TEM

TEM - Transmission Electron Microscopy

card

The TEM laboratory, established during 2002, is part of the Centre for Electron Microscopy of the TASC-INFM National Laboratory whose equipment includes the sample preparation facilities, the laboratory for simulating TEM images and spectra and the AFM facility. The Centre for electron Microscopy, (CEM) has the mission to represent a TEM facility for the scientific and industrial community and to explore and to develop new TEM methodologies to be applied to the study of the solid state matter. The latter task has to strongly take into account the location of the CME: inside the TASC laboratory and close to the synchrotron storage ring of ELETTRA.

A JEOL JEM 2010F UHR TEM/STEM is installed in a dedicated laboratory at TASC - MM building. A vibration-insulation foundation for the microscope was built by anchoring a concrete platform directly on the carsic rock with only weak links to the MM building. The TEM laboratory is kept at constant temperature with a drift rate of less than 0.1°C/min, with low noise and minimal turbulence.

TEM instrument has an accelerating voltage of 200 kV and a low spherical aberration coefficient ((0.47±0.01) mm) objective lens yielding a phase contrast resolution - at Scherzer defocus - of 0.19 nm. The electron source is a thermally assisted field emission gun (FEG) with a ZrO/W [100] filament. The FEG is a high-brightness source producing a highly coherent electron probe with a diameter smaller than 0.13 nm and a resolution limit of 0.11 nm in phase contrast. The small probe size yields sub-nanometer resolution in analytical as well as spectroscopic measurements.

FIGURE 1- a) experimental high resolution image in [110] zone axis of one Si atomic layer buried in GaAs; b) simulation of inset a) by using multislice code. The fine details in the experimental image have been successfully simulated by considering Si atomic columns in which 25% of Si atoms are replaced by Ga atoms.

The instrument is currently equipped with an energy-disperse x-ray spectrometer (EDS) with ultra-thin window to enhance detection of light elements (Z > 5). The available scanning transmission electron microscopy (STEM) attachment coupled with EDS can be used to obtain chemical profiles with high spatial resolution. Moreover, coupling between the STEM attachment and the high-angle annular dark field (HAADF) detector is used to obtain Z-contrast imaging. This technique is one of the most powerful evolutions of TEM methods for ultimate resolution. HAADF technique can be used to image a single atomic column down to a theoretical resolution of 0.123 nm and, at the same time, acquire the electron energy loss spectrum from the same single atomic column.

FIGURE 2 - Electron optical configuration as used in HAADF experiments. Energy dispersive x-rays spectra can be acquired by scanning a 0.14 nm electron probe across the region of interest. The HAADF image contrast has strong dependence on the atomic number, in the figure the Ga and As colums can be distinguished due to the different intensities.

fig 3 HAADF image with a resolution 0.126 nm

Methodological studies: CHIRALTEM

The Centre of electron microscopy if also part of the CHIRALTEM project. The aim of the project is to measure the circular dichroism by means of a TEM. Dichroism is the property of certain materials whose photon absorption spectrum depends on the polarisation of the incident radiation. In the case of  X-Ray Magnetic Circular Dichroism (XMCD) the absorption cross section of a ferromagnet or a paramagnet in a magnetic field changes when the helicity of a circularly polarised probing photon is reversed relative to the magnetisation.  Although the similarities between X-ray absorption (XAS) and electron energy loss spectra (EELS) in the transmission electron microscope (TEM) have long been recognised, it was presumed that extending such equivalence to circular dichroism would require a beam of spin polarised electrons. Recently, it was argued on theoretical grounds that this is probably wrong [1]. Within the CHIRALTEM project it has been  demonstrated the first direct experimental proof of magnetic circular dichroism in the TEM by comparing Electron Energy Loss Magnetic Chiral Dichroism (EMCD) with XMCD spectra from the same specimen together with theoretical calculations [2]. The experiment shows that chiral atomic transitions in a specimen are accessible with inelastic electron scattering under particular scattering conditions. This result bears dramatic consequences for the study of magnetism on the nm and sub-nm scale, as EMCD offers the potential of spatial resolutions down to the nanometre scale and provides depth information, overcoming the major limitations of X-ray methods.

[1] C.Hebert ,P.Schattschneider Ultramicroscopy  96(2003) 463
[2] P. Schattschneider, S. Rubino, C. Hébert, J. Rusz, J. Kune, P. Novák, E. Carlino, M. Fabrizioli, G. Panaccione, G. Rossi.
Experimental proof of circular magnetic dichroism in the electron microscope Nature 441, 486-488 (2006)

Figure 1, Scattering geometry. Simplified scattering geometry, drawn in the diffraction plane of the TEM: Bragg diffraction creates a coherent superposition of two incident plane waves (k₀ and k₀’) in the Fe crystal. A phase shift of π/2 is set between the two wave fronts by tilting the incoming beam. The detector (or a contrast aperture) selects q and q’. The dotted circle represents the points for which q is perpendicular to q’. The full circles show the two positions for which also the condition |q| = |q’| is true and indicate the actual experimental setup. As the two positions “+” and “-“ have opposite chirality, EMCD can be detected by simply acquiring spectra at the two positions and taking their difference.

Figure 2 EMCD. Measured Fe L_2,3 edges for 10 nm Fe on GaAs (001) in the two configurations shown in figure 1. The difference (magnified by a factor of 5 in the figure) is 0.07 for the measured spectra and 0.32 for the simulations. The r.m.s of the noise is ±8% of the difference. The experiment was performed at the Technical University of Vienna by the group of P. Schattschneider

^.

Methodological studies: Picometer range resolution by coherent diffraction imaging in transmission electron microscopy


The in-depth study of nanoscale matter represents a burning issue in modern materials science. Along with development of tools enabling programmable material fabrication with nanometer-level compositional and geometric precision, the unfolding of the local structure of nanomaterials with atomic resolution is increasingly emerging as a fundamental transition pathway towards control of their unique size-dependent properties and realization of their technological potential.
Coherent diffractive imaging (CDI) relies on recording the coherent diffraction pattern of an isolated object at a sampling frequency smaller than the reciprocal of its size (Nyquist sampling), on the basis of which the image of the object is deduced by computational phase retrieval, instead of using a lens to back-transform the diffraction pattern. A particular CDI technique that can be performed in a transmission electron microscope (TEM) experiment is the electron diffractive imaging (EDI), whereby some of the information missing in the electron diffraction (ED) pattern, relative to the target object, can be deduced from its corresponding phase-contrast high-resolution TEM (HRTEM) image. The latter can be used as input data in the EDI experiment, from which a final phase-retrieved picture of the specimen can be eventually extracted with a resolution higher than that characterizing the original HRTEM image.

Herein, we demonstrate an EDI methodology, performed in a Jeol 2010F UHR microscope (spherical aberration coefficient Cs = 0.47±0.01 mm), by which the crystal structure of transition-metal oxide nanocrystals can be determined at 70 pm of resolution while unambiguously revealing the presence and location of light elements atomic columns in the relevant lattice. This approach, applied as a case study to TiO₂ in the form of organic-capped nano-rods, also allows appreciating subtle alterations in the unit cell structure of the nano-crystals, relative to that inherent to the bulk material counterpart, which would not be otherwise detectable by conventional HRTEM. Such structural deviations could be at the origin of peculiar size-dependent physical-chemical properties of the concerned oxide material in the nanoscale regime. In addition, it is worthwhile to remark that this result has been achieved exposing the specimen to an electron dose as low as 106 e/nm2. The latter condition usually prevents the specimen against possible structural damages under exposure to 200 keV electrons, the induction of which remains one of the key issues in the ultimate accuracy achievable in the structural determination of materials.
Fig. 1a displays an experimental phase-contrast HRTEM image of an individual TiO₂  nanorod found on an isolated area of the TEM grid. The small dimensions of the nanocrystal assure kinematical scattering of electrons, while the optic conditions enable an interpretable image resolution of 0.19 nm; Fig. 1b is obtained as a combination of the fast Fourier transform (FFT) of a) with the nanodiffraction pattern as obtained illuminating a region of about 40 nm centered on the rod and subtracted from the contribution of the amorphous carbon substrate; Fig. 1c is a magnified view of the HRTEM contrast shown in fig. 1a. A new phase retrieval algorithm was then applied which allowed to extend the crystal information from the 0.19 nm HRTEM image (Fig. 1a) to the maximum spatial resolution of the diffraction data (here 70 pm), shown in Fig 1b. Some iterations of Hybrid-Input-Output and of Error-Reduction (ER) algorithms have been cyclically alternated for allowing a faster convergence. Any simultaneous modulus and phase updating for the unobserved data, used in the phasing process, was avoided. Missing data at high resolution were estimated only at half of the ER cycles. Among these unobserved data estimates, only the strongest ones were used in the phasing process [5]. Fig. 1d shows the phase-retrieved diffractive image of the nanorod crystal structure, which has been rotated so as to exhibit the c-axis vertically oriented. 

Figure 1: a) HRTEM image of  a  TiO₂ rod in [100] zone axis; b) Combination of the FFT of a) with the nanodiffraction pattern of the relevant rod subtracted from the contribution of the amorphous carbon substrate; c) Magnified view of the HRTEM image contrast in a); d) Retrieved image, the rectangular box is the TiO₂ in [100] projection (blue: O atoms, red: Ti atoms).

The retrieved result has been averaged over 10 cells to improve the display but producing a symmetric pattern. The elongated spots visible in fig 1d, are formed by two Ti and two O atomic columns that are not completely resolved, as schematically shown by comparison with the partially overlapped tetragonal structure of bulk anatase TiO₂ highlighted in the rectangular box. Indeed, the approach allows one to visualize the oxygen atomic column at a resolution of 70 pm [5], not detectable in the relevant HRTEM image. In conclusion, electron coherent diffraction imaging in the TEM represents an approach to improve the capability to study the matter at the highest spatial resolution and accuracy and its development can push further the limits of today microscopy.

For further details see:

Methodological studies: Quantitative atomic resolution z-contrast in STEM (also known as HAADF imaging).

HAADF imaging is a field of large interest in which the European community has some delay with respect to USA or Japan. In particular, CME is the first laboratory in Italy to achieve a resolution of 0.126 nm with this approach. It has also demonstrated how HAADF can be used to gauge the Si distribution in GaAs on the atomic scale even without any image simulation (Fig. 1).

FIGURE 1 - (a) High-resolution z-contrast HAADF image in [110] zone axis of a Si quantum well. (b) Intensity profile (a. u.) as measured on the image in Fig. 2(a)

On the side of the simulation of the results, software with multislice approach and the frozen phonon approximation is being developed starting from the codes of E. J. Kirkland. The limit of this calculation is in the huge CPU time necessary to get reliable simulation. CEM has also developed  a  procedure to reduce the calculation time by means of parallel computing. By comparison of the experimental and simulated images it has been possible to add a quantitative value to the chemical information. The quantitative agreement with STM measurement is embarassing...

The Centre for Electron Microscopy of TASC is a facility open, in first place, to the whole INFM community. Potential users are requested to address to the CEM at TASC informal but detailed expressions of interest in TEM experiments. Please contact the head of the microscopy group, Dr Elvio Carlino, 

(Elvio.Carlino@TASCdomain).

TASCdomain=tasc.infm.it 

The CEM is also responsable for the SEM machine: a new TASC  facility recently  installed and available for users. Fig. 4 shows one of the first images obtained with the new SEM: it is an image of GaAs nanowires grown at the MBE group at TASC


fig. 4