Dynamic Equipment and Process Simulation for Atomic Layer Deposition Technology

Transcription

1 Dynamic Equipment and Process Simulation for Atomic Layer Deposition Technology Wei Lei, Yuhong Cai, Laurent Henn-Lecordier and Gary W. Rubloff Department of Materials Science and Engineering and Institute for Systems Research University of Maryland, College Park

2 ALD Requirements Technology: Perfect atomic layer growth; Conformality; Uniformity; Material quality; Manufacturing: Cycle time; Advanced process control ESH (Environment, Safety, Health): High reactant utilization Energy Emissions Our work: Use dynamic simulation approach to study ALD equipment and process behavior: 1. Investigate tradeoff among technology, manufacturing and ESH metrics in ALD 2. Evaluate ALD equipment and process design strategy 3. Support our experimental program: process optimization and advanced process control of ALD

3 Methodology Current physical and chemical understanding Equipment and Process Model (physics-based, empirical, reduced order) Experimental Observation Verification Dynamic behavior through process cycle Simulation Tool Guidelines for equipment and process design Simulation-based technology, manufacturing and ESH Metrics

4 Dynamic Simulator for ALD Process Recipe Pressure, Exposure/Purge Time, Temperature etc. Operation Mode: Dynamic or Static Equipment Simulator Chamber, Mass flow controllers, Pumps, Valves; Partial and total pressure; Pressure control system; Process Simulator Process Process Sensing Sensing System System Differential Pumping Quadrupole Mass Spectroscopy ALD Reaction (Tungsten Nitride from WF 6 and NH 3 ) Gas phase transport Reactant adsorption and byproduct desorption Surface-condition-dependent reaction rates - surface kinetics Wafer State Deposition rate Film thickness Material quality Technology, Manufacturing and ESH Metrics Technology Material Quality Uniformity Conformality Manufacturing Cycle time Equipment cost, reliability ESH Reactant utilization Energy Emission

5 ALD Process Recipe Strategy Two possible operation modes for ALD process: Static Mode Half-reaction finishes starts On/off valve Gas Gas inlet inlet ALD Reactor (P < = Target P) Gas Gas outlet Molecular flow condition Less reactant in reactor Fast purge Molecular flow condition is preferred in static mode

6 ALD Process Recipe Strategy Dynamic Mode Reactant Carrier and gas and carrier only carrier gas gas Reaction Purge starts Throttle valve Viscous flow condition Gas inlet Gas inlet ALD Reactor (P = < Target P) Viscous carrier gas flow replaces reactant effectively during purge Gas Gas outlet Fast purge Viscous flow condition is preferred in dynamic mode Simulation Recipe Molecular Flow Condition ( 100 mtorr) Viscous Flow Condition ( 10 Torr) Static Mode Candidate to evaluate Dynamic Mode Candidate to evaluate

7 ALD Process Behaviors Dynamic simulation reveals fundamental science and equipment behavior 0.10 Reactor Static Mode Pressure (Torr) WF 6 Byproduct NH 3 Decreasing slope as reaction saturates surface coverage 1.0 WF 6 Pressure: 1 mtorr Time (Sec) Simulation results - Dynamic behavior; - Basis for overall time-integrated behavior metrics; Surface Coverage Delayed onset of surface coverage as reactant fills the chamber Exposure Time (Sec)

8 Technology and Manufacturing Metrics Tradeoff between technology performance and manufacturing efficiency In each ALD half-cycle, material quality may degrade if surface coverage is not quite complete Surface Coverage WF 6 Pressure: 1 mtorr Require high surface coverage for material quality Significant manufacturing throughput penalty if extend half-cycle to increase final surface coverage slightly Exposure Time (Sec) APC may be critical in ALD: use end point control to optimize coverage (material quality) vs. cycle time

9 Manufacturing Metrics Gas cycling is the limiting factor for ALD throughput Purge 0.10 V:1L V:3L V:5L V:1L V:3L V:5L Reactant Pressure (Torr) Increasing reactor volume Static Mode Reactant Pressure (Torr) Increasing reactor volume Dynamic Mode Purge Time (Sec) Purge Time (Sec) Shorter purge time can be achieved with smaller reactor volume Fill Gas filling time Reactor volume Flow rate Smaller reactor volume also leads to faster gas filling up Small reactor volume is important to achieve fast gas filling/purge in ALD

10 ESH Metrics Precursor utilization is an important factor in ESH Metrics Utilization (%) = Total Reacted Reactant Total Input Reactant Increasing utilization as surface reaction happens in the beginning of each half cycle Cycle A Purge Cycle B Purge Dynamic Mode Deceasing utilization as reaction saturates surface coverage while reactant is still being supplied Overall utilization in one cycle WF 6 Utilization (%) Time (Sec) Constant utilization after the half cycle

11 ESH Metrics 100 Cycle A Purge WF 6 Utilization (%) Static Mode Dynamic Mode Time (Sec) In static mode, utilization remains constant once surface reaction and coverage are complete In dynamic mode, precursor utilization continues to decrease because reaction saturates and reactant is still supplied Normally less reactant will be wasted in static mode than dynamic mode ; Less reactant is required to fill small reactor Small reactor also benefit ESH;

12 Equipment and Process Design Strategy Dynamic simulation also benefits equipment and process design strategy Key requirements: Small ALD reactor volume; Ability to operate under both static and dynamic modes; Our answer: Small ALD reactor embedded in big vacuum chamber Small ALD Reactor Vacuum Chamber Gas inlet 10-5 torr Gas outlet/sampling Wafer Substrate heater

13 Equipment and Process Design Strategy Two operational modes Precursor Purge B A exposure Static Mode ALD Reactor Gas inlet Gas inlet Vacuum Chamber Vacuum Chamber torr 0.1 torr Sampling (Only little gas Removal) Wafer Substrate heater Vacuum Chamber 10-5 torr Dynamic Mode Gas Inlet 10 torr Gas Outlet

14 ALD Reactor Design Motion Device 8 inch 6-way cross as the vacuum chamber Reactor cap Gas inlet Gas outlet ALD reactor Small ALD reactor volume: ~ 0.5 L; Vacuum chamber maintained at low pressure (10-5 Torr) via turbo pump; Gas inlet To gas exhaust and QMS Motion Device

15 ALD Reactor Design Main vacuum chamber Movable reactor cap Loadlock

16 Sensor System Integration Strategy Dynamic simulation benefits sensor system design for ALD Vacuum Chamber Valve Orifice QMS System ALD reactor Sampling tube e.g. optimization of orifice and sampling tube size on response time Response Time (Sec) Orifice diameter: 150 micron Orifice diameter: 200 micron Orifice diameter: 250 micron 0 1/8 1/4 1/2 Tube Diameter (inch)

17 Summary (1) Dynamic simulation is effective to identify synergies and tradeoffs among technology, manufacturing and ESH metrics (2) Dynamic simulation results provides guideline for our ALD equipment and process design: Smaller ALD reactor volume will accelerate gas cycling and increase precursor utilization, benefiting both manufacturing and environment Normally, static mode has a higher precursor utilization than dynamic mode (3) QMS sensor system integration has been optimized through dynamic simulation analysis Dynamic simulation approach has helped us to better understand ALD process and supported our experimental program