A new portable macro X‐ray fluorescence scanner has been specifically designed for in situ, real‐time elemental mapping of large painted surfaces. This system allows scanning 80 × 80 × 20 cm3 along the X, Z, and Y directions, respectively, with adaptive beam size at the energy of the Rh Ka‐line. The detection system consists of a 50 mm2 active area detector coupled to a CUBE pre‐amplifier and to the DANTE digital pulse processor (DPP) with adaptive shaping time. The system is controlled with a custom software including a graphical user interface (GUI) programmed in Python for real‐time control of the stage, DPP, and camera of the scanner. This system allows considering new ways of sampling the object surface than the usual raster scanning in serpentine as well as a live elaboration of X‐ray data; technical details and performances of the scanner are presented in this paper together with an example of its application to investigate painted surface, illustrating the value of the developed instrument.

No abstract

Previous work [1] demonstrated that pulse-by-pulse adaptive digital filtering improves the precision of X-ray quantitative analysis for a given sample electron dose, with no loss of accuracy compared to conventional pulse processing. The improvement stems from better energy resolution compared to short fixed digital filtering for the same throughput. The gain in precision is greatest for small peaks below about 5 keV near to or overlapping with larger peaks, such as Al in NIST K412 glass, which has favorable implications for analysis at low accelerating voltages.That work used only a small set of digital filters for ease of generating standards. The time intervals between X-ray arrivals are randomly unequal. Asymmetric processing, as described by Koeman [2], reduces the noise component of resolution by allowing different integration times before and after an Xray arrives. It promises further gains in analytical precision, but it also requires more sophisticated peak shape modeling to derive appropriate fitting spectra from standards. Table 1 shows the resolution gained by asymmetrical filtering. The matrix diagonal is conventional symmetric filtering (shaded). Entries in bold are the filters used for last year's results. The axes are leading and trailing integration times in nS. Entries are Al-K resolution in an Al 2 O 3 sample at 7 nA, 319 kcps input rate, 196 kcps throughput rate. Figure 1 is a resolution contour plot of these data.The matrix is symmetric within a few tenths of an eV, which is expected since resolution should not depend on arrival order for each pair of intervals. For each diagonal entry, resolution improves as we move up or to the right (increasing the integration time toward the wider interval). Table 2 gives the probability of occurrence for each of the cells in Table 1 at an input count rate of 320 kcps. The interval pair probability depends only on the input rate, regardless of the composition of the sample. At 320 kcps ICR and 38% dead time, nearly 27% of the X-rays fall in the upper right matrix cell with six times the integration period and 30 eV better resolution relative to the shortest filter, while 77% are measured with 2.4 uS integration on one side (top row or right column).The table sums to 1, and the lower-left entries are so small because the time bins are of unequal size. The rectangle with each diagonal entry at a lower left corner sums to the fraction of X-rays measured with that diagonal entry's resolution or better. Since the input rates generated from the unknown and the standards will be different in general, the procedure for generating fitting standards uses the noise components derived from Table 1 with the probabilities of Table 2, which can be readily computed for any input count rate.

Highly luminescent thin films of zinc tungstate (ZT) have been deposited on top of conventional scintillators (Yttrium Aluminum Perovskite, Yttrium Aluminum Garnet) for electron detection in order to replace the need for a top conducting layer, such as indium tin oxide (ITO) or aluminum, which is non-scintillating and electron absorbing. Such conventional conducting layers serve the single purpose of eliminating electrical charge build-up on the scintillator. The ZT film also eliminates charging, which has been verified by measuring the Duane–Hunt limit and electron emission versus accelerating voltage. The luminescent nature of the ZT film ensures effective detection of low energy electrons from the very top surface of the structure ZT/scintillator, which we call “first-surfacescintillator”. The cathodoluminescence has been measured directly with a photodetector and spectrally resolved at different accelerating voltages. All results demonstrate the extended range of operation of the first-surface scintillator, while the conventional scintillators with a top ITO layer decline below 5 kV and have practically no output below 2 kV. Scintillators of different types were integrated in a detection system for backscattered electrons (BSE). The quality of the image at high accelerating voltages is comparable with the conventional scintillator and commercial BSE detector, while the image quality at 1 kV from the first-surface scintillator is superior.