3D-Printed Porous Magnetic Carbon Materials Derived from Metal–Organic Frameworks

Here we report new porous carbon materials obtained by 3D printing from photopolymer compositions with zinc- and nickel-based metal–organic frameworks, ZIF-8 and Ni-BTC, followed by high-temperature pyrolysis. The pyrolyzed materials that retain the shapes of complex objects contain pores, which were produced by boiling zinc and magnetic nickel particles. The two thus provided functionalities—large specific surface area and ferromagnetism—that pave the way towards creating heterogenous catalysts that can be easily removed from reaction mixtures in industrial catalytic processes.


Introduction
Today, 3D printing, an additive manufacturing (AM) technique, has gone far beyond prototyping of industrial products [1]. It is now used to transform digital models into real-world objects for applications in catalysis [2], medicine [3], gas adsorption and storage [4,5], etc. [6] by layer-by-layer deposition of a polymer [7]. Of the many 3D printing processes [8], the most popular are polymer extrusion (fused deposition modeling, FDM, or directly ink writing, DIW) [9,10] and vat polymerization (stereolithography, SLA, or digital light processing, DLP) [11]. They can both produce functionalized objects from composite materials [12] containing the polymer matrix with a filler that provides the needed functionality. Inorganic nanoparticles may be added to increase the catalytic activity of the 3D-printed objects [13], such as zeolites and metal-organic frameworks, to increase their adsorption characteristics [14], and graphene, to improve their electrical conductivity [15].
Such a simple approach to creating active objects, however, suffers from a blocking of the filler by the polymer matrix that prevents it from performing its functions [16]. Possible ways of overcoming this drawback include functionalization of the objects after the 3D printing process [17] or heat dissolution of the polymer matrix to remove the binder and thereby obtain 3D-printed objects [18] with new, emergent properties.
Of particular interest are porous carbon materials obtained by their pyrolysis [33]. The MOF-based carbon materials with distributed nanoparticles of metal, metal carbide or metal oxide [34,35] feature high thermal, chemical and mechanical stability as nanotubes (CNTs) [36,37], nanowires [38], etc. Together with a large specific surface area and an adjustable pore structure [39] that allows encapsulating various compounds [40], they are finding use in catalysis [41], gas storage [42], etc. [43,44]. Nevertheless, the pyrolyzed MOFs are rather brittle, so they are very difficult to mold [45].
Here, we report a porous MOF-based carbon material doped with nickel particles that has a complex geometry obtained by stereolithography (SLA) 3D printing from a photopolymer composition containing two popular MOFs, Ni-BTC [46] and ZIF-8 [47], as functionalizing fillers and so the material can be potentially applied as a nickel-based catalyst [48].
ZIF-8: A solution of Zn(NO 3 ) 2 •6H 2 O (2.93 g, 9.87 mmol) in 200 mL of methanol was quickly added to a solution of 2-methylimidazole (6.489 g, 79.04 mmol) in 200 mL of methanol. The reaction mixture was stirred at room temperature for 1.5 h, and the resulting suspension was centrifuged at 6000 rpm for 5 min. The precipitate was washed with DMF and three times with methanol to exclude residues of 2-methylimidazole. The obtained crystalline product was dried under vacuum. Yield: 0.435 g (19.37%). Calculated for C 48  Ni-BTC: Ni(OAc) 2 •2H 2 O (3 g, 5.16 mmol) was dissolved in 100 mL solution of water, ethanol and DMF (1:1:1) at room temperature. A solution of trimesic acid (1.08 g, 5.16 mmol) in 100 mL of the same solvent mixture was added. The reaction mixture was stirred at room temperature for 2 h, and the resulting suspension was filtered, washed three times with DMF and three times with methylene chloride and was then immersed in methylene chloride. The solvent was decanted and replaced once per day over the next three days. The obtained crystalline product was dried under vacuum. Yield: 1.74 g (41.8%). Calculated for C 18  Photocurable resin preparation. 2-Phenoxyethyl acrylate (2.35 g) and trimethylolpropane triacrylate (2.35 g) were mixed with 2.4 wt.% of bis(2,4,6-trimethylbenzoyl)-phenylphosphineoxide and 4.8 wt.% of 1-hydroxy-cyclohexyl-phenyl-ketone in a polypropylene cup.
ZIF-8@polymer: ZIF-8 (5 wt.%) was incrementally added to the resulting resin until the desired loading level was reached, and the mixture was sonicated with a tip-sonicator (UZD2-0.1/22, Russia) for 5-10 min and at 50% amplitude to produce a homogeneous dispersion.
3D printing. Stereolithography was performed with a DLP/LCD/SLA 3D printer DUPLICATOR 7 PLUS (Wanhao, China) equipped with a reduced-size working bath. Creation Workshop software (1.0.0.75) was used for slicing and creating G-code files. The layer thickness was set to 100 µm, and the exposure time of the base layer was 5 s and of five initial layers, 100 s.
Pyrolysis. 3D-printed objects were washed with isopropyl alcohol and then pyrolyzed in a tubular furnace (PT-1200, Rosuniversal, Russia) under flow of H 2 and Ar (7.03 vol.% of H 2 ) of 100 mL/min at temperature of 950 • C that was reached from room temperature with a ramping rate of 3 • C/min. 3D-printed objects were calcined for 5 min at 950 • C. After cooling to room temperature, carbon residues were collected.
Scanning Electron Microscopy (SEM). SEM images for samples placed on a 25 mm aluminum stage and fixed with a conductive carbon tape were obtained in the secondary electron mode at an accelerating voltage of 15 kV and low vacuum mode with a Hitachi TM4000Plus benchtop electron microscope (Hitachi High-Technologies Corporation, Tokyo, Japan) equipped with an energy dispersive X-ray detector QUANTAX 75 (Bruker Nano GmbH, Berlin, Germany). N 2 and CO 2 adsorption studies. Adsorption-desorption of CO 2 was measured at 273 K and of N 2 , at 77 K on a Surface Area and Pore Size Analyzer System 3P Micro 200 (3P Instruments GmbH & Co. KG, Odelzhausen, Germany). Before the measurements, the samples were degassed at 473 K for 6 h under vacuum. In CO 2 adsorption-desorption experiments, micropore specific volumes and specific surface areas were calculated using the methods of non-local density functional theory (NLDFT), Dubinin-Radushkevich (DR) and grand canonical Monte Carlo (GCMC) [49] using the adsorbed CO 2 density ρ ads of 1.044 g/mL and CO 2 cross-sectional area A of 0.21 nm 2 ; affinity coefficient β was taken as 0.35 [50]. Pore volume-size distributions were obtained by the NLDFT method. In N 2 adsorption-desorption experiments, the calculations were performed using BET, Langmuir and NLDFT methods. For higher precision, BET equation was applied to the isotherms according to Rouquerol criteria [51]. Pore volume-size distributions were obtained by the NLDFT and BJH methods.
Magnetic measurements. Magnetic hysteresis curves of pyrolyzed samples were obtained with a Quantum Design PPMS-9 device by scanning magnetic field between 50,000 and −50,000 kOe at room temperature for samples sealed inside a polyethylene capsule.

Results
Pyrolysis of MOFs is known to produce highly porous carbon materials with uniformly distributed metal particles [52]. If MOFs are incorporated into a photopolymer composition [19,29], such materials can be obtained in a variety of complex geometries for use in catalysis [13]. To increase their porosity, however, the use of a larger amount of a MOF is not an option, as its content dramatically affects the processibility of the photopolymer composition. The solution for this purpose is to choose a MOF with metal nodes made of ions of the metal that evaporates at temperatures of the pyrolysis, such as ZIF-8, {Zn(mim) 2 } n (mim = 2-methylimidazolate) [53]. 2-Methylimidazolate linkers carbonize during the pyrolysis, while the zinc eventually boil off, leaving pores and channels unoccupied and thereby increasing the surface area of the pyrolyzed carbon material (Scheme 1).
To test this hypothesis, we used a custom-made photopolymer composition containing 5 wt.% of ZIF-8 to 3D-print a complex object by stereolithography (SLA) process (Scheme 1) with high spatial resolution (Figure 1a). After the pyrolysis at the highest available temperature of 950 • C in a reducing environment of Ar and H 2 , the 3D-printed object lost up to 89% of mass and shrank to half of its size (the wall thickness increased from 0.7 mm to 1-1.2 mm). Although it also became significantly more fragile, its complex geometry remained nearly unchanged (Figure 1b).  Scanning electron microscopy (SEM) images showed many pores on the surface that appeared after the pyrolysis, as well as traces left by boiling zinc (Figure 2). The metal fully evaporated from the object, as follows from its lack in the elemental composition evaluated by energy dispersive X-ray (EDX) spectrometry. As a result, the object became microporous with an average pore diameter of 0.524 nm estimated by porosimetry based on CO 2 and N 2 adsorption. The calculated surface area for CO 2 adsorption was 1352 m 2 /g, but for N 2 adsorption, it was only 47 m 2 /g ( Figure S1 of Supplementary Materials). Although the adsorption of these gases by the pyrolyzed material is not as high as by the pure ZIF-8 [47], it is still quite high given the small amount of MOF in the photopolymer composition.
As zinc completely boiled off in the pyrolysis at 950 • C, another MOF was introduced into the photopolymer composition to obtain a porous carbon material with added functionality, such as for applications in catalysis. A nickel-based MOF, Ni-BTC ({Ni 3 (BTC) 2 •12H 2 O} n , BTC = 1,3,5-benzene tricarboxylate) [46], was chosen for this purpose, as metal nickel is a popular catalyst in a variety of processes [54]. An object of the same complex geometry ( Figure S2 of Supplementary Materials) was 3D-printed from the photopolymer composition that contained 5 wt.% of ZIF-8 and 5 wt.% of Ni-BTC. Its pyrolysis in the flow of Ar and H 2 at 950 • C did not compromise the geometry of the object, although the loss of mass (94%) and changes in the linear dimensions were higher than for the object 3D-printed with no Ni-BTC, probably owing to the graphitization process boosted by the nickel particles [55].
Holes and traces left by boiling zinc on the surface of the pyrolyzed object were clearly seen on SEM images (Figure 3). Elemental analysis with EDX spectrometry also supported the lack of zinc and a significant amount (up to 50% of the total mass of the object) of nickel that did not boil off from the surface upon the pyrolysis at 950 • C. The latter resulted in the microporosity. Indeed, the calculated surface area for CO 2 adsorption was 859 m 2 /g, while for N 2 adsorption it was only 145 m 2 /g ( Figure S1 of Supplementary Materials). An average pore volume (0.286 cm 3 /g) was lower than observed for the photopolymer composition with ZIF-8 only, thus agreeing with the proposed graphitization process.
As metal nickel is a well-known ferromagnet, we also studied the magnetic behavior of the pyrolyzed object by dc-magnetometry. Narrow magnetic hysteresis loops were observed at room temperature ( Figure 4). The obtained value of magnetic coercivity (36 Oe) fell into the range expected for metal nickel [56] and was enough for a magnetic separation of the pyrolyzed object from non-magnetic substances. Indeed, it was attracted to a small permanent samarium-cobalt magnet. Note that the object kept the constant mass after numerous exposures to the magnetic field, so the nickel particles were held tightly in the carbon matrix.  To confirm the ability of the 3D-printed objects to retain their complex geometries upon the pyrolysis, another object made of a repeating series of gyroids was 3D-printed from a commercial photopolymer resin Harz Labs filled with 5 wt.% of ZIF-8 and 5 wt.% of Ni-BTC ( Figure S1 of Supplementary Materials). This object was pyrolyzed under the same conditions to lose up to 90% of mass and shrunk to a half of its size while keeping its initial geometry. SEM images of its surface featured holes and traces from boiling zinc ( Figure 5) that were not among the elements identified by EDX spectrometry. In contrast, 50% of the surface composition was nickel. The pyrolyzed object was also microporous, with the surface areas for CO 2 adsorption of 519 m 2 /g and for N 2 adsorption of 5 m 2 /g only ( Figure S1 of Supplementary Materials).

Conclusions
By combining two MOFs, ZIF-8 and Ni-BTC, with the custom-made and commercial photopolymer compositions, the new porous carbon materials containing metal particles for the added functionality were obtained by SLA 3D printing followed by the hightemperature pyrolysis. The resulting materials that retained the shape of complex objects had a large specific surface area that is sought in adsorption or catalysis. They also featured ferromagnetism provided by the nickel particles, which may greatly help to separate catalysts from initial compounds and their products in industrial catalytic processes. The reported approach based on a 3D printing technique available to many researchers around the world offers new possibilities for creating functional materials and objects of a desired shape and geometry for various applications.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/polym13223881/s1, Figure S1: CO 2 adsorption isotherms at 273 K (top), N 2 adsorption isotherms at 77 K (center) and a plot of specific surface area vs. pore size according to NLDFT from CO 2 adsorption measurements for the pyrolyzed objects 3D-printed from the custom-made photopolymer composition filled with 5 wt.% of ZIF-8 (blue circles) and filled with 5 wt.% of ZIF-8 and 5 wt.% of Ni-BTC (red triangles) and from the commercial Harz Labs resin filled with 5 wt.% of Ni-BTC and 5 wt.% of ZIF-8 (grey squares), Figure S2: Objects 3D-printed from the custom-made photopolymer composition filled with 5 wt.% of Ni-BTC and 5 wt.% of ZIF-8 (a,b) and from the commercial Harz Labs resin filled with 5 wt.% of Ni-BTC and 5 wt.% of ZIF-8 (c,d) before (a,c) and after (b,d) the pyrolysis.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare that they have no conflict of interest.