Fig. 7: Computation accuracy evaluation.

a Increasing challenge vector length narrows the voltage interval and increases the computation error rate. b Using capacitors with larger capacitance reduces mismatch and improves computation accuracy. c The averaging computation cancels the impact of the capacitor temperature drift, enabling high accuracy. d The reliable charge-domain computing mitigates the impact of \({V}_{{{{\rm{X}}}}}\) variations and ensures less degraded PUF performance. e The computation error rate increases as \({V}_{{{{\rm{DD}}}}}\) decreases due to the reduced sense margin. f Both higher or lower \({V}_{{{{\rm{R}}}}}\) narrows the on/off ratio of \({V}_{{{{\rm{X}}}}}\) and limits the performance of the proposed PUF. Source data are provided as a Source Data file.