Highly Efficient Bimetallic Catalysts Supported on Carbon Nanotubes for the NOx Reduction

Abstract

Nitrogen oxides represent a major source of concern related to atmospheric pollution, causing substantial impacts on human health. One innovative approach to reducing these emissions, and a promising alternative to conventional methods using NH₃, is selective catalytic reduction with carbon (SCR-C). The aim of this study is the development of carbon-based catalysts that are active in the reduction of NO. For that, carbon nanotubes were subjected to treatments to modify their surface chemistry, including introducing oxygen and nitrogen groups, as well as potassium (K) and copper (Cu) incorporated as metal phases. In their original form, carbon nanotubes do not exhibit catalytic activity in reducing NO. However, catalytic performance is significantly improved by the addition of surface groups and Cu. Adding K to the support notably contributes to increasing the catalytic performance. N-doped carbon nanotubes impregnated with copper and potassium (CNT_M_BM@5Cu5K) achieved complete NO reduction at 360 °C. In this catalytic system, the formation of CO₂ and N₂ was observed and CO was not identified. Furthermore, although N₂O was detected during the reaction, its amount was very low compared to the N₂ and CO₂ products. The stability of this catalyst was investigated over 87 h continuous test, revealing deactivation after 41 h of reaction.

1. Introduction

The reduction of nitrogen oxide emissions remains a significant environmental and technological challenge, given their role in atmospheric pollution and their impact on human health [1,2,3,4,5,6]. Nitric oxide (NO) is recognized as one of the most challenging gases to eliminate from the air due to its tendency to exhibit properties similar to those of a supercritical fluid properties at room temperature [7]. Nitrogen dioxide (NO₂) is easier to eliminate due to its affinity for dissolving in water. Therefore, the main obstacle to reducing NOx gases is the difficulty in eliminating NO. The main polluting component of this type of gas is NO, which accounts for around 90% of total NOx [8]. N₂O is not part of the group known as NOx, but it is still a relevant pollutant. Although present in low amounts, they are particularly important because they act as intermediaries in the reduction of NO and are produced as a by-product of NOx [9,10]. Many methods, including adsorption, catalytic oxidation, photochemical processes and selective catalytic reduction (SCR), have been developed to control air pollution [11,12].

The conventional method adopted to minimize emissions of NOx is the application of SCR technique, in which ammonia is used as a reducing agent [13,14,15]. However, the use of ammonia in this process can result in additional environmental challenges, especially due to the potential for leakage of this chemical compound. These leaks can contribute to problems such as creating fine particles in the atmosphere, negatively impacting air quality and surrounding ecosystems [16,17].

An alternative to this process is selective catalytic reduction with carbon (SCR-C) [18,19]. In this approach, carbon materials play a dual role, acting both as catalyst supports and as reducing agents [19,20,21,22]. Carbon is capable of reducing NO to N₂ without the need for an external reductant, although high temperatures are usually required and the reaction efficiency strongly depends on the structural and surface properties of the carbon material [23]. Therefore, tailoring the physicochemical characteristics of carbon-based catalysts is essential to enhance NO conversion under practical conditions.

Carbon materials, identified as highly effective catalysts to stimulate and promote this reaction, stand out for their porous structure and unique chemical properties, allowing efficient interaction with the reagents and enhancing the conversion of the compounds involved in the reaction [7,19,24,25]. Carbon nanotubes (CNT) have attracted much attention and offer distinctive advantages compared to conventional activated carbons [26], including high structural purity, excellent thermal stability, high electrical conductivity and the absence of significant microporosity, which reduces internal diffusion limitations [27].

CNT were selected in this study due to their accessible structure and negligible microporosity, which minimizes internal diffusion limitations and facilitates mass transfer during the SCR-C reaction. In addition, CNT exhibit high thermal stability and are widely recognized as efficient catalyst supports due to their graphitic nature and ability to disperse active phases on their surface [26,27]. Furthermore, CNT can be readily functionalized (e.g., by oxidation or N-doping), allowing controlled tuning of surface chemistry, a key parameter governing NO reduction on carbon-based catalysts. In the SCR-C process, carbon nanotubes play multiple roles. They function as a reducing agent, a catalyst support and a catalyst, rather than acting as inert supports.

The efficiency of NO activation and reduction is governed by their surface chemistry and interaction with the metallic phases. N-doping modifies the surface chemistry of CNT by increasing their basicity and electron density. Meanwhile, interactions with Cu and K govern the efficiency of NO activation and reduction via metal–support interactions, resulting in synergistic enhancement of catalytic performance.

Carbon nanotubes are rolled graphene sheets made up of sp² carbons [28]. They can be categorized into single-walled carbon nanotubes (SWCNT), which contain a single graphene layer, and multi-walled carbon nanotubes (MWCNT), formed by multiple graphene layers. The specific surface area values of SWCNT range between 400 and 900 m² g⁻¹, and between 150 and 300 m² g⁻¹ for MWCNT [26]. In addition to acting as catalysts, carbon-based materials can act as active support, which playing a critical role in enhancing catalytic activity. Several methodologies have been employed to optimize their performance [18,19,21,29].

Recently, catalysis using nitrogen-doped carbon materials has received increasing attention [4,30]. These materials have been used as catalysts in various fields of application [31,32,33]. The incorporation of nitrogen enhances the chemical performance of these materials, increasing the electron density on the surface and their electron donation capacity, which results in greater catalytic effectiveness [4,18,25]. Ramalho et al. [19] developed catalysts with different surface properties and demonstrated that the nitrogen functionalization enhances catalytic performance in NOx reduction.

Incorporating alkali and transition metals into carbon-based catalysts significantly reduces the energy required to initiate reactions and increases NO adsorption capacity, further enhancing the reactivity of carbon towards NOx [18,19,34]. During the interaction between carbon and NO, the impregnated metal participates in the redox cycle, undergoing oxidation by NO and reduction by carbon [19,29]. Illán-Gómez et al. [20,35] developed catalysts that combined potassium (K) with transition metals (Co, Fe, Cu and Ni) to reduce NO and concluded that bimetallic K catalysts (KCo, KFe, KCu and KNi) exhibited a positive effect on NOx reduction. Similarly, Bingnan et al. [24] studied the role of K addition to Cu and found that it influences catalytic performance in NO reduction, acting as a highly effective promoter of CuO for NO conversion.

In previous studies, Cu–K bimetallic catalysts supported on N-doped activated carbon were found to exhibit enhanced NO reduction performance, achieving complete NO conversion at lower temperatures than monometallic systems [36]. The present work explores the use of CNT as an alternative carbon support material, which has different textural and surface chemical properties, building on these findings.

The main objective of this work is to explore the potential of CNT in the catalytic reduction of NO using CNT as a support for a catalyst containing Cu and/or K. Furthermore, the impact of N-doped CNT on NO reduction was analyzed. For comparison purposes, CNT treatments were also carried out to introduce oxygenated groups.

2. Materials and Methods

2.1. Preparation of Modified Carbon Nanotubes

Multi-walled carbon nanotubes (MWCNT, 90% purity, Nanocyl NC7000, Birla Carbon, Sambreville, Belgium) were used as the starting material for the preparation of supports with distinct textural and surface chemical characteristics. Surface oxidation was carried out by refluxing in nitric acid, enabling the introduction of oxygen-containing functional groups (CNT_HNO₃), according to procedures previously described in the literature [32,37]. Nitrogen-doped CNT was achieved through a mechanothermal approach. In this process, melamine was used as the nitrogen precursor during milling. The obtained solid mixture was subsequently subjected to thermal treatment under a N₂ atmosphere at 600 °C in order to promote melamine decomposition and nitrogen incorporation into the carbon framework (CNT_M_BM) [31,38].

2.2. Preparation of Metal Catalysts

Aqueous solutions of Cu(NO₃)₂ and KNO₃ were used to prepare Cu and/or K catalysts by the incipient wetness impregnation technique. The copper content was estimated to be 10 wt.%, and for the bimetallic catalysts 5 wt.% of each metal. After impregnation, the materials were dried at 100 °C and subsequently subjected to thermal treatment. The temperature was increased to 460 °C (corresponding to the reaction temperature) at a heating rate of 10 °C min⁻¹. The catalysts were treated for 1 h under N₂, followed by 3 h under H₂ [39,40].

2.3. Characterization of Materials

The textural properties of the prepared catalysts were determined by N₂ adsorption–desorption isotherms at −196 °C using a Quantachrome NOVA 4200e multistation instrument (Anton Paar, Boynton Beach, FL, USA).

Surface chemistry composition was investigated by XPS analysis, following methodologies previously reported [18]. Thermogravimetric analysis (TGA) was performed using a STA 409 PC/4/H Luxx Netzsch thermal analyser (Netzsch, Selb, Germany). Further experimental details regarding these techniques can be found elsewhere [41].

Metal loadings were quantified by Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES) employing an ICPE-9000 spectrometer (Shimadzu, Auckland, New Zealand). The calibration curves used for copper and potassium determination are included in the Supplementary Information (Figures S1 and S2).

The morphology was examined by scanning/transmission electron microscope (S/TEM) using a Thermo Fisher Scientific model Talos F200X (Norristown, PA, USA, University of Malaga, Málaga, Spain) operated at an acceleration voltage of 200 kV. Elemental distribution maps were obtained in STEM mode using a high-angle annular dark field (HAADF) detector coupled with energy-dispersive X-ray spectroscopy (EDX).

Raman spectra were recorded at room temperature on a WITec Alpha 300 spectrometer (Oxford instruments, Abingdon, UK) equipped with a 532 nm laser source. For each sample, 60 consecutive spectra were collected with an integration time of 1 s and a laser power of 5 mW. Spectral deconvolution was performed using Lorentzian functions, and the D and G band areas were calculated by numerical integration.

2.4. Catalytic Tests

Catalytic activity tests were carried out in a fixed-bed U-shaped quartz microreactor loaded with 200 mg of catalyst. The reduction of NO was performed under a total gas flow of 100 cm³ min⁻¹, using a feed stream containing 1000 ppm of NO diluted in He.

The reaction temperature was increased from 100 °C to 460 °C at a heating rate of 3 °C min⁻¹. The concentrations of NO, NO₂ and total NOx in the outlet stream were continuously monitored using a Thermo Fisher Scientific NO–NO₂–NOx analyser.

For selected experiments, the reaction products were further analyzed by gas chromatography (GC) employing an Inficon Micro GC Fusion system equipped with two columns (Rt-Molsieve 5A and Rt-Q-Bond) and thermal conductivity detectors (TCD), with He as the carrier gas. The Rt-Molsieve 5A column was used for the detection of N₂, NO and CO, whereas N₂O and CO₂ were quantified using the Rt-Q-Bond column.

3. Results

3.1. Materials Characterization

The properties determined of the CNT samples are shown in Table 1.

Table 1. Textural properties, elemental composition and metal loadings of the CNT-based catalysts.

Sample	SBET (m2 g−1)	VP,P/P0=0.95 (cm3 g−1)	CEA * (wt.%)	HEA * (wt.%)	NEA * (wt.%)	OEA * (wt.%)	Ash † (wt.%)	CuICP-OES (%)	KICP-OES (%)
CNT	255	0.50	88.4	0.25	0.00	1.10	10.5	-	-
CNT_HNO3	310	0.70	87.9	0.35	0.01	5.10	4.79	-	-
CNT_M_BM	225	0.49	83.1	0.39	4.90	2.20	10.1	-	-
CNT@10Cu	244	0.47	n.d.	n.d.	n.d.	n.d.	n.d.	9.5	-
CNT_HNO3@10Cu	305	0.56	n.d.	n.d.	n.d.	n.d.	n.d.	9.8	-
CNT_M_BM@10Cu	207	0.42	n.d.	n.d.	n.d.	n.d.	n.d.	9.5	-
CNT@5Cu5K	187	0.43	n.d.	n.d.	n.d.	n.d.	n.d.	3.5	3.1
CNT_HNO3@5Cu5K	195	0.50	n.d.	n.d.	n.d.	n.d.	n.d.	3.9	3.6
CNT_M_BM@5Cu5K	173	0.40	n.d.	n.d.	n.d.	n.d.	n.d.	4.6	3.6
CNT@10K	164	0.41	n.d.	n.d.	n.d.	n.d.	n.d.	-	9.2

n.d.—parameter not determined. *—determined by elemental analysis. ^†—determined by thermogravimetric analysis.

Table 1 shows that the treatments carried out modify the properties of the CNT. The oxidation of the CNT with HNO₃ increases the specific surface area (S_{BET,CNT_HNO3} = 310 m² g⁻¹) compared to the original sample (S_BET,CNT = 255 m² g⁻¹) which can be attributed to the opening of the tube ends and the introduction of structural defects in the side walls [37,42]. Furthermore, treatment with HNO₃ promotes the untangling of the CNT bundles and reduces the length of the CNT [43,44], increasing the total pore volume from 0.50 to 0.70. Conversely, CNT_M_BM has a lower surface area, with S_{BET,CNT_M_BM} = 225 m² g⁻¹. This decrease can be attributed to the blocking of the pores by the N surface groups [31].

A reduction in both surface area and pore volume of the supports was observed after impregnation with metallic phases. This effect results from the introduction of metals, indicating a possible blockage of N₂ access to the internal cavities of the CNT.

Elemental analysis was performed to assess the composition of the materials, and the results are presented in Table 1. Through this technique, it was observed that the CNT had a weight fraction of approximately 90% carbon, as specified by the manufacturer. Oxidation with HNO₃ resulted in an increase in oxygen content, which reached 5.10%. The nitrogen incorporation process introduced 4.90% nitrogen into the CNT structure [38]. Through thermogravimetric analysis (TGA), it was confirmed that the CNTs presented ~10% of ashes. After oxidation with nitric acid, this percentage decreases to 4.59%. This can be attributed to the removal of impurities by the acidic treatment, resulting in a reduction in the solid residue in the nanotubes.

The metal content, specifically copper (Cu) and potassium (K), was quantified by ICP-OES and the results are showed in Table 1. The catalysts prepared with Cu and K showed metal concentrations within the expected range. For example, ICP-OES analysis confirmed that CNT@10Cu contained 9.5 wt.% Cu, which is consistent with the nominal impregnation level of 10 wt.%.

To complement the analysis of the chemical composition, Raman spectroscopy was performed to evaluate any structural changes to the CNT framework that might have been induced by mechanothermal treatment and metal incorporation. Raman analysis revealed the main spectral features typically observed in CNT-based carbon materials. Two dominant contributions were detected: the D band, centred at ~1350 cm⁻¹, which is commonly associated with structural disorder/defects in the sp² carbon framework; and the G band, located at ~1580 cm⁻¹, which is related to the in-plane vibration of graphitic domains. Additionally, a second-order band was observed at higher Raman shifts, around 2700 cm⁻¹ (the 2D or G’ band), which is characteristic of graphitic carbon materials and frequently reported for carbon nanotubes [45,46,47]. The spectra were deconvoluted using Lorentzian functions, taking into account the characteristic Raman bands of CNT materials. The D and G band areas were obtained by numerical integration. The defect indicator was calculated as the ratio between the integrated band areas (A_D/A_G). The CNT exhibited an A_D/A_G value of 0.91, while the CNT_M_BM@5Cu5K sample showed a slightly higher value of 0.99. This indicates a moderate increase in defect density following mechanothermal processing and Cu/K incorporation while preserving the graphitic structure. Similar increases in the D/G area ratio upon metal incorporation into CNTs have been reported in the literature [46]. The corresponding Raman spectra are provided in the Supplementary Information (Figure S3).

3.2. Catalytic Results

3.2.1. Catalytic Reduction of NO by Cu-Based Catalysts Supported on Functionalized Carbon Nanotube

Catalytic tests were carried out to evaluate the influence of support functionalization on NO reduction. The resulting catalytic performance profiles are shown in Figure 1.

Figure 1. Evolution of dimensionless NO concentration in the presence of Cu-based catalysts supported on functionalized carbon nanotube.

Initially, the catalysts underwent an isothermal pre-treatment at 100 °C until equilibrium was reached. During this stage, the previously adsorbed NO was released, resulting in an initial NO concentration at the reactor outlet greater than the amount initially fed [19].

Carbon nanotubes (CNT) do not present catalytic activity for the NO reduction (X_NO,CNT = 2%). However, when copper is impregnated in this material, it is possible to notice an increase in catalytic activity (X_NO,CNT@10Cu = 42%), as expected [18,19,34]. CNT_HNO₃@10Cu and CNT_M_BM@10Cu achieved NO conversions of 60% and 78%, respectively, indicating that the catalyst structure is a key factor in determining its performance. Xue et al. [48] observed that an activated carbon (AC) catalyst treated with H₂O₂ and impregnated with Cu achieved good catalytic performance, highlighting the importance of oxygen-functionalized surfaces and NO adsorption on AC in generating active sites. Additionally, Zhu et al. [49], reported that activated carbon fibres (ACF) treated with HNO₃ and doped with 9 wt.% CeO₂ showed better catalytic activity, thus showing that this type of support is promising for NO SCR.

Nitrogen functionalities also contribute significantly to catalytic performance [19,30,50,51]. Ramalho et al. [19] found that Cu impregnation in N-doped activated carbon could improve catalytic activity. The addition of N demonstrates a significant increase in the catalytic activity of activated carbons, indicating that nitrogen groups play a crucial role in forming nitrogen in this type of reaction. The importance of pyridinic nitrogen (N-6) sites in NO reduction has been reported by Wang et al. [4], who identified these species as highly active for direct NO decomposition at 500 °C. In contrast, non-doped carbon materials exhibited very limited NO conversion under similar conditions. In a related study, Lin et al. [52] examined the influence of nitrogen precursors and surface modification on SCR performance over activated carbons, showing that nitrogen-doped materials achieved significantly higher NOx conversions (up to 66%) compared to undoped samples.

A detailed study of N-doped carbon nanotubes has been reported [38]. Briefly, N-doped carbon nanotubes present 3 distinct peaks in the N1S spectrum, as revealed by XPS analysis. The N6 group (pyridine group) was identified as the most abundant species, representing around 60% of the total nitrogen, followed by the N5 group (pyrrole group) at around 30%, while NQ group (quaternary nitrogen group) is present in smaller quantities (~10%) [38].

Carbon materials with nitrogen functionalities tend to outperform their undoped counterparts in NO reduction reactions. The presence of surface nitrogen species modifies the electronic structure of the carbon matrix, thereby promoting NO adsorption and facilitating its subsequent transformation into molecular nitrogen [19,30].

It can be concluded that factors such as the introduction of a metallic phase and the inclusion of N or O-surface groups play a critical role in NO reduction performance.

3.2.2. Catalytic Reduction of NO by Bimetallic Catalysts Supported on Functionalized Carbon Nanotube

The improved NO conversion observed for Cu catalysts supported on modified CNT highlights the decisive role of surface functionalization, as the corresponding Cu/CNT system showed significantly lower activity. Potassium was found to have the ability to influence the catalytic behaviour of carbon during in the NO reaction [24,35]. Therefore, the influence of copper (Cu), potassium (K) and the combination of Cu-K was evaluated, with the aim of further enhancing their catalytic properties (Figure 2).

Figure 2. Evolution of the dimensionless concentration of NO (a) and NO conversion and temperature to reach full conversion (b) in the presence of Cu- and K-based catalysts supported on carbon nanotubes.

Analyzing Figure 2, it becomes evident that the copper (Cu) and potassium (K) monometallic catalysts demonstrate a relatively low catalytic activity for the reaction temperature of 460 °C (X_NOx,CNT@10Cu = 26% and X_NOx,CNT@10K = 60%), when compared to bimetallic catalysts, which exhibited a higher catalytic activity (X_{NOx,CNT@5Cu5K} = 79% at 460 °C). Thus, the bimetallic formulation proved to be promising for the reduction of NO.

The higher NO conversion observed for the bimetallic Cu–K catalysts compared to their monometallic counterparts indicates that potassium promotes Cu-based catalysts rather than simply contributing to them additively. This result demonstrates that potassium enhances the catalytic performance of Cu-containing catalysts.

Additionally, it is highlighted that functionalized catalysts demonstrate substantially higher NO conversion rates when contrasted with unmodified supports. The CNT@5Cu5K sample exhibited 79% NO conversion at 460 °C, while CNT_HNO₃@5Cu5K achieved 100% total conversion at 430 °C, and CNT_M_BM@5Cu5K achieved complete conversion at 360 °C. It can be concluded that the catalytic efficiency of carbon materials for NO reduction is linked to their surface chemistry. The incorporation of nitrogen into the structure has a significant effect on this relationship, highlighting its role in enhancing the catalytic performance of the material; high NO conversion (82% at 260 °C) has been achieved over nitrogen-enriched porous biochar, as described by Li et al. [53], who attributed this performance to the presence of pyridinic (N-6) functionalities acting as active sites in the denitrification process. Furthermore, the incorporation of certain metals, notably potassium, emerges as a crucial factor in optimizing performance. The combined effect of surface nitrogen functionalities and Cu–K incorporation plays a decisive role in governing NO reduction performance, underscoring the strong interplay between surface composition and catalytic behaviour.

3.2.3. Stability Test

As the stability of the catalyst is a key factor in the overall assessment of its performance, the catalyst with the highest efficiency was evaluated in this study.

Figure 3 reports the molar flowrates of the reaction products obtained over CNT_M_BM@5Cu5K. The upper temperature was fixed at 340 °C, corresponding to nearly 90% NO conversion, to define the conditions for the stability evaluation.

Figure 3. Molar flowrates vs. the reaction temperature in the reduction of NO with CNT_M_BM@5Cu5K.

Under the employed reaction conditions, NO may undergo different transformation pathways, resulting in the formation of distinct products. The main reaction routes considered are shown in Equations (1) and (2) [54]:

$$2\mathrm{N}\mathrm{O}+\mathrm{C}\to {\mathrm{N}}_2+{\mathrm{C}\mathrm{O}}_2$$

(1)

$$2\mathrm{N}\mathrm{O}+\mathrm{C}\to {\mathrm{N}}_2\mathrm{O}+\mathrm{C}\mathrm{O}$$

(2)

The results presented in Figure 3 show that when the reaction temperature increases, the amounts of CO₂ and N₂ produced increase. However, the amount of CO₂ is always slightly higher than that of N₂. Furthermore, small amounts of N₂O are formed during the reaction. However, at higher temperatures, N₂O tends to decompose into N₂, thus reducing its quantity [55]. CO was not detected in the outlet stream by GC–TCD under the experimental conditions used. The Micro-GC (TCD) response to CO was previously verified through calibration runs, in which a clear CO peak was observed. Therefore, if CO was formed during NO reduction, its concentration was below the detection limit of the GC–TCD method and/or it was rapidly converted to CO₂. If CO is formed as an intermediate, it may be rapidly converted to CO₂ via the following reaction: $\mathrm{C}\mathrm{O}\left(\mathrm{g}\right)+\mathrm{C}\left[\mathrm{O}\right]={\mathrm{C}}_{\mathrm{f}}+{\mathrm{C}\mathrm{O}}_2\left(\mathrm{g}\right)$ [56]. The equation describes how carbon monoxide (CO) interacts with an oxygen complex $\left(\mathrm{C}\left[\mathrm{O}\right]\right)$ on a carbon-based material to form carbon dioxide (CO₂) and creating a free active site $\left({\mathrm{C}}_{\mathrm{f}}\right)$ on the surface. This active site can then participate in subsequent reactions. This reaction is significant in processes involving carbon surfaces, such as catalytic reactions or carbon gasification, because it produces CO₂ as a by-product and regenerates active sites on the carbon surface for further reactions [56]. Therefore, temperature and reaction conditions play essential roles in the formation of products, leading to the predominance of CO₂ and N₂, which suggests that Equation (1) is mainly responsible for the observed products.

Xue et al. [57] reported that Cu-loaded activated carbon yielded only N₂ and CO₂ during NO reduction, with CO₂ being generated in higher quantities than N₂.

The NO conversion profile for the CNT_M_BM@5Cu5K and the products resulting from the reduction of NO over 87 h at 340 °C is shown in Figure 4.

Figure 4. Dimensionless concentration of NO on a stability test at 340 °C for 87 h (a); evolution over time of reaction products in the reduction of NO during the stability test (b).

Figure 4a shows that the sample maintained a 90% conversion during 41 h of reaction. However, after this time, the conversion of NO begins to decrease. This suggests that the carbon present in the catalyst, which acts as a reducing agent, is being consumed over time. The loss of activity can be explained by the loss of carbon in close contact with the metal catalysts. This can affect the catalyst efficiency, as it reduces the amount of material available to promote the NO reaction. Wang et al. [58] investigated catalysts composed of Fe nanoparticles incorporated into N-doped CNT, grown on the surface of ZSM-5 using an internal growth method. They verified that the catalyst achieved complete NO conversion during 20 min of reaction. However, after this period, they observed a loss of catalytic performance. This reduction in activity was attributed to the gradual consumption of CNT. With the decrease in carbon, the oxygen present in the iron was unable to be transferred to the carbon, resulting in the deactivation of the catalyst.

Figure 4b shows how the flow rates of N₂ and CO₂ changed during the stability test. No N₂O was detected under these conditions, consistent with its thermal decomposition at elevated temperatures. CO was also not observed in the outlet stream by GC–TCD. During the first 41 h of operation, the production rates of N₂ and CO₂ remained stable, indicating sustained catalytic performance. Beyond this period, however, both products gradually decreased, reflecting progressive catalyst deactivation. This behaviour aligns with the NO conversion profile presented in Figure 4a, confirming the loss of catalytic activity over an extended period of operation.

The XPS results of Cu 2p(3/2), N1S and K2p of the CNT_M_BM@5Cu5k sample before and after the catalytic reaction are shown in Figure 5 and Table 2.

Figure 5. X-ray photoelectron spectroscopy analysis of the sample CNT_M_BM@5Cu5K before (a–c) and after (d–f) the catalytic reaction in the Cu 2p(3/2) region (a,d), N1S region (b,e) and K2p region (c,f).

Table 2. Peak position (eV) and peak area for the core levels Cu 2p(3/2) N1s and K2p of the samples CNT_M_MB@5Cu5K before and after the catalytic reaction.

Element	Core-Level	Properties	CNT_M_BM@5Cu5K (Before Reaction)	CNT_M_BM@5Cu5K (After Reaction)
Cu0/Cu+	2p(3/2)	Binding energy (eV)	932.4	933.2
		Peak area (%)	45.7	55.7
Cu2+	2p(3/2)	Binding energy (eV)	934.3	935.5
		Satelite peak 1	939.5	940.5
		Satelite peak 2	943.3	943.4
		Peak area (%)	54.3	44.3
N6	1s	Binding energy (eV)	398.2	398.3
		Peak area (%)	61.8	55.6
N5	1s	Binding energy (eV)	399.8	399.5
		Peak area (%)	27.6	35.9
N-Q	1s	Binding energy (eV)	401.9	401.2
		Peak area (%)	10.6	8.5
K+	2p(3/2)	Binding energy (eV)	293.1	292.7
		Peak area (%)	0.74	1.06
	2p(1/2)	Binding energy (eV)	295.7	295.2
		Peak area (%)	0.37	0.53

Before the catalytic reaction, CNT_M_BM@5Cu5K presents a higher amount of Cu²⁺ (54.3%) on its surface and a lower amount of Cu⁰/Cu⁺ (45.7%), as shown in Figure 5 and Table 2, indicating copper oxidation in the catalyst. The Cu 2p(3/2) components at 932.4–933.2 eV and 934.3–935.5 eV are assigned to Cu⁰/Cu⁺ and Cu²⁺ species, respectively, in agreement with literature reports and standard XPS reference data [19,24,59]. After the catalytic test, the area of the Cu²⁺ peak decreased (44.3%), suggesting this may be that Cu²⁺ species during the NO reaction. Initially, the catalyst presents good catalytic activity, as it has a larger amount of Cu²⁺ on the surface [19,24]. The N1S spectrum show three nitrogen peaks on its surface, pyridine (N6), pyrrole (N5) and quaternary (N-Q). The corresponding binding energies at 398.2–398.3 eV (N6), 399.5–399.8 eV (N5), and 401.2–401.9 eV (N–Q) are consistent with previous XPS studies and standard reference data [19,59,60]. After the catalytic test, an increase in N5 (from 27.6% to 35.9%) and a decrease in N6 (from 61.8% to 55.6%) are observed. This nitrogen group (N6) provides greater basicity to the surface [19,25,60]. A decrease in N6 also affects catalytic performance, as it plays an important role in promoting catalytic activity. Li et al. [53] reported N6 played an important role in increasing denitrification activity of N-doped porous biochar. Wang et al. [4] investigated the influence of N₂ on SCR and observed a correlation between NOx conversion and the presence of the N6 group, indicating that catalytic activity is mainly benefited by the introduction of N6 groups instead of other N-groups.

Regarding potassium, the characteristic K2p doublet remains clearly detectable after the catalytic test. The K 2p(3/2) and K 2p(1/2) components at 292.7–293.1 eV and 295.2–295.7 eV, respectively, are assigned to K⁺ species in agreement with literature and standard XPS reference data [59,61]. After the reaction, an increase in the relative peak areas of K 2p(3/2) (from 0.74% to 1.06%) and K 2p(1/2) (from 0.37% to 0.53%) is observed. This increase in surface potassium content is consistent with carbon consumption and surface restructuring during the catalytic test. These processes lead to an increased exposure of surface species rather than potassium loss or chemical transformation. These results suggest that potassium remains on the catalyst surface and acts as a stable promoter during NO reduction, which is consistent with the reported role of alkali metals in heterogeneous catalysis [61].

Table 3 shows that, after the catalytic reaction, the catalyst presents changes in its properties. A mass loss of approximately 2% led to changes in textural properties, such as a reduction in specific surface area. The ICP-OES analysis showed that the Cu and K contents in the catalyst remained practically unchanged before and after the stability test, indicating that the amounts of these metals did not undergo significant variations.

Table 3. Textural properties as well as Cu, K and N content determined by ICP-OES and XPS of CNT_M_BM@5Cu5K before and after the stability test.

	CNT_M_BM@5Cu5K (Before)	CNT_M_BM@5Cu5K (After)
Catalyst (mg)	0.20	0.18
SBET (m2g−1)	173	126
VP,P/P0=0.95 (cm3g−1)	0.40	0.38
CuICP-OES (%)	4.6	4.4
CuXPS (%) *	2.4	3.1
KICP-OES (%)	3.6	3.7
KXPS (%) *	2.3	2.9
CXPS (%) *	84.8	82.4
NXPS (%) *	5.5	4.5
OXPS (%) *	5.0	7.1

* XPS values correspond to weight fractions calculated from atomic percentages provided by the XPS analysis.

The surface composition of the catalysts has been studied using XPS, which allows the composition and nature of the chemical bonds between atoms on the surface of the material to be determined down to depths of 5–12 nm. This analysis revealed a 2.4% decrease in surface carbon content, accompanied by a proportional increase in oxygen concentration. There was also a slight enrichment in the surface content of copper (Cu) and potassium (K), whereas the nitrogen content decreased after the stability test. These compositional changes are consistent with carbon consumption during the prolonged reaction, resulting in greater exposure of the metallic phases at the surface. The increase in oxygen content (2.1%) can be attributed to the formation of additional oxygen-containing surface groups during the reaction. The reduction in nitrogen content (N) may be linked to the interaction of surface nitrogen groups with the catalyst support, possibly influencing the reduction of nitrogen oxides (NO). These changes suggest that, in addition to mass loss, transformations in the composition of the catalyst surface influence its catalytic activity.

In addition, as shown in Table 3 and Figure 3, the catalyst undergoes some degradation as the temperature increases. This occurs because some of the O-containing surface groups present in the catalyst decompose when subjected to high temperatures, resulting in the formation and release of CO₂. The catalyst showed a mass loss of 0.02 mg, corresponding to the amount of CO₂ generated during the process, which represents the formation of 0.02 mg of carbon.

To further investigate possible structural modifications of the carbon framework during long-term operation, Raman spectroscopy was performed for CNT_M_BM@5Cu5K after the stability test. Compared to the fresh catalyst, the spent sample exhibited an increase in the A_D/A_G ratio from 0.99 to 1.07 (Figure S3), indicating additional disorder/defect generation in the carbon structure. This result is consistent with carbon consumption and surface restructuring under reaction conditions, which may contribute to the observed catalyst deactivation over extended time on stream.

For the CNT_M_BM@5Cu5K catalyst, detailed analyzes were performed on three different samples by transmission electron microscopy (TEM-EDX) before the reaction and at the end of the reaction (Figure 6). TEM-EDX was used to evaluate the catalyst’s morphological evolution during the stability test, with a particular focus on metal particle agglomeration and carbon consumption associated with NO reduction.

Figure 6. TEM-EDX micrograph of CNT_M_BM@5Cu5K catalysts before reaction (a) and end of the reaction (b).

Before starting the reaction, the images obtained by TEM allowed us to establish a baseline on the structure of the catalyst. After the stability test, new analyzes were conducted to identify any structural changes or degradation the catalyst may have undergone. These images are presented in Figure 6, where the morphological evolution of the material can be observed. In addition to TEM images, EDX (energy dispersive X-ray spectroscopy) was used to study the distribution of metals in the catalyst support. Using EDX, it was possible to observe how the metals are dispersed in the catalyst matrix. Elemental distribution maps obtained by EDX enable the dispersion of metals and the formation of agglomerates after the reaction to be assessed. The corresponding micrographs are presented in Figure 6.

Figure 6 shows TEM-EDX micrographs of the CNT_M_BM@5Cu5K catalyst. At the magnifications used, the images allow the identification of changes in metal dispersion and the formation of particle agglomerates after the stability test, together with evidence of carbon consumption.

In Figure 6a, the TEM image reveals that some metallic particles are dispersed throughout the support. EDX analysis confirms this distribution, showing the presence of dispersed metallic elements. After the stability test (Figure 6b), it is possible to observe an agglomeration of metallic particles. This agglomeration is correlated with C consumption during the reaction. With less C to maintain the reaction, the particles cluster on the surface, thus impairing the catalytic activity. This clustering is clearly visible on the EDX maps, which show regions densely populated with metals and at the same time the amount of C close to the metal particles appears to decrease, in contrast to the initial scattering observed.

The agglomeration of metal particles and the consumption of C have a significant impact on the catalytic activity of the material. Catalytic activity depends on the available surface area to interact with reactants. Smaller, well-dispersed particles on the support offer greater surface area. When particles clump together, the available surface area decreases. As a result, fewer active sites are available to catalyze reactions, leading to a decrease in catalytic activity.

Thus, Figure 6 shows how the initial dispersion of metallic particles transforms into agglomerates after the stability test and the consumption of C due to the reduction of NO, and how this structural change negatively affects the catalytic efficiency of the material.

The catalytic behaviour observed in this study results from the interaction between the physicochemical properties of the carbon support and the characteristics of the active metal phase. The results suggest that, in addition to textural features, combining a nitrogen-doped carbon matrix with a Cu–K bimetallic formulation is crucial for promoting NO conversion. Furthermore, stability tests reveal that catalyst deactivation is primarily due to carbon consumption and surface restructuring. These processes progressively limit the availability of reactive carbon near the metal particles and favour agglomeration of the particles, as demonstrated by XPS, Raman and TEM/EDX analyses.

To provide a broader context for the catalytic performance achieved in this work, Table 4 presents a qualitative comparison with representative bimetallic catalysts reported in the literature under different operating conditions.

Table 4. Comparison of representative carbon-based bimetallic catalysts reported in the literature for NO reduction under different operating conditions.

Catalyst	Carbon Material	NO Concentration (ppm)	Reducing Agent	Key Performance	Stability	Ref.
CNT_M_BM@5Cu5K (this work)	N-doped CNT	1000	none	100% NO conversion at 360 °C	41 h (340 °C)	-
AC_M_BM@5CuK	N-doped AC	1000	none	100% NO conversion at 410 °C	40 h (376 °C)	[36]
Cu-K-O/AC (Cu/K = 2)	AC	2000	none	100% NO conversion at 390 °C	5 h (400 °C)	[24]
Cu-Ce/CNT	CNT	250	CO	96% NO conversion at 220 °C	-	[62]
Fe-K/AC	Biomass-derived activated carbon	300	none	~100% NO conversion at 300 °C (with 3% O2)	~250 min at >80% conv.	[34]

Table 4 compares the catalytic performance obtained in this study with data from the literature, including previously reported studies on carbon-based bimetallic catalysts. The results suggest that combining Cu and K on N-doped CNT enables complete conversion of NO at a lower temperature (360 °C) than Cu–K catalysts on N-doped AC (390 °C). This demonstrates the impact of the carbon support on NO reduction performance.

4. Conclusions

This study evaluated the performance of carbon nanotube-based catalysts in SCR-C conditions, focusing particularly on the impact of surface functionalization and metal incorporation on NO reduction.

The results confirm that CNT’s catalytic behaviour is strongly influenced by surface chemistry. Nitrogen functionalization, particularly the presence of pyridinic (N-6) species, markedly improves NO conversion, and oxygen-containing groups also affect activity. Furthermore, incorporating metallic phases significantly improves performance, with the Cu–K bimetallic formulation clearly outperforming monometallic systems. Of the catalysts investigated, CNT_M_BM@5Cu5K achieved complete NO conversion at 360 °C.

Product analysis revealed that the predominant stable products were N₂ and CO₂, whereas N₂O was only present in small amounts and decreased at higher temperatures. Stability tests revealed sustained catalytic activity for 41 h at 340 °C, followed by gradual deactivation. Structural and surface analyses suggest that this loss of activity is associated with carbon consumption near the metallic particles, resulting in partial agglomeration and reduced availability of reactive carbon sites.

Overall, combining N-doped CNT with a Cu–K bimetallic system is an efficient approach for NO reduction under SCR-C conditions. These findings emphasize the importance of tailoring carbon surface chemistry and metal–support interactions to optimize catalytic performance.