PROGRESS, OPPORTUNITIES AND CHALLENGES IN MODELING OF PLASMA ETCHING Yang Yang, Mingmei Wang, Juline Shoeb and Mark J. Kushner Iowa State University Ames, IA 50011 USA mjk@iastate.edu http://uigelz.ece.iastate.edu May 2008 IITC_0508

Optical and Discharge Physics ACKNOWLEDGEMENTS Group Members Ananth N. Bhoj (Novellus) Ramesh Arakoni (Intel) Natalie Y. Babaeva Ankur Agarwal (Appl. Matls.) Yang Yang Mingmei Wang Juline Shoeb Luis Garcia (UG RA) Samantha Eischeld (UG RA) Funding Agencies 3M Corporation Applied Materials Micron, Inc. Semiconductor Research Corporation Iowa State University Optical and Discharge Physics IITC_0508 2

Optical and Discharge Physics AGENDA Modeling of Plasma Processing Philosophical and Hierarchical Basis of Hybrid Modeling Challenges and opportunities Nanoscale processing and atomic layer control Extremely high aspect ratio processing Doing without data Green processing Concluding Remarks Iowa State University Optical and Discharge Physics IITC_0508

PLASMAS FOR NANOSCALE FABRICATION Optical and Discharge Physics Plasmas are and will continue to be indispensable for etching, deposition and cleaning in microelectronics fabrication. Control of dimensions at 32 and 45 nm nodes requires resolution of a few nm or less. Required: Unprecedented control of reactant fluxes from the plasma onto the wafer: Uniformity, Composition, Energies Modeling is being challenged to provide added value to process and equipment development. Mechanisms Timeliness and Relevancy Being ahead of the curve Iowa State University Optical and Discharge Physics http://www.intel.com http://realworldtech.com, IITC_0508 4

ACTIVATION ENERGY: SUB-eV, SUB-DEGREE CONTROL Activation energy for fabricating nanostructures is largely delivered through ion bombardment. Distinguishing between materials will result from sub-eV and sub-degree control of ion energies. Intel Fin-FET Iowa State University Optical and Discharge Physics IITC_0508 5

WHY IS THIS SO DIFFICULT TO MODEL?: TYPICAL PLASMA ETCHING REACTOR  Hitachi XT ECR Iowa State University Optical and Discharge Physics http://www.hitachi-hta.com IITC_0508 6

THE TIMESCALE CHALLENGE Optical and Discharge Physics Technological plasmas have vastly different timescales that must be addressed in models. Integrating timestep (required for numerical stability): t Dynamic timescale (to resolve the evolution of phenomena): T Plasma transport: Dielectric relaxation t = /  1 ps – 10 ns T = ns - ms Surface chemistry: t = s, T = 10 s Iowa State University Optical and Discharge Physics IITC_0508

HYBRID MODELING IITC_0508

THE FUNDAMENTAL APPROACH Optical and Discharge Physics Virtually all processes in plasmas are interdependent. Electron kinetics depend on circuitry, gas flow, surfaces…. Plasma harmonics depend on cable lengths…. The most fundamental and accurate modeling technique is capturing all interdependencies in a massive set of partial differential equations (or equivalent PIC). Integrate, integrate, integrate….This is beyond the current state of the art and presents computational challenges on the par of weapons design. Iowa State University Optical and Discharge Physics IITC_0508

Optical and Discharge Physics HYBRID MODELING TECHNIQUES Hybrid models resolve multi-physics over multi-scales. Compartmentalize physical processes into modules having minimum of overlap. Time slice on physics times-scales. Iowa State University Optical and Discharge Physics IITC_0508 10

Optical and Discharge Physics MODELING PLATFORMS: HPEM, nonPDPSIM, MCFPM HPEM (Hybrid Plasma Equipment Model) More mature, more sophisticated physics Lower pressures (sub-mTorr to 10s Torr) Surface chemistry, radiation transport, circuits Virtual sensors-actuators-real time control Structured rectilinear meshes, generally semi-implicit nonPDPSIM Higher pressures (10s mTorr to multi-atmosphere) Less mature but improving physics Unstructured meshes, fully implicit MCFPM (Monte Carlo Feature Profile Model) Evolution of surface features nm-to-atomic scale Plasma and surface chemistry, charging Iowa State University Optical and Discharge Physics IITC_0508 11

Optical and Discharge Physics ELECTROMAGNETIC FIELDS AND ELECTRONS Maxwell’s Equation: Boltzmann’s equation for electron velocity distributions coupled with electron energy conservation equations. S(Te) = Power deposition from electric fields L(Te) = Electron power loss due to collisions  = Electron flux (Te) = Electron thermal conductivity tensor SEB = Power source source from beam electrons Iowa State University Optical and Discharge Physics IITC_0508

Optical and Discharge Physics PLASMA CHEMISTRY, TRANSPORT AND ELECTROSTATICS Continuity, momentum and energy equations are solved for each species Implicit solution of Poisson’s equation Iowa State University Optical and Discharge Physics IITC_0508

Optical and Discharge Physics SURFACE MORPHOLOGY AND PROFILE EVOLUTION Monte Carlo Feature Profile Model Psuedoparticles representing neutrals, ions and electrons are directed towards surface. Energy and angularly resolved fluxes from equipment scale models. Reaction mechanism determines probability of adding, removing or changing composition of materials. Atomic to a few nm resolution per mesh cell. Solution of Poisson’s equation to account for charging of feature. Iowa State University Optical and Discharge Physics IITC_0508

AFFECT PROCESS DESIGN TO ACHIEVE EXTREME SELECTIVITY MODELING CHALLENGE: AFFECT PROCESS DESIGN TO ACHIEVE EXTREME SELECTIVITY IITC_0508 15

FLUOROCARBON PLASMA ETCHING OF Si/SiO2 Optical and Discharge Physics In fluorocarbon plasma etching of Si/SiO2 CFx radicals produce a polymer passivation layer which regulates activation energy. SiO2 consumes the polymer in etching; thicker layers on Si result in slower etch rates, and the ability to achieve selectivity. Iowa State University Optical and Discharge Physics IITC_0508 16

ACHIEVING HIGH SELECTIVITY Optical and Discharge Physics High selectivity achieved by controlling the ion energy distribution to wafer. Sinusoidal bias: Broad IEAD does not discriminate thresholds. Tailored bias: Produce a narrow IEAD which may discriminate between threshold energies. Sinusoidal Bias Ref: S.-B. Wang and A.E. Wendt, J. Vac. Sci. Technol. A, 19, 2425 (2001) Iowa State University Optical and Discharge Physics IITC_0508 17

NON-SINUSOIDAL BIAS WAVEFORMS Optical and Discharge Physics Ions cross the sheath quickly compared to the rf period and arrive at wafer with nearly the instantaneous sheath voltage. By controlling the sheath potential with non-sinusoidal waveforms, one can control the IED. a Iowa State University Optical and Discharge Physics IITC_0508 18

IEDs: DURATION OF POSITIVE SPIKE Optical and Discharge Physics Custom bias produces constant sheath potential resulting in narrow IED. As  increases, IEDs broaden in energy and bias becomes less positive. 15 mTorr, 500 W, 5 MHz, 200 Vp-p, Ar/C4F8 = 75/25  Vdc:-73 -12 13 42 46 56 64 75 Iowa State University Optical and Discharge Physics IITC_0508 19

ETCH PROFILES vs VOLTAGE Optical and Discharge Physics Non-sinusoidal bias enables etching with nearly mono-energetic ions. Transition between “selective” and “fast” is unambiguous. 15 mTorr, 500 W, Ar/C4F8 = 75/25 Iowa State University Optical and Discharge Physics IITC_0508 20

Optical and Discharge Physics ETCHING RECIPES Etching usually occurs in 2 phases: Main-etch: Fast but non-selective Over-etch: Slow but selective Gas mixture is often changed to distinguish main and over-etch Slow due to gas exchange times. Unable to resolve monolayer Custom tailored voltage waveform Controlling physical component Change amplitude with ms resolution. Iowa State University Optical and Discharge Physics IITC_0508 21

ELECTRONICALLY CONTROLLED “RECIPES” Optical and Discharge Physics MASK SiO2 Si T=0.35 T=0.65 Electronically controlled recipes enable rapid etch with a “soft landing” 200 V (Slow, selective) 1500 V (Fast, non-selective) 1500/200 V (Fast, selective) Iowa State University Optical and Discharge Physics ANIMATION SLIDE-GIF IITC_0508 22

MODELING CHALLENGE: AFFECT EQUIPMENT DESIGN FOR ETCHING OF EXTREMELY HIGH ASPECT RATIO FEATURES IITC_0508 23

HIGH ASPECT RATIO CONTACT (HARC) ETCHING Optical and Discharge Physics Processes for HARC etching with aspect ratios > 50-100 are being developed for capacitors and through wafer vias. Very Challenging… Twisting or curvature of features is randomly observed. Features are so small that random fluctuations of fluxes of radicals, ions and electrons into holes produces variations. Charging of features by electrons and ions produce random fields that deviate paths. Iowa State University Optical and Discharge Physics IITC_0508 24 24

DUAL FREQUENCY CCP Ar/C4F8/O2 : CHARGING Optical and Discharge Physics High energy ions penetrate deep into feature. Electrons charge top of feature. Internal E-fields that affect trajectories. Randomness of charging leads to erratic paths. 10 mTorr, Ar/C4F8/O2 = 80/15/5, 10/40 MHz, 500 W. Iowa State University Optical and Discharge Physics IITC_0508 25 25

SiO2 / Si HARC ETCH: EFFECTS OF CHARGING Optical and Discharge Physics Etch rate higher with increasing power. Without charging: Generally straight profiles. With charging: Ion trajectories perturbed. Overcome with voltage. Some evidence of randomness due to small contact area. 10 mTorr, Ar/C4F8/O2 = 80/15/5, 10/ 40 MHz, 500 W. Without Charging With Charging Iowa State University Optical and Discharge Physics IITC_0508 26 26

SiO2/Si HARC ETCH: RANDOMNESS OF CHARGING? 6 trenches receiving “same” fluxes. Stochastic nature of fluxes produces random twisting. Similar behavior observed experimentally. Effect is amplified by finite size of particles and mesh. 10 mTorr, Ar/C4F8/O2 = 80/15/5, 300 sccm, LF 4 kW, HF 500 W. Iowa State University Optical and Discharge Physics IITC_0508 27 27

Optical and Discharge Physics IS THERE A FIX? DC-AUGMENTED RF Single (or dual) frequency CCP…with external, negative dc bias on opposing electrode. DC ion current produces dc e-beam current incident onto wafer. dc e-beam, mono-energetic and narrow in angle, penetrates deep into feature to neutralize excess positive charge. Iowa State University Optical and Discharge Physics IITC_0508

Optical and Discharge Physics 10 MHz LOWER, DC UPPER: [e] DENSITY, DC CURRENT Electron density 3 x 1010 cm-3. Dc voltage on upper electrode requires dc current path to ground. LF electrode blocking capacitor and insulators require dc current return to side wall. Dc current peaks near outer edge of electrode. Ar, 40 mTorr, LF: 10 MHz, 300 W, 440V/dc=-250V DC: 200 W, -470 V Iowa State University Optical and Discharge Physics IITC_0508

Optical and Discharge Physics 10 MHz LOWER, DC UPPER: PLASMA POTENTIAL LF electrode passes rf current. DC electrode passes combination of rf and dc current with small modulation of sheath potential. Ar, 40 mTorr, LF: 10 MHz, 300 W, 440V/dc=-250V DC: 200 W, -470 V Iowa State University Optical and Discharge Physics ANIMATION SLIDE-GIF IITC_0508

Optical and Discharge Physics 10 MHz LOWER, DC UPPER: [e], ION ENERGY DISTRIBUTIONS Ion energy distribution to wafer is many degrees, 150 eV in width. Electron energy distributions onto wafer is narrower in angle and broader in energy. E-beam reflects instantaneous potential difference between electrodes. Ar, 40 mTorr, LF: 10 MHz, 300 W, 440V/dc=-250V; DC: 200 W, -470 V Iowa State University Optical and Discharge Physics IITC_0508

Optical and Discharge Physics ETCH PROFILES: WITHOUT AND WITH E-BEAM CURRENT E-beam current neutralizes sufficient charge to prevent major twisting. Difference in etch depth results from randomness of fluxes. Ar/C4F8/O2, 40 mTorr Iowa State University Optical and Discharge Physics IITC_0508

CAN YOU BE AHEAD OF THE CURVE BY VIRTUALLY CREATING NEW PROCESSES? MODELING CHALLENGE: CAN YOU BE AHEAD OF THE CURVE BY VIRTUALLY CREATING NEW PROCESSES? IITC_0508 33

ANOTHER HARC CHALLENGE: MICRO-TRENCHING Optical and Discharge Physics Modeling will be most impactful by leading process development. As AR increases, specular ion reflection off side walls leads to micro-trenching. SiO2 remains when Si is exposed ….over-etching is required. How do you maintain critical dimension? Can modeling lead experiment? Ar/C4F8 = 75/25, 100 sccm, ICP, 15 mTorr, 500 W, 100 V at 5 MHz Si Mask SiO2 Aspect Ratio = 1:10 Iowa State University Optical and Discharge Physics ANIMATION SLIDE-GIF IITC_0508 34

NON-SELECTIVE SINUSOIDAL BIAS FOR OVERETCH Over-etch is used to clear out the remaining SiO2. Conventional etching using a sinusoidal waveform does not have enough selectivity. Underlying Si is breached. CD of atomic layer resolution is not maintained. Ar/C4F8 = 75/25, 100 sccm, ICP, 15 mTorr, 500 W, 100 V at 5 MHz Aspect Ratio = 1:10 Iowa State University Optical and Discharge Physics ANIMATION SLIDE-GIF IITC_0508 35

PLASMA ATOMIC LAYER ETCHING Optical and Discharge Physics Plasma Atomic Layer Etching (PALE) monolayer by monolayer removal of material in a cyclic, self-limiting process. Step 1: Top monolayer is passivated in non-etching plasma. Step 2: Remove top layer with lower threshold energy (self-limiting) Extreme control of energies to prevent erosion of the under-layer. Iowa State University Optical and Discharge Physics IITC_0508 36

Ar/c-C4F8 TAILORED BIAS PALE: IEADs Optical and Discharge Physics PALE of SiO2 using ICP Ar/C4F8 with variable non-sinusoidal bias. Step 1 Vp-p = 50 V Passivate single layer with SiO2CxFy Low ion energies to reduce etching. Step 2 Vp-p = 100 V Etch/Sputter SiO2CxFy layer. Above threshold ion energies. Narrow IEADs in energy discriminate between threshold energies. Ar/C4F8 = 75/25, 100 sccm, 15 mTorr, 500 W Iowa State University Optical and Discharge Physics IITC_0508 37

TAILORED BIAS PALE: EXTENDED OVER-ETCH Optical and Discharge Physics PALE is used for over-etch. Extreme selectivity of PALE enables removal of SiO2 without damage to underlying Si. 15 cycles of PALE over-etch clears the feature. More energy intensive per layer than conventional etching. Future technology nodes will be challenged to maintain lower J/cm2 of wafer. 15 cycles of PALE Iowa State University Optical and Discharge Physics ANIMATION SLIDE-GIF IITC_0508 38

ADDRESSING DAY-TO-DAY ISSUES OF TOOL DESIGN AND MAINTENANCE MODELING CHALLENGE: ADDRESSING DAY-TO-DAY ISSUES OF TOOL DESIGN AND MAINTENANCE IITC_0508 39

INTO WAFER FOCUS RING GAP Optical and Discharge Physics PLASMA PENETRATION INTO WAFER FOCUS RING GAP Wafers are beveled at edge with small gap (< 500 m) between wafer and focus ring for mechanical clearance. Penetration of plasma into gap produces particles and erodes materials. How large can gap be without having significant plasma penetration? PVD penetration persists after CMP http://www.micromagazine.com /archive/00/10/simpson.html Iowa State University Optical and Discharge Physics IITC_0508 40

Ar/CF4 CCP FOR SiO2 Etching Optical and Discharge Physics Dielectric etching performed in polymerizing capacitively coupled plasmas. Typical plasma densities are 1010-1011 cm-3. Ar/CF4 = 97/03, 10 MHZ, 90 mTorr, 300V, 300 sccm [CF3-] [F-] Iowa State University Optical and Discharge Physics MIN MAX Log scale IITC_0508 41

ELECTRON PENETRATION INTO GAP Optical and Discharge Physics  1.0 mm Gap  0.25 mm Gap Electron penetration into gaps in anode portion of cycle is nominal due to surface charging and sheath formation. Ar/CF4 = 97/03, 10 MHz, 90 mTorr, 300 V, 300 sccm ANIMATION SLIDE-GIF Iowa State University Optical and Discharge Physics MIN MAX Log scale IITC_0508 42

Ar+ PENETRATION INTO GAP Optical and Discharge Physics  1.0 mm Gap  0.25 mm Gap Ions penetrate into larger gap throughout the rf cycle provided size is commensurate with sheath width. Smaller gap receives only nominal flux. Ar/CF4 = 97/03, 10 MHz, 90 mTorr, 300 V, 300 sccm ANIMATION SLIDE-GIF Iowa State University Optical and Discharge Physics MIN MAX Log scale IITC_0508 43

ION PENETRATION vs GAP SIZE vs Optical and Discharge Physics DEBYE LENGTH Ion penetration into gap depends on size of gap relative to sheath thickness. Gap ≥ sheath thickness allows penetration. High plasma density tools produce smaller sheaths and more penetration. More erosion…more cost. Ar/CF4 = 97/03, 10 MHz, 90 mTorr, 300 V, 300 sccm Iowa State University Optical and Discharge Physics IITC_0508 44

ION ENERGY AND ANGULAR DISTRIBUTIONS Optical and Discharge Physics IEADs differ significantly on different surfaces in gap. Erosion of materials depends on energy and angle of incidence. MIN MAX Log scale Iowa State University Optical and Discharge Physics IITC_0508 45 45

CAPACITANCE OF FOCUS RING: IEAD Optical and Discharge Physics Penetration of potential into low capacitance focus ring produces lateral E-fields. IEAD on substrate is off normal… more erosion by sputtering. Iowa State University Optical and Discharge Physics IITC_0508 46 46

DEVELOPING REACTION MECHANISMS WITHOUT FUNDAMENTAL DATA MODELING CHALLENGE: DEVELOPING REACTION MECHANISMS WITHOUT FUNDAMENTAL DATA IITC_0508 47

HfO2 GATE STACK MODELING…WITHOUT DATA(!) Optical and Discharge Physics High-k metal oxides are being used as SiO2 replacements to minimize gate leakage. HfO2 has promising properties and can be integrated into current process streams. For process integration and speed, desirable to simultaneously etch entire gate stack…Success with Ar/BCl3/Cl2 plasmas. Challenge: Modeling is required to speed process development and optimization. There exists no fundamental database for process. Develop mechanism based on experience and data from literature. Iowa State University Optical and Discharge Physics IITC_0508

GOALS AND PREMISES FOR ETCH MECHANISM Optical and Discharge Physics Goal: High selectivity of HfO2 over Si at low ion energies to minimize damage. Premise 1: HfO2 etching is a multistep process requiring Hf-O bond breaking and separate removal of Hf and O. Premise 2: Si etch rate slowed by BClx polymer. BClx first produces Si-B bonding to assist deposition of polymer. Premise 3: TiN etching is analogous to other metals with volatile products (e.g., Al). Use experiments from literature to build mechanism. Material Etch Rate (A0 /min) Threshold (eV) HfO2 90 25 Si 100 29 TiN 400 _ Material Bond Bond (eV) HfO2 Hf-O 8.3 Si Si-Si 3.4 Ref: L.Sha and J. P. Chang, J. Vac. Sci. Technol. A, v. 22, 88 (2004) Iowa State University Optical and Discharge Physics IITC_0508

ETCHING MECHANISM IN Ar/BCl3/Cl2 PLASMA Optical and Discharge Physics HfO2 Etching Bond Breaking M+(g) + HfO2(s)  HfO(s) + O(s) + M(g) M+(g) + HfO(s)  Hf(s) + O(s) + M(g) Adsorption Cl(g) + Hf(s)  HfCl(s) BClx(g) + O (s)  BClxO(s) Etching M+ (g) + HFClx(s)  HfClx(g) + M(g) M+(g) + BClxO(s)  ByOClx(g) + M(g) Selectivity with respect to Si results from deposition of BClx polymer on Si BClx(g) + Si(s)  SiBClx(s) BClx(g) + SiBClx(s)  SiBClx(s) + Poly-BClx(s) Iowa State University Optical and Discharge Physics IITC_0508

HfO2 ETCH RATE AND SELECTIVITY Optical and Discharge Physics Photo-resist TiN HfO2 HfO2 etch rate increases with bias at expense of selectivity as sputtering of Poly-BClx increases. Micro-masking from PR sputtering is problematic. Si Ar/BCl3/Cl2 = 5/40/55, 5 mTorr, 300 W ICP Iowa State University Optical and Discharge Physics IITC_0508

SELECTIVITY: CALIBRATION AND COLLABORATION Complex reaction mechanism requires calibration and collabora-tion for key experimental validation. Example: Selectivity of HfO2 over Si requires layer of Poly-BClx, a competition between deposition and sputtering. BClx(g) + Si(s)  SiBClx(s) BClx(g) + SiBClx(s)  SiBClx(s) + P-BClx(s) M+(g) + P-BClx(s)  M(g) + BClx(g) In the absence of fundamental data, reaction mechanisms must be derived through sensitivity studies. Iowa State University Optical and Discharge Physics IITC_0508

MODELING CHALLENGE: GREEN PROCESSING IITC_0508 53

Optical and Discharge Physics BECOMING MORE GREEN: ROLE OF MODELING Fabs must become greener by reducing carbon footprint. Typical energy usage: 1 kW-hr/cm2 of wafer . About 1/3 of fab power is expended in process tools, most of them plasma tools. Can modeling be an enabling technology to reduce Energy/cm2 and aid in fabs becoming greener? Electricity Usage http://ateam.lbl.gov/cleanroom/Fact Hu and Chuah, Energy 28, 895 (2003) Iowa State University Optical and Discharge Physics IITC_0508

EXAMPLE: “PLASMA PHYSICS” TO MINIMIZE ENERGY USE PER WAFER Optimize plasma etching of SiO2 vs pressure to minimize energy usage. Change pressure while: Constant: Gas mixture, residence time, energy deposition/molecule Scale flow rate, total power with pressure. Inductively Coupled Plasma Base Case: Ar/C4F8/O2 = 80/18/2, 10 mTorr, 500 W, -200 Vdc Iowa State University Optical and Discharge Physics IITC_0508

PRESSURE SCALING: ETCH RATES, ENERGY… Optical and Discharge Physics Increase pressure  Increase fluxes  Increase etch rate More deposition  Thicker polymer layer  Eventual etch stop Energy savings (J/cm2) decreases with pressure. Electrical load savings by optimizing process: 200 kW Larger footprint of fab..more tools for same throughput? Base Case: Ar/C4F8/O2 = 80/18/2, 10 mTorr, 500 W, -200 Vdc Iowa State University Optical and Discharge Physics IITC_0508

Optical and Discharge Physics CONCLUDING REMARKS Modeling has enabled more rapid and economic development of plasma tools. Value added for process development is less clear though promising. Challenges and opportunities for modeling to lead process development: Controlling activation energy for nanoscale resolution Green process design “Aging” resistant tool design. Coordinated modeling-experiment collaboration early in development cycle can result in robust reactions mechanisms to optimize process. Iowa State University Optical and Discharge Physics IITC_0508