Role of precursors mixing sequence on the properties of CoMn2O4 cathode materials and their application in pseudocapacitor

In this study, the effect of oxygen vacancy in the CoMn2O4 on pseudocapacitive characteristics was examined, and two tetragonal CoMn2O4 spinel compounds with different oxygen vacancy concentrations and morphologies were synthesized by controlling the mixing sequence of the Co and Mn precursors. The mixing sequence was changed; thus, morphologies were changed from spherical nanoparticles to nanoflakes and oxygen vacancies were increased. Electrochemical studies have revealed that tetragonal CoMn2O4 spinels with a higher number of oxygen vacancies exhibit a higher specific capacitance of 1709 F g−1 than those with a lower number of oxygen vacancies, which have a higher specific capacitance of 990 F g−1. Oxygen vacancies create an active site for oxygen ion intercalation. Therefore, oxidation–reduction reactions occur because of the diffusion of oxygen ions at octahedral/tetrahedral crystal edges. The solid-state asymmetric pseudocapacitor exhibits a maximum energy density of 32 Wh-kg−1 and an excellent cyclic stability of nearly 100%.


Results and Discussion
shows the XRD spectra of T 1 and T 2 samples. The obtained Bragg's diffraction patterns were matched with the standard JCPDS data file (no #18-0408), which validated the formation of the tetragonal structure of CoMn 2 O 4 with a mixed spinel (Co, Mn) (Co, Mn) 2 O 4 11 . The strong peaks in the spectra indicated that the as-synthesised T 1 and T 2 had satisfactory crystallinity. Figure 2 shows the SEM and TEM micrographs of T 1 and T 2 . Figure 2(a,b) show SEM images of T 1 and T 2 , respectively, which indicate that T 1 comprises uniformly distributed spherical nanoparticles, whereas T 2 primarily comprises nanoflakes and a few nanoparticles with a square structure. Figure 2(c,d) illustrate TEM images of T 1 and T 2 , respectively, and T 1 and T 2 nanoparticles are aggregated with an average particle size of approximately 23 and 27 nm, respectively [insets of Fig. 2(c,d)]. Moreover, similar nanoparticle sizes were reported in spinel CoMn 2 O 4 nanoparticles supported on nitrogen phosphorus-doped graphene electrode materials by He's group, in which the range of particle sizes was 5-25 nm 13 . Therefore, T 2 has higher conductivity than T 1 , which is validated by electrochemical impedance spectroscopy (EIS). To identify the phase of the samples, high-resolution transmission electron microscopy was performed on T 1  This result was consistent with XRD results (Fig. 1).
The surface area and porosity of the electrodes are crucial in the performance of pseudocapacitor devices 14,15 ; therefore, a Brunauer-Emmett-Teller (BET) surface area measurement was performed. Figure 3 shows nitrogen adsorption-desorption isotherms at 77 K of T 1 and T 2 . According to the definite hysteresis loop of IUPAC classification, the isotherms of both spinels are classified as type IV, which corresponds to the mesoporous structure 13,16,17 . The availability of hysteresis loop intermediate P/P 0 = 0.5 to 1 confirms the mesoporous nature of both compounds 15 . A high adsorption P/P 0 at 0.9 to 1 indicates the performance of the macroporous part to the overall surface area of T 1 and T 2 14 . The BET surface areas and the corresponding pore volumes of T 1 and T 2 are 53.21 m²/g (0.32 cm³/g) and 44.47 m²/g (0.23 cm³/g), respectively. A higher surface area corresponds to a smaller grain size 18 . The pore-size distribution curve can be obtained through N 2 adsorption-desorption isotherms by using the Barrett-Joyner-Halenda (BJH) method (inset of Fig. 3). The average pore sizes of T 1 and T 2 are 20.0 and 20.8 nm, respectively.
An XPS analysis was conducted to evaluate defects and oxidation states in the CoMn 2 O 4 spinel structure. Figure 4 shows the XPS spectra of Co2p, Mn2p, Mn3s, and O1s core levels of pure T 1 and T 2 samples. Figure 4(a) presents the Co2p spectra of T 1 and T 2 , respectively, indicating that there are doublet peaks of Co 2p 3/2 and Co 2p 1/2 located at 780.46 and 795.96 eV for T 1 and 780.62 and 796 eV for T 2 , respectively. The energy splitting (ΔE) values of the divalent and trivalent Co ions are 15.4 and 15.1 eV for T 1 and T 2 , respectively, which are consistent with values reported in the literature 19 . The Co2p spectra of T 1 and T 2 are fitted into six peaks with binding energies of approximately 780.4 and 795.8 eV for Co 2+ and 782.2 and 797.3 eV for Co 3+ 5,19 . Furthermore, two satellite peaks are observed at 786.13 and 802.6 eV for T 1 and 785.84 and 802.45 eV for T 2 , respectively 19 . The concentrations of Co 2+ and Co 3+ are 63% and 37% for T 1 and 70% and 30% for T 2 , respectively(as listed in Table 1). Figure 4(b,c) show the Mn 2p and Mn 3 s spectra of T 1 and T 2 , respectively. Figure 4(b) shows the Mn 2p spectra combined with 2p 3/2 and 2p 1/2 located at approximately 641.9 and 653.8 eV for T 1 and 642 and 653.7 eV for T 2 , respectively. The Mn2p 3/2 and Mn2p 1/2 of T 1 are deconvoluted into sub-peaks located at approximately 641.39 and 652.91 eV for Mn 3+ and 642.5 and 654 for Mn 2+ states, respectively. Furthermore, those of the T 2 sample are deconvoluted into sub-peaks located at approximately 641.2 and 652.91 eV corresponding to Mn 3+ and 642.5 and 653.9 eV for Mn 2+ states 20 . The concentrations of Mn 2+ and Mn 3+ in T 1 are 73% and 27%, respectively, whereas those for T 2 are 82% and 18%, respectively (Table 1). To ensure peak splitting and doublet, the parallel spin coupling of Mn 3 s of T 1 and T 2 were analysed, and Fig. 4(c) depicts the results. Mn 3 s core level spectra show binding energies at approximately 83.9 and 89.3 eV for T 1 and 83.9 and 89.5 eV for T 2 . The energy separation for T 1 is 5.4 eV, which is lower than that for T 2 , (5.    www.nature.com/scientificreports www.nature.com/scientificreports/ Electrochemical performance examinations of T 1 and T 2 electrodes were conducted using a three-electrode cell system. The cyclic voltammetry (CV) responses of T 1 , T 2 , and bare Ni foam are presented in Fig. 5(a) at a scan rate of 20 mV s −1 in a voltage window from −0.1 to 0.6 V and indicate that both T 1 and T 2 exhibit oxidation and reduction redox peaks at approximately 0.2 and 0.5 V, respectively, because of the Faradic reaction. The quasi-reversible Faradic reaction with redox peaks for T 1 and T 2 reveals the occurrence of surface redox reactions, which are provided as follows 25,26 .  Figure 5(b,c) show the CV responses of T 1 and T 2 at a scan rate from 5 to 200 mV s −1 at −0.1 to 0.6 V, respectively. Moreover, no significant changes are observed in the shape of CV curves when the scan rate increased from 5 to 200 mV s −1 , indicating that both electrodes have satisfactory electronic conduction and low equivalent series resistance 27 . Because of the kinetics of redox reactions, the current increases with the increasing scan rate  www.nature.com/scientificreports www.nature.com/scientificreports/ in accordance with Ohm's law 26 . Because of the internal resistance between the electrode and electrolyte ions, the anodic peaks show red shift of 0.17 V and cathodic peaks blue shift of 0.1 V with the increasing scan rate for T 1 which is higher than T 2 (0.16 V for the anodic scan and 0.1 V for the cathodic scan) [ Fig. 5(b,c)] 27,28 . The reduced peak shifting of T 2 indicates that a lower overpotential is required for ionic transport because of a high current response at higher scan rates. Therefore, T 2 shows a higher rate capability than T 1 .
Specific capacitance (F g −1 ) can be estimated from the CV curve by using the following equation 14,18 : where γ, m, (V 2 − V 1 ), and I represent the scan rate, mass of active materials, voltage window, and current response, respectively. Figure 5(b,c) show that the specific capacitances (C sp ) of T 1 and T 2 calculated using Eq. (1) are 730 and 1071 F g −1 at 5 mV s −1 , respectively. The scan rate dependence C sp [Fig. 5(d)] indicates that C sp values of T 1 and T 2 decrease with the increasing scan rate, which is attributed to the decreasing redox active sites with an electrolyte at high scan rates 14,29 . The mesoporous structure in the electrodes creates low-resistance pathways for ion diffusion and increased charge transport. Therefore, electrolyte ions are soaked by the mesoporous wall and increase the capacity of the electroactive channel for high charge storage at a high scan rate 16 . The area under the CV curve of T 2 is larger than that of T 1 , thereby improving capacitive performance and ionic conductivity. Because both T 1 and T 2 electrodes are constructed using nanocrystalline materials with a mesoporous structure, electrolyte ions could be transported through their nanochannels 14 . When the scan rate increased to 200 mV s −1 , T 1 and T 2 show specific capacitance of 318 and 493 F g −1 with 43% and 46% retention of the initial capacitances, respectively. The variation in charge storage efficiency may be attributed to the difference between the crystallite size and surface morphology of T 1 and T 2 . CV performance was significant for the infliction of potential in pseudocapacitive devices. Furthermore, charge-discharge behaviour was one of the most crucial characteristics. Figure 6(a) exhibits that the charge-discharge characteristics of T 1 and T 2 are −0.2 to 0.5 V at a current density of 1 A g −1 , indicating that the discharge profile is nonlinear, which is prominent pseudocapacitance within this potential window. These discharge profiles can be divided into two parts. One part varies nonlinearly from 0.5 to approximately 0.25 V, suggesting pseudocapacitor behaviour, whereas the other parameter varies from approximately 0.25 to −0.2 V, indicating www.nature.com/scientificreports www.nature.com/scientificreports/ a double-layer capacitor mechanism 14 . This phenomenon is attributed to the quasi-reversible redox reaction at the electrode-electrolyte interface 2,14 . Figure 6(a,b) show the charge-discharge profiles of T 1 and T 2 from 1 to 30 A g −1 , which demonstrate Faraday pseudocapacity with a negligible voltage drop.
The specific capacitance (F g −1 ) can be calculated from the charge-discharge curve as follows 30 : where m, I, V 2 , V 1 , and Vdt t t 1 2 ∫ represent the mass of an active material, discharge current, upper voltage, lower voltage, and area covered by the discharge curve, respectively.
Moreover, C sp at current densities ranging from 1 to 30 A g −1 is estimated using Eq. (2), and Fig. 6(d) displays the results. Figure 6(a) reveals that the discharge time of T 2 is considerably longer than that of T 1 because of the longer time for the redox reaction; therefore, C sp of T 2 (1709 F g −1 ) is considerably higher than that of T 1 (990 Fg −1 ) at 1 A g −1 [ Fig. 6(d)]. When the current density increased to 30 A g −1 , C sp values of T 2 and T 1 are 1216 and 641 F g −1 with capacitive retention of 71% and 64%, respectively. Such high capacitive retention at high current density indicates the effect of oxygen vacancies 5 . The improved performance of T 2 is attributed to the effective sharing of oxygen vacancies causing a high degree of contact of OH − ions for Faradic reactions 31 . Note that the specific surface areas of T 1 and T 2 are nearly equal, yet, the pseudocapacitive charge storage between the two materials are distinctive; therefore, we can infer that the higher C sp and retention of T 2 than those of T 1 are due to the higher oxygen vacancies concentration in T 2 32 , as discussed below. In the CV procedure, the current response is measured with respect to voltage at constant time. However, galvanostatic charge-discharge (GCD) measurement is performed at a constant current for a varying voltage with respect to time. GCD measurement is generally more applicable to the study of the pesudocapacitive material 33 . A slightly higher specific capacitance is observed in this study for the GCD method compared with CV [Figs 5(d) and 6(d), respectively], which is attributed to the different measurement procedures 33 .
To understand the redox nature of oxygen intercalation in CoMn 2 O 4 , ex-situ XPS of T 1 and T 2 cathode materials along with carbon black are also performed after 50 and 100 consecutive CV cycles (the results shown in the Supplementary Information, Figs S1-S3). www.nature.com/scientificreports www.nature.com/scientificreports/ and Mn 2+ remain unchanged up to 50 cycles. After 100 CV cycles, interestingly, the atomic % of Mn 3+ and Mn 2+ significantly change. Mn 3+ decreases from 26 to 18 while Mn 2+ increases from 74 to 82. As for the cathode T 2 , Mn 3+ slightly increases from 19 to 21 atomic % and Mn 2+ decreases from 81 to 79 atomic % when changing to 50 cycles. After 100 cycles, Mn 3+ reaches an atomic % of 26 and Mn 2+ has 74. Therefore, the net 8% decrease of Mn 3+ for T 1 while 7% increase for T 2 after 100 CV cycles are found. In the meantime, it is shown in Fig. 7(a) 8% increase of Mn 2+ for T 1 and 7% decrease for T 2 after 100 cycles. On the other hand, under consideration of oxygen in the present cathode material, the O 2− ions from the electrolyte should have sufficient resistance to being inserted into a densely packed structure of the cathode material at room temperature to fill the oxygen vacancies without external field 32 . In addition, OH −1 ion from the electrolyte(0.5 M LiOH aqueous solution) would be absorbed by oxygen vacancy site leading to transfer its proton to a neighbouring lattice oxide after applied voltage. As a result, oxygen vacancy would be filled after 50 CV cycles. In general, the valence change of transition metal oxide(such as Mn), (Fig. 7(a)) is correlated with the concentration of charged oxygen vacancy, where atomic concentration of Mn 3+ continuously increases while that of Mn 2+ decreases up to 100 cycles in T 2 cathode material. In the mean time, the valance change in Co of T 2 cathode is not obvious(Supplemental Fig. S3(c)). In this case, the concentration of charged oxygen vacancy would be decreased. But based on Supplemental Fig. S2(c)), the concentration of oxygen vacancy gradually increases up to 100 cycles for both T 1 and T 2 cathode materials. Therefore, it seems contradiction in this case. This phenomenon may probably be attributed to the valence change of charged oxygen vacancy, complex reactions between electrolyte and composite electrode, and/or that the O1s peak contains some additional low intensity peak related to carbon material (example C=O, C-O, etc.) in the cathode affected the vacancy calculation. It needs to be proved in the future work. Figure 7(b,c) show the mechanism for the T 2 cathode material. Both T 1 and T 2 comprise multivalence Mn in lattice sites, primarily Mn 2+ and Mn 3+ [ Fig. 4(b,c)], suggesting the electron transition occurred from Mn 2+ → Mn 3+ and vice versa. Through intercalation, oxygen vacancies in T 1 and T 2 cathode material are combined with oxygen, thereby altering the oxidation states of manganese. Oxygen ions are diffused, resulting in the oxidation state of Mn 2+ to Mn 3+ , which is validated in Fig. 7(a). Therefore, it is concluded that material T 2 has almost equal surface area and more oxygen vacancies than T 1 ; however, T 2 has higher capacitance, Its charge storage is believed to be due to oxygen intercalation 32 . The cobalt oxidation states of T 1 and T 2 cathode materials are almost stable after various CV cycles. This indicates that the manganese is crucial for charge storage (Supplemental Information S3). www.nature.com/scientificreports www.nature.com/scientificreports/ Because of the prevalence of the electron transport and kinetic features of ions at the electrode-electrolyte interface, the pseudocapacitor exhibited a low internal resistance 14,34 . To understand this phenomenon, EIS measurements are performed at frequencies ranging from 1 Hz to 1 MHz. Figure 8 displays the conventional impedance spectra of T 1 and T 2 electrodes. The impedance behaviour introduces a semicircle in the high-frequency region and an inclined line in the low-frequency region, thus revealing pseudocapacitor behaviour 34,35 . The intercept along Z' represents the internal resistance of electrode R 1 . The diameter of the semicircle provides charge transfer resistance R 2 . All resistances (i.e., R 1 and R 2 ) of T 1 are slightly higher than those of T 2 . The oxygen vacancies facilitate the migration of electrolyte ions in the conducting path 11,31,36 ,. Moreover, the electron conductivity of the electrode increases with the increasing number of oxygen vacancies 11,37 . Because the concentration of oxygen vacancies is higher in T 2 than in T 1 , the R 1 and R 2 values of T 2 are 0.62 and 2.57 Ω, whereas those of T 1 are higher, 0.65 and 2.87 Ω, respectively. The Density Functional Theory method is applied to study the effect of oxygen vacancies on the catalytic performance of manganese oxide. Oxygen vacancies increase the conductivity because of compression of the band gap 36 . The R values of T 2 are lower than those of T 1 and are consistent with those in this study 36 .
Asymmetric pseudocapacitors, including T 1 //AC and T 2 //AC, are fabricated, where T 1 and T 2 are the cathode and commercial AC is anode material. Figure 9(a)and (b) present the conventional CV curves of T 1 and T 2 cells at various scan rates in the voltage ranges of 0-1.6 V, which display a nearly rectangular shape. According to the CV characteristics of the T 1 cell [ Fig. 9(a)], no redox peak is observed in a wide voltage window, whereas redox peaks are observed in the T 2 cells at approximately 1.5 and 1.0 V [ Fig. 9(b)], indicating that T 2 cell exhibits two different energy storage mechanisms because of rapid electrolyte ion transport and redox reaction kinetics. Figure 9(c,d) represent the charge-discharge behaviours of T 1 and T 2 cells, respectively, at a current density of 1-10 A g −1 . T 1 cell shows an approximately linear charge-discharge behaviour, whereas T 2 cell exhibits nonlinear behaviour, indicating pseudocapacitive characteristics.
Energy density E (Wh-kg −1 ) and power density P (W-kg −1 ) values are calculated using Eqs (3) and (4) 18 : where C sp , ΔV, and Δt are specific capacitance (F g −1 ), potential window, and discharge time (s), respectively. Figure 9(e) depicts the Ragone plots of the as-fabricated T 1 and T 2 cells, respectively. The energy density of T 2 cell decreases from 32 to 9 Wh-kg −1 with an increase in power density from 0.9 to 8.7 kW-kg −1 , whereas for T 1 cell, the energy density decreases from 5 to 0.8 Wh-kg −1 when increasing power density from 0.7 to 5 kW-kg −1 . T 2 exhibits higher energy density and power density than T 1 . For achieving high-performance pseudocapacitors, obtaining high energy density and power density are crucial. The flake-like morphological structure [ Fig. 2(b)] can improve diffusion paths for electrons and ions in the oxide materials and increase interfacial redox reactions, resulting in the increase in capacitance. Therefore, the T 2 //AC device exhibited high energy and power density [ Fig. 9(e)]. It is believed that the mesoporous structure and flake-like morphology increase the specific capacitance, thus increasing the energy and power density of the asymmetric pseudocapacitor, T 2 . Figure 9(f) presents the cyclic stabilities of T 1 and T 2 cells, which indicate that the T 1 //AC exhibits the negligible decay of capacitance (95%), whereas the T 2 //AC shows 110% retention after 1000 cycles in the PVA-LiOH polymer gel electrolyte at 2 A g −1 . This observation indicates the excellent electrochemical stability of T 1 and T 2 cells. The slight decay of www.nature.com/scientificreports www.nature.com/scientificreports/ cyclic stability for T 1 //AC is possibly attributed to the decrease in the adhesion of active materials with the current collector 27 . These results reveal that cobalt manganese oxide is a prominent electrode candidate material for making asymmetric pseudocapacitors.

conclusions
Modulation of the intrinsic defect concentration in CoMn 2 O 4 tetragonal spinel was synthesised at moderate temperatures by adjusting the mixing sequence of Co and Mn precursors. Moreover, two distinct morphologies of CoMn 2 O 4 were observed as a result of this adjustment. CoMn 2 O 4 synthesised by mixing the Mn precursor after adding the Co precursor (T 1 ) formed nanoparticle morphology. Moreover, a nanoparticle and nanoflake combined structure was formed when the Mn precursor was mixed before adding the Co precursor (T 2 ). The electrochemical evaluations revealed that the T 2 electrode had a higher specific capacitance (1709 F g −1 ) than the T 1 electrode (990 F g −1 ) at 1 A g −1 with capacitive retentions of 71% and 64% for T 2 and T 1 , respectively. A higher number of oxygen vacancies existed in the T 2 electrode, which enhanced the capacitance because of the intercalation of oxygen ions/oxygen vacancies from the aqueous alkaline electrolyte. The T 2 //AC asymmetric pseudocapacitor exhibited a maximum energy density of 32 Wh-kg −1 at a power density of 0.9 kW-kg −1 . The T 2 //AC pseudocapacitor had a higher energy density and power density than T 1 //AC. Both T 1 and T 2 cells demonstrated excellent electrochemical stability. This study not only demonstrates that the mixing sequence of the precursors during CoMn 2 O 4 synthesis is crucial in determining the performance of the pseudocapacitor but also provides insights into the mechanism of charge storage with oxygen vacancies. The present study develops a new type of CoMn 2 O 4 -based electrode for future pseudocapacitor applications.

Methods
Materials. To   www.nature.com/scientificreports www.nature.com/scientificreports/ the sample was denoted as T 1 -CoMn 2 O 4 (T 1 ). Similarly, another sample was synthesised using the same stoichiometric amounts of Co and Mn precursors as the first sample; however, 25 ml of 0.

Material characterisation.
To identify the crystal structure powder, XRD was conducted with Cu Kα radiation using a Bruker D2 PHASER. The specific surface area of nanomaterials was characterised by the BET surface area analyser (ASAP, 2020). SEM was conducted to study the surface microstructure (Hitachi SU-8010). A high-resolution TEM analysis was used to analyse the formation of the nanostructure and composition of materials (JEOL JEM-2010F). To examine the oxidation states of spinel oxides, XPS (ULVAC-PHI Quantera SXM) was performed.
Preparation of working electrodes. To fabricate the working electrode, active material powders were mixed in the N-methyl-2-pyrrolidone (NMP) solution to form a uniform slurry of 10 mg/ml. A porous nickel foam substrate was washed with acetone and etched using a 6 M HCl solution for 30 min. After being washed using deionised water (DI water), the substrate was dried at 70 °C for 6 h. Next, the slurry containing active materials was coated onto the substrate (area of 1 cm 2 ) using a brush and dried at 80 °C for 15 h under a vacuum condition to study its electrochemical properties and mechanisms. The mass loadings of T 1 and T 2 were 0.3 and 0.15 mg, respectively.
Asymmetric pseudocapacitor assembly. The asymmetric pseudocapacitor was fabricated using activated carbon (AC) as an anode material, and T 1 or T 2 was used as a cathode material. The anode electrode was prepared using a mixture of AC, a polyvinylidene fluoride (PVDF) binder, and carbon black in a weight ratio of 85:10:5 in a N-methylpyrrolidone (NMP) solution to form homogeneous dispersion for deposition on the substrate. For fabricating a cathode electrode, active materials (T 1 or T 2 ), PVDF, and carbon black were used in a weight ratio of 75:10:15 and dispersed in the NMP solution. The slurries (T 1 and T 2 ) were coated on the nickel foam substrates (current collector) and dried at 80 °C for 15 h in a vacuum to form the cathodes.
To assemble pseudocapacitor devices, the solid-state electrolyte was first prepared: 1 g of PVA and 1 g of LiOH were dispersed in 20 ml of water and heated at 90 °C to form a transparent gel. After cooling, cathode and anode electrodes were then immersed on the gel electrolyte and dried at room temperature. Both electrodes were pressed to fabricate hybrid supercapacitors and were denoted as T 1 -CoMn 2 O 4 /PVA-LiOH/AC (T 1 cell) and T 2 -CoMn 2 O 4 /PVA-LiOH/AC (T 2 cell). For an asymmetric pseudocapacitor, the charges (Q) for cathode and anode electrodes were balanced using the following equation: Q = CmΔE, where C, m, and ΔE are the specific capacitance, mass of active materials, and potential window, respectively. After calculations, the total masses of T 1 and T 2 cell devices were approximately 1.5 and 0.5 mg, respectively. electrochemical characterisation. The electrochemical performances of electrodes and pseudocapacitors were measured using an electrochemical analyser (Instruments CHI618B), which included CV, galvanostatic charge-discharge cycling (GCC), and EIS. Electrochemical characterisations of the electrode materials were performed using three-electrode cells with a saturated calomel electrode, platinum as a counter electrode, a working electrode of an active material deposited on the Ni foam, and 0.5 M LiOH serving as the electrolyte. The working electrode was immersed in the electrolyte for providing suitable contact between the electrode and electrolyte.

Data availability
Readers can access data by contacting the corresponding author.