Performance evaluation of twin-field quantum key distribution with injection-type attacks

Twin-field (TF) quantum key distribution (QKD) can overcome the fundamental rate-loss bound of repeaterless QKD, enabling secure communication over long distances. However, in practical implementations, its security relies on accurate intensity modulation, which may be compromised by source imperfections and light-injection attacks. In this work, we investigate the security of the NPP TF-QKD protocol in the presence of imperfect intensity modulation, focusing on Trojan-horse attacks and induced photorefractive attacks. By employing the reference technique, we establish a finite-key security proof that explicitly incorporates this source-side information leakage. Numerical simulations show that with realistic transmitter isolation, the NPP TF-QKD protocol can still surpass the rate-loss bound and achieve secret-key rates close to the ideal case. Our results further demonstrate strong robustness against modulation deviations induced by photorefractive effects. These findings confirm the practical security of TF-QKD under realistic source imperfections and provide quantitative guidance for the design of secure quantum communication systems.