AZtec sample navigation using imported images

Abstract

Locating the area(s)/features of interest on a sample in a scanning electron microscope (SEM) can sometimes be difficult. The reason for this relates to the high magnifications that the SEM can achieve — which can make the move from macro to micro scale difficult. When working with large or multiple samples, knowing where one is on the sample can be particularly challenging. Similar problems arise when the same sample area has been analysed using different instruments/techniques (e.g., Energy Dispersive Spectroscopy (EDS), Electron Backscatter Diffraction (EBSD), Atomic Force Microscopy (AFM)) and there is a requirement to correlate the datasets.

Here we present a solution integrated into the AZtec NanoAnalysis suite, allowing an imported image from a camera, optical microscope, or even a schematic diagram, to provide a complete sample overview, and to aid navigation on the sample. This approach utilises registered images of the sample and makes it a lot easier and quicker to navigate around the sample and ensure that the analysis in AZtec is done in the correct location.

Introduction

There are several ways in which image registration can be used to aid in the process of navigating around samples. A variety of data and image sources can be used including existing AZtec data, or images acquired using AZtecLAM (Large Area Mapping), as a reference for further analysis done on the same sample, either during the same analysis session or later.

Previously acquired data can simply be loaded into AZtec and, with the sample mounted on the SEM stage, the stage is moved to align features on the "reference image/data" with the current stage position. After having defined the stage positions for two selected reference markers (i.e., fiducials), it becomes possible to drive the stage x, y motors using the old dataset for navigation. This can be very useful if the same sample is analysed on different instruments or is revisited at a later point in time to extract more information about features of interest.

This functionality also allows images collected by other means, such as photos, optical images, or schematic diagrams, to be loaded into AZtec and used in the image registration process described above. These images can then be used to drive the SEM stage to features of interest, which are often obvious in an optical image or schematic. This is particularly useful if the sample is large, and therefore difficult to get a good initial overview from in the SEM. Similarly, if the sample is analysed using different instruments or different techniques, it can be very powerful to use an image from one system as a reference for locating features for analysis in AZtec. This application note will demonstrate how image registration works and provide several different examples where this new functionality has been used to reduce the time spent and difficulty associated with navigating to features of interest.

The AZtec Solution

One of the easiest ways of using an image for sample navigation is to take a photo of the sample before it is loaded into the SEM (an example is shown in Figure 1).

Fig 1. Photograph of a sample block.

This photo can then be loaded into AZtec. An easily identifiable feature is then located on the sample and is centered within the field of view on the SEM. The feature is centered in the SEM image and is also marked on the optical image. Once accepted, this saves the image coordinates together with the stage coordinates. The process is repeated using another identifiable feature on the sample and optical image.

Once image registration is complete, the software shows the current SEM field of view as a blue square on the reference image. The size of the square is determined by the magnification of the SEM. This means that the optical image can now be used for navigating and setting the magnification to get the features of interest positioned in the central part of the scanned electron image.

Figure 2 shows two examples of where the stage has been positioned using the imported image as a reference. First the scanned area was centred around the Faraday cup to allow measurement of the beam current, then it was moved to the manganese sample for analysis.

Fig 2. Sample stage repositioned by navigation using the imported image. Double click on the optical image and the SEM stage automatically moves to the selected location. a) Sample positioned with the Faraday cup in the centre of the field of view. b) The manganese sample is selected on the optical image and the stage is automatically moved prior to EDS spectrum acquisition.

This approach of using an imported optical image provides a simple and intuitive way of navigating on samples in the SEM, making it quick and easy to ensure that the analysis performed in AZtec is done at the correct location(s).

Application Examples

Using Image Registration for Large Area Mapping

Large area mapping is an automation function that allows the user to cover a large area on a sample by combining a sequence of stage movements and single field data analysis (e.g., electron imaging, EDS and/or EBSD). These fields can be montaged together by AZtec to produce a high-resolution, large-scale map. The total area is defined by a few predetermined stage positions, typically by manually driving the stage to the corner positions of the desired area. For large irregularly shaped features this can be difficult to setup and result in a LAM area being defined that does not fully cover the feature of interest.

In this example a photo of the sample was taken with a smartphone and then loaded into AZtec using the Image Registration step using the process described earlier.

Fig 3. Photo of beetle fossil sample. The area of interest is marked by the orange rectangle.

Getting the setup of the area correct is important as, too large an area means it will take more time than necessary to acquire the data, and too small could result in the part of the sample of interest (the beetle) not being fully covered during the analysis.

(a)

(b)

Fig 4. Montaged electron image and EDS layermap covering the entire beetle. This makes it easier to zoom in to study the details of the fossil, and to identify areas/features of interest for more in-depth analysis.

Geological Thin Section

In geosciences polished thin sections are routinely viewed and imaged on optical microscopes to study the different minerals and textures in rock samples. Both plane polarized light (PPL) and cross polarized light (XPL) are used for mineral identification, with XPL providing additional information about the minerals present. Combining optical scans of a geological sample can provide a useful roadmap for navigation when the samples are viewed and analysed in an SEM.

Here we consider a sample with large grain size variations, which can easily be seen on the XPL image shown in Figure 5b. The large variation in grain size is related to the conditions under which this rock formed.

(a)

(b)

Fig 5. Optical images of a geological sample. a) Plain polarised light, primarily showing mineral distributions. b) Cross polarised light, primarily showing grain size variation.

In order to improve the understanding of how this rock formed the sample was analysed in the SEM, using BSE imaging, EDS and EBSD. Being able to easily use the images obtained from the optical microscope to guide the analysis in the SEM to specific areas of interest is not just time saving, but also very beneficial when correlating the different datasets afterwards.

Figure 6 shows an example of BSE imaging, EDS element mapping, and EBSD from the same area of the sample. This analysis gives more in-depth information about the exact mineral phases, and how the observed microstructure was created.

(a)

(b)

(c)

(d)

Fig 6. A variety of SEM data collected from the same geological sample. a) BSE imaging, b) EDS element map of magnesium, c) EDS element map of aluminium, d) EBSD map showing the grain orientation.

AFM (Atomic Force Microscopy) and EDS Analysis on the Same Sample

In this example we look at an SRAM (Static Random-Access Memory) device analysed with AFM and EDS. The AFM analysis was done first in order to avoid issues caused by beam damage and/or contamination. On the Asylum AFM an optical image of the sample was saved in order to show where the AFM measurements were made. The same optical image was loaded into AZtec using the Image Registration step and used to aid sample navigation on the SEM. This enables the EDS analysis to be performed on the same area as the AFM analysis.

Being able to ensure that the data acquisition is taking place in the same areas is the first challenge when trying to correlate data from the same sample acquired using different techniques, on different instruments. The example in Figure 7a shows the optical image from the AFM, which is being used to show where on the sample the AFM data acquisition took place. As demonstrated, using Image Registration in AZtec, the same area can easily be found in the SEM, making it possible to relate the topography data provided by AFM data to the EDS data acquired in the SEM. The two datasets can be combined in the Relate software which makes it easy to correlate the AFM and EDS data.

(a)

(b)

(c)

(d)

Fig 7. a) Optical image taken in AFM. b) Electron image from the same area as highlighted on the optical image. c) AFM topography map. d) EDS layermap showing the chemical composition of the same area.

Failure Analysis of Solder Pads

When performing failure analysis of devices it is often possible to relate the failure mode to specific locations of interest in the device based on the design schematics. If the design schematics contain easily identifiable surface features, then it can be used in the same way as a photograph taken of the device.

Here we have used this approach when looking at poor contact pads in the device shown in Figure 8. It is known that some of the contact pads are of poor quality. Using an optical microscope it is possible to see variations between the solder pads and relate to the expected location of the failed contact. The SEM can then be used to characterize the poor-quality pads in greater detail, and subsequently, to determine the likely cause of failure.

Image Registration in AZtec makes it very easy to navigate directly to the pads of interest, saving time and obtaining a faster result. Figure 9 shows an example of the EDS data acquired from two of the damaged pads. The pads are supposed to have a gold coating on top of the nickel plating, whereas the EDS data shows that the gold layer is not homogeneous and has holes where the underlying nickel is exposed. Such "black pad defects" have been shown to relate to failure of solder joints, so being able to quickly identify and confirm devices with damaged contact pads is very important.

Fig 8. Schematic diagram showing positions of contact pads on a PCB next to a montage of optical images of the pads on a failed device.

Fig 9. EDS element maps acquired for two contact pads from the sample shown in Figure 8.

Conclusion

As shown in this application note, a wide range of samples are analysed using SEM based techniques. The Image Registration functionality in AZtec is a powerful tool, making sample navigation and analysis quicker and easier, and thereby saving the user time. Additionally, it ensures that analysis performed using different techniques (e.g., EDS, EBSD, AFM, etc.) can take place on the same sample locations, making it easy to subsequently relate results from a wide range of measurements.