In-situ Ohmic Contacts to p-InGaAs Ashish Baraskar, Vibhor Jain, Evan Lobisser, Brian Thibeault, Arthur Gossard and Mark Rodwell ECE and Materials Departments, University of California, Santa Barbara, CA Mark Wistey Electrical Engineering, University of Notre Dame, IN 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 1 Outline • Motivation – Low resistance contacts for high speed HBTs – Approach • Experimental details – Contact formation – Fabrication of Transmission Line Model structures • Results – Doping characteristics – Effect of doping on contact resistivity – Effect of annealing • Conclusion 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 2 Outline • Motivation – Low resistance contacts for high speed HBTs – Approach • Experimental details – Contact formation – Fabrication of Transmission Line Model structures • Results – Doping characteristics – Effect of doping on contact resistivity – Effect of annealing • Conclusion 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 3 Device Bandwidth Scaling Laws for HBT To double device bandwidth: We • Cut transit time 2x Tb • Cut RC delay 2x Wbc Tc Scale contact resistivities by 4:1* 1 in RC 2f f max HBT: Heterojunction Bipolar Transistor f 8 Rbb Ccb eff *M.J.W. Rodwell, CSICS 2008 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 4 InP Bipolar Transistor Scaling Roadmap 256 128 64 32 nm width 8 4 2 1 Ω·µm2 access ρ 175 120 60 30 nm contact width 10 5 2.5 1.25 Ω·µm2 contact ρ 106 75 53 37.5 nm thick 9 18 36 72 4 3.3 2.75 2-2.5 V breakdown fτ 520 730 1000 1400 GHz fmax 850 1300 2000 2800 GHz Emitter Base Collector mA/µm2 current Contact resistivity serious barrier to THz technology Less than 2 Ω-µm2 contact resistivity required for simultaneous THz ft and fmax* 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 5 Approach To achieve low resistance, stable ohmic contacts • Higher number of active carriers - Reduced depletion width - Enhanced tunneling across metal- semiconductor interface • Better surface preparation techniques - For efficient removal of oxides/impurities 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 6 Approach (contd.) • Scaled device thin base (For 80 nm device: tbase < 25 nm) • Non-refractory contacts may diffuse at higher temperatures through base and short the collector • Pd/Ti/Pd/Au contacts diffuse about 15 nm in InGaAs on annealing Need a refractory metal for thermal stability 15 nm Pd/Ti diffusion 100 nm InGaAs grown in MBE 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 7 Outline • Motivation – Low resistance contacts for high speed HBTs and FETs – Approach • Experimental details – Contact formation – Fabrication of Transmission Line Model structures • Results – Doping characteristics – Effect of doping on contact resistivity – Effect of annealing • Conclusion 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 8 Epilayer Growth Epilayer growth by Solid Source Molecular Beam Epitaxy (SS-MBE)– p-InGaAs/InAlAs - Semi insulating InP (100) substrate - Un-doped InAlAs buffer - CBr4 as carbon dopant source - Hole concentration determined by Hall measurements 100 nm In0.53Ga0.47As: C (p-type) 100 nm In0.52Al0.48As: NID buffer Semi-insulating InP Substrate 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 9 In-situ contacts In-situ molybdenum (Mo) deposition - E-beam chamber connected to MBE chamber - No air exposure after film growth Why Mo? - Refractory metal (melting point ~ 2620 oC) - Easy to deposit by e-beam technique - Easy to process and integrate in HBT process flow 20 nm in-situ Mo 100 nm In0.53Ga0.47As: C (p-type) 100 nm In0.52Al0.48As: NID buffer Semi-insulating InP Substrate 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 10 TLM (Transmission Line Model) fabrication • E-beam deposition of Ti, Au and Ni layers • Samples processed into TLM structures by photolithography and liftoff • Contact metal was dry etched in SF6/Ar with Ni as etch mask, isolated by wet etch 50 nm ex-situ Ni 500 nm ex-situ Au 20 nm ex-situ Ti 20 nm in-situ Mo 100 nm In0.53Ga0.47As: C (p-type) 100 nm In0.52Al0.48As: NID buffer Semi-insulating InP Substrate 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 11 Resistance Measurement • Resistance measured by Agilent 4155C semiconductor parameter analyzer • TLM pad spacing (Lgap) varied from 0.5-26 µm; verified from scanning electron microscope (SEM) • TLM Width ~ 25 µm 5 Resistance () 4 3 2 2 RC 1 0 0 0.5 1 1.5 2 2 C RSh W 2.5 3 3.5 Gap Spacing (m) 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 12 Error Analysis • Extrapolation errors: • Processing errors: – 4-point probe resistance measurements on Agilent 4155C – Resolution error in SEM – Variable gap spacing along width (W) – Overlap resistance 3.5 Resistance () 3 2.5 Variable gap along width (W) 1.10 µm 1.04 µm dR 2 dd 1.5 Lgap 1 0.5 0 dRc 0 1 W 2 3 4 Gap Spacing (m) 2010 Electronic Materials Conference 5 6 Overlap Resistance June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 13 Outline • Motivation – Low resistance contacts for high speed HBTs and FETs – Approach • Experimental details – Contact formation – Fabrication of Transmission Line Model structures • Results – Doping characteristics – Effect of doping on contact resistivity – Effect of annealing • Conclusion 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 14 Doping Characteristics-I Hole concentration Vs CBr4 flux 80 70 Tsub = 460 oC 60 2 hole concentration Mobility (cm -Vs) -3 Hole Concentration (cm ) 10 20 mobility 50 40 10 19 0 10 20 30 40 50 CBr foreline pressure (mtorr) 60 4 – Hole concentration saturates at high CBr fluxes – Number of di-carbon defects as CBr flux 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 15 Doping Characteristics-II 10 19 Hole Concentration (10 ) cm -3 Hole concentration Vs V/III flux 8 CBr = 60 mtorr 6 4 2 10 20 30 40 50 60 Group V / Group III As V/III ratio hole concentration hypothesis: As-deficient surface drives C onto group-V sites 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 16 Doping Characteristics-III Hole concentration Vs substrate temperature 20 Hole Concentration (cm -3) 2 10 20 1.6 10 20 1.2 10 19 8 10 CBr = 60 mtorr 19 4 10 300 350 400 450 o Substrate Temp. ( C) Tendency to form di-carbon defects as Tsub * *Tan et. al. Phys. Rev. B 67 (2003) 035208 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 17 Doping Characteristics-III Hole concentration Vs substrate temperature 20 -3 Hole Concentration (cm ) Hole Concentration (cm -3) 2 10 20 1.6 10 20 1.2 10 19 8 10 CBr = 60 mtorr 19 4 10 300 350 400 10 20 o Tsub = 350 C o Tsub = 460 C 10 19 0 450 4 o Substrate Temp. ( C) Tendency to form di-carbon defects 100 80 60 40 20 CBr foreline pressure (mtorr) as Tsub * *Tan et. al. Phys. Rev. B 67 (2003) 035208 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 18 Results: Contact Resistivity - I Metal Contact ρc (Ω-µm2) ρh (Ω-µm) In-situ Mo 2.2 ± 0.8 15.4 ± 2.6 20 15 Resistance () • Hole concentration, p = 1.6 x 1020 cm-3 • Mobility, µ = 36 cm2/Vs • Sheet resistance, Rsh = 105 ohm/ (100 nm thick film) 10 5 2 RC ρc lower than the best reported contacts to pInGaAs (ρc = 4 Ω-µm2)[1,2] 0 0 1 2 C RSh W 2 3 4 Gap Spacing (m) 1. Griffith et al, Indium Phosphide and Related Materials, 2005. 2. Jain et al, IEEE Device Research Conference, 2010. 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 19 2 Contact Resistivity, c (-m ) Results: Contact Resistivity - II 10 Tunneling 1 C exp p * Thermionic * ~ constant c Emission Data suggests tunneling 1 5 10 15 [p]-1/2 (x10-11 cm-3/2) High active carrier concentration is the key to low resistance contacts * Physics of Semiconductor Devices, S M Sze 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 20 Thermal Stability - I Mo contacts annealed under N2 flow for 60 mins. at 250 oC Before annealing After annealing 2.2 ± 0.8 2.8 ± 0.9 ρc (Ω-µm2) Molybdenum C substrate • ρc increases on annealing • Mo reacts with residual interfacial carbon?* Kiniger et. al.* *Kiniger et. al., Surf. Interface Anal. 2008, 40, 786–789 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 21 Thermal Stability - II Mo contacts annealed under N2 flow for 60 mins. at 250 oC TEM of Mo-pInGaAs interface - Suggests sharp interface - Minimal/No intermixing Au Ti Mo InGaAs 200 nm 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN InAlAs Ashish Baraskar 22 Summary • Maximum hole concentration obtained = 1.6 x1020 cm-3 at a substrate temperature of 350 oC • Low contact resistivity with in-situ metal contacts (lowest ρc=2.2 ± 0.8 Ω-µm2) Contacts suitable for THz transistors 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 23 Thank You ! Questions? Acknowledgements ONR, DARPA-TFAST, DARPA-FLARE 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 24 Extra Slides 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 25 Correction for Metal Resistance in 4-Point Test Structure Rmetal I W L ( sheet contact )1 / 2 / W sheet L / W V V I ( sheet contact )1/ 2 / W sheet L / W Rmetal / x Error term (Rmetal/x) from metal resistance 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 26 Random and Offset Error in 4155C 0.6647 • Random Error in resistance measurement ~ 0.5 m • Offset Error < 5 m* Resistance () 0.6646 0.6645 0.6644 0.6643 0.6642 0 5 10 15 20 25 Current (mA) *4155C datasheet 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 27 Accuracy Limits • Error Calculations – dR = 50 mΩ (Safe estimate) – dW = 1 µm – dGap = 20 nm • Error in ρc ~ 40% at 1.1 Ω-µm2 2010 Electronic Materials Conference June 23-25, 2010 – Notre Dame, IN Ashish Baraskar 28