Nickel-doped Ceria Nanoparticles: the Effect of Annealing on Room Temperature Ferromagnetism

Nickel-doped cerium dioxide nanoparticles exhibit room temperature ferromagnetism due to high oxygen mobility within the doped CeO2 lattice. CeO2 is an excellent doping matrix as it can lose oxygen whilst retaining its structure. This leads to increased oxygen mobility within the fluorite CeO2 lattice, leading to the formation of Ce 3+ and Ce 4+ species and hence doped ceria shows a high propensity for numerous catalytic processes. Magnetic ceria are important in several applications from magnetic data storage devices to magnetically recoverable catalysts. We investigate the effect doping nickel into a CeO2 lattice has on the room temperature ferromagnetism in monodisperse cerium dioxide nanoparticles synthesised by the thermal decomposition of cerium(III) and nickel(II) oleate metal organic precursors before and after annealing. The composition of nanoparticles pre-and post-anneal were analysed using: TEM (transmission electron microscopy), XPS (X-ray photoelectron spectroscopy), EDS (energy-dispersive X-ray spectroscopy) and XRD (X-ray diffraction). Optical and magnetic properties were also studied using UV/Visible spectroscopy and SQUID (superconducting interference device) magnetometry respectively.

Cerium(IV) oxide (ceria) and metal-doped ceria nanoparticles have been developed for their role in catalysis [30][31][32][33][34][35][36].One of the features that makes ceria attractive for catalysis is its ability to release/uptake oxygen from its lattice without losing its structural integrity [37].Cerium(IV) oxide has the fluorite structure, in which the oxygen are packed along the same plane.This allows high oxygen mobility through the lattice as the number of oxygen vacancies to increase, with cerium atoms changing oxidation state from Ce 4+ to Ce 3+ increasing the propensity for redox chemistry.The ability for ceria to readily develop oxygen vacancies is key to its effectiveness as a dopant matrix [38].
Various transition and lanthanide metals have been used to dope ceria to enhance catalytic activity [31,39], performance in solid oxide fuel cells [40] and to develop magnetic properties [41,42].Transition metals, particularly Mn [43], Fe [44], Co [45] and Ni [46] have all been doped into ceria to affect room temperature ferromagnetism and form dilute magnetic semiconductor oxides.It has been hypothesized by several groups that the room temperature ferromagnetism is due to the creation of oxygen vacancies in the CeO2 lattice.Oxygen vacancies which generate exchange interactions between unpaired electron spins and changes in surface chemical states are both postulated to affect room temperature ferromagnetism [47][48][49].Oxygen vacancies also effect the band structure of the host CeO2, often dramatically affecting optical properties [50][51][52].Studies involving the doping of non-magnetic elements into CeO2 such as Ca 2+ [53] and Cr 3+ [48] support this as they also demonstrate room temperature ferromagnetism.
Thurber et al. demonstrated that room temperature ferromagnetism in Ni-doped CeO2 is not due to the formation/coagulation of ferromagnetic nickel within the material [42].This was determined by the absence of nickel/nickel oxide diffraction peaks in X-ray diffraction pattern (and observations of shifting of the Curie temperature with different Ni doping levels showing doping throughout the material), and the fact that ferromagnetic properties did not increase with high amounts of nickel doping.Pure phase CeO2 and heavily Ni-doped samples did not show ferromagnetism, indicating that a random impurity could be eliminated as the cause of the ferromagnetism as the magnetic moment would be identical throughout the doped/pure phase samples.
In this paper, we present the synthesis of nickel-doped cerium dioxide nanocrystals by the thermal decomposition of metal-oleate complexes in the presence of surfactants at high temperature; a route which has a high success rate in producing large quantities of highly monodisperse nanocrystals.The synthesis of Ni-doped CeO2 via this route has advantages over Fe, Co or Cr doped CeO2 due to the ease in which the solid precursor, nickel(II) oleate, can be handled and stored for long periods of time in comparison to viscid iron(III) oleate or air-sensitive cobalt(II) oleate.We investigated the effect of annealing in air on the composition, crystallinity and room-temperature ferromagnetic properties using superconducting quantum interference device (SQUID) magnetometry.Nanoparticles were characterised using transmission electron microscopy (TEM) , X-ray photoelectron spectroscopy (XPS) , UV/Visible spectroscopy (UV/Vis), X-ray diffraction (XRD) and energy dispersive X-ray spectroscopy (EDS).

Results and Discussion
The thermal decomposition of metal-oleate complexes has been shown to be an excellent method for the synthesis of large quantities of monodisperse metal oxide nanoparticles.The method involves the synthesis of the metal-oleate complex by a salt metathesis reaction between the metal chloride and sodium oleate, to give the metal-oleate and sodium chloride.Aside from the formation of sodium chloride, phase transfer in the two-phase solvent system (a mixture of water, ethanol and n-hexane) provides a major driving force.
Once isolated, cerium(III) and nickel(II) oleate complexes are gold and bright green waxy solids respectively, and are easier to manipulate than their iron(III) or cobalt(II) counterparts due to their higher viscosity.Ni-doped ceria nanoparticles were synthesised by decomposing different ratios of cerium(III) and nickel(II) oleate in the presence of a surfactant (oleic acid) and a high boiling point solvent (1-octadecene) at 320 °C.The ratios for different nickel doping levels are given in Table 1.Post-decomposition, nanoparticles were isolated as black sticky residues, several ethanol washes/centrifuge cycles were used to remove excess oleate-type species and 1-octadecene.Once washed, the nanoparticles were readily dispersible in organic solvents.At this stage, the samples were divided in two, with one half kept for analysis and the other half annealed at 450 °C in air for 12 h.The annealed powders were free flowing and different in color, with the most marked example being Sample 1 changing from dark brown to yellow.The composition of nanoparticles pre-and post-anneal were characterised using: TEM, XPS, EDS and XRD.Their optical properties were also analysed using UV/Visible spectroscopy.
TEM analysis demonstrated the effectiveness of the metal-oleate decomposition method for the synthesis of large quantities of monodisperse nanoparticles (Figure 1).Sample 1 (CeO2) showed spherical nanoparticles of 1.83 ± 0.47 nm in diameter with very little shape anisotropy (on annealing 7.41 ± 2.1 nm).However, on nickel doping, shape anisotropy increased, as did overall sample polydispersity.Sample 5 gave a mean size of 2.63 nm with a standard deviation of 1.18 nm (Figure 1).TEM and therefore EDS analysis was exacting due to the amount of carbon contamination on the TEM grids from the blanket coverage of oleic acid on nanoparticle surfaces/excess oleic acid in solution.Once annealed, high resolution images were taken, as was a more accurate EDS analysis with little carbon contamination (Figure 2).
On annealing, nanoparticle size increased.Sample 5 increased from 2.63 ± 1.18 nm to 7.11 ± 3.37 nm.Annealing had a particularly marked effect on Sample 2 (NiO) which showed a very large increase in particle size (3.77± 0.90 nm to 21.49 ± 5.28 nm) and crystallinity, supported by peak narrowing in its XRD pattern (Figure 4j).
Characterisation by TEM was supported by compositional analysis by XPS and EDS spectroscopy.Results from composition analysis by quantitative XPS are fully summarised in Table 2. EDS spectra were effective in identifying the elements present in pre-annealed samples, although a quantitative comparison of compositions was achieved by analysing XPS spectra.
EDS spectra showed the presence of Ni, Ce and O in the pre-annealed samples, but the high levels of surfactants present (oleic acid) introduced increased background interference.As evidenced in Figure 2a, c and d, this was solved by examining the annealed samples, as annealing entirely removed this background.XPS was used to determine the presence and composition of elements in the doped and undoped samples.Quantitative XPS results are listed in Table 2, with all samples showing the desired ratios of Ce to Ni in accordance with the amount of precursors used (Table 1).The survey scan revealed a degree of sodium in the samples and is a possible contaminant derived from sodium oleate used in the precursor synthesis.
High resolution Ce 3d XPS spectra showed spectra typical of a Ce(IV) species, specifically for cerium   XRD patterns were easily obtained on powder samples post-annealing and demonstrated the propensity of the ceria lattice to receive metal ion dopants without significant structural re-arrangement or collapse (Figure 4).All patterns (bar NiO, Sample 2) were shown to exhibit the halite structure of cerium(IV) oxide (COD: 4343161) [50,58], with Sample 2 showing the cubic structure of NiO (COD: 1010093).When nickel doping was increased beyond 8%, Figure 3b,c shows the emergence of the nickel oxide pattern.In Sample 3 (1:1 molar ratio of cerium and nickel precursors), it is evident that a mixed phase system exists, supported by TEM micrographs which show two distinct nanoparticle populations.On annealing, Sample 2 (NiO) exhibited narrowed peaks compared to those of the non-annealed sample (Figure S1) consistent with nanoparticle sintering and increased crystallised size seen by TEM micrographs (Figure 2c,d).The diffraction pattern for Sample 2 also revealed that even after annealing, the NiO was not phase-pure; indeed there was a degree of Ni present as shown by reflections at 44.7° and 51.7° assigned as the <111> and <200> of Ni metal (COD: 1512526).However, this was not seen by high resolution Ni2p XPS spectra of Sample 2, which showed solely NiO.This is indirect evidence of a Ni/NiO core/shell type system as XPS is surface sensitive, whereas XRD analyses the entire sample.
As expected, the presence of Ni metal decreases when comparing the non-annealed and annealed states in sample 2 (Figure S1).XRD patterns were further analysed for unit cell parameter and unit cell volume (Table S1) using QualX software with standard data taken from the Crystallography Open Database (COD).However, there was little correlation between the cell parameter values or the unit cell volumes with nickel doping.Reports have suggested that cell parameter decreases with nickel doping due to the smaller size of Ce 3+ and Ce 4+ compared to Ni, Ni + or Ni 2+ [42], or that nanoscale effects overtake this, and an expansion in unit cell parameters is reported with decreasing nanoparticle size due to Ce 4+ replacement with smaller Ce 3+ [59][60][61].
One of the marked effects of introducing oxygen vacancies into CeO2 by transition metal doping is the alteration of the band structure.This can be seen using UV/visible spectroscopy, specifically diffuse reflectance spectroscopy on powder samples.A sample of the results are summarised in Figure 5.
Figure 5 shows the change in diffuse reflectance optical spectra with increased nickel doping.Using these data, it is possible to determine the optical band-gap of the samples after applying the Kubelka-Munk function (Equation ( 1)): The band gap of Sample 1 (pure CeO2) was found to be 3.81 eV, in agreement with the literature value of 3.78 eV for cerium(IV) oxide [62].With increasing nickel doping, the band gap increases to 3.7, 3.5, 3.81 eV and 3.93 eV for 1%, 2%, 3% and 8% doping respectively.This is consistent with reports that show that ceria nanoparticles of decreasing size show an increase in bad gap energy [59].TEM image analysis post annealing showed that particle size decreased slightly on Ni-doping, with the exception of sample 2 (pure NiO) which was dramatically different to any of the samples containing cerium.Thurber et al. reported changes in band gap energy of nickel doped ceria systems as a combination of particle size effects and structural changes brought about by interstitial Ni incorporation, even at the 1% level [42].The authors also show a decrease in the band gap up to x ≤ 0.04 (Ce1−xNixO2) but an increase thereafter, and correlated their findings with lattice strain induced by nickel doping.Our samples did exhibit this trend, but only up to x ≤ 0.02 after which band gap energy values increased, possibly due to smaller particle sizes obtained from our syntheses.Along with changes in the optical properties common in materials of interest in spintronics, the magnetic properties of the Ni-doped CeO2 samples were investigated using SQUID magnetometry (Figure 6).Both pre-and post-annealed NiO samples (Sample 2) have a coercive field that increase from 14 Oe to 70 Oe post-annealing.The annealing process also impacts the saturation magnetisation with a 17% increase in saturation post-annealing, commensurate with the increase in crystallinity and particle size.It is noteworthy that the non-annealed CeO2 sample (Sample 1) showed no magnetism at room temperature, with the annealed sample giving a weak magnetic moment of 1.9 × 10 −3 emu/g.Sample 6, 5% Ni-CeO2, showed weak ferromagnetic behaviour, however there was an absence of a coercive field in both annealed and non-annealed samples, presumably due to the small size of the nanoparticles, and the absence of ripening in the doped samples compared with NiO (Sample 2).Curiously there is however a decrease in the magnetic saturation post-annealing, which is contrary to expectations, however previous studies [42] noted a linear increase of ferromagnetic moment in dopant concentrations below 4%, with a dramatic decrease thereafter.Therefore, annealing in the case of Sample 6 could be detrimental to the overall moment due to increased crystallinity of the annealed product.

Methods
Metal-oleate syntheses: Nickel(II) and cerium(III) oleate complexes were synthesized according to a modified procedure by Park et al. [63].Briefly, nickel(II) chloride hexahydrate (7.13 g, 30 mmol) was dissolved in deionized water (60 mL) and added to a 250 mL three-necked flask fitted with a condenser and charged with sodium oleate (18.23 g, 60 mmol), n-hexane (105 mL), deionized water (45 mL) and ethanol (60 mL).The mixture was refluxed under stirring (~70 °C) for 4 h, giving a dark green suspension in hexane.On cooling, the organic (hexane) layer was isolated and washed with deionized water (2 × 100 mL) and evaporated in vacuo to give solid green nickel(II) oleate.
Nanoparticle synthesis: Cerium(III) oleate, nickel(II) oleate or a mixture of both (see Table 1) was added to a three-necked 250 mL flask containing oleic acid (0.64 mL, 2 mmol) and 1-octadecene (30 mL).The mixture was then heated to 320 °C at a constant rate of 3.3 °C•min −1 under stirring and kept at 320 °C for 1 h, developing a vigorous reflux.On cooling, the black suspension was subjected to several ethanol washings/centrifuge cycles (10 min at 3000× g) until a black solid was isolated.
Half the product was then placed into a ceramic crucible before being annealed in air for 12 h at 450 °C at a heating rate of 10 °C•min −1 , yielding free flowing powders free of ligands.
The remainder of the product was re-dispersed in n-hexane (ca.20 mL) giving a colloidally stable black dispersion.

Instrumentation
TEM images were recorded using a FEI Tecnai G2 20 with a LaB6 source at an acceleration voltage of 200 kV.EDS spectra were taken with an Oxford XMax 80 TLE detector running AZTEC software.UV-Vis absorption spectra were recorded using a Shimadzu UV-1800 UV/Vis spectrometer (Japan) in the wavelength range 200-800 nm.X-ray photoelectron spectra (XPS) were recorded on a Kratos Axis Supra instrument (Kratos Analytical, Manchester, UK) using a monochromated Al Kα source.All spectra were recorded using a charge neutralizer to limit differential charging and subsequently calibrated to the main CxHy carbon peak at a binding energy of 284.8 eV.Survey scans were recorded at a pass energy of 160 eV and high resolution data at a pass energy of 20 eV.Data was fitted using CASA XPS with Shirley backgrounds.XRD diffraction patterns were acquired using a Bruker D8 Advance diffractometer (Germany), using a Cu Kα source (λ = 0.154018 nm) with a Gobel mirror optic and Soller slits on the detector side.Scans were acquired at a grazing incidence of 3° over a 2θ range of 20°-80° with 0.05° steps and 3 s per step.Samples for magnetometery were prepared in polycarbonate sample holders and weighed using a 6 d.p. microbalance before mounting them in a rigid brass tube specifically designed for SQUID magnetometry measurements.A Quantum Design MPMS SQUID-VSM (San Diego, CA, USA) was used to measure M vs. H curves at room temperature (300 K) with a field range of ±7T.

Conclusions
Ni-CeO2 nanoparticles were successfully synthesised using the thermal decomposition of metal-oleate precursors.A portion of the obtained nanoparticles were annealed at 450 °C leading to some sintering (in the case of NiO) and increased crystallinity.Nanoparticle structure and composition were characterised using: TEM, XPS, EDS and XRD, giving expected levels of nickel doping and structural deviation.Optical properties were analysed using reflectance spectroscopy, with even low levels of Ni doping (ca.1%) having a marked effect on overall reflectance.SQUID magnetometry was used to compare annealed and non-annealed samples, with an increase in ferromagnetic moment observed on annealing of the pure NiO sample and indeed the pure CeO2 sample.The 5% Ni-CeO2 sample demonstrated a decrease on annealing, perhaps due to the drop-off of ferromagnetism in Ni-CeO2 samples with nickel doping levels of above 4%.
Magnetic ceria are important in several applications from magnetic data storage devices to magnetically recoverable catalysts, and weak room temperature ferromagnetism induced by transition metal doping is a facile method to realise the aforementioned applications at the nanoscale.
(IV) oxide (CeO2) [54-56].Ce 4+ exhibits six peaks comprised of 3 pin-orbit doublets due to different Ce 4f occupancies post-excitation.The peak positions and doublet separation (Δ) for Sample 3 are as follows: peak positions: 880.6, 886.5, 896.3, 899.2, 905.8 and 914.7 eV, Δ = 18.5 eV).One of the difficulties in analysing Ni-doped cerium dioxide is that the most intense signals for Ni and Ce (Ni 2p and Ce 3d) have a degree of overlap in Figure 3c. Figure 3d is the Ni 2p high resolution scan from Sample 2 (pure phase NiO) and it is clear that a substantial portion of the Ni 2p1/2 is masked by the Ce 3d3/2 peak.The full Ni2p scan is evident in Figure 3d, which is indicative of NiO (peak positions: 853.5, 855.2, 860.7, 872.0 and 879.3 eV, Δ = 18.5 eV) [54,57].

Figure 3 .
Figure 3. (a) XPS survey scan of Sample 3 (50% Ni-CeO2) showing elements present; (b) Fitted Ce 3d high resolution XPS spectrum showing the 3 doublets that are indicative of Ce(IV) and indeed CeO2; (c) Ni 2p region of Sample 3, showing the overlapping Ce 3d region which masks the Ni 2p1/2 region; (d) is a high resolution scan of the Ni 2p region in Sample 2 (pure phase NiO).

Figure 5 .
Figure 5. Diffuse-reflectance spectra of nickel-doped CeO2 samples, with reflectance intensity decreasing markedly with increased nickel doping."QS" is the reflectance spectrum of blank quartz slides.

Table 1 .
Samples synthesized and used in this study with the amounts of starting reagents listed.

Table 2 .
Nanoparticle compositional analysis by quantitative XPS spectroscopy.