Inorganic salt melts as porogen and solvent
Can we find solvents that withstand carbonization at high temperatures?
Melts of inorganic salts (SMs) and the mixtures thereof show a wide liquidus and by example allow for efficient mass transport as a so-called flux and are used in the production of high temperature ceramics. Presumably because the melting point of common salts is high compared to the reaction onset of organic polymerization/cross-linking reactions, SMs were rarely considered a potential reaction medium for carbonization [8–10]. Additionally, precursors should be stable and soluble in or miscible with the inorganic melt to guarantee homogeneous reaction conditions. In recent years, molten ZnCl2 was employed for the synthesis of covalent triazine-based organic frameworks . The crystallization of graphitic carbon nitrides was successfully obtained from a eutectic LiCl/KCl melt [12–14]. On top of that, some polymers were successfully dissolved in salt melts . Solubility of organic molecules is thus given at least in few cases. In fact, some eutectics show relatively low melting points (e.g., NaCl/ZnCl2 with T
= 230 °C), so that there is the chance to find carbonizable precursors, which are stable and soluble in molten salts at this temperature.
Very interesting precursors for such “ionothermal” syntheses are organic salts, as solubility in inorganic SMs is highly probable. Specifically, it is known that ionic liquids, i.e., low-melting organic salts, with an N-heterocyclic cation and the dicyanamide anion are carbonizable and produce nitrogen-doped carbon via thermolysis under inert atmosphere . Figure 2a shows the reaction scheme of carbonization of the pure ionic liquid 1-ethyl-3-methylimidazolium dicyanamide (Emim-dca). In Fig. 2b, the process of ionothermal carbonization is drafted. The archetype system comprises carbonization of the ionic liquid (herein Emim-dca) from solution in molten eutectic NaCl/ZnCl2. The eutectic inorganic salt is simply mixed with the ionic liquid under dry conditions, transferred to a ceramic crucible and calcined under inert atmosphere. The product is obtained as a powder after aqueous removal of the salt. Considering the recovery of the salt porogen, we have an efficient cyclic process .
The combination of properties of organic and inorganic salts indeed leads to porous nitrogen-doped carbons with very high specific surface areas of more than 2500 m2 g−1 throughout heat treatment of the simple mixture [17, 19]. The material comprises nitrogen atoms, which according to X-ray photoelectron spectroscopy (XPS) are a mixture of pyridinic, quaternary graphitic and oxidized species (Fig. 3a). The presence of the inorganic salt leads to efficient precursor conversion with yields almost double as high (~40 %) as the carbonization of the pure ionic liquid. Depending on the chosen ratio of ionic liquid and inorganic salt, microporous carbons with moderate pore volumes or fluffy carbon aerogels (Fig. 3b) with very high pore volumes (>3 cm3 g−1 based on nitrogen physisorption measurements) are obtained [17, 20]. The carbon morphology also strongly depends on the choice of the eutectic, which apparently can be attributed to the different miscibility and viscosity. These properties point to an applicable description of the process as a sol–gel process like shown in Fig. 1d. Obtained gels are desirable for multiple applications, as the hierarchical porosity—evaluated by N2-physisorption (Fig. 3c)—with concurrent presence of micro-, meso- and macropores, are advantageous in mass transport limited processes. High-resolution transmission electron microscopy imaging (HRTEM) illustrates the reason for the high specific surface areas as even the primary nanoparticles are composed of disordered (amorphous) graphene sheets and therefore comprise microporosity (Fig. 3d, inset).
The bottom-up approach allows for variation of the chemical composition of the materials. This way the nitrogen content can be tuned, but also iron- or cobalt-based nanoparticles can be embedded in one step by simple addition of the respective salts to the reaction mixture . The porogen may be removed by simple aqueous work-up, which is much easier as compared to the leaching of hard templates with sometimes hazardous chemicals. It is to mention that such nitrogen-doped carbons, because of the chemical composition, the high surface area and the hierarchical pore system show very promising performance as electrocatalysts for fuel cells [20, 21]. The applicability of such carbons is, however, not in the scope of this article, and the interested reader is referred to the original literature.
However, not only ionic liquids are suitable carbon precursors. Fischer et al.  extensively studied the solubility of cellulose in metal chloride hydrate melts. In fact, it was recently shown that cheap and abundant biomass can also be used for the synthesis of structurally very similar materials [22–26]. In other work, LiCl/KCl, cesium acetate or Na2CO3/K2CO3 melts were used as flux to obtain oligographene, carbon gels or active carbon [27–29].
Crystallization and self-assembly of carbon in SMs
Typically, carbon materials are obtained in solid-state reactions. Because carbonization/graphitization needs high temperatures even in the presence of catalysts, typically nonvolatile precursors are employed. Solid-state reaction, however, has the drawbacks of restricted mass transport and missing reorganizational ordering. Therefore, mostly disordered, but also heterogeneous products are obtained. This issue can be tackled with gas-phase carbonization of volatile precursors and not only fullerenes, nanotubes and graphene, but also amorphous, colloidal carbons can be prepared this way. Unfortunately, also gas-phase process have disadvantages such as low space–time yields, moderate heat transfer and temperature control. Due to the lack of any surface stabilizing agents, e.g., in spray pyrolysis spherical, non-porous materials (e.g., printing ink and conductive soot) are obtained . Also here the SMs are interesting reaction media as the structure formation can be expected to be different in a liquid phase, which might act as an interface stabilizing agent.
In chemical vapor deposition, special substrates are used to direct the growth of graphene or carbon nanotubes. Carbon atoms originating from decomposed precursors intermediately form solid or liquid solutions within the substrate, and carbon formation can be understood as a recrystallization from the substrates surface.
What happens if we have a solid–liquid interface instead of the solid–gas interface in a chemical vapor deposition? Investigations using the archetype system of ionothermal carbonization in the presence of a nickel foam substrate were carried out. A mixture of Emim-dca and ZnCl2 pasted on nickel foam and treated in inert atmosphere at 900 °C gives interesting results (Fig. 4a) . For comparison, blind experiments were carried out: (1) without nickel foam and (2) without ZnCl2. The product without Ni foam resembles the previously described carbon aerogels (Fig. 4b), while the pyrolysis of the pure ionic liquid on nickel foam results in a “forest” of carbon nanotubes growing from the nickel surface (Fig. 4c). In the presence of ZnCl2, however, instead of nanotubes, vertically aligned carbon nanosheets are obtained (Fig. 4d).
Like in the case of growth of carbon nanotubes from the solid–gas interface, the nickel substrate mediates directed growth from the surface inside the melt. The presence of the molten salt porogen, however, allows for the formation of an ordered superstructure. It turns out that it is worth to study the carbon formation inside a liquid medium, as such carbon architectures, strongly bound to the Ni surface, show interesting electrochemical properties .
Syntheses using a substrate as a catalyst for structure formation are limited by the number of spatial degrees of freedom, which can explain the formation of (oligo)graphenes and carbon nanotubes in classical vapor deposition. A wet-chemical carbonization/graphitization could solve aforementioned problems and gives rise to interesting “new” carbon chemistry. Most of the other nanomaterials, such as metal or metal chalcogenide nanoparticles, are typically obtained in precipitations, crystallizations or self-assembly from solution. One particular interesting method to obtain nanocrystals is the so-called hot-injection technique . Here precursors with low thermal stability are decomposed in a controlled and sometimes very selective fashion to enforce precipitation/crystallization. In analogy polymer, particles with desired morphology are prepared in carbon chemistry and carbonized afterward. The structure of the resulting carbon particles, however, cannot be directed, and it remains a solid-state carbonization.
But what is happening if thermal instable organic precursors are injected into hot salt melts?
If classic organic solvents like ethanol, acetonitrile or glycol are added dropwise into molten ZnCl2 at 550 °C under inert atmosphere, indeed carbon materials are obtained with sometimes surprising high yields [33, 34]. A schematic view of the synthesis is shown in Fig. 5. The composition of the obtained carbon materials can be varied by the choice of the precursor solvent. Acetonitrile, benzonitrile and pyridine after simple aqueous work-up are giving nitrogen-doped carbon, whereas DMSO after alkaline work-up leads to sulfur-doped carbon. The materials partly show very high specific surface areas and pore volumes (up to 1666 m2 g−1 and 2.8 cm3 g−1). In particular interesting are, however, the obtained morphologies (Fig. 5).
SEM images show that depending on the choice of the precursor solvent, three different morphologies can be obtained. First of all, we can observer the “typical” gel-like agglomerated spherical nanoparticles (Fig. 5a), which originate from solvents such as acetonitrile and benzonitrile. The second structure is composed of extended, porous layers with a thickness of ~100 nm, which, e.g., are obtained from injection of ethylene glycol (Fig. 5b). The third and most interesting structures are branched carbon nanofibers, which degree of branching and aspect ratio depends on the choice of the precursor but even more on the choice of salt melt (Fig. 5c). The interesting morphology reminds on inorganic materials, which were obtained by vectorial alignment of primary particles, i.e., by self-assembly or oriented crystallization. High-resolution TEM imaging indicated in fact that the fibers evolve from “clustering” primary sheet-like carbon nanoparticles; however, more detailed studies are necessary (Fig. 6).