Synthesis of Mesoporous Metal Oxides by Structure Replication: Thermal Analysis of Metal Nitrates in Porous Carbon Matrices

A variety of metal nitrates were filled into the pores of an ordered mesoporous CMK-3 carbon matrix by solution-based impregnation. Thermal conversion of the metal nitrates into the respective metal oxides, and subsequent removal of the carbon matrix by thermal combustion, provides a versatile means to prepare mesoporous metal oxides (so-called nanocasting). This study aims to monitor the thermally induced processes by thermogravimetric analysis (TGA), coupled with mass ion detection (MS). The highly dispersed metal nitrates in the pores of the carbon matrix tend to react to the respective metal oxides at lower temperature than reported in the literature for pure, i.e., carbon-free, metal nitrates. The subsequent thermal combustion of the CMK-3 carbon matrix also occurs at lower temperature, which is explained by a catalytic effect of the metal oxides present in the pores. This catalytic effect is particularly strong for oxides of redox active metals, such as transition group VII and VIII metals (Mn, Fe, Co, Ni), Cu, and Ce.

Here, were present the thermogravimetric analysis of the thermally induced processes during the nanocasting synthesis of various metal oxides by using mesoporous CMK-3 carbon as the structure matrix. Both the conversion of the respective metal nitrate into the metal oxide and the thermal combustion of the carbon matrix turn out to occur at temperatures different from those observed for the pure components. Figure 1 shows the thermogravimetric analysis (TGA) of pure CMK-3 carbon that was prepared from sucrose by the standard literature procedures [9]. All thermal analyses was performed at a heating rate of 10 °C min −1 . The gas atmosphere was O2/Ar (20:80 vol.) instead of air, in order to prevent interference of N2 + with CO + in gas ion detection (as both ions exhibit the same mass). The thermal combustion of the CMK-3 carbon occurs at a temperature between ca. 450 °C and 610 °C. It is marked by a mass loss (TG signal) of ca. 90%, as depicted more clearly by the first derivative of the TG curve (DTG), accompanied by the detection of the CO2 + cation (m/z = 44). Small amounts of the H2O + molecule cation (m/z = 18) are also observed during the combustion, which can be explained by some oxygen-containing organic functions in the CMK-3 carbon network, presumably at the pore wall surface. Prior to the combustion, especially in the temperature interval up to 100 °C, a mass loss of ca. 10% is observed which is attributable to the evaporation of physisorbed water, as confirmed by the detection of the H2O + ion. As stated in the introduction, mesoporous CMK-3 serves as a matrix for the synthesis of mesoporous metal oxides by structure replication (nanocasting). For this purpose an aqueous solution of a metal nitrate is typically filled into the carbon pores, followed by drying and subsequent conversion into the respective metal oxide by heating under air atmosphere. Depending on several factors, such as the metal nitrate's solubility and the metal oxide's density, this procedure typically leads to a filling of the carbon pores between 10% and 30% [2][3][4][5][6]. This procedure is applicable only if the temperature required for the conversion is lower than the temperature at which the CMK-3 carbon matrix will oxidize. This is usually the case, as will be shown below. Figure 2 displays the thermogravimetric analysis of ordered mesoporous CMK-3 carbon after infiltration of Al(NO3)3 as an example. Aluminum nitrate, like most other metal nitrates, my contain variable amounts of crystal water, depending on the drying conditions; in the following, we deliberately omit any such information, since the true amount of crystal water/residual solvent water is ill-defined. The thermal profile is marked by two distinct steps: (i) ca. 25% of the sample mass is lost between 100 and 255 °C. In this interval the NO + (m/z = 30), NO2 + (m/z = 46), and H2O + (m/z = 18) molecule ions are detected, confirming the conversion of Al(NO3)3 to Al2O3 and the release of water (presumably, mostly crystal water present in Al(NO3)3). Additionally, very weak signals of CO2 + (m/z = 44) and CO + (m/z = 28) are observed at this temperature, which are absent in the thermal profile of pure CMK-3 carbon; apparently, the presence of finely dispersed Al(NO3)3 seems to have a weak oxidative/catalytic effect on the thermal combustion of carbon. (ii) The second step of mass loss (ca. 50%) occurs between 380 and 665 °C. This is attributable to the thermal combustion of the CMK-3 carbon, as confirmed by the detection of the CO2 + (m/z = 44) and the CO + (m/z = 28) molecule cations. A weak signal is also observed in the detection of molecule cations with m/z = 46; this corresponds most likely to the 12 C 16 O 18 O + rather than the NO2 + cation. It is apparent that the temperature intervals in which the two instances (conversion of Al(NO3)3 to Al2O3 and carbon combustion) occur do not overlap. This has practical consequences in the nanocasting synthesis of Al2O3, as repeated cycles of nitrate infiltration and subsequent oxide formation turn out to be possible without premature damage of the CMK-3 carbon matrix. It may be worth mentioning that the final temperature of 800 °C may be higher than what is usually used in a true nanocasting synthesis. To preserve the nanoscopic structure and porosity, temperatures lower than 800 °C may be advantageous, in which case complete removal of the carbon matrix must be achieved by isothermic treatment for longer periods. For mesoporous Al2O3 we have previously used a maximum temperature of 450 °C with isothermic heating for 48 h [21]. Comparison of Figures 1 and 2 reveals that in the Al(NO3)3-/Al2O3-loaded sample (i) the thermal combustion of carbon sets in at lower temperature, as will be discussed in detail below, and (ii) the release of CO2 lasts up to higher temperature. The latter observation may be attributable to adsorption of CO2 to the Al2O3 phase, potentially up to the point of formation of a carbonate species; the release of CO2 would then be delayed. Additionally, the gas-phase transport of oxygen through the pores of the carbon may be less effective owing to the presence of Al2O3; thus, the oxidation of carbon would be kinetically retarded and, therefore, completed at higher temperature under the non-isothermal conditions of this measurement.

Results and Discussion
The same analysis was performed for CMK-3 carbon loaded with several other metal nitrates in its pores, namely Zr(NO3)4, Mg(NO3)2 [13], Zn(NO3)2 [15], Ni(NO3)2, Fe(NO3)3, Mn(NO3)2, Co(NO3)2, Cu(NO3)2, and Ce(NO3)3 [20]. Figure 3 shows the mass loss (TG and derivative TG) as well as the signal of the NO + (m/z = 30) and CO2 + (m/z = 44) molecule cation detection. In all cases, the thermal profiles resemble the same general picture as observed for the Al(NO3)3-impregnated sample in that two instances can be distinguished: (i) at relatively low temperature the conversion of the metal nitrate into the respective metal oxide occurs, as observed by a first step in the mass loss and by the detection of the NO + cation (as well as the NO2 + and the H2O + cations; not shown); (ii) At higher temperature the combustion of the porous CMK-3 carbon matrix occurs, as observed by a second step in the mass loss and by the detection of the CO2 + cation (as well as the CO + cation; not shown). However, it is apparent that the respective temperature intervals, in which the two instances occur, differ from case to case, as depicted more clearly in Table 1 and Figure 4.   Note: * temperature of thermal decomposition of metal nitrates according to [27].
The onset of the thermally induced, non-isothermic conversion of the metal nitrates into the respective metal oxides ranges from 70 to 125 °C. The temperature at which this conversion is finished varies more significantly, between 215 and 460 °C. The onset values are always lower than the decomposition temperatures for the pure metal nitrates (i.e., in the absence of carbon) reported in the literature [27]; the latter were measured by thermal analysis with the same heating rate, but in air instead of Ar/O2 atmosphere. In some cases, even the offset temperature is lower than the literature value. Hence, it is fair to say that metal nitrates dispersed in the nanopores of CMK-3 carbon tend to decompose at lower temperature than otherwise. A possible explanation is the high dispersity of the metal nitrate in the carbon pores. In general, any comparison of thermal analysis data with previous ones needs to be carried out with great care; apart from the heating rate, thermal profiles may sensitively depend to a high degree on such experimental parameters as sample amount or crucible design.
More interestingly, the temperature interval for the thermal decomposition of the CMK-3 carbon matrix also varies substantially, depending on which metal nitrate is used. The onset temperature is always substantially lower than for the pure, i.e., metal nitrate-free, CMK-3 carbon material (450 °C); is varies between 180 °C (for the Ce(NO3)3-loaded sample) and 380 °C (for the Al(NO3)3-loaded sample). The offset temperature follows the same trend; in most cases it is lower than for the pure CMK-3 carbon (610 °C). It is also worth noticing that the combustion of carbon rarely overlaps with the conversion of the metal nitrate into the respective metal oxide; such overlap is observed only for the three samples with the lowest onset temperature of carbon combustion (Co(NO3)2, Cu(NO3)2, and Ce(NO3)3), as well as for the Ni(NO3)2-loaded sample. The thermal behavior of the Ni(NO3)2-loaded sample is more complex than that of the other samples. Apparently, the reaction of (hydrous) Ni(NO3)2 to NiO occurs in several distinct steps, as also confirmed in the literature [28]. As a general outcome the data clearly show that the oxidative combustion of CMK-3 carbon is facilitated by the presence of the metal oxides that have formed inside the carbon pores. Obviously, the finely dispersed metal oxides catalyze the combustion, which results in lower combustion temperature. Moreover, this effect is correlated with the redox activity of the respective metal: a strong catalytic effect is observed for the oxides of transition group VII and VIII metals (Mn, Fe, Co, Ni), as well as of Cu and Ce. The oxides of less redox-active metals (Al, Mg, Zn, Zr) show weaker catalytic effect. Similar findings were observed for metal nitrate-impregnated activated carbon materials, although the effect was assigned to catalytic activity of the metal nitrates rather than the metal oxides [29]. Our data do not suggest that the metal nitrates themselves (instead of the metal oxides) have a predominant impact on the combustion of the carbon matrix, since in most cases the metal nitrates are completely converted to the respective metal oxides beforehand.

Synthesis of SBA-15 Silica
Mesoporous SBA-15 silica (serving as the structure matrix for CMK-3 carbon) was synthesized by a standard procedure [30]. Pluronic P123 triblock copolymer (16.00 g; Sigma-Aldrich, St. Louis, MS, USA) was dissolved in deionized water (480 mL) and hydrochloric acid (48 mL, 37%, Stockmeier, Bielefeld, Germany) und stirred overnight at 35 °C. After the addition of 32 mL tetraethyl orthosilicate (99%, abcr, Karlsruhe, Germany), the solution was stirred for 24 h at the same temperature. The dispersion was transferred into a glass-lined autoclave and hydrothermally treated at 140 °C for 24 h. After cooling down to room temperature the solid product was filtered off and washed with deionized water. After drying at 120 °C overnight the powder was treated in a tube furnace at 550 °C with a heating rate of 2.5 °C min −1 for 6 h to remove the polymer.

Synthesis of CMK-3 Carbon
Mesoporous CMK-3 carbon was synthesized by a modified standard procedure [9]. One gram of SBA-15 silica was mixed in 6 mL of an aqueous solution of 1.25 g sucrose (99.5%; Sigma-Aldrich) sulfuric acid (3%). The silica was added in small portions and stirred until a homogenous mixture was obtained. The mixture was treated in a drying cabinet at 100 °C for 6 h and again at 160 °C for the same time. The solid was ground and the procedure was repeated but with a reduced amount of sucrose (1.00 g). Finally, the resulting material was heated to 900 °C in a tube furnace under N2 flow for 4 h. To remove the silica matrix the composite material was treated with hydrofluoric acid two times, washed intensively with water and ethanol and dried at 120 °C overnight.

Thermal Analysis
Thermal analysis (TG/MS) was carried out in a Al2O3 crucible (ca. 10-25 mg sample mass) under O2/argon flow (20/80 volume ratio; flow rate: 50 cm 3 ·min −1 ) with a Netzsch STA409PC thermobalance connected to a Balzers QMG421 quadrupole mass spectrometer by a custom-made coupling device. The heating rate was 10 °C min −1 . Onset and offset temperatures were determined from the combined DTG curves and MS signals; any tailing of the respective signals was disregarded when its contribution to the overall peak was below 10%.

Conclusions
In summary, the thermally induced conversion of metal nitrates into the respective metal oxides inside the pores a mesoporous CMK-3 carbon matrix and the subsequent thermal combustion of the carbon matrix has previously shown to provide a versatile means of preparing mesoporous metal oxides. The processes can be monitored by thermogravimetry (TG), coupled with mass ion detection (MS) of the decomposition products (water, nitrogen oxides, carbon oxides). The decomposition of the metal nitrates dispersed in the porous carbon matrix turns out to occur at lower temperature than under carbon-free conditions. In most cases the combustion of the carbon matrix sets in only after the metal oxides have formed. However, the carbon combustion always takes place at lower temperature than in case of metal nitrate-/metal oxide-free CMK-3 carbon. This can be explained by a catalytic effect of the metal oxides, which is particularly strong for oxides of redox-active metals.

Author Contributions
Jan Roggenbuck prepared the samples and analyzed experimental data. Jan Hanss carried out the thermoanalytical measurements. Christian Weinberger and Michael Tiemann further analyzed and interpreted the experimental data and wrote the manuscript.