An interpretation for the components of 2p 3/2 core level x-ray photoelectron spectra of the cations in some inverse spinel oxides

In an effort to reconcile the various interpretations for the cation components of the 2p 3/2 observed in x-ray photoelectron spectroscopy (XPS) of several spinel oxide materials, the XPS spectra of both spinel alloy nanoparticles and crystalline thin films are compared. We observed that different components of the 2p 3/2 core level XPS spectra, of these inverse spinel thin films, are distinctly surface and bulk weighted, indicating surface-to-bulk core level shifts in the binding energies. Surface-to-bulk core level shifts in binding energies of Ni and Fe 2p 3/2 core levels of NiFe2O4 thin film are observed in angle-resolved XPS. The ratio between surface-weighted components and bulk-weighted components of the Ni and Fe core levels shows appreciable dependency on photoemission angle, with respect to surface normal. XPS showed that the ferrite nanoparticles Ni x Co1−x Fe2O4 (x = 0.2, 0.5, 0.8, 1) resemble the surface of the NiFe2O4 thin film. Surface-to-bulk core level shifts are also observed in CoFe2O4 and NiCo2O4 thin films but not as significantly as in NiFe2O4 thin film. Estimates of surface stoichiometry of some spinel oxide nanoparticles and thin films suggested that the apportionment between cationic species present could be farther from expectations for thin films as compared to what is seen with nanoparticles.

In an effort to reconcile the various interpretations for the cation components of the 2p 3/2 observed in x-ray photoelectron spectroscopy (XPS) of several spinel oxide materials, the XPS spectra of both spinel alloy nanoparticles and crystalline thin films are compared.We observed that different components of the 2p 3/2 core level XPS spectra, of these inverse spinel thin films, are distinctly surface and bulk weighted, indicating surface-to-bulk core level shifts in the binding energies.Surface-to-bulk core level shifts in binding energies of Ni and Fe 2p 3/2 core levels of NiFe 2 O 4 thin film are observed in angle-resolved XPS.The ratio between surface-weighted components and bulk-weighted components of the Ni and Fe core levels shows appreciable dependency on photoemission angle, with respect to surface normal.XPS showed that the ferrite nanoparticles Ni x Co 1−x Fe 2 O 4 (x = 0.2, 0.5, 0.8, 1) resemble the surface of the NiFe 2 O 4 thin film.Surface-to-bulk core level shifts are also observed in CoFe 2 O 4 and NiCo 2 O 4 thin films but not as significantly as in NiFe 2 O 4 thin film.Estimates of surface stoichiometry of some spinel oxide nanoparticles and thin films suggested that the apportionment between cationic species present could be farther from expectations for thin films as compared to what is seen with nanoparticles.

Introduction
Spinel oxide thin films and nanoparticles are among some of the most extensively studied oxide systems using various characterization techniques for several electric and magnetic properties [1][2][3][4][5].One of the frequently used characterization techniques for such systems is x-ray photoelectron spectroscopy (XPS)  which is performed mainly for the study of surface chemistry and surface electronic properties.Although there have been numerous XPS of several inverse spinel oxide materials, there has been no consensus on the interpretation for the cation components of the 2p  1.The spectrum contains three components: P 1 , P 2 , and S.
Although the feature labeled S (in figure 1) is consistently identified as the commonly observed core level satellite feature, in the XPS spectra of many transition metal oxides, there has been various interpretations of two component peaks of type P 1 and P 2 as shown in figure 1. Apparently, for the inverse spinel oxide systems of type AB 2 O 4 , there appear to be four interpretations that are frequently mentioned in the literature, when P 1 and P 2 type component XPS peaks (as shown in figure 1) of the A-type and B-type elements are assigned.One explanation put forth is that P 1 is the +2 cationic state and P 2 is the +3 cationic state of both A-type and B-type elements [6][7][8].Alternatively, P 1 is assigned to the +3 cationic state and P 2 is the +2 cationic state of both A-type and B-type elements [9][10][11][12].Then there is the assignment that P 1 is the +2 cationic state and P 2 is the +3 cationic state for one of the elements between A and B, but for the other element P 1 component corresponds to the +3 state while the P 2 component corresponds to the +2 state [13][14][15][16][17][18][19][20].By far the most common explanation that has been put forward has been that the P 1 XPS component is a result of the metal cation occupancy in the octahedral site while the P 2 XPS component is the result of occupancy of the metal cation in the tetrahedral site, both components with the same cationic states [13,[21][22][23][24][25][26][27][28][29][30][31][32].Clearly this represents a situation where there is no general agreement on the assignment of the component peaks seen in the XPS spectra of the transition metal constituents.
Complicating an assessment that leads to an ambiguous assignment of the different core level binding energy features seen in the XPS 2p core level spectra is that photoemission is a final state spectroscopy and contains contributions from both initial state and final state effects [33][34][35][36][37][38].If we just consider initial state effects, which mainly include the effects of chemical bonding, charge transfer and neighbor interactions on binding energy before the photoemission event, binding energies for different cationic states should have the order of A 0 < A 1+ < A 2+ < A 3+ …. [34,35].This would favor an assignment where P 1 is the +2 cationic state and P 2 is the +3 cationic state, in the transition metal 2p core level XPS spectra.In the photoemission, however, emission of the electron is always followed by screening of the core hole [33][34][35][36][37][38].Final state effects, which are mainly the result of screening of the core hole (called relaxation or rearrangement of charges) on binding energy after the photoemission event, must be included in the analysis and the inclusion decreases the initial state binding energies [34,[36][37][38].It is therefore possible that order for the measured core level binding energies of cationic species in the XPS may not strictly follow A 0 < A 1+ < A 2+ < A 3+ .....For instance, the order of the measured 2p 3/2 core level binding energies for multiple cationic states of cobalt could increase as Co 0 < Co 3+ < Co 2+ , as a result of final state effects, particularly if we note that the Co XPS core level binding energies for CoO and Co 2 O 3 might only differ within the margin of error so effectively that the binding energies for Co 2+ is approximately the same as Co 3+ [39].Such a picture makes the possibility more likely that P 1 is assigned to the +3 cationic state and P 2 is the +2 cationic state or that P 1 is the +2 cationic state and P 2 is the +3 cationic state for one of the elements between A and B, but for the other element P 1 component corresponds to the +3 state while the P 2 component corresponds to the +2 state.As it is known that for an inverse spinel oxide of type AB 2 O 4 , the cations B 3+ occupy tetrahedral sites and both cations A 2+ and B 3+ occupy octahedral sites, but practically speaking the tetrahedral and octahedral sites might contain multiple cationic states of both A and B elements.So, while in principle, all aforementioned assignments of the various cation component features might be possible, these different explanations cannot be reconciled.Importantly none of these explanations acknowledge that the surface or surface oxide might differ from the bulk and the resulting initial and final state effects might lead to different core level binding energies as a result of the surface being distinctly different from the bulk.
The surface of a solid material is chemically different from the bulk due to the lack of symmetry at the surface.The bonding environment of the atoms at the surface is different from the environment of the atoms in the bulk of the material, and such difference can cause shifts in core energy levels of the surface atoms compared to the bulk atoms [40].Using the angle resolved x-ray photoelectron spectroscopy (ARXPS), Co and Fe 2p 3/2 core levels of CoFe 2 O 4 thin films have been shown to have surface weighted and bulk weighted components [41] in x-ray photoelectron spectra.Furthermore, the ratio between the components of the Co and Fe 2p 3/2 core level spectra was found to depend on photoelectron emission angle (or photoemission angle) with respect to surface normal of the CoFe 2 O 4 thin films, showing in a compelling fashion that there are surface weighted and bulk weighted components [41] in the x-ray photoelectron spectra of core levels.That study provided the evidence for the surface-to-bulk core level shifts in the binding energies of the Co and Fe 2p 3/2 core levels in CoFe 2 O 4 thin films.This surface-to-bulk core level shift does not, per se, exclude a surface oxide that differs from the bulk, so may include initial state effects, not just the expected surface enthalpy difference combined with final state effects [37,40].This prior study of CoFe 2 O 4 thin films [41] does imply that the dependency of the ratio between surface weighted and bulk weighted components on the photoemission angle might be different for different spinel and inverse spinel oxide thin films.Furthermore, this prior work does not provide insight as to the role of the surface for a nanoparticle counterpart where the surface to volume ratio is significantly larger than for the spinel or inverse spinel oxide thin films.

Materials and methods
Epitaxial NiFe 2 O 4 (111), CoFe 2 O 4 (111), and NiCo 2 O 4 (111) thin films, of 5 nm, 5 nm and 10 nm thicknesses respectively, were grown on Al 2 O 3 (0001) substrates using pulsed laser deposition, as described elsewhere [4,41].For the inverse spinel oxides discussed here, there appears to be little variation in the XPS for the films of thicknesses of 5 nm or more [4,41].We have observed some deviations in the XPS, from the behavior seen with thicker films, in films of substantially less thickness than studied here (⩽2 nm) [4,41] but the XPS on various thicknesses of samples is beyond the scope of this work and the much thinner films were not studied here.The Ni x Co 1−x Fe 2 O 4 (x = 0, 0.2, 0.5, 0.8, 1) nanoparticles were synthesized using a microwave assisted process.Nickel nitrate (Ni(NO 3 ) 2 .6H 2 O), cobalt nitrate (Co(NO 3 ) 2 .6H 2 O), and iron acetate (Fe(CO 2 CH 3 ) 2 ) from Sigma Aldrich were used as the precursor materials for the synthesis of the nanoparticles.Stoichiometric ratios of precursor materials were dissolved in 40 ml of ethylene glycol followed by continuous stirring and heating at 65 • C for 3 h to create a homogenous solution.Next, the resulting homogeneous solution was kept in the CEM microwave reactor and irradiated for 30 min at 180 • C with 100 mW power.Distilled water was then added to the solution and kept overnight for precipitation.The precipitates were washed with distilled water and separated by centrifugation.The precipitates were then dried in an oven at 200 • C. The dried precipitates were crushed into fine powder and calcined at 600 • C for 5 h in a tube furnace.To analyze the particle size of nanoparticles scanning electron microscopy (SEM) was done on Helios NanoLab Dual Beam 660.First, 0.2 mg of nanoparticles were dispersed in 15 ml ethanol and sonicated for 10 min.Then a small amount of the dispersion was drop-cast on carbon tape using a dropper.The films of nanoparticles on carbon tape were analyzed on SEM.It was found that the cobalt ferrite nanoparticles follow a normal distribution of particle size i.e. particle diameter with mean diameter of 73 nm as shown in figure 2. But the nickel ferrite nanoparticles followed a bimodal distribution of particle diameter with mean of 21 nm and 69 nm (see figure S1 of the supplementary information).These particle sizes are in much the same range as prior studies (20-70 nm) of cobalt nickel ferrite nanoparticles [42].The Ni x Co 1−x Fe 2 O 4 (x = 0, 0.2, 0.5, 0.8, 1) nanoparticles crystallite sizes, determined from x-ray diffraction (see the supplementary materials figure S2), were 14.4 nm, 11.9 nm, 11.7 nm, 9.0 nm and 12.4 nm for x = 0, 0.2, 0.5, 0.8, 1 respectively.The crystallite size calculated from xray diffraction data are smaller than the particle size determined from the SEM, but as we know the crystallite size depends upon the coherently diffracted domain size which may often be smaller than the particle size.
The magnetic hysteresis loops were measured for the CoFe 2 O 4 and NiCo 2 O 4 nanoparticles in a super conducting quantum interference device (SQUID) magnetometer (MPMS XL SQUID magnetometer from Quantum Design), as a function of temperature, while the entire range of Ni x Co 1−x Fe 2 O 4 (x = 0, 0.2, 0.5, 0.8, 1) nanoparticles were also characterized by vibrating sample magnetometry (VSM).VSM magnetometry data for the five samples (CoFe 2 O 4 , NiFe 2 O 4 , Ni 0.2 Co 0.8 Fe 2 O 4 , Ni 0.5 Co 0.5 Fe 2 O 4 , Ni0 0.8 Co 0.2 Fe 2 O 4 ) were collected using a Quantum Design VersaLab 3 Tesla cryogenfree VSM at 300 K.In the ferrimagnetic systems at zero magnetic field, spin moments with larger and smaller magnitudes are oppositely aligned.When the magnetic field is applied, changing from zero field, the rotation of the oppositely aligned spin moments results in a characteristic feature for a ferrimagnet at higher magnetic fields, much higher than the apparent coercive magnetic field.These effects that broaden the magnetic hysteresis to higher fields was observed in the hysteresis measurements of the nanoparticles, and the feature is much noticeable especially in the hysteresis of CoFe 2 O 4 nanoparticles (see figure S3 in the supplementary information).For both CoFe 2 O 4 and NiCo 2 O 4 nanoparticles, the moments in general saturated at lower field and saturation magnetization was larger at lower temperatures.The measured remanent magnetization of the Ni x Co 1−x Fe 2 O 4 (x = 0, 0.2, 0.5, 0.8, 1) nanoparticles decreases with increasing nickel concentrations, as was seen in prior cobalt nickel ferrite nanoparticles studies [42].
The x-ray photooemission spectra for the CoFe 2 O 4 thin films and CoFe 2 O 4 nanoparticles were acquired using a SPECS x-ray Mg Kα anode (hv = 1253.6eV) source The limited electron mean free path ensures that by changing the photoelectron emission angle, the ARXPS can be used to distinguish components in the XPS spectra that have greater surface weight, i.e. larger photoemission angles with respect to the surface normal of the sample result in core level photoelectron spectra with greater surface weights (see figure S4 of the supplementary information) [43].

XPS of selected Ni and Co spinel ferrites: thin films versus nanoparticles
Figure 3 shows the ARXPS of a NiFe 2 O 4 (111) thin film.The Ni 2p 3/2 core level spectra were fitted for three components: P 1 at 854.7 eV, P 2 at 856.5 eV, and S (satellite) at 861.6 eV.The main core level peak of the Ni 2p 3/2 , at normal photoemission, was observed to be at a binding energy of 855.2 eV which is in excellent agreement with the value reported elsewhere [39].The Fe 2p 3/2 core level spectra are fitted for two components: P 1 at 710.5 eV and P 2 at 712.3 eV.The main Fe 2p 3/2 core level spectral feature, at normal emission, was observed to be at 711.5 eV, which is found to be higher than the value found elsewhere [44].For the component fits to the ARXPS spectra, the widths of all the component peaks and peak positions of the components P 1 and P 2 are fixed at all angles so that any changes in the spectral features, as photoemission angle changes, could be ascribed to surface weighted or bulk weighted feature.As the photoemission angle increases, we observed that intensity of P 1 component relative to P 2 of both Ni and Fe core levels increases, showing that P 1 (P 2 ) is a surface (bulk) weighted component of the core levels.The binding energy shift between components P 1 and P 2 , which is 1.8 eV, is therefore an effective surface-to-bulk core level shift.Figure 4 shows that the P 1 /P 2 intensity ratios for both Ni and Fe core levels increase, as photoemission angle increases, and become maximum at the largest take-off angle, with respect to the surface normal, measured here (50 • ).The XPS characterization at larger (smaller) photoemission angles in this work gives the more surface (bulk) weighted character (see figure S4 of the supplementary information).The consequence of the effective surface-to-bulk core level shifts observed for the NiFe 2 O 4 thin film is that the main peaks of the XPS spectra of both Ni and Fe core levels at 50 • , when compared with the main XPS spectral peaks at 0 • , are shifted toward the side of surface weighted component, which in this case is lower binding energy side (see figure S5 of the supplementary information).
ARXPS is most meaningful where the change in the photoelectron emission angle is representative of a change of the electron mean free path component along the surface normal.For the systems like nanoparticles, the ARXPS technique is physically less meaningful because of inhomogeneities between different particles and more random multiple surface directions of different particles.A comparative study between thin films and their nanoparticle counterparts is best done by comparing XPS spectra with the emission angle along the surface normal as this contains both surface and more bulk components, and for the nanoparticles, might be considered more of an average of the chemical or electronic properties representative of nanoparticles.that main 2p 3/2 core level peaks for both Fe and Ni of all nanoparticles are shifted to lower binding energies compared to the corresponding peaks of NiFe 2 O 4 (111) thin film.These core level shifts suggests that the nanoparticles, on average, chemically and electronically behave similar to the atoms at the surface of NiFe 2 O 4 thin film, i.e. dominated by the more surface transition metal 2p 3/2 XPS P 1 components, so the nanoparticle on average should, electronically, be dominated by the surface and thus more resemble the surface of their thin film counterparts.In another words, the Ni x Co 1−x Fe 2 O 4 (x = 0.2, 0.5, 0.8, 1) type nanoparticles are chemically or electronically different from the bulk of NiFe 2 O 4 thin films.In the comparison between the thin films and nanoparticles, as shown in figure 5, the P 1 components (see figure 3) of the 2p 3/2 core level spectra, for the nanoparticles, exhibit a much larger intensity than the intensity of P 2 component of the Fe and Ni p 3/2 core level spectra.This relative enhancement of the P 1 components, of the nanoparticle 2p 3/2 core level spectra, is evident in the fitting of the XPS core level spectra components for the NiFe 2 O 4 thin films and nanoparticle specimens (see figure S6 of the supplementary information).
Similar ARXPS measurements were carried out on CoFe 2 O 4 (111) thin films.Figure 6 shows some representative ARXPS Co and Fe 2p 3/2 core level spectra from a CoFe 2 O 4 (111) thin film.The Co 2p 3/2 core level spectra contain three component peaks: P 1 at 779.4 eV, P 2 at 781.4 eV and S (satellite) at 785.9 eV binding energies.At normal emission (0 • ), the main peak position of the Co 2p 3/2 core level was found to be at a binding energy of 779.7 eV which is in excellent agreement with the 779.9 ± 0.2 eV reported elsewhere [39].The Fe 2p 3/2 core level spectra contains two component peaks: P 1 at binding energies of 709.6 eV and P 2 at 711.6 eV.The main peak position of Fe 2p 3/2 core level at normal emission was found to be at a binding energy of 710.3 eV which is in good agreement with the binding energy value of 710.6 ± 0.2 eV, as reported elsewhere [43].As the photoelectron emission angle increases with respect to the normal, the ratios of the XPS core level component peaks change: the P 1 /P 2 intensity ratio values for the Co (Fe) 2p 3/2 core level decreased (increased) when the photoemission angle was increased, as shown in figures 7 (a) and (b), indicating that P 1 (P 2 ) of Co (Fe) core level is a bulk weighted component, and P 2 (P 1 ) of Co (Fe) core level is a surface weighted component.In simple terms, the surface must again be somewhat distinct from the bulk in CoFe 2 O 4 (111) thin films.
Although an intensity ratio (P 1 /P 2 ) dependency was observed for the P 1 and P 2 components of the Co and Fe 2p 3/2 core level spectra of CoFe 2 O 4 (111) thin films, as a function of photoemission angle as illustrated in figures 7(a) and (b), the intensity ratio changes from normal emission to a large take-off angle, 0 • to 50 • , were not as large as compared to the ratio change observed for NiFe 2 O 4 thin films.The change in the P 1 /P 2 component ratio for CoFe 2 O 4 (111) is not large enough to appreciably show shifts of the spectra taken at 50 • relative to the spectra taken at 0 • , as is evident in figures 7(c) and (d).Unlike the ARXPS of the NiFe 2 O 4 (111) thin film, no noticeable shifts of the core level spectra taken at 50 • towards the surface component side of the spectra taken at 0 • was observed.When the 2p 3/2 core level XPS spectra of the CoFe 2 O 4 (111) thin film are compared to those of CoFe 2 O 4 nanoparticles, as shown in figures 7(e) and (f), the CoFe 2 O 4 thin film and nanoparticles imply that the nanoparticles have XPS signatures more like the bulk of CoFe 2 O 4 thin film because the nanoparticle spectra are apparently shifted to now the P 1 bulk component of the Co 2p 3/2 spectra.These signatures are, however, not overly compelling.
Table 1 lists the estimated core level binding energies and the stoichiometry of all oxide thin films and nanoparticles studied here.From the apparent stoichiometry, estimated from the core level intensities after corrections for cross section and the transmission of the electronic energy analyzer, several general features are observed: the surface is Ni rich compared to Co in the Ni x Co 1−x Fe 2 O 4 type ferrite nanoparticles, and the surface concentration ratio of Fe to the other cationic species together in the system is higher in ferrite nanoparticles than in thin films.Roughly, the estimated stoichiometry, from table 1, suggests that the ferrite nanoparticles have the concentration ratio of Fe to all other cationic species closer to the expected ratio of 2:1 as compared to the thin film ferrites.Ideally, ratio of concentration of Fe relative to other cationic species should be 2:1 in these ferrites, but the ratio is much less than one in the stoichiometry for thin film ferrites, as estimated from the experimental intensities obtained from XPS.
In order to provide a comparison between ferrite thin films and another spinel oxide thin film of similar structure, we present the study of ARXPS on NiCo 2 O 4 thin film in the following section.

ARXPS of the NiCo 2 O 4 thin film
For the purpose of comparison, a closer examination was given to the NiCo 2 O 4 crystalline thin films.Like the ferrites under the study, NiCo 2 O 4 also exists in the inverse spinel structure.Figure 8 shows the 2p 3/2 core level XPS spectra of Ni and Co for NiCo 2 O 4 (111).The Ni 2p 3/2 core level spectra contain three component peaks: P 1 at 854.4 eV, P 2 at 856.2 eV and the satellite feature S at 861.1 eV.At normal emission (0 • ), the main peak position of the Ni 2p 3/2 core level is found to be at a binding energy of 855.0 eV.The Co 2p 3/2 level contains three components too: P 1 at 779.7 eV, P 2 at 781.5 eV, and S (satellite) at 785.5 eV.The main peak position of the Co 2p 3/2 core level at normal emission is found to be at a binding energy of 780.1 eV.The main peak positions of both Ni and Co 2p 3/2 core levels are comparable but somewhat larger than the values of core level binding energies reported elsewhere [45].
An effective surface-to-bulk core level shifts was observed for the NiCo 2 O 4 thin films.This is evident as there are higher values for the component ratios P 1 /P 2 of both Ni and Co 2p 3/2 core levels at higher photoemission angles, on average, as shown in figure 9(a) and (b).The component P 1 (P 2 ) is a surface (bulk) component of the core levels and the surface-to-bulk core level shifts was observed to be 1.8 eV, which is similar to the shifts observed for the NiFe 2 O 4 thin films discussed above.The NiCo 2 O 4 thin film surface was found to be Ni rich, as indicated by the estimated surface    stoichiometry of Ni Although clear dependencies of the component intensity ratios (P 1 /P 2 ), for each cation of three inverse spinel oxide thin films, were observed on the photoemission angle, indicating surface-to-bulk core level shifts, the degree of the dependency was not the same for each of the thin films.In order to better understand the nature of the thin film surfaces, adventitious carbon (AdC) 1s and oxygen core level O 1s were analyzed for all three thin films.The analyses showed that NiFe 2 O 4 thin film surface had lower number of adsorbed contamination species compared to the CoFe 2 O 4 and NiCo 2 O 4 thin film surfaces (see figures S7-S9 of the supplementary information).It is possible that relatively lower degree of dependencies of the component intensity ratios (P 1 /P 2 ) on photoemission angle for CoFe 2 O 4 and NiCo 2 O 4 thin films could be due to larger number of contamination species at the surfaces.
In the XPS spectra of all the inverse spinel oxides studied here, the satellite feature intensity for the core level spectra of nominally second transition metal cation B 3+ is significantly suppressed.For instance, the core level spectra of Fe in ferrites and Co in NiCo 2 O 4 do not have an appreciable satellite intensity, although some non-zero intensity for the Co satellite component was inferred from the data for the NiCo 2 O 4 thin film.Satellite features are ascribed to the presence of two-hole final state [36], one of the multiple final states of the system in photoemission, which effectively increases nuclear charge and decreases the kinetic energy of the photoelectron, thereby leading to the increased binding energy component for the satellite feature [36].Suppression of the satellite intensity indicates that the two-hole final state is suppressed for +3 cationic species.The XPS spectra show suppressed satellite intensities for the Fe 2p 3/2 core levels of NiFe 2 O 4 and CoFe 2 O 4 where the Fe cation is nominally 3+ (3d 5 ) and for the Co 2p 3/2 core levels of NiCo 2 O 4 where the Co cation is nominally 3+ (3d 6 ) [3].We found especially for Co in NiCo 2 O 4 that annealing the system at certain higher temperature could enhance the intensity for satellite components (see figure S10 of the supplementary information), suggesting that two-hole final state could be activated by annealing.

Conclusions
We observed, using ARXPS, that a surface and a bulk weighted components can be distinguished in the 2p 3/2 core level spectra of the various cations in some inverse spinel oxide thin films.That some 2p 3/2 core level spectral components are more surface or bulk weighted lead to what may be considered as an effective surface-to-bulk core level shifts in the transition metal 2p 3/2 core level spectra from these inverse spinel oxide thin films.In the ARXPS of NiFe 2 O 4 thin film, an appreciable dependence of the core level surface and bulk weighted component ratios, on the photoemission angle, indicates that the surface of NiFe 2 O 4 thin film is chemically different from the bulk.The nanoparticles of type Ni x Co 1−x Fe 2 O 4 (x = 0.2, 0.5, 0.8, 1) may also chemically be similar to oxides seen at the surface of NiFe 2 O 4 thin films.From ARXPS, the dependencies of the 2p 3/2 core level component ratios on photoemission angle for CoFe 2 O 4 and NiCo 2 O 4 thin films were not observed to be as appreciable as was observed for NiFe 2 O 4 thin film, possibly due to higher contaminant species present at the surface of CoFe 2 O 4 and NiCo 2 O 4 thin films.In addition, the surfaces could be richer in A-type atoms in the inverse spinel oxide thin films of type AB 2 O 4 , and proportion between the cationic species in the thin film systems could be farther from the ideal proportion, compared to the inverse spinel oxide nanoparticle systems.The deviations from the expected composition, as seen in XPS, are because the surface composition differs from the bulk, as is so often the case with oxides [46][47][48].This is consistent with the fact that some of the various transition metal 2p 3/2 core level components have surface weight.The overall stoichiometry of the inverse spinel oxide films is as expected, as is the structure of the bulk of the film.There can be now no guarantee that the surface oxidation state is the same as the bulk.

Figure 1 .
Figure 1.The XPS spectrum of the Ni 2p 3/2 core level of a 10 nm thick NiCo 2 O 4 (111) thin film.The raw spectrum (red circles) is fitted (green solid line) for three component peaks (blue solid lines): P 1 , P 2 , and S (satellite) peaks.

Figure 2 .
Figure 2. (a) The scanning electron microscopy (SEM) images of CoFe 2 O 4 nanoparticles.(b) The distribution of the CoFe 2 O 4 nanoparticles sizes, for the nanoparticles studied here.

Figure 3 .
Figure 3.The ARXPS of a NiFe 2 O 4 (111) thin film.Shown are the XPS spectra of (a) Ni 2p 3/2 and (b) Fe 2p 3/2 core levels at six different photoemission angles with respect to the surface normal of the sample.Red circles, green solid lines, and blue solid lines represent raw spectra, fit of the data, and components of the spectra respectively.Spectra at higher angles are normalized to the normal emission (0 • ) photolectron spectra.

Figure 4 .
Figure 4.The dependency of the ratio between P 1 and P 2 components of the Ni 2p 3/2 core level (blue upright triangles) and the Fe 2p 3/2 core level (red inverted triangles) spectra on photoemission angle supporting a surface-to-bulk core level shifts in NiFe 2 O 4 (111) thin films.

Figure 5
Figure 5 shows the comparison of normal emission XPS (photoemission angle is 0 • ) core level spectra of some ferrite nanoparticles of type Ni x Co 1−x Fe 2 O 4 (x = 0.2, 0.5, 0.8, 1) with that of NiFe 2 O 4 (111) thin film.This comparison shows

Figure 5 .
Figure 5.A comparison of normal emission XPS spectra of the (a) Fe and (b) Ni 2p 3/2 core levels in some spinel oxide nanoparticles with that observed for the NiFe 2 O 4 (111) thin film.The blue spectra are the XPS core level spectra for (i) NiFe 2 O 4 , (ii) Ni 0.2 Co 0.8 Fe 2 O 4 , (iii) Ni 0.5 Co 0.5 Fe 2 O 4 , and (iv) Ni 0.8 Co 0.2 Fe 2 O 4 nanoparticles, and red spectra are the normal emission XPS core level spectra for a NiFe 2 O 4 thin film.

Figure 6 .
Figure 6.Some representative XPS core level spectra of the (a) Co 2p 3/2 and (b) Fe 2p 3/2 of a CoFe 2 O 4 (111) thin film for photoemission angles of 0 • and 50 • with respect to the surface normal are shown.In the figure, red circles, green solid lines, blue solid lines, and black solid lines are raw data, fitted lines, component peaks, and background levels respectively.

Figure 7 .
Figure 7.The P 1 /P 2 component intensity ratios for the (a) Co and (b) Fe 2p 3/2 core level as a function of photoemission angle.The comparison of the (c) Co and (d) Fe 2p 3/2 core level spectra of a CoFe 2 O 4 (111) thin film (TF) between small and large, 0 • and 50 • , photoemission angles.The comparison of (c) Co and (d) Fe 2p 3/2 core level spectra between a CoFe 2 O 4 (111) thin film and the corresponding nanoparticles (NP).In (e) and (f) red and blue spectra are for CoFe 2 O 4 thin film and nanoparticles respectively.

Figure 8 .
Figure 8.The angle resolved x-ray photoelectron spectroscopy of a 10 nm thick NiCo 2 O 4 (111) thin film.(a) The Ni 2p 3/2 and (b) Co 2p 3/2 XPS core level spectra taken at photoemission angles of 0 • and 50 • with respect to the surface normal of the sample.

Figure 9 .
Figure 9.A summary of the effective surface-to-bulk core level shifts in NiCo 2 O 4 (111) thin film.(a) Dependency of the ratio between P 1 and P 2 of the (a) Ni and (b) Co 2p 3/2 core levels on photoemission angle.(c) Ni and (c) Co 2p 3/2 XPS spectra at 40 • photoemission angle, compared to the corresponding XPS spectra taken at 0 • photoemission angle.
3/2 XPS spectra for the inverse spinel oxides of type AB 2 O 4 (examples NiCo 2 O 4 , CoFe 2 O 4 , NiFe 2 O 4 , etc).A typical XPS spectrum of Ni 2p 3/2 core level in NiCo 2 O 4 , one of the inverse spinel oxides, is shown in the figure

Table 1 .
The estimated surface stoichiometry and core level binding energies of the thin film (TF) and nanoparticles (NP) samples under the study.
1.3 Co 1.8 O 3.8 .NiCo 2 O 4 thin film, similar to the ferrite thin films reported here, is observed to have surface-to-bulk core level shifts and, on comparing NiCo 2 O 4 to AB 2 O 4, the 'A' type cation is observed to be rich at surface, which is again observed for ferrite thin films under study here.