The technology development of Heat Assisted Magnetic Recording (HAMR) is a primary focus of both the industry and academia to ensure the HDD storage capacities follow the roadmap established by the Advanced Storage Research Consortium (ASRC). One important limitation arises from media noise, which results from structure or magnetic defects of the FePt grains that constitute the HAMR medium. This includes in-plane grains with easy axes pointing in-plane instead of out-of-plane. While the presence of these grains has been established in literature through experiments, there is a lack of understanding of how these grains can impact the HAMR performance. In this modeling study, we focus on the impact of in-plane grains specifically on the HAMR performance, by varying their proportion throughout the media and capturing the pre-channel and post-channel performance parameters. We extend this analysis using the THMap process to understand the effect of in-plane grains on room temperature switching (RTS), magnetization variance and the readback signal variance. Additionally, we compare the signal, noise, low frequency SNR (LF SNR) and jitter captured through micro magnetic simulations with values obtained from an analytical grain counting model. Finally, we explore ways to alleviate the impact of in-plane grains by optimizing media parameters and identifying the sensitivity for each media parameter.

We present micromagnetic simulations at various grain sizes of four-layer exchange-coupled composite microwave-assisted magnetic recording (MAMR) systems that have maximized areal density capability using the Nelder-Mead simplex algorithm. We restrict our media to have crystalline KuV/kT > ~100. Our optimized MAMR designs show roughly 120 Gb/in 2 areal density growth per nanometer of grain pitch reduction and achieve an areal density of 37% larger than a publicly available 700 Gb/in 2 conventional perpendicular recording reference at a similar grain size and write width. We analyze all 26 optimization variables to identify common trends indicative of high-performance MAMR media. Our principle goal is to help media designers optimize media for MAMR performance, and to project how this optimization changes with the grain size. Balancing write-ability, noise performance, thermal stability, and resonance-matching characteristics is most important.

We present a micromagnetic modeling study of microwave assisted magnetic recording (MAMR) on a typical exchange-coupled four-layer media. With the help of an alternating current (ac) magnetic field, the switching field can be reduced, especially at the media resonant frequency. Typical hysteresis loop measurements employ a depth-invariant ac field; however, the field generated from a spin-torque oscillator (STO) varies with spacing. Considering this depth variance, the switching field reduction becomes less than observed with a depth-invariant ac field. Furthermore, we find an even smaller switching field reduction when using a realistic magnetic writer and STO fields to record a dc track on randomly magnetized grains. Consequently, we conclude that MAMR hysteresis loops employing a depth-invariant ac field significantly overestimate the ac field's contribution to grain switching in an actual recording situation.

In this paper, a general energy expression is developed for a cluster of multilayer exchange spring media grains. We expand the rotation of the magnetization vectors to second order in elements of an exponential unitary transformation. The first and second derivatives of the energy are determined and a damped optimization procedure is used to efficiently locate minima and transition states on the potential energy surface. In this method, the interaction of neighboring grains is taken into account and thus the effect of inter-granular exchange on the energy barrier is explicitly taken into consideration. Thus, cluster formation is seen as coherent switching of a group of grains and can be studied as a function of anisotropy and exchange distributions in the media. The lowest energy barrier for a grain or cluster is then used to construct a rate constant. We then determine rate constants for an ensemble of interacting grains from which M-H loops and magnetization decay are simulated as a function of an applied field. Both kinetic Monte Carlo and direct integration of the rate equations are used to study the kinetics of these systems. We find that thermal decay rates have better correlation with energy barriers than with nucleation fields.

The use of a wider writing shingled magnetic recording (SMR) head in spinstand experiments allowed different regions of the SMR track to be analyzed and described. Here, we break the track into curved, best, and trimmed areas. The curved area is written with the writer edge, which has high curvature and poor write field gradient, resulting in low resolution and high noise. The trimmed region is defined by head fields combining with disk demagnetization fields. Together these fields attack the low frequency magnets, boosting resolution, increasing noise, and generating a narrow out of phase copy of the trimmed track (the negative sliver). This is explained via measurements and modeling. We also explain the measured magnetic write width dependence on linear density based on the frequency response of the reader as well as writer and media effects. We employ linear deconvolution to extract the binary impulse responses of the SMR data track and its neighboring tracks to characterize intertrack interference. The writer and media effects, which give rise to the negative sliver observed, are shown to explain the interference behavior with linear density.

A systematic investigation of the switching behavior and writeability characteristics of perpendicular magnetic oxide media was performed by varying the number of oxide layers as well as the anisotropy K(u) gradient in multilayered oxides. The media switching behavior highly depends on the magnetic volume of laterally exchange-coupled cap layer as well as the vertical exchange coupling between oxide and cap layers. The media switching field is significantly reduced with incoherent switching mode. The media writeability is further enhanced by employing the multilayered oxides whose anisotropy values vary. Higher K(u) graded oxide media exhibits improved resolution, sharper transition, and higher signal-to-noise ratio. An analytical 9-spin model is used to study the impact of anisotropy grading, intergranular and interlayer exchange coupling on writeability and noise performance. The 9-spin model is found to qualitatively agree with the experimental results. Both model and experiment indicate that optimal tuning of the media anisotropy and exchange to the head design is critical in attaining good recording performance.

The effect of interlayer thickness (t IL ) on recording performance of the media showing relatively close to coherent switching behavior is investigated. The range of t IL was widely controlled from 20 to 45 nm by inserting an amorphous CrTa filling layer between NiW(6 nm)/Ru(14 nm) and amorphous FeCoTaZr soft underlayer (SUL). Minor heating effects at a thicker CrTa on a glass substrate are detected, but the change in magnetic properties and microstructure even at t IL = 45 nm is small enough to study the head-to-SUL spacing effect. A thinner IL clearly improves both reverse overwrite (ROW) and signal-to-noise (SoNR 2T ) at 2 T, but it reduces resolution (RES). Less dependence of magnetic core width (MCW) on t IL is shown due to compensation of wider magnetic write width (MWW) and narrower erasure band (EB) at a thinner IL. According to simulation results, a thinner IL enhances field strength underneath the center of the writer and more field penetration into the media but the maximum effective field located near the gap changes very little. These field profiles improve writeability at low frequencies, but they do not affect much at mid-to-high frequencies. This situation can explain better ROW, lower RES, and wider MWW at a thinner IL. Higher crosstrack field gradient at a thinner IL corresponds to narrower EB. The improvement of SoNR 2T at a thinner IL is caused by the increase in signal at low frequency and the further decrease in dc noise. All the recording parameters with increasing t IL are qualitatively consistent with simulation results.

This paper examines trends in magnetic spacing in the hard disk drive industry for the last 15 years or so, as we transitioned from inductive to magnetoresistive (MR), giant magnetoresistive (GMR), and tunneling magnetoresistive (TMR) heads, and from longitudinal to perpendicular recording. Historical data show surprisingly good scaling relationships between areal density, linear density, track density, bit aspect ratio (BAR), and magnetic spacing. These trends have held from below 1 Gb/in 2 until today's densities, with no significant discontinuities when crossing major technology changes. In particular, it is found that magnetic spacing has scaled with the length of the bit, and more specifically has equaled about half the bit length. Magnetic spacing components, on the other hand, show larger degree of deviations from the trend. Projections to 1 and 10 Tb/in 2 are made.

Recent angular VSM measurements indicate that the remanent coercivity of various types of longitudinal media has an applied field angular dependence that approximates the behavior of Stoner–Wohlfarth particles with easy axes oriented isotropically in-plane. This is in contrast to the Kondorsky style switching (in-plane field, domain wall switching) inherently assumed in most models of the write process. In this article, we incorporate the measured angular dependent switching into a Williams–Comstock style write model. We find that the vertical spacing dependence of the transition width or “a” parameter can be much lower than that expected when only the in-plane component of the head field participates in switching the media. This phenomena is verified experimentally by writing at different spacings and reading at a constant spacing. We find that the experimental data can only be matched if the measured angular dependent switching is taken into account.