Electrical Physically Unclonable Function via Stochastic-Defect-Driven Invisible Current Pathways

With the growing demand for secure authentication in IoT and other resource-constrained environments, Physically Unclonable Functions (PUFs) have gained attention as an appealing approach that leverages intrinsic device randomness. Recently, helper data and error-correction systems have been widely adopted to compensate for the limited intrinsic stability of device fingerprints. In contrast, the breakdown (B.D.) fingerprint introduced in this work demonstrates a robust and lightweight PUF platform that operates without any such compensation. Despite the uniformity of Atomic Layer Deposition (ALD)-grown amorphous HfO₂, intrinsic film and interface defects generate subtle stochastic variations in insulation, which─under our tailored key-generation scheme─can be reliably binarized through the breakdown process, yielding a perfectly balanced bit uniformity of 0.5. Intra/interdevice Hamming distances (HD) converge to 0 and 0.5, and correlation-coefficient analysis further verifies the high entropy of the generated security code. Moreover, the device exhibits excellent electrical retention and thermal stability over a wide operating range from +20 to 150 °C. These characteristics collectively enable a wide and reliable authentication threshold margin, maintaining stable performance under temperature variation, device aging, and operational noise. In conclusion, the proposed B.D. fingerprint provides robust and lightweight device identities with a broad authentication window.