CA2175044A1

CA2175044A1 - Memory material and method for its manufacture

Info

Publication number: CA2175044A1
Application number: CA002175044A
Authority: CA
Inventors: Shimon Gendlin
Original assignee: Individual
Current assignee: Kappa Numerics Inc
Priority date: 1993-10-29
Filing date: 1993-10-29
Publication date: 1995-05-04

Abstract

A composition of materials having ferromagnetic and piezoelectric properties is disclosed. In the preferred embodiment, the composition of materials comprises a first layer (100) of Pb(1-x-y)CdxFey and a second layer (110) of Cr(1-z-w)ZnzTew where x, y, z and w are values within the ranges of 0.38 < x < 0.042, 0.08 < y < 0.094, 0.38 < z < 0.41, 0.28 < w < 0.31 and 0.25 < (1-z-w) < 0.32. Additionally, each of the layers (100, 110) contain the elements of Bi, O and S.

Description

WO 95111799 21 7 5 ~ ~ ~ PCTIUS93110403 MEMORY MATERIAL AND METHOD
FOR ITS MANUFACTURE
~ACKGROU~ND OF THE INV~NTION
Computer technology requires memories having large storage capacity and high speed. Typically, in a modern computer, a semiconductor memory is employed as high-speed primary memory and magnetic disks are used for a large volume secondary memory.
Prior to the development of semiconductor memories, high-speed primary memory was implemented using a magnetic core memory. A magnetic core memory comprises a matrix of ring ~ ped feLL I_ic cores.

Each memory cell of the magnetic core memory includes 20 a fe~LLI ~I~etic core having two or more wires passing through the center of the core and a sensing coil installed around the core.

When a current I is applied to a wire that 25 passes through the core, a magnetic field, having a magnetic field strength H which i8 a function of the current I is generated . The magnetic f ield produced by the current I causes a p~ nPnt magnetization of the core. The magnetization is measured by the magnetic induction B. The relationship between the magnetic induction B and the f ield s. LL ~ y ~1~ H ( i . e . a 35 plot of B versus H) is known as a magnetizatiOn curve WO 95/11799 ~ PCT/I~S93/10403 0 ~1 7~o4~i or BH loop. In the core memory, the magnetization curve i5 substantially square.
The magnetic induction B in the core has two states, B, and ~Br~ that cuLL~r~und to the opposite directions of the magnetic f ield . Accordingly, each core can store a bit of binary data by associating one lO state with a "l" and the other state with a "O", for example, +B, may be associated with a binary "l" and -B, with a binary "O".
The binary data is written into a core 15 memory cell by applying appropriate ~iULL~l~LS to the wires. If the total current passing through the core is greater than a critical current Ic, the magnetic induction of the core changes from -B to +B.
~ r Similarly, if the current is less than -I~, the magnetic induction switches from +B, to -B,. In an array of magnetic cores, switching is performed when 25 polarities of currents on two or more wires coincide.
Thus, if the magnetic induction of a given core is initially -B, which COLLt:~OIIdS to a binary "O", to change the induction to +B" i . e. to store a binary 30 "l", a current I>IC/2 is applied to each of the two wires, so that the total current pas5ing through the core is greater than +Ic.
The data stored in the core is retrieved by sensing the voltage across the coil induced by ~ Wo 95/11799 21 7 ~ 0 4 ~ PCTIUS93/10403 switching between the two magnetic states described above. The polarity of the induced voltage indicates the magnetic state of the core prior to switching.
Although the magnetic core memory described above is random accessible and non-volatlle, such memory is large, cv~ a large amount of power, lO operates at a slow speed and can not be manuf actured to have a high storage density. To VV~L~ these problems, magnetic thin f ilm memory devices have been developed. A magnetic thin f ilm memory consists of a 15 5trip of ferromagnetic thin film, two or more wires formed on the film for storing data and a coil around the film for retrieving data.

In the thin f ilm memory, the magnetic moment M of the film L~:~L~Ge~ S the stored information. The magnetic moment M is oriented primarily in the plane of the film, and has two discrete orientations or states, namely M and -M, that represent binary "l" and "O". To store a bit of binary data, currents are applied to the wires formed on the thin film. These ~:Ur Lel~ ; induce a magnetic field that is sufficient 30 for changing the direction of the magnetic moment M.
The stored inf ormation is retrieved by applying currents to the wires and measuring the induced voltage in the coil. As in` magnetic core memory, the 35 currents are typically selected such that a single current has insufficient amplitude to reverse the WO 9~111799 PCr/US93/10403 ~1~5~
magnetic moment of the f ilm so that at least two coincident currents are required f or storing data .
There are significant drawbacks associated with magnetic thin film memory technology. First, thin film devices have an open magnetic flux structure and therefore the BH loop is smeared by a self-lO demagnetizing effect. To reduce this effect, the filmi8 typically fabricated as a rectangle which length is much greater than its width. Since the induced voltage in the coil around the film is proportional to 15 the cross-sectional area of the film, reducing the width of the film also reduces the induced voltage.
As a result, the readout signal is easily affec'ced by noise .
Second, in the existing magnetic films, the magnetic moment has a preferred in-plane direction.
Thus the device is complicated by the necessity of 5 applying currents of differeDt amplitude for storing and retrieving data in the selected orientations. In addition, the thin film devices are not sufficiently small to achieve high storage densities.

In compari60n to magnetic core and thin film memories, 6emiconductor memory is faster, Cu~ s less power, and can have higher storage densities.
Typical sPmi con~ ctor memories include Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), and Read Only Memory (ROM).
_ .. . . . ..... .

~ Wo 95111799 2 ~ 7 a ~ 1~ PCTIUS93/10403 5 -- ~

DRAM of f ers relatiYely high speed, high storage density, low power ~:vl~S,u~ ion, and is readable and writable. However, both DRAM and SRAM
are volatile, that is, they lose the stored information when the power i6 turned off. In addition, DRAM requires a constant refresh of the lO stored data which necessitates complex circuitry.
While SRAM does not require a refresh, it has a higher power c V~ l , tion and lower storage density than DRAM.
ROM is non-volatile but the information 15 stored in a ROM cannot be updated , i . e ., data cannot be easily written into a ROM.
In a typical di~k storage system, ferromagnetic material having a substantially square BH loop is coated on a disk. A magnetic head read is provided f or reading and writing inf ormation on the disk as the disk rotates past the head . The di sk is divided into circular tracks. Each track is further divided into small regions in which a magnetic moment has two states that represent binary values. An external magnetic f ield introduced by the read/write 30 head changes the magnetic moment of each small region 80 as to store a binary value in the region. Thus, to write data, the magnetic head magnetizes an adjacent small region of the rotating disk material. The 35 stored data is retrieved in the form of a voltage Wo 95/11799 PCT/US93/10403 ~17~0~ 6-induced in the head by the magnetic moment of the small region as it moves past the head.
Magnetic disk storage sy5tems can store high volumes of data, e.g., 500 Megabytes or more. The magnetic disk storage systems, however, are not random accessible, operate at slow speed due to the 10 requirement of mechanical ~ L, and require complex mechanical and electronic assemblies.
As will be apparent, none of the above-described memory technologies provides all the 15 featureS that are desirable in a memory storage system. Thus, there is a present need to develop a non-volatile, high speed, high capacity, random accessible, static, and updatable storage system.
The parent application serial No.
07/88g,025, entitled "Memory Material and Method of Its Manuf acture", discloses a new composition of materials and a non-volatile, high speed, high capacity, and random accessible memory built on the basis of this composition of materials.
Specifically, in the preferred: '_'; L, 30 the composition of materials disclosed in the parent application comprises sequentially formed layers of Pbo~oCdo1osiolol SeO90S0lO and FeO7~CrOl~. In addition, these layers contain one or more of the following elements:
35 Bi, Ag, O, and N. This composition of materials displays ferromagnetic, piezoelectric and electro-optical properties.
The memory disclosed in the parent application comprises two sets Or parallel address lines di6posed orthogonally on the opposite sides of a planar silicon substrate. The layers o~ the lO composition o~ materials, described above, are disposed on both sides of the substrate above the address lines such that the FeCr layers are the outermost, and each FeCr layer is connected to an 15 electrode. In this device, each memory cell is def ined by each crossing point of the address lines on the opposite sides of the substrate. Typically, in this memory device, each address line is approximately

2,um wide and l,llm thick, and the spacing between the adjacent lines is approximately 9~m. Thus, each memory cell occupies an approximately ll ,um square.
Each cell of this memory is capable of storing two ;n~lPrPnr~Pnt bits of information. Accordingly, the storage density of the memory device disclosed in the parent application is typically about l. 65 mega bits 30 per Bquare centimeter.
- An object of thi5 invention is to realize a new composition of materials for b~ i;n~J a memory with a storage density higher than provided by the 35 memory disclosed in parent application.

Wo 95/ll799 ~17 5 0 q ~ PCT/US93/10~03 ~

S~IMMARY OE THF INVF~N~QN
According to the present invention, a new compo6ition of materials, that exhibits f erromagnetic and piezoelectric properties and can be employed as a memory material is invented. This invention also relates to a non-volatile random accessible memory l0 built on the basis of the invented composition of materials . A method f or storing and retrieving two independent bits of information in a single cell of the memory of the present invention is also disclosed.
The composition of materials of the present invention comprises two layers of material . A f irst layer inrl~ the elements of Pb, Cd, and Fe. The ratio between Pb, Cd, and Fe is (l-x-y) :x:y, where 0 S

x S l, 0 S y S l, and 0 S (l-x-y) S l. A second layer is formed on the first layer. The second layer includes the elements of Cr, Zn, and an element 25 selected from the group consisting of Te and Tl (denoted by "R"). The ratio between Cr, Zn and R is (l-z-w) :z:w, where 0 S Z S l, 0 S w S l, and 0 S (l-z-w) S 1.
In the preferred ~ o~lir-~lt, the ratio values set forth above are within the following ranges: 0.038 S ~ ~ 0.042, 0.08 S y S 0.094, 0.28 S Z

S 0.41, 0.28 S w S 0.31 and 0.25 S (l-z-w) < 0.32; and both layers also contain the elements of S, O, and Bi.

~W095/11799 ~ a44 PCT/USs3/10403 _ g _ In a memory device of the present invention, two sets of parallel address lines are disposed orthogonally on the opposite sides of a planar substrate. The layers of the novel composition of materials, as described above, are disposed on both sides of the substrate above the address lines such lO that the second layers are the outermost, and each second layer is connected to an electrode. Each individual cell of the memory is def ined by an intersection of the address lines of the two sets.
15 The storage density of the memory is approximately eight mega bits per l. 2 cm x l. 2 cm chip.
In the memory of this invention, two in~pc.nr1-.nt bits of information can be magnetically stored in a single memory cell. This information is retrieved as a piezoelectric voltage bQtween the electrodes generated in response to current pulses 25 applied to the two address lines.
More ~pP~!;f;C~lly, to store and retrieve a f irst bit of inf ormation in a memory cell, two ~.y~ L~ ized current pulses having the same amplitude 30 and polarity are applied to two orthogonal address - lines of the memory cell. The second bit of the same cell is stored and retrieved by applying two synchronized pulses of the same amplitude but opposite 35 polarity to the same two address lines. The current pulses employed for storing binary information are WO 95/11799 ~ 1 ~ 5 0 4 ~ PCT/U593/10403 such that the amplitude of a single pulse is not suf f icient to alter the state of the stored information and two concurrent pulses are n~c~:s~ry $or storing data . To retrieve the stored inf ormation non-destructively, the combined current pulses employed f or retrieving the stored inf ormation should lO be insufficient to alter the stored information.
The memory cell of the present invention is non-volatile, random accessible, static, operates at high speed, requires low power, is readable and 15 writable, and can be made in high density arrays.

-- .

Wo 95111799 PCTIUS93110403 ~17~Q'~'l BRIEF DESCRIPTION OF THE DRAWINGS
These and other objects, features and advantages of the invention are more fully set forth in the accompanying Detailed Description in which:
Fig. l shows the cross-section of a preferred l~mho~; ~ of the composition of materials of the present invention;
Fig. 2 illustrates the magnetization curve (BH loop) of a conventional ferromagnetic material;
Fig. 3 shows a substantially square BH loop Of the composition of materials of the present invention;
Figs. 4 (a)-(h~ illustrate the process of generating piezoelectric voltage in the composition of materials;
Figs. 5 (a) and (b) are the cross section and top-view of the preferred ' ~;r-nt of the memory device of the invention;
Fig. 6 depicts a top view of address lines f ormed on a substrate .
Fig. 7 (a) and (b) illustrates the cross section of a wafer, before and after polishing.
Figs. 8 (a) and (b) show plots of voltage waveforms and current amplitude for the electrochemical process.
Figs. 9 (a) and (b) illustrate selecting bits of inf orm~tion in a memory cell;

~17~44 Wo 95111799 PCT~593/10403 Figs. lO (a) and (b) illustrate storing a f irst bit of inf ormation;
Figs. 11 (a) and (b) illustrate retrieving the first bit of information;
Figs. 12 (a) and (b) illustrate storing a second bit of inf ormation;

Fig. 13 illustrates the current pulses used for retrieving the second bit of information, and Fig. 14 symbolically summarizes the storing and retrieving information from the memory device.

WO 95/11799 ~ PCT/I~S93110~03 DETAILED DESCRIPTIOI~
The present invention relates to a new composition of materials having ferromagnetic and piezoelectric properties, and to a random accessible non-volatile memory utilizing the new composition of materials. The memory of this invention is also o capable of storing two independent bits of information in each storage cell of the memory in a similar f ashion to the memory disclosed in the parent application .
In accordance with the invention, the composition of materials comprises two layers of material. A first layer includes the elements of Pb, Cd, and Fe. The weight ratio between Pb, Cd, and Fe is (l-x-y) :x:y, where 0 x S 1, 0 s y s 1, and 0 5 ( l-x-y) S 1. A second layer is f ormed on the f irst layer. The second layer includes the elements of Cr, Zn, and an element selected from the group consisting of Te and Tl (denoted by "R" ) . The weight ratio between Cr, Zn and R is (l-z-w) :z:w, where 0 S z S 1, 0 S w S 1, and 0 S (l-z-w) S 1. Preferably, the layers of the composition of materials also contain the elements of S, 0, and Bi.
Hereinafter, the followiny notation will be used: the first layer will be referred to as Pb(l~
y~CdzFey (or simply as "PbCdFe") and a second layer will be referred to as Cr~lwz)ZnwTez (or simply as "CrZnTe"), WO 95/11799 PCr/US93/10403 ~17~

and the parameters x, y, z, and w are as described above .
Note that the boundary between the two layers are not precisely def ined and the elements of one layer are likely to penetrate another one. Due to the fine dimensions of the structure, the extent of 10 intPr~ ; n~ of the elements of the layers has not been previou51y det~rminP~l. Accordingly, although this discussion refers to two distinct layers, it ~hould be emphasized that a composition of materials where the 15 elements of the layers are substantially int~rm; ~ is within the scope of this invention.
Preferably, in the first layer, the values of x and y are within the ranges of o. 038 ~ x < 0 . 042 and 0 . 08 S y < 0 . 094 . In the second layer, the values of w, z, and (1-w-z) are preferably within thc ranges of 0.34 S w S 0.45, 0.25 S z S 0.34 and 0.25 S (1-w-z) S 0.32. More preferably, the values of w, z, and (1-w-z) in the second layer are within the ranges of 0 . 38 w S 0.41, 0.28 S z S 0.31 and 0.25 S (1-w-z) S 0.32.
As depicted in Fig . 1, in the pref erred 30 embodiment, the composition of materials comprises a first layer 100 (PbO~7CdO04Feoo~) and a second layer 110 (CrO3ZnO4Te03). Additionally, both layers 100 and 110 are saturated with 5. The proportion of S in the 35 first layer is about 4% compared to 96% of PbCdFe and in the second layer the proportion of S is also 4% as .. . . . . . .. _ .. _ . . . _ . . . . . . , _ _ _ . _ .

Wo 95/11799 PCr/US93/10403 - 15 - ~17~Q~
compared to 96% of Cr5nTe. In addition, both the f ir6t and the second layers are saturated with O .
Both the f irst and second layers also contain si, which is provided to the layers in the proper amount according to the process described bQlow.
In the f irst layer, S bonds with at least lO portions of Pb and Fe to form PbS and FeS, and O
combines with at least a portion of Fe to form Fe2O3.
In the second layer, S bonds with Zn to f orm ZnS . In the preferred composition of materials, both the first 15 and second layers are saturated with S and O, and both layers display high resistivity.
Bi is present in both layers of the preferred composition of materials to improve the sensitivity of the composition to applied fields such as a magnetic field. The amount of Bi incuL~uLc~ed in the layers can be adjusted according to the desired sensitivity. The sensitivity increases by adding Bi to the layers. However, an excess amount of Bi makes the composition undesirable sensitive and susceptible to environmental conditions such as mechanical 30 vibrations or temperature variations. Thus the amount of Bi in the composition is typically detPrm; nPd by the reguired sensitivity and signal-to-noise ratio.
As indicated, the composition of materials 35 of the present invention has f~L ~ letic and piezoelectric properties . Due to the f t~ lletic Wo 95/11799 PCT/US93/10403 ~7~

properties, the composition has two stable magnetic states. The preferred composition of materials is very sensitive to external fields, such as a magnetic f ield generated by an electrical current. For example, if a conducting wire is formed immediately below the first layer of the composition, the portion lO of the composition of material in the vicinity of the wire can be switched between its two stable magnetic states by applying an electrical current on the order of only nano-amperes. In addition, the switching 15 between the two stable magnetic states generates a detectable pi''70F'~ eC-tl-ic voltage vertically acro6s the layers in the order of ten micro-volts. As will be apparent, these properties are essential for realizing a high capacity, non-volatile, and random accessible memory of the present invention.
Physical properties of the invented compo6ition of materials are described below.
Understanding of these properties will help to understand the operation of the memory device of the parent invention.

3 0 By way of ba. l~L vu~.~l, a f ~L L ~.letic material exhibits a p~rrqnqnt magnetic field in the absence of an external magnetic f ield. Such materials can be described in terms of a large number of small magnets known as magnetic dipoles. An external magnetic field, applied to a feLL~ gnq~ic material WO 95/11799 ~ 1 7 ~ PCT/US93/10403 aligns the magnetic dipoles within the material in the direction of the applied field, so that the totAl magnetic f ield within the material is the sum of the external field and the field generated by the aligned magnetic dipoles . When the inf luence of an external magnetic f ield is discontinued, the orientation of 10 magnetic dipoles does not change, resulting in a constant magnetic f ield in the material . Magnetic information storage is based on this property of f erromagnetic materials .
Fig. 2 shows an exemplary maqnetization curve of a typical f~LL Atic material. The magnetization curve is also referred to as a BH loop.
The y axis in this f igure represents r- jnet; ,A, induction B, which is an overall magnetic f ield in the material, and the x axis represents the magnetic f ield strength H of the external magnetic field Thus, the BH loop shows the change in the magnetic induction B
with ~AhAn~;ng magnetic field strength H.
Let us consider the BH loop of Fig. 2 in further detail. A'AA~rlnj that initially the 30 orientations of magnetic dipoles of the feLL~ t.ic material are evenly distributed in all directions, the total value of B in the absence of the external field is zero (point "a" on the curve). When an external 35 magnetic field is applied to the f~LL, gnAtic material, the ,value of B gradually increases as H

217~4 '~

increases until it reaches a point where magnetic induction B begins to saturate (point "b" on the curve). In other words, when H reaches a certain value, B remains substantially at Bo even if H is being increased. If, after saturation, the external magnetic f ield is decrea6ed to H=O, magnetic f ield lO induction B does not return to the point "a" (B=O).
Instead, the value of B remains approximately at B=Bo t point " c" on the curve) .
At point "c" the direction of the external 15 magnetic field H is I~V~:L~ed. At approximately H=-Hc~
the external f ield H changes the polarity of the f ield B, and, at point "e", the field saturates at the opposite polarity B=-Bo-Increasing the field strength H causes B to change from point "e" on the curve to point "f " and then "b", as illustrated in Fig. 2.

Fig. 3 illustrates the BH loop of the composition of materials of the present invention. As in Fig . 2, the x-axis indicates the external f ield strength H and the y axis indicates the magnetic induction B. It is important to note that for the invented composition of the materials, the shape of the BH loop is substantially square with angle c~
between the y-axis and the BH loop at B=0 being approximately greater than 1 but less than 2.

WO 95/11799 2 ~L 7 ~ 0 ~5 ~ PCT/US93/1o4o3 Because the magnetization curve i8 substantially square, the magnetic induction B is almost invariably at one of the two discrete, stable states, namely ~Bo and -Bo~ Accordingly, the novel composition of materiels is suitable for storing binary information.
As indicated, the composition of materials Of the present invention also has piezoelectric properties. In general, a piezoelectric voltage is generated as a result of mechanical pressure being applied to a piezoelectric material. In the present invention, a AnirAl ~re6~UL~:, applied to the composition of materials outwardly from within the composition so as to stretch the composition in a direction substantially perpendicular to the plane of the layers, results in a piezoelectric voltage generated across the layers. In the memory device of the present invention, the mechanical pressure is produced by a changing in the magnetic state of the composition of materials as ll;Rrllc~d in further detail below.
To aid in understanding the p;~Y~lectrical properties of the composition of materials, an illustrative structure is shown in Fig. 4 ta). The properties of the composition of materials are explained in conjunction with Figs . 4 (b) - (h) .
Fig. 4 (a) illustrates a structure l90 comprising two layers of the composition of materials ~175~4~

of the present invention. Specifically, the structure comprises a first PbCdFe layer 220, a first CrZnTe layer 230, a second PbCdFe layer 210, a second CrZnTe layer 200. In addition, all the layers also contain S, o, and 8i. A wire 260 runs parallel to the layer through the middle of the structure ( i . e . between two lO compositions of materials).
As shown in Fig. 4 (b), an electrical current applied to wire 260 in a direction pointing down onto the page generates a substantially circular 15 magnetic f ield around the wire ( indicated by a circle Br) in the clockwise direction ( indicated by the arrow). Arrows 270 illu5trate the directions of the magnetic dipoles in the layers under the influence of this external field. At this point, the structure can be viewed as divided into two sections 275, 280 that are symmetric about a vertical axis 26~ perp~nr~ r 25 to wire 260 as shown in Fig. 4 (b). I'he dipole arrangement in the section5 275, 280 is equivalent to two magnets of the same strength having north and south poles as indicated by arrows 282 and 284 in Fig.

30 4 (c). The length of each arrow represents the amplitude of the magnetic induction B of the in~ magnet. Due to the attraction between the South pole S and the North pole N of each magnet, 35 the structure is mechanically compressed in the WO 9Sll 1799 PCT/lJS93/10~03 ~17~4 direction perpendicular to the layers of the structure .
The BH loop of the magnetic induction Br is shown in Fig. 4 (d). As described previously the BH
loop is substantially 6quare, exhibiting two discrete, stable magnetic states, +Bo and -Bo.
Additionally, the magnetic field has a critical f ield strength Hc which is def ined as the amplitude of the magnetic f ield strength that causes switching between +Bo and -Bo~ Consequently, if H is 15 greater than Elc, the magnetic induction Br has a value +Bo~ If H is less than -H~, Br has a value -Bo~
Assume that initially under the inf luence of the applied external field, the magnetic state is described by point "a" on the curve of Fig. 4 (d), where the induction is +Bo~ To change the magnetic state of the composition of materials from +Bo to -B
25 the current passing through wire 260 must be reduced to reduce the magnetic f ield strength H . When the current is zero the magnetic strength H is also zero (point "b" on the BH loop). As noted, due to its 30 f~:lL~ ~n,~tic properties, even without the external field, the magnetic state of the storage media remains at Bo~ l.e., the information represented by the magnetic induction Bo is retained.

Wo 9~/11799 PCT/US93/10403 ~17~4~

When the direct1on of the current is reversed, the magnetic field strength continues to decrease. At point "c", the magnetic induction B
reaches the value B~, which is less than Bo~ At this point, the dipole moment has been reduced, as shown in Fig. 4 (e), because the dipoles began to realign in lO the opposite direction. Accordingly, the m~ :~nic~l IOL eSDUL t: on the layers due to the attraction of the FeCr layers 200, 250 has been reduced. The change in the pLe:S~UL~ on the layers causes piezoelectric 15 voltage to be generated perpendicularly across the layers. At point "d", where H=-HC and the magnetic induction B is zero, the plesDuL~ applied to the layers is minimal because the dipoles are aligned in different directions. At this point, the induced piezoelectric voltage reaches its maximum value, due to the maximum change in the ~Lt:DDuL~ on the layers.
As }/ continues to decrease below -Elc, the magnetic state switches from point "d" to the point "e" and then to point "f" where it reaches the second stable state B=-Bo. Fig. 4 (f) shows that at point "~"
the pole8 of the magnet8 have been Lesvt:LDed. Thus, at point "f", the mechanical ~L~:~DULe on the layer returns to its initial value, reducing the piezoelectric voltage. Increasing the LeveLDed current further (from point "f" to point "g") does not Wo95~11799 2~ PCT/US93/10403 increase the magnitude of the dipole moment and therefore does not increase mechanical yr~s~ul~ on the 5 layers.
In Fig. 4 (g), the piezoelectric voltage generated in response to a current pulse is illustrated in the time domain. As illustrated, a lO piezoelectric voltage pulse is delayed from the time of the application of the current pulse. Note that the amplitude of the current pulse applied to the wire (-I) should be sufficient to switch Bo to -Bo~
l5 similarly, switching from the magnetic state -Bo to +Bo also generates a piezoelectric voltage pulse.
As indicated in Fig. 4 (d), a current that generates a f ield having an amplitude greater than Hc is required for switching between the magnetic states.
If, however , a current of a lesser amplitude causes B
to assume a value indicated by point "c" in Fig. 4 25 (e). At this point, the magnetic state is unstable.
In such a case, the magnetic induction B tends to oscillate between the values of Bo (point "b") and B~
(point "c"). The amplitude V2 of this piezoelectric 30 voltage pulse is smaller than the amplitude of the pulse generated as a result of switching f rom +Bo to Bo .
Fig. 4 (h) shows a current pulse I which 35 causes B to assume the value B~. The piezoelectric Wo 95~11799 PCrrUS93110403 ~
~7~ 24-voltage pulse, generated in response to this current is 6hown in the lower portion of the f igure . The shaded area ref lects the oscillation between the two states (B~ and I Bo), as it would be observed on an osr~ 11 n~cope As de6cribed subsequently, the piezoelectric voltage generated in response to the lO current that disturbs but does not switch the magnetic states, can be employed for reading magnetically stored information.
Figs. 5 (a) and (b) illustrate the cross-15 section (not to scale) and the top-view of the preferred A~l~o~ t. of a portion of the memory device 290 of the present invention. The memory device comprises a silicon planar substrate 500, first address lines 510 formed on one surface of the substrate, and second addres5 lines 520 orthogonal to the first lines formed on the opposite surface of the substrate. A first 530 and a second 540 of the composition of materials of the present invention are disposed on the opposite side5 of the substrate over the address lines. Electrodes 550, 560 are connected 30 to the compositions of materials 530, 540 respectively. Note that, in the composition of materials 530 and 540, the portion located between the address lines comprise both layer lO0 of PbCdFe and 35 layer llO of CrZnTe. However, only the CrZnTe layer lO0 is present; above the address lines.

~Wo95/11799 r~ Y.sll~
217~04~

The f irst and second address lines are silver strips approximately l~m wide and approximately l~m thick. The spacing between the adjacent address lines is approximately 4 . 75~Lm. Each composition of materials 530 and 540 comprises the first layer and the second layer of the present invention described lO above. The first and second layers of 530 and 540 are sequentially formed on the Si substrate with the two second layers being the outermost on both sides of the substrate . Pref erably, each layer is approximately 0.7 ~m thick so that each combination of materials 530 and 540 is approximately l . 4 ,(Lm thick. The substrate is approximately 140 ~m thick and the electrodes are about l~m thick 6ilver layers formed on the second layers of 530 and 540.
In the memory device of the present invention, the intersection and its neighboring area of the first and second address lines, formed on the opposite sides of the substrate, constitute a single storage cell. As in the memory disclosed in the parent application, each storage cell of the memory of 30 the present invention is also capable of storing two independent bits of binary information. Given the tl;r -ionC: of the address lines and the spacing between them, the storage density of the memory of the 35 present invention is approximately eight mega bits per l . 2 cm x l . 2 cm chip .

_ WO 95111799 PCTNS93/10~03 217a~ 26-Next the fabrication of the combination of materials and memory device of the present invention are described. Preferably, the fabrication takes place in a class lOo clean room environment.
The f abrication of the memory device begins with building the first and second address lines onto 10 a substrate. Fir6t, approximately 1,um thick silver layers are deposited onto the opposite surfaces of a silicon planar substrate which is approximately 150 ~m thick and three inches in ~ tPr. Alternatively, 15 other materials having low electrical resistivity, such as gold, aluminum or platinum, can be used instead of silver to form the address lines. The deposition of the silver layer is conducted by a conventional technique such a6 thermal evaporation, e-beam evaporation, or sputtering. The deposited silver layers are then photolithographically patterned and etched to form a series of metal strips, each having a width of approximately 1,um and the series of strips on one side of the substrate being orthogonal to the strips on the opposite side so that the strips on both 30 6ides of the substrate form a cross-barred ~U~;~ule.
The top view of the substrate after this step is depicted in Fig . 6 where the stripes are f ormed in the centered square portion of a circular silicon 35 substrate 600 which is approximately three inches in dianeter. Preferably, after this step, the silver Wo 9S111799 ~ 1 7 ~ PCr/Uss31l0403 layers outside the square that contains the strips remain unetched so that the end5 of all the strips are connected in this peripheral region of the wafer.
For convenience, the term "wafer" is used hereinafter to refer to the substrate together with various structUres built upon the substrate at various 10 steps during the fabrication process.
The next step is to deposit a thin layer of sulfur (S) onto both sides of the wafer with address lines formed on its two surfaces. Sulfur is deposited 15 by exposing the substrate to sulfur vapor in a sealed chamber. The chamber has an air-tight partition dividing the chamber into upper and lower portions.

Prior to the deposition of sulfur, the wafer is placed in the upper portion of the chamber and a desired amount of sulfur is placed in the low portion. Then, the sulfur is heated to 450 to 500C so that sulfur vapor is formed in the lower portion of the chamber.
Next, the partition is removed and sulfur vapor is introduced into the upper chamber. The wafer is then expose~ to the sulfur vapor for about one minute. As 30 a result, a layer of sulfur is formed on each surface of the wafer. Note that the exact thi- knl~cc of sulfur formed on the wafer is not important as long as the amount of ~ulfur is sufficient to saturate the 35 subsequently deposited layers.

WO95/11799 F~IIIJ~9~ C~3 ~ 1 7~ 28 -Following the deposition of sulfur, layers PbO~7CdOo4Feoo7 and CrO3ZnO4Te03 are deposited sequentially onto both sides of the wafer. Prior to the deposition, a deposition source for each layer is prepared to contain the proper amount of components of each layer accordinq to the ratios 6pecif ied above .

lo Specifically, the deposition source is prepared by f irst mixing together proper amounts of powder of each required element to form a mixture.
The typical size of particles in the powders of each 15 element is about 2~m. Following the mixing, the mixture is pressed and baked to form the desired 6hape and rli~ nq For example, for the deposition of layer PbOI7CdOo4Feoo9r powders of Pb, Cd and Fe are mixed in the proportions of 87: 4: 9 . After the powders of Pb, Cd, and Fe are well mixed, the mixture is pressed in a mold to form a 1.5 x 1.5 x 1.5 cm3 cube. Typical 25 ~ ~S~UL~ applied to form the cube is between 1. 5 and 2. 0 tons. When the ~les~uL~ is removed, the cube remains in the mold where it is heated to a t~ _LCILUl~ between 450 and 500C u5ing a microwave 30 source. The temperature reaches 450 to 500OC in about one to two seconds. Thereafter, the heat is turned off and the cube is left in the mold to cool down to room temperature in the room temperature 35 environment. In the preferred f"nhOr~ nt, the same system is employed for pressing the mixture of powders WO 95/11799 ~ l ~ 5 ~ ~ ~ PCr/USs3/10403 in a mold and for heating it to the temperature specif ied above . Such system which integrates pressing and heating are known in the art. A cube for depositing CrO3ZnO4Te03 layer is formed in the same f ashion .
To reduce the grain size of the materials in lO the cube5, the cube for PbO~7Cdo04FeoQ~ i5 then immersed in liquid oxygen and the cube for CrO3ZnQ~TeO3 is immersed in liquid nitrogen for about three hours.
Thereafter, the cubes are removed from the liquid lS nitrogen or liquid oxyyen and are placed in the room temperature environment to warm up to room t~ _L~LuL ~.

Next, each cube is employed for depositing the respective layer. The cubes are referred to as a deposition source for the corresponding layer. The Pbo 57Cdo 04Feo Q~ and CrO 3 ZnO 4Teo 3 l ayers are then 25 sequentially deposited onto both sides of the substrate. The deposition can be accomplished using well-known methods. For example, in the preferred embodiment, a conventional plasma sputtering 30 techniques is used. The sputtering is performed in a vacuum chamber utilizing Ar gas. In the vacuum chamber, the deposition source is placed at a distance above the center of the wafer. During the deposition 35 of each layer, the temperature of the wafer is not WO95/11799 PCr~S93/10403 30 _ controlled, however, it has been estimated that the temperature varies approximately between 9O and 220C.
As described previously, prior to the deposition of the PbO~~CdOO4FeOo9 and CrO3ZnO4Te03 layers, the waf er has been covered with a thin layer of lo 6ulfur- During the deposition of PbOI7CdOoAFeu~7 and CrO3ZnO4Te03 layers, sulfur diffuses into and reacts with the deposited materials. As a result, after the deposition, both PbOa7cdoo1Feo~o9 and CrO3ZnOATe03 layers are 15 saturated with sulfur, forming PbS and FeS in the f irst layer and ZnS in the second layer . The col~c~ L<Ition of sulfur in both layers has been estimated to be 4% of weight compared to about 96% of Pb, Cd, and Fe combined in the first layer and 96% of Cr, Zn, and Te combined in the second layer.
During the deposition, a thickness sensor in 25 the chamber is utilized to detect the thickness of the deposited layers. As indicated, each PbOa7CdO0AFeoo7 and CrO3ZnO4TeO3 layer should be approximately 0.7 I~m thick, forming two structures, each approximately 1.4 ,tLm 30 thick, on the opposite surfaces of the wafer with the two CrO3ZnO4Te03 layers being the uu~ ~. Fig. 7 (a) illustrates the cross-sectional view of the wafer after this step. Since the silver address lines are 1 35 ,um thick, the total thickness of the materials above WO 9~i/11799 PCT/US93/10 ~03 the surfaces of the silicon substrate is 2 . 4 am in the areas having the silver address lines and 1. ~ ,~Lm in the areas between the silver address lines.
Both sides of the resultant structure are then polished to create smooth and planar surfaces such that the total thickness of materials deposited 10 on both sides of the wafer is approximately 1. 4~m. As illustrated in Fig. 7 (b), after polishing, the deposited materials between the address lines include both PbO87CdOO4FeOQ~ and CrO3~nO~Te0Q layers and remain 15 approximately 1. 4 ,um thiclc. The deposited materials above the address line has been reduced to 0 . 4 ~m thick, and consi6t of essentially PbQ87CdO04FeOQ~ layer.

Thereafter, Bi and 0 are added to the layers by an electro~ hPm;~-~l process that employs heated water containing Bi2o3 . Specif ically, high purity water is first heated to approximately 97C in a 25 st~; nl ~s~:: steel container with a continuously stirring device on the bottom of the container. Bi~o3 powder is then aclded to the heated water. Preferably, the weight of Bi20~ powder is about 4-6% and the weight of 30 water is about 94-96%.
Prior to the electro~h~mi~l process, the address lines on both sides of the substrates should be con~ected to a single electrode. Recall that in the preferred ~mho~ , the address lines are etched WO 95/11799 PCrlUS93/10403 2 ~ 7 ~

such that the lines on each side are connected around the perimeter of the waf er . The 5ubstrate is then immersed in the water containing Bizo3 powder. The water containing Bizo3 powder is maintained at a temperature of 97C and i6 continuou61y 6tirred. A
voltage having 6pecif ic characteristic6 de6cribed 10 below i6 then applied between the container and the electrode that is connected to all the address lines.
Under the influence of the applied voltage, Bi and O
diffuse into the deposited layers. As a result, Bi 15 and o are incuL ,uu:~ ~ted into the depo6ited material6 .
The electrochemical proce66 de6cribed above takes twenty-eight day6 during which the voltages are applied in the following manner. During the initial f orty-f ive hours, high voltage pulses illustrated in Fig. 8 (a) are applied to the bath. The applied voltage waveforms of Fig. 8 (a) comprise a series of 2 voltage pul5e5 having amplitudes of two thousand volts and alternating polarities. Each pulse has a shape of a triangle and a duration of about one 6econd. The time intervals between centers of the adj acent pulses 30 is about six seconds. After the initial forty-five hours, and until the end of the proce6s, an electrical potential of sixty volts is applied to the substrates.
Every four hour5 the polarity of the sixty volt 35 potential is changed. During this proces6, the StA; nl ~ stQel container is kept at a ground WO95/11799 21~ PCTIUS93/10403 potential and the potential applied to the wafer alternates between positive and negative sixty volts.
Throughout the process the liquid in the container is maintained at 97C and is continuously stirred.
The amplitude of the current flow in the liquid is continuously monitored. Fig. 8 (b) lO illustrates the current I (in Amperes~ in the water during the entire twenty-eight days of the process f or one hundred three inch wafers processed concurrently in the electrode bath. In Fig. 8 (b), the days are 15 indicated on the horizontal axis as "t" and the current is indicated on the vertical axis as "I".
At the end of the twenty-eighth day, when the current passing the liquid is about 1.5 Amperes, the electrode is disconnected from the substrates and the substrates are removed f rom the bath . At this point, the layers are saturated with O and a sufficient amount of Bi has been incorporated in the layered structure . Note that in a dif f erent - L, ion implantation techniques can also be employed for introducing these elements into the 30 layered structure.
Subsequently, approximately l~Lm thick layers of silver are deposited onto both surfaces of the wafer to form electrodes 550 and 560 of Fig. 5 (a) and 35 (b). Thus, the structure as shown in Fig. 5 has been f ormed .

. . .

Wo 95/11799 PCTrUSs3/10403 0 2 ~ ~ 0 l~ 34 ~

The wafer is then diced into 12cm x 12cm square chips . Each surf ace of a chip contains two ol~c~n~ and forty-eight address lines. The dimensions and spacing between the address lines are discussed above. Since each intersection of the two address lines on the opposite surfaces of the chip o forms a storage cell, there are approximately four million storage cells in each chip. since two independent bits of information can be stored in each storage cell of the chip, each 12cm x 12cm chip is 15 therefore capable of storing approximately eight mega bits of inf ormation .
Thus, at the completion of this procedure, a novel two-dimensional memory array has been manuf actured, where the area in the vicinity of each intersection of the orthogonal address lines 510 and S20 is an individual cell of memory. The storage and retrieval of binary data in each storage cell is discussed in detail below.
As shown in Figs. 5 (a) and 5 (b), the set of metal strips on the lower surface of the Si 30 substrate forms a first set of addressing lines S20, (referred to as X lines), and the set of metal strips on the upper surface of the substrate forms a second set of addressing lines 510, (referred to as Y lines).
35 Nhen two electrical currents Ii and Ij are simultaneously applied to a given line Xi of the X

WO 9S/11799 ~ 1 7 ~ ~ ~ 4 PCr/US93110403 -- 35 ~

lines and a line Yj of the Y lines, respectively, a storage cell ( i, j ) at the intersection of X; and Yj is selected. By properly choosing the magnitudes and polarities of the currents I; and IJ, information can be stored or retrieved from the storage cell (i, j) .
Thus, the memory array, comprising the storage cells lO of the present invention, is random accessible.
The process of storing inf ormation in a cell will be apparent from Figs. 9 (a), 9 (b), lO (a), lO (b), ll (a), ll (b), 12 (a) and 12 (b), 13 and 14 . Figs. 9 (a) 15 and (b) illustrate top views of a single storage cell.
As illustrated in Fig. 9 (a), the orthogonal address lines 925, 945 divide the cell into four quarters 970, 975, 980 and 985. As discussed below, a first bit of information is magnetically stored in the quarters 970 and 980 and a second bit is magnetically stored in the quarters 975 and 985. For simplicity, 25 the quarters 970 and 980 where a first bit of information is stored, are collectively referred to as carrier "a" and the quarters 975 and 985 where a second bit is stored are collectively referred to as a 3o carrier "b".
- To store a bit of inf ormation in one of the carriers of the storage cell, two electrical currents having specified amplitudes and polarities are applied 35 to first and second address lines, 925 and 945, respectively. Information is retrieYed from one of Wo 95/11799 PCT/US93/10403 21~ 36 -the carriers by applying two electrical currents to the address lines and measuring a piezoelectric voltage generated between the upper and lower electrodes .
The current applied to the f irst address line is denoted by Ij, and the current applied to the lO second address line is denoted by Ij. The directions of I; and I; are indicated by the arrows at the address lines. In the preferred ~ t, the currents I;
and I~ have the same amplitude, Io~ Each current 15 generates an induced circular magnetic f ield around the cuLLe~uu.,ding address line as illustrated by the arrows 9 9 0 and 9 9 5 .

The directions of the magnetic f ields B; and Bj induced by I; and I; in each quarter is illustrated in Figs. 9 (a) and (b) . A dot (-) indicates that the field is in the "up" direction and a cross (x) 25 indicates that the f ield is in the opposite or "down"
direction .
As illustrated in Fig. 9 (a), in the quarters 985 and 975 (carrier "b"), Bj and Bj have the opposite 30 directions and thus cancel each other out. For this reason the currents illustrated in Fig. 9 (a) do not affect the information stored in carrier "b". On the other hand, in the quarters 970 and 980 (carrier "a"), 35 the fields Bi and Bj are induced in the same direction.

W0 95111799 ~ 1 7 5 ~ ~ ~ PCTnJS93~l0403 -- 3 7 ~

Accordingly, these fields enhance each other and, thus, can alter the stored information.
Thus two currents having the same polarities and amplitudes applied to the address lines, affect only the magnetic state of carrier "a" and thereby select ~his carrier. Likewise, two negative pulses - lO also select and can store data in carrier "a". Note also that the amplitudes of the currents that select carrier "a" do not have to be equal, as long as their combined effect does not change the magnetic state in 15 carrier "b".
Fig. 9 (b) illustrates the process of selecting carrier "b". Two ~ ULL~ ;, having opposite polarities, I1=+Io and Ij=-Io, are applied to the first and second address lines respectively. As illustrated using the dot and cross convention described above, in carrier "a", the f ields generated by these currents cancel each other out, without affecting the magnetic statc . In carrier "b", however, the f ields generated by these currents enhance each other so that carrier "b" is selected.
Similarly, two currents I'---Io and Ij=+I~
applied to the two address lines also select carrier "b". Thus two :ULL~11t~ with the same amplitude but the opposite polarities select carrier "b" for storing or retrieving information.

To store information, the amplitudes of the two currents combined should also be 6uf f iciently large to switch the magnetization of a carrier between the magnetic states Bo and -Bo. In addition, the ~mplitudes of the two currents combined should be sufficiently small so that a single current alone is lO unable to change the magnetic state of a carrier.
This is n~ Ary to assure that only one carrier in the memory array i8 selected by two signals applied on two address lines.
To non-destructively retrieve the stored information, the amplitudes of the two ~:uLLe11L~
combined should be small enough that the induced f ield is insuf f icient to change the magnetic state of the carrier. The combined amplitudes, however, should also be suf f icient to disturb the magnetic state of the carrier so as to generate a pie20electric voltage across the device structure. As discussed above, the direction of this piezoelectric voltage L~:pL~:se~1~s the binary data stored in the carrier.
By way of illustration, Fig. lO(a) depicts 30 the process of writing a binary "l" into carrier "a"
using eiy~1~;11L~11oLIs current pulses on the two address lines. Initially, all the cells of the array are assumed to be in the "o" state which ~ILL~ ollds to a 35 magnetic induction of -Bo~ To write a binary "l", two ~iy~ lL~llized current pulses, Ij=5.01LA and Ij=5.0uA are , . .. . ... , _ .. , ,, .. . _ _ _ , _ , ,,,, _ _, ,, ,, _ _ _ Wo 95/11799 ~ 5 ~ ~ ~ pcrluss3/lo4o3 applied to the two address lines, respectively. This generates a magnetic field H which magnetizes the compositions of materials of a memory cell. The magnetic induction B, of these layers is depicted in Fig. lO (a) as a closed loop with an arrow. For the structure of Flg. 5 having the dimensions described lO above, the amplitude of the critical current Ic necessary to generate the critical field 6trength Hc required for switching between the two discrete states is approximately 8 . 0 ~A. At the cell where the two 15 pulses coincide, the two +5 . 0,uA currents create a f ield H that would be generated by applying a lO . 0,uA
current. Since this current is greater than Ic, the magnetic induction becomes Bo so that a binary "l" is stored. As explained previously, after the pulses have been applied, the magnetic induction at the cell B~ remains equal to Bo without requiring additional 25 refresh or power, so that a binary "l" is retained in carrier "a".
As shown in Fig . lO (b), to store a binary "O" into carrier "a", two ~y~ u~-ized current pulses, 30 I,---5 . 0,uA and I~=-5 . 0,uA are applied to the address lines respectively . Since the sum of these ~ L t~ is -10 . 0,uA, which is less than -Ic, the current pulses switch the magnetic state from +Bo to -B

Wo 95111799 ; PCrlUS93/10~03 Switching between +Bo and -Bo generates a piezoelectric voltage pulse between the f irst and second electrodes after a delay ~t from the time of the application of the current pulses. The piezoelectric pulse is positive ~or switching from +Bo to -Bo and is negative for switching from -Bo to +B
lO If the magnetic state does not change, no piezoelectric pulse is generated. Accordingly, the generated piezoelectric voltage pulses can be employed to verify that a bit of binary data has been stored.
To non-destructively retrieve information stored in carrier "a" of the cell, two ~y-,~ l.Lu..ized current pulses, I,-3.33,1LA and Ij=3.33,uA, are applied to the address lines. Since the critical current I~--8 . O~A, the 6 . 66~A sum of these currents cannot switch the magnetic states from +Bo to -Bo~ This current, however, is suf f icient to disturb the magnetic state 25 without complete switching. As shown in Fig . ll (a), assuming a binary "l" is stored in carrier, the applied current pulses change B, from a value corresponding to a point such as "a" on the 30 magnetization curve to the value ~_VL r e~ vl~ding to point "b" on the curve. Because of the previously discussed piezoelectric properties of the cell, this change in the magnetic induction generates a po8itive WO 95/11799 Z, 1 ~ ~i O ~ ~s PC~IUS93/10~03 piezoelectric voltage of approximately +lO,uV, indicating that a "l" is stored in carrier.
If a "O" is stored in carrier "a", the applied current pulses will change B, from a value corresponding ~o point "c" to the value corresponding to point "d" on the curve. In this case, however, the lO magnetic induction of carrier "a" remains at B,=-Bol so that no piezoelectric voltage is generated, indicating that a "0" is stored. The information stored in the memory cell is not changed during the reading process 15 because I is less than I~.
This process of non-destructively retrieving data is illustrated in the time domain in Fig. ll (b) .

The delay ~t between the piezoelectric voltage pulse and the input synchronized current pulses is theoretically estimated to be less than l. ons. This delay measured f rom the chip described above is about 1-2ns .
The storage and retrieval of data in carrier "b" is performed similarly. As shown in Fig. 12 (a), two synchronized current pulses, Ijs-5 . O~A and Ij=~5 . O~LA
30 are applied to the address lines to store a "l" in carrier "b". As discussed above, such current pulses do not affect carrier "a". At the point where the pulses coincide, a field which is equivalent to the 35 field induced by the O . l,uA current ls induced. Since WO95~11799 PCrlUS93/10403 2~
this value is greater than I~=8.0~A, a "l" is stored in carrier "b". As shown in Fig. 12 (b), the current pulses Ii=~5 ~ 0,uA and Ij=-5 . O~A are applied to store a "O" in carrier "b".
Switching from "l" to "O" in carrier "b"
generates a piezoelectric voltage pulse between the 10 electrodes at a delayed time ~t, and switching from "0" to "l" also generates a piezoelectric voltage pulse. If the state does not change, no piezoelectric voltage is generated.

The data stored in carrier "b" is non-destructively retrieved in a similar fashion as discussed in conjunction with carrier "a". As shown in Fig. 13, to non-destructively retrieve data stored in carrier "b", two synchronized current pulses, I;=3 . 33~A and Ij=-3 . 33~A are applied. The combined magnitudes of these pulses are not sufficiently large 25 to switch the magnetic state in carrier "b". If a "0"
is stored, the current pulses would not change the magnetic induction Bb f carrier "b", and no piezoelectric voltage between the electrodes would be 30 generated. If a "l" is stored in carrier !'b", the applied current pulses would disturb the magnetic state Bb=Bo but not change it, generating a positive piezoelectric voltage pulse after a delay l\t. Thus, 35 for carrier "b", the absence of piezoelectric voltage WO95/11799 2 17 ~ 0 ~ PCT~S93/10403 pulse indicates that a "0" is stored, and a positive piezoelectric voltage indicates that a "l" is stored.
In the above paragraphs, a non-destructive method of retrieving data from carriers "a" and "b"
has been described. Alternatively, information can also be destructively retrieved from the carriers.
Recall that changing the magnetic state of a carrier generates a piezoelectric voltage between the two electrodes. This piezoelectric voltage identifies the information stored in carrier. For example, to destructively retrieve data from carrier "a", two synchronized current pulses Ij=-5.0~A are applied to the two address lines respectively. These pulses write a "0" in.carrier "a". If a "0" has been stored in carrier "a", these current pulses would not change the magnetic state of carrier "a" and therefore, no piezoelectric voltage will be detected between the electrodes. However, if a "l" has been stored in carrier "a", the current pulses will change the magnetic state of carrier from the state corresponding to "l" to the state corresponding to "0", thereby generating a piezoelectric voltage between the electrodes. Note that in such retrieve process, the stored information may be changed. Therefore, a restoring process to rewrite the stored information back to the carrier is normally employed after the destructive r~etrieving.

WO 95/11799 PCTn~Ss3110403 44_ Fig. 14 symbolically summarizes the above-described methods of storing and retrieving data from carriers "a" and "b" of the memory device. Note that in this figure the wave forms are shown as symbolic representations and do not depict the actual signals.
Two synchronized currents Ij=3 . 33~LA and Ij=3 . 33~A are lO utilized for non-destructively retrieving information from carrier "a". Similarly, two synchronized currents I;=-3.33~1A and Ij=3.33~A can be used to retrieve data from carrier "b". Also a method of 15 destructive readout can be employed. For example, two synchronized current pulses Ij=5 . o~A and Ij=5 . 0,uA, that write a "l" in carrier "a", can be applied and the piezoelectric voltage generated in response to these pulses would identify the previously stored data, thereby destructively retrieving data from carrier "a" .
Another property of the memory of this invention is that a sufficient voltage applied between the two electrodes prevents storing information in this memory. This property can be utilized to selectively store information in a memory when several memory chips are employed c.",~iuLLently in the same storage device . The voltages suf f icient to prevent storing inf ormation in the memory of the invention is 35 approximately 200~V. When this voltage is applied to WO95/11799 ~ 17 ~ ~ ~ ~ PCT~S93/10403 the electrodes, the application of currents to the address lines does not change the stored information.
One of the advantages of the memory device of the present invention is its low power consumption in comparison with the prior art non-volatile magnetic memory devices. Since the storage media employed in this device is highly sensitive to the magnetic field generated by the driving currents, it can quickly store binary values at relatively small driving currents, about 5.0~A on each line. Consequently, the power consumption is low for storing and retrieving data.
Note also that the retrieval of information as a piezoelectric voltage generated between the electrode is intrinsically faster than generating induced electromagnetic voltage as it was done in prior art magnetic memory devices.

Typically, the delay between the current pulses and corresponding piezoelectric voltage is in the range of subnanoseconds. Switching between a "1"
and a "0" usually takes a few nanoseconds.
Thus a memory device which is random accessible, non-volatile, and operates in static mode has been described. This memory device offers high-speed operation, low power consumption, and storage of information at high density.

Wo 95/11799 PCTiU593110403 ~
2~o~

The claims which follow are to be interpreted to cover all the equivalent structures and methods. The invention is, thus, not to be limited by the above exemplary disclosure, but only by the following claims.

:~5

Claims

WHAT IS CLAIMED IS:

1. A composition of materials, comprising:
a first layer of material comprising the elements of Pb, Cd, and Fe, wherein the ratio between Pb, Cd, and Fe in the first layer is (1-x-y) :x:y where the values of (1-x-y), x and y are within the ranges of 0 (1-x-y) 1, 0 x 1, and 0 y 1; and a second layer of material formed on the first layer, the second layer of material comprising the elements of Cr, Zn, and R where R is an element selected from the group consisting of Te and Tl, wherein the ratio between Cr, Zn and R in the second layer is (1-z-w):z:w where the values of (1-z-w), z and w are within the ranges of 0 (1-z-w) 1, 0 z 1, and 0 w 1.

2. The composition of materials of claim 1 wherein the first layer further comprises an element selected from the group consisting of S and Se.

3. The composition of materials of claim 1 wherein the first layer further comprises the element S, and at least portions of Cd and Fe bond with S to form PbS and FeS.

4. The composition of materials of claim 2 wherein the first layer further comprises the element O, and at least a portion of Fe bonds with O to form Fe2O3.

5. The composition of materials of claim 4 wherein the first layer further comprises the element Bi.

6. The composition of materials of claim 2 wherein the value of x is within the range of 0.038 x 0.042.

7. The composition of materials of claim 6 wherein the value of y is within the range of 0.080 y 0.094.

8. The composition of materials of claim 2 wherein the value of x is substantially 0.04.

9. The composition of materials of claim 8 wherein the value of y is substantially 0.09.

10. The composition of materials of claim 1 wherein the second layer comprises the elements of Cr, Zn, and Te.

11. The composition of materials of claim 1 wherein the second layer further comprises an element selected from the group consisting of S and Se.

12. The composition of materials of claim 1 wherein the second layer further comprises the element S, and at least a portion of Zn bonds with S to form ZnS.

13. The composition of materials of claim 11 wherein the second layer further comprises the element O.

14. The composition of materials of claim 13 wherein the second layer further comprises the element Bi.

15. The composition of materials of claim 11 wherein the value of z is within the range of 0.34 Z 0.45.

16. The composition of materials of claim 15 wherein the value of w is within the range of 0.25 w 0.34.

17. The composition of materials of claim 16 wherein the value of (1-z-w) is within the range of 0.25 (1-z-w) 0.32.

18. The composition of materials of claim 17 wherein the value of z is within the range of 0.38 z 0.41.

19. The composition of materials of claim 18 wherein the value of w is within the range of 0.28 w 0.31.

20. The composition of materials of claim 11 wherein the value of z is substantially 0.4.

21. The composition of materials of claim 20 wherein the value of w is substantially 0.3.

22. The composition of materials of claim 1 wherein the first layer is substantially 0.7µm thick.

23. The composition of materials of claim 22 wherein the second layer is substantially 0.7µm thick.

24. A composition of materials having ferromagnetic and piezoelectric properties, comprising:
a first layer of material comprising the elements of Pb, Cd, Fe, S, and O, wherein the ratio between Pb, Cd, and Fe is (1-x-y):x:y and the values of x and y is within the ranges of 0.38 x 0.042 and 0.080 y 0.094, wherein the first layer is saturated with S and O, and wherein at least portions of Pb and Fe bond with S to form PbS and FeS and at least a portion of Fe bonds with O to form Fe2O3; and a second layer of material formed on the first layer, the second layer of material comprising the elements of Cr, Zn, Te, S, and O wherein the ratio between Cr, Zn, and Te is (1-z-w):z:w and the value of (1-z-w), z, and w are within the ranges of 0.25 (1-z-w) 0.32, 0.38 z 0.41 and 0.28 w 0.31, wherein the second layer is saturated with S and O, and wherein at least a portion of Zn bonds with S to form ZnS.

25. The composition of claim 24 wherein the first and second layers further comprise the element Bi.

26. A method of manufacturing a composition of materials having ferromagnetic and piezoelectric properties comprising the steps of:
forming a first layer of material on a substrate, the first layer comprising the elements of Pb, Cd, and Fe, wherein the ratio between Pb, Cd, and Fe is (1-x-y):x:y and wherein the values of (1-x-y), x and y are within the ranges of 0 (1-x-y) 1, 0 x 1, and 0 y 1; and forming a second layer of material on the first layer, the second layer comprising the elements of Cr, Zn, and R, where R is an element selected from the group consisting of Te and Tl, wherein the ratio between Cr, Zn and R is (1-z-w):z:w and wherein the values of (1-z-w), z and w are within the ranges of 0 (1-z-w) 1, 0 z 1, and 0 w 1.

27. The method of claim 26 further comprising the step of adding an element selected from the group consisting of S and Se into the first and second layers.

28. The method of claim 26 further comprising the step of adding S into the first and second layer by exposing the substrate surfaces to sulfur vapor prior to forming the first and second layers.

29. The method of claim 27 further comprising the step of adding the elements of Bi and O
to the first and second layers.

30. The method of claim 29 wherein the step of adding the elements of Bi and O comprises immersing the composition of materials in liquid comprising Bi2O3 and water and performing an electrochemical process.

31. The method of claim 30 wherein the weight percentage of Bi2O3 in the liquid is approximately between four and six percent.

32. The method of claim 31 further comprising the step of continuously stirring the liquid.

33. The method of claim 32 further comprising the step of heating the liquid.

34. The method of claim 33 further comprising the step of maintaining the liquid at substantially 97°C.

35. The method of claim 30 comprising the step of applying a series of voltage pulses having triangular shapes to the composition of materials with respect to the liquid.

36. The method of claim 35 wherein the pulses are applied for forty-five hours and the amplitude of each pulse being substantially two thousand volts.

37. The method of claim 36 wherein the duration of each pulse is essentially one second and the time interval between centers of two adjacent pulses is essentially six seconds.

38. The method of claim 35 further comprising the step of alternatively applying positive and negative electrical potentials to the composition of materials with respect to the liquid after applying the voltage pulses.

39. The method of claim 38 wherein, during an eight hour interval, the negative potential is substantially sixty volts and is applied continuously for substantially four hours, and the positive potential is also substantially sixty volts and is applied for substantially four hours.

40. The method of claim 39 wherein the electrochemical process is performed for substantially 28 days.

41. The method of claim 29 wherein the value of x is within the range of 0.038 x 0.042.

42. The method of claim 41 wherein the value of y is within the range of 0.08 y 0.094.

43. The method of claim 42 wherein the value of z is within the range of 0.34 z 0.45.

44. The method of claim 43 wherein the value of w is within the range of 0.25 w 34.

45. The method of claim 42 wherein the value of (1-z-w) is within the range of 0.25 (1-z-w) 0.32.

46. The method of claim 29 wherein the value of x is substantially 0.04.

47. The method of claim 44 wherein the value of y is substantially 0.09.

48. The method of claim 45 wherein the value of z is substantially 0.4.

49. The method of claim 46 wherein the value of w is substantially 0.3.

50. A non-volatile, random-accessible memory device comprising:
a substrate having first and second surfaces;
a first address line formed on the first surface of the substrate;
a second address line formed on the second surface of the substrate;
a first composition of materials formed on the first surface of the substrate, the first composition of materials having ferromagnetic and piezoelectric properties and comprising the elements of Pb, Cd, Fe, Cr, Zn and an element selected from a group consisting of Te and Tl; and a second composition of materials formed on the second surface of the substrate, the second composition of materials having ferromagnetic and piezoelectric properties and comprising the elements of Pb, Cd, Fe, Cr, Zn and an element selected from a group consisting of Te and Tl.

51. The device of claim 50 wherein the second address line is substantially orthogonal to the first address line.

52. The device of claim 50 further comprising:
a first electrode formed on the first composition of materials; and a second electrode formed on the second composition of materials.

53. The device of claim 50 wherein the first and second compositions of materials are substantially identical.

54. The device of claim 53 wherein the first and second compositions of materials comprise essentially two sequentially formed layers.

55. The device of claim 48 wherein the first and the second compositions of materials comprise a first layer of material formed on the substrate and a second layer of material formed on the first layer, the first layer of material comprising the elements of Pb, Cd, and Fe, wherein the ratio between Pb, Cd, and Fe in the first layer is (1-x-y):x:y where the values of (1-x-y), x and y are within the ranges of 0 (1-x-y) 1, 0 x 1, and 0 y 1, and the second layer of material comprising elements of Cr, Zn, and R where R is an element selected from the group consisting of Te and Tl, wherein the ratio between Cr, Zn and R in the second layer is (1-z-w):z:w where the values of (1-z-w), z and w are within the ranges of 0 (1-z-w) 1, 0 z 1, and 0 w 1.

56. The device of claim 55 wherein the first and second layers further comprises an element selected from the group consisting of S and Se.

57. The device of claim 56 wherein the first and second layers further comprise the element O.

58. The device of claim 57 wherein the first and second layers further comprise the element Bi.

59. The device of claim 58 wherein x, y, z, and w are values within the ranges of 0.038 x 0.042, 0.08 y 0.94, 0.34 z 0.45, and 0.25 w 0.34.

60. The device of claim 59 wherein z, w, and (1-z-w) are values within the ranges of 0.38 z 0.41, 0.28 w 0.31, and 0.25 (1-z-w) 0.32.

61. The device of claim 60 wherein the value of x is substantially 0.04, the value of y is substantially 0.09, the value of z is substantially 0.40, and the value of w is substantially 0.30.

62. The device of claim 55 wherein the substrate is a silicon substrate.

63. The device of claim 55 wherein the first and second address lines are made of a conductive material.

64. The device of claim 63 wherein the conductive material comprises silver.

65. The device of claim 63 wherein the address lines are metal strips substantially 1µm wide and substantially 1µm thick.

66. A method of manufacturing a non-volatile, random-accessible memory device comprising the steps of:

forming a first address line on a first surface of a substrate;
forming a second address line on a second surface of the substrate;
forming a first composition of materials on the first surface of the substrate, the first composition of materials having ferromagnetic and piezoelectric properties; and forming a second composition of materials on the second surface of the substrate, the second composition of materials having ferromagnetic and piezoelectric properties.

67. The method of claim 66 further comprising a step of polishing the outermost surfaces of the first and second compositions of materials until the surfaces become substantially planar.

68. The method of claim 67 further comprising the steps of:
forming a first electrode on the first composition of materials; and forming a second electrode on the second composition of materials.

69. The method of claim 66 wherein the second address line is substantially orthogonal to the first address line.

70. The method of claim 66 wherein the first layer of material comprises the elements of Pb, Cd, and Fe, wherein the ratio between Pb, Cd, and Fe is (1-x-y):x:y where the values of (1-x-y), x and y are within the ranges of 0 (1-x-y) 1, 0 x 1, and 0 y 1; and wherein the second layer of material comprises elements of Cr, Zn, and R where R is an element selected from the group consisting of Te and Tl, wherein the ratio between Cr, Zn and R is (1-z-w):z:w where the values of (1-z-w), z and w are within the ranges of 0 (1-z-w) 1, 0 z 1, and 0 w 1.

71. The method of claim 70 wherein the values of x, y, z, and w are within the ranges of 0.038 x 0.042, 0.08 y 0.094, 0.34 z 0.45, 0.25 w 0.34.

72. The method of claim 71 wherein the values of z, w and (1-z-w) are within the ranges of 0.38 z 0.41, 0.28 w 0.31, and 0.25 (1-z-w) 0.32.

73. The method of claim 72 wherein the value of x is substantially 0.04, the value of y is substantially 0.09, the value of z is substantially 0.40, and the value of w is substantially 0.30.

74. The method of claim 66 wherein the step of forming the address lines comprises depositing conductive layers onto the surfaces of the substrate and etching the conductive layers.

75. A method of storing two independent bits of binary data in a single storage cell of a non-volatile, random accessible memory comprising composition of materials having ferromagnetic and piezoelectric properties and comprising the elements of Pb, Cd, Fe, Cr, Zn, Te, S and O, each such cell being addressable by a first and a second address lines, comprising the steps of:
applying two electrical currents having same polarities to the first and second address lines, respectively, for storing a first bit of binary data in the storage cell; and applying two electrical currents having opposite polarities to the first and second address lines, respectively, for storing a second bit of data in the storage cell.

76. The method of claim 75 wherein the currents applied for storing the first bit of data have essentially identical amplitudes.

77. The method of claim 76 wherein the currents applied for storing the second bit of data have essentially identical amplitudes.

78. The method of claim 77 wherein the amplitudes of the electrical currents applied for storing the first and the second bits of data are such that a single current is not sufficient to change the binary value of the stored data.

79. The method of claim 78 wherein the amplitude of each current applied for storing the first bit of binary data is substantially 5.0µA.

80. The method of claim 79 wherein the amplitude of each current applied for storing the second bit of binary data is substantially 5.0µA.

81. The method of claim 75 further comprising the step of detecting a piezoelectric voltage induced in the storage cell, the piezoelectric voltage indicating that a binary data has been stored in the cell.

82. The method of claim 75 wherein the step of applying two electrical currents for storing the first bit of data includes storing a first binary value by applying two electrical currents both having a first polarity and same amplitude to the first and second address lines, respectively, and storing a second binary value by applying two electrical currents having a second polarity which is opposite to the first polarity and same amplitude to the first and second address lines, respectively.

83. The method of claim 82 wherein the two electrical currents for storing the first bit of data are applied as two synchronized electrical current pulses.

84. The method of claim 75 wherein the step of applying two electrical currents for storing the second bit of data includes storing a first binary value by applying two electrical currents having a first and second polarities and the same amplitude to the first and second address lines, respectively, and storing a second binary value by applying two electrical currents having the second and first polarities and same amplitudes to the first and second address lines, respectively.

85. The method of claim 84 wherein the two electrical currents for storing the second bit of data are applied as two synchronized electrical current pulses.

86. A method of non-destructively retrieving first and second independent bits of binary data stored in a single storage cell of a non-volatile, random-accessible memory comprising composition of materials having ferromagnetic and piezoelectric properties and comprising the elements of Pb, Cd, Fe, Cr, Zn, Te, S, and O, each such cell being addressable by a first and a second address line, comprising the steps of:
applying two electrical currents having the same polarity to the first and second address lines, respectively, for retrieving the first bit of data, wherein the amplitudes of the currents are substantially the same and the combined amplitudes of the currents are insufficient for changing the binary value of the stored data; and applying two electrical currents having opposite polarities to the first and second address lines, respectively, for retrieving the second bit of data, wherein the amplitudes of the currents are substantially the same and the combined amplitudes of the currents are insufficient for changing the binary value of the stored data.

87. The method of claim 86 further comprising the step of sensing a piezoelectric voltage generated in the storage cell in response to the applied currents.

88. The method of claim 87 wherein the amplitudes of the two electrical currents applied for retrieving the first or the second bits of data are substantially 3.33µA.

89. The method of claim 86 wherein the two electrical currents are applied for retrieving the first bit of data as two synchronized electrical current pulses having the same amplitudes and polarities.

90. The method of claim 89 wherein the two electrical currents are applied for retrieving the second bit of data as two synchronized electrical current pulses having the same amplitudes and opposite polarities.

91. A method of destructively retrieving first and second independent bits of binary data stored in a single storage cell of a non-volatile, random-accessible memory comprising composition of materials having ferromagnetic and piezoelectric properties and comprising the elements of Pb, Cd, Fe, Cr, Zn, Te, S, and O, each such cell being addressable by a first and a second address line, comprising the steps of:
applying two electrical currents having a first polarity to the first and second address lines, respectively, for retrieving the first bit of data, wherein the amplitudes of the currents are substantially the same and the combined amplitudes of the currents are sufficient for changing a binary value stored in the first bit of binary data to a first binary value, and sensing a first piezoelectric voltage; and applying two electrical currents, one having a first polarity to the first address line, and the other having a second polarity which is opposite to the first polarity to the second address line, respectively, for retrieving the second bit of data, wherein the amplitudes of the currents are substantially the same and the combined amplitudes of the currents are sufficient for changing a binary value stored in the second bit of binary data to a first binary value, and sensing a second piezoelectric voltage.

92. The method of claim 91 wherein the first piezoelectric voltage has a non-zero value, indicating the first bit of data having a second binary value; and wherein the second piezoelectric voltage has a non-zero value, indicating the second bit of data having a second binary value.

93. The method of claim 92 wherein the amplitudes of the two electrical currents applied for retrieving the first or the second bits of data are substantially 5.0µA.

94. The method of claim 91 wherein the two electrical currents are applied for retrieving the first bit of data as two synchronized electrical current pulses having the same amplitudes and polarities.

95. The method of claim 94 wherein the two electrical currents are applied for retrieving the second bit of data as two synchronized electrical current pulses having the same amplitudes and opposite polarities.