Adsorption induced modification of in-plane magnetic anisotropy in epitaxial Co and Fe/Co films on Fe(110)

Engineering of magnetic anisotropy (MA) of ferromagnetic films is one of the key tasks in modern spintronics. A great progress has been achieved in tailoring the perpendicular magnetic anisotropy, while a control of in-plane MA^1–5 was usually accomplished either by means of atomic steps induced uniaxial MA^6,7 or by tuning the balance between contributions from the thickness and temperature-dependent volume and surface (interface) MA contributions.^8–13 Another possibility, direct modification of interface or surface MA, can be attempted by gas adsorption on the film surface¹⁴ or by deposition of both magnetic¹⁵ and nonmagnetic^14,16,17 capping materials. In the present report we combine two of the above mentioned concepts in order to precisely tune the in plane MA in Co and Fe/Co films epitaxially grown on Fe(110) surface. Motivated by huge MA recently reported for Co films on Fe(110)¹⁵ we followed its dependence on both Co and Fe overlayers thickness and we documented its evolution upon adsorption of residual gases.

Epitaxial Co films and Fe/Co bilayers were in situ deposited on a Fe(110)/W(110) surface using molecular beam epitaxy. High-quality epitaxial Fe films, with a thickness from several to several tens of nanometers were grown on a W(110) single crystal at room temperature and post-annealed at 675 K to obtain atomically smooth (110) surfaces. In the following text we term this thick Fe films grown directly on W(110) as the “base” Fe layers in order to avoid confusion and to distinguish them from thin Fe films deposited on top of Co/Fe/W. Next, on the Fe(110) base, Co films and Fe/Co bilayers were deposited at room temperature. As discussed in our previous report, on the Fe(110) surface Co crystalizes in a metastable bcc structure in the whole investigated thickness range.¹⁵ In order to precisely determine the full Co and Fe thickness dependence of in plane MA, we prepared a specially designed sample divided in two parts that we will further refer as “Co part” and “Fe/Co part”. Schematic picture showing a design of the sample is presented in Fig. 1.

The Co part was a double wedge area, with two orthogonal wedges of Co and base Fe. The thicknesses of the base Fe (d_{base Fe}) and Co (d_Co) wedges were continuously varied from 110 Å to 440 Å and from 0 to 20 Å, respectively. The Fe/Co part of the sample was also a double wedge area, but in this case the thickness of the base Fe layer was fixed at d_{base Fe} = 440 Å, while two orthogonal wedges of Co and Fe, ranging (0, 20 Å) and (0, 12 Å), respectively, formed a Fe/Co bilayer area on base Fe(110).

The magnetic properties of the Co films and Fe/Co bilayers were imaged in situ using the longitudinal magneto-optic Kerr effect (MOKE). Fig. 2a shows a differential MOKE image of a selected area of the sample. To enhance the magnetic contrast and to highlight characteristic features present in the sample, we subtracted a reference image taken at a small external magnetic field along [1$1¯$0] from the image taken at remanence (H = 0). On the Co part of the sample, the dark area is where the remanence magnetization remained along the saturation direction, [1$1¯$0], whereas the brighter area corresponds to the [001] magnetization direction in the remanent state, as concluded from respective hysteresis curves presented for regions of interest (ROI) marked as ROI#1 and ROI#2. The hard axis hysteresis loops, like the one for ROI#2 area in Fig. 2c, exhibit a characteristic jump of magnetization at the switching field H_s. The border between the dark and bright areas, shown by the white dotted line, visualizes the in-plane spin reorientation transition (SRT) in the (d_{base Fe}, d_Co) space, and its shape reflects an oscillatory dependence of the magnetic surface anisotropy (MSA) on the Co thickness, as previously reported.¹⁵ The entire Fe/Co part, is characterized by typical hard axis hysteresis loops, even in the dark area, where, for thicker Co films, the hard hysteresis curves are measured with very small values of the switching field H_s. The switching field H_s, determined from the hard axis hysteresis loops, is a good measure of the uniaxial anisotropy,^18,19 so the in-plane uniaxial magnetic anisotropy can be characterized quantitatively from single-loop measurements across the two-dimensional (d_Co, d_Fe) space. In order to determine the full dependence of H_s in (d_Co, d_Fe) space, the sample area was divided into a (80 x 80) matrix of ROIs. For each ROI a hysteresis loop was extracted by analysis of a series of MOKE images taken as a function of the external magnetic field, H, applied along the [1$1¯$0] in-plane direction. In this way H_s, and therefore the in-plane magnetic anisotropy, can be analyzed for any combination of Co and Fe thicknesses. The size of a ROI was 50×80 μm², which corresponds to the averaging of magnetic properties over a finite thickness intervals ∆d_Co = ∼0.25 Å and ∆d_Fe = ∼0.30 Å for the Co and Fe wedges, respectively.

Results of the hysteresis curves analysis are presented in Fig. 3, where two-dimensional maps of H_s are shown for the entire investigated sample area. Fig. 3a presents H_s map for the freshly deposited sample. The white areas in Figs. 3a, b mark the regions where typical easy axis square hysteresis loops are observed. In the Co part of the sample (left part of the H_s map in Fig. 3a) d_Co oscillations of H_s are visible with the most intense peak at d_Co = 5 Å. For any given d_{base Fe} (any particular vertical profile of the map) H_s oscillates with d_Co with exactly the same period, and only the H_s magnitude depends on d_{base Fe}. An exemplary H_s(d_Co) profiles, derived from the map presented in Fig. 3a for d_{base Fe} = ∼260 Å (dashed vertical line) is shown in Fig. 4a. Following the detailed analysis presented in Ref. 15 we interpret oscillations of switching field as resulting from a strong and non-monotonic dependence of MSA on Co coverage. Exposure to residual gases leads to the drastic modification of the corresponding H_s map, as presented in Fig. 3b. This effect can be also seen by comparing the one-dimensional H_s profiles in Fig. 4 a and b. The main change in the H_s(d_Co) dependence is its flattening induced by adsorption of residual gases; especially the peak observed at d_Co = 5 Å for the as-deposited sample is strongly suppressed by adsorption. To better visualize the adsorption driven modifications of H_s and MSA, a differential map of the switching field was calculated as ∆H_{s diff} = H_{s before} - H_{s after} (Fig. 3c). Two regions in the Co part are marked (contours 1 and 2 in Fig. 3c), where the adsorption effects are especially pronounced and exhibit a strong d_Co sensitivity. This sensitivity of adsorption induced MSA on d_Co is also reflected in the ∆H_{s diff}(d_Co) profile in Fig. 4c (see corresponding peaks “1” and “2”). The adsorption driven change of the switching field ∆H_s has the opposite sign for peaks “1” and “2”, which means that MSA can be either enhanced or suppressed, respectively. This suggests a controllable method of tuning the MSA magnitude.

We now focus on the H_s maps and profiles determined from the Fe/Co part, shown in Fig. 3a-c and Fig. 4d-f, respectively. At first glance, one may judge that the H_s(d_Co, d_Fe) maps determined for the as-deposited and adsorbed states of the sample are almost identical, see Fig. 3 a and b. The blue areas correspond to hard axis hysteresis loops with a small switching field, while the red areas are characterized by the highest H_s values. The difference between the as-deposited and adsorbed states of the sample becomes visible in the differential ∆H_{s diff} map shown in Fig. 3c, where the area that is most sensitive to adsorption effects is marked by contour 3. Corresponding exemplary H_s(d_Fe) profiles, as derived from H_{s before}, H_{s after} and ∆H_{s diff} maps, are shown in Fig. 4 d, e and f, respectively. These profiles are determined for fixed d_Co = 4.5 Å, along the horizontal dashed lines in Fig. 3a–c. The adsorption driven change of H_s (and therefore induced MSA) have a clear maximum at d_Fe = ∼1 Å, which is shifted with respect to the maxima observed for as-deposited and adsorbed states of the sample.

To conclude this part, Fig. 3a-c and Fig. 4d-f show how the MSA sensitivity to adsorption of residual gases can be tuned by a proper choice of the point in (d_Co, d_Fe) two-dimensional space.

In order to identify adsorbates responsible for the reported modifications of MA, we used Auger Electron Spectroscopy (AES) to study the surface cleanliness. We found that upon exposure to residual gases, C and O AES peaks showed on the initially clean surface with an ultimate intensity corresponding to a monolayer. Taking into account the composition of the residual gas atmosphere, carbon oxides and hydroxyl groups are the most probable adsorbates responsible for the observed modification of MA. In addition, the intensity of C and O AES peaks stayed constant over the entire area of our sample and we can presume that these adsorbates are spatially homogeneous. Another and more difficult problem is the expected adsorption of hydrogen, the main constituent of the residual UHV atmosphere, which is not detectable in AES. At the present stage of research we refrain from a deeper discussion of the adsorption effects, leaving it to detailed studies of the surfaces purposely modified by adsorption of specific gases.

In-plane magnetic surface anisotropy of Co films and Fe/Co bilayers on Fe(110) was investigated. Our results indicate that MA sensitivity to the surface chemical state depends strongly on the thickness of Co and Fe. This can be exploited in future experiments with controlled adsorption of specific gases, i.e. CO, O₂, O or H₂.

The authors acknowledge the CERIC-ERIC Consortium for financial support through the research grant MAG-ALCHEMI.