Patent Application
20120299132
The invention relates to non-volatile magnetic storage devices and more particularly to a magnetic disk drive including a tunneling magnetoresistance (TMR) read sensor with low-contact-resistance interfaces.
One of many extensively used non-volatile magnetic storage devices is a magnetic disk drive that includes a rotatable magnetic disk and an assembly of write and read heads. The assembly of write and read heads is supported by a slider that is mounted on a suspension arm. The suspension arm is supported by an actuator that can swing the suspension arm to place the slider with its air bearing surface (ABS) over the surface of the magnetic disk.
When the magnetic disk rotates, an air flow generated by the rotation of the magnetic disk causes the slider to fly on a cushion of air at a very low elevation (fly height) over the magnetic disk. When the slider rides on the air, the actuator moves the suspension arm to position the assembly of write and read heads over selected data tracks on the magnetic disk. The write and read heads write and read data in the selected data tracks, respectively. Processing circuitry connected to the write and read heads then operates according to a computer program to implement writing and reading functions, respectively.
The write head includes a magnetic write pole and a magnetic return pole that are magnetically connected with each other at a region away from the ABS, and an electrically conductive write coil surrounding the write head. In a writing process, the electrically conductive write coil induces magnetic fluxes in the write head. The magnetic fluxes form a magnetic write field emitting from the magnetic write pole to the magnetic disk in a direction perpendicular to the surface of the magnetic disk. The magnetic write field writes data in the selected data tracks, and then returns to the magnetic return pole so that it will not erase previously written data in adjacent data tracks.
The read head includes a read sensor that is electrically connected with lower and upper ferromagnetic shields, but is electrically separated by insulation layers from longitudinal bias layers in two side regions. In a reading process, the read head passes over data in a selected data track, and magnetic fields emitting from the data modulate the resistance of the read sensor. A change in the resistance of the read sensor is detected by a sense current passing through the read sensor, and is then converted into a voltage change that generates a read signal. The resulting read signal is used to decode data in the selected data track.
A tunneling magnetoresistance (TMR) read sensor is typically used in the read head. The TMR read sensor includes a nonmagnetic insulating barrier layer sandwiched between a ferromagnetic reference layer and a ferromagnetic sense layer. The thickness of the barrier layer is chosen to be less than the mean free path of conduction electrons passing through the TMR read sensor. The magnetization of the reference layer is pinned in a direction perpendicular to the ABS, while the magnetization of the sense layer is oriented in a direction parallel to the ABS. When passing the sense current through the TMR read sensor, the conduction electrons are scattered at lower and upper interfaces of the barrier layer. When receiving a magnetic field emitting from data in the selected data track, the magnetization of the reference layer remains pinned while that of the sense layer rotates. Scattering decreases as the magnetization of the sense layer rotates towards that of the reference layer, but increases as the magnetization of the sense layer rotates away from that of the reference layer. This scattering variation induces a tunneling effect characterized by a change in the resistance of the TMR read sensor in proportion to the magnitude of the magnetic field and cos θ, where θ is an angle between the magnetizations of the reference and sense layers. The change in the resistance of the TMR read sensor is then detected by the sense current and converted into a voltage change that is processed as a read signal.
The TMR read sensor has been progressively miniaturized for magnetic recording at higher linear and track densities. Its thickness, which defines a read gap, is reduced by utilizing thinner reference, barrier, sense or other layers, in order to increase linear densities. Its width, which defines a track width, is reduced by patterning with an advanced photolithographic tool, in order to increase track densities. In this miniaturization progress of the TMR read sensor, its resistance will progressively increase so that electronic noises may becomes significant and electrostatic discharges may occur. It is thus crucial to control the resistance to below a safety margin to ensure the feasibility of the TMR read sensor miniaturized for performing magnetic recording at higher linear and track densities.
The invention provides a TMR read sensor with low-contact-resistance metal/metal, metal/oxide and oxide/metal interfaces. The low-contact-resistance metal/metal interfaces in a reference or sense layer structure are in-situ formed in a high-vacuum deposition module of a sputtering system, without exposures to low vacuum in a transfer module and damages caused by a plasma treatment conducted in an etching module. The low-contact-resistance metal/oxide interface is formed by utilizing a thin Co—Fe—B reference layer and a thick Co—Fe reference layer to reduce boron diffusion and segregation caused by annealing. The low-contact-resistance oxide/metal interface is formed by replacing a Co—Fe—B sense layer with a Co-rich Co—Fe sense layer to eliminate boron diffusion and segregation caused by annealing. With the low-contact-resistance metal/metal, metal/oxide and oxide/metal interfaces, the TMR read sensor exhibits a junction resistance-area product of below 0.6 Ω-μm2, while maintaining a low ferromagnetic coupling field and a high TMR coefficient.
These and other features and advantages of the invention will be apparent upon reading of the following detailed description of embodiments taken in conjunction with the figures in which like reference numerals indicate like elements throughout.
For a fuller understanding of the nature and advantages of the invention, as well as the preferred mode of use, reference should be made to the following detailed description read in conjunction with the accompanying drawings which are not to scale.
Table 1 is a table listing HF, RJAJ, ΔRT/RJ and FoM for TMR read sensors without a plasma treatment and with Co—Fe—B and Co—Fe reference layers of various thicknesses;
Table 2 is a table listing HF, RJAJ, ΔRT/RJ and FoM for TMR read sensors without a plasma treatment and with Co—Fe and Co—Fe—B sense layers of various thicknesses; and
Table 3 is a table listing various methods of attaining low-contact-resistance metal/metal, metal/oxide and oxide/metal interfaces in accordance with the invention, and their evaluation based on ΔδMgOxN, ΔHFN and ΔFoMN.
The following description is of the best embodiments presently contemplated for carrying out the invention. This description is made for the purpose of illustrating general principles of the invention and is not meant to limit inventive concepts claimed herein.
Referring now to
At least one slider 113 is positioned near the magnetic disk 112, each slider 113 supporting one assembly of write and read heads 121. As the magnetic disk 112 rotates, the slider 113 moves radially in and out over the disk surface 122 so that the assembly of write and read heads 121 may access different data tracks on the magnetic disk 112. Each slider 113 is mounted on a suspension arm 115 that is supported by an actuator 119. The suspension arm 115 provides a slight spring force which biases the slider 113 against the disk surface 122. Each actuator 119 is attached to an actuator means 127 that may be a voice coil motor (VCM). The VCM comprises a coil movable within a fixed magnetic field, the direction and speed of the coil movements being controlled by the motor current signals supplied by a control unit 129.
During operation of the magnetic disk drive 100, the rotation of the magnetic disk 112 generates an air bearing between the slider 113 and the disk surface 122 which exerts an upward force or lift on the slider 113. The air bearing thus counter-balances the slight spring force of the suspension arm 115 and supports the slider 113 off and slightly above the disk surface 122 by a small, substantially constant spacing during sensor operation.
The various components of the magnetic disk drive 100 are controlled in operation by control signals generated by the control unit 129, such as access control and internal clock signals. Typically, the control unit 129 comprises logic control circuits, storage means and a microprocessor. The control unit 129 generates control signals to control various system operations such as drive motor control signals on line 123 and head position and seek control signals on line 128. The control signals on line 128 provide the desired current profiles to optimally move and position the slider 113 to the desired data track on the magnetic disk 112. Write and read signals are communicated to and from the assembly of write and read heads 121 by way of a recording channel 125.
The TMR read sensor 201 includes an electrically insulating MgOX barrier layer 210 sandwiched between lower and upper sensor stacks 212, 214. The MgOX barrier layer 210 is formed of a 0.1 nm thick oxygen-doped Mg (Mg—O) film, a 0.6 nm thick oxide (MgO) film, and another 0.1 nm thick oxygen-doped Mg (Mg—O) film.
The lower sensor stack 212 comprises a buffer layer 216 formed of a 2 nm thick nonmagnetic Ta film, a seed layer 218 formed of a 2 nm thick nonmagnetic Ru film, a pinning layer 220 formed of a 6 nm thick antiferromagnetic 23.2Ir-76.8Mn (composition in atomic percent) film, a keeper layer structure 222, an antiparallel-coupling layer 226 formed of a 0.4 nm thick nonmagnetic Ru film, and a reference layer structure 224. The keeper layer structure 222 comprises a first keeper layer 223 formed of a 1.8 nm thick ferromagnetic 72.5Co-27.5Fe film and a second keeper layer 225 formed of a 0.4 nm thick ferromagnetic Co film. The thicknesses of the first keeper layer 223 and the second keeper layer 225 are selected to attain a total saturation moment of 0.32 memu/cm2 (corresponding to that of a 4.6 nm thick ferromagnetic 88Ni-12Fe film sandwiched between two Cu films). The reference layer structure 224 comprises a first reference layer 252 formed of a 0.4 nm thick ferromagnetic Co film, a second reference layer 254 formed of a 0.4 nm thick ferromagnetic 75.5Co-24.5Hf film, a third reference layer 256 formed of a 1.2 nm thick ferromagnetic 65.5Co-19.9Fe-14.6B film, and a fourth reference layer 258 formed of a 0.4 nm thick ferromagnetic 46.8Co-53.2 Fe film. The thicknesses of the first reference layer 252, the second reference layer 254, the third reference layer 256 and the fourth reference layer 258 are selected to attain a total saturation moment of 0.30 memu/cm2 (corresponding to that of a 4.3 nm thick ferromagnetic 88Ni-12Fe film sandwiched between two Cu films).
The upper sensor stack 214 comprises a sense layer structure 228 and a cap layer structure 230. The sense layer structure 228 comprises a first sense layer 262 formed of a 0.4 nm thick ferromagnetic 46.8Co-53.2Fe film, a second sense layer 264 formed of a 1.6 nm thick ferromagnetic 79.3Co-4.0Fe-16.7B film, a third sense layer 266 formed of a 1.2 nm thick ferromagnetic 75.5Co-24.5Hf film, and a fourth sense layer 268 formed of a 5.6 nm thick ferromagnetic 96Ni-4Fe film. The thicknesses of the first sense layer 262, the second sense layer 264, the third sense layer 266 and the fourth sense layer 268 are selected to attain a total saturation moment of 0.56 memu/cm2 (corresponding to that of a 8 nm thick ferromagnetic 88Ni-12Fe film sandwiched between two Cu films). The cap layer structure 230 comprises a first cap layer formed of a 1 nm thick nonmagnetic Ru film, a second cap layer formed of a 1 nm thick nonmagnetic Ta film, and a third cap layer formed of a 4 nm thick nonmagnetic Ru film.
A typical insulation layer 202 in each side region is formed of a 2 nm thick nonmagnetic, amorphous Al2O3 film. A typical longitudinal bias stack 204 in each side region comprises a seed layer 232 formed of a 4 nm thick nonmagnetic Cr film, a longitudinal bias layer 234 formed of a 25.6 nm thick hard-magnetic 82Co-18Pt film, and a cap layer 236 formed of a 10 nm thick nonmagnetic Cr film. The thickness of the Co—Pt longitudinal layer 234 is selected to attain a remnant moment of 2.24 memu/cm2 (corresponding to that of a 32 nm thick ferromagnetic 88Ni-12Fe film sandwiched between two Cu films).
In the fabrication process of the read head 200, the TMR read sensor 201 is deposited on a wafer with a lower shield 206 formed of a 1 μm thick ferromagnetic 80Ni-20Fe film in various deposition modules of a sputtering system, and is annealed in a magnetic field of 50,000 Oe for 5 hours at 280° C. in a high-vacuum oven. In the deposition process of the TMR read sensor 201, the wafer is frequently transferred through a transfer module not only to the various deposition modules, but also to an etching module to apply a plasma treatment to a metal/metal interface in the reference layer structure 224. The plasma treatment mildly etches the metal/metal interface in the reference layer structure 224 for 72 seconds at a target power of 20 W, thereby smoothening the surface of the lower sensor stack 212 and facilitating the MgOX barrier layer 210 to grow thereon with less waviness.
The plasma treatment may be applied to the Co reference layer 252 or the Co—Hf reference layer 254. However, its mild etching effect may penetrate downwards into the adjacent Ru antiparallel coupling layer 226, thus deteriorating the antiparallel coupling strength. Alternatively, the plasma treatment may be applied to the Co—Fe—B reference layer 256. However, its mild etching effect may slightly damage the metal/metal interface between the Co—Fe—B reference 256 and the Co—Fe reference layer 258, thus deteriorating the tunneling effect that relies on scattering of conduction electrons in the Co—Fe—B reference layer 256 and the Co—Fe reference layer 258. Alternatively, the plasma treatment may be applied to the Co—Fe reference layer 258. While its smoothening effect may be maximized, its mild etching effect may slightly damage the metal/oxide interface where coherent scattering of conduction electrons induces the most significant tunneling effect. It is thus recommended to apply the plasma treatment to the Co—Fe—B reference layer 256 and slightly reduce its thickness from 1.32 to 1.2 nm to alleviate concerns on the deteriorations of the antiparallel coupling strength and the tunneling effect.
The TMR read sensor 201 is patterned in a photolithographic process to produce sensor front and rear edges, and then patterned again in another photographic process to produce sensor tails at the two side regions. The Al2O3 insulation layer 202 and the longitudinal bias stack 204 are then deposited into the two side regions. The photoresist is then removed and a chemical-mechanical-polishing process is conducted. The TMR read sensor 201, the Al2O3 insulation layer 202, and the longitudinal bias stack 204 are then covered by the upper shield 208 also formed of a 1 μm thick ferromagnetic 80Ni-20Fe film, and by a gap formed of a 100 nm thick ferromagnetic Al2O3 film. A read gap 240 is defined by the thickness of the TMR read sensor 201, or a distance between the lower shield 206 and the upper shield 208. After completing the read head fabrication process, the write head fabrication process starts.
The keeper layer structure 222, the antiparallel-coupling layer 226 and the reference layer structure 224 form a flux closure where four magnetic interactions occur. First, antiferromagnetic/ferromagnetic coupling occurs between the pinning layer 220 and the keeper layer structure 222, thus increasing the easy-axis coercivity (HC) of the keeper layer structure 222 and inducing a unidirectional anisotropy field (HUA). Second, ferromagnetic/ferromagnetic coupling occurs across the antiparallel-coupling layer 226 and between the keeper layer structure 222 and the reference layer structure 224, thus inducing a bidirectional anisotropy field (HBA). Third, ferromagnetic/ferromagnetic coupling also occurs across the barrier layer 210 and between the reference structure 224 and the sense layer structure 228, thus increasing the easy-axis coercivity (HCE) of the sense layer structure 228 and inducing a ferromagnetic-coupling field (HF). Fourth, magnetostatic interaction occurs in the sense layer structure 228 due to stray fields that stem from the net magnetization of the keeper layer structure 222 and the reference layer structure 224, thus inducing a demagnetizing field (HD). To ensure proper sensor operation, HUA and HBA must be high enough to rigidly pin the magnetizations of the keeper layer structure 222 and the reference layer structure 224 in opposite transverse directions perpendicular to the ABS, while HF and HD must be small and balance with each other to orient the magnetization of the sense layer structure 228 in a longitudinal direction parallel to the ABS.
When a sense current flows in a direction perpendicular to interfaces of the TMR read sensor 201, the TMR read sensor 201 acts as a series circuit. Its extrinsic junction resistance (RJ), that depends on a sensor geometry, is a sum of RM, RMgOx and RC, where RM is the total resistance of all the metallic layers, RMgOx is the resistance of the MgOX barrier layer 210, and RC is the total contact resistance of all the interfaces. Since the resistivity of the MgOX barrier layer 210 (ρMgOx) is higher than 1×105 μΩ-cm while resistivities of all the metallic layers are lower than 200 μΩ-cm, the MgOX barrier layer 210 acts as the highest-resistance path in the series circuit. When the thickness of the MgOX barrier layer 210 (δMgOx) is large enough to exhibit a significantly high RMgOx, RM is negligible and RJ is thus the sum of RMgOx and RC. In other words, the intrinsic area resistance of the TMR read sensor 201 (RJAJ, where AJ is a junction area) is a sum of ρMgOx δMgOx and RC AJ.
When the sense current quantum-jumps across the MgOX barrier layer 210 and a magnetic field rotates the magnetization of the sense layer structure 228 from the same direction as that of the reference layer structure 224 to an opposite direction, scattering of conduction electrons at lower and upper interfaces of the MgOX barrier layer 210 induces the tunneling effect and causes an increase in the resistance from RJ to RJ+ΔRT. The strength of this tunneling effect can be characterized by a TMR coefficient (ΔRT/RJ).
It is desirable to attain a high ΔRT/RJ at a low RJAJ for ensuring high read signals from a miniaturized TMR read sensor 201 without high electronic noises and electrostatic discharges. The low RJAJ requires low ρMgOx δMgOx or RCAJ. ρMgOx has reached an intrinsic value after optimizing the deposition of the MgOX barrier layer 210 to ensure no residual Mg atoms and no excessive oxygen atoms. δMgOx has reached a minimal value, below which more pinholes will deteriorate the tunneling effect. AJ is fixed after defining a track width in a photolithographic process and a stripe height in a chemical-mechanical-polishing process. RC thus remains as the only parameter ignored in any methods of reducing RJAJ in the prior art. In details, RC is a sum of RC1, RC2 and RC3, where RC1 is the total contact resistance of all the metal/metal interfaces, RC2 is the contact resistance of the metal/oxide interface, and RC3 is the contact resistance of the oxide/metal interface.
In the prior art, the Co—Fe reference layer 258 with a thickness of as small as 0.4 nm separates the Co—Fe—B reference layer 256 with a thickness of as large as 1.2 nm from the MgOX barrier layer 210. The thin Co—Fe reference layer 258 acts as a diffusion barrier layer to reduce boron diffusion through the metal/metal interface and boron segregation at the metal/oxide interface, thereby decreasing RC1 and RC2, respectively.
In the prior art, the Co—Fe sense layer 262 with a thickness of as small as 0.4 nm also separates the Co—Fe—B sense layer 264 with a thickness of as large as 2.0 nm from the MgOX barrier layer 210. The thin Co—Fe sense layer 262 also acts as a diffusion barrier layer to reduce boron diffusion through the metal/metal interface and boron segregation at the oxide/metal interface, thereby decreasing RC1 and RC3, respectively.
In spite of the uses of the thin Co—Fe reference layer 258 and the thin Co—Fe sense layer 262 as diffusion barrier layers, the TMR read sensor 201 can exhibit ΔRT/RJ of as high as 72% at RJAJ of 0.68 Ω-μm2 after annealing for 5 hours at 280° C. This strong tunneling effect originates not only from a transformation in the Co—Fe—B reference layer 256 and in the Co—Fe—B sense layer 264 from amorphous to polycrystalline phases after annealing, but also from the maintenance of a Co—Fe—B(001)[110]//MgO(001)[100]//Co—Fe—B(001)[110] epitaxial relationship across the thin Co—Fe reference layer 258 and the thin Co—Fe sense layer 262. The phase transformation and the epitaxial relationship ensure coherent scattering of conduction electrons at the MgOX barrier layer 210, thereby inducing the strong tunneling effect.
In the invention, methods of further decreasing RC1, RC2 and RC3 are proposed, as described below.
The reference layer structure 324 comprises a first reference layer 252 formed of a 0.2-0.6 nm (or 0.4 nm) thick ferromagnetic Co film, a second reference layer 254 formed of a 0.2-0.6 nm (or 0.4 nm) thick ferromagnetic 75.5Co-24.5Hf film, a third reference layer 356 formed of a 0.4-1.0 nm (or 0.6 nm) thick ferromagnetic 65.5Co-19.9Fe-14.6B film, and a fourth reference layer 358 formed of a 0.4-1.2 nm (or 0.8 nm) thick ferromagnetic 46.8Co-53.2 Fe film. The thickness of the Co—Fe—B reference layer 356 is minimized while that of the Co—Fe reference layer 358 is maximized correspondingly to attain a total saturation moment of 0.30 memu/cm2 (corresponding to that of a 4.3 nm thick ferromagnetic 88Ni-12Fe film sandwiched between two Cu films), and to reduce boron diffusion through the metal/metal interface and boron segregation at the metal/oxide interface, thereby decreasing RC1 and RC2, respectively.
While exact atomic ratios of the layers 252, 254, 356, 358 have been described above, this by way of providing a best mode contemplated by the inventor. More generally the atomic ratios of the layers 252, 254, 356, 358 can be described as follows. The layer 254 can be 66-86 atomic percent Co and 14-34 atomic percent Hf. The layer 356 can be 55-75 atomic percent Co, 10-30 atomic percent Fe and 5-25 atomic percent B. The layer 358 can be 37 to 57 atomic percent Co and 43 to 63 atomic percent Fe.
The sense layer structure 328 comprises a first sense layer 362 formed of a 0.4-1.2 nm (or 0.8 nm) thick ferromagnetic 46.8Co-53.2Fe film, a second sense layer 364 formed of a 0.4-1.2 nm (or 1.2 nm) thick ferromagnetic 79.3Co-4.0Fe-16.7B film, a third sense layer 266 formed of a 0.6-1.8 nm (or 1.2 nm) thick ferromagnetic 75.5Co-24.5Hf film, and a fourth sense layer 368 formed of a 2.4-7.2 nm (or 4.8 nm) thick ferromagnetic 96Ni-4Fe film. The thickness of the Co—Fe sense layer 362 is maximized while that of the Co—Fe—B sense layer 358 is minimized correspondingly to attain a total saturation moment of 0.56 memu/cm2 (corresponding to that of a 8.0 nm thick ferromagnetic 88Ni-12Fe film sandwiched between two Cu films), and to reduce boron diffusion through the metal/metal interface and boron segregation at the oxide/metal interface, thereby decreasing RC1 and RC3, respectively.
While exact atomic ratios of the layers 362, 364, 266, 368 have been described above, this by way of providing a best mode contemplated by the inventor. More generally the atomic ratios of the layers 362, 364, 266, 368 can be described as follows. The layer 362 can be 37-57 atomic percent Co and 43-63 atomic percent Fe. The layer 364 can be 69-89 atomic percent Co, 0-14 atomic percent Fe and 7-27 atomic percent B. The layer 266 can be 66-86 atomic percent co and 14-34 atomic percent Hf. The layer 368 can be 86-100 atomic percent Ni and 0-14 atomic percent Fe.
Alternatively, the sense layer structure 328 comprises a first sense layer 362 formed of a 0.4-1.2 (or 0.8 nm) thick ferromagnetic 46.8Co-53.2Fe film, a second sense layer 364 formed of a 1.0-3.0 nm (or 2.0 nm) thick ferromagnetic 90.4Co-9.6Fe film, a third sense layer 266 formed of a 0.6-1.8 nm (or 1.2 nm) thick ferromagnetic 75.5Co-24.5Hf film, and a fourth sense layer 368 formed of a 2.4-7.2 nm (or 2.8 nm) thick ferromagnetic 96Ni-4Fe film. The thickness of the Co-rich Co—Fe sense layer 364 is maximized while that of the Ni—Fe sense layer 368 is minimized correspondingly to attain a total saturation moment of 0.56 memu/cm2 (corresponding to that of a 8.0 nm thick ferromagnetic 88Ni-12Fe film sandwiched between two Cu films). The Co—Fe—B sense layer is not used at all to completely eliminate boron diffusion through the metal/metal interface and boron segregation at the oxide/metal interface, thereby substantially decreasing RC1 and RC3, respectively.
While exact atomic ratios of the layers 362, 364, 266, 368 in this above described alternative embodiment have been described as exact atomic ratios, this by way of providing a best mode contemplated by the inventor. More generally the atomic ratios of the layers 362, 364, 266, 368 in this alternative embodiment can be described as follows. The layer 362 can be 37-57 atomic percent Co and 43-63 atomic percent Fe. The layer 364 can be 80-100 atomic percent Co and 0-20 atomic percent Fe. The layer 266 can be 66-86 atomic percent Co and 14-34 atomic percent Hf. The layer 368 can be 86 to 100 atomic percent Ni and 0.14 atomic percent Fe.
In a method of further decreasing RC1 in accordance with the invention, the reference layer structure 324 is in-situ formed without a plasma treatment. In other words, the Co reference layer 252, the Co—Hf reference layer 254, the Co—Fe—B reference layer 256 and the Co—Fe reference layer 258 are sequentially in-situ deposited on a wafer in a deposition module of a sputtering system. Without transfers through a transfer module to different deposition modules for depositions and to an etching module for the plasma treatment, low-RC1 metal/metal interfaces are immediately in-situ formed in the reference layer structure 324. It should be noted that the term “in-situ” is strictly defined in the invention by processes conducted only in a deposition module, without exposures to other vacuum in different transfer, deposition and etching modules, instead of to air in general. To immediately in-situ form more low-RC1 metal/metal interfaces in the lower sensor stack 312 the seed layer 218, the pinning layer 220, the keeper layer structure 222 and the antiparallel coupling layer 226 may also be sequentially in-situ deposited in the same deposition module. However, it is difficult to conduct in this way since in general there are only five or six targets in one deposition module, and it is less crucial since these layers do not affect the tunneling.
In another method of further decreasing RC1 in accordance with the invention, the sense layer structure 328 is also in-situ formed. In other words, the Fe-rich Co—Fe sense layer 362, the Co—Fe—B or Co-rich Co—Fe sense layer 364, the Co—Hf sense layer 266 and the Ni—Fe sense layer 368 are sequentially in-situ deposited on a wafer in a deposition module of a sputtering system. Without transfers through a transfer module to different deposition modules for depositions, low-RC1 metal/metal interfaces are immediately in-situ formed in the sense layer structure 328. To immediately in-situ form more low-RC1 metal/metal interfaces in the upper sensor stack 314, the cap layer structure 230 may also be in-situ deposited in the same deposition module. However, it is difficult to conduct in this way since in general there are only five or six targets in one deposition module, and it is less crucial since these layers do not affect the tunneling.
In a method of further decreasing RC2 and RC3 in accordance with the invention, the Co—Fe reference layer 358, the MgOX barrier layer 210 and the Fe-rich Co—Fe sense layer 362 are also sequentially in-situ deposited on a wafer in a deposition module of a sputtering system. Without transfers through a transfer module to different deposition modules for depositions, low-RC2 metal/oxide and low-RC3 oxide/metal interfaces are immediately in-situ formed.
The elimination of the plasma treatment leads to substantial decreases in RC1 and RJAJ, as described below. The TMR read sensors with and without the plasma treatment are deposited on bare glass substrates and wafers. The TMR read sensor with the plasma treatment comprises Ta(2)/Ru(2)/Ir—Mn(6)/Co—Fe(1.8)/Co(0.4)/Ru(0.4)/Co(0.4)/Co—Hf(0.4)/Co—Fe—B(1.2)/Co—Fe(0.4)/MgOX/Co—Fe(0.4)/Co—Fe—B(1.6)/Co—Hf(1.2)/Ni—Fe(4.8)/Ru(1)/Ta(1)/Ru(4) films (thickness in nm). The Co—Fe—B reference layer is originally 1.32 nm thick, but becomes 1.2 nm thick after the plasma treatment. The TMR read sensor without the plasma treatment comprises identical multilayer films. The only difference is that the Co—Fe—B reference layer is originally 1.2 nm thick.
After annealing in a magnetic field of 50,000 Oe for 5 hours at 280° C. in a high-vacuum oven, the TMR read sensor deposited on the bare glass substrate is measured with a vibrating sample magnetometer to determine HCE and HF. The TMR read sensor deposited on the wafer with the lower shield 206 is coated with Cu(75)/Ru(12) top conducting leads (not shown), and is probed with a 12-point microprobe in a magnetic field of about 160 Oe. Measured data from any four of the microprobe are analyzed with a current-in-plane tunneling model to determine RJAJ and ΔRT/RJ.
In addition, HF decreases sharply as RJAJ increases. For example, HF decreases sharply from 229.9 to 124.1 Oe as RJAJ increases from 0.48 and 0.78 Ω-μm2 for the TMR read sensor without the plasma treatment. Therefore, after attaining RJAJ comparable to that (0.79 Ω-μm2) of the TMR read sensor with the plasma treatment and the 0.695 nm thick MgOX barrier layer 210, the elimination of the plasma treatment causes an HF increase from 112.4 to 124.1 Oe (by as small as 11.7 Oe). This small HF decrease can also be realized by comparing two hysteresis loops as shown in
In summary,
In order for the TMR read sensor 301 to exhibit an even lower RJAJ, the metal/oxide interface at the reference layer structure 328 is preferably formed without boron diffusion and segregation. It is thus suggested in the invention to completely eliminate the Co—Fe—B reference layer 356 in the reference layer structure 328. However, the Co—Fe—B reference layer 356 with a thickness of at least 1.2 nm is needed to exhibit the desired tunneling effect in accordance with the prior art. It is speculated though that without the plasma treatment which may cause slight damages into the Co—Fe—B reference layer 356, a thinner Co—Fe—B reference layer 356 might function enough to exhibit the desired tunneling effect. In addition, a thicker Co—Fe reference layer 358 may be used to further suppress the boron diffusion and segregation at the metal/oxide interface.
Table 1 lists HF, RJAJ, ΔRT/RJ and FoM for TMR read sensors 301 without the plasma treatment and with Co—Fe—B and Co—Fe reference layers of various thicknesses. The TMR read sensor 301 comprises Ta(2)/Ru(2)/Ir—Mn(6)/Co—Fe(1.8)/Co(0.4)/Ru(0.4)/Co(0.4)/Co—Hf(0.4)/Co—Fe—B/Co—Fe/MgOX(0.74)/Co—Fe(0.4)/Co—Fe—B(2.0)/Co—Hf(1.2)/Ni—Fe(4.8)/Ru(1)/Ta(1)/Ru(4) films. To maintain the saturation moment of the reference layer structure 328 unchanged for a fair comparison, a 0.34 nm decrease in the thickness of the Co—Fe—B reference layer 356 with a saturation magnetization (MS) of 1,018 emu/cm3 requires a 0.2 nm increase in that of the Co—Fe reference layer 358 with MS of 1,734 emu/cm3. Therefore, when the Co—Fe—B reference layers are 1.2, 0.86, 0.52, 0.18 and 0 nm thick, the Co—Fe reference layers are 0.4, 0.6, 0.8, 1.0 and 1.1 nm, respectively. With the 1.2, 0.86, 0.52, 0.18 and 0 nm thick Co—Fe—B reference layers 356, the TMR read sensors 301 exhibit HF values of 107.3 to 106.7, 109.5, 135.8 and 137.7 Oe, respectively, RJAJ values of 0.73, 0.70, 0.66, 0.57 and 0.50 Ω-μm2, respectively, and ΔRT/RJ values of 80.0, 78.2, 71.4, 36.4 and 10.0%, respectively.
RJAJ gradually decreases from 0.73 to 0.66 Ω-μm2 as the thickness of the Co—Fe—B reference layer 356 decreases from 1.2 to 0.52 nm. The uses of a thinner Co—Fe—B reference layer 356 and a thicker Co—Fe reference layer 358 thus lead to less boron diffusion and segregation at the metal/oxide interface, thereby causing the RJAJ decrease. Unexpectedly, HF remains nearly unchanged as RJAJ decreases from 0.73 to 0.66 Ω-μm2, instead of increasing sharply as predicted from an inverse relationship between HF and RJAJ. The uses of a thinner Co—Fe—B reference layer 356 and a thicker Co—Fe reference layer 358 thus improve the surface flatness of the reference layer structure 328 due to less boron diffusion segregation at the metal/oxide interface, thereby maintaining the nearly identical HF at lower RJAJ. ΔRT/RJ decreases from 80.0 to 71.4% as RJAJ decreases from 0.73 to 0.66 Ω-μm2. Since FoM remains nearly unchanged, the tunneling effect in fact remains nearly identical. It is thus confirmed that without the plasma treatment which may cause slight damages into the Co—Fe—B reference layer 356, the Co—Fe—B reference layer 356 can be as thin as 0.52 nm to act as a nucleus for inducing the desired tunneling effect.
In addition, RJAJ sharply decreases to 0.50 Ω-μm2 after eliminating the entire Co—Fe—B reference layer 356. It is thus concluded that without boron segregates containing boron, which block the scattering path at the metal/oxide interface, RC2 substantially decreases and thus RJAJ reaches a minimal value. However, without the Co—Fe—B reference layer 356, the TMR read sensor exhibits ΔRT/RJ of as low as 10.0%.
Table 1 thus suggests a decrease in the thickness of the Co—Fe—B reference layer from 1.2 to 0.6 nm, and an increase in that of the Co—Fe reference layer from 0.4 to 0.8 nm. For a TMR read sensor with a designed RJAJ of 0.6 Ω-μm2, δMgOxN (calculated from
In order for the TMR read sensor 301 to exhibit an even lower RJAJ, the oxide/metal interface at the sense layer structure 328 is preferably formed without boron diffusion and segregation caused by annealing. It is thus suggested in the invention to completely eliminate the Co—Fe—B sense layer 264 in the sense layer structure 328. However, the Co—Fe—B sense layer 264 with a thickness of at least 1.6 nm is generally used to exhibit the desired tunneling effect in accordance with the prior art. The feasibility of eliminating the Co—Fe—B sense layer 264 is described below.
Table 2 lists HF, RJAJ, ΔRT/RJ and FoM for TMR read sensors 301 without the plasma treatment and with Co—Fe and Co—Fe—B sense layers of various thicknesses. The TMR read sensor 301 comprises Ta(2)/Ru(2)/Ir—Mn(6)/Co—Fe(1.8)/Co(0.4)/Ru(0.4)/Co(0.4)/Co—Hf(0.4)/Co—Fe—B(1.2)/Co—Fe(0.4)/MgOX(0.74)/Co—Fe/Co—Fe—B/Co—Hf(1.2)/Ni—Fe(4.8)/Ru(1)/Ta(1)/Ru(4) films. To maintain the saturation moment of the sense layer structure 328 unchanged for a fair comparison, a 0.4 nm decrease in the thickness of the Co—Fe—B sense layer 364 with MS of 868 emu/cm3 requires a 0.2 nm increase in that of the Co—Fe sense layer 362 with MS of 1,734 emu/cm3. Therefore, when the Co—Fe—B sense layers are 2.0, 1.6, 1.2 and 0.8 nm thick, the Co—Fe sense layers are 0.4, 0.6, 0.8 and 1.0 nm, respectively. With the 2.0, 1.6, 1.2 and 0.8 nm thick Co—Fe—B sense layers, the TMR read sensors 301 exhibit HF values of 103.5, 101.5, 98.8 and 96.8 Oe, respectively, RJAJ values of 0.73, 0.74, 0.74 and 0.73 Ω-μm2, respectively, and ΔRT/RJ values of 79.4, 79.5, 78.4 and 71.7%, respectively.
HF, RJAJ and ΔRT/RJ remain nearly unchanged as the thickness of the Co—Fe—B sense layer 364 decreases from 2.0 to 1.2 nm. The uses of a thicker Co—Fe sense layer 362 and a thinner Co—Fe—B sense layer 364 thus appear to function well in maintaining the strong tunneling effect. It is understood though that as the Co—Fe sense layer 362 is thicker than 1.0 nm and the Co—Fe—B sense layer 364 is thinner than 0.8 nm, the tunneling effect starts to deteriorate. This deterioration in fact originates from an insufficient scattering length (a sum of thicknesses of the Co—Fe sense layer 362 and the Co—Fe—B sense layer 364), instead of the use of the thinner Co—Fe—B sense layer 364.
Table 2 thus suggests an increase in the thickness of the Co—Fe sense layer from 0.4 to 0.8 nm, and a decrease in that of the Co—Fe—B sense layer from 2.0 to 1.2 nm. For a TMR read sensor with a designed RJAJ of 0.6 Ω-μm2, δMgOxN will stay at 0.734 nm, HFN will decrease from 125.9 to 121.8 Oe (by 3.1 Oe), and FoMN will decrease from 109.4 to 106.3 (by 3.1). In addition, the thickness of the sense layer structure 328 will decrease by 0.4 nm, and thus the read gap will also decrease by 0.4 nm.
A comparison between Tables 1 and 2 reveals different diffusion behaviors of boron atoms in the Co—Fe—B reference layer 356 and in the Co—Fe—B sense layer 364. It seems easy for boron atoms in the Co—Fe—B reference layer 356 of as thin as 0.52 nm to diffuse upwards through the Co—Fe reference layer 358 of as thick as 0.8 nm, but difficult for boron atoms in the Co—Fe—B sense layer 364 of as thick as 2.0 nm to diffuse downwards through the Co—Fe sense layer 362 of as thin as 0.4 nm.
To further explore the feasibility of eliminating the Co—Fe—B sense layer 264, the Co—Fe—B sense layer 264 is replaced by a new second sense layer 364 preferably formed of a ferromagnetic 90.4Co-9.6Fe film. To maintain the saturation moment of the sense layer structure 328 unchanged for a fair comparison, a 2.0 nm thick Co—Fe—B sense layer 364 with MS of 868 emu/cm3 is replaced by a 0.9 nm thick Co-rich Co—Fe sense layer 364 with MS of 1,442 emu/cm3.
The Co-rich Co—Fe sense layer 364 differs from the Fe-rich Co—Fe sense layer 362 in that its lower Fe content may lead to a more negative saturation magnetostriction (λS), and thus it can be thicker for extending the scattering length (a sum of thicknesses of the Fe-rich Co—Fe sense layer 362 and the Co-rich Co—Fe sense layer 364). To maintain the saturation moment of the sense layer structure 328 unchanged for a fair comparison, a 0.2 nm increase in the thickness of the Co-rich Co—Fe sense layer 364 with MS of 1,442 emu/cm3 requires a 0.53 nm decrease in that of the Ni—Fe sense layer 368 with MS of 543 emu/cm3.
However, to maintain the sense-layer saturation moment unchanged, the replacement of the 1.6 nm thick Co—Fe—B sense layer 264 with the 2.0 nm thick Co-rich Co—Fe sense layer 364 requires the Ni—Fe sense layer 368 to be thinner than the Ni—Fe sense layer 268 by as large as 2.8 nm. The Ni—Fe sense layer 368 does not affect the tunneling, but plays a key role in attaining good soft ferromagnetic properties. For example, by decreasing the thickness of the Ni—Fe sense layer 368 from 5.6 to 2.8 and 0.8 nm for designs of the sense-layer saturation moment of 0.56 and 0.42 memu/cm2, respectively (corresponding to that of 8 and 6 nm thick ferromagnetic 88Ni-12Fe films sandwiched between two Cu films, respectively), λS increases from −4.17×10−6 to −1.41×10−6 and 1.61×10−6, respectively. To maintain a more negative λS, Fe atoms which dominates λS may be eliminated by replacing the 2.8 and 0.8 nm thick Ni—Fe sense layers 368 with 3.2 and 1 nm thick Ni sense layers with MS of 463 emu/cm3, respectively, and atomic mixing at an interface between the sense layer structure 328 and the cap layer structure 230 may be reduced by replacing the Ru first cap layer with a Pt cap layer.
Since the reference layer structure 324, the barrier layer 210 and the sense layer structure 328 are deposited independently in three different deposition modules in accordance with the invention, all the metal/metal interfaces in the reference layer structure 324 and the sense layer structure 328 are in-situ formed, but the metal/oxide and oxide/metal interfaces are still ex-situ formed. In a method of further decreasing RC2 and RC3 in accordance with the invention, the Co—Fe reference layer 358, the MgOX barrier layer 210 and the Fe-rich Co—Fe sense layer 362 are also sequentially in-situ deposited on a wafer in a deposition module of a sputtering system. Without transfers through a transfer module to different deposition modules for depositions, low-RC2 metal/oxide and low-RC3 oxide/metal interfaces are immediately in-situ formed. The in-situ formed metal/metal, metal/oxide and oxide/metal interfaces may ensure the cleanness of the scattering path in the reference layer structure 324, the MgOX barrier layer 210 and the sense layer structure 328, so that saturation moments can be precisely controlled and conduction electrons can be effectively scattered to attain a strong tunneling effect.
In summary, Table 3 lists various methods of attaining low-RC1 metal/metal, low-RC2 metal/oxide and low-RC3 oxide/metal interfaces in accordance with the invention, and their evaluation based on changes in δMgOxN, HFN and FoMN (δMgOxN, ΔHFN and ΔFoM, respectively). To attain low-RC1 metal/metal, low-RC2 metal/oxide and low-RC3 oxide/metal interfaces in accordance with the invention, the plasma treatment is eliminated, thinner Co—Fe—B and thicker Co—Fe reference layers are used, the Co—Fe—B sense layer is replaced by the Co-rich Co—Fe sense layer, and in-situ metal/oxide/metal interfaces are formed. For a TMR read sensor with a designed RJAJ of 0.6 Ω-μm2, δMgOxN will increase by 0.088 nm, HFN will decrease by 35.9 Oe, and FoMN will decrease by 6.1.
While various embodiments have been described, it should be understood that they have been presented by way of example only, and not limitation. Other embodiments falling within the scope of the invention may also become apparent to those skilled in the art. Thus, the breadth and scope of the invention should not be limited by any of the above-described exemplary embodiments, but should be defined only in accordance with the following claims and their equivalents.
