November_EDFA_Digital

edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 22 NO. 4 4 EDFAAO (2020) 4:4-8 1537-0755/$19.00 ©ASM International ® STEM EBIC: TOWARD PREDICTIVE FAILURE ANALYSIS AT HIGH RESOLUTION William A. Hubbard The Aerospace Corporation, El Segundo, California william.a.hubbard@aero.org INTRODUCTION Failure analysis is almost necessarily a post-mortem investigation, where clues to the cause of failure are gleaned from its aftermath. Ideally, the dynamics leading up to failure could be studied before the failure occurs. Controlled stressing of a component can produce elec- trically defective, but not destroyed, regions, but such regions can be difficult to locate because subtle electrical degradation often does not manifest as obvious physical damage. Even when a region of probable failure can be located, standard characterization tools are often blind to the electronic and thermal dynamics at play, particularly at small length scales. Recent progress toward directly imaging electronic structure using scanning transmis- sion electron microscope (STEM) electron beam induced current (EBIC) imaging is discussed in this article. By identifying electronic device features that are at higher risk of failure, STEMEBICmay provide a path to predictive failure analysis at high resolution. The transmission electron microscope (TEM) is a workhorse instrument for characterizing microelectron- ics, with modern systems routinely capable of achieving atomic resolution. A TEM image is formed by illuminating an electron-transparent sample with a high-energy (typi- cally 80-300 keV) electron beam and collecting transmit- ted beam electrons. A variety of contrast mechanisms are accessible depending on the geometric distribution of the incident and detected electrons. [1] TEM is remark- ably sensitive to the physical structure of a sample, i.e . , the amount, arrangement, and types of atoms imaged. It is often paired with spectroscopic attachments such as energy dispersive x-ray and electron energy loss spec- troscopies (EDS and EELS), to more precisely deter- mine composition. The ability to discern the composition and placement of atoms makes TEM an extremely powerful characteriza- tion tool for microelectronic components. However, for many devices the dynamics underlying normal operation do not displace atoms. Device function is often, instead, mediated by electronic and thermal processes that have a miniscule effect on physical structure, and thus are invis- ible to TEM, until the onset of failure. For example, TEM cannot detect whether a transistor is in the ON or OFF state or determine whether or not a wire is Joule heating. This article describes EBIC and the electronic informa- tion it provides as a complement to the physical-based contrast of TEM. EBIC MODES While TEMcontrast is ameasure of beamelectrons that pass through the sample, EBIC measures currents within the sample itself that are produced by the beam interac- tion. Beam electrons can generate current in a sample through various processes, each of which maps to differ- ent electronic properties. The most common, “standard” form of EBIC is current produced when electron-hole pairs generated by the beam are separated by an electric field within the sample. Standard EBIC reveals the loca- tion of strong electric fields and provides information about carrier recombination properties. [2] Electron beam absorbed current (EBAC) consists of beam electrons absorbed by the sample, [3] with a strong signal indicating regions in the circuit that are electrically connected to the EBIC amplifier. While these two EBIC modes have been in use for decades and are usually observed in a scanning electron microscope (SEM), a recently reported [4] third mode, calledsecondaryelectronemissionEBIC, or SEEBIC, has only been demonstrated in STEM. In SEEBIC, beam- induced ejection of electrons from the sample generates the current, which is generallymuch smaller inmagnitude (a few pA) than currents in EBAC and standard EBIC (nA or greater). SEEBIC has been demonstrated at atomic resolution [5] and, as shown in the next section, is capable of mapping electronic transport properties. IN SITU TEM SAMPLE CHALLENGES Transmission electron microscope imaging requires