Investigating the Role of Fe‐Pyrrolic N4 Configuration in the Oxygen Reduction Reaction via Covalently Bound Porphyrin Functionalized Carbon Nanotubes

Atomically dispersed iron–nitrogen–carbon catalysts are promised, low‐cost, and high‐performance electrocatalysts for the Oxygen Reduction Reaction (ORR) in fuel cells. However, most Fe–N–C materials are produced via pyrolysis at a high temperature and it is difficult to characterise the precise Fe–N configurations. This can lead to confusion surrounding the best chemical and coordination environment for Fe and understanding the subsequent ORR mechanisms. In this work, Fe porphyrin is used to produce a specific Fe–N environment, therefore allowing the role and activity of this environment to be studied. Carbon nanotubes (CNTs) are covalently functionalized with iron 5,10,15,20‐triphenylporphyrin (FeTPP) motifs via aryl diazonium methodology, enabling the exact role of only the Fe‐Pyrrolic N4 configuration of FeTPP in ORR to be studied and better understood. Upon covalent functionalization, a high electrochemical active site density of 1.12 × 1015 sites cm−2, approximately six‐fold more than that of noncovalently functionalized samples with 12.7% electrochemical active site. The heightened active site density and superior electrochemical active site utilization (12.7%) lead to the more favorable 4‐electron pathway for the ORR. Furthermore, a preliminary discussion regarding the selectivity of the ORR pathway is initiated.


Introduction
The Oxygen Reduction Reaction (ORR) via a four-electron transfer mechanism plays a crucial role in energy storage and conversion technologies such as fuel cells and metal-air batteries. [1]OI: 10.1002/adfm.202311086Despite exhibiting efficient performance during the ORR process, the high cost and poor poison tolerance of platinum group metal (PGM) catalysts present a significant challenge for large-scale applications. [2]In the last decades, atomic dispersed Fe atoms anchored in nitrogen-doped carbon materials (Fe-N-C) have emerged as promising candidates to replace Pt-based materials on the cathode of proton exchange membrane fuel cells (PEMFCs) and anion-exchange membrane fuel cells (AEMFCs). [3]Besides, many studies have reported the outstanding ORR performance of Fe nanoparticles/Fe-N-C composite materials as well as the synergistic effects of Fe-based dual atom electrocatalysts. [4]However, most of these Fe single-atom catalysts (SACs) were obtained by high-temperature pyrolysis, which brings some disadvantages.These include complex structural evolution, diverse coordination environments around the Fe-N centers, various structural defects and unselective N-type generation in the N-doped carbon supports. [5]These factors present significant challenges in understanding the associated ORR mechanisms and debates regarding the active catalytic sites of Fe SACs persist. [6]nstead of the uncertain Fe-N x species in the pyrolysed Fe-N-C materials, Fe macrocycles, which mimic hemoglobin and cytochrome c oxidase catalysts with well-defined Fe-N 4 environments, have also been applied as heterogeneous ORR catalysts. [7]Since the initial report by Jasinski using phthalocyanines (Pc), [8] other organic molecules such as porphyrins, [9] corroles, [10] and their derivative covalent organic framework, [11] have also been used to study reaction mechanisms and associated structure-function relationships. [12]Unfortunately, these macrocyclic systems do not exhibit high utilization of the metallic active sites, due to ease of aggregation, which also causes poor conductivity and instability of the catalysts. [13]xtensive efforts have been made to minimize this observed macrocycle aggregation, to improve conductivity and stability.Immobilizing molecules on carbon supports via - interactions has been found to offer moderate improvements in the conductivity and stability of the catalyst, however, it does not solve the problem of aggregation to achieve a uniform distribution on the carbon support. [14]Constructing porous frameworks such as covalent-organic frameworks (COFs), metal-organic frameworks (MOFs) and aerogel from these macrocycles can form a long-range - interaction between frameworks and carbon support, which exhibits good stability during the ORR. [15]he design of covalent-organic polymers (COPs) and metalorganic polymers (MOPs) with porphyrin or phthalocyanine are also effective and convenient approaches for synthesizing organic polymer electrocatalysts with improved stability. [16]owever, the process of polymerization to form frameworks is difficult to control, so these frameworks typically show a layered structure, through - stacking, that limits the exposure of the active sites. [17]In addition, studies have confirmed that the conventional stacking FePc framework on graphene causes the 3d orbits of the Fe center to shift to graphene, which leads to the thermodynamically unfavorable ORR process. [18]ovalent grafting of molecules onto the surface of a carbon support is an alternative method that provides better electrochemical stability by forming a covalent bond between the macrocycle and carbon support. [19]Furthermore, a covalent functionalization approach enables a complete dispersion of the molecules across the carbon surface, thereby avoiding aggregation. [20]By promoting a uniform distribution of active sites, covalent functionalization facilitates efficient catalytic reactions and improves the utilization of the catalyst, resulting in improved performance and higher turnover frequency (TOF) values. [21]Azide-alkyne cycloadditions and amide bond-forming reactions are two common methods to achieve covalent functionalization that both exploit the carboxyl groups (─COOH) at oxidized carbon defects of nanomaterials. [22]However, two or three steps are needed to treat the carbon supports by these two methods, and some harsh reactants such as H 2 SO 4 , HNO 3, and SOCl 2 are necessary for some steps.These reactants strongly etch the surface of the carbon material, leading to further defects and oxygencontaining groups forming on the surface that result in multiple additional structural contributions to ORR performance. [23]s such, there is still a need for a simple approach to covalent functionalisation that simultaneously minimizes macrocycle aggregation, and maximizes active site utilization and catalyst stability.
Herein, carbon nanotubes (CNTs) were covalently functionalized with iron-complexed porphyrin rings (iron 5-(4aminophenyl)−10,15,20-(triphenyl) porphyrin (FeTPP)) via an aryl diazonium methodology that has previously been used to covalently functionalize CNTs and graphene substrates with different arene components. [24]With the advantage of high conductivity, low-defect CNTs were chosen as the carbon substrate in this work. [25]Porphyrins that did not undergo covalent functionalization could still become associated with the CNT via - stacking; importantly any unreacted porphyrin was removed from the samples by ultrasonic cleaning from the surface of CNTs due to the low affinity from its non-planar structure. [26]For comparative purposes, a non-covalently functionalized reference sample (noncov FeTPP/CNT) was prepared by mixing FeTPP with CNTs when preparing the ink.The ORR performance of catalysts was evaluated in the alkaline media, where the covalently functionalized sample showed higher site density and a more selective 4 e− pathway.

Preparation and Characterization of the Covalent FeTPP-CNT Sample
5-(4-Aminophenyl)−10,15,20-triphenylporphyrin (TPP-NH 2 ) and the corresponding iron complexes (FeTPP-NH 2 ) were synthesized according to a previous report (synthetic details provided in the Supporting Information). [27]The identity and purity of TPP-NO 2 and TPP-NH 2 were established by 1 H NMR and 13 C NMR spectroscopy (Figures S1-S4, Supporting Information) and high-resolution mass spectrometry (HRMS) (Figures S5 and  S6, Supporting Information).The NMR spectroscopic results and HRMS results of these two molecules match the results in previous reports and indicate the purity of organic molecules.The paramagnetic iron complexes can not be analysed by NMR spectroscopy, hence Matrix-assisted laser desorption/ionizationtime of flight (MALDI-TOF) mass spectrometry, in addition to FTIR and UV-vis spectroscopy, was instead used for characterization.Iron 5-10,15,20-triphenylporphyrin (FeTPP) was also synthesized, lacking the primary aryl amine, to prepare the analogous non-covalently functionalized CNT material.The MALDI-TOF mass spectrometry results of FeTPP and FeTPP-NH 2 match the theoretical mass (Figures S7 and S8, Supporting  Information).
To further confirm the structure of the Fe complexes, FTIR and UV-vis spectroscopy tests were performed.28a,29] In addition, a very weak and broad peak around 3360 cm −1 , which is seen on the curve of TPP-NH 2 and FeTPP-NH 2 , can be ascribed to the primary amine on the benzene ring.Other bands at ≈1350 cm −1 , 1400-1700 cm −1 , and 2900-3000 cm −1 can be ascribed to stretching vibrations of C═N, C═C, and C─H bonds and the band at ≈800 cm −1 is attributed to the aromatic C─H Bending. Figure S10 (Supporting Information) shows the UVvis spectrum of four different compounds in dichloromethane.27a,30] Finally, the range of characterization techniques presented provides strong evidence to support the synthesis of FeTPP-NH 2 and FeTPP.Electron paramagnetic resonance (EPR) spectroscopy was used to study the unpaired electrons within FeTPP and FeTPP-NH 2 (Figure S11, Supporting Information).The ferric high spin (S = 5/2) square pyramidal coordination (according to g = 5.8 and g = 2) with some contribution of low spin (S = 1/2) (according to g = 3.3-3.4)signals appear in our FeTPP sample spectrum.This matches well with the results of FeTPPCl in previous reports, indicating there is possibly an axial Cl − coordination structure around the FeN 4 in our sample. [29]FeTPP-NH 2 with a sharp axial signal (g = 5.8) closely resembles FeTPP.To stabilize the Fe center in the porphyrin structure, coordination with Cl − or other ligands to become Fe (III) is necessary.This is because Fe(II) is inherently unstable and prone to rapid oxidation to ferric porphyrin in the presence of air. [32]However, there is some evidence that the axiallycoordinated Cl − will leave the FeN 4 site during the first few CV cy-cles of the electrochemical test with the resulting oxidation state changing from III to II. [29,33]he covalently functionalized CNTs based on the high-purity TPP-NH 2 and FeTPP-NH 2 macrocyclic systems described above were achieved via an aryl diazonium methodology (Figure 1a).SEM was used to observe and compare the surface morphology of CNT samples before and after the functionalization.The SEM results (Figure 1b and Figure S12, Supporting Information) revealed that the functionalized CNTs retain the morphology of pristine CNTs.We did not observe any small particles or crystals from the aggregated molecules under the TEM (Figure S13 Supporting Information).Moreover, HRTEM in Figure 1c clearly shows a thin organic layer (≈1 nm) covering the surface of the CNTs, suggesting the majority of molecules are dispersed on the surface of CNTs as a thin layer, without large-scale aggregation or polymerization.The high-angle annular dark field-scanning transmission electron microscopy (HAADF-STEM) image and energy-dispersive X-ray spectroscopy (EDS) mapping confirmed the molecule was uniformly dispersed on the surface of CNTs.Furthermore, as revealed by XRD patterns in Figure S14 (Supporting Information), one strong peak located at about 26.6°and one weak peak at about 54.1°in the functionalized CNT sample is attributed to the (002) peak and (004) peak of CNTs. [34]There are no peaks observed from Fe atoms or macrocycles in these spectra, indicating no aggregated crystals became physically mixed within the functionalized CNTs.Overall, we cannot observe any molecular aggregation from the SEM, TEM and XRD results.To confirm the thin organic layer on the surface of CNTs is achieved by covalent functionalization and to probe its stability, a control experiment was designed wherein diazotization could not occur.To this end, FeTPP-NH 2 and CNTs were exposed to the standard functionalization protocol, except in the absence of isoamyl nitrite.The reaction mixture was then washed by five centrifugal-ultrasonic cycles with DMF and finally collected by suction filtration with MeOH.The XPS characterization was employed to characterize the surface elements and their binding energy.Their XPS results (Figure S15a,b, Supporting Information) clearly show that there is no N peak in the N 1s spectrum and no Fe peak in the Fe 2p spectrum of the control experiment sample.This suggests that any FeTPP-NH 2 molecules that might have bound to the CNT surfaces, in the absence of diazonium formation, via alternative - stacking interactions were effectively removed during the protocol. [26]he XPS N 1s spectra of covalently functionalized FeTPP-CNTs and TPP-CNT samples were further analyzed for information on the types of bonding exhibited.In Figure 2a,b, the peak at about 399.7 eV in the FeTPP-NH 2 and TPP-NH 2 organic powder sample is attributed to the NH 2 group. [35]This peak disappears in the FeTPP-CNT and TPP-CNT samples, following the covalent functionalization reaction, which suggests the amino groups are successfully diazotized by isoamyl nitrite and may form a C─C bond between porphyrin and CNTs. The 0.2 eV energy shift may come from the strong - interaction by the CNTs. [37]The pyrrolic N peak at about 400.2 eV and the ─C═N─ peak at 398.3 eV are typical for the porphyrin structures. [38]The peak at about 401.4 eV of FeTPP-NH 2 and TPP-NH 2 is assigned to a - * satellite peak.35b,36] Furthermore, this peak disappears after the molecule is dispersed and grafted onto CNTs, suggesting the aggregation was limited after the reaction.The Fe 2p peak in Figure S16b (Supporting Information) is not further fitted and analyzed, due to its highly complex nature, whilst the peak of FeTPP-CNT is very weak.However, there is a clear Fe 2p3/2 peak at about 710 eV and a Fe 2p1/2 peak at about 723 eV in the FeTPP-CNT sample, illustrating that FeTPP molecules were successfully functionalized on CNTs (Figure S16b, Supporting Information).From the XPS results, the amino group of FeTPP-NH 2 can be diazotized by isoamyl nitrite and subsequently lost as nitrogen gas during the reaction.Combined with the control experiment results, it can be concluded that the FeTPP and TPP molecules were grafted onto the surface of CNTs via the aryl diazonium reaction.
This functionalization was further confirmed by various spectroscopy techniques.For the Raman spectrum of CNT, TPP-CNT, and FeTPP-CNT in Figure 2c, the two bands located at 1350 and 1590 cm −1 can be assigned to the disordered carbon (D band) and the graphitic carbon (G band), respectively, and the intensity ratio of I D /I G represents the carbon type change after the reaction.A slight increase of the I D /I G value from 0.38 of pristine CNT to ≈0.43 of TPP-CNT and 0.45 of FeTPP-CNT was observed after the reaction, which likely arises from the surface lattice after the covalent grafting. [39]The UV-vis spectra are shown in Figure 2d: the pristine CNT shows no absorbance peaks from 350 to 750 nm while the TPP and FeTPP samples show a strong Sband at ≈406 nm.The FeTPP-CNT and TPP-CNT samples show a weak S-band with a slight red shift, which can be attributed to the strong interaction between organic molecules and CNT, suggesting the molecule is covalently grafted onto the CNT. [39]-ray absorption spectroscopy was further utilized to assess the oxidation state and coordination environment in FeTPP-CNT.By comparing with the model sample Fe (III)Pc-Cl in Figure 2e, the FeTPP-CNT shows a similar adsorption edge indicating the +3 oxidation state of the Fe.A small peak is also observed at about 7114 eV in both FeTPP-CNT and Fe (III)Pc-Cl, owing to the 1s to 4p z electronic transition, suggesting a distorted square planar geometry which is possibly caused by the axial Cl − group on the FeTPP. [40]Fourier transformed EXAFS comparison in Figure 2f indicates the presence of the Fe─N bond in the FeTPP-CNT in the first shell, suggesting that most of the Fe species in the FeTPP-CNT sample exist as isolated single atoms.It is noted that there is some small difference between FeTPP-CNT and Fe(III)PcCl, which may be caused by the difference in molecule structure. [36]he macrocyclic loading on the CNT surface was confirmed by various tests.The iron content of FeTPP-CNT was determined to be ≈0.29 ± 0.05 wt% and 0.17 ± 0.06 at% using both ICP-MS and XPS, respectively (Table S1, Supporting Information).The N content of TPP-CNT is found to be close to that of FeTPP-CNT (Table S1, Supporting Information), and the TGA results (Figure S17, Supporting Information) also show similar weight loss at 600 °C, suggesting a similar molecular loading between FeTPP-CNT and TPP-CNT.

ORR Performance of the Covalent FeTPP-CNT Electrode
The electrochemical performance of covalently functionalized FeTPP-CNTs electrodes was investigated in a three-electrode system with a rotating disk electrode (RDE) in N 2 and O 2 -saturated 0.1 m KOH.To assess the impact of the covalent bond on the electrode's electrochemical performance, a non-covalent functionalized FeTPP/CNT sample (noncov FeTPP/CNT) was also prepared by simply mixing CNT with FeTPP molecules (mass ratio 31:1) when preparing the electrode ink.The theoretical Fe content of noncov FeTPP/CNT was approximately 0.26 wt%, which is comparable to the Fe content of the covalently functionalized FeTPP-CNT sample (cov FeTPP-CNT).The ICP test result of this noncov FeTPP/CNT sample shows a similar Fe content of 0.21 wt%.The cyclic voltammetry (CV) curves of CNT, cov FeTPP-CNT, and noncov FeTPP/CNT show clear oxygen reduction peaks at about 0.6 V versus RHE in O 2 -saturated 0.1 m KOH (Figure S18, Supporting Information).However, when preparing inks without CNT, the FeTPP powder electrode shows very sluggish CV results (Figure S18c, Supporting Information).A reduction/oxidation peak at about 0.25 V versus RHE in the CV curve of cov FeTPP-CNT and noncov FeTPP/CNT sample corresponds to the change between Fe 3+ and Fe 2+ . [41]Interestingly, as shown in Figure 3a, the CV plot in N 2 -saturated KOH of the cov FeTPP-CNT sample was found to be significantly larger than that of FeTPP/CNT and pristine CNT, indicating more FeTPP molecules were exposed on the surface of the CNT via covalent functionalization.The Fe 2+ is assumed as the electrochemical active sites, and the active metal site density (SD cv ) of both samples was estimated by integrating the peak area of Fe 3+ to Fe 2 (Figure S19, Supporting Information). [41]The results show the SD cv of cov FeTPP-CNT is ≈1.12 (±0.1) × 10 15 sites cm −2 which is almost six-fold higher than the SD cv of FeTPP@CNT (1.8 (±0.12) × 10 14 sites cm −2 ).The SD ICP of cov FeTPP-CNT is ≈8.75 × 10 15 sites cm −2 and SD ICP of noncov FeTPP-CNT is ≈6.34 × 10 15 sites cm −2 .The active site utilization was calculated based on the SD CV and ICP results.The utilization CV/ICP of cov FeTPP-CNT is 12.7 (±3) %, while that of noncov FeTPP/CNT is only ≈2.8 (±0.5) %, suggesting covalent functionalization helps to increase the surface area available for the catalytic reaction by allowing more active sites to be exposed.The double-layer capacitance (C dl ) of the three samples was determined by analyzing the non-Faradaic region of the CV curves obtained at different scan rates (Figure S20, Supporting Information).The results indicate that the C dl of the cov FeTPP-CNT sample is four times higher than that of the noncov FeTPP/CNT sample.An oxygen control sample was prepared by refluxing CNT under air without Fe complexes (O-CNT) and its oxygen content is ≈1.15 + 0.06 at% tested by XPS.Compared with pristine CNT, the O-CNT shows little increased C dl of ≈0.09 uF and ≈0.5 mA cm −2 increasing in diffusion-limited current densities of LSV, which suggests the oxygen contain group and defects on the CNTs only give a small contribution to the C dl .This indicates that the enhanced electrochemical activity of the cov FeTPP-CNT catalyst is mainly due to the highly isolated dispersion of the FeTPP molecules achieved through covalent functionalization, and the oxygen-containing groups and defects also give a small contribution to their ECSA.
The Linear sweep voltammetry (LSV) curves were carried out to evaluate the ORR performance.As expected, both materials with FeTPP molecules show better ORR activity than pristine CNT (Figure 3b).The cov FeTPP-CNT electrode provides better ORR activity with an onset potential (E at 0.1 mA cm −2 ) of 0.79 V versus RHE and halfwave potential (E 1/2 ) of 0.64 V versus RHE, while the noncov FeTPP/CNT sample only exhibits an E 1/2 value of ≈0.51 V versus RHE.Although the noncovalent functional sample shows the same onset potential of 0.79 V versus RHE, the reaction becomes sluggish after the reaction begins.To determine this performance difference, the ORR kinetic current density (J k ) of these two samples at 0.75 V versus RHE and 0.7 V versus RHE was calculated.The J k of cov FeTPP-CNT is 0.36 mA cm −2 at 0.75 V versus RHE and 1.08 mA cm −2 at 0.70 V versus RHE, while the J k of noncov FeTPP/CNT is 0.24 mA cm −2 at 0.75 V versus RHE and 0.46 mA cm −2 at 0.70 V versus RHE, suggesting the high ORR kinetic of covalently functional catalyst.To fairly evaluate these two materials, the average intrinsic TOF was further extracted based on ICP results, as shown in Figure 3f.The cov FeTPP-CNT shows a higher TOF ICP of ≈0.71 e − site −1 s −1 than noncov FeTPP/CNT of 0.45 e − site −1 s −1 at 0.7 V versus RHE.The TOFs based on their SD CV are also calculated.The TOF CV of cov FeTPP-CNT is ≈6.02 e − site −1 s −1 and that of noncov FeTPP-CNT is ≈15.9 e − site −1 s −1 The reduced TOF CV for the cov FeTPP-CNT is caused possibly a combination of higher site density and strain. [42]Moreover, the Tafel slope was calculated to reveal the ORR reaction kinetics.The cov FeTPP-CNT catalyst presents a relatively lower Tafel slope at both low and high overpotentials (Figure 3d), indicating the high kinetic ORR performance.The slope change from low overpotential to high overpotential only takes place in the noncovalent functional catalyst, suggesting the changes in the rate-limiting step during the ORR process. [43]Moreover, the chronoamperometry test showed the covalent functional methods can improve the stability of the catalyst during the ORR process (Figure 3e) and the covalently functional sample shows good stability after 5000 CV cycles in N 2 -saturated condition (Figure S21, Supporting Information).Compared with other reported porphyrin-based molecular ORR electrocatalysts, the covalently functionalized FeTPP-CNT shows good performance (Table S2, Supporting Information).
To reveal the difference in electrochemical active site density ORR performance between covalent and non-covalent methods samples, the N 2 sorption measurements were carried out to characterize the specific surface area of the samples.As shown in Figure S22 (Supporting Information), according to the Brunauer-Emmett-Teller (BET) analysis, the specific surface area of CNT and functional CNT remain similar after the diazonium reaction.Moreover, control experiment 2 was designed wherein diazotization without CNT and its results (Scheme S1 and Figure S23, Supporting Information) demonstrate that dimers or covalent organic framework (COF) type species do not generate during the synthesis.This result is also supported by the TEM results where only ≈1 nm thin layers of molecules can be observed on the surface of the CNT (the sur-face area of each FeTPP is ≈1-1.25 nm 2 ).In addition, during the development of the functionalization method, it was also found that the FeTPP loading on the CNT surface reached a maximum limit and could not be increased by further increasing the ratio of starting material to CNT (Table S3, Supporting Information).Although the presence of the CNT makes the reaction more complicated, based on the above results, we hold the opinion that FeTPP did not participate in large-scale aggregation and polymerization during the reaction.Focusing on the noncovalent functional ink, although the CNTs have provided enough surface area to support the dispersed molecule catalysts from the theoretic calculation (Table S4, Supporting Information), particles could be found on the surface of the dried noncov sample (Figure 4a; Figure S24, Supporting Information).We did not observe any particle structure on the surface of the covalent functional sample and CNT sample ink (Figures S25 and S26, Supporting Information).According to the EDS results, it is observed that the particles present in the noncovalent functional ink are aggregates of FeTPP molecules (Figure 4b).This aggregation phenomenon can be attributed to the low solubility of FeTPP in water, [44] and the lack of dispersion of the FeTPP in the ink can not be easily addressed by the sonication.The formation of these aggregates significantly reduces the active site density and electrochemical surface area, thereby resulting in a sluggish performance of ORR.
The rotating ring disk electrode (RRDE) result shows that both covalent and noncovalent functional samples seemingly convert O 2 into H 2 O by the 4-electron pathway at a high overpotential with a low HO 2 − yield (<5%) and the pristine CNT electrode with an HO 2 − yield ≈60% during the whole reaction.Similar electron transfer number results were also confirmed based on the calculated Koutecky-Levich plots (Figure S27, Supporting Information).However, an obvious bump in the ring current curve of noncov FeTPP/CNT can be observed at the mixed controlled region (from 0.7 to 0.4 V vs RHE).In contrast, this is not observed in the covalently functionalized sample, and the calculation also demonstrates the covalent sample will directly conduct the 4-electron reaction from the ORR onset potential.In the ring current curve, the onset potential for both CNT and the noncovalent functional sample is at ≈0.7 V versus RHE.This suggests the bump is possibly caused by the uncovered CNT in the noncov FeTPP/CNT samples due to the limited active sites.In diffusioncontrolled regions, the reaction rate of dissolved oxygen becomes very fast and efficient.As a result, a reduced ring current of noncov FeTPP/CNT was observed, which is possibly due to the dissolved oxygen reacting on limited Fe active sites or the peroxide formed on the CNT possibly passing to Fe sites to further convert into H 2 O.The limited oxygen transfer was further confirmed by the electrochemical impedance spectroscopy (EIS) (Figure 4e), which was carried out at around the halfwave potential of each sample in the mixed diffusion+kinetic region, where samples show differences in LSV.All the samples show three semi-circles in the Nyquist plot, and they all show almost the same first semicircle in the high-frequency region, which was the diffusion resistance due to the carbon/Nafion matrix. [45]Compared with CNT and noncov FeTPP/CNT, cov FeTPP-CNT shows the smallest second and third semi-circles in the low-frequency region, which represent the charge transfer resistance and the resistance due to external mass-transport of oxygen, respectively. [44]The reduced second and third semi-circles suggest the charge transfer resistance and oxygen transport resistance could be reduced by achieving covalent functionalization.For the noncovalently functionalized sample, we assume the sluggish charge and oxygen transport are possibly caused by the aggregation of the molecule and the uneven film observed in SEM results.Besides, it should be noted that the better thin film is beneficial to ORR performance and stability of catalysts while the rough thin film surface of noncov FeTPP/CNT may lead to suboptimal ORR performance. [46]PP was also covalently functionalized onto the surface of CNT via the same method and reaction conditions.Compared with CNT, although the ORR performance of TPP-CNT improved to some extent due to the potential nitrogenous structures in molecules and defects possibly generated during the experiment, and it is less selective to 4 e-at higher overpotential (Figure S28, Supporting Information).Especially, the TPP-CNT also shows the 2-electron pathway with an HO 2 − yield of ≈40%.Based on all the above results, it is believed that the Fe centers are active sites in the 4-electron pathway.Although, both the covalently functionalized samples and noncovalent functionalized samples should possess the same active center, it is important to highlight that cov FeTPP-CNT exhibits ≈6.2 times higher active site density and 1.6 times higher TOF ICP compared to noncov FeTPP/CNT.The aggregation of molecules may limit oxygen diffusion from the surface macrocycles to those beneath and hinder electron transfer from the CNTs, resulting in a sluggish ORR.This discrepancy could lead to variations in ORR intermediates and potentially different ORR pathways.
An interesting observation from the RRDE results of noncov FeTPP/CNT is the presence of an HO 2 − bump at 0.4 to 0.7 V versus RHE, suggesting a potential 2e − + 2e − mechanism in alkaline electrolyte.However, it is crucial to acknowledge that reaction mechanisms are influenced by numerous factors, including active site density, O 2 diffusion rate, intermediate residence time, and more. [47]Future research endeavors will focus on developing innovative strategies to prevent macrocycle catalyst aggregation and delving deeper into their reaction mechanisms.

Conclusion
In summary, we have synthesized covalently functionalized iron porphyrin carbon nanotube materials using an aryl diazonium methodology for the first time.The ORR performance of the Fepyrrolic N structure was investigated and then compared with the analogous non-covalently bound FeTPP CNT system.The results demonstrate that the covalently functionalized material shows approximately six times the electrochemical active site density compared to its noncovalent counterpart, resulting in an enhanced TOF across a wide potential range while maintaining similar macrocycle loading.We anticipate that the insights gained from this comparative study of functionalization methods, active site density, TOF, and reaction pathways will contribute to the design of effective macrocycle catalysts for various reactions, including but not limited to ORR, CO 2 or nitrate reduction, and will deepen our understanding of reaction selectivity.

Figure 1 .
Figure 1.a) Schematic illustration of the overall synthetic procedure for FeTPP-NH2 and FeTPP-CNT.b) SEM images of FeTPP-CNT.c) TEM image of FeTPP-CNT.d) HADDF-STEM of FeTPP-CNT and EDS elemental mapping of Fe.

Figure 3 .
Figure 3.The ORR performance of cov FeTPP-CNT, noncov FeTPP/CNT, CNT, and 20% Pt/C.a) CV curves in N 2 -saturated 0.1 m KOH with a rotation rate of 0 rpm at 100 mV s −1 scan rate.b) LSV curves obtained by subtracting N 2 -saturated LSV from the O 2 -saturated result (cathodic scan in 0.1 m KOH at with a rotation rate of 1600 rpm recorded at 10 mV s −1 ).Error bars represent the standard deviation from five separate measurements.c) Comparison of SD CV and Utilisation CV/ICP .d) Tafel plots e) chronoamperometric responses in O 2 -saturated 0.1 m KOH at 0.4 V versus RHE and 1600 rpm.f) Comparison of J kin and TOF ICP at 0.7 V versus RHE.

Figure 4 .
Figure 4. a) The SEM image, b) The EDS mapping of noncov FeTPP/CNT ink.c) Cathodic scan RRDE measurement in O 2 saturated 0.1 m KOH at 1600 rpm, 10 mV s −1 .d) H 2 O 2 production and electron transfer numbers calculated.e) Nyquist plots of the samples obtained at the voltage near the E1/2 in O 2 -saturated 0.1 m KOH solution at a rotation rate of 1600 rpm.