Printed circuit boards (PCBs) are a key component when it comes to making progress in electronic science and technology. As such, they are expected to function flawlessly and as designed. There is no tolerance or margin of error, no matter how minor it seems, as PCBs are employed in applications in industries that require cutting-edge technology, such as aerospace, biotechnology, automotive, military, and many more.
![An Insight into Electronics: How Spectral CT Can Be Used to Create Failure-Free Devices](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_17079222377531668.jpg)
Image Credit: TESCAN USA Inc.
While numerous PCB testing protocols exist, including in-circuit testing or bare board testing, these tests may not detect any performance issues that fall outside of their testing ranges. PCBs are most usually subjected to further optical and internal validation protocols.
One technique that enables visualization of internal PCB materials is X-Ray imaging, which can be leveraged to formulate a strong understanding of the material’s internal structure and any potential defects. Examples of defects that can occur are an excess of solder, component misalignment, micro-cracks, and voids.
As components become smaller and multiple layers of sophisticated material mixes are added, the area on PCBs becomes crowded and a single radiography approach has its limitations. This article highlights the use of the entire X-Ray energy spectrum through the application of spectral radiography to identify the individual materials present.
With spectral radiography, components in a mixed or multi-layered electronic device become clearly visible. Herein, conventional and spectral X-Ray imaging techniques are not limited to the analysis of PCBs. For instance, a multimodal conventional and spectral CT study was conducted on a smart ring wearable electronic device.
Wearable electronic devices are becoming more commonplace in today’s world. These devices feature a variety of sensors to track temperature, heart rate, blood oxygen, movement etc. These devices usually require other sophisticated components, such as batteries, antennas, or gyroscopes, to monitor and transmit data relative to the user.
Given the unique shape of some of these components, it is vital to be able to assess and understand the internal connections once placed into the final device. This article also outlines how, in the rapidly expanding market of electronics, traditional micro-CT, in combination with spectral CT, can carve out a path toward a future with failure-free electronic devices.
Materials and Methods
A dual In-Line Memory Module (DIMM), typically referred to as a RAM stick or RAM module, has a structure similar to a PCB and was consequently chosen to demonstrate the capacities of spectral radiography.
The entire sample was reviewed at 160 kV accelerating voltage, 20 W target power, and a resolution of 50 µm. This produced an image identifying and distinguishing singular components across the sample, as overlaying components impede the visibility of underlying structures.
While micro-CT facilitates 3D sample analysis, time constraints and other limitations mean it is not always possible or the desired method. Spectral radiography makes it possible to resolve each material layer individually without having to perform to a full 3D CT scan.
By leveraging the multi-energy information from a single spectral radiography acquisition, the various materials in the X-Ray path can be distinguished. From this full spectral information, an unlimited number of materials can be seen and differentiated without needing to pre-tune the acquisition parameters, which is usually required in conventional dual-energy approaches.
The spectral radiograph of the DIMM allows for separate visualization of the Cu and polyimide board signals. The spectrum can also be fine-tuned for filtering K-edges. Since any given element has its own unique absorption edge, it is possible to differentiate the materials used in the sample.
The lower detection limit for K-edges sits just below 20 keV, meaning that the K-edge detection method can be used for materials that contain molybdenum or heavier elements. Throughout this study, a 2.5 cm diameter smart ring wearable device, which features sophisticated electronics and a battery pack, was subject to analysis.
The TESCAN UniTOM XL SPECTRAL was used to perform a standard micro-CT scan of the entire sample, using a voxel size of 13 µm and acceleration voltage of 180 kV. In the next step, a complete spectral CT scan was conducted by sliding the spectral detector into the field of view.
The detector tallies every X-Ray photon that hits its surface after interacting with the sample. Simultaneously, the energy of each of those photons is measured, generating a spectrum between 20 and 160 keV. This enables an analysis of how the photons of varying energies interact with the various materials in the sample. This information is acquired in just a single CT scan.
Using spectral CT to analyze the spectrum, a comprehensive identification of the main atomic number in a specific area of the sample using K-edge detection is in reach, even without knowing any compositional information before scanning.
The spectral CT scan of the smart ring acquisition time was 12.4 hours and was captured with a voxel size of 40 µm. The standard CT scans were reconstructed with the support of Panthera™ software for 3D visualization and analysis, in addition to VGSTUDIO MAX (Volume Graphics) for 3D rendering.
The TESCAN PolyDET II detector was used for capturing spectral CT scans. For reconstruction, visualization, and analysis the TESCAN Spectral Suite™ was used. For the acquisition of both spectral and standard CT scans, Acquila™ was used.
Regions of interest were identified from the reconstructed volume based on the data acquired from the conventional overview and the spectral CT scans of the smart ring. Panthera™ makes it possible to highlight regions of interest directly on the reconstructed slices.
The coordinates of these volumes of interest (VOIs) can be transmitted into Acquila™ acquisition software. Consequently, using a voxel size of 4.5 µm, batch scanning was run on the VOIs. Correlations of the five high-resolution VOI scans were made with the overview volume by leveraging the shared coordinate systems.
![A) Conventional radiograph of the DIMM. B) Radiograph only showing the Cu signal. C) Spectral radiograph with Cu signal subtracted, revealing connections not visible in the other images](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_17079783567845894.png)
Figure 1. A) Conventional radiograph of the DIMM. B) Spectral radiograph only showing the Cu signal. C) Spectral radiograph with Cu signal subtracted, revealing connections not visible in the other images. Image Credit: TESCAN USA Inc.
Results and Discussion
DIMM Radiographs
The results of traditional radiography performed on the DIMM are displayed in Figure 1A. This shows the combination of all material interactions across the sample. As previously mentioned, the brightest regions in the image may be indicative of either extremely high atomic numbers typically related to something like solder material, or a combination of stacked layers of materials with lower atomic numbers.
The areas that are darkest in the image are made up of a relatively thin layer of a material with a low atomic number. This is the polyimide board itself. As displayed in Figure 1B, the PCB’s polyimide board signal is deducted from the radiograph, this leaves only Cu tracks visible in the image.
This offers a clearer picture on the connections between components and facilitates the observation of features that were otherwise hidden in the traditional radiograph. When the Cu signal is deducted from the radiograph, features that were otherwise obscured by the Cu signal become visible, as displayed in Figure 1C.
K-edge detection on the radiograph identifies absorption edges for barium (capacitors), tin (solder bumps) and gold (connectors and wire bonds). While it is possible to distinguish the capacitors and solder bumps from other features in the sample when observing the results from conventional radiographs, the gold connectors and coatings really jump out in the spectral radiographs.
Although gold is considered to be a high attenuating material, the limited dimensions of the wire bonds in contrast to the other components in the DIMM and the very thin gold coating on the connectors makes it almost impossible to differentiate in a traditional radiograph.
Figure 2 displays the K-edge filtered images for the gold absorption edge at 80.7 keV. The gold in the IC wire bonds is revealed at several locations across the DIMM and at the gold-plated contact points that connect the DIMM to a PC’s motherboard.
Gold is used sparingly at these contact points to protect the underlying nickel, preventing it from oxidizing. Figure 2A reveals a more detailed, higher resolution radiograph which enhances visualization of the small wire bonds.
![K-edge filtered image showing only the gold signal on the DIMM. The gold is present in wire bonds (A) and on the connectors (B).](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_17079783697179074.png)
Figure 2. K-edge filtered image showing only the gold signal on the DIMM. The gold is present in wire bonds (A) and on the connectors (B). Image Credit: TESCAN USA Inc.
Smart Ring
While radiographs can be useful for analyzing flat samples such as the DIMM, more intricate and sophisticated electronic devices such as the smart ring call for a 3D approach. The initial basic CT scan of the ring offers a clear overview of its various components as shown in Figure 3.
At a resolution of 13 µm, this scan provides an overview revealing the location of the different areas of interest in the sample. These areas of interest include components such as the battery, sensors, antennas, etc., as well as possible defects inside these components. For instance, delamination defects inside the smart ring’s battery can be seen in Figure 4.
![Renders of the smart ring, acquired using conventional CT at a voxel size of 13 µm, visualized using different opacity settings.](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_1707978384527627.png)
Figure 3. Renders of the smart ring, acquired using conventional CT at a voxel size of 13 μm, visualized using different opacity settings. Image Credit: TESCAN USA Inc.
![Horizontal cross section through the conventional CT scan of the smart ring. The battery shows delamination defects at different locations](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_17079783927184164.png)
Figure 4. Horizontal cross section through the conventional CT scan of the smart ring. The battery shows delamination defects at different locations. Image Credit: TESCAN USA Inc.
To determine other potential areas of interest, traditional micro-CT was run in combination with spectral CT. The spectral detector was positioned into the field of view to obtain a full spectrum 3D volume. This volume contains complete spectral data in every voxel of the dataset. Resultingly, the volume can be viewed as a combination of 140 monochromatic tomograms ranging between 20 and 160 kV.
In Figure 5 a standard CT slice (left) and a spectral CT slice (right) are shown side by side for a comparison of the central part of the ring. By way of K-edge detection, areas holding high-Z elements, including gold or tungsten can be automatically identified and mapped across the 3D spectral volume. The insert in Figure 5 indicates the presence of tungsten and gold respectively as locations inside the infrared LED of the heart rate sensor where K-edges at 69.5 and 80.7 keV are shown.
![Conventional (left) and spectral (right) CT slice through the center of the smart ring. Peaks at 69.5 and 80.7 keV indicate the presence of tungsten and gold.](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_17079784015742215.png)
Figure 5. Conventional (left) and spectral (right) CT slice through the center of the smart ring. Peaks at 69.5 and 80.7 keV indicate the presence of tungsten and gold. Image Credit: TESCAN USA Inc.
Both the conventional CT and spectral CT overview scans were analyzed with Panthera™, and areas of interest were marked for scanning at higher resolution to resolve additional detail. The standard CT scan was used to identify volumes of interest for the two battery defects.
Where there is a presence of gold, the spectral CT scan was used to select VOI locations: the gyroscope and the two infrared LEDs from the heart rate sensor. These locations are displayed on a horizontal slice and on a 3D render as shown in Figure 6.
The resultant high-quality scan of each region, at 4.5 µm voxel size, is useful for studying different components at high resolution using traditional CT. In case of the battery pack, VOI scanning creates a comprehensive image of the delamination in the battery.
This shows crucial information relative to the degradation of batteries, which is of vital importance for wearable devices that come into close contact with the human body, as battery malfunctions could lead to serious injury. A comprehensive view of one of the delamination defects in the smart ring’s battery pack is displayed in Figure 7.
![Location of the volumes of interest for detailed CT scanning: gyroscope (A), infrared LED (B,C), battery defects (D,E).](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_17079784095937154.png)
Figure 6. Location of the volumes of interest for detailed CT scanning: gyroscope (A), infrared LED (B,C), battery defects (D,E). Image Credit: TESCAN USA Inc.
![Detailed view of battery lamination by VOI scanning on the smart ring](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_17079784194683188.png)
Figure 7. Detailed view of battery lamination by VOI scanning on the smart ring. Image Credit: TESCAN USA Inc.
Leveraging the other volume of interest scans, it is possible to correlate signals of gold, tungsten, or even tin obtained from the spectral CT scan to the high-quality structural information.
On the high-resolution VOI scans, the precise origin of the different K-edge signals can be located for a clear picture of the sample’s true structure. Figure 8 exhibits the locations of the different K-edge signals and pinpoints the location of one of the gold signal origins.
The gyroscope is comprised of a series of three small gold connections, which are resolved and clearly differentiated on the VOI scan. The smart ring contains only one tungsten component and as the image shows, the majority of the soldering material is made up of tin.
![Left: Location of Au, Sn and W in the smart ring sample. Right: detail of the VOI scan of the gyroscope showing the gold connectors used in this component](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_17079785389416578.png)
Figure 8. Left: Location of Au, Sn and W in the smart ring sample. Right: detail of the VOI scan of the gyroscope showing the gold connectors used in this component. Image Credit: TESCAN USA Inc.
Conclusions and Outlook
Combining traditional CT and radiography methods with their spectral counterparts can be a powerful tool for gaining a deeper understanding of sophisticated electronic components. Where CT can be seen as the cornerstone of non-destructive 3D characterization, spectral CT opens the way for identifying gold in electronics.
By leveraging the multi-energy information from a single spectral radiography acquisition, there is the potential to discern the presence of different materials in the X-Ray path of a radiograph.
The number of materials that can be distinguished from the full spectral information is unlimited, even without having to pre-tune the acquisition parameters as is usually the case when conducting classical dual-energy approaches.
The spectral information facilitates the identification and selection of specific regions of interest that may benefit from further investigation with high resolution CT imaging. Such advancements may close the gap in the failure analysis track and ultimately usher in an era of failure-free devices.
![](https://d12oja0ew7x0i8.cloudfront.net/images/Article_Images/ImageForArticle_23357_45336407746365746816.png)
This information has been sourced, reviewed and adapted from materials provided by TESCAN USA Inc.
For more information on this source, please visit TESCAN USA Inc.