February_EDFA_Digital
edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 20 NO. 1 In a marked change from previous ISTFA events, the FIB User Group meeting was held concurrently with the Contactless and Nanoprobing User Group during the two-hour lunch break on Thursday, November 9. Despite some concern, the “lunch and learn” experiment seemed to work. Overall attendance was up from last year, and audience participation was strong after each paper and during the panel discussion at the end. A peak attendance of 97 was recorded just before 1 p.m. The organizers also wish to thank Thermo Fisher Scientific for their sponsor- ship of the event. ISTFA 2017 CONTACTLESS OPTICAL/NANOPROBING EFA USER GROUP Moderators: Dan Bockelman, Intel Corp.; Sweta Pendyala, GLOBALFOUNDRIES; and Nebojsa Jankovic, NXP Semiconductors dan.bockelman@intel.com , sweta.pendyala@globalfoundries.com , nebojsa.jankovic@nxp.com D r. Daminda Dahanayaka (GLOBALFOUNDRIES, Essex Junction, Vt.) gave the first presentation, “Challenges of FEOL Sample Preparation for Nanoprobing of Advanced Nodes.” The goals in sample prep are good edge definition of conductors, protection of underlying structures, and good surface electrical con- ductivity for probe contacting. Because scaling reduces line widths as well as film thicknesses, specialized sample preparations for both scanning electron microscopy (SEM)- and atomic force microscopy (AFM)-based probers are required. First, the IC is delayered down to the layer of interest. Next, topographical relief of the contacts is performed 2 nm below the surface with a diamond slurry. Colloid slurry relief is not sufficient for AFM probers but is sufficient for SEMnanoprobing. An alternative to diamond slurry is HF reactive ion etch (CF 4 , CHF 3 , CH 2 F 2 , etc.). The HF yields poor images, and CF 4 etches SiN faster than SiO 2 , but C-F leaves polymer deposits. Finally, a third alternative is ion milling (argon, N 2 , O 2 ). The N 2 etches a little faster than argon or O 2 . Overall, this presentation emphasized how important sample prep has become as devices have continued to scale aggressively. The second presentation, “Recent Advances in Nano- probing,” was given by Dr. Martin Von Hartman (Intel, Hillsboro, Ore.). The presentation discussed nPIII upgrades comparedwith the nPII system. The xProber using the Zeiss SEM is the precursor to the nPIII, which uses a substage. Faster nanoprobing is needed, especially for future tech- nologies. Intel has seen increased demand for nanoprob- ing year after year, which is more than was anticipated. Essential capabilities of a nanoprober are minimal drift; maximum time on contact; SEM image quality; quality probe tips; good sample prep; automation; capability to detect opens, shorts, resistive shorts, and so on; and minimal e-beam invasiveness. Primary benefits of the nPIII system are 50% reduced beam current from nPII, a lower workingdistanceof less than4mm, andgreater than10min time on contact. Pulse probing was aimed at finding resis- tive gate failures, but it is too slow and not done routinely in the Intel labs. There is need for an alternating current test to investigate gate resistance, but it must be fast. It can take hours to set up for pulse probing. The third presentation, “Advantages of Electrical Opti- cal Voltage Probing (EOP) Using a Digitizer,” was given by Brett Adler (Hamamatsu, San Jose, Calif.). Sampling long-duration test loops enables detectability of timing behavior not possible with an oscilloscope. The technique uses a continuous-wave laser. Carrier concentrations under the transistor gate result in reflection changes as the transistor switches, and the changes are felt in the reflected beam. The reflected beam is sampled through the use of a detector, and the waveform information is derived. It allows zooming into any fast glitch in a wave- form. The EOP mapping combined with a digitizer works well for SRAMs. These enable new analysis with 31.25 ps resolution. The technique uses up to four Giga-samples per second, and 500,000 points can be sampled at 250 ps resolution. The fourth presentation, “Fault Localization—LTP Tech- niqueUsingHigh-SpeedDigitizer versus Oscilloscope,”was presentedbyDr. MikeBruce (Semicaps). Laser timing probe (LTP) measurements have been limited by the acquisition speed of oscilloscopes. Three spectrum analyzers have been used in parallel, due to the triggers in short test loops. S - 4
Made with FlippingBook
RkJQdWJsaXNoZXIy MjA4MTAy