Silicon Drift Detectors (SDD) [1] are rapidly replacing Si(Li) detectors for EDX microanalysis in SEM, but have yet to have an impact in the S/TEM world. Main reason for this difference is the low count rate created by thin S/TEM samples compared to the bulk samples in SEM . These low count rates make EDX mapping a very slow process in S/TEM. However, the recent introduction of higher brightness electron sources [2] and probe Cs-correctors has led to significantly increased beam currents in small electron probes and, potentially, to higher EDX count rates. Since a key advantage of the SDD is the high count rate capability, the throughput improvement compared to the Si(Li) detectors will be considerable in these new instruments. Compared to SEM, the smaller excited volumes obtained with the atomic-scale probes in the new S/TEM instruments can lead to radiation damage of beam-sensitive materials before the analysis is completed. Therefore S/TEM microanalysis needs not only the higher count rate capability, but also higher collection efficiency of the X-rays generated, in order to reduce the dose on the sample. In this paper we present a new prototype EDX detector system for an FEI 200kV TEM/STEM, in which FEI has integrated a detector system consisting of multiple SDDs, placed symmetrically around the electron beam axis in the objective lens chamber without affecting the S/TEM resolution. The SDDs with a total active area of 120 mm 2 were designed by PN Sensor to fit into the FEI design to achieve a quantum leap in solid angle of collection compared to previous designs in S/TEMs. The SDDs are cooled to achieve the optimum energy resolution, typically below 130 eV. The windowless design allows for better sensitivity for light-element detection than conventional thin-window detectors. The specially designed front-end electronics and ultra fast multi-channel pulse processor are provided by Bruker AXS MA in collaboration with FEI. The processor is capable of fast mapping with pixel dwell times down to a few microseconds and >100 kcps count rates per channel. Compared to currently available Si(Li) detectors the anticipated count rates will be an order of magnitude higher with the new detector. Additionally the new high brightness gun of FEI (X-FEG) [2] increases the brightness of the electron source compared to conventional Schottky sources, leading to a further increase in count rate, and an equivalent significant decrease in mapping time at the same spatial resolution. This improvement is illustrated in Fig. 1 where the relative minimum detectable mass MDM ~ (t.P.P/B) -1/2 (t=analysis time, P=elemental peak counts, P/B = peak-to-background ratio) [3] is shown for conventional and new EDX detector count rates at the same spatial resolution. Fig.1 also compares the MDM with EELS and, for the specific case of strontium titanate, shows that the new EDX detector is expected to be more sensitive than EELS. Further results will be reported at the conference.

Extended abstract of a paper presented at Microscopy and Microanalysis 2010 in Portland, Oregon, USA, August 1 – August 5, 2010.

Lorentz microscopy is used to image magnetic materials in field-free conditions to reveal the magnetic structure on a nanometer scale. Insight into magnetic properties on this scale help to understand and optimize the magnetic properties of functional magnetic material, such as sensors (GMR) and magnetic storage material. In conventional S/TEM imaging the magnetic structure is destroyed due to the high magnetic fields at the sample area. In the current instruments the resolution in field-free imaging is limited by the performance of the weak Lorentz lens with its high spherical aberration (Cs~10m) to 2nm point resolution at 300kV acceleration voltage. In this contribution an image Cs-corrector on a Titan80-300 column is used to reduce the Cs of the Lorentz lens by 3 orders of magnitudes from 8m to~10mm at 300kV acceleration voltage. A special optical set-up is used to compensate the high spherical aberration of the Lorentz lens using the hexapole elements of the image Cs-corrector. The chromatic aberration of the special set-up has been measured to~93mm by determining the focus change, when the energy of the electrons has been altered. These optical parameter improvements enhanced the resolution in field-free imaging to the Sub-Nanometer level and an information limit of 0.7nm has been obtained. (figure 1). In addition to the compensation of the optical parameters using the image Cs-corrector the stability of the Lorentz lens has been improved compared to conventional Lorentz lenses. The flexibility of the Titan80-300 is not impaired by this solution and the column with its Cscorrector can be used in Cs-corrected HR-TEM, HR-STEM mode at different acceleration voltages. Switching between the normal Cs-corrected objective lens on-operation mode (non field-free) and the special Cs-corrected Lorentz mode (field-free mode) can be achieved as a push-button function. The same Zemlin tilt-tableau method to measure and correct for aberrations used in objective lens on`mode can be used in field-free Lorentz mode. This makes the method easy to use and reliable in performance. (figure 2) An important performance measure in Lorentz microscopy is not only the optical performance, but the smallest magnetic field at the specimen area in the Lorentz imaging mode. Measurements to determine the magnetic field strength result in field strengths smaller than 2 Oe (figure 3) at the specimen area and the dependence of the field in respect to the Lorentz lens, objective lens and Mini condenser lens are presented. These measurements allow the user to apply magnetic fields to the sample and to determine changes of the magnetic structure in a reproducible and quantitative way.The CEOS authors are thankful to Prof. U. Kaiser for the opportunity to perform first alignment experiments using a Titan80-300 with image Cs-corrector installed at University of Ulm.

We demonstrate chemical mapping at the atomic level using energy dispersive x-ray spectroscopy (EDS) in Cs-corrected scanning transmission electron microscopy (STEM). The combination of the increase in current in an atomic sized probe by Cs-correction and the increase in sensitivity of the Super-X detector [1] allows acquisition of such results within minutes and at high sampling rates. The high speed of the software and excellent S/N ratio of the EDS detector enables ultra fast mapping at > 10.000 spectra per second and minimizes artifacts like sample drift and beam damage.To benchmark the technology we show results at the atomic level using different materials like perovskites and semiconductors. For the first time atomic chemical information on interfaces is obtained using EDS spectroscopy (figure 1). The sampling rate, field of view (200×200 pixels = 5×5 nm) and signal strength presented is much higher when compared to previously reported results using conventional EDS detectors [2], [3]. The dependence of performance on experimental parameters and the absorption edges (L and K edges) on different elements is discussed.Theoretical Bloch wave simulations of a PTO-STO interface are shown in figure 2. The simulations for EDS mapping assume that the cross section for x-ray emission (when filling a K or L shell hole) is proportional to the total cross section for K or L shell ionization for a detector which samples the full solid angle and all possible energy losses above the ionization threshold. This ensures that the ionization interaction is as localized as possible and that, like high-angle annular dark field (HAADF) imaging, EDS is an incoherent form of imaging and directly interpretable. The simulations in figure 2 for an interface with no diffusion across the boundary do not exhibit the same intermixing of the Sr signal at the Pb-Sr interface seen in experiment (figure 1). Assuming that the EDS imaging is directly interpretable, this indicates that there are both Pb and Sr atoms in the interface. Further simulations are currently underway investigating this.Finally, as the number of elements in the periodic table accessible for chemical mapping by EDS is much higher than for EELS, EDS mapping provides chemical information at the atomic level, on complex multiphase materials, which was not accessible via EELS in the past. Lett. 104, 196101 (2010) 598

No abstract