To obtain volumetric information on the chemical composition inside objects, a full-field micro X-ray fluorescence (microXRF) imaging technique has been implemented. Opposed to the traditional point-by-point methods using confocal imaging, this new method collects complete 2D XRF projections of 3D objects without any scanning. And in contrast with EDX surface mapping in an SEM, having complete information from the entire volume allows creating 3D XRF images non-destructively by tomographic reconstruction.

Introduction

The most common approach to microXRF imaging is the confocal approach using a collimated primary X-ray beam and a photon counting detector with a focused field of view [1, 2]. To create a 2D or 3D map of the emitted fluorescence X-rays, the confocal volume is shifted in the specimen by moving the sample, i.e. the elemental image is built up point-by-point. This method has the advantage that the measurements directly yield an image and no image reconstruction is required. However, building up a high-resolution 3D image pixel-by-pixel takes a very long time and requires very intensive excitation, like synchrotron radiation, to get a reasonable photon counting rate at every local position.

Alternatively 3D imaging can also be achieved using tomography. This technique is based on the acquisition of 2D angular projections (also called views) of an object. When a sufficient number of angular views over all directions are obtained, a 3D image of the entire volume of the specimen can be computed using various image reconstruction algorithms.

Full-field Micro XRF Imaging

In full-field micro-XRF imaging the complete sample is irradiated with primary X-ray sources to create fluorescence. Hence there is no need for X-ray focusing optics to produce a narrow beam. The fluorescence X-rays are captured by a CCD after passing through a narrow pinhole. Just like in the early ‘camera obscura' photographic instruments, the pinhole creates an upside down projection of the fluorescence X-rays emitted by the object (fig.1). Different angular projections can be obtained by rotating the specimen.

Pinhole optics also keep all parts of the object "in focus" and avoid losing information out of the focal plane which is the case with any other optics, such as Fresnel lenses or capillary waveguides.

The diameter of the pinhole determines the spatial resolution and efficiency. Smaller pinholes result in a higher resolution but they also reduce the number of X-ray photons collected by the CCD. The CCD is operating in photon-counting mode by choosing the exposure time such that the probability for more than one X-ray photon hitting the same pixel is negligible within one acquisition frame. In this case the signal produced by a given pixel is proportional to the energy of the fluorescence X-ray photon impinging on it. All fluorescence photons that were detected are read out and stored in a list-mode format, i.e. for each individual photon we store the position where it was detected together with its energy. This format retains a maximum amount of information and allows the most flexible post-processing. An XRF list-mode data set is built up by repeatedly acquiring frames until sufficient statistics have been reached.

During a scan the energy spectrum of the fluorescence X-rays recorded in all projections is shown and the elements present are identified. After scanning, the operator can select from which elements he wants to create a set of 2D XRF projections, which are then used to reconstruct 3D elemental images.

Combined Micro XRF-Micro CT System with 3D Capability

The principle of full-field XRF imaging was implemented in the new Sky‑Scan2140 system [3], which combines a microXRF and a microCT imaging modality. Two powerful X-ray sources (Mo anodes) operating at 50 kV/200W irradiate the specimen uniformly. The primary X-ray spectrum is filtered to reduce the continuous background due to backscattering of the Bremsstrahlung on the specimen. The spectral peak due to the elastic backscattering of the primary Mo characteristic X-rays is used to auto-calibrate the energy scale.

A pinhole changer allows the user to select from three pinhole sizes. The smallest pinhole with a diameter of 30 µm yields the best resolution (40 micron) while the largest 100 µm diameter pinhole allows the shortest scan times (10x faster).

The fluorescence X-rays passing through the pinhole are detected by a cooled 1Kx1K deep-depleted CCD with an energy resolution of 170 eV at 5.9 keV. The imaging geometry has a 13.3x13.3 mm FOV in the object plane with a 13 µm image pixel size.

The system also includes the possibility for microtomography (microCT) of the same specimen using a separate micro-focus X-ray tube and a scintillator-CCD combination. The availability of a microCT modality (12 µm resolution) allows co-registering morphological information with the 3D chemical maps obtained by the XRF scan. Since the 3D microCT images represent the local attenuation coefficients in the sample, this information can also be used to correct for object self absorption of the primary and fluorescence X-rays during the image reconstruction.

Full-field MicroXRF Application ­Examples

Full-field microXRF can be used to examine larger objects at a high resolution. Figure 2 shows a screenshot of the elemental image, X-ray absorption image, combined image and energy spectrum obtained from an electronic board (e.g. for ROHS analysis or quality control). The entire object is imaged with a 60-micron resolution in a few hours. This resolution is sufficient to detect and visualize the thin 25-micron Au bonding wires inside the chip package.

Elemental imaging using microXRF can also be used in geology or mining. Figure 3a shows an image of a fine powder sample (2.5x9 mm) prepared with epoxy resin. The particle sizes ranged from 40 to 80 micron. The object contains local traces of heavy elements, like Pt, with concentration from 50 to 100 ppm. Elements present at trace level are detected using local spectral analysis. The regions where they are found can be indicated on the XRF image for verification and further analysis (fig. 3b and c).

3D Application Example

If the main matrix of a specimen consists of relatively X-ray transparent or porous materials, it becomes possible to perform 3D microXRF imaging. This was demonstrated by mapping the strontium distribution in a rat vertebra. The vertebra specimen (6.7 mm x 6.2 mm x 2.8 mm) was taken from a rat used in a preclinical study on osteoporosis treatment in which the animal received a daily dose of strontium-containing drugs during a certain time. The specimen was scanned using both microCT and microXRF modalities of the SkyScan 2140. Figure 4a shows a 3D rendered microCT image. The strontium distribution is illustrated in the volume rendered, fused microCT-microXRF image (fig. 4b) where the strontium concentration is displayed as a color in a blue-red scale

Conclusions

Developed and implemented in a commercially available SkyScan2140 instrument, the full-field microXRF imaging principle has shown the capability to provide 2D and 3D elemental maps of different objects at spatial resolutions down to 40 micron and detail detectability down to 5-25 microns. This 3D chemical information can be co-registered with 3D micromorphology obtained from the microCT set-up integrated into the same instrument.

References
[1] B.M. Patterson et al.: X-ray Spectrom. 39, 184-190 (2010)
[2] U. Fittschen et al.: Anal Bioanal chem 400, 1743-1750 (2011)
[3] http://www.skyscan.be/products/2140.htm

Authors
Dr. Peter Bruyndonckx (corresponding author)
Dr. Alexander Sasov
Dr. Xuan Liu
Dr. Bart Pauwels
Skyscan
Kontich, Belgium