Nanoelectronics and Nano-Architectonics
Transcription
Nanoelectronics and Nano-Architectonics
MARCO-FENA Nanoelectronics and Nano-Architectonics - Towards Robust Integrated Nanosystems Kang L. Wang MARCO Center on Functional Engineered Nano Architectonics -- FENA University of California, Los Angeles Westwood, CA 90095-1594 MARCO - FENA MARCO Focus Center -- FENA 2 The term “Architectonics” is derived from a Greek word meaning the Science of master builder – Mastering the building of Nanoelectronics and Nanosytems – NanoLegos 2 Grand Challenge in Nanoelectronics 1968-2003 (CMOS) throughput by 108-109 ¾ Transistor count increasing to 108-109 transistors/chip ¾ Price per transistor decreases 7 orders of magnitude ¾ Volume decreasing 6-7 orders of magnitude ¾ Power decreases 7 orders of magnitude ¾ Moore’s second observation – increasing cost of manufacture -- Self assembly (G. Moore, ISSCC 2003) 3 Performance Space MARCO - FENA ¾ Functional The next? 10 nm 100nm Technology node V. V. Zhirnov, D.J.C. Herr: New Frontiers: Self-Assembly and Nanoelectronics. IEEE Computer 34 (1): 34-43 (2001) 3 MARCO - FENA Current Issues, Challenges & Opportunities Efficient logic functionality -- Low power Nonideal Fundamental limits From functional Power dissipation nanomaterials and new device Increase number of with alternate state variables interconnects Utilization of quantum effects Slow increase of density Novel Architectures for new and functionality devices Complexity in design Small current drive Higher functional throughput Device variations – Reduced or having min Reliability and number of interconnects reproducibility Homogeneous cells Low Power and robust • Increase of manufacture cost Directed Self Assembly 4 4 Nanoelectronics -- Next Generations • Fewer electrons • Massive number of interconnects • Power consumption Gate Length (nm) Number of electron (per Gate) MARCO - FENA 106 1000 100 10 1970 1980 1990 2000 2010 10 19 10 18 10 17 10 16 10 15 10 2020 14 -3 10000 Doping Concentration (cm ) Major Issues Year 104 Ultimate CMOS Nanoelectronics Quantum computing/ information Room temperature Si based devices Low temperature devices Alternate architectures 102 “Single” Electron Devices 1 0.1 1970 1990 2010 2030 2050 Year 5 Single electron detection made 5 CMOS Device Scaling Continues ---20 0 3 45nm Node –20 1 2007 32nm Node 2 –– – 2009 50nm Length (IEDM2002) 22nm Node 2011 30nm Prototype (IEDM2000) 25 nm 20nm Prototype (VLSI2001) 15nm 16 nm node ---201 2013 1-3nm 3– 201 11nm node 7 2015 15nm Prototype (IEDM2001) 10nm Prototype 7nm (DRC 2003) Limit ~ 5-7 nm 6 Na Na notu no be wi res s, , .. MARCO - FENA 90nm Node 2003 65nm Node 2005 8 nm node 2017 5nm Si CMOS becoming less Si Heterogeneous integration 3nm 6 MARCO - FENA Mission Statement of FENA : 7 To create and investigate new nanoengineered functional materials and devices, and computational architectures for new information processing systems beyond the limits of conventional CMOS technology. 7 FENA - Participating Universities Distributed Research Model Multi-Institutional University of Minnesota Dept. Electrical and Computer Eng. University of California at Berkeley MIT MARCO - FENA Dept. Electrical Eng. and Computer Sciences Dept. Biological Engineering UC at Santa Barbara State University of New York at Stony Brook Dept. Materials Dept. Electrical and Computer Eng. Dept. Physics UCLA North Carolina State University Dept. Electrical Eng. Dept. Mathematics Dept. Chemistry and Biochemistry Dept. Material Science and Eng. Dept. Electrical Eng. University of California at Riverside Arizona State University Dept. Electrical Eng. Dept. Mechanical Eng. Dept. Chemistry Mathematics Dept. Electrical Eng. University of Southern California Dept. Electrical Eng and Electrophysics Dept. Chemistry California Institute of Technology Dept. Chemistry Dept. Materials Science Dept. Applied Physics MULTI-DISCIPLINARY MULTI-DISCIPLINARY TEAM TEAM MATERIAL MATERIAL SCIENTISTS SCIENTISTS CHEMISTS CHEMISTS 8 PHYSICISTS PHYSICISTS BIOENGINEERS BIOENGINEERS MATHEMATICIANS MATHEMATICIANS FENA Quick Facts • • • • Principal Investigators: 28 Students & Researchers: 60 Universities Involved: 11 Admin Headquarters: UCLA ELECTICAL ELECTICAL ENGINEERS ENGINEERS 8 MARCO - FENA Funding Stream Semiconductor Industry Association Semiconductor Industry Supplier Sponsors Management MARCO - FENA Funding FCRP Governing Council Focus Centers 9 9 FENA – Research Theme MARCO - FENA TOP-DOWN VISONS Theme 5: New computational and information processing technologies/paradigms suitable for novel nanodevices – Richard Kiehl Theme 4: Novel devices for high functionality and heterogeneous integration - Kang Wang Theme 2: Synthesis & manufacturing methods for nanoscale ordered materials & structures -Evelyn Hu Theme 3: Simulation and computations of novel engineered nanomaterials and devices – Russ Caflisch Theme 1: Novel materials from atomic and molecular levels – Fraser Stoddart 10 BOTTOM-UP SOLUTIONS 10 Building Blocks: Novel Materials from Atomic and Molecular Levels Aim: To use bottom-up synthesis and fabrication approaches to create and assemble new nanoscale functional materials and nano structures. Nanostructured Material O O O O +N +N O O N+ O S S S S O O O N+ S S S N S S N S S P N N + N + O N + O Ro tary Mo le c ular S witc h O OMe S Line ar Mo lec ular S witc h O O CH2 O O a) Nanopatterning Self assembly - Chemical and N O S O O Nano Patterning O O + MARCO - FENA S O d) Atomic Interfacial Layers biological self-synthesis b) Nano-patterning self-assembly polymers c) Nanowires and Nanogrids Utilizing Self-assembled Templates d) Virus Engineering to Grow and Self Assemble Interconnects S Me N S Me S S + N Functional Molecular NanoMaterials O O O memory using SiGeC/Si hetero-nanocrystals a) Structural ordering of organic polymers b) Nanowire Sensors c) Functional Molecular Materials (c) (b) O + a) Molecular Recognition b) Fabrication of Nanowires/Nanotubes c) Evaporation Induced Self-Assembl Single electron (a ) O O O O O O O OMe Me O Me O Mo le c ular Dio de O O CH2 O O Me O Me O Functional molecular electronic components, 2 switches: a) Catenane b) Rotazene and c) a diode. GCAT CGTA Organic polymer thin film transistors: nanoscale transistor functionality. CT G A A CG T CG TA TGC A 11 Polyaniline/Au conductive polymers. Applications: Molecular recognition allowing polymers to selfsensors, conductive connecstors, nanogrids, assemble into nanowires, grids or nanopatterns. 11 Self Assembly: Principles Physical self assembly ¾ MARCO - FENA ¾ ¾ ¾ Ge Wires Chemical and Bio chemical self assembly ¾ ¾ 12 Mechanical Field – templating, strain, etc. Use of structured strain Electrical and magnetic (including photon) fields Surface energy – catalyst seeding Chemical bonding Conjugating - e.g., triple conjugation of QDs will be achieved at the Y-Junction, while QDs are trapped at the junction 12 Self-Assembled Molecular Memories 10 Au electrodes -4 2 10 nd bias cycle Pd nanowire -5 ON Current (A) MARCO - FENA 10 st 1 bias cycle -6 NC N 10 -7 rd 3 bias cycle 10 -8 10 -9 NC NH2 N H 2-amino-4,5-imidazoledicarbonitrile (AIDCN) OFF Pd Vswitch -10 Au 10 0 13 1 2 3 4 (V) LocalizationVoltage -- regimented Field assisted Physical, chemical, and biochemical 5 SiO2 Si Yang Yang of UCLA 13 Nanowires - In2O3 NO2 Sensor 1 .0 I (µΑ) 0 .8 0 .6 5 p pb 0 .4 10 ppb 0 .0 0 2 4 6 10 0 p p b T im e (s) 500 ppb 8 1 0 x1 0 3 30 1000 25 Dopant: 20 15 10 Before 5 20 30 500 HCl 40 D ia m e te r s (n m ) I (nA) 0 Polyaniline Nanofibers 2.4 OFF 2.0 -1000 -1.0 R/R0 1.8 1.6 After NO2 In -0.5 0.0 V (V) 0.5 1.0 On/off ratio: 1.2 × 106; Lowest detectable level: 5 ppb; Response time: 5 s for 100 ppm NO2, 17 min for 5 ppb 1.4 1.2 1.0 0.8 0 -500 0.9 ppm NH3 Vapor 2.2 14 50 pp b 35 Counts MARCO - FENA 0 .2 ON 0 100 200 300 Time (s) 400 500 14 MARCO - FENA Chemical and Biochemical process for Integrated Circuit Architectonics NANO-ASSEMBLED CIRCUIT COMPONENTS •SYNTHESIS OF CIRCUIT BUILDING BLOCKS A circuit block •DROP-IN ONTO EXISTING TECHNOLOGY 15 Chemical Conjugated bonding Nano self assembly of nano particles Anchoring the components – to nanowires Mihri Ozkan 15 Assembly of Organic and Inorganic Nano Legos (A bottom-up approach) MARCO - FENA (A) (B) Example: Nano particles with Biotin-Avidin Example: Nano particles with with amine groups bridged with dialdehydes Two particles nonwith complementary Twonano individual nanowith particles recognition Bivalent linker that directly recognition groups, whichto can bridged usingof nano groups complementary eachbe other recognizes the surface (C) bispecific linker moleculesparticles Example: Nano Au assembly with disulfide groups 16 16 FET as Demonstration (1) Drain CNFET (B*) Gate (B) Gate N-type SWCNT C* (A) (A*) MWCNT Source Pd atoms (metalized DNA) MWCNT MARCO - FENA MOSFET Source 17 17 CMOS Demonstration (2) Combinational CNFET Inverter CMOS Inverter MARCO - FENA Vdd 18 Vin Vout 18 QD Functionalization at Nanotube-ends for Short and Long CNTs (Configuration 1) QDs (B) QDs at the end of the CNT QDs bundles Variable band gap (A) (B) QDs at the end of the CNT bundles MARCO - FENA (A) (2.1eV) CNT (4.0 eV) CNT CdSe CdSe 200 nm CNT bundle CNT CNT bundle ZnS ZnS ZnS ZnS ZnS ZnS 2 µm CdSe 200 nm CNT CdSe 2 µm ZnS capped CdSe at the ends of a CNT AQD CNT ~500 nm in withfor QDtubes conjugation at both conjugation at length CNT-ends as long as 4µm.ends Band diagram of a QD-CNT-QD heterostructure 19 19 MARCO - FENA Nanomaterials – Enabling Heterogeneous Integration Nanostructures enable the reduction of defects and Free from the constraint of crystalline substrates High performance, large area Heterogeneous nanosystems!! 20 20 21 New nanoelectronics ¾ Spin as a variable Ultimate CMOS and beyond Nano photonics Molecular devices Spintronics ¾ Nanomagnetics ¾ Coherent wave Nano bio devices Sensors and Fluidics; Nanotransducers Quantum coherence devices (and Systems) Performance space MARCO - FENA What else can be put on Si?? in Integrated Nanosystems Market sectors Application space (Heterogeneous Integration) Self Assembly 21 CG T A TG C A CT G A A CG T MARCO - FENA GCAT CGTA Heterogeneous integration of nanosystems O O O OOO O O N C C N O O Chromophore N+ CH2 OH HO • Increasing functionalities N+ CH2 Coupling layer 2)n (CH 2)n (CH SiO2 SiO2 •Self adapting and correcting •Reduced cost • Interface The schematic of a structure with covalently coupled single chromophore molecule and its on SiO • Process compatibility attached nano -scale MOSF ET. 22 • Self assembly 2 22 Power Density, Consumption and Clock Frequency MPU physical gate length (nm) 9 13 18 -15 10 12 10 28 37 700 350 1500 53 130 500 1000 70 200 3000 6000 10000 10 10 9 10 8 10 7 10 Nuclear Reactor 10 Pentium Pro Pentium -18 10 Extrapolated value: i486 3x10-20J@5nm kT ln2 ≈ 3x10-21 J 4x1011Hz@5nm -19 10 -20 10 5 Year 10 Hot Plate 8086 80286 8080 i386 00 99 97 95 93 89 82 85 10 1 4004 0.1 100 1000 Technology node (nm) 2016 13 10 06 04 02 23 100 Pentium IV Pentium III Pentium II -17 2 10 -16 10 Power Density (W/cm ) On-chip clock frequency (Hz) MARCO - FENA 11 Power-delay product (J/device) 1000 10000 78 74 71 Sources: ITRS 2002 and Osman Unsal and Isreal Koren, IEEE Proceedings., Vol. 91, No. 7, 2003.23 MARCO - FENA MSI K7T266 Pro-2 Raid Motherboard Done by Trubador on 26 Feb 2002 24 24 Power Consumption Power (W/sq cm) MARCO - FENA Small static power consumption • Subthreshold current Power consumption = fs*Cout*Vdd2 10 3 10 2 10 1 10 0 10 Major problems ITRS Switching Tunneling Off Current -1 10 -2 10 -3 10 Gate Drain 100 Feature Size (nm) 25 25 MARCO - FENA I think that there’s a world market for about 5 computers. 26 Thomas J. Waston, Sr. IBM Chairman of the Board, ca1946 The ENIAC machine occupied a room 30 x 50 ft. (van Pelt Library, U Penn) Size: 7.44mm x 5.29mm; 174,569 transistors; 0.5 um CMOS Created by Jan Van der Spiegel August 27, 1997. 26 Robustness For an activation energy of ∆ −1 − ∆ / 2kT Z e MARCO - FENA b a −1 ∆ / 2kT Z e Probability ratio of the two states : Pa = exp ( −∆ / kT ) Pb At 125 C (400K) Assume Activation Energy: 1 eV (Si-Si bond of 3.37 eV) The error: exp ( −∆ / kT ) = 6.9x10-14 For an activation Energy (hydrogen bonding) 0.168 eV The error: exp ( −∆ / kT ) = 6.1x10-3 27 27 Nanoscale Architectures and Information Processing Paradigms Aim: To investigate novel information processing architectures based on the unique features of nano devices Cellular Nonlinear Network Architectures MARCO - FENA a) b) Neuromorphic Architectures a) b) Theoretical and Experimental study on New Computational System Cellular Architectures Based on Nonlineaer Phase Dynamics in 2DArrays Associative pattern recognition based on CMOS/SET hybrid arrays CrossNet Arrays using CMOL Information Processing Paradigms Based on the Dynamics of Biological Systems: dynamic associative memory METALL CONTACTS SiO2 Output Neuron B2 B1 B3 SiO2 EDGE CMOSs B4 BM B5 (CMOS) QUANTUM BARRIER QUANTUM WELL Synapse (SET) 28 Schematic of a CrossNet switching plaquette. Green dots with arrows are three-terminal single electron switches. A1 A2 A3 A4 A5 A6 AN Input A neuromorphic model, investigate fan out through capacitance SET devices & robustness. QUANTUM BARRIER QUANTUM WELL QUANTUM BARRIER COMMON CONDUCTIVE SUBSTRATE Cellular automata or CNN based on semiconductor tunneling nanostructure array 28 New computational and information processing technologies/paradigms suitable for nanodevices • Reversible computing MARCO - FENA • Homogeneous Cells: Cellular automata consisting of 2-dimensional arrays of 3dimensional lattice array of nanocells. • Self adaptive, fault tolerant, or error correction automata using nanodevices (e.g., single-electron transistors, single molecules, single spin,etc.) • Automata integrating CMOS and other nano devices including large molecules 29 Identified the critical problems Large number of wiring ¾ ¾ Design complexity ¾ ¾ Common problems/Challenges for Nano devices Low drive current ¾ ¾ Leakage, defects, errors ¾ ¾ Power issue ¾ ¾ Unique features – Quantum effects ¾ ¾ Interactive neighbors ¾ ¾ Cannot have many global wirings ¾ ¾ Complexity of fabrication 29 In A Nutshell FENA: To explore radically new approaches to giga and tera scale architectures for future nanosystems MARCO - FENA Parallel information processing Defect and Fault tolerance : Self repair, self learning systems Top view to provide our vision of the Center and for planning of the research Bottom up approach from functional materials and nanoarchitecture Novel Nano materials and building blocks by design Robust nanodevice structures ¾ ¾ Ultimate CMOS and SOI CMOS leading to seamless transition to quantum devices and other new devices Molecular devices Process compatibility with Si – Directed Self Assembly ¾ 30 High level processing NanoElectronics >> Heterogeneous NanoSystems (new state variables) 30 Methods for Self-assembly Physical self assembly MBE, GSMBE, CVD, etc. ¾ Templates: Electrochemical, mechanical, Sol gel, etc. ¾ MARCO - FENA Chemical self assembly Molecular self assembly, polymer self assembly, protein, DNA, bio-molecular, etc. ¾ Colloidal self assembly ¾ Bio self assembly ¾ Peptide, Protein and Virus engineering User defined surface dip pen Key Issues: Uniform size Controlled placement Directed processes ¾ 31 Physical mechanisms, Chemical Mechanisms, and Biochemical Processes 31