Jose, California 95135, USA 94035, USA

Transcription

1 Extracting magnetic cluster size and its distributions in advanced perpendicular recording media with shrinking grain size using small angle x-ray scattering Virat Mehta 1, Tianhan Wang 2,3,Yoshihiro Ikeda 1, Ken Takano 1, Bruce D. Terris 1, Benny Wu 3,4, Catherine Graves 3,4, Hermann Dürr 3, Andreas Scherz 3, Jo Stöhr 3 and Olav Hellwig 1 1 San Jose Research Center, HGST a Western Digital company, 3403 Yerba Buena Rd., San Jose, California 95135, USA 2 Department of Materials Science and Engineering, Stanford University, Stanford, California 94035, USA 3 Stanford Institute for Materials & Energy Science (SIMES), SLAC National Accelerator Laboratory, 2575 Sand Hill Road, Menlo Park, California, 94025, USA 4 Department of Applied Physics, Stanford University, Stanford, California 94035, USA We analyze the magnetic cluster size (MCS) and magnetic cluster size distribution (MCSD) in a variety of perpendicular magnetic recording (PMR) media designs using resonant small angle x- ray scattering (SAXS) at the Co L 3 absorption edge. The different PMR media flavors considered here vary in grain size between 7.5 and 9.5 nm as well as in lateral inter-granular exchange strength, which is controlled via the segregant amount. While for high inter-granular exchange the MCS increases rapidly for grain sizes below 8.5 nm, we show that for increased amount of segregant with less exchange the MCS remains relatively small, even for grain sizes of 7.5 and 8 nm. However, the MCSD still increases sharply when shrinking grains from 8 to 7.5 nm. We show evidence that recording performance, such as signal-to-noise-ratio on the spin stand correlates well with the product of magnetic cluster size and magnetic cluster size distribution. 1

2 Historically, areal density increases in longitudinal hard disk drive media technology have been driven by reduction of grain size. However, since its introduction in 2006, the perpendicular magnetic recording (PMR) media grain size has remained more or less constant at ~9 nm [1]. PMR media consist of CoCrPt-based granular magnetic thin films, which are phase segregated with a non-magnetic oxide at the grain boundaries [2,3]. Along with other improvements this geometry allows for a tighter packing of bits compared to longitudinal media and has resulted in a six-fold increase in areal density (150 Gb/in Gb/in 2 ) from 2006 till today without significant changes in grain size. To drive this technology to its sustainable limits in terms of areal density, a deeper understanding of the trade-offs between structural grain size and magnetic cluster size and their corresponding distributions is required. This understanding is especially important for the management and optimization of inter-granular exchange coupling, which could potentially inhibit any gains in areal density capability as grain sizes and grain boundary widths are reduced. Micromagnetic modeling shows that instead of grain size alone, it is the exchange and the magnetostatic fields from grain size distributions and exchange coupling between grains that have more relevant effects on overall recording performance and thermal stability [4]. So it is not surprising that most of the improvements in the media that have led to areal density progress were driven by an increasingly complex and more functional magnetic layer structure. Specifically, the addition of a lower anisotropy magnetic cap layer (MCL) [5,6] that is deposited on top of the higher anisotropy granular recording layer (GRL) provides a more continuous and uniform reversal assist [7,8]. By separately tuning the magnetic properties and microstructure in the GRL and the MCL [9,10,11] to influence the anisotropy and the inter-granular exchange in 2

3 the media we can control the magnetic correlation length also referred to as magnetic cluster size (MCS) and its distribution (MCSD) to achieve the best overall media performance. In a previous study we compared various generations of PMR media and have shown that using a finely-tuned degree of lateral exchange via a MCL, the MCS can be reduced by decreasing the lateral exchange at the cost of an increased MCSD [12]. From this earlier work, we found that reducing the grain size alone is no longer a viable approach towards achieving higher areal densities. In addition a careful control of MCS and MCSD is needed as well. Here, we demonstrate for a systematic set of PMR media flavors that the joint variation of structural grain size and inter-granular exchange in the GRL may be used to tune the MCS and MCSD in a controlled manner. All samples were deposited onto soft x-ray transparent 100 nm thick SiN membrane substrates in order to perform small angle x-ray scattering (SAXS) experiments in transmission geometry. A similar set of samples deposited on regular hard disk drive glass substrates has been explored earlier [13] using the spin stand and a number of different traditional laboratory techniques. Such techniques, which include transmission electron microscopy [14] and minor hysteresis loop analysis [15] can provide information on grain size, grain size distributions, and indirectly also MCS [15], but yield little information on MCSD. Meanwhile, SAXS is a powerful synchrotron based characterization technique that provides information about magnetic and structural correlation length within the various layers of the recording media [16-19], so that we can simultaneously probe all of the critical sizes and their distributions in one set of experiments. To achieve this, we probe resonant magnetic as well as charge (non-magnetic) scattering with linearly polarized x-rays at the Co L 3 absorption edge in a transmission geometry as described in our previous work [12]. 3

4 The details of the SAXS setup used for measurements taken at BL13.3 at the Stanford Synchrotron Radiation Light source (SSRL) are described in a previous publication [12]. For this study we focus on better understanding limitations in the current PMR media design in order to develop future optimization strategies. We explore in detail two key parameters, namely grain size and lateral magnetic exchange, which among others control areal density capability. The first parameter involves controlling the seed layer for the media layer to investigate the role of different grain size and grain size distribution of the GRL. The second parameter involves control over the segregant to influence the lateral exchange in the GRL layer. For this study we explored 5 nominal GRL grain sizes (7.5, 8, 8.5, 9 and 9.5 nm) matrixed with three different degrees of inter-granular exchange controlled by using different amounts of segregant, namely high exchange (HE), medium exchange (ME), and low exchange (LE). In order to understand the effects of the MCL on top of these structures and for separating out its effects on the GRL grain size, we explored both, samples grown with and without the MCL. Corresponding resonant scattering example profiles for low, medium and high degree of lateral exchange are shown in Fig. 1 for the largest (9.5 nm) and smallest (7.5 nm) nominal grain sizes. After measuring the q-space profiles from each of the 30 different AC-demagnetized [20] media samples, the grain size, MCS, grain size distributions, and MCSD were extracted by fitting the grain and cluster size peaks as illustrated in Fig. 1c to log normal distributions. The data was then compiled and is summarized in Fig. 2. As expected, the samples nominal target grain size and their measured grain size extracted from SAXS data match up fairly well and the addition of the MCL does little to alter the measured grain size from SAXS. The measured relative grain size distribution gets broader at smaller grain size within each series, and here as well the MCL has little effect on that trend. The 4

5 LE, ME, and HE segregant does not alter the structural grain size and grain size distribution, so all three series show a similar dependence with targeted grain size as shown in Fig. 2. For the MCS there are differences in the trends as a function of nominal target grain size across the three different inter-granular exchange coupling series. The LE, ME, and HE series all show an increased MCS and an increased MCSD as the nominal target grain size is reduced. The LE series shows the weakest MCS and MCSD trend as a function of grain size. Most dramatic changes of MCS and MCSD for shrinking grain size are observed for the HE segregant series. Here the MCS for the GRL only shows a dramatic step like increase, when moving from 8.5 nm to 8 nm nominal target grain size, indicating that the exchange de-coupling between adjacent grains breaks down at this point due to the low amount of segregant in this series. Once adding the MCL this dramatic step-like effect in MCS is mostly masked by the large amount of additional lateral exchange supplied by the MCL. However the lateral exchange intrinsic to the GRL introduces a wider MCSD than the exchange introduced by the MCL, as can be seen in Fig. 2b, where MCSD with MCL reaches the highest value for small grain size (7.5 nm). Unlike for structural grain size and grain size distribution the MCL appears to have a significant impact on the MCS and MCSD in each of the three sample series. The addition of the MCL increases the MCS and significantly reduces the MCSD for all samples. It is worth noting that the overall relative trend for MCS as a function of nominal target grain size is roughly maintained with and without the MCL layer. But on the other hand the absolute values of the MCSD are strongly reduced by addition of the MCL, as the amount of exchange provided to each grain with respect to its neighbors can be precisely controlled by the thickness and magnetization of the cap layer. The reduction in MCSD due to the MCL is even more pronounced at smaller grain size. The lowest MCS with the tightest MCSD at a given grain size 5

6 is highly dependent on the amount of exchange in the GRL and whether we consider a structure with or without MCL. Fig. 3 summarizes the trends in MCS and MCSD as a function of lateral exchange strength for small and large grain size. For large grains the impact of intergranular exchange is weaker, since the MCS is largely governed by the strong demagnetization of the larger grains and thus tracks closely with the overall structural grain size. But for small grains the demagnetization field from each grain is much smaller and the intergranular exchange between grains becomes more important. Only by keeping the intergranular exchange low, the MCS and MCSD can be kept low as well. When the exchange is high multiple grains switch together and the MCS and MCSD get large (much larger than twice the structural grain size). Intriguingly, there is no clear advantage in getting a smaller MCS when moving to smaller grains. All three series show a larger MCS with decreasing grain size. However, the MCSD is also affected by the intergranular exchange and the grain size. So depending on how important the MCSD and especially its shape and its tails are for the recording performance, it may be advantagous to choose an intermediate grain size with moderate amount of intergranular exchange to keep the MCSD low, while maintaining an overall small MCS. When the grain size gets too small (nominal target grain size 7.5 nm), the exchange between grains seems to be difficult to control and the MCS and the MCSD increase for any amount of intergranular exchange probed in this study. However for both low and medium exchange the MCSD reaches a clear minimum at 8 nm grain size for full media structures. In order to take into account that both, MCS and MCSD are important factors for recording performance, we define in Fig.4 an effective magnetic cluster size by multiplying the measured MCS with a normalized MCSD factor that is 1 for the sample with the lowest 6

7 MCSD and correspondingly higher for the other samples (MCSD sample /MCSD min ). Fig. 4a confirms that a minimum effective magnetic cluster size is reached for intermediate nominal grain sizes of 8 nm for the low exchange and medium exchange sample series and 9 nm for the high exchange sample series. Furthermore Fig. 4b reveals an almost linear correlation between relative signal to noise (SNR) of spin stand disk sister samples and effective magnetic cluster size from SAXS, While low exchange yields better recording performance in Fig. 4b it is not possible to increase the amount of segregant arbitrarily when shrinking grain size. Keeping the grain boundary width constant for low exchange and reducing the grain size at the same time yields a lower magnetic (grain) filling factor. This reduces the signal to noise ratio and therefore areal density capability. For current PMR media and current read and write heads our study shows that pushing the grain size below 8 nm will be very challenging [13]. Therefore while PMR technology is approaching areal densities of Tb/in 2, it will not be able to provide densities significantly beyond that. In conclusion we have shown that based on the media structures considered here additional media areal density gains [13] originate from grain size and MCS reduction, with tight MCSD around 30%. Furthermore we have shown that this type of reduction is not a simple matter of shrinking the structural grain size, but instead can only be achieved by a careful tuning of intergranular exchange as the grain size is reduced. Here is seems important that the intergranular exchange originates from the MCL rather than the GRL in order to keep the MCSD small. Therefore it is critical to maintain low intergranular exchange i.e. well isolated grains within the GRL, which compromises in turn magnetic filling factor and therefore readback signal as well as thermal stability K U V, due to a reduced magnetic core size. 7

8 Overall achieving areal density gains has been very challenging within the current PMR recording paradigm, so that technologies such as Heat Assisted Magnetic Recording (HAMR) or Bit Patterned Recording (BPR) may be necessary in order to push areal densities significantly beyond 1 Tb/in 2. SAXS is clearly a powerful technique that provides simultaneous information about the charge and magnetic lateral microstructure in advanced PMR media systems. In particular, one key advantage of SAXS originates from being able to obtain information about the MCSD, which cannot be obtained easily from any other method and is clearly very important for achieving high areal density [12]. Comparing MCS and MCSD in the GRL only with the GRL+MCL full composite media system for media with different amounts of intergranular exchange helps not only to determine the direction for future media generations, but also reveals critical limitations of PMR technology. Research at SIMES is supported by the Department of Energy, Office of Basic Energy Sciences, under Contract DE-AC02-76SF Portions of this research were carried out at the Stanford Synchrotron Radiation Lightsource, a Directorate of SLAC National Accelerator Laboratory and an Office of Science User Facility operated for the U.S. Department of Energy Office of Science by Stanford University. 8

9 References [1] G. Bertero, R. Acharya, S. Malhotra, K. Srinivasan, E. Champion, G. Lauhoff and M. Desai, The 21st Magnetic Recording Conference (TMRC) 2010: Double exchange-break PMR media structures. [2] S. N. Piramanayagam, J. Appl. Phys. 102, (2007). [3] A. Moser, K. Takano, D. T. Margulies, M. Albrecht, Y. Sonobe, Y. Ikeda, S. Sun and E. E. Fullerton, J. Phys. D: Appl. Phys. 35, R157 (2002). [4] J. J Miles, IEEE Trans. Magn. 43, 955 (2007). [5] K. Tang, X. Bian, G. Choe, K. Takano, M. Mirzamaani, G. Wang, J. Zhang, Q.-F. Xiao, Y. Ikeda, J. Risner-Jamtgaard, and X. Xu, IEEE Trans. Magn. 45, 786 (2009). [6] R. H. Victora and X. Shen, IEEE Trans. Magn. 41, 2828 (2005). [7] D. Suess, J. Lee, J. Fidler, T. Schrefl, J. Magn. Magn. Mater (2009). [8] T. Hauet, S. Florez, D. Margulies, Y. Ikeda, B. Lengsfield, N. Supper, K. Takano, O. Hellwig, and B. D. Terris, Appl. Phys. Lett. 95, (2009). [9] T. P. Nolan, B. F. Valcu, and H. J. Richter, IEEE Trans. Magn. 47, 63 (2011). [10] G. Choe, J. Park, Y. Ikeda, B. Lengsfield, T. Olson, K. Zhang, S. Florez, and A. Ghaderi, IEEE Trans. Magn. 47, 55 (2011) [11] G. Choe, Y. Ikeda, K. Zhang, K. Tang, and M. Mirzamaani, IEEE Trans. Magn. 45, 2694 (2010). 9

10 [12] T. Wang, V. Mehta, Y. Ikeda, H. Do, K. Takano, S. H. Florez, B. D. Terris, B. Wu, C. Graves, M. Shu, R. Rick, A. Scherz, J. Stöhr, and O. Hellwig, Appl. Phys. Lett. 103 (2013) [13] Y. Ikeda, S. H. Florez, F. Q. Zhu, K. Takano, H. Do, T. Hennen, and B. D. Terris, IEEE Trans. Mag. 48, 3185 (2012). [14] K. Tang, X. Bian, G. Choe, K. Takano, M. Mirzamaani, G. Wang, J. Zhang, Q.-F. Xiao, Y. Ikeda, J. Risner-Jamtgaard, and X. Xu, IEEE Trans. Magn. 45, 786 (2009). [15] H. Nemoto, I. Takekuma, H. Nakagawa, T. Ichihara, R. Araki, Y. Hosoe, Jour. Magn. Magn. Mat. 320 (2008) [16] O. Hellwig, J.B. Kortright, D.T. Margulies, B. Lengsfield, and E. E. Fullerton, Appl. Phys. Lett. 80, 1234 (2002). [17] E. E. Fullerton, O. Hellwig, Y. Ikeda, B. Lengsfield, K. Takano and J. B. Kortright, IEEE Trans. Mag. 38, 1693 (2002). [18] E. E. Fullerton, O. Hellwig, K. Takano and J. B. Kortright NIM B 200, 202 (2003). [19] J. B. Kortright, O. Hellwig, D. T. Margulies, and E. E. Fullerton, Jour. Magn. Magn. Mat. 240, 325 (2002). [20] Demagnetization was performed by reducing the amplitude of an external magnetic field aligned normal to the sample surface from initially 15 koe down to 10 Oe in 0.1% steps, while switching the field polarity back and forth between positive and negative at about 0.5 Hz. The AC demagnetized state is commonly used to evaluate media noise and cluster size, as it should capture the media in a state with minimum possible cluster size. 10

11 Figure Captions FIG. 1: (a) SAXS profiles of the GRL layer for small (nominal grain size 7.5 nm, solid lines) and large (nominal grain size 9.5 nm, dashed lines) target grain sizes of PMR recording media designed with low (black), medium (red) and high (blue) exchange between grains. (b) Corresponding profiles from the full media structures (GRL+MCL). SAXS features from the structural grains are visible at periods between 7 and 9 nm, while SAXS features representing magnetic correlations, i.e. magnetic clusters are visible at periods of nm. The intensity scale between large and small grain media flavors has been shifted for better comparison. (c) Illustration of on-resonant and off-resonant scattering profiles, as well as the corresponding "extracted" magnetic scattering. Grain size and its distribution are extracted from off-resonant data (red), while magnetic cluster size and its distribution is obtained from extracted magnetic scattering. The profiles are offset in intensity for a clearer comparison and a linear scale is used for better visual comparison. FIG. 2: Summary of SAXS results for 5 targeted grain sizes, 9.5 nm nm, with different amounts of exchange (high=he, medium=me, low=le) between grains for GRL only and GRL+MCL full media structures. (a, c, e) Average grain and magnetic cluster sizes (MCS) determined from SAXS peak positions for HE, ME and LE sample series respectively. (b, d, f)) 11

12 Corresponding grain and magnetic cluster size distributions (MCSD) determined from fitted widths of the SAXS peaks for HE, ME and LE sample series respectively. FIG. 3: (a) Average full media (GRL+MCL) MCS for various grain sizes as a function of decreasing magnetic exchange between grains. (b) Corresponding MCSD for various grain sizes as a function of the magnetic exchange between grains. The corresponding nominal grain sizes are given in the two legends in nanometers. FIG. 4: Definition of an effective magnetic cluster size as the product of MCS and MCSD normalized by the overall minimum MCSD for the current sample series. The parameter takes into account that MCS and MCSD are both important for recording performance. (a) Effective magnetic cluster size versus average nominal grain size for the three media series. (b) Correlation between spin stand signal to noise ratio (from recording disk sister samples) and effective magnetic cluster size from SAXS for the three media series. 12

13

14

15 Magnetic Cluster Size (nm) (a) HE Series ME Series LE Series Magnetic Cluster Size Distribution (%) (b) HE Series ME Series LE Series

16 "effective" magnetic cluster size [nm] (a) high exchange medium exchange low exchange nominal grain size [nm] SNR [db] (b) high exchange medium exchange low exchange "effective" magnetic cluster size [nm]