The knowledge of how oxygen atoms are distributed at a magnetic-metal/oxide, or magnetic-metal/non-magnetic-metal interface, can be a useful tool to optimize device production. Multilayered Ni81Fe19/Ta samples consisting of 15 bilayers of 2.5 nm each, grown onto glass substrates by magnetron sputtering from Ni81Fe19 and Ta targets, have been investigated. X-ray absorption near edge structure, extended X-ray absorption fine structure, small angle X-ray diffraction, and simulations were used to characterize the samples. Oxygen atoms incorporated onto Ni81Fe19 films during O2 exposition are mainly bonded to Fe atoms. This partial oxidation of the Ni81Fe19 surface works as a barrier to arriving Ta atoms, preventing intermixing at the Ni81Fe19/Ta interface. The reduction of the Ni81Fe19 surface by the formation of TaO
In tunneling junctions the tunnel magnetoresistance (TMR) is mainly determined by the spin polarization at the magnetic surfaces touching the insulating barrier, and the presence of a nano-oxide layer (NOL) at the metal/barrier interface decreases the TMR ratio. In multilayered spin valves, the giant magnetoresistance (GMR) ratio can be improved by the insertion of a NOL at the interfaces [
At thermodynamic equilibrium, the distribution of oxygen atoms at a metal/metal interface will be determined by the difference in oxidation energy of the metals. During the deposition of very electronegative materials like Ta onto another metal’s surface, the reduction of this surface can occur if it presented some initial degree of oxidation [
The direct measurement of oxygen states is not simple and has been restricted to a few studies with X-ray photoelectron spectroscopy [
In tunneling junctions, the insulating barrier can be produced by the oxidation of a metal film, exposed to a natural or plasma assisted oxidation process. The interface prevents the oxidation of the bottom electrode, and the insulator has a higher oxygen content than in the more stable stoichiometry [
In magnetic sensors, the presence of a NOL allows for the annealing-induced improvement of the multilayered microstructure; the NOL acts as an interdiffusion barrier and as an oxygen-reduction agent [
The knowledge of how oxygen atoms are distributed at a metal/metal interface can be a useful tool to optimize device production. In this work, we have investigated the Ni81Fe19/Ta interface in multilayered samples produced under a controlled atmosphere, where we have deliberately introduced O2 at different stages of the sample preparation process.
In order to measure XAFS signal-to-noise ratios sensitive to the oxidation at the interfaces, multilayered samples of very thin Ni81Fe19/Ta layers were produced. The multilayers consisted of 15 bilayers of 2.5 nm each. The films were grown onto glass substrates from Ni81Fe19 (Py) and Ta targets using RF or DC magnetron sputtering, respectively, in a 0.7 Pa argon atmosphere. The thickness of the Py and Ta layers were fixed at 2.0 and 0.5 nm, respectively. Ta was chosen because it presents an absorption edge that could be detected under our experimental conditions.
The sample PyTa was grown without any deliberate exposition to oxygen during growth. The samples PyO2Ta and PyTaO2 were grown with an additional step; oxygen was admitted into the deposition chamber (13.3 Pa for 300 s after each Py or Ta layer, respectively), and then the chamber was pumped back down to base pressure (∼2 × 10−8 Pa) before the growth of the next layer. The sample PyO2 was grown with the same exposition to oxygen after the deposition of each Py layer, and with no deposition of Ta. All samples were exposed to ambient atmosphere after growth.
The chemical modulation perpendicular to the substrate surface was observed using small angle X-ray reflectometry (XRR). In order to obtain a proper standards for comparison, Py and Ta2O5 films were grown at room temperature both onto Si or glass substrates. The interference patterns were acquired at small angles
The XAFS experiments were performed at the XAFS1 beam-line of the Laboratório Nacional de Luz Síncrotron, near room temperature. XANES and EXAFS of the samples were measured at the
Figure
(a) Normalized Fe (left) and Ni (right)
It can be clearly seen from the spectrum from sample PyO2 that the Py exposition to the oxygen atmosphere led to oxidation. The oxidation signatures are present Fe and Ni absorption edges and are much more evident in the Fe spectrum in both Figure
The growth of the Ta layer on top of the Py layer produces some degree of disorder, as can be seen in the spectrum from sample PyTa; the degree of disorder can be inferred from the Debye-Waller factors shown in the last rows of Table
XAFS analysis results using WinXas program [
Sample | Fe2O3 (wt.%) | Py (wt.%) | NiO (wt.%) | Py (wt.%) | Disorder Fe-edge |
Disorder Ni-edge |
---|---|---|---|---|---|---|
PyTa | 11.22 ± 0.43 | 88.78 ± 0.47 | 2.82 ± 0.04 | 97.18 ± 0.04 | 8.32 ± 0.098 | 5.28 ± 0.078 |
PyTaO2 | 28.51 ± 0.05 | 71.49 ± 0.06 | 1.18 ± 0.04 | 98.82 ± 0.42 | 5.40 ± 0.017 | 5.16 ± 0.015 |
PyO2Ta | 20.29 ± 0.52 | 79.71 ± 0.57 | 1.68 ± 0.04 | 98.32 ± 0.04 | 4.60 ± 0.019 | 3.66 ± 0.070 |
PyO2 | 50.24 ± 0.30 | 49.76 ± 0.32 | 6.13 ± 0.04 | 93.87 ± 0.04 | 4.28 ± 0.020 | 3.31 ± 0.017 |
In samples PyO2Ta and PyTaO2, the Ta layers clearly extract oxygen atoms from the Py layers. To estimate the degree of oxidation in the multilayers, we can fit a calculated spectrum (obtained by the linear combination of spectra from oxide and metal phases) to the experimental spectrum, obtaining the Fe2O3, NiO, and Py weight contribution [
Fe
Figure
(a) Fe
The FT spectra shown in Figure
In accordance with what was indicated in the XANES measurement, the degree of disorder in the Py layer is lower in sample PyO2Ta than in samples PyTa or PyTaO2, where Ta is grown directly onto the metallic Py surface. It has been assumed that all metal atoms show a fix coordination number for the first shell (first peak in the FT); thus, disorder is the key feature to distinguish between different deposition processes, shown in the last columns of Table
The deposition of Ta onto the metallic Py surface leads to a defective interface, as indicated by the last two columns in Table
Figure
(a) Ni
Figure
Normalized Ta
Figures
XRR measurement from sample PyTa and simulation.
XRR measurement from sample PyTaO2 and simulation.
XRR measurement from sample PyO2Ta and simulation.
XRR measurement from sample PyO2 and simulation.
XRR analysis results using IMD program [
Sample | Density Py (g/cm3) | Density Ta (g/cm3) | Roughness Py (nm) | Roughness Ta (nm) | Thickness Py (nm) | Thickness Ta (nm) |
---|---|---|---|---|---|---|
PyTa | 9 ± 1 | 17 ± 1 | 1 ± 1 | 0.7 ± 0.1 | 2.0 ± 0.1 | 0.5 ± 0.1 |
PyTaO2 | 9 ± 1 | 7 ± 1 | 0.7 ± 0.1 | 0.4 ± 0.1 | 2.0 ± 0.1 | 1.0 ± 0.1 |
PyO2Ta | 9 ± 1 | 8 ± 1 | 0.3 ± 0.1 | 0.3 ± 0.1 | 2.0 ± 0.1 | 0.9 ± 0.1 |
PyO2 | 9 ± 1 | 4 ± 1 |
0.8 ± 0.1 | 0.8 ± 0.1 |
1.7 ± 0.1 | 0.4 ± 0.1 |
The roughness of the PyO2Ta sample is the smallest, in accordance with the absorption experiments. The degree of disorder in the Py layer is lower when it is exposed to oxygen prior to the deposition of Ta than when Ta is grown directly onto the metallic Py surface.
We studied the oxygen distribution at Ni81Fe19/Ta interfaces. The exposition of a Ni81Fe19 surface to an O2 atmosphere bonds oxygen atoms mainly to Fe atoms. This oxide layer limits intermixing during the deposition of a very thin Ta film, acting as a barrier to the diffusion of Ta atoms into the Ni81Fe19 layer; during Ta deposition oxygen atoms are extracted from the Ni81Fe19 surface to form TaOx. On the other hand, when the Ta film is deposited onto a Ni81Fe19 surface that has not been deliberately exposed to oxygen, intermixing or alloying takes place; this Ni81Fe19/Ta interface layer is less effective in incorporating oxygen from the atmosphere.
The XAS and XRR data used to support the findings of this study have been deposited in the Google Drive repository at
The authors declare that they have no conflicts of interest.
This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-Brasil (CAPES)-Finance Code 001, FAPESP, Laboratório Nacional de Luz Síncrotron (Brazilian Synchrotron Light Laboratory-LNLS), Brazil, and FAPERGS (grant PRONEX 2014). L.S.D. acknowledges the financial support from CNPq (grant no. 302950/2017-6).