Ferromagnetic Gate Memory

This invention pertains in general to nonvolatile memories, and more particularly to nonvolatile memories which store information in a ferromagnetic gate region.

Nonvolatile memories retain their stored information after their power supply has been turned off, and have many applications in digital information systems, control apparatus, and other electronic systems. Although several forms of nonvolatile memories exist, each has serious shortcomings.

Over the last ten years, much work has been performed developing magnetic bubble memories. However, bubble memories remain complex, and require a large current drive for their operation. Additionally, the cost of bubble memory greatly exceeds that of other memory devices.

EPROMs and EEPROMs are capable of retaining information over the extended periods of time without losing data, but the charge, and therefore the data, within an EPROM or an EEPROM will eventually be lost. Furthermore, changing the information stored in an EPROM or an EEPROM cannot be performed rapidly. Static RAMs and dynamic RAMs, while not nonvolatile, can be maintained by a battery during periods when system power is turned off. However, batteries have a relatively short useful life, and thereby present the danger of losing valued information as the batteries wear out.

While important strides have been made in improving the storage capacity of random access memories, it is still desirable to increase the storage capacities further, especially in nonvolatile devices.

Thus, a need has arisen in the industry for a fast, nonvolatile memory having a high storage capacity.

The present invention disclosed and claimed herein describes a method and apparatus for storing information which substantially eliminates problems associated with prior memory devices.

In one aspect of the invention, information is magnetically stored in a memory cell by passing a current beneath a ferromagnetic region, thereby magnetizing the ferromagnetic region. Current in a first direction causes a predetermined magnetic field in the ferromagnetic region, and current in the opposite direction will create an opposite magnetic field. To read information from the ferromagnetic memory cell, a small current is passed beneath the magnetized ferromagnetic region. Depending on the orientation of the magnetic field of the ferromagnetic region, the electrons will be deflected towards or away from the surface of the substrate. Electrons deflected towards the surface of the substrate will be scattered by surface impurities, thus reducing the mobility of the electrons, and thereby decreasing the magnitude of the resultant current. A current subjected to a magnetic field pushing the electrons away from the surface will not exhibit a mobility loss due to surface scattering effects. Hence, the magnetic field orientation representing the stored information can be detected by measuring the magnitude of the current and comparing it against a reference value to determined whether surface scattering was in affect.

It is a techanical advantage of this aspect of the invention that nonvolatile memories capable of storing information over a long period of time can be produced.

It is a further advantage of the present invention that the resulting memory will have relatively fast access times, limited only by how quickly the domain can be magnetically changed in the ferromagnetic region. It is another advantage of the present invention that a simple structure for the memory cells may be obtained which allows a high density of memory cells to be formed on a single chip.

For a more complete understanding of the present invention and the advantages thereof, reference is now made to the following description taken in conjunction with the accompanying drawings in which:

FIG. 1 illustrates a perspective view of a memory cell storing information in a ferromagnetic gate;

FIGS. 2a-b illustrates a perspective view of the memory cell of FIG. 1 during a read operation;

FIG. 3 illustrates the surface scattering effects inherent in a magnetic orientation forcing electrons towards the substrate surface;

FIGS. 4a-f illustrate embodiments for producing an array of ferromagnetic gate memory cells; and

FIG. 5 illustrates a block diagram of a integrated circuit using the present invention.

For a more complete understanding of the present invention, reference is now made to FIGS. 1-5 of the drawings, like numerals being used for like and corresponding parts of the various drawings.

FIG. 1 illustrates a perspective view of a memory cell having a ferromagnetic gate. The ferromagnetic gate memory cell (hereinafter "memory cell") 10 comprises a gate 12 formed from an electrically conducting ferromagnetic material. A thin insulating layer 14 separates the gate 12 from the substrate 16. Two N+regions serve as the source 18 and drain 20 of the memory cell; for illustration purposes, the regions will be denoted as source 18 and drain 20, although the actual configuration of source and drain will depend upon the voltages applied to each region. An inversion layer or channel 22 is created in the substrate 16 when a voltage is applied to the gate 12. A current between the source 18 and drain 20 is illustrated by the reference character I.

The gate 12 is formed from a ferromagnetic material which is operable to conduct an electrical current. An example of an electrical conducting ferromagnetic material would be "permalloy", although many other similar materials are known to exist. It is important that the gate 12 be electrically conducting so that the channel 22 may be formed.

The remaining aspects of the memory cell 10 are similar to those found in a MOS transistor. For example, source and drain regions 18 and 20 may be formed by diffusing N+ regions into a P type substrate.

Information is written into or stored in the memory cell 10 by creating a local magnetic field in the gate 12. The magnetic field is created by passing a high current between the N+ source/drain regions 18 and 20. As illustrated, a high electron flux passed from source 18 to drain 20 (by convention, current I is shown from drain 20 to source 18) will create a magnetic field in gate 12 having a particular orientation. This orientation may represent a specified logic state, such as a logic "one." A high electron flux passed in the opposite direction, from "drain" 20 to "source" 18, would result in a magnetic field in gate 12 having the opposite orientation. This orientation would thus represent the opposite logic state, such as a logic "zero." Thus, digital information is stored in the memory cell by the orientation of the magnetic field produced in the gate 12. This produces the technical advantage that the information stored in the memory cell 10 will be nonvolatile, since the magnetic field of the gate 12 is not dependent upon the power supply.

FIGS. 2a-b illustrate the memory cell of FIG. 1 during a "read" operation. In FIG. 2a, a magnetic orientation on the gate 12 is such that the front of the ferromagnetic gate 12 has a north pole, and the back of the gate 12 has a south pole, as illustrated by reference characters N and S. This north/south orientation would be created by a large electron flux passing from drain 20 to source 18.

To read information from the memory cell 10, a small electron flux creating a current "i" is passed from source 18 to drain 20 as a result of a voltage between the source 18 and a drain 20. With the magnetic orientation illustrated in FIG. 2a, electrons passing from source 18 to drain 20 will be pulled upwards towards the surface of the substrate 16 just below the insulating layer 14. Hence, the electron flux from source 18 to drain 20 will flow substantially along the substrate surface.

As will be discussed in more detail in connection with FIG. 3, the mobility of the electrons travelling at the surface of the substrate 16 will be impeded, due to surface scattering mechanisms. Consequently, the flow of electrons from source 18 to drain 20 will be impeded and reduced by the magnetic field illusrated in FIG. 2a due to surface scattering effects.

FIG. 2b illustrates a perspective view of the memory cell 10 during a read operation in which the ferromagnetic gate has a magnetic field orientation opposite to that of FIG. 2a. This magnetic field orientation could be created by passing a large electron flux from source 18 to drain 20 as illustrated in FIG. 1.

Once again, a small predetermined voltage is generated between source 18 to drain 20 in order to perform a read operation. In this instance, the magnetic field present on gate 12 pushes electrons in the resultant current towards the bottom of the channel 22. The bottom of the channel 22, being in the bulk of the substrate 16, does not impede the flow of electrons to the degree exhibited by the surface of the substrate 16. Hence, in response to an applied voltage, a larger current will pass between source 18 and drain 20 with the gate 12 having the magnetic orientations specified in FIG. 2b than with the gate 12 having the magnetic orientations specified in FIG. 2a. This difference in current level may thus be detected to read either a logic "one" or "zero" stored in the cell 10. Suitable current level detection circuitry may be coupled to the cell to detect the electron mobility.

FIG. 3 illustrates a mechanism by which the mobility of the electrons in FIG. 2a is reduced. FIG. 3 illustrates a cross section of the gate 12, insulating layer 14, and substrate 16. The surface 24 of the substrate 16 will be irregular, impeding movement of the electrons 25 at the surface 24 as illustrated by impeded electrons 25a. Also, impurities 26, which may be added to the area just below the surface 24, will further deflect the electrons 25 away from a straight path between source 18 and drain 20. By roughening the surface 24, by means such as laser treatment, crevices 28 may be formed which will further immobilize electrons 25b near the surface 24. While some irregularities may exist at the bottom of the channel 22, their number will not be nearly as great in magnitude, and thus their effect on the magnitude of the current will be much less.

Studies have shown that a ferromagnetic field at the surface may act vary the magnitude of the current by as much as 1%. By further reducing the mobility of the electrons by controlled fracturing of the surface and adding impurities, the differential between a current travelling at the surface 24 of the substrate 16 and a current travelling at the bottom of the channel 22 should be even greater than 1%. Other techniques are available for reducing mobility of electrons at the surface, such as providing an amorphous layer at the surface 24. Alternatively, the substrate could be made of a substance having much higher mobility, such as InSb, which will proportionally affect the difference in mobility and facilitate reading of the information.

FIGS. 4a-d illustrate various embodiments of an array 29 of memory cells 10. FIG. 4a illustrates a top view of an array of memory cells 10, comprising strips of ferromagnetic gate material 32 in parallel on top of the insulating layer 14. Memory cells 10 are shown in an embodiment where N+ channels 30 are shared between cells 10 to increase cell density. In FIGS. 4a-b, N+ channels 30, which form the source and drain regions 18 and 20, lie beneath the insulating layer 14. Strips of the ferromagnetic gate material 32 are positioned normal to the N+ channels 30. Regions of the insulating layer 14 between the ferromagnetic strips 32 form a channel stop. Portions of the strips of ferromagnetic gate material 32 lying between the N+ channels 30 comprise the gates 12 of the individual memory cells 10. The N+ channels 30 are shared between memory cells 10, i.e., the source of one memory cell 10 is the drain of the adjacent memory cell, in order to increase the density of the memory cell 10.

To address a memory cell 10, N+ channels 30 on one side of the memory cell 10 are held to a high voltage, while N+ channels 30 to the other side of the memory cell 10 would be held to a low voltage, normally ground. A voltage applied to the strip of ferromagnetic gate material 32 will create channels beneath the strips, through which current can flow. Thus, a potential would exist between two of the N+ channels 30 resulting in a current being passed between them. A write operation would vary the direction of the current flow depending on whether a "1.revreaction. or a "0" was being written to the memory cell. Normally, the current will always be passed in the same direction during a read operation. The magnitude of the voltage applied to the N+ channels 30 would also depend upon whether a read or a write operation was being performed, since a write operation passes a high current and a read operation passes a low current.

FIG. 4c illustrates an alternative embodiment for the memory array 29. In this embodiment, a thick oxide layer 34 is formed above the N+ channels 30 on top of the insulating layer 14. This embodiment substantially reduces the capacitance of the ferromagnetic gate strips 32.

Another embodiment of the present invention is illustrated in FIG. 4d, wherein a channel stop region 36 is placed between every other N+ channel 30. In this embodiment, N+ channels 30 are not shared between memory cells. Thus, to address a particularly memory cell 10, a high voltage is applied to one N+ channel of the memory cell and a low voltage is applied to the other N+ channel of the memory cell.

Another embodiment of the present invention is illustrated in FIGS. 4e-f, wherein an electrically insulating layer 37 is formed to encase the first layer of ferromagnetic material 32 and a second layer of ferromagnetic material 37a is formed into parallel strips positioned to cover the space between the strips of the first layer of ferromagnetic material 32. In this embodiment, only the N+ channel regions 30 are inactive memory locations, thus providing the highest memory density possible.

FIG. 5 illustrates a block diagram of a memory architecture particularly suitable for use with the present memory cells. The memory array 29 is addressed by an address buffer and decode section 38 which receives address inputs 40 from outside circuitry. The address buffer and decode section 38 drives the N+ channels 30 and ferromagnetic gate material 32 to address particular cells as described in connection with FIGS. 4a-d. Data inputs 42 are received by the address buffer and decode section 38 to determine the direction of the current for each of the memory cells addressed. A read/write control input 44 is connected to the address buffer and decode section 38 indicating whether the operation is to be a read operation or a write operation.

During a read operation, the output of the memory cells address is received by a differential current sense amplifier section 46 which compares the current output from the memory cell to a reference current to determine whether the stored information is a "1" or a "0". The data from the differential current sense amplifier 46 is sent to an output buffer 48 which is connected to data outputs 50 for interaction with outside circuitry.

Since the geometries of the memory cells can be made as small as photolithography techniques will allow, the memory array 29 may have a very high density. Using a two micron spacing between N+ channels 30 and two micron strip widths for both ferromagnetic gate regions 32 and 37 illustrated in FIG. 4e, approximately 1.6 million memory cells may be provided on 20,000 sq. mils. Allowing another 20,000 sq. mils for the remaining circuitry, a 1.6 megabit memory chip may be provided for on less than a 40,000 sq. mil integrated circuit.

Since the information is stored by the magnetic field, the data stored in memory of the present invention should not deteriorate for many years, so long as another magnetic field does not influence the ferromagnetic gates. This provides the technical advantage of a longer storage period than existing technologies.

This invention also provides an advantage that relatively low currents can be used to read and write to the memory cells, as opposed to magnetic bubble memories in which a relatively high current is necessary.

Although a preferred embodiment of the present invention has been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.