The electronic structure, surface properties, and in situ N 2 O decomposition of mechanochemically synthesised LaMnO 3 †

The use of mechanochemistry to prepare catalytic materials is of significant interest; it oﬀers an environmentally beneficial, solvent-free, route and produces highly complex structures of mixed amorphous and crystalline phases. This study reports on the eﬀect of milling atmosphere, either air or argon, on mechanochemically prepared LaMnO 3 and the catalytic performance towards N 2 O decomposition (deN 2 O). In this work, high energy resolution fluorescence detection (HERFD), X-ray absorption near edge structure (XANES), X-ray emission, and X-ray photoelectron spectroscopy (XPS) have been used to probe the electronic structural properties of the mechanochemically prepared materials. Moreover, in situ studies using near ambient pressure (NAP)-XPS, to follow the materials during catalysis, and high pressure energy dispersive EXAFS studies, to mimic the preparation conditions, have also been performed. The studies show that there are clear diﬀerences between the air and argon milled samples, with the most pronounced changes observed using NAP-XPS. The XPS results find increased levels of active adsorbed oxygen species, linked to the presence of surface oxide vacancies, for the sample prepared in argon. Furthermore, the argon milled LaMnO 3 shows improved catalytic activity towards deN 2 O at lower temperatures compared to the air milled and sol–gel synthesised LaMnO 3 . Assessing this improved catalytic behaviour during deN 2 O of argon milled LaMnO 3 by in situ NAP-XPS suggests increased interaction of N 2 O at room temperature within the O 1s region. This study further demonstrates the complexity of mechanochemically prepared materials and through careful choice of characterisation methods how their properties can be understood.


Introduction
Developing catalytic processes that are both economically viable and sustainable represents a significant challenge. To meet this ambition, the development of new technologies can no longer rely on iterative trial and error approaches and must instead be design-led using sophisticated characterisation methods. 1 Synchrotron techniques, such as X-ray absorption spectroscopy (XAS), have multiple benefits, which aid the in-depth understanding of chemically important yet complex systems. [2][3][4] Previously, we have reported the advantages of XAS for studying the preparation of metal oxides prepared through mechanochemistry; this solvent-free synthesis route produces less waste than traditional methods and prepares materials with enhanced catalytic properties. 5 Our recent study focussed on the mechanochemical synthesis of LaMnO 3 from its single oxide precursors. During this process there was a significant amount of amorphous material produced at different stages of milling. Traditional characterisation methods, such as X-ray diffraction (XRD), provided an incomplete understanding of the underlying chemistry. Using XAS, which provides structural insights on all length scales, allowed us to learn more about the principal chemical steps within the milling process. Furthermore, these studies also increased the understanding of how these materials promote the catalytic decomposition of the environmental pollutant N 2 O (deN 2 O).
However, even after performing conventional XAS at the Mn K-edge information such as the local charge density and electronic configuration still eluded us. In this work we discuss the use of further advanced characterisation techniques, such as high energy resolution fluorescence detection (HERFD) to improve the XANES spectral resolution, 6,7 and X-ray emission spectroscopy (XES) to attain information on electronic structure and oxidation state. 8,9 Furthermore, we endeavoured to mimic the environmental conditions experienced during planetary ball milling by performing an in situ XAS study at high pressure in a diamond anvil cell (DAC). This was performed in conjunction with in situ vibrational ball milling by XRD. These new structural insights have then been coupled with in situ near ambient pressure X-ray photoelectron spectroscopy (NAP-XPS) to further understand how ball milled materials have improved catalytic activity at lower temperatures during deN 2 O.
By combining these local structural and electronic techniques, along with surface sensitive characterisation, it can be a powerful tool for understanding how structural properties, induced by the syntheitic route, can affect catalysis. Herein, we report an in-depth analysis during the mechanochemically synthesised LaMnO 3 by XES and HERFD, not previously reported. Furthermore, we can now link these structural properties of the final material to the enhancement of catalytic activity for the in situ deN 2 O combined with NAP-XPS.

a. LaMnO 3 sample preparation
High energy planetary ball milling was performed at room temperature in a 4-station Fritsch Pulverisette 5 Planetary Ball Mill. ZrO 2 vessels were prepared with precursors Mn 2 O 3 and La 2 O 3 at the correct proportions to synthesise stoichiometric LaMnO 3 , along with 5 mm ZrO 2 spheres at a sphere : powder ratio of 10 : 1. Milling was then conducted up to 4 hours at 400 rpm in sessions of 20 min, in both atmospheric and inert environments. For the sol-gel synthesis stoichiometric amounts of Mn(NO 3 ) 2 Á4H 2 O, and La(NO 3 ) 3 Á6H 2 O were dissolved in deionised water and added to a gel prepared by mixing equimolecular amounts of citric acid (99.5%) and ethylene glycol (99.5%) as a polydentate ligand. The excess water was slowly removed on a hot plate until a viscous liquid was obtained. Subsequently, the final slurry was slowly heated in air (1 1C min À1 ) from room temperature up to 700 1C and kept at this temperature for 4 h. These conditions are essential to obtain a crystalline material.

b. Characterisation
XES/HERFD measurements were performed on the scanning branch of the I20 beamline at the Diamond Light Source, Didcot, UK. Measurements were taken using a Si(111) fourbounce monochromator using a high-resolution X-ray emission spectrometer. 10 For this experiment, the spectrometer was equipped with three Ge(440) analysers to allow the Kb spectra to be measured. 10 Mn K-edge XES spectra were fitted using three Voigt functions which represent the Kb 1,3 , Kb x and Kb 0 peaks. 11 Prior to the fitting, spectra were normalized, and linear background was subtracted. During the fitting, peak splittings were kept constant at theoretical values; 12 (i) splitting between Kb 1,3 and Kb 0 peaks E 14-16 eV and (ii) Kb x was lower than main, Kb 1,3 , peak by B3 eV. The same fitting procedures were applied on spectra taken for Mn standard materials for comparison purposes. HERFD were measured by scanning the incident energy and detecting the fluorescence intensity at the maximum of the Mn Kb 1,3 emission line with time resolution of B20 min per spectrum with 3 spectra collected. Data processing, background submission and normalisation was performed using Athena package. In situ XRD ball milling was performed at the ID15A beamline at the European Synchrotron Radiation Facility using a modified MM200 Retsch mill reported by Halasz et al. 13,14 The set-up comprised of a PMMA milling jar containing one 5 mm WC milling media with Mn 2 O 3 and La 2 O 3 precursor powders. The mill operated at 30 Hz for B19 h. The in situ XRD was collected at 69 keV, l = 0.17971 Å with a beam size of 300 Â 300 mm 2 . A Dectris Pilatus X area detector was used with an exposure of 5s and a readout time of 120 s. The incident wavelength and detector distance (430 mm) were calibrated using a NIST CeO 2 . The Raw data frames were integrated using the internal ESRF Matlab.Eva 5.0.0.22/PDF-4+ 2020 software package was used for phase identification and data plotting. In situ high pressure EDE-XAFS were performed at I20-EDE, the energy dispersive beamline at the Diamond Light Source, Didcot, UK. Energy-dispersive EXAFS was performed at the Mn K-edge (6539 eV) in transmission mode using a Si(111) polychromator. A diamond anvil cell (DAC) was fitted with 0.5 mm thick diamonds in order to reduce the diamonds absorption and enable XAS measurements at the Mn K-edge. The singlecrystal diamond mini-anvils were mounted on fully perforated diamonds. Pressure was measured by the ruby fluorescence method. 15 The dispersive X-ray beam was focussed on the sample in the DAC with a spot size of 0.05 mm diameter. Diamond single crystal reflections were deglitched following a similar procedure to Hong et al. 16 Spectra were recorded every 2 GPa of pressure. XRD was collected on a Bruker AXS D8 diffractometer with Cu Ka radiation over a range of 2y = 10-1301 with 0.0441 step size (Johnson Matthey, Sonning Common, UK). Phase identification was performed using Bruker-AXS Diffrac Eva V4.2 with Rietveld refinement performed using Bruker-AXS Topas 4.2. XAS measurements were performed at the B18 Beamline, Diamond Light Source. XAS measurements at the Mn K-edge (6539 eV) and La L 3 -edge (5483 eV) were performed in transmission mode using QEXAFS setup with fast scanning Si(111) double crystal monochromator. An appropriate foil was placed between I t and I ref and measured concurrently with XAS spectra acquisition. At the Mn K-edge XAS spectra was acquired with a time resolution of 20 min per spectrum (k max = 14) averaged over 3 scans. At the La L 3 -edge XAS spectra were acquired with a time resolution of 5 min per spectrum (k max = 10) averaged over 3 scans. The catalytic reaction was performed on 3 samples; mechanochemically prepared LaMnO 3 , milled under air and argon gas atmospheres for 4 h, and compared to sol-gel synthesised LaMnO 3 . NAP-XPS was performed at the beamline B07-C (VerSoX) at the Diamond Light Source. Spectra were recorded at the La 3d, Mn 2p, O 1s, C 1s and Mn 3s levels at 834.5-855 eV, 641-655 eV, 529-532 eV, 284-289 eV and 82-89 eV binding energies, respectively, during the deN 2 O. Samples were pressed into 8 mm pellets containing 50 mg of LaMnO 3 . The reaction was performed under 0.5% N 2 O in He, with scans performed at room temperature in vacuum, room temperature with 10 mbar of N 2 O and then subsequently at 400 1C and 600 1C (still under 10 mbar of N 2 O). The measurements were performed at specific incident energies for each XPS region to ensure the same kinetic energy/depth of the emitted electron (B350 eV). To achieve a variable depth profile the Mn 3s region was recorded with an incident energy of 1200 eV. Each XPS region was aligned to a reference C 1s peak at 284.8 eV 19 to compensate for charging effects. Peak positions and FWHM determined after using Shirley background subtraction on CasaXPS, with all residual standard deviations of the curve fitting reported to be under 1.

Results and discussion
3.1 Understanding the mechanochemical synthesis of LaMnO 3 3.1.1 HERFD-XANES. XAS benefits from analysing both the amorphous and crystalline content, as it does not rely on periodic ordering, to give a combined bulk structure. Performing XAS at both the Mn K-edge and La L 3 -edge on ex situ time-slices during the mechanochemical synthesis of LaMnO 3 allowed for further insights into the mechanochemical synthesis of LaMnO 3 , which suggested that initially La dispersed over Mn 2 O 3 as soon as 1 h of milling. However, traditional Mn K-edge XAS spectra often results in broad features in the XANES region (Fig. 1). The major cause of this is the lifetime of the core-hole that is created during the excitation of the 1s electron. 20 There is also the added This journal is © the Owner Societies 2020 complication of the overlap of multiple transitions. The different transitions arise as a consequence of both local and non-local excitations; i.e. there is promotion to the hybridised orbitals located on the absorber or there is charge transfer to a neighbouring site. 7 This study has collected HERFD-XANES using the Kb 1,3 emission line, which probes the decay from 3p -1s, to assess the changes to the electronic structure of the milled materials as a function of milling time. Fig. 1 shows the XANES and HERFD-XANES data, with expanded pre-edge regions for the milled materials and the Mn oxide reference compounds. Assessing the Mn K-edge HERFD-XANES, the level of spectral resolution has vastly improved compared to the conventional XANES data. This increased resolution is more clearly pronounced in the pre-edge region, which shows more defined transitions that are well separated from the main edge. The HERFD-XANES spectra of the reference Mn oxides (Mn 2 O 3 , Mn 3 O 4 , MnO 2 and LaMnO 3 ) and their interpretation allow for an increased understanding of the ball milled intermediates and final material.
For the Kb 1,3 HERFD-XANES data the main edge results from an allowed 1s -4p transition. The position of this main edge transition is dependent on the oxidation state of the system; an increased formal charge on the Mn results in the transition shifting to higher energy. In the pre-edge region, there are multiple transitions that give rise to the complex structure observed. The 1s -3d transitions, that are normally forbidden by dipole selection rules, in this region can be quadrupolar or dipolar in nature, dependent on the local mixing of 3d/4p wavefunctions and the presence of non-local excitations; these occur with 3d states of neighbouring metal sites, through an oxygen-mediated intersite hybridisation, e.g. Mn(4p)-O(2p)-Mn 0 (3d). 21 The size and shape of the pre-edge peaks is therefore highly dependent on the geometry of central absorbing atom. 7 Assessing the pre-edge region for Mn 2 O 3 two clear peaks can be observed at 6540.8 eV and 6543.0 eV. These different features arise as a consequence of the structure of Mn 2 O 3 that contains two different Mn(III) coordination sites. Both sites are 6 coordinate, with respect to oxygen, however, one site adopts a more regular centrosymmetric O h geometry, whereas the other displays a large degree of Jahn-Teller distortion. The work by Farges showed that the lower energy peak can be assigned to the centrosymmetric Mn site and the higher energy peak to the Jahn-Teller distorted Mn site. 22 This can further be applied to the pre-edge features for the mixed valent Mn 3 O 4 , which contains both Mn(II) and Mn(III). The intense peak at 6540.5 eV results from the tetrahedrally coordinated Mn(II), allowing for 3d-4p local hybridisation, resulting in an intense dipole transition. A low intensity peak is also observed at 6542.6 eV, originating from Mn(III). It should be noted that the intensity of these two peaks, at 6540.5 eV and 6542.6 eV, do not relate to the proportion of Mn(II) and Mn(III). Work by Radu et al. shows the Mn(II) peak as a single feature, with the Mn(III) feature extending below the Mn(II) peak due to the multiplet effect of the final state (1s 1 3d 5 ). 23 For MnO 2 , the Mn(IV) cation is positioned in an octahedral site, however, the pre-edge for Mn(IV) is extremely complex and even HERFD-XANES struggles to resolve the features. 22 Previous work shows that the broad feature observed at 6542.9 eV is a combination of 1s -3d quadrupole transitions and non-local excitations. 22 LaMnO 3 , synthesised here by the Pechini method, 24 has two peaks in the pre-edge region observed at 6542.3 eV and (a shoulder at) 6547.7 eV. The peak at 6542.3 eV has been demonstrated to be two overlapping peaks from both a 1s-3d quadrupole transition and a 1s -3d4p non-local transition. This non-local transition occurs from the 3d states of neighbouring metal sites, Mn(4p)-O(2p)-Mn 0 (3d), through the vertex linked octahedral units. 25 The shoulder observed at 6547.8 eV is often overlooked within the Mn K-edge XANES of LaMnO 3 . However, work by Ignatov et al. showed they could not effectively model this shoulder within LaMnO 3 XANES without considering the contribution of d and f orbitals, suggesting this shoulder arises due to Mn 4p and La 5d hybridisation. 26 Using this analysis of reference Mn oxides we can now effectively assess the changes within the pre-edge region of Mn during the evolution of LaMnO 3 by mechanochemical synthesis. After 1 h of milling there are already clear differences in the HERFD-XANES spectrum compared to Mn 2 O 3 . The key differences are a decrease in intensity of the transitions at (1)  The peak at (3) is assigned to a 1s -3d4p non-local hybridised transition to neighbouring Mn sites, the strength of this transition is dictated by the metal-oxygen bond length and the metal-oxygen-metal bond angle. 27 The strongest transition, and ideal hybridisation, is achieved by a short bond length and a linear bond angle. 25 In our previous study, the Mn-O bond lengths calculated by EXAFS for both the sol-gel and mechanochemically prepared LaMnO 3 have comparable distances within the error range of the technique. Moreover, the final ball-milled LaMnO 3 is a complex mixture of both amorphous and crystalline content, consistent with multiple metal-oxygen-metal bond angles. The crystalline LaMnO 3 has a more linear metal-oxygen-metal bond angle and it has a higher intensity peak for this transition at (3). Furthermore, the absence of the shoulder at (4) for the ball milled materials suggests the lack of Mn4p La5d hybridisation. 26 In our previous work, we reported that the EXAFS fitting model for the ball milled material required two Mn-La scattering paths at 3.24(1) Å and 3.37(1) Å, which is significantly different to that observed for the crystalline sol-gel sample. These changes in Mn-La coordination have disrupted the Mn 4p La 5d hybridisation and affect the transition at (4).

XES.
The assessment of the HERFD-XANES pre-edge region has provided a greater detailed understanding of changes to the local geometry during the mechanochemical synthesis of LaMnO 3 and the differences compared to the sol-gel synthesised perovskite. The pre-edge features suggest a purely Mn(III) state within the ball milled materials, with no Mn(II) or Mn(IV) features observed. However, as we have seen, the XANES is strongly influenced by the local geometry around the absorbing atom and this needs to be accounted for when assessing oxidation state. We now detail Mn Kb 1,3 emission spectroscopy in order to obtain more reliable information on the Mn charge during the mechanochemical synthesis. Fig. 2 shows the Mn Kb mainlines, consisting of a strong Mn Kb 1,3 peak and with a Mn Kb 0 satellite at a lower energy, for the 3p -1s transition of reference Mn oxides and a sol-gel synthesised LaMnO 3 . The splitting of these peaks originate from the strong coupling interaction between the 3p hole and the total spin of the 3d electrons. 28,29 Any energy shifts of the Mn Kb 1,3 peak is expected to result from a combination of effective nuclear charge (Z eff ) and spin state. Work by Beckwith et al. showed that for Mn complexes the Z eff and 3d spin effects oppose one another. 30 Therefore, an increase in oxidation state corresponds to an decrease in total spin, which tends to shift the Mn Kb 1,3 to a lower energy. 21 This is strongly reflected in the Mn reference materials reported here, which have fitted Mn Kb 1,3 peaks for MnO 2 , Mn 2 O 3 and M 3 O 4 species at 6490.26, 6490.66 and 6490.73 eV, respectively (Table 1) (Fig. S1, ESI †). When considering the reference sol-gel synthesised LaMnO 3 , the fitted Mn Kb 1,3 peak position is reported at 6490.55 eV. This indicates that the Mn(III) oxidation state shows a small variation in the Mn Kb 1,3 peak position.
Assessing the position of the fitted Mn Kb 1,3 peak for reference Mn oxide materials as a function of oxidation state allows us to extract oxidation state information (Fig. 2C). The positions of the Mn Kb 1,3 emission lines suggests that the milled materials have an average oxidation state higher than the Mn(III) oxidation state expected. This is consistent with our previous XPS analysis which identified Mn(IV) at the surface of the mechanically prepared LaMnO 3 . However, there is an associated error in determining the positions of deconvoluted peaks. When this is taking into consideration, alongside the variance Mn(III) reference materials (Mn 2 O 3 and sol-gel LaMnO 3 ), all that can be reliably confirmed is that the net oxidation state for all milled samples is very similar to that of the sol-gel prepared LaMnO 3 .
3.1.3 Investigating the formation of LaMnO 3 by in situ milling and high pressure studies on La 2 O 3 and Mn 2 O 3 . We have used ex situ analysis to provide further insights into the mechanochemical synthesis of LaMnO 3 . Though in situ milling has been successfully monitored in real-time using XRD on materials such as MOFs with a vibrational mill and a PMMA jar, it was not possible to observe the evolution of the perovskite phase LaMnO 3 using this set-up (Fig. 3) with phase ID reporting only La 2 O 3 , La(OH) 3 and Mn 2 O 3 diffraction peaks after B19 h of milling. 14 Mechanochemistry results in extremely localised high temperature and pressure spots on the powdered materials due to collisions between milling media and the milling jar. 31,32 High thermal synthesise (1300-1500 K), known as the ceramic method, is commonly known as a traditional preparation method for perovskites and are already well documented in literature. 33,34 Though high pressures have been applied to already synthesised mixed metal oxides, there is limited information on using pressure alone to synthesise the LaMnO 3 phase. 35,36 In order to mimic the forces transferred to the powdered materials through   (Fig. 4). This indicates that either higher pressures or a combination of increased pressure and temperature are needed to mimic the conditions that drive mechanochemical transformations. Ultimately, this information combined with the results from the in situ vibrational mill, confirms that the mechanochemical synthesis of these metal oxide systems are extremely complicated to monitor in situ.

Comparison of milling environment
The La oxide precursor is significantly hydroscopic and readily forms a hydroxide on exposure to air. To understand the effect this change of precursor has on the final milled material, we have conducted the milling procedure in an inert atmosphere of argon. 37 The mechanochemical production of LaMnO 3 under argon was performed for 4 h, with the final material then subsequently exposed to air. The XRD data was analysed using Rietveld refinement 38 to confirm the fraction of crystalline phases in the final milled material (Fig. 5). The data confirmed that for the mechanochemically prepared LaMnO 3 under inert conditions a milling time of 4 h was not sufficient to achieve complete conversion to perovskite LaMnO 3 , with 77% crystalline perovskite calculated (100% was achieved when milling in air). The residual 23% comprises unreacted crystalline precursor phases. Here, the excess moisture/oxygen present in the milled material prepared in air is clearly beneficial in forming a crystalline final product.
To assess the crystalline and amorphous content of the argon milled LaMnO 3 EXAFS analysis was also performed (Fig. 6) (Table S1, ESI †). The argon milled LaMnO 3 EXAFS data is well modelled, using a single Mn-O and two Mn-La scattering paths. The two Mn-La scattering paths were calculated to be 3.13 Å and 3.30 Å, similar to that found for the air milled LaMnO 3 ; the shorter and longer distance La-Mn scattering paths arise due to amorphous and crystalline content, respectively. In general the EXAFS data of the air and argon milled samples are similar, however, there are differences in the Fourier transform resulting from changes in the amorphous/crystalline LaMn phases present. This model also exhibits a degree of oxygen deficiency, as with the air milled sample. To further investigate this, HERFD-XANES and Mn Kb XES studies have been performed.
As seen in Section 1.1, broad unresolved features are observed within the traditional XANES, with HERFD-XANES showing a much greater resolution (Fig. 1). Fig. 7 shows the XANES and HERFD-XANES data, with expanded pre-edge regions for LaMnO 3 catalysts synthesised by mechanochemistry (in air and argon) and sol-gel, compared to the Mn 2 O 3 precursor. The argon milled sample clearly shows an unshifted peak (1) at 6540.8 eV, which arises as a consequence of unreacted Mn 2 O 3 . The absence of peak (2) and the presence of peak (3) at 6542.6 eV suggest the formation    The observed shift in the main edge for the argon catalyst could suggest a reduction in Mn charge. However, the edge position at the Mn K-edge is not a suitable measure for oxidation state as geometry and symmetry effects also dominate the XANES and therefore Kb 1,3 emission spectroscopy has been performed in order to achieve more reliable information.
Using the fitting analysis stated in Section 3.1.2, the Mn Kb mainlines have now been assessed for the argon milled LaMnO 3 , compared to the air milled and sol-gel prepared perovskite (Fig. 8) (Fig. S3, ESI †). Both air and argon milled LaMnO 3 have fitted Mn Kb 1,3 peak positions within error of one  This journal is © the Owner Societies 2020 another, which is consistent with the sol-gel LaMnO 3 , as stated earlier.
Previously, we reported the importance of assessing the surface of the LaMnO 3 catalysts to identify possible active sites for catalysis by X-ray photoelectron spectroscopy (XPS). 5 The atomic percent ratios of La, Mn and O species at the surface of LaMnO 3 show the argon milled to have higher ratio of La to Mn on the surface compared to the air milled and sol-gel synthesised samples (Table S2, ESI †). Fig. 9A shows the O 1s XPS signals of mechanochemically and sol-gel synthesised LaMnO 3 . The argon milled sample follows a vastly different spectral profile compared to that of both the air milled and solgel synthesised LaMnO 3 . Previously, we reported using curve fitting to assess the change in adsorbed species on the catalyst surface. 5 Using curve fitting here, three features were deconvoluted within the O 1s XPS region at 529.3-529.6 eV, 531.1-531.7 eV and 533.4-535.0 eV. The lowest energy binding peak at 529.3-529.6 eV is assigned to lattice-type oxygens in the perovskite (O 2 2À ), with the peak at 531.1-531.7 eV assigned to adsorbed species, such as O 2À , O À or OH À . [39][40][41] The peak at B533.5 eV corresponds to adventitious carbon, with analogous C-O peaks confirmed within the C 1s XPS region at 286.2 eV and 288.6 eV. [39][40][41] Commonly it is assumed that active oxygen species adsorb on the surface to compensate for lattice oxygen vacancies. 41 Firstly, the Brunauer-Emmett-Teller (BET) surface areas were calculated to be within range of one another; B5 m 2 g À1 for the ball milled samples and B8 m 2 g À1 for the sol-gel LaMnO 3 . Therefore, comparing the adsorbed oxygen to lattice oxygen can give an indication of the changes to adsorption centres, independent of surface area. [42][43][44] The argon milled sample shows a greater intensity for the higher energy binding peak, indicating more adsorbed species on the surface compared to the other LaMnO 3 samples. The relative intensity of the adsorbed oxygen feature decreases in the order; argon milled 44 h milled E sol-gel preparation.
The XPS Mn 3s region spectra (Fig. 9B) indicate a mixed Mn oxidation state with peak splitting at 4.8-5.0 eV suggesting the presence of Mn(IV) and Mn(III) species (Table S3, ESI †). Analysis in the La 3d region (Fig. 9C) indicate a 3+ oxidation state for all samples, however, the argon milled LaMnO 3 presents a different multiplet splitting (Table S3, ESI †). A reduced multiplet splitting of 3.7 eV is indicative of La(OH) 3 species, consistent with the higher surface ratio of La (Table S2, ESI †), and the higher percentage of adsorbed species, -OH, within the O 1s region. This XPS analysis signifies the importance of surface sensitive techniques, as well as bulk structure analysis, XAS and XES, in order to understand how different synthetic routes effect the structural and catalytic properties of these LaMnO 3 catalysts.    (Fig. 10). Good reproducibility of this performance is shown in Fig. S4 (ESI †). Air milled and the sol-gel synthesised LaMnO 3 show similar activity, which begins B350 1C with 100% conversion achieved by 550 1C. The argon milled LaMnO 3 , however, shows activity starting below 300 1C. This indicates that the catalytic activity cannot be explained by analysis of the bulk structures alone, as both ball milled catalysts show similarities in structure by HERFD-XANES and XES. The surface analysis of the O 1s region by XPS, however, indicates changes at the surface for the argon milled LaMnO 3 . With the desorption of oxygen well understood to be the rate controlling step within the mechanism of deN 2 O this supports our previous work, that proposes the higher amount of surface adsorption centres is linked to a higher conversion of N 2 O to N 2 at lower temperatures. 5,45,46 Furthermore, this argon milled catalyst shows a change in the light-off curve at 550 1C, which has previously been correlated to an increase in crystalline perovskite phase for other ball milled species during deN 2 O. Here, the high reaction temperatures have resulted in annealing of the ball milled perovskite, changing the catalyst structure and thus the catalytic activity (Fig. S5, ESI †). However, the argon milled catalyst still shows to be largely effective at lower temperatures, before the phase transformation takes place. Furthermore, with XPS analysis showing the argon milled catalyst to have a higher proportion of La at the surface, it raises a very interesting feature as traditionally it is thought that the Mn is the active species for catalysis in this system. 47 Due to La electronic configuration it does not possess redox capabilities and is most stable in a 3+ oxidation state. However, differences do occur in the multiplet splitting within the XPS La 3d region, with the argon milled catalyst having a higher percentage of adsorbed oxygen species. In order to understand the role of La during deN 2 O and how the other XPS regions are affected during the reaction in situ near ambient pressure (NAP)-XPS will now be discussed.

In situ NAP-XPS deN 2 O
In the previous sections, we were able to link the catalytic activity towards deN 2 O to the surface properties of the LaMnO 3 catalysts. It is critical to understand how the deN 2 O process effects the surface properties and stability of our LaMnO 3 catalysts during N 2 O adsorption, dissociation and desorption of N 2 . Our previous lab-based XPS was performed under vacuum with a fixed incident beam. This ex situ analysis of the surface of the catalysts often presents vastly different properties compared to that under reaction conditions. 48 Furthermore, controlling the incident energy allows for the same specific penetration depth of the surface for each XPS region. 49 Table 2). [51][52][53] On subsequent heating to 400 1C and 600 1C the peak positions are observed to remain AE0.2 eV within each other. Assessing the area of the deconvoluted peaks within Mn 2p 3/2 region can give an indication of the relative abundance of Mn(III) and Mn(IV) contributions (Table 2). 40   This is indicative of surface reactions occurring during catalysis that change the net oxidation state.
The Mn 3s region provides complimentary information and was recorded at an increased kinetic energy of B770 eV, for enhanced depth profiling. The splitting of the Mn 3s peak (Fig. 11B) at B83.5 and B88.5 eV (Table 2) is a final state effect due to the parallel and anti-parallel coupling between the spins of the remaining 3s electron and the 3d electrons. 54 On increasing oxidation state, i.e. a decrease of d electrons, the separation between these two peaks is known to decrease. 55 For the argon milled catalyst a Mn 3s peak splitting value of 5.1 eV is observed at room temperature in vacuum. This suggests, a mixed Mn(IV)/Mn(III) valency, with literature reporting a peak splitting of 5.4 eV for Mn 2 O 3 and 4.4 eV for MnO 2 . 52,53 No change is observed for the Mn 3s peak splitting on the exposure to N 2 O gas. However, on increasing temperature, this splitting increases to 5.4 eV, which is indicative of the Mn(III) oxidation state. 53 This increased depth-profiling observes the diffusion of sub-surface oxygen towards the exterior of the particle, which is responsible for the observed change in oxidation state. 56,57 The lattice oxygen that has migrated towards the surface is then available to participate in the deN 2 O process. 58,59 With the mixed Mn(III) and Mn(IV) valences still observed at 600 1C at a lower depth penetration (Mn 2p region) it suggests this is a highly surface sensitive reaction.
This behaviour, of the splitting energy increase due to elevated temperatures, within the Mn 3s region was also found for the other LaMnO 3 catalysts (Tables S4 and S5) (Fig. S10 and S11) (ESI †). Understanding the Mn 2p and Mn 3s XPS regions highlights that, though the differently synthesised LaMnO 3 catalysts have varying bulk properties, at the surface there are similarities in how the Mn environment acts under working conditions.
The La 3d XPS region ( Fig. 12 and Fig. S12, ESI †) shows the typical doublet splitting, with the lower energy doublet assigned to La 3d 5/2 and the higher, La 3d 3/2 . 60 The peak splitting of the La 3d 5/2 doublet has been used to understand the nature of the La species. 61 The asymmetric nature of the peaks within the La 3d region required three peaks to deconvolute each doublet ( Fig. 12 and Fig. S12, ESI †). 62,63 For the argon milled catalyst, all the values of La 3d 5/2 peak splitting during deN 2 O are typical of La(III) compounds, which is further confirmed by the spin-orbit coupling values of 16.8-16.9 eV for all species (Table 2). 41 Initially, at RT under vacuum, the La 3d 5/2 peak splitting of 3.9 eV is indicative of La(OH) 3 (Table 2). This is further confirmed by the presence of La(OH) 3 diffraction peaks in Fig. 5. Increasing the temperature during deN 2 O results in an increase in the peak splitting value, suggesting the formation of an oxide species at the surface. 64 Both the air milled and sol-gel LaMnO 3 start with a higher La 3d 5/2 splitting value, of 4.1 eV and 4.2 eV respectively, compared to that of the argon milled catalyst, indicating a more oxide surface initially (Tables S4 and S5) ( Fig. S13 and S14) (ESI †). 65 The ball milled samples record a decrease in the La 3d splitting on exposure to N 2 O (B0.1 eV). This could be indicative that the adsorbed N 2 O is interacting with surface La species. Fig. 13 shows the deconvoluted fitted peaks for the O 1s region in argon milled LaMnO 3 at RT under a N 2 O atmosphere and then at 600 1C (full data set Fig. S15, ESI †). On deconvolution of the O 1s region three oxygen species can be identified ( Table 2). In Table 2, peak 1 is assigned to lattice-type oxygens (O 2 2À ), peak 2 to adsorbed species such as O 2 À , O À or OH À and peak 3 and 4 is associated with adsorbed adventitious carbon or molecular water. [39][40][41] By monitoring in situ deN 2 O by NAP-XPS clear changes can be observed within the O 1s region for the argon milled catalyst, which have not been previously reported. Whilst remaining at RT, on the introduction of N 2 O an increase in the proportion of adsorbed species can be observed, along with the presence of a higher binding energy peak 4 at 534.6 eV (Fig. S15A, ESI †). This is indicative that N 2 O is adsorbed on the surface at room temperature, in agreement with La 3d XPS data. A change in

View Article Online
This journal is © the Owner Societies 2020 Phys. Chem. Chem. Phys.
the area of peak 1, arising due to lattice-type oxygens, is observed on exposure to N 2 O. This could indicate a possible rearrangement of the surface structure due to N 2 O interaction. On increasing the temperature to 400 1C the surface adsorbed species decrease. Further increase of the temperature to 600 1C shows a significant change in the peak shape, along with a 0.5 eV energy increase in the binding energy of peak 1.
Here, the change in relative oxygen species abundance and peak positions could relate to either the hydrothermal removal of adsorbed oxygen species from the surface, as seen in the La 3d region (Table 2), or changes due to the Mn enivronment. 66,67 The Mn 2p and Mn 3s XPS data confirm that under these conditions the surface Mn species are relatively unchanged, whilst at greater depths oxygen is transferred towards the surface. Considering, that the O 1s data is acquired at the same depth penetration as the Mn 2p, it infers that the changes observed are correlated to changes in La speciation. Both the air milled and sol-gel prepared LaMnO 3 also recorded a higher proportion of adsorbed species at the surface on exposure to N 2 O (Tables S4 and S5) ( Fig. S16 and S17) (ESI †). Significantly, the air milled and sol-gel catalysts do not have peak 4, fitted at 534.6 eV for the argon milled LaMnO 3 , present within the O 1s region. This suggests that there are distinct surface sites for the argon milled LaMnO 3 that are involved in the deN 2 O process at lower temperatures, increasing the catalytic activity (Fig. 10). 58,59 Furthermore, the air milled catalyst also suggests a rearrangement in surface structure due to exposure of N 2 O, as a reduction in peak 1, corresponding to lattice-type oxygen, is observed.
The deN 2 O catalytic testing observed an abrupt change to the light off profiles for the ball milled prepared samples that have been assigned to a phase transformation (Fig. 10). This behaviour mirrors what is found in the O 1s XPS data; between 400 1C and 600 1C there are significant changes within the O 1s profile for the balled milled samples. Conversely, the sol-gel prepared LaMnO 3 shows a similar peak profile within the O 1s region at both 400 1C and 600 1C, suggesting a relatively stable oxygen environment.
Although the role of Mn towards the catalytic activity is clearly important, in this work the Mn 2p region shows minimal changes during deN 2 O. 63 However, there are significant changes, within the La 3d and O 1s regions that have provided additional value in understanding the catalytic activity of these LaMnO 3 systems. Furthermore, variations within the in situ deN 2 O NAP-XPS regions compared to the ex situ XPS reported in Section 3.2 show the importance in tuning the incident beam in order to achieve equivalent data for each XPS region at the same penetration depth.