Description:
FIELD OF THE INVENTION
This invention relates to information storage arrangements of the type in which information is stored as magnetic single wall domains.
BACKGROUND OF THE INVENTION
A magnetic single wall domain is a magnetic domain of a type encompassed by a single domain wall which closes on itself in the plane of a host magnetic layer forming typically a right circular cylinder in the layer. Inasmuch as a single wall is self-defined in a plane of movement, it is free to move in two dimensions in such a plane as is now well known.
A typical material in which single wall domains can be moved is an epitaxially grown layer of Garnet. Such material have a preferred direction of magnetization along an axis out of the plane of the layer, nominally normal to that plane. A single wall domain, in a material of this type, is magnetized in one direction along that axis whereas the remainder of the layer is magnetized in the opposite direction, the domain appearing as a dipole oriented normal to the plane of the layer.
Single wall domains are known to be stable for a range of bias values for which the diameter of domains in a host layer varies by a factor of three between a maximum value at which a domain strips out and a minimum value at which the domain collapses, values determined for layers the areas of which are large compared to the maximum diameter for domains in the layer. These areas are such that field contributions from the edges of the layer are negligible, a condition met for example, by layers having dimensions ten times the maximum diameter of domains in the layer. The bias field is directed nominally normal to the plane of the host layer and is of a polarity to constrict a domain. Consequently, a relatively high value of bias maintains a domain at its collapse (minimum) diameter whereas a relatively low value of bias maintains a domain at its strip-out (maximum) diameter.
Typically, the bias field is set to maintain a domain at a diameter in the middle of its stable range in order to achieve optimum operating margins. An average bias field range corresponding to an average range of stable domain diameters is about 20 oersteds. Naturally, an increase in the bias field range would lead to increased operating margins and the elimination for the need of an externally supplied bias field would reduce the cost of arrangements of this type.
BRIEF DESCRIPTION OF THE INVENTION
The invention is based on the realization that a layer of material in which single wall domains can be moved can be formed with dimensions to supply the equivalent of a bias field for maintaining a domain at a preselected diameter. Specifically, a layer of material in which domains can be moved is formed with a width w such that the width of the layer is about the order of and typically about equal to the maximum domain diameter D m for the layer. For epitaxially grown layers having thicknesses of about 4 microns, the maximum domain size is typically about 8 microns and the width of the stripe of material is about 10 microns.
A variety of domain patterns are found to coexist in such stripes, a semicircular variety which remains attached to one side of the stripe or the other, a circular (viz: single wall) domain, and a two-wall domain where in the last case the walls extend from one edge of the stripe to the other. An input arrangement including a pair of conductors which overlie the surface of the stripe and extend beyond opposite edges at an input stage provide one of the four possible domain configurations when pulsed synchronously depending upon the polarity of the pulses.
BRIEF DESCRIPTION OF THE DRAWING
FIG. 1 is a schematic representation of magnetic domain arrangements in accordance with this invention;
FIG. 2 is a cross-sectional view of one of the magnetic elements of FIG. 1;
FIGS. 3 through 8 are schematic representations of portions of the arrangement of FIG. 1; and
FIGS. 9, 10, and 11 are schematic representations of portions of alternative magnetic domain arrangements in accordance with this invention .
DETAILED DESCRIPTION
FIG. 1 shows an illustrative arrangement 10 including a single crystal substrate 11 on the surface of which stripes 12, 13, 14, and 15 of magnetic material are grown epitaxially, for example, from the liquid phase. Typically, a single crystal epitaxial film is formed over the entire surface of substrate 11 and thereafter divided into stripes of width w by well-known photolithographic techniques or by ion or electron beam milling. FIG. 2 shows a cross-sectional view of the substrate and stripes of FIG. 1 taken along broken line 2--2'. The substrate 11 in FIG. 2 can be seen to include stripes grown to a thickness h having widths w.
A suitable epitaxial layer of relatively large area is capable of maintaining stable domains having a maximum diameter which is a function of the material parameters and the thickness h as is now well understood. A domain of such maximum diameter is represented by a pair of broken vertical lines in stripe 13 in FIG. 2, the diameter (maximum) being designated D m . In large area layers, domains are maintained at the maximum diameter by a bias field of a given minimum value. In accordance with this invention, the width w ≅ D m for a given stripe for providing a bias field for constricting single wall domains within the stripe in the absence of an externally supplied bias field.
It has been discovered that a number of different types of domains in addition to single wall domains are stable in a stripe having a width in accordance with this invention and that all of these domains are movable simultaneously by a propagation arrangement of the type shown in FIG. 3.
FIG. 3 shows a representative stripe 12 along with a pair of conductors 31 and 32 which may be recognized as operative to move single-wall domains in stripe 12. The conductors are of serpentine geometry offset from one another for producing magnetic fields for moving such domains in stripe 12 when pulsed. The conductors are connected between a propagation pulse source 33 of FIG. 1 and ground and are pulsed alternatively first with pulses of one polarity and then with pulses of the opposite polarity as indicated by the pulse forms 34 to the left as viewed in FIG. 3.
Alternatively, conductor 32 is replaced either by magnetically soft dots or by forming the stripe itself with a serrated edge. In either case, the dots or the relatively wide ports of a (serrated) stripe are arranged with respect to the period of conductor 31 to offset a domain from any position to which it is moved by a pulse in conductor 31 thus ensuring domain movement along a selected direction in the channel. Representative dots for one period are shown at 35 in FIG. 3. Representative serrated edges are shown dotted at 36 in FIG. 3. Typically the wide portion of the film (with serrated edges) is 45° out of phase with the conductor period.
The propagation arrangement of FIG. 3 is also operative to move other types of domains in stripe 12 as well. For example, semicircular (SC) domains can be formed, attached to the top or the bottom edges of stripe 12 as represented by the broken parabolic lines designated D sct and D scb in FIGS. 5 and 6, respectively. A conventional two-wall (TW) domain also is stable in stripe 12 as represented by the pair of broken vertical lines designated D tw in FIG. 7.
The broken lines D tw in FIG. 7 are not to be confused with the broken lines in stripe 13 of FIG. 2. The latter represents the maximum diameter of a domain such as single-wall domain D sw of FIG. 4. The former represents a domain bounded by two spaced-apart walls which intersect the edges of the stripe.
In as much as domains of all of these types are moved in response to pulses on the conductors 31 and 32 of FIG. 3, they are useful for representing information. Such a use requires the controlled introduction and detection of such domains.
The controlled introduction of domains into stripe 12 occurs at an input stage coupled by a pair of input conductors 41 and 42 and a conventional conductor domain generator designated G as shown in FIG. 4 illustratively operative to introduce a domain with two spaced-apart walls. Each of conductors 41 and 42 is connected between an input pulse source 43 of FIG. 1 and ground and has a width approximately twice that of the diameter of a single wall domain in stripe 12 in order to move the domains uniformly into the desired geometry at the input position as is described hereinafter. The input pulse source is operative under the control of control circuit 44 to apply pulses of a first or second polarity to each of conductors 41 and 42.
Arrows i are shown adjacent each of the input conductors indicating the direction of current flow in the conductor during those pulses for arrangements where the conductors are separated from the substrate 11 by stripe 12. The direction of the arrow in each instance may be taken as the direction of the thumb in using the right-hand rule for determining the direction of the resulting field. To be consistent with this convention, all domains herein are taken to have flux directed toward the viewer (positive) and the remainder of the stripe has flux directed away from the viewer (negative).
For arrows (currents) i directed as shown in FIG. 4, a positive field is generated between the two conductors (41 and 42) and a two wall domain generated at G is converted into a single wall domain D sw . When pulses are applied to conductors 41 and 42 to produce currents as represented by arrows i in FIG. 5, a domain D sct adjacent the top edge of stripe 12 results as shown in the figure.
FIGS. 6 and 7 show the direction for arrows i for converting a two wall domain at G into a domain D scb adjacent the bottom edge of stripe 12 and into a domain D tw which extends from edge to edge of stripe 12, respectively. It should be clear, then, that the polarity of the pulses applied concurrently to conductors 41 and 42 determines the type of domain provided at the input stage for propagation.
FIG. 8 shows all of the different types of domains and the information which they may be taken to represent. To be specific, as viewed from right to left in FIG. 8, a domain D scb (++) may be taken to represent a binary "1" and the associated absence of a domain (--) at the top edge of stripe 12 may be taken to represent an associated binary "0." On the other hand, a domain D sct may be taken to represent a binary "1" at the top edge of stripe 12 and a binary "0" at the bottom. A single-wall domain D sw and a two-wall domain D tw present negative poles in the first instance and positive poles in the second instance at both edges of the stripe and accordingly may be taken to represent 0--0 and 1--1, respectively. The domains of FIG. 8 accordingly may be be taken to represent the information 0101 at the top edge of strip 12 and 1001 at the bottom edge, as viewed from right to left in FIG. 8, as the information is moved along the channel into an output stage.
Of course, all the different types of domains need not be employed. For example, a semicircular domain and a two-wall domain could be employed to represent the two binary values.
The output stage is defined at a selected propagation stage illustratively by the presence of first and second magnetoresistive elements 80 and 81 as shown in FIG. 8. Magnetoresistive elements and their use for providing signals indicative of the presence of a magnetic field provided by magnetic domains are now well understood in the art. Therefore, the elements are not discussed in detail here. Suffice it to say that illustratively, the magnetoresistive elements are coupled to the edges of stripe 12 for generating signals under the control of control circuit 44 responsive to the fields associated with the domains of FIG. 8. The resulting signals are amplified by amplifiers 82 and 83 of FIG. 8 and applied to a utilization circuit represented by block 84 of FIG. 1.
It is frequently advantageous for every stage of a single-wall domain propagation arrangement to be occupied by a domain. This is particularly true in such arrangements where closed-loop information channels are employed as described in my copending application Ser. No. 49,273 filed June 24, 1970 now U.S. Pat. No. 3,636,531. That application describes an arrangement in which single-wall domains move along one side or the other of a rail in what is commonly known as a lateral displacement coding (LDC) arrangement. The rail is defined by a closed loop strip of magnetically soft material on the surface of the layer in which the domains move. Movement of domains along the rail is effected by propagation conductors much as is shown in FIG. 3.
FIG. 9 shows one such closed loop propagation channel, represented by annulus 90, comprising a stripe of material in which single-wall domains can be moved. The strip of magnetically soft material in this instance overlies the stripe along the center line thereof. When every stage of such an arrangement is occupied by a domain, repulsion forces which exist between domains, are operative to reduce the dimensions of domains along the axis of domain movement in opposition to exchange fields which tend to enlarge domains. These repulsion fields aid bias fields, which tend to constrict domains also, resulting in a reduction in the bias field which would otherwise be necessary to maintain a single-wall domain at a selected diameter. In accordance with this invention, relatively wide stripes (viz: wider than the maximum domain diameter) thus can be used when every stage of arrangements defined in stripe 12, for example, is occupied.
Lateral displacement arrangements, in which all stages are normally occupied, are realized in accordance with another aspect of this invention conveniently by providing a groove 100 in a stripe 101 as shown in FIG. 10 rather than by providing a magnetically soft strip. FIG. 11 shows a top view of the structure of FIG. 10 where domains DA and DB indicate the zero and one positions for domains (viz. in consecutive stages) for movement along the channel in response to pulses applied to propagation conductors substantially as described in connection with FIG. 3.
In arrangements where every stage is occupied, the input configuration of FIGS. 4 through 7 is operative normally to reshape an existing single-wall domain rather than to provide a domain anew and the input at G is not necessary. The former requires an advantageously reduced drive field. The drive fields necessary for the latter are reduced by providing, for example, a localized area of reduced coercive force at G in the input stage.
Mathematically, it is clear also that single-wall domains are stable in (etched or milled) stripes of magnetic material as shown in FIG. 1 in the absence of an externally supplied bias field. The theory of single-wall domains is described by A. A. Thiele in the Bell System Technical Journal, Vol. 50, No. 3, dated March 1971 at page 725 et seq. In accordance with this theory, values of H b /M can be calculated for domain stability as a function of h/l where H b is a bias field, M is the magnetization of the domain layer, and h is the thickness of the layer. The term 1 is a material parameter l = ρ W μ O /M 2 where cl ρ W = wall energy (Joules/meter 2 )
μ O = permeability of free space
M = magnetization (Tesla's)
and is called the intrinsic (or material) length as defined in the above Thiele publication. But the field H b provided by a stripe having a width-to-thickness ratio W/h and a thickness to intrinsic length ratio of h/l is given by:
From equations (1 ) and (2 ) and from Thiele's results, Table I is computed:
TABLE 1 TA h/l
Domain Diameter The effective bias W/h h(min-max) Field M O H b /M due (min-max) W/h ____________________________________________________________
______________ 111 10 - 33 0.020 - .032 20 -32 1.5 3.5 - 12.8 0.09 - 0.06 7 - 10 2.4 2.1 - 7.2 0.16 - 0.10 4 - 6.4 3.0 1.3 - 4.0 0.25 - 0.18 2.5 - 3.5 4.0 1.0 - 3.0 0.33 - 0.23 1.5 - 2.2 5.0 0.8 - 2.4 0.38 - 0.28 1.2 - 2.0 10.0 0.5 - 1.4 0.49 - 0.42 0.8 - 0.9 ____________________________________________________________
______________
Note that the larger width-to-thickness (W/h) values give effective bias fields which correspond to domains of maximum stable diameter which diameter in turn is larger than the width of the stripe. This means that isolated domains of maximum diameter cannot be maintained as a practical matter in stripes of widths to provide bias fields for them. Domains of lesser diameter are used in such cases.
Since a domain diameter is a function of layer thickness for any given material, the criteria for the width of a stripe in accordance with this invention may be expressed in terms of the maximum stable domain diameter D m . When every stage of a channel is occupied as described above, the resulting repulsion forces add to the effective bias field and increase the permissible width.
For stripes of orthoferrite platelets such as yttrium orthoferrite, 250 microns wide and 75 microns thick, a 9 oersted bias field is adequate to maintain single-wall (viz: circular) domains. In practice, the platelet (before etching into stripes) requires about 27 oersteds (1,800- 2,200 amperes per meter) for maintaining stable single-wall domains. The effective bias, due to the edge of the stripe in this instance where W/h = 3.3, is 1,400 amperes per meter.
For a representative epitaxially deposited garnet film such as Europium Erbium Gallium garnet (Eu 2 Er 1 Ga .7 Fe 4 .3 O 12 formed from the liquid phase on a substrate of nonmagnetic gadolinium gallium garnet, single-wall domains are stable over a range of diameters of from 4 to 12 microns for a film thickness of 4 microns. An externally supplied bias field of 60 oersteds maintains domain diameters at 6 microns. When the film is etched into stripes having widths of 7.2 microns, the bias field is supplied by the edges of the stripes in the absence of an external bias field source.
What has been described is considered merely illustrative of the principles of this invention. Therefore, various modifications can be devised by those skilled in the art in accordance with those principles within the spirit and scope of this invention. For example, it should be clear that conductors 31 and 32 as shown in FIG. 3 may be extended to couple all the stripes of FIG. 1 for providing propagation fields simultaneously for moving domains in a plurality of propagation channels. Also, the cross-sectional shape of a stripe may be varied to control the positioning of domains. For example, the thickness of the stripe can be reduced at the edges as indicated by broken lines 110 in FIG. 10 for permitting operation of an LDC arrangement with stripes having widths of three to four D m. The domains in this instance are confined to the center of the stripe which now in cross section appears as a truncated pyramid, domains being confined to the region in which opposite faces of the stripe are parallel. In addition, it may be advantageous, to form the stripes from a continuous epitaxial film without completely removing the material between stripes. If, for example, the thickness of the film is reduced by one third between stripes, the bias contribution in such a case would be about one third of that contributed by a like arrangement in which the film is entirely absent between stripes.