rod-type replicas show abundant contributions of micropores; (iii) deposition of carbon source in the silica pores proceeds starting from the pore walls, and the polymer chains grow towards the pore center.This means that the rod-type structures can be formally considered as the hollow-type ones filled with an additional portion of carbon precursor [61].
Combining the above, one may conjecture that each rod constituting the rod-type replicas is built of a non-microporous shell, that envelops an appreciably microporous core.The pictorial illustration of the elucidated pore structures of carbon replicas is presented in Fig. 6.
Plausibly, this effect can arise from the following three reasons: (i) disparities in the packing of the carbon precursor directly on the silica-polymer interface and in the bulk of the PFA deposit, which is very likely in view of our previous research [16,17,61], (ii) the pressurizing effect of silica that undergoes shrinkage during carbonization, leading to the compression of the carbonaceous matter and, thus, the extinction of its indigenous microporosity, nearby the silica-polymer interface; (iii) the stabilizing role of silica during heat treatment of PFA (again, on the polymer-silica interface) [75].To the best of our knowledge, such nuance of the internal porosity of carbon replicas has not been reported to date.

Morphology and phase purity of carbon replicas
The morphology and arrangement of the pore system of silica matrices and carbon replicas were investigated by scanning and transmission electron microscopy imaging.The collected SEM and TEM micrographs are shown in Figs.7 and 8, respectively.
The S15 silica exhibits a sausage-like morphology with particles of roughly 2 mm in length and ~1 mm in diameter, highly aggregated along their longitudinal axis [76,77] (Fig. 7; top panel).The S16 matrix shows larger, well-formed particles (~3 mm) with a single crystal rhomb-dodecahedral morphology characteristic for this silica [78e80].Unlike the previous ones, the K6 material is characterized by irregularly shaped particles varying in size, which is consistent with other reports [13,81].As seen in Fig. 7; middle and bottom panels, the carbon replicas inherited the matrix grain morphology, notwithstanding whether the silica pores were incompletely or fully filled with carbon precursor.This means that the studied carbons do not show the presence of any phase impurities as was observed in the case of other synthesis procedures [16].This is interesting in view of synthesis in an aqueous medium, in which the excess carbon precursor forms an amorphous carbon shell that clads the external surface of the matrix grains, as mentioned above [16,17,56,57].The long-range honeycomb 2D ordering of the pore structure of the S15 matrix assigned to the p6mm space group is seen in Fig. 8; left panel.This arrangement was perfectly preserved in the corresponding carbon replicas; the CMK-3 material is an exact inverse structure built of carbon rods which are merged by thinner crossbars, while the CMK-5 replica observed along the [0 0 1] incident shows a hexagonally arrayed bundle of hollow nanopipes, also linked by the spacers (Fig. 8).In addition, both carbons show a perfectly parallel arrangement of the nanorods/nanopipes alongside their longitudinal axes (i.e.along the [1 1 0] direction; (cf.Scheme 1)).The TEM images collected for the S16 silica (Fig. 8; central panel) show a characteristic body-centered cubic array ascribed to the Im3m space group.Again, both daughter carbon materials cast from this matrix retained the pore array of pristine silica.It is noteworthy that, although confusing at first glance, several images of the CMK-6 and CMK-7 replicas resemble to some extent the hexagonal arrangement of S15-derived replicas.This becomes clear in view of the fact that these are the cubic-symmetry body-centered structures observed along the [1 1 1] direction.Despite its irregular morphology on a micrometer scale (cf.Fig. 7; right panel), the K6 silica shows a perfectly ordered bicontinuous cubic porous mesostructure (double gyroid) classified as Ia3d symmetry (Fig. 8 and Scheme 1).Similarly to the others, both the CMK-8 and CMK-9 carbon replicas turned out to be high-fidelity negative structures of their mother silica.The seemingly hexagonal arrangement of the CMK-8 material poses in fact to the image of the Ia3d structure observed alongside the [1 1 0] axis (see Fig. 8).It should be noted that, at first glance, the TEM images of the CMK-6 and CMK-8 carbons look like hollow-type replicas, while the CMK-7 and CMK-9 resemble rod-type materials.However, it is known that care must be taken when analyzing TEM images of porous carbons [82].Besides the impact of mutual orientation of the electron beam direction and a given crystallographic axis, the second effect affecting the image is the probability that the electron beam passes through two (or more) nanosheets (or flat grains) of material superimposed on each other when randomly oriented.This may cause the symmetry observed on the TEM images to seemingly change [83].It is also pertinent to mention that the TEM images of all carbon replicas additionally evidence the lack of superficial carbon islets on the external surface of their grains.
The phase purity of the carbon replicas was investigated by means of high-resolution thermogravimetric measurements under an air atmosphere [3,61].The relevant DTG curves are gathered in Fig. S2.The CMK-5 and CMK-9 mesostructures show narrower DTG profiles than their rod-type relatives.Furthermore, in both cases, the maximum oxidation rates are shifted towards a lower temperature, roughly 13 C.This is understandable considering the openwork structure of the hollow-type replicas that features a facilitated intraparticle diffusion of oxygen during the TG measurements.The narrow DTG curves are indicative of a high phase purity of the materials (i.e.lack of disordered carbonaceous agglomerates).Interestingly, such a well-distinguished difference cannot be seen for the CMK-6 and CMK-7 materials, although the latter one shows a bit narrower DTG profile.All the rod-type carbons show a bit broader DTG due to their more "bulky" character compared to the hollow-type replicas.The hollow-and rod-type carbons also differ in their elemental composition.The average empirical formula of the former can be expressed as C 10.0 H 2.3 O 0.8 , while for the latter it is C 10.0 H 1.3 O 0.5 .These formulas are in good consistency with those presented elsewhere [1].Though, the differences in O and H contents are not surprising in view of our previous reports [16,17].Therein, we found that a prominent fraction of the oxygen-containing surface moieties is formed when the freshly carbonized sample comes into contact with air.Therefore, the higher the apparent surface area (as in the case of hollow-type carbons herein), the larger the amounts of oxygen-containing species that saturate the highly reactive surface of the turbostratic carbon.The aforementioned TG measurements were also used to calculate the efficiency of the removal of silica matrices from the carbonizates.The determined content of the residual silica in the carbon replicas did not exceed 0.5 wt%.

Conclusion
A new, facile and versatile method of synthesis of both hollowand rod-type CMK-type carbon replicas has been proposed.The concept relies on simple, one-pot incorporation of the carbon precursor (poly(furfuryl alcohol)) into the hard template pore structure by in situ precipitation polycondensation of the monomer in the toluene suspension of the silica.The basis of this method is the homogeneous surface-selective deposition of the polymer onto the silica matrix driven by the preliminary preferential preadsorption of a polar monomer attracted by the superficial silanols.Thereby, the formation of perfectly ordered carbon negative structures is feasible.
Employing the developed procedure, six CMK-n replicas (n ¼ 3, 5, 6, 7, 8, and 9) were synthesized by the nanocasting of SBA-15, SBA-16, and KIT-6 silicas.The intentional structures were received by changing just two synthesis parameters, i.e. choosing a suitable silica matrix and adjusting the amount of monomer used.A systematic characterization of the materials has evidenced that the proposed pathway allows the synthesis of carbon replicas that feature excellent structural and textural parameters while maintaining the morphology of the mother silica.In fact, the literature provides only a handful of examples of such perfectly ordered carbon mesostructures (particularly those hollow ones).It was found that the hollow-type replicas have an exclusively mesoporous nature, while their rod-type relatives show mixed micro-mesoporous character.This discloses the core-shell structure of these latter ones, in which each carbon rod is built of the outer nonporous shell that envelops the microporous core.In contrast, hollow-type carbons are built of non-microporous turbostratic carbon.To the best of our knowledge, herein we report on this for the first time.
The proposed procedure is operated under ambient pressure and at moderate temperature.Moreover, the preliminary modification of silica as well as the use of an advanced laboratory setup and carbonization in vacuo are no longer needed.Therefore, it is believed that the synthesis and application of carbon replicas can pass beyond the laboratory scale.In the forthcoming research, we investigate the feasibility of the present concept for nanocasting of silicas with narrower pores (i.e.MCM-48 and SBA-1) leading to the formation of CMK-1, CMK-2, and CMK-4 replicas.

Fig. 4 .
Fig. 4. Argon adsorption-desorption isotherms recorded at À196 C for the carbon replicas.The nitrogen adsorption isotherms added for the sake of comparison.Insets: respective semilogarithmic plots.

Fig. 5 .
Fig. 5. Carbon dioxide adsorption isotherms recorded at 0 C for the carbon replicas.The solid red lines represent the fitted curves calculated based on the 2D-NLDFT model for carbons with heterogeneous surfaces [65,66].Insets: corresponding PSDs.

Fig. 6 .
Fig. 6.Schematic representation of the internal porous structure of the synthesized carbon replicas.

Fig. 7 .
Fig. 7. SEM images of the silica matrices (top panels), and corresponding carbon replicas with hollow-and rod-type structures (middle and bottom panels, respectively).

Fig. 8 .
Fig. 8. TEM images of the silica matrices (top panels), and corresponding carbon replicas with hollow-and rod-type structures (middle and bottom panels, respectively).The Miller indices of the incidences along which the images were taken are provided in the images.Additionally, respective Fourier diffractograms are displayed.

Table 2
Textural and structural parameters of silica matrices and carbonizates calculated from nitrogen adsorption and X-ray diffraction.

Table 3
Textural and structural parameters of carbon replicas calculated from nitrogen adsorption and X-ray diffraction.S me1 ; S me2 e shares of inter-and intratubular specific surface area, respectively, computed from the corresponding mesopore volume contributions.For CMK-5 and CMK-9 Micropore volume calculated from CO 2 isotherms based on the Dubinin-Radushkevich equation (Figs.5 and S3).c V me1 ; V me2 e shares of inter-and intratubular mesopore volume, respectively, calculated by the Gaussian deconvolution of PDSs.V me1 ¼ Vt À V me2 .Average diameter of the carbon rods assumed to be equal to the mesopore diameter of the corresponding silica reference calcined at 850 C(D sil:850 a b d BET means the S BET for the respective silica reference calcined at 850 C; v mat: sil: is the volume of silica walls in 1 g of the respective silica.Based on the density of amorphous silica, v mat: sil:¼ 0:45 cm 3 g À1 .For hollow-type replicas RFI ¼ me2 refer to the replica's intratubular surface and pore volume, respectively.g