WO2001033585A1

WO2001033585A1 - Synthesis and magnetoresistance test system using double-perovskite samples for preparation of a magnetoresistance device

Info

Publication number: WO2001033585A1
Application number: PCT/US2000/041816
Authority: WO
Inventors: Ivan Bozovic
Original assignee: Oxxel Oxide Electronics Technology, Inc.
Priority date: 1999-11-05
Filing date: 2000-11-02
Publication date: 2001-05-10

Abstract

A position-addressable combinatorial library of double-perovskite materials is disclosed. The library includes a plurality of multi-region substrates, and formed on the substrate regions are double-perovskite compounds having the formulae (AxA'1-x)2(ByB'1-y)CzC'1-z)O6±$g8d), wherein A is a 2+ cation, A' is a 3+ or 1+ cation, B and B' are each paramagnetic 3+ cations, C and C' are each paramagnetic 5+ cations, x, y and z are whole or fractional number including zero, and δ represents a small deviation from ideal oxygen stoichiometry. The stoichiometry or structure of one pair of elemental components of the double-perovskite materials varies systematically in a first direction and that of another pair varies systematically in a second direction. Also disclosed are an apparatus and a method of producing such a library, namely the combinatorial molecular beam epitaxy (COMBE), the use of the library in identifying double-perovskite compounds with desired properties, e.g., spin polarization or high Curie temperature, and the fabrication and testing of TMR devices comprised of double-perovskite compounds.

Description

junctions with 1.6-nm-thιck barriers," Applied Physics Letters 74: 290-292 (1999).

[8] K.I. Kobayashi, et al., "Room-temperature magnetoresistance m an oxide material with doubles ' perovskite structure," Nature 395:677-680 (1998).

[9] T. Manako, et al., "Epitaxial thin films of ordered double perovskite Sr_∑FeMoOβ," Applied Physics Letters 74:2215-2217 (1999).

[10] X.D. Xiang et al., "A Combinatorial Approach to 0 Materials Discovery," Science 2_68: 1738-1740 (1995).

[11] G. Briceno, et al., "A Class of Cobalt Oxide Magnetoresistance Materials Discovered with Combinatorial Synthesis," Science 270:273-275 (1995).

[12] E. Damelson et al., "A Combinatorial Approach to 5 the Discovery and Optimization of Luminescent Materials , " Nature 389:944-948 (1997).

[13] E. Damelson, et al., "A Rare-Earth Phosphor Containing One-Dimensional Chains Identified through Combinatorial Methods," Science 279:837-839 (1998). 0 [14] I. Bozovic, et al., "Superconducting Oxide Multilayers and Superlattices : Physics, Chemistry , and

Nanoengineeπng," Physica C 235-240:178-181 (1994).

[15] I. Bozovic, et al., "Atomic Layer Engineering of Cuprate Superconductors," Journal of Superconductivity 7:187-195 (1994) .

[16] I. Bozovic, et al., "Atomic-Level Engineering of Cuprates and Manganites , " Applied Surface Science 113- 114:189-197 (1997).

[17] I. Bozovic, et al., "Reflection High-Energy Electron Diffraction as a Tool for Real Time Characteriztion of Growth of Complex Oxides," Chapter 3 in CHARACTERIZATION OF THIN FILM GROWTH PROCESSES VIA IN SITU TECHNIQUES, A. Kraus and 0. Auciello, editors (John Wiley and Sons, New York, 1999) m press. [18] R. J. Soulen Jr., et al., "Measuring the Spin Polarization of a Metal with a Superconducting Point Contact," Science 282:85-87 (1998). [19] S. K. Upadhyay, et al . , "Probing Ferromagnets wi th Andreev Reflection , " Physi cal Review Letters 81 : 3247-3250 (1998) .

[20] J. H. Park, et al . , "Direct evidence for a half- filled ferromagnet , " Na ture 392:794-796 (1998).

Background of the Inventxon

Silicon-based dynamic-random-access memory (DRAM) devices are used today in most consumer electronic devices. A DRAM may contain millions of cells, each consisting of a transistor and a capacitor. The latter needs some back-up current in order to be continually recharged, and from this drawback stems an incentive to develop non-volatile memory devices. Several candidate technologies are being actively pursued, including the so-called flash memory, the electrically erasable programmable read-only memory

(EEPROM) , ferroelectric random-access memory (FRAM), and several versions of magnetic random-access memory (MRAM) .

One particularly interesting and recent implementation of MRAM is based on the so-called magneto-resistance (MR) effect, described below.

The traditional magnetic reading and writing heads employ inductive coils, which cannot be made very small; this limits the maximum density of the recordings. To achieve higher density, such coils are replaced by solid- state magnetic sensors, which can be made substantially smaller. Such sensors are made of materials whose electrical resistance changes under the effect of a magnetic field. [1] The larger the relative change of resistance, ΔR = [R(H=0) - R(H=H_op)] / R(H=0), at the operational magnetic field (H_op) , the larger is the change in the measured electrical quantity (i.e., voltage or current), and the more sensitive is the sensor. For a practical device, the field H_op should be reasonably small, on the order of 100 Gauss. In systems commercially available today, such sensors are based on metallic multilayers that show the so-called Giant Magneto-Resistance (GMR) effect. Despite the name, the actual values of ΔR m

GMR sensors hardly exceed a few percent. Clearly, materials or heterostructures with much larger ΔR would be rather desirable, since this would provide higher device sensitivity. [1]

For this reason, great interest has been raised by the discovery of the so-called Colossal Magneto-Resistance (CMR) effect, m a class of manganites such as Lao 7Sro ₃Mn0₃ and Lao 7Cao ₃Mn0₃. [2] In these compounds, ΔR can be as large as 100,000%. However, there are two drawbacks: the effect attains this maximum only at cryogenic temperatures, and only at rather high operational fields, 5-10 Tesla, which is too high for the intended applications of these magnetic sensors. [2] To remedy these drawbacks, several strategies have been employed. The one with the greatest technological potential is to use artificial tπ-layer (or multi-layer) heterostructures, where the electrons tunnel from one ferromagnetic electrode to another, across a non-magnetic barrier. If the spins of mobile electrons m the electrodes are polarized - as is the case in the manganites mentioned above - the tunneling current through the device is larger when the top and the bottom electrodes are polarized in parallel, and it is smaller in the case of anti-parallel polarization. Such devices are known as Spin-Valve Transistors (SVTs), and the effect itself is called Tunneling Magneto-Resistance (TMR) . [3-7] . As a magnetic field sensor, the mam advantage of a SVT is that it can operate at low fields: devices made with Lao 7Sro ₃Mn0₃ electrodes have shown ΔR = 500-1,000% at H_op = 100 Gauss.

This is a great enhancement of the low-field sensitivity compared with either the bulk Lao 7Sro ₃Mn0₃ crystals or thin films, or with the conventional GMR sensors. Furthermore, the electrical reading of a TMR device provides information on its polarization state, without changing it. Hence, it is expected that such devices may play a role in future non-volatile MRAM electronics [3-7]. However, the second drawback, the need for cryogenic operation, is even more serious here: most of the manganite-based TMR devices made so far required cooling down to 50 K and below for a major MR effect to occur. This constraint indeed reduces the range of conceivable implementations m commercial systems.

For this reason, interest has been renewed in certain ferromagnet compounds that show a substantial level of spin polarization and have a high Curie temperature T_c (i.e., the temperature below which the material is ferromagnetic) . These two features make them candidate materials for TMR devices capable of operation at room temperature. Examples of such compounds are Ca₂FeReθ6 with T_c = 538 K, Sr₂FeMoθ6 with T_{c ∞} 450 K, etc. [8, 9]. These compounds have a so- called double-perovskite structure, A₂BCO_D, where A is a large divalent cation in the 12-fold oxygen coordination, while B and C are a small trivalent cation and a small pentavalent cation, respectively, both n the 6-fold coordination. Clearly, there is a large number of conceivable double-perovskite compounds; if one allows for partial chemical substitution, it becomes huge. Since only a few of these compounds have been synthesized and characterized so far, it is extremely unlikely that we have already reached the limits m the properties of interest such as the Curie temperature T_c, the degree of spin polarization, etc .

Even with various simplifications and restrictions imposed, the number of possible stoichiometric combinations is enormous. This means that the number of conceivable stable or metastable compounds within this general composition is also huge. If one wanted to check each of them, the traditional approach of searching for new inorganic materials - mixing predetermined quantities of the starting elements or compounds, making them react, performing some characterization measurements, using this feedback to synthesize another sample of a different composition, and so on - is clearly totally impractical here .

Hence, there is an apparent need for a methodical search for better materials, by means of synthesis and characterization of a series of compounds with systematic variation of the chemical composition, and in small increments of stoichiometric ratios.

To further exacerbate the problem, the traditional sequen tial search is slow and inadequate when dealing with complex compounds. For a faster and more systematic search, one would like to employ some parallel processing, i.e., to synthesize simultaneously batches of many (100, 1,000 or even more) samples, with a systematic and known variation of the composition across the series. Ideally, one would also like to test the samples in parallel, i.e., all at once. If the samples are coded somehow - e.g., by the position in a one- or two-dimensional matrix - one could then tell which composition provided the optimum value of the property under study. This approach, known as combinatorial search for new materials, or combina torial chemistry, has been utilized in pharmaceutical industry for quite some time. It has recently been implemented also in inorganic chemistry, and more specifically, in search for novel complex oxides. [10, 11] Several new phosphors, of various colors and more efficient than the ones currently used in cathode-ray tubes, have been discovered in this way. [12, 13]

Atomic Layer-by-Layer Molecular Beam Epitaxy (ALL- MBE) , is a technique developed in the last decade for deposition of single-crystal thin films of cuprate superconductors and other complex oxides. [14-16] An ALL- MBE system comprises an ultra-high vacuum chamber housing a number of atomic sources - most frequently thermal effusion sources or Knudsen cells - equipped with computer- controlled shutters. To monitor the atomic fluxes, atomic absorption spectroscopy has been proved the most useful, since this is accurate enough to detect changes of less than one per cent and fast enough to allow real-time feedback control. By using a pure ozone beam, sufficient oxidation can be achieved under high vacuum conditions, which permits in -si tu monitoring of the surface structure by RHEED (reflection high-energy electron diffraction) and other surface analytical tools. [17] Thin films grown by ALL-MBE typically show atomically flat surfaces and, in the case of superlattices and multi-layers, virtually perfect interfaces. This allows fabrication of tri-layer ("sandwich") tunnel junctions with state-of-the-art properties and uniformity. [14-16]

A novel, combinatorial ALL-MBE (COMBE) system which has been implemented at Oxxel is the subject of a pending provisional application, serial no. 06/156,264, entitled "COMBINATORIAL MOLECULAR BEAM EPITAXY (COMBE) APPARATUS,

METHOD, AND LIBRARY ARRAYS," filed on September 27, 1999 in the name of Ivan Bozovic and assigned to the assignee of this application, the disclosure of which is incorporated by reference herein. The COMBE system disclosed in that application has the capability to vary all of the key parameters that influence the outcome of an ALL-MBE deposition experiment - the shuttering sequence, the stoichiometry (i.e., the chemical composition), the substrate temperature T, the ozone partial pressure p and, for multi-layer structures, the thickness of the constituent layers - either one at a time or all at once, in a combinatorial way. This is made possible by use of the following devices: (a) transmission (sieve) masks in front of each elemental metal source, allowing one to control the composition gradient separately for each element; (b) substrate shadow masks; (c) a multi-filament substrate heater, providing for a temperature gradient; and (d) a multi-nozzle ozone delivery system, providing for a gradient in ozone partial pressure. In addition, the Oxxel COMBE system is equipped with a set of analytical tools including a novel scanning Reflection High-Energy-Electron Diffraction (RHEED) system, a scanning quartz crystal monitor (QCM) system, a Low- Energy Electron Microscope (LEEM) , and Time-of-Flight Ion Scattering and Recoil Spectroscopy (TOF-ISARS) system, which enable m-si tu , real-time monitoring of the chemical composition, the crystallographic structure, and the morphology of the growing film surface. These tools have never been integrated with an MBE system before. They make it possible to determine the optimal conditions (e.g. T, p, the shutter sequencing, etc.) for deposition of a given compound, in a single growth experiment.

Summary of the Inventxon

It is therefore an object of this invention to provide a combinatorial library of double-perovskite materials, with the library constructed and the materials selected to expedite search for new oxide ferromagnet and ferπmagnet compounds. The library comprises a substrate having a plurality of surface regions covered with thin films of a family of distinct materials, all of which can be described by a generic chemical formula.

It is another object of this invention to provide a method and apparatus for synthesizing the libraries using combinatorial molecular beam epitaxy (COMBE) .

It is a further object of this invention to provide a method for fast, parallel testing of the materials in the libraries .

It is yet another object of this invention to provide a method of fabrication and testing of combinatorial libraries of tunneling magneto-resistance (TMR) devices. It is still a further object of this invention to provide a method for testing and optimizing magneto- transport properties of such TMR devices.

In one aspect, the invention includes a position- addressable combinatorial library of double-perovskite materials. The library is formed on a substrate having a substrate surface bounded by two or more pairs of opposite side regions, and defining multiple surface regions at known positions with respect to the side regions. Formed on these surface regions are double-perovskite materials having the formulae (A_XA' ι-_x) ₂ (B_yB' ι-_y) (C_ZC ι-_z) Oβ₊δ, wherein A is a 2+ cation, A' is a 3+ or 1+ cation, B and B' are each paramagnetic 3+ cations, C and C are each paramagnetic 5+ cations, x, y and z are whole or fractional numbers including zero, and δ represents a small deviation from ideal oxygen stoichiometry. The stoichiometries or structures of the double-perovskite materials vary systematically and in a known way in at least one elemental component, progressing between one side region and another in a pair of opposite side regions.

In preferred embodiments, A is selected from the group consisting of Ba, Sr, Ca and Pb; A' is selected from the group consisting of Ce, Na and K; B and B' are each selected from the group consisting of Mn, Fe, Co, Cr and Ni; and C and C are each selected from the group consisting of Mo, W and Re .

Preferably, one pair of elemental components varies systematically, progressing between one side region and another in a first pair of opposite side regions, while another pair of elemental components varies systematically, progressing between one side region and another in a second pair of opposite side regions. In a preferred embodiment, the double-perovskite materials are formed on a film that is partitioned into multiple areas, each area corresponding to a respective one of the multiple regions on the substrate surface and each area containing a square array of pixels. With this arrangement, x varies in predetermined increments from pixel to pixel, progressing between one side region and another in a first pair of opposite side regions, while y varies in the predetermined increments from pixel to pixel, progressing between one side region and another in a second pair of opposite side regions.

The substrate may be partitioned into a set of 16 areas (or comprise a set of 16 substrates) arranged in a 4x4 matrix. Each film area corresponds to a respective one of the multiple regions on the substrate surface, with x, y or z varying approximately 40% across the film, in the current Oxxel COMBE system. This range can, of course, be made larger or smaller, without representing an essential deviation from the present invention.

Each of the variables x, y and z may vary systematically from 0 to 1 in predetermined increments or in a continuous manner. All three of the variables may vary, or one of the variables may be fixed while the other two vary, or two of the variables may be fixed while the third one varies .

In another general embodiment, a plurality of double- perovskite materials identified as optimal based on Curie temperature and degree of spin polarization are used to form two electrode magnetic layers. A barrier layer comprised of an insulator perovskite is formed in between.

In another aspect, the invention includes a method of identifying at least one double-perovskite material having a desired solid-state property, such as ferromagnetism or ferrimagnetism. The method involves forming a position- addressable combinatorial library of double-perovskite materials, as described above. The individual double- perovskite materials on the substrate are tested for the desired solid-state property, and the structure of each double-perovskite material having the desired property is idnetified according to its position on the substrate and the known synthetic history of that substrate position.

The testing may be carried out at different temperatures to identify the Curie temperature for each double-perovskite material. The testing may also, or alternatively, include determining the degree of spin polarization of each double-perovskite material.

To reduce the full three-dimensional search, a first variable of x, y and z may be varied in a predetermined increment and the corresponding double-perovskite materials tested to identify a first subset of materials having a

10 desired characteristic. Then, a second variable of x, y and z may be fixed while a first sublibrary of double-perovskite materials is formed, varying the first and third variables, after which the first sublibrary of materials is tested to identify a second subset of materials having the desired characteristic. Then, the third variable of x, y and z may be fixed while a second sublibrary of double-perovskite materials is formed, varying the first and second variables, after which the second sublibrary of materials is tested to identify a third subset of materials having the desired characteristic .

In another embodiment, for use in fabricating and testing a tunneling magneto-resistance (TMR) , double- perovskite materials identified as optimal based on Curie temperature and degree of spin polarization are used to synthesize electrode magnetic layers. Individual double- perovskite materials on the electrodes are tested to identify the double-perovskite material on the electrodes that optimizes the thickness of an insulating perovskite barrier layer formed in between the electrodes, as well as thermodynamic parameters for growth of the barrier layer.

In another embodiment, TMR devices are tested m parallel. The testing includes applying a magnetic field to switch the polarization of one electrode of each device while maintaining the polarization of the other electrode of each device in a fixed orientation, measuring the relative change of electrical resistance for each device, and identifying each device exhibiting one or both of the highest change of electrical resistance and the highest operational temperature.

In yet another aspect, the invention includes a method of producing a combinatorial library of double-perovskite materials. The library is formed by placing in a vacuum chamber containing elemental sources with shutters, a substrate having a substrate surface bounded by two or more pairs of opposite side regions, and defining multiple surface regions at known positions with respect to the side

11 regions. Successive atomic layers are deposited on the substrate surface regions in the presence of a gaseous source of oxygen atoms. During the depositing, the deposition conditions are varied m a known and systematic way for at least one process variable, progressing between one side region and another in a pair of opposite side regions. The processing variable may be: (1) the quantity of at least one elemental component deposited from one or more elemental sources; (n) the sequence of deposition from the elemental sources; (m) the temperature of the substrate regions; (ιv) the partial pressure of the source of atomic oxygen at the substrate regions; and (v) the thickness of barriers layers m multilayered structures.

The deposition conditions for at least one process variable may be varied under the control of a program- controlled processor.

Varying the process variable of the quantity of material deposited from an elemental source may be accomplished by placing between the source and the target, a sieve mask effective to produce a desired gradient of atomic flux through the mask, progressing between one side region and another in a pair of opposite side regions of the substrate, or effective to produce a uniform atomic flux through the mask. Additionally, or alternatively, the process variable of the quantity of material deposited from an elemental source may be varied by placing a movable shutter between the source and the target, at each of a plurality of selected positions. The placing of the movable shutter may be controlled by a program-controlled processor.

Another way is to arrange that the two sources of the same element, placed one opposite to another, have different atomic fluxes, e.g., by adjusting the temperature of each source. Yet another and the simplest way is to use different shuttering times on the two opposing sources of the same element, i.e., to keep one of them open longer than the other, according to the desired gradient. The current

12 Oxxel COMBE system has 16 sources at low angle (about 20°) with respect to the substrate, which is favorable for implementing this scheme. It is apparent to one skilled in the art that one could implement an MBE system with a smaller or a larger number of elemental sources.

Varying the temperature of the substrate regions may be accomplished by heating the substrate so as to produce a temperature gradient, progressing between one side region and another in a pair of opposite side regions of the substrate. For example, this can be accomplished by using a plurality of heater elements, such as quartz lamps, each one independently powered and controlled, as is the case in the current Oxxel COMBE system.

Varying the partial pressure of the source of atomic oxygen at substrate regions may be accomplished by directing ozone from a source positioned at one side of the substrate holder, at a nearly grazing angle with respect to the substrate surface.

The combinatorial library of double-perovskite materials is preferably produced with a combinatorial molecular beam epitaxy apparatus.

These and other objects and features of the invention will become more fully apparent when the following detailed description of the invention is read in conjunction with the accompanying drawings.

Brief Description of the Drawings

Fig. 1 is a fragment of a combinatorial library in accordance with the present invention. Fig. 2 is a schematic representation of a TMR device, constructed in accordance with embodiments of the present invention.

Fig. 3 is functional block diagram of a typical computer that may be used to implement various aspects of the present invention.

13 Detailed Description of the Invention

A. Combinatorial Library of Double-Perovskite Ferromagnets

The combinatorial library of compounds of the present invention is constructed and the compounds selected to expedite search for new oxide ferromagnets and ferπmagnets . The library is comprised of a number of substrates covered with thin films of family of distinct materials, all of which can be described by the common chemical formula:

(A_xA'ι-_x)₂(B_yB'ι-_y) (CzC'i-zJOe±_δ, (1)

where A is a large 2+ cation, in 12-fold coordination; A' is 3+ or 1+ cation, 12-fold coordination; B and B' are small paramagnetic 3+ cations, in 6-fold coordination, and C and C are small paramagnetic 5+ cations, in 6-fold coordination. Preferably, A= Ba, Sr, Ca or Pb; A' = Ce, Na or K; B, B' = Mn, Fe, Co, Cr or Ni, and C, C = Mo, W or Re, and δ ^1S possible small deviation from the ideal oxygen stoichiometry. Preferably, the indices x, y, z, and take on only a discrete set of values, say 0, 0.01, 0.02,..., 0.99, 1.00. The rationale for this simplification is twofold: (l) the physical properties of interest change very little on a finer scale, and (n) m the current thin- film deposition practice, it is hardly possible to control the stoichiometry beyond this accuracy. This is to say, when we write, e.g., Bao 65, what we really mean is Bao 65_±o oi, etc. As will be apparent to one skilled in the art, it is also possible, and actually easier, to implement a quasi- continuous stoichiometry variation of one or more elements across a wafer or a substrate array, as expounded above. A library having the same combination of elements as the one disclosed here, and varied in the same overall range, a quasi-continuous manner, is just another, and simpler, embodiment of the present invention. It can be made using

14 the same apparatus and method, but without the need to use the shutters and masks, just the elemental atomic flux gradients across the wafer or the substrate array.

The library disclosed here consists of a large number (to be specified below) of lxl cm² thin films deposited onto suitable lattice-matched perovskite substrates (e.g., SrTi0₃ , LaAl0₃ or NdGa0₃ single crystal) . Each film is partitioned into 10x10 pixels, with each pixel having an area of lxl mm². The typical film thickness is 1,000 angstroms. Substrates are grouped into sets of 16, and each set is arranged as a 4x4 matrix, on a 3" substrate holder. The film is deposited simultaneously over the entire set of 16 substrates, but with composition gradients along two axes, say x and y. In this way, the relative amounts of four elements are varied, in steps of 0.01 from one pixel to the next one, so that each pixel corresponds to a different and known stoichiometry. The sets are chosen in such a way as to cover all the possible elemental and stoichiometric combinations. An alternative embodiment, also possible with the above-described COMBE system, is to use one 2" or 3" wafer, and dice it subsequently into lxl cm² chips for testing. Alternatively, the testing apparatus could also be modified to accommodate such larger wafers. Further, it is possible to also use other substrates, e.g., Si, with an appropriate epitaxial oxide buffer layer, e.g., SrTi0₃.

In the library disclosed here, there are 4x3x10x3 = 360 combinations with six different metal atoms present. For each of these combinations, there is a three- dimensional matrix of stoichiometric indices x, y, z. In one deposition, we keep one of these (say z) fixed, and vary the other two (say x, y) across the wafer.

As an example, Fig. 1 illustrates a fragment of the combinatorial library (A_XA' ι-_x) 2 (B_yBι-_y) Ci.ooOe, which is a special case (z=0, δ=0) of formula (1) .

Actually, it is not necessary to carry out the entire three-dimensional search, which would require 101 growths

15 for one elemental combination, if one could cover the full two-dimensional range (101x101 samples) in one growth. Instead, we first scan the third coordinate (say z) in coarser steps, of say 0.1. Then, we measure the physical property of interest - here, the Curie temperature T_c - and find the set xi, yi, zi for which T_c is the largest. Than we fix another index, say x=xι, and deposit a film with a spread m y, z. Next, we find the set xi, y₂, z₂ with the largest T_c . Finally, we fix y = y₂, and deposit a film with a spread x, z, and find the new maximum at x₂, y₂, z₃. This should already be quite close to the maximum T_c, since this variable is expected to be a relatively smooth function of the stoichiometry, and should not vary much if the latter is changed on the ±1% scale. The point here is that in this way a full three-dimensional search can be accomplished in less than 15 deposition runs, if each one would cover the full 101x101 two-dimensional matrix. With our current implementation (one 3 inch substrate holder, carrying one 4x4 matrix of 1 cm² substrates at a time) , we need about 9 runs for each z, i.e., roughly 100 deposition runs to accomplish one full three-dimensional combinatorial search. The entire library thus consists of about half million films, 1 cm^" each.

An important part of the optimization procedure consists of monitoring the surface stoichiometry, crystallographic structure, and morphology using the unique set of surface-analysis tools (scanning RHEED, LEEM, and TOF-ISARS) with which the COMBE machine is equipped. In this way, for each deposition, the wafer temperature T and the oxidation power (quantified e.g., in terms of the partial pressure of ozone p) are optimized to get the smoothest surface possible. This also sets the optimal value of δ- To double-check the latter, it is possible to perform a control deposition experiment, by scanning δ in a combinatorial way, for selected samples.

The procedure to generate the combinatorial library disclosed here could be reproduced by an experienced MBE

16 practitioner, provided one has the proper equipment. Using the current Oxxel COMBE system at its maximum capacity, the whole library can be synthesized in about 1-2 years. The largest-capacity manufacturing MBE machines commercially available today as a standard product, such as, e.g., the VG Semicon (England) Model V150 system, enable one to deposit films on three 6" wafers at the same time. Such a machine can produce the over 30,000 such large wafers in one year. With such or similar equipment, the entire library disclosed here can be reproduced in a few months.

On the other hand, the primary innovation of the present disclosure resides in (a) developing a novel apparatus and a novel synthesis technique, COMBE; (b) using it to produce a novel object, the combinatorial library of double-perovskite compounds of the present invention, that contains a very large number of novel and so far uninvestigated compositions, and (c) testing every sample from this library to determine its magnetic properties. This was not possible using prior art and techniques. Indeed, it would be possible to impose various further restrictions and reduce the size of the library, or to break it into several or many smaller sub-libraries, and synthesize only these; such variations are considered to be within the scope of the present invention.

B. Parallel Testing of Magnetic Properties As previously noted, a further object of this invention is to provide a method of fast, parallel magnetic characterization of the present combinatorial library of double-perovskite compounds.

Once the films with combinatorial spread of stoichiometry are generated as described above, and unloaded from the COMBE system, they are tested for ferromagnetism or ferrimagnetism, as follows. The substrate holder carrying a multitude (currently, 16) of substrates, is placed onto a metal block and thermally anchored. The block is placed in a cryostat or a

17 cryo-cooler, for measurements at cryogenic temperature; alternatively, for measurements at elevated temperature, it is mounted onto or inside a heater, or exposed to heating from the outside. At any fixed temperature, one measures the magnetic susceptibility or magnetization of the sample, by either a single scanning probe, or by a multi-probe array of magnetic sensors. Examples of such sensors are (1) magnetic force sensors similar to those used in magnetic force microscopy (MFM) , (2) the Hall-effect sensors, (3) smgle-coil (self-inductance) or double-coil (mutual inductance) susceptometers, (4) SQUID (superconducting quantum interference device) sensors, (5) magneto-optic imaging, etc.

The measurement is repeated at different temperatures, e.g., scanning the temperature systematically in small intervals (say 0.1-1 K) at a time, until the Curie temperature is identified for each pixel. A cruder but more practical method, aimed ust at identifying the pixels with the highest Curie temperature, is to make the magnetic measurement at just few selected temperatures Ti, T₂, , T_n, and then bisect the interval between the highest temperature Ti at which some pixels are still magnetic, and the next higher temperature T₁₊ι, and perform the next measurement at the temperature (T_x + T₁₊ι)/2, etc. One can proceed in this way until the maximal Curie temperature is determined to the desired accuracy.

Another test that we can perform is to determine the degree of spin polarization the candidate electrode material using the so-called Andreev reflection, which is manifested as a specific feature of the I-V characteristic tunneling between a ferromagnet and a superconducting tip. [18, 19] Yet another test is sp -polarized X-ray Photoemission Spectroscopy (XPS) , which is possible by virtue of ultra-high vacuum (UHV) character of the COMBE system, and in particular by the capability to transfer the sample UHV to the characterization chamber using the so- called vacuum suitcase. [20]

18 C. Fabrication and Testing of TMR Devices

Yet another object of this invention is to provide a method of fabrication and fast, parallel testing of combinatorial libraries of TMR devices, based on previous optimization of the magnetic electrode material as expounded above.

The next step in this combinatorial search for optimal TMR devices is ALL-MBE synthesis of tπ-layer heterostructures consisting of the bottom electrode magnetic layer Ml, the barrier layer B, and the top electrode magnetic layer M2. Alternative embodiments may contain additionally some or all of the following layers:

(a) a bottom magnetic anchoring layer, (b) a top magnetic anchoring layer, (c) a buffer layer to promote nucleation of layered (two-dimensional) growth, (d) a highly conductive top layer for improving the electrical contact to the top electrode, etc. The ma magnetic electrode layers Ml and M2 are synthesized out of magnetic double- perovskite compounds selected from the present library, and identified as optimal based on the highest Curie temperature, the degree of spin polarization, and conceivably other relevant physical or chemical properties.

The barrier material is selected from the known insulator perovskites such as SrTι0₃, CaTι0₃, DyTι0₃, etc. Other insulating oxides may also be used, particular those with well lattice-matched crystal structure, and which are chemically stable enough that there is little bulk ter-diffusion at the relevant growth temperature. Doping on selected sites can also be used to vary the lattice constant of the barrier material. In si tu monitoring analytical tools (RHEED, LEEM, TOF-ISARS), available the present Oxxel COMBE system, make it possible to observe and optimize the interface properties and ensure smooth heteroepitaxy . A combinatorial search is made to optimize the barrier thickness, as well as the thermodynamic parameters for the growth of this layer.

19 It is important to point out here the main difference between the ALL-MBE deposition method, on one hand, and other, faster but cruder, deposition methods such as sputtering, laser ablation, or chemical-vapor deposition, on the other. ALL-MBE is tuned to deposition of single- crystal films, to achieving perfect homo- and hetero- epitaxy, and atomically smooth film surfaces and hetero- interfaces, and to fabrication of precise multilayers and superlattices . It is thus the technique of choice when dealing with device structures with ultrathin (1-2 nm thick) barriers.

To this purpose, the deposited tri-layer or multilayer films are subsequently patterned into so-called c- axis or vertical transport devices, where the tunneling current is running trough the barrier in a mesa structure, perpendicular to the substrate plane. An example of such a device is schematically depicted in Fig. 2. A variety of circuit designs are possible. The simplest one is to have each device separated and having separate contacts allowing for a 4-point-contact c-axis transport measurement, as shown in Fig. 2. Indeed, more complex circuit implementations are also conceivable.

Next, chips containing such tunnel junction devices are placed in a multi-probe electrical transport tester, such as the 64 pogo-pin contact system already implemented at Oxxel, which allows for fast, parallel (64-channel) testing of a multitude of devices. An external magnetic field can be used to switch the polarization of the "free" electrode (e.g. the top electrode), while the polarization of the other electrode (e.g. the bottom electrode), is kept anchored in a fixed orientation. [This can be accomplished, e.g., by using a buffer layer made of a hard ferromagnet, which is polarized separately using a strong external field. ] The relative change of the electrical resistance ΔR is measured and recorded for each device, and displayed within a two-dimensional matrix. The devices with the best performance, i.e., the highest value of Δ

20 and the highest operational temperature, are identified in this way. At the same time, one can specify the optimal deposition parameters, the choice of the material for the electrodes, the barrier, and other layers, as well as the optimal layer thickness.

As an extension of this work, it is possible to fabricate stacks of two, three, or more such devices, on top of one another. The ALL-MBE capability of the COMBE system allows also fabrication of superlattices in which the ferromagnet layers are alternated with insulating layers. This is considered within the scope of the present invention, insofar as one uses the same ferromagnet compounds from the present library, and the same methodology. As previously noted, various aspects of the library synthesizing process, such as placement of the moveable shutters or control of deposition conditions, may be controlled using a processor-controlled device, such as a digital computer. Fig. 3 is a functional block diagram of a typical computer system that may be used to implement these aspects of the invention. As shown, this computer system includes a bus that interconnects a central processing unit (CPU) which represents processing circuitry such as a microprocessor, system memory comprised of various memory components such as random-access memory (RAM) and read-only memory (ROM) , and several device interfaces (i.e., controllers). Input controller represents interface circuitry that connects to one or more input devices such as a keyboard, mouse, track ball or the like. Display controller represents interface circuitry that connects to one or more display devices such as a video display device. I/O controller represents interface circuitry that connects to one or more I/O devices such as a modem or a network connection. Storage controller represents interface circuitry that connects to one or more storage devices such as a magnetic disk drive, magnetic tape drive, optical disk drive or solid-state storage

21 device. Printer controller represents interface circuitry that connects to one or more printer devices such as a laser printer, ink-jet printer or plotter. No particular type of computer system is critical to practice the various aspects of the present invention that may be computer- implemented. Any suitable computer system or processor- controlled device may be used.

In one embodiment, the computer system carries out the computer-controlled aspects of the present invention by using CPU to execute a program of instructions residing RAM that may be fetched from ROM, storage device or obtained from a network server or other source through I/O device. The program of instructions (i.e., software) may be conveyed by any medium that is readable by the computer or other suitable processor-controlled device. Such media include various magnetic media such as disks or tapes, various optical media such as compact discs, as well as various communication paths throughout the electromagnetic spectrum including base-band or broadband signals and a carrier wave encoded to transmit the program of instructions .

Alternatively, the computer-controlled aspects of the present invention may be implemented with functionally equivalent hardware using discrete components, application specific integrated circuits (ASICs), or the like. Such hardware may be physically integrated with the CPU or may be a separate element. Where the hardware is a separate element, it may be embodied m the computer itself or on a computer card that can be inserted into an available card slot in the computer.

Thus, the term "program-controlled processor" as used herein is intended to cover both a processor operating under the control of appropriate software and a device having the appropriate programming hardwired therein. While embodiments of the invention have been described, it will be apparent to those skilled in the art m light of the foregoing description that many further

22 alternatives, modifications and variations could be made to the above-described system. The invention described herein is intended to embrace all such alternatives, modifications, applications and variations as may fall within the spirit and scope of the appended claims.

23

Claims

I T I S CLAIMED :

1. A position-addressable combinatorial library of double-perovskite materials, comprising: a substrate having a substrate surface bounded by two or more pairs of opposite side regions, and defining a plurality of surface regions at known positions with respect to the side regions; and formed on the surface regions, a plurality of double- perovskite materials having the formulae (A_XA' i-_x) ₂ (B_yB' i- _y) (C_ZC i-z) Oε_+δ, wherein A is a 2+ cation, A' is a 3+ or 1+ cation, B and B' are each paramagnetic 3+ cations, C and C are each paramagnetic 5+ cations, x, y and z are whole or fractional numbers including zero, and δ represents a small deviation from ideal oxygen stoichiometry, and the stoichiometπes or structures of the double-perovskite materials vary systematically and a known way in at least one elemental component, progressing between one side region and another in a pair of opposite side regions.

2. The library of claim 1, wherein:

(I) A is selected from the group of elements consisting of Ba, Sr, Ca and Pb;

(II) A' is selected from the group of elements consisting of Ce, Na and K;

(III) B and B' are each selected from the group of elements consisting of Mn, Fe, Co, Cr and Ni; and

(IV) C and C are each selected from the group of elements consisting of Mo, W and Re.

3. The library of claim 1, wherein a first pair of elemental components varies systematically, progressing between one side region and another in a first pair of opposite side regions, and a second pair of elemental components varies systematically, progressing between one side region and another in a second pair of opposite side regions .

24

4. The library of claim 1, wherein the plurality of double-perovskite materials is formed on a film that is partitioned into a plurality of areas, each area corresponding to a respective one of the plurality of regions on the substrate surface and each area containing a square array of pixels, wherein x varies predetermined increments from pixel to pixel, progressing between one side region and another in a first pair of opposite side regions, and wherein y varies in the predetermined increments from pixel to pixel, progressing between one side region and another in a second pair of opposite side regions.

5. The library of claim 4, wherein the film is partitioned into a set of 16 areas arranged m a 4x4 matrix on the surface of the substrate, each film area corresponding to a respective one of the plurality of regions on the substrate surface, wherein x or y varies approximately 40% across the film.

6. The library of claim 1, wherein one of x, y and z is fixed, and the other two of x, y and z each vary systematically from 0 to 1 in predetermined increments, or in a quasi-continuous manner.

7. The library of claim 1, wherein x, y and z each vary systematically from 0 to 1 n predetermined increments, or in a quasi-continuous manner.

8. The library of claim 1, wherein a plurality of double-perovskite materials identified as optimal based on Curie temperature and degree of spin polarization are formed on first and second substrates to form first and second electrode magnetic layers, and a barrier layer comprised of an insulator perovskite is formed in between.

9. A method of identifying at least one double- perovskite material having a desired solid-state property,

25 comprising: forming, on a substrate having a substrate surface bounded by two or more pairs of opposite side regions and defining a plurality of surface regions at known positions with respect to the side regions, a library of double- perovskite materials having the formulae (A_XA' ι-_x) 2 (B_yB' _x- _v) (CC'i -006₊§, wherein A is a 2+ cation, A' is a 3+ or 1+ cation, B and B' are each paramagnetic 3+ cations, C and C are each paramagnetic 5+ cations, x, y and z are whole or fractional numbers including zero, and δ represents a small deviation from ideal oxygen stoichiometry, and the stoichiometries or structures of the double-perovskite materials vary systematically and in a known way at least one elemental component, progressing between one side region and another a pair of opposite side regions; testing the individual double-perovskite materials on the substrate for the desired solid-state property; and identifying the structure of each double-perovskite material having the desired property according to its position on the substrate and the known synthetic history of that substrate position.

10. The method of claim 9, wherein the desired solid- state property is ferromagnetism or ferπmagnetism.

11. The method of claim 10, wherein the testing is carried out at different temperatures to identify the Curie temperature for each double-perovskite material.

12. The method of claim 10, wherein the testing includes determining the degree of spin polarization of each double-perovskite material.

13. The method of claim 9, wherein a first variable of x, y and z is varied m a predetermined increment and the corresponding double-perovskite materials tested to identify a first subset of materials having a desired characteristic,

26 then a second variable of x, y and z is fixed and a first sublibrary of double-perovskite materials is formed varying the first and third variables and the first subliorary of materials is tested to identify a second subset of materials having the desired characteristic, then the third variable of x, y and z is fixed and a second sublibrary of double- perovskite materials is formed varying the first and second variables and the second sublibrary of materials is tested to identify a third subset of materials having the desired characteristic.

14. The method of claim 9, for use fabricating and testing a tunneling magneto-resistance (TMR) device comprised of first and second electrode magnetic layers formed with a plurality of double-perovskite materials identified as optimal based on Curie temperature and degree of spin polarization, and a barrier layer comprised of an insulator perovskite formed in between, wherein individual double-perovskite materials on the electrodes are tested to identify the double-perovskite material on the electrodes that optimizes thickness of the barrier layer and thermodynamic parameters for growth of the barrier layer.

15. The method of claim 14, wherein a plurality of TMR devices are tested in parallel by: applying a magnetic field to switch the polarization of the first electrode of each device while maintaining the polarization of the second electrode of each device in a fixed orientation; measuring the relative change of electrical resistance for each device; and identifying each device exhibiting one or both of the highest change of electrical resistance and the highest operational temperature.

16. A method of producing a combinatorial library of double-perovskite materials, comprising:

27 placing, in a vacuum chamber containing a plurality of elemental sources with shutters, a substrate having a substrate surface bounded by two or more pairs of opposite side regions, and defining a plurality of surface regions at known positions with respect to the side regions; depositing successive atomic layers on the substrate surface regions in the presence of a gaseous source of oxygen atoms; during said depositing, varying deposition conditions in a known and systematic way for at least one process variable, progressing between one side region and another in a pair of opposite side regions, said process variable being selected from the group consisting of:

(i) the quantity of at least one elemental component deposited from one or more elemental sources;

(ii) the sequence of deposition from the elemental sources;

(iii) the temperature of the substrate regions; (iv) the partial pressure of the source of atomic oxygen at the substrate regions; and

(v) the thickness of barrier layers in multilayered structures; and producing, by said depositing, a plurality of double- perovskite materials having the formulae (A_XA' ι-_x) ₂ (B_yB' i- _y) (C_zC'ι-_z)θ6_±(1, wherein A is a 2+ cation, A' is a 3+ or 1+ cation, B and B' are each paramagnetic 3+ cations, C and C are each paramagnetic 5+ cations, x, y and z are whole or fractional numbers including zero, and δ represents a small deviation from ideal oxygen stoichiometry, and the stoichiometries or structures of the double-perovskite materials vary systematically and in a known way in at least one elemental component, progressing between one side region and another in a pair of opposite side regions.

17. The method of claim 16, wherein varying the process variable of the quantity of material deposited from an elemental source includes placing between the source and

28 the target, a sieve mask effective to produce a desired gradient of atomic flux through the mask, progressing between one side region and another in a pair of opposite side regions of the substrate.

18. The method of claim 16, wherein varying the process variable of the quantity of material deposited from an elemental source includes placing between the source and the target, a sieve mask effective to produce a uniform atomic flux through the mask.

19. The method of claim 16, wherein varying the process variable of the quantity of material deposited from an elemental source includes placing a movable shutter between the source and the target, at each of a plurality of selected positions.

20. The method of claim 16, wherein varying the temperature of the substrate regions includes heating the substrate, or a set or a matrix of substrates, so as to produce a temperature gradient, progressing between one side region and another in a pair of opposite side regions of the substrate, or from one substrate to another withm the matrix.

21. The method of claim 16, wherein varying the partial pressure of the source of atomic oxygen at substrate regions includes directing ozone from a source positioned at one side of the substrate holder, at a nearly grazing angle with respect to the substrate surface.

22. The method of claim 16, wherein the deposition conditions for at least one process variable is varied under the control of a program-controlled processor.

23. The method of claim 19, wherein the placing of the moveable shutter between the source and the target at each

29 of the plurality of selected positions is controlled by a program-controlled processor.

24. The method of claim 16, wherein the combinatorial library of double-perovskite materials is produced with a combinatorial molecular beam epitaxy apparatus.

30