Transmission electron microscopy in its modern high-resolution analytical form has great potential for elucidating mechanisms of environmental cracking in structural alloys and other materials. Apart from a few pioneering efforts, the difficulties of preparing suitable samples with the original corrosion structures preserved in narrow crack tips have limited contributions of TEM in this area. Recent work involving cross-sectional crack preparation has focused on intergranular stress-corrosion cracking of Fe- and Ni-base stainless alloys in high-temperature water. This presentation will illustrate preparation methods that have proven successful, and describe limitations of the analysis.

The method for preparing cross-section samples suitable for high-resolution TEM characterization is based on earlier work by Lewis et al. It involves preparing a thin section containing cracks filled with epoxy, dimple grinding selected locations near the crack tip, and ion milling. Figure 1 schematically illustrates this method. The first step is to preserve the crack opening characteristics present under loading and protect the crack surfaces during thinning.

The combination of focused ion beam (FIB) sputtering with transmission electron microscopy (TEM) offers new opportunities for the nanoscale characterization of materials. The FIB may be used to prepare membranes for TEM imaging which are: (i) Site selective, i.e. the membranes may be placed with sub-micron precision in all three dimensions, (ii) Largely free of differential sputtering artifacts, such that membranes may be prepared which are of constant thickness from structures with very dissimilar materials, and (iii) Of precisely known geometry.

The challenges associated with FIB specimen preparation will also be discussed and are summarized in Figure 1: (a) Surface amorphization damage, (b) Residual differential sputtering effects, (c) Redeposition of sputtered material and (d) Membrane bowing due to internal or beaminduced stresses. It will be demonstrated that each of these effects can be sufficiently controlled to allow high quality diffraction contrast imaging in a wide range of materials.

The use of focused ion beams (FIB) to prepare site specific specimens for transmission electron microscopy (TEM) and scanning electron microscopy (SEM) has been well documented. The 5 to 7 nm resolution on the latest FIB instruments has enhanced the ability to locate and expose features of interest. The use of the lift-out technique of FIB specimen preparation removes the need for prior thinning of the sample for TEM analysis and permits the study of materials that were difficult or impossible to do previously. Use of high current (10 nA) FIB instruments makes specimen preparation possible in less than one hour; automatic operation of FIB instruments will further reduce this time. After milling in the FIB, specimens are micromanipulated in air onto a 3 mm diameter TEM sample grid coated with carbon. The ability to analyze lift-out specimens using other analytical techniques that can take advantage of this site specific capability was previously suggested.

There is a trend in electron microscopy towards centralization of the more sophisticated EM systems. The rationale is that the techniques associated with such instrumentation are more efficiently practiced and developed by dedicated specialists in a regional or national centre surrounded by an extensive infrastructure. A similar argument can be used for the preparation of TEM specimens, especially as the demand grows for more precise location of the thin area to be examined. A new trio of techniques - tripod polishing, ultramicrotomy, and focused ion beam (FIB) thinning - can locate the thin area to within a micron or less. In the first two, a linear feature first located via light or scanning electron microscopy can be cross-sectioned by controlled material removal via mechanical polishing or diamond knife sectioning, respectively. With FIB, feature location and material removal are carried out in-situ, first imaging with a low intensity beam, then milling trenches on either side of the feature with a high intensity beam that is decreased as the feature is approached.

The focused ion beam (FIB) miller has been widely accepted as a powerful tool in the semiconductor industry. However, it is now finding applications in more general materials science applications. The high resolution, energetic gallium ion beam can rapidly and precisely section materials to reveal their internal structure; one particularly valuable application being the preparation of thin foils for TEM examination, especially from heterogenous materials.

To date, TEM sample preparation using FIBs has concentrated on semiconductor cross-sections [1], powders [2], and surface treated materials, e.g. galvanized steels [3]. However, thin foils of grossly heterogeneous materials, such as metal-matrix composites, are also difficult to prepare using conventional methods and are therefore well suited to sectioning using the FIB. In this study, thin foils were prepared from two composite materials: a 7075 aluminium alloy containing a 20% volume fraction of SiC particles and a FeAl alloy containing a 60% volume fraction of WC particles.

As features on an IC chip become smaller than the resolution power of an optical microscope and of the size of the grinding particles, the trend for preparing cross-sectional transmission electron microscopy (TEM) samples at specific locations (bits) is moving towards using a focused ion beam (FIB) machine. Details on how to use a FIB machine to prepare cross-sectional TEM samples have been outlined in many references.The general procedure is to first mark the specific location (bit) in the FIB machine and then grind the sample down to about 20 microns, 10 microns on each side of the feature of interest. After grinding, the sample is mounted on a pre-cut TEM grid and thinned with the FIB to about 0.1 micron in the region containing the feature of interest. There are several disadvantages to this method. First, the sample goes into the FIB machine at least twice—once for FIB marks on the location and once again for the final thinning.

Scanning Electron Microscopy/Energy Dispersive Spectroscopy (SEM/EDS) has become less useful for semiconductor samples as feature dimensions decrease. It has become increasingly difficult to get sufficient signal from small features because of the large interaction volume of the primary beam. As the interaction volume is dependent on the accelerating energy of the primary electron beam, decreasing the accelerating voltage will decrease the interaction volume. This increases the percentage of signal from the feature of interest, but also lowers the number of peaks available for interpretation, reducing the sensitivity of the analysis. Several elements commonly used in the fabrication of semiconductors have overlapping peaks at lower energies, and the ability to distinguish between them is lost at low accelerating energies. It is possible to modify samples using Focused Ion Beam (FIB) to improve EDS resolution without decreasing the accelerating voltage.

The samples were prepared in an FIE 800 FIB and imaged in an Hitachi S-4700 SEM.

1. Introduction

The FIB (focused ion beam) is now widely accepted as the most site-specific TEM preparation tool and as such proves to be highly valuable when analysing ULSI devices. However, using high-energy Gallium ions for milling induces amorphization of the crystal surfaces. A method able to quantify this surface alteration on silicon using a combination of CBED (convergent beam electron diffraction) and EELS (electron energy loss spectroscopy) is presented. CBED is a powerful tool that also can generate an accurate measure of crystal thickness. EELS can yield the total sample thickness, so from the difference the combined amorphous layers can be assessed. Two sets of application results are presented: the first one is obtained on a FIB thinned sample using an ion energy of 50 keV and the second set of results confirms the validity of the proposed method on a mechanically polished specimen with no subsequent ion milling.

1. Introduction.

With continuing reductions in semiconductor device dimensions high spatial resolution physical and chemical analysis techniques will be more and more required for defect analysis and process development in the microelectronics field. Transmission Electron Microscopy (TEM) analysis is now extensively used thanks to the fast Focused Ion Beam (FIB) specimen preparation technique which has furthered its development. Recently, we have shown the advantages of adding Electron Energy Loss Spectroscopy (EELS) to FIB-TEM analysis for semiconductor process characterization. In this paper we extend the EELS technique using FIB sample preparation to Energy Filtering TEM (EFTEM) observations. The EFTEM analysis allows high-resolution compositional mapping using spectroscopic imaging of core level ionization edges3. We show some applications of FIB-EFTEM to defect analysis and process development.

2. Experimental details.

The FIB system is a MICRION model 9500 EX using a gallium ion beam of 50 keV maximum energy with a 5 nm minimum spot diameter.

Focused ion beam (FIB) thinning of materials to electron transparency is now a routine procedure for preparation of specimens for transmission electron microscopy (TEM) of microelectronic materials and devices. The nano-scale structural damage, including implantation and amorphization due to this ion milling process has been well investigated and documented. In this paper, we discuss the micro-scale structural damage observed in copper/low-k materials and our efforts to minimize the extent of the damage without compromising the overall specimen preparation time.

Figure 1 shows an area-specific cross-sectional specimen prepared from a copper/low-k via-chain test structure using the FIB-milling technique. The procedure involved mechanical thinning of a transverse wafer sliver followed by FIB-milling the area of interest to electron transparency according to conventional steps and conditions' using a liquid Ga+ ion source FIB system. The evidence of structural damage in terms of melting and/or sputtering of the metallization is visible at different areas.

Preparation of precision planar TEM specimen for failure analysis in microelectronics industry has become increasingly difficult as the critical dimensions of integrated circuits (ICs) continue to shrink down to deep sub-micron regime. Accurate defect localization and endpoint control of specimen final thinning are critical steps leading to successful specimens and defect characterization. Conventional dimple and ion-mill method becomes inadequate for this class of specimen preparation due to poor optical imaging resolution, non-planar polishing, lack of flexible ion beam control and in situ process monitoring for endpoint control. These problems are exacerbated by the high density and a wide range of materials found in IC device structures. In this paper we present a novel method employing dual-beam focused ion beam (FIB) which significantly improves upon these difficulties.

The Scanning Transmission Electron Microscope (STEM) can provide images beyond Scanning Electron Microscope (SEM) capability, on samples which do not have to be prepared to the degree required for Transmission Electron Microscopy (TEM). For some applications, such as semiconductor production, speed of analysis is critical. This paper explores Focused Ion Beam (FIB) sample preparation for STEM.

The samples were prepared using an Hitachi FB-2000A FEB tool. Two analysis regions of similar thickness were cut. The rough sample was milled using a 500 pA beam and no polish cut. The smooth sample was prepared by cutting first with a 500 pA beam, and then using additional, smaller diameter beams, concluding with a 30 pA beam. This resulted in a sample surface polished to a degree similar to that used for TEM specimens. The samples were imaged in an Hitachi HD-2000 STEM at 200KeV. Transmission electron images taken at 100 kx are shown in Figure 1 of the rough sample and of the smooth sample.

Focused Ion Beam (FIB) specimen preparation for both scanning and transmission electron microscopy (SEM and TEM respectively) has seen an increase in usage over the past few years. The advantage to the FIB is that site specific cross sections (or plan view sections) may be fabricated quickly and reproducibly from numerous types of materials using a finely focused beam of Ga+ ions [1,2]. It was demonstrated by Prenitzer et al. that TEM specimens may be acquired from individual Zn powder particles by employing the FIB LO specimen preparation technique [3]. In this paper, we use the FIB LO technique to prepare TEM specimens from Mount Saint Helens volcanic ash.

Volcanic ash from Mount Saint Helens was obtained at the Microscopy and Microanalysis 1998 meeting in Atlanta. TEM analysis of the ash was performed using the FIB lift out technique [1]. Ash powders were dusted onto an SEM sample stud that had been coated with silver paint.

It is well known that the “channeling contrast” and EBSD pattern formation, which is the basis for “orientation imaging microscopy” (OIM), is highly sensitive to specimen surface quality in SEM studies. Typically, specimens for these studies are prepared by a combination of mechanical grinding then chemical etching or electropolishing. As an alternative to the chemical step(s) in this technique, we report a method based upon broad ion beam etching equipment [1]. This alternative method is more universal and reliable for enhanced channeling contrast and EBSD/OIM of metallic materials.

A variety of polycrystalline metallic specimens were etched in an instrument dedicated for etching and coating SEM specimens [1] and examined in the SEM for channeling contrast and OIM mapping. Fig. la is a SEM image of an annealed titanium sheet specimen in as-received condition showing a well-defined surface texture. It can be seen that channeling contrast is absent, presumably due to the existence of a surface layer formed during the manufacturing process.

Argon ion beam etching has established itself as an alternative technique to “wet chemical” etching for the preparation of cross sectional SEM specimens of semiconductors [1]. Complementing this technique, we are reporting the results of an iodine RIBE method for improved etching/cleaning capabilities with a measurable increase in etching rates as compared to argon ion beam etching. RIBE systems have been used for decades in the semiconductor research/industry for wafer processing, patterning and surface cleaning. This same technique has also been used for high quality TEM specimen preparation of certain semiconductor materials [2,3]. The beneficial aspects of the iodide RIBE technique for surface etching for a variety of semiconductor structures along with the related instrumentation will be discussed. The semiconductor specimens include traditional ICs and more advanced copper technology devices.

The design and construction of the original system used in this work has already been reported [4].

A method for site specific characterization of the materials using a dedicated focused ion beam(FIB) system and an analytical transmission electron microscope (TEM) was developed. Needless to say, in TEM specimen preparation using FIB system, stability of a specimen is quite important. The specimen stage employed in the developed FIB system is the one designed for high resolution TEM, and the specimen drift rate of the stage is less than lnm/min. In addition, FIB-TEM compatible specimen holder which allows milling of a specimen with the FIB system and observation of the specimen with the TEM without re-loading was developed. To obtain thin specimen from the area to be characterized correctly, confirmation of the area before final milling is needed. However, observation of cross sectional view in a FIB system is recommended because it causes damage by Ga ion irradiation. To solve this problem, we used a STEM unit as a viewer of FIB milled specimen.

I present details of novel sample preparation techniques used for delineating two-dimensional dopant profiles in silicon devices. These techniques are based on selective chemical etch of doped silicon in a mixture of hydrofluoric and nitric acid, or simply in buffered HF. The altered topography of the etched surface is imaged by transmission electron microscopy (TEM). Two different strategies will be presented by focusing on their sensitive, resolution, and field of application.

In the first case the silicon device is subjected to the conventional thinning procedure for TEM observations in cross-sectional configuration. The final thickness is obtained by Ar ion milling and it can vary between 50 to 500 nm. Sample is then immersed in a chemical solution containing HF (40%), HN03 (65%), and CH3COOH (95%) in the ratio 1:10:10. In presence of an intense illumination this mixture preferentially etches those device regions which are doped with boron, while in order to delineate n-type regions, the etching procedure must be performed in the dark.

Preparation of lead and lead alloy specimens for metallographic examination is more difficult than preparation of the majority of other metals and their alloys. The troublesome difference is that lead and lead alloys are soft (lead has a Vickers hardness of 4-5 [1,2]), and therefore readily develop a cold-worked surface layer and disturbances in the subsurface layers during mechanical polishing. The specimen softness also allows embedding of the abrasive particles. If not removed, these mechanically introduced artifacts obscure the true structure of the specimen, and in some cases could develop a pseudostructure, especially if excessive frictional heat is involved [1]. Usually etchpolishing, consisting of cycles of alternately etching, polishing, and microscopic examination, is required to remove the above artifacts. This treatment either might not completely remove these artifacts or it might overetch the sample. In addition, etching could introduce its own artifacts, such as tarnishing, etching pits, or both.

Sample substrates for x-ray microanalysis of particles should be chemically simple and composed of low atomic number elements. Substrates meeting these requirements will generate a minimum number of low energy x-ray peaks that should not interfere with spectra generated by the particles of interest. Elemental carbon fulfills these requirements and carbon planchets are used for the majority of work involving x-ray microanalysis of particles. When particles of interest are carbonaceous, alternate substrates must be sought. Beryllium planchets have traditionally been used for the analysis of carboncontaining particles. However, concerns about beryllium toxicity along with regulations from the Occupational Safety and Health Administration (OSHA) have resulted in diminished use of beryllium as a substrate material. These toxicity issues prompted us to explore the possibility of using elemental boron as a substrate for microanalysis of carbon-containing particles.

Elemental boron is available from chemical supply houses and the material we purchased was in the form of irregularly shaped pieces, approximately 3 cm in the longest dimension.

Remote ice core, snow, and air filter samples can potentially profile past and present atmospheric characteristics, acting as media to preserve populations of particulate matter. Major constituents of these aerosols derive from combustion processes (fossil and biomass) and airborne minerals. Extraction and characterization of these particles with scanning electron microscopy (SEM) and energy-dispersive x-ray spectroscopy (EDS) provide morphological and elemental data used in determining the presence and sources of atmospheric aerosols. Due to the minute quantity of carbonaceous particles in the snow and ice samples, ca. 1-10 ng g-1, the microscopist faces several principal challenges to prepare and analyze these samples. First, methodology must be established for particle removal in a clean and quantitative fashion. Secondly, since carbon is contained in many of the particles under analysis, techniques must be established to achieve quantitative carbon X-ray data. The SEM requires optimization in order to analyze the greatest number of particles in the least time.