Issue link: https://iconnect007.uberflip.com/i/831357
72 SMT Magazine • June 2017 All X-ray inspection systems (2D and 3D) rely on absorption contrast imaging, where the contrast is generated by the stopping power of the sample. As such, higher density and/or thicker regions of a sample produce darker re- gions in a grayscale image. X-ray inspection sys- tems use this method to image features such as wire bonds and ball grid arrays down to a fea- ture recognition size of 100 nm (0.1 µm). While contrast imaging is a very powerful and widely used technique, there is significant- ly more information present within the X-ray beam, which, until now, has not been exploit- ed in electronics inspection. Instead of simply measuring the total absorption of the X-ray beam, a physical structure known as a multi ab- sorption plate (MAP) can be placed in the beam path. This, coupled with machine learning al- gorithms, enables material type and thickness information to be acquired alongside the stan- dard grayscale image. Quantitative composition and thickness in- formation can then be used to provide more de- tailed diagnostics in PCB and semiconductor inspection. Applications include, but are not limited to, analysis of solder types, track thick- ness measurements, and conformance to qual- ity standards. An overview of the theory behind the oper- ation of the MAP is given below, as well as some simple examples showing the benefits of using the MAP technology to add additional capabil- ities to 2D X-Ray inspection. MAP technology also has applications in the security, medical and food processing industries. Nordson Dage has partnered with IBEX to bring the benefits of this technology to the electronics inspec- tion, including inspection of PCBs and semi- conductor devices. A range of examples is pre- sented and the authors aim to start active dis- cussions that would generate new needs and ideas for use of the technology for non-destruc- tive inspection of electronics systems and com- ponents. Basic Principles of the Map Technology X-ray Interaction with a Sample X-rays are generated using a tungsten target which produces a continuous spectrum, known 2D X-RAY INSPECTION WITH MATERIALS AND THICKNESS IDENTIFICATION as Bremβtrahlung radiation, as well as charac- teristic peaks at specific energies. Adding a sample such as a printed circuit board into the beam attenuates the X-ray spec- trum in the following way: I(E) = I 0 (E)exp(-µ(E)t) (1) where I 0 (E) and I(E) are the intensities of the X-ray spectra before and after the sample, re- spectively; µ(E) is the material-dependent linear attenuation coefficient; and t is the thickness of the sample. If I 0 (E) and I(E) can be determined, it is pos- sible to extract the parameter µ(E) which relates to the material type and the material thickness. A standard CMOS detector integrates the to- tal energy deposited into each pixel over the us- er-selected integration time. As such, the detec- tor is able to measure total energy deposited per pixel but not the actual energy spectrum. This means that it is not possible to decouple the material and thickness terms in equation (1), meaning that thin, high density materials are indistinguishable from thick, low density mate- rials in a single 2D projection. The Multi Absorption Plate (MAP) acts like a complex color filter for the X-rays by impos- ing a repeating modulation to the X-ray beam over a few neighboring pixels. This modulation results in a variation in the energy distribution of the X-ray beam incident on the neighboring pixels in a way that enables unique materials in- formation to be obtained. Extracting Materials Information— Simple Examples Figure 1 shows the word IBEX constructed using a series of copper tiles with a background made up of foils of silver. The thicknesses of the foils have been selected in order to randomize the grayscale intensity transmitted by the tiles and hence obscure the word IBEX in a standard X-ray image. Analyzing the image with a MAP in place recovers the materials information. The intensities of the coloring in the bottom image of Figure 1 reflect the relative thicknesses of the tiles. This materials information can be displayed to the user in terms of a basic mask identify-