News

In-situ metrology for pad surface monitoring in CMP

CMP, Consumer Electronics

Thanks to S mart CMP it has been clearly demonstrated that CMP pads are largely under-utilized and are often discarded with more than half of the useful lifetime still remaining.

Chemical Mechanical Planarization (CMP) is one of the most critical processes in the semiconductor, hard disk and LED wafer manufacturing segments. The process is applied to assure the required planarity of the substrate wafer and for re-planarization at some intermediate points following deposition and lithographic processing for the structures built upon the wafer. In general, planarization is necessary to ensure functionality of the mutli-level interconnects within the structures, but the technique is also commonly used to reduce wafer thickness whilst maintaining uniformity.
During the CMP process, the wafer is held on a rotating fixture and pressed against a rotating polishing pad while an abrasive chemical liquid (slurry) is distributed between the wafer and the pad. Pad rotation and concentric grooves engineered into the pad surface serve to transport the slurry across the interface between the pad and wafer.  The chemical slurry weakens the surface of the wafer, thus enabling removal of material by asperities of the pad1 1-. J. McGrath, C. Davis. Polishing pad surface characterization in chemical mechanical planarization.( Journal of material processing technology, 153&154-2014.) .
The quantity of material removed from the wafer and the quality of the overall process is highly influenced by the surface properties of the polishing pad. The surface of the pad degrades as a result of the polishing action, so it needs to be constantly reconditioned. Conditioning involves an abrasive process on the pad surface, where a rotating abrasive or conditioning disk recovers the pad surface. These disks are usually made of stainless steel or electroplated diamond.

cmp_diagram
Figure 1. CMP process schematic diagram
cs_cmp-diagram-2
Figure 2. Wafer manufacturing. Without CMP (a) and with CMP (b)

Application note
Sensofar

pdf_download_icon

The need for surface metrology for the CMP process is evident given that mechanical interaction between the surfaces is a critical variable for all stages of the process. The surfaces that need to be periodically characterized throughout the process include the conditioning disk surface, the wafer surface and the pad surface. In high volume production environments, however, non-destructive in-situ pad characterization at natural pauses in the polishing process (for example, when changing wafers) is desired. This would permit detection of drifts in key pad parameters and assist in validation of process changes. The goal in all cases is to extend the lifetime of consumables and to improve the yield of the whole process.

cs_cmp-3-topography_50x_3x3_sp2x_disc75
Figure 3. 3D topography of conditioning disk surface

There are two main factors affecting pad surface degradation and lifecycle – pad groove occlusion and pad glazing.

Groove occlusion
During the polishing process the material removed from the wafer deposits in the pad grooves, thus occluding them. This prevents uniform slurry distribution across the water, thus causing non-uniform removal between the center and edge of the wafer. Monitoring of groove occlusion enables prediction of the need to clean the pad grooves and determining the best point in time to do it.  This cleaning operation can extend pad lifetime by up to 20%2 2-. T. Moore, N. Schwarz. NEOX Ex-Situ CMP Pad. NCCAVS, CMP Users Group Proceedings (2013). .

Pad glazing
Glazing is a more complex phenomenon where the polishing ability of the pad is reduced due to surface degradation. This phenomenon enhances wear between wafer and pad, thus increasing process temperature and that may result in material selectivity during polishing. The phenomenon cannot be predicted as easily as groove occlusion and requires constant monitoring in order to guarantee ideal performance of the CMP process.
In order to enable in-situ pad surface monitoring, the metrology approach used must be able to work under wet conditions. Only immersion metrology can meet these requirements. The main benefit of this approach is that the pad does not need to be removed from the polisher in order to be characterized. This makes it possible to monitor pad glazing and groove occlusion in-situ at various points in the pad’s life-cycle. In-situ metrology has proven to extend pad lifetime, allowing operators to utilize the pads to the end of their useful life.

Immersion objective
Figure 4. Immersion objective

The need for surface metrology for the CMP process is evident given that mechanical interaction between the surfaces is a critical variable for all stages of the process. The surfaces that need to be periodically characterized throughout the process include the conditioning disk surface, the wafer surface and the pad surface. In high volume production environments, however, non-destructive in-situ pad characterization at natural pauses in the polishing process (for example, when changing wafers) is desired. This would permit detection of drifts in key pad parameters and assist in validation of process changes. The goal in all cases is to extend the lifetime of consumables and to improve the yield of the whole process.

Smart Sensorhead, S mart CMP, 3D optical sensor
Figure 5. S mart CMP, 3D optical sensor
(a) Optical diagram of S mart sensor; (b) Pixel axial responser
Figure 6. (a) Optical diagram of S mart sensor; (b) Pixel axial responser

S  mart CMP is provided with Sensofar patented microdisplay technology and a high intensity blue LED, allowing combination of Confocal microscopy and Focus Variation technologies within a single sensor [3]. A software plug-in built around the CMP application provides all of the necessary tools and analysis for this application within the S mart CMP control interface.
The microdisplay enables projection of different fringe patterns onto the surface to be inspected, generating a confocal image of the surface. The confocal image is a greyscale map of the surface, where pixels have brighter signal if they are in the focus plane, while those out of focus appear completely black. By performing a vertical scan on the surface, it is possible to collect a stack of confocal images and obtain an axial response of the focus for each pixel. Processing the axial responses makes it possible to generate a 3D reconstruction of the surface.
By combining this metrology approach with an appropriate immersion objective, the S mart sensor can measure the pad asperities while the pad is still on the polisher.
Using the same immersion objective with Focus Variation technology, groove depth and width can be quickly determined in order to detect and monitor groove occlusion.

3D measurement of a new pad groove
Figure 7. 3D measurement of a new pad groove

Provided with Focus Variation technology, S mart CMP is able to measure a pad groove within a few seconds. This allows characterization of the groove width and depth, enabling groove occlusion monitoring along the process. Automatic analysis is made possible for production environments thanks to a function within the CMP software plug-in that determines both the width and depth of the groove independently of groove orientation.
following conditioning and polishing phases. This not only allows monitoring and thus determination of the right point in time for conditioning, but also permits characterization of the optimal conditioning time required to regenerate the pad surface. All these aspects help to extend usable (and valuable) pad lifetime, reduce the need for control wafers and additionally optimize the process – thus also avoiding re-processing of otherwise completed wafers.
Image showing evolution of the pad asperity height during the polishing process. After several hours of polishing a ‘glazing peak’ appears in the surface distribution. By monitoring the pad surface at intervals during the polishing process, the pad surface can be duly conditioned (intervention) in order to revert the surface condition to the initial state.

Image showing evolution of the pad asperity height during the polishing process.
Figure 8. Image showing evolution of the pad asperity height during the polishing process. After several hours of polishing a ‘glazing peak’ appears in the surface distribution. By monitoring the pad surface at intervals during the polishing process, the pad surface can be duly conditioned (intervention) in order to revert the surface condition to the initial state.
Image showing comparison for the pad before intervention (black) vs the pad following intervention (red)
Figure 9. Image showing comparison for the pad before intervention (black) vs the pad following intervention (red). The image on the left side was for the surface after 6 hour polishing process and the right one is after 12 hours.
topgraphy comparison
Figure 10. (a) New pad; (b) Good pad after 10h use; (c) Bad pad after 3h use

To simplify surface parameter control, the analysis of the acquired data can be automated. The operator then simply places the sensor over the pad surface and focuses on it, and once the acquisition has started the software will automatically show the target parameter value.
Special analysis algorithms developed for this application can automatically detect groove width and depth to monitor groove occlusion, while a different algorithm is designed to analyze the critical parameters of the asperities on the pad surface in order to detect pad glazing.

SensoPRO example
Figure 11. Software screenshot example

Related
products

Related products

Confocal technology has been successfully used in this study to investigate the surface topography of dental implants. This non-destructive measuring technique allows characterizing different locations of the complex threated dental implants with high resolution. Qualitative and quantitative analysis were done to evaluate the surface modifications. The following main points can be concluded from the various components of this study:
Surface damage varied at different evaluation sites for each implant system (neck, top, valley and flank) and also along the implant within each site.
The developed interfacial ratio (Sdr) variation correlated positively with the material volume variation (Vm) and is considered to be the best damage indicator.
The use of functional volume parameters of ISO 25178 standard is suggested rather than the Rk family parameters defined in the European report EUR 15178N due to the higher sensitivity encountered.
The 3D optical profilometer S neox (Sensofar, Spain) has been shown to be an accurate tool to investigate the surface topography of dental implants before and after insertion into bone. More advanced tools such as the S neox Five Axis, would not only enable faster and more automated high resolution measurements, but also the 3D reconstruction of the complete dental implant’s shape, which can be of interest for wear analyses or quality control in further studies.

The University of Mondragon (Surface Technology Group) Located in the Basque Country (Spain), the mission of the Surface Technology Group of this university is focused on the component surface analysis and development with the aim of functional optimization. To that end, the research is mainly focused on advanced surface metrology studies, on the effect of surface topography on functional behavior (tribology, fatigue, fretting, corrosion and biological response, among others) using both; experimental and numerical approaches.

Leave a comment

Contact Us

How can we help?

Start typing and press Enter to search