Asymmetric nanoparticle oxidation observed in-situ by the evolution of diffraction contrast

The use of transmission electron microscopy (TEM) to observe real-time structural and compositional changes has proven to be a valuable tool for understanding the dynamic behavior of nanomaterials. However, identifying the nanoparticles of interest typically require an obvious change in position, size, or structure, as compositional changes may not be noticeable during the experiment. Oxidation or reduction can often result in subtle volume changes only, so elucidating mechanisms in real-time requires atomic-scale resolution or in-situ electron energy loss spectroscopy, which may not be widely accessible. Here, by monitoring the evolution of diffraction contrast, we can observe both structural and compositional changes in iron oxide nanoparticles, specifically the oxidation from a wüstite-magnetite (FeO@Fe3O4) core–shell nanoparticle to single crystalline magnetite, Fe3O4 nanoparticle. The in-situ TEM images reveal a distinctive light and dark contrast known as the ‘Ashby-Brown contrast’, which is a result of coherent strain across the core–shell interface. As the nanoparticles fully oxidize to Fe3O4, the diffraction contrast evolves and then disappears completely, which is then confirmed by modeling and simulation of TEM images. This represents a new, simplified approach to tracking the oxidation or reduction mechanisms of nanoparticles using in-situ TEM experiments.


Introduction
Transmission electron microscopy (TEM) is a powerful tool for the real time observation of structural and compositional changes in nanomaterials and has been employed widely in heterogeneous catalysis and for the tracking of dynamical evolution of nanostructures, under controlled atmospheric conditions [1][2][3][4].When coupled with aberration correction, TEM performed in bright-field conditions can allow for real-time tracking of atomic movement and facet reconstruction at exposed crystalline surfaces to provide information on nanoparticle evolution with unprecedented temporal and spatial resolution.This has been helped significantly by the development of next-generation direct electron detection cameras that can image at incredibly high speeds and low electron dose rates [5][6][7][8][9].Further, the increased field of view in such cameras increases the chances of detecting and tracking the dynamical evolution of a nanoparticle of interest.Although atomic-scale information is captured in real time it can typically require post-experimental processing such as drift correction, frame integration and denoising to increase the signal to noise ratio and be observable.Therefore, the tracking of a macroscopical structural event such as surface reconstruction, particle movement (i.e.along a substrate), or particle growth is typically used to identify the region of interest for high resolution observation and analysis.While this is a reasonable approach for many experiments, nanomaterials that experience subtle structural, crystallographic, or compositional changes can be difficult to identify in bright field TEM, especially given the often-present drift issue.Furthermore, unless the microscope is equipped with in-situ electron energy loss spectroscopy (EELS), observing changes in oxidation state is not feasible.
In the present study, we track the evolution of diffraction contrast as an effective identifier of structural and compositional changes during an in-situ TEM experiment.When a small spherical inclusion is included in a thin film matrix and there is coherent strain across the interface the TEM image can be dominated by strain-field contrast [10,11].Ashby and Brown first observed this contrast phenomenon under a two-beam weak field imaging condition in the TEM.When imaged under a two-beam condition the contrast appears as a pair of lobes approximately symmetrical across a 'line of no contrast' .The line of no contrast is aligned with the image of the particle perpendicular to the active g vector [12].The two-fold symmetry of the contrast is not due to two-fold symmetry of the particle itself, rather it is associated with the principal diffracting plane causing the contrast.When imaged under ideal and calibrated conditions the contrast can be used to measure the coherent strain-field, if no misfit dislocations are present [10].Ashby-Brown contrast has been observed in Au@Pd core-shell nanocubes, with a lattice mismatch of 4.6%, and CdSe@CdS core-shell nanoparticles, with a lattice mismatch of 4.4% [13,14].
Iron oxide nanoparticles are ideal nanomaterials for studying the evolution of structural and compositional changes since they can exhibit different oxidation states within a single nanoparticle, e.g. in wüstite-magnetite (FeO@Fe 3 O 4 ) core-shell nanoparticles.The lattice mismatch between FeO and Fe 3 O 4 has been theoretically and experimentally shown to range from approximately 2.3%-4.6%,falling within the range required for coherent strain conditions [15].Elucidating the mechanisms of oxidation for iron oxide nanoparticles is an important problem as their magnetic properties depend on their crystal structure, and the formation of defects during oxidation can significantly affect their performance [16].For example, it has been shown that the presence of antiphase boundaries can decrease the saturation magnetization (M sat ) by up to 50%, even in nanoparticles of the same size and overall composition [17].
In this study, we observe the dynamic evolution of Ashby-Brown contrast during the in-situ oxidation process of FeO@Fe 3 O 4 core-shell nanoparticles, leading to the formation of single-crystalline Fe 3 O 4 nanoparticles.Through experimental, atomic modeling and simulation of TEM images, we establish a correlation between the progression of light and dark contrast and the corresponding structural and compositional changes.We therefore show that diffraction contrast can be used as an effective identifier of nanoparticle transformations in-situ.

Sample preparation
FeO@Fe 3 O 4 nanoparticles were prepared by first synthesizing FeO nanoparticles, following a previously reported synthesis, and then letting the surface oxidize passively [18].To synthesize FeO nanoparticles 1.34 g of Fe(acac) 3 (Aldrich, 97%) was combined with 20 ml of dried and degassed oleic acid (Aldrich, 90%) in a 50 ml flask held in a molten metal heating bath.A constant flow of dry N 2 gas was maintained at 10 sccm over the reaction solution.The temperature was then increased to 380 • C where it was held for 4 h.The FeO nanoparticles were stored in their crude reaction solution in the freezer inside the N 2 glovebox, to prevent uncontrolled oxidation.To begin the in-situ experiment the FeO nanoparticles were removed from a N 2 glove box and washed with several washing steps using degassed isopropanol and hexanes with centrifugation, to remove excess organic capping ligands.This caused partial oxidation of the nanoparticle surface, resulting in FeO@Fe 3 O 4 core-shell nanoparticles.A 4 µl aliquot of the FeO@Fe 3 O 4 core-shell nanoparticles in hexanes was then dropcast onto a Protochips Heating Chip (E-FHDC-ENV-10) and then promptly inserted into the high vacuum of the TEM to prevent any further uncontrolled oxidation.

In-situ analysis
High-resolution TEM (HRTEM) imaging was performed on a FEI Titan environmental TEM (ETEM) with Image Cs corrector operating at 300 keV, equipped with a Gatan K3-IS direct detection camera and Protochips Aduro heating holder and custom gas-delivery cart.The FeO@Fe 3 O 4 core-shell nanoparticles were then imaged in bright-field conditions with a cumulative e − dose of 3000 e − Å −2 with no change, indicating there are no appreciable effects from the e − beam.Oxygen (O 2 ) gas flow was then set to 10 sccm and pressure controlled to 100 torr at the delivery point to the ETEM.The leak valves were adjusted until a steady column pressure of 2.5 mbar O 2 was achieved.No further oxidation was observed under these conditions at room temperature.
The sample temperature was then increased at a ramp rate of 0.33 • C s −1 to minimize the drifting until it reached 175 • C, at which point the column valves were opened and the sample exposed to the e-beam at a dose rate of 10 e − Å −2 s −1 .In-situ dataset collection was started once the temperature reached 220 • C (using the 5 s lookback feature) and stopped once complete oxidation was observed at a temperature of 265 • C. Data was recorded at 25 fps using the full field of view of the K3-IS camera (5760 × 4092 pixels).
To increase signal to noise ratio 10 frames of the in-situ dataset were summed and aligned, then the region of interest was systematically reduced to isolate and track a single FeO@Fe 3 O 4 nanoparticle oriented close to the [110].All manipulation of the in-situ data was performed using the In-Situ Editor of Gatan DigitalMicrograph.A final video was then exported at 20 fps, sped up by 16× and compressed for ease of data handling and viewing using a common video editor.To measure the contrast profiles each image was loaded into ImageJ and a vertical contrast profile extracted using the Profile tool.The plots were then smoothed in Origin using Adjacent Averaging with Weight Average and a Reflect boundary condition.

FeO@Fe 3 O 4 core-shell nanoparticle models
To create a model of the core-shell system, we first create a spherical Fe 3 O 4 nanoparticle with a diameter of 10 nm.For the uniform system, a spherical void of diameter 8 nm was carved out.To create the low and high skew systems, the void center was moved to 0.5 and 0.25 of the void radius along the y-axis.The void diameters were also reduced from 8 nm to 5 and 3 nm for the low-and high-skew systems, respectively.After creating the voids, a FeO nanoparticle was created from the FeO unit cell.The radius of the FeO nanoparticle was less than that of the void by 1%.This was done to ensure relaxation in the interlayer region during conjugate gradient.Fe 3 O 4 has a spinel structure and the Fd-3m space group.FeO 4 tetrahedra are formed when Fe 2+ are bonded to four O 2− ions.The corners of the tetrahedra are shared with 12 FeO 6 octahedra.The unit cell is cubic with a lattice constant of 8.44 Å which results in a unit cell volume of 602.62 Å 3 .FeO is a rock salt crystal structure in which Fe 2+ is bonded to six O 2− ions.This leads to a formation of FeO 6 octahedra.The lattice constant of FeO is 4.26 Å with a unit cell volume of 77.44 Å 3 .(However, a more appropriate comparison is to 8 cells of FeO, which equal a volume of 619.52 Å 3 ).The entire system was placed in the center of a 20 nm 3 box and periodic boundary conditions were applied in all the three directions.This creates a standalone nanoparticle which was then thermalized in molecular dynamics simulation.
We used a reactive (ReaxFF) forcefield for the simulations [19].The reactive forcefield consists of both valence and non-valence interactions.The ReaxFF potential energy consists of the following terms: The non-covalent interactions include E vdw and E coul , both of which are screened by a taper function.The charges in ReaxFF are quantified by electronegativity equalization method.The initial forcefield was developed by Kubicki and co-workers [20].The forcefield describes the interaction of α FeOOH (Geothite) and water.We used this forcefield as the starting point to train a Reax forcefield for our simulation.DFT based calculations of energy and lattice constant were carried out on both FeO and Fe 3 O 4 structures and the two and three body terms were updated.We used VASP to carry out our DFT calculations.Generalized gradient approximation GGA-PBE were used with a cutoff of 500 eV [21].

Image simulation
Atomic-resolution phase-contrast image simulations are carried using the multislice algorithm within the Java-EMS (JEMS) program [22].Multislice simulations have been performed to generate the thickness-defocus maps of the FeO embedded in Fe 3 O 4 matrix nanoparticle structures along [110] direction.These nanoparticles structures are generated by a customized protocol with FeO insertions in Fe 3 O 4 with (i) uniform strain (ii) low skewed (iii) high skewed and (iv) pure Fe 3 O 4 .The simulated images have been generated for a defocus window of −20 nm to −40 nm of the objective lens and in the thickness raging from ∼25 nm to 35 nm.Simulated images of a single nanoparticle are presented with a thickness of ∼25 nm and defocus values of −35 nm.They match well with experimentally observed high-resolution phase-contrast images of the nanoparticles.During the recording of the HRTEM images in the aberration-corrected TEM, the defocus value of the objective-lens system was maintained at approx.−25 nm, while the fine focus control was achieved with a piezo-controlled goniometer-stage movement with nanometer precession.However, the simulation is performed with defocus window ranging −20 to −40 nm to accommodate any deviation during recording of the images.

Results and discussion
The FeO nanoparticles were synthesized using a previously reported thermal decomposition method and possessed an average size of 21.1 nm with a size distribution of 6.3% (figure S1) [18].The nanoparticles are partially oxidized to form FeO@Fe 3 O 4 core-shell during TEM sample preparation.They were then dropcast on a Protochips in-situ-heating chip and loaded into the ETEM in a Protochips Aduro heating holder.The sample was first imaged with a high e − dose, with no observable changes, indicating the beam-sample interactions were negligible.Then, O 2 gas was introduced into the column until a steady pressure of 2.5 mbar was achieved.The sample was then heated at 0.33 • C s −1 and recording of the in-situ dataset was started at the first observable change to the nanoparticles at 220 • C (t 0 ).TEM images were taken at an e-dose of 10 e − Å −2 s −1 and 25 frames per second (fps) using the Gatan K3-IS camera (figures S2 and S3, SI).Recording of the in-situ dataset was stopped following complete oxidation of the nanoparticles once the temperature reached 265 • C (t f ).
Figure 1(a) shows a TEM image of an ensemble of FeO@Fe 3 O 4 core-shell nanoparticles at t 0 , before the introduction of O 2 gas.To increase signal to noise ratio and contrast 10 frames of the in-situ dataset were summed and aligned using gatan digital micrograph (GMS).Several nanoparticles are oriented to show parallel alternating light and dark diffraction contrast, so called 'Ashby-Brown' contrast [10,23].This diffraction contrast is attributed to a local deformation of the crystal lattice and coherent strain across the interface of FeO and Fe 3 O 4 and indicates a core-shell structure [10,14].
Figure 1(b) shows a selected area diffraction (SAD) pattern of FeO@Fe 3 O 4 core-shell nanoparticles at t 0 .At t 0 we identify rings corresponding to the (111), (200), and (220) reflections of FeO.There is also a weak The conversion of FeO@Fe 3 O 4 core-shell nanoparticles to pure Fe 3 O 4 nanoparticles is confirmed by SAD pattern in figure 1(d).Following complete oxidation of the nanoparticle ensemble (t f ), there is an increase of the brightness of the Fe 3 O 4 (220) reflection as well as the disappearance of the FeO (220) reflection and emergence of the (422), (522), and (440) reflections of Fe 3 O 4 .While there are enough differences in the SAD patterns from the ensemble to discern between FeO and Fe 3 O 4 , this becomes much more difficult on the single particle level, as the main reflections for FeO and Fe 3 O 4 can overlap depending on the nanoparticle zone axis.However, at the single particle level the diffraction contrast can provide a unique pathway to identify a structural or compositional transformation in-situ, and in real time.We therefore tracked a single nanoparticle that showed Ashby-Brown contrast at t 0 through to complete oxidation at t f .We used GMS to down-select a region of interest of the in-situ dataset, to align and track the single FeO@Fe 3 O 4 nanoparticle highlighted in red in figure 1.The complete in-situ transformation is given as a video in supporting information, video S1.
Figure 2(a) shows the down selected FeO@Fe 3 O 4 core-shell nanoparticle taken at t 0 , along with the corresponding fast Fourier transform (FFT) in figure 2(b).The nanoparticle shows alternating light and dark diffraction contrast perpendicular to the (111) diffraction spot, indicating (111) is the principal diffracting plane causing the contrast.From FFT analysis the nanoparticle is oriented along the [110] zone axis.There appears to be a slight tilt due to the non-uniform brightness of the (111) spots in the FFT. Figure 2(c) shows a TEM image of the same nanoparticle now fully oxidized to Fe 3 O 4 at t f , with the corresponding FFT in figure 2(d) indicating no change in crystallographic orientation following oxidation.However, the alternating light and dark contrast has evolved into uniform contrast across the nanoparticle.To track the evolution of the light and dark contrast over time, snapshots of the in-situ video at various time points were taken corresponding to the most striking changes in particle appearance.Figure 3(a) shows the FeO@Fe 3 O 4 nanoparticle taken at t = 23 s, with figures 3(c), (e) and (g) showing nanoparticles at varying stages of oxidation taken at t = 96 s, t = 100 s, and t = 148 s, respectively.To quantify the change in diffraction contrast we summed the pixel intensity profiles across the white highlighted areas which are perpendicular to the direction of the alternating contrast (along the [111] direction of the dominant reflection), with the profiles given alongside the respective TEM image.In figure 3(a), the nanoparticle at t = 23 s shows an alternating dark and light contrast, which is represented in figure 3(b) as a pseudo-sinusoidal intensity profile.For the nanoparticle at t = 96 s, the diffraction contrast in figure 3(e) is somewhat flatter, with the profile intensity in figure 3(f) likewise reduced along with a change in peak position.At t = 100 s (figure 3(i)) we again observe alternating light and dark contrast.However, the peak position has now shifted when compared to the initial nanoparticle and the diffraction contrast intensity profile is inverted (figure 3(j)).We also observe an additional third peak of contrast emerge.Finally, after complete oxidation at t = 148 s we observe uniform contrast in the nanoparticle (figure 3(m)), which is confirmed in the extracted intensity profile (figure 3(n)).Such an evolution of diffraction contrast was reported by Ashby and Brown [10].They showed that a small spherical inclusion contained within a matrix with coherent strain displays alternating contrast directed perpendicular to a line of no contrast.Further, they showed that the diffraction contrast displayed is dependent on the size of the inclusion, its shape, and its depth in the matrix with respect to the upper surface.It was also theorized that as the inclusion increased its depth with respect to the foil surface, the contrast profile would evolve and invert in both bright and dark field conditions.
To understand the structural contribution to the change in diffraction contrast we made a qualitative comparison to simulations of TEM images of model FeO@Fe 3 O 4 core-shell nanoparticles with changing core sizes.Models were formed with a FeO@Fe 3 O 4 core-shell structure containing different core size and shell thickness.To simulate a change in the depth of the inclusion the core was offset within the shell to create uniform, low-skew, and high-skew core-shell structures (figure S4).The systems were relaxed and then heated to 300 K in the canonical ensemble (NVT) over 10 000 steps and quenched using the conjugate gradient method to attain the zero-force configuration, using a timestep of 1 fs to run the calculations.The open source LAMMPS software was used to run the simulations [24].Then, atomic resolution phase contrast image simulations were carried out using the algorithm within the JEMS program.The atomic models were viewed down the [110] direction (figure S5) and simulated in JEMS to compare directly to experiment.The contrast intensity profiles of the simulated HRTEM images from the atomic models show a similar pattern as the intensity profiles obtained from the in-situ experiment.The intensity profile of the uniform core-shell model (figure 3(d)) possesses the same pseudo-sinusoidal profile comparable to the intensity profile of the experimental TEM image (figure 3(b)).The intensity and phase of the peaks differs, however exactly matching the size and depth of the FeO core between simulation and experiment is not possible.In figure 3 As the shape and symmetry of Ashby-Brown contrast in bright-field images depends on the depth of the inclusion in the upper surface of the foil, and contrast symmetry only occurs in bright-field images when the inclusion is at center height, these results would indicate that the shell is not oxidizing uniformly.Further, the inversion of alternating light and dark contrast bands indicates that at t = 96 s the inclusion is towards the lower surface of the matrix, as observed by Ashby and Brown [10].In our case, this arises from asymmetric oxidation due to the deposition of FeO@Fe 3 O 4 nanoparticles onto the carbon substrate, as illustrated in  It has previously been shown that a single line of no contrast flanked by two lobes of contrast was indicative of a fully coherent and symmetrical strain field across the interface, whereas multiple contrast lines suggested the presence of misfit dislocations and the loss of coherency [25].However, the nanoparticles used in this study have an average size of 21.1 nm which is not expected to be large enough to support misfit dislocations.The thickness of a pseudomorphic layer of magnetite on a substrate of wustite is at least 16 nm, indicating a larger nanoparticle diameter would be needed for misfit dislocations to emerge [26].In fact, in a uniform thin film particles as large as 35 nm in diameter appear to possess coherent strain fields [11,23].However, if the foil matrix was deformed by 5% in tension the characteristic contrast of a coherent strain field was lost, and multiple lines of no contrast were observed [11].Core-shell nanoparticles possess considerable tensile strain at their interface, with extended facets showing increased tension compared to corner or edge sites [27,28].In a spherical inclusion the strain is expected to be relatively uniform throughout the interface, however it will increase with increasing curvature i.e. decreasing core size.This is observed in figure 3

Conclusion
In conclusion, our study demonstrates that diffraction contrast can be used as a simple approach to monitor the structural and compositional evolution of nanoparticles, as shown by tracking the complete oxidation of FeO@Fe 3 O 4 core-shell nanoparticles to single crystal Fe 3 O 4 nanoparticles.The distinct light and dark contrast observed at the center of the nanoparticles corresponds to the characteristic 'Ashby-Brown' contrast, indicating the presence of a FeO@Fe 3 O 4 core-shell structure.As the nanoparticles undergo oxidation under flowing O 2 gas in an in-situ TEM the contrast profile evolves, inverts and then gradually disappears, indicating the transformation into single-crystal Fe 3 O 4 .This concept presents a new pathway to identify and track oxidation or reduction of nanoparticles without relying on high-resolution TEM, scanning transmission electron microscopy (STEM), or EELS spectral imaging.

Figure 1 .
Figure 1.TEM (a) and SAD (b) images of FeO@Fe3O4 core-shell nanoparticles taken at t0 along with images of fully oxidized magnetite (Fe3O4) nanoparticles taken at t f (c) and (d).The Fe3O4 (220) ring is faintly visible in (b) indicating the presence of a thin magnetite shell at t0, which increases in intensity after oxidation (d).The individual nanoparticle tracked and aligned for in-situ observation is highlighted in red.
Fe 3 O 4 (220) reflection emerging from the thin shell created by surface oxidation during sample preparation.FeO has a rock-salt-type structure where O 2− anions are arranged in fcc closed-packed T h sites and Fe 2+ ions occupy the O h sites.The Fe 3 O 4 (220) diffraction line is associated with the cations ordered in T h sites characteristic of spinels and confirms the presence of an oxide shell.The same ensemble of nanoparticles following complete oxidation into Fe 3 O 4 is shown in figure 1(c), where we observe the disappearance of Ashby-Brown contrast, and each nanoparticle possesses uniform contrast.The evolution of diffraction contrast is indicative of a structural or compositional change, as adherence to the carbon substrate precludes any rotational or translational movement.From the introduction of O 2 gas and increasing temperature we observed appreciable drift in the ensemble which is indicated by the single nanoparticle in red (figures 1(a) and (c)).

Figure 2 .
Figure 2. TEM (a) and corresponding FFT (b) images of a single FeO@Fe3O4 core-shell nanoparticles taken at t0.From (b) the particle is oriented along the [110] direction, with a slight tilt observed due to the non-uniform brightness of the spots.The alternating light and dark contrast observed in (a) is positioned perpendicular to the strongest (111) reflection.TEM (c) and corresponding FFT (d) images of a magnetite nanoparticles taken at t f .
(h) the profile of the low-skew model flattens, mimicking the behavior seen at t = 23 s in experiment.The profile intensity of the high-skew simulation is given in figure 3(l), and shows an inversion of the contrast intensity profile when compared to figure 3(d) which again matches the experimental profile change at t = 96 s.Finally, a simulation of a single crystal Fe 3 O 4 nanoparticle (figure 4(p)) displays a uniform contrast profile which is similar to the intensity profile of the experimental TEM image (figure 4(n)).

Figure 4 .
Figure 4. FeO@Fe3O4 core-shell nanoparticle undergo asymmetric oxidation in an in-situ TEM experiment.The carbon support limits oxidation occurring from the lower side of the nanoparticle.This means the FeO core can be considered an inclusion that reduces in size, and increases in depth, from the surface of the Fe3O4 shell.

figure 4 .
figure 4. Oxygen gas flows across the upper surface of the nanoparticles and oxidizes the nanoparticles from the top-down.It has previously been shown that a single line of no contrast flanked by two lobes of contrast was indicative of a fully coherent and symmetrical strain field across the interface, whereas multiple contrast lines suggested the presence of misfit dislocations and the loss of coherency[25].However, the nanoparticles used in this study have an average size of 21.1 nm which is not expected to be large enough to support misfit dislocations.The thickness of a pseudomorphic layer of magnetite on a substrate of wustite is at least 16 nm, indicating a larger nanoparticle diameter would be needed for misfit dislocations to emerge[26].In fact, in a uniform thin film particles as large as 35 nm in diameter appear to possess coherent strain fields[11,23].However, if the foil matrix was deformed by 5% in tension the characteristic contrast of a coherent strain field was lost, and multiple lines of no contrast were observed[11].Core-shell nanoparticles possess considerable tensile strain at their interface, with extended facets showing increased tension compared to corner or edge sites[27,28].In a spherical inclusion the strain is expected to be relatively uniform throughout the interface, however it will increase with increasing curvature i.e. decreasing core size.This is observed in figure3(j); the contrast has inverted when compared to figure3(b) indicating the FeO core position has moved downwards with respect to the upper Fe 3 O 4 nanoparticle surface, and the number of contrast bands has increased indicating an increase in strain due to the smaller core size.
figure 4. Oxygen gas flows across the upper surface of the nanoparticles and oxidizes the nanoparticles from the top-down.It has previously been shown that a single line of no contrast flanked by two lobes of contrast was indicative of a fully coherent and symmetrical strain field across the interface, whereas multiple contrast lines suggested the presence of misfit dislocations and the loss of coherency[25].However, the nanoparticles used in this study have an average size of 21.1 nm which is not expected to be large enough to support misfit dislocations.The thickness of a pseudomorphic layer of magnetite on a substrate of wustite is at least 16 nm, indicating a larger nanoparticle diameter would be needed for misfit dislocations to emerge[26].In fact, in a uniform thin film particles as large as 35 nm in diameter appear to possess coherent strain fields[11,23].However, if the foil matrix was deformed by 5% in tension the characteristic contrast of a coherent strain field was lost, and multiple lines of no contrast were observed[11].Core-shell nanoparticles possess considerable tensile strain at their interface, with extended facets showing increased tension compared to corner or edge sites[27,28].In a spherical inclusion the strain is expected to be relatively uniform throughout the interface, however it will increase with increasing curvature i.e. decreasing core size.This is observed in figure3(j); the contrast has inverted when compared to figure3(b) indicating the FeO core position has moved downwards with respect to the upper Fe 3 O 4 nanoparticle surface, and the number of contrast bands has increased indicating an increase in strain due to the smaller core size.