CHARACTERIZATION OF LASER INDUCED PROTRUSION IN HAMR BY THE BURNISH METHOD

Transcription

1 A3 1 CHARACTERIZATION OF LASER INDUCED PROTRUSION IN HAMR BY THE BURNISH METHOD Zhenyi Zhang 1, Kowang Liu 1, Eric X. Jin 1, Moris M. Dovek 1, James D. Kiely 2, Osamu Nakada 3, Pak Kin Wong 4, Vincent Man Fat Chiah 4, Xiao Ming Liu 4 1 Headway Technologies, Milpitas, CA, USA 2 Seagate Technologies, Bloomington, MN, USA 3 TDK Asama Techno Factory, Saku-shi, Nagano, Japan 4 SAE Magnetics, Hong Kong, China Heat-assisted magnetic recording (HAMR) has become one of the leading approaches to increase the areal density in hard disk drives to several Tb/in 2. Laser-induced protrusion significantly impacts the HAMR heads' performance and lifetime. Nonetheless, the pointy local protrusion amount is hard to accurately quantify by touch down sensors. Based on previous indirect thermal-mechanical studies with different sensors and techniques, a new methodology is developed to measure the laser induced protrusion by burnishing the head during writing with laser powered on and by utilizing atomic force microscopy (AFM) scans. This burnish method requires the air bearing surface (ABS) to physically contact the rotating disk without excessive overdrive, which relies on precise settings of touch down sensors and detection methods. The time dependence of the burnish behavior is studied. The results show that its time constant is much longer than the pure thermal and mechanical time constants. The repeatability of the experiment is validated. The local protrusion induced by the laser is in the range of 2nm ~ 5nm based on current head design. Accordingly, detailed studies about heater power effects and protrusion amount as a function of laser power have been conducted. In addition, the impact of varying degrees of stiffness of the air bearing surface is discussed. Index Terms Atomic force microscopy (AFM), heat assisted magnetic recording (HAMR), near field transducer (NFT), protrusion. T I. INTRODUCTION HE areal density of the hard disk drive based on heatassisted magnetic recording (HAMR) [1] has been gradually improving with fine-tuned near field transducer (NFT) designs and well-engineered reader and writer structures. Thus, HAMR has become one of the most promising techniques in the hard disk drive industry for the near future. This technique technology requires the magnetic recording layer of the medium to be heated to the Curie temperature (Tc). Iron platinum (FePt) alloys have been used in recent HAMR media and are the most promising for products. The Tc for this material is around 750K [2] [3]. In order to achieve the temperature during writing, a large optical output is radiated by the NFT continuously and the power is transferred partially to the recording layer within an extremely confined area. With the help of the tiny thermal spot in the magnetic recording layer, the areal density based on HAMR can be significantly improved compared with conventional products. One challenge in HAMR is that the NFT has a very low efficiency and more than half of the energy will be distributed around the NFT in the form of heat. Thus the NFT area and all the surrounding components will be heated. Compared to perpendicular magnetic recording, this additional heat source causes an additional protrusion and affects headmagnetic spacing (HMS). The pointy protrusion and the unusually high temperature would not only affect the writing performance [4] because of the decrease in the transducer efficiency [5], but can also impact reliability due to unintentional head-to-disk contact. So far, many methods have been implemented to measure various HAMR related protrusions. In a large scale, conventional touch-down sensors (Acoustic Emission, thermal proximity sensors, etc.) can be utilized to measure the broad surface protrusion due to the heat distribution. Writer pole tip protrusion (WPTP), induced by the write current, is well known and because of its relatively large protrusion area, can also be detected by touch-down sensors. Reader protrusion can be accurately measured based on the read back signal and Wallace spacing loss equation [6]. Due to the very small area of the NFT local protrusion, conventional touch-down methods cannot accurately detect it, and because the localized protrusion of the NFT is spatially separated from the reader, reader spacing methods cannot detect it accurately, either. Hence, measuring local laser induced protrusion has become one of the most critical issues for HAMR s reliability. Xu et al. [4] have utilized a HAMR spin stand tester and laser Doppler vibrometer (LDV) to investigate the thermal effect on the slider flyheight. This method can give indirect information about the local laser induced protrusion. Li et al. [7] have developed a new atomic force microscopy (AFM)- based technique that allows surface protrusion measurements in the subnanometer level with typical AFM lateral resolution. However this experiment is carried out under non-flying conditions and requires metal coverage over the NFT area to block the laser from hitting the AFM tip. The cooling condition of this method is quite different from that of normal (flying) conditions and the errors from the different conditions are unknown. In this paper, we present a new methodology to measure the local laser induced protrusion that maintains the cooling conditions during flying and does not require metallization. The method uses product HAMR heads, and uses HAMR spin stand testers and AFM scans. The results indicate that the local laser induced protrusion is around ~2-5nm with current NFT, heat sink, and magnetic pole designs. In addition, more

2 A3 2 detailed studies regarding the time dependence, repeatability, heater power effect, laser power effect, and ABS stiffness effect have shown reasonable results consistent with our current understanding of the head disk interface (HDI). II. EXPERIMENTAL SETUP 1) Spin stand Tester Setup The experiment was carried out based on HAMR spin stand testers and atomic force microscopy. Before the test, a large area (20µm 10µm) of the fresh head s air bearing surface (ABS) was scanned and recorded by atomic force microscopy. Once loaded on the spin stand tester, as shown in figure 1, the head would go through three major steps: touch down, laser current sweep, and burnish stress. The touch down step was very important for this experiment because it helped us establish the appropriate heater power at which the head was physically contacting the disk. To be specific, we used a motor jitter sensor as the touch down sensor for this step so that the contacting area would include the HDI sensor position and the write head. Because the motor jitter sensor required relatively larger touch down area compared with other touch down sensors, it gave us appropriate touch down heater power with reasonable success rate. The second step was to find the appropriate laser current for writing performance. Although the actual protrusion amount was unknown during this step, we estimated that the total protrusion amount would not be above 8nm for the selected heads. As a result, we reduced the heater power to give 8nm reader spacing from the touch down point and swept the laser current to find the best SNR condition with squeezed track. The squeeze condition was two sided with around 63nm track pitch. The last step was to burnish the head s protruded portion using the rotating disk. We used the heater power from the 0% LDI, step 1 touch down and the laser current selected from step 2. The stress time was 4 minutes. At the touch down heater power, the HDI sensor and the writer structure were already physically contacting the disk and by adding the laser power, the protrusion induced by the laser would be aggressively contacting the rotating disk and would be quickly burnished off. Fig. 1. The burnish method requires three major steps: touch down, laser current sweep, burnish stress. As shown above, typically the overall protrusion of the head should not exceed 8nm and during laser current sweep step, 8nm is mostly used as a safe clearance. (LDP = laser current) Fig. 2. AFM topography (20µm 10µm) of a fresh head and a burnished head. The AlTiC material at both leading and trailing side is used for plane fit. The height measurement is along the downtrack direction through the center of HDI sensor and writer. After this testing, a second AFM scan for the same area was conducted in order to measure wear, which established the protruded amount during the stress procedure with the selected heater power and laser current. 2) Atomic-Force Microscopy Setup As discussed above, every head required two AFM scans in the experiment. The first AFM scan was for recording the topography of the fresh head as a reference and the second AFM scan was to measure the local worn space which represented the local laser induced protrusion. The atomic force microscopy was used in tapping mode and the scan size was 20µm 10µm with the NFT centered. All the height measurements crossed the center of HDI sensor and writer position along the downtrack direction (figure 2). In order to get a reasonable height measurement, the AFM resolution was set at 512 by 256 pixels. After every scan, we performed an accurate plane fit for the image for the purpose of making the AlTiC material of both the leading trailing sides at the same height level. As shown in figure 2, the plane fit region for the trailing side was 3µm 10µm and for the leading side was 2µm 10µm. Both regions were on top of the AlTiC material. In this case, the measured height was well defined by reference to the fitted AlTiC substrate. Normally, we would test several heads and take the average value of the local worn space to quantify the local laser induced protrusion. 3) Hypothesis The experiment required the head s closest point to touch the rotating disk, which could cause small heating and cooling differences from normal writing operation. The local protrusion amount should be dominated by the temperature around NFT area and we make an assumption that the temperature change was relatively small from the normal writing condition to this experimental condition based on a certain hypothesis: Friction and transducer inefficiency could be regarded as two heat sources that affect the temperature in the NFT area. Normally, the slider s passive flyheight can range from 10nm to 20nm. During normal writing condition, if spacing is well controlled, the clearance between the NFT and the rotating disk could be as small as 1 to 2nm. With relatively rough medium rotating under the head s close point at a linear

3 A3 3 velocity range of m/s [8], there is a high chance of collision between thermal asperities from the medium and the close point of the head. This kind of collision can be regarded as a source of friction and it would generate heat. Although many nano-scale heat transfer methods have been discussed, friction related heat generation is hard to quantify. In this experiment, we assume the amount of heat generation from friction to be relatively small compared with heat generation from the plasmons. This implies that the temperature around the NFT area should not change significantly when the close point contacts the rotating disk (versus being in close proximity). In addition, we also assume the transducer efficiency to be about the same when the clearance is very small. The heat generation from the NFT would therefore remain constant as well. Based on the two assumptions above, the protrusion amount induced by laser would remain the same both in the normal writing condition and in this experimental condition. III. RESULTS AND DISCUSSION Before discussing detailed results, we would like to point out the definition for local laser induced protrusion in this experiment. Figure 3 shows height information based on AFM scans for a head s topography before and after the experiment. The curves indicate the height along the scan direction. Two positions are identified to define local protrusion: the position of the head disk interface (HDI) sensor and the position of the NFT. The NFT area has higher temperature which results in larger protrusion and, as shown in Fig. 3, larger wear. The HDI sensor is farther from the transducer and has smaller wear. The local laser induced protrusion is defined here as the change in the difference of these two heights. By measuring the change in height difference between the NFT area and the HDI sensor position, we can conclude the local protrusion amount to be around 4.1nm for this tested head. A. Local Wear Depth Time Dependence Based on the definition discussed above, the protrusion amount difference between the NFT area and the HDI sensor position is converted into the local wear depth, which represents the local laser induced protrusion. The time constants of the mechanical protrusion could range from a few microseconds to more than 100µs. The shortest thermal time constant can be much smaller than some mechanical time constants. The numbers are strongly dependent on different heat sink designs and different NFT structures. Localized mechanical and thermal time constants are small enough for NFT local area to respond within one sector. However, it doesn t mean the local protruded portion can be worn off within microseconds only. Indeed, the local protruded portion needs much longer time to be burnished off. In figure 4, a relationship between local wear depth and the burnish time is plotted. In this plot, the burnish times used are: 10s, 20s, 60s, 120s, 240s and the resulting local wear depth increases with longer burnish time. The local wear depth increases very fast in the first half minute and it slows down as the burnish procedure progresses. After 120s burnish, the local wear depth Fig. 3. The head s fresh (black) and burnished (red) topographies height information is shown above. By measuring the height difference between HDI sensor and the NFT area (fresh: 1.13nm, burnished: -2.95nm), we can calculate the local wear depth as ~4.1nm. Fig. 4. (a) Height information based on AFM scan of fresh heads and burnished heads is shown above. The local wear depth is getting larger with increasing burnish time; (b) Assuming the local wear depth would saturate at ~4.1nm, we can fit the experimental data to an exponential line and the time constant for the burnish behavior based on the equation would be 1/0.024 = 41.7s.

4 A3 4 is almost saturated as 240s case gives very similar results. The trend line is fitted as an exponential line with the assumption that the upper limit of the local wear depth is 4.1nm. According to this plot, the 4 minute stress should be enough for the locally protruded area to be burnished off completely. This approach could be applied to different head designs to understand wear rates ensure that the wear limit has been reached B. Repeatability Of great interest is also the repeatability of the burnish method. Variation in mechanical protrusion is a significant concern for HDI clearance setting and it is even harder to measure the local protrusion in such a small area around the NFT for each head. However, within one kind of NFT design and heat sink structure, similar thermo-mechanical behavior is expected and the local laser induced protrusion should only vary within a small range. By repeating the burnish method for several heads with same design, we should expect similar burnish topography. Figure 5 shows the AFM results for a set of burnish experiments using four randomly selected heads. All four heads have the same design and we repeat the burnish method with exactly the same procedure and same tester, media, and track position. Based on the AFM scan results, we can measure the local wear depth for each head. The results show that these four heads require different laser currents to maximize each head s own SNR under squeezed conditions. However, the local wear depth for every head is close to each other with an average value of 4.3nm. By definition, the local wear depth is calculated by two contributors: fresh head s topography height information and burnished head s topography height information. We can see that the variation of the height difference between the HDI sensor and the NFT area from the fresh heads topographies is comparable to the burnished heads topographies. If the fresh heads topographies can be more consistent, a better repeatability could be expected. In addition, all the heads burnished topographies show the close point being the NFT or the Fig. 5. A set of burnish method experiment results throughout four head with same design. (a) laser current ~ 57mA, local worn space ~ 4.1nm; (b) laser current ~ 59mA, local worn space ~ 4.4nm; (c) laser current ~ 53.5mA, local worn space ~ 4.4nm; (d) laser current ~ 52.5mA, local worn space ~ 4.2nm. Fig. 6. (a) topo cali = modeling topography calibrated by experimental data; topo pred = topography by model prediction; reader cali = modeling reader position calibrated by experimental data; reader pred = reader position by model prediction. NFT or trailing edge of the magnetic pole is the close point in both case; (b) 3D plots for the surface protrusion. trailing portion of the NFT and the trailing magnetic pole area. This close point finding matches the finite element modeling predictions very well (figure 6). So far, the results of the burnish method give us good repeatability and a reasonable match with the model in terms of close point. C. Heater Power Effect Although the burnish method gives us a reasonable measurement for the local laser induced protrusion, exact matching of the experimental condition with the cooling under normal writing condition as provided by the air bearing is very difficult. At this point of our understanding, measuring the absolute local protrusion amount is very difficult and we can only try to improve our experimental predictions by fine tuning our settings. Among all the factors, heater power plays a very important role for the burnish method. As discussed above, the touch down heater power based on motor jitter sensor is used during the stress step. This heater power setting is much higher than the setting under the normal writing condition and the high bias in heater power would result in very different topographies over a large scale. However, this burnish method focuses on local area laser induced protrusion. Although elevated heater power would affect large area topography change, local area protrusion should not deviate greatly. In this section, a special approach using different values for heater and laser power for burnish experiments are described to verify the effect of the heater power on local laser induced protrusion. Table I lists a summary of the results for the special experiment. Four groups of heads with same design are tested under different conditions in the stress step. Group 1 is burnished under touch down heater power and zero laser power; group 2 is burnished under touch down heater power with 10mW of additional heater power and zero laser power; group 3 is burnished under touch down heater power and full laser power; and group 4 is aggressively burnished under

5 A3 5 TABLE I HEATER POWER EFFECT Laser Power 0% Laser Power 100% Laser Power Heater Power TD Power TD Power TD Power TD Power + 10mW + 10mW Reader Wear Depth (nm) NFT/MP Wear Depth(nm) HDI Sensor Wear Depth (nm) Local Wear Depth (nm) Table I lists the results of a special burnish method experiment. Group 1 uses touch down heater power and 0% laser power for stress step; group 2 uses touch down plus 10mW heater power and 0% laser power; group 3 uses touch down heater power and 100% laser power; group 4 uses touch down plus 10mW heater power and 100% laser power. (TD = Touch Down) touch down heater power with 10mW additional heater power and full laser power. Figure 7 shows typical topographies of the four groups of heads based on the AFM scans. As the stress condition becomes more aggressive from group 1 to group 4, the concave shape generated due to the stress becomes deeper. By comparing group 1 with 2, and group 3 with 4, we can see the 10mW heater power overdrive could generate ~ nm wear depth difference in HDI sensor position as well as the NFT and magnetic pole region. This indicates that 10mW additional heater power would generate large area protrusion covering the HDI sensor and the writer area together and push the whole surface within ~15-20µm length in the downtrack direction toward the rotating disk by around 2nm. This result is comparable to our understanding about the heater power efficiency. Under flying conditions, the most accurate tool to measure spacing change is the Wallace Spacing Loss measurement. Based on the read back signal from the reader, spacing change in absolute nanometers with different heater power can be calculated. For this desuign, the heater efficiency ranges from ~ nm/mW based on reader spacing change, and this value is consistent with the finding from the above experiment. However, the reader is the Fig. 7. Heater power would greatly affect the worn space over a large area from reader to writer. Based on current heater design, the protrusion driven by heater power would be relatively flat and the local laser induced protrusion wouldn t deviate greatly. (DFH: heater power; TD: Touch Down) TABLE II ABS EFFECT AND LASER POWER EFFECT Local Wear Depth (nm) Laser Power 40% Laser Power 100% Laser Power Soft ABS Stiff ABS Soft ABS shows obviously larger local wear depth. 40% laser power case has smaller local wear depth than 100% laser power case as expected. The numbers are not strictly proportional to the laser power due to the possible plastic deformation and the over coat combustion. only available sensor for utilizing the Wallace Spacing Loss measurement and typically the downtrack separation from reader to NFT is around 5µm. We have to rely on other methods, such as static measurement under non-flying conditions [7], to achieve the spacing information at locations other than reader position. Consolidating our understanding of the heater behavior, the burnish method gives us another way to measure the spacing information in a large scale under flying condition. It is worth pointing out another observation in this experiment. Referring to group 1, we expect the local wear depth without laser power to be small. In particular, the value is around 4 Angstroms. However, it gives us information that the close point without laser is still in the NFT area even though the heater is closer to the reader side. There are several explanations: more metal concentrated in the writer side would make the writer expand more; the write current during stress would add to more writer protrusion; and the thicker initial head overcoat in the NFT area also would lead to bigger wear depth in the NFT/magnetic pole region. In summary, the close point is of high interest from the HDI perspective as it determines the minimum spacing between head and disk. It is very hard to control because it can be affected by multiple factors. The burnish method gives us information not only about protrusion localized to the NFT area, but also over a wider scale. D. Air Bearing Surface Effect It is well known that different air bearing surface designs are used in various projects for the purpose of solving different HDI issues. HAMR s biggest concern from the HDI point of view is the pointy protrusion in the NFT area. In order to accommodate this protrusion during writing, higher passive fly height is required. In addition, different designs are used to achieve different air pressures and stiffnesses that can be tailored for particular platforms. In this section, we would like to discuss the impact of air bearing stiffness on local laser induced protrusion. All previous results and discussions are from a relatively low-stiffness ABS. We selected a group of heads with the same NFT and transducer design but integrated into two different ABS designs: the low-stiffness design and a stiffer design. These two groups are labeled as soft ABS and stiff ABS. Both groups were testing using the approach described earlier using the same touch down and laser sweep steps. In the third step, each group was separated into two small sub groups with different laser power settings. One used 100% laser power as usual and the other used 40% laser power.

6 A3 6 Table II lists the results of this particular experiment. If compared within rows, 40% laser power shows smaller local worn space than 100% laser power as expected. The difference between these two cases for both ABS designs is around 1.1nm. If compared within columns, it is obvious that the local wear depth is very different based on different ABS stiffness. The softer ABS shows larger local wear depth. The offset between two different ABS under both laser power conditions is around 1.6nm. This is a relatively large change in the local laser induced protrusion, considering that the only difference is the ABS design. The stiffness would affect the touch down procedure and the resulting touch down power. If well controlled, both groups of heads should be stressed under similar cooling conditions by physically contacting the rotating disk. As we expected, the thermo-mechanical heater response should also be similar. There are several other factors that must be taken into consideration that might be possible explanations for the observed differences between ABS designs. By recalling the assumptions of the burnish method, that the protrusion is an elastic deformation, we might expect the ABS stiffness to affect the magnitude of protrusion. We assume the protruding area is burnished off during the stress step and the remainder would elastically recess to the original position. If this assumption stands, the elastic property could also be utilized to explain the stiffness effect. A stiffer ABS has higher air pressure and normally the highest pressure is located at the protruded NFT area. If the higher air pressure is applied to the local area during the stress, the NFT area would be elastically pushed back. As a result, the same laser power might generate similar local protrusion in the absence of pressure differences, but a stiffer ABS would show smaller local wear depth due to the local push back effect. Similarly, if the pressure is applied to the large area, both the HDI sensor and writer may be pushed back at the same time and the overall amount of wear may be less. In this case, although the heater power is set for the heater touch down condition, the stiffer ABS may not protrude as much as soft ABS might. Due to the large area push back effect, only a small portion of the local laser induced protrusion could be burnished off by the rotating disk. This mechanism may lead to the much smaller local wear depth of stiffer ABS as Table II illustrates. Another possible explanation would be that the softer ABS generated much higher spacing modulation during stressing. Conversely, the stiffer ABS may have maintained a more uniform clearance during stress. The local protruded portion with the soft ABS would have much higher chance to interact with the media surface due to the modulation, which could result in a more seriously burnished surface. This is a third reason why a stiff ABS may show less local protrusion. In summary, all three mechanisms may be in operating and it is difficult to decouple these factors. Eventually, the protrusion measurement affects the clearance setting greatly and will significantly affect the lifetime of the head. IV. CONCLUSION A spin stand and AFM based technique to measure the local laser induced protrusion has been developed. Using this technique, we have performed a series of experiments to study the local surface protrusion induced by laser-induced heating of the near field transducer. Our observations show the time constant of the burnish behavior to be much longer than the pure thermal and mechanical time constants. For the selected heads, a 4 minute stress with the heater and laser on is required for burnish of the locally protruded area to reach steady state. Different head designs may require different time durations to reach steady state. Also, the repeatability of the burnish method is found to be reasonable; the measured local worn space varies only within a very small range. One of the contributors to the variation could be the starting topography difference among the fresh heads. Another important observation is that heater power difference wouldn t deviate the local laser induced protrusion greatly because the heater distributes heat over a large area and both reader and writer including HDI sensor and NFT area would be expanded toward the disk at the same time with a similar distance. In addition, our results show that a stiffer air bearing design generates less protrusion an smaller local wear depth. The cause may be that the elastic deformation under high air pressure may counteract some of the thermally-induced protrusion either at the very local NFT area or at a larger area. Our clearance setting based on the protrusion amount measured by this burnish method is reasonable and a further study based on hard disk drive level evaluation would greatly benefit the improvement of the technique and make it more universal. REFERENCES [1] M. H. Kryder, E. C. Gage, T. W. McDaniel, W. A. Challener, R. E. Rottmayer, G. P. Ju, et al., Heat assisted magnetic recording, Proc. IEEE, vol. 96, no. 11, pp , Nov [2] T. W. McDaniel, Ultimate limits to thermally assisted magnetic recording, J. Phys.: Condens. Matter, vol. 17, pp. R315-R332, [3] O. Mosendz, S. Pisana, J. W. Reiner, B. Stipe, and D. Weller, Ultrahigh coercivity small-grain FePt media for thermally assisted recording, J. Appl. Phys., vol. 111, p. 07B729, [4] B. X. Xu, H. X. Yuan, J. Zhang, J. P. Yang, R. Ji, and T. C. Chong, Thermal effect on slider flight height in heat assisted magnetic recording, J. Appl. Phys., vol. 103, no. 7, pp. 07F F525-3, [5] B. X. Xu, Y. T. Toh, C. W. Chia, J. Li, J. Zhang, K. Ye, and C. An, Relationship between near field optical transducer laser absorption and its efficiency, IEEE Trans. Magn., vol. 48, no. 5, pp , May [6] R. L. Wallace, The reproduction of magnetically recorded signals, BellSyst. Tech. J., vol. 30, pp , Jul [7] D. Li, M. Staffaroni, E. Schrech and B. Stipe, A New AFM-Based Technique to Detect the NFT Protrusion on HAMR Head, IEEE Trans. Magn., vol. 49, no. 7, pp , Jul [8] E. Schreck, D. Li, S. V. Canchi, L. Huang, G. P. Singh, B. Marchon et al., Thermal Aspects and Static/Dynamic Protrusion Behaviors in Heat- Assisted Magnetic Recording, IEEE Tran. Magn., vol 50, no. 3, Mar