Interface time:  Finding the source of perpendicular magnetic anisotropy

Interface time: Finding the source of perpendicular magnetic anisotropy

As touched upon in last week’s Pick of the Week, perpendicular magnetic anisotropy (PMA) may be important for spin transfer torque MRAM (STT-MRAM) as it allows for a smaller size of each memory bit, greater thermal stability, and smaller write current.  It is therefore crucial to gain a deeper understanding of PMA in relevant magnetic stack layers in order to be able to enhance its properties for memory device applications.  This week’s pick, “Contribution of individual interfaces in the MgO/Co/Pd trilayer to perpendicular magnetic anisotropy upon annealing”, by Kim et al. in Applied Physics Letters reveals new insight into the source of PMA in the commonly used oxide/ferromagnet/heavy metal (O/FM/HM) stack structure.


Interface the facts

The O/FM/HM stack has been used in various experiments to investigate three different phenomena: PMA, spin-orbit torques (SOTs), and the Dyaloshinskii-Moriya interaction (DMI).  All of these share the same common origin of interfacial spin-orbit coupling (SOC).  This interfacial SOC is induced by charge transfer at the O/FM and/or FM/HM interface(s) due to orbital hybridization between the different layers.  One of the key questions addressed by Kim and co-workers in their paper is whether the PMA originates from the O/FM or FM/HM interface for the commonly studied MgO/Co/Pd stack.


XXX viewing

For the experimental investigation, MgO(5nm)/Co(tnm)/Pd(3nm) stacks were grown by UHV sputtering on Si substrates.  The thickness (t) of the Co layer varied from 0.8nm-2.4nm.  Interface SOC depends on the chemical and crystalline structure of the interface, and this can be varied by annealing the stack.  Stacks were annealed at either 200oC, 300oC, or 400oC in an out of plane magnetic field of 4.5kOe.  Magnetic annealing is a big business and a common step in MRAM chip production, and therefore its use to tailor PMA in this experiment could be compatible with mass-scale MRAM fabrication.  The paper does not state how the PMA of the samples was measured, but presumably it was by a standard magnetometry technique.  Three types of x-ray measurements were also made on the samples to obtain chemical (x-ray absorption spectroscopy –XAS), structural (x-ray diffractometry – XRD), and spin and orbital moment (x-ray magnetic circular dichroism – XMCD) information.


The incredible bulk

All of the as-grown samples had in-plane magnetic anisotropy, whereas all of the samples annealed at 400oC had PMA.  For the samples annealed at 200oC and 300oC, samples with a thinner Co layer (less than around 2nm) had PMA.  While the PMA is induced by an interface, it can be enhanced by a bulk like contribution of the Co layer which can be phenomenologically separated from the interface contribution.  By making XAS and XRD measurements on the various samples, Kim came to the following conclusions:

  • The bulk PMA contribution increases with annealing (though is similar for 200oC and 300oC annealed samples). This increase occurs as annealing reduces the number of defects in the MgO layer.  Reducing the defects in MgO reduces its density, and this leads to an in-plane tensile stress arising in the adjacent Co layer which changes its magnetic anisotropy due to the magnetoelastic effect.
  • Unlike the bulk PMA contribution, the interface contribution increases by around 50% between samples annealed at 200oC and 300o While the MgO/Co interface appears chemically the same for these two annealing temperatures, the Co/Pd interface does not, and so must be the interface responsible for this increase.
  • For annealing at 400oC, the interfacial PMA contribution decreases from that of the as-grown (and hence 200oC and 300oC) samples. This is believed to be because the annealing temperature is sufficient to form a CoPd alloy.


Not looking on the oxide of life

XMCD measurements were made at the Co L-edge.  It was found that the interfacial structure of MgO/Co plays no significant role in changing PMA, and therefore the origin of the PMA is attributable to the Co/Pd interface.  Application of the sum rules showed that the ratio of Co’s orbital to spin magnetic moment increases with annealing.  This result is expected as the ratio is reflects the SOC energy of the Co atoms – annealing increases SOC which in turn enhances PMA.


The moment to stand up

What I like about this study is the use of the three different spectroscopic techniques to piece together the observed differences in PMA for the various samples.  While results show that the PMA is bought about at the FM/HM interface, they also show that it can be significantly enhanced by the O layer.  This information could be important when considering suitable stack structures for use in STT-MRAM, where O commonly forms the tunnel barrier needed for large signal reading.  The characteristics of O should now not only be considered for electronic purposes on the tunnel magnetoresistance read-out, but also as a tool for enhancing the PMA of FM.  The addition of a HM layer to initiate the PMA in an STT-MRAM stack also now appears more important than ever, with its characteristics with annealing better defined.  Other HM layers should be used in similar studies of trilayer stacks to that of Kim’s work in order to help find the optimum stack for an STT-MRAM bit with PMA.


Image: Jacob Earl, some rights reserved

Kim, M., Kim, S., Ko, J., & Hong, J. (2015). Contribution of individual interfaces in the MgO/Co/Pd trilayer to perpendicular magnetic anisotropy upon annealing Applied Physics Letters, 106 (10) DOI: 10.1063/1.4914497