Carbon Nanotubes in Interconnect Applications Page 1 What are Carbon Nanotubes? What are they good for? Why are we interested in them? - Interconnects of the future? Comparison of electrical properties - Au-nanowires versus Carbon nanotubes: resistance and current carrying capacity Where do we stand - our results - how to make them... -...in the right place... -...first results: single tubes, vias and contact holes Conclusion European Workshop on Materials for Advanced Metallization Location: VAALS, NETHERLANDS Date: MAR 03-06, 2002
What are Carbon Nanotubes? just roll up a graphene sheet and you will get a single wall carbon nanotube (SWNT) Page 2
What are Carbon Nanotubes? Page 3 multi-wall tubes (MWNT)
What are Nanotubes good for? prominent features: Page 4 length: microns to millimeters diameter: 0.4-100 nm electrical conductivity: metallic (ballistic transport) or semiconducting depending on chirality energy gap depends on diameter operating current density > 10 9 A/cm 2 (Cu~ 10 6 A/cm 2 ) thermal conductivity ~ 6000 W/Km (Cu~ 400W/Km) gravimetric surface: > 1500 m 2 /g maximum strain: 0.11 % @ 1Volt (Ferroelectr. 0.11 % @ 100 V) e-modulus 1000 GPa, elastic elongation of 40% i.e 100-times stronger than steel at 1/6 weight huge hydrogen storage capacity of ~ 6 weight percent
Some Carbon Nanotubes Applications transistor 3d-wiring via-interconnect actuators, sensors carbon-nanotubes Page 5 nanocontainer hydrogen storage biomolecular applications gas sensors batteries flat panel displays
Why are CNTs interesting for us? Nanotubes: a possible solution for roadmap problems? red areas: no known solutions! Page 6 source:http://public.itrs.net/files/2001updatefinal/2kudfinal.cfm
Comparison: Gold nanowires vs. Carbon nanotubes two approaches to compare CNTs with metal wires: bottom up approach: what is know from mesoscopic / atomistic theory? top down approach: what do we expect from classical scaling (including size-effects)? 100 Au effective rho (Fuchs/Sondheimer) Au bulk value of rho rho [µωcm] 10 Page 7 why to compare with Gold -wires? : 1 0 10 20 30 40 50 diameter [nm] get rid of troublesome oxidation / interface layers theoretical results are available
Comparison: Gold nanowires vs. Carbon nanotubes bottom up approach DFT-calculations of structural and electronic properties of Au-nanowires Page 8 Wang et al. PRL 86, 2046 (2001)
Comparison: Gold nanowires vs. Carbon nanotubes bottom up approach Au-nanowires Carbon Nanotubes Wang et al. PRL 86, 2046 (2001) tight-binding calc. Nardelli PRB 86, 7828 (1999) tight-binding calc. (4,4)-SWNT Page 9 => Max. conductance is below Fermi energy! => number of conductance channels scale linearly with diameter => strong structure dependence => no influence of bending
Comparison: Gold nanowires vs. Carbon nanotubes bottom up approach Conductance of Au-nanowires and Nanotubes conductance / (2e 2 /h) 13 12 11 10 9 8 7 6 5 4 3 2 1 0 Au Nanowires (theory) multiwalled CNT 1 2 3 diameter /nm conductance / G 0 10 2 CNT-Array Au Nanowires (theory) 10 1 10 0 0.7 nm 0 1 2 3 4 5 diameter /nm Page 10 at the smallest dimensions the conductance is almost equal but the life-time of a 2 nm long Au-nanowire is about 1 sec (even without applied current)
Comparison: Gold nanowires vs. Carbon nanotubes top down approach Conductance of Au-nanowires and Arrays of Nanotubes without Size Effects resistance [Ohm] 1M 100k 10k 1k 100 10 Au-wire 100 nm CNT-array Au-wire 300 nm Au-wire 500 nm Au-wire 700 nm 0.7 nm TEM-image(Smalley) Page 11 0 5 10 15 20 25 30 diameter [nm] values for Au are classically evaluated, i. e. bulk value without size-effects! for long wires (>100 nm) CNT-arrays are the better conductors
Comparison: Gold nanowires vs. Carbon nanotubes top down approach Conductance of Au-nanowires and Arrays of Nanotubes with Size Effects resistance [ Ohm ] 1M 100k 10k 1k 100 10 CNT-array Au-wire 100 nm Au-wire 300 nm Au-wire 500 nm Au-wire 700 nm 0.7 nm TEM-image(Smalley) Page 12 0 5 10 15 20 25 30 diameter [ nm ] values for Au are evaluated with size-effect (Fuchs-Sondheimer) CNT-arrays are the better conductors
Gold nanowires vs. MW-CNT: reliability and current density Nanotube I Current densities up to 10 10 A/cm 2 without heat sink (not imbedded in SiO 2 ) equivalent Au-, Cu-, Al-wires deteriorate at 10 7 A/cm 2 Nanotube II Page 13 Nanotube I equivalent Au-wire I Nanotube II equivalent Au-wire II Diameter 8.6 nm 8.6 nm 15.3 nm 15.3 nm Length 2.6 µm 2.6 µm 2.5 µm 2.5 µm Voltage 25 V 17 V Resistance 2.4 kω 5.6 kω 1.7 kω 1.0 kω Current density 1.8 10 10 A/cm 2 ------ 5.4 10 9 A/cm 2 ------
Where we stand - our results how to make them CVD-growth of Carbon Nanotubes Page 14 The catalyst support, i.e., the substrate is also very important.
Where we stand - our results how to make them Task Identify: suitable catalyst suitable catalyst support suitable CVD-process to get good quality nanotubes with processes compatible with semiconductor backend of line Page 15
Where we stand - our results how we make them Preparation of the samples by lift-off technique CVD-growth of Carbon Nanotubes i-line or e-beam resist substrate resist patterning by i-line or e-beam substrate deposition of catalytic materials by evaporation, sputtering or spin-on methods substrate lift-off with solvents and ultrasonification CVD-growth of multiwall carbon nanotubes Page 16
Pure Metal (PVD) Where we stand - our results how we make them CVD of Multi-walled Nanotubes - Table of catalysts tested Carbonyl Chloride Acetylacetonate (spin-on) Oxalate Carbonyl in cellulose nitrate (spin-on) (spin-on) (spin-on) (spin-on) Fe Excellent Excellent Excellent Excellent Good None Ni None - None Poor - - Co None None Single None - None Ru - None None - - - Mo - None - - - - Ir - - - - - - Pt None - - - - - Page 17 Suitable substrates include: Silicon, Silicon Oxide, Silicon Nitride, Poly-Si,Tantalum, Tantalum Nitride, Titanium, Aluminum, Nickel (poor). Unsuitable Substrates: Gold, Platinum, Tungsten. The interaction between substrate and catalyst is crucial.
Where we stand - our results how we make them Aligned nanotube growth on 6-inch wafers in a batch process up to 25 wafers in parallel contamination < 10 11 atoms/cm 2 1 uro Page 18
Where we stand - our results how we make them time dependence of the growth rate Page 19
Where we stand - our results how we make them 100µm x 100µm blocks of carbon nanotubes on a SiO 2 substrate Page 20
Where we stand - our results how we make them Selective Growth of Carbon Nanotubes on Metal Electrodes Cross-section SEM-image Page 21 We grow nanotubes on: Ta, Al, TaN, Ti, TiN, W Ta-layer
Carbon-Nanotubes Characterization of single multi-walled carbon nanotube 12nm CNT Pt Contact Pt Contact SiO 2 /Si Page 22 20 µm long MW-CNT adsorbed on 2 Pt-contacts 12 nm in diameter TEM-image: showing individual layers
Carbon-Nanotubes Characterization of a single multi-walled carbon nanotube linear, ohmic I(V)-characteristic of a 20 µm long MWCNT 200 kω resistance (mainly due to contact resistance ) 20 current [ µa ] 10 0-10 rho [µωcm] 100 10 Au effective rho (Fuchs/Sondheimer) Au bulk value of rho Page 23-20 -4-2 0 2 4 voltage [ V ] 1 0 10 20 30 40 50 diameter [nm] yields a specific resistance of ρ MWCNT 10 µω cm 4*ρ bulk of Au
Our Vision Vision: smallest contact hole of the world filled with highly conductive carbon nanotube Vision Metal 2 CNT Dielectric Catalyst Metal 1 Page 24
Where we stand - our results how we make them- first attempt Objective: Vision Use selective growth of Carbon Nanotubes for high aspect ratio nanoscale vias Status Metal 2 400 nm CNTs Dielectric Catalyst Metal 1 Page 25
Where we stand - our results Electrical characterization of self-aligned grown CNT-Via Array of vertically aligned CNTs grown out of a 30x5 µm 2 wide via. The tops of the nanotubes are contacted by tungsten. R CNT W-contact Total Resistance = 11 Ω Contact Resistance =10 Ω Resistance of CNT Bundle = 1 Ω Contact Area = 5 x 30 µm 2 Number of CNTs ~ 10,000 i.e. ~3% CNTs in volume Quantum Resistance = 13 kω (one state, electron spins up and down) R sum Nanotubes Resistance per CNT ~ 10 kω Page 26 ~10 R Ta+Si Si-oxide Ta & catalyst Si- Substrate
Carbon Nanotubes contact holes (CH) I(U) Tungsten-contact CNT-array 5µm x 5µm SiO 2 Si-wafer GaIn-back-contact current [ ma ] 1.5 1.0 0.5 0.0-0.5 Top-view of the FIB-deposited -1.5 W-contact -400-200 0 200 400 voltage [ mv ] -1.0 Page 27 Before RTP (700 C, 30sec), the resistance was in the MOhm-range After RTP the resistance of 2 CH is about 339 Ohm / 370 Ohm The catalyst (Fe) still will form a nanoscale Schottkybarrier (~0.5 ev) and a native oxide adds to a series resistance
Carbon Nanotubes: two contact holes in series W-contact I(U) CNT-array 5µm x 5µm SiO 2 Si-wafer 300µm current [ ma ] 0.4 0.2 0.0-0.2-0.4-300 -200-100 0 100 200 300 voltage [ mv ] Page 28 Before RTP (700 C, 30sec), the resistance was in the MOhm-range After RTP the resistance of 2 CH is about 710 Ohm i.e the sum of the individual resistances The catalyst (Fe) still will form a nanoscale Schottky-barrier (~0.5 ev) and the native oxide adds to a series resistance
Carbon Nanotubes Contact Holes: bias dependent resistance Page 29 resistance [Ohm] 1000 900 800 700 600 370 360 350 340 330-0.0005 0.0000 0.0005 current [A] bias-dependent resistance due to Schottky barrier? contact hole chain series resistance of I +II single contact hole II single contact hole I
Carbon Nanotubes...outlook Present: Vertical growth of multi-walled CNTs between two contacts Relatively low contact resistance per nanotube achieved for vias but not for contact holes (formation of interface?) High current densities through isolated multi-walled nanotubes suggest ballistic transport Metallic behavior of isolated multi-walled CNTs suggests that single nanotubes vias are possible Future: Page 30 Improve nanotube density in vias & contact holes > 2% Improve contact resistance per tube: contact all layers/ contact metal Narrower vias 10-100 nm, high aspect ratio...lateral growth...temperature budget for process integration
Summary basics of Carbon Nanotubes possible general applications of CNTs Nanotubes as interconnects comparison of Au-wire versus CNT world wide first growth of CNT on 6-inch wafers First electrical characterization of single MW-CNT MW-CNT-Vias MW-CNTS contact holes outlook on future integration challenges Page 31
The people involved in the presented work Andrew Graham Eugen Unger Maik Liebau Georg Düsberg Werner Steinhögl Page 32