Advances in Information Storage Systems

The following sections are included:

• Preface

• ADVANCES IN INFORMATION STORAGE SYSTEMS

Flying heights (the mean separation between magnetic head slider and magnetic disk) are in the range of 50–100 nm, comparable to the surface roughness of the mating surfaces. The need for increasingly higher recording densities requires that surfaces be as smooth as possible and the flying heights be as low as possible. Number of drive designs are under development in which slider is supposed to be in contact or near contact the disk surface. Ultralow flying heights can result in head-disk contacts during operation, consequently deteriorating head-disk interface tribological performance and reliability. Friction and wear issues are addressed either by reducing the slider mass, size and loading or by applying a relatively thin layer of low viscosity lubricants in order to develop a liquid lubricated bearing. Tribology and mechanics of head-disk interface are critical technologies, and the fundamental understanding of which is crucial to the future of the fast-growing magnetic recording industry. Micro/nanotribological tools and techniques are commonly used today to develop fundamental understanding of the tribological processes at the head-disk interface. (A new book titled Handbook of Micro/Nanotribology edited by B. Bhushan will be published by CRC Press in early 1995.) The papers in Part I address important issues relevant to tribology and mechanics of magnetic rigid disk drives…

Laser Doppler anemometry measurements of the tangential velocity were obtained for the flow between shrouded corotating centrally clamped disks with through-hub ventilation. The radial profiles along the centerline between the disks show the existence of three distinct radial regions to the flow, the jet entry region, the inner region, and the outer region. The axial profiles at varying radial locations show the axial flow structure to be divided into two regions, the core region and the disk boundary layers. In addition, spectral measurements of the fluctuating velocities were taken to quantify the behavior of the vortices that characterize the outer region.

The orientation of the magnetic recording head suspension gimbal, with respect to the “roll” axis of the head in a direct access storage device (DASD), can be controlled in the manufacturing process by intentionally inducing compensating bias. This can be done by skewing the mandrel, which contacts the suspension during the final forming step in the manufacturing process, in the plane of the suspension. The goal of the present study is to determine how inducing such a static roll bias in the suspension will affect the structural resonances of the head-gimbal assembly (HGA). Induced suspension roll bias is determined for each HGA studied, using the included angle between incident and reflected laser beams directed at the end of the suspension adjacent to the gimbal. Mobility frequency response functions (FRFs) are determined from laser Doppler vibrometry (LDV) measurements. The technique of modal analysis is employed to facilitate identification of modes and nodal locations of the vibration modes of the suspension using measured FRFs. Resonance frequency dependence on small changes in mandrel skew, and on the induced suspension roll bias, is observed. The effect of mandrel skew angle on suspension modal frequencies is found to be symmetric about the zero skew angle axis, and to vary from mode to mode. Modal frequencies increase, as mandrel skew is increased, for modes comprised of mostly out-of-plane motion, such as torsion or bending. Conversely, the sway mode frequency is observed to decrease, as mandrel skew is increased. These behaviors are explained in terms of how the crease, or bend left by the mandrel in the manufacturing process, affects the bending moments of inertia of the suspension.

Experimental investigations of the dynamic loading process of a 2.5″ hard disk drive with a ramp loading system are presented. The dual beam Polytec LDV is successfully applied to the measurement of slider-disk relative motion during single load events. An AE system is used to confirm the slider-disk contacts. The effects of different head-load speeds and of different initial pitch and roll angles are examined. It is observed that the following three parameters: (1) Initial loading velocity, which is determined by the actuator swing velocity as well as the disk runout, (2) Initial pitch, and (3) Initial roll, strongly affect the occurrence of the slider-disk contact. An apparent inconsistency between the LDV and AE measurements disappeared after all sliders except the LDV measured one were removed from the drive. Critical ramp speeds below which slider-disk contacts do not occur were established for two different sets of initial pitch and roll for the tested drive.

Some advanced magnetic heads may require sputtering a thin layer of film on the air-bearing surface (ABS) for the protection of the magnetic head and the improvement of the head/disk interface tribological performance. However, the conventional interferometric flying height tester does not take account of the film in its flying height determination. This will result in erroneous measurement data if the layer of film is transparent or translucent. This investigation proposes a methodology to address and circumvent this problem. In particular, a glass-air-film-substrate model is compared with a glass-air-substrate model where the film is not considered. The flying height measurement error due to the neglect of film is evaluated as a function of the film thickness for carbon, diamond and quartz films coated on top of Al₂o₃-TiC and MnZn ferrite substrates. It is found that depending on the optical properties of the substrate and the film, the error can be significant at certain film thickness. One way to resolve this problem is to use the glass-air-film-substrate model. Another approach for the existing flying height testers that already employ the glass-air-substrate model is to adopt a pseudo homogeneous material consisting of a pair of equivalent index of refraction and coefficient of extinction in the conventional interferometric flying height tester. Some unrealistic optical property values, however, may need be used for this equivalent material.

The aim of this numerical investigation is the systematic determination of the effects of surface texture patterns on the performance of ultrathin, finite-width gas bearings. The two dimensional Reynolds Equation of lubrication with first-order molecular slip effects is solved by the use of a second-order accurate, factored implicit, finite difference technique. Sixteen texture patterns are analyzed and compared in order to assess their individual and combined effects. Then, the consequences of accounting for molecular slip in the analysis are evaluated for the various texture configurations in order to understand the load behavior of a more realistic gas bearing. The extensive parametric studies conducted include the effects of side leakage, texture amplitude, and bearing number on the bearing's performance indicators including the load, maximum and minimum pressures, friction drag force, mass flow rates, and location of the center of pressure. The results indicate that a combination of transverse texture on the stationary surface and longitudinal texture on either surface gives rise to lower loads than those generated with transverse stationary texture alone.

Both static behaviors and dynamic responses of slider when flying over an embossed disk with discrete tracks and servo marks are examined in this paper. It is shown that the averaged Reynolds equation may be used to design a reliable slider-disk interface for grooved discrete track disk. A mechanical resonance problem associated with slider traveling over a periodic pattern of embossed servo segments is analyzed and resolved. It is also indicated that there is no critical change in flying height fluctuation for an embossed discrete track disk compared with the conventional flat surface disk.

A technique for studying the limits of low flying height is developed by using simultaneous measurements of friction force and acoustic emission (AE) at gradually reduced speeds/flying heights. Three flying zones are identified with boundaries defined by the acoustic emission and friction avalanches. In addition, a technique is derived for characterizing the AE and friction avalanche growth to help better understand the phenomenon. Experimental results are encouraging for near-contact recording. It is shown that safe flying is possible not only at the commonly used zone above the AE avalanche, but also at the vast part of the zone between the AE and friction avalanches. This opens the way for significantly lowering flying heights and suggests an alternative to the glide height criterion for disk surface quality inspection. Smoother disks show improved durability at lower flying height but also increased stiction. The trade-off between smoothness and stiction on lubricated disks is illustrated.

The phenomena taking place at the slider-disk interface are a topic of long term fundamental interest as well as practical significance. A novel accelerated wear tester for studying the slider in close proximity to the disk at high velocity is described. This incorporates a pneumatic loading cylinder on the suspension to increase the load while the slider is flying. Slider-disk interaction is detected by accelerometers mounted on the suspension arm. As the flying height is lowered, the resonant vibration of the arm is recorded to characterize the interface. Preliminary results showing the effect of disk velocity and roughness on the vibration-flying height curve are presented along with qualitative interpretation.

The effects of thickness and porosity-inducing additives on the lubricant areal density are investigated. Topical lubricant applied to the media provides a long lifetime for the head disk interface. It is essential to control the lubricant areal density in the manufacturing process, which requires understanding the relation between the media thickness and porosity. In this study, both the porosity and media thickness were adjusted in a regular manner, and the lubricant areal density was measured. A pore depth distribution model is developed to assist with interpretation of the results. The results, in conjunction with the pore distribution model, indicate that the majority of the lubricant resides in pores within the media. These pores are unevenly distributed through the depth of the media, with porosity increasing near the surface. The effective shape of the pores is hemispherical. Sliding wear tests on the same media demonstrate that lubricant held within the pores is effective for lubrication of the asperities.

Friction Force Microscopy (FFM) was used to study the microscopic friction between a sharp diamond tip and single-crystal silicon, C⁺-implanted silicon and thin-film magnetic rigid disks with a few nanometer-thick lubricant layer. Thin-film magnetic disks with silicon dioxide and diamond like carbon overcoats were lubricated with perfluoropolyether (PFPE) lubricant layers with and without functional end groups. In the low normal force region (0 to 25 μN) for single-crystal silicon, the friction force increases linearly with an increase in the normal force. However, the coefficient of friction is not stable once the normal force exceeds a value of about 25 μN. The normal force at which the coefficient of friction starts to increase is defined as the “critical force”. The increase in the coefficient of friction above the critical force occurs as a result of the surface damage of sample during the microfriction measurements. The C⁺-implanted silicon exhibits a lower value of coefficient of friction than the unimplanted sample and implanted silicon does not exhibit the evidence of surface damage in the entire range of normal force. The magnetic disk sample lubricated with PFPE lubricant exhibits a lower coefficient of friction value than the unlubricated one. An increase in the lubricant (PFPE with hydroxyl functional end groups) thickness from 1.3 nm to 11.2 nm results in an increase of the coefficient of friction value from 0.10 to 0.15. The critical force for the surface damage/scratching or depletion of the lubricant layer increases with an increase in the lubricant thickness in the range of 1 nm to 11 nm. The PFPE lubricant with hydroxyl functional end groups exhibits higher critical force as compared to the PFPE lubricant without functional end groups. Limited tests on the lubricated disks with carbon overcoat show that the coefficient of friction value is same and the critical force is lower as compared to the silicon dioxide overcoated magnetic disks.

The Point Contact Microscope is used to conduct indentation hardness tests on thin film magnetic disks. Determinations of the indentation depth and area are critical to the measurement of hardness when the indentation depth is on the order of the surface roughness. An algorithm is developed to first locate the overlap part between the indented and original surfaces and then subtract the two to more precisely determine the indentation depth and area. This technique is used to determine the hardness of the nominally 20–40 nm thick carbon/zirconia protective overcoats of thin film disks. It is found that the zirconia overcoat is somewhat softer than the carbon overcoat, and the carbon overcoat is about 4 to 5 times harder than the magnetic layer.

Variations in surface roughness, average material removal rate, and surface texture were measured for nickel-phosphorus plated aluminum substrates polished in the double-sided polishing process. A kinematics process model was developed to explain the genesis of polished surface texture. For realistic polishing conditions, applied load and polishing pad condition had little effect on polished surface roughness. Material removal rate varied with applied load and slurry application. The most notable result was that in the apparently symmetric process, the average material removal rate was different on the two sides of the substrate. Further, material removal rate varied over sequential polishing tests to the extent that the substrate side of maximum material removal changed. The polished surface textures predicted by the process model correspond to measured textures for substrates which were constrained in the carriers during polishing. Substrates which were free to move in the carriers during polishing exhibited large variations in surface texture between substrates and between the two sides of individual substrates.

Glass and glass-ceramic substrates are beginning to be used in the construction of magnetic thin-film rigid disks for their rigidity, dent resistance, and smoothness over commonly used Ni-P coated aluminum-magnesium (Al-Mg) substrates. Elastic-plastic deformation behavior, hardness, Young's modulus of elasticity, and scratch resistance of Ni-P coated Al-Mg, glass and glass-ceramic substrates as well as finished disks made of these three substrates were measured using a depth-sensing nanoindenter. The glass-ceramic substrate and the corresponding disk exhibited the highest hardnesses and elastic moduli as compared to other substrates and disks examined. Hardness and elastic modulus of the glass-ceramic disk at small indentation depths of about 25 nm is comparable to those of single-crystal silicon. Based on the friction data, width and depth of scratches, the amount of debris generated, and scratch morphology in scratch testing with a diamond tip, the glass substrate and disk exhibited a lower coefficient of friction and a higher resistance to scratch than the glass-ceramic substrate and disk.

The flying characteristics of the head/disk interface in a liquid lubricated “wet” interface is determined using intensity-based interferometry and magnetic signal analysis. The non-Newtonian characteristics of the liquid interface is investigated for power-law and Winer-Bair fluids, and the pressure distribution and frictional force as a function of velocity are calculated. The experimental data for the variation of spacing as a function of velocity are compared with numerical calculations based on non-Newtonian fluid models. In addition, the frictional drag of the head/disk interface is determined numerically for different fluid models.

Since the early 1990's, track density of magnetic recording rigid disk drives has gone up approximately 30% every year. In fact, track density has gone up much faster than bit density in recent products and together, they enable the industry to maintain the steady annual growth of areal recording density on the order of 60%. Although this by itself is not enough to increase the profit margin, at least it keeps the industry alive and competitive and most of us employed. Currently, the industry is shipping products with track density in excess of 5,000 tracks-per-inch (TPI) which is expected to increase to 12,000 by the year 1998 and 25,000 by 2001. At a track density of 25,000, the track-to-track spacing is 1 micron and the allowable-off-track error (or track misregistration, TMR) is less than 120 nm. Without a doubt, this has posed and will continue to pose extreme challenges to electromechanical/servo engineers who are responsible for designing the tracking sound systems…

The trend towards smaller form factors in the magnetic disk storage industry is analyzed and discussed. Miniaturization is the result of small systems' storage requirements and advances in magnetic recording technology. A scaling methodology for determining disk drive sizes is described and used to project potential future form factors. Some limiting characteristics of disk drives in these form factors are examined.

The spindle is one of the major components of a disk drive. Its performance is critical to the function of the drive, affecting among other things, track density, data access rates, power dissipation, and acoustics. This work focuses on the dynamic performance of the spindle and its potential for affecting track misregistration (TMR) through self-induced and externally induced vibrations. A finite element model of the spindle is developed to demonstrate its natural modes of vibration, and further, to predict its dynamic performance under external vibrations and the effect this has on TMR. It is also shown experimentally that the structural modes have a significant effect on the nonrepeatable runout of a ball bearing spindle. An experimental method for measuring transfer functions of the spindle is demonstrated and is used to obtain data to verify the finite element model. Measurements of the spindle vibration show excellent agreement with the analysis prediction. The paper employs a 3.5″ ball bearing spindle as the working example of the methodology, but also includes data measured on a 3.5″ air bearing spindle for purposes of comparison of the two bearing technologies.

Precoded servo patterns are effective for narrowing track pitch of magnetic disk drives. Current magnetic disk drives use a servo track writer (STW) to record the positioning signal on assembled disks. Observation of a magnetic servo pattern written by STW shows that random deviations exist. Such deviations, in the order of several μm, are primarily due to mechanical vibrations and magnetization pattern edge shape irregularities. While a pit type servo pattern fabricated using photo-lithography is effective for reducing such mechanical deviations, it produces an uncertain magnetization transition width shape, resulting in pattern deviation. We propose a pre-coded with magnetic burst pattern (PM pattern), which has a straight sharp edge. We analyze the positioning signal through computer simulation and evaluate the track following performance when the precoded disk is assembled into a disk drive, taking into account the mechanical and electrical conditions of the drive system. We find that the servo pattern recorded by STW tends to be random. However, a precoded pattern tends to be continuous. With the precoded PM pattern, it is possible to achieve a submicron track pitch on magnetic disk.

An Adaptive Finite Impulse Response (AFIR) method is presented for use in repetitive tracking systems. The basic algorithm employs a learning and correction scheme, that continually attempts to reduce the system tracking error using past error measurements. Because of the repetitive nature of the system, a periodic correction signal can be added to the basic control to aid in tracking the anticipated input signal. The feed-forward correction signal is a discrete sequence, but the feedback controller may be either continuous or discrete. The correction sequence is generated using an impulse response estimate of the system and the tracking error signal. Slow time-varying changes in the input signal are detected through the error signal, and corrected through the learning correction sequence. This method also allows for adaptation to changes in the dynamics of controller or plant by regularly computing an estimate of the impulse response of the system. Summary results for stability and learning rate are supplied. Algorithm performance is also evaluated using a computer simulated tracking system.

Reducing structural vibration is essential for achieving high-accuracy, high-speed head positioning for small magnetic disk drives. In conventional disk drives, reaction forces from the head-positioning actuator cause structural excitation in the base. Rotating the carriage by a conventional single-drive actuator also causes structural vibration in the carriage mechanism. Taking such phenomena into consideration, we developed a nonreacting, twin-drive actuator, which excites the base and carriage mechanism much less than conventional ones. Using the nonreacting, twin-drive actuator, residual vibration is one fourth of that with the conventional actuator.

Parallel data transfer is a promising way to increase data transfer rates. In order to achieve parallel data transfer that is stable, all of the beams should be positioned on their respective tracks with equal precision. This paper presents a new multibeam positioning system and a new dove prism actuator for precise and equal positioning. The positioning system consists of a conventional two-stage tracking control system and a beam array rotation control system with an interaction compensator, which equalizes the positioning errors. The dove prism actuator for rotation control has a high mechanical resonance of 4 kHz and good rotary balance. Experimental results show the effectiveness of this positioning system and dove prism actuator for precise and equal positioning.

Mechanical positioning time, or seek time, is a key factor in determining the performance of data storage devices. Average seek time is often used as the only measure of data storage system performance, however, methods to calculate average seek time vary. This paper first presents some common methods to measure average seek distance and develops a new method based on exponential seek distributions. Secondly, the relationship between seek distance and time is explored. Third, the relationship between seek time and rotational latency is documented. Finally, a method is outlined to model mechanical access time using only logical block addresses and turn-around times.

For the past 25 years, improvements in CPU performance have overshadowed improvements in the access time performance of disk drives. CPU performance has been slanted towards greater instruction execution rates, measured in millions of instructions per second (MIPS). However, the slant for performance of disk storage has been towards capacity and corresponding increased storage densities. The IBM PC, introduced in 1982, processed only a fraction of a MIP. Follow-on CPUs, such as the 80486 and 80586, sported 5-10 MIPS by 1992. Single user PCs and workstations, with one CPU and one disk drive, became the dominant application, as implied by their production volumes. However, disk drives did not enjoy a corresponding improvement in access time performance, although the potential still exists. The time to access a disk drive improves (decreases) in two ways: by altering the mechanical properties of the drive or by adding cache to the drive. This paper explores the improvement to access time performance of disk drives using cache, prefetch, faster rotation rates, and faster seek acceleration.

A simple, low cost, yet effective device has been developed for immobilizing the head-arm assembly in a disk drive or similar mechanism during power-off conditions. The latching scheme also provides a consistent means of releasing the head-arm assembly from the immobilized position upon power up of the disk drive. The latch uses no electrical power in either immobilized or released state. This design is immune to extreme torque and linear shock forces applied to the disk drive case. The latch system can use the energy stored in the spinning disks to drive the head-arm assembly toward a safe position while simultaneously arming the latch mechanism to secure the head-arm assembly in the safe position upon arrival. A low energy five msec pulse of current drives the latch from one state to the other. Solenoids as presently used in latch mechanisms are bulky, expensive, have variable force characteristics, and often generate contaminants. The latch described in this paper is expected to replace such solenoids. It may also replace small magnet latches, which have limited latch force and apply unwanted torque to a proximate head positioning arm.

This part contains papers that discuss the mechanics of flexible media in information storage and processing systems. Applications here include floppy disk drives, linear- and rotary-format tape drives, copiers, facsimile machines, and printers. The distinction between rigid and flexible media systems is somewhat artificial, but is traditionally made to account for the fact that in the latter case, the media can be subjected to substantial structural deformation during its use. As a result, these systems give rise to challenging problems (and opportunities) in the areas of lubrication, contact mechanics, solid mechanics, and dynamics, to name a few disciplines…

A transversely slotted magnetic recording head for high density tape drives is characterized theoretically for geometry and experimentally for wear and durability under different environmental conditions, tape substrate thicknesses, recording head substrate materials, and wear resistant overcoat materials. The triple bump recording head structure, which allows read-after-write in both directions of tape motion, has transverse slots placed in the head to produce contact between the tape and the read/write gap locations. A through slot “video” style contour design is also presented and analyzed.

The effect of head/tape design parameters on the head/tape spacing and pressure of an infinitely wide tape is simulated numerically by coupling the Reynolds equation with the tape deflection equation. Contact between head and tape is considered using an experimentally determined relationship between contact pressure and spacing. The head/tape spacing is found to increase if tape tension or bending stiffness is decreased, or tape velocity or head radius is increased. The contact pressure is found to decrease if tape velocity, or head radius is increased, or tape tension or bending stiffness is decreased.

Most head-disk interface of flexible disk drives function under a mixed lubrication regime. In this regime, both solid contact pressure and air film pressure act on the bearing surfaces of head slider and disk. This paper describes a unique numerical method capable of simulating the mixed lubrication mechanism. In order to obtain both fluid and contact load, two equations have been adopted; the linearized Boltzmann equation combined with air flow factor by surface roughness effect and the elastic contact equation of rough surfaces presented by Greenwood and Williamson. These equations have been solved simultaneously with the rigid body equations of slider motion, and then spacing, air pressure and contact pressure distributions have been calculated as a result. And the relation between friction force and relative speed has also been obtained assuming that friction force is mainly affected by contact force. Comparing with the experimental coefficient of friction on Stribeck curve, it is concluded that the numerical model of mixed lubrication for flexible disk drives is validated by the experimental results.

A laminate disk which is freely rotated at high speed is analyzed in this paper. The governing differential equilibrium equations are developed and solved using a finite difference approach. The analysis assumes that displacements are small eliminating the nonlinear terms in the tension-displacement equations. The axial equilibrium equation includes a bending stiffness reduction factor which accounts for the arbitrary position of the coordinate system used to calculate the elastic parameters. Displacement results are presented in terms of normalized quantities which can be used to approximate the displacements resulting from the laminate structure of the disk.

In order to stabilize a self-induced vibration of a magnetic recording flexible disk drive, the effects of damping on the instability caused by dynamic frictional forces are analyzed using a disk-heads coupling model. From parameter studies of the eigenvalue analysis, the following results are obtained: first, the internal and external dampings acting uniformly on the disk make all real parts of the eigenvalues shift towards a negative direction. Second, the transverse and pitch dampings of the head-suspension system can stabilize unstable natural modes near the transverse and pitch resonances of the head-suspension system respectively. When the transverse and pitch resonances are close to each other, there are two types of unstable natural modes near the resonance, whose instability can be suppressed mainly by either transverse damping or by pitch damping. Third, the instabilities caused by dynamic frictional forces are hardly affected by the transverse and pitch contact dampings.

A very thin and smooth magnetic coating of metal particulate tape is made possible by a simultaneous double coating method. This MP tape consists of a magnetic upper layer and a nonmagnetic underlayer composed of very fine Tio₂ powder. We observed the anticipated advantages of decreasing the thickness of the magnetic layer; the thinner the magnetic layer, the smoother the surface and the higher the electromagnetic characteristics. Thinning the layer to below 0.3 μm increased the reproduced output to a level as high as that of Hi-8 ME media. This result is impossible to obtain with present MP media. We also researched how to make these smooth surface media stable in a VCR. The nonmagnetic layer under the magnetic layer played an important role in the reduction of the friction force by acting as a lubricant to the surface. This method is effective for compatibility with highly reproduced output and durability.

This paper investigates the handling of ink films for thermal transfer facsimiles. The film tension is shown to be the most important factor to avoid wrinkles in the ink film, and can vary within a design area defined by the minimum film handling tension and the film handling limit. The precision of shaft assembly, such as the guide rollers, and the pressure applied to the thermal head are experimentally examined. Shear deformation tests of a membrane with a tensile stress suggest limits to the formation of wrinkles. The minimum tension for stable film handling can be estimated from the buckling limit obtained from shear tests and numerical analysis.

The following sections are included:

• Author Index

• Subject Index

• INSTRUCTIONS FOR AUTHORS