Origin of the interlayer exchange coupling in $[\mathrm{Co}∕\mathrm{Pt}]∕\mathrm{Ni}\mathrm{O}∕[\mathrm{Co}∕\mathrm{Pt}]$ multilayers studied with XAS, XMCD, and micromagnetic modeling

The origin of the oscillatory interlayer exchange coupling in $[\mathrm{Co}∕\mathrm{Pt}]∕\mathrm{Ni}\mathrm{O}∕[\mathrm{Co}∕\mathrm{Pt}]$ multilayers is investigated using advanced microscopy and spectroscopy techniques and micromagnetic modeling. X-ray magnetic circular dichroism (XMCD) measurements show the presence of the canting of Ni spins in the NiO film being greater for antiferromagnetically coupled multilayers than for ferromagnetically coupled ones. This behavior is consistent with the model, which assumes a different sign of the exchange coupling at the two interfaces and the antiferromagnetic layer-by-layer coupling in the NiO film. An unexpectedly short attenuation length of $4\mathrm{\AA }$ for secondary electrons in NiO is measured, which has implications for the interpretation of XMCD data. Domain images obtained using XMCD-photoemission electron microscopy at the Co and Ni resonances indicate that the canting of the Ni spins occurs on both a microscopic and macroscopic scale. The average size of the domains is shown to increase with exchange coupling strength. In antiferromagnetically coupled samples, the competition between magnetostatic and interlayer exchange effects gives rise to a region of overlapping domains. The size of this region scales inversely with coupling strength. Finally, the temperature dependence of the interlayer coupling shows both reversible and irreversible effects. The irreversible effects stem from oxidation∕reduction reactions at the $\mathrm{Co}∕\mathrm{Ni}\mathrm{O}$ interface. The reversible effects stem from the temperature dependences of the many factors that play a role in the interlayer coupling and exhibit nonmonotonic temperature dependence.

Figure 1

(a) Room temperature XAS spectra at the Ni and Co $L_{3,2}$ resonances indicate no evidence of multiplet splitting at the Co $L_{3,2}$ peaks, as would be expected if CoO were present, and the Ni $L_2$ peak reveals a doublet splitting, as expected for ${\mathrm{Ni}}^{2+}$ in NiO. (b) The major and minor hysteresis loops at room temperature for a sample with a NiO thickness of $8\mathrm{\AA }$, which was measured using polar MOKE. For measurement of the minor loop, a large field was applied to saturate the sample; the field was then decreased until the top $\mathrm{Co}∕\mathrm{Pt}$ multilayer reversed. Then the field was again increased to complete the loop, during this entire loop the bottom $\mathrm{Co}∕\mathrm{Pt}$ multilayer does not switch. (c) The room temperature minor loop shift is a measure of the interlayer exchange coupling, thus the variation in coupling with NiO thickness is given. The coupling oscillates with NiO thickness from antiferromagnetic to ferromagnetic coupling.

Figure 2

(a) Element-specific magnetization loops for Co and Ni taken with x-ray magnetic circular dichroism in total electron yield mode at $175\mathrm{K}$ on a ferromagnetically coupled, series 2 sample with a $10\mathrm{\AA }$ thickness of NiO. (b) Possible NiO spin configuration for a ferromagnetically coupled sample in the antiferromagnetic state $\left(①\right)$ and ferromagnetic state $\left(②\right)$, leading to the net Ni out-of-plane magnetization presented in (a) (See Sec. 3C in the text for details on exchange coupling at the $\mathrm{Co}∕\mathrm{Ni}\mathrm{O}$ interface.) (c) Element-specific magnetization loops for Co and Ni taken with x-ray magnetic circular dichroism in total fluorescence yield mode at $154\mathrm{K}$ on an antiferromagnetically coupled, series 1 sample with a $12\mathrm{\AA }$ thickness of NiO. (d) Possible NiO spin configuration for an antiferromagnetically coupled sample in the antiferromagnetic state $\left(③\right)$ and ferromagnetic state $\left(④\right)$, leading to the net Ni out-of-plane magnetization presented in (c).

Figure 3

Measurement of the secondary electron attenuation length in NiO. The total electron yield signal arising from the Co $L_3$ resonance on a ${[\mathrm{Co}∕\mathrm{Pt}]}_3∕\mathrm{Ni}\mathrm{O}∕\mathrm{Cu}$ sample is measured at room temperature, where the NiO thickness varies with position. The exponential fit indicates an attenuation length for secondary electrons in NiO of $3.9\mathrm{\AA }$. The inset shows the profile of the NiO wedge, characterized with low-angle x-ray reflectivity.

Figure 4

XMCD signal at the Ni $L_3$ resonance for the 8, 10, 11, and $12\mathrm{\AA }$ samples, taken at $175\mathrm{K}$. The integral over this signal gives a measure of the net out-of-plane magnetization in the antiferromagnetic NiO. Inset: The magnitude of the integral over the dichroism signal vs the magnitude of the coupling strength for the given sample. A larger out-of-plane signal arises for the AFM coupled samples when compared to the FM coupled sample.

Figure 5

X-ray magnetic circular dichroism-photoemission electron microscopy images taken at room temperature at the Co and Ni $L_3$ resonances on a virgin, antiferromagnetically coupled, $8\mathrm{\AA }$ sample from series 2. This technique images ferromagnetic domains in both the top Co and buried NiO layers. There is exact coincidence in the domain structure of the Co and NiO. Arrows indicate the position of coincident domains.

Figure 6

XMCD-PEEM images taken at room temperature at the Co $L_3$ resonance on virgin samples representing various NiO and Pt thicknesses. Due to attenuation this measurement is only sensitive to the top Co layers. The top two images represent two Pt thicknesses (5.1 and $11.8\mathrm{\AA }$), where the NiO thickness was set to $8\mathrm{\AA }$. The lower five images represent varying NiO thicknesses (8, 9.5, 10.5, 11, and $12\mathrm{\AA }$). The plot shows the average domain size of these samples as a function of their coupling strength. The average domain size tends to increase with increasing coupling strength, no matter how this variation in coupling is attained (varying NiO or Pt thickness).

Figure 7

(Color online) Room temperature MFM images of samples with 8, 10.5, 11, and $12\mathrm{\AA }$ NiO thicknesses. The 8 and $12\mathrm{\AA }$ samples are antiferromagnetically coupled, up and down domains disappear and only a domain overlap region is observed. The $10.5\mathrm{\AA }$ sample is ferromagnetically coupled and only up and down domains are observed. The $11\mathrm{\AA }$ is very weakly coupled (slightly antiferromagnetic). The domain overlap that occurs in the antiferromagnetically coupled samples grows with decreasing coupling strength, where the $8\mathrm{\AA }$ is the strongest and $12\mathrm{\AA }$ is the most weakly coupled sample. The orientation flips from up to down along the overlap to minimize magnetostatic energy.

Figure 8

The model domain structure for the two $\mathrm{Co}∕\mathrm{Pt}$ multilayers. The view is in the plane of the film along the stripe domains. The dimensions are defined in the text.

Figure 9

The dependence of the equilibrium domain overlap $\delta$ on the interlayer exchange coupling $J_{\mathrm{IEC}}$. The open circles are results of the numerical calculation for periodic stripe domains, and the curve calculated directly from Eq. (11). Arrows indicate specific values for $J_{\mathrm{IEC}}$ as defined in the inset. The closed circles show the measured $\delta$ values for the 12 and $8\mathrm{\AA }$ samples, as defined in the text. Inset: The variation in the total energy versus the overlap $\delta$ for three $J_{\mathrm{IEC}}$ values (a) 0.015, (b) 0.033, and (c) $0.1\mathrm{erg}∕{\mathrm{cm}}^2$, where (a) and (b) correspond to the coupling for the 12 and $8\mathrm{\AA }$ samples, respectively.

Figure 10

(a) A plot of the room temperature minor loop shift after heating to a specified temperature, indicated on the horizontal axis. This indicates permanent, irreversible changes in the exchange coupling due to heating, where these changes increase with increased heating. (b) Low-angle x-ray reflectivity taken on an $8\mathrm{\AA }$ sample from series 2 before and after a $468\mathrm{K}$ heating showing minimal change in the multilayer structure due to diffusion and no evidence of increased roughness. The inset of (b) shows no change in the intensity of the x-ray diffraction at the NiO fcc(111) peak before, at and after a $468\mathrm{K}$ heating. (c) XAS data taken at the Co $L_3$ resonance before and after (d) heating to $468\mathrm{K}$. The presence of CoO after heating is evident in (d).

Figure 11

Interlayer exchange coupling as a function of temperature for the 8 and $12\mathrm{\AA }$ samples, which couple antiferromagnetically. Two plots are given for the $8\mathrm{\AA }$ sample. The $H_{\mathrm{REV}}$ data are the reversible component of the temperature dependence, obtained as explain in the text. The 8 and $12\mathrm{\AA }$ samples exhibit a decrease in interlayer exchange coupling with increasing temperature.