Meetings & Events

Publishing Alliance

MRS publishes with Springer Nature



spring 1997 logo1997 MRS Spring Meeting & Exhibit

March 31 - April 4, 1997 | San Francisco
Meeting Chairs: Linda G. Griffith-Cima, David J. Eaglesham, Alexander H. King

Symposium Z—Workshop on Specimen Preparation for Transmission Electron Microscopy of Materials IV


Ronald Anderson 
Dept of Analytical Service 
IBM East Fishkill Facility 
B/630 IBM ZIP E40 
Hopewell Junction, NY 12533 

Scott Walck
Materials Directorate
Air Force Wright Laboratory
WL/MLBT Bldg 654
Wright Patterson AFB, OH 45433-7750

Symposium Support
*Allied High Tech Products, Inc.
*Delaware Diamond Knives 
*Diatome U.S.
*E.A. Fischione Instruments, Inc.
*Electron Microscopy Sciences
*FEI Company
*Gatan, Inc.
*IBM Analytical Services 
*IBM East Fishkill Facility
*Leica, Inc.
*LEO Electron Microscopy Inc.
*Micrion Corporation
*NSA Hitachi
*Philips Electronic Instruments Co.
*South Bay Technology Inc.
*Ted Pella, Inc.
*VCR Group, Inc.
*Wright Laboratory, Materials Directorate

1997 Spring Exhibitor

Proceedings published as Volume 480
of the Materials Research Society
Symposium Proceedings Series.

In sessions below "*" indicates an invited paper.

Chairs: Ronald M. Anderson and Scott D. Walck 
Wednesday Afternoon, April 2, 1997
Salon 13

1:30 PM *Z1.1 
TRANSMISSION ELECTRON MICROSCOPY SPECIMEN PREPARATION OF COATED CERAMIC FIBERS, Michael K. Cinibulk, Air Force Wright Laboratory, Materials Directorate, Wright Patterson AFB, OH; John R. Welch, Randall S. Hay, Air Force Wright Laboratory, Wright Patterson AFB, OH.

The typical ceramic fiber tow has 400 filaments; each filament is 10 m in diameter. The uniformity of applied ceramic coatings and their microstructure on filaments near the center of the tow can be different than those on filaments near the outside, especially in the case of a multicomponent coating. Determination of the morphology, structure, and composition of a 0.1-0.5 m thick coating on the fibers requires transmission electron microscopy. The preparation of the electron transparent specimens is a challenging task. Problems encountered in the preparation of TEM specimens of coated fibers by conventional methods include loss of the coating, destruction of the fiber-coating interface, differential polishing of the composite samples, limited thin area production, and low specimen yield. We have developed a technique for the preparation of electron-transparent sections, containing a representative sampling of fibers to meet the following requirements. First, the epoxy-impregnated tows must contain a high volume fraction of coated fibers. Second, differential polishing of the fiber-epoxy composite must be avoided to allow for a flat, polished surface with no relief to minimize differential thinning in the ion mill. This necessitates thinning the specimens to l-5 m thickness on diamond lapping films to minimize the time for ion-beam thinning. By meeting these criteria, a single TEM specimen can be prepared containing hundreds of coated filaments suitable for characterization by electron microscopy in a comparatively short time. The technique can be applied to any multiphase sample and can also be used for preparing very flat specimens for SEM characterization. The use of this method for preparing specimens of ceramic fibers with ceramic coatings, and their characterization by TEM, will be discussed.

2:00 PM *Z1.2 
FOCUSED ION BEAM MILLING AND THE MICROMANIPULATION LIFT-OUT TECHNIQUE FOR SITE SPECIFIC CROSS-SECTION TEM SPECIMEN PREPARATION, Lucille A. Giannuzzi, J. L. Drown, Univ of Central Florida, Dept of Mech, Matls & Aerospace Engr, Orlando, FL; S. R. Brown, Kirk Resources, Orlando, FL; R. B. Irwin, F. A. Stevie, Bell Labs, Lucent Technologies, Orlando, FL.

Focused Ion Beam (FIB) Milling has been used to prepare electron transparent specimens from electronic, metallic, and ceramic materials. A TEM specimen with a nominal dimension of 5 m wide by 20 m long by 150 nm thick is machined from a bulk sample using an FIB instrument. The electron transparent specimen slice is lifted-out of the bulk sample using a modified glass rod attached to a micromanipulator arm. The specimen is then placed onto a formvar-coated copper grid for TEM analysis. Using this technique, site specific cross sections with submicron location accuracy of complex geometry and chemistry have been prepared for TEM analyses in only 5-8 hours. The advantage to the FIB/Micromanipulation lift-out technique is that little or no sample preparation is required prior to inserting the sample into the FIB, which extends the applicability of TEM to new arenas.

2:30 PM *Z1.3a 
AN UPDATED ION POLISHING SYSTEM FOR TEM SPECIMEN PREPARATION OF MATERIALS17405, Reza Alani, Richard J. Mitro, Peter R. Swann, Gatan Inc, Pleasanton, CA.

The construction and performance of an improved ion milling instrument are described. The instrument is based upon an earner version, implementing two fixed low angle ion guns capable of high milling rates on TEM specimens. The updated system employs two improved guns having the ability of independent variable milling angle adjustment, while operating at high voltage in the vacuum system. The milling rate of these ion guns is typically >90 m/hr/2 guns for copper at 5 keV and 10 . Designed to complement the ion gun variable angle feature is a new specimen post which permits low angle double-sided milling. The combination of these two features offers numerous single- or double sided milling conditions to facilitate the production of higher quality TEM specimens in less time. The range of ion gun angles encompassed is 10 to 0 for single-sided milling and 10 to 0 for double-sided milling. The main instrument also incorporates an electronic mechanism to generate directional sector milling which benefits primarily the production of cross sectional TEM specimens and heat sensitive materials since the ion guns ar actually turned off for a portion of the time during each revolution. Another new feature of the instrument is chemical ion milling which facilitates the preparation of certain compound semiconductors. The ability of this improved ion milling instrument for rapid ion polishing and improved specimen quality is demonstrated by a number of TEM images of semiconductors, metals, ceramics, and composites.

3:00 PM *Z1.3 

As the role of transmission electron microscopy (TEM) in materials research as well as in device development increases, the preparation of satisfactory TEM samples becomes more important. Ion-milling has been widely adopted for the preparation of cross-sectional TEM samples. However, the different thinning rates for each layer during the ion-milling of multilayer structures usually cause large variations in electron transparency. It is critical to understand the process of sputtering and to develop methods of eliminating or alleviating the differential thinning problems. We have suggested that the differential thinning problems can be effectively solved by optimizing both specimen rotation mode and ion beam incidence angle. In the present paper, we will define a parameter which can be used as a guideline for successful ion-milling of cross sectional samples of layered systems.

3:30 PM *Z1.4 
FOCUSED ION BEAM SAMPLE PREPARATION OF NON-SEMICONDUCTOR MATERIALS, M. W. Phaneuf, Chipworks Inc, Ottawa, ON; G. Sundaram, Micrion Corporation, Peabody, MA; G. J. C. Carpenter, CANMET, Materials Technology Lab, Ottawa, ON; N. Rowlands, Micrion Corporation, Peabody, MA.

Focused Ion Beam (FIB) systems have been steadily gaining acceptance as specimen preparation tools in the semiconductor industry. This is largely due to the fact that such instruments are relatively commonplace as failure analysis tools in semiconductor houses, and are commonly used in the preparation of cross-sections for imaging under the ion beam or using an electron beam in an SEM. Additionally, the ease with which cross sectional TEM specimens of semiconductor devices can be prepared using FIB systems has been well demonstrated. However, this technology is largely unknown outside the semiconductor industry. Relatively few references exist in the literature on the preparation of cross-sectional TEM specimens of non-semiconductor materials by FIB. This paper discusses the use of FIB technology in the preparation of cross-sectional TEM specimens of non-semiconductor samples that are difficult to prepare by conventional means. One example of such materials is commercial galvannealed steel sheet that is used to form corrosion- resistant auto bodies for the automobile industry. Cross-sectional TEM specimens of this material have proved difficult and time-intensive to prepare by standard polishing and ion milling techniques due to galvanneal's inherent flaking and powdering difficulties, as well as the different sputtering rates of the various Fe-Zn intermetallic phases present in the galvannealed coatings. TEM results from cross-sectional samples of commercial galvannealed steel coatings prepared by conventional ion milling and FIB techniques are compared to assess image quality, the size of the electron-transparent thin regions that can be readily prepared, and the quality of sample produced by both techniques. Specimen preparation times for both techniques are reported.

4:00 PM *Z1.5 
SURFACE SCIENCE ASPECTS OF CONTAMINATION IN TEM SAMPLE PREPARATION, John T. Grant, Univ of Dayton, Research Inst, Dayton, OH; Scott D. Walck, Air Force Wright Laboratory, Materials Directorate, Wright Patterson AFB, OH; Frank J. Scheltens, UES Inc, Dayton, OH; Andrey A. Voevodin, Wright Patterson AFB, Materials Directorate, Wright Lab, Wright-Patterson AFB, OH.

With the advent of the modern analytical electron microscope with its ultrahigh vacuum conditions, extremely high current density in a very small probe, and its light element analysis capability, it is imperative that contamination of the TEM sample, particularly by hydrocarbons, is eliminated. The degree of cleanliness required for analysis using surface sensitive techniques is also much greater than for other forms of analysis. Contamination of samples can occur (a) prior to analysis, due to the way samples are prepared or handled, or (b) during analysis. Examples of surface contamination prior to analysis and during analysis will be given, in order to illustrate their effect on an analysis. Polished Si, Ni, and Ti bulk samples and pre-thinned TEM Ni and Ti samples were intentionally contaminated by two methods. One set of samples were put on a liquid nitrogen cooled stage for one hour in a diffusion pumped Gatan Duomill in which the liquid nitrogen cold trap above the diffusion pump was intentionally left warm. The other set was contaminated by dipping them into a 100 ml solution of acetone containing 0.04 g of an acetone soluble low temperature wax. Two commercially available plasma cleaners, specifically designed for eliminating hydrocarbon contamination, were used to clean the samples. The bulk samples were examined with x-ray photoelectron spectroscopy and the TEM samples were examined in a 200 keV analytical TEM. The results from these experiments will be discussed. This work was sponsored by the Materials Directorate, Wright Laboratory, Air Force Materiel Command, United States Air Force, Wright-Patterson AFB, Ohio 45433-6533, USA.

4:30 PM Z1.6 
OPEN FORUM FOR TEM SAMPLE PREPARATION, Ronald M. Anderson, IBM East Fishkill Facility, Dept of Analytical Services, Hopewell Junction, NY; Scott D. Walck, Air Force Wright Laboratory, Materials Directorate, Wright Patterson AFB, OH.

Following the platform presentations, an open forum will be held for topics relating to specimen preparation for transmission electron microscopy of materials. The panel will consist of the invited speakers for this symposium and the forum will be moderated by the organizers. An overhead projector with blank viewgraphs and pens and a slide projector will be available for the audience and panel members to present problems, solutions, and ideas. The topic areas are open, but are limited to the area of TEM specimen preparation, cleaning, or handling. Suggested topics include but are not limited to the following subjects: FIB milling, ultramicrotomy, dimpling, ion milling, cleaving, electropolishing, reactive ion etching, instrumentation, plasma cleaning, specimen artifacts, and specific material systems. Audience members are encouraged to come prepared with slides or viewgraphs.

Chairs: Ronald M. Anderson and Scott D. Walck 
Wednesday Evening, April 2, 1997
8:00 P.M. 
Salon 7


A small-angle cleavage technique, based on a method originated by McCaffrey[Mat. Res. Soc. Symp. Proc. Vol.254 (1992) P109], has been developed to prepare both cross-sectional and plan-view samples of both semeconductors and thin films of amorphous and crystalline materials on silicon substrates. In many cases, cross-sectional and plan-view samples of the same material can be produced and mounted together on the same TEM grid greatly facilitating TEM characterisation of the materials. Examples of specimens cleaved from both ion implanted bulk silicon and thin films of a number of materials on silicon substrates are presented to illustrate the versatility of the technique.


Characterizing dopant profiles in semiconductor devices is becoming more important for fabricating highly-integrated ULSIs. Various efforts to characterize two-dimensional dopant profiles have been reported; however, the spatial resolution reported in previous studies was insufficient for characterizing commercial Si devices. Here, we present a high spatial resolution technique for observing two-dimensional dopant profiles in commercial silicon devices using FIB and TEM. 
This method is based on the preferential etching of doped region in silicon by HF/HNO mixture. As a result of this etching, a two-dimensional dopant profile can be observed as a thickness fringe in the TEM image. 
To investigate the etching conditions and detection limit of this method, preferential etching was performed on a cross-sectional TEM sample of an As-implanted Si wafer. The contrast of observed thickness fringe agreed well with a simulation based on two-wave approximation using a dopant profile obtained by SIMS. The detection limit of the dopant concentration in this method was determined to be about 1 x 10 atoms/cm. This method was applied to a commercial Si device, and the thickness fringe was also observed in its diffusion layer. The TEM image indicated a dopant profile with a spatial resolution better than 100 nm in both the lateral and vertical directions. This spatial resolution is sufficient for characterizing commercial Si devices. In addition, we examined sample preparation using a focused ion beam (FIB) instead of Ar ion milling. The dopant profile in the commercial device was then observed in the sample prepared by FIB. Applying FIB enables us to specify certain areas, such as the faulted cell of DRAM. We also discuss the detailed sample preparation conditions for FIB.

(Abstract transferred to Z1.3a)


XTEM investigations are an important tool to characterize geometry, structure and microchemical composition of semiconductor devices. Over the last 25 years, several thinning procedures have been used working with Ar ions at about 5 keV. Common for all is that the thinning takes several hours or the ion thinning has to be combined with troublesome mechanical preparation steps, like dimpling. The newly developed technique uses Ar ions up to 15 keV and with a total ion current up to 0.2 mA. A maximum sputtering rate of 25 m/h for Si is reached at an incident angle of 6. A low energy thinning step at 3.5 keV is used at the end to reduce the amorphization and heating of the thinnest regions. The combination of this fast ion thinning with simpler mechanical preparation requiring only a final lapping step, without the need for dimpling and polishing, resulting in a yield of nearly 100 and a preparation time of about 3h.


In tool industry, as in many other industries, modification of the outmost layers have become one of the only ways to prolong the lifetime of the individual components. There exist several ways to modify the surface. One way is to grow thin layers of a hard ceramic coating on top of the base material. In the development of the hard coatings, knowledge of the microstructure and the local chemical composition are of great importance. One of the most powerful tools to obtain this information is cross-section TEM. The sample preparation time is often a factor that limits the use of X TEM. However, a proper design of the substrate can cut the total time used to prepare the X-TEM samples. This paper present a substrate sample design which, along with the low angle ion milling technique by Barna, has proved to be fast, reliable, and giving large areas appropriate for X-TEM of both the coating, the interface, and the substrate. The way the X-TEM samples are prepared has reduced the problems with the magnetic properties of high-speed steel when the X-TEM sample is investigated in the electron microscope.

A DETAILED PROCEDURE FOR RELIABLE PREPARATION OF TEM SAMPLES USING FIB MILLING, David Hung-I Su, Philips Semiconductors, Materials Analysis Group, Sunnyvale, CA; Howard Takeshi Shishido, LSI Logic, Milpitas, CA; Feng Tsai, Philips Semiconductors, Materials Analysis Group, Sunnyvale, CA; Liya Liang, Precision TEM Inc, Fremont, CA; Florence Carmen Mercado, Philips Semiconductors, Materials Analysis Grp, Sunnyvale, CA.

Although many papers have been written about focused ion beam (FIB) preparation of TEM samples, few have presented a detailed, step-by-step milling procedure. This is a summary of the techniques that have evolved over the past few years in our laboratory. In addition to describing more traditional mechanical prethinning techniques, we introduce a method to prethin samples within 1 hour down to thicknesses on the order of 20 m using a wafer dicing saw. We then discuss different ways to handle mechanically difficult samples such as those prone to delamination. Our approach to FIB milling is designed to minimize the effects of ion-beam spreading which is responsible for most of the failures in preparing good TEM samples. The technique is presented in a step-by-step fashion, including a simple yet reliable method to terminate FIB milling. Examples are shown to illustrate applications to different types of problems, including precision cross-sectioning of integrated circuit (IC) devices, cross-sectioning of samples prone to delamination, and cross-sectioning of specific defect sites. Finally, we discuss the effect of artifacts in the quality of TEM samples.

THE PREPARATION OF SUBMICRON PRECISION CROSS SECTIONS BY DIMPLING WITH A 'FLATTING TOOL', Helen L. Humiston, Philips Semiconductors, Dept Matls Analysis Group, Sunnyvale, CA.

The complex materials systems in VLSI devices require specialized preparation techniques for TEM microstructural analysis. For this purpose, it is desirable to obtain electron transparency in all material layers from the oxides used in dielectrics to refractory metals such as tungsten. The primal advantage of dimpling these materials is that ideal specimens are obtained for low angle ion milling. By dimpling both sides of the cross section with a padded flatting tool, a thicker specimen of 130 m at the outer rim of the 3 mm disc is produced that narrows to the 125 nm thickness fringes in the center of the disc. These samples do not require a copper support grid, thereby allowing for a lower milling angle of 2.5 degrees. This technique provides a cross section that is electron transparent in all layers without the loss of the oxides due to differential thinning rates of various materials at higher ion milling angles. It is generally thought that precision dimpling through submicron features is not possible on the dimpler. However, a simple step-by-step procedure for this technique will be demonstrated and discussed.

APPLICATIONS OF REACTIVE GAS PLASMA CLEANING TECHNOLOGY IN MINIMIZING CONTAMINATION OF SPECIMENTS DURING TRANSMISSION AND ANALYTICAL ELECTRON MICROSCOPY, Shane P. Roberts, South Bay Technology Inc, San Clemente, CA; Nestor J. Zaluzec, Argonne National Laboratory, Materials Science Div, Argonne, IL; Scott D. Walck, Air Force Wright Laboratory, Materials Directorate, Wright Patterson AFB, OH.

The generation and usage of gaseous plasma for a wide range of applications has been sited since the early 1970's (Thomas, 1974). More recently the use of a plasma generating system has been applied to analytical transmission electron microscopy to minimize and, in some cases, eliminate the problems associated with various contamination sources, including the specimen holder and the specimen itself. (Zaluzec, US Patent 5,510,624). Although the technology is well-known, no definitive characterization of process parameters has been developed for specimen and specimen holder cleaning applications. An investigation of the effects that power levels and gas mixtures have upon contamination rates and removal were done using a Philips EM420T. Measurements of contamination rates both prior to and following plasma cleaning were done to characterize the effects of various parameter changes. Results of different process parameters and contamination rates will be reported.

ULTRAMICROTOMY: MINIMIZING THE CUTTING FACE OF DIFFICULT SAMPLES WITH THE TRIPOD POLISHER, Scott D. Walck, Air Force Wright Laboratory, Materials Directorate, Wright Patterson AFB, OH; Shane P. Roberts, South Bay Technology Inc, San Clemente, CA; Pamela F. Lloyd, UES Inc, Dayton, OH; John T. Grant, Univ of Dayton, Research Inst, Dayton, OH.

Ultramicrotomy as a technique for preparing cross sectional TEM samples in the physical sciences has become more common in characterization laboratories. It is a relatively straightforward technique and most researchers with friends in the life sciences have access to it with the ususal caveat that they buy their own diamond knives. Substrates which are both hard and tough are difficult to microtome because too much material is exposed in the cutting face of the sample. The typical result is that the sample jumps during the cut and often damages the knife. For successful microtoming of difficult samples, both the thickness of the substrate and the length along the stroke should be minimized. To minimize the stroke length, samples to be prepared were cut to an apex using a diamond cutoff saw. The apex can be carefully placed on the sample, adding a degree of site-specific capability to the technique. To minimize the thickness of the substrates, the samples were polished to a wedge at this apex using a Tripod Polisher. Results of different apex angles used and materials systems will be reported.

THE EFFECTS OF GAS COMPOSITION ON THE ION MILLING OF CROSS SECTIONAL TEM SAMPLES CONTAINING CARBON LAYERS, Scott D. Walck, Air Force Wright Laboratory, Materials Directorate, Wright Patterson AFB, OH; Frank J. Scheltens, Scott D. Apt, UES Inc, Dayton, OH; Josekutty J. Nainaparampil, Air Force Wright Laboratory, Materials Directorate, Wright PattersonAFB, OH.

Two separate groups in our laboratory developed a common problem with respect to the preparation of cross sectional TEM samples with carbon coatings. One sample was diamond- like carbon films deposited on silicon by pulsed laser deposition and the other was a carbon coating on silicon carbide fibers. The problem was during ion milling, the carbon has a much slower milling rate than the Si, SiC, and the epoxy resin. As a result, the substrates were thinned much more rapidly than the carbon films. A solution suggested by several people on the Microscopy Listserver among others was to reactively ion mill the samples with a 20-25% oxygen-argon gas mixture. Is this the optimal ion milling condition and is Ar the best inert gas to use? The maximum relative energies imparted to a Si and C atom by an Ar ion is 97% and 71%, respectively, while for Ne, it is 97% and 93.5%. To determine whether the mass of the inert gas is important in balancing the milling rates, five gases were used to mill layers of pulsed laser deposited diamond-like coatings on Si. The gases were Ar, Ne, 25% O%Ar, 25%O%Ne, and dry air. Samples polished flat using the Tripod Polisher were ion milled using typical conditions and the surfaces examined by atomic force microscopy and compared. TEM samples were prepared first by dimpling and then by using the same ion milling conditions as the bulk samples. These were then examined in a JEOL 2000FX TEM.

THE SMALL ANGLE CLEAVAGE TECHNIQUE: AN UPDATE, Scott D. Walck, Air Force Wright Laboratory, Materials Directorate, Wright Patterson AFB, OH; John P. McCaffrey, National Research Council of Canada, Inst for Microstructural Sciences, Ottawa, CANADA.

The small angle cleavage technique is a relatively simple and inexpensive method of producing superior cross sectional TEM specimens. For speed of preparation, it is unsurpassed. One limitation is that the technique does require the substrate material to cleave or fracture. For this reason, it has been applied almost exclusively to semiconductor materials. Recently, the technique has been extended to other substrates such as glass, silicon carbide, quartz, sapphire, and other hard materials. It is particularly well suited for rapidly examining coatings and thin films very soon after they are deposited. This paper will report on the applicability of the technique to these types of materials. Examples of the technique applied to multilayer films will also be shown. Several procedures have been added or modified to simplify the technique. For example, a method for mounting the cleaved samples utilizing a commercially available grid is presented. New procedures for preparing amorphous or non-cubic substrates will also be discussed. In addition, the advantages that the special geometry of the prepared samples have when mounted properly in a double-tilt holder will be discussed with respect to the angular range of tilting experiments that are now possible in the TEM.


Use of the focused ion beam (FIB) systems for preparation of site-specific TEM specimens is currently a well established technique The procedure commonly includes two distinct steps: a) preparation of a 20-50 m thick sliver of the original sample containing the site of interest for TEM analysis through conventional grinding/polishing procedures, and b) FlB thinning of the prepared sliver to electron transparency at the above site. The first step of the procedure by itself is generally elaborate and in most cases requires up to few hours of preparation time. In this paper, a new FlB-based specimen preparation strategy is presented which eliminates the above mentioned first step (step a) of the procedure completely. The new technique is mainly based on the formation of the FlB-thinned ''wall'' at the site of interest within the as-received bulk sample followed by its removal and direct mounting on a support grid for TEM study. The new strategy requires a number of modifications to the routine FlB-thinning procedure as well as some new instrumentation for the FIB system. These details as well as other critical information have been presented and discussed.


Wedge-polishing or angle-lapping of materials to electron transparency for the purpose of TEM analysis is now a routine and popular TEM specimen preparation technique. Although initially employed primarily for the preparation of specimens from semiconductor materials and devices, the technique soon proved to be applicable to a wide variety of other materials and material systems. The original tool for this purpose was designed and developed about a decade ago. While this tool is still adequate for its intended application, a number of improvements have been suggested by different users since its introduction. The most notable among these modifications have been initiated by the group involved in development of the original tool. 
A new and improved commercial version of the original tool for angle lapping to electron transparency has been developed and presented in this paper. These improvements were conceived and implemented mainly as a result of a systematic analysis of the original tool design, the shortcomings and complications encountered during the routine hands-on use of the tool, requirements for dedicated major auxiliary equipment in conjunction with application of the tool and, finally, the input from other users of the tool. The basis for each added and/or improved feature of the new tool has been individually discussed and the resulting advantages have been reviewed.

COMBINED TRIPOD POLISHING AND FIB METHOD FOR PREPARING SEMICONDUCTOR PLANAR SPECIMENS, Ronald M. Anderson, Stanley Klepeis, IBM East Fishkill Facility, Dept of Analytical Services, Hopewell Junction, NY.

Tripod polishing and FIB methods are both suited to preparing specific-site TEM specimens of semiconductor devices in cross section. Combining the two methods by polishing a specimen to less than 10 microns in thickness before final polishing in the FIB has been shown to reduce or eliminate many problems associated with traditional FIB specimen preparation. We have developed a new tripod and FIB method for planar TEM specimen preparation that allows the production of truly unique specimens that can not be fabricated by any other means. 
Two tripod polishing steps are performed: The first to polish a cross section plane near, but not into, the desired area. The second polishing operation removes material from the back of the device region so that the specimen is less than 10 microns in thickness. After mounting on a half-grid, the specimen can be FIB processed to TEM transparency in any number of locations on the exposed cross section face-yielding one or many plan views. By using this protocol, a succession of plan views can be prepared next to each other that show different levels of a complex semiconductor device from the metallurgy / insulator stack above the substrate down into the substrate at various levels. This technique has proven invaluable for imaging crystal defect or ion implantation effects at known depths below the substrate surface.

A NEW TRIPOD POLISHER METHOD FOR PREPARING TEM SPECIMENS OF PARTICLES AND FIBERS, Ronald M. Anderson, Stanley Klepeis, IBM East Fishkill Facility, Dept of Analytical Services, Hopewell Junction, NY.

The tripod polisher's original application was the preparation of cross section TEM specimens from silicon semiconductors. One of the reasons that made this protocol a success is the fact that thin Si displays a series of colors with which a preparer can judge the thickness and wedge angle of the specimen as it is being polished. Optical fringes in the Si are used in the final stages of polishing to finish a specimen that often does not need ion milling. When non-Si specimens are prepared, there are often no visual clues to signal specimen thickness and wedge angle. The technique that we have developed overcomes this difficulty as well as the difficulties attendant with handling small fragments, particles, and fibers. 
The method consists of ultrasonically drilling holes in Si substrates using a modified dimpling accessory on an ultrasonic disc cutting tool. The size of the drilled hole is adjusted as a function of the dimensions of the specimen being prepared. The specimen is inserted into the hole and cemented in place with epoxy. Fragments and particles are inserted into blind holes in an epoxy slurry. Fibers plus epoxy are bundled and pushed into through holes. Tripod polishing proceeds as normal using the color changes in the surrounding Si to judge polishing progress. We will show examples of preparing very small laser diodes using this method as well as powders and both longitudinal and cross sections of fiber specimens.

PREPARATION OF CROSS-SECTION TEM SPECIMENS OF SEMICONDUCTORS CONTAINING TUNGSTEN INTERCONNECTS USING CHEMICAL MECHANICAL POLISHING AND CHEMICALLY ASSISTED ION BEAM MILLING, AND USE OF AFM TO EVALUATE THE SUCCESS OF THESE PROCEDURES, Robert Jamison, Univ of California-Berkeley, Dept of Matls Science, Berkeley, CA; John Mardinly, David Susnitzky, Jian Duan, Intel Corp, Dept of Matls Technology, Santa Clara, CA; Carmen Matos, Sharon Darknell, Intel Corp, Santa Clara, CA.

CVD Tungsten interconnects have provided a significant technological advantage in improving semiconductor devices. However, evaluation of these structures by TEM presents two significant problems: The high atomic number (74) makes penetration by the electron beam much more difficult than in the surrounding matrix of silicon, SiO and aluminum-copper, and the high hardness and low sputtering yield of tungsten result in TEM specimens in which the interconnects are significantly thicker than the surrounding silicon, SiO and aluminum-copper. The ideal TEM specimen would have high atomic number phases thinned preferentially to the matrix of low- atomic number materials so that the electron transparency of the specimen had a minimal variation among the various phases. Chemical mechanical polishing (CMP) (using a tungsten-specific compound), and Chemically Assisted Ion Beam Etching (CAIBE) (using a perfluorinated gas) have the potential to overcome these problems, but it is difficult to evaluate and optimize these approaches when the only feedback is TEM observation of the finished specimen. We have used Atomic Force Microscopy (AFM) measurements to characterize the differential thinning of tungsten during conventional dimpling, chemically assisted dimpling, conventional argon ion milling, and Chemically Assisted Ion Beam Etching. AFM measurements from TEM samples prepared using these techniques demonstrate that in some cases the differential thinning of tungsten interconnects can be eliminated and even reversed. AFM data and TEM images from tungsten interconnects prepared using these preparation approaches will be included.

CROSS-SECTIONAL TEM SAMPLE PREPARATION METHOD USING FIB ETCHING FOR THIN-FILMS TRANSISTOR, Katsuhiro Tsujimoto, ITES Co Ltd, Shigaken, JAPAN; Nobuhito Miura, ITES Co Ltd, Dept of Surface & Material Science, Shigaken, JAPAN; Koutarou Kuroda, Nagoya Univ, Dept of Quantum Engr, Aichi-ken, JAPAN; Satoshi Tsuji, IBM Japan Ltd, Display Technology, Kanagawa, JAPAN; Hirosasu Saka, Nagoya Univ, Dept of Quantum Engr, Nagoya, JAPAN; Hiroshi Takatsuji, IBM Japan Ltd, Dept of Display Technology, Shiga-ken, JAPAN.

A rapid and precise sample preparation method using focused ion beam (FIB) etching has been developed for cross-sectional TEM analysis of thin film transistor (TFT). The Ga ions accelerated at different voltages and at different incident-beam angles have been applied during FIB etching. We have successfully prepared TEM specimens of long and fragile aluminum whiskers formed on thin aluminum films deposited on glass substrate, where a strong charge-up is built up during FIB etching. The effect of ion-beam-assist tungsten deposition prior to FIB etching is discussed. A whisker as long as 10 m can be successfully etched to the thickness of 200nm while keeping the original shape. Finally, the performance of this technique is discussed with special reference to application to other fragile locations to be etched on fragile substrates.

CROSS SECTIONAL TEM SAMPLE PREPARATION USING EBEAM LITHOGRAPHY AND REACTIVE ION ETCHING, Hyun Jin Cho, Stanford Univ, Cntr for Integrated Systems, Stanford, CA; Peter B. Griffin, James D. Plummer, Stanford Univ, Integrated Circuits Laboratory, Stanford, CA.

Simple method for obtaining cross sectional TEM sample of semiconductor devices using ebeam lithography and reactive ion etching was developed. The basic idea o f this technique is to form pillar or line type pattern thin enough to be transp arent with electron beam. The technique used the same procedure as in semiconduc tor process. In order to make this method simple, etching condition of silicon a nd silicon dioxide with photo resist mask was developed. Using this method XTEM sample of various kinds of silicon devices such as MOS transistor and isolation were obtained. Since the thickness of sample was thin enough, it was possible to get HREM image to observe an atomic level defects in the device. Using align function in ebeam machine, cross sectional TEM sample at specific lo cation was successfully achieved. We developed the method to obtain XTEM sample at specific location of silicon devices such as source drain region and gate oxi de region of MOS transistor and birds beak region of LOCOS isolation. The advant age of this technique is its ability to examine specific device which location a re pre determined by electric measurement.

THE USE OF A LOW-ENERGY GAS PLASMA FOR THE FINAL PROCESSING OF CONTAMINATION -FREE TEM SPECIMENS, Paul Fischione, E.A. Fischione Instruments Inc, Export, PA; Jan Ringnalda, Philips Electronic Instruments Co, Mahwah, NJ; Hendrik Colijn, Ohio State Univ, Electron Optics Facility, Columbus, OH; Michael J. Mills, Jorg M. Wiezorek, Ohio State Univ, Dept of MS&E, Columbus, OH.

The issue of specimen contamination is increasing at a rate proportionate to the utilization of high-brightness electron source Transmission Electron Microscopes. The trend in the transmission electron microscopy of materials science specimens is to utilize higher-voltage TEMs incorporating field emission gun technology. These FEG TEMs are capable of generating large electron beam current in a small area probe. Microanalysis of materials, especially semiconductor specimens, necessitates fine electron probes which tend to increase the migration rates of hydrocarbon contamination on the specimen's surface to the impingement point of the electron probe. The subsequent formation of carbon deposits oftentimes obstructs imaging and prevents acceptable analytical results from being achieved. By plasma cleaning the specimen, contamination is removed and the results obtained by high resolution electron microscopy (HREM) and analytical electron microscopy (AEM) using EDS or PEELS are greatly enhanced. Recent instrumentation developments have resulted in the application of a low energy, reactive gas plasma that chemically removes hydrocarbon contamination from both the specimen and the TEM specimen holder. Critical aspects of the high-frequency plasma creation, ion energy, location of the electrodes external to the plasma, process gas, and oil free vacuum technology will be discussed. Results for a wide range of specimen processing plasma parameters and their effects on various materials research specimens will be presented.

A NEW PROCEDURE FOR MAKING TEM SPECIMENS OF SUPERCONDUCTOR DEVICES, Yi Huang, Karl I. Merkle, Argonne National Laboratory, Matls Science Div, Argonne, IL.

The goal of the TED sample preparation is to obtain a sufficiently large thin area at the point of interest. This is not easy for superconductor devices with isolated junctions where the area of interest is a couple of microns in size. The conventional dimpling method can provide only relatively small thin area near the dimple. Therefore, it is very difficult to align the thin area with the tiny point of interest. To overcome this problem, instead of thinning only the center part of the sample, we use a special procedure to polish the whole sample disk to 5-15 micron, employing a precision polishing apparatus. After lifted off from the polishing sample holder, the sample foil is mounted between two single hole or slot grids and then ion-milled until the desired thin area is obtained. Because the whole disk is thinned, any area on the disk can be milled to an electron-transparent thickness. Hence the point of interest is relatively easy to reach. This procedure has been applied to prepare both cross-section and plan view samples of various superconductor devices. One example is the plan view sample of a YBCO grain boundary junction.