High surface stability of magnetite on bi-layer Fe3O4/Fe/MgO(0 0 1) films under 1 MeV Kr+ ion irradiation

We investigate the stability of the bi-layer Fe3O4/Fe(0 0 1) films grown epitaxially on MgO(0 0 1) substrates with the layer thickness in the range of 25–100 nm upon 1 MeV Kr+ ion irradiation. The layer structure and layer composition of the films before and after ion irradiation were studied by XRR, RBS and RBS-C techniques. The interdiffusion and intermixing was analyzed. No visible change in the RBS spectra was observed upon irradiation with ion fluence below 1015 Kr cm−2. The bi-layer structure and the stoichiometric Fe3O4 layer on the surface were well preserved after Kr+ ion irradiation at low damage levels, although the strong intermixing implied a large interfacial (FexOy) and (Fe, Mg)Oy layer respective at Fe3O4–Fe and Fe–MgO interface. The high ion fluence of 3.8  ×  1016 Kr cm−2 has induced a complete oxidization of the buffer Fe layer. Under such Kr fluence, the stoichiometry of the Fe3O4 surface layer was still preserved indicating its high stability. The entire film contains FexOy -type composition at ion fluence large than 5.0  ×  1016 Kr cm−2.


Introduction
Iron oxides, magnetite (Fe 3 O 4 ), maghemite (γ-Fe 2 O 3 ), hematite (α-Fe 2 O 3 ) and wüstite (FeO x ), form naturally of Fe-containing rocks. They play an important role where they exist everywhere (in rocks (both on lands as well as in oceans), soils and dust in the atmosphere) and have integrated in many biological systems including the human body tissues. Magnetite (Fe 3 O 4 )-the oldest known magnetic material has been studied since the early days of science. Since the 1960s magnetite has been investigated extensively due to its high potential for technological applications in many important fields, e.g. high-density recording media and catalysis [1]. Magnetite crystallizes into a cubic inverse spinel structure and is a ferrimagnet below 858 K (T N ) with a net magnetic moment of 4.1 µ B per formula unit. The coexistence of Fe 2+ and Fe 3+ ions in the octahedral sublattice leads directly to many interesting properties, e.g. the Verwey transition around 123 K [2][3][4]. In recent years, magnetite has attracted again much attention, since it is referred as a half-metallic mat erial having a predicted full spin polarization at the Fermi level and thus considered as a promising candidate for spintronic devices at room temperature utilizing the spin polarised current [5,6]. Many attempts have been focused on determining Advances in Natural Sciences: Nanoscience and Nanotechnology
In the last few decades, an increasing interest is focused on the iron oxide surfaces, since they play a major role in e.g. corrosion, catalysis, spintronics and biomedicine [19][20][21]. In particular a large attention is paid to magnetite films due to their potential application as spin dependent transport devices. Especially, up to 72% mobile electrons were found to be polarized in Fe 3 O 4 (0 0 1)/MgO(0 0 1) films [22]. Besides, using thin films opens an exciting road for future research of surface engineering through absorption, which may permit tailoring of the interfacial spin polarization in magnetitebased spintronic devices [23]. The best understood iron oxide surface at present is probably Fe 3 O 4 (0 0 1) surface; its structure is known and the major properties are well characterised [24]. The best technique to grow Fe 3 O 4 (0 0 1) thin films is the reactive molecular beam epitaxial (MBE) deposition utilizing MgO(0 0 1) substrate [25]. For a review of surface termination and reconstruction of Fe 3 O 4 surfaces, see references [7,[24][25][26][27][28].
The physical and structural properties of interfaces play a crucial role in obtaining well-orientated thin films. Not only the surface but also the interface can be performance-limiting the practical applications of magnetite. Besides, while a large effort is focusing on the magnetite film surfaces, their interfaces have been less studied. It is thus important to investigate the interfaces and their correlation to the entire-film properties. Moreover, regarding the possible technical applications, it is worth obtaining the knowledge of the stability of the magnetite films in the external conditions, such as high temperature (annealing) and ion irradiations. In fact, there is a lack of the knowledge of interdifussion and intermixing effects in magnetite films. To the best of our knowledge, there is no report on such a research for Fe 3 O 4 thin films, except of our publications [29][30][31]. There existed some reports devoted to using the swift heavy ions to modify the magnetic and transport properties of magnetite films [32,33].
The goal of the present work is to use MeV ion beams to study and modify the structure, composition and properties of magnetite films especially at the interfaces. We are interested in finding the ion fluence upon which the film structure and the Fe 3 O 4 surface layer can be well preserved under ion irradiations. We have exploited Rutherford backscattering (RBS) and RBS-channeling (RBS-C) methods, since they allow the detection of elements in the sub-monolayer range with an atomic depth resolution typically in the nanometers range. In particular, RBS-C can provide the information about the crystalline quality (crystalinity) of the films. The main subject of our investigations are the bi-layer Fe 3 O 4 /Fe/MgO(0 0 1) films. Our results supported the earlier observations [27,34] that by using a two-step deposition procedure, in which the Fe 3 O 4 layer was deposited reactively on an epitaxial Fe(0 0 1) buffer layer grown epitaxially on MgO(0 0 1) substrate, we could always obtain the stoichiometric Fe 3 O 4 on the film surface. Moreover, such a layer was found to remain stable upon thermal annealing as well as in exposing to air. However, RBS-C data revealed that the large lattice mismatch at both Fe 3 O 4 /Fe and Fe/MgO interface (of about 4%, in a comparison with 0.31% for Fe 3 O 4 /MgO one) has degraded the film crystallinity [31]. Our up-to-date investigations have been mostly carried out for very thin layers (with a layer thickness of 10-20 nm). In this work, we extend our investigations of the intermixing on the bi-layer Fe 3 O 4 /Fe/MgO(0 0 1) films with a larger film thickness (in the range of 25-150 nm) under 1 MeV Ar + and Kr + ion irradiation. We expect to induce a significant intermixing (i.e. to tailor the films and their interfaces) by Kr + ions. It allows gaining a better understanding of the film stability and the interface stoichiometry and properties.

Experimental
We prepared and performed investigations on the following films: . By analyzing and comparing the results obtained for films with either a similar layer structure but with different layer thicknesses or with a similar layer thickness but with a different layer structure, we expect to gain a deep understanding of the atomic transport process. These single-layer magnetite films possess a high crystalinity. However, Mg out-diffusion from MgO substrate into the Fe 3 O 4 film implies that entire film or a part of the film has a spinel composition. Indeed, Mg was found to segregate to the surface for the thin film with a thickness <25 nm.
The chemical composition and layer structure of the films in the as-grown state have been determined by a combined Rutherford back-scattering (RBS) and x-ray reflectometry (XRR) analysis. The RBS spectra by using 2 MeV He + ions at a backscattering angle of 171° were collected and evaluated at different tilt angles ϕ in the range of 0-45° [28]. SIMNRA simulation program was used for RBS data analysis, taking into account the electronic stopping power data by Ziegler and Biersack, Chu Yang's theory for electronic energy-loss straggling and Andersen's screening function to Rutherford cross-section [35]. XRR was performed by using a Seifert two-circle diffractometer (using a rotating anode with 40 kV and 120 mA, and a LiF monocromator and a slit system for separating the Cu-K α1 line) with the Seifert reflectivity software for the data analysis. One of the best techniques to study the film crystalline quality (crystallinity) is the RBS-C. In our study, the RBS-C spectra (by using 1.6 MeV He + ions and a backscattering angle of 160°) were collected and evaluated at every tilt-angle step of 0.04°. Such experiments have been performed only on selected films exhibiting a high crystallinity and channeling effect.
The ion beam mixing experiments were carried out using 1 MeV Kr + beam with ion fluence (φ) in the range of 1.6 × 10 15 -3.8 × 10 16 Kr cm −2 (some φ -values are listed in table 1). The sample temperature was kept at −50 °C during irradiations. Each film after each ion irradiation was analyzed by RBS. We chose such ion type, energy and ion fluence to optimize the ion beam mixing effect at the Fe 3 O 4 -MgO, Fe 3 O 4 -Fe and Fe-MgO interfaces in the investigated films with chosen layer thickness mentioned above. Prior to the experiments, all irradiation parameters have been pre-estimated by using the stopping and range of ions in matter (SRIM) simulation programs [36]. Namely, the values for electronic energy loss (S e ), nuclear energy loss (S n ) and the range of ion (R p ) are found to be 1.01 keV nm −1 , 1.47 eV nm −1 and 334 nm for Fe 3 O 4 , respectively. For Fe, the values S e , S n and R p are 1.36 keV nm −1 , 1.99 eV nm −1 and 241 nm, respectively. For both materials, S e and S n is in the same magnitude and R p is larger than the film thickness. Such ion fluences used in our experiments produced low levels of irradiation damage with evaluated displacement-per-atom (dpa) in the range of 0.00007-0.0017. The results of XRR and RBS indicate that the initial composition and thickness of the films remain unchanged after irradiation with φ < 10 15 Kr cm −2 . Using Kr + ions, for obtaining a similar intermixing level, the irradiate time would be much shorter than Ar + ions and thus we could avoid the contamination to the films (since the pressure of the standard irradiated chamber is maintained at only around 10 −6 mbar).
The RBS, RBS-C and irradiation experiments were performed at the Institute of Nuclear Physics of the Johann von Goethe University in Frankfurt/Main (IKF-Frankfurt). We also used the second RBS equipment 5 with 1.7 MeV He + ions from a Tandetron accelerator at a scattering angle 170° and the detector out-of-plane arrangement. The sample can be rotated around the ϕand θ-axis in the small range of angles enabling to visualize the image scan of the planar channels and indicate the axial one in order to find some aligned RBS spectra. With such an arrangement, we could perform additional investigations on selected irradiated films for a structural analysis focusing on the changes that occurred after Kr + ion irradiation.

Results and discussions
Each film was irradiated 1, 2, 3 or 4 times marked respectively as Kr1, Kr2, Kr3 and Kr4 (given in table 1 and in the figures), with the ion fluences in the range of 1.6 × 10 15 -1.9 × 10 16 ions cm −2 for Fe 3 O 4 /Fe(25-25 nm) and Fe 3 O 4 / Fe(100-50 nm), whereas for Fe 3 O 4 /Fe(50-50 nm) film we used higher ion influences in the range of 6.7 × 10 15 -3.8 × 10 16 ions cm −2 (Kr4). Kr + irradiations with similar ion fluencies were performed on two single-layer film Fe 3 O 4 (25 nm) and Fe 3 O 4 /Fe(50 nm). The layer-thickness and composition of different layers of films in the as-grown state and after Kr + ion radiation (by different ion fluencies) estimated from RBS data analysis are given in table 1. We present only selected data revealing a visible effect of intermixing. In general, irradiations of the magnetite films by 1 MeV Kr + ions induce a strong Fe-Mg intermixing leading to a large change in the layer composition and thickness and especially a formation of large interface zones. We present in figure 1 the comparison of RBS spectra in the as-grown state and after Kr + ion irradiation for Fe 3 O 4 /Fe(50-50 nm) film, for which the ion beam mixing effect is most visible to eyes. We notice here that, not only single-layer magnetite film, but also bi-layer Fe 3 O 4 films exhibited a strong channeling effect set at untilted angle (ϕ = 0°) observed even if by means of a standard RBS apparatus [37], when it happened that the incoming ion beam could go forwards along the channels formed along the main crystallographic (0 0 1) direction of magnetite, implying a large decrease of the backscattered signal in the RBS spectrum. For ensuring the randomness, most of RBS spectra were collected   The absolute values of the layer-thickness and density were determined from XRR measurements. In general, a good agreement for the layer thickness was obtained between RBS and XRR results. In all cases, the densities of the magnetite layers and Fe layers are found to be respectively ρ (   We notice here that, despite of the fact that we could observe visible features in the RBS spectra, we should be careful in getting a reliable analysis in nm range, i.e. in the range of the depth resolution of a standard RBS method. On the other hand, both RBS and XRR results provided no evidence for an excess Fe within the Fe 3 O 4 or for a stoichiometry gradient in the as-grown films. One might expect that growing the Fe 3 O 4 film on an Fe buffer layer would lead to a film with excess interstitial Fe, but it is certainly not our case (for thick films). The excess Fe was found to exist, but only in a thin interfacial layer (3)(4)(5)  Due to the formation of the mixed layers and large interface zone as a consequence of ion-induced intermixing, the films have no longer any clear layer-layer separation between different layers. Namely, Fe, O and Mg are present everywhere. (Mg is absent only in the stoichiometric Fe 3 O 4 layer on the surface of the bi-layer films). The only difference between different layers or sub-layers is the difference in the element composition. However, the difference in the mass density of those mixed layers is small. Thus we could not obtain any reliable XRR results for irradiated films. Besides, the crystallinity is low and we were unable to obtain any reliable RBS-C results.
The pure Fe 3 O 4 layer and the bi-layer structure of other films were found to survive under Kr + ion irradiation, although the layer thickness decreases largely (see table 1). The damage together with a large layer roughness, inhomogeneous mixed elements in the layer and between different layers, etc would certainly be the source of a large uncertainty of our RBS dataanalysis for irradiated films. In other words, one should not trust completely the values themselves, but rather the meaning which they brought (the possible changes). Our investigations indicated that RBS measurements and data analysis provided a very clear picture of the ion mixing effect. We show in figure 4 the RBS spectra of three irradiated films after third or fourth irradiation. For a comparison, we show the normalized spectra and also the data for the single-layer Fe 3 O 4 (50 nm) film. For the Fe 3 O 4 (50 nm) as well as Fe 3 O 4 /Fe(25-25 nm) film, only one (overlapped) Fe peak was revealed, the ion beam mixing after the third irradiation was exhibited by a large decrease of the Fe peak intensity and a visible non-zero background between the Fe peak and Mg edge. For Fe 3 O 4 / Fe(100-50 nm) film, the distinguished Fe peak (maximum) and the wide shoulder feature were still observed after the fourth irradiation, indicating that the main bi-layer structure was still existed, despite that the layer thickness decreases largely. Namely, the layer-thickness remains as 63% and 69% respectively for Fe 3 O 4 and Fe layer with respect to 100% as that in the as-grown state (table 1). No such distinguished Fe   peak was observed for Fe 3 O 4 /Fe(50-50 nm) after the fourth irradiation by a twice larger ion influence, since the pure Fe layer was no longer existed (or in other words the Fe layer was oxidized completely). The layer-thickness of Fe 3 O 4 layer remains as 20%, while a value of 0% was found for Fe layer. For both those two films, a much larger non-zero background between the Fe peak and Mg edge was observed. Some small step-decreases were revealed attributed to a small deviation of element composition in different sub-layers (i.e. Fe redistribution into the MgO depth).
RBS and XRR data analysis for Fe 3 O 4 /Fe(100-50 nm) after (Kr4) irradiation provided three distinguished layers: Fe 3 O 4 layer on the surface, the Fe layer and in between those two pure layers is the Fe x O y layer (table 1). Since each layer has a large enough layer-thickness, we decided to perform additional RBS experiments with titled samples using a different RBS equipment 7 on this irradiated film to look for a possible aligned RBS spectrum, despite of the fact that the film certainly has a low crystallinity. The results were shown in figure 5. The random RBS spectrum has revealed similar features as that collected by using the first RBS equipment revealing only Fe peak and a wide shoulder at the right hand side. By turning the sample, at a certain proper titled angle, we have observed a channeling effect through different layers. Namely, we obtained the aligned RBS spectrum exhibiting three peaks structure attributed to the three identified layers.
It is worth simulating the damage depth profiles induced by Kr + ion irradiations. An energetic ion penetrating through a Fe 3 O 4 and/or Fe generates vacancies in the Fe and O sublattices as well as Fe and O interstitials. The vacancies depth distribution in the particular layers can be evaluated by the Monte Carlo simulation of the ion interaction with matter using SRIM code [35]. Despite of the fact that SRIM theoretical calculations of atomic displacements take into account only ballistic processes and completely neglect dynamic ones as well as do not take into account crystallinity, SRIM simulations show the general trend and the damage propagation through the layers. In figure 6 we presented SRIM simulations with the full damage cascade for 1 MeV Kr + ions in Besides, the Fe vacancies depth profile in Fe buffer layer has the same shape as that we observed in the RBS spectrum in figure 5, taking into account the fact that SRIM included two layers only (Fe 3 O 4 (100 nm) and Fe(50 nm)) and not Fe x O y intermixed layer created during the irradiation. The O and Fe vacancies depth profile in the first layer has in fact the same shape, thus are not recognizable in the figure 5, but it can be deduced from the total vacancies depth profiles, that the total number of vacancies is about twice higher as the sum of Fe and O vacancies.
For a clear illustration of the ion beam mixing effect, i.e. the change in the composition and thickness of the layers, we show in figure 7 the film diagram before and after (selected) 7 See footnote 5. irradiations. We draw the layer thickness in a millimeter scale while keeping the same proportional portion with respect to the values estimated from SIMNRA in nm for different layers of the films (see table 1). The original separation (i.e. in the as-grown state) between the film and MgO substrate is indicated by the solid lines. Different colours indicate different layer compositions. In order to have a good view regarding the change of the layer thickness and composition, we show only the main layers. For instance, we presented only one mixed Fe x O y layer, although the RBS data analysis provided that such a layer consisted of several sub-layers with a small deviation of x-and/or y-component. Besides, we used the straight lines to show the real values of the estimated layer thickness of the mixed layers deeply into the MgO substrate, while for a large layer-thickness of the mixed layers we show the reduced thickness presented by curved lines. It shows clearly that the pure Fe 3 O 4 layer in all investigated bi-layer films were wellpreserved, while that for the single layer films could not survived under 1 MeV Kr + ion irradiations.
The dependence of the relative changes of the layer thickness (in %) with increasing Kr fluence is shown in figure 8. As we mentioned earlier, irradiation with Kr fluence below 10 15 Kr cm −2 did not lead to any visible change in the RBS spectra. However, the relative layer thickness was found to decrease rapidly upon Kr fluence higher than 10 16 Kr cm −2 . The ion fluence upon which the Fe buffer layer disappeared determined experimentally was 3.8 × 10 16 Kr cm −2 . From a polynomial fits through the data points, the stoichiometry of the surface Fe 3 O 4 layer is expected to be destroyed (i.e. the composition is changed into Fe x O y , i.e. x ≠ 3, y ≠ 4) upon irradiation with a larger fluence >5.0 × 10 16 Kr cm −2 .

Conclusions
Our investigations showed that the stoichiometry of the Fe 3 O 4 layer on the surface of the bi-layer film (the magnetite-on-Fe film) is well preserved under 1 MeV Kr + ion irradiation at a large ion fluence φ = 3.8 × 10 16 ions cm −2 , indicating a high stability of magnetite layer in external conditions such as Kr + ion irradiation. This information is crucial for using the magnetite-based devices in spintronics. Besides, the bi-layer structure (Fe 3 O 4 /Fe film-structure) was also preserved under ion irradiation with lower ion fluencies.
1 MeV Kr + ion irradiation has induced a large interfacial zone resulting in a more than double thickness of the films. It indicates a possibility of using the high-energy ion beam in controlled experiments for tailoring of the magnetite films and their interface engineering to obtain the layers or the interfaces with required thickness and desired properties for the practical applications.