Lecture 6 Alternative storage technologies All optical recording Racetrack memory Topological kink solitons Flash memory Holographic memory Millipede Ferroelectric memory
All-optical recording It is possible to reverse the magnetisation of a medium without using a magnetic field. Experiments using femtosecond laser pulses demonstrate the feasibility of this approach, although a practical data storage application remains to be developed. In all-optical recording the magnetisation direction of a medium can be controlled using circularly polarised light. Circularly polarised light contains photons that carry angular momentum. Some of this momentum is transferred to the magnetic moments in the medium to induce magnetisation reversal. Apparatus for all-optical recording [1]
All-optical recording II The circularly polarised light creates domains in the medium magnetised up or down depending on the sense of the light polarisation. Magnetisation switching is ultra-fast, on the order of the duration of the laser pulse (40 fs). Image of all-optical recording In magneto-optics the medium magnetisation changes the polarisation of incident light. Here an inverse, optomagneto effect allows light to change the magnetisation direction.
Racetrack memory Magnetic nanowires are used in this memory device. Domains in the wire are used to represent bits of information. The domains can be moved by passing a current through the wire [2].
Racetrack memory II The domain wall velocity depends on the current. The length of the current pulse determines the amount the domains move. The image shows a 12 mm long wire.
Topological kink solitons These are similar in some ways to racetrack memory. A thin-film stack is formed consisting of alternating magnetic and non-magnetic layers. The non-magnetic layer creates anti-ferromagnetic coupling between adjacent magnetic layers. Solitons are used to store information in the stack [3]. The solitons are created when parts of the stack with different order parameters meet. The solitons can be propagated along the stack by applying a rotating magnetic field. Placing a sensor somewhere along the stack allows the solitons to be detected as they pass by.
Topological kink solitons II The sense of rotation of the external magnetic field determines whether the soliton moves up or down the stack. Arrays of stacks may be formed to create a high-density memory. Propagation of a soliton An array of thin-film stacks
Flash memory Flash memory cells are similar to MOSFETs, but with two gates: a control gate and a floating gate. Oxide layers separate the gates, preventing current from flowing through the device. The source and drain are n-type and the substrate is p-type. A single flash memory cell
Flash memory II Applying a positive voltage to the word line and the bit line allows a current to flow from the source to the drain. Some electrons also tunnel through the lower oxide layer and are trapped in the floating gate. The stored charge represents a 1. Applying a negative voltage to the word line empties the floating gate: this state represents a 0. Writing a 1 state to a flash memory cell
Flash memory III Flash memory cells are coupled together so that they can be erased in blocks. Subsequently, individual cells can be written. This makes the memory very fast. However, after repeated writes the oxide layers can degrade and become leaky, leading to eventual cell failure. Longevity is typically from 10000 to 1000000 writes, depending on the type of cell. An array of flash memory cells
Holographic memory Holographic memories have the potential to store vast amounts of information due to their 3D nature. However, implementation has so far lagged behind other memory technologies. Data are recorded in pages. The data are encoded on a spatial light modulator. The hologram is stored as the interference pattern of the reference and signal beams. Recording data The storage medium is usually a lithium niobiate crystal or photopolymer. Each page of data is stored in a different volume of the medium, depending on the incident angle of the reference beam. The medium can be stationary and the angle of the reference beam varied by a mirror.
Holographic memory II To retrieve data only a reference beam is needed. The angle of the reference beam determines which data are retrieved. Data retrieval can also be very fast as whole pages of data are imaged on the detector at the same time. The reference beam must be precisely aligned to obtain the correct page of data. Data retrieval The cost of the recording medium is the major factor holding back holographic memory.
Millipede The millipede is (was) a nanomechanical AFM-based data storage system. An array of cantilevers thermally record data on a thin polymer film. The cantilevers can be positioned with nanometer-scale accuracy over the surface of the medium.
Millipede II Writing data is achieved by heating a tip and melting the medium to form a pit. Reading data is accomplished by measuring the resistance of the tip, which depends on the tip temperature. This, in turn, depends on the area of the tip in contact with the medium: more when the tip is in a pit, less when it isn't.
Ferroelectric memory This uses a single crystal of ferroelectric material, such as LiTaO 3 (lithium tantalate). Recording of data is achieved using a probe and applying voltage pulses [4]. Recording information on a ferroelectric medium
Ferroelectric memory II Readback uses scanning non-linear dielectric microscopy (SNDM) to observe the polarisation distribution in the medium. Data recorded on LiTaO 3
Conclusions At the moment hard disc drives form the vast majority of storage devices used in data centres. However, there are many alternative data storage technologies which may, or may not, be used in the future. Novel ideas are constantly being invented: technologies which can take advantage of all three dimensions to maximise the amount of data stored in a unit volume should ultimately be sucessful. References [1] C. D. Stanciu, Phys. Rev. Lett. 99, 047601, (2007). [2] S. S. P. Parkin, Science 320, 190, (2008). [3] R. Lavrijsen, Nature 493, 647, (2013). [4] Y. Cho, Proc. Symp. Ultrason. Electron. 27, 21, (2006).