Standard Article

You have free access to this content

High-Resolution 3D Imaging and Material Analysis with Transmission X-ray Microscopy and Nano-CT

X-Ray Techniques

  1. Yuxin Wang

Published Online: 18 MAY 2012

DOI: 10.1002/0471266965.com133

Characterization of Materials

Characterization of Materials

How to Cite

Wang, Y. 2012. High-Resolution 3D Imaging and Material Analysis with Transmission X-ray Microscopy and Nano-CT. Characterization of Materials. 1–10.

Author Information

  1. Argonne National Laboratory, Advanced Photon Source, Argonne, IL, USA

Publication History

  1. Published Online: 18 MAY 2012

1 Introduction

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

Although x-ray radiation has been used for imaging since its discovery (Röntgen, 1896), x-ray microscopy is a relatively new imaging technique that experienced rapid development during the past two decades (Schmahl et al., 1994; Kirz et al., 1995; Howells et al., 2008). The key difference between x-ray microscopy and the widely used projection imaging techniques used in medical imaging and security scans is that x-ray microscopes use x-ray lenses to magnify the images and achieve higher resolution. To date, close to 10 nm has been achieved with sub-kiloelectron-volt “soft” x-rays (Chao et al., 2005; Attwood, 2000) and better than 20 nm has been achieved with multi-kiloelectron-volt “hard” x-rays (Vila-Comamala et al., 2009). Compared with established imaging techniques based on electrons and visible light, x-ray techniques offer several advantages that make them uniquely well suited for nondestructive material characterization:

  1. Large penetration depth of up to millimeters through ceramics or light metals (Henke et al., 1993) (Fig. 1) so that devices can be imaged with minimum need of preparation and modification.

  2. No charging effect and much lower radiation damage than electrons. Devices can often remain fully functional when imaged at micrometer to tens of nanometer resolution (Colangelo, 2002; Sun et al., 1986; Chen et al., 1988).

  3. When combined with a computed tomography (CT) technique, they reproduce distortion-free 3D images that represent the devices' full structures, including both superficial and internal features (X-Ray Computed Tomography) (Haddad and Trebes, 1997).

  4. Sample's elemental compositions can be accurately mapped in 3D with the use of their absorption edges. Furthermore, chemical compositions can also be mapped by combining the x-ray CT and x-ray absorption near-edge spectroscopy (XANES) techniques (Liu et al., in press).

thumbnail image

Figure 1. Attenuation length (1/e) of several materials as a function of x-ray energy.

The combination of these attributes provides a unique set of capabilities to directly image and analyze three-dimensional nanostructures in situ, and track its dynamic structural changes in elemental and chemical composition. The strengths of this technique complement those of electron (Electron Techniques, Introduction) and visible light (Optical Microscopy) techniques in resolution scale, sample preparation requirements, sample environment, and material analysis capabilities as summarized in Table 1.

Table 1. Comparison of X-Ray Imaging Techniques Based on Visible Light, X-Rays, and Electrons
 Light MicroscopyTXMTEM
Imaging beam energyA few electron volts250 eV to 50 keV10 keV to 1 MeV
2D resolution200 nm∼25 nmAtomic resolution
3D resolution (transverse × longitudinal)Optical sectioning (200 × 500 nm)Computed tomography (50 × 50 nm)Computed tomography (5 × 5 nm)
Sample thicknessDepends on sample transparencyUp to hundreds of micrometers to many millimeters<1 μm
Sample preparationSimple, routine in situ and in operando observationSpecialized. Sectioning and metal coating required
Radiation damageNoneNegligible for inorganic samplesHigh
Imaging environmentRoom condition for most samples. Compatible with in situ high-temperature, cryogenic, high-pressure, and aggressive chemical environmentVacuum required
Material analysisIndirect observation by color and polarizationElemental mapping, XANES spectroscopyEELS
  X-ray fluorescence 
  X-ray diffraction 

Currently, several commercial companies produce high-resolution x-ray imaging systems with submicrometer resolution. The most common approach uses a projection imaging scheme where samples are placed very close to a nanofocused x-ray source and far from the detectors to achieve high magnification (Mayo et al., 2005; Roth et al., 2008). The resolution of such a geometry is roughly the size of the x-ray source. Because of the lack of a lens to collect the x-ray beam within a finite numerical aperture, the practical resolution is typically limited to 250 nm. With advances in nanofabrication technology, high-resolution Fresnel zone plate lenses have become practical for focusing x-ray radiation with tens of nanometer resolution. First developed at synchrotron radiation facilities in Europe (Schmahl et al., 1994) and the United States (Kirz et al., 1995), lens-based transmission x-ray microscopes (TXM) have been installed in several synchrotron radiation facilities around the world and commercial turnkey system using rotating anode x-ray sources are being marketed (Wang et al., 2002). Besides directly imaging nanostructures, the high brilliance and continuous tunability of synchrotron radiation also allows high-resolution images to be acquired at close energy intervals to produce an “energy scan.” Such spectral imaging techniques provide a unique means of mapping the elemental composition of samples as well as certain chemical bonding states (XAFS Spectroscopy). These synchrotron-based facilities are generally funded by government agencies around the world with the mission to support public research. Commercial TXM systems using laboratory x-ray sources provide better than 40 nm resolution in practical use, with much lower throughput and limited material analysis capability. However, they offer the convenience and availability of in-house systems, and are better suited for performing proprietary and classified imaging tasks.

Both scanning and full-field imaging x-ray microscopes have been developed. Scanning transmission x-ray microscopes (STXM) provide more versatile simultaneous imaging modalities such as absorption contrast, phase contrast for low-absorption materials, x-ray fluorescence for trace element mapping, and microdiffraction for structural imaging, but at significantly lower throughput and limited 3D imaging capability (Jacobsen and Kirz, 1998). Furthermore, because of the requirement of high-brilliance illumination, they are typically only operated at synchrotron radiation sources. Full-field TXM technique, however, is compatible with both laboratory and synchrotron sources, and generally provides three to five orders of magnitude higher throughput but with relatively lower 1%-level material analysis sensitivity. Typical data rate is within 1 s per 2D image at full resolution with synchrotron radiation facilities and 1 min with laboratory sources. For material characterization, particularly with inorganic materials and in situ characterization applications, multi-kiloelectron-volt hard x-ray radiation is generally needed. This article will primarily focus on the application of hard x-ray TXM systems.

2 Principles of the Method

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

2.1 Lens-Based X-Ray Imaging

The majority of the x-ray imaging systems in use today operate in the direct-projection geometry used by Röngten to record the first radiograph. In this setup, the sample is placed in the x-ray beam emitted from microfocus or nanofocus sources while the detector is placed behind the sample to record the x-rays passing through the sample. The resolution of these systems depends on the x-ray source spot size, magnification, and detector pixel size (Bonse, 2004; Brockdorf et al., 2008; Wang, 2004). Most commercial systems provide up to 1-µm resolution for 2D imaging and several micrometer resolution for 3D CT imaging. They have been widely used in imaging and analysis applications including medical imaging, security scans, industrial R&D, and scientific research. In order to achieve significantly better resolution in practical application, an x-ray lens is needed to magnify the x-ray images.

Figure 2 shows a typical TXM system setup. As a direct analog to the visible light microscope and Transmission Electron Microscopy, it consists of a condenser lens to project the x-ray beam onto the sample and an objective lens that creates a magnified image of the sample on the detector plane. The x-ray source can be an x-ray tube, rotating anode source, microfocused x-ray source, or synchrotron radiation facility. The detector typically consists of a scintillator screen that converts an x-ray beam into visible light, and a fiber-optic taper or visible light microscope objective to relay the visible light to a CCD camera. The x-ray lens typically provides 10–100 times magnification while the visible light system in the detector assembly provides an additional factor of 5–20 times magnification, thus giving a typical total magnification of between 100× and 2000×. For example, the TXM system at the Advanced Photon Source at Argonne National Laboratory has an x-ray magnification of 80× and visible light magnification of 10×, resulting in an overall 800× magnification.

thumbnail image

Figure 2. A schematic illustration of a TXM imaging system consisting of a condenser lens, objective lens, and detector system.

To date, the highest resolution was achieved with diffraction optics, Fresnel zone plate lenses. Illustrated in Figure 3, they are circular diffraction gratings consisting of a series of opaque or phase-shifting rings with decreasing spacing at larger radius in order to increase the diffraction angle and thus produce the focusing effect. As a grating, they produce multiple diffraction orders, and the first order produces the highest focusing efficiency and is generally used in focusing applications. The placement of the nth zone is described as

  • mathml alt image(1)

where inline image is the x-ray wavelength and inline image is the focal length of the first diffraction order. The numerical aperture of a zone plate lens is determined by the diffraction angle of the outermost zones:

  • mathml alt image(2)

where inline image is the outermost zone width. Therefore, the resolution of a zone plate lens based on Rayleigh criterion is inline image, and the focal length is

  • mathml alt image(3)
thumbnail image

Figure 3. Illustration of a Fresnel zone plate. A Fresnel zone plate is a circular diffraction grating with the grating period decreasing with increased radius. (a) The geometric structure of a zone plate and (b) an SEM image of a zone plate.

The depth of focus at diffraction-limited resolution is

  • mathml alt image(4)

before and after the focal plane. For example, a typical zone plate with 40 nm outer zone width and 80 µm diameter will provide roughly 50 nm resolution, 25.8 mm focal length, and 51 µm depth of field with 10 keV x-rays. The peculiar effect is that the resolution does not depend on the wavelength, but the focal length and numerical aperture do. But this is in fact what the Abbe theory predicts, as the zone plate behaves as a grating spatial filter and the same grating spatial frequency is required to analyze the same size features in the sample. The numerical aperture of this lens, 0.0015, is very small compared with typical visible light objective lenses. Thus, zone plate lenses are nearly ideal thin lenses with negligible primary aberration, except for chromatic aberration.

Since they focus the x-rays by diffraction, Fresnel zone plates require monochromatic illumination. As we see in the expression for zone plate focal length, it is a highly chromatic focusing device. The bandwidth required to achieve diffraction-limited resolution is inversely proportional to the number of rings (called zones) in the zone plates. For example, the same zone plate with 40 nm outer zone width and 80 µm diameter has

  • mathml alt image(5)

so that a bandwidth of 0.2% is required to achieve diffraction-limited resolution.

The zone plate shown in Figure 3 is a binary zone plate consisting of simple circular gratings in a stepped profile. At low sub-kiloelectron-volt x-ray energy range, they primarily act as opaque rings to block x-ray radiation, and the lens primarily functions as an amplitude zone plate. With the loss from the blocked x-ray beam and different diffraction orders, the maximum theoretically achievable efficiency is 10% in the first order. At higher x-ray energies, the phase shifting property of the rings becomes dominant and an ideal phase zone plate consisting of zones that shift the phase of the x-ray radiation by inline image can achieve a maximum theoretical efficiency of up to 40%. If the zones are shaped to the surface of a concave lens modulo inline image thickness of the zone material, the resulting lens is a Fresnel lens. In an ideal phase-shifting Fresnel lens, additional diffraction orders are suppressed so that up to 100% efficiency can be achieved at the designed x-ray energy.

As the zones must be able to block or phase shift the multi-kiloelectron-volt hard x-ray beam, they must be made of heavy metal such as gold or tungsten with sufficient thickness, often greater than 1 µm. Compared with the tens of nanometer zone width, this requires fabricating nanostructures with large height-to-width aspect ratio, as shown in Figure 3. In contrast, skyscrapers such as Willis Tower in Chicago (former Sear's Tower) typically have an aspect ratio of 15. Therefore, the challenge of fabricating zone plate lenses lies in fabricating such high-aspect-ratio nanostructures that are able to maintain stability during operation. The state-of-the-art zone plate techniques can achieve an aspect ratio of 15–20, and these high-aspect-ratio structures usually have relatively poor long-term stability and their efficiency often degrades with use.

The TXM systems typically operate in absorption-contrast mode, where the transmitted x-ray beam passing through the sample is directly magnified by the objective lens and recorded by the detector. In this mode, the images represent the integrated absorption through the sample. This absorption map is highly dependent on the x-ray energy as shown in Figure 1, and different material compositions can often be identified easily. Figure 4 shows two images of an integrated circuit sample used in electromigration testing acquired by an Scanning Electron Microscopy and TXM. Because of the lower penetration depth, the SEM images show primarily the superficial features so that damage to the metallization lines and via is not clearly visible. The TXM system, on the other hand, images through the entire sample structure to reveal both the top-level and embedded structures and defects. Furthermore, the tungsten plugs and copper lines can be readily identified by their attenuation differences.

thumbnail image

Figure 4. An integrated circuit test structure imaged (a) with SEM and (b) TXM. Slices extracted from the 3D reconstruction are shown in (c) lines, (d) vias, (e) pad, and (f) tungsten plugs. X-ray fluorescence images showing the Cu lines (g) and Ta/TaN layers (h). Resolution is 50 nm for all images. Images (b)–(f) from Xradia, Inc. and (g) and (h) acquired at 2-ID-B beamline at Advanced Photon Source, Argonne National Laboratory.

2.2 Three-Dimensional Imaging at Nanometer Scale with CT Technique

By combining the TXM with the CT technique, the full 3D structure of the sample can be obtained without physically modifying the sample (X-Ray Computed Tomography). This approach is similar to that of the medical CT, but with five orders of magnitude higher resolution. When the sample is contained in the depth of field, each image represents an integrated absorption map, that is, a parallel projection through the sample. A series of projections can then be acquired at different projection angles by rotating the sample relative to the optical axis of the imaging system. These projections provide perspectives of the sample from different view angles that fill the Fourier space along the projection angles (Binwks, 1980). They can then be mathematically reconstructed to form a 3D volume that represents the 3D structure of the sample.

The parallel projection nature of the TXM is the simplest CT imaging geometry. The high data acquisition speed of a TXM system allows many projections to be acquired within a short time. The combination of these two factors means that high-quality 3D data can be produced easily using established CT imaging techniques, and for in situ applications, at sufficient speed to capture real-time dynamics in 3D. Figure 5 shows a series of “virtual cross-sections,” that is, slices extracted from the 3D reconstruction of the sample in Figure 4 at different depth levels showing the metallization, via, pad, and tungsten plug layers. The defect location can then be clearly identified from these 3D images.

thumbnail image

Figure 5. Images of an SOFC sample acquired (a) below and (b) above Ni K edge. (c) Volume rendering of the 3D distribution of Ni structure determined from a series of differential absorption projections such as (a) and (b). (d) XANES spectra of pure Ni and several of its compounds (Grew et al., 2010).

2.3 Mapping Elemental and Chemical Composition

A powerful feature of x-ray techniques is their ability to identify specific elemental composition by their absorption edges and, in many cases, chemical composition by the XANES spectra. Referring to Figure 1, for example, at the nickel K absorption edge of 8.3 keV, pure Ni or Ni compounds exhibit an abrupt absorption change. This results from the core electron excitation by x-ray radiation: increasing the x-ray photon energy across the K edge provides the photons with sufficient energy to eject K-shell electrons from Ni atoms, thus leading to a rapid increase in the absorption cross-section. At the same time, the attenuation length of other elements remains relatively constant. Therefore, acquiring two images above and below the absorption edge and taking their ratio gives the map of Ni within the sample with this differential absorption technique.

An example from a solid oxide fuel cell (SOFC) is shown in Figure 5 with two images acquired below and above the absorption edge of Ni (Grew et al., 2010; Goodarzi and Huggins, 2001). As shown in Figure 1, just below the edge, the Ni attenuation length is about 26 µm but it suddenly decreases to about 3 µm just above the edge with a 20 eV shift, or 0.25% increase in energy. The ratio shows the map of the Ni within the sample with a precision up to the resolution of the optical system, roughly 30 nm in this case.

In addition to mapping the elemental composition, the fine spectral structure change resulting from an orbital shift when Ni atoms form different bonds, known as XANES, can be used to uniquely identify chemical bonding states, and subsequently track chemical reactions in in situ imaging applications (XAFS Spectroscopy, In situ X-ray Measurement Methods). This technique is illustrated in Figure 5d, where the absorption images of pure Ni foil and several of its compounds are compared. Although the absorption spectra of Ni and its compounds show the general trend of the Ni edge, their detailed structure differs significantly during the transition. For example, a useful feature for identifying Ni from NiO is an NiO peak at 20 eV and a valley at 30 eV from the edge that lead to a contrast reversal so that images acquired at these two energies will show an opposite intensity change. The ratio of these two images containing both materials will yield an elemental map image in which Ni and NiO show different signs.

When combined with the CT techniques, the elemental and chemical composition of samples can be mapped in 3D at tens of nanometer resolution. This technique provides a powerful means of nondestructive characterizations of not only device structures but also their functions by tracking composition changes and chemical reactions.

2.4 Complementary X-Ray Imaging Modalities

Additional x-ray imaging techniques such as x-ray fluorescence (XRF) and x-ray diffraction (XRD) techniques provide valuable information on the distribution of trace elements as well as structural and strain information that complements the 3D images obtained with the TXM system (X-ray Microprobe for Fluorescence and Diffraction Analysis). These modalities are often available only at synchrotron radiation facilities, such as the Advanced Photon Source operated by the same division at the TXM imaging system, so that a sample can be examined by multiple techniques within short duration.

The TXM technique typically provides 20–40 nm resolution at various facilities and 1%-level material sensitivity using elemental mapping and XANES imaging techniques. It is not sensitive to trace elements or structural features such as grain or strain. In contrast, XRF and micro-XRF techniques can provide up to parts per million (ppm) sensitivity in elemental mapping while structural information can be obtained by XRD and micro-XRD techniques. Combining information from these complementary techniques can therefore provide a more complete picture of the sample's structure and function. Because micro-XRF and micro-XRD techniques are performed in scanning mode, they are much slower (typically requiring 1000 times longer than TXM), and their resolution is often poorer—typically 100–300 nm in practical use today.

Figure 4g and h shows micro-XRF images of the integrated circuit electromigration test sample showing the Cu interconnects and Ta/TaN liners. The nano-CT images show similar Cu lines, but at higher resolution. However, the Ta/TaN liners are not visible in the nano-CT images. Combining the nano-CT and fluorescence techniques can provide a more complete picture that shows the behavior of different constituents.

3 Practical Aspects of the Method

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

3.1 Sample Selection Issues

Compared with other microscopy techniques, TXM techniques offer many unique and often complementary capabilities, but access is currently limited to several universities and national laboratories. Therefore, selecting the suitable types of sample for TXM studies and preparing them correctly is an important issue. The nondestructive nature of x-ray techniques allows a wide range of samples to be studied in great detail with little preparation and modification. However, samples best suited for TXM studies are the ones with extensive complex internal structures, samples that are very difficult to cross-section without introducing artifacts or damage, and whose structures must be studied in situ or environments not suitable for other techniques. Typical examples include embedded nanofabricated structures, porous structures such as fuel cell and battery devices, and biological system at the single-cell level.

Small surface structures are typically difficult to study with transmission x-ray techniques (but those in certain size range can be effectively studied with reflective x-ray techniques; Fenter et al., 2006), and samples with primarily superficial features are better studied with atomic force microscopy (AFM) or SEM with much better resolution and convenience of easy access. Thin samples or those that can be easily cross-sectioned can be studied with both TXM and TEM systems. TEM provides up to two orders of magnitude higher resolution as well as better access, but requires much thinner sections and a vacuum environment. In contrast, samples can be studied in a TXM system under a wide range of extreme environments such as a cryogenic state, high temperature and high pressure resembling earth's core conditions, aggressive chemical baths, and high magnetic and electric fields. With these different and often complementary imaging characteristics, an imaging study will depend heavily on the sample type and imaging conditions. Users considering using TXM and other microscopy techniques should consider their relative strength and limitation.

3.2 X-Ray Energy Selection

Since the TXM image contrast depends primarily on the absorption while the absorption is strongly energy dependent, optimizing the x-ray energy is critical for obtaining the best imaging quality. Energy selection is typically not possible with laboratory source–based TXM systems, but is a necessary process with synchrotron-based systems. For an example, to optimize the energy for imaging a sample containing nickel and copper, we look at the attenuation length graphs in Figure 1. Before the Ni absorption edge at 8.3 keV, Ni and Cu have comparable attenuation length so that there is little relative contrast between them. When the energy increases to between their absorption edges 8.3 and 9.0 keV, the Ni attenuation length decreases dramatically while the Cu attenuation length increases slightly. Therefore, the two materials exhibit excellent contrast within this energy range. However, when the energy is further increased to above the Cu edge, both Ni and Cu become highly absorbing, and the contrast between them diminishes. This example demonstrates that the energy selection is a critical process of TXM imaging and the user should study the contrast as well as overall sample thickness requirements as a function of energy to optimize the imaging conditions.

3.3 Access to Laboratory- and Synchrotron-Based TXM Systems

Commercial high-resolution TXM systems first entered the market in 2003 and the installed base is relatively small compared with other high-end microscopes such as TEM. Currently Xradia, Inc. is the only supplier of lens-based TXM systems. A large proportion of these systems is available at university microscopy centers, such as University of Illinois at Urbana-Champaign and Virginia Tech. These systems are generally open to both university and industrial users. Laboratory-based TXM systems typically use rotating anode x-ray sources with Cr (5.4 keV) or Cu targets (8 keV). Up to 40 nm resolution has been achieved routinely with exposure times of several minutes for each 2D image and hours for each 3D data set.

Synchrotron-based TXM systems provide much faster, often subsecond, exposure time with the additional spectral imaging capabilities. Synchrotron TXM systems are operated as public research facilities for both academic and industrial users free of charge for nonproprietary work and at a modest fee for proprietary use. Several TXM facilities with different energy ranges are available in the United States and around the world, for example, Advanced Light Source (ALS) at Lawrence Berkeley National Laboratory (LBNL) with 200–500 eV range, Stanford Synchrotron Radiation Laboratory (SSRL) with 4–14 keV range, and Advanced Photon Source (APS) at Argonne National Laboratory (ANL) with 8–17 keV range. Access to these facilities is based on the scientific merit of the project through a proposal process reviewed by an independent Scientific Advisory Committee (SAC). The time between proposal submission and access to the instrument is typically 1–2 months for proposals receiving best scores, but can be as long as 12 months for others. These facilities provide full staff support during operation and a user is typically allotted 3–5 days of access during each 3-month operations period.

4 Method Automation

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

The level of automation and the user-friendliness varies significantly from commercial instruments to synchrotron systems. Commercial systems provide highly automated instrumentation and a graphical user interface (GUI) similar to that of modern SEMs and confocal microscopes. Major functions such as alignment of key components, focusing, and different data acquisition modes operate either fully automatically or semiautomatically with minimum need of operator input. Synchrotron-based instruments are typically designed to maximize flexibility in order to accommodate a wide range of researchers. The level of automation is generally lower and users need to undergo extensive training before they can operate independently.

A key step of the nano-CT data acquisition that requires significant user input today is the process of aligning the projection data to a common rotation axis. Mechanical rotation stages typically have an asynchronous (or random) runout error of over 1 µm, significantly larger than the resolution of the TXM. As a result, as the sample is rotated to acquire projection data, the rotation error will lead to a shift in each image. Since reconstruction algorithms assume a common fixed rotation axis, these shifts will lead to significant resolution loss, artifacts, and distortions in the 3D image. In order to minimize these problems, the projections must be aligned before the reconstruction step. While many automated alignment algorithms have been developed, none has been demonstrated to perform with accuracy comparable to the imaging resolution for arbitrary samples. The only high-precision and general technique today is using fiducial marks, such as gold nanodots, on the sample. After acquiring the projection data, the fiducials are identified and marked within each image and aligned to a common coordinate. In effect, the fiducials are used as a proxy for a common rotation center. During this process, extensive operator input is needed to identify the fiducial mark in the image and this is often the most tedious part of the nano-CT imaging process. Several attempts have been made to overcome this problem. Xradia, Inc. has developed a metrology-based solution that constantly monitors the position of the rotation axis spindle and provides feedback on its shifts during rotation. This system has demonstrated measurement accuracy of 100 nm. Designs based on a high-precision air-bearing stage with runout limited to 20 nm have been developed at ESRF and Argonne National Laboratory. These developments have the potential to fully automate the nano-CT data acquisition and 3D reconstruction process with significantly improved 3D resolution and accuracy.

5 Data Analysis and Initial Interpretation

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

The key advantage of the TXM and nano-CT system is that it provides direct visualization of samples being imaged. In the absorption-contrast mode, the images represent an integrated absorption map through the sample, while in the phase-contrast mode typically implemented in the TXM system in the Zernike scheme, the image intensity is a monotonic function of integrated phase shift through the sample at each point. Nano-CT is typically performed in the absorption-contrast mode.

Three-dimensional reconstructions obtained in the nano-CT process can be analyzed as a volume rendering (Fig. 6a) or as cross-sections in arbitrary planes (Fig. 6b). The volume rendered 3D images provide a direct visualization of the 3D structure, where the morphology and relative location of features can be observed easily. However, cross-sectional slices are typically used to make more detailed observations, such as of a feature's exact size and margins. Modern 3D data viewing and analysis packages such as Amira and VG Studio typically provide both viewing options and convenient image adjustment and measurement tools.

thumbnail image

Figure 6. Micro-CT image of a cricket shown (a) as a volume rendering and (b) in cross-sections.

When analyzing 3D data with a high signal-to-noise ratio, features in the volume data can often be automatically segmented with intensity thresholds. However, when the signal-to-noise ratio is not sufficient for accurate automatic segmentation or when the sample contains complex structure with slowly varying intensity levels, manual segmentation is often necessary. In this process, the operator must work with each slice, identify features of interest in each slice, and digitally label all the pixels within the feature. Although most visualization software packages provide integrated tools to streamline this process, it is often tedious and time consuming.

6 Sample Preparation and Specimen Modification

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

Because of the high penetration power of hard x-ray radiation, sample preparation is relatively simple. The primary requirement of sample preparation is that x-rays transmitted through the sample's region of interest can provide sufficient flux for adequate signal-to-noise ratio for data analysis. A rule of thumb is that the total thickness of the sample does not greatly exceed the effective combined attenuation length of the sample including its enclosures:

  • mathml alt image(6)

where inline image are the attenuation lengths of each constituent in the sample and inline image are their proportions through the sample's region of interest.

In the cases where the samples are micrometer-sized particles, a convenient way to image them in a TXM system is to first deposit the particles on a thin substrate, such as a thin Si3N4 window or aluminum foil, etc. Aggregates of such materials can also be injected into a thin-wall glass tube. Organic substrates and adhesive, such as Kapton tape and epoxy, are prone to damage from the x-ray radiation during imaging at synchrotron radiation sources, but are generally reliable using laboratory sources. The samples can then be placed in specialized environments such as vacuum, inert gas, and chemical baths.

Thick samples can often be reduced by removing material from the substrate while leaving the active area intact. For example, microelectronics and MEMS devices can be prepared by removing most of the Si wafer substrate just under the region of interest without affecting the active devices, and SOFC devices can be thinned by removing most of the support without affecting the three-phase boundary, where the electrochemical reaction takes place. When samples are intended for 3D nano-CT imaging, a cylindrical sample is preferred so that it can be imaged at full 180° rotation. A planar sample shape will limit the angular sampling range depending on its thickness and the x-ray energy. For certain samples, the information from the missing angles can be supplemented with a priori knowledge so that the artifacts in the reconstructed 3D structure will be negligible. For example, integrated circuit samples are fabricated in layers and the interconnects generally have rectangular cross-sections. Using this knowledge, their 3D structures can be reconstructed accurately without a full 180° angular sampling range. When the thinning process is applied locally and accurately, the devices can often remain functional ex situ or in many cases in situ during the imaging process.

Commonly used techniques to locally thin the sample include the following:

  1. Dimpling is a mechanical grinding technique with precisely controlled depth and grinding rate. It can be used to prepare many devices at relatively high speed. A sample can be prepared with 10-µm scale precision within several hours. This technique has been highly refined and is widely used in the semiconductor industry and analytical services laboratories. Dimpling machines are also very affordable. A drawback of this technique is the risk of damaging fragile components, particularly when the final thickness is reduced to less than 50 µm.

  2. Focused ion beam (FIB) is able to shape a sample at tens of nanometer precision. This is currently the most commonly used technique to prepare precise cylindrical-shaped samples for nano-CT imaging. The main drawback of this approach is the low throughput, as removing a small (100 µm3) volume may take many hours. Although FIB systems are becoming increasingly more common in most universities and industry laboratories, the relatively limited access and high operating cost can limit the practical number of samples being processed.

  3. Laser ablation is often used to rapidly remove material at micrometer-scale precision. The main limitation of this technique is that the heat generated at the sample surface makes it unsuitable for a wide range of samples.

7 Problems

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

The process of imaging with TXM is a direct visual process very similar to visible light microscopes and TEM. However, the key difference is that contrast depends on the attenuation length and contrast reversals can often occur across elemental absorption edges. For example, referring to Figure 1, just below the Fe absorption edge at 7.1 keV, Ni is more absorbing than Fe, so that in an absorption-contrast image with a mixture of Fe and Ni, nickel components will appear slightly darker than iron. Tuning the energy above the iron edge with less than 0.5% energy change, however, will make iron far more absorbing, and almost significantly reverse the contrast in the images. Particularly when using TXM systems at synchrotron radiation sources, users are advised to calculate the relative contrast between different materials in the sample as a function of energy to both optimize image contrast and avoid interpretation errors.

8 Protocols

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

Samples imaged with TXM samples generally require a minimum level of preparation and since the imaging process is typically performed in air, the protocol is very simple for most users. The preparation steps are highly sample dependent, but in the majority of cases are simply intended to minimize absorption near the region of interest or local thinning. Prior to imaging experiments, the image contrast based on the absorption properties of the materials of interest should be calculated as a function of energy to optimize imaging conditions. With the imaging energy optimized, the sample is placed into the TXM sample stage for imaging. Focusing and field of view adjustment processes in the TXM system are generally highly automated and often assisted with an in situ visible light microscope.

When acquiring tomographic projections for 3D imaging, an additional alignment step is required to ensure the region of interest remains in the field of view at all rotation angles. With sample stage designs that include translation stages above the rotation, this process includes first aligning the rotation axis to the center of the field of view, and then moving the sample's region of interest to the field of view. If no translation stage is placed above the rotation stage for centering, users must determine the offset distance between the rotation axis and the region of interest, and move the sample stage to undo this offset for each rotation angle. This is typically done in a highly automated iterative process with both commercial and synchrotron instruments.

9 Summary

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

High-resolution x-ray microscopy is a new approach for nanoscale and mesoscale imaging. Its nondestructive nature and comprehensive material analysis capabilities make it uniquely well suited for in situ studies on dynamic behavior of nanostructures, including their elemental and chemical composition, while in their real operation conditions. These functionalities also provide valuable complementary analytical imaging capabilities to widely used visible light and electron microscopy techniques. In particular, coordinated application of these approaches to a single material can potentially lead to unprecedented insight into materials behavior over a wide span of length scales from atomic scale to mesoscale.

10 Acknowledgment

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References

Use of the Advanced Photon Source, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science by Argonne National Laboratory, was supported by the U.S. DOE under Contract No. DE-AC02-06CH11357.

Literature Cited

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References
  • Attwood, D. 2000. Soft X-Rays and Extreme Ultraviolet Radiation, Chapter 9. Cambridge University Press.
  • Binwks, R. 1980. Computational principles of transmission CT. In Medical Physics of CT and Ultrasound, pp. 3752. American Institute of Physics, New York.
  • Bonse, U. (ed.) 2004. Developments in X-Ray Tomography IV. SPIE, Wellingham.
  • Brockdorf, K., et al. 2008. Sub-micron CT: Visualization of internal structures. In Developments in X-Ray Tomography VI (S. Stock, ed.), Proceedings of SPIE, Vol. 7078, San Diego.
  • Chao, W., Harteneck, B. D., Liddle, J. A., Anderson, E. H., and Attwood, D. T. 2005. Soft x-ray microscopy at a spatial resolution better than 15 nm. Nature 435: 12101213.
  • Chen, T. C., Norum, J. P., and Ning, T. H. 1988. Comparison of effects of ionizing radiation and high-current stress on characteristics of self-aligned bi-polar transistors. In Proceedings of the 20th Conference on Solid-State Devices and Materials, Tokyo, p. 523.
  • Colangelo, J. 2002. Microelectronic Failure Analysis Desk Reference, 4th ed. ASM International, Ohio.
  • Fenter, P., Park, C., Zhang, Z., and Wang, Y. 2006. Observation of molecular-scale structures with x-ray reflection phase-contrast microscopy. Nat. Phys. 2: 700704.
  • Goodarzi, F. and Huggins, F. 2001. Monitoring the species of arsenic, chromium and nickel in milled coal, bottom ash and fly ash from a pulverized coal-fired power plant in western Canada. J. Environ. Monit. 3: 16.
  • Grew, K. N., Chu, Y. S., Yi, J., Peracchio, A. A., Izzo, Jr., J. R., Hwu, Y., De Carlo, F., and Chiu, W. K. S. 2010. J. Electrochem. Soc. 157: B783.
  • Haddad, W. S. and Trebes, J. E. 1997. Developments in limited data image reconstruction techniques for ultrahigh-resolution x-ray tomographic imaging of microchips. In Developments in X-Ray Tomography. SPIE, San Diego, CA.
  • Henke, B. L., Gullikson, E. M., and Davis, J. C.July 1993. X-ray interactions: Photoabsorption, scattering, transmission, and reflection at E = 50-30000 eV, Z = 1–92. Atomic Data Nuclear Data Tables 54 (No. 2): 181342.
  • Howells, M., Jacobsen, C., Warwick, T., and van den Bos, A. 2008. Principles and applications of zone plate x-ray microscopes. In Science of Microscopy (P. W. Hawkes and J. C. H. Spence, eds.), Chapter 13, pp. 835926. Springer, New York.
  • Jacobsen, C. and Kirz, J. 1998. X-ray microscopy with synchrotron radiation. Nat. Struct. Biol. 5 (Suppl.): 650653.
  • Kirz, J., Jacobsen, C., and Howells, M. Q. 1995. Soft x-ray microscopes and their biological applications. Rev. Biophys. 28: 33130.
  • Liu, Y., Andrews, J. C., Meirer, F., Mehta, A., Gil, S. C., Sciau, P., Mester, Z., and Pianetta, P. Applications of hard x-ray full-field transmission x-ray microscopy at SSRL. In Proceedings of the 10th International Conference on X-Ray Microscopy, Chicago, in press.
  • Mayo, S. C., Miller, P. R., Sheffield-Parker, J., Gureyev, T., and Wilkins, S. W. 2005. Attainment of <60nm resolution in phase-contrast x-ray microscopy using an add-on to an SEM. In Proceedings of the 8th International Conference on X-Ray Microscopy, Himeji, Japan, pp. 343345.
  • Röntgen, W. C. 1896. On a new kind of rays. Nature 53: 274276.
  • Roth, H., He, Z., and Paul, T. 2008. IC package inspection with nanofocus x-ray tubes and nanoCT. In 15th International Symposium on the Physical and Failure Analysis of Integrated Circuits, Singapore, pp. 13.
  • Schmahl, G., Rudolph, D., Schneider, G., Guttmann, P., and Niemann, B. 1994. Phase contrast x-ray microscopy studies. Optik 97: 181182.
  • Sun, J. Y. C., et al. 1986. Effects of x-ray irradiation on the channel hot-carrier reliability of thin-oxide re-channel MOSFETs. In Conference on Solid State Devices and Materials, Japan, p. 479.
  • Vila-Comamala, J., Jefimovs, K., Raabe, J., Pilvi, T., Fink, R. H., Senoner, M., Maassdorf, A., Ritala, M., and David, C. 2009. Advanced thin film technology for ultrahigh resolution x-ray microscopy. Ultramicroscopy 109 (11): 13601364.
  • Wang, Y. 2004. X-ray microtomography tools for advanced IC packaging failure analysis. In Microelectronic Failure Analysis Desk Reference, 5th ed., p. 261. ASM International, Ohio.
  • Wang, Y., et al. 2002. A transmission x-ray microscope (TXM) for non-destructive 3D imaging of integrated circuits at sub-100 nm resolution. In Conference Proceedings of the 29th International Symposium for Testing and Failure Analysis, Santa Clara, CA, pp. 227233.

Key References

  1. Top of page
  2. Introduction
  3. Principles of the Method
  4. Practical Aspects of the Method
  5. Method Automation
  6. Data Analysis and Initial Interpretation
  7. Sample Preparation and Specimen Modification
  8. Problems
  9. Protocols
  10. Summary
  11. Acknowledgment
  12. Literature Cited
  13. Key References
  • Howells et al., 2008. See above. Book chapter providing a general introduction to x-ray optics and microscopy techniques.

Liu et. al., 2010. See above.

Describes with elemental mapping and chemical speciation methods using nano-CT techniques along with several applications.

Sun et al., 1986. See above.

Provides discussion on radiation damage issues with high-resolution x-ray imaging.

Vila-Comamala et. al., 2009. See above.

Article detailing state-of-the-art high-resolution x-ray optics fabrication techniques and resolution measurement methods.