Systemic Evaluation of Microwave Assisted Magnetic Recording

Microwave Assisted Magnetic Recording (MAMR) is a type of energy assisted recording technology that uses microwaves to assist in the recording process. This allows for the use of high anisotropy media to maintain good thermal stability at high areal densities [1]. In MAMR, a spin torque oscillator (STO) is placed between the main pole and trailing shield to generate a magnetic field at microwave frequencies. The magnetic recording trilemma [2] is overcome by reducing the coercive field of the media via this microwave field, thereby aiding in the switching of media grains. Appropriate media stack configurations for MAMR and STO optimization have been investigated by both analytical theory and micromagnetic simulation [3], [4], [5]. In this paper, MAMR performance is evaluated on a reference design using our MAMR micro-magnetic model and the results are reported using EWSNR metrics [6]. In this study, we compare performance metrics for MAMR with other recording technologies such as Perpendicular Magnetic Recording (PMR) and Heat Assisted Magnetic Recording (HAMR). Recording media dynamics are modeled using the Landau-Lifshitz-Gilbert (LLG) equation for PMR and MAMR, and utilizing the renormalized LLG method for HAMR [7]. The DC magnetic write field is a current generation PMR writer design with 55 nm physical pole width and 1.0 T peak field strength. The STO stack has an optimized geometry that applies a 3D circularly polarized AC field to the media. STO field strength (∼0.1Hk) and oscillating frequency (∼35GHz) are determined by the magnetic properties of the field generation layer (FGL) and the magnitude of the injected spin current. The magnetic head to media spacing (HMS) is fixed at 6.0 nm and head velocity is modeled to be 20 m/s. For MAMR, single layer media with varying anisotropy are considered. The PMR ECC multilayer media model, which is calibrated based on current PMR products, contains multiple magnetic layers and non-magnetic break layers that provide optimal exchange coupling. The HAMR media is a single layer L10 FePt media with a Curie temperature distribution of 3% and anisotropy field distribution of 10%. The down track thermal gradient used in the HAMR model is around 8K/nm, which is consistent with common near field transducer designs. The average media grain size is around 8.0 nm with 17% grain size distribution. A magnetoresistive reader with 30 nm width is used for obtaining the play back signal. Magnetic information is encoded using pseudo random bit sequences (PRBS), which mimic real user data. Figure 1 shows a comparison between PMR, HAMR and MAMR with thick free layer designs in terms of common performance metrics. Ensemble waveform analysis is used to calculate the total spatial SNR, the breakdown between transition and remanence SNR contributions, and the channel bit density (CBD) [6]. Bit error rate (BER) is calculated using a pattern dependent Viterbi detector [8]. The use of an NFT in HAMR allows the recording of much narrower tracks than PMR, with similar CBD and reasonable SNR and BER. In general, MAMR exhibits a large CBD, which may be related to intersymbol interference and track edge erasure caused by the demagnetization field. This high CBD results in much worse BER than both HAMR and PMR at all track width and linear density combinations considered. For the same ADC (shown in the last two table items in figure 1), SNR and BER are much better when the linear density is higher and the track width is larger. Since track edge demagnetization effects seem to have such a large effect in MAMR, this data suggests MAMR should aim to increase ADC via higher linear density and lower track density. For thick free layers, multi domains and multi-scattering contributions to electron’s propagation can greatly deteriorate STO performance. Therefore, SNR versus recording media Hk for MAMR with thin free layer designs below 15 nm and STO width of 40 nm is shown in Figure 2. The figure suggests an optimal for MAMR media of around 25 kOe. MAMR SNR degradation at 20 kOe is due to erasure effects for low Hk material [9]. For high anisotropy media ∼30 kOe, the transition SNR is reasonable (around 12 dB), but remanence noise is high, causing a 2 dB loss in total SNR.