Catalytic Properties of High Nitrogen Content Carbonaceous Materials

Abstract

The influence of structural modifications on the catalytic activity of carbon materials is poorly understood. A collection of carbonaceous materials with different pore networks and high nitrogen content was characterized and used to catalyze four reactions to deduce structure–activity relationships. The CO₂ cycloaddition and Knoevenagel reaction depend on Lewis basic sites (electron‐rich nitrogen species). The absence of large conjugated carbon domains resulting from the introduction of large amounts of nitrogen in the carbon network is responsible for poor redox activity, as observed through the catalytic reduction of nitrobenzene with hydrazine and the catalytic oxidation of 3,3′,5,5′‐tetramethylbenzidine using hydroperoxide. The material with the highest activity towards Lewis acid catalysis (in the hydrolysis of (dimethoxymethyl)benzene to benzaldehyde) is the most effective for small molecule activation and presents the highest concentration of electron‐poor nitrogen species.

Carbon materials are attractive alternatives to metal catalysts in terms of price and reduced environmental impact. However, the influence of structural modifications on the catalytic activity of carbon materials is poorly understood. General trends in the catalytic activity of nitrogen‐doped carbons are identified using a holistic approach to carbocatalysis.

Introduction

Carbon‐based materials are cheap, high surface area supports for metal catalysts. Activated charcoal has been known for over a century to promote the aerobic oxidation of organic molecules, such as ethanol, in the gas phase. ^[1] Nanostructured materials, including graphene, ^[2] graphene oxide, ^[3] carbon nanotubes, and nanodots ^[4] act as acid/base catalysts and promote redox reactions (“Carbocatalysis”).[ ¹ , ⁵ ] Manipulating the electron density via heteroatom doping is among the most studied strategies to tune the catalytic properties of carbonaceous materials.[ ¹ , ⁶ ] Nitrogen doping results in the withdrawal of electron density from adjacent carbon atoms. ^[7] A graphene monolayer with nitrogen atoms inserted in the carbon backbone was studied using scanning tunneling microscopy, ^[8] and Mulliken charges of nitrogen‐doped carbon materials were calculated.[ ^6a , ⁹ ] While the numbers differ, a negative charge localized on nitrogen atoms (between −0.2 and −0.6) and a positive charge on adjacent carbon atoms (between +0.05 and +0.58) were found in all cases.

For catalytic applications, nitrogen atoms with low oxidation states act as Lewis base, while the adjacent carbon atoms have Lewis acidic character.[ ^7b , ¹⁰ ] Carbon atoms close to nitrogen are involved in activating small electron‐rich molecules, such as hydrazine or hydrogen peroxide, for redox reactions. The reduction of nitrobenzenes to anilines with hydrazine is facilitated by nitrogen atoms due to improved substrate adsorption.[ ^7d , ¹¹ ] The performance of N‐doped carbonaceous materials as Lewis acid/base catalysts and catalytic redox reactions has not been comprehensively investigated to date to understand the parameters that influence the catalytic activity of such materials.

Preparation of nitrogen‐containing carbonaceous networks relies on nitrogen‐containing carbon precursors or post‐treatment of carbon materials with nitrogen‐rich molecules. The resulting materials typically have low nitrogen content (<10 wt%), which limits the effect of heteroatom doping on the catalytic activity. Here, we present the preparation and characterization of nitrogen‐rich nanocarbons from cytosine at different temperatures via a scalable and cost‐efficient salt melt template synthesis. The catalytic activity in acid‐, base‐ and redox catalysis was evaluated and compared with the properties of the carbonaceous materials.

Results and Discussion

A set of materials was synthesized from cytosine at different temperatures (500 °C, 600 °C, 700 °C, and 800 °C) using a salt melt (LiCl, ZnCl₂) template synthesis that promotes the formation of a porous network (Figure 1a). ^[12] In all cases, a mixture of LiCl, ZnCl_2, and cytosine (10 : 10 : 1 wt/wt/wt ratio) was heated to the target temperature at a heating ramp of 1 °C/min. After cooling to room temperature, the materials were washed with hydrochloric acid (0.1 M) to remove the salt template and obtain the nitrogen‐rich carbon materials C5LZ (prepared from c ytosine at 5 00 °C in a L iCl/ Z nCl₂ melt), C6LZ, C7LZ, and C8LZ.

Figure 1.

a) Synthesis of nitrogen‐rich carbonaceous materials from cytosine in a LiCl/ZnCl₂ melt; b) N₂ adsorption‐desorption isotherms at 77 K, c) CO₂ adsorption isotherm at 273 K, and d) SEM‐EDX composition of the prepared samples.

The pore structure of the materials was analyzed by N₂ adsorption/desorption isotherms at 77 K (Figure 1b). The nitrogen isotherm shape of C8LZ indicates a hierarchical pore structure comprising micro‐, meso‐, and macropores. ^[13] C7LZ and C6LZ showed a sharp increase in the volume adsorbed at low relative pressures, characteristic of materials that have mainly a microporous structure. In addition, the small hysteresis loop at 0.5 P/P₀ in the nitrogen isotherms indicated a small fraction of mesopores. C5LZ showed poor nitrogen sorption, which is rationalized by the absence of a porous network. ^[7b] Specific surface areas were calculated by applying the Brunauer–Emmett–Teller (BET) equation to the adsorption branch of nitrogen adsorption isotherms and gave 44, 749, 1207, and 1701 m² g⁻¹ for C5LZ, C6LZ, C7LZ, and C8LZ, respectively (Table S1). Pore size distribution, calculated using quenched solid density functional theory (QSDT), confirmed these observations (Figure S1). While N₂ adsorption increases linearly with carbonization temperature, CO₂ adsorption isotherms recorded at 273 K showed a different trend (Figure 1c). The highest pore volume was calculated for C8LZ (0.180 cm³ g⁻¹), and the lowest value was obtained for C5LZ (0.103 cm³ g⁻¹). The pore volume calculated from CO₂ sorption for C5LZ is higher than the pore volume observed by N₂ sorption (Table S1), indicating a higher affinity for CO₂ binding due to its large nitrogen content. Despite its lower micropore volume, C6LZ had a higher CO₂ uptake than C7LZ. This is indicative of a higher affinity of carbon dioxide to the surface of C6LZ. Such an increase in affinity is typically ascribed to the presence of more pyridinic nitrogen functionalities. ^[14] These differences motivated the textural and chemical characterization of the samples.

Scanning electron microscopy (SEM) revealed that the materials present a texture formed by colloidal particles (Figure S2). This morphology indicates that the LiCl/ZnCl₂ mixture melts before condensation of the carbonaceous precursor starts such that the carbonaceous precursor dissolves in the ionic melt. Upon further heating, the carbon precursor condenses and precipitates as oligomer seeds, similar to a sol‐gel mechanism. At higher temperatures, the seeds grow and form colloidal carbonaceous particles. A similar effect was observed in the salt melt template synthesis of carbon materials using ZnCl₂. ^[15] Transmission electron microscopy (TEM) confirmed that lower target temperatures result in less porous materials (Figure S3). A dense solid phase was observed for C5LZ, while increasing the target temperature results in more porous materials. Powder X‐ray diffraction (XRD) analysis showed a broad band at 26.5° for C5LZ and C6LZ, which indicates layer stacking (Figure S4a). This signal was not observed for C7LZ and C8LZ, suggesting that de‐stacking occurs at higher temperatures typical for highly porous materials. The tendency of the diffraction feature at 26.5° to widen into a weak broad band at 2θ ≈25–26° is characteristic of disordered frameworks. ^[16] Moreover, the development of large pore volumes fosters an increase in intensity at diffraction angles below 10°. On the contrary, when cytosine is directly carbonized in the presence of no salt melt at different temperatures (samples C5d, C6d, C7d, and C8d), the peak at 26° increases with the increase of the carbonization temperature and the increase in the intensity at angles below 10° is not as intense (Figure S4b).

Composition analysis via energy‐dispersive X‐ray spectroscopy (SEM‐EDX) (Figure 1d, Table S2) showed that the carbon content increases from 62.2 wt% to 83.0 wt% with increasing target temperature. In contrast, the nitrogen content decreases from 29.9 wt% to 11.1 wt%. This translates to an increase in the C/N ratio from 2.4 (C5LZ) to 8.7 (C8LZ) (Figure S5). The oxygen content slightly varies from 4.8 wt% to 3.8 wt%. Here it is important to mention that, to avoid the overestimation of the oxygen content of the samples, each material was degassed at 150 °C in a vacuum oven for 12 h to remove the contribution of weakly bonded water molecules. ^[17] All materials contain small amounts of chlorine (0.3–3.2 wt%) and zinc (0.1–0.3 wt%) that originate from the salt melt template synthesis, which was further confirmed by inductively coupled plasma analysis (ICP) (Table S3).

X‐ray photoelectron spectroscopy (XPS) was used to investigate the chemical structure of all materials in detail (Figure S6). The deconvoluted high‐resolution N1s XPS spectrum showed peaks centered at 398 eV, 400 eV, 401 eV, and 402 eV, which indicates that the materials contain nitrogen atoms with low (“electron‐rich”) and high (“electron‐poor”) oxidation states (Table S4, Figure S7). Even though the high nitrogen content and the amorphous nature of the materials renders difficult an assignment to specific functionalities, the conventional assignments will be used for clarity purposes. The peaks at 398 eV, 400 eV, 401 eV, and 402 eV will be ascribed to pyridinic, pyrrolic, quaternary, and N‐oxide nitrogen atoms, respectively.[ ^7c , ¹⁸ ] Because the materials significantly differ in their total nitrogen content, the amount of each nitrogen species was calculated in mol per 100 g of material for better comparison using the data obtained from SEM‐EDX and XPS experiments (Figure 2a). A slightly higher amount of nitrogen atoms with high oxidation states (i.e., quaternary nitrogen at 401 eV and 402 eV) was observed for C6LZ compared to C5LZ and stayed almost constant at higher synthesis temperatures. It is worth noting that these nitrogen species still carry a negative Mulliken charge even though they present a high oxidation state. ^[9] On the contrary, the number of nitrogen atoms with low oxidation states (i.e., pyridinic and pyrrolic) decreases gradually with rising target temperatures.

Figure 2.

a) Concentration (mol/100 g) of nitrogen species as obtained from N1s XPS and SEM‐EDX composition, b) concentration (mol/100 g) of carbon species as obtained from C1s XPS and SEM‐EDX composition, and c) NH₃‐TPD profile of C5LZ, C6LZ, C7LZ, and C8LZ.

The deconvoluted spectra of the C1s peak showed three bands centered at 284 eV, 286 eV, and 288 eV (Figure 2b, Figure S6). This relates to carbon atoms with low (sp² C−C bonds, 284 eV) and high (bound to one or more heteroatoms, 286 eV and 288 eV) oxidation states. Although the total carbon content increases at higher target temperatures, the amount of carbon atoms at 284 eV remains similar for C5LZ, C6LZ, and C7LZ (Table S5, Figure 2b). The deconvolution peaks centered at 286 eV and 288 eV indicated an increase of electron‐poor carbon atoms from C5LZ (1.6 mol/100 g) to C7LZ (2.3 mol/100 g). C8LZ showed significant differences compared to other materials. The total amount of carbons with high oxidation states (286 eV and 288 eV) is low (1.5 mol/100 g), whereas a comparably high amount of carbons at 284 eV was calculated (5.4 mol/100 g). This observation agrees with the high total carbon content for C8LZ, which translates to more C−C bonds in the material. The deconvoluted spectra of the O1s peak showed that the proportion of C=O and C−O−C groups does not change significantly by increasing the temperature (the C=O/C−O−C area ratio 1:0.60±0.02 is for all the samples). Taking into account the very steady amount of oxygen in the samples, we will not consider oxygen contributions as a possible factor influencing the catalytic performance of these samples.

Temperature‐programmed desorption (TPD) was used to study the interaction of ammonia with all materials (Figure 2c). C8LZ had a desorption band with low intensity from 100 °C up to 250 °C. From C5LZ to C7LZ, the desorption band became sharper and more intense, and the desorption maximum shifted to lower temperatures. The higher intensity of the NH₃‐TPD traces can be rationalized by a higher amount or better accessibility of Lewis acidic groups, and the desorption maximum at lower temperatures indicates a lower Lewis acidity. The uptrend observed for C8LZ is ascribed to ammonia slowly desorbing from its much larger pore volume as TPD traces of the sample prior to ammonia adsorption, and TGA analysis in nitrogen atmosphere shows that that the sample is stable in that temperature range (Figure S8).

Correlating properties of nitrogen‐doped carbonaceous materials with catalytic activity is challenging because varying the amount of nitrogen is accompanied by other changes, such as a different surface area and pore size distribution. We decided to investigate the structure–activity relationship of all materials for four different reactions to study the influence of each parameter. The catalytic reduction of nitrobenzene in the presence of hydrazine (N₂H₄) and the catalytic oxidation of 3,3′,5,5′‐tetramethylbenzidine (TMB) using hydrogen peroxide (H₂O₂) were chosen as model reactions to study the factors influencing the activation of small, electron‐rich molecules in redox reactions. The ability of the materials to serve as acid and base catalysts was studied for the synthesis of cyclic carbonates from CO₂ and epichlorohydrin (ECH), as well as the hydrolysis of (dimethoxymethyl)benzene (DMB) to benzaldehyde (BA) followed by a Knoevenagel reaction with malononitrile in a one‐pot two‐step procedure.

First, we studied the catalytic reduction of nitrobenzene with N₂H₄ at room temperature using a reaction time of 48 h (Figure 3a). It was previously proposed that carbon materials catalyze this reaction by creating a pool of electrons through the adsorption of hydrazine ^[19] and that the catalyst activity can be improved by tailoring the adsorption energy by nitrogen and sulfur doping. ^[20] In addition, in 2018, nitrogen doping was shown by Chanjuan Liao et al. as an effective strategy to generate active sites for reductions in the presence of hydrazine. ^[21] Accordingly to these results, the desired product was obtained in excellent yield when C8LZ was used. Nonetheless, a small amount of the desired product was detected in the case of C7LZ, While C6LZ and C5LZ did not yield in the formation of aniline.

Figure 3.

a) Catalytic reduction of nitrobenzene in the presence of hydrazine (N₂H₄) and molar fraction of aniline and nitrobenzene after 48 h obtained by ¹H NMR (5 mg catalyst, ethanol as solvent, and N₂H₄ ⋅ H₂O). b) Catalytic oxidation of 3,3′,5,5′‐tetramethylbenzidine (TMB) using hydrogen peroxide (H₂O₂).The kinetic profile was determined via the absorbance at 652 nm (5μg L⁻¹ of catalyst, 0.25 mM of TMB, 40 mM H₂O_2, and 1 mL water).

Next, the catalytic activity of the materials was evaluated for the oxidation of TMB with hydrogen peroxide by tracking the reaction progress using UV/Vis spectroscopy (Figure 3b, Figure S9). Similar to the results obtained for the nitro reduction described above, C8LZ had the highest catalytic activity. The use of C7LZ resulted in a moderate yield of the oxidation product. C5LZ and C6LZ showed almost no catalytic activity. The reaction using C8LZ was analyzed using Michaelis–Menten kinetics (Figure S10, Table S7). ^[22] The maximum rate at a saturating substrate concentration (V _max) was 0.8 μM s⁻¹. The substrate concentration at which the reaction rate is half of v_max (K _M) was 28.0 mM. A comparison with previously published carbonaceous materials as catalysts for the same transformations shows that C8LZ presents the highest v_max and the second‐highest K _M reported to date (Table S7).[ ²³ , ²⁴ ]

Due to the low catalytic activity of C5LZ, C6LZ, and C7LZ in both reactions, few conclusions can be drawn. Although electron‐poor carbons are, to a certain degree, responsible for the adsorption of small molecules, their amount and Lewis acidity is likely not the critical parameter to promote these reactions. The catalytic activity of C5LZ, C6LZ, and C7LZ, which contain many electron‐poor carbon atoms in their carbon network, was low. A large surface area and extensive pore structure do not guarantee high catalytic performance. C6LZ presents a larger surface area than previously reported nitrogen‐doped carbonaceous materials, but its catalytic activity in both model reactions was low. ^[23] Only when the surface area is increased over 1000 m² g⁻¹ does the effect of the extensive pore network affect the catalytic activity (see the performance of sample C7LZ). Here, it is important to point out that oxygen functionalities might also act as active sites as described by Shuchang Wu et al. ^[25] However, the rather constant oxygen content in the different samples suggests that they might not be the main catalytic active site of the samples.

The fact that C5LZ, C6LZ, and C7LZ do not show high catalytic activity suggests that a very high nitrogen content is detrimental for redox catalysis, likely because it lowers the dimension of conjugated carbon domains. These domains are critical for electron transfer processes. ^[26] The high K _M of C8LZ in the TMB oxidation indicates that a high H₂O₂ concentration is necessary to saturate the catalytically active sites. Other studies using carbon‐based materials show that K _M decreases upon nitrogen doping, rationalized by a higher affinity to H₂O₂.[ ^23a , ^24b ] We assume that the high nitrogen content in C8LZ, in combination with the porous morphology, results in a situation where catalytically active sites are accessible but slowly saturated.

We studied the ability of high nitrogen‐content carbonaceous materials to serve as catalysts in the synthesis of cyclic carbonates from CO₂ and epichlorohydrin at 130 °C (Figure 4a, Table S8). The use of C6LZ resulted in the highest conversion (58 %) of epichlorohydrin, but the selectivity towards the desired product was only moderate (61 %). No side products were detected using ¹H NMR analysis (Figure S11), suggesting that polymerization reactions are responsible for the low selectivity. ^[27] A lower conversion (27 %) but excellent selectivity (95 %) was observed for C8LZ. C7LZ showed low activity (36 % conversion) and selectivity (46 %). The use of C5LZ did not result in any reaction of ECH. This reaction requires the activation of CO₂ with a nucleophile (Lewis base). ^[28] Consequently, the results confirm the presence of Lewis basic sites within the structure of C6LZ, C7LZ, and C8LZ. According to the accepted mechanism, a Lewis acid is not required but can improve carbonate formation. ^[28] Our results suggest that C6LZ (highest ECH conversion) and C8LZ (highest selectivity) have the highest amount of Lewis acidic sites of all tested materials.

Figure 4.

a) CO₂ cycloaddition to epichlorohydrin and conversion values obtained by ¹H NMR in the presence of an internal standard using the different materials as catalysts. b) Acid‐catalyzed hydrolysis of (dimethoxymethyl)benzene (DMB) to benzaldehyde (BA) followed by a base‐mediated Knoevenagel condensation and conversions for the acidic (red) and basic (blue) reaction as obtained by ¹H NMR in the presence of an internal standard using the different materials as catalysts.

For an in‐depth analysis, we studied the acid‐catalyzed hydrolysis of (dimethoxymethyl)benzene (DMB) to benzaldehyde (BA) followed by a base‐mediated Knoevenagel condensation that yields benzomalononitrile (BMN) (Figure 4b). ^[29] The two‐step one‐pot reaction was monitored by ¹H‐ NMR analysis (Figure S12, Figure S13). The amount of DMB that was consumed after one hour showed that C8LZ (80 % DMB conversion) is a better Lewis acid catalyst than C7LZ (65 % DMB conversion) and C6LZ (52 % DMB conversion) (Figure 4b). The base‐catalyzed conversion of BA to BMN follows a negative trend for the same set of materials due to the decreasing Lewis basic sites from C6LZ to C8LZ. C5LZ shows good activity for Lewis acid catalysis but performs poorly as a Lewis base catalyst. Likely the low surface area of C5LZ is responsible for that finding. In the case of C6LZ, C7LZ, and C8LZ, the obtained results are in agreement with speciation via XPS analysis. The activity in Lewis base catalysis decreases in line with the decrease of electron‐rich pyridinic and pyrrolic nitrogen atoms. In contrast, the activity towards Lewis acid catalysis increases with a higher amount of electron‐poor carbon atoms. This is in agreement with our observation from NH₃‐TPD. The samples showing larger amounts of electron‐poor carbon atoms also have a higher ratio of electron‐poor over electron‐rich nitrogen atoms. These nitrogen electron‐poor atoms can promote the formation of more positive charges on the adjacent carbon atoms. ^[9] These results confirm that strategic nitrogen doping tailors carbonaceous materials Lewis acidic and basic properties.

Our findings reveal that CO₂ and ECH reaction conversion follows a similar trend as the Knoevenagel reaction, indicating its dependence on Lewis basic sites. All materials show almost complete conversion in the two‐step, one‐pot BA synthesis (Figure S12), confirming the absence of large conjugated carbon domains as the main factor in the low activity towards the activation of N₂H₄ and H₂O₂ of C5LZ, C6LZ, and C7LZ. The material with the highest Lewis acid catalysis activity (C8LZ) is also the most effective material for small molecule activation and presents the highest concentration of quaternary nitrogen. These are key structural features for Lewis acidity of nitrogen‐doped carbon‐based materials.

Conclusion

We present a simple, straightforward synthetic method for carbonaceous materials with high nitrogen content (up to 34 wt% nitrogen content) and a high surface area (up to 1701 m² g⁻¹). The materials were characterized by N₂ and CO₂ sorption, SEM‐EDX, TEM, PXRD, XPS, and TPD to study the influence of the synthesis temperature on the surface area, composition, and surface electron density. The materials are active catalysts for the reduction of nitrobenzene with N₂H₄, the oxidation of 3,3′,5,5′‐tetramethylbenzidine H₂O₂, the synthesis of cyclic carbonate from CO₂ and epichlorohydrin, and a two‐step one‐pot synthesis of benzomalononitrile that involves Lewis acid and Lewis base catalysis. Our results identified the critical physicochemical parameters of such materials for catalytic applications. N₂H₄ and H₂O₂ can be activated on the surface of nitrogen‐doped carbonaceous materials through Lewis acidic sites. If the nitrogen content becomes too high, a detrimental effect was observed due to the loss of a conjugated structure. Cyclic carbonate synthesis mainly depended on Lewis basic sites identified that activate CO₂. The amount of electron‐poor quaternary nitrogen atoms could be correlated to the activity in Lewis acid catalysis and was ascribed to the presence of electron‐poor carbon atoms next to electron‐poor nitrogen species.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Lepre E., Rat S., Cavedon C., Seeberger P. H., Pieber B., Antonietti M., López-Salas N., Angew. Chem. Int. Ed. 2023, 62, e202211663; Angew. Chem. 2023, 135, e202211663.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.