Ionizing Radiation Effects Spectroscopy for Analysis of Total-Ionizing Dose Degradation in RF Circuits

摘要

Ionizing radiation effects spectroscopy (IRES) for the analysis of total-ionizing dose (TID) degradation in radio frequency (RF) circuits is presented. IRES is used to image the change in the operating parameters with respect to TID and for classification of the circuit's state based on image features. Measured data from two voltage-controlled oscillators (VCOs) and two phase-locked loops (PLLs) exposed to TID are used to demonstrate the utility of IRES for radiation effects analysis. The VCO and PLL circuits were designed in a 130-nm bulk CMOS technology. The circuits were irradiated up to 300 krad(SiO2) using an Aracor X-ray source, and IRES images were developed at various TID levels based on the instantaneous frequency of the output waveforms. The IRES technique is based upon RF-distinct native attribute fingerprinting for specific emitter identification using RF waveforms in communication systems. The technique employs statistical time-frequency analyses of various signal features. The resulting statistical measures are utilized to identify the operational state of the circuit (i.e., the TID level and bias voltage) using machine learning (ML) classification. While minor increases in operating frequency were measured with dose, the technique resulted in up to 100% prediction accuracy in the identification of the operational state following an ML training set of at least 50 measured samples. Moreover, the IRES technique enhances standard measurements by capturing global parametric shifts as well as transient variation. The technique, which exploits the subtle characteristics inherent in the waveforms, shows promise in radiation dosimetry applications as well as in-situ monitoring of device and circuit operational health.