Menny
Shalom
*,
Debora
Ressnig
,
Xiaofei
Yang
,
Guylhaine
Clavel
,
Tim Patrick
Fellinger
and
Markus
Antonietti
Max Planck Institute of Colloids and Interfaces, Department of Colloid Chemistry, Research Campus Golm, 14424 Potsdam, Germany. E-mail: menny.shalom@mpikg.mpg.de
First published on 9th March 2015
Efficient, robust and low cost materials as electrocatalysts for energy-related applications are highly desired for the future of renewable energy production. Here we show a simple method to fabricate nickel nitride (Ni3N) on nickel (Ni) foam for electrocatalytic applications. The Ni3N/Ni-foam exhibits extremely low overpotential (∼50 mV), high current density and excellent stability for the hydrogen evolution reaction (HER) in alkaline solution. In addition, the modified foam demonstrates enhanced activity in the oxygen evolution (OER) and reduction (ORR) reaction compared to original Ni-foam. The activity enhancement can be attributed to the facile formation of a Ni(OH)2 layer on the nitride layer due to improved lattice matching. The formation of the Ni3N/Ni(OH)2 catalyst results in lower overpotentials due to easier water dissociation on the nickel hydroxide layer. In addition, the HER is further improved due to stronger adsorption of hydrogen to the metal nitride than to the pure metal. We believe that the utilization of nickel nitride as an electrocatalyst opens opportunities for energy-related devices such as batteries and fuel cells.
As an alternative to metals and their oxides, metal nitrides attract a lot of attention thanks to their excellent catalytic activities in a variety of reactions for energy related applications.2,15,16 Metal nitrides (Ni–Mo–N or Co–Mo–N) show particularly high activity as electrocatalysts towards the HER.17 Moreover, theoretical calculations predict that metal nitride electrocatalysts possesses many advantages over the pure metals.2 In the hydrogen evolution reaction, the superior activity of the metal nitride is explained on the one hand by the high binding energy for hydrogen, which results in higher hydrogen adsorption to the metal nitride surface. On the other hand, the presence of nitrogen strongly influences the electronic properties of the metal by increasing the density of electrons on the surface. Consequently, metal nitrides are found to have higher (electro)catalytic activities in reduction reactions, compared to the corresponding pure metals.2,16–18 Another strategy to enhance the performance of Ni in the HER through synergistic effects, was shown recently by the deposition of Ni(OH)2 onto Ni electrodes.19,20 The improvement of the catalytic performances was explained by better water dissociation by the Ni(OH)2 near the metal which leads to higher hydrogen adsorption on the Ni.
Here, we show the facile, direct growth of nickel nitride (Ni3N) on Ni-foam by solid state synthesis for electrocatalytic applications. The growth of the material was achieved by using supramolecular complexes of cyanuric acid, melamine and small amounts of barbituric acid (referred here as CMB). The modification of the Ni-foam results in the formation of a rough Ni3N layer at the electrode surface, covered with carbon nitride, which can be removed with repeated cyclic voltammetry scans. The modified foam exhibits high electrocatalytic activity in the OER, ORR and especially in the HER.
In order to study the synthesis intermediates and products, we heated the mixture to different temperatures, from 475 °C up to 600 °C. Fig. S2† shows temperature-dependent X-ray diffraction (XRD) patterns of the modified Ni-foam. At the lowest temperature (475 °C), an amorphous carbon-nitride structure is formed on the Ni-foam by the condensation of CMB complexes to melem.22 Increasing the reaction temperature to 520–550 °C results in the growth of β-Ni3N with lattice parameters calculated for a hexagonal system of a = 4.581 Å and c = 4.344 Å (Fig. 1a and S3†). Stoichiometric deviations and carbon impurities likely lead to lattice adaptions and shifted reflections compared to other Ni3Ns.23 Increasing the synthesis temperatures above 550 °C results in the decomposition of the metastable metal nitride through loss of the nitrogen, hence only the reflections of metallic nickel remain (Fig. S2†). It is also important to note that during the standard synthesis (at 550 °C), the Ni-foam slightly loses its flexibility, while at higher temperatures (<∼600 °C) the elasticity is completely lost making it less usable for electrochemical applications.
We found that the use of the supramolecular precursor is mandatory for the formation of Ni3N. Fig. S4† compares images of Ni-foams, either modified with the CMB complex or melamine. The use of melamine results in the growth of a thick carbon-nitride layer on the surface of the Ni-foam which could not be removed with common physical methods. However, the CMB complex carbon-nitride forms a loose, removable powder while the surface of Ni-foam is modified. Strong intermolecular hydrogen bonds within CMB apparently allow smooth modification of the foam without massive intergrowth of carbon-nitride by-products with the foam surface. The synthetic approach is flexible and equally allows growth of nickel sulfide by exchanging melamine with trithiocyanuric acid in the complex. The modification of the foam with this complex results in the formation of a homogenous Ni/Ni3S2 layer as shown in Fig. S5.† The electrochemical characterization of the Ni/Ni3S2 is however beyond the scope of this paper.
SEM images of the modified foam in comparison to the pristine Ni-foam are shown in Fig. 1, S6 and S7.† While the unmodified Ni-foam surface is rather smooth the modified Ni is covered with a thin layer (∼300 nm) that is composed of Ni3N and amorphous carbon-nitride according to XRD (Fig. 1a), XPS and HRTEM. More detailed, HRTEM analysis shows that the layer contains small particles (<10 nm) with lattice fringes corresponding to Ni3N (Fig. 1a, inset).
XPS measurements of the foam further confirm the formation of carbon-nitride along with Ni3N on the surface (Fig. 2). For the N 1s spectrum several binding energies can be distinguished. The main signal shows occurrence of C–N–C groups (398.9 eV) and tertiary nitrogen groups (399.6 eV), which can be attributed to carbon nitride,22 alongside the metal nitride peak at smaller binding energy (398.1 eV).24 In addition, the C 1s spectrum shows two main peaks for aromatic C–C and C–N–C bonds at 284.8 eV and 287.9 eV, respectively.22 The final step in the preparation of the electrocatalyst includes the removal of the carbon nitride that covers the foam by several cyclic voltammetry scans at positive voltages. Fig. S8a† shows the 10 first oxidation cycles of the modified Ni-foam in which a strong oxidation peak evolves at ∼1.5 V vs. RHE.
Fig. 3a and S8b† compare the 5th and the 500th first CV cycles of the pristine and the modified foam. Only with the later, a strong oxidation (at ∼1.5 V vs. RHE) and reduction peak (at ∼1.35 V vs. RHE) evolve during the cycling while the initially transparent KOH solution changes to yellowish, due to carbon-nitride dissolution from the electrode (Fig. S9†). The oxidation of water with Ni in alkaline solution requires several pre-steps. The first step (0.5–1.35 V vs. RHE) is the formation of stable Ni(OH)2, which is an irreversible process and occurs only in the first cycles. At higher potentials (E > 1.35 V), the Ni(OH)2 species is reversibly oxidized to NiOOH.25 The formation of NiOOH is required to catalyze the OER that sets in at higher potentials (E > 1.53 V). In our measurements, we found that the intensity of the NiOOH oxidation peak reaches a maximum and does not change after ∼500 CV cycles (Fig. S8a and b,†3a). The strong intensity of the NiOOH formation peak indicates that the modification increases the amount of the active Ni species on the surface.
![]() | ||
Fig. 3 (a) Cyclic voltammetry measurements (25 mV s−1 scan rate) (b) cathodic and anodic charging currents measured at 0.86 V vs. RHE plotted as a function of scan rate (the determined double-layer capacitance values of the system are given in Table S1†) (c) linear sweep voltammetry curves (LSV, 1 mV s−1 scan rate) for the oxygen evolution reaction and (d) hydrogen evolution reaction of the Ni/Ni3N-foam compared to the original Ni-foam. (e) Oxygen reduction reaction with the foams recorded at different scan rates and (f) a performance comparison in nitrogen and oxygen saturated solution (25 mV s−1 scan rate). All the measurements were performed in 1 M KOH solution. |
Ni-foams exhibit high ECSAs, which is one of their advantages for the utilization in electrochemical devices. The ECSA of the foams were determined by double-layer capacitance measurements (Fig. 3b and S8d). The measurements show a 5-times higher ECSA for the modified foam, by assuming identical specific capacitance for both (Table S1†). This high electrochemically-active surface area (ECSA) likely results from surface restructuration processes and etching during the growth of Ni3N. Moreover, the integration of the Ni2+/3+ oxidation wave indicated that the amount of active Ni species for water oxidation was enhanced by a factor of 52. Therefore, the number of potential catalytically active sites for water splitting is dramatically increased upon modification. It should be noted that all further measurements in this paper were conducted on the modified foam after 500 oxidation cycles.
As mentioned, the intense oxidation peak strongly suggests formation of a nickel hydroxide layer on the Ni3N/Ni surface. XPS measurement of the foam after 500 oxidation cycles clearly confirms the formation of Ni(OH)2 on the surface19,26 (Fig. 2c). Moreover, the N 1s signal totally disappeared (Fig. S10a†) and a new hydroxide peak (530.5 eV) evolved in the O 1s spectrum,26 suggesting that the Ni(OH)2 replaces parts of the Ni3N layer. In addition, the removal of the carbon-nitride also becomes evident from the C 1s spectrum (Fig. S10b†), in which the peak for C–C bond dominates (284.7 eV) along with oxidized carbon (288.6 eV).
The partial removal of the side product, the amorphous carbon-nitride layer is also evidenced by the HRSEM images of the foam after the CV cycles (Fig. S11†). In addition, cross section images show that the thickness of the layer decreases to ∼100 nm, which is probably due to the removal of the carbon nitride layer (Fig. 1c and S7†), and EDX measurements of the modified foam (before and after 500 oxidation cycles) support the creation of a hydroxide layer by the detection of a clear oxygen signal (Fig. S12†). The surface of the activated foam is rougher and covered with small particles (<50 nm). In order to confirm that the nanoparticles visible in SEM (Fig. S11†) are Ni hydrate/hydroxide on the Ni/Ni3N surface we conducted EDX line scans. Fig. S11 and S13† show the SEM images and the corresponding line scans of the modified and the original Ni/Ni3N foams, respectively. This proves a rather homogeneous increase of surface oxide concentration along with the diameter (∼50 nm) of the patches. However, despite the conversion of the surface to Ni(OH)2, the XRD data after the electro-oxidation, approves the presence of the metal nitride in the bulk sample (Fig. S14b†).
Linear sweep voltammetry scans (LSV) of the activated Ni-foam towards positive potentials show higher current density in the formation of NiOOH, which leads to a higher catalytic current density for the OER than the pristine foam. In addition, the overpotential for the OER is ∼60 mV lower with the modified Ni-foam. However, the lower overpotential and the enhanced current density in the OER region can mainly be attributed to the extensive formation of Ni(OH)2 that further oxidize to active NiOOH.
As discussed before, the oxidation of Ni/Ni3N to Ni(OH)2 and subsequently to NiOOH on the electrode surface is thought to be the essential pre-step to oxygen evolution. According to the above data the presence of Ni3N seems to promote the formation of these hydroxy/oxohydroxy phases. The atomic displacement caused by incorporation of nitrogen to the nickel matrix goes in hand with structural convergence towards these hydroxide phases (fcc-Ni → β-Ni3N) and weakening of metal–metal bonds, hence likely supports subsequent oxidation during electrolysis from an energetic viewpoint (Fig. S3 and S15†). Furthermore, the increase of the electrochemical activity for the OER can be also attributed to the physical roughening of the Ni surface and to the formation of Ni3N nanoparticles (Fig. 1a).
Fig. 3d compares LSV curves for the HER of the modified and the pristine Ni-foam. It is important to note that without electrochemical activation that goes in hand with carbon nitride removal (Fig. S9†) the activity of the modified foam is lower (Fig. S16†) and does not change after 500 cycles toward negative potentials (from 1 to −0.5 V vs. RHE). For the pristine Ni-foam the current density also increases after 500 cycles (Fig. S17†), and the overpotential remains almost constant (∼150–180 mV). In contrast the modified foam exhibits a remarkably low overpotential of ∼50 mV and the current density is twice as high. To the best of the authors' knowledge this overpotential is amongst the lowest for the HER with non-noble metal catalysts27 and slightly higher than that of Pt (Fig. S18†).
To confirm that Ni3N is essential for the electrocatalytic activity in the OER and HER we modified a Ni foam with Ni(OH)2, as previously reported (Fig. S19†).19 In the OER, the Ni(OH)2 exhibited similar performance as the Ni3N/Ni electrodes (Fig. S19a†). We note the changes in the current densities can be attributed to the amount of Ni(OH)2 on the surface. However, the importance of the Ni3N layer for the HER activity is clearly shown in Fig. S19b.† Although the Ni/Ni(OH)2 foam showed slightly higher activity than the pristine Ni, the Ni3N/Ni foam demonstrated significantly smaller overpotential along with higher current densities in the HER. In addition, the Tafel slope for both, Ni3N/Ni and Ni/Ni(OH)2 exhibited classical Volmer–Heyrovsky behavior (∼120 mV dec−1), which is similar to most Ni-based materials in alkaline water electrolysis (Fig. S20†). The Tafel slope of the Ni/Ni3N for the OER side (Fig. S21†) showed a typical value of ∼60 mV dec−1.28
Furthermore, we characterized the other CMB modified foams, which were prepared at 475 and 550 °C that are not modified with Ni3N. In all cases both, the OER and HER performances (current densities and onset potentials) were lower than with the pristine Ni-foam and did not show any significant improvement with cycling (Fig. S22†). Hence, the superior performance of the modified foam in HER is likely due to the presence of the Ni3N/Ni(OH)2 junction and not because of the carbon-nitride layer.
For the alkaline HER, water dissociation and hydrogen formation rates are decisive for the productivity. M/M(OH)2 catalysts show good productivity because water dissociation is readily achieved with the metal hydroxide, while the protons diffuse to the nearby electron-rich metal nitride surface for the reduction.19 Recent activity evaluations of different M/M(OH)2 couples unveiled the Ni/Ni(OH)2 pair to be the most active for the alkaline HER. Here, we show that the activity of Ni3N/Ni(OH)2 is even higher than the one of the pure nickel electrode, owing to the advantageous electronic configuration of the metal nitride for the hydrogen adsorption and liberation.
It was recently shown by Boettcher et al. that Fe impurities in the KOH solution can play a critical role in the activity enhancement of Ni-based electrocatalysts for OER.29 To study the influence of the Fe impurities on our catalysts we measured the HER and OER in Fe-free KOH solution that was prepared as previously described29 (Fig. S23†). For the HER, the electrocatalyst did not demonstrate any noticeable change in activity in the Fe-free KOH solution while the Fe impurities seem to have a small influence on the OER (by slightly quenching the oxidation of Ni2+ to Ni3+). The influence of the Fe impurities on the electrocatalytic performance of the Ni-foam seems minor in our study, because of the foams high surface area and the low Fe concentration in the KOH solution, which results in relatively low Fe content in the foam compared to a smooth/thin Ni layer.
The advantage of the modified Ni-foam is also demonstrated in the oxygen reduction reaction (ORR). In general, ORR measurements are carried out using a rotating disk electrode (RDE) to increase the diffusion rate of oxygen to the electrode.30 However, due to the monolithic character of the foam it was not possible to use an RDE setup. Moreover, due to strong measurement noise caused by the movement of the electrode in the oxygen stream, it was not possible to bubble oxygen during the measurement, but only before. Consequently, the presentation of ORR results solely compare the activity of the modified foam to the pristine Ni-foam, without detailed quantitative analysis. Fig. 3e shows the LSV profiles of the foams, collected at different scan rates in oxygen- or nitrogen-saturated solution. While the unmodified Ni-foam has no electrocatalytic activity in the ORR, the LSVs of the modified foam clearly shows favorable electrocatalytic behavior. At this stage, it cannot be excluded that remaining carbon impurities (which is a well-known electrocatalyst for ORR) contribute to the electrocatalytic activity.30 The ORR activity is also demonstrated in Fig. 3f, in which both LSV curves (25 mV s−1) in either nitrogen or oxygen saturated solution are shown. The modified Ni/Ni3N foam exhibits low overpotential for the ORR together with high current densities. However, we note again that due to the monolithic nature of the foam it is not possible to further analyze the mechanism of this reaction in the current work.
In order to test the stability of the modified Ni-foam as both HER and OER catalyst, we recorded 500 cyclic voltammetry scans, plus for the HER the changes of the overpotential at 10 and 100 mA cm−2 for two hours. For the OER, only the most critical measurements at 100 mA cm−2 were performed. Fig. 4a shows HER chronoamperometry measurements at 10 and 100 mA cm−2 and compares the initial and final overpotentials of the two foams. At a current density of 10 mA cm−2 the overpotentials remain almost constant for both foams. The overpotential needed to achieve this current density is remarkably lower for the modified foam (∼100 mV) than for the pristine foam (∼200 mV). The superior performance of the modified foam becomes even more evident at higher current density (100 mA cm−2) where the overpotential is only 260 mV compared to 400 mV. Moreover, the overpotential remains almost constant, with a drop of only 13 mV after two hours while the overpotential for the original foam increases by ∼60 mV in 2 h. Given these values, the performance of the modified foam is at the forefront of non-noble metal based HER electrocatalysts. Additional information on the stability of the modified foam is given by the measurement of the current voltage properties for 500 cycles at a scan rate of 25 mV s−1. Fig. 4b shows the first and the five-hundredth scan of the modified foam. The electrode demonstrates high stability, with negligible change in the onset potential along with just ∼15% current loss. However, these current density losses can be mainly attributed to the formation of gas bubbles, which accumulated during the scanning.
Fig. 4c compares OER chronoamperometry measurements of the modified and the pristine foam at 100 mA cm−2 for 1 h. The advantage of the modified foam is demonstrated with the lower overpotential which is required to support this high current (∼0.48 V vs. 0.52 V) and even more by its high stability and the low overpotential changes during the measurement. While the overpotential of the original foam increases by ∼90 mV after one hour, the overpotential of the modified foam increases by only ∼15 mV with more than 90% Faradaic efficiency. The stability of the modified Ni-foam was also tested by measuring the current voltage properties for 500 cycles at a scan rate of 25 mV s−1 (Fig. 4d). The current density and the overpotential of the modified foam did not change after 500 CV scans, demonstrating the high stability of the modified foam. In addition, longer stability measurements for HER and OER (100 mA cm−2 at 25 °C for 12 h and 250 mA cm−2 at 61 °C for 2 h) which further confirmed the high stability of the modified foam are shown in Fig. S24 and S25.† Recent work by Chorkendorff et al. showed that the electrochemical stability is not enough to judge the total stability of the material due to partially dissolution of the electrocatalyst.31 In order to confirm the stability of the modified foam we tracked the Ni content with ICP-OES (inductively coupled plasma optical emission spectrometry) of the KOH solution before and after 500 oxidation scans. The Ni content was in both cases below 1 ppm and with no noteworthy differences of the signals, indicating insignificant material losses through dissolution processes.
The stability of the Ni/Ni3N foam was also confirmed by the XRD measurements and the SEM images of the modified foam after all electrochemical measurements (Fig. S13†). The XRD pattern of the nickel nitride remained unchanged, suggesting that the core material is still composed of nickel nitride on Ni.
The properties of the modified foam were also measured in extreme alkaline environment (6 M KOH). Fig. S26a† compares the LSV curves of the hydrogen evolution reaction of the modified foam and the original foam. The overpotential of the modified foam remains low (∼85 mV) and the current densities very high. In contrast, the original foam exhibits an overpotential of around 200 mV and 2 times lower current densities. Most importantly, the modified foam exhibits high stability in the chronoamperometry study also under these “accelerated ageing” conditions (Fig. S26b†). The overpotential of the modified foam remained almost constant for 1 h at 100 mA cm−2. In addition, the overpotential for the OER is reduced down to 280 mV, and high current densities can be obtained (Fig. S26c†). It is important to note that in this strong alkaline solution the original foam also exhibits low overpotential for the OER.
Ni(OH)2-modified electrodes were prepared by chemical deposition as described elsewhere.32 The pristine foam was immersed to an aqueous NiCl2 solution (0.1 M) for 12 h, and washed thoroughly with water before being introduced into the electrolyte. For the final oxide coverage the foam was cycled from 1 to −0.5 V vs. RHE until no change in the activity was observed. Purification of the KOH buffer was achieved according to the procedure that was described by Boettcher et al.29
ic = νCDL |
For the oxygen reduction reaction (ORR), the solution was purged with N2 for 20 min before each control measurements. Afterward, the solution was purged with O2 for 20 min before each measurement. However, due to strong noise, during the measurement the gas flow was turned off. All measurements were referred to the reversible hydrogen electrode (RHE) by using the relationship: VRHE = VSCE + 0.244 V + pH (0.059 V).33 Tafel slopes were obtained by recording the current in different voltages after 10 min chronoamperometry at each voltage. The O2 evolution (only in the solution) was monitored by an oxygen probe (PreSens Precision Sensing GmbH, Fibox 3 fiber optic oxygen transmitter) in a three neck flask (not sealed). The Faradaic efficiency was calculated by
Footnote |
† Electronic supplementary information (ESI) available: Further chemical (XRD, SEM, FIB and XPS) and electrochemical characterizations. See DOI: 10.1039/c5ta00078e |
This journal is © The Royal Society of Chemistry 2015 |