Use of Spectrograph-based OES for SiN Etch Selectivity and Endpoint Optimization

Transcription

1 Use of Spectrograph-based OES for SiN Etch Selectivity and Endpoint Optimization F. G. Celii and C. Huffman Texas Instruments, Inc., Dallas, TX, USA J. Hosch* and K. Harvey Verity Instruments, Carrollton, TX, USA *Presenter 1

2 Goals of the Project Develop a new process for etching SiN NO + F are active etch species in SiN etch Kastenmeier, Matsuo and Oehrlein, J. Vac. Sci. Technol., A17 (1999) 3179 Use N2 + O2 with CH2F2 in Ar to generate NO in the chamber Determine optimum gas mix for rapid etch Use OES to monitor NO concentration Use OES to detect endpoint 2

3 Ar AND Ar/N 2 PLASMA OES (v',v") = ER42B, Ar or Ar/N2, 500 W N 2 (C 3 Πu B 3 Πg) N2 (D 3 Σ u + B 3 Πg) Ar/N 2 Ar OES Intensity (a.u.) (0,0) (0,1) (0,2) (0,3) (4,1) (3,1) (2,1) (1,2) (1,3) (1,4) (1,5) (3,0) (2,0) (1,0) (0,0) (0,1) (0,2) (0,3) N2 + (B 2 Σ u + X 2 Σ g + ) (2,3) (2,4) (2,5) Ar* Ar* x10 N2 (B 3 Πg A 3 Σ u + ) Wavelength (nm) 3

4 OES TRANSITIONS Emission Diagram v" C v' Energy (cm -1 ) A B X Molecular Spacing (A) 4

5 Ar AND Ar/O 2 PLASMA OES ER42B, Ar or Ar/O2, 500 W (considered O, O 2, O 2 +, SiO) OES Intensity (a.u.) CO (b 3 Σ a 3 Π) (0,0) (0,1) (0,2) (0,3) (0,4) Ar/O 2 (x3) Ar Wavelength (nm) 5

6 Ar/N 2 /O 2 PLASMA OES ER42B, N2-O2-Ar vs. binary mixtures, 500 W N 2 (C 3 Π u B 3 Π g ) Ar/N 2 OES Intensity (a.u.) NO (A 2 Σ + X 2 Π) (0,0) (0,1) (0,2) (0,3) (0,4) Ar/O 2 (x2) Ar/N 2 /O (2,0) (1,0) (1,2) (1,3) (1,4) (1,5) Wavelength (nm) 6

7 PLASMA OES DURING SiN ETCH OES COMPARISON OVER Si/SiN C 3 Π u B 3 Π g ) Si; Ar/N2/O2/CH2F2 SiN; Ar/N2/O2; x SiN; Ar/N2/O2/CH2F2; x2 OES Intensity (a.u.) NO (A 2 Σ + X 2 Π) Wavelength (nm) 7

8 CH 2 F 2 & CF 4 in Ar Etching Silicon CF (B 2 Ä X 2 ð) CF (A 2 X 2 ð) CF 2 (Strong Band Heads) 8

9 Spectrograph Resolution Effects on Spectra Extremely High Resolution * * The Identification of Molecular Spectra, R.W.B. Pearse and A.G Gaydon, Halsted Press, 1976 Simulated 1.7 nm Resolution 9

10 CH 2 F 2 & CF 4 in O 2 & Ar Etching Silicon CF (B 2 Ä X 2 ð) CF (A 2 X 2 ð) CF 2 (Strong Band Heads) 1 0

11 CH 2 F 4 & CF 4 in Ar Etching Photoresist CF (B 2 Ä X 2 ð) CF (A 2 X 2 ð) CF 2 (Strong Band Heads) 1 1

12 CH 2 F 2 & CF 4 in O 2 & Ar Etching Photoresist CF (B 2 Ä X 2 ð) CF (A 2 X 2 ð) CF 2 (Strong Band Heads) 1 2

13 CH CH2F2 & CF4 in O2 Ar Etching Protoresist 2 F 2 & CF 4 in O 2 & Ar Etching Photoresist OES Intensity (a.u.) (282.4 = O2 + ) (297.0 = O2 + ) (312.9 = C2) ( = O2 + ; = NO; = CF2) (308.9 = OH) ( = N2 +; = N2 + ) (=?) ( = F) ( =?) ( = F) (731.1 = F) ( = N2 +; = OH + ) ( = CN; = Ar; = Ar) ( = NO) (407.2 = Ar + ) ( = Ar) ( = Ar) (433.4 = Ar; = SiF) (451.1 = CO; = Ar) (476.5 = Ar) (486.1 = H) (488.0 = Ar + ) ( =?) ( = CO + ) ( = Ar; = O2 + ) ( = O2 + ) ( = Ar) ( = O2 + ) ( =?) ( = H; = H) ( = F) (727.3 = Ar) (738.4 = Ar; = N2) (772.4 = Ar) ( = O; = O; = O) (794.8 = Ar) (750.4 = N2) (763.5 = Ar) ( = Ar; =CN ) X Wavelength (nm) 1 3

14 nm Emission Lines of Species that Change Concentration at Endpoint W L Species Intensity W L Species Intensity W L Species Intensity W L Species Intensity CO CF O CF CO SiF SiF O CO SiF CO CF CO CO CF SiO Si CO N CO CO SiO NO CO CO CO CF CF CO SiF SiO SiO CO SiF CO NO Si NO O OH CO CO SiO OH CO CO CF OH CO CO SiF O NO SiF NO OH CO N CO CO CO SiF NO CF CO CF NO C CO CO CF NO NO CO CO CF CO CF O Si CO CO CO CO CO C CF CO NO NO N O CO SiO CO NO SiO CF SiO CF CO O CO SiF CO CO CF CO C NO NO CO CO CO CF O CF Si SiO CF CO CO CO N SiO Si O N N Si CF C CO CF NO SiF SiO Si CO O NO Si CO N CO Si CF CO SiF CO NO SiO NH SiO 6 1 4

15 Lack Of Fit NitrideER Blanket Etch Rate Results DOE with Power, CF 4, Flow 600 NitOxSelect N 2 /O 2 = 450/50 Pwr=300 W Press=200 mt ECHIP Pwr ECHIP Pwr Power CF4Flow 50 OxideER CF4Flow 50 ECHIP Power Pwr Power CF4 Flow CF4 Flow CF4 Flow 50 FG Celii & JW Hosch CF4Flow 8/10/01 1 5

16 Lack Of Fit NitrideER Blanket Etch Results DOE with Power, CH 2 F 2 Flow Lack Of Fit 600 NitOxSelect N 2 /O 2 = 450/50 Pwr=300 W Press=200 mt ECHIP Pwr ECHIP Pwr Power CH2F2Flow OxideER CH2F2Flow 50 ECHIP Pwr Power Power CH 2 F 2 Flow CH 2 F 2 Flow CH 2 F 2 Flow 50 Survey lower power CH2F2Flow 1 6

17 Blanket Etch Results: Survey of N 2 /O 2 Flow Blanket Nit/Ox Selectivity, Etch Rates Nitride ER Oxide ER Selectivity Etch Rate (A/min) CH 2 F 2 = 50 Pwr=300 W Press=200 mt Nit/Ox Etch Selectivity O2/(N2 + O2) Flow Ratio 1 7

18 Methods for Developing an Endpoint Detection Algorithm for a New Process The Plasma Process: Etch SiN stopping on silicide using CH 2 F 2 in Ar with O 2 Similar to other SiN etches Previous endpoint wavelength = 387 nm CN Difference Spectrum Ratio of Wavelengths NeuralPCA Multivariate Method 1 8

19 387 nm Emission of CN from Etching a Blanket Film of SiN on Silicide SMA Window Snap Shot Before Endpoint SMA Window Snap Shot After Endpoint Difference Spectrum = SS_After Endpoint - SS Before Endpoint 1 9

20 Regions Defined Using Difference Spectrum 2 0

21 Expanded View of Regions 2 1

22 Dominant Normalized Intensity Changes at Endpoint 2 2

23 Ratio Endpoint on Wafer Without Resist 2 3

24 Ratio Endpoint on Patterned Wafer 2 4

25 First NPCA Calibration Before Endpoint During After Endpoint Magnet Periods 2 5

26 Results of First NPCA Calibration Used on the Calibration Wafer Ratio NPCA Four Components of the Ratio 2 6

27 Second NPCA Calibration During Before Endpoint After Endpoint 2 7

28 Results Second NPCA Calibration Used on the Calibration Wafer Ratio NPCA Four Components of the Ratio 2 8

29 Comparison NPCA Calibration v.s. Ratio Endpoint Etching a Patterned Wafer Ratio NPCA Four Components of the Ratio 2 9

30 SRAM, center ENDPOINTED ETCH RESULTS x-sem OF PATTERNED WAFERS 39.5 s etch SRAM, edge Center not clear, Edge has cleared N 2 /O 2 = 300/200 CH 2 F 2 = W, 200 mt 3 0

31 Conclusions N 2 and O 2 in Ar with CH 2 F 2 do enhance the SiN etch rate. Flow ratios for maximum etch rate and maximum selectivity. Overlapping spectral lines may adversely affect tracking NO concentration using actinometry. More work is required. The NeuralPCA multivariate endpoint technique can accommodate unresolved spectra from closely spaced emission lines. NPCA is easier to implement than selected wavelength endpoint techniques. Anticipate NPCA endpoint sensitivity similar to SiO 2 contact etch. 3 1