SNR improvement by variation of recording and media parameters for a HAMR exchange coupled composite media

An exchange coupled composite media structure proposed previously seems to address both the issue of Tc variation in FePt as well as poor SNR/User Density during the HAMR process. Here we examine a thinner 3-6 nm structure that is likely easier to fabricate than the previous 13.5 nm thick structure. We find that increasing the damping within the write (superparamagnetic) layer and introducing intergranular exchange within the grains in the write layer are both successful approaches to improve the recorded SNR. Ensemble waveform analysis that allows the breakdown of the total SNR into transition SNR (due to AC noise) and remanence SNR (due to DC noise) helps identify the leading causes for this SNR improvement. Further studies indicate that varying the peak heat spot temperature in the HAMR write process is also a successful approach for improving the recorded SNR. This lends credence to the idea that a thinner composite media may still be used successfully to realize significant enhancements of SNR and th...

An exchange coupled composite media structure proposed previously seems to address both the issue of T c variation in FePt as well as poor SNR/User Density during the HAMR process.Here we examine a thinner 3-6 nm structure that is likely easier to fabricate than the previous 13.5 nm thick structure.We find that increasing the damping within the write (superparamagnetic) layer and introducing intergranular exchange within the grains in the write layer are both successful approaches to improve the recorded SNR.Ensemble waveform analysis that allows the breakdown of the total SNR into transition SNR (due to AC noise) and remanence SNR (due to DC noise) helps identify the leading causes for this SNR improvement.Further studies indicate that varying the peak heat spot temperature in the HAMR write process is also a successful approach for improving the recorded SNR.This lends credence to the idea that a thinner composite media may still be used successfully to realize significant enhancements of SNR and the corresponding user density.© 2017 Author(s).All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).https://doi.org/10.1063/1.5007072 The most promising technology for ensuring high-density (up to 5 Tb/in 2 ) recordings is HAMR (Heat Assisted Magnetic Recording). 1 The demand for higher density recording suggests a reduction of grain size down to 4.5 nm and more uniform grain size distributions (GSD) down to 15%.3][4] Fabrication issues leading to T c and H k variation among the grains of the recording media can cause great variation in the intrinsic magnetic parameters of individual grains. 5n exchange coupled composite media recently proposed by Liu and Victora 6 addresses both the issues of T c variation of FePt and poor SNR/User density in the recording process.Simulations show that the 4.5nm (write layer) -9nm (storage layer) is an optimal structure considering jitter variation and corresponding Bit Error rate (BER) and User Density (UD). 8However fabrication issues force us to restrict layer dimensions to 3nm (write layer) -6nm (storage layer) structure with a consequent increase in transition jitter and a decreases in recorded SNR values.The subsequent part of the paper focuses on different techniques used to improve the recorded SNR.In addition, the recorded SNR has been broken down into transition SNR (affected by the transition noise where transitions are written at erroneous places) and remanence SNR (affected by DC noise or thermal fluctuations).This breakdown helps illustrate the effect of each of the recording parameters on the recorded SNR and suggests the cause for improvement in the same.
The design of the composite structure is different from other proposed media structures in two aspects (i) The write layer has substantially high Curie temperature (> the Curie temperature of the storage layer) and moderate anisotropy ensuring successful writing at high temperatures (ii) The writing process explicitly uses superparamagnetic writing, which means writing with thermal fluctuations. 7This means that the optimization techniques for the composite media design can involve both the write layer and the storage layer.Based on previous simulations, the effects of varying some of the factors implicit in the recording process like the magnetic fly height of the head from the medium, shield-to-shield spacing, bit length, applied head field magnitude, and the head velocity, on the recorded SNR are already known. 8This paper focuses on factors internal to the media design like the damping, the intergranular exchange, and some external recording conditions like the peak temperature of the heat spot incident on the media.To explore the SNR improvement techniques in greater detail, micromagnetic simulation 9 based on the Landau-Lifshitz-Gilbert (LLG) equation is implemented.The renormalized cell size is 1.5nm×1.5nm×1.5nm.The composite structure has two layers, a 3 nm superparamagnetic writing layer and 6 nm FePt storage layer.The intrinsic properties of the storage layer are maintained similar to Ref. 7. The magnetic profiles of the write layer are similar to Ref. 7 except that a small typo has been corrected.The arguments for M s,wl/sl , K u,wl/sl & A ex,wl/sl are defined to be (T-300)/(T c(wl/sl) -300) where 300 is room temperature.For example, a temperature of 300 • corresponds to 0. Using this notation, we can write the storage layer properties: M s,sl (0) = 922.3emu/cm 3; K u,sl (0) = 4.11 × 10 7 erg/cm 3 and A ex,sl (0) = 1.1×10 -6 erg/cm and the write layer profile becomes In this paper, we use T c,wl = 900K, T c,sl = 700K, M s,wl (0) = 550.0emu/cm 3 ; K u,wl (0) = 1.0 × 10 7 erg/cm 3 .The exchange coupling between these two layers is taken to be A ex,wl * A ex,sl .In this paper, all the calculated SNR values are based on the playback signals obtained by writing on 8 different magnetic media with an average grain pitch (GP) of 5.5 nm.A 31 bit Pseudo Random Bit Sequence (PRBS) generated by a generator polynomial x 5 + x 3 + 1 represents the head field profile.Each bit of the polynomial is denoted by either 1 (positive applied field) or 0 (negative applied field) with the field being applied at an angle of 22.5 • to the easy axis direction (+z).Appropriate padding bits are used at either end of the sequence to avoid edge effects of the media.
To understand the reasons behind the variations in the calculated SNR, the total SNR (calculated SNR/spatial SNR) is further broken down into transition SNR and remanence SNR. 10 The spatially noise-free signal is simply an average of all spatially noisy signals.The total spatial noise is calculated by subtracting each noisy signal from the averaged noise free signal.Total SNR is calculated from the definition using the power of the noise free signal and the total power of the different spatially noisy signals.The transition and remanence contributions to the total SNR are obtained by applying appropriate windowing functions to the total spatial noise.To obtain the transition component of the spatial SNR, the windowing function for transition noise is non-zero close to the transitions and zero everywhere else.The remnant noise is the difference between the total noise and the transition noise.
Damping variation in write layer: To understand the effect of varying the damping in the write layer of the composite media, the damping is scaled uniformly across the write layer.This scaling is denoted as the damping scaling factor in Fig. 1(a) The effect of damping increase can be understood as follows: within the limited time that the heat spot is incident on a certain bit on the media, increasing the damping ensures that the grain is switched as the Zeeman energy due to the applied head field overcomes the anisotropy energy at high temperatures.Higher damping values ensure that more grains switch at these temperatures.This increases the calculated SNR as shown, by almost 1.5 dB thus indicating a decrease in the noise power by almost 30%.However, increasing the damping after a certain value may delay the write process at the write temperatures where damping is already enhanced. 9This results in decreasing the SNR to a certain extent (within error bars).The total SNR calculated from the ensemble waveform analysis is denoted as spatial SNR in the same figure.A comparison between the two shows accuracy at every data point within the given error bars of the calculated SNR.It can be seen that the total SNR values agree to a good degree even for the limited number of simulations.The SNR variation clearly denotes that the transition SNR follows the trend of the total SNR thus indicating that the SNR variation process is largely affected by the transition/jitter noise.The strong variations in the remanence SNR is likely caused by inadequate number of bits to fully represent this much smaller noise.
The effect of damping variation on SNR is extended to include the entire composite structure (both the write and the storage layer) in Fig. 1(b).The trends are very similar except at the largest values, which are no longer detrimental.The results suggest that optimization of damping only need be done in the write layer, which is helpful given that the write layer also has more design flexibility, and thus opportunity for optimization.
Write layer intergranular exchange variation: As the intergranular exchange is increased, adjacent grains start coupling together to form clusters.In the presence of the heat spot, these grains that form clusters have a higher thermal stability factor (K u ×V/k B T) and hence are less susceptible to thermal fluctuations and superparamagnetic effects due to low Ku values.This leads to successful switching under the applied head field and ensures fewer stray transitions.It should reduce remanence noise.However as the intergranular exchange increases beyond a certain value, the particles form larger clusters that create noise, particularly jitter.To observe the effect of varying intergranular exchange coupling (IGC) in the write layer, the IGC is uniformly scaled for all the grains in the write layer only.As expected, from Fig. 2(a), the calculated SNR shows an increasing trend to start with, with a significant increase in the SNR of almost 2.2 dB and a consequent decrease as the IGC goes beyond 10%.The agreement between the calculated SNR values and the spatial SNR values remains.Both the transition SNR values and the remanence SNR values follow similar trends as the spatial SNR value.However, their absolute magnitudes help us ascertain that the effect of the transition/jitter  noise is the deciding factor yet again in causing the aforementioned SNR variation.To extend our understanding, the intergranular exchange is varied throughout the composite structure uniformly in Fig. 2(b).The calculated SNR initially increases as the intergranular exchange is increased to 5% of the exchange coupling values in each of the layers.The increase is almost 1.5 dB denoting a decrease of almost 30% in the noise power of the playback signal.The SNR then decreases as the IGC value increases.Interestingly, the peak is found at a lower value of IGC, presumably because all layers are now exchange coupled.This limits the overall SNR increase because the higher values of exchange are no longer accessible: it suggests that exchange coupling should be limited to the write layer and the FePt grains should continue to be magnetically isolated.
Variation of peak heat spot temperature: In HAMR, a laser source is incorporated in and moves together with the write head, and heat energy is delivered to the recording layer through a near-field transducer.1 A small rate of temperature change is important for successful grain switching.Usually, in HAMR, we also desire to have a large temperature gradient for sharp transitions.These competing factors cause the most optimum heat spot temperature to be greater than the writing temperature of the medium within a few hundred Kelvin.In the micromagnetic simulation, the peak heat spot temperature is varied up to a maximum value of 900 K as based on previous simulations for single layer FePt.Three different cases are addressed here: (i) Media with no T c & H k variation (ii) Media with 3% T c variation and no H k variation (iii) Media with 3% T c variation and 15% H k variation.The variation of the calculated SNR for no T c and H k variation values shows a slight increase in the SNR as the peak spot temperature reaches 800K and then a slight fall (Fig. 3(a)).Extending the simulation to include the media with T c & H k variation produces an interesting observation: the maxima in SNR is observed to shift to the right as the peak temperature is varied.
The effect is shown in Fig. 3(b) where, in both cases, the maximum SNR is observed to shift towards a peak temperature of 850 K as compared to 800 K in the absence of any T c /H k variation.Increasing the peak temperature and maintaining the same write temperature (which happens when intrinsic media variations are included) ensures that writing takes place at a higher thermal gradient (farther away from the peak temperature point) in the temperature profile as compared to the case with no media variations.This helps explain the shift of the SNR maxima towards the right at a higher peak temperature.
In conclusion, a reduction of the media thickness affects both the SNR as well as the jitter of the composite structure.Optimizing the intergranular exchange and damping can increase the SNR values to make them comparable to the ones originally obtained for the thicker 4.5nm-9nm composite structure.Calculations show that the remanence SNR is largely negligible in this structure.Finally, the role of granular variation is identified in optimizing the peak temperature.
FIG. 1.(a) Variation of calculated SNR and individual components vs write layer damping (b) Variation of calculated SNR and individual components vs damping of the whole structure.
FIG. 2. (a) Variation of calculated SNR and individual components vs write layer intergranular exchange (b) Variation of calculated SNR and individual components vs intergranular exchange of the whole structure.

FIG. 3 .
FIG. 3. (a) Variation of calculated SNR and individual components vs heat spot peak temperature (no T c /H k variation in the media) (b) Effect of T c and H k variation on SNR.