Carbon Derived from Soft Pyrolysis of a Covalent Organic Framework as a Support for Small-Sized RuO2 Showing Exceptionally Low Overpotential for Oxygen Evolution Reaction.

Electrochemical water splitting is the most energy-efficient technique for producing hydrogen and oxygen, the two valuable gases. However, it is limited by the slow kinetics of the anodic oxygen evolution reaction (OER), which can be improved using catalysts. Covalent organic framework (COF)-derived porous carbon can serve as an excellent catalyst support. Here, we report high electrocatalytic activity of two composites, formed by supporting RuO2 on carbon derived from two COFs with closely related structures. These composites catalyze oxygen evolution from alkaline media with overpotentials as low as 210 and 217 mV at 10 mA/cm2, respectively. The Tafel slopes of these catalysts (65 and 67 mV/dec) indicate fast kinetics compared to commercial RuO2. The observed activity is the highest among all RuO2-based heterogeneous OER catalysts-a touted benchmark OER catalyst. The high catalytic activity arises from the extremely small-sized (∼3-4 nm) RuO2 nanoparticles homogeneously dispersed in a micro-mesoporous (BET = 517 m2/g) COF-derived carbon. The porous graphenic carbon favors mass transfer, while its N-rich framework anchors the catalytic nanoparticles, making it highly stable and recyclable. Crucially, the soft pyrolysis of the COF enables the formation of porous carbon and simultaneous growth of small RuO2 particles without aggregation.

Introduction

Covalent organic framework (n class="Chemical">COF)-derived carbons can be highly structured with substantial surface areas.[1−9] This makes them good supports for the catalyst/electrocatalyst.[7−9] The mesopores in the COF can accommodate active nanoparticles.[10−15] Of the different approaches to developing COFs as supports for electrocatalytic nanoparticles,[16−21] utilizing them as a single-source precursor for generating nitrogen-functionalized (pyridinic, pyrollic, etc.) porous carbons is quite effective. Such carbons can have high surface areas,[22] and the N- and the C-centers in their framework can provide appropriate binding sites for the catalytically active metal, and their inherent micro-mesoporosity favor high active-site accessibility and mass transfer via shortened diffusion paths. Most importantly, if the carbons can be synthesized from the framework solids using mild conditions (T < 500 °C), they can be expected to have less microstructural defects and potential high carbon-electrolyte interfacial energy.[23−26] This makes them superior to typical amorphous carbons obtained from high-temperature pyrolysis.[27−29]

Catalyzed electrochemical oxygen evolution reaction (OER) involves the release acceptance of the electron and proton.[30,31] However, most catalysts display torpid kinetics, making n class="Chemical">water-splitting less translatable in practice.[30,32] Developing highly active and durable OER catalysts can make the technology lucrative.[30,33−41] Their performance is normally quantified by overpotential (η), the extra voltage required over the thermodynamic potential (1.23 V) for the 4e-process. Most OER catalysts are oxide-based and require >300 mV overpotential (η10) to deliver 10 mA/cm2 of geometrical current density.[30,42−44] Only a countable number of them manifests <250 mV overpotential to reach a current density of 10 mA/cm2 (see Table S1).[16,19,30,40,45−48] Precious metal oxides such as IrO2 and RuO2 are known to catalyze OER effectively.[30,49−51] Though precious metals can easily be ruled out as an expensive choice for an OER catalyst, their superior performance still makes them promising.[52,53] For these to reach realistic applications, their concentrations in the active catalyst need to be kept low, the catalyst need to be recyclable, and the catalyst support has to be cheap. Most importantly, if such expensive catalysts can be made to run for long cycles, the cost may not be a huge factor over extended operation periods. These advantageous features are the hallmark of a heterogeneous catalyst. In many works, the OER catalysts are benchmarked against Ru- and Ir-based ones.[38,47−55] However, a scrutiny of the literature reveals that the touted Ru containing heterogeneous catalysts using different carbonaceous and metal foam supports have high overpotentials for OER particularly under alkaline conditions (Table S1).[54−58] This leaves a clear missing piece. What is the best performance achievable from the heterogeneous Ru-based OER catalyst?

Our study identifies a soft pyrolysis method that supports the formation of two materials that require contrasting synthesis n class="Chemical">conditions. Formation of RuO2 nanoparticles requires the use of surfactants and annealing under oxidizing conditions. However, even 500 °C is known to cause significant growth of RuO2 nanoparticles (∼6 nm).[59] Meanwhile, the synthesis of graphenic carbon supports from various sources including COFs generally requires temperatures >600 °C. Thus, identifying soft chemical routes to synthesize these carbons under conditions that favor the growth of small metal-derived nanoparticles can be of immense value. Here, we have utilized the COF as a sacrificial reaction pot for the growth of RuO2 under mild annealing temperatures (<400 °C). Our approach yields a homogeneous dispersion of RuO2 in ordered micro-mesoporous N-rich carbons derived from two different N-rich COFs with related structures. RuO2 grow as extremely small-sized nanoparticles (∼3–4 nm). The RuO2@C composites catalyze OER with a η10 of only 210 and 217 mV in alkaline medium. The overpotentials observed for these composites are very close to the recently reported high-performing (191 and 200 mV) electrocatalysts,[38] wherein the elevated activity emerge only upon supporting the catalyst on highly conducting nickel–gold foam as against the all-organic carbon host presented here.

Results and Discussion

Synthesis and Structure of the IISERP-COF6

IISERP-COF1 (1) was prepared acn class="Chemical">cording to our earlier reported procedure[17] with slight modification to the solvent combination. IISERP-COF6 (2), reported here for the first time, was synthesized by reacting 1,3,5-triazine-2,4,6-triyltris(oxy))tribenzaldehyde and hydrazine hydrate in a mixture of mesitylene, ortho-dichlorobenzene (ODCB), and ethanol (EtOH) in a sealed Pyrex tube under solvothermal conditions (90 °C for 3 days) (Scheme S1).

We described the structure of IISERP-n class="Chemical">COF1 (1) elsewhere.[17] Similar to 1, IISERP-COF6 (2) also has a 2D hexagonal honeycomb structure (Figure A). The structures of the COFs were solved employing crystal building simulations and periodic DFT-based geometry optimizations using the Material Studio program. The structure was geometry-optimized using tight binding density functional theory (DFTB). In the optimized structure of 2, the layers stack in an AAA···fashion, generating hexagonal-shaped pores with a diameter of 26.0 Å along the ab-plane (Figures B and S1, not factoring the van der Waals radii). The powder X-ray diffraction pattern (PXRD) for 2 was refined in the P6/m space group using the Pawley routine, which yielded an excellent fit. Refined cell parameters: a = b = 31.395(5); c = 3.386(7); Rp = 3.77%; wRp = 5.1% (Figure C). The COF forms pure as confirmed from their PXRD patterns (Figure C). The COF retains its crystallinity even after being exposed to basic conditions (1 M KOH, 12 h, Figure D). From 77 K N2 adsorption–desorption isotherms, the Brauner–Emmett–Teller (BET) surface area of 1290 and 1440 m2/g was estimated for 1 and 2, respectively.[17] A model-independent Barrett–Joyner–Halenda (BJH) fit to the desorption branch yields pore sizes of 25 Å for 1 and 23 Å for 2. A ∼40% drop in the porosity of 2 is observed when it is treated with KOH (Figure E).

Figure 1

(A) Connolly representation of the structure of IISRP-COF6 showing the presence of uniform hexagonal channels. (B) Mesopores in IISRP-COF6 and their AAA···stacking with optimal interlayer separation for π–π interactions. Color code: C-gray; N-blue; O-red. (C) Pawley fit for IISERP-COF6. (D) Scalability and stability of IISERP-COF6 from powder X-ray diffraction. (E) Porosity comparisons of the samples refluxed in water and soaked in KOH (1 M, 12 h).

(A) Connolly representation of the stn class="Chemical">ructure of IISRP-COF6 showing the presence of uniform hexagonal channels. (B) Mesopores in IISRP-COF6 and their AAA···stacking with optimal interlayer separation for π–π interactions. Color code: C-gray; N-blue; O-red. (C) Pawley fit for IISERP-COF6. (D) Scalability and stability of IISERP-COF6 from powder X-ray diffraction. (E) Porosity comparisons of the samples refluxed in water and soaked in KOH (1 M, 12 h).

Synthesis and Characterizations of the COF-Derived Carbons

A family of composites was synthesized by heating 1/2 loaded with n class="Chemical">Ru(acac)3 at different temperatures (250, 280, 310, and 350/370 °C). At this elevated temperatures, Ru(acac)3 oxidizes, while the COFs undergo carbonization (Figure S3). The trapping of Ru(acac)3 in the nanopores of the COFs helps to grow truly small-sized (<5 nm) RuO2 nanoparticles in this capping agent-free method. The soft pyrolysis restrains RuO2 from growing into larger particles. We have used the COF, as they are known to pyrolyze to high surface area carbons with accessible porosity.[22] The best-performing catalysts were IISERP-COF1_RuO2@370 (Composite-I) and IISERP-COF6_RuO2@350 (Composite-II); they are mostly discussed here (Scheme ).

Scheme 1

Schematic Representation of the Entire Process of the Catalyst Preparation

From the PXRD patterns and the thermogravimetric analysis (TGA) plots of the pyrolysis products, we find that the complete n class="Chemical">carbonization of the RuO2-loaded COF occurred at 350–370 °C (Figures S4–S7). The COF pyrolysis was carried out in the absence of RuO2 to establish the nature of the carbon that was produced. We noticed that the resulting carbon was micro-mesoporous with high surface areas (BET = 517 m2/g; Langmuir: 840 m2/g; BJH pore size: 27 Å). A DFT fit (C-spherical pore model) indicates the presence of hierarchical pores. Notably, this was found to be graphenic carbon with a significantly ordered structure as confirmed by PXRD and Raman (Figures S8 and S9). Formation of such graphenic carbon from tartaric acid at these temperatures (<400 °C) is known,[60] but we reproduced this by carrying out the pyrolysis of tartaric acid at 370 °C and found that the resulting graphenic carbon was nonporous (Figure S10). These observations emphasize the importance of having to use the COF as a single source precursor for generating porous graphenic carbons.

The fluffy needle-shaped morphology of the carbons was very similar to those of the respective parent n class="Chemical">COFs as can be seen from the field emission scanning electron microscopy (FESEM) images (Figure ). The elemental mapping confirms the homogeneity of the Ru distribution in both the composites (Figures S11–S40). A comparison of the adsorption isotherms of the composites prepared at different temperatures reveals a systematic drop in the porosity with increasing temperature (Figure ).

Figure 2

(A) FESEM Images of IISERP-COF1. (B)TEM images of IISERP-COF1 (C) N2 sorption at 77 K of IISERP-COF1 and RuO2-loaded IISERP-COF1. (D) FESEM image of IISERP-COF1_RuO2@370. (E) TEM Images of IISERP-COF1_RuO2@370. (F) HRTEM image of IISERP-COF1_RuO2@370 showing the lattice fringes. (G) FESEM Images of IISERP-COF6. (H)TEM images of IISERP-COF6 (I) N2 sorption at 77 K of IISERP-COF6 and RuO2-loaded IISERP-COF6. (J) FESEM image of IISERP-COF6_RuO2@350. (K) TEM Images of IISERP-COF6_RuO2@350. (L) HRTEM image of IISERP-COF6_RuO2@350 showing the lattice fringes. B and H show the presence of uniform micropores all along the surface of the COFs.

(A) FESEM Images of IISERP-n class="Chemical">COF1. (B)TEM images of IISERP-COF1 (C) N2 sorption at 77 K of IISERP-COF1 and RuO2-loaded IISERP-COF1. (D) FESEM image of IISERP-COF1_RuO2@370. (E) TEM Images of IISERP-COF1_RuO2@370. (F) HRTEM image of IISERP-COF1_RuO2@370 showing the lattice fringes. (G) FESEM Images of IISERP-COF6. (H)TEM images of IISERP-COF6 (I) N2 sorption at 77 K of IISERP-COF6 and RuO2-loaded IISERP-COF6. (J) FESEM image of IISERP-COF6_RuO2@350. (K) TEM Images of IISERP-COF6_RuO2@350. (L) HRTEM image of IISERP-COF6_RuO2@350 showing the lattice fringes. B and H show the presence of uniform micropores all along the surface of the COFs.

From the high-resolution transmission electron microscopy (HRTEM) images, the majority of the n class="Chemical">RuO2 particles are in the size limit of ∼2.5–4 nm (Figures and S41). SAED patterns of the composites reveal their poor long-range crystallinity (Figures S42 and S43), yet there are crystalline regions giving rise to lattice fringes. The indexing of the lattice fringes from the HRTEM shows that the different crystal facets corresponding to the [200], [101], and [110] reflections of the tetragonal RuO2 (ICDD: 00-040-1290) are exposed in the composites (Figure F,L). Some of these facets are known to be highly active for OER.[52,53,59]

Interestingly, the Raman and infrared spectra (IR) (Figures S44–S47) showed that the many of the functionalities of the COF were nearly intact in this soft pyrolysis-generated n class="Chemical">carbon. The aforementioned fact corroborates well with the large C/N ratio (neat COF = 3; RuO2@C = 4.25) obtained from the elemental analysis. The Raman spectra show the characteristic graphenic D and G band being present in this carbon (Figure S47). To verify the electronic conductivity of this graphenic composite, we carried out four-probe measurements. The conductivity of IISERP-COF1_RuO2@370 was 2 × 10–5 S/cm (Figure S48), which is quite moderate compared to the traditional graphenic carbon (∼103 S/cm) synthesized at high temperatures (>800 °C).[61,62] This means we do not get any major advantage of electronic modulation from this COF-derived carbon support.[19]

The X-ray photoelectron spectroscopy (XPS) analyses on IISERP-n class="Chemical">COF1_RuO2@370 reveal the presence of characteristics peaks for RuO2 (Figure S67, Ru 3p1/2: 485.0 eV and Ru 3p3/2: 462.9 eV).[63] In the C 1s spectra, the peak at 280.9 corresponds to RuO2, and the peak at 282.5 eV is assignable to RuO3, occurring as a surface defect.[64,65] The other higher binding energy peaks can be assigned to C 1s 285 and 287. N 1s spectra have two features of note. The peaks at 399.4 and 400.7 eV signify the presence of pyridinic and pyrrolic nitrogen in the carbon framework, and these are positively shifted, potentially from weak Ru–N interactions.[66] O 1s spectra show the presence of a peak at 532.2 eV which corresponds to Ru–O bonds.[63] This was consistent with the 459 and 506 cm–1 bands observed in the IR spectra (Figure S46), which is assigned to the Ru–N and Ru–O stretching frequency.[67,68]

Electrochemical Measurements

Electrochemical measurements were carried out using a traditional three-electrode setup using Hg/HgO and platinum flag as the ren class="Chemical">ference, counter, and catalyst-coated glassy carbon as a working electrode, respectively. All potentials measured by using Hg/HgO were converted into the reversible hydrogen electrode (RHE) scale by calibrating Hg/HgO in 1 M KOH solution, saturated with H2. The catalyst mass loading was maintained to be 5 μg for all electrochemical studies. There was a sharp rise in current after 1.4 V due to the oxygen evolution. The onset potentials for composite-I and composite-II were 1.37 and 1.38 V, respectively. The overpotentials (η) at a current density of 10 mA/cm2 calculated for composite-I and composite-II were 210 and 217 mV, respectively, (Figure A) which is to the best of our knowledge is the lowest reported overpotential for a RuO2 catalyst till date.

Figure 3

(A) LSV curves showing the overpotentials for composites prepared from IISERP-COF1 and IISERP-COF6. (B) Comparison of the Tafel slope for composite-I and composite-II. (C) Nyquist plots showing the resistivity for composite-I and composite-II. (D) Comparative LSV plots for composite-I showing the systematic increase in activity with the increase in their synthesis temperature. This is related to the controlled formation of graphenic carbon.

(A) LSV curves showing the overpotentials for composites prepared from IISERP-n class="Chemical">COF1 and IISERP-COF6. (B) Comparison of the Tafel slope for composite-I and composite-II. (C) Nyquist plots showing the resistivity for composite-I and composite-II. (D) Comparative LSV plots for composite-I showing the systematic increase in activity with the increase in their synthesis temperature. This is related to the controlled formation of graphenic carbon.

The overpotentials to achieve a current density of 50 mA/cm2 for composite-I and n class="Chemical">composite-II were 274 and 269 mV, respectively, which are quite remarkable. Not only the low overpotential but these composites also deliver current density as high as 200 mA/cm2 at a potential of only 1.6 V which is highly desirable for faster kinetics. From a plot of log I versus potential, the Tafel slope (with 65% IR-compensation) was calculated for composite-I and composite-II to be 65 and 67 mV/dec, respectively (Figure B), which indicates trouble-free mass and charge transfer at the electrode and electrolyte contact yielding fast kinetics for OER. Electrochemical impedance spectra comparisons show that composite-I has less resistance than composite-II (Figure C).

As composite-I has better performance as n class="Chemical">composite-II (Figure A,D), the electrochemical stability studies were carried out with Composite-I. Highly reproducible loops over continuous 500 CV cycle measurements at 50 mV/s in the Faradaic region (1–1.4 V, Figure A) confirmed Composite-I’s cyclic stability. The cyclic voltammetry (CV) profile indicates well-featured redox peaks in the cathodic and anodic scans, corresponding to the Ru(IV)–Ru(V) redox couple. The characteristic features of the CV profile in the first cycle were completely intact even after 500 cycles. Also, there was some small contribution from the capacitive current in the non-Faradaic region. To further establish the electrochemical stability of the composite, chronoamperometry measurements were carried out. A moderately stable current was generated over 12 h (Figure A). The Faradaic efficiency of the composite was calculated from the rotating ring-disk electrode experiment carried out by applying a series of current (Figure S68). Faradaic efficiency for composite-I was evaluated to be 98% at 0.25 mA which further systematically reduces to 25% at 2.65 mA disc current. This reduction in Faradaic efficiency at higher current was due to the rapid release of oxygen at the disc electrode, and all of which do not get collected on the ring, as some escapes. The same is the cause for the drop in the activity during the chronoamperometry, something we noticed consistently during multiple measurements. In more detail, the decline could occur due to two possibilities, either RuO2 leaches out from the COF or RuO2@C shows poor adhesion with the electrode. To test this, we carried out the experiment with a larger electrode and with a larger sample, and then, we collected the postcatalysis sample and measured the PXRD which showed RuO2 being intact in the carbon (Figure ). Thus, it appears that the rapid evolution of oxygen witnessed by the vigorous bubbling most likely weakens the adherence of RuO2@C with the electrode. This disintegration of the catalyst from the electrode due to rapid oxygen bubbling could represent a practical problem for all high activity materials and this leaves sufficient room for the optimization of the material/electrode preparation.

Figure 4

(A) Chronoamperometry plot showing moderate stability in the current outputs over 12 h. Inset: the Ru(IV)–Ru(V) redox couple in composite-I showing stability over 500 CV cycles. (B) Quantification of the evolved oxygen from GC over two subsequent cycles. (C) Comparison of overpotential for composite-I and composite-II with some of the top performing OER catalyst. (D) FE-SEM, PXRD, and XPS characterizations of the spent catalyst.

(A) Chronoamperometry plot showing moderate stability in the current outputs over 12 h. Inset: the Ru(IV)–n class="Chemical">Ru(V) redox couple in composite-I showing stability over 500 CV cycles. (B) Quantification of the evolved oxygen from GC over two subsequent cycles. (C) Comparison of overpotential for composite-I and composite-II with some of the top performing OER catalyst. (D) FE-SEM, PXRD, and XPS characterizations of the spent catalyst.

As another realistic estimation of the catalyst efficiency, we quantified the oxygen evolved during the OER. A systematic increase of the n class="Chemical">oxygen amount from 0 to 248 mmol/h/g was observed (Figure B); this is higher than the amounts reported for other COF nanoparticle-based OER catalysts.[16,19,69] Cyclic stability was confirmed by repeating the same experiment. Each time, before the measurement, the cell was fully degassed for 1 h. Some drop in the oxygen amount in the second cycle is due to the bubble formation at the electrode surface. Despite this, the superior performance of the composites compared to many top-performing OER catalysts can be seen from Figure C. Composite-I had good structural integrity under these electrochemical conditions (1 M KOH, 1–1.6 V, 500 cycles), which was confirmed by PXRD, scanning electron microscopy (SEM), and XPS studies (Figures D and S69–S71). Notably, the presence of Ru-3p peaks (Ru 3p1/2: 489.9 eV and Ru 3p3/2: 463.8 eV) in the spent catalyst means, at the end of the redox cycles, RuO2 is recovered.

The TOF for the OER was calculated to be 0.103 at 300 mV for composite-I. This TOF is notably higher than the neat n class="Chemical">RuO2,[50] RuO2@CNx,[50] and Ir/C[58] but is lower than some of the Fe-based amorphous catalysts[38,39,45] and the recently reported Ni3N-COF19-based catalyst (Figures S72, S73, and see Table S3). However, we do see a more vigorous evolution in this case, which is explained by the superior kinetics (Tafel slopes: 65 mV/decade for composite-I). The observed kinetics is inferior only to Gelled FeCoW oxy-hydroxide[38] and Fe(PO3)2/Ni2P.[45] A comparison of the OER overpotentials for selected high-performing materials with our composites at current densities of 10 and 50 mA/cm2 is shown in Figure C and listed in Table S1. The superiority of the performance of this composite originates not only from the small-sized RuO2 nanoparticles grown at a precise temperature (Figures , 3D, S41, and S74–S78) but also from the high degree of crystallinity evidenced from the lattice fringes arising from the stacked structure of the graphitic sheets (Figure S8), quite comparable to the recently reported COF-derived carbons.[9]

Conclusions

To summarize, we have developed a heterogeneous RuO2-based OER catalyst using the soft-pyrolysis method by employing the n class="Chemical">COF as a sacrificial precursor in open air. This favors the growth of small-sized RuO2 nanoparticles in the nanoconfinement of porous graphenic carbon. This brings both energy benefits and opens up the possibility of making superior catalysts via soft pyrolysis of the COF. Considering that ruthenium compounds are the benchmark catalysts for OER, the performance reported for this new RuO2 catalyst clearly raises the bar.

Experimental Section

Synthesis

All chemicals were purchased from Sigma-Aldrich and used without further purification. 4,4′,4″-((1,3,5-triazine-2,4,6-triyl)tris(n class="Chemical">oxy))tribenzaldehyde was synthesized according to the reported procedure.[70]

Synthesis of IISERP-COF1

This was synthesized via reported procedure S2 with slight modification. In a 20 ml pyrex tube, 0.112 g (0.25 mmol) of 4,4′,4″-((1,3,5-trin class="Chemical">azine-2,4,6-triyl)tris(oxy))tribenzaldehyde and 0.044 g (0.4 mmol)1,4-phenylenediamine were dissolved in a mixture of 3 mL Mesitylene, 3 mL ODCB, and 5 mL EtOH. The mixture was then stirred at room temperature for 60 min. During stirring, off-white colored slurry was formed. About 0.3 mL of aqueous acetic acid (6 M) was added to this slurry, and contents were stirred for another 30 min. The tube was then purged with N2 and flash frozen in liquid nitrogen. Then, the tube was sealed under a blanket of nitrogen and heated to 120 °C for 3 days in a programmable oven. When the reaction mixture was cooled to room temperature, an off-white colored fluffy powder was obtained, which was isolated by vacuum filtration. The product was subjected to vigorous washing with dimethylformamide (DMF), dimethylacetamide (DMA), acetone, and finally with plenty of tetrahydrofuran (THF). It was dried in hot air oven before further characterization. The yield was 76% with respect to the aldehyde. Formula: C66N12O6H30; molecular weight: 1087.04 g/mol. CHN values (calculated values within bracket): C: 68.2 (72.92%); H: 4.18 (2.78%); N: 16.06 (15.46%). Note that the COF utilizing these monomers are highly reproducible, and many COFs are being reported by other groups using this trialdehyde monomer since our first report of IISERP-COF1.[17]

Synthesis of IISERP-COF6

In a 20 mL pyrex tube, 4,4′,4″-((1,3,5-trin class="Chemical">azine-2,4,6-triyl)tris(oxy))tribenzaldehyde 0.112 g (0.25 mmol) was dissolved in a mixture of 3 mL mesitylene, 3 mL orthodichlorobenzene (ODCB), and 5 mL EtOH. Following this, hydrazine hydrate (35%) 0.03 mL was added to the mixture under vigorous stirring at room temperature. Contents were stirred for 60 min. During stirring, off-white colored slurry was formed. About 0.25 mL of aqueous acetic acid (6 M) was added to the mixture and stirred for another 30 min. The tube was then purged with N2 and flash frozen in liquid nitrogen. Then, the tube was sealed under a blanket of nitrogen and heated to 90 °C for 3 days in a programmable oven. After the reaction mixture was cooled to room temperature, the product, an off-white colored fluffy powder, was isolated by vacuum filtration. It was subjected to vigorous washing with DMF, DMA, acetone, and finally with plenty of THF. It was dried in a hot air oven before further characterization. The yield was 78% with respect to the aldehyde. Formula: C48N12O6 H30; molecular weight: 870.842 g/mol. CHN values (calculated values within bracket): C: 66.61(66.20); H: 3.87 (3.47); N: 20.29 (19.30). This is a cost-effective COF and forms reproducibly in good yields and is quite crystalline. The discrepancies in the calculated and observed CHN values most likely stem from the guest solvent molecules in the pore and the presence of some unreacted terminal aldehydes. These are consistent with the observations from the TGA and the 13C-SSNMR.

Synthesis of IISERP-COF1_RuO2@370

The synthesis of IISERP-COF1_n class="Chemical">RuO2 was carried out in two steps. A solution of Ru(acac)3 (100 mg) in 10 mL of MeOH was slowly added to a sonicated suspension of IISERP-COF1 (250 mg) in 50 mL THF. After the inorganics addition, the mixture turned to a bright pink color, and it was allowed to stir for 24 h at room temperature. Following this, the suspension was allowed to settle down, and the solid was recovered by centrifugation. This solid was thoroughly washed with THF and MeOH [note that Ru(acac)3 is highly soluble in these solvents]. Thus, isolated pink-colored solid, Ru(acac)3 loaded in COF, was heated at 370 °C for 3 h under air. After cooling to room temperature, the black powder was collected. CHN values: C: 74.64%; H: 1.61%; N: 8.84%.

Synthesis of IISERP-COF6_RuO2@350

The synthesis of IISERP-COF6_RuO2 was carried out in two steps. A solution of n class="Chemical">Ru(acac)3 (100 mg) in 10 mL of MeOH was slowly added to a suspension of IISERP-COF6 (250 mg) in 50 mL THF. After this inorganics addition, the mixture turned from a yellow suspension to a bright pink color. It was allowed to stir for 24 h at room temperature. Following this, the suspension was allowed to settle down, and the solid was recovered by centrifugation and washed. The pink-colored solid, Ru(acac)3 loaded in COF was the heated at 350 °C for 3 h under air. When cooled to room temperature, a homogeneous black powder was obtained. This was washed vigorously with water, methanol, and THF. No colored compounds were seen in the postwash solutions. CHN values: C: 75.52%; H: 0.57%; N: 7.66%.

Synthesis of IISERP-COF1@370

To verify the role of COF-derived n class="Chemical">carbon in the synthesis, IISERP-COF1 was pyrolyzed at 370 °C under air. The black powder was washed with water, THF, and methanol. CHN values: C: 59.77%; H: 1.86%; N: 14.54%. Note that the presence of substantial quantities of nitrogen in these carbons is generated from COF pyrolysis.

Note that the optimal pyrolysis temperature for the preparation of a homogeneous n class="Chemical">carbon-supported RuO2 catalyst was established from systematic analysis using the PXRD and TGA (Figures S4–S7). The pyrolysis temperature was chosen based on the decomposition temperatures of the COF.

Analytical Characterizations

Powder X-Ray Diffraction

Powder XRD was carried out using a full-fledge Bn class="Chemical">ruker Pxrd instrument. In some cases, the data were recorded using a Rigaku Miniflex-600 instrument and processed using PDXL software.

Thermogravimetric Analysis

Thermogravimetry was carried out on the NETSZCH TGA–DSC system. The routine TGAs were carried out under n class="Chemical">N2 gas flow (20 mL/min) (purge + protective), and samples were heated from 25 to 550 °C at 4 K/min.

IR Spectroscopy

IR spectra were obtained using a Nin class="Chemical">colet ID5-attenuated total reflectance IR spectrometer operating at ambient temperature. The KBr pellets were used for IR data collection.

Field Emission SEM

FESEM images were n class="Chemical">collected using the Ultra Plus field emission scanning electron microscope with the integral charge compensator and embedded EsB and AsB detectors. Oxford X-max instruments 80 mm2. (Carl Zeiss NTS, Gmbh), Imaging conditions: 2 kV, WD = 2 mm, 200 kX, Inlens detector. For SEM images, as an initial preparation, the samples were ground thoroughly, soaked in THF for 30 min, and were sonicated for 5 min. These well-dispersed samples were drop-casted on a silicon wafer and dried under vacuum for 12 h.

High-Resolution Transmission Electron Microscopy (HRTEM)

The FEI (Jeol n class="Chemical">FEG 2100F is the model) high-resolution transmission electron microscope (HR-TEM) equipped with a field emission source operating at 300 KeV was used.

Adsorption Studies

All adsorption studies were carried out using a 3-FLEX pore and surface area analyzer, micromeritics and few cases “n class="Chemical">Autosorb IQ”, Quantachrome.

X-ray Photoelectron Spectroscopy (XPS)

XPS measurements were carried out using the Thermo K alpha + spectrometer using microfocused and monochromated Al Kα radiation with an energy of 1486.6 eV. The base pressure of the instrument was 1 × 10–9 Torr. The pass energy for spectral acquisition was kept at 50 eV for individual n class="Chemical">core levels. The electron flood gun was utilized for providing charge compensation during data acquisition. The peak fitting of the individual core levels was done using Origin Pro software with a Shirley-type background.

The electrochemical measurement was carried out using a Bio-Logic instrument with techniques such as CV, linear sweep voltammetry (LSV), and chronoamperometry. Typically, a three-electrode test cell was fabricated using Hg/HgO and n class="Chemical">platinum flag as the reference and counter electrode, respectively. The catalyst-coated glassy carbon electrode was used as the working electrode.

Preparation of Working Electrode

This was prepared by coating the catalyst on the n class="Chemical">carbon disk and with subsequent drying under an IR lamp. The catalyst mass loading was maintained to be 5 μg for all electrochemical studies.

Electrolyte

For all electrochemical measurements, the degassed 1 M KOH was used as the n class="Chemical">electrolyte. All potentials were converted to RHE according to the following equation

iR Correction

Excluding the resistances arising from the electrolyte, ren class="Chemical">ference and counter electrodes are necessary. For this, iR compensation needs to be done on the LSV data. However, excluding the electrolyte resistance completely from the system is not realistic. This is due to internal resistance from the electrochemical system. In this study, we have corrected the LSV data using 65% iR compensation. The Tafel plot is derived from the LSV data, which is already iR compensated.

Evolved Oxygen Quantification Setup

The testing was carried out in an air-tight electrochemical cell. A chronoamperometric study at 1.44 V (vs RHE) followed by quantification of headspace gas by gas chromatography (GC) was carried out. This measurement was carried out over an hour via injection of the headspace gas at a 20 min time interval.