Process Analysis and Control of 200 mm Sputter Etch Equipment

Transcription

1 Process Analysis and Control of 200 mm Sputter Etch Equipment Gernot Bauer 1, Robert Fischer 1, Michael Klick 2 1 Texas Instruments Deutschland GmbH, 2 Plasmetrex GmbH 1

2 Outline: Process issue and understanding RF characterization of chamber parts Failure Analyses Solution by RF system refurbishment Results - Failure statistic Summary 2

3 Process Issues: Reflected Power Warnings / Aborts Our Aviza Hot Sputter Etch chamber (HSE) frequently suffered from: Platen RF Reflected Power Warnings/ or Aborts Platen RF Reflected Power Volt Watts Accompanied by instable etch plasmas and therefore Platen RF DC Bias Warnings/ or Aborts Platen RF DC Bias 3

4 Arcing Opened Chambers showed signs of Arcing On the chamber lid Or in strong cases even black residuals on the chamber walls Resulting in increased DD and single wafer Scrap 4

5 Chamber Setup - Phase Angle How it works Synchronization of coil and platen generator: Coil generator Master ~ Δφ coil cable CEX OUT Δφ MB matchbox coil Platen generator Δ φ sync switch box ~ Slave CEX IN Δφ platen cable matchbox platen Δφ MB Δφ Δφ chamber 5

6 Why is the Phase Angle important for the Etch Rate & Damage? Voltage [V] Phase Angle between Coil and Platen impacts Bias voltage Ion energy Etch rate & Damage measured by TI recalculated with corrected values HSE Chamber with real phase angle Vp [V] Vbias [V] Phase chamber [ ] 6

7 Why is the Phase Angle Important for Process Stability? RF Hardware stability RF current in chamber Process stability Stable Wafer RF current ion energy etch rate no big influence Nonlinear plasma model for Aviza HSE by Plasmetrex High RF current variations at chamber lid Process instability, ignition problems, aborts, arcing 7

8 RF Characterization of Chamber Parts The main Principle Generators Phase shifter (SSB = Synchron Switch Box) Entire system 8

9 Phase Angle How it works Synchronization of coil and platen generator: Coil generator Master ~ Δφ coil cable CEX OUT Δφ MB matchbox coil Platen generator Δ φ sync switch box ~ Slave CEX IN Δφ platen cable Δφ MB Δφ matchbox platen Δφ chamber 9

10 What is needed for stable RF Hardware and Process? The control signal from the Master Generator (CEX OUT) must be well defined in shape and phase angle: ideally a rectangular signal. Transmission of the control signal must not influence neither shape nor phase significantly. The Slave Generator must use the input signal (CEX IN) correctly. 10

11 Generator: Shape of Control Signal at Output Variations across different Generators: Gen.1 Gen.2 Gen.3 Gen.4 Gen.5 Gen.6 Gen.7 Gen.8 The ideal shape of the control signal should be a rectangular signal. This signals are recorded with dummy loads - real loads are even worse Possible reason: Generator aging. Measurement Equipment used: Digital 500 MHz Scope HP 54615A with various hf probes. 11

12 delay of CEX OUT control signal [ ] Generator: Phase Variance of Control Signals The shown variance in the control signal referred to RF power output is directly transferred to the same phase variance at the chamber! The tool offers adjustment in steps of 11! Any change of a generator can lead to a nonfunctional chamber as often experienced Measurement: Falling edge of power output to falling edge of control signal Gen.1 Gen.2 Gen.3 Gen.4 Gen.5 Gen.6 Gen.7 Gen.8 Generator influence on control signal 12

13 SSB = Synchron Switch Box The SSB is designed to delay the control signal (=phase shifter); With dip switches cables of different length are combined It offers adjustments in increments of 11,25 13

14 Poor Transmission due to Signal Reflections at Switches inside the SSB 1 (= Synchron Switch Box 1) Measurement: Reflection with open end Network analyzer FSH 3 with VSWR bridge FSH-22 (Rohde & Schwarz) Length of feed cable 1 Reflections at end normal Issue: Signal reflections at end of feed cable and multiple ones inside at switches Reflections lead to bad signal transmission and so to phase deviations 14!

15 . SSB 2 (= Synchron Switch Box 2) Measurement: Reflection with open end Network analyzer FSH 3 with VSWR bridge FSH-22 (Rohde & Schwarz) Length of feed cable 2 Reflections at end normal Issue: Signal reflections at end of feed cable and multiple ones inside at switches Reflections lead to bad signal transmission and so to phase deviations! 15

16 Control Signal before and after SSB Each SSB modulates the control signal differently! Before SSB: Control signals After SSB: Tool1 Tool2 Tool3 16

17 Worse Control Signals when SSB is connected to CEX OUT of Coil Generator Tool1 Tool2 control signal Δφ Tool3 Phase angle between coil control signal and Platen RF power output (falling edge to falling edge) Platen RF power output 17

18 Chamber1 Chamber3 Phase difference of coil and platen generator [ ] Chamber2 Total Phase Variance of three Chambers Phase deviations of Generators alone Total phase from two measurements = (control signal output of coil generator) + (platen output control signal) Test conditions: BKM recipe, 400W platen, 300 W coil, Ar Range of phase error: 56 Can lead to different process results and unstable process (aborts) Phase error finally depends on the combination of generator, SSB, and SSB adjustment. 18

19 Baseline State Failure Analysis (Summary) Problem #1: poor control signal on CEX OUT Master Master: Coil RF Generator 13,56 MHz Coax 50 Ohm Coil Match Plasma Argon line Coil Wafer Problem #2: degraded control signal behind synchron switch box Sync Switch Box = phase shifter Chuck Problem # 3: each SSB modulates the signal differently Problem # 4: 0 or 180 phaseshift on CEX IN Slave: Platen RF Generator 13,56 MHz Coax 50 Ohm Platen Match Cryo pump 19

20 Solution by Refurbishment of RF System Implementation of external Master Oscillator with stable and good shaped control signal Change the generator synchronization from Master/Slave to Slave/Slave Exchange of the SSB with fixed phase delay RF cable of defined length 20

21 Solution by Refurbishment of RF System Argon line Slave: Coil RF Generator 13,56 MHz Coax 50 Ohm Coil Match Plasma Coil Wafer Master oscillator 13,56 MHz Chuck phase shift cable Cryo pump Slave: Platen RF Generator 13,56 MHz Coax 50 Ohm Platen Match 21

22 Modified Chambers Increased Stability Current Status: All chambers are now modified. Baseline: 135 warnings 78 aborts mostly due to high reflected power. Modified Chambers: 2 warnings 2 aborts Compare of three Chambers on high Utilization Timeframe: cumulative 13 months before and after modification 22

23 Summary Issue: Aging of generators, low quality of cables, and switches in phase shifter leads to unstable synchronization of coil and platen generator. Range observed at three chambers: 55. This results in risk of: Process aborts / scrap Chamber down after generator is changed Decreased uptime and OEE and increase of work load and costs of repair Solution: Control by external master control signal generators Replacement of all SSBs by customized cable set Test method already verified 23