Nitrogen‐doped Carbon–CoOx Nanohybrids: A Precious Metal Free Cathode that Exceeds 1.0 W cm−2 Peak Power and 100 h Life in Anion‐Exchange Membrane Fuel Cells

Abstract Efficient and durable nonprecious metal electrocatalysts for the oxygen reduction (ORR) are highly desirable for several electrochemical devices, including anion exchange membrane fuel cells (AEMFCs). Here, a 2D planar electrocatalyst with CoOx embedded in nitrogen‐doped graphitic carbon (N‐C‐CoOx) was created through the direct pyrolysis of a metal–organic complex with a NaCl template. The N‐C‐CoOx catalyst showed high ORR activity, indicated by excellent half‐wave (0.84 V vs. RHE) and onset (1.01 V vs. RHE) potentials. This high intrinsic activity was also observed in operating AEMFCs where the kinetic current was 100 mA cm−2 at 0.85 V. When paired with a radiation‐grafted ETFE powder ionomer, the N‐C‐CoOx AEMFC cathode was able to achieve extremely high peak power density (1.05 W cm−2) and mass transport limited current (3 A cm−2) for a precious metal free electrode. The N‐C‐CoOx cathode also showed good stability over 100 hours of operation with a voltage decay of only 15 % at 600 mA cm−2 under H2/air (CO2‐free) reacting gas feeds. The N‐C‐CoOx cathode catalyst was also paired with a very low loading PtRu/C anode catalyst, to create AEMFCs with a total PGM loading of only 0.10 mgPt‐Ru cm−2 capable of achieving 7.4 W mg−1 PGM as well as supporting a current of 0.7 A cm−2 at 0.6 V with H2/air (CO2 free)—creating a cell that was able to meet the 2019 U.S. Department of Energy initial performance target of 0.6 V at 0.6 A cm−2 under H2/air with a PGM loading <0.125 mg cm−2 with AEMFCs for the first time.

Polymer electrolyte membrane fuel cells have long been considered as the future power source for transportation systems and portable devices due to their environmental friendly operation and high energy conversion efficiency. [1,2] Proton exchange membrane fuel cells (PEMFCs) are currently widely accepted as the most promising alternatives to internal combustion engines,and PEMFCs are already being used to power thousands of fuel cell electric vehicles (FCEVs). However,t he widespread application of PEMFCs has been limited by high costs,i ncluding the use of platinum group metal (PGM) electrocatalysts,w hich account for approximately one-quarter of the cost of PEMFC-based FCEV systems. [3] In recent years,a nion exchange membrane fuel cells (AEMFCs) have been highly touted as apossibly much lower cost electrochemical powerplant than PEMFCs.From acatalytic perspective,the alkaline environment means fundamentally enhanced kinetics for the oxygen reduction reaction (ORR) in AEMFCs compared to PEMFCs.This can allow for the use of ORR catalysts at the AEMFC cathode that are PGM-free,l ike Ag, [4] or even precious metal (PM)-free and hence less costly.T he alkaline AEMFC environment also widens the possible materials chemistries throughout the FCEV system, which can allow for the use of more affordable bipolar plates and less expensive membranes.I na ddition, many of the highest performing AEMFCs in the literature have operated at low cathode pressures,meaning that the air loop in the balance of plant could possibly be simplified. [3] In recent years,asignificant amount of research has been conducted and great progress has been made in the search for aP GM-free (or preferably PM-free) ORR electrocatalyst in both acid and alkaline media. Much of the work in this area has focused on nonprecious transition metal-based materials or metal-free nitrogen-carbon catalysts for the ORR in alkaline media, including metal oxides, [5] graphitic carbons [6] and metal-carbon composites. [7][8][9] Among these,h eteroatomdoped carbon materials coupled with transition metals such as nickel, cobalt, iron and manganese have been widely accepted as the most promising candidates to replace Pt, with several catalysts showing very high ORR activity in ex-situ rotating disk electrode (RDE) experiments.U nfortunately,t od ate, high ex-situ activity has not been translated in the literature into high performance in operating AEMFCs.Infact, the best performing AEMFCs in the literature using aP M-free cathode have been able to achieve ap eak power density of only 0.20-0.70 Wcm À2 and maximum achievable current density of less than 2.0 Acm À2 . [6,9,10] Accompanying these lower than desired performance metrics is the fact that the in-cell stability of PM-free catalysts generally remains unexplored. [6,9] Thus,P M-free catalysts are currently not able to compete with PGM-based catalysts in AEMFCs in terms of single cell performance (1.5-1.9 Wcm À2 )a nd durability (ca. 500 h). [11][12][13][14][15] PM-free cathode catalysts have also not been able to even come close to meeting the 2019 U.S. Department of Energy (DOE) targets for performance (> 0.6 Va t6 00 mA cm À2 on H 2 / air;m aximum pressure of 1.5 atm a )o re nabling AEMFCs with the target total PGM loading of 0.125 mg PGM cm À2 . [16] Thea bility to meet these DOE targets in an operating fuel cell is not only af unction of the intrinsic activity of the catalyst, but also due to the accessibility of active sites during operation since mass transport is more complex in fuel cell catalyst layers (CLs) than liquid-based thin film ex-situ RDE experiments. Therefore,i ti si mportant to consider the final application in designing the chemistry and structure of new catalysts for fuel cell applications.
TheORR activity of PM-free M-N-C (M-Fe, Co,Mn, Ni, etc.)catalysts is generally thought to come from defect-laden active sites [17] where the transition metal (M), nitrogen (N) and carbon (C) coexist. Thec hallenge with traditional structures,where the transition metal is supported by nitrogen doped carbon (N-C), is that the catalysts tend to have very low active site density.This means that even though acatalyst active site may have ahigh turnover frequency,the volumetric reaction rate can be quite low.L ow volumetric active site density translates directly into thick fuel cell CLs with notoriously poor mass transport properties,and, by extension, lower in-cell performance than desired. [9,10] Therefore,t o achieve PM-free catalysts with high in-cell AEMFC performance,itisnecessary to develop electrocatalysts where their morphology may facilitate ah igher density of M-N-C sites, and facilitate facile bulk mass transport.
Carbon shell embedded nanomaterials have recently attracted strong interest for energy conversion and storage applications. [18][19][20][21] In this design, ac arbon shell is deposited onto the active material, providing physicochemical protection, electronic conductivity,a sw ell as high interaction area between the active material and the carbon. This is exactly the family of properties that are expected to be advantageous for PM-free AEMFC cathodes (as long as the inter-particle pore structure can be controlled to allow for facile mass transport). Here,w er eport the synthesis and excellent properties of aw ell-defined 2D,p lanar-structured ORR electrocatalyst with cobalt oxide (CoO x )embedded into acasing of nitrogendoped graphitic carbon, denoted as N-C-CoO x .T he N-C-CoO x was fabricated via af acile,s calable heat treatment in aNaCl template,which simultaneously decomposed glucose, cobalt nitrate and ethylenediaminetetraacetic acid (EDTA) at 700 8 8C. TheNaCl template was chosen because it provides athermally stable supporting surface that is inexpensive and easy to recycle.G lucose was used as the carbon source because it is known to form graphitic carbon at relatively low temperature,below 750 8 8C, [19] which helps to avoid significant loss of nitrogen from the EDTAd uring calcination. An illustration of the synthesis process is provided in Figure 1, and the synthesis details are provided in the Supporting Information.
Ther esulting catalyst was comprised of 75 wt. %C oO x (determined by thermogravimetric analysis,F igure S1 in the Supporting Information), 2% nitrogen (determined by X-ray photoelectron spectroscopy,X PS,F igure S2 and Table S1) and the balance carbon. TheX-ray diffraction pattern for N-C-CoO x is shown in Figure S3. Theb road peak observed at 23.58 8 can be attributed to the graphite (002) reflection from the carbon. [22] Thed iffraction peaks at 31.  (Figure S4a,b,c) showed planar, thin (20-50 nm) carbon nanosheets with embedded CoO x nanoparticles ( Figure S4d). To show the importance of EDTAi nf orming the 2D nanostructure,the synthesis procedure was performed in the absence of EDTA, and though the resulting C-CoO x catalyst still exhibited an embedded nanoparticle morphology,t he sheets were no longer planar ( Figure S5).
Thec arbon nanosheet structure was further investigated by TEM. Lower resolution TEM images (Figure 2a,b) confirmed the 2D planar structure with densely packed CoO x Overlaying the Co and Cs ignals (Figure 2i)a lso clearly showed that the CoO x nanoparticles were confined in the N-C matrix. This structure allows for the broad distribution and high density of reactive metal-nitrogen moieties.
Theelectrocatalytic activity of the N-C-CoO x catalyst was evaluated by cyclic voltammetry (CV) and RDE measurements in O 2 -saturated 0.1m KOHs olution, and compared to control materials that were:i )C oO x embedded in N-free carbon (C-CoO x ); ii)r aw CoO x without any encapsulation (CoO x ); iii)s ulfuric acid treated N-C-CoO x (leaving elemental Co at the N-C sites,but removing bulk CoO x ), denoted as SA-Co-N-C;a nd iv) N-C prepared without any CoO x .A s shown in Figure 3, all three components of the N-C-CoO x catalyst was necessary to achieve high activity.I ncreased activity was shown not only by the more positive position of the ORR cathodic peaks in the CV in Figure 3a, [7] but by more positive half wave and onset potentials in the RDE voltammograms (Figure 3b). In the RDE environment, CoO x showed the lowest ORR activity,t hough after the CoO x was coated with carbon (C-CoO x )t he charge transfer resistance was significantly decreased ( Figure S7), which resulted in considerably improved activity.
After the inclusion of Ninthe carbon matrix, the N-C-CoO x catalyst showed the highest activitywith the highest half-wave (0.84 V vs.R HE) and onset (1.01 Vv s. RHE, Figure 3b)p otentials-and the lowest Tafel slope (62 mV dec À1 ,F igure 3c)-very promising results.E ven when compared to commercial Pt/C (BASF,5 0%), the highest performing ORR catalyst in AEMFCs,t he N-C-CoO x performed very well. In fact, the N-C-CoO x half-wave potential was only % 20 mV more negative than Pt/C ( Figure S8), making N-C-CoO x one of the most active PGM-free electrocatalysts reported in the literature to date. [9,[23][24][25] TheORR mechanism on N-C-CoO x was further investigated by rotating ring disk electrode (RRDE) voltammetry at several rotation rates between 400-2500 rpm (Figure 3d)a nd performing aK outecky-Levich analysis. [26] TheK outecky-Levich plots from the RRDE disk are shown in Figure 3e,a nd the peroxide yield from the ring is shown in Figure 3f.T he average number of electrons transferred (n)w as 3.9 and the HO 2 À yield was stable at ca. 3% over the entire potential window of interest (0.30-0.80 Vv s. RHE). Thus,t he overwhelmingly dominant ORR mechanism on the N-C-CoO x catalyst is the four-electron (4e À )r eduction of O 2 ,w ith the first electron transfer being the rate-determining step.T he combination of high activity,l ow production of unwanted peroxide,a nd its high surface area and open structure make N-C-CoO x an ideal candidate material for the AEMFC cathode.
Next, the N-C-CoO x catalysts were mixed with ETFE solid powder ionomers, [27] dispersed in solvent and sprayed onto gas diffusion layers to create PM-free gas diffusion electrodes (GDEs). [14] SEM images of the GDEs (Figure S9a,b) showed au niform distribution of catalyst and ionomer particles as well as avery porous architecture,which is highly beneficial to operando reactant and product mass transfer. [28] Energy-dispersive X-ray spectroscopy (Figure S9d) clearly showed that the ionomer was well integrated with the catalyst (Figure S9c), which suggests that these electrodes are likely to have aw ell-formed triple-phase boundary in operating AEMFCs. Thec athode GDEs were used to construct lab-scale, single-cell AEMFCs that could be used to test the in situ N-C-CoO x activity and stability under realistic operating conditions.T he membrane used in this work was low-density polyethylene (LDPE) (25 mm, IEC = 2.87 AE 0.05 mmol g À1 ) with covalently-bound benzyltrimethylammonium (BTMA) cationic head-groups. [11] First, the operating AEMFCs were fed with pure H 2 and O 2 reacting gases,where afew important observations were made.F irst, the N-C-CoO x catalyst was able to achieve very high kinetic current in the operating AEMFC,100 mA cm À2 at 0.85 V. This in-cell kinetic behavior compares extremely well to the existing state-of-the-art in both AEMFCs [6,9,29] (Figure S10a) and PEMFCs [30,31] (Figure S10c), even though many of the previous works were done at higher temperature (particularly PEMFC). Second, the N-C-CoO x cathode was able to achieve amass transport limited current density of 3Acm À2 and amaximum power density of 1.05 Wcm À2 -both of which are the highest reported values for aP M-free cathode in AEMFCs to date (Table S3, Figure S10b). [6,9,10,15,29,[32][33][34][35][36] Such high power density and achievable current density shows that the reported catalyst, integrated with the ionomer,e nabled the creation of catalyst layers that:i )h ave much lower mass transport resistance than previous PM-free cathodes in operating AEMFCs;a nd ii)a re competitive with PM-free PEMFC cathodes.T hird, by surpassing the 1.0 Wcm À2 threshold, this is the first PM-free electrode that can compete with state-of-the-art Pt/C cathodes in operating AEMFCs.
To further explore its feasibility for commercial use,t he behavior of the N-C-CoO x GDE was evaluated under several additional conditions.F irst, air (CO 2 -free) was used as the oxidant, and the N-C-CoO x GDE was able to support am ass transport limited current of 2.5 Acm À2 and achieve ap eak power density of 0.66 Wcm À2 .Second, the total PGM loading of the MEA was reduced to 0.10 mg PtRu cm À2 (which is below the DOE target of 0.125 mg PGM cm À2 )b y coupling the N-C-CoO x GDE with at hin PtRu/C anode GDE. [32] This very low PGM-loading cell was able to support high peak power density of 0.73 Wcm À2 under H 2 /O 2 reacting gases,equating to aremarkable specific power output of 7.4 Wmg À1 PGM -the highest of any AEMFC to date (Figure S11).
Despite the high performance,a n obvious limiting factor in this cell was water management, which is gaining attention as ac ritical consideration for AEMFC performance and durability, [15,28] which is indicated by the small difference in performance under O 2 and air feeds as well as the curvature of the polarization curve at high current densities.F inally,t he N-C-CoO x GDEs were subjected to short-term stability testing under both H 2 /air (Figure 4c)a t 600 mA cm À2 and H 2 /O 2 ( Figure S12) at 300 mA cm À2 .Under H 2 /air reacting gas flows,t he cell showed promising shortterm stability,w ith as mall voltage loss (ca. 15 %) over the 100 hexperiment. Apolarization curve was collected after the 100 htest, which also showed avoltage loss of around 15 %at 600 mA cm À2 ( Figure S13). After the durability test, the most significant change was in the mass transport regime where the mass transport limited current was surprisingly reduced by 40 %, which will have to be investigated further in future work. Operating under H 2 /O 2 reacting gas flows,the cell also did not see significant degradation over 70 hours,t hough adequate water management was unfortunately not achieved, indicated by spikes in the cell voltage that are most likely due to the accumulation and quick release of water within the low loading, hence thin, anode. After AEMFC stability testing,the cell was disassembled and N-C-CoO x was collected by abrasive removal from the CL and subjected to TEM to examine the evolution in its morphology during testing and to evaluate possible degradation mechanisms.A ss hown in Figure S14, an overwhelming majority of the N-C-CoO x was able to preserve its 2D nanosheet structure (Figure S14 a, b) with embedded CoO x (Figure S14 c, d), providing evidence for excellent operando stability for this cathode.T here is also no obvious interparticle agglomeration between CoO x particles.H owever, as mall amount of cobalt oxide dissolution was observed, which resulted in the formation of void spaces in the Ncontaining carbon. There was also evidence of cobalt redeposition (and re-oxidization) in weak affiliation with the N-C( Figure S15 a-c). However,itshould be noted that the N-C was very stable ( Figure S14 d-f), and an overwhelming majority of the CoO x remained embedded in the N-C matrix. In combination, the ex-situ RDE and operando AEMFC performance and stability of the N-C-CoO x catalyst represents ap romising new pathway to creating high performing PM-free catalysts for AEMFCs and other electrochemical devices.
In conclusion, af acile,l ow cost and scalable method was used to fabricate ah ighly active and stable N-C-CoO x catalyst, which was comprised of CoO x nanoparticles confined in 2D nitrogen-doped carbon nanosheets.The N-C-CoO x had very high ex situ ORR activity,w hich was translated into an operating AEMFC where the catalyst was able to achieve ahigh in situ activity of 100 mA cm À2 at cell voltage of 0.85 V. Additionally,w hen combined with the ionomer in operating AEMFCs,the N-C-CoOx cathode was able to support apeak power density of 1.05 Wcm À2 -unprecedented for apreciousmetal(PM)-free AEMFC electrode.A EMFCs were also prepared where the N-C-CoO x cathode was paired with av ery low PGM loading anode,0 .10 mg cm À2 PtRu, and the cells were able to achieve an ew record for specific power: 7.4 Wmg À1 PGM .T he N-C-CoO x catalyst also showed very promising stability over 100 hofoperation. These results not only significantly narrow the gap between high performing PGM-containing and PM-free ORR cathodes,b ut also point to the promise of AEMFCs as al ow-cost alternative to PEMFCs for both stationary and mobile applications as well as provide anew direction for the design of ORR catalysts in alkaline media. .c )Stability of H 2 /air (CO 2 -free) AEMFC operatingat600 mA cm À2 ;c athode: 2.4 mg cm À2 of N-C-CoO x ,0 .2 MPa backpressure;anode:0 .70 mg cm À2 of PtRu, 0.2 MPa backpressure (data presented with iR-correction).T he membrane in this work was aLDPE-BTMA AEM (IEC = 2.87 AE 0.05 mmol g À1 ) [11] and the ionomer was ETFE-BTMA powder. [27] and J.R.V.todevelop and synthesize the LDPE/ETFE-based membrane/anionomer materials was funded by the UKs Engineering and Physical Sciences Research Council (EPSRC grant EP/M014371/1). We also acknowledge Dr. Lichun Zhang from the University of Connecticut for collecting the TEM/STEM images and Dr.S tavros G. Karakalos at University of South Carolina for collecting and analyzing the XPS data. We thank Rebecca Sisk for creating the cover art.

Conflict of interest
Theauthors declare no conflict of interest.