Muhamedsalih, Hussam (2013) Investigation of Wavelength Scanning Interferometry for Embedded Metrology. Doctoral thesis, University of Huddersfield.
Abstract

The tremendous growth in the manufacture of a wide range of deterministic and complex free form surfaces, has made surface metrology an essential part of the manufacturing process. The mass production of these surfaces at the micro and nano-scales is stimulating the development of new metrology instrumentation. Current instrumentation still suffers from one fundamental limitation, namely the inability to perform embedded (on-line) measurement in the manufacturing environment. The capability to measure surfaces on-line can improve yields through better process control, thus reducing costs and enabling the commercialisation of more hi-tech products. Better embedded metrology can also enable the manufacture of next generation surfaces with high precisions and more complex artefacts. Embedded measurement requires the development of precise on-line measurement instruments capable of a large measurement range with immunity from environmental vibrational disturbance.

This thesis introduces a new optical interferometry system for fast areal surface measurement at the micro and nano-scales which is robust against the effects of vibration. Wavelength Scanning Interferometry (WSI) together with an acousto-optic tuneable filtering technique has been developed which can measure surfaces with large step heights. The illumination source of the WSI tunes the wavelength provided from a broadband light (halogen-tungsten lamp) over a 93 nm range with a linewidth of approximately 2 nm. This light source is combined with a Linnik interferometer to generate an apparatus that can measure large discontinuous step heights absolutely and without any 2π phase ambiguity and with nano scale resolution.

A reference interferometer which acts as a position feedback sensor is multiplexed with the WSI in order to stabilise the system against environmental disturbances. The reference interferometer is integrated with an active closed loop control system to track the optical path length mechanically using a piezoelectric translator. The stabilised WSI can effectively perform surface measurements under conditions of vibration that would make measurement impossible for an unstabilised interferometer.

High speed measurement is of high importance for embedded metrology instrumentation as this has a direct effect on manufacturing throughput. In order to increase the speed of measurement calculation, this thesis presents a parallel programming model using large numbers of threads processed by many cores of a graphic processing unit (GPU). This model can accelerate the computing time up to 27 times compared to a conventional single threaded CPU calculation. This parallel programming model is based on CUDATM (Compute Unified Device Architecture) program structure developed by NVIDIA and harnesses the cheap computing power of commercial graphics cards. The system design, implementation, operation, fringe analysis algorithms and parallel programming model are all reported in this thesis.

The results of measurements for standard step height samples and a V-groove structured surface are also presented. For the standard step height specimens the system achieved large measurement ranges up to 96 μm with nanometre measurement uncertainty. The measurement results of a step height sample, under mechanical disturbance, showed that the system can successfully withstand environmental vibration up to 300 Hz. For the V-groove structure, the system successfully measured the sample by rotating the interferometer to face the groove side walls using an accurate rotary stage. The measurement result shows that the system has the potential to be used for on-line/in-process measurement on a shop floor.

Library
Documents
[thumbnail of Hussam_Muhamedsalih_-_Final_Thesis.pdf]
Preview
Hussam_Muhamedsalih_-_Final_Thesis.pdf - Accepted Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (4MB) | Preview
Statistics

Downloads

Downloads per month over past year

Add to AnyAdd to TwitterAdd to FacebookAdd to LinkedinAdd to PinterestAdd to Email