edfas.org 5 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 25 NO. 4 spatial resolution require producing and detecting vastly more x-ray photons and confining them to an ever-smaller, resolution-sized spot. The extreme demands for x-ray intensity mean that published 3D analysis of ICs at nanoscales has generally been the province of research based at synchrotron beamlines. Since the first IC imaging at a synchrotron by absorption contrast in 1999,[1] the technique has been further refined. The complementary technique of x-ray ptychography further exploits both the high x-ray intensity available at synchrotrons and the coherent nature of their radiation. Valuable as they are, synchrotrons are scarce resources that cannot be moved into industrial or other typical research settings. There remains a strong interest in the development of laboratory instruments that can also tomographically analyze semiconductor devices. Some promising work applies a focused-ion beam to delaminate layers of a microelectronic device, alternating with acquisition of 2D surface images from a scanning-electron microscope (SEM).[2] The accurate synthesis of many thousands of such images into the full 3D reconstruction is a challenging mathematical problem, however. The process also destroys the IC, preventing follow-up measurements by spectroscopic or other analytic techniques. We have lately been exploring whether nondestructive tomography of ICs can be accomplished by harnessing the extremely high energy-resolving power of superconducting x-ray detectors. This power can dramatically reduce unwanted x-ray backgrounds to achieve the most efficient possible use of the limited supply of x-ray photons. It also opens the door for discrimination among multiple chemical elements in a sample. In our approach (Fig. 1), a small x-ray spot is achieved by focusing a SEM beam onto a platinum film only 100 nm thick. The sub-micrometer features of an IC are magnified onto an imaging x-ray spectrometer by generating the x-rays in the thin-film target mere micrometers from the layers of interest in the IC. By placing the electron-tophoton converter so close to the attenuating sample, we can perform the measurement in a compact instrument only a few meters across. It is the size not of a synchrotron facility, but of a typical laboratory instrument based on a vacuum chamber. Discrimination of the signal x-rays from a large background is aided by good energy resolution. This is where superconducting sensors enter the picture. TRANSITION-EDGE SENSORS: SUPERCONDUCTING X-RAY MICROCALORIMETERS Microcalorimeters are energy-resolving sensors that detect x-ray photons, or in fact, photons all the way from the optical band to gamma-ray energies. Like any calorimeter detector, they convert photon energy to heat, then take advantage of some temperature-dependent property to measure that thermal energy electrically. In order to make precision measurements of each x-ray photon’s energy, the sensors generally must be made small and kept cryogenically cold. Many designs harness the unusual properties of cryogenic materials: the metallic magnetic calorimeter uses the temperature-dependent magnetization of a metallic paramagnet, while the superconducting tunnel junction uses the quantum tunneling of electrons freed by the absorbed energy. One of the most fully developed and widely used types of microcalorimeter is the superconducting transitionedge sensor (TES),[4] such as those shown in Fig. 2. In a TES, the temperature is sensed by a material held in the transition between its superconducting and normal states. At temperatures in the narrow range of this superconducting phase transition, the material’s electrical resistivity is extremely sensitive to tiny temperature changes. The desired temperature can be reached by placing a constant electrical potential (typically, a few microvolts) across a Fig. 2 Left: A TES array “snout.” The sensors (top), the multiplexing SQUID amplifiers (4 of 8 are shown below the flexible wiring runs), and other support electronics are cryogenically cooled in this assembly. Right: An array approximately 9 mm in diameter containing 240 TES detectors (bottom), similar to the array used in the first version of Tomcat. Courtesy of Daniel Schmidt.
RkJQdWJsaXNoZXIy MTMyMzg5NA==