Brigid Mullany

Material Removal Mechanisms in Lapping and Polishing

Polishing processes are critical to high value production processes such as IC manufacturing. The fundamental material removal mechanisms, howeve, are poorly understood. Technological outputs (e.g., surface finish, sub-surface damage, part shape) and throughput of lapping and polishing processes are affected by a large number of variables. Individual processes are well controlled within individual enterprises, yet there appears to be little ability to predict process performance a priori. As a first step toward improving process modeling, this paper reviews the fundamental mechanisms of material removal in lapping and polishing processes and identifies key areas where further work is required.

The effect of slurry viscosity on chemical–mechanical polishing of silicon wafers

Abstract The chemical–mechanical polishing (CMP) process planarises wafers with a high degree of success; however, many of the fundamental mechanisms of the process are yet to be fully understood and defined. Much of the theoretical analysis to date has focused on kinematics [J. Electrochem. Soc. 138 (6) (1991) 1778], von Mises stress distributions [CIRP Ann. Manuf. Technol. 48 (1) (1999) 143; Thin Solid Films 308–309 (1997) 533; J. Electrochem. Soc. 144 (3) (1997) 1121] and hydrodynamic aspects [J. Electrochem. Soc. 141 (6) (1994); J. Electrochem. Soc. 146 (2) (1999) 761]. This paper is concerned with the hydrodynamic aspect of CMP. It details, both experimentally and theoretically, how changes in slurry viscosity change removal rates and influence the contact mode between the wafer and the polishing pad.

The Effect of Pad Wear on the Chemical Mechanical Polishing of Silicon Wafers

Abstract The Chemical Mechanical Polishing process planarises wafers with a high degree of success, however wear on the polishing pad causes the planarisation rate and the post-process planarity to deteriorate. To date, there has been no method of predicting the effect of this wear on the wafer planarity. Using finite element models of the process for new and worn pads the wafer stress distribution on the wafer surface can be predicted. Equating high stresses to high material removal rates these models predict that the process should become ‘centre slow’ as the pad wears. This correlates well with experimental data.

Using quantum dots to tag subsurface damage in lapped and polished glass samples

Grinding, lapping, and polishing are finishing processes used to achieve critical surface parameters in a variety of precision optical and electronic components. As these processes remove material from the surface through mechanical and chemical interactions, they may induce a damaged layer of cracks, voids, and stressed material below the surface. This subsurface damage (SSD) can degrade the performance of a final product by creating optical aberrations due to diffraction, premature failure in oscillating components, and a reduction in the laser induced damage threshold of high energy optics. As these defects lie beneath the surface, they are difficult to detect, and while many methods are available to detect SSD, they can have notable limitations regarding sample size and type, preparation time, or can be destructive in nature. The authors tested a nondestructive method for assessing SSD that consisted of tagging the abrasive slurries used in lapping and polishing with quantum dots (nano-sized fluorescent particles). Subsequent detection of fluorescence on the processed surface is hypothesized to indicate SSD. Quantum dots that were introduced to glass surfaces during the lapping process were retained through subsequent polishing and cleaning processes. The quantum dots were successfully imaged by both wide field and confocal fluorescence microscopy techniques. The detected fluorescence highlighted features that were not observable with optical or interferometric microscopy. Atomic force microscopy and additional confocal microscope analysis indicate that the dots are firmly embedded in the surface but do not appear to travel deep into fractures beneath the surface. Etching of the samples exhibiting fluorescence confirmed that SSD existed. SSD-free samples exposed to quantum dots did not retain the dots in their surfaces, even when polished in the presence of quantum dots.

Optical polishing pitches: impact frequency responses and indentation depths

The short-term dynamic properties of a wide range of commercially available optical polishing pitches were determined with an impact frequency response test. A simple dynamic model is presented to evaluate the relevance of the measurement variations across the range of pitches tested. Test results show that harder pitches have higher natural frequencies and lower damping ratios. Strong correlations between pitch hardness (as measured by the indent test) and the natural frequency only existed among pitches from the same series, i.e., the Gugolz or Acculap series. This implies that pitches with similar hardness values from different manufacturers have different dynamic properties.

Fabrication of aspheric optics: process challenges arising from a wide range of customer demands and diversity of machine technologies

Different applications of aspheric optics and their related fabrication methods are firstly summarized and then discussed using Carl Zeiss examples. This is done in order to highlight the potential and challenges of fabricating aspheric optics. The need to stimulate new ideas for extending manufacturing capabilities, process improvements and new fabrication technologies will also be outlined.

Positioning sensor by combining photogrammetry, optical projection and a virtual camera model

A novel, low cost, non-contact measurement technique is proposed to realize an optical positioning sensor that enables real-time measurement of a small lightweight modules location. The technique is based on a combination of photogrammetry, optical projection and a virtual camera model. The module generates an optical pattern that is observable on the surrounding walls, photogrammetry is used to track the motion of the projected optical pattern, and this indirectly allows the motion of the module to be tracked. This is done by treating the module as a virtual pinhole camera, generating a virtual image that carries the angular characteristics of the optical pattern. Images of the projected optical pattern on the walls taken by cameras around the room, together with the virtual image, are then processed through a photogrammetry-based bundle adjustment. This results in a position and orientation estimate of all cameras, including the virtual pinhole camera. The position and orientation of the virtual camera is therefore also the position and orientation of the module. Both experiment and optical ray-trace simulations are performed to validate the proposed technique. Experimental agreement of 3 parts in 104?was obtained by translating the module over 0.9?m. The incorporation of the virtual camera model is innovative, leads to a simple solution, and is more robust than the previously published approach.

Using Optical Projection in Close-range Photogrammetry for 6DOF Sensor Positioning

A novel, low cost, non-contact, six degrees of freedom (DOF) measurement technique is proposed that enables real-time measurement of a small lightweight module’s location. Straight forward applications of the proposed technique include robot calibration by installing the module to the end effector of a robot arm, and head-tracking in a typical virtual reality environment by attaching the module to a human head. The technique is based on a combination of photogrammetry and optical pattern projection. The module generates an optical pattern that is observable on the surrounding walls, and photogrammetry is used to measure the absolute coordinates of features in the projected optical pattern with respect to a defined global coordinate system. By combining these absolute coordinates with the known angular information of the optical projection beams, a minimization algorithm can be used to extract the absolute coordinates and angular orientation of the module itself. Experimental agreement of 1 to 5 parts in 103 was obtained by translating the module over 0.9 m and by rotating it through 60°. Numerical simulations were conducted to demonstrate that optimum design of the projected pattern gives a lower associated measurement uncertainty than is possible by direct photogrammetric measurement with traditional tie points. This paper documents the proof of principle and describes how the measurement can be further improved.

VORTEX MACHINING: LOCALIZED SURFACE MODIFICATION USING AN OSCILLATING FIBER PROBE

This paper demonstrates for the first time a method for surface modification of a substrate material based on the generation of localized vortices of abrasive slurry using slender oscillating fibers. In experiments presented in this paper, the abrasive slurry is a water based suspension of 1 µm alumina particles. This is pumped onto, and flows across, the specimen surface. A fiber (typically 7 µm in diameter and between 3.5 to 5 mm long) is immersed into this flowing slurry and oscillated at frequencies around 30–40 kHz to produce a small rotational flow (vortex) that results in the locally accelerated particles. Using such a system, it has been possible, over machining times of 6–24 hours, to produced localized depressions in the surface of a silicon substrate with typical depths of around 60–700 nm and widths of around 10–300 µm. Based on these initial studies the material removal rate is estimated to be approximately 40 nm per hour. Using white light interferometry and stylus profilometry the surface deviations (roughness) of these features have a root mean square variation in the region 1–2 nm, which is comparable to that of the surface remote from the machined feature.

Evaluation of fiber-based tools for glass polishing using experimental and computational approaches

Polymeric pad or pitch-based tools combined with loose abrasive slurries are typically used in the polishing of optical materials. In this paper, the potential of fiber-based tools to both remove material and provide high quality surface finishes on BK7 glass is explored. The potential advantage of fiber-based tools over traditional tools is their inherent compliance, which could accommodate varying workpiece surface curvatures as found in aspheres and freeforms. To evaluate the new tools, both experimental and finite element (FE) modeling approaches were taken. A FE model consisting of a single fiber engaged with the workpiece surface was used to estimate the shape and magnitude of the pressure distribution exerted by the fiber on the workpiece surface. Two different tool configurations, yielding two different Fes, predicted pressure distributions, were used to polish BK7 samples, and the material removal profiles were interferometrically measured. The resulting profiles and the predicted pressure distributions share the same v-shape. While differences in scale exist between the experimental and FE-predicted profiles, the tool generating higher material removal had the greater predicted pressure distribution, thus demonstrating the ability of the FE model to provide insights into tool design. Additional testing was conducted to determine if the tools removal rate can be predicted by Prestons equation. Initial results indicate the equation is valid within the range of parameters tested. The surface roughness of BK7 samples processed by this tool was measured and some deterioration on the Sq value was noted; the surface roughness increased from 1.89 to 3.66 nm Sq. Over several hours of continuous use, the load applied by the fibers decays in a repeatable manner, and little wear was observed on the fibers after 5.33 h of polishing.