Metal Microstructures in Advanced CMOS Devices

As advanced semiconductor device features shrink, grain boundaries and interfaces become increasingly more important to the properties of thin metal films. With film thicknesses decreasing to the range of 10 nm and the corresponding features also decreasing to sub-micrometer sizes, interface and grain boundary properties become dominant. In this regime the details of the surfaces and grain boundaries dictate the interactions between film layers and the subsequent electrical properties. Therefore it is necessary to accurately characterize these materials on the proper length scale in order to first understand and then to improve the device effectiveness. In this talk we will examine the importance of microstructural characterization of thin metal films used in semiconductor devices and show how microstructure can influence the electrical performance. Specifically, we will review Co and Ti silicides for silicon contact and gate conductor applications, Ti/TiN liner films used for adhesion and diffusion barriers in chemical vapor deposited (CVD) tungsten vertical wiring (vias) and Ti/AlCu/Ti-TiN films used as planar interconnect metal lines.

EELS Measurement of the Local Electronic Structure of Copper Atoms Segregated to Aluminum Grain Boundaries

It has long been known that the addition of Cu to Al interconnects improves the resistance to electromigration failure. It is generally accepted that this improvement is the result of Cu segregation to Al grain boundaries. The exact mechanism by which segregated Cu increases service lifetime is not understood, although it has been suggested that the formation of thin layers of θ-CuA12 (or some metastable substoichiometric precursor, θ’ or θ”) at the boundaries may be necessary. This paper reports measurements of the local electronic structure of Cu atoms segregated to Al grain boundaries using spatially resolved EELS in a UHV STEM. It is shown that segregated Cu exists in a chemical environment similar to that of Cu atoms in bulk θ-phase precipitates.Films of 100 nm thickness and nominal composition Al-2.5wt%Cu were deposited by sputtering from alloy targets onto NaCl substrates. The samples were solution heat treated at 748K for 30 min and aged at 523K for 4 h to promote equilibrium grain boundary segregation. EELS measurements were made using a Gatan 666 PEELS spectrometer interfaced to a VG HB501 STEM operating at 100 keV. The probe size was estimated to be 1 nm FWHM. Grain boundaries with the narrowest projected width were chosen for analysis. EDX measurements of Cu segregation were made using a VG HB603 STEM.

Facet Topography and Dislocation Structure as a Possible Source for Electrical Heterogeneity in [001] TILT Bicrystals of YBa2Cu3O7-δ

A current theme in the study of the critical current density behavior of YBa2Cu3O7-δ (YBCO) grain boundaries is that their electromagnetic properties are heterogeneous on various length scales ranging from 10s of microns to ˜ 1 Å. Recently, combined electromagnetic and TEM studies on four flux-grown bicrystals have demonstrated a direct correlation between the length scale of the boundaries’ saw-tooth facet configurations and the apparent length scale of the electrical heterogeneity. In that work, enhanced critical current densities are observed at applied fields where the facet period is commensurate with the spacing of the Abrikosov flux vortices which must be pinned if higher critical current density values are recorded. To understand the microstructural origin of the flux pinning, the grain boundary topography and grain boundary dislocation (GBD) network structure of [001] tilt YBCO bicrystals were studied by TEM and HRTEM.

Comparison of Techniques for EELS Mapping in Biology

Elemental mapping of biological specimens by electron energy loss spectroscopy (EELS) can be carried out both in the scanning transmission electron microscope (STEM), and in the energy-filtering transmission electron microscope (EFTEM). Choosing between these two approaches is complicated by the variety of specimens that are encountered (e.g., cells or macromolecules; cryosections, plastic sections or thin films) and by the range of elemental concentrations that occur (from a few percent down to a few parts per million). Our aim here is to consider the strengths of each technique for determining elemental distributions in these different types of specimen.On one hand, it is desirable to collect a parallel EELS spectrum at each point in the specimen using the ‘spectrum-imaging’ technique in the STEM. This minimizes the electron dose and retains as much quantitative information as possible about the inelastic scattering processes in the specimen. On the other hand, collection times in the STEM are often limited by the detector read-out and by available probe current. For example, a 256 x 256 pixel image in the STEM takes at least 30 minutes to acquire with read-out time of 25 ms. The EFTEM is able to collect parallel image data using slow-scan CCD array detectors from as many as 1024 x 1024 pixels with integration times of a few seconds. Furthermore, the EFTEM has an available beam current in the µA range compared with just a few nA in the STEM. Indeed, for some applications this can result in a factor of ~100 shorter acquisition time for the EFTEM relative to the STEM. However, the EFTEM provides much less spectral information, so that the technique of choice ultimately depends on requirements for processing the spectrum at each pixel (viz., isolated edges vs. overlapping edges, uniform thickness vs. non-uniform thickness, molar vs. millimolar concentrations).

Direct imaging of charge transfer

The electron holes confined to the CuO2-plane are the charge carriers in high-temperature superconductors, and thus, the distribution of charge plays a key role in determining their superconducting properties. While it has been known for a long time that in principle, electron diffraction at low angles is very sensitive to charge transfer, we, for the first time, show that under a proper TEM imaging condition, it is possible to directly image charge in crystals with a large unit cell. We apply this new way of studying charge distribution to the technologically important Bi2Sr2Ca1Cu2O8+δ superconductors.Charged particles interact with the electrostatic potential, and thus, for small scattering angles, the incident particle sees a nuclei that is screened by the electron cloud. Hence, the scattering amplitude mainly is determined by the net charge of the ion. Comparing with the high Z neutral Bi atom, we note that the scattering amplitude of the hole or an electron is larger at small scattering angles. This is in stark contrast to the displacements which contribute negligibly to the electron diffraction pattern at small angles because of the short g-vectors.

Probing vacancies and structural distortions at individual defects in ceramics using EELS

In electron energy loss spectroscopy (EELS), the near-edge region of a core-loss edge contains information on high-order atomic correlations. These correlations give details of the 3-D atomic structure which can be elucidated using multiple-scattering (MS) theory. MS calculations use real space clusters making them ideal for use in low-symmetry systems such as defects and interfaces. When coupled with the atomic spatial resolution capabilities of the scanning transmission electron microscope (STEM), there therefore exists the ability to obtain 3-D structural information from individual atomic scale structures. For ceramic materials where the structure-property relationships are dominated by defects and interfaces, this methodology can provide unique information on key issues such as like-ion repulsion and the presence of vacancies, impurities and structural distortion.An example of the use of MS-theory is shown in fig 1, where an experimental oxygen K-edge from SrTiO3 is compared to full MS-calculations for successive shells (a shell consists of neighboring atoms, so that 1 shell includes only nearest neighbors, 2 shells includes first and second-nearest neighbors, and so on).

Time-resolved high-resolution electron microscopy of structural stability in mgo clusters

Structure and stability of atomic clusters have been studied by time-resolved high-resolution electron microscopy (TRHREM). Typical examples are observations of structural fluctuation in gold (Au) clusters supported on silicon oxide films, graphtized carbon films and magnesium oxide (MgO) films. All the observations have been performed on the clusters consisted of single metal element. Structural stability of ceramics clusters, such as metal-oxide, metal-nitride and metal-carbide clusters, has not been observed by TRHREM although the clusters show anomalous structural and functional properties concerning to solid state physics and materials science.In the present study, the behavior of ceramic, magnesium oxide (MgO) clusters is for the first time observed by TRHREM at 1/60 s time resolution and at atomic resolution down to 0.2 nm.MgO and gold were subsequently deposited on sodium chloride (001) substrates. The specimens, single crystalline MgO films on which Au particles were dispersed were separated in distilled water and observed by using a 200-kV high-resolution electron microscope (JEOL, JEM2010) equipped with a high sensitive TV camera and a video tape recorder system.

Site occupancy in gamma alumina

Gamma (γ) alumina is referred to as a defect spinel because it has a tetragonally distorted spinel structure (AB2O4) and an insufficient number of cations to fill all cation sites. In the spinel structure, the oxygen lattice is cubic close packed with A- and B-site cations in tetrahedral and octahedral coordination, respectively. The 2l⅓ Al atoms per unit cell of γ alumina can distribute themselves across 16 octahedral and 8 tetrahedral sites.The literature differs on where the 2⅔ cation vacancies per unit cell are located. Wilson and McConnell proposed that the vacancies in γ alumina, as first formed by calcining boehmite, are predominantly on the tetrahedral lattice but, with further heat treatment, move to occupy random positions on both octahedral and tetrahedral lattices. One study using NMR showed that the vacancies lay exclusively on the tetrahedral lattice, independent of the calcination temperature. A more-recent study using Rietveld refinement of powder neutron diffraction data suggested that both octahedral and tetrahedral lattices were partially occupied.

Construction of plastic contact deformation maps on ceramics: a case study on aluminum nitride

Modern polishing, precision machining and microindentation techniques allow the processing and mechanical characterization of ceramics at nanometric scales and within entirely plastic deformation regimes. The mechanical response of most ceramics to such highly constrained contact is not predictable from macroscopic properties and the microstructural deformation patterns have proven difficult to characterize by the application of any individual technique. In this study, TEM techniques of contrast analysis and CBED are combined with stereographic analysis to construct a three-dimensional microstructure deformation map of the surface of a perfectly plastic microindentation on macroscopically brittle aluminum nitride.The bright field image in Figure 1 shows a lg Vickers microindentation contained within a single AlN grain far from any boundaries. High densities of dislocations are evident, particularly near facet edges but are not individually resolvable. The prominent bend contours also indicate the severity of plastic deformation. Figure 2 is a selected area diffraction pattern covering the entire indentation area.

Use of digital microscopy in process control during the manufacture of rubber-modified thermoplastics

The production of rubber modified thermoplastics can exceed rates of 30,000 pounds per hour. If a production plant needs to equilibrate or has an upset, that means operating costs and lost revenue. Results of transmission electron microscopy (TEM) can be used for process adjustments to minimize product loss. Conventional TEM, however, is not a rapid turnaround technique. The TEM process was examined, and it was determined that 50% of the time it took to complete a polymer sample was related to film processing, even when using automated equipment. By replacing the conventional film portion of the process with a commercially available system to digitally acquire the TEM image, a production plant can have the same TEM image in the control room within 1.5 hours of sampling.A Hitachi H-600 TEM Operated at 100 kV with a tungsten filament was retrofitted with a SEMICAPS™ image collection and processing workstation and a KODAK MEGAPLUS™ charged coupled device (CCD) camera (Fig. 1). Media Cybernetics Image-Pro Plus software was included, and connections to a Phaser II SDX printer and the network were made. Network printers and other PC and Mac software (e.g. NIH Image) were available. By using digital acquisition and processing, the time it takes to produce a hard copy of a digital image is greatly reduced compared to the time it takes to process film.