Control of Spatial Uniformity in Microelectronics Manufacturing: An Integrated Approach

Transcription

1 Control of Spatial Uniformity in Microelectronics Manufacturing: An Integrated Approach H. Zhang & M. Nikolaou Chemical Engineering Department University of Houston

2 Real-Time Feedback Control of Spatial Uniformity: Limitations Spatial uniformity sensors Plasma sensors Models Feedback control on-line computations

3 Objectives Overall Develop and experimentally implement an integrated methodology to model, measure, and control spatial uniformity in semiconductor processing This presentation Simulation study on real-time feedback control of plasma etching uniformity

4 Modeling Use Modular Plasma Reactor Simulator (MPRES) Develop control-oriented reduced-order models

5 Sensors CCD interferometry camera (Leybold- Inficon, LES 1200) Need Time resolution Spatial resolution

6 Feedback Control Control strategy: completion of the task within finite time precise arrival at a final target satisfaction of intermediate constraints Analogs: Aircraft landing Batch chemical reactor control

7 Feedback Control (Cont d( Cont d) Wafer axis Photo resist Etched layer i Substrate Etch target line

8 Case Study Chlorine etching of polysilicon Empirical model: output k = f(input k-1 )

9 Model Reduction The basic idea: y=mz Mz c Examples: Principal component regression Wavelet compression

10 Model Reduction (Cont d( Cont d) 5 x wavelet bior x PCA

11 Reduced Model from First-Principles Model Network model

12

13

14

15

16

17

18 Control Objective At end of run, t K Etch wafer completely: y i 0 Minimize non-uniformity: y i 0 Wafer axis Photo resist Etched layer i Substrate Etch target line

19 Feedback Strategy At each time step, t k : minimize v( t ),, v( t ), t k K 1 K N E yˆ ( t t ) i= 1 i K k p 1/ p subject to Py [ ˆ ( t t) 0] 1 α i K k umin v( t j ) u max t t t K K K min i = 1,2,..., N max

20 Results Etch rate (A/s) Time, s

21 Results (Cont d( Cont d) input power (W) flow rate (sccm) Time, s Time, s DC bias amplitude (Volt) pressure (mtorr) Time, s Time, s

22 Results (Cont d( Cont d) No noise and no input move/output constraints Show 150 more simulation results etch rate (Angstrom /Second) time (Second) overetch (Angstrom) -0.7 x radius

23 Results (Cont d( Cont d) input power (W) pressure (mtorr) flow rate (sccm) bias amplitude(volt)

24 Input move [ ]; etch rate constraint [45 to65];noise case etch rate (Angstrom/second) overetch (Angstrom) radius

25 Results (Cont d( Cont d) input power (W) flow rate (sccm) pressure (mtorr) bias amplitude(volt)

26 3000 etch profile every 5 seconds waferthickness (Angstrom) radius

27 Input move [ ];etch rate [10 80]; noise case ERpediction/ERreal =0.8 etch rate (Angstrom) overetch depth (Angstrom) radius

28 Results (Cont d( Cont d) input power (W) pressure (mtorr) pressure (flow rate (sccm)) bias amplitude (volt))

29 3000 etch profile at every five seconds 2500 waferthickness (Angstrom) radius

30 Input move [ ];etch rate [10 80]; noise case ERp/ERr =0.5; the whole etch done in 53 seconds etch rate (Angstrom/second) overetch depth (Angstrom) radius

31 Results (Cont d( Cont d) input power (w) pressure (mtorr) flow rate (sccm) bias amplitude(v)

32 3000 etch profile at every five seconds waferthickness (Angstrom) radius

33 Input move [ ];etch rate [10 80]; noise case ERprediction/ERreal =1.2; the whole etch done in 62 seconds etch rate (Angstrom /Second) over-etch depth (Angstrom) radius

34 Results (Cont d( Cont d) rf power flow rate time time pressure bias amplitude time time

35 waferthickness (Angstrom) radius

36 Conclusions Real-time control of uniformity: Integrated approach Model, sensor, control Feedback control: Batch process approach

37 Future Work Experimental validation of proposed approach Refinement of control strategy: Model Sensor Control

38 Acknowledgments Prof. D. Economou Applied Materials