Using Phase Changes to Store Information in the Magnetic Permeability

Authors: Alan S. Edelstein^*, John Timmerwilke, Jonathan R. Petrie 

Affiliation: US Army Research Laboratory, Adelphi, Maryland, USA

*Email: aedelstein@cox.net

Finding better methods for storing information was a serious problem for the earliest computers; considerable effort is still being devoted to decreasing the cost and increasing the density and lifetime of stored information. Initial methods of storage included using acoustic delay lines and pixels on the display of cathode ray tubes [1]. Though the current methods for storing information, which include magnetic hard disks, magnetic tape, and various forms of random access memory (RAM) are far superior to these early methods, they still have significant limitations.

As an example, the information on hard disks or magnetic tape is written by a magnetic field and stored as regions, i.e. bits, having different directions for their remanent magnetization. Thus, this information can be erased by exposure to a magnetic field. Furthermore, to avoid thermal upsets of the spin direction of the bits there is a trade-off between how long the information can be stored without acquiring too many incorrect bits and the information density on hard disks. Maintaining a magnetic hard disk lifetime of about seven years has made it difficult to increase the information density in hard disk drives. Magnetic tape degrades in about 20 years.

To avoid these problems, we have been developing a new approach for storing information that we call Magnetic Phase Change Memory (MAG PCM). Instead of using the direction of magnetic remanance, information is stored in bits of soft ferromagnetic material having different values for their magnetic permeability. The initial magnetic permeability of a soft ferromagnetic material is reversible and dependent only on its atomic structure. Thus, it is independent of the magnetic field. The permeability of iron rich FeGa single crystal alloys [2]  is an interesting example of soft magnetic behavior.

The property we use is that high permeability bits attract magnetic fields and low permeability bits do not. To read the information stored in bits with high or low permeability we measure their effect on a probe field. We have primarily focused on writing and changing the permeability of bits of amorphous ferromagnet, 2826 MB Metglas, films by thermally heating with a laser. The nearest neighbor atoms in amorphous ferromagnets materials do not have long range order. Thus, they do not have much crystalline anisotropy or coercivity and are soft ferromagnets with large values for their permeability. When amorphous materials are heated above their glass temperature, they crystalize and have larger values for their coercivity and smaller values for their permeability. The glass temperature of 2826 MB Metglas is 410^oC. At temperatures as high as 200^oC, 2826 MB Metglas will retain its high permeability for hundreds of years.

Figure 1: a) Microscope view of three 50 micron wide crystallized lines written into amorphous Metglas. b) Voltage of the MTJ reader as it moved over the three crystalline lines written in the presence of a 32 Oe probe field.

We have read the effect on a probe magnetic field near each bit using both magnetic tunnel junction (MTJ) sensors [3,4], and spin transfer oscillators [5]. Figure 1a shows a microscope image of three crystalline lines in Metglas written by a focused 1.966 micron (Tm fiber) laser. The output voltage of an MTJ sensor is plotted in Figure 1b as it is swept over the crystallized Metglas lines shown in Figure 1a. One sees that the crystallized lines affect a 32 Oe probe field. The crystallized lines do not attract the probe field as much as the amorphous Metglas film does which causes the magnetic sensor to measure a larger field. Smaller nanometer-sized bits were created using e-beam lithography. Figure 2a is a microscope image of two square 0.9 mm arrays of 300 nm bits of amorphous Metglas. Figure 2b shows a scanning electron microscope (SEM) image of the 300 nm diameter bits. Laser heating was used to crystallize all of the bits in the left array. Figure 3 shows the voltage output of the MTJ sensor when it is moved over the two arrays before and after the laser heating. One sees that the bits in the left array no longer attract the magnetic field lines of the probe field.

Figure 2. (a) Microscope image of two square 0.9 mm arrays of 300 nm diameter bits of Metglas; (b) scanning electron microscope image of the 300 nm diameter bits of Metglas.

This new approach for storing information has several advantages. One can write bits with at least three different values [6]  for their permeability. The bits will not be corrupted by a magnetic field or thermal upsets and therefore should last decades. It should be possible to write nm sized bits economically by an adaption of heat assisted magnetic recording (HAMR) [7], a technology that is being developed by hard disk companies such as Seagate and Western Digital to cope with the problem mentioned above of maintaining stability against thermal upsets. In HAMR a laser and a near field transducer is used to heat nm sized regions to 710^oC to decrease the coercivity so that they can be written without using as large a magnetic field. What we need to do is simpler, in that we do not need a magnetic field and for archiving we do not need to rewrite.

Figure 3: Magnetic tunnel junctions scans before (red, o) and after (blue, x) the left array of 300nm Metglas bits in Fig. 2 were crystallized by heating with a laser.

MAG PCM has the potential for a combination properties not found in other storage technologies. It should have decades of longevity and the rapid access and high density of future hard disks. We have a clear path for developing MAG PCM into commercial products for long term storage applications such as archiving. Though it is unnecessary for archiving, we have found that we can rewrite our bit [8], i.e., return a crystallized bit to an amorphous state.

This work was done while we were at the US Army Research Laboratory.

References:
[1] D.R. Hartree, M.H.A. Newman, M.V. Wilkens, F.C. Williams, J.H. Wilkinson, A.D. Booth, ” A discussion on computing machines”, Proceedings of the Royal Society of London Series A- Mathematical and Physical Sciences, 195:1042, 265 (1948). Abstract.
[2] Harsh Deep Chopra, Manfred Wittig, “Non-Julian magnetostriction”, Nature, 521, 340-343 (2015). Abstract.
[3] J.R. Petrie, K.A. Wieland, R.A. Burke, G.A. Newburgh, J.E. Burnette, G.A. Fischer, A.S. Edelstein, “ A non-erasable magnetic memory based on the magnetic permeability”, Journal of Magnetism and Magnetic Materials, 361, 262 (2014). Abstract.
[4] John Timmerwilke, J.R. Petrie, K.A. Wieland, Raymond Mencia, Sy-Hwang Liou, C.D. Cress, G.A. Newburgh, A.S. Edelstein, “Using magnetic permeability bits to store information”, Journal of Physics D: Applied Physics, 48, 405002 (2015). Abstract.
[5] J.R. Petrie, S. Urazhdin, K.A. Wieland, A.S. Edelstein, “Using a spin torque nano-oscillator to read memory based on the magnetic permeability”, Journal of Physics D: Applied Physics, 47, 055002 (2014). Abstract.
[6] J.R. Petrie, K.A. Wieland, J.M. Timmerwilke, S.C. Barron, R.A. Burke, G.A. Newburgh, J.E. Burnette, G.A. Fischer, and A.S. Edelstein, “A multi-state magnetic memory dependent on the permeability of Metglas”, Applied Physics Letters, 106, 142403 (2015). Abstract.
[7] M.H. Kryder and Soo Kim Chang, “After hard drives—what comes next?” IEEE Transactions on Magnetics, 45, 3406 (2009). Abstract.
[8] Unpublished data.