Observation of Skyrmions at Room Temperature in Co2FeAl Heusler Alloy Ultrathin Film Heterostructures

Magnetic skyrmions are topological spin-textures having immense potential for energy efficient spintronic devices. Here, we report the observation of stable skyrmions in unpatterned Ta/Co2FeAl(CFA)/MgO thin film heterostructures at room temperature in remnant state employing magnetic force microscopy. It is shown that these skyrmions consisting of ultrathin ferromagnetic CFA Heusler alloy result from strong interfacial Dzyaloshinskii-Moriya interaction (i-DMI) as evidenced by Brillouin light scattering measurements, in agreement with the results of micromagnetic simulations. We also emphasize on room temperature observation of multiple skyrmions which can be stabilized for suitable combinations of CFA layer thickness, perpendicular magnetic anisotropy, and i-DMI. These results provide a significant step towards designing of room temperature spintronic devices based on skyrmions in full Heusler alloy based thin films.

A novel magnetic pattern in which the magnetic moments exhibit a characteristic swirling configuration is referred to as a skyrmion [1][2][3] . Remarkably, the skyrmions are topologically stable/protected 4 structures such that once formed they cannot be deformed/distorted into a ferromagnetic (FM) spin texture or any other magnetic state(s). Due to these properties, skyrmions have immense potential for energy efficient spintronic devices, such as skyrmion based ultra-dense storage, race-track memories, magnetic random access memories (MRAMs), magnetic tunnel junctions, and skyrmion based spin torque oscillators, etc [5][6][7][8] . In addition to skyrmions, other magnetically ordered complex spin configurations, viz., vortices [9][10][11][12] and magnetic bubbles have also been studied intensively at the nanoscale by the magnetic community [13][14][15] . Very recently, the skyrmionic bubbles (which possess helicity and vorticity in contrast to chirality in the skyrmions) are proposed in non-centrosymmetric system Fe 3 Sn 2 which are of large dimensions compared to the skyrmions and have different textures 14,16 . Though skyrmions were first theoretically predicted in the early sixties, their experimental evidence had to wait until 2009 when their presence was reported in non-centrosymmetric B20 compounds like MnSi 1 , FeCoSi 17 , MnGe 18 and FeGe 19 . However, in them the film thicknesses were several tens of nm and Curie temperatures (T C ) were lower than 300 K, i.e., 35 K, 29 K, 270 K and 280 K, respectively. On the other hand, the present work is based on the Co 2 FeAl (T C ~1000 K 20 ) thin films having perpendicular magnetic anisotropy and skyrmions are observed in a robust manner at room temperature. Further, in all these non-centrosymmetric cases, the samples were cooled to a low temperature and/or required the presence of a high magnetic field to reveal the existence of the skyrmionic state. It was understood that the non-centrosymmetric structures present in these compounds resulted in a non-centrosymmetric spin exchange or the so-called Dzyaloshinskii-Moriya interaction (DMI) 21,22 . From the atomic scale perspective, the DMI energy (E DM ) is expressed as, , where  → D ij is the DMI exchange constant characterizing the interaction between the nearest neighbor spins possessing the atomic Scientific RepoRts | (2019) 9:1085 | https://doi.org/10.1038/s41598-018-35832-3 moments → S i and → S j . Initially, the DMI was considered to be of bulk origin, however, it was first proposed by Fert et al. 23 that the DMI can be induced in a heavy metal doped system or later by Bogdanov et al. 24 in multilayer structures by breaking the inversion symmetry at the interfaces between FM and nearby non-magnetic ultrathin layers. The so-induced DMI is referred to as interfacial Dzyaloshinskii-Moriya Interaction (i-DMI). Since the most striking effect induced by i-DMI is the spatial twist of the magnetization leading to the formation of skyrmions even at room temperature 25 , thus, i-DMI is expected to be more effective than the bulk DMI in the case of ultrathin films.
Several techniques have been employed to directly confirm the existence of skyrmions. These include sophisticated techniques such as neutron scattering, Lorentz transmission electron microscopy, X-ray holography, spin-polarized STM, etc. 1,[26][27][28][29][30] . The studies clearly suggest that the topological nature of the skyrmions makes them very exciting for applications in spintronic devices, primarily due to the giant reduction in the threshold current density by five to six orders of magnitude compared to the current density reported in currently employed FMs for driving the domain walls 31,32 . Even at low current density, the skyrmions are expected to move with high speed especially in materials with low Gilbert damping 33 such as Heusler alloys which are established for low Gilbert damping 34 . Fert's group has theoretically predicted the formation of skyrmions by means of micromagnetic simulations in confined media 3 . This non-trivial magnetic spin structure so formed is commonly defined by the skyrmion number, x y n n 1 4 , where n is the magnetization unit vector. Based on the magnetization evolution structure, the value of N sk is 1 for the existence of skyrmions, e.g., spin spiral 1 or hedgehog type 35 .
Recently, various groups have focused on the ultrathin magnetic film/non-magnetic heavy metal interface 36,37 to explore the formation of skyrmions. In these studies, there exists a plenty of room for varying the various control parameters, viz., material, thickness and the structure of the interfacial layer employed for the control of i-DMI, exchange and anisotropy constants, etc. By appropriate tuning of the control parameters, the strength of i-DMI can be tuned significantly 37 . Rohart et al. theoretically predicted the confinement of skyrmions in Pt/Co/AlO x ultrathin films nanostructures exhibiting i-DMI 38 . In recent reports, room temperature magnetic skyrmions in Co/Pd 26 , Pt/ Co/MgO 25 and Co/Ni 39 ultrathin magnetic multilayer structures were reported. For spintronic applications, it is inevitably important to stabilize the skyrmions at room temperature and in absence of external magnetic field. Few reports are also available on the room temperature and zero magnetic field skyrmion [40][41][42] . To the best of our knowledge, the existence of skyrmions in high spin-polarized materials, i.e., full Heusler alloys such as Co 2 FeAl (CFA) which possess low damping constant 34 , high Curie temperature 20 , large spin polarization 43 and tunable spin-dynamics properties 44,45 is yet to be reported. Despite the rapidly increasing number of experimental studies for gaining further insight into understanding the stabilization of skyrmions in ultrathin films, the robust formation of skyrmions at room temperature in large continuous thin films continues to be an exciting area.
In this report, the direct imaging of stable skyrmions at room temperature employing magnetic force microscopy (MFM) in trilayer stacks containing CFA full Heusler alloy ultrathin films is presented. There exits a few reports on the observation of skyrmions using MFM technique at low temperature as well as few at room temperature 39,[46][47][48][49] . In the present study, a Heusler alloy ultrathin film based heterostructure Ta/CFA(t)/MgO has been used in truly unpatterned form (see methods for growth and measurement details). The motivation for this work lies in the prediction that i-DMI originates at the interface and is inversely proportional to the FM layer thickness 37 . The micromagnetic magnetization and MFM simulations performed in the present study also corroborate the existence of skyrmions in these trilayer ultrathin film heterostructures. The simulations clearly demonstrate that the realization of skyrmionic state is indeed conceivable as a result of the competition between the i-DMI, anisotropy, exchange interaction and dipolar interaction in Ta/CFA/MgO heterostructure.

Results and Discussion
Figure 1(a) shows the XRD pattern recorded on Ta(10)/CFA(1.8)/MgO(2) trilayer ultrathin sample. Thickness and elemental analysis of the each layer in the stack have been confirmed by X-ray reflectively (XRR) and X-ray photoelectron spectroscopy (XPS) measurements, respectively (see Supplementary Information S2 and S3). In XRD pattern, it can be seen that while diffraction from Ta(110) is clearly observed, two broad humps appear at diffraction angle of 65.5° and 82.5°, whose position match with the (400) and (422) superlattice reflections, respectively of the CFA [See Fig. 1(a) in ref. 34 ].
In order to confirm the chemical/elemental composition in Ta/CFA/MgO trilayer ultrathin films, the secondary ion mass spectroscopy (SIMS) profiles are recorded on the two samples with different CFA thicknesses. Figure 1(b,c) show the depth profiles (normalized intensity) of each element as a function of sample thickness. Profiling was done through the film from the top to down (vertically). As expected, the 2 nm thin MgO can be seen to be the topmost layer (c.f. the left most Mg peak). This is followed by the peaks due to constituent elements Co, Fe and Al of the next 1.8 nm [ Fig. 1(b)] and 1.0 nm [ Fig. 1(c)] Co 2 FeAl ultrathin films in trilayer heterostructures. Finally, a broad peak due to thicker Ta(10 nm) layer is clearly seen in both the profiles. In the absence of suitable standard reference sample, the absolute atomic concentrations could not be determined. However, a comparison of the intensity of the measured peaks shows that the signals for each element are present in the CFA ultrathin films. The appearance of small shoulder/peak just below 1 nm depth of the sample is quite common and is attributed to the 'sputtering transition' associated with the different sputtering yields of elements present in the sample under investigation. It is clear that Co and Fe are uniformly distributed over the film thickness. A small shift of Al-peak could be due to its well-known dominant surface diffusion on account of intermixing due to atomic displacement and minor diffusions that occur within the collision cascade region created along the track of high energy (~2 keV) Cs + ion used for sputter depth profiling. Such interfacial mixing is known to result in the artificial broadening of the profile 50 as is indeed observed in the recorded depth profiles. From the foregoing, it is thus concluded that in these heterostructures, the CFA exists as an alloy albeit with the possibility of slightly different composition owing to such small film thicknesses. The NEXAFS spectra have been recorded on Ta(10)/CFA(1.8)/MgO(2) trilayer and 50 nm thick single layer CFA (hence-after SL-CFA) thin films at room temperature. The spectra corresponding to the L 2,3 edge of Fe and Co have been measured in total fluorescence yield (TFE) mode (see Fig. 2). The clear correspondence of each of the Co-and Fe-edge spectra between SL-CFA (top panel) and Ta(10)/CFA(1.8)/MgO(2) trilayer (bottom panel) can be clearly seen. In particular, the L 2 and L 3 transition peaks appear at same energy with alike broadening in the L 2,3 -edge spectra of thick (SL-CFA) and ultrathin CFA film (trilayer), which is the indication of the identical chemical environment of Co in both the films 51 . Similarly, in Fe L 2,3 -transition peaks in these spectra are identical in (both) thick and ultrathin CFA films together with the presence of shoulder (indicated by arrows in Fig. 2(b)). Consistent with the existing literature, this shoulder suggests the partial oxidation of Fe, i.e., Fe-O formation [52][53][54] which is likely since MgO layer is grown over CFA layer. The clear prominence of this shoulder in the Fe L 2,3 -edges, compared to that in Co L 2,3 -edges, further suggests that at the CFA/MgO interface, the 1.8 nm CFA film terminates within the unit cell such that the terminating plane is rich in Fe. Figure 3(a,b) show the X-ray absorption near edge structure (XANES) at Co and Fe K-edge on SL-CFA (50 nm) and Ta(10)/CFA(1.8)/MgO(2) trilayer, respectively. The qualitative similarity such as the maximum absorption at the same energy in the spectra (Co-as well as Fe-edge) among the trilayer and SL-CFA indicates the same local structure in them. The lower definition (low intensity) in the spectral feature of ultrathin CFA in trilayer, as compared to thick SL-CFA, is a consequence of the reduced average number of scatterers in the different coordination shells for a film at such a low thickness. Figure 3   The small intensity oscillations occurring in between these two prominent peaks arise due to the known backscattering contributions from low Z element such as Al, Mg and O. The ATOM and FEFF packages (see method for measurement and simulated fitting details) were used to generate the structural input for EXAFS data analysis 55,56 . The simulated curves for SL-CFA (solid lines, Fig. 3(c,d)) are found to be well fitted with the experimental data which indicates that the CFA single layer thick sample is quite ordered. In Co K-edge FT spectra, the χ(R)-oscillations have a similar shape in both single and trilayer structures and the simulated parameters are also comparable which indicate the alike bonding configuration in both thick and ultrathin CFA films. In the Fe K-edge FT spectra of trilayer, the intensity variations and position of the two main peaks are different compared to that in SL-CFA. These are ascribed to the joint contributions arising as a result of the atomic disordering between Fe-Al expected to be prominent at low thickness, and the finite oxidation of Fe as is evident from the development of a clear shoulder in NEXAFS spectra (Fig. 2). The former possibility stems from the fact that Fe-Al disorder will cause local changes in the orientation of atomic planes which could be critical in films of few monolayer thickness and leads to the suppression or shifting of the main peaks 57,58 . Furthermore, since the 1.8 nm thick CFA film does not have complete unit cell (at its surface) while it terminates in contact with the MgO layer, the chemical environment is therefore quite different at its two interfaces (Mg and O at the top and Ta at the bottom). Therefore, the interference effect between the scattered waves from the two different surfaces/interfaces cannot be ignored in this trilayer structure which is known to cause multiple scattering processes, thereby contributing to the intensity and broadening of the absorption peaks 59 . The Fe-edge spectrum was therefore simulated by considering one Fe-O shell and ordered CFA structure. The sky-blue color box in the radial distribution curves represents the range (window) used for fitting simulation. The coordination numbers, bond lengths, and effect of disorder which is approximated by the Debye-Waller factor (σ 2 ) (mean-square fluctuations in the path length) have been used as fitting parameters. The simulated parameters are shown in Table-II in Supplementary Information S4. At Co and Fe K-edges in CFA single layer, the coordination number and the bond length distances matched very well with the theoretical values within the uncertainty. These values are slightly different (lower) in the ultrathin film which is due to insufficient coordination shell in the ultrathin film of few monolayer thickness regime, as discussed above. Moreover, the σ 2 is small in these films which is comparable to the reported results for ordered CFA and other Heusler compounds 60,61 . Figure 4 shows the magnetization hysteresis (M-H) curves for the Ta/CFA/MgO trilayer heterostructure recorded both in the in-plane and the out-of-plane (OOP) magnetic field orientations. The latter M-H loop exhibits lower saturation field compared to the former suggesting that the CFA film possesses perpendicular magnetic anisotropy. The saturation magnetization is found to be 838 ± 40 kA/m which is smaller than the value of 1000 kA/m reported for the bulk samples 34 which is due to the finite size effect in ultrathin films. The perpendicular uniaxial magnetic anisotropy energy, evaluated from the difference in the measured values of saturation fields, was found to be 2.90(±0.05) × 10 5 J/m 3 . In the inset, the micromagnetic simulated magnetization curve, for the OOP magnetic field orientation is also shown for comparison with the experimental MH curve. Both the experimental and simulated MH curves show qualitatively similar characteristics, except for the different saturation fields which is attributed to the different demagnetization effects on account of different sample size/ geometry in the experiment and simulation. The simulation details are given in Supplementary Information S1.
Quantitative assessment of i-DMI at ta/CFA/Mgo Interface. We employ Brillouin light scattering (BLS) technique to determine the i-DMI strength in these Ta/CFA/MgO ultrathin films. The advantage of using the BLS technique is that the wave-vector of the spin wave is uniquely determined by the wavelength and the angle of incidence of the laser beam and it can simultaneously detect the propagating spin-wave excitations at +k and −k  63 where the spin-wave wave-vector and magnetization both lie in the sample plane and are mutually perpendicular. The i-DMI energy density can be determined either by modeling the full spin-wave frequency vs. wave-vector dispersion with a modified dispersion relation for DE spin waves by introducing the DMI term or from the frequency difference (Δf) between spin-waves with opposite (+k and −k) wave-vectors, which is given by 37 , Where, Δf is the frequency difference, k x is the x-component of the wave vector, γ is the gyromagnetic ratio, D is the i-DMI energy density and M s is the saturation magnetization. In our case, we have chosen the latter method for determining D, as the estimation of D here is primarily determined by the experimentally measured quantities Δf, k x and M s (=1000 ± 50 kA/m is used for thick films 44 ). Figure 5(a,d) show typical BLS spectra, recorded from Ta/CFA(t)/MgO thin film heterostructures with t = 10 and 5 nm, respectively. It is clearly observed that the positions of both the Stokes and anti-Stokes peaks in the BLS spectra move to the higher frequency values with increase in k x . Δf was extracted from Lorentzian fits to the BLS spectra and is plotted vs. k x in Fig. 5(c,f). By fitting the experimental results by using equation (1) (2) thin film recorded at room temperature in remnant state (see methods for MFM measurement details). The image displays a well-developed characteristic feature which is present throughout the whole (large) imaged area of thin film. The very uniform/confined texture with a black core is suggestive of the presence of radially chiral spin configuration in the remnant state. These structures are similar to those reported previously by other groups in different systems 25,46 . It is emphasized here that the data recorded in MFM represents the gradient of magnetic force with respect to z-component of the magnetic field originated from the sample. Therefore, these dense textures developed in remnant state must be chiral in nature 19,48,65,68 which is further confirmed from the MFM simulations in the upcoming section. Further, the line (lateral) scan profiling was performed for the confirmation and characterization of the size and the periodicity of the chiral nanostructures observed in these Ta/CFA/MgO thin film. Figure 6(c) shows the zoomed area (labeled as '1' in Fig. 6(a)) for a clear view of the chiral structure and Fig. 6(d) shows the corresponding 3D plot where the different color contrast corresponds to the different orientations of the z-component of the force gradient. Figure 6(e) shows the line scan profile performed on one such spin-texture structure within the region labeled as '1' in Fig. 6(a) (circles represent data points). The magnetic contrast variation has been identified as skyrmions and to confirm our consideration we performed the identical line scan profiling on the simulated skyrmions (in the forthcoming section) and found the chirality in the observed structures; as a result, we identify/endorse them as skyrmions in these unpatterned Ta/CFA/MgO thin  Fig. 7(b-d) revealing the formation of skyrmions texture together with a few strip-like domains which indicate that the system exhibits perpendicular magnetic anisotropy. The well-developed skyrmionic feature is clearly visible in Fig. 7(d) and the line scan profile (shown in the inset) highlights the typical size of a skyrmion.
Moreover, the skyrmions with opposite polarity are also visualized on reaching the remnant state by applying a large negative field (−0.5 T) and removed and then performed the MFM measurement. Figure 8 shows the MFM images of the Ta(10)/CFA(1.0)/MgO(2) thin film after magnetic saturation in a opposite direction as compared to the remnant state in Fig. 7. It is clearly seen that the polarity of skyrmions is reversed. Moreover, the line scan profile is shown in Fig. 8(d) indicates opposite contrast/profile (compared to the previous) in the remnant state reached after saturation in a negative field. Here, it is also realized that the domain boundaries are not distinctly resolved because of the limited MFM lateral resolution, but the skyrmionic-state is identical even in the case of a multi-skyrmions, which signifies that the structure is topologically invariant. Thus, the skyrmion formation in Ta/CFA/MgO thin films is robust against defects and shape imperfections.
To validate the experimental interpretation of room temperature skyrmions, micromagnetic simulations were carried out using the Mumax3 simulation code 69 (see Supplementary Information S1 for simulation details). As pointed above, although the i-DMI could be measured only in slightly thicker films [i.e., 0.2 ± 0.01 mJ/m 2 for CFA (5 nm)]. Assuming that the i-DMI strength increase inversely with the FM layer thickness 37 , for the CFA films of thickness ≤1.8 nm, the i-DMI strength is expected to be of ≥1 mJ/m 2 . On the other hand, in our simulations, the skyrmions were formed at the smallest DMI value of 1.2 mJ/m 2 . Thus to be consistent with experimental findings, the i-DMI value has been chosen to be 1.2 mJ/m 2 for the simulations, whose results are presented in    (2) thin films (area ~2 × 2 μm 2 ) at room temperature. Fig. 9 shows the results of these simulations in zero external magnetic field at 300 K. Figure 9(b) shows the formation of multiple skyrmions in an extended area which clearly demonstrates the features in accordance with  our experimental findings (Figs. 6-8). The zoomed view of the dotted squared region of Fig. 9(a) is shown in Fig. 9(b) for clarity. A simulated 3D-image (area 100 × 100 nm 2 ) of the magnetization is shown in Fig. 9(c), which clearly reveals that the observed spin texture confirms Néel type skyrmion. Further, MFM simulation was also performed in these trilayer heterostructure, which also yielded the formation of skyrmion texture (see Fig. 9(d)). Figure 9(e) shows the line scan profile of single skyrmion (shown by dotted line in Fig. 9(b)) together with a fit using the equation proposed by the Romming et al. 29 to verify the magnetization variation within the skyrmion. The diameter is found to be sub ~100 nm and the statistical size distribution of the skyrmions in the image shown in Fig. 9(f) suggests a large number of skyrmions lies in the range of sub-100 nm which are comparable to the size of the skyrmions estimated experimentally from line scan profiling. The line scan profile and 3D surface plot (see Fig. 9(g)) where each color represents a different orientation of the magnetization, both reveal the chiral texture of magnetization in this system. Thus, Fig. 9(e,g) offer the wisdom that the magnetic moments point downward in the center of the skyrmion (black region) and that a re-orientation of the magnetic moments occurs as one moves towards the periphery of the skyrmion where the magnetic moments point upwards. This is consistent with the experimentally observed MFM images (Figs. [6][7][8] which similarly revealed the presence of annular bright regions surrounding the central dark cores. A simulated results for the sample Ta(10)/CFA(1)/MgO (2) is discussed in the Supplementary Information S1. The skyrmion number N sk is also evaluated directly from the simulations for these structures. The value of N sk is found to be equal to 1 which corresponds to the existence of skyrmions. Hence, the simulated magnetization and MFM structures identically match with those experimentally observed using MFM imaging performed in remnant state. Hence, it is evident that the experimentally observed spin texture in the full Heusler alloy CFA consists of clusters of skyrmions. Further, it is to be noted that skyrmion clusters have been previously demonstrated both experimentally [71][72][73] as well as theoretically 73 in patterned structures. These theoretical studies show that the single, double and even multiple skyrmions may coexist together due to skyrmion-skyrmion interactions via competing ferromagnetic exchange and DMI both of which are short-range interactions 74 .
In summary, the observation of room temperature skyrmions in ultrathin Ta/Co 2 FeAl/MgO trilayers is reported based on magnetic force microscopy imaging, and is supported by micromagnetic simulations and Brillouin light scattering measurements. The line scan profiling clearly reveals the nano-scale chiral texture. The topological state is corroborated by micromagnetic simulations by considering the experimental parameters such as interfacial Dzyaloshinskii-Moriya Interaction, exchange constant and anisotropy energy magnitudes. The study highlights the prospects of room temperature evolution of skyrmions in simple trilayer extended thin films without requiring the additional cost and time intensive film patterning constraints. The skyrmions observed in these CoFe 2 Al (1.8 nm) films were of the sub -100 nm size and even sub -50 nm in thinner Co 2 FeAl(1.0 nm) films. It is envisaged that this robust formation of skyrmions at room temperature is a vital significant experimental advancement and will pave the way towards the industrial development of the skyrmion based devices for ultra-dense memory storage and logic applications, etc.

Methods sample preparation (Ion-assisted ion-beam sputtering). The trilayer thin film stacks consisting of
Si(100)substrate/Ta(10)/CFA(10,5,1.8,1.0)/MgO(2) (numbers in parenthesis represent film thickness in nm) were in-situ deposited in sputter-up configuration using ion-assisted ion-beam sputtering technique (after removing the native surface oxide layer of the Si substrates with HF (10:1 ratio) solution for 60 s). Prior to deposition, the vacuum chamber was evacuated by turbo molecular and cryo-pumps to a base pressure lower than ~1 × 10 −7 Torr. A buffer layer of Ta (10) was grown at room temperature and in situ annealed at 773 K for 30 min in order to achieve a very flat surface. CFA layers with different thicknesses were then grown at room temperature followed by a thin layer of MgO on top of the CFA layer. A single layer CFA(50 nm) thin film was also grown for EXAFS analysis to compare with the ultrathin CFA films stacked in tri-layer heterostructure. During the growth of the Ta buffer and CFA layers, a working pressure of ~8.5 × 10 −5 Torr was maintained during sputtering by flowing 4sccm of Ar gas. This Ar was fed through a high energy RF-ion source comprising of two-grid assembly for extraction of ~4.5″ dia Ar-ion beam. The high energy RF ion source was operated at 75 W with inner grid voltage V + = 500 V (i.e., Ar ion-energy) and the outer grid voltage V − = −270 V (for beam extraction/acceleration). The high-energy Ar ion beam so extracted was incident at an angle of 45° on 6″ dia target mounted on a water-cooled target turret. A maximum of four targets, each of 6″ dia can be mounted on the 4 sides of the turret which is rotatable in-situ to position any of the targets under the Ar-beam for sputtering. The target-substrate distance is ~27 cm.
For growing the MgO film, while the Mg target was sputtered using high energy RF-ion source (RF power = 100 W, V+ = 500 V and V − = −270 V, Ar flow rate = 3 sccm), the growing film was simultaneously irradiated (or to say assisted) with a low energy (50 eV) oxygen ion beam (at 45° on the growing film). This oxygen beam was extracted from a low energy RF-ion source (operated at RF power = 75 W, V + = 50 V and V − = −30 V, oxygen flow rate = 12 sccm). Accordingly during the growth of MgO, whereas the total working pressure was slightly higher ~1.0 × 10 −4 Torr, the O 2 partial pressure was ~1.5 × 10 −5 Torr. The deposition rates during the growth of Ta, CFA and MgO were 0.03 nm/sec, 0.03 nm/sec and 0.02 nm/sec, respectively. After the deposition of all the 3 layers, the trilayer stack was in-situ post-annealed at a temperature of 523 K for 1 hour in high vacuum (6 × 10 −7 Torr).
Characterizations. For the crystalline phase identification, a glancing incidence X-ray diffraction pattern of Ta/CFA/MgO trilayer film was recorded at a fixed glancing angle of 1° by Philip's make X'pert(PRO) diffractometer which utilized Cu-K α radiation. The deposition rates were accurately calibrated individually and then also confirmed after deposition of the stack using data from X-Ray reflectivity measurements. The thicknesses of individual layers, their density, the interface width and surface roughness were accurately estimated by simulating the XRR data. Imaging of skyrmions in these trilayer thin films was performed in the remnant state (zero external magnetic field during imaging) by employing the Bruker make MFM (Model -Dimension icon with ScanAsyst). The MFM imaging was performed on the present samples to determine their remnant states after saturating (ex-situ) the magnetization with a perpendicular field of 0.5 T. A Co-Cr coated Si 3 Ni cantilever tip was used to map out the magnetic spin structure with a lateral resolution of 30 nm. The MFM imaging performed in this work favors the study of perpendicularly oriented magnetic domain structures. The MFM measurement consists of two steps, first tapping mode (topographical information) followed by lift mode (for magnetic domain information; positioned at the height of 70 nm from the film surface). The tip of the magnetic probe was magnetized parallel to the downward vertical direction using a permanent magnet. The resonant frequency of the cantilever (spring constant = 2.8 Nm −1 ) was 73 kHz. Magnetization vs. magnetic field (M-H) of ultrathin films was recorded using a Quantum Design make SQUID magnetometer in both in-plane and out-of-plane configurations. The chemical valence state and the interface hybridization were examined by employing XPS. The XPS spectra on Ta(10)/CFA(1.8)/MgO(2) trilayer system were recorded using a SPECS make system with an Al-K α x-ray source (1486.6 eV) and a hemispherical energy analyzer with pass energy of 40 eV and a resolution of ~1 eV. Brillouin Light Scattering (BLS) spectra were recorded using a Sandercock-type (3 + 3) pass tandem Fabry-Pérot interferometer and a p-polarized (wavelength of 532 nm with 300 mW power) single longitudinal mode solid state laser. The details of the BLS set up can be found elsewhere 75 . The spectra were recorded in the conventional backscattering geometry at various wave vector orientations selected by mounting the sample on the angle-controlled sample holder providing a range of 10°−60° incident angles corresponding to wave vectors k x lying in the range of 0.004-0.0204 nm −1 . The free spectral range of 50 GHz and a 2 9 multi-channel analyzer have been used for recording the spectra. An in-plane magnetic field of 0.1 T was applied during measurements. All measurements were performed at room temperature. X-ray Absorption spectroscopy (XAs). Extended X-ray absorption fine structure (EXAFS). EXAFS spectra were at the Co and Fe K-edges in the step scanning mode in 1D KIST-PAL beamline of the Pohang Accelerator Laboratory, Pohang, South Korea 76,77 . To measure these spectra, higher harmonics were removed by detuning of incident beam to 60% of maximum intensity and beam energy was calibrated using reference foils (Co and Fe). Three ionization chambers filled with He and N 2 -gas were used to record the intensity of the incident and the transmitted X-rays. A fluorescence detector was also placed near to the sample holder in order to measure fluorescence current. The samples (making an angle of 45° from direction of incident X-rays) were placed between the first and second ionization chambers. The data measured in fluorescence mode was used for investigating the local atomic order. simulation details for eXAFs. In order to investigate the atomic site specific information of structure and relative bond-lengths with respective to absorbing atoms, the EXAFS data was converted into radial distribution function χ(R) (Fourier transformed into R-space) after background subtraction as a function of effective distance R (in Å). Here, R includes not only the interatomic distances but also the scattering phase shifts also. IFEFFIT/ ARTEMIS 78 were used to simulate/process the recorded data with the graphical interface Athena 55 . The theoretical structure for fully ordered Co 2 FeAl full Heusler compound was generated using the ATOM and FEFF 55,56 . The errors in the fit parameters σ 2 were obtained from 90% of happiness factor.
Near edge X-ray absorption fine structure (NEXAFS). The NEXAFS spectra were recorded at soft X-ray beam line 10D KIST beam line 79,80 . The system was evacuated to ~1.5 × 10 −8 Torr. The XAS spectra from the single layer CFA(50 nm) and trilayer Ta(10 nm)/CFA(1.8 nm)/MgO(2 nm) ultrathin films were recorded in total fluorescence yield (TFY) mode. Fixed voltage (1.75 kV) was applied to micro-channel plate for recording the spectra in TFY mode and the fluorescence detector (40 mm diameter) was placed at an angle of 45° from direction of beam axes. The incident photon flux was measured by inserting the gold (Au) mesh in the path of X-ray beam.