Selection of ESH solvents for cleaning applications in semiconductor manufacturing

Transcription

1 Selection of ESH solvents for cleaning applications in semiconductor manufacturing E. Kesters 1, M. Claes 1, Q. T. Le 1, K. Barthomeuf 2, M. Lux 1, G. Vereecke 1*, T. Bearda 1, and J.B. Durkee 3** 1 IMEC, Kapeldreef 75, 3001 Leuven, Belgium 2 INSA, 20 av. A. Einstein, Villeurbanne, France 3 Precisioncleaning, PO Box 847, Hunt TX 78024, USA * guy.vereecke@imec.be ** jdurkee@precisioncleaning.com

2 Outline Introduction Materials & methods Results & first discussion Analysis of the Hansen spherical approach to solvent selection Conclusions imec

3 Micro-electronics < 100 nm imec

4 Semiconductor manufacturing Introduction of new materials Cu wires Embedded in a porous dielectric SiOC, Encapsulated in a barrier TaN/Ta, SiC, WCN, Ru, On a substrate (wafer) Si, Ge, Ga/As, etc Transistors, capacitors, etc Novel gate dielectric materials HfO, HfSiON, LaO, DyO, Novel gate electrode materials TiN, Ta 2 C, imec

5 Uses of organic photoresist (PR) layers Transistor level As implantation poly-si crust resist metal gate high-k nfet pfet Patterning of materials by etching Local modification of substrate conductivity by ion doping Protection of undoped areas imec

6 Uses of organic photoresist (PR) layers Cu wiring Patterning of dielectric layers by plasma etching Before Cu filling photoresist dielectric etch substrate imec

7 Removal of used PR Issues with plasma ashing Materials compatibility: damaging of exposed materials Degradation of properties E.g. porous dielectrics Material loss in subsequent process steps Renewed interest in wet organic stripping Must meet new ESH criteria E.g. n-methyl pyrrolidone (NMP), a good solvent, will be re-classified as reprotoxin (cat. 2) imec

8 This work Find ESH replacement solvents Making use of Hansen approach Focus on pristine PR Model for bulk under crust Crust Bulk PR removed imec

9 Outline Introduction Materials & methods PR materials Solvents selection Experimentals Results & first discussion Analysis of the Hansen spherical approach to solvent selection Conclusions imec

10 PR materials: 193 nm DUV PR1: polymethacrylate (PMA) adamantane lactone PR2: made from acrylate and methacrylate monomers /H /H adamantane lactone imec

11 ESH selection criteria Criteria set according to EEC classification Safety: flash point FP > 55 C No F+, F, R10 solvents Health: no toxic solvents T+ (very toxic) and T (toxic) discarded No R23-28, R39, R48 solvents Carcino/mutagenic & reprotoxic discarded (of all cat.; T, Xn) No R40, R45-46, R49, R60-63, R68 solvents Environment: no toxic solvents Aquatic & non-aquatic environment No R50-R59 solvents No N label imec

12 Selection of solvents Method: based on Hansen theory Hansen solubility parameters (HSP) δ = δ + δ + δ p d h Hydrogen bonds Polar interactions Dispersive interactions (VdWaals) δ p δ h General principle of use: like dissolves like δ d imec

13 Experimental procedure Dissolution of pristine PR Wafer pieces with PR films Beaker tests at RT, afo time Visual inspection for complete removal Final check by FTIR CHx lactone +ester clean Wavenumber (cm -1 ) Wavenumber (cm -1 ) imec

14 Outline Introduction Materials & methods Results & first discussion Selected solvents & mixtures Evaluation of organics solvents & mixtures Analysis of the Hansen spherical approach to solvent selection Conclusions imec

15 Selected solvents & mixtures (1) Polar parameter MIBK PC CHex NMP Ac PGME DCM EA THF Tex BA EG Very few solvents meeting selection criteria & close to good solvents mixtures HSP mixture = volume fraction weighted average of HSP components Mixtures of PC with: BA, Tex NMP & Ac 0 TCE good solvents from initial selection Hydrogen bonding parameter EG: used in commercial chemistries Acronyms: see Annex imec

16 Selected solvents & mixtures (2) 20 good solvents from initial selection Mixtures of THFA with: Polar parameter DMSO Cap NP NMP? NEP Ac THFA CHex DCM PGME MIBK EA THF EHA? HFE TCE? HFC? Hansen parameters unknown Cap, NEP, NP NMP CHex, EHA, MIBK DCM EA, HFE, HFC THF & TCE Mixtures of DMSO with: HFC, HFE NMP, Ac, CHex, MIBK Hydrogen bonding parameter Acronyms: see Annex imec

17 Dissolution of PR PC Cap? Hansen parameters unknown DMSO H2O Hansen plot All solvents & mixtures Acronyms: see annex Color scale for dissolution time Polar parameter CHex MIBK? HFE? HFC NP? NEP EHA Ac DCM TCE NMP EA THF Tex PGME unlabelled data points are mixtures THFA Hydrogen bonding parameter BA EG 0.5 min min 1-2 min 2-5 min > 5 min Discussion Hansen approach does not seem to work No well defined solubility domain Solvents & mixtures with similar parameters do not show same behavior See DCM & NMP imec

18 Reasons for apparent failure Hansen parameters are thermodynamics Dissolution tests data gives kinetics Discrepancy points to kinetic factors Effect of size of solvent molecules? See later Microstructure of PR PR is a (blend of) copolymer(s) PR contains additives Hansen parameters are 3-D 2-D analysis See later imec

19 Dissolution of PR2 20 PC? Hansen parameters unknown DMSO H2O Hansen plot All solvents & mixtures Acronyms: see annex 15 Cap Color scale for dissolution time Polar parameter MIBK? HFE? HFC NP EHA? NEP DCM NMP Ac EA THF Tex unlabelled data points are mixtures THFA Hydrogen bonding parameter BA EG 0.5 min min 1-2 min 2-5 min > 5 min Discussion Very large solubility domain Very different behavior compared to PR1 Possible reasons Different copolymers Different additives in PR mixture imec

20 Summary of solubility tests on pristine PR Best pure solvents Benzyl alcohol (1-nitropropane) 1 NEP (NMP) 2 DMSO Propylene carbonate 1 FP = 33 C 2 Reprotoxin cat.2 Mixtures with HFC & HFE Not miscible with DMSO (down to 10 %) Not miscible with THFA (down to 10 %) Ranking of the solvent mixtures 1. Propylene carbonate/benzyl alcohol 2. THFA/1-nitropropane 3. THFA/NEP 4. THFA/NMP 2 5. THFA/MIBK 6. THFA/Cyclohexanone 7. THFA/Ethyl Acetate 8. Propylene carbonate/ethylene glycol 9. Propylene carbonate/texanol 10.THFA/2-ethylhexyl acrylate 11.THFA/ε-caprolactam imec

21 Outline Introduction Materials & methods Results & first discussion Analysis of the Hansen spherical approach to solvent selection Conclusions imec

22 Outline Review of Method Based on matching impact of intermolecular forces between solvent and soil Presentation format is spherical geometry An optimization routine & a two-dimensional plot Application to PR1 and PR2 Data Sets Analysis of specific results Key Learnings About Spherical Approach Maintain realistic optimization goals Recognize / accept false results Beware of two-dimensional plots Conclusions imec

23 Hansen Approach to Solvent Selection For polymers, soils, or solvents [solutes]: One matches intermolecular forces of solvent & solute Like dissolves like Three intermolecular forces: Dispersed throughout the molecule Localized between poles of electrostatic charge Localized among hydrogen bonds Each force characterized by a thermodynamic parameter: δ disperse δ Polar δ H2-Bonding Energy σ = Volume Total force characterized by the Hildebrand parameter: δ = δ + δ + δ Hildebrand Polar Dispersion Hydrogen bonding imec

24 A Geometric Vision by Hansen & Coworker Each solubility parameter is independent Impacting in one of three independent coordinates A polymer / soil / solvent is located in a three-dimensional space at the values of its solubility parameters. The limit of action with another chemical is a radial dimension called R O. There will be differentiable outcomes GOOD and BAD solvency are binary outcomes Experimental data are of solution time Solubility Grade = solution time <> value imec

25 Not-Plain Geometry The separation between any two chemicals in threedimensional Hansen space is another radial dimension called R A. In terms of solubility parameters for solvents and solute: [{ } ] { } Equation derived from data plots The 4 has theoretical and empirical significance Not R A δ Hildebrand Mutual solubility is expected when R O overlaps R A I.e., The data lies within the solubility sphere 2 [ ] [{ 2 2 } ] R 2 2 A = disperse disperse + polar polar + h bonding h bonding δ δ δ δ δ δ imec

26 The Meaning of Data Analysis Data analysis is to inquire: When R A < R O, is there solubility? i.e., Good and IN? When R A > R O, is there not solubility? i.e.., Bad and OUT? False Negatives are: When R O > R A, & there is solubility! i.e., Good and OUT? False Positives are: When R O < R A, & there is not solubility? i.e., Bad and IN? imec

27 Data Processing Wanted are the parameters which characterize the material being solubilized Will allow selection of new solvents/mixtures Optimization identifies the solubility parameters which place all the data where it should be placed: So all are Good and IN; or Bad and OUT. The parameters are known for each solvent: δ disperse δ Polar δ H2-Bonding δ { } any = X i δ i R A X i = Volume Fraction imec

28 Optimization Methodology; One Needs: To define an error: [ ( ) { }] df i = exp R R FACTOR O A i Df i = 1 for Good and IN, or Bad and OUT, because then FACTOR = 0. For False results, FACTOR = 1 Without regard to values of R A and R O To choose an error function: The Desirability Function (DF) [ ( ) { }] { exp O A 1} { exp [ ( O A) { }] 2}... { exp[ ( O A) { }] } DF = R R FACTOR R R FACTOR R R FACTOR i To choose an optimization routine Linear (and non-linear) programming in spreadsheets To pay attention To the distribution of outcomes 1 i imec

29 Distribution of Outcomes PR1 Targets were methylene chloride and n-methyl pyrrolidone HSP Optimization with PR1 Photoresist Solution Time, Min RA ð Dispersion Optimized MPa^(1/2) ð Polar ð H2 Bonding Bad & Out Good & IN Bad & IN Good & OUT DF < < < < > The goal optimum (DF = 1) includes all data, i.e. No optimization being done A significant amount of data is properly < 0.5 min 69 of 80 data points imec

30 Analysis of the False Outcomes Molar volume is often a significant parameter in solubility performance: Low volume solvents like methylene chloride, water, and acetone often perform better than expected. That s not the case here. But, all but 3 of the False outcomes for PR1 are within ± 1 R A unit of expected performance imec

31 Analysis of Four False Outcomes Solvent Combinations Performing Considerably Outside of Expectations Solvents, or Mixtures ð Dispersion ð Polar ð H2 Bonding Excess HSP R A Molar Volume, cc/mole Vs Expectations MPa^(1/2) Benzyl Alcohol (0.17 min) Better (PR1) 'Tetrahydrofurfuryl alcohol (THFA), (5 min) Poorer (PR1) 'THFA & 50 % THFA (0.17 min) Better (PR1) Caprolactam (>100 min, but it s a solid at room temperature ) Poorer (PR2) Benzyl alcohol is far outside the sphere and performed very well! The two Tetrahydrofurfuryl alcohol materials performed differently? THE REASONS FOR THIS ARE NOT KNOWN! imec

32 Distribution of Outcomes PR2 HSP Optimization with PR2 Photoresist Solution Time, Min RA ð Dispersion ð Polar ð H2 Bonding Bad & Out Good & IN Bad & IN Good & OUT DF Optimized MPa^(1/2) < < < < > All but three other False results are within ± 1 RA unit of expected performance 80 of 81 results are as expected There Are Fewer Anomolies With PR2 imec

33 Solvents Selected for Low Molar Volume imec

34 Surprising Outcome Hydrogen-bonding HSP is about zero for PR1. And not for PR2. It s not clear, based on these structures, why this should be so. Removal of the False outcomes doesn t change this outcome. adamantane lactone /H /H adamantane lactone imec

35 Beware Two-Dimensional HSP Plots Commonly used in analysis of solubility problems. Based on assumption that δ disperse doesn t vary among solvents But it does. Plot makes it appear that some GOOD & Out are actually in. They re not δ disperse optimum = 18.9 imec

36 A Two-Dimensional Plot Re-plotted Plot shows that some GOOD & Out are actually out. HSP optimum + 4 = 22.9 Shows just the end of the sphere Don t use twodimensional plots to make final judgments Use them to select solvents for test imec

37 Outline Introduction Materials & methods Results & first discussion Analysis of the Hansen spherical approach to solvent selection Conclusions imec

38 Conclusions Several ESH solvents & mixtures identified that can dissolve the PRs Selection is dependent on PR Kinetics vs. thermodynamics PR is a (blend of) different copolymer(s) PR is a complex mixture (with different additives) imec

39 Conclusions Hansen spherical approach shows measured solubility data set thermodynamically consistent. 18 for PR1 (< 0.5 min solution time) 62 for PR2 (< 0.5 min solution time) The HSP and R A values are known for the PRs For additional evaluations Care must be taken with the spherical approach. It can produce unexplained outliers DF =1 is not the optimization target Properly-located assignments Vs Solubility Grades are Don t make judgments from 2-D plots imec

40 imec

41 Solvents acronyms Ac: acetone BA: benzyl alcohol Cap: ε-caprolactam CHex: cyclohexanone DCM: dichloromethane DMSO: dimethyl sulfoxide EA: ethylacetate EHA: 2-ethyl hexyl acrylate EG: ethylene glycol HFC: Vertrel MCA HFE: HFE-7100DL MIBK: methyl-isobutylketone NEP: n-ethyl pyrrolidone NMP: n-methyl pyrrolidone NP: 1-nitropropane PC: propylene carbonate PGME: 1-Methoxy-2- propanol TCE: trichloroethylene Tex: texanol THF: tetrahydrofuran THFA: tetrahydrofuran alcohol imec