Silicon Bulk Micromachined Hybrid Dimensional Artifact Characterization Meghan Shilling Sandia National Laboratories NCSLI 2009 July 27, 2009 Co-Authors: Hy D. Tran, Todd M. Bauer, Andre A. Claudet, SNL This work is a product of the Laboratory Directed Research and Development program at Sandia National Laboratories. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.

2 / 28 Research Objective Mesoscale metrology commonly uses video probing Accuracy of video systems are typically limited by calibration artifact, not resolution –Calibration artifact accuracy ~1  m –System resolution ~0.1  m Objective: To create a calibration artifact for a video-based measurement system which can be certified to better than 0.1  m accuracy.

Artifact Manufacturing and Design

4 / 28 Hybrid Dimensional Artifact Commercially available CMMs have relatively low uncertainty Must have features that allow for both touch and video probing –Vision systems good at detecting edges –Intersection line of two well-defined planes Bulk silicon micromachining chosen to fabricate the artifact

5 / 28 Si Bulk Micromachining

6 / 28 Bulk Silicon Micromachining Angles created by anisotropic etching are intrinsic –Defined by the crystal structure –Consistent over a wafer using single crystal material –Sidewalls create a sharp and straight intersection line with top surface silicon used –Sidewalls at a 54.74º angle from vertical –1.5 mm thick, etched to 1 mm depth

7 / 28 Artifact Design Touch probe measures points on sidewall and top plane –Finds best-fit planes –Calculates intersection line of two planes –Very low uncertainty (<50 nm) of intersection line location possible Vision probe measures intersection line directly during calibration –Top-down lighting –Change in illumination level doesn’t change distance between lines

8 / 28 Geometric Design Fabricate artifact which contains miniature versions of “macro” metrology Step gage –1D and 2D performance evaluation Ball plate –3D performance evaluation Other objects for investigation

9 / 28 Artifact Fabrication Parameters were varied to optimize processing steps –Different rinse agents and etchants (chemistry, concentration, temperature) were tested –19 trials run –Approximate etch time is 17 hours –Sidewall roughness not accessible, roughness on bottom of trench measured, used to choose optimal parameters

10 / 28 Produced Artifacts

11 / 28 Final Products

Artifact Characteristics

13 / 28 Sidewall Roughness Initial Processing Parameters Want to optimize processing to minimize roughness Rough sidewalls lead to greater uncertainty in edge location Initial processing parameters lead to visibly rough sidewalls –~1  m peak to valley using stylus-based instrument –~17  m peak to valley using non-contact instrument

14 / 28 Sidewall Roughness “Optimized” Parameters Roughness trace using stylus instrument 2  m radius tip, N measuring force 15 mm total scan length, peak-to-valley = 22.5  m For 3 mm local area (shown above) –Ra =  m –Peak-to-valley = 0.97  m

15 / 28 Sidewall Roughness “Optimized” Parameters Roughness trace using non-contact instrument 15 mm total scan length –Ra = 5.59  m –Peak-to-valley = 57.4  m

16 / 28 Sidewall Roughness “Optimized” Parameters

17 / 28 Sidewall Roughness No apparent improvement with “optimized” processing parameters Measurement of bottom of trench used to optimize parameters –AFM measurement of 30 mm x 30 mm patch –Rq ranged from 5 nm to 156 nm for all conditions Bottom trench roughness does not appear to correlate to sidewall roughness New processing options available Further process optimization is needed

18 / 28 Edge Sharpness Edge sharpness measured using fiber-probe system 5 traces Measurement of sharp edge with round probe gives round trace Appears that we have a sharp edge

Dimensional Characterization by CMM

20 / 28 CMM Properties Leitz Infinity PMMC Volumetric accuracy of ±(0.3 + L/1000)  m Repeatability of approx 100 nm for this test 0.5 mm diameter ruby probe used N trigger force Environment good to 20 ± 0.03 ºC

21 / 28 Measurement Plan 19 intersection lines calculated –Top plane scanned (~60 points) –Sidewall point-to-point (16 points) Results reported –Form/flatness of planes –Distance between neighboring line centers –Angles between lines x

22 / 28 Measurement Results Flatness

23 / 28 Measurement Results Distance Between Lines

24 / 28 Measurement Results Repeatability

25 / 28 Measurement Results Angle of Lines

26 / 28 Measurement Results Flatness for sidewalls is not as good as we would like, efforts are being made to further optimize manufacturing process Mask quality appears sufficient to achieve evenly spaced squares with good angular alignment Contamination is a concern, careful cleaning is very important

27 / 28 Conclusions Optimization of silicon artifact manufacturing process is ongoing –Sidewall roughness is of critical importance, needs to be improved –In-process trench roughness does not correlate to final sidewall roughness Edge sharpness is good CMM measurement shows good spacing and angular alignment

28 / 28 Acknowledgements MESAFab at Sandia Sandia, Lawrence Livermore National Lab, and Kansas City Plant Measurements Groups NIST