A. Nickel–manganese oxide (Ni _x Mn_1−x O _y )

The mixed metal oxide is in general represented by Ni _x Mn_1−x O _y where 0.0 ≤ x ≤ 1.0 and y are dependent on the type of phase present. Throughout the full range of composition, 0.0 ≤ x ≤ 1.0, the phase(s) present depends mainly on the x composition in the sample. Two clear phase regimes can be identified from XRD: (i) solid solution of Ni _k Mn_1−k O (0.9 < x ≤ 1) (ii) a solid solution of Ni _m Mn_3−m O₄ (0.05 < x ≤ 0.9), where the indices k and m defines the rock-salt and spinel solid solution series, respectively. In the entire range, there is no segregation of NiO and MnO _x phases as evidenced by the absence of clear α-Mn₂O₃ (221) peaks and the shrinking of NiO (222) peak from x = 1.0 until it disappears at x = 0.4.

B. Solid solution of Ni _k Mn_1−k O

In this compositional range 0.9 < x ≤ 1, a solid solution is formed whose composition can be tuned. The resulting solid solution has the rock-salt phase (space group: $Fm\bar 3m$ ). The XRD data in Fig. 1(a) show pure phases for NiO and α-Mn₂O₃ cubic phases and Fig. 1(b) shows the (200) peak of the rock-salt phase (at 43.3°) for Ni _k Mn_1−k O solid solutions. The 43.3° peak position shifts monotonically to lower angles with the increase of manganese concentration. Figure 1(c) shows the plot of the variation of lattice parameter with Mn mol% doping in Ni lattice. Clearly, the lattice parameter increases nonlinearly until it reaches the solubility limit which is about 10 mol% Mn composition or x = 0.90. It is noteworthy that at compositions above the solubility limit of Mn in Ni lattice, the speciation of the spinel phase is more evident. The crystal structure of sample with x = 0.90 was examined with HRTEM and the SAED, and the result is in agreement with the XRD results (Fig. 2).

FIG. 1.

(a) XRD of Ni _x Mn_1−x O _y (0.0 ≤ x ≤ 1.0) showing the phases present as x is varied, 2 theta range of 42–44°marked by (200) peak is expanded in (b) for clarity. (b) XRD showing the (200) plane of NiO lattice shifting to a lower angle with Mn doping. (c) Variation of the NiO lattice parameter with Mn doping concentration.

FIG. 2.

(a) HR-TEM of sample x = 0.9, showing grains with single lattice fringes having d-spacing indexed to (101) of the tetragonal spinel lattice, (b) a close-up view of the lattice fringes showing d ₁₀₁ to be 4.80 Å, (c) SAED of sample x = 0.9 (inset) showing the diffraction rings. The intensity line profiles measured from these patterns are as shown, assigned to the rocksalt phase with some spinel peaks, which is in agreement with XRD data, (d) HR-TEM of sample x = 0.9, showing grains with single lattice fringes having d-spacing indexed to (200) of the rocksalt lattice, (e) a close-up view of the lattice fringes showing d ₂₀₀ to be 2.08 Å, and (f) bright field TEM of composition x = 0.9.

C. Solid solution of Ni _m Mn_3−m O₄ spinel

In this compositional range 0.05 < x ≤ 0.9, the XRD data for samples were indexed to spinel solid solutions (Ni _m Mn_3−m O₄). The sample with composition x = 0.30 showed well-resolved diffraction lines corresponding to a tetragonal spinel phase (ICDD PDF-01-070-9039) with space group number I4₁/amd (141). XRD data for samples with the following x composition and the corresponding m values in bracket: 0.50 (1.50), 0.40 (1.20), 0.30 (0.90), 0.25 (0.75), 0.20 (0.60), show that phase pure Ni _m Mn_3−m O₄ bimetal spinel solid solutions exist in the composition range 0.20 < x ≤ 0.50 (or 0.60 < m ≤ 1.50). The series of oxides with composition 0.05 < x ≤ 0.20 shows peak deviation from nickel manganite peaks, and peaks at (112) and (103) were highlighted in Fig. S2 (Supplementary information) for emphasis. There is a widening of the gap between the centerlines of these two peaks coupled with peak broadening until manganese oxide (Mn₃O₄ and α-Mn₂O₃) peaks are evolved at x = 0.05. Peak broadening can be ascribed to the non-uniform strain on the spinel lattice. It is interesting to note that at nickel composition x ≤ 0.05, there is not enough Ni(II) to diffuse into the solid lattice of Mn₃O₄ to occupy the manganese vacancies or partially replace manganese ions in the tetrahedral or octahedral holes. ^{Reference Burns28} In the compositional range 0.5 < x ≤ 0.9, there is a co-existence of spinel solid solutions and rock-salt (Ni_0.9Mn_0.1O), however; at x > 0.5, only phase pure Ni _m Mn_3−m O₄ can be seen. Similar bimetallic oxide systems were reproduced for the nickel ferrite, available in the Supplementary information Fig. S3.

The spinel manganites are known to exhibit large tetragonal distortions. ^{Reference Ko, Li, Podraza, Dickey and Trolier-McKinstry29,Reference Noh, Yeo, Kang, Zhang, Cheong, Oh and Johnson30} The tetragonal distortion is a consequence of Jahn–Teller effect of the cubic spinel associated with a high density of Mn³⁺ cations. ^{Reference Miyahara31} Large amounts of Mn³⁺ are found in Mn-rich nickel–manganese spinels due to the reduced amount of Ni²⁺ on the octahedral sites of the spinel lattice. ^{Reference de Györgyfalva and Reaney32} The Mn valence states are strongly dependent on Ni occupancy in the octahedral sublattice, which is a measure of inversion parameter of the spinel lattice. Ni cations moving to the octahedral sublattice are balanced by the Mn going in the opposite way (tetrahedral sites). The tetrahedral Mn has a valence state of +2 since Mn³⁺ is a less suitable state in a 4-fold coordination. To attain charge neutrality, an equal amount of Mn⁴⁺ is formed in the octahedral sites to compensate for the Mn²⁺ on the tetrahedral sites, resulting into internal disproportionation. ^{Reference Díez, Schmidt, Sagua, Frechero, Matesanz, Leon and Morán33}

The XRD data are further corroborated with diffraction and EDX analysis using HRTEM. A representative sample x = 0.2 is used as an example to confirm the morphology and crystal structure. Figure 3(a) shows the bright field TEM image for the solid solution with composition x = 0.2 where a rapidly changing strong diffraction contrast can be noticed, an indication of randomly oriented grains and polycrystallinity of sample. The SAED patterns obtained for sample x = 0.2 are shown in the inset of Fig. 3(b). The intensity line profiles measured from these patterns are shown in Fig. 3(b). About ten diffraction rings are visible which correspond to the family of planes of the tetragonal spinel nickel manganite; the other rings were not clearly detectable due to their low intensity. Formation of tetragonal spinel phase is further confirmed from the sample with composition x = 0.2 by the HRTEM analysis. We identified two of lattice fringes, one with the d-spacing of 4.85 Å and the other one with the d-spacing of 3.00 Å, which are consistently observed in HRTEM images of the nanoparticles [Fig. 3(c)]. Knowing that the typical accuracy of HRTEM analysis is usually not better than a few percent, the observed lattice fringes can be assigned to the (101) and (200) planes of tetragonal spinel Ni_0.6Mn_2.4O₄. As expected, the HRTEM data did not show any presence of α-Mn₂O₃ and NiO phases in this sample composition. SEM image [Fig. 3(d)] of the solid solution with sample composition x = 0.2 has a morphology resembling cubes interconnected by a channel. Figure 4 shows TEM-EDX compositions of four randomly selected grains of samples with precursor compositions x = 0.1 to x = 0.9, showing strong agreement with the precursor composition.

FIG. 3.

(a) Bright field TEM of composition x = 0.2 with a crystallite size of 50–80 nm, (b) SAED of sample x = 0.2 (inset) showing ten diffraction rings. The intensity line profiles measured from these patterns are as shown, assigned to the tetragonal spinel, which is in agreement with XRD data, (c) HR-TEM showing grains with two lattice fringes having d-spacing indexed to (101) and (200) planes of the tetragonal spinel crystals, and (d) SEM micrograph of samples x = 0.2 showing its morphology.

FIG. 4.

TEM-EDX composition of the four selected nanoparticle grains with Ni at.% on the vertical axis and Mn at.% on the horizontal axis.

D. Comparison between plasma and thermal oxidation

To understand whether one can obtain similar results with thermal oxidation, a sample with composition x = 0.2 is investigated using thermal oxidation route. The phases obtained were the ilmenite (NiMnO₃), bixbyite (α-Mn₂O₃), and the bunsenite (NiO) phases, which are completely different from the result obtained with plasma oxidation (refer to Fig. 5). The difference in the phases can be ascribed to the dissimilar mechanistic steps obtainable in the two oxidation processes, given that the reaction temperature and pressure are same. As can be seen from the results, plasma oxidation better reflects the reaction off-stoichiometry, producing solid solution of nickel manganite spinel. However, the thermally oxidized sample are segregated into thermodynamically stable monometal oxide (bunsenite and bixbyite) and mixed metal oxide (ilmenite) phases in agreement with Xiao-Xia and co-workers. ^{Reference Xiao-Xia, Manthiram and Goodenough34}

FIG. 5.

X-ray diffraction of the sample with nominal composition of x=0.2 prepared via plasma and thermal oxidation.

To explain the differences between thermal and plasma oxidation, a proposed mechanism as elucidated by the schematic presented in Fig. 6 is used to illustrate various stages of oxidation process. The following are the stages of oxidation: first of all, upon either plasma exposure or thermal heating, the liquid droplet will lose solvent by evaporation until a densified amorphous phase is formed. In the case of thermal oxidation, heating and cooling rates are slow and so atoms have enough time to diffuse, inducing a phase transformation toward thermodynamically stable phases. So, irrespective of precursor composition inside liquid droplets, the resulting materials would consist of thermodynamically stable phases. The oxidation proceeds with diffusion of molecular oxygen into the amorphous mass containing nitrates and the thermal decomposition of the amorphous mass to release NO _x . After an induction period, nuclei are formed. In the case of plasma oxidation, several processes take place in a very short period of time of a few seconds and they include the following: oxygen radicals can diffuse into the core of the densified amorphous mass of precursor; faster reaction kinetics with oxidation of nitrates with oxygen radicals; and fast heating and cooling rates of the core for nucleation and transport limited ordering of atoms make for the speciation of metastable phase nuclei. As the growth proceeds with transport limited region, the composition of resulting solid solutions depends directly on the composition of elements in the surrounding environment. The composition of the surrounding nutrients is as homogenous and similar to the supersaturation; thus, the growing crystals reflect this compositional homogeneity and on rapid cooling, the diffusion of atoms is suppressed, which leads to a non-equilibrium solid solution phase growth. Average size of crystallites obtained in plasma oxidation is about twice that of thermal oxidation indicating fast growth kinetics compared to thermal oxidation.

FIG. 6.

Proposed mechanism for the formation of nanoparticles of thermally oxidized and plasma oxidized precursor solutions.

E. Ternary solid solutions of NiMn _z Fe_2−z O₄ (0 ≤ z ≤ 1)

Ternary spinel oxides comprising manganese substituted nickel ferrite, with nominal series formula NiMn _z Fe_2−z O₄ (where z = 0.0, 0.20, 0.40, 0.60, 0.80, and 1.0) have been successfully synthesized by the plasma spray method. The XRD reflections [Fig. 7(a)] show highly crystalline nanoparticles with exceptional phase purity. Figure 7(b) shows an expanded view of (311) peak position. As can be seen, there is a monotonic shift of the 311 peak position to the left with increasing manganese doping, an indication of the formation of a spinel solid solution. The same family of peaks indexed to cubic spinel phase (space group: $Fm\bar 3m$ ) appears in all six samples and no foreign peak was observed, an indication that no other phase co-existed with sample phase. The domain size of microstrain was estimated from Williamson–Hill (W–H) plots (Supplementary information S4) and tabulated as shown in Table I. After a linear fit to the W–H plot, the microstrain and domain size were extracted from the slope and vertical axis intercept, respectively. Results reveal that the strain portion of the W–H plot has no contribution to the line broadening (Supplementary Fig. S4) except for the sample z = 1.0 with negligible microstrain (0.01%) as shown by Table I. The lattice parameter and unit cell volume expand isotropically as the density of the manganese dopant increases, and the crystal size is in the range 500–590 Å. For uniform strain, the unit cell dimensions change in an isotropic manner which leads to changes of the lattice parameters and a shift of the peaks with no accompanying broadening of the XRD reflections. The shape of peaks are unchanged from z = 0.0 to z = 1.0 which confirms the absence of anisotropic microstrain.

FIG. 7.

(a) XRD reflections of ternary spinel oxides comprising manganese substituted nickel ferrite NiMn _z Fe_2−z O₄ (0 ≤ z ≤ 1) showing a single cubic spinel phase. (b) Zoom-in of (311) peak position showing a monotonic shift to a lower angle with increasing Mn doping concentration. (c) TEM-EDX spectra obtained at 0–20 kV, along with peak assignments for Mn, Ni, and Fe elements for various compositions (0 ≤ z ≤ 1). (d) Ternary plot of NiMn _z Fe_2−z O₄ (0 ≤ z ≤ 1) showing an excellent match between the nominal precursor composition (∆) and the TEM-EDX measured alloy composition (○), in normalized atomic percent.

TABLE I.

Structural parameters of NiMn _z Fe_2−z O₄ (0.0 ≤ z ≤ 1.0) showing how lattice parameter, unit cell volume and anisotropic strain vary υ composition.

Figure 7(c) shows the typical EDX spectra obtained at 0–20 kV, along with peak assignments for Mn, Ni, and Fe elements for various compositions (0 ≤ z ≤ 1). The peak areas fairly match the precursor compositions, confirming the formation of ternary solid solutions [Fig. 7(d)]. After integrating the peak areas of Mn–K_α, Ni–K_α, Fe–K_α, and Fe–K_β, we converted the x-ray counts into the elemental weight percentages using quantification software. The elemental analysis of the nanoparticles confirms that the proportion of the elements is consistent with the precursor composition (Table II). Figure 8 shows the XRD patterns and corresponding Rietveld analysis of as-prepared sample NiMn_0.2Fe_1.8O₄, confirming the formation of spinel phases. Rietveld refinement and Bertaut method ^{Reference Weil, Bertaut and Bochirol35} are used to determine the cation distribution (Supplementary information, Table S1). Combining the refined results and TEM elemental analysis, the structural formula of NiMn_0.2Fe_1.8O₄ was determined to be (Ni_0.1Mn_0.15Fe_0.75)_8a [Ni_0.9Mn_0.05Fe_1.05]_16d O₄, revealing a high degree of inversion in agreement with other results for ferrite spinel, ^{Reference Birajdar, Shirsath, Kadam, Patange, Mane and Shitre36} where the parenthesis refers to the tetragonal sites while octahedral sites are represented by square brackets. For an AB₂O₄ type spinel, the extent to which sublattices are disordered is determined by the inversion parameter i which is the fraction of B³⁺ occupying the tetragonal sites, as such i can vary from 0, which corresponds to normal spinel and 1, which refers to a structure where all the tetragonal sites are occupied by B³⁺.

TABLE II.

TEM-EDX analyzed composition of NiMn _z Fe_2−z O₄ (0.0 ≤ z ≤ 1.0).

FIG. 8.

Structural analysis of the synthesized nanocrystalline spinels NiMn_0.2Fe_1.8O₄ showing Rietveld refined XRD patterns with experimental data (black line), calculated profiles (red dots), allowed Bragg diffraction positions (vertical bars) and difference curve (blue line).

F. OER kinetics

A series of NiMn _z Fe_2−z O₄ (z = 0, 0.20, 0.40, 0.60, 0.80 and 1.0) mixed metal oxide films were tested for OER catalysis by studying the cyclic voltammetry at 20 mV/s scan rate. Two parameters were selected to understand the catalyst performance: (i) overpotential (ƞ) required to reach a current density of 10 mA/cm²; (ii) steady state Tafel slope (mV/decade).

The data in Fig. 9(a) show a decrease in ƞ for OER catalysts with increase in Mn content in the binary metal oxide nickel ferrite. Pure nickel ferrite shows higher overpotential of 0.46 V compared to the ternary metal oxides. Among all the six compositions, z = 0.20 demonstrates the lowest ƞ at 0.39 V. There still remains the challenge of impartially comparing the efficiencies of electrocatalysts, largely due to complications arising from both inadequate standardizations and deviations in average particle size and/or surface areas. It is also notable that the manganese doped ternary metal oxides are more active catalysts than stand-alone NiO _x , FeO _x , or MnO _x catalysts. The NiO _x -type catalyst has been previously reported to give 10 mA/cm² at an overpotential value of ∼0.42 V, ^{Reference Chekin, Tahermansouri and Besharat37,Reference Yeager, Su, Marinković and Teng38} and the reported overpotential value at a current density of 1.0 mA/cm² for FeO _x and MnO _x based catalyst are 0.41 and 0.51 V, respectively, ^{Reference Trotochaud, Ranney, Williams and Boettcher39} which are expected to be at a higher overpotential value when adjusted to the benchmarking current density of 10 mA/cm². Substitution of cations into complex metal oxide crystal lattice has been known to induce oxygen vacancies, and according to Bao et al., ^{Reference Bao, Zhang, Fan, Zhang, Zhou, Yang, Hu, Wang, Pan and Xie40} there is a positive correlation between oxygen vacancy and electrocatalytic activity in OER of nickel-cobalt oxide based catalysts, of which they adduced low H₂O adsorption energy at the defect site to this outcome.

FIG. 9.

(a) Cyclic voltammetry of the conditioned NiMn_zFe_2-zO₄ (0≤z≤1) films collected in 1 M NaOH solution at a scan rate of 20 mV s^-1; (b) Steady state Tafel plot for the linear region from 0.35 – 0.50 V.

The linear part of the Tafel plots is shown in Fig. 9(b) obtained for all the six compositions in the overpotential window of ƞ = 0.35–0.50 V. The electrodes with lower z value shows smaller Tafel slopes less than 45 mV/decade compared to the electrodes with higher content of Mn doping indicating an improvement in the electrocatalytic activity of the catalyst. The data presented in Table III show the variation of the overpotential and linear portion of the Tafel slope with composition. The values of the Tafel slope found on the manganese substituted nickel ferrite solid solutions are comparable to those found elsewhere: MnFe_2−x Cr _x O₄ (∼40 mV/decade) ^{Reference Singh, Singh, Nguyen Cong and Chartier41} and NiFe_2−x Cr _x O₄ (∼40 mV/decade). ^{Reference Singh, Singh, Lal, Thomas and Bera42} Strasser and co-workers ^{Reference Strasser, Koh, Anniyev, Greeley, More, Yu, Liu, Kaya, Nordlund, Ogasawara, Toney and Nilsson43} found through experimental and first principle studies that compressive strain in Pt–Cu solid solution have a strong correlation with its electroactivity in oxygen reduction reaction. Also, the improved activity of the alloy may have been due to the favorable adjustment of the electronic properties of the alloy as a direct effect of the combination of the completely filled d orbitals (Ni) with semifilled orbitals (Mn and Fe) according to the Brewer–Engel theory. ^{Reference Jakšić44}

TABLE III.

Electrode kinetic parameters for the OER of NiMn _z Fe_2−z O₄ (0.0 ≤ z ≤ 1.0) in 1 M NaOH at 298 K.

Besides high catalytic activity, an optimal catalyst material should be stable for their implementation in any operating system. The stability of these catalysts under catalytic conditions was determined using controlled-current electrolysis. The catalyst material was held at a constant current density of 10 mA/cm² per geometric area for 2 h, while the operating potential was measured as a function of time. Figure S5 from the Supplementary information shows the results of the controlled current electrolysis measurements of these catalysts. Clearly, ƞ (t = 0) and ƞ (t = 2 h) are similar, an indication that the catalyst is stable under the operating conditions for at least a 2 h period.

The concept of plasma oxidation of precursor solution mentioned here offers excellent metastable phase formation possibilities due to its fast reaction kinetics and cooling rates. In general, short reaction times allow for the formation of either transient phases in the case of monometal oxides or solid solutions in the case of mixed metal oxides. ^{Reference Brinley, Babu and Seal45} The process is highly adaptable to any precursor chemical (inorganic, organic, etc.) and any solvent (aqueous, polar, nonpolar, organic, etc.) which can be easily incorporated into the spray feedstock. Precursor solution mixtures can be used to produce multicomponent systems such as bimetallic and ternary oxides and their solid solutions. Since each droplet contains precursor chemicals in the same stoichiometric composition as desired in the produced particle, the synthesized particles will have exceptional compositional uniformity.