Obtaining Accurate Surface Measurements. Bruker Nano Surfaces Division Dr. Erik Novak February 2012

Obtaining Accurate Surface Measurements Bruker Nano Surfaces Division Dr. Erik Novak February 2012

Outline Overview of precision, stability, accuracy Factors affecting data quality and metrology results Conclusion 2/28/2012 2

Overview of Precision, Stability, and Accuracy Precision measurement tool dependent/internal characteristic Resolution? Stability Measurement tool dependent/internal characteristic influenced by environment and other factors Accuracy Calibrate to an external standard (usually) - how well does my instrument do with respect to THE TRUTH? Best metrology is had when tools are precise, stable and accurate! 2/28/2012 3

Several Factors Influence Data Quality for Surface Metrology Instrument/Environment Internal mechanics, noise floor of detection mechanism, stability of hardware and precision of motion (if motion is present) Calibration Temperature/vibration considerations Test Surface/Sample Fixturing! Surface features (structure, heights, slopes) Surface roughness, form, waviness components THESE ARE NOT ALWAYS INDEPENDENT!!! Analysis and Computations with Data Measurement produces a representation of the test surface Filtering and computation methods 2/28/2012 4

Accurate Metrology is a Key Component of Product Success for Many Applications Precision Machining Optics MEMS and Semiconductor Data Storage MEMS Cantilevers Microfluidic Channel Knee Implant Cotton Cloth Holographic Film Currency Clutch Plate 2/28/2012 5

Instrument Factors Lateral Calibration is Key to Accuracy Lateral Instrument Calibration With optimized internal mechanics overall accuracy depends on calibration of lateral and vertical motion Lateral calibration can be accomplished several ways: Scanning stage (or making single FOV measurement) and measuring periodic sample of known pitch (optical or stylus) Measuring a known radius part and ensuring the radius of curvature calculation is correct 2/28/2012 6

Instrument Factors Vertical Calibration is Key to Accuracy Vertical Instrument Calibration Measurement accuracy can be verified by setup of measurement of known height sample (step standard, for example) The motor steps/unit measurement are computed based on measurement result for a known step with associated uncertainty Use a step that is close to the feature heights of interest. Ensure you calibrate the same way every time 2/28/2012 7

Bruker ContourGT-X8 Continuously Calibrates for Highest Accuracy ContourGT-X8 Offers Continuous Self Calibration Laser tracks scanner motion by interference with reference signal reflection Accuracy traceable to known He-Ne wavelength Second-level traceable standard Measurement Signals CCD Reference signal module Mirror Reference signal detector(s) Laser Beamsplitter Reference mirror Illuminator Mirror on the scanner Sample 2/28/2012 8

Bruker ContourGT-X8 Continuously Calibrates for Highest Accuracy ContourGT-X8 Offers Continuous Self Calibration Minimizes impact of irregularity in scan mechanism Minimized impact of drift of scanner Uncertainty in nm of 50 um step measurement Uncertainty in % of 50 um step measurement 300 0.60% Uncertainty (nm) 250 200 150 100 50 Uncertainty (%) 0.50% 0.40% 0.30% 0.20% 0.10% 0 Continuous Calibration Without Continous Calibration 0.00% Continuous Calibration Without Continous Calibration 2/28/2012 9

Continuous Calibration Reduces Uncertainty in Step Measurement Result 2/28/2012 10

Environmental Factors - Temperature Effects Minimized via Control or Calibration Operating environment control minimizes effects Continuous calibration provides excellent correction 2/28/2012 11

Instrument Factors Slope and Lateral Resolutions Vary With Options Key parameters are available for 3D microscopes and stylus that help understand tradeoffs of different instrument options Matching the instrument settings to the target is key to obtaining accurate and repeatable results 200nm lines 70 degree sloped screw threads 12

Instrument Factors - Height Capability Can Vary 3D Microscopes determine a signal peak as you move through focus Pictures below show signal for a single measurement line as you move through focus Traces show a smooth surface measured with 3D microscopes using interferometry (top) and confocal (bottom) technologies 1X 2.5X 5X 10X 20X 50X 100X Ra=4 nm Ra=4 nm Ra=4 nm Ra=4 nm Ra=4 nm Ra=4 nm Ra=4 nm Not Usable Not Usable Ra=472nm Ra=74nm Ra=12nm Ra=7nm Ra=4nm 13

Environmental Factors - Vibration Effects Should Be Understood Vibration can cause fringes in WLI based instrument to print through Typically results in errors of a few 10 s of nm to a few 100 s of nm if severe Isolation table or damping mechanism employed to minimize Avoid drafting from HVAC units, clean hoods 2/28/2012 14

Noise in measurements random noise loses against averaging With random (most) noise, noise will reduce by the square root of the number of averages. Averaging can help see finer detail than is otherwise possible. Difference measurements can tell you the noise floor you are achieving. Averaging may not help in loud or high vibration environments ContourGT-X8 can achieve a 0.015nm noise floor Ra of Difference Measurement vs. # of averages Ra (nm 0.3 0.25 0.2 0.15 0.1 0.05 0 0 50 100 150 200 250 # of Averages 2/28/2012 15

Smooth surfaces internal optical reference subtraction is key Smooth surfaces with small variation in shape/roughness benefit from instrument reference subtraction Use minimum 4 locations, 4 averages Subtracts out common element between measurements from future ones Essential for stitching super-smooth objects (wafers, mirrors, etc.) 2/28/2012 16

Smooth surfaces internal optical reference for spheres works well With steep slopes, errors from the optics will have some effect Effect is typically <80nm For very smooth objects, this error can affect stitching or certain surface calculations User can generate a reference using a random ball method Measure multiple locations on a sphere of the correct target radius Average the results Subtract the base curvature and save the residual as the reference file Reduces shape effects to <5nm 2/28/2012 17

Sample Considerations for Accurate Metrology How Should I Fixture? Vacuum is excellent choice where possible Provides stability and holds reproducibly if set up with kinematic contacts Bruker offers quick release dovetail slides with vacuum fixtures for easy on and off handling

Sample Considerations for Accurate Metrology Where Should I Measure? Wide range of crystal structure apparent across PV cell 2/28/2012 19

Data Analysis - Filtering is a critical component of accurate, reproducible results 2D Stylus filtering according to ISO 4287/4288 standards Filtering separates different portions of data of interest depending on specific criteria Make sure you report data that you care about! 2/28/2012 20

Robust Gaussian Filtering Better Separates Form from Finish Leads to More Reproducible Metrology 2 1 um 0-1 -2-3 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Unfiltered Data Non Robust Robust -4-5 -6-7 mm Robust Gaussian Filter Created to Filter Form from Waviness and Roughness Without Surface Distortions 2/28/2012 21

3D filtering analogous critical component of accurate metrology Filter Unfiltered data 3D Areal filtering works in analogous way according to ISO 25178-2 standards Waviness + form Again, filtering is key to reporting data of interest! Roughness 2/28/2012 22

3D areal parameters Accurate results with specialized computations 3D extension of R parameters from 2D stylus metrology (Sa, Sq, Sz) Skew, kurtosis, bearing area, peak density, slopes are computed Sds: Summit density Ssc: Mean summit curvature 2/28/2012 23

ISO standard computations enable excellent 2D to 3D correlation as well 2/28/2012 24

Agreement between optical and stylus results is excellent Method Nominal Optical Stylus Ra 100 nm 105 nm 108 nm Stylus profiler; Dektak Optical profiler; Contour GT <2um size tip Single 55 um profile 2/28/2012 115X 0.8 NA objective, XLI Single 55 um profile Know your Standard! This is a sinusoidal standard but deviates greatly from an ideal sine wave 25

Which is Accurate? Comparing Results Between Systems Creates Challenges New System X measures a part 10nm differently than our old system. How do we offset System X I measured some parts across the two systems and the correlation is terrible! How was each system calibrated? How do results vary within and across systems of each type? Can the two systems detect the same features? Are you examining the same areas on each system? Do the analysis algorithms on the two systems match? Is there sufficient range in the values for correlation to be meaningful? Ra nm Veeco WLI Vs Contact Stylus - PEEK y = 0.9663x R 2 = 0.9939 8000 7000 6000 5000 4000 3000 2000 1000 0 0 1000 2000 3000 4000 5000 6000 7000 8000 Ra nm Optical Value (nm) -10-11 -12-13 -14-15 y = 0.8916x - 10.726 R² = 0.796-16 -6-5 -4-3 -2-1 0 AFM value (nm) Optical Value (nm) -10-11 -12-13 -14-15 y = 0.7337x - 11.494 R² = 0.4111-16 -6-5 -4-3 AFM value (nm) 2/28/2012 26

Standard Error is Used to Evaluate Agreement Between Two Systems Good for parts with small range in values compared to the average Assumes measurement of the same features Two methods are considered agreeable to twice the calculated standard error Avoids having to know the true sample standard deviation required by the correlation coefficient R σ T Correlation coefficient = ( )( ) 2 2 2 2 σ + σ σ + σ sys1 σ T 2 T sys2 Is the true variation in the sample set Standard error: standard deviation of the difference 2 SE 2 sys1 σ = σ + σ 2 sys2 T 27

Summary Accurate surface metrology depends on many factors Verify performance on known samples if possible Proper fixturing and filtering are significant contributors to obtaining accurate surface metrology Ensure data reported is the data representative of need for test Bruker offers instrumentation which provides fast, accurate 2D stylus and 3D optical metrology for virtually all applications needs Partner with your metrology provider to ensure proper results! 2/28/2012 28