1. Introduction
The sol-gel process is an important synthesis method for the development of nanostructured materials with different applications [
1,
2,
3]. The great advantage of the sol-gel synthesis is the ability to fit the properties and nanostructure of materials by properly combining the synthesis conditions and the interactions between the different components. Carbon gels (such as, aero-, xero- or cryogels) can be obtained in different formats and with different micro/nanostructure from a wide number of monomers, although they are commonly prepared from resorcinol-formaldehyde (R-F) polymerization [
4,
5,
6,
7,
8]. That reaction is typically catalyzed by weak bases (Na
2CO
3) and starts with the formation of anions by deprotonation of the R molecule [
8,
9,
10] favoring the subsequent F addition and then, the crosslinking in the RF-macromolecule for the formation of the organic gel. In acidic media, reactions take place with the formation of carbocations by protonation of F molecules [
11,
12].
The control of the porous structure of carbon gels is strongly important when these materials are designed to be used as adsorbents or catalysts. The structure of these materials consists of a 3D-network of spherical particles, leaving between them large volumes of macro or mesopores [
5,
10,
12]. The size and arrangement of the primary particles, and consequently the macro-mesoporous structure of the samples, is controlled during the sol-gel process and determined by the composition of the starting solution (i.e., monomer, catalyst or solvent nature and concentration) or synthesis conditions (i.e., pH, temperature), and should be preserved during the drying process [
10], supercritical drying being the best option to avoid the collapse of the porous structure. Microporosity is developed inside the primary particles during the carbonization, activation or both processes, by the release of pyrolysis gases and it is influenced by parameters such as, temperature, atmosphere or soak time.
Surfactants (S) can also be used as template or shape-directing agents of carbon gels [
13,
14,
15,
16,
17]. The electrostatic interactions between the components of the starting solutions change depending on surfactant nature (cationic, anionic or non-ionic) [
13,
14,
15,
16,
17]. The surfactant molecules can be incorporated into the RF-chemical structure forming RFS composites depending on the surfactant nature [
15,
16]. The micellar geometry is also influenced by parameters like, the pH value due to the protonation/deprotonation of acidic/basic groups, co-surfactants or metal precursors [
18,
19].
In order to improve the application of carbon gels in catalysis, an appropriate control of the porous texture, together with the development of specific active sites is needed. Carbon gels can be chemically modified with different strategies including the introduction of heteroatoms (i.e., N, B, S, P, O), but mainly metallic active phases [
7,
20,
21]. When heterogeneous catalysts based on metallic compounds are developed, the active phases can be supported by classical deposition or impregnation methods [
21], but also by doping the starting solutions of organic monomers with the soluble metal precursors [
22], directly obtaining suitable metal-doped carbon gels to be used as catalysts. Both procedures have advantages and drawbacks which have been reported [
5,
23].
In our previous manuscripts [
4,
5,
16,
23,
24], different series of heterogeneous catalysts based on carbon gels were prepared in different formats (powders, pellets, coatings, etc.) either by doping or impregnation with metal precursors. When metal precursors are solubilized in the starting solutions, ions can induce additional electrostatic interactions with the charged R-F intermediates [
25,
26], which determine the sol-gel transition, namely the primary particle size and packing and consequently, the morphology and textural properties of the RF organic gel. These parameters change as a function of the polarizing power of the counter ions, when even low amounts of different alkaline carbonates (M
2CO
3, M = Li, Na, K, Rb, Cs) are used as polymerization catalysts [
26]. The pore size distribution changes from narrow mesopores to macropores, with increasing the alkali cation size. Larger effects are provoked using transitions metals and/or high metal loadings [
22,
25]. In addition, metallic nanoparticles are formed during the synthesis procedure interacting with the organic fraction and also undergo changes according the experimental conditions of these steps (polymerization, carbonization/activation, graphitization), which determine the dispersion and nature of the metallic phases, and their catalytic performance consequently [
22,
24].
When surfactants and metal precursors are combined in the starting solutions, interactions become progressively more complex. Inorganic ions are believed to link with the hydrophilic head group of the surfactant molecule, modifying the characteristics of the micelle [
27]. The morphology of the organic RF-polymer, the nature, shape, size and distribution of the metallic nanoparticles, as well as the transformation during thermal treatment, including graphitization [
28] or carburization [
17] processes, depend on the balance of these electrostatic interactions and the possible links between phases [
16,
17,
28]. There are only a few references in this regard.
Molybdenum was selected as active phase in the corresponding catalysts because of their good performance in different environmental or energetic catalytic processes [
29,
30].
The aim of this work is to prepare a series of Mo-catalysts by polymerization of RF by adding a cationic (Mo5+) or anionic (MoO42–) metallic phase (Mo) to the starting RF solution in presence of a cationic (hexadecyltrimethylammonium bromide, CTAB) or anionic (sodium dodecyl benzene sulfonate, DCBS) surfactant at different pH values. The Mo-loading was of 1 wt% regarding reactant (R+F). The influence of these combinations on the development of the sol-gel process, the subsequent transformations during carbonization and consequently, the final physicochemical properties of Mo-doped carbon xerogels were analyzed by several characterization techniques.
3. Results and Discussion
The development of metal-doped carbon xerogels for catalytic applications aims the control of the porous texture of the carbon supports, the chemical nature and dispersion of the supported metal-phase as well as the interactions between both phases. We developed a synthesis procedure of metal-doped carbon xerogels in order to avoid some of the metal nanoparticles possibly being trapped by the organic matrix. Because the network structure of the polymeric gel is previously formed, the metal particles can be formed only in the surface. Nevertheless, impregnation is carried out before curing with the aim to avoid long aging or high temperature treatment, thus preserving the reactivity of the surface groups on the polymeric surface of the gel previously formed and favoring the Mo-anchoring. For that purpose, the experimental conditions of each synthesis step should be controlled. The morphology of the samples is defined during polymerization, although shrinkage occurs during drying and carbonization, defining the macro-meso structure of the samples. However, the thermal treatments induce the development of the microporosity, as a consequence of the release of the pyrolysis gases, and the transformations of both organic and metallic phases [
5,
28].
The chemical structure of the organic gels was analyzed by XPS. Each spectral region was de-convoluted and the atomic surface composition quantified. Results are collected in
Table 1 for the prepared samples at pH = 7. It is noteworthy that the surface Mo-content strongly depends on the cationic or anionic character of surfactants and Mo-precursors. The Mo-loading obtained is lower using the anionic surfactant, but also the anionic metal precursor regarding the cationic one, which denotes the importance of the electrostatic interactions in the doping process. The presence of heteroatoms (N or S) also confirms the incorporation of the surfactant molecules (CTAB and DCBS, respectively) to the chemical structure of the organic gels.
The components of each spectral region were fitted and assigned according to the literature. The C1s XPS spectra for both MoC-SC-pH7 and MoC-SA-pH7 are shown in
Figure 1a,b, respectively. Three components were used to deconvolute the C1s core level spectra for these organic polymers [
32]. The main C1s component is observed at a binding energy (BE) of ≈286.9 eV, corresponding to C-O bonds of aliphatic methylene ether bridges between aromatic rings and unreacted hydroxyl groups of the phenolic monomers [
10], together a smaller contribution of the aromatic and aliphatic C=C bonds at ≈285.0 eV and C=O bonds at ≈288.4 eV. It is also noteworthy that the peak of MoC-SA-pH7 is quite symmetric, while that corresponding to MoC-SC-pH7 presents a marked shoulder at ≈288.4 eV denoting a higher concentration of C=O bonds.
The O1s spectral region was fitted using two components associated to the organic phase (MoC-SC-pH7 and MoC-SA-pH7 in
Figure 1c and d, respectively). The pure MoO
3 exhibits a sharp O1s peak at BE ≈ 530.6 eV [
33], but this signal is negligible at these BE values. Thus, the oxygen content is mainly due to the organic phase, the first component at ≈532.7 eV is assigned to the C=O bonds and the second one at ≈534.3 eV to the C-O ones [
32]. In the MoC-SC-pH7 polymer the maximum of the O1s profile is located at ≈532.6 eV, while in case of MoC-SA-pH7 is observed at ≈534.1 eV, denoting the preferential C=O bonds over C-O bonds are more abundant in the second one, which is in agreement with the previous C1s spectral region fitting. As commented, the presence of heteroatoms on the organic gel structure due to the incorporation of the corresponding surfactants was also detected. In the case of CTAB (MoC-SC-pH7,
Figure 1h) the N-signal is enough to analyze the N1s core level spectra showing the formation of different N-surface groups, such as, N-pyridinic (N6), N-pyrrolic (N5), N-quaternary (NQ), which were also identified according to the literature [
34].
The Mo3d XPS spectra for MoC-SC-pH7, MoC-SA-pH7 and MoA-SC-pH7 are shown in
Figure 1e–g, respectively. The region Mo3d was also fitted using two components. The Mo
6+ in MoO
3 presents a characteristic d
5/2 and d
3/2 doublet at ≈232.9 and ≈236.0 eV, respectively, with a separation between peaks of 3.2 eV [
35], in good agreement with the results obtained for MoA-SC-pH7 (
Figure 1g), where the Mo3d
5/2 and Mo3d
3/2 components are centered at ≈233.3 and ≈236.4 eV, respectively. When cationic Mo
5+ was used as precursor, the analysis indicates that Mo
5+ was oxidized during the sol-gel synthesis, being as Mo
6+ in both MoC-SC-pH7 and MoC-SA-pH7 samples (
Figure 1e,f, respectively). Moreover, in both cases, the Mo3d
5/2 and Mo3d
3/2 components are centered at ≈234.2 and ≈237.3 eV, respectively, at BE even higher than those expected for Mo
6+. These results indicate a different chemical environment of the Mo-nanoparticles and different interactions between the Mo-organic phases. Previous studies showed that when MoO
3 is supported on organic polymers, electronic transfers between phases significantly modify the position of the XPS peaks in the different spectral regions [
36].
The carbonization process was simulated by TG, as summarized in
Figure 2. Typically, the carbonization of RF-organic gels is complete after 650–700 °C. TG curves showed the main weight loss between 300–400 °C associated with the breaking and release of weaker methylene and methylene-ether links between aromatic rings of the RF chemical structure, with a second step between 450–600 °C, where the breakage of C-H bonds and the formation of C-C ones occur [
9,
16].
Figure 2a shows the influence of the surfactant nature on the transformation of chemical structure of Mo
5+ doped samples (i.e., MoC-SA-pH7 and MoC-SC-pH7 samples). When using anionic surfactant (MoC-SA-pH7) there is a certain initial weight loss below 100 °C due to the loss of solvent and unreacted products, although the main decomposition process occurs at ≈420 °C. Moreover, a small shoulder is formed at ≈320 °C, and finally, another weight loss at around 650 °C. The differential thermogravimetric (DTG) profile obtained for the MoC-SC-pH7 sample, prepared using cationic surfactant, showed the main weight loss also at around ≈420 °C, mainly denoting a similar bond breakage and thermal stability of the RF-structure; however, other significant weight loss can be observed at ≈240 °C and ≈810 °C regarding the MoC-SA-pH7 sample. The first one is due to the decomposition of the groups generated by the incorporation of the CTAB surfactant into the chemical structure of RF-gel, which decompose at this temperature range [
16]. The last sharp DTG-peak at high temperature (ca. 810 °C) can be associated with the reduction of the molybdenum oxide phases, as it will be discussed below. Using the combination of CTAB and MoO
42– precursors to point out the influence of pH on the chemical stability (MoA-SC-pH7 and MoA-SC-pH5 samples,
Figure 2b), it is observed that increasing pH of the starting solution, the DTG profile is maintained but the position of the main DTG bands are slowly shifted to higher temperatures, denoting some chemical reinforcement. This shifting could also be influenced by the different porosity of the samples which can determine the removal of the pyrolysis gases (
Figure 2b).
The morphology of Mo-doped carbon xerogels were analyzed by SEM (
Figure 3). The images pointed out the strong influences of the different combinations of cationic and anionic species in the starting solution. Materials obtained using the anionic surfactant (MoA-SA-pH7 and MoC-SA-pH7 samples,
Figure 3a,b, respectively) are composed of big particles with a smooth and compact surface, regardless of the cationic or anionic nature of the metal precursor used. However, using the cationic surfactant, (MoC-SC-pH7,
Figure 3d), the structure obtained resembles the typical RF structure [
10], consisting in a network of highly interconnected spherical particles (primary particles formed during the sol state), leaving a developed porosity between them. This structure depends on the synthesis conditions, thus in the MoA-SC-pH7 sample (
Figure 3c) the primary spherical particles can form filaments, while changing the pH of solution either to acidic (the MoA-SC-pH5 sample,
Figure 3f) or basic (the MoA-SC-pH10 sample,
Figure 3e), the primary particles are not observed, and in both cases the coalescence of these particles is favored forming larger structures that leave large macropores between them. In the case of MoA-SC-pH5, the acidic media also induces a greater coalescence of these structures leading to a continuous solid exhibiting a foam structure. Thus, a wide variety of morphologies were observed depending on the reactant combinations.
The change in the morphology of the samples that also influences their porous texture was analyzed by physical adsorption of N
2 and CO
2. The corresponding N
2-adsorption isotherms of carbonized samples are recorded in
Figure 4 and the textural parameters obtained from them are collected in
Table 2. CO
2 isotherms are shown in
Figure S1a and b of the
Supplementary Materials. In general, all isotherms correspond to the type IV associated to the mesoporous character of carbon gels. After the micropore filling at low P/P
0, the slope of the curve and the formation of a certain hysteresis loop denotes the formation of mesopores. The CO
2 adsorption isotherms (not shown) provide information about the narrowest microporosity [
37]. In all cases, there is a strong diffusion restriction to the narrowest microporosity, thus the micropore volume and the surface area determined from the CO
2-adsorption isotherms are always larger than those obtained from the N
2-isotherms (
Table 2).
The porosity of the samples decreased in both acidic and basic media (MoA-SC-pH5 or MoA-SC-pH10), regarding neutral pH values (MoA-SC-pH7), as shown in
Figure 4b. However, the volume of the narrowest micropores is favored by increasing the pH of the starting solution, although the mean micropore size is maintained at around 0.55 nm, that is clearly inaccessible to N
2 at –196 °C. Note that the mean micropore width (
L0) determined by N
2-adsorption is larger than 1.1 nm in all cases and that the mesopore volume (
Vmeso) decreased regarding the sample obtained at neutral pH (MoA-SC-pH7), probably associated with the transformation of mesopores into micro and macropores in acidic or basic conditions, as previously denoted by SEM. These results are in agreement with those previously published [
9,
38]. Job et al. [
38] obtained pure micro-mesoporous carbon materials at pH ranging from 5.5–6.3, but this pH range changes in the presence of metals (Ni) to 6.5–7.3 for metal-doped carbon xerogels, but they obtained exclusively microporous materials in basic media, which showed a strong porosity reduction after carbonization, depending on metal and complexing agent.
MoC-SC-pH7 was the most porous sample obtained using the cationic surfactant and cationic metal precursor at neutral pH. On the contrary, the less porous sample (i.e., MoA-SA-pH7), was obtained using both anionic surfactant and metal precursor. Intermediate behaviors were obtained using cationic surfactant and anionic metal precursor (MoA-SC-PH7) or anionic surfactant with cationic metal precursor (MoC-SA-pH7). As previously commented, the RF polymerization using soft basic catalysts, develops forming resorcinol anions which favor the F-addition reactions and consequently the crosslinking in the growing polymeric macromolecule. These results indicate that: (i) electrostatic interactions between the chemical species are important to control the crosslinking during the polymerization of RF, (ii) anionic species of the growing macromolecular polymer interact with the rest of the charged species in solutions, (iii) cationic species produce attractive interactions facilitating the incorporation into the molecular structure of the polymers (including surfactants and metallic phases, as previously denoted) and (iv) repulsive interactions with anionic species lead to non-porous samples with lower metal-loadings.
The chemical state and dispersion of the metallic phase in the carbonized samples were analyzed by DRX, XPS and HRTEM in order to analyze the transformations during carbonization and the stabilization of the corresponding Mo-active phases. The XRD-patterns of the carbonized samples are showed in
Figure 5. The large bands at around 25° and 42° are due to the (002) and (101) diffractions of graphite and denote certain but poor ordering of the carbon support. Clearly, molybdenum possesses a weak influence on the graphitization of the carbon phase regarding other metals like Fe or Ni [
28]. In general, no diffraction peaks corresponding to the Mo-phase were detected using molybdate species as metal precursor, only small peaks at ≈39.5 and ≈61.7°, assigned to the most intense peaks of β-Mo
2C, are detected in MoA-SC-pH5. Nevertheless, in our previous studies, the formation of highly crystalline β-Mo
2C in Mo-doped carbon xerogels was detected using MoA-SC combinations in acidic conditions, however, a higher TMB/t-BuOH concentration was used [
16]. When cationic CTAB and Mo
5+ were used (MoC-SC-pH7 sample), intense diffraction peaks at ≈34.4, ≈37.9, ≈39.5, ≈52.2, ≈61.7 and ≈69.6° corresponding to the (100), (002), (101), (110) and (103) planes of hexagonal β-Mo
2C (JCPDS 35-0787) were observed, denoting the high crystallinity of this phase [
35]. Nevertheless, none of the possible intermediate molybdenum oxides (MoO
x) were detected, pointing out that these phases are mainly amorphous. Small diffraction peaks were also observed in sample MoC-SA-pH7 (using DCBS) at ≈9.5, ≈14.5, ≈16.1, ≈27.2, ≈32.7 or ≈39.4°. These peaks can be due to the formation of sodium molybdate species, as reported by Wang et al. [
39]. They used surfactants as unique carbon precursors and ammonium heptamolybdate as Mo-precursor to obtain β-Mo
2C. They obtained aggregates of β-Mo
2C crystallites with different structures, from hollow rods to hollow spheres when using polyethylene glycol (PEG) or CTAB, respectively, and big platelets using DCBS. These platelet structures showed low surface and are contaminated by the formation of Na
2Mo
3O
3 (JCPDS 28-1111) as impurity, related to the significant amount of Na
+ added as a counter-ion with the anionic surfactant [
37].
XPS analyses of Mo-doped carbon xerogels were also performed and the C1s, O1s and Mo3d spectral regions were properly deconvoluted, as shown in
Figure 6. The chemical surface composition of the different samples is summarized in
Table 1, together with those corresponding organic precursors. The carbon content increases as the heteroatom content decreases (both O and N) during carbonization because they are released as pyrolysis gases (CO
x, NO
x). The metal content is also modified as a consequence of sintering or reaction with the organic phase. Analogously to the organic gels, a great difference in the metallic surface concentration present on carbon xerogels is obtained using the anionic surfactant regarding those obtained using the cationic one.
Figure 6 depicted the Mo3d region for carbon samples obtained with different combinations of anionic and cationic surfactants and metal precursors, the different chemical state of Mo in each sample being obvious at a glance. The Mo3d spectra of the MoA-SA-pH7 sample (both anionic surfactant and metal precursor,
Figure 6a) was fitted using only a doublet, the peaks at ≈232.9 and ≈236.0 eV corresponding to the d
5/2 and d
3/2 components of Mo
6+ in MoO
3 [
35], as previously described. The oxide should be mainly amorphous, taking into account the previous XRD (
Figure 5). In the MoC-SA-pH7 sample (anionic surfactant and cationic Mo
5+ precursor), the Mo spectra is more complex (
Figure 6b). As described for the organic xerogels, the main peaks at ≈234.7 and ≈237.3 eV denote the formation of highly oxidized Mo-species. These MoO
3 particles are in a more withdrawing environment than in MoA-SA-pH7, because the components of Mo
6+ are shifted to significantly higher BE. The new peaks, not observed in the organic xerogels (
Figure 1f), were located at ≈231.4 and ≈234.4 eV and assigned to the d
5/2 and d
3/2 signals of Mo
5+, respectively [
39]. Moreover, a small peak located at ≈228.3 eV can be assigned to Mo
2+ [
40]. In the case of the MoA-SC-pH7 sample (
Figure 6c), i.e., combining cationic surfactant and anionic Mo-precursor, the main peaks located at ≈233.0 and ≈236.2 eV showed that molybdenum remains as Mo
6+ in MoO
3, but in this case the peak at ≈228.4 eV also confirms the Mo
6+ reduction to Mo
2+. The reduced Mo-fraction corresponds to a 10% of the total Mo-content observed. Finally, the combination of the cationic Mo
5+ precursor with cationic surfactant (MoC-SC-pH7,
Figure 6d) leads to a strong increase of the peak at ≈228.6 eV, denoting the significant reduction of Mo-phase to Mo
2+ (reaching 60% of the total Mo-content observed). Although this peak can be associated with the presence of an intermediate oxide (MoO
2) [
40], the obtained XRD patterns of this sample (
Figure 5) provides unequivocal proof of the formation of β-Mo
2C crystallites, thus the peak at ≈228.6 eV can be assigned to Mo
2+ in molybdenum carbide (Mo
xC); also in agreement with the literature [
39,
41]. Similar XPS spectra (not shown), were also detected for samples obtained at different pH (MoA-SC-pH10 and MoA-SC-pH5), Mo
2+ and Mo
6+ species being detected, although the proportion of Mo
2+ changed from acidic media (37%) to basic media (43%). The Mo-reduction and the fraction of Mo-carbide formed strongly depend on the electrostatic interactions during polymerization that determine the nature of metallic–organic phase links and therefore, the transformations during carbonization.
The shape, dimension and distribution of Mo-particles were analyzed by HRTEM. Representative HRTEM images of the Mo-doped carbon xerogels are shown in
Figure 7. When both anionic DCBS and MoO
42– (MoA-SA-pH7) are combined, the formation of big Mo-particles was observed together with a significant distribution of small nanoparticles (
Figure 7a,b). These large particles present an ordered structure formed by the stacking of platelet particles, as previously described by Wang et al. [
39]. The selected area electron diffraction (SAED,
Figure 7b inset) clearly showed a certain crystallinity, although it was not detected in XRD patterns (
Figure 5). The formation of such large structures can be related with the low Mo-content detected by XPS and a poor porosity. When using Mo
+5 (MoC-SA-pH7) needle shaped particles are formed instead of the platelet ones (
Figure 7c–d), together at a good concentration of smaller nanoparticles. Using the CTAB surfactant, a high concentration of small homogeneously distributed Mo-nanoparticles is always observed (
Figure 7e–h), independently of the morphology of the carbon supports, the nature of Mo-precursor or the pH value, which is also consistent with the XPS results. It is interesting to note that the carbon foam structure observed for MoA-SC-pH5 or the carbon filaments obtained in the case of MoA-SC-pH7 are highly coated with metallic nanoparticles (
Figure 7e,g respectively).
As summary, a scheme with all possible combinations concerning surfactants and metal precursors with the organic RF matrix is showed in
Figure 8. Overall, both cationic and metal precursors are desirable because an enhanced porosity and Mo-reduction and dispersion are obtained in the final Mo-doped carbon xerogel.