The role of automatic defect recognition
BY VIKRAM BUTANI
The introduction of optical components in electronics assembly has demanded an urgent response from test and inspection suppliers. Major electronic manufacturing service (EMS) providers have set up optoelectronics R&D facilities to develop manufacturing techniques for these products.
An important step toward process verification and quality control in optoelectronics manufacturing is the availability of high-resolution X-ray inspection. As material densities have decreased, and the inspection requirements have increased, the resolution of X-ray systems has improved to accommodate new specifications.
This article describes two such improvements in X-ray technology that are a direct response to optoelectronic components. The first is the use of digital detectors to achieve more accurate X-ray images. The second is automatic defect recognition (ADR), an integrated X-ray technology that represents the best in platform, detector and image analysis algorithms.
Present Day Optoelectronic Components
The material content of the various optoelectronic components makes X-ray inspection a challenge. Optoelectronic components consist of optical fibers with very low material density, combined with the III-V family of materials. III-V materials are binary crystals with one element from the metallic Group III of the periodic table and one from the non-metallic Group V. The family includes gallium arsenide (GaAs), indium phosphide (InP), gallium nitride (GaN), indium antimonide (InSb) and indium arsenide (InAs).
Optoelectronic devices are typically planar, thus enabling easy fabrication onto GaAs or silicon wafers. In turn, these wafers can be applied in the manufacture of multiplexers, demultiplexers and high-speed optical switches for telecommunications applications. VCSEL (vertical cavity surface emitting laser) technology is also the basis for many new photonic components. Essentially, it is a semiconductor micro-laser diode that emits a vertical cylindrical beam of light from a wafer's surface.
Figure 1a. An image intensifier based X-ray inspection system.
X-ray inspection of optoelectronic components is typically used to detect voids in the solder and defects in the actual fiber. Inspecting for solder voids around the optical fiber is important because these fibers are very delicate, and they could be damaged in the field after installation if they are not properly protected by the solder. This is a challenging X-ray scenario because of the large variation in the material densities of the voids, the solder material and the optical fiber.
X-ray technology works on the concept of penetration of the samples under inspection. The higher the density of the sample, the higher kV level that is required. However, at high kV levels, images of low-density materials get washed out. Hence, the material content of optoelectronic packages, made up of low-density optical fibers combined with high density III-V compounds, makes X-ray inspection of optoelectronic packages a challenge.
Present day X-ray systems are ranked primarily on the basis of their magnification capabilities, or spatial resolution. Three parameters – spatial resolution, contrast resolution and system speed – placed at edges of a triangle with a central pivot conceptually represents that increasing any two parameters decreases the third. Adding a fourth parameter – energy resolution – extends the parameter space in ways that are not as straightforward to describe. We do know that an increase in energy resolution can result in an increase in contrast resolution but with a likely decrease in spatial resolution and system speed. However, the appropriate measure of the system is the system resolution, which consists of both contrast and magnification. At high-resolution levels, the ability to image minute features is substantially enhanced by the ability to capture subtle changes in density.
Figure 1b. A digital detector based X-ray inspection system.
Magnification capabilities of contemporary X-ray systems are enhanced by the use of open tube based X-ray sources, in which the filament is closer to the window. This results in a lower source-to-object distance as compared to closed tube based systems. This allows greater magnification as compared to the one or two micron size of the focal spot. The focal spot size at any given time is directly proportional to the wattage being emitted by the source at that time. The rule of thumb under 10 microns is 1 micron of focal spot size per watt of output. Accordingly, at 80 kV and 0.1 mA, the source output is 8 watts, and the spot size is closer to 8 microns than to the starting point of 1 or 2 microns. Similarly, at 160 kV and 250 mA, the output is 40 watts, and the spot size is much larger than 1 or 2 microns. The difference between an open and closed source is thus reduced primarily to the closeness of the filament to the output window in an open source, as compared to a closed source.
The X-ray industry has primarily used image intensifiers as the detector of choice for real-time X-ray inspection systems (Figure 1a). This analog detector produces an 8-bit signal that is reduced to 4 to 5 bits of data by the time it goes through the video chain and is reproduced on the monitor. Therefore, the contrast levels are limited to 16 to 32 shades of gray.
New Detector Technology
Recently, the X-ray industry has started using a digital detector system for advanced inspection of low-contrast applications (Figure 1b). Amorphous silicon (a-Si) imaging technology, which was developed by medical equipment manufacturers for digital radiography, has gained importance by recent breakthroughs in thin film transistor arrays, similar to those found in notebook computer screens. As a result, the a-Si detectors, the latest generation that produces images in a digital 16-bit format yielding more than 65,000 shades of gray for analysis, achieve the resolution needed for optoelectronic applications.
Figure 2. A cross section of the thin-film structure of a digital detector based on amorphous silicon.
The technology is based on a two-dimensional, solid state, amorphous (non-crystalline) silicon imaging array that contains hydrogen (Figure 2). A scintillator is deposited directly onto the surface of the arrays. X-ray photons striking the phosphor are converted to visible light, which is absorbed and converted to an electric charge by the photodiodes. The charge is integrated on each photodiode so that each pixel collects a signal proportional to the local flux of the X-ray beam. When the array circuitry scans the diodes, the charge is converted into a video signal.
This video signal is transmitted directly onto the monitor, which reproduces the X-ray image. There is very little loss of signal as there is no A-to-D conversion or transfer between lenses and cameras. This image is far richer in information because the spectrum between black and white can now be divided into more than 65,000 shades of gray, instead of 256 shades of gray with the traditional imaging systems. Subtle changes in density can now be captured very effectively in the X-ray image.
Assessing Detector Performance
The most accurate and easily ascertained image quality indicator (IQI) is “detective quantum efficiency” (DQE) because it takes into consideration the combined effects of signal plus noise. DQE may be defined as the ratio of the square of the signal-to-noise ratio (SNR) of the detector image, to that of the input image (for instance, the original image in the form of X-ray flux) as presented if the system were perfect. To obtain information about overall signal and noise performance of any imaging detector, DQE is studied as a function of spatial frequency. The spatial frequency dependant DQE is an effective means of comparing the imaging results of different detectors. Recent technical literature has shown the effects on DQE of commonly encountered parameters, such as pixel fill factor and modulation transfer function (MTF) plus various sources of electronic noise.
Automatic Defect Recognition (ADR)
Once the detector has transmitted the image onto the monitor, it is a greater challenge to get the maximum amount of information from the image in the shortest amount of time; it is on the manufacturing floor where the speed of the system is most relevant. In a failure analysis lab, the engineer can take an extended amount of time to understand the complexity of the image and the relevant issues. On the manufacturing floor, the operator does not have the time or the engineering skills to tackle the minute details of the problem – the operator needs pass/fail feedback right away.
Figure 3. Three individual X-ray images of connectors using (clockwise from the upper left) a digital detector, an image intensifier, and a digital detector with an automatic defect recognition filter applied to it.
A number of software filters for X-ray image analysis have been developed, including sharpen, smoothen, emboss, laplacian, etc. These filters enhance the image quality to provide more information to the operator.
Recently, as part of ADR technology, more sophisticated image enhancement filters have been implemented to help the system pick out defects in samples. With the help of digital detector technology, information about the various density levels is available in a single image because of the high 12-16 bit image depth of the detectors. However, this information cannot be visible all at one time. ADR filters make full use of the high bit depth available from the digital detectors. (Figure 3 is an example of the relative clarity of different X-ray systems.) This not only reduces inspection time, but because the filter conducts this inspection automatically, it removes the operator subjectivity from the inspection process. Overall, these filters help to reduce inspection cycle time, resulting in high productivity and consequent cost savings.
ADR technology can now be applied not only to optoelectronics, where there is a large difference in material density, but also in inspection of area array packages (Figure 4), where operator subjectivity plays a crucial role. Thus, a large optoelectronic backplane board containing optical components placed beside a myriad of area array packages can all be inspected more objectively.
In the fast advancing world of X-ray technology where system designs are being driven by the development of new components, it is important to remember the operator's perspective when considering one system over another. “Nanofocus” technology and “oblique views” may help to get a higher magnification, but if there is not enough data being transferred by the detector, there isn't enough data for the engineer to analyze. Simply integrating a digital detector into an X-ray system will not supplant the value of extensive experience with digital detector technology. It is also important to have the right set of tools to obtain a clear interpretation of the data. AP
Vikram Butani, general manager, can be contacted at VJ Electronix, 89 Carlough Road, Bohemia, NY 11716; 631-589-8800; Fax: 631-589-8892; E-mail: email@example.com.