Developing Benign Ni/g-C3N4 Catalysts for CO2 Hydrogenation: Activity and Toxicity Study.

This research discusses the CO2 valorization via hydrogenation over the non-noble metal clusters of Ni and Cu supported on graphitic carbon nitride (g-C3N4). The Ni and Cu catalysts were characterized by conventional techniques including XRD, AFM, ATR, Raman imaging, and TPR and were tested via the hydrogenation of CO2 at 1 bar. The transition-metal-based catalyst designed with atom-economy principles presents stable activity and good conversions for the studied processes. At 1 bar, the rise in operating temperature during CO2 hydrogenation increases the CO2 conversion and the selectivity for CO and decreases the selectivity for methanol on Cu/CN catalysts. For the Ni/CN catalyst, the selectivity to light hydrocarbons, such as CH4, also increased with rising temperature. At 623 K, the conversion attained ca. 20%, with CH4 being the primary product of the reaction (CH4 yield >80%). Above 700 K, the Ni/CN activity increases, reaching almost equilibrium values, although the Ni loading in Ni/CN is lower by more than 90% compared to the reference NiREF catalyst. The presented data offer a better understanding of the effect of the transition metals' small metal cluster and their coordination and stabilization within g-C3N4, contributing to the rational hybrid catalyst design with a less-toxic impact on the environment and health. Bare g-C3N4 is shown as a good support candidate for atom-economy-designed catalysts for hydrogenation application. In addition, cytotoxicity to the keratinocyte human HaCaT cell line revealed that low concentrations of catalysts particles (to 6.25 μg mL-1) did not cause degenerative changes.

Introduction

In recent years, technologies that facilitate the creation of clean energy from renewable and sustainable resources have attracted increased attention.[1−3] Much research and pilot work is dedicated to developing efficient, cost-effective synthetic fuel/biofuel production methods in a carbon-neutral or carbon-negative manner.[3−9] In search of potential starting candidates for future chemical building blocks, CO2, one of the greenhouse gases (GHGs), has captured industrial interest because it is an abundant, inexhaustible, and inexpensive carbon feedstock.[2,10−13] At the same time, its hydrogenation to energy-rich products can offer diverse hydrocarbon compounds such as methane, methanol, formic acid, formaldehyde, and C1–C2 ethers.[7−9,12,14,15] Because of the variety of possible applications, including the rubber industry, pharmacy, agriculture, and food technology, they are considered to be highly attractive products. They can also be directly employed as a hydrogen-rich source/energy vector for fuel cells.[8,15−17]

CO2 hydrogenation is a highly exothermic process, where the reaction equilibrium is significantly influenced by pressure and temperature.[8,14,16,18] The hydrogenation reaction research concentrates mainly on C1 or short-chain products CO, HCOOH, CH3OH, CH4, and C2–C4 olefins.[12,19] For that, two kinds of mechanisms have been proposed. The first one involves CO2-to-CO conversion via a reverse water–gas shift (RWGS) reaction and subsequently CO methanation according to eqs and 2, respectively.[2,14]

The second proposed mechanism considers the direct hydrogenation of CO2 (eq ) via intermediate species formation, i.e., formate species.[18]

Materials development plays an essential role in extensive technological innovation for clean energy conversion and storage. Mostly used in the hydrogenation process, transition-metal-based catalysts are loaded with active phases and are considered to be toxic materials, with dedicated disposal protocols.[20] Considering the vast amounts of spent catalysts worldwide, especially fluid catalytic cracking (FCC) catalysts, an alternative is needed to prevent extensive traditional landfill disposal, reduce environmental contamination, and reduce hazardous solid wastes.[21] Current research activities focus on a new catalyst preparation methodology that applies solvent-free protocols ideally, atom economy principles, new catalyst compositions, catalyst stability, and durability. Both noble-metal-based catalysts, i.e., Pt, Rh, Ru, and Ir, and transition-metal monometallic catalysts, i.e., Ni and Ni–V bimetallic systems, were found to be effective for the CO2 valorization process.[8,14,17,22−25] However, non-noble-metal small clusters or single-atom catalysts (SACs) for the hydrogenation of CO2 to gas or liquid products, including fuels/drop-in fuels, remain rare.[5,26,27] For CO2 methanation reactions, Ni-based catalysts (usually up to 20–30% Ni content) are usually applied in real-scale industrial lines because of their good catalytic performance and for economic reasons.[14] However, their application potential is limited, both at high and low temperatures, because of catalyst coking and Ni nanoparticle (NPs) sintering and poor activity, respectively.[14] Recently, vanadium-modified Ni 2D nanocatalysts have been reported, delivering exceptional conversions for low-temperature hydrogenation.[8] Moreover, the NiV catalyst, obtained from a hydrotalcite precursor, outperformed the best-known catalyst and presenting at the same time high durability and selectivity (the equilibrium conversion occurred at 623 K, and the primary product of the reaction was CH4 (>97% CH4 yield)). Copper is another transition metal considered for hydrogenation reactions both as (1) a Cu-based catalyst with a high specific surface area and semiconductor properties (usually up to 60–70% Cu content), i.e., Cu–ZrO2 and Cu/ZnO/ZrO2 for the CO2-to-methanol reaction, and (2) Cu complexes, i.e., copper(I) complex LCu(MeCN)PF6 for CO2 hydrogenation to formate/formic acid in the presence of DBU as the base.[28,29] Applying higher pressure, usually 5–40 atm, supported Cu-based catalysts being used in various industrially relevant hydrogenation processes, i.e., methanol synthesis, the low-temperature water–gas shift, and various organic compounds’ selective hydrogenation or fine chemical synthesis.[30] Here, the metal/oxide interface, the synergy between Cu0 sites and −OH groups, and basic Lewis sites shape the overall catalyst activity and selectivity.[31,32] For example, in the case of Cu/ZnO/ZrO2 catalysts, hydrogen adsorption and dissociation are carried out by Cu sites. The basic sites of ZnO and ZrO2 are responsible for CO2 adsorption (as a mixture of carbonate and bicarbonate species). Because both sites are in close proximity to each other, the reaction is facilitated. The atomic hydrogen spills over from the Cu0 surface to the surface of basic sites, hydrogenating the adsorbed carbonate species to formate, methoxide species, and methanol.[33]

For the efficient adsorption and activation of CO2, various carriers are investigated that exhibit the desired structural properties due to oxygen vacancies and reversible valence shifts. Carbon materials, including doped carbons, graphene oxide (GO), reduced graphene oxide (rGO), and carbon nanostructures (i.e., carbon nanofibers CNFs), are also studied because of their exceptional morphological and chemical properties, adjustable surface chemistry, high thermostability, electrical conductivity, and high adsorption of hydrogen.[3,34−37] Among those materials, graphitic carbon nitride (g-C3N4) possesses unique chemical properties and great mechanical and chemical stability. g-C3N4 sites that can serve as Brønsted acids (N–H sites) and Lewis bases (N lone electron pairs) are present, which can be further modified by alkanization or protonation.[15,38] Besides, within this material, the catalyst nanoparticles can be anchored, and sulfur, boron, or metal particles (e.g., Fe, Cu, and Ni) can be doped to enhance the electrical conductivity as well as the photo- and electrocatalytic activity[15,38] Pristine g-C3N4 is a semiconductor with a 2.5–2.8 eV band gap. Structurally, g-C3N4 is similar to some degree to the N-substituted graphite framework with a very high level of nitrogen doping. It comprises π-conjugated graphitic planes formed from sp2-hybridized carbon and nitrogen atoms.[13,15,35,37,38] Pristine or doped g-C3N4 is tested in various applications in hydrogen evolution, the oxidation of C–H, C–C, N–H bonds, the degradation of organic pollutants, sensing, bioimaging, and gas storage.[15,38]

In this work, graphitic carbon nitride (g-C3N4) has been used to prepare a transition-metal-based catalyst (Ni, Cu, and Cu–Ni) with improved stability in CO2 hydrogenation. The hypothesis is that, similar to other carbon materials, g-C3N4 offers a variety of sites type and can promote the reaction by stabilizing metal clusters/nanoparticles and providing good CO2 adsorption and H2 storage. Moreover, in the current work, the Cu influence on the catalysts’ selectivity and durability in CO2 hydrogenation is shown. The catalysts’ performances were compared with those of commercially available best-performing reference material (NiREF).[39,40] Their toxicities were evaluated using an in vitro study of keratinocyte cell lines from histopathologically normal adult human skin. Because of numerous reports on the cytotoxicity of Cu, Ni catalysts, and their oxides, we decided to estimate the influence of these particles on human skin cells (HaCaT line). The morphological changes were determined microscopically, and metabolic disorders of cells were estimated in the MTS assay over a wide range of concentrations (3.125–500 μg mL–1). This allowed for assessing the biological safety of the use of synthesized catalysts particles.

Results and Discussion

Physicochemical Characterization

The physicochemical data that apply the procedures described in the SI are presented in Table and Figure . All homemade g-C3N4 samples were characterized by a roughly constant SBET of ±20 m2g–1, while the SBET of NiREF industrial full-grain samples was very low 2 m2 g–1 (specific for α-Al2O3).

Table 1

Physicochemical Properties Data Resume of Studied Catalysts

catalyst	description	Ni or/and Cu content, %	SBET, m2/gcat	H2 consumption T, K	DpNiO, nma
Ni-V/α-Al2O3	NiREF	13.2	2.0	603, 673, >750	<12
g-C3N4	CN	0	179
Cu/g-C3N4	Cu/CN	4	157	650, >730	<20
Cu-Ni/g-C3N4	Cu-Ni/CN	4	161	>730	<4/130
Ni/g-C3N4	Ni/CN	4	167	>730	<1

Obtained from a XRD study.[38]

Figure 1

(A) XRD patterns of the calcined samples with (B) the assignment of the corresponding XRD peaks (based on SI(15)). (C) HRTEM and (D) ATR spectra of CN-supported catalysts.

The XRD patterns of the catalysts displayed in Figure are dominated by a sharp peak at 2θ = 27.4°, characteristic of graphitic materials and ascribed to π-conjugated layers stacked with index (0 0 2).[37,41] Consistent with previous studies, the interlayer spacing of the stacked nanosheets is 0.32–0.35 nm, and the size range is 1.1–1.5 nm.[38,42,43] Moreover, in all prepared hybrid catalysts, fine-scale nanostructures have been evidenced, as all corresponding patterns exhibit the broad features in the low 2θ region. In Figure B, the peaks assignment of the respective CuO, NiO, and CuNi(1– phases is summarized. Clear signals from the transition-metal-containing phase were observed for Cu/CN and Cu–Ni/CN catalysts, suggesting some extent of segregation and agglomeration of CuO and NiO particles, which was also confirmed by HRTEM Figure C. For Ni/CN, the data suggested a fine distribution of Ni/NiO over the CN support. In general, those phases were beyond the XRD detection limit. The obtained particles sizes are shown in Table . In all studied materials, 2D sheetlike structure with wrinkles characteristic of pure CN (Figure C) was observed via HRTEM. This structure was unchanged for the samples modified with Ni, Cu, or both incorporated metals (Figure C), in agreement with the literature.[13,15,38]

For the CN support (Figure D) and all investigated hybrid catalysts, five main regions were identified as being indicative of (i) −N–H stretching (3000–3200 cm–1), (ii) terminal C–N bonds (2100–2700 cm–1), (iii) aromatic conjugated C–N stretching (1400–1200 cm–1), and (iv) C=N stretching vibrations (1650–1540 cm–1) and CN heterocycle vibrations of the triazine ring (800–700 cm–1).[15,44,45] The broad bands at 3000–3500 cm–1, most pronounced for the Cu–Ni/CN catalyst and assignable to the OH stretching mode ν(OH)−, were due to physisorbed water on the catalyst surface.[33,36,46]

Figure A shows the optimized structure of g-C3N4. The heptazine (tri-s-triazine) units are highlighted in pink in Figure A. They are linked by N–H groups on their edges and bridging −NH– groups to form condensed heptazine rings. In bare CN, the interplanar pores and the 6-fold cavity size stay close to 3.2 and ca. 7.3 Å, respectively.[7,28,29] The dimension of Ni NPs was shown to be <1 nm, indicating possible Ni insertion and NP stabilization within the CN cavity, as was suggested before.[38] A similar effect was observed for Pd and Pt atoms deposited on g-C3N4.[47] The 6-fold cavity of g-C3N4 was estimated by DFT calculation to stabilize Pd and Pt atoms, and the binding energies were given as −2.17 and −2.95 eV, respectively. Furthermore, it was shown and evidenced by STM microscopy that, very similar in chemical structure to CN, the porphyrin cavity (i.e., comprising N, C and H), although smaller in dimensions compared to the CN cavity, was able to stabilize Ni and Co.[30] However, within the CN cavity, atoms such as V, Cr, Mn, and Fe were shown to be more stable than their bulk phases.[7] At the same time, preferential segregation and metal cluster formation for Co, Ni, Cu, and Zn were reported.[7] This tendency has also been evidenced in this work by an AFM microscopy study for fresh and after-reaction catalyst samples. The AFM topography images of the graphitic nitride support deposited on the HOPG surface are displayed in Figure B for fresh samples and in Figure B′ for after-reaction samples (samples after catalytic study in the hydrogenation reaction). Graphitic nitride islands of hundreds of nanometers are evidently distinguished in the presented images. They are characterized by relatively smooth surfaces with characteristic linear steps suggesting a layer structure (with a layer thickness in the range of 5–10 nm). These typical elements are also visible for all studied samples containing Cu, Ni, and both transition metals (Figure C–E for Ni/CN, Cu/CN, and Cu–Ni/CN, respectively). Moreover, the investigated layers consist of another element with more undefined (cloudlike) character, which can be related to the amorphous carbon residue. The amount of this phase increases (is more visible) in the samples after reaction (Figure C′–E′ for Ni/CN, Cu/CN, and Cu–Ni/CN, respectively). Moreover, for the Cu–Ni/CN catalyst, segregated Cu and Cu–Ni phases in the form of nanotubes/nanowires were evidenced at sizes of hundreds of nanometers (Figure F,G).

Figure 2

(A) Optimized structural elements of graphitic carbon nitride: heptazine (tri-s-triazine) units (marked) linked by bridging −NH– groups and N–H groups on their edges forming condensed heptazine rings. (B−G) AFM topography images of CN before (B) and after reaction (B′), Ni/CN before (C) and after reaction (C′), Cu/CN before (D) and after reaction (D′), and Cu−Ni/CN before (E−G) and after reaction (E′). The films were deposited on HOPG, images were acquired in air at room temperature.

The chemical and structural changes in the CN sample were evaluated by Raman imaging spectroscopy, and the results are presented in Figure . The typical bands for sp2 carbon materials were assigned to G (for graphite) and D (for disordered) at ca. 1570 and 1370 cm–1, respectively.[5,15,38] Using the ID/IG peak intensity ratio, the level of disorder in the bare CN and hybrid material can be characterized as described previously.[15,16,38] Moreover, from the I750/I705 ratio (or the I543/I479 ratio), the degree of exfoliation (extent of exfoliation) can be obtained.[38] In all studied catalysts, the I750/I705 ratio indicated the presence of bulk g-C3N4; i.e., in the case of pure CN and hybrid Ni/CN, the I750/I705 ratios were ca. 0.2–0.5 but they increased to 0.4–0.5 for Cu/CN and Cu–Ni/CN samples.

Figure 3

Raman images of catalyst nanoparticles with a 50 μm pinhole aperture and 50× magnification. In the ROI region, distributions of 705 and 750 cm–1 for the D band (1347 cm–1) and G band (1580 cm–1) were visualized with intensity ratios of extent of exfoliation (I705/I750) and level of disorder (I1347/I1580). Intensity ratios are presented with 3D images on the right side of the panel.

The structural morphology with the surface samples (Figure ) and its changes for after-reaction samples were studied by detailed Raman mapping. This allows for visualizing any chemical changes in the studied material. High scattering for bands located at 705 and 750 cm–1 was evidenced for all samples, which are indicative of π-conjugated graphitic planes in g-C3N4, specifically for the overall layer numbers, layer-stacking configurations, and interlayer coupling.[15] As shown in Figure , the intensity of the band at 705 and 750 cm–1 was the same within the whole sample (described as the ratio of bulk coplanar g-h-heptazine-C3N4), confirming the homogeneous distribution of investigated surface species within gCN material. The derived maps (Figure ), acquired with the pinhole aperture, enabled the surface samples to be characterized at a specific depth of about ca. 2 μm. These maps (Figure ) demonstrate periodic spots, and the uniform distribution of bands confirms the interval variation of the intensity of each band and thus the homogeneity of the samples throughout the region as well as the lack of significant changes in morphology for the postreaction samples. This is consistent with the sample morphology results obtained by HRTEM and AFM analysis.

The temperature-programmed reduction (TPR) profiles for hybrid catalysts and reference samples are presented in Figure . The reducibility of the catalysts was examined in the temperature range of 298–773 K at a ramp rate of 10 K min–1. In Figure , the H2 consumptions as a function of temperature are presented. The TPR profiles demonstrated considerable differences between the CN-supported catalyst samples and the reference catalyst described elsewhere.[8,16] It can be noticed that the catalyst support determines the TPR profiles in general, particularly the onset of the reduction temperature and the amount of hydrogen consumed. For the NiREF reference catalyst, in the TPR profile, peaks with maxima at 600, 673, and above 773 K are visible. They can be linked to the reduction of selected phases, namely, (i) NiO bulk and small NiO crystallites at ca. 600 K, (ii) noncrystalline NiO species at ca. 673 K, and (iii) NiAl2O4 above 773 K.[10,11,39] It is recognized that the reducibility of the sample is strongly governed by metal–support interactions.[3] Free NiO that interacts weakly with the support is usually readily reducible at lower temperatures (α-type species; low reduction temperature of 573–623 K), and the nickel species that are not fully attached to the spinel are reducible at intermediate temperatures (β-type species; intermediate temperature, 623–773 K). Nickel aluminate spinels (γ-type Ni species) undergo reduction at high temperatures, i.e., above 973 K.[16]

Figure 4

TPR profiles of the catalysts (5% H2 in He).

Figure 5

(A) CO2 conversion. (B) MS profiles of CH4 and CO. (C) Catalyst activity measured as CO2 conversion at 600 K with (C′) the product distribution for Cu/CN and Cu–Ni/CN. (D) Concentrations as a function of temperature and for Ni/CN- and NiREF-supported catalysts (NiREF performance given for comparison). Reaction conditions GHSV = 1.2 × 104 h–1, 1 atm, 293 K, and CO2/H2 in a 1/5 molar ratio.

The observed reduction temperature for CN-supported Ni (Figure ) differs from that reported for NiREF. This discrepancy can most probably be linked to the catalyst composition, namely, the metal–support interaction and the presence of small Ni particles <1 nm.[14,16] The peak of H2 consumption approaches maxima above 773 K. This is in good agreement with the literature, where the NiO bulk and noncrystalline NiO reduction are typically observed at temperatures of ∼650 and ∼750 K, respectively.[23,48,49] For Cu-containing samples, the peak maxima related to H2 consumption are less intense. For Cu/CN, maxima at 625 and ∼750 K can be distinguished, while for Cu–Ni/CN, the broad signal at ∼750 K occurred again. This can be connected to noncrystalline CuO and Cu(Ni(1– binary phase reduction in Cu/CN and Cu–Ni/CN, respectively. The evidenced slight shift of the reduction peak for Cu/CN can be indicative of a weakening NiO–support interaction, which improves the Ni reducibility.[50]

Usually, one broad band above 523 K is observed for a Cu-containing carbon-supported catalyst, indicating CuO to Cu reduction. Typically, this band can be deconvolved into two α and β peaks, attributable to the highly dispersed CuO species reduction at a lower temperature (α-peak) and the bulk CuO species reduction at a higher temperature (β peak), respectively.[51] Apparently, for the Cu/CN sample, studied, the percentage of the highly dispersed CuO species is significantly lower than that of the bulk species, which is consistent with microscopic and XRD studies.

Catalyst Activity

The catalytic activity results for the CN-supported catalysts and NIREF samples reduced at 773 K are given in Figure . The calculated equilibrium data for the CO2 methanation reaction for NiREF are displayed together with the MS profiles of the reactor exit gases. Before hydrogenation, all catalyst samples were activated by in situ NiO-to-Ni reduction in a stream of H2. An identical procedure was used for Cu/CN for comparative purposes. Online MS was used for the continuous monitoring of relevant C species, such as CH4 (m/z = 16), CO (m/z = 28), CH2O (m/z = 30), CO2 (m/z = 44), and H2O (m/z = 18). For the studied process, the primary products were selected on the basis of the reaction outlined in eqs –3. These reactions were taken into account owing to their higher relevance resulting from the enthalpy values.[8,16] For Ni/CN, CO2 conversion is initiated at 500–520 K, while in Cu-containing catalysts (Cu–Ni/CN and Cu/CN), on the other hand, it is shifted to temperatures higher than 550 K. It must be noted that CO2 conversion is kinetically constrained below 573 K. Because of that, all observed CO2 conversion profiles are significantly below the thermodynamic equilibrium curve (Figure A). At the same time, for the NiREF sample, the activity increases, reaching almost equilibrium values.

For Ni/CN and NiREF catalysts, the maxima in CO2 conversion and CH4 yield are observed between 620 and 750 K (Figure A,B) and remain below the thermodynamic limits. Moreover, for all CN-supported catalysts, RGWS is significant, as indicated by the respective CO profiles. According to the former study, small cubic Ni nanocrystals (nanoparticles NPS) are found to be highly selective in the hydrogenation of CO2 to methane, while larger Ni particles are reported to be more active in the RWGS reaction leading to CO formation.[8,14,16] However, this is valid for the particular support, which is characterized by rather weak basicity, as in the case of the NiREF sample, where alumina is used as a support.

In the cases Cu–Ni/CN and Cu/CN, although small CuO, the CuNi(1– mixed phase, and NiO crystallites were formed, the overall activity toward methanation is low (Figure C). CO2 conversions stayed far below the values allowed by equilibrium even in the higher temperature range (Figure A). As shown in Figure A,B compared to the Ni/CN- for Cu–Ni/CN-based catalysts, an increase in temperature has led to a CH4 formation drop, which was accompanied by enhanced CO formation.[16] For these catalysts, CO formation commences at 500–550 K and increases with increasing temperature, reaching a maximum above 720 K. During the CO2 methanation process, CO has the potential to be formed mainly via the reverse water–gas shift reaction (RWGS) (eq ).[16] Compared to the alumina-supported NiFER catalyst, for Ni/CN the CN support decreases the selectivity to methane. Besides, in the presence of a CN support, the RWGS reaction onset was shifted at a temperature lower by 200 K. Significant amounts of CO were detected from 600 K for all Ni/CN, Cu–Ni/CN, and Cu/CN catalysts. For these catalysts, the active phase is composed of very fine and stable nanoclusters/nanoparticles stabilized within the g-C3N4 cavity, as shown by the XRD study. The NPs size follows the order Cu (20 nm) > Cu–Ni (4, 130 nm) > Ni (ca. 1 nm). Besides, in Ni/CN and Cu/CN, Ni and Cu phases are defined by XRD as mainly Ni(111) and Cu(111), respectively. For those types of surfaces presented, the DFT study considered another hydrogenation mechanism for the difference between surface and subsurface H.[17,52] Over an alumina-supported catalyst, the CO2 methanation reaction is accepted to occur through an adsorbed CO intermediate via (i) the transformation of CO2 to CO prior to methanation or (ii) gas-phase transformation pathways not first requiring the transformation of CO2 to CO on the catalyst surface,[53] with the WGS reaction possibly occurring via (i) the formate, (ii) the redox, or (3) the carbonate mechanism.[54] Ni(111) CO2 hydrogenation to a formate intermediate was found to be more favorable than hydrogenation to a carboxyl intermediate.[17,52] Furthermore, the hydrogenation to formate proceeds through a univalent structure that is promptly transformed to a bivalent structure.[17,52] The formation of the formate species (HCOO–) involves at least three bond rearrangements, namely, (i) the cleavage of the O–H bond followed by (ii) the formation of the O–C and (iii) C–H bonds.[55] The once-formed HCOO– is relatively stable because its subhydrogenation to formic acid (FA) is energetically not as favorable as the formation of formate. Similar monovalent formate has been shown to be stable on Cu(111).[32,56] Despite these species being difficult to further hydrogenate to FA, they undergo almost spontaneous transformation to bidentate formate, as the energetic barrier for this reaction was reported ca. 0.06 eV.[17,52] Nevertheless, the presence of formate species is suggested to decrease the number of available sites for H2 dissociation as well as partially/completely block the path for hydrogen surface transport.[57] The formate coverage decreases with increasing temperature because of formate decomposition.[57,58] Accordingly, two different pathways have been proposed for that: (i) dehydrogenation producing CO2 and H2 and (ii) dehydration where CO and H2O are formed.[12] It was suggested that CO is usually released during dehydration while terminal OH (OH groups) are restored.[59] Our activity profiles over a CN-supported catalyst (Figure A,B) suggest that a similar mechanism can govern the CO2 hydrogenation because the dehydroxylation process is observed in a similar temperature range, also with CO present in the gas phase.[55]

The comparison of catalyst activity at 600 K is shown in Figure C. A relatively high activity of NiREF catalysts was related to the support basicity and its CO2 adsorption ability.[14,16] The Cu/CN catalyst has the lowest activity in CO2 hydrogenation among the CN-supported samples studied. For Cu-containing catalysts, an increase in CO selectivity was observed with traces of methanol and FA species (Figure C′). Cu is considered to be a relatively efficient catalyst for selective C–O bond hydrogenolysis and an excellent WGS catalyst, facilitating the conversion of CO to CO2 and vice versa.[32,56] It is also reported as an efficient methanol synthesis catalyst, using both CO2 and/or CO as a carbon source, even at atmospheric pressure.[33]

Although the CN catalysts’ activity is much lower than that noted for NiREF (Figure C), i.e., with a 40% drop in CO2 conversion for CN-supported material, the Ni content is far less in all studied CN-supported catalysts. In comparison to NiREF, Ni content is 70% lower for Ni/CN, Cu–Ni/CN, and Cu/CN catalysts. Besides, compared to standard commercial hydrogenation catalysts, the content of Ni is lowered by almost 90%, but still the CO2 conversion is maintained at ca. 20% at 600 K.[14,16] This, on the contrary, would suggest a much higher atom economy efficiency for CN-supported catalysts. This issue could also most likely be connected to the ease of hydrogen diffusion and access to the Ni or Cu sites, as was mentioned earlier for the 2D g-C3N4 structure with well-described cavities, porous structure, and interlayer distances.[13,15,19,27,30,38] Similarly, graphitic carbon nitride nanotubes were considered to be a promising hydrogen sorbent because of their highly porous structure and doubly bound nitrogen at the edges of the pores.[60,61] These reasons may be responsible for facilitating hydrogen diffusion into the interior of the nanotubes, thus providing active sites for hydrogen adsorption and/or functionalization with metal catalysts.

The stability test for Ni/CN was carried out at 623 K over 72 h, and the results were compared with NiREF (Figure D). Ni/CN catalysts exhibit a CO2 conversion barely reaching 50%, with CH4 selectivity declining from 43.0 to ca. 41.0% over the whole on-stream test (Figure D). The catalysts were characterized after the 72 h on-stream reaction. An AFM study showed no relevant changes in support morphology (Figure B′). More amorphous material was detected in the samples after reaction (Figure C′,D′), suggesting some extent of coking. However, no agglomeration of active phase/crystallites was observed.

Generally, nickel-based catalysts show high activity and selectivity in the CO2 methanation reaction, but they are easily sintered under the reaction conditions. On the other hand, Cu-based catalysts are found to be efficient for the hydrogenation of CO2 to hydrocarbons.[62] Regardless of the type of catalyst used, at low CO2 conversions, high methanol selectivity is reported, according to eq . The methanol synthesis described by eq is competitive with RWGS (eq ), and the increase in reaction temperature favors the endothermic RWGS reaction.A previous study shows that the Cu facets exposed to the reaction determine the dominant path of CO2 methanation. On the pristine Cu(100) surface, the CO2 adsorption was found to be thermodynamically unstable, yet the Cu(100) surface for the adsorption of CO2 molecules is reported to be more favorable than the Cu(111) surface.[63] Besides, the preadsorbed H atoms are accessible at each step of CO2 hydrogenation, as H2 molecule dissociation proceeds easily over Cu under the reaction conditions. According to the DFT study, a low activation barrier was predicted for thermocatalytic CO2-to-methanol reduction on small Cu nanoclusters. Introducing a second metal, i.e., Zn, Ni, or Co, onto the Cu surface can enhance the catalytic activity. Here, a promoting effect was demonstrated for Zn doped on the Cu surface for Zn coverage below 0.2 mL. The increased yield of methanol production was also observed for Ni/Cu(100) and Co/Cu(100) surfaces as well as for Rh-, Pt-, Pb-, and Au-substituted Cu atoms on the top layer of the Cu(111) surfaces. For Cu–Zn–Zr catalysts, it has been demonstrated that the CO2–adsorbent interaction can be weakened by Cu–O bond formation.[63] Therefore, the low conversion rates and CO2 reduction selectivities appear to be due to the thermodynamically unstable adsorption of crucial intermediates such as H2CO and H2COO on the clean Cu surface. Besides, both Cu+ and Cu0 species were reported to be essential for methanol selectivity, while the Cu+/Cu0 ratio determines the specific activity.[64] The second metal that is doped can inhibit or promote the desired reaction pathway. For example, Zn in the case of the Cu/ZnO-based catalyst was shown to have a considerable promoting effect in methanol synthesis, whereas RWGS was not influenced. In the case of CN-supported catalysts for all systems, the formation of FA or methanol stayed below 5 ppm, and Ni addition to Cu enhanced the CO2 methanation and increased the amount of CO formed at a higher temperature likely via RWGS. This could also suggest that methane is mainly formed from CO2 rather than CO hydrogenation.

Catalyst Toxicity Study

Influence of Catalysts on the Morphology of HaCaT Cells

The microscopic images showed that the HaCaT cells incubated for 24 h with catalyst samples demonstrated dose-dependent alterations in the morphology and that at the highest concentrations (250 and 500 μg mL–1) there was a reduction in their number (Figure ). In contrast, the untreated control cells were characterized by morphology that was normal and typical for this cell line and a proper high density (Figures –9). Cells treated with a low concentration of studied catalysts particles (3.125–25 μg mL–1) elicited only minor morphological changes compared to the control cells, which showed only a reduced number of cell-division-dependent spherical floating cells. At higher concentrations, HaCaT cells also demonstrated medium deviations in size, shape, volume, and structure, and cytoplasmic vacuolization occurred (Figure ), indicating a moderate cellular degeneration process. Some cells are swelled and even lysed; thus, catalysts could induce the permeability disorders of the cell membrane. The precipitation of catalyst particles in the form of aggregates (dark spots) is also visible. These morphological abnormalities were the most pronounced at a 100 μg mL–1 concentration, but they can also be seen at 25 μg mL–1. The results (images) are presented for the Cu/CN sample, but other samples caused similar changes, except the CN sample, where no symptoms of cell degeneration were observed.

Figure 6

Determination of HaCaT cell morphology. The cells were incubated for 24 h alone or treated in the presence of increasing concentrations of nanoparticles. The white scale bar corresponds to 100 μm.

Figure 7

HaCaT cell viability determined using the MTS assay after 24 h of coculture with increasing concentrations of catalyst particle solutions compared to control cells cultured in SFM. *p < 0.05, **p < 0.01, and ***p < 0.001 in comparison to the control.

Figure 9

Influence of catalysts on HaCaT cell viability/proliferation (dose–response curves). The data are expressed as the mean ± SD, n = 18. The IC50 values are marked with a dashed line. *p < 0.05, **p < 0.01, and ***p < 0.001 in comparison to the control.

Impact of Catalysts on the Viability/Proliferation of HaCaT Cells

The viability of the HaCaT cells treated with catalysts was determined with the MTS assay after 24 h of incubation (Figure ). The HaCaT cells cultured by the addition of catalysts particles (Cu/CN, Cu–Ni/CN, and Ni/CN samples) in the 3.125–500 μg mL–1 concentration range showing a dose-dependent proliferation decrease in comparison to the control, which constitutes untreated cells. They contrasted with cells subjected to the vehicle (CN sample), which did not induce a cytotoxic effect. A statistically significant effect (p < 0.05) was observed at concentrations of 50 μg mL–1 in the Cu/CN sample, 12.5 μg mL–1 in the Cu–Ni sample, and 25 μg mL–1 in the Ni/CN sample.

Interestingly, a statistically significant increase in cell proliferation was reported in the case of treatment with the CN sample at the highest tested concentration (500 μg mL–1). Incubation with 100 and 250 μg mL–1 Cu/CN, Cu–Ni/CN, and Ni/CN resulted in a decrease in the number of viable/proliferating HaCaT cells by 76.82–33.63, 21.56–12.47, and 55.48–39.70%, respectively. A 50% reduction (IC50) in HaCaT cell viability was observed at catalyst particle concentrations of 157.00 μg mL–1 (log IC50 = 2.196), 40.10 μg mL–1 (log IC50 = 1.603), and 53.93 μg mL–1 (log IC50 = 1.732) in Cu/CN, Cu–Ni/CN, and Ni/CN samples, respectively. IC50 values were calculated on the basis of the log dose–response (Figure ) and dose–response curves (Figure ). In the CN sample, the determination of IC50 was impossible because of the lack of 50% growth inhibition. The most pronounced cytotoxicity of catalysts was observed at doses from 100 to 500 μg mL–1 in tested samples. The HaCaT cells were the most sensitive to the Cu–Ni/CN treatment.

Figure 8

Influence of catalysts on HaCaT cell viability/proliferation (log dose–response curves). The data are expressed as the mean ± SD, n = 18. The log IC50 values are marked with a dashed line. *p < 0.05, **p < 0.01, and ***p < 0.001 in comparison to the control.

Contact with catalyst particles may occur through inhalation, ingestion, and dermal contact,[65,66] and for this reason, we have chosen the HaCaT human keratinocyte cell line derived from normal skin. Ni nanoparticles (NPs) are the most frequently studied among the metallic NPs in relation to cell culture.[67]

Previously, it was found that Ni/NPs induced cytotoxicity in a dose-dependent manner in the 10–100 μg mL–1 concentration range in human breast carcinoma MCF-7 cells.[68] The Ni/NPs caused mostly cyto-genotoxicity and oxidative stress, manifested by ROS production and GSH depletion.[68] It was also demonstrated that nickel oxide nanoparticles (NiO NPs) at levels above 10 μg mL–1 exhibited cytotoxicity in human liver cancer HepG2 cells via an apoptotic pathway and reactive oxygen species (ROS) generation.[69] For Ni/NPs, a concentration of approximately 30 μg mL–1 evoked a concentration-dependent morphology alteration and a cytotoxic effect after 24 h of treatment with bluegill sunfish BF-2 cells.[70] Various abnormalities have been reported there, including lysosomal, mitochondrial, and lactate dehydrogenase activity and oxidative stress manifested by the escalation in the peroxidation of lipids (LPO), protein carbonyl (PC), glutathione sulfotransferase (GST), and glutathione peroxidase (GPX). However, the catalase (CAT) and total glutathione content (TGSH) were dose-dependent and diminished when the dose decreased.[70] Pietruska et al. proved that Ni-containing NPs are cytotoxic to human lung carcinoma H460 and human bronchial epithelial NHBE cells at doses of up to 20 μg cm2 by activating the HIF-1α pathway and inducing hypoxia.[71] Interestingly, metallic Ni nanoparticles showed less toxicity than NiO NPs, whereas Ni microparticles were nontoxic toward these cells. Furthermore, the cytotoxicity evoked by all Ni species was associated with the apoptotic response activation, which can suggest substantial carcinogenic potential due to the dose- and time-dependent caspases and polypolymerase (ADP-ribose) cleavage.[71]

Nevertheless, Ni NPs at concentrations of as low as 2 μg mL–1 induce the occurrence of the cytotoxic effect (oxidative stress and apoptosis activation) in human lung epithelial A549 cells.[68,69] It has been reported that in normal human immortalized bronchial epithelial HBEC3-kt cells Ni and NiO NPs at a concentration above 10 μg mL–1 caused a release of inflammatory cytokines from exposed macrophages and inflammation-driven genotoxicity as a consequence.[72] Ni NPs have been shown to enhance platelet-derived growth factor (PDGF) activity in modulating the production of chemokines in average rat pleural mesothelial 2 NRM2 cells via a mechanism that includes ROS generation and prolonged activation of protein kinase ERK-1,2.[73] Ni NPs caused high toxicity starting at concentrations of 1–10 μM in human coronary artery endothelial cells (hCEC) and human coronary artery smooth muscle cells (hCASMC), disturbing their metabolic activity.[74] In the human cervix epithelioid carcinoma cell line (HeLa), NiO NPs caused 20% of cells apoptosis.[75] The cytotoxicity was determined in the concentration range of 50–200 μg mL–1 for 16 h, whereas the time was shortened by up to 2–6 h for the highest (i.e., 400–500 μg mL–1) concentration. A 400 μg mL–1 concentration of NiO NPs caused the lysis of the cell membrane and cell detachment from the culture plate surface due to the induction of apoptosis and necrosis.[75] In leukemia, K562-cell-functionalized Ni NPs with positively charged groups enhanced the permeability of the cell membrane, and besides causing apoptosis, they also induced necrosis.[76] In human fibroblasts, WI-38 cells demonstrated that nickel cylindrical nanostructures nanowires (Ni NWs) induced endoplasmic reticulum (ER) stress and, as a result, ER swelling in concentrations ≥2.25 μg mL–1 after 72 h of incubation.[77]

On the other hand, Cu NPs induce mutagenic changes and cause a significant increase in the number of binucleated cells with micronuclei.[77] This can be indicative of a genotoxic risk associated with Cu/NPs exposure because bare Cu/NPs have the potential to promote DNA strand breaks and cause oxidative DNA damage. The in vitro studies suggest that Cu/NPs can decrease cell viability, leading to cell death. Thus, the Cu/NPs’ possible neurotoxic potential and neurodegenerative activity were demonstrated. This effect has been related to dopamine depletion, the alteration of dopaminergic gene system expression, and oxidative stress in the neurons of rats. Apart from that, Cu/NPs exert toxicological effects on the liver, kidney, and spleen in mice.[77] However, the IC50 dose for Cu/NPs against ca. 1.71 μg mL–1 has been reported for human skin cancer cell A-375. Cu/NPs reduced the cell membrane rigidity, causing a genotoxic effect via DNA degradation and chromosomal condensation. They were found to induce cell cycle arrest in the G2/M phase, leading to depolarization of the mitochondrial membrane and finally to cells apoptosis.[78] A similar action model was observed in CuO NPs toward TIC-enriched PANC1 human pancreatic cancer cell cultures. The IC50 value was 10 μg mL–1.[79] A549 lung adenocarcinoma cells were found to be considerably more sensitive to the cytotoxic effects of CuO/NPs than HBEC human bronchial epithelial cells.[80] The in vitro toxic potential of increasing concentrations of the Cu/NPs obtained using green chemistry (1–500 μg mL–1) in the proliferation and morphological characteristics of the human HepG2 cancer cell line was estimated to be 54.5%.[81] However, another study over surface-modified Cu/NPs with broccoli green extract did not exhibit cytotoxicity in the 0.5–1.5 μM concentration range for the prostate PC-3 cancer cell lines.[82]

Bimetallic Cu–Ni NPs possess antibacterial activity and could be used in dental materials, although there are some reports that monometallic Cu/NPs also exhibit high antibacterial potency.[83−85]

The data obtained for the CN-containing material are consistent with literature data because reduced viability and morphological changes in HaCaT cells were observed from concentrations of 12.5, 25, and 50 μg mL–1 for Cu–Ni/CN, Ni/CN, and Cu/CN NPs samples, respectively. Indeed, in most studies that evaluated the cytotoxicity of Ni NPs or NiO NPs, cells from mammals, especially cells derived from human lungs, have been involved. This is because Ni is well recognized to cause several pulmonary diseases, including fibrosis and lung cancer.[86,87] It is worth remembering that different cells have different cell viability under comparable conditions when exposed to Ni and Cu/NPs, but the mode of NP action seems to be similar. NP-mediated toxicity includes oxidative stress, inflammation, genetic damage, and the inhibition of cell division, consequently leading to cell degeneration and apoptosis.[77,88] Generally, Cu/CN NPs were proven to be highly cytotoxic compared to other metal NPs.[88] The next step should be to trace the mode of action of Ni and Cu/CN NPs in HaCaT cells and to determine whether in the case of this cell line NPs induce the same effects.

By exerting cytotoxicity on cancer cells, they exhibit great potential in oncology clinical and (bio)medical applications; they may constitute promising drug carriers by enhancing the proper cellular uptake.

Conclusions

The technology needed to valorize CO2 via hydrogenation has been identified as a possible pathway to transform one GHG gas into value-added products such as chemicals, fuel feedstocks, and drop-in fuel. This makes CO2 a likely alternative in climate change mitigation. Although the CO2–methanol process is recognized as a reasonably well-established and mature approach, a vital concern for increasing the process efficiency is to identify a durable and nontoxic catalyst that ideally demonstrates high low-temperature activity.

The results presented by 2D graphitic carbon nitride-supported catalysts revealed a stable performance. All studied hybrid catalysts with active phase loading of ∼4 (wt %) were active in the CO2 hydrogenation reaction. Although the Ni loading in Ni/CN is lower by more than 90%, compared to the reference NiREF catalyst a conversion of close to 20% was achieved at ∼623 K, with CH4 being the primary reaction product (>80% CH4 yield).

Moreover, considering the toxicological profiles, the hybrid catalysts provide a direction for catalyst design with less toxicity to the environment and health. Twenty-four hours of treatment with Cu/CN, Cu–Ni/CN, and Ni/CN reduced the viability of HaCaT keratinocytes cells, causing degenerative morphological changes. Only CN bare g-C3N4 did not disturb the appropriate cell structure but positively affected their proliferation, even at the highest studied concentrations. Surprisingly, it shows great potential for clinical and (bio)medical applications in the fields of oncology and pharmacy as a drug carrier. All samples were nontoxic toward HaCaT cells at a concentrations of 6.25 μg mL–1. The highest toxicity was found for the Cu–Ni/CN sample, which caused a decrease in the viability of cells at a concentration of 12.5 μg mL–1. At the same time, the least toxic was the Cu/CN sample, where the antiproliferative effect was present at a 50 μg mL–1 concentration.