Metallurgical Cross Sectioning of Microelectronic   Packages for Optical Inspection and Electron Beam Analysis(2)

Robert James Burgoyne
Analog Devices, Inc.

SEM/EDS Analysis: Reflected light microscopes are the analytical tool most frequently used for the qualitative, and, at times, quantitative, examination of metallurgical cross sections. These instruments do, however, have limitations. The wavelength of the light source defines a useful magnification of about 1,000X, and a single focal plane makes imaging samples with significant topographical details nearly impossible. Superior depth of focus, considerably higher spatial resolution, and elemental identification and mapping capabilities make the scanning electron microscope the ideal complement to the reflected light microscope. This section of the paper will focus on the practical application of the scanning electron microscope to the examination of microelectronic package cross sections.

Specimens subjected to electron beam irradiation undergo several complex interactions with the electrons from the primary beam. This interaction volume yields a variety of emission signals that may be collected and processed by one of several chamber mounted detectors. Figure 9 offers a pictorial representation of a typical interaction volume and emission products that are created when a sample is subjected to primary beam excitation. A brief description of the signals that originate from within the interaction volume follows:

Figure 9

Secondary electrons (SE): Low energy electrons emitted from a given atom due to the ionization event that occurs when the primary beam electron collides with that atom. Created throughout the interaction volume as a result of inelastic (electron-electron) collisions, SE's typically have electron energy levels less than 50eV. As such, only those SE's created near the surface escape the sample; the remainder are absorbed by adjacent atoms.

Backscattered electrons (BSE): Electrons that pass close to, or encounter, the nucleus of a specimen atom and are subsequently deflected through large angular trajectories with a minimal loss of electron energy. Created throughout the interaction volume as a result of elastic (electron-nucleus) collisions, BSE's have an energy level that is, on average, about 20% lower than their initial accelerating potential. Because of this negligible loss of energy, BSE's are able to escape to the surface from deep within the interaction volume.

Characteristic x-rays: As the incident beam electron strikes an atom, a weak inner-shell electron is ejected. To maintain atomic stability, a higher energy, outer-shell electron fills the vacancy and simultaneously emits radiation whose value is equal to the difference between the energy levels of the two shells. The portion of energy lost by the decay of electrons as they move from outer to inner shells reflects the unique electron transitions of a given element and, as such, are referred to as a characteristic x-ray.

Continuum x-rays: When an incident beam electron interacts with the nucleus of an atom, it is decelerated and inelastically scattered. The resulting loss of energy can range from 0eV up to the initial accelerating potential of the electron at the emission source. This continuum, or white radiation, constitutes the major part of the x-ray spectrum onto which the characteristic x-rays are superimposed.

The scanning electron microscope is an effective analytical tool that can provide a wealth of useful information. A strong understanding of how a sample will react to electron beam irradiation is the only surefire way to achieve optimal results. Examination of the completed sections is fairly straightforward; however, scanning electron microscopy is subjective in that each analyst has their own way of conducting analyses to achieve the desired results. Nonetheless, the following guidelines and helpful hints should be considered when examining the sections.

Sample preparation: Ensure that the sample is clean and dry prior to insertion in the SEM vacuum chamber. If a conductive coating is not tolerable, apply conductive carbon or silver paint to the bulk of the epoxy and down the side to the metal sample holder. This practice will significantly reduce excess charging of the bulk mounting material. Further, understand that prolonged exposure to high vacuum (typically 10-7 Torr) will result in outgassing and shrinking of the epoxy, thus creating small gaps along the periphery of the package. This could be of concern if the sample is to be re-worked after examination of that specific plane of interest.

Coating: There are four interconnected reasons for coating a sample. In short, these are to: 1) improve the signal, 2) increase the spatial resolution, 3) provide conductivity so that excess surface charging can be shunted to ground potential and 4) minimize thermal damage. Coatings fall into two categories: carbon (C) and precious metals (Au, Au/Pd, Pt, etc.). The decision to apply a coating largely depends on the type of analysis being performed. (Remember that the features within microelectronic packages are huge when compared to the submicron features on the integrated circuit. Imaging is often performed at magnifications less than 20,000X; consequently, a 50Å - 100Å thick layer of conductive coating is imperceptible and recommended when X-ray analysis is not being performed.

SE Imaging: The majority of sample "viewing" will be accomplished using secondary electron imaging. This mode of operation is invaluable when the identification of surface structure is the principal objective for the analysis. When performing SE imaging, it is often best to use lower beam accelerating potentials (<7.5kV) to maximize surface topology resolution. This will not only reduce the amount of local charging on the surface of uncoated samples, but will provide better contrast in the image and photomicrograph due to the decrease in penetration of the incident beam. The application of a precious metal film to semiconductor and insulator materials will further enhance the quality of the image and photomicrograph.

Figures 10A and 10B illustrate the effect that accelerating potential has on image resolution and contrast. Shown is silver filled epoxy die attach that has been etched in oxygen plasma for five minutes. Figure 10A was imaged at 3.0kV accelerating potential, and Figure 10B was imaged at 25.0kV accelerating potential. Note the increase in surface resolution, decrease in edge charging effects and inherently better image detail at 3.0kV. Both images were taken at a magnification of 5,000X and at a 40° tilt angle to improve depth of field.

Figure 10A

Figure 10B

BSE Imaging: BSE imaging provides both atomic number contrast and topographical information about the sample being examined. Because BSE yield is significantly lower than SE yield, higher beam accelerating potentials (>15kV) are often needed to increase signal levels. Although charging is not an issue in BSE imaging, a conductive coating is recommended for the reasons described earlier. Any concerns that the image will be of a single contrast (i.e. that of the coating) are unfounded; the vast majority of BSE's are generated well below the thin conductive coating layer.

The number of BSE's generated in a given area of a sample is directly proportional to the average atomic number of the material in that area. BSE images of non-homogeneous samples will therefore inherently show dramatic contrast variations. For example, in BSE images of gold-silicon (Au-Si) eutectic, the Au phases (Z=79) will appear much brighter than the Si phases (Z=14). This atomic number contrast information can prove invaluable when "mapping" different materials in a sample or conducting EDS analysis.

Figures 11A and 11B illustrate the difference between BSE atomic number contrast (11A) and BSE topography (11B) modes of operation. Shown is the plastic multi-chip module sectioned for this paper. Note the more uniform grey shade and "textured" appearance of Figure 11B, whereas Figure 11A has significant contrast variations but lacks depth. Both images were taken at a magnification of 300X.

Figure 11A

 Figure 11B

Figure 12 shows the various layers found within a ceramic sidebraze package. Shown, from top to bottom, is the silicon chip, gold/silicon die attach, nickel and cobalt plating layers, sintered tungsten refractory and the ceramic package. The sample was imaged in a SEM, in BSE mode, at a magnification of 1,250X.

Figure 12

Figure 13 shows a plastic leadless chip carrier (PLCC) soldered onto a multi-layer printed circuit board (PCB). This assembly was sectioned in an area that did not include the silicon chip, thus making it an "all soft materials" sample. Shown clearly is the bottom of the plastic package; the solder dipped, copper core package lead; the multi-layer PCB; and the solder joint. (Note that the epoxide fully encapsulated the assembly with no air bubble formation in the restricted area between the lead and package body. This further illustrates the benefits of using both a vacuum impregnation chamber and a pressurization vessel during encapsulation.) The sample was imaged in a SEM, in BSE mode, at a magnification of 250X.

Figure 13

X-ray Analysis: When elemental identification (EDS) is required to complete an analysis, several factors must be taken into consideration. First is the accelerating potential. A general rule of thumb is to select an accelerating potential that is at least 2.5 times the primary x-ray line (in keV) of the heaviest element being detected. For the vast majority of packages, including the one that will be used as an example, that element is lead (Pb), common to most solders and frit seals. The Pb La line appears at 10.55keV; thus, based on this line, the accelerating potential should be set at approximately 26.5kV. However, lead has numerous emission peaks due to its high atomic weight. The Pb Ma line appears at 2.34keV, thus reducing the recommended accelerating potential substantially. This is beneficial when analyzing samples that are mainly nonconductive. Severe local charging can actually deflect the beam away from the primary area of interest, resulting in erroneous data collection. Lowering the kV setting, without compromising the detectability of all potential elements, is therefore highly desirable. If appropriate, a high accelerating potential can initially be used to identify all potential elements present. The accelerating potential can then be lowered to not only reduce charging, but to improve the spatial resolution of x-ray energies at the lower end of the spectrum.

Figures 14A and 14B represent the spectrum obtained when a ceramic sidebraze package (14A) and plastic multi-chip module (14B) were subjected to a whole package EDS analysis.

Figure 14A

Figure 14B

Conductive coating is another factor that needs to be addressed. It is always best to perform an initial x-ray analysis without a conductive coating. This preliminary analysis provides a true and accurate qualitative assessment of all elements contained within the package. When coating is necessary, select carbon, as it is the preferred conductive coating when performing x-ray analysis. It generates only a single peak (~0.25keV) at the low end of the spectra.

A third issue is the scanning mode of the microscope. Most SEM's have three basic modes of operation. These are: 1) Normal scan: the area being scanned is that which is displayed on the screen; 2) Window or partial field: the area scanned is a variable percentage of the total screen area; and 3) Spot: the X & Y deflection coils are fixed at a controllable DC level, thus creating a moveable, single point electron microprobe. Normal scan is generally used when analyzing whole package sections, whereas window or spot mode is used when specific areas, such as plating layers, intermetallics, etc. need to be analyzed.

X-ray mapping is a versatile mode of x-ray analysis used to detect and display the spatial distribution of one or more specific elements in a non-homogeneous sample when multiple elements are being detected. In this application, a specific element is initially selected for detection and display. When the given element is detected, a bright spot is displayed on the SEM screen at the location that correlates to the detection site. Numerous scans will generate a pattern of light regions representing the relative distribution of the selected element. This procedure can be repeated for several different elements, resulting in a series of x-ray maps showing the distribution of all elements of interest.

Figure 15 illustrates the results of an x-ray map performed on the die attach area of the plastic multi-chip module sectioned for this paper. Shown are: 1) a SE image of the site, 2) a silicon (Si) map, 3) a copper (Cu) map and 4) a nickel (Ni) map.

Figure 15

Artifacts: Microelectronic packages are no different than any other complex sample prepared using metallographic techniques in that improper preparation can lead to induced artifacts. Table 5 details the artifacts that microelectronic packages are more likely to incur and suggestions on how to correct them. The recommended corrective actions listed below are by no means exhaustive and the analyst may have to resort to a "trial and error" method to determine the appropriate course of action.

Table 5

Conclusions: Proper implementation and utilization of metallographic sample preparation techniques will provide valuable information about the overall quality of microelectronic package components and the assembly process. The data obtained by metallographic sectioning, and subsequent reflected light and electron beam analysis, can be a critical part of failure analysis activities, process and package qualifications and overall quality control. By always considering the properties of all materials in a microelectronic package throughout the sample preparation sequence, the analyst is able to produce deformation free cross sections that can provide a wealth of information not easily obtained from any other analytical technique.

Acknowledgements

The author would like to take this opportunity to thank James Griffin, Hardie Macauley and Andrew Olney at Analog Devices, Inc. and James Nelson at Buehler, LTD., for their editing services, and Bruce Fried at Analog Devices, Inc., for his ultrasonic imaging and radiography support.

References

1) B. Bousfield, "Surface Preparation and Microscopy of Materials", John Wiley & Sons, LTD., West Sussex, England, (1992).
2) "Metallography Principles and Procedures", Leco, Corp. (1992).
3) "Buehler Dialog, Microstructural Analysis Reference Manual", Buehler, LTD.
4) "Metalog Guide, Your Guide to the Perfect Materialographic Structure", Struers Corp. (1992).
5) S.N. Chanat, A.M.H. Trujillo, D.C. Zipperian, Buehler Digest, Vol. 27, No. 1, "Advanced Techniques for High Speed Precision Sectioning", Buehler, LTD, (1993).
6) B.L. Gabriel, "SEM, A User's Manual for Materials Science", American Society for Metals, (1985).
7) A.M. Glauert, A.W. Agar, R.H. Anderson, D. Chescoe, "Principles and Practices of Electron Microscope Operation", North-Holland Publishing Company, Amsterdam, (1974).

Appendix
　

Table 3 (Process Description for Ceramic and Metal Alloy Packages)

Table 4 (Process Description for Plastic Packages)

(1)  (2)