Gamma Ray Backscatter Radiometry

Everyone today is familiar with transmission mode X-ray radiography, in which X-rays are used to cast shadow images of teeth, bones, and luggage in airports. Less familiar is reflection mode tomography, in which X-rays or gamma rays are scattered back to a detector array for producing an image. Since the X-ray response to scattering differs from that of absorption and transmission, reflected images can reveal details that shadow images miss. In addition, reflection mode only requires access to one surface. This can have advantages where the high density bulk of an object obliterates all transmission shadows but a volume of interest is near one surface.

About a decade ago, we investigated some of the potential of closely related gamma ray backscattering inspection in a scanning mode for a range of applications including polymer composite materials, silica tiles, and metal welds, possibly even for microscopy. Related wide aperature configurations have used gamma ray backscatter radiometry (GRBR) to monitor surface coatings such as paint, enamel, and hardfacings.  The new approach combined the scanning geometry of near field microscopy with carefully collimated photon sources and detectors. Such collimation only views a selected volume under a surface, potentially resolving microscopic features. For rough surfaces or inhomogeneous materials, a dual detector system can eliminate spurious responses to surface features.  Energy resolution by sodium iodide spectrometry further refines system responses to changes in density in the sensitive volume.

Collimation and energy resolution together promise greater sensitivity to delaminations, cracks, voids, and inclusions for GRBR in comparison with related X-ray backscatter tomography.  A computer code adapted to model GRBR performance for realistic material configurations and collimator designs made predictions verified in later bench-scale tests. The verified code allowed selection of optimal gamma ray energies for various applications. For inspecting through up to 2.5 inches of polymer type material, optimal response energy was found to be below 0.200 MeV, which is conveniently low for shielding and source handling.
  The distinguishing design feature of GRBR is the use of converging plate collimators, constructed of tungsten foils.  Scattered gamma rays are isolated from the rest of the pulse height spectrum by selecting a specific energy region of interest based on Compton's scattering formula.  A primary detector is collimated to view in depth, while a less finely collimated reference detector eliminates response to surface features by accepting scattered events from a must larger volume but with a concurrent surface area of intersection.

     Above, basic converging plate collimator configuration using thin tunsten plates.

     Left, arrangement of concurrent collimated detectors and gamma ray source.

In several years of bench testing, digitized signals were counted individually and compared to determine relative intensity from two detectors, but a rapidly-scanning system could operate in "current mode" with readily available electronic comparator circuitry.  A (256 channel) multichannel analyzer system was used to create the pulse height spectrum for discrete events from each detector.  Even simpler pulse height gating or capacitance-discharge current mode would be suitable for a prototype hand-held instrument.
  Experimental mapping of a reference detector collimation made by moving a 30 microcurie Co-57 source in front of the collimator slits.

  Experimental mapping of primary detector collimator with vanes that converge to touching, i.e., a completely closed collimator. Nonvanishing acceptance is due to transmission through corners at the front edge of each vane. Thus, this is a measure of leakage and the limit of resolution.

Calculated counting rates for two-inch thick silica tile (Space Shuttle protective covering).

Both computer predictions and laboratory tests confirmed feasibility of constructing a reliable scanning instrument based on GRBR for inspection of composite materials, welds, and various other applications.  Effective designs all included a low-energy gamma radionuclide source (exact energy optimized for the different applications), sodium iodide detectors, single-channel analyzers, comparator circuitry, and small collimators weighing about one pound.

    Pattern of scattering probabilities for energetic photons.

The computer code predicted sensitivity to hidden voids in polymer composites to be optimized for a gamma ray energy range of 0.070 to 0.200 MeV.  With a 200 µCi Co-57 source, sufficient sensitivity was demonstrated in bench tests to easily map delaminations smaller than 0.005 inches through 0.125 inches of simulated material.  Larger sources and tighter collimation can potentially increase sensitivity by an order of magnitude if desired.

Representative backscatter spectrum. The two fluorescence peaks are excited by a combination of the incident primary signal and high energy gamma contamination from outside the system, including cosmic rays. The boxed area indicates the kinematic region on which measurements were based.

Counting statistics, sufficient for one percent resolution in counting rate, were achieved in 15 minutes of data collection for static configurations, using source strengths which do not require special handling.  Results and theoretical estimates indicate that safety should not be a problem, although a radioactive materials license will be required.  Although stronger sources will be required for a rapidly scanning instrument, this should not create problems because the low energy gamma rays are easily shielded and collimated.  NRC approved containers and devices for such sources are readily available for common medical and industrial radiography applications.

   Depth scan of two Lexan sheets, 3.2 mm thick separated by a 1.6 mm gap. The curve represents the response of a solid piece of Lexan. Similar scans easily resolved gaps as small as one micrometer.

Potential advantages of NDI and calculated performance estimates confirmed by tests strongly warrant further bench testing and laboratory development including detailed calibrations of actual defect densities and sizes.  We plan construction of a prototype instrument as a student project with thorough testing, both in laboratory mock-ups and in field settings.  System modifications required for portable hand-held inspection should also be carefully evaluated.

     Scan of a 0.38 mm tantalum wire sandwiched between two 3.2 mm thick aluminum plates.