Controlling Roughness on Polished Glass Surfaces

Transcription

1 Controlling Roughness on Polished Glass Surfaces APOMA Workshop Tucson, Arizona November 10, 2016 Tayyab Suratwala R. Steele, R. Dylla Spears, M. Feit, R. Desjardin, L. Wong, N. Shen, P. Miller Lawrence Livermore National Laboratory LLNL-PRES This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under contract DE-AC52-07NA Lawrence Livermore National Security, LLC

2 Schematic model of the parameters that affect roughness during polishing Ensemble Hertzian Mult-Gap (EHMG) Model Beilby Layer Properties vs Depth Workpiece Bulk (E 1, 1 ) P A Redeposition Removal function (Plastic, Chemical, dissolution) z Workpiece Beilby Layer (E 4 (z), 4 (z)) Slurry Stability & Interface Interactions H 2 O gap Pad (E 2, 2 ) Active Particles (E 3, 3 ) Particle Size Distribution Particle Composition Pad Roughness Pad Mechanical Prop. 2

3 Polishing was conducted using the Convergent Polishing Method (ceria or silica slurry on various glasses using a polyurathane pad) CISR0 polisher CISR1 polisher 3

4 SIMS measurements show Ce penetration into polished surface is not due to diffusion & K penetration is consistent with diffusion [Ce] profile on polished fused silica surface as fn of polishing velocity [K] profile on polished fused silica surface as fn of polishing velocity [Ce] (atoms/cm 3 ) dh/dt V1: 40 rpm (1.00 m/hr) V2: 20 rpm (0.88 m/hr) V3: 10 rpm (0.45 m/hr) V4: 5 rpm (0.09 m/hr) [K] (atoms/cm 3 ) dh/dt V1: 40 rpm (1.00 m/hr) V2: 20 rpm (0.88 m/hr) V3: 10 rpm (0.45 m/hr) V4: 5 rpm (0.09 m/hr) Depth (nm) Depth (nm) SIMS (note Si 2x10 22 atom/cm 3 ) 4

5 [Ce] s increases with polishing removal rate & is weakly dependent on other polishing parameters [Ce] of polished surface layer for variety of polishing conditions Correlation between [Ce] s and removal rate (dh/dt) [Ce] (atoms/cm 3 ) S1: Colloidal ceria slurry P1: MHN pad P2: Chem-Poly pad P3: Optivision pad P4: Polytex 1 pad P5: Polytex 2 pad NP1: Soak in Ce [Ce] s (atoms/cm 3 ) V1 S2 P4 H1 V2 P2 H3 H4 H2 V3 P1 V4 S1 H5 P3 V: Velocity Series P: Pad Series S: Slurry Series H: ph Series Hydrolyis ratio Model Depth (nm) NP Polishing Removal Rate (dh/dt) ( m/hr) T. Suratwala et. al., J. Am. Cer. Soc 98(8) (2015)

6 The penetration of Ce into silica surface during polishing is proposed to be a competition of hydrolysis reactions Condensation Si-OH + HO-Ce Si-O-Ce + H 2 O glass particle Mechanism 1) Removal rate increases 2) Interface temperature increases 3) Arrhenius increase to r 4) Greater Ce surface deposition Silica Hydrolysis Si-O-Si-O-Ce-O-Ce + H 2 O Si-OH + HO-Si-O-Ce-O-Ce Ceria Hydrolysis Si-O-Si-O-Ce-O-Ce + H 2 O Si-O-Si-O-Ce-OH + HO-Ce r =Ceria Hydrolysis rate/ Silica Hydrolysis rate 6

7 Combining all these measurements, a structural polishing model at particle workpiece interface has been proposed LLNL-PRES-xxxxxx 7

8 Schematic Model of the parameters that affect roughness during polishing Ensemble Hertzian Mult-Gap (EHMG) Model Beilby Layer Properties vs Depth Workpiece Bulk (E 1, 1 ) P A Redeposition Removal function (Plastic, Chemical, dissolution) z Workpiece Beilby Layer (E 4 (z), 4 (z)) Slurry Stability & Interface Interactions H 2 O gap Pad (E 2, 2 ) Active Particles (E 3, 3 ) Particle Size Distribution Particle Composition Pad Roughness Pad Mechanical Prop. 8

9 Nanoscratching at ultralow loads offers exploring mechanical properties of Bielby layer and the particle removal function Load vs Corresponding Depth on Fused Silica Elastic Plastic and/or Densification Fracture Hertzian Cracking (Blunt) 4,5 mm Depth (nm) LateralCracking (Vickers) 4,5 Radial Cracking Nanoindentation (Vickers) 4,5 Multi-pass Brittle Chipping (nanoscratch; =150 nm) 2 Shear-band Cracking (Vickers) 3 Single-pass Brittle Chipping (nanoscratch; =150 nm) 2 Typical loads during Polishing Ultra-low load Nanoscratching Plastic deformation/densification (nanoscratching; =150 nm) Applied Load (mn) m nm 1 This study (2014); 2 Thongoom J. Mat. Sci 40 (2005); 3 Miller, Optics Letters 35(16) (2006); 4 Lawn, Fracture of Brittle Solids (1993); 5 Suratwala, JNCS 354 (2008) 9

10 Fused silica and BK7 show little load dependence on permanent deformation; changes in Bielby layer of fused silica influences depth AFM images of nanoscratches on different surfaces at various loads Cross-section of nanoscratches at various loads on various substrates N 150 N 110 N 80 N 50 N 20 N Relative Height (nm) LHG-8 Phosphate BK7 ur20 Fused Silica B4 Fused Silica B1 Fused Silica Position ( m) 10

11 Comparison of nanoscratching on different materials (function of # of passes & environment) Relative Height (nm) Nanoscratch cross sections on various substrates # of Passes Sapphire Fused Silica (S1) Borosilicate (B1) Phosphate (P1) Position ( m) air water air water air water air water 11

12 The removal volume for a single polishing particle was determined from multi pass nanoscratching to account densification effects LHG-8 phosphate glass: scratches at 110 N Passes: nm N. Shen et. al., J. Am. Cer. Soc (2016)

13 A detailed description of the removal function has been determined for various glasses aiding to the prediction of roughness Determined removal function for single particle on various glasses Depth (nm) Fused Silica Borosilicate glass (BK7) Phosphate glass (LHG-8) Molecular Plastic Load/particle (N) N. Shen et. al., J. Am. Cer. Soc (2016) 1-8 Removal occurs over two regimes during polishing (molecular and plastic) Fused silica and BK7 have similar removal functions Removal function for phosphate glass is higher Combining removal function with load/particle distribution allows for predicting roughness 13

14 Schematic model of the parameters that affect roughness during polishing Ensemble Hertzian Mult-Gap (EHMG) Model Beilby Layer Properties vs Depth Workpiece Bulk (E 1, 1 ) P A Redeposition Removal function (Plastic, Chemical, dissolution) z Workpiece Beilby Layer (E 4 (z), 4 (z)) Slurry Stability & Interface Interactions H 2 O gap Pad (E 2, 2 ) Active Particles (E 3, 3 ) Particle Size Distribution Particle Composition Pad Roughness Pad Mechanical Prop. 14

15 Slurry s PSD* strongly correlates with workpiece roughness and removal rate Particle Count, N Measured PSD of ceria slurries Accuplane E92 Stabilized Hastilite E134 Unstabilized Hastilite E133 Ultra-sol 3005 E135 Ultra-sol Particle size, d ( m) The tail end of each slurry follows a single exponential distribution *Particle size distribution T. Suratwala et. al., J. Am. Cer. Soc 97(1) (2014) 81 AFM Exponent images constant of fused in silica PSD workpieces of slurry after vs RMS polishing roughness with different of polished ceria surface slurries Accuplane 10 1 Stab.Hast. Unstab.Hast. Exponent Constant in PSD, d o ( m) 10 0 RMS= nm 10-1 Ultrasol RMS=0.653 nm Ultrasol3030 d n =0.008 m o =0.20 nm RMS= 0.99 nm d o =d n e - / Full scale= nm 1.4to 41.6 nm RMS roughness (AFM 50 um), (nm) The RMS= slope 1.12 nmof the RMS= slurry s 1.27 nmpsd quantitatively scales with the rms 50 m roughness 15

16 Pad topography during polishing strongly influences removal rate MHN Pad Surface Topology (Confocal Microscope Images) with various surface treatments Pad Height Histograms m/hr 0.31 m/hr Cum Area Fraction New MHN (T1) Used MHN (T2) 5 min DC (C) 45 min DC (T3) Relative Height ( m) 0.82 m/hr 2.10 m/hr Tall pad asperities (100 s m) are removed with diamond conditioning pad treatment Removal rate increased from 0.08 m/hr to 2.10 m/hr; 26x increase 16

17 EHMG (Esemble Hertzian Multi Gap) polishing model accounts for both slurry PSD & pad topology to determine RR and roughness EHMG Model Setup Key Inputs: Slurry PSD & Pad Topology Using pad height histograms: Pad asperities compress leading to single value gap of pad (g p ) based on load balance Fraction of pad area making contact is calculated Each asperity compresses by height (h i ) resulting in stress ( i ) Using slurry PSD at each asperity land workpiece interface, slurry particles are loaded with a unique gap (g i ) following load balance Load/particle distribution is calculated from summing all pad asperities T. Suratwala et. al., J. Am. Cer. Soc (2016) accepted 17

18 EHMG model compared with experiments expands our insight to the diverse factors affecting material removal rate Experiment EHMG Model Removal Rate ( m/hr) Fraction of Active particles with plastic removal (f p ) Measured removal rate & EHMG model Comparison dh dt S4 NanoArc NanoArc Ceria Slurry PSD C Stabilized Hastilite Stabilized Hastilite S3 Unstabilized Hastilite Unstabilized Hastilite N f p t f A Active Particles (N T *f r ) (10-11 cm -2 ) Fraction of load on particles (f L ) f Slurry Concentration N2 N1 N3 N4 N5 N Baume N t f r f L Areal # density of Particles (N t ) (cm -2 ) L f r V r f p d Fraction Pad Contact Area (f A ) p Pad Topology T1 T2 f A 2a p C T2 New Used 5 min DC 45 min DC New Used 5 min DC 45 min DC f m Average Depth of Plastic Removal (<d p >) d m F1 Silica Silica Glass Type B1 Borosilicate <d p > Borosilicate 2a m P1 Phosphate Phosphate Widening PSD increases load/particle & fraction of removal by plastic removal (f p ) Increasing slurry conc increases active particles density (N t f r ) and fraction of load carried by particle (f L ) Increasing pad flatness increases fraction of pad area making contact (f A ) Change in glass type change removal depth by plastic removal (d p ) 18

19 Load/particle distribution calculated using EHMG model, combined with measured removal function, gives the removal amount for each slurry particle Single particle removal function & calculated load/particle distribution Depth (nm) or Fractional Distribution* Fused Silica Borosilicate glass (BK7) Phosphate glass (LHG-8) Stabilized Hastilite Colloidal Silica or NanoArc Molecular Unstabilized Hastilite Plastic Load/particle (N) This can be now used to calculate both removal rate and roughness during polishing 19

20 EHMG model also simultaneously simulates trends in observed AFM roughness over a variety of polishing parameters Experiment EHMG Model Simulation 20 AFM roughness (nm) Ultra-Sol S27 (S6) Stabilized Hastilite (C) Unstabilized Hastilite (S3) New MHN Pad (T1) 45 min DC MHN Pad (T3) Fused Silica (F4) Borosilicate Glass (B4) Phosphate Glass (P4) Measured Roughness vs EHMG model T. Suratwala et. al., J. Am. Cer. Soc 97(1) (2016) 81

21 Schematic model of the parameters that affect roughness during polishing Ensemble Hertzian Mult-Gap (EHMG) Model Beilby Layer Properties vs Depth Workpiece Bulk (E 1, 1 ) P A Redeposition Removal function (Plastic, Chemical, dissolution) z Workpiece Beilby Layer (E 4 (z), 4 (z)) Slurry Stability & Interface Interactions H 2 O gap Pad (E 2, 2 ) Active Particles (E 3, 3 ) Particle Size Distribution Particle Composition Pad Roughness Pad Mechanical Prop. 21

22 Probing roughness over different scale length: factors affecting u roughness are not necessarily the same as those affecting AFM roughness Power Spectra for various spatial band on a typical fused silica optic -roughness vs AFM roughness PSD (nm 2 mm) Gradient PSD1 PSD2 -rough AFM1 AFM Full Aperture Interferometry 2 White Light Interferometry (10 mm x 11.6 mm) 3 White Light Interferometry (800 m x 1600 m) 4 AFM (50 m x 50 m) 5 AFM (5 m x 5 m) Frequency (mm -1 ) Spatial scale length (mm) -roughness ( m) Fused Silica (this study) Phosphate (this study) Fused Silica (ref [x]) Phosphate (ref [x]) AFM2 roughness ( m) 22

23 Little change in AFM roughness suggests plastic removal function is unaffected by ph; Large change in roughness suggest ph is influencing slurry agglomeration at larger scale lengths -roughness AFM Surface roughness of Phosphate glass (LHG-8) polished with Stabilized Hastilite at different phs Note same behavior observed with Stabilized & Unstabilized Hastilite for LHG-8 23

24 The uniformity of slurry on the pad is greatly improved at lower ph, likely leading to lower u roughness Confocal image of pad surface after polishing ph=2 ph=13 P ppt Suratwala NAS March 07,

25 To understand interface interactions, the Zeta potential of slurry, pad & workpiece have been measured as function of ph Zeta Potential is a measure of the surface charge on a particle or surface Zeta Potential vs ph of slurry, pad & workpiece Zeta Potential (mv) Ceria (Stabilized Hastilite) Ceria (Hastilite PO) Colloidal Silica (S27) -30 Zeta Potential (mv) ph Fuses Silica Phosphate Glass (LHG-8) Polyurethane Pad (MHN) ph 25

26 Impact of glass products on zeta potential is very different depending on the nature of the glass product Zeta Potential (mv) Zeta Potential of Stabilized Hastilite PO as a fn of ph and [K 3 PO 4 ] [K 3 PO 4 ] 0 M 0.06 M 0.19 M 0.13 M 0.25 M SH Baume M 0.19 M 0.13 M 0.09 M Addition of glass product surrogate for phosphate glass (K 3 PO 4 ) make the zeta potential positive with little change in ph ph [K 3 PO 4 ] 0 M [K 3 PO 4 ] 0.25 M 0.19 M 0.13 M 0.06 M 0 M Zeta Potential (mv) Zeta Potential of Stabilized Hastilite PO as a fn of ph and [Si(OH) 4 ] SH Baume ph [Si(OH) 4 ] Addition of glass product surrogate for silica glass (Si(OH 4 )) has little impact to zeta potential 26

27 A model to determine the electrostatic double layer interaction forces between the 3 components at the interface (as a function of ph and glass products) has been developed Electrostatic double-layer interaction forces (two dissimilar surfaces of different radii)* F Workpiece + i j Lap Particle-Workpiece Force (F P-WP ) Particle-Particle Force (F P-P ) Particle-Lap Force (F P-L ) ri rj 4 o r i rj i je = dielectric constant for slurry solution r = Radius of surface of interest = Debye-Hunkel Parameter h = Separation distance = Surface potential of surface of interest (zeta potential) h *S. Carnie, D. Chan, J. Gunning Langmuir 10 (1994) Force Calculation Results (using measured zeta potential vs ph and glass product conc) Force (10-12 N) No glass products M K 3 PO 4 Particle-Workpiece Particle-Pad Particle-Particle Particle-Pad Particle-Workpiece ph Repulsion Attraction 27

28 Using the IDG model, simulated roughness compares well with measured data suggesting that slurry spatial distribution is an important contributor to roughness Comparison between measured and simulated -roughness Schematic of Island Distribution Gap (IDG) Model 28

29 Schematic model of the parameters that affect roughness during polishing Ensemble Hertzian Mult-Gap (EHMG) Model Beilby Layer Properties vs Depth Workpiece Bulk (E 1, 1 ) P A Redeposition Removal function (Plastic, Chemical, dissolution) z Workpiece Beilby Layer (E 4 (z), 4 (z)) Slurry Stability & Interface Interactions H 2 O gap Pad (E 2, 2 ) Active Particles (E 3, 3 ) Particle Size Distribution Particle Composition Pad Roughness Pad Mechanical Prop. 29

30 Polishing near isoelectric point of workpiece results in strong attraction of slurry particles to workpiece Colloidal Silica (S27) polishing on Fused Silica AFM2 image after polishing at ph=2 AFM2 image after polishing at ph=9 ur61 ur62 X,Y: 5 m x 5 m Z : -2.5 nm to 2.5 nm 30

31 Best fit parameters to deposition kinetics and equilibrium suggest deposition flux scales with C (settling controlled) and desorption scales as C 0.6 Simulation vs Experiment Monte Carlo Deposition Rate Model Setup Model Parameters b J z V m P ad P d AFM Measurement Simulation 0.1x (3.2*10 4 /nl) 1x (3.2*10 5 /nl) 5x (1.6*10 5 /nl) Substrate 10x (3.2*10 6 /nl) C =bulk particle concentration (#/cm 3 ) r = particle radius V m = fluid velocity b = channel width P ad = adsorption probability P d = desorption probability 25x (7.9*10 6 /nl) 31

32 Schematic model of the parameters that affect roughness during polishing Ensemble Hertzian Mult-Gap (EHMG) Model Beilby Layer Properties vs Depth Workpiece Bulk (E 1, 1 ) P A Redeposition Removal function (Plastic, Chemical, dissolution) z Workpiece Beilby Layer (E 4 (z), 4 (z)) Slurry Stability & Interface Interactions H 2 O gap Pad (E 2, 2 ) Active Particles (E 3, 3 ) Particle Size Distribution Particle Composition Pad Roughness Pad Mechanical Prop. 32

33 Increase in pressure resulted in expected removal rate increase and little change in roughness as predicted by the EHMG model Removal Rate ( m/hr) Plot of measured removal rate & roughness on Fused Silica S6 S6 S5 S5 T3 L3 ~3x ~2.5x V r S4 T3 S4 L2 L1 L1 L2 S2 L3 S2 N6 Slurry Series Slurry (S) Series (S) Slurry Concentration Slurry Concentration Series (N) Series (N) Pad Type Pad Series Type (P) Series (P) N4 Pad Topology Pad Topology Series (T) Series (T) C N3 S3 N2 N6 N4 N3 T2 S3N2 C T1 T2 0 T rms AFM roughness (nm) N1 N1 S1 S1 0.6 psi 20 rpm 1.9/2.5 psi 50 rpm Narrow PSD (S6) rms=1.3 Ang Flat Pad (T3) Flat Pad (ur171) rms=2.6 Ang These results have large practical implications since it is largely believed that low roughness surface are only achieved at low removal rates 33

34 Adding literature values also suggest that our recent results are in a regime that has not been published to date. Removal Rate ( m/hr) Plot of measured removal rate & roughness on Fused Silica 10 5 T3 L5 L1 S4 S6 S5 L9 L1 L2 N6 N4 N3 L3 S3N2 C L6 S2L10 T2 T1 Slurry Series (S) Slurry Concentration Series (N) Pad Type Series (P) Pad Topology Series (T) Other studies (literature) S1L11 L2 L rms AFM roughness (nm) N1 L4 L12 L3 L8 Reference Lap Roughness (nm) Removal Rate (um/hr) L1 Tesar (1992) Pitch L2 LLE Review Pitch L3 LLE Review Pitch L4 Berggren Pad L5 Camp Pitch L6 Camp Pitch L7 Cumbo Pitch L8 Cumbo Pitch L9 Jungling Float L10 Jungling Float L11 Li (2008) Pad L12 Li (2008) Pad These results have large practical implications since it is largely believed that low roughness surface are only achieved at low removal rates 34

35 Strategies to reduce roughness and increase removal rate during polishing 1) Establish a narrow load/particle distribution Use slurry with narrow particle size distribution (especially at the tail) Use a compliant lap 2) Remove asperities from lap Example: Correctly diamond condition polyurethane pad 3) Stay within molecular removal regime (avoid plastic regime) i.e., increase load up until plastic regime is reached 4) Control slurry chemistry such that slurry is uniformly distributed at the interface e.g., ph control and glass products removal 35