Composite Cathodes with Oxide and Nitride Phases for High-Temperature Electrocatalytic Ammonia Production from Nitrogen and Water

Lanthanum ferrite-type perovskites which have their A-sites doped with Sr and B-sites doped with Ni and Co, were synthesized using an EDTA-citric acid complexation route. ²⁸ Stoichiometric amounts of La(NO₃)₃.6H₂O, Sr(NO₃)₂, Ni(NO₃)₂.6H₂O, Co(NO₃)₂.6H₂O and Fe(NO₃)₃.9H₂O were first dissolved in deionized water. EDTA in a stoichiometric ratio of 1.5:1 with respect to the metal ions was then added to this solution under uniform stirring and heating. Citric acid was added with a molar ratio of 1:1 with the total metal ions once all the EDTA particles get dissolved. pH was adjusted to 6 by the dropwise addition of NH₄OH. Ethylene glycol was also added at this point to the solution. The complex was then kept at 90 °C under constant stirring for gelation. Once the gel was formed, it was dried at 150 °C for 12 h to obtain an amorphous material. The powder was ground and calcined at 1000 °C for 5 h to obtain perovskite oxide samples. The different compositions of samples synthesized for the present study are: La_0.9Sr_0.2FeO₃ (A-site excess LSF92), La_0.8Sr_0.2FeO₃ (A-site stoichiometric LSF82), La_0.7Sr_0.2FeO₃ (A-site deficient LSF72), La_0.7Sr_0.2Ni_0.2Fe_0.8O₃ (Ni-doped LSNF7228) and La_0.7Sr_0.2Co_0.2Fe_0.8O₃ (LSCF7228).

In a typical synthesis, 0.1 g of iron (III) oxide (γ-Fe₂O₃) nanopowder (<50 nm particle size) was put in a ceramic boat placed inside an alumina tube in a tubular furnace. The alumina tube was flushed with N₂ at room temperature for 10 min to get rid of all the oxygen molecules from the gas environment and then the sample was heated to a nitridation temperature at a heating rate of 10 °C min⁻¹. Three different nitridation temperatures were investigated: 400 °C, 450 °C and 500 °C. Pure ammonia stream was introduced into the furnace when the nitridation temperature was reached. The samples were kept at the nitridation temperature under ammonia flow for different time intervals (45 min, 2.5 h and 4 h). After nitridation was done, samples were cooled down to room temperature under N₂ flow. Since iron is pyrophoric in nature, the nitride samples were passivated at 30 °C by injecting small amounts of air to the nitrogen stream.

Ex-situ XRD studies were performed on both perovskite and oxynitride phases in order to confirm the presence of the desired crystal structures. A Bruker D8 Advance Diffractometer with a Cu Kα radiation source was used for XRD experiments.

Temperature-programmed desorption (TPD) technique was used to characterize the perovskite phase of the composite cathode. The materials were loaded into a quartz tube placed inside a tubular furnace and treated in situ with air at 1000 °C for 1 h to remove the surface impurities and to fill all the vacancies with oxygen. After the pre-treatment, the catalyst bed was cooled to room temperature under air and then the gas environment was switched from air to He. The catalyst bed temperature was increased from room temperature to 1000 °C with 10 °C min⁻¹ under He. During the controlled heating, the gas outlet was fed to an MKS Cirrus benchtop residual gas analyzer to monitor the m/z = 32 (oxygen) for the comparison of oxygen desorption related to vacancy formation/ion conductivity.

TPD technique was also used to compare the amount of lattice nitrogen in the iron oxynitride structure. The same procedure was followed except the pre-treatment part where m/z = 28 (nitrogen) signal was monitored to compare the nitrogen desorption amounts from different iron oxynitride samples.

Temperature-programmed reaction (TPRxn) technique was used to investigate the ability of the lattice nitrogen in the oxynitride structure to interact with the reactant H₂O to form NH₃. In this experiment, 3% H₂O/He was flowed over the catalyst during a linear temperature ramp and the effluent from the reactor was continuously analyzed with the on-line benchtop mass spectrometer.

The surface structure and composition of the iron oxynitride samples were studied using X-ray photoelectron spectroscopy (XPS) technique. A Kratos Axis Ultra XPS instrument equipped with a monochromated Mg Kα source (1254 eV, 12 kV, 10 mA) and a charge neutralizer at 2.1 A, 1.3 V bias, and a charge of 2.6 V was used to perform surface analysis of iron nitride samples. In order to analyze the composition of the inner surface layers, Ar etching of the outer surface layers was performed.

The iron oxynitride samples were studied via ⁵⁷Fe Mössbauer spectroscopy. Spectra were recorded at the CNRS IRCE-Lyon in France using a 2GBq ⁵⁷Co/Rh γ-ray source and a conventional constant acceleration spectrometer operated in triangular mode. The relative amounts of the iron species in the catalysts have quantitatively been evaluated by comparing the relative areas of the corresponding spectral components and assuming equal recoil-free fractions for the corresponding species.

The RAMAN spectroscopy is sensitive to the vibrations of oxygen and nitrogen ions which gives information about the changes in the cation-anion bonds and the anion sublattice. Iron nitride/oxynitride phase on the spent composite cathode (0–3 mA for 72 h) was characterized with this technique using a Horiba LabRAM HR-800 Raman Spectrometer equipped with an asymmetric Czerny Turner spectrometer. A 100X was used to focus an argon ion laser beam (514.5 nm wavelength, Coherent) onto the sample and collect the backscattered photons.

The electrical conductivity of the perovskite oxide materials was measured using the four probe DC van der Pauw method. The sample pellets were prepared by first uniaxially pressing sample powder in a hydraulic press and then sintering at 1300 °C for 5 h. Four silver wires (0.1 mm dia, 99.997% metals basis, Alfa Aesar) were connected to four points on the pellet using a conductive silver paste (Pelco™). Two of these leads were used to apply 10 mA of current by a Keithley 6220 current source, and the other two leads were used to measure the voltage drop using a Keithley 6182 sensitive nanovoltmeter. The electrical conductivity was measured by the formula [σ = (I/V)(l/A )] where I = current, V = voltage, l = distance between voltage measurement points and A = cross sectional area available for current flow. Conductivity experiments on sample pellets were performed under a flow of 100 sccm (standard cubic centimeter per min) of air at a temperature range of 25 °C–800 °C with a 30-min stabilizing time at each temperature before performing measurements.

Structural Verification of Iron Nitride/Oxynitride Samples.—The phase identification of the iron nitride samples synthesized at different temperatures was performed using XRD technique. As shown in Fig. 4, iron nitride formation starts at 450 °C. In the iron nitride sample synthesized at 450 °C for 2.5 h, ε-Fe₃N phase dominates the structure. Further increase in nitridation temperature to 500 °C causes the formation of γ'-Fe₄N which is in the nitrogen-deficient range of the Fe-N phase diagram. ²⁹ This phenomenon is expected because at high ammonolysis temperatures, ammonia starts decomposing to a mixture of H₂ and N₂ which has a lower nitridation potential compared to NH₃.

Zoom In Zoom Out Reset image size

Mössbauer spectroscopy was also used to characterize the bulk properties of the nitrided iron oxide. Figure 5 shows the Mössbauer spectra recorded at room temperature for Fe-N (400 °C–4 h), Fe-N (450 °C–2.5 h) and Fe-N (500 °C-4 h). In good agreement with the XRD pattern of Fe-N (400 °C-4 h), Mössbauer spectrum of this sample shows mainly a doublet with a small quadrupole splitting which is attributed to γ-Fe₂O₃ with foreign ions replacing the octahedral vacant sites in the structure and probably N³⁻ replacing O²⁻ during nitridation. ³⁰ The Mössbauer spectrum of Fe-N (450 °C-2.5 h) shows two new sextets with 21 and 30 T hyperfine fields (relative intensity 65%) confirming the presence of ε-Fe₃N. ³¹ The remaining doublet is again attributed to substituted γ-Fe₂O₃, although the XRD pattern of the sample does not show any corresponding diffraction peaks. In the spectrum of Fe-N (500 °C-4 h), two magnetic sextets are identified with low and high hyperfine fields and quadrupole splitting of 0.00–0.02 mm s⁻¹ for the low hyperfine field component that can be attributed to ε-Fe₃N and γ-Fe₄N. ³² The doublets identified in this last spectrum are attributed to paramagnetic oxynitride species with vacancies in the first coordination sphere of the iron sites. ³³

Zoom In Zoom Out Reset image size

Surface Characterization of Iron Oxynitride Samples —In addition to the analysis of the bulk structure, the surface structure of the oxynitride samples was studied using X-ray photoelectron spectroscopy (XPS) technique. In order to analyze the composition of the inner surface layers, Ar etching of the outer surface layers was performed. Figure 6 shows N 1 s region of the photoelectron spectra of three different iron nitride catalysts along with the deconvoluted components. In all the samples, there are three features in N 1 s region at 396 eV, 397.5 eV and 399.2 eV which belong to nitrogen in oxynitride phase (Fe_xN_yO_z), nitride phase (Fe_xN) and surface oxidized nitrogen (N-O), respectively. ³⁴ As shown, iron oxynitride phase dominates the outer surface and with etching of the outer surfaces, nitride phase becomes the dominant phase in the inner layers. The nitride-to-oxynitride ratio (N/oxyN) increases as we go into sublayers and after 150 s of etching, iron nitride synthesized at 450 °C for 2.5 h shows the highest nitride to oxynitride ratio.

Zoom In Zoom Out Reset image size

Amount/Mobility/Activity of Lattice Nitrogen in Oxynitride Phases —TPD experiments were performed to investigate the mobility of nitrogen ions in the iron oxynitride materials. Figure 7a shows the m/z = 28 signal collected during the TPD of the iron oxynitride samples under helium flow. Due to the higher density of active nitrogen species, the sample prepared at 450 °C was considered for the activity experiments discussed later.

Zoom In Zoom Out Reset image size

TPRxn experiments were also performed to examine the ability of the lattice nitrogen to interact with H₂O to produce NH₃. In this experiment, 3% H₂O/He was flowed over the Fe-N (450 °C–2.5 h) catalyst during a linear temperature ramp. Figure 7b shows the m/z = 18 (H₂O) and m/z = 16 (NH₃) signals collected during temperature-programmed reaction of H₂O over the Fe-N (450 °C–2.5 h) sample. Two peaks of NH₃ were observed at 467 °C and 545 °C and the corresponding H₂O signal showed consumption at the same temperatures, suggesting a wide temperature range over which the lattice N could be accessed by H₂O molecules for NH₃ formation.

An oxygen ion conducting solid oxide electrolysis cell (described in the "Button Cell Fabrication and Electrocatalytic Activity Measurements" section) was used for the electrochemical production of ammonia from N₂ and H₂O at 600 °C. Figure 8 shows the comparison of NH₃ production rates obtained with LSNF7228 and Fe-N promoted LSNF7228 composite cathode catalysts. Although the production rates of H₂ over both cathodes were very similar (at 1 mA, H₂ production rate over LSNF7228 and Fe-N promoted LSNF7228 cathodes were 2.2 × 10⁻⁴ and 2.8 × 10⁻⁴ mol s.m⁻², respectively), the NH₃ formation rates were different by orders of magnitude. While LSNF7228 cathode produced negligible amounts of NH₃, the Fe-N promoted cathode (composite cathode) exhibited substantial activity. This demonstrates that the concept of a composite cathode can be a promising catalyst design strategy for the purpose of electrochemical ammonia production at atmospheric pressure. It is observed from Fig. 8 that, as the applied current increases, the ammonia production rate also increases up to 0.5 mA and then starts decreasing with higher currents. This change in the trend of ammonia production rate can be explained by the competitive adsorption between N₂ and H₂ on the active sites. In order to investigate the source of nitrogen in the ammonia produced, a blank experiment with 3% H₂O/Ar was conducted under the same reaction conditions at OCV and −1mA. The reaction was performed for 8 h at OCV and −1 mA to utilize most of the lattice nitrogen in the iron oxynitride structure. As seen in Fig. 8, the ammonia production rate in the blank experiment is negligibly low compared to the one conducted in the presence of nitrogen which proves that during the electrochemical reaction, the nitrogen vacancies are created as the lattice nitrogen reacts with H₂O to produce NH₃ and the vacant sites are replenished by the nitrogen in the feed gas.

Zoom In Zoom Out Reset image size

In order to study the stability of the composite cathode, the post-reaction (0–3 mA, 72 h) button cell's working electrode (cathode) was characterized using RAMAN spectroscopy technique (Fig. 9). Under RAMAN instrument microscope, the post-reaction composite cathode exhibited three distinct features which are pink-colored, green-colored, and gold-colored areas that represent iron nitride/oxide, LSNF7228-type perovskite oxide and gold paste, respectively (Fig. 9c). The iron nitride/oxide (pink-colored areas) phases were divided into two different areas which are; F1: iron nitride/oxide phase surrounded by LSNF7228 perovskite phase (P), and F2: iron nitride/oxide phase near gold paste (electrical lead). For comparison, the RAMAN spectra of commercial Fe₃N and Fe₂O₃ were collected (Fig. 9b). As shown in Fig. 9a, while iron nitride (Fe₃N) phase dominates the area F2, area F1 is mostly composed of iron oxide (Fe₂O₃). This result suggests that the environment of Iron nitride on the composite cathode determines its final state after the reaction. While the iron nitride surrounded by LSNF7228 turns to iron oxide due to the replacement of lattice nitrogen with lattice oxygen from LSNF7228 phase, iron nitride near the gold paste preserves its nitride structure due to the replenishing of nitrogen vacant sites by gas phase nitrogen via electrochemical route. Although additional operando characterization experiments are needed to prove this theory, one thing is clear that the composite cathode partially preserves its nitride phase after 72 h of electrocatalytic operation.

Zoom In Zoom Out Reset image size