UPGRADE OF THE EXTERNAL BEAMLINE AT THE MICROANALYTICAL CENTER OF THE JOŽEF STEFAN INSTITUTE

The existing external beamline at the Microanalytical Center of the Jožef Stefan Institute (JSI) in Ljubljana has been currently upgraded in order to extend the capabilities towards in-air PIXE imaging across relatively large sample surface area (  1 cm 2 ) with lateral resolution of sub 100 µm. For that purpose, a doublet of magnetic quadrupole lenses was installed on the beamline together with the newly designed exit nozzle with a 200 nm Si 3 N 4 exit window resulting finally in  50×50  m 2 lateral resolution. Additionally, an in-vacuum beam chopper was installed before the exit nozzle to measure the proton dose deposited on the sample. The beam characterization results are presented together with the first application from the field of biology, namely quantitative concentration maps of a longitudinal section of a whole thick wheat grain.


I. INTRODUCTION
Because of its high sensitivity, non-destructive character, and capability to perform full quantitative analysis, Particle Induced X-ray Emission (PIXE) technique is commonly used to analyze multi-elemental composition of the studied samples from various research disciplines. Using magnetic quadruple lenses, a MeV proton beam can be effectively focused down to  1 m 2 size and scanned across the sample using the scanning coils to record quantitative concentration maps with micrometer lateral resolution [1]. At the Microanalytical Center of the Jožef Stefan Institute (JSI) in Ljubljana hosting a 2 MV Tandem accelerator, micro-PIXE analysis is regularly performed at the microprobe beamline based on the triplet quadrupole lens system from Oxford Microbeams (OM-150) [1] and applied successfully to biomedical applications [2][3][4]. While in these applications we are usually looking for maximal lateral resolution and the size of the scanning area is usually of the order of mm 2 and lower, we have been recently faced with the requirements to increase significantly the scanning area to allow PIXE imaging of larger objects (few cm 2 ) at the expense of moderate sub 100 m spatial resolution, which would complement detailed micro-PIXE analysis. For that purpose, we have decided to upgrade the external beamline which has been so far mainly used for in-air PIXE archaeometry studies employing point analysis with a broad ( 1 mm) proton beam [5][6][7][8]. The main goal was to achieve inair beam focusing down to few tens of m, which would allow us to perform in-air PIXE analysis of larger biological objects with very little sample preparation.
The manuscript presents the main technical specifications of this upgrade and the results of characterization measurements providing actual experimental lateral resolution. Finally, the new setup is used to analyze a longitudinal section of a whole thick wheat grain illustrating the capabilities of this new setup for in-air PIXE imaging of large objects and surface areas with sub 100 µm lateral resolution.

II. EXPERIMENTAL SETUP
The external proton beam is situated at the -30 o exit port of switching magnet of the 2 MV Tandem ion accelerator. During the current upgrade the beamline was equipped with water-cooled object slits positioned at the beginning of the beamline after the switching magnet. Next, a second set of collimator slits was installed further down in front of the quadrupole doublet (manufactured by Drusch et Cie), which is used to focus the beam on the target. The distance between the object slits and quadrupole doublet is 550 cm while the distance between the quadrupole doublet and the exit window is 74 cm.
Before the exit nozzle an in-beam chopper device [9,10] has been installed. The rotating chopper (gold plated carbon blades) periodically intersects the beam in vacuum just before it exits through the exit window into the ambient atmosphere. The spectrum of the backscattered protons is recorded simultaneously with the PIXE spectra and the integrated Au Rutherford Backscattering Spectrometry (RBS) peak is used to measure the proton dose, which serves for normalization. Finally, the exit nozzle terminating the beamline was redesigned to optimize the detection geometry and minimize the sampleexit window working distance. For this purpose, it accommodates a 3×3 mm 2 Si frame with an ultrathin 200 nm Si3N4 window necessary to diminish the lateral beam scattering in the foil. The redesigned exit nozzle also allows to use smaller exit window-sample distance, which is crucial for achieving good lateral resolution. The sample is placed on the motorized computer controlled XYZ stage (manufactured by ISEL automation), which is used to move and position the sample with steps of 12 m. The relative position of the beam on the sample window is adjusted using a microscope. The microscope is used also to adjust properly the sample distance from the exit window by positioning the sample into the focal plane of the microscope. The end station is equipped with two Si(Li) detectors for PIXE analysis positioned at 135° with respect to the primary beam direction. The Si(Li) detector (manufactured by PGT, active area 30 mm 2 , 7.5 m thin Be window, 138 eV FWHM resolution at 5.9 keV) positioned above the beam at a distance of 3.2 cm from the sample is used for in air PIXE imaging (label 1 in Figure 2) presented later on in the manuscript. The second PIXE detector (label 6 in Figure 2) was used previously for in-air PIXE archaeometry studies with a broad proton beam [5][6][7][8] and is in fact not applied in current imaging applications. Beside both PIXE detectors, the end station is equipped with an additional high-purity Ge detector (P-type coaxial manufactured by ORTEC) for Particle-Induced Gamma Emission (PIGE) analysis. The schematic layout of the beamline is presented in Figure 1 and the photograph of the exit nozzle and the detector configuration at the beamline end station is shown in

III. CHARACTERIZATION RESULTS
The most important parameter we have first tried to quantify is the actual size of the proton beam spot on the sample. In order to achieve optimal focusing we have first inserted a scintillator in the proton beam and roughly adjusted the magnetic fields of both quadrupoles by observing the shape and size of the fluorescence spot under the microscope. After that, the beam size has been measured using the knifeedge method [11]. In this measurement a metallic blade is moved across the beam in 15 m steps and the X-ray yield is recorded as a function of sample stage position. This is used to perform the fine tuning of the magnetic fields of both quadrupoles required to reach optimal focusing.
Besides the focusing capability of the quadrupole doublet the actual size of the beam on the target is   Another very important part of the upgrade is the beam chopper used to determine the proton dose deposited on the sample, which is necessary to perform quantitative analysis. After the installation we have tested the stability and accuracy of the proton dose measurement by observing the ratio of the Xray yield recorded from a homogeneous Ni metallic target and the Au RBS yield from the chopper. The ratio was found very stable, the values measured within first 5000 s are presented in Figure 5. The standard deviation of the measured yield ratio was found 1.4% which is consistent with the statistical error of the measured RBS yield. In order to finally perform full quantitative PIXE analysis we need to calibrate the proton dose measurement together with the efficiency and solid angle of the X-ray detector. This calibration was done by measuring a set of K X-ray spectra from several mono-elemental homogeneous thick samples.
The absolute calibration factor between the proton dose and the Au RBS chopper yield was determined by analyzing the measured PIXE spectra with the GeoPIXEII software package [13]. This calibration factor was found 3.5 10 -2 nC/count yielding the overall beam chopper sensitivity of approximately 0.3 cts/sec within RBS signal per pA proton current on the sample. Standard deviation of the calibration factor measured for different calibration targets was found 5.4% and this value was assigned as the final experimental uncertainty. After this calibration, absolute quantitative in-air PIXE analysis can be performed without using any reference material. However, just to check all experimental parameters and confirm the reliability of the overall quantification procedure a PIXE spectrum from reference material is recorded before every experimental run. Generally, any external reference standard could be used but we usually try to use a material of composition similar to the studied sample. Finally, the calibration procedure was checked by measuring two standard reference materials from National Institute of Standards and Technology (NIST), NIST 1107 (naval brass) and NIST 1573a (tomato leaves) and analyzing the measured spectra with the GeoPIXE. The measured concentrations are presented in Table 1

III. FIRST APLICATION
Mineral deficiency is a widespread nutritional disorder especially prevalent in developing countries with a cereal dominant diet [14,15]. Mineral nutrients (Mg, S, P, K, Ca, Fe, Zn) are present in cereals but are unevenly distributed across the grains with the highest accumulation in tissues which are generally removed in the grain processing for the human consumption (seed coat, aleurone layer and embryo). Since understanding the localization and accumulation of elements in plants is an important topic in a range of fields, different techniques have been used for analysis of biological samples [16].
Elemental distribution patterns among different grain genotypes can be distinguished using micro-PIXE technique which provides precise localization of minerals and their accumulation levels. Therefore, the investigation of mineral co-localization and nutrient transport in different grain tissues is crucial for nutritional improvement of grain crops and selection of the appropriate genotypes. While the elemental distribution in wheat grain tissues was previously examined by in-vacuum micro-PIXE [17,18], micro-PIXE in ambient atmosphere enables analysis of thick samples and whole plant organs (leaf, root, grain etc.) and hydrated biological specimens with minimal sample preparation.
The first application of the updated external beam was the micro-PIXE mapping of a thick longitudinal section of a wheat grain in ambient environment. The morphology of the grain as seen under the optical microscope is presented in Figure 6. A 3 MeV proton beam was used in these measurements, the sample distance from the exit nozzle was 3 mm and the overall scanning area was 3.3 × 7.8 mm 2 . The X-ray detector was positioned 3 cm from the target to attain a solid angle of about 0.022 sr; in order to stop scattered proton to reach the detector crystal it was equipped with a 100 µm kapton absorber. Mapping was performed by mechanical scanning using a dedicated home-made software written in the LabVIEW programming environment which controls the positioning stage and reads consecutive PIXE spectra from the XiA DPP used to process the signal from X-ray detectors. The acquisition time for single point was 3 s. During PIXE mapping, RBS spectra from the chopper was simultaneously recorded in parallel with PIXE spectra and used to determine quantitatively final concentrations and also to normalize separate points in the map due to possible fluctuations of the proton beam intensity. Since the RBS signal from the chopper is too weak to provide statistically high enough The final quantitative elemental distribution maps are presented in Figure 7. Due to in air X-ray absorption, some of the elements which are common in plant tissues (Na, Mg, P, S), were below the detection limit of in-air micro-PIXE. The elemental distribution maps indicate that K is predominant element in the grain, mainly present in aleurone layer, embryo and scutellum. Fe and Zn seem to colocalize in the aleurone layer and embryo with varying concentrations. Mn was predominantly measured along the crease of the grain while Ca was present in the whole grain with higher concentrations measured around the pigment strand and in 'hotspots' within the endosperm. Finally, we have also calculated average section-specific concentrations for four 5×5 pixel sub regions (300 x 300 µm 2 ) using GeoPIXE and these are tabulated in Table 3. Next, we zoomed in on embryo region of the grain and performed another more detailed scan with 20 m step size over smaller area (2.4 × 1.7 mm 2 ). Mapping with a smaller step size allowed for a more precise localization of higher concentrations of specific elements in the region. Generally, the final lateral resolution is limited by the size of the proton beam and this defines the smallest scanning step used in the PIXE imaging. However, in case of smaller scan area we can apply also smaller step since approximately 100×100 pixels are required to build an image. The resolution is not enhanced but we acquire larger statistics over smaller area yielding more accurate average concentrations in specific sub regions.  The elemental distribution maps gave us an overall image of localization of minerals in the whole grain and, with lateral resolution of 50 m, we were able to distinguish different regions of the wheat grain such as aleurone layer which has an average thickness of 30-70 m [19]. The average concentrations of measured elements are in agreement with those previously reported [17,18]

IV. CONCLUSIONS
In