Magnetic response of FeRh to static and dynamic disorder

Atomic scale defects generated using focused ion as well as laser beams can activate ferromagnetism in initially non-ferromagnetic B2 ordered alloy thin film templates. Such defects can be induced locally, confining the ferromagnetic objects within well-defined nanoscale regions. The characterization of these atomic scale defects is challenging, and the mechanism for the emergence of ferromagnetism due to sensitive lattice disordering is unclear. Here we directly probe a variety of microscopic defects in systematically disordered B2 FeRh thin films that are initially antiferromagnetic and undergo a thermally-driven isostructural phase transition to a volatile ferromagnetic state. We show that the presence of static disorder i.e., the slight deviations of atoms from their equilibrium sites is sufficient to induce a non-volatile ferromagnetic state at room temperature. A static mean square relative displacement of 9 × 10−4 Å−2 is associated with the occurrence of non-volatile ferromagnetism and replicates a snapshot of the dynamic disorder observed in the thermally-driven ferromagnetic state. The equivalence of static and dynamic disorder with respect to the ferromagnetic behavior can provide insights into the emergence of ferromagnetic coupling as well as achieving tunable magnetic properties through defect manipulations in alloys.


Introduction
Equiatomic FeRh is well-known for its unique properties, including a rst-order metamagnetic isostructural transition from a low temperature antiferromagnetic (AFM) to a high temperature ferromagnetic (FM) phase, at the transition temperature T tr of approximately 370 K. Potential applications include heat-assisted magnetic recording (HAMR), 1 magnetotransport, 2 antiferromagnetic spintronics 3 and magnetic refrigeration. [4][5][6] For this purpose, efforts are made to engineer the hysteresis, for example by growing epitaxial thin lms on different substrates, 7-10 varying the strain affected volume by modifying the lm-thickness 11,12 or doping the system to shi the phase transition towards lower or higher temperatures. 13,14 During this metamagnetic phase transition of equiatomic FeRh the local Fe and Rh moments increase from AE3 m B and 0 m B in the AFM phase to 3.3 m B and 1.0 m B , respectively, in the FM phase, accompanied by a volume increase of 1%. [15][16][17] Apart from the isostructural phase transition, B2-FeRh also possesses a disorder-induced phase transition, which can be driven, for example, by ion or laser irradiation. 18 This process has been known for FeRh and was initially shown by Iwase et al., 19 where irradiating a bulk FeRh sample with Ni, Kr, Xe or Au ions, leads to a nite orbital polarization measured by X-ray magnetic circular dichroism (XMCD) at the Fe K edge below room temperature. Systematic disordering is achieved via irradiation with light noble gas ions, such as He + or Ne + with sufficient energy to cause displacements of the Fe and Rh atoms from their ordered sites without articially doping the system with materials possessing free electrons. Recently, similar investigations have been performed for FeRh thin lms irradiated with He + (ref. 20) and Ne + (ref. 21 and 22) showing, for low irradiation uence, an increase of the magnetization at low temperature. For bulk FeRh alloys similar investigations have been performed, where the magnetization depends strongly on the ion species. [23][24][25] Previously, only microscopic investigations of the disorder-induced ferromagnetic phase in FeRh were performed with measurement techniques, which integrate all magnetic contributions, making it dicult to separate contributions from the different crystallographic sites or phases in metallic systems.
In comparison, a similar effect occurs in Fe 60 Al 40 , where the recombination of vacancies in ion irradiated B2-ordered Fe 60 Al 40 thin lms leads to the formation of a disordered bcc structure (A2 structure) with ferromagnetic ordering at room temperature. 26,27 Extended X-ray absorption ne structure spectroscopy (EXAFS) measurements at the Fe K edge reveal the formation of Fe and Al-rich regions with increasing irradiation uence, accompanied by an expansion of the lattice. 28 Additional investigations of FeRh showed the existence of a new monoclinic ground state, 29-31 due to a lattice instability. As evident from the Fe-specic vibrational (phonon) density of states (VDOS, g(E)) the low energy phonon modes of the low temperature AFM phase have a higher g(E) compared to those of the FM phase. 31 This means that (at the same temperature) the lattice of the AFM phase is soer with respect to phonons than the lattice of the FM phase. 31 This is also revealed by the Lamb-Mössbauer factor (f-factor) and by the average atomic force constant, which both are lower for the AFM phase than for the FM phase. 31 The lattice soening was also revealed in a temperature-dependent EXAFS study performed by Wakisaka et al. 32 by observing a discontinuity of the mean square relative displacement along the phase transition. Recently, FeRh gained increased interest due to the possibility to tailor the metamagnetic phase transition by inducing external strain. This can be achieved by growing FeRh thin lms on a ferroelectric substrate (for example on PMN-PT or BaTiO 3 ). [33][34][35][36][37][38] In a recent work of Keavney et al., it could be shown by a combination of XMCDphotoemission electron microscopy (PEEM) and nano-XRD measurements, that the magnetostructural phase transition exhibits a defect-driven domain nucleation behaviour. 39 Similar effects of an inhomogeneous phase transition have been observed in a transmission electron microscopy (TEM) study of Gatel et al., 40 where the lm-surface and the lm-substrate interface have a lower transition temperature than the centre of the lm. Saidl et al. 41 observed that the optical properties of different microscopic regions possess different transition temperatures leading to a distribution of transition temperatures T tr . Furthermore, these results can be used to explain a stable FM phase located at the surface, which has been shown by Pressacco et al. 42 In this work, we investigate the effect of static disorder compared to dynamic disorder on the metamagnetic phase transition of FeRh. Dynamic disorder is dened as the motion of nuclei for example by means of zero-point vibrations at low temperatures or at higher temperatures by lattice vibrations (phonons), leading to an oscillating atomic displacement. This kind of disorder is strongly dependent on temperature. In contrast, static disorder is temperature independent and is created, for example, at the time of the crystallization process during sample preparation, effectively minimizing the formation energy. Static disorder can, for example, occur in the form of mono-vacancies, vacancy cluster, anti-site contributions (nuclei on a different crystallographic site) or a grain boundary. These kinds of defects can lead to a change of the microscopic physical properties, for example, an anti-site contribution or a vacancy effectively change the exchange interaction. This can result in the formation of ferromagnetism in an otherwise non ferromagnetic system. 43 On the other hand, a grain boundary or void can generate a distortion of the lattice, effectively varying the nearest neighbour distance. In a time-averaged microscopic picture, these static variations of the nearest neighbour distance have the same effect as lattice vibrations leading to a blurred distribution of bond lengths.
For a detailed investigation of the difference or similarities of dynamic and static disorder, it is necessary to investigate the system on a microscopic scale by considering the electronic structure, local structure and the defect structure. We combine magnetometry measurements with conversion electron Mössbauer spectroscopy (CEMS) in order to relate changes in macroscopic magnetization to the local magnetic properties and the electronic structure of the system. With this approach, we can separate changes of the electronic structure on the microscopic scale and correlate these with changes of the macroscopic scale. Also, changes of the local structure by element-specic extended X-ray absorption spectroscopy at the Fe K edge will be presented, while an interlink between defects and the disordering process will be highlighted by positron annihilation spectroscopy (PAS). We show that for an ion irradiation uence below 0.4 Ne + per nm 2 an equivalence of the AFM-FM transition occurs compared to a temperature-induced phase transition. By freezing in the ferromagnetic state, the mechanism for the emergence of the ferromagnetism can be explored in greater detail.

Macroscopic magnetic properties
Field dependent measurements performed at 300 K ( Fig. 1a) show, for a MBE grown 40 nm FeRh thin lm, with a residual A1 phase, a small saturation magnetization of 25 emu cm À3 consistent with an AFM-ordered system with a small fraction of non-compensated Fe-spins particulary at the surface. 42,44 For the irradiated samples, a hysteresis divided into two parts is observed. In small applied magnetic elds a hysteresis occurs with a coercive eld below 50 mT, while for higher magnetic elds a second loop occurs which closes in the case of applying a eld of 9 T. This loop can also be observed in the ordered samples at temperatures below the phase transition by applying a large magnetic eld (e.g. H z 11 T for T z 300 K) 5,45 and can be interpreted as the eld induced isostructural AFM-FM phase transition. 46,47 Temperature-dependent magnetization measurements show, that for the initial B2-ordered sample the rst-order phase transition occurs at 370 K with a symmetric thermal hysteresis width of 10 K (see Fig. 1b). The increase of the magnetization at low temperatures originates from defects and impurities 48 in the MgO substrate. With increasing ion irradiation uence, the initial sharp transition occurs at lower temperature and the hysteresis width increases, while additionally, the magnetization at low temperatures increases.

Microscopic magnetic properties
Mössbauer spectroscopy probes slight deviations in the energy levels of the 57 Fe-nuclei due to the presence of hyperne interactions, and measures the valence state of iron and the orientation of the Fe-spin relative to the incident g-ray. Since the hyperne eld may be roughly proportional to the magnetic moment 49 changes of the microscopic Fe moment can be probed. Room temperature zero-eld measurements show two distinct states in the B2-FeRh lm (see Fig. 2i). The majority phase can be described by a magnetic split sextet state with peaks of the lines 1 and 6 at À4.2 mm s À1 and +4.1 mm s À1 respectively, corresponding to a hyperne eld B hf ¼ 25.4 T. This is in good agreement with different Mössbauer investigations of AFM B2-ordered FeRh. 15,50 The intensity ratio of lines 2 (5) and 3 (4), also referred to as A 2,3 -ratio describes the average angle q between Fe-spin and incident g-ray, by the formula.
Therefore, the A 2,3 -ratio exhibits values between 0 (Q ¼ 0 with spin orientation out of plane) and 4 (Q ¼ 90 with spin orientation in-plane). From the performed measurements, we determine an A 2,3 -ratio of 3.9, corresponding to an almost inplane spin orientation. The second contribution, (a single peak at about À0.1 mm s À1 ) can be attributed to A1-FeRh, 15,51,52 which exhibits paramagnetism at RT. Both spectral contributions show a small negative isomer shi d iso of À0.01 mm s À1 , close to that of bulk bcc-Fe, which represents the metallic character of the sample and no oxide contribution (d iso (oxide) > 0.3 mm s À1 ) is present. Upon rising temperatures ( Fig. 2(ii)), the hyperne splitting of the sextet decreases and close to the transition temperature at 380 K a third contribution appears, which indicates the isostructural AFM-FM transition. As is seen for temperatures below 380 K a decrease of the hyperne eld occurs, expected for an AFM-system approaching its Néel temperature T N , and at 375 K the third phase occurs, leading to an increase of the average hyperne eld by 1.4 T. The third phase with the increased hyperne eld is assigned to the FM phase of B2-FeRh. 15 The magnetic splitting of this third magnetic state also decreases upon heating. By using a Brillouin-function to describe the temperature dependence of the average B hf , one can determine an extrapolated Néel temperature T N ¼ 615 AE 17 K for the rst phase and a Curie temperature T C ¼ 662 AE 13 K for the third phase, the latter being in reasonable agreement with the Curie temperature obtained by magnetometry on thin lms (T C ¼ 670 K) 53 and bulk materials (T C ¼ 675 K). 54 In the chemically disordered system aer irradiation with 25 keV Ne + with a small ion uence ( Fig. 2(iii)), changes in the magnetically ordered state are observed. From the initially ordered state, it is seen that an additional spectrum arises leading to a broadening of lines 2 and 5 combined with an additional ne structure at lines 1 and 6 corresponding to a hyperne eld of 27.4 T. With increasing irradiation uence the relative spectral area of this phase increases, which leads to an average hyperne eld of 27.4 T for the maximum irradiated sample. Hence, anti-site Fe (Fe on an initial Rh site with B hf (Fe II ¼ 38 T)) is not created, as one can observe for a Fe-rich B2-FeRh sample, and also the formation of the A2 phase 15 (B hf (A2) ¼ 35 T) does not occur. As discussed for the Mössbauer results, one can see that the samples have different amounts of A1-FeRh (blue central singlet), showing an inhomogeneous distribution of this impurity phase, 52 while for the present low irradiation uences a formation of additional A1-FeRh does not occur. [20][21][22] All of the measured magnetically split spectra were described by a hyperne eld distribution, which is presented for the temperature-driven phase transition and irradiation induced transition in Fig. 3. By dening the macroscopic remanent magnetization as an order parameter of the individual phase transition, one can compare the changes of the microscopic 57 Fe hyperne eld for a system with a thermally driven phase transition (Fig. 3a) to a systematically structural disordered system (Fig. 3c). The plots for the dependence of the microscopic hyperne eld as a function of the macroscopic magnetisation are shown in Fig. 3b and d. From this comparison we nd that by a systematic increase of the structural disorder, a FM-phase with a hyperne eld of 27.4 T is induced (Fig. 2(iii)), which corresponds to the hyperne eld found at room temperature for ferromagnetic B2 ordered Fe 51 Rh 49 . [15][16][17] These measurements demonstrate, that the metamagnetic isostructural phase transition can be driven by ion irradiation with low uences, as the changes of the microscopic moment as a function of the macroscopic remanent magnetisation show a similar trend compared to a temperature-driven system (Fig. 3).

Microscopic local structure
EXAFS measurements have been performed at the Fe K edge at low temperatures (T ¼ 5 K) to identify changes of the microscopic local structure. For the analysis of the measured spectra and to resolve the ne structure c(k) the DEMETER package tool 55 has been used. For the Fourier transformation of the signal a Kaiser-Bessel window function (Fig. 4a) has been used starting at k min ¼ 2.3Å À1 and ending at k max ¼ 13.3Å À1 (Dk ¼ 11 A À1 ) with a width of dk ¼ 0.1Å À1 . The Fig. 4a shows a small decrease of the amplitude in the ne structure, while no clear change of the oscillations can be discovered by comparing an as-grown lm with a lm irradiated with 0.4 Ne + per nm 2 .
In comparison, a similar c(k) has been observed for FeRh in the work of Wakisaka et al. 32 by temperature-dependent EXAFS measurements at Fe and Rh K edge. The amplitude of the Fourier Transform is illustrated in Fig. 4b for the two different states. The proles are very similar, with an overall slight decrease in the intensity of the peaks. The rst peak between 1.5-3Å originates from back-scattering from the nearest

Characterisation of open volume defect concentration
Doppler broadening variable energy positron annihilation spectroscopy (DB-VEPAS) probes the open volume defects in a solid, where one detects changes of width and intensity of the 511 keV positron annihilation line. The S-parameter is the fraction of positrons annihilating with low momentum valence electrons. It represents vacancy type defects and their concentration. Plotting the calculated S parameter as a function of positron implantation energy, S(E), provides depth dependent information on the defect concentration. 58 For the analysis of positron diffusion length, L + , which is inversely proportional to defect concentration, the VEPFit code 59 has been utilised, which permits to t S(E p ) curves for multilayered systems and to acquire thickness, L + , and the specic S-parameters for each layer within a stack. L + has been calculated as 19.7(1), 7.27(4), and 5.68(4) nm for the as-grown, irradiated with 0.1 Ne + per nm 2 , and 0.125 Ne + per nm 2 lms, respectively. The thickness of the lm was xed to 42.8 nm for all three samples, utilizing a material density of r FeRh ¼ 9.76 g cm À3 for FeRh and r MgO ¼ 3.6 g cm À3 for MgO respectively. L + for MgO was xed to 35 nm for all the samples (the tted values: 33-37 nm). The as-grown sample, because of a large L + , has a relatively low defect concentration nearly as low as the MgO crystal. Ion irradiation produces damage to the lm structure introducing vacancy like defects. Most likely, the size of defects remains the same and only the defect concentration increases (by a factor of 4 for the largest uence). DB-VEPAS clearly shows an increase in the defect concentration, Fig. 5. This implies that the fraction of the open volumes, the empty spaces in the lattice increase with increasing uence. However, the defect type, whether purely static disorder or vacancy like defects, cannot be ascribed from the Doppler Broadening results alone. Nevertheless, the increase of the open volumes is consistent with the increased static disordering observed in the ion-irradiated lms, as determined from the EXAFS measurements, discussed in the next section.  From the performed EXAFS measurements it is evident that a change in the chemical composition of the nearest neighbour shell upon ion irradiation is not detected. A variation of the chemical composition that can give rise to the onset of Fe-rich regions due to intermixing and would lead to changes in the relative intensity of specic features in the Fourier transform, as it is the case for example in ion irradiated Fe 60 Al 40 thin lms does not occur in FeRh. 28 This is in contrast to XRD measurements, 22 where intermixing has been suggested due to an observed decrease of the long-range order parameter S with rising ion irradiation uence. In fact, the spectra of the disordered sample can be simulated by analyzing the effect of a static mean square relative displacement (MSRD) s stat 2 of 9 Â 10 À4 A À2 on the spectra of the ordered (as-grown) sample, while changes of the dynamic contributions can be neglected due to the low measurement temperature (T ¼ 5 K). This shows that the Ne + -irradiation causes an increase of the MSRD. This can be due to the increased structural disorder and lattice distortions, for example, by trapped Ne ions inside the lattice 20 or a decreased grain size, increasing the static disorder. DB-VEPAS reveals such an increase of the defect concentration leading to an increase of open volume.
On the other hand, in temperature-dependent magnetisation measurements, a ferromagnetic ordering at low temperatures is present for the ion irradiated samples, while this is accompanied by a broadening of the hysteresis and a decrease of T tr . From the eld and temperature-dependent measurements, it is seen that a coexistence of two different magnetically ordered states occurs. As shown in Fig. 1, the rst magnetic phase is a so ferromagnetic phase and can be attributed to the disorder-induced phase, while the second phase can be described to the initial hard magnetic B2 phase with an AFM ordering, where an increased eld breaks the anti-parallel spin alignment and induces the formation of FM ordering. A similar increase of the magnetisation at low temperatures has been observed previously, where, for higher irradiation uences, 21,22,60 a suppression of the AFM-FM transition and even a suppression of the ferromagnetic ordering was observed, while forming the paramagnetic A1 phase (fcc). The deviation in the increase of the high eld magnetisation as a function of the ion uence is due to the fact, that the two samples exhibit a different amount of a secondary A1-FeRh phase (as discussed before in Section 2.2). The secondary phase has not been considered in the magnetisation normalisation due to the unknown composition of the secondary phase. 61 Dening the remanent macroscopic magnetisation M r as an order parameter of the phase transition in the temperature-driven or ion-irradiation induced AFM-FM transition and comparing the dependence of B hf as a function of the remanent magnetisation, a similarity between the two distinct phase transitions is present (shown in Fig. 3c and d). For the temperature-driven phase transition for small M r the hyperne eld B hf decreases, while a secondary magnetic contribution occurs at 10 emu cm À3 from the ferromagnetic phase with the coexistence of both B hf contributions (AFM and FM) up to 100 emu cm À3 . From the temperature dependence for elevated temperatures using a Brillouin function a hyperne splitting B hf ¼ 27.4 T at 300 K is obtained, which corresponds to the hyperne eld splitting of FM Fe 51 Rh 49 at 300 K. 15 In the ion irradiated samples, an additional magnetic contribution occurs at 27.4 T, whose spectral contribution increases with rising ion uence and M r . The occurrence of this ferromagnetic phase can be explained, for example, by a change of the chemical composition and the formation of Fe-rich regions (Fe nucleus on a Rh-site). Such a change is not persuasively present in the Mössbauer spectra (maximum 0.6 Fe-at%), consistent with the EXAFS measurement. Therefore, the formation of the ferromagnetic ordering needs to be explained by a different process.
In the thermally induced phase transition a discontinuous increase of the short-range disorder (MSRD 32 or Lamb-Mössbauer factor 31 ) occurs, while in the ion irradiated sample an increase of the static disorder can be seen, indicated by the increased MSRD s 2 (EXAFS). This is due to induced grain boundaries or lattice distortions owing to ion irradiation. This increase of the defect concentration was conrmed in the positron annihilation spectroscopy (decreasing L + ). Combining the observed changes of the hyperne eld splitting towards an identical splitting in the FM phase at 300 K and an increase of the static disorder in EXAFS, while changes of the chemical composition are absentthe effect can be explained in such a way that a structural defect breaks the local symmetry. Hence, a modication of the electronic structure takes place, effectively lowering the transition temperature of the surroundings. This concept of a defect-driven domain nucleation growth in B2 FeRh was previously suggested by Keavney et al. 39 In summary, we found an increase of the static structural disorder in FeRh thin lms irradiated with low ion uences, while no substantial intermixing of Fe and Rh atoms can be observed on the local scale by microscopic probes such as CEMS or EXAFS. Besides, PAS-measurements show an increase of open volume defect concentration, while the size of the defects does not change. 57 Fe-CEMS measurements reveal a second magnetic phase with an increased relative spectral area with rising irradiation uence. The hyperne eld splitting of this second magnetic phase is identical to that of ferromagnetic B2-FeRh. By comparing the thermally-driven and the structural disorder-driven phase transition a jump-like behaviour of the remanent magnetization dependence of the hyperne eld can be observed. Thus, in B2 Fe 50 Rh 50 , slight deviations of the Fe and Rh atoms from equilibrium lattice positions are sufficient to induce a ferromagnetic onset. These lattice deviations can be dynamic, as realized through thermal excitations above the metamagnetic transition. The ferromagnetic onset can also be realized through irradiation-induced static disorder, which is a snap-shot equivalent to the thermally driven dynamic disorder. Further investigations, for example by positron annihilation lifetime spectroscopy, may reveal the defect type and provide further understanding of the formation of irradiationinduced ferromagnetism in B2-FeRh.

Experimental
FeRh thin lms were grown by molecular-beam epitaxy (MBE) by co-deposition of 57 Fe-metal (95% enriched in the isotope 57 Fe) and Rh in ultrahigh vacuum (p growth ¼ 4 Â 10 À9 mbar) on a MgO(001) substrate. Before the deposition of the lm, the MgO(001) substrate was cleaned using isopropanol and was heated at 300 C for 60 min in a pressure of 1 Â 10 À9 mbar to remove contaminants from the surface structure. During the deposition, the temperature of the substrate was 300 C, while the deposition rates of 57 Fe and Rh were measured and controlled by independent quartz-crystal oscillators. Aer deposition the lm was in situ annealed at 700 C for 90 min to ensure the formation of the B2 structure. The thickness of the sample was conrmed to be 42.8 nm by X-ray reectivity. It is worthwhile to mention, that due to slight deviations of the composition it is possible to stabilize the FM phase below room temperature (for example in an Fe-rich sample consisting of Fe 51 Rh 49 ) a stable FM phase is present even at low temperatures. Therefore, it is necessary to have an optimal control of the deposition rates of the individual constituents. Furthermore, the application of a capping layer to prevent oxidation has been avoided based on the observations made in different works, 20,[62][63][64] where the capping layer inuenced the magnetic properties of the B2-ordered system in such a way that a ferromagnetic phase was observed at the interface between the lm and capping layer. Based on the performed CEMS and PAS measurements, we can neglect the formation of an oxide layer. In Mössbauer spectroscopy, an Fe-oxide (for example Fe 3 O 4 ) would be present with a hyperne splitting B hf close to 50 T and an isomer shi d iso around 0.3 mm s À1 (Fe 3+ ) or 1.0 mm s À1 (Fe 2+ ) at room temperature. 65 At the same time, FeO x (Wustite) is characterized by a paramagnetic asymmetric feature with an isomer shi of 1.0 mm s À1 . 66 This contribution is not detected in our Mössbauer spectra. Furthermore, from the positron annihilation spectroscopy, an additional oxide layer would result in a different dependence of the S parameter at positron energies E p below 1 keV, e.g. in the form of a plateau, a minimum, or a maximum. 67 Summarizing, all these spectroscopic features, which would hint towards an Fe-oxide layer, are not present in the studied samples.
The as-grown sample was cut into three pieces, while the rst piece was used to perform magnetometry and temperaturedependent 57 Fe-CEMS measurements, the second and third sample pieces were used for an initial irradiation with 0.05 Ne + per nm 2 and 0.075 Ne + per nm 2 respectively. Both irradiated samples were then further irradiated with 0.05 Ne + per nm 2 to achieve a total irradiation of 0.1 and 0.125 Ne + per nm 2 respectively. In addition, a sample with an irradiation dose of 0.4 Ne + per nm 2 was used for EXAFS measurements. For the different irradiation steps an ion energy of 25 keV has been chosen based on SRIM 68 simulations to fully penetrate the lm volume while minimizing defects in the substrate.
Magnetic characterization was performed using the vibrating sample magnetometer (VSM) option of a Quantum Design PPMS DynaCool, providing magnetic elds up to AE9 T applied parallel to the lm surface in a temperature range between 4.3 K and 400 K. Temperature-dependent magnetization measurements were performed in an external eld of 10 mT using the ZFC-FC protocol. To achieve a temperature of 450 K a ceramic sample holder and the VSM oven option (temperature range between 300 and 1000 K) was used, resulting in an offset at 300 K due to different sensitivity ranges. 57 Mössbauer spectroscopy at perpendicular incidence of the g-rays onto the lm surface was performed by detection of conversion electrons. For the detection of the electrons, the sample was installed in a proportional gas counter, i.e. housing with a continuous He gas ow mixed with 4% CH 4 to avoid ionization processes. For the measurement, a constant acceleration Mössbauer driving unit was used with a 57 Co source embedded in an Rh matrix, while the velocity of the spectrometer was calibrated with a a-Fe foil reference sample at room temperature. The experimental spectra were evaluated by a least-squares tting routine using the Pi program package. 69 Element-specic EXAFS measurements have been performed at the XAS beamline BM23 at the ESRF. 70 Spectra have been recorded at the Fe K absorption edge in the total uorescence yield detection mode at 45 incident geometry using an energy dispersive detector to detect the signal originating from the Fe K a -radiation.
Doppler broadening variable energy positron annihilation spectroscopy (DB-VEPAS) measurements have been conducted using the apparatus for in situ defect analysis 67,71 of the slow positron beamline 72 located at the Helmholtz-Zentrum Dresden-Rossendorf. Positrons were extracted from a radioactive 22 Na source, moderated down to several eV and magnetically guided to the accelerator unit, where they were subsequently accelerated in discrete voltage values in the range of E p ¼ 0.04-35 keV. This positron energy allows depth proling of the lms from the surface down to about 2 mm for FeRh, while a mean positron implantation depth can be approximated by a simple material density-dependent formula: z mean ¼ 3.