Electrocatalytic composites of carbon quantum dots and anion exchange ionomers for the oxygen reduction reaction

Abstract

Heteroatom-doped carbon quantum dots (CQD) were synthesized by 3 preparation methods: pyrolysis, microwave irradiation, hydrothermal synthesis. Their properties as electrocatalysts for the oxygen reduction reaction (ORR) in alkaline conditions were studied by linear sweep voltammetry. The best properties for N-doped CQD were obtained for hydrothermally synthesized CQD with an onset potential of 0.85 V vs RHE. Co-doping with a second heteroatom, B or S, improved the electrocatalytic properties further with the best performance obtained for B-N co-doping with an onset potential of 0.87 V vs RHE. The influence of an anion exchange ionomer (AEI) on the properties was also studied using 5 AEI with different backbone hydrophilicity and side-chain length. The best properties were observed for a composite electrode combining B-N co-doped CQD and a poly(2,6-dimethylphenylene oxide) with quaternary ammonium groups grafted on pentyl side-chains with an onset potential of 0.94 V vs RHE, which is among the highest in the literature.

Keywords:

• Electrochemistry
• carbon materials
• solid ionic conductors
• fuel cells

Introduction

The electrocatalytic performances of materials are a complex function of multiple factors (Costamagna & Srinivasan, 2001; Gasteiger, Kocha, Sompalli, & Wagner, 2005; Jaouen et al., 2011). In particular, the interaction of a molecule with the surface of an electrocatalyst leads to physi- or chemisorption phenomena (Thommes et al., 2015) that weaken the intramolecular bond susceptible to be broken by the electrochemical reaction. The interactions between the molecule and the electrocatalyst should be strong enough to allow the breaking of the bond, but not too strong so that the reaction product can be desorbed easily (Gasteiger et al., 2005). In that sense, an ideal electrocatalyst should behave according to the Sabatier principle (“interactions not too weak, not too strong” (Che, 2013)) The interactions can often to some extent be predicted from chemical considerations, including acid-base (such as amine-CO₂ (Yadav et al., 2022)) or particular chemical affinities (such as copper ions-ammonia (Poyet, Knauth, & Llewellyn, 2002)).

The electrocatalytic reduction of oxygen is a particularly difficult process, given that the oxygen molecule does not present a permanent dipole moment and shows low interactions with most substrates. Furthermore, the O = O double bond is particularly strong (bond enthalpy in the order of 500 kJ/mol) and difficult to break, especially in acidic conditions. The oxygen reduction reaction (ORR) requires consequently a potent electrocatalyst, generally from expensive Pt-group metals (Costamagna & Srinivasan, 2001; Gasteiger et al., 2005), which are critical raw materials. Finding a replacement electrocatalyst for the ORR (Coutanceau, Croissant, Napporn, & Lamy, 2000) is therefore a challenging but highly rewarding objective of current research.

In alkaline conditions, where the ORR is less energetically demanding, non-noble metal electrocatalysts are available (Ge et al., 2015; Jaouen et al., 2011; Lin, Miao, Wallace, Chen, & Allwood, 2021; Ma et al., 2019; Nunes et al., 2015; Wang & Su, 2014); however, their implementation requires to address several relevant needs, including high catalytic activity, but also high stability and durability, a large availability (no critical raw materials) and a low cost. The ORR can be implemented by two pathways in alkaline conditions. In the four-electron route, the reduction proceeds right to the formation of hydroxide ions: (1) $${\mathrm{O}}_2+{\text{4 e}}^{-}+{\text{2 H}}_2\mathrm{O}è{\text{4 HO}}^{-}\mathrm{E}°=0.40\text{V vs SHE}$$(1)

In the two-electron route, the reduction leads to intermediate hydroperoxide anions according to the reaction: (2) $${\mathrm{O}}_2+{\text{2 e}}^{-}+{\text{H}}_2\mathrm{O}è{\text{HO}}_2{}^{-}+{\text{HO}}^{-}\mathrm{E}°=-0.0\text{7 V vs SHE}$$(2)

Hydroperoxide can be reduced in a second step to hydroxide ions. The four-electron pathway is evidently favourable, because it allows a better use of energy and also avoids the production of potentially corrosive hydroperoxide anions. The latter reaction can be prevented by the use of a potent electrocatalyst (Lin et al., 2021).

The electrocatalytic reaction takes place at the triple-phase boundary between the gas phase, the electrocatalyst and the ionomer (Cao et al., 2021; Jinnouchi et al., 2021; Zhang, Zhang, et al., 2021); the addition of a hydroxide ion-conducting ionomer is useful for the removal of hydroxide ions from the electrode (Chen & Lee, 2021; Zhang, Zhang, et al., 2021). The electrode reaction kinetics are therefore very sensitive to the presence and the distribution of the ionomer (Adhikari, Pagels, Jeon, & Bae, 2020). An optimized microstructure with well-distributed catalyst and ionomer particles increases the triple phase boundary area and the electrode kinetics (Lin et al., 2021). The intrinsic properties of the ionomer, including its hydrophilicity, its gas permeability, its compatibility with the catalyst particles (e.g., in terms of wetting and avoiding delamination with the ion exchange membrane) and its dispersivity in the casting solvent used for the electrode fabrication are further paramount factors for an optimized electrocatalytic electrode (Jinnouchi et al., 2021). They depend in a complex way on the macromolecular backbone and the position of the functionalizing quaternary ammonium groups for anion exchange ionomers (AEI) (Adhikari et al., 2020).

In this work, the electrocatalytic performances of carbon quantum dots (CQD) for the ORR are analyzed (Tian et al., 2021; Wang, Feng, Dong, & Huang, 2019; Zhai et al., 2022). CQD are nanostructured, spherical particles, made up by several graphitic planes containing generally a variety of oxygenated groups at the plane edges (Lim, Shen, & Gao, 2015), contributing prominently to the electrocatalytic properties for the ORR, especially hydroxyl and carboxyl groups (Kordek et al., 2019). Further doping by heteroatoms, especially by nitrogen, is acknowledged to improve the catalytic properties of carbon materials (Kim et al., 2022; Sarapuu, Kibena-Poldsepp, Borghei, & Tammeveski, 2018; Scardamaglia et al., 2017; Sibul et al., 2019); nitrogen can be inserted on various sites in the carbon lattice, including pyrrolic, pyridinic (on edge positions or near a carbon vacancy) and so-called graphitic sites, where the nitrogen atom is included on substitutional lattice positions (Palaniselvam, Kashyap, Bhange, Baek, & Kurungot, 2016). Furthermore, aminic or amidic sites can exist, which show strong interactions with acidic gases, such as CO₂ (Yadav et al., 2022) (Figure 1a).

Figure 1. Schematic representation of carbon quantum dots. a) N-doped CQD with oxygen (red) and nitrogen atoms (light blue). One can differentiate pyrrolic, pyridinic, graphitic and aminic N sites; b) B-N co-doped CQD (B: grey); and c) S-N co-doped CQD (S: yellow). One can observe boronic acid, sulphide and sulfone sites.

Co-doping by adding a second heteroatom, such as boron and sulphur, to nitrogen can further improve the catalytic properties (Liu, Song, Ning, & Xu, 2015; Sun et al., 2018; Tian et al., 2021; Van Tam et al., 2017). Boron dopant atoms are essentially inserted on edge positions in a boronic acid environment (Figure 1b), as evidenced by ¹¹B NMR spectroscopy (Nallayagari, Sgreccia, Pasquini, Sette, Knauth, & Vona, 2022); sulphur dopants can be present in various oxidation states, including sulfoxide and sulphide, mostly on edge sites (Figure 1c) (Nallayagari, Sgreccia, Pasquini, Vacandio, Kaciulis, Di Vona, & Knauth, 2022). Furthermore, the interactions with ionomers are explored using a variety of anion exchange ionomers (AEI) having different backbone structures and different side chain lengths (Figure 2). Mechanistic interpretations of the electrocatalytic performances based on bonding and molecular structures are given wherever possible.

Figure 2. Repeat units of anion exchange ionomers (AEI).

Materials and methods

Citric acid, D-(+)-glucose, D-(+)-glucosamine hydrochloride, sucrose (≥ 99.5%), N-octylamine, boric acid (≥ 99.5%), L-cysteine (97%), polyaniline (emeraldine base, PANI, MW >20000 g/mol), dimethylformamide (DMF), and other chemicals were used as received from Sigma-Aldrich. Carbon paper (AvCarb EP55) and Pt/C 60% cloth gas diffusion electrode (GDE, 0.5 mg/cm²) were purchased from the Fuel Cell Store.

Synthesis of CQD

Three synthesis methods were used for the CQD preparation: pyrolysis (abbreviated P), microwave irradiation (M) and hydrothermal synthesis (H).

Pyrolysis

Here the precursor was citric acid (Nallayagari, Sgreccia, Pizzoferrato, Cabibbo, Kaciulis, Bolli, et al., 2022). The precursor was introduced in a round flask heated to 200 °C and left 10 min at this temperature. The melt became yellow and then orange; it was then cooled down and the pH adjusted to 9 to control the average CQD size (Naik, Sutradhar, & Saha, 2017). The precipitate was centrifuged and the sedimented CQD (called CAP) were purified in a dialysis bag (pore dimension: 2 kDa).

CAP-DMA was obtained by a similar procedure, but before solidification of the orange melt, dimethylamine was added in molar ratio citric acid/dimethylamine 1:1.

Microwave

Here the precursor was glucose (Nallayagari, Sgreccia, Pizzoferrato, Cabibbo, Kaciulis, Bolli, et al., 2022). 1 g of glucose was dissolved in 20 mL 0.1 M HCl. The solution was placed in a microwave oven at a power of 800 W during 4 minutes. After this time, the black precipitate was dispersed in 20 mL water and centrifuged and the sedimented CQD (called GM) were purified in a dialysis bag.

GM-Try was obtained by a similar procedure, except that tryptophan was added initially in molar ratio glucose/tryptophan 1:1.

Hydrothermal

Here the precursor was glucosamine hydrochloride for N-doped CQD (Nallayagari, Sgreccia, Di Vona, Pasquini, Vacandio, & Knauth, 2022). 0.45 g was dissolved in I mL water and the solution was transferred into an autoclave, which was placed in a furnace at 190 °C during 12 h. The precipitate was then dispersed in water, centrifuged and placed in the dialysis bag (called GAH).

GAH-Oct was prepared by adding N-octylamine in 1:1 molar ratio with glucosamine.

For co-doped CQD, sucrose (S) was used as precursor (Nallayagari, Sgreccia, Pasquini, Vacandio, Kaciulis, Di Vona, & Knauth, 2022).

B-N co-doped CQD (SH-B-N) were synthesized adding 0.5 wt% boric acid and dimethylformamide (DMF) as solvent and nitrogen source. S-N co-doped CQD (SH-S-N) were prepared from sucrose and L-cysteine, used as sulphur source, in molar ratio 1:1.5 and DMF. The solutions were stirred for about 30 min and heated in an autoclave at 190 °C for 12 h. The precipitates were then dispersed in water, centrifuged and placed in the dialysis bag.

The CQD were characterized by many spectroscopic techniques, including NMR, FTIR, Raman and XPS spectroscopy (Nallayagari, Sgreccia, Pasquini, Sette, Knauth, & Vona, 2022; Nallayagari, Sgreccia, Pasquini, Vacandio, Kaciulis, Di Vona, & Knauth, 2022; Nallayagari, Sgreccia, Pizzoferrato, Cabibbo, Kaciulis, Bolli, et al., 2022; Nallayagari, Sgreccia, Di Vona, Pasquini, Vacandio, & Knauth, 2022).

The XPS analyses were made by depositing a drop of CQD dispersion on an Au foil using an Escalab 250Xi spectrometer (ThermoFisher Scientific, East Grinstead, UK) with a monochromatic Al Kα (1486.6 eV) excitation source. The binding energy scale was corrected by positioning the C 1s peak of aliphatic carbon at 285.0 eV and controlling if the Fermi level was at 0 eV. Spectroscopic data were processed by the Avantage v.5 software (ThermoFisher Scientific Ltd).

Scanning Electron Microscopy (SEM) images were recorded with a ZEISS Gemini SEM 500 at 5 kV. Samples for Transmission Electron Microscopy (TEM) were prepared from a droplet on carbon film dried under infrared radiation. The carbon film had a mean thickness of 50 nm. All TEM images were made with a PhilipsTM CM-20 working at 200 KV.

FT-Raman spectra were collected using a Micro Raman Optosky ATR 8300 spectrometer with an excitation wavelength of 785 nm. Spectra were collected for at least 10000 scans using a laser power of 85 mW.

Anion exchange ionomer (AEI)

The synthesis and characterization of AEI (Figure 2) were reported before: PPO LC (Becerra-Arciniegas et al., 2020), PPO-Br SC and PPO-Cl SC (Becerra-Arciniegas et al., 2019), PSU SC (Di Vona, Narducci, Pasquini, Pelzer, & Knauth, 2014), ABBA LC (Narducci et al., 2022). The ion exchange capacity (IEC in mmol/g of dry AEI) was measured by acid-base titration after ion exchange with 2 M NaOH at room temperature for 2 days and washing in bidistilled water at 60 °C for 2 days to remove residual salt. After drying over P₂O₅ for 72 h, the weighed membranes were introduced in 0.02 N HCl solution, which was then backtitrated with 0.02 N NaOH solution. The water uptake WU (%) was determined at 25 °C following the equation: WU = 100·(m_wet-m_dry)/m_dry. m_dry and m_wet are respectively the masses of dried and wet sample (after drying over P₂O₅ or immersion in bidistilled water for 48 h). The hydroxide ion conductivity was determined by impedance spectroscopy with an a.c. amplitude of 20 mV in a frequency range between 1 Hz and 1 MHz.

Composite electrode preparation

The carbon paper substrate was activated adapting the procedure reported in Ref. (Kordek et al., 2019) Commercial carbon paper was punched into 8 mm diameter pieces and then treated in a mixture of 15 mL concentrated H₂SO₄, 15 mL concentrated HNO₃ and 15 mL distilled water in a Teflon bottle in a preheated oven at 70 °C for 24 h. The samples were rinsed twice with distilled water and one time with double distilled water and dried overnight in the oven at 70 °C and then using a rotary vane pump (Edwards) for 4–5 h. The carbon paper disks were finally treated 2 h at 500 °C.

The catalytic ink composition was adapted depending on the objective of the study.

For the investigation of the catalytic properties of N-doped CQD, a relatively large amount of 150 mg GAH (23 wt%) and 70 mg of PANI (10 wt%) in 0.6 g of solvent DMF were added to 450 mg PSU SC (67 wt%) in 2.0 g of DMSO. In the case of GAH-Oct, 150 mg GAH-Oct with 0.7 g NMP were added to disperse the materials well. PANI was added in some cases, because it was reported to improve the electrocatalytic ORR also in alkaline conditions (Jiang, Yu, Huang, & Sun, 2018; Matseke, Munonde, Mallick, & Zheng, 2019). In all cases, the solutions were kept overnight under stirring before coating carbon paper with 5 μL of these suspensions using a microsyringe. The coated electrodes were dried at 80 °C for 24 h.

For the study of heteroatom co-doped CQD, a large percentage of CQD was used: (75 ± 5) mg of the co-doped CQD (90 wt%) and (8 ± 1) mg of PSU SC (10 wt%) were dispersed in 0.5 mL DMF and stirred overnight. 5 µL of the above-prepared slurry were deposited on a carbon paper substrate and dried at 80 °C for 24 h.

For the study of the influence of the AEI, a relatively large amount of AEI (20 wt%) was added to accentuate their effect: 10 mg of AEI was dispersed in 1 mL DMSO together with 40 mg of B-N co-doped CQD. The mixture was stirred at room temperature and then sonicated for the entire night. 5 µL of the ink was drop-cast on the activated carbon paper. To avoid disrupting the surface of the catalyst, the electrodes were carefully vacuum-dried at 40 °C for about 4 hours using a rotary vane pump.

The composite electrodes were attached to the 6 mm diameter glassy carbon disc of a rotating disk electrode (RDE, OrigaTrod, OrigaLys) used as working electrode. The electrochemical measurements were performed using a Biologic potentiostat VMP3 with a standard three-electrode system. The counter electrode was a 4 cm² platinum foil and the reference electrode was Ag/AgCl/KCl saturated type (E = 0.197 V vs SHE). Linear sweep voltammograms (LSV) were recorded at room temperature in a potential range of −0.1 to −0.7 V vs Ag/AgCl at a scan rate of 5 mV/s in 0.1 or 1 M KOH after a 40 min oxygen purge. The potential values vs Ag/AgCl can be transformed into values vs the reversible hydrogen electrode (RHE) by the equation: (3) $$E\left(\mathit{\text{RHE}}\right)=E\left(\mathit{\text{Ag}}/\mathit{\text{AgCl}}\right)+197+59·\mathit{\text{pH}}\left(3\right)$$(3)

Results

Synthesis and characterization of CQD

Table 1 shows the O/C ratio by XPS spectroscopy of various CQD made by pyrolysis (P), microwave irradiation (M) and hydrothermal synthesis (H) and a comparison with the oxygen-carbon atom ratio in the precursor molecules (citric acid CA, glucose G, glucosamine GA and sucrose S).

Table 1. O/C ratio of various CQD observed by XPS analysis.

Samples	O/C ratio CQD	O/C ratio precursor
CAP	0.56	1.2
CAP-DMA	0.59	1.2
GM	0.42	1.0
GM-Try	0.37	1.0
GAH	0.24	0.8
GAH-Oct	0.12	0.8
SH-S-N	0.25	0.9
SH-B-N	0.23	0.9

The O/C ratios of CQD are strongly dependent on the precursor, but also on the synthesis technique. The highest ratio is observed for pyrolysis of citric acid, indicating that C-O bonds, which are sensitive to hydrolysis, are protected in the absence of water in this process. The microwave irradiation technique leads to a lower O/C ratio than in the glucose precursor; some alcoholic groups are lost in acidic conditions during the process even if the microwave conditions are mild. Hydrothermal synthesis leads instead to distinctly lower O/C ratios. A large part of the hydrophilic precursor groups in sucrose and glucosamine gets lost by hydrothermal reaction at high temperature, leading to scission of hydrolysis-labile C-O bonds. The particularly low O/C ratio of GAH-Oct can be attributed to the long alkyl chain of N-octylamine that is preserved in the CQD (Nallayagari, Sgreccia, Pizzoferrato, Cabibbo, Kaciulis, Bolli, et al., 2022). The low O/C ratio increases the electronic conductivity of hydrothermally prepared CQD, which is an important asset for electrocatalysis. For that reason, hydrothermal samples were exclusively used for the electrocatalytic studies.

The nitrogen insertion site is also strongly dependent on the precursor for hydrothermal synthesis: when a molecule with preexisting carbon-nitrogen bonds is used, the nitrogen is inserted mainly in graphitic sites (Nallayagari, Sgreccia, Pasquini, Vacandio, Kaciulis, Di Vona, & Knauth, 2022), whereas with the solvent DMF as nitrogen source, the nitrogen atoms are present mainly in pyridinic (means at edges or with an adjacent carbon vacancy) and pyrrolic sites (Nallayagari, Sgreccia, Pasquini, Sette, Knauth, & Vona, 2022) (Table 2).

Table 2. Binding energy values and atomic concentrations of nitrogen species in CQD.

Nitrogen species	Pyridinic eV (at %)	Pyrrolic/Amine/Amide eV (at %)	Graphitic eV (at %)	Pyridinic oxide eV (at %)
Sample
GAH	398.0 (2.5 at%)	399.7 (3.7)	401.6 (1.6)	–
GAH-Oct	–	399.0 (1.0)	401.1 (7.4)	402.8 (1.6)
GM-Try	–	400.1 (3.1)	–	402.0 (2.3)
SH-S-N	–	399.8 (7.4)	–	402.8 (1.9)
SH-B-N	–	400.0 (3.1)	–	402.4 (1.8)

Table 3 summarizes carbon XPS data. The high ORR activity of co-doped samples (SH-B-N and SH-S-N) is probably also linked to a higher amount of catalytically active –C = O bonds, introduced by the precursor cysteine or formed with boron oxide. The low amount of C-O/C = N bonds in GM is simply related to the absence of nitrogen in this sample.

Table 3. Binding energy values and atomic concentrations of carbon species in CQD.

Carbon species	C = C/C-C bonds eV (at %)	C-O/C = N bonds* eV (at %)	-C = O bonds eV (at %)
Sample
GM	285.0 (55.6)	287.0 (6.4)	–
GM-Try	285.0 (55.3)	286.7 (17.6)	289.3 (4.0)
GAH	285.0 (54.9)	286.8 (19.8)	288.7 (3.6)
GAH-Oct	285.0 (56.4)	286.6 (17.3)	–
SH-S-N	285.0 (36.1)	286.5 (18.6)	288.4 (9.8)
SH-B-N	285.0 (35.7)	286.2 (26.9)	287.9 (11.0)

*C = N exists in GM-Try, GAH, GAH-Oct, SH-S-N, SH-B-N.

The heteroatom dopant concentration in SH-S-N and SH-B-N was also measured by XPS: C-B-O_2: (192.4 eV) 0.6 at %; C-S-C (163.9) 5.0 at %, C-SO₂-C (167.6 eV) 1.7 at % (Nallayagari, Sgreccia, Pasquini, Vacandio, Kaciulis, Di Vona, & Knauth, 2022).

Figure 3 shows the Raman spectra of heteroatom co-doped CQD with the typical D and G peaks (Wu, Wang, Wang, & Fang, 2018). The D band is associated with defects in amorphous carbons here related to the presence of a high ratio of “C-edge” and structural modifications near doping atoms (Dychalska et al., 2015; Thang et al., 2021). The G band is related to sp² carbon in double bonds and aromatic rings. In our samples the I_D/I_G value is around 1.3, which is in agreement with the literature and indicates a relatively high edge over graphene ratio (Pan, Zhang, Li, & Wu, 2010; Thang et al., 2021).

Figure 3. Comparison of Raman spectra of heteroatom co-doped CQD showing D and G bands.

A typical TEM image of CQD is shown in Figure 4. The GAH nanoparticles have a spherical shape and an average size of 4 ± 1 nm.

Figure 4. TEM micrograph of GAH CQD.

Characterization of AEI

The ion exchange capacity and water uptake of the ionomers are reported in Figure 5. One notices that the water uptake is as expected strongly correlated with the ion exchange capacity, whereas the ionic conductivity shows no simple trend. In fact, the ionic conductivity depends largely on the tortuosity and percolation situation of the ion conduction channels inside the ionomer matrix (Knauth et al., 2021), which is related to the nanophase separation. One can recognize that the PPO LC ionomer presents a large ion conductivity, indicating that the flexible side chain improves the nanophase separation and therefore reduces the related tortuosity and percolation threshold of the ionomer.

Figure 5. Water uptake and ionic conductivity at 25 °C as function of the ion exchange capacity (IEC) of various AEI (see Scheme 1).

Nitrogen-doped CQD electrocatalysts for the ORR

Figure 6 shows the typical microstructure of a composite CQD-AEI electrode here containing PPO LC. It is possible to observe the fibres of carbon paper with a composite layer deposited on top.

Figure 6. SEM image of SH-B-N/PPO LC composite electrode.

Figure 7 shows a comparison of linear sweep voltammograms of pristine carbon paper and a Pt/C cloth GDE benchmark with several N-doped CQD on carbon paper. The onset potential for N-doped CQD is situated at an average value of 0.85 V vs RHE. However, the limiting current for GAH samples is very low, showing strong mass transport limitations, which indicate problems in the microstructure of the electrodes. The much-improved limiting current and the higher half-wave potential of GAH-Oct are attributable to a better electrode morphology, presumably due to the much-reduced agglomeration of CQD, due to the passivating layer of N-octylamine chains. The properties of electrodes without PANI are even better so that PANI was removed in forthcoming electrode design.

Figure 7. Linear sweep voltammograms of ORR in 1 M KOH at 1500 rpm rotating speed with various N-doped CQD with PSU SC and comparison with carbon paper and Pt/C cloth GDE.

Koutecky-Levich plots reported in Nallayagari, Sgreccia, Di Vona, Pasquini, Vacandio, & Knauth (2022) show that the 4-electron mechanism is predominant for electrodes containing GAH-Oct, whereas a significant amount of 2-electron reduction is observed for carbon paper.

Heteroatom co-doped CQD electrocatalysts for the ORR

In the linear sweep voltammograms of the ORR (Figure 8), the first part of the curve is dominated by the electron transfer reaction giving an exponential shape described by the Butler-Volmer equation (Hamann, Hamnett, & Vielstich, 2007); this region is where electrocatalytic effects are observed and the onset potential is an appropriate parameter to describe the electrocatalysis. The final part of the curve, where a limiting current plateau can be observed, is determined by mass transport phenomena, including oxygen gas and hydroxide ion transport. This domain is controlled by the porosity and tortuosity of the catalytic layer. At intermediate overpotentials, the Ohmic drop due to the electrode resistance can be observed, which is determined by the electronic conductivity of the CQD and the ionic conductivity of the AEI. The half-wave potential is a relevant parameter for assessment of the Ohmic and diffusion resistances.

Figure 8. Linear sweep voltammograms of ORR in 1 M KOH at various rotating speeds in rpm with B-N co-doped CQD and PSU SC.

Figure 9 shows a plot of half-wave vs onset potentials for the ORR from which the electrocatalytic activity can be deduced to increase in the order N-doped < S-N co-doped < B-N co-doped.

Figure 9. Electrocatalytic performance of doped CQD with PSU SC ionomer for the ORR in 1 M KOH.

B-N co-doped CQD with various ionomers for the ORR

The linear sweep voltammograms of Figure 10 demonstrate the high electrocatalytic activity of composites of B-N co-doped CQD with PPO LC ionomer, which attains a limiting current density near the one of Pt/C cloth GDE.

Figure 10. Linear sweep voltammograms of ORR in 0.1 M KOH at 2000 rpm rotating speed with SH-B-N CQD and various ionomers and comparison with a Pt/C cloth GDE.

A comparison of onset and half-wave potentials for the various AEI is presented in Figure 11. One can recognize that the onset potential is around 0.94 V and independent of the AEI inside the limits of uncertainty, whereas long-chain ionomers show the best performance in terms of half-wave potential, especially PPO LC in agreement with its high limiting current density.

Figure 11. Electrocatalytic performance for the ORR in 0.1 M KOH of electrodes combining SH-B-N CQD with various ionomers (see Figure 2).

Finally, Figure 12 shows a long-time cycling test of the best performing SH-B-N/PPO LC electrode. There is about 10% decrease of the limiting current after 3000 cycles showing a change of the electrode microstructure probably related to the swelling of the ionomer. Instead, the onset potential shows only a small change indicating a good stability of the CQD.

Figure 12. Long term stability test by cyclovoltammetry (3000 cycles at 1500 rpm) of SH-B-N/PPO LC.

Discussion

Oxygen adsorption and dissociation

The oxygen molecule has no permanent dipole moment, but dipoles can be induced by neighbouring surface or edge dipoles. The electronegativity increases in the order boron < carbon < nitrogen leading to local polarization effects that establish favourable conditions for oxygen adsorption on heteroatom-doped CQD.

The better electrocatalytic performance of B-N co-doped CQD is generally attributed to the asymmetry between more electropositive boron and more electronegative nitrogen (Sun et al., 2018; Tian et al., 2021; Zhang, Zhang, et al., 2021). Furthermore, boron has only three valence electrons so that insertion on graphitic sites is difficult, because no participation in aromatic bonds is anticipated. For that reason, boron is commonly located on edge sites as boronic acid; this position was confirmed by ¹¹B NMR and XPS spectroscopy studies (Nallayagari, Sgreccia, Pasquini, Sette, Knauth, & Vona, 2022).

Nitrogen has instead five valence electrons and it can be placed on several positions inside the carbon lattice, especially on pyrrolic, pyridinic (necessarily on edge sites or near a carbon vacancy) but also on graphitic positions. The relative amount of the various positions is strongly related to the precursor: if glucosamine is used, the presence of nitrogen in the precursor molecule leads to an important amount of nitrogen in graphitic positions (Nallayagari, Sgreccia, Di Vona, Pasquini, Vacandio, & Knauth, 2022), whereas with sucrose as precursor and dimethylformamide solvent as nitrogen source pyrrolic and pyridinic nitrogen predominate (Nallayagari, Sgreccia, Pasquini, Vacandio, Kaciulis, Di Vona, & Knauth, 2022).

The B-N co-doped samples (SH-B-N) present consequently a combination of pyrrolic and pyridinic N sites together with boronic acid edge sites. The best electrocatalytic properties of B-N co-doped CQD are therefore attributable to the prevalence of edge sites for both boron and nitrogen atoms that establish particularly favourable conditions for oxygen adsorption and dissociation with formation of epoxy species (Scardamaglia et al., 2017).

Impact of AEI

Figure 11 confirms that the onset potential of the ORR does not depend on the AEI, because it is attributable to the intrinsic catalytic activity of the CQD for the electron transfer reaction. However, the half-wave potential is strongly dependent on the presence of the AEI, because the mass transport limitations are observed here. The mass transport restrictions are attributable to the oxygen transport to the catalytic sites and the hydroxide ion transport away from the catalytic sites, according to the ORR Eq. (1)(1) $${\mathrm{O}}_2+{\text{4 e}}^{-}+{\text{2 H}}_2\mathrm{O}è{\text{4 HO}}^{-}\mathrm{E}°=0.40\text{V vs SHE}$$(1) .

The oxygen transport is limited by the porosity and tortuosity of the catalyst layer (CL). These factors can be modified by the use of the casting solvent for the catalytic ink; a low boiling solvent increases the CL porosity. Furthermore, the oxygen permeability of the AEI is important, which is related to the product of oxygen solubility and oxygen diffusion coefficient. Most work done on the influence of an ionomer on the catalytic activity was done on proton exchange ionomers (Jinnouchi et al., 2021) and experiments on AEI are rarer (Chen & Lee, 2021). It is generally accepted that the oxygen solubility is enhanced in ionomers with hydrophobic backbone, whereas the diffusion of oxygen is augmented in hydrophilic media. The increased solubility in hydrophobic media is consistent with data on the oxygen solubility in organic solvents, showing a clear decrease of O₂ solubility with decreasing aliphatic chain length, e.g., in the order 1-butanol > 1-propanol > ethanol > methanol (Kretschmer, Nowakowska, & Wiebe, 1946; Sato, Hamada, Sumikawa, Araki, & Yamamoto, 2014). For these reasons, the combination of a moderately hydrophilic backbone, like PPO, with long aliphatic side chains supporting the grafted quaternary ammonium groups presents the best compromise in terms of oxygen permeability leading to a clearly higher half-wave potential. The second-best value is observed by ABBA LC, but the very hydrophobic AEI backbone is detrimental for the oxygen diffusivity so that only a limited improvement is observed vs the short-chain ionomers.

The second effect of AEI is on the hydroxide ion transport, which is correlated with the ionic conductivity of the ionomers that depends on the hydroxide ion mobility. The correlation of hydroxide ion mobility with the tortuosity and percolation of AEI was recently treated phenomenologically (Knauth et al., 2021). Figure 3 shows that PPO LC has the highest ionic conductivity, which reduces the Ohmic drop in the catalytic electrode, probably due to the better nanophase separation, as previously observed for many other ionomers with long flexible side chains (Pan et al., 2021; Zhu, Yu, & Hickner, 2018). The impact on the half-wave potential of the ORR is observed in Figure 11: PPO LC attains clearly the highest value (Nallayagari, Sgreccia, Pasquini, Sette, Knauth, & Vona, 2022). The better performance of the other long side chain AEI (ABBA LC) is less evident, probably due to the lower ionic conductivity, which can be attributed to a lower nanophase separation.

Conclusions

The electrocatalytic properties of heteroatom-doped CQD in conjunction with an AEI are investigated for the ORR in alkaline conditions. The CQD are prepared by three different methods: pyrolysis, microwave irradiation and hydrothermal synthesis. The latter method gives CQD with higher electronic conductivity, due to a lower O/C ratio. Linear sweep voltammograms showed that N-doped CQD have an onset potential around 0.85 V vs RHE. The performances can be further improved by co-doping with a second heteroatom; B-N co-doped CQD performed the best with an onset potential around 0.87 V vs RHE. The influence of an AEI on the electrocatalytic performance was studied using 5 AEI with different backbone hydrophilicity and side-chain length. The ionomers with long side-chain worked better than those with short side-chain; the best properties were obtained adding B-N co-doped CQD with an AEI based on poly(2,6-dimethylphenylene oxide) with long side-chain giving onset potentials of 0.94 V vs RHE, among the highest reported in the literature. Promising perspectives of this work are the application of composite heteroatom-doped CQD/ionomer electrocatalytic electrodes for other environmentally relevant reactions, such as the electrochemical CO₂ reduction.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Philippe Knauth

Philippe Knauth is Professor of Chemistry of Materials at Aix Marseille University.

Emanuela Sgreccia

Emanuela Sgreccia is Postdoctoral Fellow at Tor Vergata University of Rome.

Ashwini R. Nallayagari

Ashwini R. Nallayagari made her PhD in cotutela at Aix Marseille University and Tor Vergata University of Rome.

Luca Pasquini

Luca Pasquini is Maitre de Conférences at Aix Marseille University.

Riccardo Narducci

Riccardo Narducci is Researcher and Maria Luisa Di Vona is Professor of Chemistry at Tor Vergata University of Rome.