An underrated variable essential for tailoring the structure of xerogel: the methanol content of commercial formaldehyde solutions

Resorcinol-formaldehyde xerogels are polymers whose porosity can be designed for a specific application by the selection of appropriate physical and chemical synthesis variables. Until recently, the methanol content of commercial formaldehyde solutions has never been considered as a chemical variable that must also be taken into account. However, it has been demonstrated that the proportion of methanol might be even more important than other variables for tailoring the porosity of xerogels. Different reaction mechanisms are proposed to explain this heavy dependence of the final structure of RF xerogels. Organic and carbon xerogels synthesized with low concentrations of methanol showed a higher porosity with much larger pore sizes (up to two orders of magnitude) than when using formaldehyde with high concentrations of methanol. This means that extra caution needs to be shown in choosing these commercial products since formaldehyde solutions differ greatly in methanol content from one supplier to another.

Microwave radiation is the most common method used to synthesize this type of material since it is a highly competitive process that requires only 3 h instead of the 5 days needed for conventional heating.
The porosity of the samples studied was characterized by means of mercury porosimetry on an AutoPore IV 9500 (Micromeritics) device, which is able to measure from atmospheric pressure up to 228 MPa. To characterize mesoporosity the lowest limit of the apparatus considered was 5.5 nm.
Likewise, Vmacro refers to porosity ranging from 50 to 10000 nm. The surface tension and contact angle were 485 mN m -1 and 130º, respectively, while the stem volume was between 45-58 % in all the analyses performed. Initially, in the low pressure step, the samples were evacuated at 6.7 Pa and the equilibration time chosen was 10 seconds. Subsequently, the pressure was gradually increased to its maximum value, after which the mercury intrusion was evaluated. This characterization yielded data about pore size distribution, pore volume and bulk density.

Microscopy
Micrographs of the carbon xerogels were obtained on a Zeiss DSM 942 scanning electron microscope (SEM). For this characterization the samples were previously dried overnight and then attached to an aluminum tap using a conductive double-sided adhesive tape. A voltage of 25 kV and a secondary electron detector EDT Everhart-Thornley, were used in all the analyses.
Transmission electron microscopy (TEM) was carried out on a Jeol 2000 at 180kV. A very dilute ethanolic suspension was prepared and then deposited on a 200 mesh copper grid with carbon membrane.
The samples were analyzed directly after synthesis and also after being dried overnight under vacuum at room temperature by 1 H and 13 C solid state NMR spectroscopy and 1 H NMR imaging (MRI).
The 1 H and 13 C NMR measurements were performed on a Bruker Avance 400 spectrometer (Bruker BioSpin, Rheinstetten, Germany) equipped with a 89 mm wide bore, 9.4 T superconducting magnet (at 1 H and 13 C Larmor frequencies of 400.14 and 100.61 MHz, respectively). All the reported data were obtained at room temperature. Powdered samples were placed in 4 mm zirconia rotors and a standard Bruker double resonance 4 mm cross-polarization (CP)/magic angle spinning (MAS) NMR probe head was used. The 1 H MAS spectra were recorded using single pulse excitation and the 13 C CP/MAS spectra were obtained by applying a 3 ms CP contact time and sideband suppression (seltics). The recycle delays in the 1 H and 13 C experiments were 5 and 3 s, respectively. The MAS spinning rates varied between 5.0 and 7.0 kHz. 13 C NMR spectra were obtained by high-power proton decoupling at 75 kHz.
In all cases, the NMR spectra were evaluated using the manufacture's Top-Spin™ spectrometer software package. All the free-induction decays were subjected to the standard Fourier transformation procedure with line broadening ( 13 C: 50-100 Hz; 1 H: 20 Hz) and phasing. The chemical shifts were externally referenced to adamantane ( 13 Cδ: 29.5 ppm; 1 Hδ: 1.63 ppm) secondary to tetramethylsilane ( 13 C and 1 H δ: 0.0 ppm).
The proton NMR imaging measurements were performed on the spectrometer described above fitted with a microimaging accessory. The maximum possible amplitude available for the magnetic field gradients was 97.3 G cm -1 . Cylindrical monoliths of each sample cut into pieces of about 10 mm in length were prepared and analyzed by 1 H MRI after being dried overnight under vacuum at room temperature ( Figure 1). Glass tubes with an outer diameter of 20 mm and a height of 8 cm were employed as sample holders. The cylindrical samples were fully impregnated with cyclohexane under vacuum, and loaded into an NMR probe fitted with a radiofrequency coil insert of diameter 25 mm. All measurements were performed at 25 ± 0.2 ºC. Scout images were recorded using a two-dimensional multislice gradient-echo pulse sequence with an echo time (TE) of 1.86 ms and a repetition time (TR) of 300 ms. The images were obtained in three orthogonal planes, 1 mm thick, with a field of view (FOV) of 25 x 25 mm 2 and an isotropic resolution of 128 x 128 pixels. To characterize the morphological features of the xerogels, fully relaxed (TR > 5 x T1) multislice/multispin-echo images at echo times between 5 and 40 ms were recorded. Eight transaxial images with a slice thickness of 0.5 mm, a gap between slices of 0.4 mm, and an in-plane isotropic resolution of 78 x 78 μm 2 were obtained. T 1 measurements were performed using an inversion-recovery pulse sequence. The images were processed using ImageJ software (National Institute of Health, Bethesda, US).

Fourier Transform Infrared Spectroscopy
The chemical surface characterization of the powdered samples was carried out by Fourier Transform Infrared Spectroscopy (FTIR). Samples were dried overnight before characterization. The spectra were recorded in the 525 and 4000 cm -1 range on a Nicolet IR 8700 spectrometer fitted with a DTGS detector (deuterated triglycine sulphate) at a 4 cm -1 resolution in 64 accumulated scans. The samples were analyzed twice. In order to prepare the tablets, the samples and KBr (previously dried overnight) were mixed in a proportion of 1:100 in an agate mortar for 10 minutes until a homogeneous mixture was obtained. Around 0.125 g of this mixture was subjected to 8 tons of pressure in a 13 mm diameter matrix.

Results and Discussion
Two samples synthesized under the same conditions, except for the methanol content in the formaldehyde solution (i.e., 0.7 and 12.5), were studied by means of different techniques. The pore size distribution of each sample, both before and after carbonization: OX and CX respectively, is shown in Figure 2. It can be seen that a decrease in the methanol content of the formaldehyde solution causes an increase in the pore diameter of up to two orders of magnitude. Moreover, an increase in the methanol content produces an increase in the bulk density of the material from 0.34 to 0.41 g/cm 3 for OX-0.7 and OX-12.5, respectively (Table 1), leading to radically different material. These variations may result from a change of supplier or if a different batch of formaldehyde from the same supplier is used. The differences in methanol content are of crucial importance since the main advantage of organic and carbon gels are that they allow porosities to be tailored for specific applications.
To investigate possible differences in the chemical structure of xerogels, solid state NMR spectroscopy and FTIR measurements were carried out. Figure 3a shows the 13 C CP/MAS NMR spectra corresponding to samples OX-0.7 and OX-12.5. The spectra are not significantly different and show 5 main resonances centered at 151.7; 132.4; 120.5; 103.7, and 28.3 ppm. The 151.7 peak corresponds to phenolic carbons from the substituted resorcinol. Due to the fact that aromatic rings can be substituted by one, two or three bridges with respect to the adjacent rings, this gives rise to asymmetric resonances.
The peak at 132.4 corresponds to carbons in meta position with respect to both phenols, whereas the peak at 120.5 ppm corresponds to mono and bi-substituted aromatic carbons in ortho position relative to the phenol group. The peak at 103.7 ppm corresponds to the carbons between the phenolic OH and the wide peak at 28.3 is assigned to different CH 2 bridges [22].
The 13 C CP/MAS NMR spectra of samples CX-0.7 and CX-12.5 are shown in Figure 3b. In this case, there is only one peak centered at ca. 127 ppm that can be assigned to C-C sp 2 bonds (the other peaks in the top trace are spinning sidebands that have not been completely suppressed). As CP/MAS experiments favor the detection of carbons with nearby protons, a comparison of the spectra of samples OX and CX, suggests that the CX samples have a lower proton density than the OX ones, especially, sample CX-12.5 (which has lower signal intensity than the rest). It should also be pointed out that the spectra of the OX samples are the result of averaging 1,000 scans, while those of CX represent an average of 10,000 scans. Figure 4 shows the 1 H MAS spectra of the organic samples (OX) analyzed before and after drying. The spectra in Figure 4a correspond to sample OX-0.7. A narrow peak at 4.62 ppm only appears in the sample prior to drying (top trace), and the broad peak at 3.94 ppm shifts to 3.49 after drying (bottom trace). Likewise, Figure 4b illustrates the spectra corresponding to sample OX-12.5. In this case, only one peak at 4.06 ppm is visible before drying. After drying, there is only one broad peak at 3.87 ppm is observed. In both cases, the narrow linewidth of the peaks observed in the xerogels before drying suggests that they are associated to protonated species with a high mobility. The broad peaks observed in the dried samples (bottom traces in Figure 4) could be associated to protons with a low mobility (protons bound to the polymer network). The carbonized samples were also characterized. However, the differences between the pristine and dried samples are very small, reflecting the weak hydrophilic character of the CX samples due to the absence of surface groups (Figures S1 and S2 in the supplementary material).
The FTIR spectra of the OX xerogels are presented in Figure 5. Both spectra exhibit a broad band between 3700 and 3000 cm -1 associated with O-H stretching vibrations, due to phenol groups with a high concentration of hydroxyl groups on their surfaces, providing these materials with a high hydrophilicity. The band at 2931 cm -1 can be assigned to aliphatic stretching vibrations (CH 2 ). The band located at 1474 cm -1 corresponds to CH 2 deformation vibrations and the one at 1613 cm -1 to the aromatic ring stretching vibration (C=C). Stretching vibration bands at 1217 and 1092 cm -1 indicate the presence of methylene ether bridges C-O-C), although they should be not dominant as they were not detected by NMR [23,24]. It should also be noted that, as in the case of NMR, FTIR revealed identical bands in both organic samples, indicating similar chemical compositions. Elemental chemical analysis (Table 2) also corroborated the similar composition of the OX samples, independently of the amount of methanol used. In fact, according to the values presented in Table 2 the C/H and C/O ratio are 1.0 and 2.5, respectively, for OX-0.7; and 1.1 and 2.6, respectively, for OX-12. 5. These values are also in good agreement with the theoretical ratios from the proposed mechanisms in the bibliography for RF xerogels (i.e. 1.0 and 2.6 for C/H and C/O, respectively). The carbonized samples (CX) were also analyzed, and again their chemical composition seems to be independent of the amount of methanol used ( Table 2).
The CX samples have a carbon content of ca. 95 wt%. The FTIR spectra of the CX samples reveal a poor surface chemistry, a feature typical of a carbonized material (see Figure S3). It is also worth noting that in no case do these xerogels (OX and CX samples) show signs of impurity content.
It is clear from Figure 4 that all of these organic materials are sensitive to air humidity since water is easily absorbed by their surfaces. Furthermore, Figure 5 shows a high hydroxyl group content on the surface of these organic materials which endows them with a high hydrophilicity. What is more, their porosity provides them with a good water storage capacity, giving rise to a material with outstanding desiccant properties [25]. The FTIR spectra of the CX samples (data not shown) were characteristic of high temperature carbon materials, i.e., there being no bands detected.  Table 2.
This higher surface oxygen content helps to depict some slightly differences between the two CX materials, hence on the possible influence of the methanol content on the final properties of the materials. The deconvolution process of the high resolution C1s profiles is shown in Figure 6.  Table 3.
It was notice that both samples (CX-12.5 and CX-0.7) present a very similar chemistry. The most relevant difference between them is the relative contribution of O=C-OR functionalities, with a little higher presence in the sample CX-12.5. These would suggest that this carbon material, prepared using the formaldehyde solution containing a higher content of MeOH, leads to the formation of a slightly higher number of ester bonds which are most likely related to high levels of crosslinking between the nodules. This would agree with a more condensed structure, leading narrow pores as it is observed in the porosity characterization mentioned above, besides the SEM and TEM observations that would be discussed later. cyclohexane. It can be seen that the images corresponding to the xerogels prepared with an initially lower content of methanol (0.7 % w/w) have a courser appearance than those prepared with higher methanol content (12.5 % w/w). This is in agreement with the electron microscopy observations (TEM and SEM) described previously, and with the data in Table 1, since an increase in the methanol content indicates an increase in density, which naturally entails a decrease in porosity.
Different reaction mechanisms are proposed based on a detailed chemical and structural analysis. First, however, it should be mentioned that methanol is added to the formaldehyde solutions in order to prevent the formaldehyde from polymerizing itself and subsequent precipitation. As illustrated in Figure   9, the reaction of methanol with formaldehyde is favored in acidic media, giving rise to mainly hemiacetals ( Figure 9a Furthermore, the greater the amount of free formaldehyde available, the smaller the number of interconnections formed between large-size nodules, as is corroborated by Figures 5-6. On the other hand, the smaller the concentration of free formaldehyde, the larger the number of interconnections formed between the small size nodules. It can also be said that, a similar pattern is observed when there is a reduction in the R/F molar ratio [14]. The lower the R/F ratio is, the higher the proportion of methanol present since more formaldehyde is consumed in the reaction media which will lead to the effects of the methanol being more prominent.
It is well known that all chemical variables involved in carbon xerogel synthesis are interdependent.
However, although the effect of methanol is always apparent, the extent of the effect depends on the other variables. For example, the start of the reaction is favored when the pH of the initial precursor solution (ca. 3) is increased due to the addition of the basic catalyst. The formation of the resorcinol anions to initiate the addition reaction is favored (see Figure 10). As the kinetics of the reaction increases, small clusters are formed, giving rise to materials with a small pore size. If the kinetics of the reaction is increased, the methanol content is not as important as it would be at a low pH. In this case, crosslinking reactions that counteract the growth of nodules are favored. Consequently the availability of formaldehyde due to methanol content has less influence on the porosity of the material. In short, methanol content has more effect when the kinetics of the reaction is slow or, in other words, when low amount of catalyst is used.
The effect of the dilution ratio has also been widely studied in the literature. However, there are some differences between authors possibly due to the different synthesis methods or drying methods employed [13, 32, 33]. Under microwave heating, the sol-gel reaction proceeds faster than under conventional heating, leaving the solvent no time to evaporate and so it is retained inside the structure and this may alter the structure somewhat. A lower dilution degree implies a higher proportion of reactants in the solution and, therefore, an increase in the number of smaller size nodules formed. On the other hand, an increase in the dilution ratio will cause the formation of a smaller number of nodules of larger size. Furthermore, a high dilution ratio will lead to a lower degree of crosslinking and materials with poorer mechanical properties. If the dilution ratio is increased even further, gel will not be able to form since the nodules will be too far apart to crosslink [34].

Conclusions
The concentration of methanol present in commercial formaldehyde solutions has a significant influence on the porosity of RF xerogels. The bulk chemical composition remains analogous in spite of some differences in the structure and possibly some functional groups. Different reaction mechanisms have been proposed to explain the strong dependence of the final structure of RF xerogels on the presence of methanol. Methanol reacts with formaldehyde, generating hemiacetal molecules which are not able to react with resorcinol. Therefore, the greater the amount of methanol, the smaller the amount of formaldehyde that will be available to participate in reactions, which means that resorcinol molecules will have difficulty finding formaldehyde in its free form to react with. In short, organic and carbon xerogels synthesized with a lower concentration of methanol show a higher level of porosity with larger pore sizes (by up to two orders of magnitude) than when formaldehyde with high concentrations of methanol is used. Consequently, in order to obtain accurate bespoke materials, apart from the already established chemical variables (i.e., pH of the precursor solution, R/F ratio, dilution ratio, etc) it is necessary to take into account the composition of the formaldehyde solution and especially the percentage of methanol.                    drying. The spectra were acquired with the same conditions and are plotted with the same vertical scale.
A broad peak centered at about 6 ppm is observed in both spectra. The narrow peak shows changes after drying, the peak at 0.13 ppm is observed in the two spectra, while the peak at -0.71 ppm is only visible in the untreated sample. Figure S2. 1 H MAS spectra corresponding to sample CX-12.5 before (top trace) and after (bottom trace) drying. The spectra were acquired with the same conditions and are plotted with the same vertical scale. A broad peak centered at about 5.5 ppm is observed in both spectra. The intensity of the narrow peak decreases after drying, and shifts from -0.99 to 0.02 ppm. Figure S3. FTIR spectra for the carbonized samples CX-0.7 and CX-12.5 (note, it was used the same scale as Figure 5 for comparison proposes).