Chemical 3D tomography of 28nm high K metal gate transistor: STEM XEDS experimental method and results.

A new STEM XEDS tomography technique is proposed thanks to the implementation of multi EDX SDD detectors in analytical TEMs. The technique flow is presented and the first results obtained on a 28nm FDSOI transistor are detailed. The latter are compared with 2D XEDS analysis to demonstrate the interest of the slice extraction in all directions from a large analyzed volume without any 3D overlap effect issues.

Coherent Bremsstrahlung effect observed during STEM analysis of dopant distribution in silicon devices using large area silicon drift EDX detectors and high brightness electron source.

In this paper, during dopant analysis of silicon devices, we have observed a phenomenon generally neglected in EDX analysis: the coherent Bremsstrahlung (CB). We discussed the reason why and came to the conclusion that the analytical TEM used for these experiments presents a configuration and performances, which makes this equipment very sensitive to the CB effect. This is due to large collection solid angle and high counting rate of the four silicon drift EDX detectors (SDD), a high brightness electron source providing large probe current and moreover a geometry favorable to on axis crystal observations. We analyzed silicon devices containing Si [110] and Si [100] crystal areas at different energies (80-120-200keV). We also observed relaxed SiGe (27 and 40at% of Ge). The CB effect, whose intensity is maximum near zone axis beam alignment, manifests as characteristic broad peaks present in the X-ray spectrum background. The peak energies are predicted by a simple formula deduced for the CB models found in the literature and that we present simply. We evaluate also the CB peak intensities and discuss the importance of this effect on the detection and quantification traces of impurities. The CB peaks also give information on the analyzed crystal structure (measurement of the periodicity along the zone axis) and allow, in every particular experiment or system, to determine the median take off angle of the EDX detectors.

Quantitative evaluation of process induced strain in MOS transistors by Convergent Beam Electron Diffraction.

Convergent Beam Electron Diffraction (CBED) experiments and simulations associated with Finite Element calculations were performed in order to measure strain and stress in a complex device such as periodic MOS transistors with a spatial resolution of about 2 nm and a sensitivity that could reach 50 MPa. A lamella of a thickness of about 475 nm was extracted from the wafer with the transistors by Focus Ion Beam (FIB) and was observed in cross-section in a Transmission Electron Microscope (TEM). When approaching the transistors, the HOLZ lines of the CBED patterns acquired in the silicon substrate, become broader and broader. This HOLZ line broadening, which is due to the stress relaxation in the thin foil, was used to determine quantitatively the strain and stress in the lamella and then in the bulk device. We showed that this procedure could be applied to a complex device. Two parameters, the intrinsic material strains--or equivalently the intrinsic material stresses--in the nickel silicide (NiSi) and nitride (Si(3)N(4)) layers on the top of the transistors gate, were successfully fitted by trial and error, in the procedure.

Two-dimensional quantitative mapping of arsenic in nanometer-scale silicon devices using STEM EELS-EDX spectroscopy.

Field emission gun (FEG) nanoprobe scanning electron transmission microscopy (STEM) techniques coupled with energy dispersive X-ray (EDX) and electron energy loss spectroscopy (EELS) are evaluated for the detection of the n-type dopant arsenic, in silicon semiconductor devices with nanometer-scale. Optimization of the experimental procedure, data extraction and the signal-to-noise ratio versus electron dose, show that arsenic detection below 0.1% should be possible. STEM EDX and EELS spectrum profiles have been quantified and compared with secondary ion mass spectrometry (SIMS) analyses which show a good agreement. In addition, the arsenic doping level found inside large and small epitaxial devices have been compared using STEM EDX-EELS profiling. The average doping level is found to be similar but variable interface segregation has been observed. Finally, STEM EDX arsenic mapping acquired in a BiCMOS transistor cross-section shows strong heterogeneities and segregation in the epitaxially grown emitter part.

Dielectric and optical properties of nanometric nickel silicides from valence electrons energy-loss spectroscopy experiments.

Valence and Core Electron Energy Loss Spectroscopy (VEELS and CEELS) experiments are performed from nanocrystallized nickel silicide thin films. Three different silicide compounds are identified in the films. Their chemical compositions are determined from Ni-L(2,3) to Si-K core edges quantification. The results obtained are coherent within less than 2% error with the pure Ni2Si, NiSi and NiSi2 phases. The analysis of the shape and energy position of Ni-L(2,3) near edge structures and volume plasmon peaks indicates that both are reliable signatures to identify unambiguously each compound. Nickel silicides low-loss spectra have been submitted as references to the EELS database (www.cemes.fr~eelsdb). Low-loss spectra are processed to extract single scattering spectra and determine the dielectric function. The results show that nickel silicides dielectric functions deduced from VEELS are in quite good agreement with epsilon1 and epsilon2 deduced from ellipsometry experiments. The optical properties (refractive index (n), absorption coefficient (k), reflectivity (R%) and resistivity (rho(opt))), calculated from VEELS dielectric function are then compared in details with the data resulting from others techniques available in the literature. We show that, except some minor divergences, the nickel silicides optical properties are generally well reproduced. This indicates that VEELS is a relevant technique for accessing reliably to physical properties and can be a successful alternative to conventional techniques when high spatial resolution is needed.

Comparison of Si and Ge low-loss spectra to interpret the Ge contrast in EFTEM images of Si(1-x) Ge(x) nanostructures.

Recently, an EFTEM imaging method, exploiting the inelastically scattered electrons in the 60-90eV energy range, was proposed to visualise Ge in SiGe alloys [Pantel, R., Jullian, S., Delille, D., Dutartre, D., Chantre, A., Kermarrec, O., Campidelli, Y., Kwakman, L.F.T.Z., 2003. Inelastic electron scattering observation using Energy Filtered Transmission Electron Microscopy for silicon-germanium nanostructures imaging. Micron 34, 239-247]. This method was proven to be highly more efficient in terms of noise, drift and exposure time than the imaging of the weak and delayed ionization GeL2,3 edge at 1236eV. However, the physical phenomenon behind this Ge contrast was not clearly identified. In this work, we explain the origin of this Ge contrast, by comparing in details EELS low-loss spectra (<100eV) recorded from pure Si and Ge crystals. High resolved low-loss experiments are performed using analytical Field Emission Gun Transmission Electron Microscopes fitted or not with a monochromator. Low-loss spectra (LLS) are then deconvoluted from elastic/quasi-elastic and plural scattering effects. The deconvolution procedure is established from Si spectra recorded with the monochromated machine. The absence of second plasmon and the measurement of a band gap (1.12eV) on the Si single scattering distribution (SSD) spectrum allowed us to control the accuracy of the deconvolution procedure at high and low energy and to state that it could be reliably applied to Ge spectra. We show that the Ge-M4,5 ionisation edge located at 29eV, which is shadowed by the high second plasmon in the unprocessed Ge spectrum, can be clearly separated in the single scattering spectrum. We also show that the front edge of Ge-M4,5 is rather sharp which generates a high intensity post edge tail on several tens of eV. Due to this tail, the Si and Ge EELS signals in the 60 to 100eV energy window are very different and the monitoring of this signal gives information about the Ge concentration inside SiGe alloys. It is now evident that the EFTEM imaging technique proposed to quantify Ge (90eV/60eV image ratio) in Si-Ge nanostructures is valid and is a relevant way of exploiting the Ge-M4-5 ionisation edge.

Chemical analysis of nanometric dielectric layers using spatially resolved VEELS.

Chemical analysis of structures in the nanometre range is a challenge even with modern analytical transmission electron microscopes (TEM). In this work we demonstrate that it is possible to measure chemical variations in the monolayer scale and identify compounds formed at the interfaces by using Valence Electron Energy Loss Spectroscopy (VEELS) in STEM line-scan mode. We discuss the impact of valence electrons delocalisation on the spatial resolution of our experiments. The method is tested first on a model sample containing a 4 nm HfO(2) layer. The limitations of VEELS to provide chemical analysis are then explored and discussed by applying this technique to a real semiconductor device containing a 2 nm HfO(2) layer.

Inelastic electron scattering observation using energy filtered transmission electron microscopy for silicon -- germanium nanostructures imaging.

This paper presents a new technique using energy filtered TEM (EFTEM) for inelastic electron scattering contrast imaging of Germanium distribution in Si-SiGe nanostructures. Comparing electron energy loss spectra (EELS) obtained in both SiGe and Si single crystals, we found a spectrum area strongly sensitive to the presence of Ge in the range [50-100 eV]. In this energy loss window, EELS spectrum shows a smooth steeply shaped background strongly depending on Ge concentration. Germanium mapping inside SiGe can thus be performed through imaging of the EELS background slope variation, obtained by processing the ratio of two energy filtered TEM images, respectively, acquired at 90 and 60 eV. This technique gives contrasted images strongly similar to those obtained using STEM Z-contrast, but presenting some advantages: elastic interaction (diffraction) is eliminated, and contrast is insensitive to polycrystalline grains orientation or specimen thickness. Moreover, since the extracted signal is a spectral signature (inelastic energy loss) we demonstrate that it can be used for observation and quantification of Ge concentration depth profile of SiGe buried layers.

Convergent beam electron diffraction extinction distance measurements for quantitative analysis of si(1-x),Ge(x).

A new method to determine the concentration of germanium in Si(1-x) Ge(x) single crystals is presented. It is based on extinction distance measurements by means of convergent beam electron diffraction (CBED). The two-beam condition CBED intensity oscillation (the so-called rocking curve) is measured for the 004 diffracted beam and compared with a numerical simulation. Using the two-beam dynamical diffraction approximation theory, this approach yields very precise values for both specimen thickness and effective extinction distance (Ultramicroscopy 87 (2001) 5). First a theoretical extinction distance zetag(x) for strain relaxed Si(1-x)Ge(x) is calculated assuming a solid solution and using tabulated atomic scattering factors of silicon and germanium atoms. It is found that for single crystals zetag(x) decreases from 156 nm in pure silicon to 90 nm in pure germanium. Measurements on calibrated strain relaxed SiGe layers with variable germanium concentrations show an excellent agreement between experimental and calculated extinction distances zetag(x). As a consequence the experimental extinction distance zetag(x) becomes an indirect measure of the germanium concentration with a 1-2 at % sensitivity. The method turns out to be insensitive to strain as experimental zetag(x) values obtained on strained SiGe layers fit the theoretical extinction distance curve calculated for strain relaxed SiGe.

Crystal thickness and extinction distance determination using energy filtered CBED pattern intensity measurement and dynamical diffraction theory fitting.

A new method for measuring thickness and extinction distance of single crystals based on computed adjustment of measured and calculated CBED pattern intensity profiles is presented and discussed. The experimental beam intensity distribution is measured from an energy filtered CBED pattern recorded on a CCD camera. The calculated profile is based on dynamical diffraction theory, and with the two-beam approximation the analytical expression contains only two free parameters: specimen thickness t and extinction distance xig. Parameter refinement through minimization of the difference between experimental and calculated intensity profiles is carried out using Origin 5.0 software from Microcal. The iterative procedure always converges to a unique solution in a few seconds, yielding an accurate value for both thickness and extinction distance. The method is extensively tested on silicon using the (0 0 4) Bragg reflection. On specimens in the usual TEM thickness range, the method gives result similar to the conventional (P.M. Kelly et al., Phys. Stat. Sol. A31 (1975) 771; S.M. Allen, Philos. Mag. A 43 (1981) 325) graphical methods, both based on the measurement of fringe spacing. Moreover, it is shown that the calculation matches perfectly both the positions of the minimums and maximums as well as the amplitude of maximums. For any single intensity profile, specimen thickness and extinction distance can be determined with a precision of about 0.2%. A statistical comparison of our method with the Kelly and Allen techniques, based on more than 50 experiments, shows an improvement in measured extinction distance dispersion. Using 197 keV electrons, and liquid-nitrogen cryo-holder, the new technique yields an experimental value of 161+/-3 nm for the extinction distance for silicon with the (0 0 4) Bragg reflection. The equivalent tabulated value at 0 K is about 156 nm. Using the Kelly and Allen methods, the extinction distance is found to be 162+/-6 nm. The improvement in precision is a direct consequence of matching the intensity profile envelope, which contains information on the extinction distance. Also the accuracy of thickness determination is improved and is around 0.5 to 1% for common specimen thickness. The minimum measurable sample thickness is shown to be two to three times thinner than with the Kelly and Allen methods (0.3 xig, as opposed to 0.8 xig). With no independent calculation of the extinction distance needed, the method is also applicable on unknown crystals. The method is fast, simple and can be easily automated.