Polybenzoxazine-Derived N-doped Carbon as Matrix for Powder-Based Electrocatalysts

In addition to catalytic activity, intrinsic stability, tight immobilization on a suitable electrode surface, and sufficient electronic conductivity are fundamental prerequisites for the long-term operation of particle- and especially powder-based electrocatalysts. We present a novel approach to concurrently address these challenges by using the unique properties of polybenzoxazine (pBO) polymers, namely near-zero shrinkage and high residual-char yield even after pyrolysis at high temperatures. Pyrolysis of a nanocubic prussian blue analogue precursor (Km Mnx [Co(CN)6 ]y ⋅n H2 O) embedded in a bisphenol A and aniline-based pBO led to the formation of a N-doped carbon matrix modified with Mnx Coy Oz nanocubes. The obtained electrocatalyst exhibits high efficiency toward the oxygen evolution reaction (OER) and more importantly a stable performance for at least 65 h.


Introduction
Electrocatalytic splitting of water into hydrogen and oxygen is limited by high overpotentials of the oxygen evolution reaction (OER). [1] Low-cost, non-precious, earth-abundant metals and their oxides and oxyhydroxides are subject of efforts in catalyst research aiming to reduce the overpotential needed to drive the kinetically sluggish OER. [2] However, the obtained materials are typically catalyst powders, which have to be brought in contact with the electrode surface maintaining fast electrontransfer pathways especially among the powder particles as well as stable fixation under vigorous gas evolution at high current densities. The lack of strategies for the fixation of powder catalysts leads to suboptimal catalyst performance es-pecially with respect to stability during operation at high current densities. Organic polymers are typically used as binder materials, although they might be insulating if used in too high concentrations, thus blocking catalytically active sites and decreasing the intrinsic activity of the catalyst. [3] Even in alkaline electrolytes, in which abundant transition metal-based catalysts can be used &&ok?&&, Nafion is often used as binder material, which can be considered unsuitable under these conditions due to covering surface active sites and its intrinsic insulating properties. Evidently, it is important to enhance the intrinsic conductivity of electrocatalysts, especially when poorly conducting materials are used or poorly conducting binder materials are added. The addition of carbon nanomaterials such as carbon black, carbon nanotubes, or graphene oxide/reduced graphene oxide is commonly used to enhance the conductivity of related slurry electrodes. [4] However, mixtures of carbon additives with powder-based electrocatalysts do not overcome limitations imposed by the poor adhesion on the electrode surface and typically an unsuitable binder material such as Nafion is used to glue the catalyst/carbon composite powder to the electrode surface.
We present the first application of polybenzoxazines (pBO), a class of high-performance thermosets, as new binder matrices in electrocatalysis to concurrently address the aforementioned issues in powder-catalyst fixation. pBOs consist of highly crosslinked networks of benzoxazine (BO) monomers or oligomers. [5] They arise from Mannich-type reactions of primary amines, aldehydes, and phenol derivatives. [5] Especially, two unique features of this specific polymer class make pBOs highly suitable as precursors for electrocatalyst matrices: i) near-zero shrinkage with respect to the deposited monomer film during polymerization, which allows for maintaining a homogeneous electrode coating after polymerization whereas other types of polymers shrink upon solvent evaporation [a] S. Barwe, Dr. C. Andronescu, Dr. J. Masa In addition to catalytic activity, intrinsic stability, tight immobilization on a suitable electrode surface, and sufficient electronic conductivity are fundamental prerequisites for the long-term operation of particle-and especially powder-based electrocatalysts. We present a novel approach to concurrently address these challenges by using the unique properties of polxbenzoxazine (pBO) polymers, namely near-zero shrinkage and high residual-char yield even after pyrolysis at high temperatures.  1  1  2  2  3  3  4  4  5  5  6  6  7  7  8  8  9  9  10  10  11  11  12  12  13  13  14  14  15  15  16  16  17  17  18  18  19  19  20  20  21  21  22  22  23  23  24  24  25  25  26  26  27  27  28  28  29  29  30  30  31  31  32  32  33  33  34  34  35  35  36  36  37  37  38  38  39  39  40  40  41  41  42  42  43  43  44  44  45  45  46  46  47  47  48  48  49  49  50  50  51  51  52  52  53  53  54  54  55  55  56  56  57  57 hence weakening the catalyst film adhesion on electrode surfaces. ii) Highly crosslinked pBO networks are well known as flame-retardant materials providing high residual char yields after pyrolysis, an important prerequisite for the formation of a stable carbon-based binder matrix. [6] Thus, subsequent polymerization and pyrolysis transform the insulating organic pBO matrix into a conducting carbon matrix [7] with incorporated electrically interconnected catalyst powder particles exhibiting high physical stability. Our strategy is to embed the prussian blue analogue (PBA) potassium manganese hexacyanocobaltate (K m Mn x [Co(CN) 6 ] y n H 2 O) as a model catalyst precursor into a benzoxazine-based precursor matrix. A sequence of polymerization and pyrolysis converts the composite into Mn-and Co-containing nanocubes embedded into a pBO-derived carbon matrix forming a highly stable OER active film on the electrode.

Results and Discussion
Three-step electrode preparation Catalyst particles embedded in a porous N-doped carbon matrix were prepared according to the three-step procedure shown in Scheme 1 and described in detail in the Experimental Section. Thermally induced polymerization of a mixed film of a bisphenol A and aniline-based BO monomer and K m Mn x [Co(CN) 6 ] y drop-coated on a glassy carbon electrode surface led to a polymer layer with incorporated PBA particles. Subsequent pyrolysis transformed the deposited pBO/PBA film into a conducting N-doped carbon matrix incorporating mildly oxidized Mn and Co nanocubes (carbon-Mn x Co y O z ; Scheme 1).
The unique near-zero shrinkage property of pBOs led to a homogenously distributed carbon matrix over the electrode surface even after pyrolysis ( Figure S1 in the Supporting Information). The modified electrode was entirely covered with Mn x Co y O z nanocubes and with relatively larger particles representing agglomerates of the nanocubes ( Figure S1). Polymerization and pyrolysis were performed under Ar flow to prevent complete degradation of pBO. However, although the thermal treatment was performed under inert gas atmosphere, Mn was oxidized by traces of oxygen at higher temperatures. This was also observed after pyrolysis of the pure PBA precursor. The Mn core-level XPS spectrum shows a Mn 2p 3/2 peak with a binding energy of 641.48 eV, representing an oxidized manganese (Mn 2 + /Mn 3 + ) species ( Figure S2). Thermogravimetric analysis (TGA) of the two precursors (PBA and pBO) was employed to gain further insight into the influence of the temperature during the polymerization and pyrolysis processes on the decomposition behavior (Figure 1). In a first step, the temperature was increased stepwise from room temperature to 240 8C inducing BO polymerization, resulting only in a minor weight loss of the pure pBO owing to evaporation of residual solvent. A further increase of the temperature to 600 8C during pyrolysis led to an additional drastic weight loss of 68 % caused by loss of volatile fragments from the degradation of the polymer chain (i.e., aniline). The pBO gave a residual char yield of 27 % after the complete thermal treatment. The behavior of the pure PBA was dominated by three major weight changes during the thermal treatment. First, a weight loss of 23 % occurred during the first heating step, caused by evaporation of crystal water of the PBA. A second weight loss of 35 %, after reaching 350 8C, was owed to the release of cyanide groups of the PBA as cyanogen. The observed behavior conforms well to the pyrolysis of PBAs as previously reported. [8] A weight increase of 5 % further indicates the formation of metal oxides as supported by the previously mentioned XPS results. A mixture of the pBO and PBA with a ratio of 1:1 (wt %) followed a similar behavior as the two starting materials, but the decomposition at high temperature was hampered compared to the pure PBA leading to a residual char yield of 30 %.

Influence of the PBA-to-pBO ratio
The influence of the pBO content on the OER activity of the PBA-derived electrocatalyst was investigated by means of hydrodynamic cyclic voltammetry using electrodes prepared with 0, 2, 10, 17, 28, or 50 wt % pBO with respect to the overall solid content in the drop-coated ink (Figure 2 a, b). The voltam-Scheme 1. Electrode preparation procedure and transformation of the precursor materials into the final catalyst.  1  1  2  2  3  3  4  4  5  5  6  6  7  7  8  8  9  9  10  10  11  11  12  12  13  13  14  14  15  15  16  16  17  17  18  18  19  19  20  20  21  21  22  22  23  23  24  24  25  25  26  26  27  27  28  28  29  29  30  30  31  31  32  32  33  33  34  34  35  35  36  36  37  37  38  38  39  39  40  40  41  41  42  42  43  43  44  44  45  45  46  46  47  47  48  48  49  49  50  50  51  51  52  52  53  53  54  54  55  55  56  56  57  57 mograms of the as-prepared samples indicate a clear decline of the catalytic activity with increasing pBO:PBA ratio ( Figure 2 a). These observations suggest that during the pyrolysis step a catalytically inactive carbon matrix is formed from pBO masking the catalytically active sites on the embedded metal oxide particles. However, consecutive potential cycling (50 cycles, scan rate 50 mV s À1 ) substantially activates the catalytic activity of the modified electrodes prepared with 28 and 50 wt % pBO whereas all other electrodes lost activity because of insufficient catalyst fixation on the electrode surface caused by the low &&ok?&& amount of pBO during film formation (Figure 2 b). Samples before and after electrochemical cycling are referred to as "as-prepared" and "pretreated", respectively. The pretreatment process apparently eliminates amorphous carbon layers, thus exposing more highly active sites to the electrolyte, concomitantly reducing the required overpotential for OER. The carbon removal can be considered as self-limiting. After activation, the catalysts with a pBO content of 50 wt % showed the highest activity. The activation effect explains the different surface morphologies seen in the SEM images of the as-prepared and pretreated electrodes with 50 wt % pBO ( The SEM image of the as-prepared electrode shows the Mn x Co y O z nanocubes covered with an additional layer, whereas the image of the pretreated electrode reveals uncovered and porous Mn x Co y O z nanocubes. Transmission electron microscopy (TEM) of the pretreated samples revealed the presence of a sufficient amount of graphitic carbon providing on the one hand high conductivity and on the other hand a matrix of improved stability ( Figures S3-S5). Mn x Co y O z nanocubes obtained without any addition of pBO showed an initial high activity during the first voltammogram, however, the catalyst detached easily from the electrode surface leading to drastic activity loss (Figure 2 a, b). Stabilization of the Mn x Co y O z nanocubes using Nafion as a binder leads to a substantially less active and less stable film ( Figure S6) compared to the activated sample containing a PBA/pBO ratio of 1:1, thus clearly confirming the benefit of using pBO-derived carbon networks for incorporating catalyst particles compared to the most commonly used binder. SEM revealed the preservation of the cubic structure of the precursor material during the heat-treatment processes (Figure 2 c-e).
Additional STEM EELS maps also show a partial segregation of Mn and Co, Mn-rich oxides and Co/MnCoO core-shell particles ( Figures S5, S10, and S11). Furthermore, XRD and XPS revealed the existence of metal carbide phases ( Figures 5 and 4), which are in good agreement with the literature about pyrolysis of PBA in inert atmosphere at 600 8C. [8] The C 1s spectrum (Figure 4 c) was deconvoluted into five peaks with binding energies at 283.52, 284.22, 285.36, 288, and 291.57 eV corresponding to metallic carbides, sp 2 CÀC, CÀOÀC/CÀN, O=CÀO, and a shake-up feature, respectively. [9] The peak at 283.52 eV can be safely assigned to carbides of Co and Mn owing to its good conformance to the C 1s energies of other transition metal carbides, and is therefore in accordance with the MnCoC phase in Figure 3 e. [10] Deconvolution of the N 1s spectrum revealed the presence of pyridinic N (399.09 eV) and pyrrolic N (401.06 eV) groups (Figure 4 e), and hence the formation of Ndoped carbon as described previously by Zhang et al. for the pyrolysis of the used pBO. [11] The presence of Co 2 + is difficult to conclude from the binding energies of the Co 2p 1/2 and Co 2p 3/2 peaks (Figure 4 b). [12] However, both Co 2p peaks have characteristic satellite features at higher binding energies, supporting the presence of Co 2 + ions. [13] The absence of satellite features in the Mn 2p (Figure 4 a) spectrum and the peak splitting of 5.5 eV in the binding energy of the Mn 3s spectrum (Figure 4 f) confirms the presence of oxidized Mn species. [12,14] The non-symmetric Mn 2p 3/2 peak indicates an overlapping of Mn 2 + and Mn 3 + contributions. The binding energy at 529.18 eV in the O 1s spectrum further confirms the existence of transition metal oxides (Figure 4 d).

Electrochemical characterization
A PBA/pBO ratio of 1:1 revealed the highest activity. Therefore, this ratio was chosen for a detailed study of the electrochemical performance of MnCo nanocubes embedded in a pBO derived N-doped carbon matrix with respect to activity and stability towards OER. Carbon-Mn x Co y O z showed high OER activity affording a current density of 1 mA cm À2 at 1.54 V versus RHE, and a current density of 10 mA cm À2 at 1.60 V versus RHE (Fig-ure 6 a), whereas the pure pyrolyzed pBO does not exhibit significant OER activity. The overpotential of 0.37 V required to reach a current density of 10 mA cm À2 is significantly lower than other Co x Mn y O z -based catalysts, which have overpotentials in the range of 0.39-0.63 V. [15] The Tafel slope of carbon-Mn x Co y O z was 59 mV dec À1 (Figure 3 b), which indicates very fast kinetics and a &&charge&& transfer coefficient (a) very close to 1 (a = 2.303 RT/nF, &&please check equation, I changed a to n and added definitions&&with R the universal gas constant, T the temperature, n the number of electrons transferred, and F the Faraday constant). According to the Tafel slope, the discharge of OH À is the rate determining step. [16] In situ Raman spectroscopy ( Figure S16) indicates that the transition metal oxides are mainly responsible for catalyzing the OER whereas the N-doped carbon matrix facilitates electrical interparticle connection. A recently developed stability assessment methodology [17] was applied to evaluate the carbon-Mn x Co y O z catalyst and showed a negligible change in activity at current densities of 1.4 and 10 mA cm À2 (Figure 6 c). Additional stability measurements were conducted using carbon-Mn x Co y O z -modified Ni foam in a custom-built flow-through cell. [17] During the accelerated galvanostatic stability test with an applied current density of 10 mA cm À2 (with respect to the exposed surface of ø = 8 mm), the corresponding potential increased from an initial value of 1.61 to 1.68 V versus RHE after 67 h (Figure 6 d). After the first 19 h, the potential remained constant without any visible change until the end of the measurement indicating a stable state of the catalyst layer. Taking into account that Ni foam is by itself active for catalyzing the OER, analogous measurements were performed with pure Ni  1  1  2  2  3  3  4  4  5  5  6  6  7  7  8  8  9  9  10  10  11  11  12  12  13  13  14  14  15  15  16  16  17  17  18  18  19  19  20  20  21  21  22  22  23  23  24  24  25  25  26  26  27  27  28  28  29  29  30  30  31  31  32  32  33  33  34  34  35  35  36  36  37  37  38  38  39  39  40  40  41  41  42  42  43  43  44  44  45  45  46  46  47  47  48  48  49  49  50  50  51  51  52  52  53  53  54  54  55  55  56  56  57  57 foam as control. The activity of pure Ni foam increased at the beginning of the measurement owing to the formation of NiO x /NiOOH, however, it later remained constant at an overpotential of about 0.17 V higher than the carbon-Mn x Co y O z -modified Ni foam electrode (Figure 6 d). The observed activity and stability can hence safely be attributed to the properties of the suggested carbon-Mn x Co y O z catalyst.

Conclusions
We suggest a novel approach for the preparation of highly active and stable catalyst-modified electrodes for the oxygen evolution reaction (OER). Benzoxazine (BO) monomers were deposited together with the prussian blue analogue (PBA) K m Mn x [Co(CN) 6 ] y n H 2 O on an electrode surface followed by temperature-induced polymerization, leading to the formation of a homogeneous polymer/PBA composite film owing to the near-zero shrinkage property of polybenzoxazine (pBO). Subsequent pyrolysis formed an N-doped carbon matrix embedding catalytically active Mn x Co y O z nanocubes. The pBO content showed a significant impact on the catalytic activity. Owing to the intrinsic stability of the pBO, only 5 % loss in activity during accelerated galvanostatic stability tests at 10 mA cm À2 on carbon-Mn x Co y O z sprayed on Ni foam after 67 h was observed.

Experimental Section
Electrode preparation Carbon-Mn x Co y O z modified glassy carbon electrodes were prepared according to the following procedure: a glassy carbon rod (ø = 3 mm, L % 1 cm; HTW, Germany) was polished successively with polishing pastes of decreasing particle sizes (3 mm diamond paste, 1 mm and 0.3 mm Al 2 O 3 paste; Leco, USA) to obtain a mirrorlike surface. The polished electrodes were immersed in a 1:1 mixture of ethanol and ultra-pure water and sonicated for 15 min. The dried electrodes were drop-coated with an ink (7.2 mL) containing K m Mn x [Co(CN) 6 ] y n H 2 O (5 mgmL À1 ) and Araldite 35600 CH (Huntsman, USA) (0, 0.1, 0.5, 1, 2, or 5 mg mL À1 ) in acetonitrile (J. T. Baker, Netherlands), leading to an initial loading of 0.5 mg cm À2 . The catalyst inks were sonicated for 20 min prior to drop-coating. The drop-coated electrodes were placed into a three-zone tube furnace and a multistep heating program was applied (160 8C/2 h; 180 8C/ 2 h; 200 8C/2 h; 220 8C/1 h; 240 8C/1 h; 600 8C/2 h) under Ar atmosphere. Electrodes prepared with a pBO/PBA ratio of 1:1 and 5 mgmL À1 PBA had an estimated final loading of 0.15 mg cm À2 calculated from the weight loss of 70 % as shown in TGA. Ni foam electrodes (1.6 mm thickness, A geometric. = 0.64 cm 2 ; Goodfellow, Germany) were spray-coated with a PBA/Araldite ink (ratio 1:1) and heat-treated as described above. The resulting electrodes had a catalyst loading of 0.625 mg cm À2 .

Electrochemical measurements
All electrochemical measurements were performed in a one compartment three-electrode configuration with the modified glassy carbon electrode (GCE, ø = 3 mm; A geometric = 0.071 cm 2 ) as working electrode (WE), a platinum mesh as counter electrode (CE), and a Ag/AgCl/3 m KCl as reference electrode (RE) using an Autolab potentiostat/galvanostat (Metrohm Autolab, Netherlands) at room temperature. Modified glassy carbon rods were fixed onto a stainless-steel holder for rotating disk electrode (RDE) measurements. The glassy carbon rods were insulated by wrapping a PTFE sealing tape very tightly around the rod so that only the disk-shaped catalyst-modified electrode was exposed to the electrolyte. The OER activity of the catalysts was investigated using RDE voltammetry in oxygen-saturated KOH (0.1 m) at room temperature at a scan rate of 5 mV s À1 to reduce capacitive current contributions. Catalyst activation was done by cycling the potential 50 times between 0.0 and 1.0 V vs. Ag/AgCl/3 m KCl at a scan rate of 50 mVs À1 . The longterm measurement procedure consisted of cycles of electrochemical impedance spectroscopy (EIS), cyclic voltammetry (0.0-1.0 V vs. Ag/AgCl 3 m KCl, 5 mV s À1 ) and galvanostatic chronopotentiometry (900 s). [17] The measurements were done either with a RDE (GCE, 1600 rpm) or an electrochemical flow-through cell (Ni foam electrodes) in oxygen-saturated KOH (0.1 m, flow rate: 30 mL min À1 ). [17] In situ Raman spectra were recorded during chronoamperometric measurements at different potentials in KOH (0.1 m). All potentials are referred to the reversible hydrogen electrode (RHE) and converted according to: E RHE = E appl + E 0 Ag/AgCl + 0.059 pH. The pH value was calculated using: pH = 14 + log OH À + log g &&ok?&&, where the OH À concentration was corrected using the KOH purity (85 %) and the activity coefficient of a 0.1 m KOH g = 0.766 (mean of the activity coefficients from Ref. [18]). By this, the pH value for the used 0.1 m KOH solution was determined to be 12.81. The potential was further corrected for the uncompensated ionic resistance of the electrolyte according to E corr RHE = E RHE Ài R. The uncompensated resistance was determined by means of EIS using an AC perturbation of 10 mV (rms) at the open-circuit potential of the respective electrode in the frequency range from 100 kHz to 10 Hz.
A new matrix for powder-based #electrocatalysts based on #polybenzoxazine @ruhrunibochum @icn2nano @icreacommunity SPACE RESERVED FOR IMAGE AND LINK Share your work on social media! ChemSusChem has added Twitter as a means to promote your article. Twitter is an online microblogging service that enables its users to send and read text-based messages of up to 140 characters, known as "tweets". Please check the pre-written tweet in the galley proofs for accuracy. Should you or your institute have a Twitter account, please let us know the appropriate username (i.e., @accountname), and we will do our best to include this information in the tweet. This tweet will be posted to the journals Twitter account @ChemSusChem (follow us!) upon online publication of your article, and we recommended you to repost ("retweet") it to alert other researchers about your publication.