Carefully designed lens systems can probe structures down to angstrom resolution, enabling 3-D nondestructive inspection of optically opaque objects and highly detailed medical imaging.
William McGuigan and Robert E. Fischer
Several “big” lens systems have been developed and are being used in research and industry, allowing scientists to see features with greater resolution than ever before. The systems are used to re-image electron-, neutron-, and x-ray-bombarded scintillation screens onto low-light charge-coupled-device (CCD) detectors. Application areas cover a broad range of nondestructive testing and high-resolution medical-research imaging, including x-ray crystallography, electron-beam microscopy (specifically for brain imaging), and x-ray and neutron-beam inspection of optically opaque parts. The design of these lens systems is challenging and involves careful consideration of many competing requirements before launching into the lens and system design.
Common requirements
All applications for these systems have common requirements and tradeoffs that must first be considered prior to initiating the design. A typical single-lens layout can consist of as many as ten or more individual lens elements-including aspheric lenses-working together to relay the image of the phosphor-based scintillation screen onto the CCD image plane. The x-ray, electron or neutron beam interacts with the phosphor or scintillation screen and produces visible photons that are re-imaged by the lens to the low-light-sensitive CCD.
Key requirements that must be balanced are the numerical aperture (NA), field of view, magnification, and the resolution/modulation transfer function (MTF). Because the system uses a fixed, single 61 mm square (with 15 µm square pixels)back-illuminated silicon detector, the lens resolution/MTF must ensure that the system is detector-limited. In addition, the lens designer must consider screen blur, color correction requirements (related to the phosphor emission bandwidth), and radiation damage. All these factors influence the end lens complexity, overall architecture, and, of course, the system cost.
Certain parameters can be considered as being fixed, such as the detector/image size and scintillation screen/object size. However, the implementation of a pyramid splitting architecture scheme allows a superior resolution to be obtained by limiting the demagnification and thus the NA at the image plane (see Fig. 1). It should be noted that these lenses are quite far away from the diffraction limit and the NA requirement (at the object) is driven by light gathering rather than resolution. For example, the typical specification frequency is around 33 cy/mm (Nyquist of the detector that relates to a 15 µm pixel)-which compares to a typical cutoff in the region of 1000 cy/mm. Despite being far away from the cutoff, the MTF is generally quite challenging and results in multielement lenses that often include aspheres as large as 130 mm in diameter.
FIGURE 1. A “big” lens imaging system built for use with an electron-beam microscope at the University of California at San Diego shows a splitting architecture that splits the image into four quadrants (top). A single arm of the system shows the complex path that the photons take through multiple lens elements (bottom). (Courtesy of UCSD and OPTICS 1)
The lenses image onto a back-illuminated 16-megapixel CCD from Fairchild Imaging (Milpitas, CA) (4096 × 4096 pixels equating to an active area of 61.4 × 61.4 mm). The back-thinning exercise improves the quantum efficiency of the CCD and makes it approximately four times more efficient than front-illuminated CCDs. For a number of the systems, the chip is housed in a Spectral Instruments (Tucson, AZ) camera and cooled (typically to -27°C) to limit noise contributions.
For the pyramid splitting architecture the four images from the cameras are seamlessly stiched together using software. This translates into a very low optical distortion requirement (less than 0.5%) and the need for an overlap region between the four images, typically 1%. With many conflicting requirements, it is important to appreciate all of these system requirements and interactions in the early stages of the design process.
X-ray inspection
The x-ray inspection system designed and built by OPTICS 1, in collaboration with Lawrence Livermore National Labs (LLNL; Livermore, CA), for the Pantex nuclear weapons assembly and disassembly facility near Amarillo, TX, is aptly named CoLOSSIS (Confined Large Optical Scintillation Screen Imaging System). The truly colossal system weighs almost 13 tons and was designed to collect three-dimensional (3-D) digital x-ray images of industrial objects and test components. The lead shielding is responsible for the bulk of the mass and is required to protect the lenses and sensor from the 9 MeV x-ray source. Radiation damage, system size, remote access during operation, and room door dimension became rather unusual driving specifications that influenced the end architecture, which is essentially a four-lens system separated with a pyramid (see Fig. 2). The remote access requirement resulted in the need for remote control off the lens and CCD to precisely set focus and magnification. The mirrors are also under remote gimbal control to finely align the four images to within a pixel resolution. The shielding comprises a core lead “exhaust” and an array of shields that are attached to an exoskeleton structure.
The lenses are designed to ensure that the system is essentially detector-limited and distortion-free so that the 16-megapixel images can be effectively stitched together into a 64-megapixel composite image. The 3-D image is accumulated by rotating the object under test, collecting data snapshots, and reconstructing the 3-D image with image-processing software developed by LLNL.
X-ray Crystallography
An x-ray crystallography system in routine operation with the Advanced Light Source (ALS) synchrotron at Lawrence Berkeley National Laboratory (Berkeley Lab; Berkeley, CA) has unlocked the structure of hundreds of crystals (see Fig. 3).1 The lens build was funded by the Molecular Biology Consortium (Chicago, IL). A lens coupling method was chosen over the more conventional fiber coupling method for several reasons, including lower noise and higher efficiency, improved point spread function (better resolution), negligible distortion, and for the ability to tile systems. Lenses eliminate the “chicken wire” effect in which the projection of the honeycomb pattern in the images can occur because of the inherent design of the fiber taper. From the outset, the lens was designed in a “Lego” fashion so that it could grow to a 2 × 2 arrangement, thus enabling the re-imaging of a larger field-of-view screen and therefore larger crystal areas (see Fig. 4).
FIGURE 3. Lens “Noir One” is part of an x-ray crystallography system in routine operation with the Advanced Light Source (ALS) synchrotron at Lawrence Berkeley National Laboratory (Berkeley Lab; Berkeley, CA). (Courtesy of MBC and Berkeley Lab)
Electron-beam imaging
Another pyramid system was built to support transmission-electron-microscopy work being conducted at the National Center for Microscopy and Imaging Research (NCMIR) at the University of California at San Diego (UCSD; La Jolla, CA). The NCMIR team is typically interested in 3-D reconstruction of structures within the brain, with an aim to resolve structures across large image fields to establish how the brain is “wired” at resolutions on the order of 3 to 4 Å. The lens and digital recording of this data will allow researchers the ability to see detail that was previously not resolvable using conventional electron beam imaging techniques (see Fig. 5).
FIGURE 5. A montage (1000 pictures) of a rat’s brain can be constructed from lens-captured images on an electron-beam microscope. (Courtesy of UCSD)
The lens system MTF is therefore critical in ensuring that no detail is lost in relaying the scintillation-screen image onto the CCD. The MTF challenges resulted in the need to build the lens using an active-centering technique. Comprising nine lens elements, including one aspheric lens element, the mechanics are designed to support the active-centering method that set the individual lens decenter tolerance at less than 15 µm. This, combined with appropriate melt and fabrication adjustments, allows the lens to be built to near-nominal design performance.
Neutron-beam imaging
A neutron-imaging system intended for use in nondestructive inspection applications relevant to U.S. Department of Energy programs-with particular emphasis on detecting voids, cracks, or other significant structural defects in heavily shielded low-atomic-mass materials-is currently in development at OPTICS 1. Such a system could detect defects in plastics, ceramics, or other materials that are shielded by steel or lead. Neutron and conventional x-ray imaging are complementary techniques. Neutron imaging offers lower resolutions than x-ray imaging (on the order of hundreds of microns rather than tens), but neutrons pass through high-atomic-mass materials much more readily than x-rays and interact more strongly in the low-atomic-mass materials; thus, they are very well suited for inspecting heavily shielded low-atomic-mass materials and in many situations-there is no other way to see them.
The interaction of the neutron beams with structures and the scintillation screen results in a blur function that is greater than some of the other systems described here. The demand on the lens is therefore decreased so that the system can afford to operate at higher demagnifications; in addition, the system MTF requirements are reduced and the optical design can be all spherical. This point emphasizes the need to understand the full system requirements so that the lens is not overdesigned.
OPTICS 1 has worked in close collaboration with a number of organizations in the development of big lens systems from system concept through detailed design to final build and installation. Early and continued close collaboration is required to allow the system tradeoffs be fully understood before embarking on the detailed optical and mechanical design. The end result of this work is a variety of lens systems that have enabled 3-D nondestructive inspection of optically opaque objects and medical imaging that can help us understand the inner workings of the brain.
REFERENCE
1. T.J. Madden, W. McGuigan, M.J. Molitsky, I. Naday, A. McArthur, E.M. Westbrook, 2006 IEEE Nuclear Science Symposium Conf. (Oct. 29-Nov. 4, 2006; San Diego, CA).
WILLIAM McGUIGAN is director of engineering and ROBERT E. FISCHER is founder and CEO of OPTICS 1, Commercial Systems Division, 3050 E. Hillcrest Dr., Suite 100, Westlake Village, CA 91362; e-mail: [email protected];www.optics1.com .
Imaging electromagnetic waves
When we think about imaging, we usually think about viewing a visible-light scene on a video display or through the lens of a telescope or microscope. But how do we image those tiny photons and electrons that flow all around us? In this Optoelectronics World special section, technologists at OPTICS 1 describe systems that can probe structures down to angstrom resolution, allowing researchers to look at electron, neutron, and x-ray interactions for applications as diverse as x-ray crystallography and neural brain function. And with recently introduced breakthroughs in near-field optical systems, scientists at Nanonics Imaging and the Hebrew University of Jerusalem are imaging plasmonic and evanescent waves of light toward applications in semiconductor-device physics and biology.
Gail Overton
Associate Editor