==== Front Sci Rep Sci Rep Scientific Reports 2045-2322 Nature Publishing Group UK London 77650 10.1038/s41598-020-77650-6 Article Competing ferro- and antiferromagnetic exchange drives shape-selective Co3O4 nanomagnetism Shepit Michael shepitm@myumanitoba.ca 1 Paidi Vinod K. 1 Roberts Charles A. 2 van Lierop Johan johan.van.lierop@umanitoba.ca 13 1 grid.21613.370000 0004 1936 9609Department of Physics and Astronomy, University of Manitoba, Winnipeg, MB R3T 2N2 Canada 2 grid.467593.aToyota Motor Engineering and Manufacturing North America Inc., 1555 Woodridge Avenue, Ann Arbor, MI 48105 USA 3 grid.21613.370000 0004 1936 9609Manitoba Institute for Materials, University of Manitoba, Winnipeg, MB R3T 2N2 Canada 2 12 2020 2 12 2020 2020 10 2099027 8 2020 13 11 2020 © The Author(s) 2020Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.We have synthesized three different shapes of Co3O4 nanoparticles to investigate the relationships between the surface Co2+ and Co3+ bonding quantified by exploiting the known exposed surface planes, terminations, and coordiations of Co3O4 nanoparticle spheres, cubes and plates. Subsequently this information is related to the unusual behaviour observed in the magnetism. The competition of exchange interactions at the surface provides the mechanism for different behaviours in the shapes. The cubes display weakened antiferromagnetic interactions in the form of a spin-flop that occurs at the surface, while the plates show distinct ferromagnetic behaviour due to the strong competition between the interactions. We elucidate the spin properties which are highly sensitive to bonding and crystal field environments. This work provides a new window into the mechanisms behind surface magnetism. Subject terms Magnetic properties and materialsFerromagnetismNanoparticleshttp://dx.doi.org/10.13039/501100000196Canada Foundation for Innovationhttp://dx.doi.org/10.13039/501100000038Natural Sciences and Engineering Research Council of Canadaissue-copyright-statement© The Author(s) 2020 ==== Body Introduction The nanomagnetism in spinel oxides is wonderfully sensitive to the surface ions bond nature, thus providing a probe to the interactions at the surface. Ideally, one wishes to examine the surface spin magnetism by removing the magnetic contributions from the core. Via structural and magnetic characterizations from bulk and of the different surface terminations, the surface magnetism can provide a quantitative measure of the interionic exchange, and can then be mapped onto the surface chemistry e.g. bonding and coordination environments. This allows one to track exchange interactions and magnetism, such as ferromagnetism and the competition between ferro- and antiferromagnetic exchange interactions at the surface. In brief, Co3O4 has a normal spinel structure with magnetic Co2+ ions located at tetrahedral sites and non-magnetic Co3+ ions at octahedral sites. Even though Co3+ is non-magnetic, it is essential to the overall magnetism by virtue of its presence in the extended superexchange pathway of Co2+-O-Co3+-O-Co2+, with Co3+ ions located at the intermediary sites between O2- ions. Co3+ ions are known to be key players in the magnetic properties, so the intrinsic behaviour should be highly susceptible to changes in the Co2+:Co3+ ratio1,2. Quantifying the exchange interactions present in bulk and at different surface terminations can give us new insights into the ions’ behaviours and coordinations present. This is particularly true for Co3O4 where the exchange is a result of both Co2+ and Co3+ ions located at the different sites. In Co3O4 there are four superexchange pathways: The typical superexchange interaction through Co2+-O-Co2+ ions, and three other exchange paths that involve Co2+–O–Co3+–O–Co2+. The Co2+–O–Co2+ exchange is weak due to the larger distances associated with some of the O2- and Co2+ ions, as first identified by Roth2. On the other hand, individual Co2+–O–Co3+–O–Co2+ paths also have weak exchange, the multiplicity of the different paths between Co2+ ions leads to an overall exchange strength that is not negligible—we define the multiplicity to be the total number of possible interaction paths from a Co2+ ion to all neighbours of a given type (e.g. nearest or next nearest neighbours). This description follows that by Roth3 using a range of spinel oxides. For example, the bulk structural analogue of Co3O4, CoAl2O4, has a much smaller exchange (Jij/kB=0.4K) compared to bulk Co3O4 (Jij/kB=4K). However, CoAl2O4 contains Co2+ ions located at tetrahedral sites and Al3+ ions located at octahedral sites (instead of Co3+ in Co3O4). The intervening Al3+ ion p-orbitals get repelled to a higher energy state by the p-orbitals of the O2- ions, and the exchange pathway through the Al3+ becomes unavailable. By comparison to the CoAl2O4 analogue, the Co2+–O–Co3+–O–Co2+ paths account for ∼90% of the total exchange in Co3O4. By examining the different surface terminations—facilitated by the exchange interactions that propagate through the non-magnetic Co3+ ion, characterization of the nanomagnetism results in direct quantification of the different site environments that are correlated to surface Co2+:Co3+ ratios. The balance of Co2+ and Co3+ ions on the surface and their coordinations in these extended exchange pathways becomes very important as a crystal’s size decreases into the nanoscale regime. Here, we would expect measurable contributions from the surface atoms having a marked effect on the overall magnetic properties4,5. For example, at the nanoscale (e.g. in thin film and nanocrystallite form) broken exchange pathways at surfaces and interfaces result in significant changes in the Co2+:Co3+ ratios, and the local coordination around these ions affecting crystal field effects. This results in a decrease in the strength of the overall antiferromagnetic exchange, competition between the intrinsic antiferromagnetic and ferromagnetic exchange paths (as they can be quite different from the bulk) can induce a wide range of magnetic properties, from an antiferromagnetic spin-flop to ferromagnetic-like behaviour6–8. With the ability to control the shape of Co3O4 nanoparticles, one can examine the impact on the magnetic properties from the exposed surfaces that present different Co2+/Co3+ ratios via the resultant surface terminations. That is, shape-selective control over the intrinsic properties. Specifically, by changing the surface terminations, we expect to reveal how the different magnetic properties are a result of the re-tuned environments, and how these environments are realized through the magnetism. Results Shape and composition Typical transmission electron microscopy (TEM) images of the nanoshapes are shown in Fig. 1a–c. The spheres and cubes have an average diameter and side length of 10 and 15 nm, respectively. The hexagonal plates are 70 nm across with a thickness of 8 nm. The visible lattice planes for the cubes are the (022) and (02¯2) planes (Fig. 1a) occurring with a 90° angle between them, with a distance of 0.28 nm between successive planes. These planes are perpendicular to the {100} family of planes exposed at the surface. The plates show the (22¯0) and (2¯02) planes 60° apart, with a separation of 0.28 nm between different planes (Fig.  1c), perpendicular to the {111} family. From TEM we can see that the plates are textured with domains with sizes of roughly 7–8 nm.Figure 1 (a) TEM image of the cubes. The inset shows the planes at a 90° angle to one another and the fast fourier transform (FFT). (b) TEM image of the spheres is shown. (c) TEM image of the plates. The inset shows the planes at a 60° angle to one another and their FFT. (d) Size distributions of the different shapes from both TEM and SEM images. (e) XRD patterns for all three of the nanoshapes with residuals (blue lines) and peak markers (red ticks) from Rietveld refinement. (f) Lorentzian fits of the XRD intensities over the (111) reflection as discussed in the text. It has been established from surveys of shapes that cubes predominantly expose the (100) planes while (hexagonal) plates expose the (111) planes1,9–12. The size distributions of the different shapes have lognormal distributions with the mean sizes and standard deviations reported in Fig. 1d). The size distribution for the plates was obtained from scanning electron microscopy (SEM) images while additional TEM images for the spheres and cubes are found in the supplemental information (SI). X-ray diffraction results are presented in Fig. 1e). Rietveld refinements were done with GSAS-II13. The patterns were all phase pure Co3O4 with the expected Fd3¯m space group. Occupancies were unchanged between the different shapes, indicating that the structure is the anticipated normal spinel. The crystallite sizes obtained from refinements match those from TEM image analysis; plates are composed of many crystallites (9(1) nm) while the cubes (25(1) nm) and spheres (15(1) nm) are single nanocrystallites. All lattice constants (spheres a=8.075(3) Å, cubes a=8.078(3) Å and plates a=8.062(3) Å) are close to the bulk value of a = 8.065(3) Å2. Preferred orientation was found for the the plates along the (111) reflection. Since preferred orientation is due to surface planes, the magnitude is small, and is most easily shown with normalized intensities for the (111) plane (Fig. 1f). The data most clearly shows the difference in intensity caused by preferred orientation for the (111) plane, with no other reflection displaying this difference (other than (222), etc.). Solid lines are Lorentzian curves to guide the eye for each shape over the (111) reflection, where the plates show an increase of roughly 25% in normalized intensity over the other shapes due to preferred orientation. Surface environment and cation distribution To obtain information about the different surface terminations for the shapes, we can use a program to reconstruct the different exposed planes. Wherein the surface terminations can elucidate the magnetic behaviour caused by the different shapes. Visualization of the shapes with their surface planes was performed using VESTA14 to quantify the total surface areas, the surface areas of planes, and the Co2+ to Co3+ ratios of those planes. The Co2+:Co3+ ratio found at the surface of the shapes are 1.8:2 for the plates, 1.4:2 for the spheres, and the cubes matching the Co2+:Co3+ ratio from bulk Co3O4 (1:2). For the spherical particles there are no specific exposed planes, but an average of equal amounts of three different families of 100, 110, and 111 planes, which gives 26 exposed planes; six from the 100, 12 from the 110, and eight from the 111 family. As a way to correlate certain properties to the different shapes, we look at the surface area of the (111) plane for the different shapes. This, along with the total calculated surface areas (m2/g) and the Co2+:Co3+ ratio at the surface for the different shapes are shown in Table 1.Table 1 Co3O4 exposed planes. Shape (Plane) Surface area (m2/g) (111) Surface area (m2/g) Co2+:Co3+ surface ratio Spheres (100), (110), (111) 110.0 33.9 1.37:2 Cubes (100) 58.5 0.0 1:2 Plates (111) 25.0 20.2 1.75:2 XAS characterizes the partial density of states just above the occupied density of states (Fermi energy) that can be influenced by hybridization with the ions’ ligand orbitals, thus identifying the different species of ions present, their coordinations and vacancies. In perovskite structures, oxygen vacancies can lead to a change in the lattice parameter. The vacancies promote charge transfer resulting in a measurable change in oxidation state for the cations16. Figure 2a shows the Co L3,2-edge spectra for each of the shapes, where the large peak at 779 eV corresponds to Co3+ and the shoulder at 777 eV corresponds to Co2+. All spectra are essentially the same for the shapes and describe a Co2+:Co3+ ratio of 1:2. Co L-edge XAS reveals no change to the cation occupations. Thus, for the different shapes, we find no change that would be caused by oxygen vacancies in the structure. Keep in mind that while TEY is a surface probe (2–3 nm probe depth for these L3,2 energies17, we would expect to see at most an intensity change of 2% between the spheres and plates (see SI).Figure 2 (a) Co L3,2-edge XAS and (b) XAS normalized XMCD spectra at 10 K, obtained under applied fields of 50 kOe using TEY detection. (c) O K-edge XAS spectra at 10 K. The t2g and eg peaks are labelled I1 and I2, respectively; the asterisk marks the peak from the oxygen on the carbon tape15. (d) The variation of I2/I1 between the shapes is shown. Indicating a difference in occupation of the Co3+ ions at the surface. The O K-edge XAS for the different shapes presented in Fig. 2c quantifies the hybridization of the metal Co3+ 3d orbitals with the O2- ligand 2p orbitals via splitting of the major (spin-up) and minor (spin-down) eg and t2g energies in addition to crystal field splitting effects18. The observed changes in peak intensities at 530 and 533 eV provides a measure of the available 3d hole states on the Co ions hybridized with the O ions, while energies above 535 eV indicate hybridization with 4s orbitals. Quantitatively, the plates have a higher number of available eg/t2g states on the Co3+ caused by a concomitant decrease in the coordination at the surface1,18. O K edge measurements provide a great link between the different surface terminations for the shapes through the different bonding environments of the oxygen ions. Correlating the ratios of the eg/t2g peaks, along with surface areas can provide us with the number of ions and coordinations present at the surface. The variation of I2/I1 vs Co3+/Co2+ ratio shown in Fig. 2d clearly identifies the differences between the spheres, cubes, and plates from the majority of oxygen ions at the surfaces; directly related to the Co-O bond combinations for the different shapes, following the Co2+:Co3+ surface ratios identified using VESTA. At the surface of the cubes the (100) plane contains Co3+ ions coordinated to five oxygen ions (CoO5), while the (111) plane contains three fold coordination on the exposed Co3+ ion (CoO3). From cubes, to spheres, to plates there are changes in the O 2p-to-Co 4s hybridization. This is the result of the exchange paths of the exposed planes resulting in changes to the Co-O bonding between the different shapes, affecting the measured hybridization. The cubes show more defined peaks due to hybridization from 537 to 545 eV (consistent with highly coordinated transition metal ions), indicating the stability of the (100) planes at the surface of the cubes1. Probing only the magnetic ions, ferromagnetism is most cleanly observed via X-ray magnetic circular dichroism (XMCD). Using both right and left circularly polarized X-rays to measure both spin-up and spin-down populations, an XMCD spectrum is not observable for antiferromagnets. For Co3O4, crystal field splitting dictates that all spins should be paired in the lower t2g orbital of the Co3+ ions, while Co2+ contains three unpaired spins (S=3/2) in the t2g orbital. The XMCD spectra in Fig.  2b collected over the Co L3,2-edge for all three nanoshapes provide irrefutable evidence of ferromagnetic behaviour—an antiferromagnet with equal spin up and down populations presents no XMCD signal. We find that 30–50% of the signal is due to the surface ions on the shapes. It should be noted that an XMCD signal occurs with the surface-probe total electron yield (TEY) measurements; total fluorescence yield (TFY) spectra that are bulk sensitive show a vastly decreased XMCD signal (for further discussion, see SI). The differences between the shapes’ XMCD spectra over the L3 edge at 779 eV are due to the different Co3+ environments at those surfaces. A decrease in the Co3+ coordination at the surface causes a change in the crystal field environments resulting in an additional magnetic moment on the Co3+ ions of the (111) plane. This can be further identified through the overall magnetic properties. Magnetism of the surface terminations We can quantify the differences in the Co ions’ surroundings (bonding, exchange interactions) on the different surface terminations of the nanoshapes by the straight-forward differences in the overall magnetism. For example, the different shapes’ low-field DC susceptibilities (χDC(T)) in Fig.  3a gives the ordering temperature of each shape and the magnetic moment per formula unit—allowing the quantification of the (overall) superexchange present in each shape. Keep in mind that bulk Co3O4 is antiferromagnetic with TN=40 K; the shapes show variations in TN where the spheres and cubes have TN’s of 31(1) and 33(1) K, respectively, while the plates have TN=25(1) K (The Néel temperatures were take from the peak in χDC and when dχDC/dT=0, Fig. 3c. Quantification of the exchange and the magnetic moments were from analysis of 1/χDC(T) (Fig. 3b) above the Néel temperature, where in this temperature regime the systems are all paramagnetic so χDC(T) can be represented by the well known modified Curie–Weiss Law χDC(T)=χ0+CT-θ, where χ0 is a temperature independent component to the susceptibility (Van Vleck paramagnetism), C is the Curie constant, and θ is the Weiss temperature (for an antiferromagnet θ<0). The results of the fits are in Table 2.Figure 3 (a) Molar susceptibility (χm) of the shapes from 20 to 300 K with the susceptibility of bulk shown in orange. (b) Fits (red lines) of the inverse susceptibility 1/χm(T) in the region of 170–280 K. (c) Derivative dχm/dT shows the Néel temperatures for the different shapes. (d) Zero field cooled (χZFC(T), closed symbols) and field cooled (χFC(T), open symbols) susceptibility of the Co3O4 nanoshapes. We can now quantify any differences from bulk Co3O4 and the variations betwen the samples related to the surface areas and exposed planes. Since the temperature independent susceptibility, χ0, of bulk Co3O4 was shown to have near ionic-like values of Van Vleck paramagnetism for Co2+ and Co3+ ions19, we would expect χ0 to remain little affected by the size and shapes. Indeed, all three shapes have a χ0 essentially identical to the bulk value. θ, the Weiss temperature, is related to the exchange interaction strength (bulk values for θ are reported between − 53 and − 110 K2,20–22), and all three shapes have θ values consistent with the bulk values. μeff gives a quantitative measure of the overall magnetic moment (μeff=2.83×C1/2), and there is an increase in the Curie constant C (and μeff) that depends on the shape of the particle (Table 2). C is largest for the spheres due to the highest overall surface area, followed by plates and cubes. But, even though the cubes have a larger per particle surface area than the plates, their magnetic moment is near bulk. The spheres expose the (100), (110), and (111) planes equally, however the plates expose the surface (111) planes, and the increase in C can be related to this plane. The change in the overall Co moment of the nanoparticle between the different shapes is related to the surface coordination and configuration of the ions on those planes.Table 2 Magnetic properties. Shape χ0 (emu/mol Oe-1) C (emu K/mol Oe-1) θ (K) μeff (μB) Jij (K/kB) Spheres 7(3)×10-3 4.0(1) − 93(5) 5.7(1) 1.5(2) Plates 1(1)×10-3 3.1(1) − 107(5) 5.0(1) 1.6(1) Cubes 1(1)×10-3 2.1(5) − 83(5) 4.1(5) 3.0(1) Bulk22–22 0.7×10-3 2.1 − 85 4.1 4.0 Furthermore, μeff and TN from the DC susceptibility provide a measure of the exchange interaction strengths from the surface. Since between magnetic ions Si and Sj exchange is described by the Hamiltonian, Hex=-2JijSi·Sj where Jij=3kBTN2zJ(J+1), and z is the number of nearest neighbours. For bulk Co3O4 with TN=40K, z=4 and J=S (the spin only value with S=32) we find Jij=4.0K/kB. Due to a large crystal field splitting and small atomic number, the 3d transition metal Co ions have quenched orbital angular momentum (L=0). For the different shapes we do not use the spin-only value for J(J+1), instead we obtain the total angular momentum from μeff2=4J(J+1)μB2 (obtained from magnetometry). This is to account for μeff shown to describe contributions from the surface spin magnetism23, where broken symmetry and low coordination provide mechanisms for the presence of unquenched angular momenta for surface atoms, and to account for the possibility of a small magnetic moment present on the Co3+ ion24. The values for the exchange Jij for the shapes are presented in Table 2. The exchange in the cubes are most similar to that of the bulk. Overall, the plates have a μeff larger than bulk and a much lower TN, indicating a weaker Jij (factor of two weaker), linked to the (111) plane exposure, and the spheres have the weakest exchange (largest overall moment) due to the combination of highest surface area and a (111) plane exposure (Table 1). Low coordination on the Co3+ ions and uncompensated surface ferromagnetism (as revealed by the Co and O XAS and XMCD) lead to a larger magnetic moment from the spins on the (111) planes. We can examine the magnetism below TN as a function of applied field and temperature, to further identify characteristic behaviour. Figure 3d shows the zero field cooled (ZFC, χZFC(T)) and field cooled (FC, χFC(T)) susceptibilities for the shapes below 50 K, where the spheres and cubes display very similar behaviour in both ZFC and FC susceptibilities—an increase in χZFC(T) and χFC(T) as temperatures cooled below 12 K. This is the result of a reconfiguration of the spins at the surface. It is characteristic behaviour seen in other systems displaying an antiferromagnetic spin-flop6,25,26. An antiferromagnetic spin-flop occurs at a critical value of the applied magnetic field in antiferromagnets where the spins parallel to the magnetic field undergo a (90°) reorientation producing a net magnetic moment in the direction of the applied field. For the spins in the shapes near the surface, a decrease in the strength of the exchange interactions cause the spin-flop to occur at a lower critical applied field than the core of the particles5,7. An illustration of a surface spin-flop is shown in the SI. By contrast, the nanoplate’s strong response with cooling of χFC(T) below TN is representative of ferro-/ferrimagnetism. The plates present a large bifurcation below TN revealing the dominant ferromagnetic interactions at the surface. The high-field susceptibility χHF is obtained by fitting a linear component above fields of H>30kOe in the M(H) data. The overall magnetization can be represented by an equation that describes all the contributions to the magnetization27,28: M(H)=Msurf(H)+χHFH. Generally, the high-field susceptibility contains contributions from the bulk and the subtracted hysteresis loops, Msurf(H), represent the remaining ferromagnetic contributions to the magnetization from the particles’ surface. Temperature dependencies of the high-field susceptibility χHF(T) in Fig.  4a identifies core behaviour for the samples and aids in assigning the remaining contributions to the different surface terminations. χHF(T) for all three shapes show values of the magnetic moment and exchange constant nearly identical to the bulk χDC(T). Thus, the core of each of the shapes can be described by bulk Co3O4, lending credence to our interpretation that the surface magnetism dominates the low-field susceptibility. Additionally, we find that TN for the plates in χHF(T) peaks at 30 K, different from χDC(T) and closer to the bulk value. This is further evidence that the core orders differently from the surface. Figure 4 (a) χHF vs T is shown for all the shapes. The high-field susceptibility is from the core of the particles. For the plates, χHF peaks at 30 K and above 50 K all the shapes behave similar to bulk (orange). (b) Coercivity (Hc) and exchange bias (Hex) below the ordering temperature TN. (c) Surface component of the M versus H loops, obtained from subtraction of the high-field susceptibility. The hysteresis loops, Msurf(H), describing the surface are shown in Fig. 4c are obtained by subtracting χHFH from the M(H) data using the above equation27,28. The spheres and cubes display “inverted” hysteresis loops at temperatures between 10 and 25 K, the clearest indication that a spin-flop transition has occurred, and surface antiferromagnetic interactions are still dominant5–7,25,26. As temperature decreases, the thermal activation energy decreases leading to stronger competition at the surface layer. This lowers the field at which the spin-flop can occur. By 2 K, the spin-flop occurs at such a small field that the magnetization of the surface (spin-flop) layer saturates. It has been long established29, that when two adjacent ferro- and antiferromagnetic systems are coupled an exchange occurs between the two materials known as exchange bias. The exchange between the materials can result in enhanced coercivities, horizontal, and vertical loop shifts. For the shapes, the presence of exchange bias reveals the existence of distinct populations of spins at the surface. The surface spins (the spin-flop layer of the spheres and cubes and the ferromagnetic layer of the hexagonal plates) couple to the antiferromagnetic core enabling a coercivity (Hc, in an overall antiferromagnet), a vertical loop shift (ΔMs) and exchange bias (Hex). Hc(T) and Hex(T) are shown for all of the shapes in Fig.  4b. The magnetization for the plates show ferromagnetic behaviour, the with magnetic saturation remaining positive and constant throughout the temperature region below TN. The plates have a larger (4–5×) coercivity than the other shapes that also remains relatively constant, indicating the strong ferromagnetic interactions and anisotropy occurring from the surface Co ions (Fig. 4b). The surface of the plates cannot be described by spin-glass behaviour. As the surface spins freeze we would expect to see a large increase in the coercivity and exchange bias field30,31. The plates show a constant coercivity at temperatures below TN, and all shapes show similar behaviour in Hex indicating that no freezing is occurring at the surface32–34. Discussion The superexchange interaction is the dominant mechanism for exchange between the magnetic Co2+ ions in Co3O4. The different exchange mechanisms present are shown in Fig. 5a–c). Where the more direct Co2+–O–Co2+ (Fig. 5a) superexchange interaction only accounts for a small (∼10%) portion of the total exchange strength. The three different Co2+-O-Co3+-O-Co2+ paths are the reason for the strong antiferromagnetic order in the bulk. The virtual electron exchange interactions that occur in the superexchange interaction manifest in the extended interaction (Fig. 5b). Although the exchange interactions are extended over three intervening ions, the general rules for superexchange can be applied to Co3O4, with one main distinction: The differences in the exchange paths in Co3O4 come from the O-Co3+-O bond rather than the usual metal–oxygen–metal configuration35–37. Using the convention of ↑↓180, ↑↓90 and, ↑↑90 for the three different superexchange paths where the arrows indicate the interaction is either antiferromagnetic (↑↓) or ferromagnetic (↑↑) and the subscript denotes the O-Co3+-O bond angle.Figure 5 (a) Shows the interaction path of the Co2+-O-Co2+ configuration. O2- orbitals are shown in red and orange; Co2+ and Co3+ orbitals are shown in purple and blue, respectively. (b) Shows one of the possible configurations for the Co2+-O-Co3+-O-Co2+ pathway. (c) Shows all of the extended Co2+-O-Co3+-O-Co2+ exchange pathways. (d) Illustration of the (100) and (e)  (111) surface terminations for the cubes and plates, respectively. The three different exchange paths are shown in Fig.  5c). ↑↓180 is antiferromagnetic due to the coupling of the O2- ions through the same Co3+ orbital. For the two 90° exchange paths, one is ferromagnetic (↑↑90) due to the coupling through separate Co3+ orbitals, and one is antiferromagnetic (↑↓90) because the two interactions are always 180° to each other through the same Co3+ orbital (Fig. 5c)35–37. This leads to opposing interactions from the Pauli exclusion principle. With competing ferro- and antiferromagnetic interactions present at the surface there is the potential for weakened antiferromagnetic or dominant ferromagnetic interactions. Paths ↑↓90 and ↑↑90 have the same bond distances, angles, and multiplicities. To first order, the interaction strengths cancel2. Thus, path ↑↓180 is then the only uncompensated interaction path and is responsible for the strong antiferromagnetic exchange in bulk Co3O4. At the surface, we can quantify this competition between ↑↓90 and ↑↑90 exchange paths, using the multiplicities for exchange paths given in Table 3 we formulate the quantity δ given by: δ=1+(M↑↑90-M↑↓90)M↑↓180, where M is the multiplicity of the path given in the subscript, and we find that a δ>1.1 characterizes dominant ferromagnetic interactions at the surface, δ<1.1 characterizes dominant antiferromagnetic interactions at the surface, and δ=1/12 is the value for bulk Co3O4. For the different shapes we find that the cubes have a value of δ100=0.83, the plates have a value of δ111=2.17, and the spheres have a value of δsphere=0.92 (obtained from an average of δhkl for the three families of planes at the surface of the spheres). These three values for δ are in agreement with the analysis presented for the magnetic structure shown above, where the spheres are shown to display antiferromagnetic behaviour similar to the cubes, despite the large exposure of the (111) plane similar to the plates.Table 3 Co3O4 surface exchange pathways. Multiplicity (111) Plane (110) Plane (100) Plane Bulk ↑↓180 6 7 6 12 ↑↓90 6 12 12 24 ↑↑90 18 12 16 24 Visualizations from VESTA of the surface of the (100) plane at the surface of the cubes, and (111) plane at the surface of the plates, are shown in Fig. 5d,e, respectively. The (100) plane shows the formation of the Co3+–O planes with a much higher Co3+ coordination leading to the stability of the (100) plane. The nature of the (111) plane is very different, with Co3+ ions threefold coordinated to oxygen ions in a zig–zag like structure, where the Co3+ ions directly at the surface can only promote the ‘W-shaped’ ↑↑90 exchange path. Examining the surfaces for all three planes—(100), (110), and (111), the exchange interactions present are shown in Table 3, where the multiplicity is the total number of possible interaction pathways from a Co2+ ion to all neighbours of a given type (e.g. nearest or next-nearest neighbours). We can now begin to examine the impact of the surface ions of the different terminations on the overall magnetic properties. Examining the magnetic behaviour for the cubes, we find the (100) plane displays a much stronger antiferromagnetic exchange relative to the (111) plane of the plates. This is a result of the configuration of the Co2+ and Co3+ ions at the different surfaces that was found through the structural and magnetic properties, such as the effective magnetic moment of the different planes and the direct variation in bonding observed from XAS and XMCD. The O K edge XAS provides us with a direct link between the magnetic properties that arise from the bonding between the shapes. With the ability to describe the coordination and environments for the Co ions, this allows us to correlate the magnetic properties for the different planes and shapes. Broken exchange pathways at surfaces result in significant changes to the Co2+:Co3+ ratios and the local coordination, as we have quantified. This changes the local exchange structure of the surface enabling interesting magnetic behaviour such as an antiferromagnetic spin-flop and distinct ferromagnetism. We found that only the (111) plane contains Co3+ with a magnetic moment corresponding to the low coordination of Co3+ ions (3/6) and dominant ferromagnetic exchange. The coordination of Co3+ ions on the (100) and (110) planes are closer to bulk (5/6 and 4/6, respectively)1. The quantity δ provides a good measure of the interactions at the surface—but these interactions are based on the available coordination for the ions. Thus, we find that δ is than related to the overall coordination of the ions at the surface (through exchange interactions). Individually, the cubes and plates allow us to characterize the magnetic behaviour for both the (100) and (111) plane, respectively. The weakening of the antiferromagnetic exchange paths at the surface of the plates is much greater. The cubes contain a larger surface area over the plates, but still display hindered magnetic behaviour. The surface of the cubes contain a highly coordinated plane of Co3+ and O ions, promoting antiferromagnetic interactions, but display a spin-flop showing ferromagnetic-like behaviour due to the coupling with the antiferromagnetic core. The surface of the plates contains an additional magnetic moment arising from the Co3+ ions. Low coordination alters the crystal field environment increasing the anisotropy of the surface and allowing for the appearance of a magnetic moment. The dominant ferromagnetic interactions of the (111) plane promote ferromagnetism at the surface of the plates. Through XMCD and magnetometry, it was found that the core of each shape is well described by bulk Co3O4. As an antiferromagnet, the absence of a net magnetic moment in the core precludes the possibility of shape anisotropy38. Thus, any increase in the effective anisotropy comes from the different surface environments such as steps, kinks, and exchange interactions38,39. Examining the (111) plane (Fig. 5e), we find that the ions at the surface align in a stepped configuration which greatly increases the anisotropy of a surface. In addition, there is a complementary increase in anisotropy introduced by the exchange interactions involving the lowly-coordinated Co3+ ion39. This results in the increased coercivity that was obtained for the hexagonal plates. The plates have the smallest overall surface area, so the large coercivity reveals that the anisotropy of the (111) plane is greatly enhanced over the other planes. For the spheres the magnetic properties can be shown to be a combination of the different exposed planes, where the (100), and (111) plane were characterized using cubes and plates, respectively. The exchange constant Jij for the spheres is the smallest, with the extra contribution of the effective magnetic moment arising from the exposed (111) plane. Based on surface area arguments the value for the effective magnetic moment, and thus, exchange constant Jij for the spheres can be correlated to the exposed planes of the surface, as described above. We were also able to characterize the overall magnetic behaviour with the parameter δ that is correlated to the surface exchange and coordination of the ions. Methods Synthesis For the spheres, ϵ-Co was synthesized by first dissolving 0.1g of trioctylphosphine oxide (TOPO) into 12ml of o-dichlorobenzene and 0.2ml of 99% oleic acid. The solution is then heated to 180 °C under an argon atmosphere. In a separate solution, 0.54 g of dicobalt octacarbonyl (Co2(CO)8) was dissolved into 3 ml of o-dichlorobenzene (DCB) and injected into the heated (180 °C) solution of TOPO, oleic acid, and DCB. The obtained Co was washed using ethanol. The precipitate was dried in a furnace at 60 °C for 12 h in air. To obtain Co3O4 the ϵ-Co was placed in a tube furnace with a initial heating rate of 5 °C/min until 300 °C was reached, and the temperature was held for 3 h. Subsequently, the temperature was ramped down at a rate of 2 °C/min. Oxidation of Co causes transformations to first CoO then finally Co3O440. For the cubes, a blue/green precipitate of cobalt hydroxide (α-Co(OH)2) was synthesized by dissolving 9.52g of cobalt chloride CoCl2·6H2O into 100ml of distilled (DI) water. 25ml of ammonium hydroxide NH4OH was added and the solution was stirred for two hours. Drying and calcination procedures were the same as the spheres. For the plates, β-Co(OH)2 was similarly synthesized, dissolving 4.76g of CoCl2·6H2O into 100ml of DI water, while 3g of sodium hydroxide (NaOH) is dissolved in 25ml of DI water. The hydroxide solution was added to a heated CoCl2 solution (80 °C) and was stirred for two hours at 80 °C. Drying and calcination procedures were the same as the cubes and spheres above. β-Co(OH)2 contains divalent Co ions octahedrally coordinated to hydroxyl ions in a layered hexagonal brucite structure41. α-Co(OH)2 forms a layered hexagonal hydrotalcite-like structure, but with Co ions in both tetrahedral and octahedral sites41. For the synthesis of the plates, the initial colour of the precipitate starts blue/green as an α-Co(OH)2 but immediately changes to the characteristic pink colour of β-Co(OH)2. This occurs as the β phase is the more thermodynamically stable of the two forms42. Over the course of the two hour synthesis the colour gradually changes from pink to a light brown. This indicates a transformation at the surface of the plates from β-Co(OH)2 to CoOOH (cobalt oxyhydroxide) which contains trivalent Co ions. The plate shape is formed at the β-Co(OH)2 stage, after calcination the shape is retained with crystallites that form to compensate for the structural change. The α-Co(OH)2 (cubes) contains both octahedrally and tetrahedrally coordinated Co2+ ions at the surface of the as-synthesized particles, while the β-Co(OH)2 (plates) contain only Co3+ octahedrally coordinated at the surface of the particles (due to the oxyhydroxide surface). It is for these reasons that we end up with the surface terminations and shapes in Co3O4 i.e. the cubes have a higher Co2+:Co3+ ratio (1:2) at the surface with highly coordinated ions, and the plates have Co3+ ions with a very low coordination and a small Co2+:Co3+ ratio (1:1). Electron microscopy Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were performed at the Manitoba Institute for Materials. TEM was performed using a FEI TALOS TEM with an accelerating voltage of 200keV. SEM was performed on a FEI Nova NanoSEM 450. Images were processed in Image J43. X-ray diffraction X-ray diffraction (XRD) was performed with a BRUKER D8 Discover diffractometer. With a Cu Kα (λ=1.5405 Å) X-ray source at a voltage and current of 40kV and 40mA, respectively. Patterns were collected from 10 to 90° with a resolution of 0.02° on a rotation stage with a knife edge and Ni Kβ filter. Visualization of the surface terminations by VESTA are performed using information obtained from the crystallography information files (CIF) from XRD refinements. We also utilize information from previous characterizations of Co3O4 surface planes characterized experimentally using high resolution TEM and theoretically using density functional theory (DFT) calculations1,9–12. Magnetometry Magnetic measurements such as DC susceptibilities and magnetization vs field (M vs H) loops were performed using a Quantum Design MPMS-XL SQUID magnetometer. Samples were mounted in supracil quartz tubes to minimize background interference. DC susceptibility measurements were performed from 2 to 300 K in an applied field of 100Oe. M vs H loops were performed at temperatures of 2, 5, 10, 15, 20, 25, 30, 50, 100, 200, and 300K with maximum applied fields of ±50kOe. Field-cooled hysteresis loops were performed at the same temperatures, with a cooling field of +50kOe. X-ray absorption spectrocopy (XAS) and X-ray magnetic circular dichroism (XMCD) X-ray absorption spectroscopy and X-ray magnetic circular dichroism is collected with total electron yield (TEY) and total fluorescence yield (TFY) at the Advanced Photon Source at Argonne National Laboratory, utilizing the beamline 4-ID-C. Samples were placed on carbon tape mounted on the cold finger of the 70kOe superconducting magnet. Co L3,2 edge spectra were normalized to the peak at 779.5eV and O K-edge spectra normalized to the peak at 529.5eV. XMCD spectra were taken under applied fields of ±50kOe, and from these an artifact-free (AF) signal is obtained. Artifact-free spectra are obtained by collecting spectra under both positive and negative fields, then subtracting to obtain the portion present under both fields. This eliminates saturation from the TEY detection technique. Supplementary information Supplementary material 1 Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information is available for this paper at 10.1038/s41598-020-77650-6. Acknowledgements The authors thank Dr. Freeland from the Advanced Photon Source at Argonne National Laboratory for aid with the X-ray absorption spectroscopy. Use of the Advanced Photon Source at Argonne National Laboratories was supported by the US DOE under contract DE-AC02-06CH11357. The authors acknowledge funding from the Natural Sciences and Engineering Research Council of Canada (RGPIN-2018-05012) and the Canada Foundation for Innovation. Author contributions J.v.L. and C.A.R. conceived the experiments, M.S., V.K.P. and J.v.L. conducted the experiment(s), M.S. analysed the results. M.S. wrote the manuscript. J.v.L., V.K.P., and C.A.R. contributed to the manuscript. All authors reviewed the manuscript. Competing interests The authors declare no competing interests. ==== Refs References 1. Su D Dou S Wang G Single crystalline \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 nanocrystals exposed with different crystal planes for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ LiO}_{2}$$\end{document} LiO 2 batteries Sci. Rep. 2014 4 5767 10.1038/srep05767 25169737 2. Roth W The magnetic structure of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 J. Phys. Chem. Solids 1964 25 1 10 10.1016/0022-3697(64)90156-8 3. Roth, W. Magnetic properties of normal spinels with only A–A interactions. J. Phys. France25, 507 (1964). 4. Ichiyanagi Y Yamada S The size-dependent magnetic properties of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 nanoparticles Polyhedron 2005 24 2813 2816 10.1016/j.poly.2005.03.158 5. Tobia D Winkler E Zysler RD Granada M Troiani HE Size dependence of the magnetic properties of antiferromagnetic \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Cr}_{2}\text{O}_{3}$$\end{document} Cr 2 O 3 nanoparticles Phys. Rev. B 2008 78 104412 10.1103/PhysRevB.78.104412 6. Zhu H Synthesis and magnetic properties of antiferromagnetic \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 nanoparticles Physica B Condens. Matter 2008 403 3141 3145 10.1016/j.physb.2008.03.024 7. Li D Spin canting and spin-flop transition in antiferromagnetic \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Cr}_{2}\text{O}_{3}$$\end{document} Cr 2 O 3 nanocrystals J. Appl. Phys. 2009 106 053913 10.1063/1.3213100 8. Fontaíña-Troitiño N Room-temperature ferromagnetism in antiferromagnetic cobalt oxide nanooctahedra Nano Lett. 2014 14 640 647 10.1021/nl4038533 24467516 9. Sun H Ang HM Tadé MO Wang S \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 nanocrystals with predominantly exposed facets: Synthesis, environmental and energy applications J. Mater. Chem. A 2013 1 14427 14442 10.1039/c3ta12960h 10. Gao R Facet-dependent electrocatalytic performance of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 for rechargeable \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ LiO}_{2}$$\end{document} LiO 2 battery J. Phys. Chem. C 2015 119 4516 4523 10.1021/jp511363p 11. Tsukiyama K Takasaki M Oaki Y Imai H Evolution of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 nanocubes through stepwise oriented attachment Langmuir 2019 35 8025 8030 10.1021/acs.langmuir.9b00342 31145617 12. Yu, X.Y. et al. Facet-dependent electrochemical properties of Co3O4 nanocrystals toward heavy metal ions. Sci. Rep.3, 2886 (2013). 13. Toby BH Von Dreele RB Gsas-II: The genesis of a modern open-source all purpose crystallography software package J. Appl. Crystallogr. 2013 46 544 549 10.1107/S0021889813003531 14. Momma K Izumi F Vesta: A three-dimensional visualization system for electronic and structural analysis J. Appl. Crystallogr. 2008 41 653 658 10.1107/S0021889808012016 15. Lin F Influence of synthesis conditions on the surface passivation and electrochemical behavior of layered cathode materials J. Mater. Chem. A 2014 2 19833 19840 10.1039/C4TA04497E 16. Enriquez E Oxygen vacancy-tuned physical properties in perovskite thin films with multiple b-site valance states Sci. Rep. 2017 7 46184 10.1038/srep46184 28417954 17. Nakajima R Stöhr J Idzerda YU Electron-yield saturation effects in L-edge X-ray magnetic circular dichroism spectra of Fe Co, and Ni Phys. Rev. B 1999 59 6421 6429 10.1103/PhysRevB.59.6421 18. de Groot FMF Oxygen 1s X-ray-absorption edges of transition-metal oxides Phys. Rev. B 1989 40 5715 5723 10.1103/PhysRevB.40.5715 19. Cossee, P. Magnetic properties of cobalt in oxide lattices. J. Inorg. Nucl. Chem.8, 483–488 (1958). Proceedings International Symposium on the Chemistry of the Co-ordination Compounds. 20. Dutta P Seehra MS Thota S Kumar J A comparative study of the magnetic properties of bulk and nanocrystalline \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 J. Phys. Condens. Matter 2007 20 015218 10.1088/0953-8984/20/01/015218 21. Madelung O Rössler U Schulz M Landolt-Börnstein—Group III Condensed Matter—Semiconductors: Non-Tetrahedrally Bonded Binary Compounds II 2007 Berlin Springer 542 22. Makhlouf SA Magnetic properties of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 nanoparticles J. Magn. Magn. Mater. 2002 246 184 190 10.1016/S0304-8853(02)00050-1 23. Ueno T Enhanced orbital magnetic moments in magnetic heterostructures with interface perpendicular magnetic anisotropy Sci. Rep. 2015 5 14858 10.1038/srep14858 26456454 24. Chen J Selloni A Electronic states and magnetic structure at the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ Co}_{3}\text{O}_{4}$$\end{document} Co 3 O 4 (110) surface: A first-principles study Phys. Rev. B 2012 85 085306 10.1103/PhysRevB.85.085306 25. Díaz C Synthesis and magnetic properties of nanostructured metallic Co, Mn and Ni oxide materials obtained from solid-state metal-macromolecular complex precursors RSC Adv. 2017 7 27729 27736 10.1039/C7RA00782E 26. Wang G Hydrothermal synthesis and optical, magnetic, and supercapacitance properties of nanoporous cobalt oxide nanorods J. Phys. Chem. C 2009 113 4357 4361 10.1021/jp8106149 27. Popkov SI Size effects in the formation of an uncompensated ferromagnetic moment in nio nanoparticles J. Appl. Phys. 2019 126 103904 10.1063/1.5109054 28. Alves Oliveira A Core-shell behavior and exchange bias of fe-doped cuo nanoparticles AIP Adv. 2020 10 065009 10.1063/5.0008366 29. Nogués J Schuller IK Exchange bias J. Magn. Magn. Mater. 1999 192 203 232 10.1016/S0304-8853(98)00266-2 30. Winkler E Surface spin-glass freezing in interacting core–shell NiO nanoparticles Nanotechnology 2008 19 185702 10.1088/0957-4484/19/18/185702 21825698 31. Nadeem K Krenn H Traussing T Letofsky-Papst I Distinguishing magnetic blocking and surface spin-glass freezing in nickel ferrite nanoparticles J. Appl. Phys. 2011 109 013912 10.1088/0957-4484/18/45/455704 32. Shendruk T Desautels R Southern B Van Lierop J The effect of surface spin disorder on the magnetism of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma $$\end{document} γ -Fe\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document} 2 O\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3$$\end{document} 3 nanoparticle dispersions Nanotechnology 2007 18 455704 10.1088/0957-4484/18/45/455704 33. Desautels R Increased surface spin stability in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma $$\end{document} γ -Fe\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document} 2 O\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3$$\end{document} 3 nanoparticles with a Cu shell J. Phys. Condens. Matter Inst. Phys. J. 2012 24 146001 10.1088/0953-8984/24/14/146001 34. Alkadour B Mercer JI Whitehead JP van Lierop J Southern BW Surface vacancy mediated pinning of the magnetization in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma $$\end{document} γ -Fe\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document} 2 O\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_3$$\end{document} 3 nanoparticles: A micromagnetic simulation study Phys. Rev. B 2016 93 140411 10.1103/PhysRevB.93.140411 35. Anderson, P. W. Antiferromagnetism. theory of superexchange interaction. Phys. Rev.79, 350–356 (1950). 36. Kanamori J Superexchange interaction and symmetry properties of electron orbitals J. Phys. Chem. Solids 1959 10 87 98 10.1016/0022-3697(59)90061-7 37. Goodenough JB Loeb AL Theory of ionic ordering, crystal distortion, and magnetic exchange due to covalent forces in spinels Phys. Rev. 1955 98 391 408 10.1103/PhysRev.98.391 38. O’Grady K Anisotropy in antiferromagnets J. Appl. Phys. 2020 128 040901 10.1063/5.0006077 39. Chuang DS Ballentine CA O’Handley RC Surface and step magnetic anisotropy Phys. Rev. B 1994 49 15084 15095 10.1103/PhysRevB.49.15084 40. Puntes VF Krishnan KM Alivisatos AP Colloidal nanocrystal shape and size control: The case of cobalt Science 2001 291 2115 2117 10.1126/science.1057553 11251109 41. Ma R Tetrahedral CO(II) coordination in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\alpha } {{-}}$$\end{document} α - type cobalt hydroxide: Rietveld refinement and X-ray absorption spectroscopy Inorg. Chem. 2006 45 3964 3969 10.1021/ic052108r 16676955 42. Deng T Atomic-level energy storage mechanism of cobalt hydroxide electrode for pseudocapacitors Nat. Commun. 2017 8 15194 10.1038/ncomms15194 28480885 43. Rueden CT Image J2: ImageJ for the next generation of scientific image data BMC Bioinform. 2017 18 529 10.1186/s12859-017-1934-z