Quantum tunneling devices and circuits with lattice-mismatched semiconductor structures

Abstract

Structures include a tunneling device disposed over first and second lattice-mismatched semiconductor materials. Process embodiments include forming tunneling devices over lattice-mismatched materials.

Claims

1. A device comprising: an analog-to-digital converter comprising: a first voltage divider, an input node of the first voltage divider being electrically coupled to a circuit analog input node, a first multi-state resonant tunneling bipolar transistor having a first base, a first collector, and a first emitter, a voltage-divided node of the first voltage divider being electrically coupled to the first base, the first collector being a first output and being electrically coupled to a first node of a first impedance element, a second node of the first impedance element being electrically coupled to a first power node, the first emitter being electrically coupled to a second power node, a second voltage divider, an input node of the second voltage divider being electrically coupled to the circuit analog input node, and a second multi-state resonant tunneling bipolar transistor having a second base, a second collector, and a second emitter, a voltage-divided node of the second voltage divider being electrically coupled to the second base, the second collector being a second output and being electrically coupled to a first node of a second impedance element, a second node of the second impedance element being electrically coupled to the first power node, the second emitter being electrically coupled to the second power node; wherein: the first multi-state resonant tunneling bipolar transistor is on a first crystalline material, the first crystalline material being on a substrate crystalline material of a substrate, the first crystalline material being lattice-mismatched to the substrate crystalline material, a majority of defects in the first crystalline material arising from the first crystalline material being lattice-mismatched to the substrate crystalline material terminating at sidewalls of the first crystalline material, and the second multi-state resonant tunneling bipolar transistor is on a second crystalline material, the second crystalline material being on the substrate crystalline material, the second crystalline material being lattice-mismatched to the substrate crystalline material, a majority of defects in the second crystalline material arising from the second crystalline material being the lattice-mismatched to the substrate crystalline material terminating at sidewalls of the second crystalline material.

2. The device of claim 1, wherein the first power node is a power supply node, and the second power node is a ground node.

3. The device of claim 1, wherein: the first voltage divider comprises a first resistor and a second resistor, a first node of the first resistor being the input node of the first voltage divider, a second node of the first resistor being the voltage-divided node of the first voltage divider and being electrically coupled to a first node of the second resistor, a second node of the second resistor being electrically coupled to the second power node, and the second voltage divider comprises a third resistor and a fourth resistor, a first node of the third resistor being the input node of the second voltage divider, a second node of the third resistor being the voltage-divided node of the second voltage divider and being electrically coupled to a first node of the fourth resistor, a second node of the fourth resistor being electrically coupled to the second power node, the first resistor and the third resistor having a same resistance value, the second resistor and the fourth resistor having a different resistance value.

4. The device of claim 1, wherein the sidewalls of the first crystalline material are defined by first sidewalls of a dielectric material, and the sidewalls of the second crystalline material are defined by second sidewalls of the dielectric material.

5. The device of claim 1, wherein each of the first multi-state resonant tunneling bipolar transistor and the second multi-state resonant tunneling bipolar transistor comprises a stack of dual tunnel structures.

6. The device of claim 1, wherein each of the first multi-state resonant tunneling bipolar transistor and the second multi-state resonant tunneling bipolar transistor comprises: a first n+ doped layer over the first crystalline material or the second crystalline material, respectively, an contact being on the first n+ doped layer, a first n doped layer over the first n+ doped layer, a p+ doped layer over the first n doped layer, a base contact being on the p+ doped layer, a second n doped layer over the p+ doped layer, a first tunnel structure over the second n doped layer, a second n+ doped layer over the first tunnel structure, a second tunnel structure over the second n+ doped layer, and a third n+ doped layer over the second tunnel structure, an emitter contact being on the third n+ doped layer.

7. The device of claim 6, wherein: the first tunnel structure comprises: a first layer of a first undoped material over the second n doped layer, a second layer of a second undoped material over the first layer, and a third layer of the first undoped material over the second layer, and the second tunnel structure comprises: a fourth layer of a third undoped material over the second n+ doped layer, a fifth layer of a fourth undoped material over the fourth layer, and a sixth layer of the third undoped material over the fifth layer.

8. A device comprising: a comparator comprising: a first resonant tunneling diode (RTD) having a first anode and a first cathode, the first anode being electrically coupled to a first power node, the first cathode being electrically coupled to an output node, a first high electron mobility transistor (HEMT) having a first gate, a first source/drain, and a second source/drain, the first gate being electrically coupled to an input node, the first source/drain being electrically coupled to the output node, a second RTD having a second anode and a second cathode, the second anode being electrically coupled to the second source/drain, the second cathode being electrically coupled to a second power node, and a second HEMT having a second gate, a third source/drain, and a fourth source/drain, the second gate being electrically coupled to a clock node, the third source/drain being electrically coupled to the first power node, the fourth source/drain being electrically coupled to the output node; wherein: the first RTD is on a first crystalline material, the first crystalline material being on a substrate crystalline material of a substrate, the first crystalline material being lattice-mismatched to the substrate crystalline material, a majority of defects in the first crystalline material arising from the first crystalline material being lattice-mismatched to the substrate crystalline material terminating at sidewalls of the first crystalline material, and the second RTD is on a second crystalline material, the second crystalline material being on the substrate crystalline material, the second crystalline material being lattice-mismatched to the substrate crystalline material, a majority of defects in the second crystalline material arising from the second crystalline material being lattice-mismatched to the substrate crystalline material terminating at sidewalls of the second crystalline material.

9. The device of claim 8 further comprising a dielectric layer over the substrate, the dielectric layer having a first opening and a second opening, the first crystalline material being in the first opening, sidewalls of the first opening defining the sidewalls of the first crystalline material, the second crystalline material being in the second opening, sidewalls of the second opening defining the sidewalls of the second crystalline material.

10. The device of claim 8, wherein a distal portion of the first crystalline material distal from the substrate has a dislocation defect density of less than 10.sup.6 cm.sup.2, and a distal portion of the second crystalline material distal from the substrate has a dislocation defect density of less than 10.sup.6 cm.sup.2.

11. The device of claim 8, wherein a distal portion of the first crystalline material distal from the substrate has a dislocation defect density of less than 10.sup.6 cm.sup.2, and a distal portion of the second crystalline than 10.sup.3 cm.sup.2.

12. The device of claim 8, wherein each of the first RTD and the second RTD comprises: a first doped layer of a first material over the first crystalline material or the second crystalline material, respectively, a first undoped layer of the first material over the first doped layer, a first barrier layer over the first undoped layer, a second undoped layer of the first material over the first barrier layer, a second barrier layer over the second undoped layer, a third undoped layer of the first material, and a second doped layer of the first material over the third undoped layer.

13. The device of claim 8 further comprising: a first contact on a top portion of the first RTD; a second contact on the first crystalline material; a third contact on a top portion of the second RTD; and a fourth contact on the second crystalline material.

14. A device comprising: a clock-generator comprising: a first resonant tunneling diode (RTD) having a first anode and a first cathode, the first anode being electrically coupled to an output node, the first cathode being electrically coupled to a first power node, a second RTD having a second anode and a second cathode, the second anode being electrically coupled to the output node, the second cathode being electrically coupled to a second power node, a third RTD having a third anode and a third cathode, the third anode being electrically coupled to the first power node, the third cathode being electrically coupled to a complementary output node, and a fourth RTD having a fourth anode and a fourth cathode, the fourth anode being electrically coupled to the second power node, the fourth cathode being electrically coupled to the complementary output node; wherein each of the first RTD, the second RTD, the third RTD, and the fourth RTD is on a respective first crystalline material, the first crystalline material being on a substrate crystalline material of a substrate, the first crystalline material being lattice-mismatched to the substrate crystalline material, a majority of defects in the first crystalline material arising from the lattice-mismatch to the substrate crystalline material terminating at sidewalls of the first crystalline material.

15. The device of claim 14 further comprising a dielectric layer over the substrate, the dielectric layer having respective openings, the first crystalline material being in the respective openings, sidewalls of the respective openings defining the sidewalls of the first crystalline material.

16. The device of claim 14, wherein respective distal portions of the first crystalline material distal from the substrate has a dislocation defect density of less than 106 cm-2.

17. The device of claim 14, wherein each of the first RTD, the second RTD, the third RTD, and the fourth RTD comprises: a first doped layer of a first material over the respective first crystalline material, a first undoped layer of the first material over the first doped layer, a first barrier layer over the first undoped layer, a second undoped layer of the first material over the first barrier layer, a second barrier layer over the second undoped layer, a third undoped layer of the first material, and a second doped layer of the first material over the third undoped layer.

Description

BRIEF DESCRIPTION OF THE FIGURES

(1) FIG. 1 is a cross-sectional view of an RTD or RITD device structure, prior art;

(2) FIG. 2 is a perspective view of a double electron layer tunnel transistor (DELTT), prior art;

(3) FIG. 3 illustrates a circuit of a tunnel diode SRAM and its associated IV curve, prior art;

(4) FIG. 4 is a cross-sectional view of a GaAs/AlGaAs RTD formed on a Si substrate with the use of ART in accordance with an embodiment of the invention;

(5) FIG. 5 is a cross-sectional view of an RTD formed with the use of ART and including III-V materials in accordance with an embodiment of the invention;

(6) FIGS. 6a-6b is a cross-sectional view of an RTD formed with the use of ART and including contacts thereto in accordance with an embodiment of the invention;

(7) FIGS. 7a-7b and 8a-8c are cross-sectional views of device structures with non-flat top surfaces formed with the use ART in accordance with embodiments of the invention;

(8) FIG. 9 is a cross-sectional view of a triple-barrier RTD formed with the use of ART in accordance with an embodiment of the invention;

(9) FIGS. 10a-10e are cross-sectional views of five different RITD structures formed with the use of ART in accordance with embodiments of the invention;

(10) FIG. 11 is a cross-sectional view of a magnetic resonant tunneling diode as a spin selector for spintronic applications in accordance with an embodiment of the invention;

(11) FIGS. 12a-12b are a pair of cross-sectional views illustrating a process for forming an RTFET in accordance with an embodiment of the invention;

(12) FIGS. 13a-13b are cross-sectional views of a QWBRTT formed with the use of ART in accordance with an embodiment of the invention;

(13) FIGS. 14a-14c are cross-sectional views of tunneling quantum structures integrated within a bipolar transistor, formed with the use of ART in accordance with embodiments of the invention, as well as the corresponding energy-band diagrams;

(14) FIG. 15 is a cross-sectional view of a three-terminal bipolar quantum resonant tunneling transistor formed over GaAs grown by ART on a Si substrate in accordance with an embodiment of the invention;

(15) FIG. 16a is a cross-sectional view of a multi-state RTBT formed by ART in accordance with an embodiment of the invention; FIG. 16b is an energy-band diagram of the device of FIG. 16a;

(16) FIGS. 17a-17b are a cross-sectional view of an RHET formed by ART and its energy-band diagram in accordance with an embodiment of the invention;

(17) FIG. 18 is a cross-sectional view of a typical DELTT structure disposed over GaAs material formed by ART in accordance with an embodiment of the invention;

(18) FIG. 19 is a cross-sectional view of a building block for various logic circuit units in accordance with an embodiment of the invention;

(19) FIGS. 20a-20c and 21 are diagrams illustrating circuit units incorporating RTD formed by ART in accordance with embodiments of the invention;

(20) FIG. 22 is a cross-sectional view of an SRAM structure including two RTDs and an n-FET in accordance with another embodiment of the invention;

(21) FIGS. 23a-23e, 24a-24e, and 25 are diagrams illustrating SRAMs incorporating RTD devices formed in accordance with embodiments of the invention;

(22) FIG. 26 is a cross-sectional view of a non-volatile memory cell including an RTD and one memory storage material (such as phase change material or ferroelectric material) in accordance with another embodiment of the invention;

(23) FIG. 27 is a cross-sectional view of a non-volatile magnetic random access memory including an RITD and a magnetic tunneling structure in accordance with another embodiment of the invention;

(24) FIG. 28 is a diagram of a monostable-bistable transition logic element including an RTD formed by RTD in accordance with an embodiment of the invention;

(25) FIG. 29 is a diagram of an analog-to-digital converter circuit employing multi-state RTBT devices formed by ART in accordance with an embodiment of the invention;

(26) FIG. 30 is a diagram of a RTD/HEMT comparator incorporating an RTD into a HEMT formed by ART in accordance with an embodiment of the invention;

(27) FIG. 31 is a diagram of an RTD-based clock generator formed by ART in accordance with an embodiment of the invention;

(28) FIG. 32 is a diagram of a field programmable gate array incorporating RTD-based SRAMS formed by ART in accordance with an embodiment of the invention;

(29) FIGS. 33 and 34 are block diagrams illustrating RTD-based or RTT-based microprocessors formed by ART in accordance with embodiments of the invention;

(30) FIG. 35 is a block diagram of an RTD-based digital signal processor formed by ART in accordance with an embodiment of the invention;

(31) FIG. 36 is a block diagram illustrating an RTD-based graphics processing unit formed by ART in accordance with an embodiment of the invention; and

(32) FIG. 37 is a cross-sectional view of a device including two RTDs connected to a drain of a MOSFET, formed by ART in accordance with an embodiment of the invention.

DETAILED DESCRIPTION OF THE INVENTION

(33) Although the following examples are illustrated with particular III-V material systems for tunneling device on Si substrates, embodiments of this invention may include other materials systems, such as SiGe, Ge, III-V and/or II-VI. The described structures may be fabricated on various semiconductor substrates, such as Si substrates, Ge substrates, semiconductor-on-insulator (SOI) substrates, strained-semiconductor-on-insulator (SSOI) substrates, and other substrate systems. Embodiments of the invention also include other similar tunneling device structures and circuit units. Structures may include various tunneling devices and various circuit units; methods are described herein for producing such structures or circuit units on a first type of substrate, e.g., from group IV such as a Si or Ge substrate, with improved material systems, e.g., III-V and II-VI epitaxial materials. The III-V or II-VI epitaxial layers may be grown selectively on conventional substrates such as Si, SOI, or SSOI substrates in selected areas (such as the device active area). Such layers may be formed by a heteroepitaxy technique, e.g., by ART heteroepitaxial techniques, as described in pending U.S. patent application Ser. Nos. 11/436,198 and 11/436,062.

(34) Cross-sectional TEM micrographs were taken of an experimental sample of lattice-mismatched semiconductor material grown within dielectric windows on a Si substrate by the ART heteroepitaxial technique. The micrographs demonstrated that defects, e.g., dislocations, are trapped by the dielectric sidewalls, resulting in high quality material suitable for device fabrication.

I. Individual Devices

(35) Referring to FIG. 4, the use of an ART technique enables the integration of low-defect III-V or II-VI materials on a Si substrate for tunneling device 400 fabrication. An exemplary RTD device structure is illustrated in FIG. 5. After a window is opened in a dielectric layer over a substrate 405, ART is used to grow a low-defect high-quality semiconductor material 410 (e.g., GaAs) on a Si surface, such that defects 415, e.g., dislocations and stacking faults, are trapped by the dielectric side walls 420, resulting a low-defect upper region of GaAs material. Next, multiple layers 425 of GaAs and AlGaAs are grown on the low-defect GaAs material by epitaxy. These layers may be used to fabricate an RTD device. This RTD device may be contacted via first contact 430 and second contact 435. In some embodiments, the first contact 430 may be the anode terminal and the second contact 435 may be the cathode terminal of tunneling device 400, and in other embodiments, the first contact 430 may be the cathode terminal and the second contact 435 may be the anode terminal.

(36) A detailed preferred fabrication process is as follows. The substrate 405 may be, for example, a bulk silicon wafer, a bulk germanium wafer, a semiconductor-on-insulator (SOI) substrate, or a strained semiconductor-on-insulator (SSOI) substrate. The substrate 405 may include or consist essentially of a first semiconductor material, such as a group IV element, e.g., germanium or silicon. In an embodiment, the substrate 405 includes or consists essentially of (100) silicon.

(37) A non-crystalline material, e.g., a dielectric layer 440, is formed over the semiconductor substrate 405. The dielectric layer 440 may include a dielectric material, such as silicon nitride or silicon dioxide. The dielectric layer 440 may be formed by a method known to one of skill in the art, e.g., thermal oxidation or plasma-enhanced chemical vapor deposition. As discussed below, the dielectric layer 440 may have a thickness t.sub.1 corresponding to a desired height of crystalline material to be deposited in a window 445 formed through the dielectric layer 440. In some embodiments, the thickness t.sub.1 of the dielectric layer 440 may be selected from a range of, e.g., 20-50000 nm.

(38) A mask (not shown), such as a photoresist mask, is formed over the substrate 405 and the dielectric layer 440. The mask is patterned to expose at least a portion of the dielectric layer 440. The exposed portion of the dielectric layer 440 is removed by, e.g., reactive ion etching (RIE) to define the window 445. The window 445 extends to a surface of the substrate 405 and may be defined by at least one sidewall 420. The sidewall 420 is formed from the dielectric layer 440 and is, therefore, non-crystalline. The sidewall 420 may have a height h at least equal to a predetermined distance H from the surface of the substrate 405. It has been observed experimentally that dislocations 415 in a mismatched cubic semiconductor grown on a Si (100) surface in the near-vicinity (e.g., within approximately 500 nm or less) of a vertical dielectric sidewall 420 surface bend toward that surface at approximately 30 degrees through 60 degrees. For example, the dislocations 415 may bend toward that surface at approximately a 45 degree angle to that surface. Based on this relationship, one may typically expect the predetermined distance H necessary to trap defects 415 to be approximately equal to a width between w and 2 w, where w is the width of the window 445.

(39) The window 445 may be substantially rectangular in terms of cross-sectional profile, a top view, or both, and have a width w that is smaller than a length l (not shown) of the window. For example, the width w of the window may be less than about 5000 nm, e.g., about 20-1000 nm. In some embodiments, the width of the window is about 150 nm. A length l of the window may exceed each of w and H. A ratio of the height h of the window to the width w of the window may be >1, preferably between about 1 and about 50.

(40) A second crystalline semiconductor material 410 is formed in the window. The second crystalline semiconductor material 410 may include or consist essentially of a group IV element or compound, a III-V compound, or a II-VI compound. Examples of suitable group IV elements or compounds include germanium, silicon germanium, and silicon carbide. Examples of suitable III-V compounds include gallium arsenide, gallium nitride, indium arsenide, indium antimonide, indium aluminum antimonide, indium aluminum arsenide, indium phosphide, and indium gallium arsenide. Examples of suitable II-VI compounds include zinc selenide and zinc oxide.

(41) The second crystalline semiconductor material 410 may be formed by selective epitaxial growth in any suitable epitaxial deposition system, including, but not limited to, metal-organic chemical vapor deposition (MOCVD), atmospheric-pressure CVD (APCVD), low- (or reduced-) pressure CVD (LPCVD), ultra-high-vacuum CVD (UHCVD), molecular beam epitaxy (MBE), or atomic layer deposition (ALD). In the CVD process, selective epitaxial growth typically includes introducing a source gas into the chamber. The source gas may include at least one precursor gas and a carrier gas, such as, for example, hydrogen. The reactor chamber may be heated by, for example, RF-heating. The growth temperature in the chamber may range from about 300 C. to about 900 C., depending on the composition of the epitaxial region. The growth system may also utilize low-energy plasma to enhance the layer growth kinetics. CVD has a number of advantages, including the capability for depositing films with low defect densities and rapidly varying alloy compositions, as well as high quality regrowth capability. CVD may also provide improved manufacturability due to relatively higher throughput, relatively short downtimes, and scalability to very large reactors.

(42) The epitaxial growth system may be a single-wafer or multiple-wafer-batch reactor. Suitable CVD systems commonly used for volume epitaxy in manufacturing applications include, for example, an Aixtron 2600 multi-wafer system available from Aixtron, based in Aachen, Germany; an EPI CENTURA single-wafer multi-chamber systems available from Applied Materials of Santa Clara, Calif.; or EPSILON single-wafer epitaxial reactors available from ASM International based in Bilthoven, The Netherlands.

(43) Dislocation defects 415 in the second crystalline semiconductor material 410 reach and terminate at the sidewalls 420 of the window in the dielectric material 440 at or below a vertical predetermined distance H from the surface of the substrate, such that dislocations 415 in the second crystalline semiconductor material 410 decrease in density with increasing distance from the bottom portion of the window. Accordingly, the upper portion of the crystalline material 410 is substantially exhausted of dislocation defects 415. Various dislocation defects 415 such as threading dislocations, stacking faults, twin boundaries, or anti-phase boundaries may thus be substantially eliminated from the upper portion of the epitaxial region. A density of such dislocation defects 415 may be less than, for example, 10.sup.6/cm.sup.2, preferably less than 10.sup.3/cm.sup.2. The second crystalline semiconductor material 410 may be either substantially relaxed or strained.

(44) A high-quality tunneling device 400 may be fabricated on top of the low-defect second crystalline semiconductor material 410. The tunneling device may be, for example, an RTD including GaAs and AlGaAs. The RTD device 400 may include multiple thin semiconductor layers 425 such as, from bottom to top, a relatively thick n-type layer 446 (e.g., n-type gallium arsenide with a thickness of, e.g., 10-25000 nm); an undoped layer 447 (e.g., undoped gallium arsenide with a thickness of, e.g., 0-50 nm); a thin barrier layer with a large bandgap 448 (e.g., undoped aluminum gallium arsenide with a thickness of, e.g., 0-50 nm); a thin two-dimensional resonant electron layer 449 (e.g., undoped gallium arsenide with a thickness of, e.g., 0-50 nm); another thin barrier layer with large bandgap 450 (e.g., undoped aluminum gallium arsenide with a thickness of, e.g., 0-50 nm); another undoped layer 451 (e.g., undoped gallium arsenide with a thickness of, e.g., 0-50 nm); and another relatively thick n-type layer 452 (e.g., n-type gallium arsenide, 10-25000 nm). All of these layers 425 may be grown epitaxially in sequence in any suitable epitaxial deposition system, including, but not limited to, MOCVD, APCVD, LPCVD, UHCVD, MBE, or ALD. This exemplary RTD device has two external contacts: a first contact 430 contacts the n-type layer 446, and a second contact 435 contacts the n-type layer 452. Both contacts 430 and 435 typically include metal contacts. The first contact 430 may be made by etching a small hole (not shown) on one side of the device to reach the layer 446. Methods and variations for the fabrication of this RTD 400 device will be apparent to one of skill in the art.

(45) In the above examples, the layers 425 of the RTD include or consist essentially of AlGaAs and GaAs. Thus, the substrate may include a group IV element such as Si, and the RTD may include a III-V compound. In some embodiments, the RTD may include a II-VI compound.

(46) Although only one exemplary RTD is illustrated in FIG. 4, embodiments of the invention also include other varieties of resonant tunneling device structures with a variety of semiconductor material combinations, layer structures, layer thicknesses, doping levels in each layer, epitaxy growth conditions (e.g., precursors, growth temperature, etc), and contact configurations.

(47) For example, referring to FIG. 5, the layers 425 of RTD may include various combinations of pairs of III-V materials. The layers may include a heavily doped materials 1 layer 546, an undoped materials 1 (spacer) layer 547, an undoped materials 2 barrier layer 548, an undoped materials 1 (2D resonant electron layer) layer 549, a second undoped materials 2 barrier layer 550, an undoped material 1 spacer layer 551, and a heavily doped materials 1 layer 552, where materials 1/materials 2 combinations may be, e.g., GaAs/AlGaAs, GaAs/InGaAs, GaInAs/AlInAs, GaAs/AlAs, InP/InGaAs, InGaAs/AlAs, InAs/AlSb, or InSb/AlInSb. Other material systems may also be used. In addition, the two doped layers 546 and 552 may include different materials. The 2D resonant electron layer 549 may be undoped and may have a thickness of 5 nm, and the barrier layers 548, 550 may be undoped and may have a thickness of 1 to 5 nm. The doped layers 546 and 552 may be, e.g., n-type with a doping level of 210.sup.18 to 110.sup.19/cm.sup.3.

(48) Referring to FIGS. 6a and 6b, a contact to a tunneling device 400 may be formed by the following approach. The steps in this approach are also applicable to other tunneling devices and structures of other embodiments of the inventions described herein. In a first step, III-V heteroepitaxial layers 410 are grown selectively in a window 445 defined by a dielectric layer 440 disposed over a semiconductor substrate 405. Such dielectric windows may be utilized in some heteroepitaxial growth technologies such as ART technology. In a second step, the dielectric 440 is etched back, e.g., approximately halfway back to form a recess structure. The etch-back of the dielectric 440 partially exposes the layers 425 within the dielectric window, allowing, e.g., a first contact 430 to also be made to a bottom layer disposed in the dielectric window (e.g., to the doped layer 546 in FIG. 6b).

(49) In most of the drawings discussed herein, the semiconductor epitaxial layers of the tunneling device structures are depicted with flat layers, for the sake of simplicity. Nevertheless, those drawings should also be considered to encompass structures with non-flat layers. One experimental sample of Ge material grown in silicon oxide windows on a Si substrate by ART technique illustrates that the surface of a semiconductor epitaxial layer is typically non-flat, due to faceting during the epitaxial growth. Therefore, the simplified illustrations such as FIG. 5 should be interpreted as including variations having non-flat epitaxial layers, as depicted in FIG. 7a. Another approach is depicted in FIG. 7b, where some layers 425 of the device structure are grown within the dielectric windows and are non-flat, while other layers are grown above the dielectric layer 440.

(50) Referring to FIGS. 8a-8c, another embodiment of an epitaxial layer and device structure is formed as follows. In the first step, FIG. 8a, III-V material 410 is grown within and above dielectric windows 445 by ART technique, this III-V material 410 having a non-flat surface due to faceting. In the second step, FIG. 8b, the structure is planarized to achieve a flat surface, for example by CMP using the dielectric layer 440 as a polishing stop. In the third step (FIG. 8c), epitaxial device layers 425 are grown on the flat surface. Contacts to both doped layers 546 and 552 may be made, by, e.g., first and second contacts 430 and 435, as shown in FIG. 8c, as both doped layers 546 and 552 are disposed above the dielectric layer 440 and are at least partially exposed.

(51) Although only one exemplary device structure is illustrated in both FIGS. 7 and 8a-8c, these approaches and methods are also applicable to other tunneling devices and structures of other embodiments of the inventions described herein.

(52) In addition to the basic RTD structures discussed above, some other exemplary tunneling devices are further described below. In some embodiments, the structures described below may be fabricated on III-V or II-VI epitaxial layers that are grown selectively on conventional substrates such as Si, SOI, or SSOI substrates in selected areas (such as the device active area). Such layers may be formed by heteroepitaxy, such as with ART techniques.

(53) Referring to FIG. 9, a triple-barrier tunneling diode includes three barriers 548, 550, 550 and two 2D resonant electron layers 549, 549. This structure gives rise to multiple current peaks on I-V characteristics. A multiple-barrier tunneling diode may be formed by incorporating more than three barrier layers separating multiple 2D resonant electron layers (not shown).

(54) Typical RTDs are often intra-band tunneling diodes. A variation of a tunneling diode is the RITD device. FIGS. 10a-10e illustrate five different RITD structures 1000. In each example, the III-V epitaxial layers 410, e.g., n.sup.+-InAs, are grown on a semiconductor substrate 405, e.g., an Si wafer, in the device area by heteroepitaxy. Referring to FIG. 10a, an RITD 1000 may include an ohmic contact 1005 and an injecting junction 1010 that is a polytype heterostructure, i.e., including a barrier layer 1020, 1020 (e.g., i-AlSb layer having a thickness selected from a range of about 1.5 nm to 3 nm) inserted in a type-III (or so-called broken-gap) heterojunction, including e.g., a p-GaSb layer 1030 and an n-InAs layer 1040. Referring to FIG. 10b, an RITD 1000 may include a type-III (or broken-gap) heterojunction including, e.g. a p-GaSb layer 1030 and an n-InAs layer 1040.

(55) Referring to FIG. 10c, an RITD 1000 may include two delta-doped layers n.sup.+, and p.sup.+ 1050, 1060. The delta-doped layer sheet charges are typically on the order of 110.sup.13/cm.sup.2. With delta-doped layers 1050, 1060, a single material may be used to make the diode, e.g., intrinsic 1061 and p.sup.+ doped 1062. Referring to FIG. 10(d), an RITD 1000 may include three delta-doped layers, e.g., n.sup.+, .sup.+, and n.sup.+ 1050, 1060, 1050, between two intrinsic layers 1061 and an n.sup.+ layer 1063. Referring to FIG. 10(e), an RITD 1000 may include a double-quantum-well heterostructure with layers including, e.g., n.sup.+-InAlAs 1070, i-InGaAs 1075, i-InAlAs 1080, i-InGaAs 1085, and p.sup.+-InAlAs 1090.

(56) One advantage of RITDs is a high peak-to-valley current ratio (PVCR). Among the illustrated devices, the RITD 1000 with double-quantum-well heterostructure (FIG. 10(e)) exhibits the highest PVCR.

(57) Referring to FIG. 11, an example of a magnetic resonant tunneling diode 1100 as a spin selector for spintronics applications is illustrated. It is fabricated in two phases. The first phase includes heteroepitaxy growth of a buffer layer 410 of either GaAs or ZnBeSe material using ART techniques to achieve high-quality low defect material. The second phase includes fabricating a typical magnetic resonant tunneling diode 1100. The magnetic resonant tunneling diode 1100 includes layer structure 425 that may include, for example, Zn.sub.0.97Be.sub.0.03Se layer (300 nm thick, n-type doping at 810.sup.18/cm.sup.3) 1105, a first ZnSe layer (100 nm thick, n-type doping at 1.510.sup.19/cm.sup.3) 1110, Zn.sub.0.97Be.sub.0.03Se layer (10 nm thick, n-type doping at 110.sup.18/cm.sup.3) layer 1115, a second ZnSe (10 nm, intrinsic) layer 1120, Zn.sub.0.7Be.sub.0.3Se layer (5 nm thick, intrinsic) 1125, Zn.sub.0.96Mn.sub.0.04Se layer (9 nm thick, intrinsic) 1130, Zn.sub.0.7Be.sub.0.3Se layer (5 nm thick, intrinsic) 1135, a third ZnSe layer (10 nm thick, intrinsic) 1140, Zn.sub.0.97Be.sub.0.03Se layer (15 nm thick, n-type doping at 110.sup.18/cm.sup.3) 1145, and a fifth ZnSe layer (30 nm thick, n-type doping at 1.510.sup.19/cm.sup.3) 1150. Metal contacts 1160 are formed to the first ZnSe layer 1110 and the fifth ZnSe layer 1150. With magnetic Mn incorporated into the quantum well material in the magnetic resonant tunneling diode 1100, a strong splitting of well resonance is obtained as a function of external magnetic field. The device 400 can thus be used as a spin selector.

(58) Referring to FIGS. 12 to 17, besides various 2-terminal tunneling diodes, a number of 3-terminal transistor devices incorporate quantum tunneling structures. The advantages of 3-terminal tunneling devices over the 2-terminal tunneling diodes may include: (i) providing a high-speed tunable negative differential resistance/transconductance, and (ii) providing isolation between the input and output.

(59) Referring to FIGS. 12a-12b, an exemplary fabrication process for forming an RTFET on ART semiconductor material is as follows. Multiple layers 425 are grown by heteroepitaxy, e.g., by an ART process, to form a quantum tunneling device structure 400. The fabrication process and materials used may be similar to those discussed with respect to FIG. 5, and include a source 1200 and a drain 1205. The dielectric layer 440 is at least partially etched back, e.g., halfway, and then a gate structure 1210 is formed to control the device. The gate 1210 is typically either a metal contact (Schottky diode), or a p-n junction. The function of the gate is to control the depletion width to vary the net cross-section area of the diode.

(60) Devices may have one of two types of gate structures. In a two-sided gate structure, the gate material may form on two opposite sides of the diode, resulting in a 1-D resonant-tunneling device, or quantum wire device. In a four-sided gate structure, the gate material may form around all sides of the diode, resulting in a 0-D resonant-tunneling device, or quantum dot (or single-electron) transistor.

(61) Referring to FIGS. 13a-13b, a QWBRTT is an ultrafast device in which carrier conduction from an emitter 1300 to a collector 1305 takes place via a tunneling structure. It is essentially an RTD device with a third terminal (a base) 1310 that contacts the quantum well to control its potential. Two types of QWBRTT include a bipolar device with p-type base that has a doping type (e.g., p-type) opposite to that of the emitter 1300 and the collector 1305 (e.g., n-type) (FIG. 13a), and a unipolar device with an n-type base that has a doping type (e.g., n-type) that is the same as the emitter 1300 and collector 1305 (e.g., n-type) (FIG. 13b). Terminals for an emitter contact 1320 and a base contact 1325 are provided.

(62) Referring to FIGS. 14a-14c, an RTBT or RBT integrates a tunneling quantum structure within a bipolar transistor. Three exemplary structures vary by the location of where the quantum tunneling structure is incorporated.

(63) Referring to FIG. 14a, an RTBT structure may include a tunneling quantum structure in the base 1310 region. Its energy band diagram is shown. As illustrated, a GaAs material system may be used; another exemplary material system is GaInAs/AsInAs. Referring to FIG. 14b, an RTBT structure may include a tunneling quantum structure at the base-emitter junction. Referring to FIG. 14c, an RTBT structure may include a tunneling quantum structure in the emitter 1300 region.

(64) Referring to FIG. 15, a three-terminal bipolar quantum RTT may be disposed over lattice-mismatched material 410, e.g., GaAs, formed by ART growth on a substrate 405. This bipolar quantum resonant tunneling transistor operates in a manner similar to a conventional bipolar transistor; the tunneling current may be controlled by modulating the potential inside a quantum well that functions as a base. The fabrication process has two phases. In the first phase, a second semiconductor material 410 (e.g., GaAs) is grown in a window 445 on a Si substrate 405 using ART. In the second phase, multiple layers 425 of GaAs and AlGaAs quantum wells are grown epitaxially on this structure. These QWs may be formed by typical processes familiar to those of skill in the art, and may include, e.g., the following layers: GaAs buffer 410 formed on Si substrate 405 by ART, undoped GaAs 1510 (1 m thick); contact layer 1515 of GaAs (2 m thick, Si-doped to 210.sup.18/cm.sup.3); supperlattice collector 1520 (50 periods of 8 nm GaAs and 2 nm of AlGaAs Si-doped to 210.sup.8/cm.sup.3), 3 periods of the same superlattice 1525 (8 nm GaAs and 2 nm AlGaAs, undoped), tunnel barrier 1530 (5 nm undoped AlGaAs), quantum well 1535 (15 nm with center 5 nm Be-doped to 110.sup.19/cm.sup.3), tunnel barrier 1540 (5 nm undoped AlGaAs), 3 periods of the same superlattice 1545 (8 nm GaAs and 2 nm AlGaAs, undoped), supperlattice collector 1550 (50 periods of 8 nm GaAs and 2 nm of AlGaAs Si-doped to 210.sup.11/cm.sup.3), and contact layer 1555 (2 m GaAs, Si-doped to 210.sup.18/cm.sup.3).

(65) Referring to FIG. 16a, a multi-state RTBT 1600 may include a stack of two Al.sub.0.48In.sub.0.52As (5 nm)/Ga.sub.0.47In.sub.0.53As (5 nm)/Al.sub.0.48In.sub.0.52As (5 nm) resonant tunneling double barrier structures incorporated into the emitter 1300 region of a Ga.sub.0.47In.sub.0.53As n-p-n transistor. Its corresponding energy band diagram is shown in FIG. 16b. This multi-state RTBT exhibits multiple negative differential resistance and transconductance characteristics.

(66) The RTBT 1600 includes n.sup.+ layer 1610 (GaInAs, 5000 angstroms), n-type layer 1615 (GaInAs, 3000 angstroms), p.sup.+ layer 1620 (GaInAs, 3000 angstroms), n-type layer 1625 (GaInAs, 500 angstroms), undoped layer 1630 (AlInAs, 50 angstroms), undoped layer 1635 (GaInAs, 50 angstroms), undoped layer 1640 (AlInAs, 50 angstroms), n.sup.+ layer 1645 (GaInAs, 1000 angstroms), undoped layer 1650 (AlInAs, 50 angstroms), undoped layer 1655 (GaInAs, 50 angstroms), undoped layer 1660 (AlInAs, 50 angstroms), and n.sup.+ layer 1665 (GAInAs, 5000 angstroms). Undoped layers 1660, 1655 and 1650 form one tunneling structure, and undoped layers 1640, 1635 and 1630 form a second tunneling structure.

(67) Referring to FIG. 17, an RHET 1700 incorporates a tunneling diode structure between the emitter 1300 and the base 1310 regions. This figure shows an exemplary RHET on a heteroepitaxial layer 410 grown on Si, and its energy-band diagram. One distinct characteristic is that the RHET has heavily doped emitter 1300 and base 1310 regions, except for the AlGaAs/GaAs/AlGaAs tunneling diode region layers in the lower portion of emitter 1300. It combines resonant tunneling with hot-electron ballistic transport in the base, resulting in super high speed. More specifically, the RHET includes a GaAs buffer layer 410 formed by heteroepitaxy, a collector 1305 including n.sup.+-GaAs 1710 and i-AlGaAs 1720, a base 1310 including n.sup.+-GaAs 1725, and an emitter 1300 including i-AlGaAs 1730, i-GaAs 1735, i-AlGaAs 1740, and n.sup.+-GaAs 1745.

(68) Referring to FIG. 18 a DELTT structure 1800 may be disposed over lattice-mismatched material, e.g., GaAs region 410. The GaAs region 410 is formed by ART growth on a substrate 405. An AlGaAs region 1850 is formed above the GaAs region 410. In this example, GaAs is used as narrow bandgap material and AlGaAs is used as wide bandgap material. Other narrow/wide bandgap material systems may be also used to fabricate DELTT devices in a similar manner.

(69) In an exemplary fabrication process, a dielectric layer over a semiconductor substrate is patterned to define a window for the DELTT region. Other devices, such as CMOS devices, may be fabricated on another region of the substrate. In the DELTT region, undoped GaAs is epitaxially grown by ART techniques, such that a top portion of the GaAs is substantially defect-free. Next a DELTT device is fabricated on top of the GaAs by, for example, a typical fabrication process such as that described in U.S. Pat. No. 5,825,049, FIGS. 6A-8D. The DELTT device may include the following features: a metal emitter contact 1842, a metal collector contact 1854, a two dimensional quantum well emitter layer 1844 electrically contacting the emitter contact 1842, a two dimensional quantum well resonant tunneling layer 1846 (e.g., about 7 nm of GaAs), a collector layer 1848 disposed opposite the resonant tunneling layer 1846 from the emitter layer 1844 (e.g., a 15 nm thick, 2-dimensional emitter layer for electrons), the collector layer 1848 electrically contacting the collector contact 1854, a first barrier layer 1845 (e.g., including two layers including a 20-30 nm GaAs voltage drop layer over the collector layer 1848 and a 5 nm AlGaAs wide bandgap barrier layer under the resonant tunneling layer 1846) disposed between the collector layer 1848 and the resonant tunneling layer 1846, a second thin tunneling barrier layer 1843 (e.g., a 7.5 nm AlGaAs layer) disposed between the resonant tunneling layer 1846 and the emitter layer 1844. The device may also include a back gate 1852 formed by implantation, an insulator 1860 (60 nm), an opening 1856 for insulation, and a metal control gate 1858. At least some of these features may also include GaAs. Thus, the substrate may include a group IV element such as Si, and the DELTT device may include a III-V compound. In some embodiments, the DELTT device may include a II-VI compound.

(70) The various transistor structures illustrated in FIGS. 13a-18 show three terminal devices (i.e., a transistor with terminals for the base, emitter, and collector) that include a tunneling structure and in which ART techniques are used to trap defects.

II. Integrated Circuit Units with Tunneling Devices and Conventional Devices

(71) Functional circuit units may incorporate both quantum tunneling devices, such as those described above, and conventional Si-based devices such as a MOSFET, MESFET, or conventional III-V-based device such as a HEMT, etc., to form a hybrid system. Such circuit units may function as, e.g., a logic circuit, an amplifier, an SRAM or a DRAM, a microprocessor, etc.

(72) A generic integrated circuit building block is disclosed in FIG. 19, a structure where tunneling devices are a building component for a variety of logic circuitry units. The tunneling device and/or the conventional device may be formed over desirable III-V or II-VI semiconductor material structures grown on common semiconductor substrates by ART techniques.

(73) Referring to FIG. 19, a structure including tunneling devices may be used as a building block for a variety of logic circuitry units. An exemplary generic logic circuit building block includes at least one tunneling device(s) 400 (e.g., RTD, RTT, RITD, RTFET, QEBRTT, RTBT, RHET, DELTT, etc.) and at least one conventional device(s) 1900 (e.g., MOSFET, MISFET, HEMT, etc.). Either or both of the tunneling device 400 and the conventional device 1900 may be formed over semiconductor material 410 grown on Si substrate 405 by ART techniques. The tunneling devices 400 may be fabricated in accordance with the fabrication process described above, or other exemplary devices discussed above. The conventional device may be fabricated in accordance with typical processes. An exemplary fabrication process may include fabricating a quantum tunneling device structure after the typical front-end process for a conventional Si-based CMOS process, but before the typical back-end process of the conventional Si-based CMOS process. Other similar processes may be also used for device and circuit fabrication.

(74) Such monolithic integration of conventional devices with high-speed tunneling devices enable the formation of high-speed logic circuits by increasing output drive current levels, logic fan-out, and output-to-input isolation, etc, while reducing both the power consumption and the number of devices used in the circuit. This hybrid circuit unit, fabricated on Si or other common substrates using ART techniques, may be used as a building block for various functional logic circuitry for various applications, such as digital logic, mixed signal, analog, etc. Several examples are illustrated below. Embodiments of the invention are not limited to those exemplary circuits, and may include other types of circuit units or circuit systems.

Example 1

(75) Referring to FIGS. 20a-20c, basic circuit units may incorporate an RTD fabricated on III-V epitaxy layers grown on Si. Alternatively, other types of tunneling diodes may also be used, such an RITD. Referring to FIG. 20a, an exemplary circuit unit 2000 includes an amplifier that provides an AC voltage gain. Referring to FIG. 20b, another exemplary circuit unit 2010 includes a simple memory cell comprising one RTD. The negative differential resistance of an RTD gives rise to two bi-stable operating states that enable a memory function. Referring to FIG. 20c, yet another exemplary circuit unit 2020 includes a logic inverter.

Example 2

(76) Referring to FIG. 21, two RTDs 400 and one MOSFET 1900 may be integrated to form an SRAM (TSRAM) 2100 with III-V heteroepitaxial layers on Si. This circuit unit SRAM 2100 may be constructed with the structure illustrated in FIG. 22, where two RTD (or RITD) devices are integrated on a semiconductor substrate 405 with a conventional n-MOSFET 1900 to construct an SRAM memory cell unit. The RTD or RITD devices may be fabricated in a process flow using ART techniques. The two RTD devices may be connected in series by an underlying doped region 2205, e.g., a p+ Si region defined by implantation. Both RTD devices may be disposed over an n-well 2210 that is electrically isolated from an adjacent p-well 2215 underlying the n-FET. Interconnections (not shown) may be defined between the n-FET and the two RTD devices to form SRAM memory cell unit. The resulting structure illustrated in FIG. 22 show two tunneling structures (in the two RTDs) formed via ART and coupled to a MOSFET 1900, which as illustrated in FIGS. 23, 24 and 25 can be coupled to a terminal (gate, source, or drain) of the transistor.

(77) Such a SRAM memory cell has two stable operation points that are substantially the same as those illustrated in FIG. 3. One may arrange multiple SRAM memory cells together within one chip to form a large scale SRAM array.

(78) Because fewer devices are used in the SRAM memory cell unit (see FIG. 21) in comparison to conventional six-transistor SRAM units, and also due to the small footprint of RTD devices that can be made much smaller than a MOSFET, this novel hybrid SRAM unit 2100 may provide a 3 to 5 times area reduction compared to the conventional six-transistor SRAM. It is refresh-free and has fast read and write. Its low-voltage operation and low leakage current enable it to serve as a low power solution. It may also be used to replace the conventional DRAM function.

(79) In summary, this exemplary RTD-based SRAM has the following features: about 3-5 area reduction in comparison to a conventional 6-transistor SRAM (the size of RTD can be much smaller than that of a conventional transistor); refresh free; fast read and write; and low voltage operation.

(80) In the example of FIG. 21, a load 2110 of the memory cell is another RTD device. Alternatively, a resistor or a conventional MOSFET can serve as a load as well. Various exemplary alternative binary T-SRAM configurations 2300 are illustrated in FIGS. 23a-23e.

(81) In the examples of FIG. 21 and FIGS. 23a-23e, the SRAM logic is binary logic. RTDs can also be used to construct multi-valued SRAM unit as well. Referring to FIGS. 24a-24e, various exemplary alternative multi-valued T-SRAM cell configurations 2400 include two or more RTDs connected in series to realize multi-value logic.

(82) In comparison to conventional SRAM circuits that typically use six transistors, when forming transistors of a given gate length, SRAM circuits according to FIGS. 21-24, i.e., incorporating RTD(s) with transistors may be implemented in at least about a 33% smaller cell area and typically in a cell area at least 50% smaller than the area required for conventional SRAM circuits. In other words, the RTD-based SRAM cell area is at least 33% smaller than a surface area of an SRAM circuit constructed using six transistors having the same gate length as the transistor in the RTD-based SRAM cell area.

(83) Referring to FIG. 25, as a variation of the memory cell of FIG. 21, two RTDs 400 are connected to the gate 2500 of the MOSFET 1900, instead of connected to the drain as in the case of FIG. 21. This circuit also functions as a memory cell.

(84) Referring to FIG. 26, a memory cell unit may include a tunneling device 400 with a memory storage film 2600 disposed thereover. The memory storage film may be a phase change material film, i.e., a phase change layer. In an alternative embodiment, the memory storage film may include a ferroelectric material film.

(85) Referring to FIG. 27, a non-volatile RITD/magnetic random access memory (RITD/MRAM) may include (1) an RITD device 400 fabricated in a window in a dielectric layer (not shown), and (2) a tunneling structure or magnetic tunnel junction (MTJ) 2700 disposed directly on top of the RITD. MTJ 2700 has, from bottom to top, an anti-ferromagnetic exchange bias layer 2710, a magnetic pinned layer 2715, a Ru spacer layer 2720, another magnetic pinned layer 2725, a tunnel barrier layer 2730, and a magnetic free layer 2735. The MTJ 2700 structure may include a conventional MRAM memory cell structure, and may have different variations By stacking more than one MTJ in series in this structure, the memory cell can have multi-value memory function.

(86) Another memory example includes a non-volatile memory cell including an RITD and a phase change chalcogenide resistor memory storage unit. By connecting a chalcogenide resistor as the load for the RITD, this memory cell may be used to realize a smaller multi-valued RAM memory. A typical chalcogenide resistor material is GeSbTe.

(87) In the manner described above, various memory circuitry, such as SRAM circuitry, DRAM circuitry, and non-volatile memory circuitry, may be fabricated incorporating RTDs.

Example 3

(88) Referring to FIG. 28, another basic circuit unit, a monostable-bistable transition logic element (MOBILE) 2800, has an edge-trigger and latching characteristics and can be used to build circuit units. For example, it can be used as an inverter. This circuit unit uses either two three-terminal resonant tunneling transistors 400, 400 or one two-terminal RTD and one three-terminal resonant tunneling transistor 400, 400. For example, a first RTD 400 may have a controlled gate, such as a RTFET, an RTBT/RBT or a RHET. The second RTD 400 may or may not have a control gate. For detailed structure of three-terminal resonant tunneling transistor, refer to, for example, structures in I (Individual Devices).

Example 4

(89) Referring to FIG. 29, an exemplary analog-to-digital converter circuit 2900 with an analog input 2910 and a binary output 2920 employs multi-state RTBT devices 1600, rather than standard RTBT devices, with an analog input 2910. For more details regarding the structure of three-terminal multi-state RTBT devices, refer to FIG. 16.

Example 5

(90) Referring to FIG. 30, an exemplary RTD/HEMT comparator 3000 incorporates an RTD 400 into a HEMT 1900. This comparator can significantly reduce the number of devices used in a circuit, reduce power consumption, and increase speed, compared to a conventional HEMT-based comparator.

Example 6

(91) Referring to FIG. 31, an exemplary RTD-based clock generator 3100 allows a great reduction in the number of devices used in the circuit by employing RTDs 400.

Example 7

(92) Referring to FIG. 32, an exemplary field programmable gate array (FPGA) 3200 incorporates a large number of RTD-based SRAMs fabricated on III-V epitaxial layer structures on Si, e.g., by using ART techniques.

(93) In a typical SRAM-based FPGA system, SRAM cells throughout the FPGA determine the functionality of the device. It typically includes:

(94) (a) configurable logic block (CLB) 3210. The fixed arrays of CLBs are connectable by a system of pass transistors, driven by SRAM cells. Each CLB typically has two lookup tables (LUT) and two registers;

(95) (b) switch matrix 3220; and

(96) (c) input/output block 3230.

(97) SRAM cells are primarily used for three purposes: As look-up tables (LUTs) for implementing logic (i.e., as a truth-table). As embedded block RAM blocks (for buffer storage, etc.). As control to routing and configuration switches.

(98) For the LUT, in addition to their use as general logic gates, LUTs may alternatively be used as general purpose RAM. Each combination of four LUTs can become a 161-bit RAM array.

Example 8

(99) Referring to FIGS. 33 and 34, an exemplary RTD-based or RTT-based microprocessor 3300 is a VLSI device that carries out a range of instructions on data transferred from a register array. Similarly to the previous examples of RTD-based circuitry, the use of RTD or RTT devices significantly reduces the number of devices needed in a circuit, reduces power consumption, and increases speed, in comparison to conventional FET-based microprocessors. The microprocessor 3300 comprises (1) arithmetic logic units (ALUs) that execute a variety of arithmetic and logical function; (2) storage locations (called registers) and memory units which are used to store data, addresses, instructions and control programs, etc; and (3) control systems and interconnection systems that provide links among different components, accept and generate external control signals (such as read and write), and provide timing signals for the entire system, etc.

(100) Using tunneling devices such as RTDs or RTTs for all of three of these components of the microprocessor can provide performance benefits including increased speed and reduced power consumption, thereby improving the basic functionalities of a microprocessor, which include: retrieving data from memory; manipulating data; and storing data in memory.
FIG. 33 shows an exemplary block diagram of a modern microprocessor 3300, and FIG. 34 shows its micro-architecture.

Example 9

(101) Referring to FIG. 35, an RTD-based digital signal processor (DSP) 3500 is a specialized microprocessor designed specifically for processing signal in digital form, generally in real-time. DSPs have at least four major subfields: audio signal processing, control engineering, digital image processing, and speech processing. The baseline DSP includes one arithmetic logic unit (ALU), dual memory interfaces, and a control unit (instruction decoder, branch control, task control). An analog signal 3510 enters the analog-to-digital converter (ADC) 3520. The analog signal is converted to a digital signal 3530 and is processed by a signal processor 3540. The digital signal is transmitted to a digital-analog converter 3550, and an analog signal 3560 is transmitted to an audio amplifier 3570.

(102) A DSP typically has many advantages over its counterpart analog processing system. It is able to provide far better levels of signal processing than is possible with analog hardware alone. It is able to perform mathematical operations that enable many of the spurious effects of the analog components to be overcome. In addition to this, it is possible to easily update a digital signal processor by downloading new software.

Example 10

(103) Referring to FIG. 36, an RTD-based graphics processing unit (GPU) 3600 (also called a visual processing unit (VPU)) is a dedicated graphics rendering device for a personal computer or game console that is very efficient at manipulating and displaying computer graphics. Modern GPUs also have support for 3D computer graphics, and typically also include digital video-related functions. GPU 3600 components include an EDRAM 3605, unified memory 3610, texture/vertex fetch 3615, unified shader 3620, primitive setup clipper rasterizer hierarchical Z/S 3625, index stream generator tessellator 3630, and an output buffer 3635, with memory export from the output buffer 3635 to the unified memory 3610. The vertex pipeline 3640, pixel pipeline 3645, and display pixels 3650 are illustrated

III. Structures and Processes for Integrating Tunneling Devices and Conventional Devices into Circuit Units on Si or Common Substrates

(104) In various integrated circuit units described in II (Integrated Circuit Units with Tunneling Devices and Conventional Devices), the tunneling devices may be integrated with conventional devices by several different structures and processes, using ART techniques in accordance to one aspect of this invention.

(105) In one embodiment, for example referring to FIG. 22, two RTD (or RITD) devices 400 may be integrated on a semiconductor substrate 405 with a conventional n-FET 1900 to construct an SRAM memory cell unit 2100. The two RTD devices 400 may be connected in series by an underlying doped region 2205, e.g., a p+Si region defined by implantation. Both RTD devices 400 may be disposed over an n-well 2210 that is electrically isolated from an adjacent p-well 2215 underlying the n-FET 1900. Interconnections (not shown) may be defined between the n-FET 1900 and the two RTD devices 400 to form SRAM memory cell unit 2100.

(106) A typical fabrication process for forming the SRAM memory cell unit 2100 may be summarized as follows. Substrate 405, including a first semiconductor material, is patterned to define an n-FET region and RTD (or RITD) regions. The n-FET device 1900 is fabricated first using the conventional front-end Si CMOS processes. Then, windows in a dielectric 440 layer are defined in RTD/RITD regions. The windows are filled with a second crystalline semiconductor material 410, e.g., a III-V or a II-VI compound, employing ART heteroepitaxy. RTD device layers 425 are formed over the second crystalline semiconductor material 410. Finally, back-end CMOS processing is used to define interconnections between the RTD devices 400 and the n-FET 1900. Issues regarding the thermal budget are mitigated by performing the CMOS front-end processing before the formation of the RTD devices. This sequence is preferable because front-end n-FET fabrication involves high process temperatures, while RTD material may only sustain a low thermal budget. This process is also highly compatible with current Si process.

(107) Referring still to FIG. 22, details of an exemplary fabrication process are now given. The fabrication process may be divided into three phases. n-FETs 1900 are fabricated during the first phase, which essentially is similar to a conventional front-end process of Si CMOS processing; RTDs or RITDs 400 are fabricated in the second phase; and interconnections are made during the third phase, which essentially is similar to the conventional back-end Si CMOS process. As used herein, the terms front-end processing and back-end processing have the meanings generally understood for these terms. Front-end processing generally refers to all the steps up to and including the formation of a gate and source/drain regions. Back-end generally refers to all the steps performed thereafter to fabricate a CMOS device, i.e., formation of inter-level dielectrics and conductive interconnects.

(108) During the first phase, the substrate is cleaned using a standard RCA clean. A pad oxide of 50 nm is grown on the surface, followed by deposition of a 150 nm layer of silicon nitride. Then windows are etched through the silicon nitride and oxide layers to expose the Si surface in the area of n-well 2210 on which the RTD devices will be defined in third phase. These areas are implanted to form n-well 2210 by, for example, phosphorus implantation to a dose of 5.5E12/cm.sup.2 at an implantation energy of 100 keV.

(109) Next, a p-well 2215 is similarly formed in the area where the n-FET 1900 will be fabricated. To form a p-well, boron implantation may be used, for example with a dose of 6E12/cm.sup.2 at 55 keV.

(110) After both n- and p-wells are implanted, a well drive-in anneal is carried out, for example at 1100 degrees C. in a nitrogen ambient. The remaining oxide is etched away.

(111) Next, another pad oxide of 50 nm and another silicon nitride layer 150 nm are deposited over the substrate. A window is etched through the pad oxide and silicon nitride layers to expose the underlying silicon surface only in the n-FET area. In this window, an n-FET 1900 is fabricated. Fabrication of the n-FET includes forming a silicon channel region with appropriate channel implantation, source/drain regions with appropriate source/drain implantation, a gate dielectric layer, and a gate electrode layer with appropriate gate implantation. At least one high temperature annual is conducted to activate the dopant, for example at 1000 C. for 30 minutes.

(112) In the second phase in which the RTD devices 400 are fabricated, and in the third phase in which interconnections are defined, the processes typically have a much lower thermal budget than the first phase.

(113) In the second phase, in an embodiment, the RTD devices 400 are grown after a FET 1900 is defined and after a pre-metal dielectric (PMD) layer is formed, but before a first metal layer is defined. Windows are etched through the dielectric layers on the areas in which RTD devices will be formed to expose the underlying silicon surface. In these windows, a second crystalline semiconductor (e.g., GaAs) 410 and then RTD devices 400 are grown epitaxially using essentially the methods described above. A relatively low temperature anneal may be carried out, for example 700 degrees C. for 1 minute.

(114) In phase three, a back-end process similar to conventional CMOS processing is performed. First, contact windows are etched through the dielectric layer to reach the contact layers of the RTD devices. Another set of contact windows are also made to reach the source/drain/gate contacts of the n-FET. A metal layer, such as Al or Cu, is deposited over the whole surface and into the contact windows. The metal layer is patterned and selectively removed by conventional lithographic and etch processes, leaving behind a pre-determined metal interconnection network that connects the n-FET 1900 and RTD devices 400 into a functional circuit system.

(115) In another embodiment, referring to FIG. 37, two RTD (or RITD) 400 devices may be integrated on a semiconductor substrate 405 with a conventional n-FET 1900 to construct an SRAM memory cell unit 2100. In this structure, the two RTDs 400 are directly connected with the drain region 3700 of the conventional MOSFET 1900. Alternatively, one or both of the RTDs 400 may be coupled to a source 3720 region of the MOSFET 1900. The other process steps may be substantially similar to that in FIG. 22, i.e., a MOSFET 1900 fabricated in the first phase; an RTD 400 fabricated in the second phase; and metal interconnection in the last phase. The resulting structure illustrated in FIG. 37 includes two tunneling structures (in RTDs 400) formed via ART above the drain region 3700 of MOSFET 1900, with the MOSFET bound by a shallow trench isolation region 3730.

(116) In the configuration illustrated in FIG. 37, the size of the RTD 400 is similar to the size of traditional contact holes 3710 for the source and drain. In a typical MOSFET layout, which may have multiple contact holes 3710 formed through a PMD layer 3705 over the source 3720 and/or drain 3700 areas, incorporating the RTDs does not require an additional significant increase in source/drain area.

(117) In yet another embodiment, the tunneling devices 400 and other devices may be fabricated on top of each other, for example, as in the structures in FIGS. 26 and 27. In those structures, the RTDs are fabricated first, before the fabrication of other device structures on the top.

(118) The invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The foregoing embodiments are therefore to be considered in all respects illustrative rather than limiting on the invention described herein. Scope of the invention is thus indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are intended to be embraced therein.