New technologies in modern microelectronic devices always ask for improvements in the methods of design verification and failure analysis: Here the problems arising in the application of the focused ion beam (FIB) for device modification, probing point preparation and failure analysis are discussed. The main problems are caused by the use of multi-level wiring in combination with chemical mechanical polishing techniques CMP and new design style in modern sub-µm-CMOS technologies. FIB-specific problems and solution are discussed.
The etching properties of ion beams are used to prepare thin sample for observation using an transmission electron microscope. A good TEM image is obtained when faces are flat and parallel. The flatness is not fully achieved using a single pass etching mode of a FIB due to the gaussian shape of the ion beam. We show that using a raster scan in a rectangular window associated with a constant scanning rate, a flat etching profile is achieved. We have calibrated the angle of the flat face according to the experimental conditions : ion current, window size and scanning rate. Using this data, we can tilt the sample at an appropriate angle and fabricate a thin film in which both faces are flat and parallel giving uniform contrast in the TEM image of the film.
In order to optimise TEM observations of FIB prepared samples, the thickness must be measured in the FIB. For film thickness above 300 nm, this can be directly achieved using the ion beam in a high-resolution mode. For thinner films, an electron beam of a dual beam system can be used. We show that the secondary electron contrast is sensitive to specimen thickness and beam energy. The primary electrons can pass trough the film at high energy and are stopped at low energy giving a brighter image for a thinner film. After calibrating this phenomena, the data can be used to directly evaluate the thickness of a film in the range of 80 to 300 nm with electron beam energy between 2 and 6kV.
Sample preparation technique for TEM routine analysis has become increasingly important as technologies continue to advance. Therefore, focussed ion beam (FIB) techniques are now more and more used as a preparation tool to guarantee a high throughput and to meet high demands for accuracy and reliability. Flip-chip packaging technology is a driving force to develope backside analysis techniques owing to the limited access from the topside of the wafer. Consequently, the backside preparation technique is challenged. Furthermore, the demands concerning localized area preparation increase because the layout recognition on the polished backside surface is more difficult. In addition to the technical arguments, a more rapid sample preparation is required for reducing the cycle time for process development and yield improvement.
We have used a backside preparation technique combining a conventional mechanical polishing and FIB to prepare TEM cross sections of the under bump metallization stack of C4 bumps. The aim of the TEM investigation was the microstructural and microanalytical characterization of the metallurgical layer stack as well as of the interfaces.
Recent improvements in TEM sample preparation techniques are reviewed with particular emphasis on the following: automation, stress control and damage. The ability to prepare defined sections unattended is discussed. Drift control software routines are required to eliminate problems of positioning accuracy and an efficient scripting language gives flexibility. Results over numerous samples show a high correlation between required thickness of sample and actual thickness. Micromachining is used to relieve the stresses which could lead to bending and warping of thin membranes. Highly stressed specimens such as oxidised metals require more than simply freeing the membrane from its side supports. A spring, micromachined on each side of the membrane, allows lateral relief without the excessive curling that can be expected a cut. The point defect distribution within a membrane is discussed. Suggestions are given for techniques which minimise the number and effect of the defects.
Atom probe field ion microscopy is used to investigate small microstuctural features such as grain and phase boundaries and nanometer-scale precipitates. An intrinsic requirement of this technique is the production of a very high electric field. (20 - 50 Vnm-1) at the surface of the specimen in order to achieve field ionisation and evaporation. The field is produced by applying a high voltage (typically 5-15kV) to a needle shaped specimen with an apex radias of curvature whiskers or blanks cut from bulk material. For samples with certain geometries or ones which cannot be favourably electropolished other methods have to be used to produce the fine needles required. Examples include certain multi-phase alloys, semiconductors, ceramics and multilayerd thin film materials.
Here we will show our work using FIB to produce atom probe samples as a way of overcoming some of these difficulties. After outlining two technbiques ( which depend largely on the type of starting sample) that we have developed to produce needles, we will pay particular attention to the way in which Ga doping of the specimen surface affecht the quality of the final sample. Our results will show implantation depths as a function of the accelerating voltage used and prove that FIB is a viable tehnique for producing a variety of atom probe samples including ones from a Cu/Co multilayer, TiAl and Stainless Steel.
Water based selective carbon milling has been introduced recently in FIB systems. Its possible applications for silicon technology are related to e.g. cross-sectioning of resist masks for etch profile studies, removal of resist passivation layers, and in future generations of devices the drilling of via holes through organic interlevel dielectrics for device modification work.
Some practical examples of these applications will be examined. The milling conditions giving optimal etching of the carbon-based materials will be discussed. It will be shown that under certain conditions, residues are formed which cannot be removed. The limitations for resist profile studies will be illustrated. Etch rates and enhancements will be discussed for different kinds of materials.
