Acoustic micro imaging uses a moving transducer that pulses ultrasound into materials and receives the return echoes from material interfaces. Images made from the echo signals may show anomalies such as delaminations or other cracks, since gaps send back stronger echoes than well-bonded interfaces.
In manufacturing that involves bonded layers of polymers, ceramics, or metals, engineers often check bond quality by making images of internal interfaces. They may also image a single layer between interfaces to search for voids or cracks within it. When destructive physical analysis is planned, the acoustic image shows exactly where to section the sample.
Imaging a Multi-layer Composite
In an acoustic micro imaging system, the ultrasonic transducer and other hardware provide the reflection-mode or transmission-mode data about the sample being examined, but it is the software that does the work of manipulating that data to solve specific problems.
In the case reported here, acoustic micro imaging was used to determine whether a particular laminated composite material could withstand the various stresses that it would encounter in service. The material, imaged at Sonoscan’s headquarters and applications laboratory in Elk Grove Village, IL, was a 4.9 mm thick graphite epoxy composite having 27 layers of fine fibers laid down at 0°, 90°, +45° and -45°.
For this application, a small hole was drilled vertically through the sample, and the hole was used to mill a long narrow channel. The acoustic micro imaging system could then gather data concerning the internal damage caused by these two operations.
Acoustic Micro Imaging Hardware
The imaging system’s transducer raster-scans one flat surface of the sample while pulsing ultrasound ranging in frequency from 5 MHz to 400 MHz into the sample and receiving the return echoes. The speed of ultrasound through most production materials is so high that the pulse-echo function can be carried out at each of several thousand x-y coordinates per second as the transducer moves across the sample.
The transducer itself contains two primary elements: a piezoelectric crystal that generates the pulses and collects the incoming echoes, and a spherical lens that focuses the pulse. As the transducer scans the sample, it is coupled to the top surface of the sample by water or another fluid, since ultrasound at these frequencies is propagated poorly or not at all through air.
In samples that consist of layers of solid materials, pulsed ultrasound is reflected by the interfaces between different materials (see Figure 1). The essentially flat material interfaces in this sample are the air-filled delaminations caused by the destructive test. All gaps (delaminations, voids, cracks) reflect virtually 100% of the ultrasonic pulse, unlike solid-to-solid interfaces, where a portion of the pulse crosses the gap and travels deeper (see gray arrow in Figure 1). In the acoustic images of this study’s particular composite material (see Figure 2), epoxy-graphite interfaces are represented by gray (reflections of modest amplitude), and gaps are represented by white (reflections of very high amplitude).
Software Controls the Imaging Process
Figure 2 is an acoustic image known as a C-Mode image, produced by raster-scanning a transducer pulsing 30 MHz ultrasound over the surface of the composite sample and collecting echoes from a depth of interest. Three features are immediately evident:
- The cross-hatch pattern of the graphite fibers is visible.
- The dark diagonal line at center is the slit milled into the composite.
- The white features adjacent to the slit are delaminations between the layers of the composite.
This image does not encompass the full thickness of the composite. In this sample, echoes can be reflected from each of the 27 layers, which would produce a very complex image. For this reason, a C-Mode image is typically “gated” on a narrow time window — only echoes arriving within that window are used to make the image. Here the gated depth is the top few layers of the composite.
Software Takes Over
The ultrasonic echoes in Figure 1 travel through the lens of the transducer and into the piezoelectric element, and the mechanical waves are converted into RF electrical signals. These analog signals are then converted into digital signals; at this point software takes over from hardware. Each signal is first analyzed to determine its polarity — positive if the interface was from lower to higher acoustic impedance, and negative if the opposite.
From this point there are two general functions that software can carry out with the millions of incoming signals. First, it can classify and sort them to make one or more of the many types of acoustic images. Second, it can analyze the results of classifying and sorting.
Software operations begin by assigning a gray-scale value to some attribute of the signal — often its amplitude, but for some purposes, its frequency or location or polarity. At a single x-y coordinate within a single gate, not one but many echoes are returned. Suppose, for example, that a 100 MHz transducer is being used. The ultrasound in a single pulse from this transducer will actually contain ultrasonic frequencies from perhaps 70 to 120 MHz. In producing a C-Mode acoustic image, software selects the single echo at that x-y location that has the highest amplitude, and then assigns to that echo a gray-sale value between 1 and 256. This value can then, if desired, be converted into a color by any one of numerous color maps that assign colors to gray-scale values. The other signals from this x-y coordinate are discarded. White areas in Figure 2, for example, have the highest amplitude and indicate gaps. Black areas have very low amplitude, or returned no signal at all. Many areas, especially among the fibers, are some shade of gray.
One newly developed technique deviates sharply from C-Mode imaging by discarding both amplitude and polarity of echoes. Instead software assigns grayscale values based only on the echo arrival time, to measure the precise distance from the transducer to the surface of the part at each x-y coordinate. The result is a contour map of the surface (see Figure 3). Magenta indicates the highest points on the surface, while green indicates the lowest points. The color map in Figure 3 shows the relative altitude of each coordinate. The area around the drilled hole is raised, a feature that can also be seen in the cross-sections at the right and bottom.
In some samples, it is important to seek out very subtle features at the interface between two materials. The feature of interest might be, for example, greater or lesser degrees of bonding at various spots on the interface. Each x-y coordinate at the interface will reflect ultrasound at many different frequencies, as mentioned above. In C-Mode imaging, only the single highest-amplitude echo is used, and the others (there may be 20 or 30) are ignored. One type of software regime accepts all of the echoes of all frequencies from an x-y location and produces a series of planar images, one for each frequency. The process is known as frequency-domain imaging, to distinguish it from time-domain (C-Mode) imaging, where signals are classified by their arrival time. If the range of frequencies is from 80 to 115 MHz, software may produce twenty or so acoustic images, each providing a different view of the same interface at a single frequency. A feature that is absent or ambiguous at 92 MHz may be crisp and sharp at 87 MHz (see Figure 4).
For some samples the most useful type of acoustic image is a non-destructive cross section. Visually, this acoustic image is the equivalent of viewing a physical cross section, but is done without destroying the part. Often it is preceded by a planar C-Mode acoustic image. The planar images display the whole area of the part, which might be a ceramic substrate containing internal traces. The planar image can locate in x and y specific anomalies or defects that can then be non-destructively cross sectioned.
Scanning proceeds along a single line in the planar image that marks the sectioning plane that intersects the anomaly or defect. The transducer scans back and forth along this line, going deeper at each pass. The return echoes are collected by the transducer for processing. The key parameters are x-y location, elapsed time, amplitude, and polarity. Software assigns gray-scale values and arranges the resulting pixels into a display showing the cross-section.
The non-destructive cross section and the planar image of the composite are shown in Figure 5. The sectioning plane passes vertically through damaged areas but not through the milled slit. In the cross section at bottom, irregular white features form an inverted U shape; these are delaminations and cracks caused by drilling and milling. The U shape seen here is generally considered acceptable because delaminations travel along fiber lengths where the material is strongest. The smaller, more regularly arranged features in the top layers are minor delaminations, probably caused by lack of wetting between the fiber and the epoxy.
There are numerous other modes in which hardware can collect and software can manipulate acoustic signals to solve specific problems, such as displaying the 3-dimensional structure of an internal feature like a crack, or preserving the entire acoustic content of a sample to permit comparison to its original condition after failure in testing or in service. These modes, most of which were developed and patented by Sonoscan, have greatly broadened the scope of problems that can be solved.
This article was written by Tom Adams, consultant at Sonoscan, Inc. (Elk Grove Village, IL). For more information, Click Here .