Measurement Accuracy in Silicon Photonic Ring Resonator Thermometers: Identifying and Mitigating Intrinsic Impairments

TL;DR

This paper investigates intrinsic impairments in silicon photonic ring resonator thermometers, quantifies their effects, and proposes mitigation strategies to enhance temperature measurement accuracy beyond 10 mK.

Contribution

It identifies and models key on-chip impairments affecting accuracy and demonstrates methods to mitigate their impact, improving thermometer precision.

Findings

Self-heating can cause up to 200 mK error at 1 mW power.

Longer ring cavities effectively suppress self-heating effects.

Selecting resonance orders with minimal distortion reduces measurement errors.

Abstract

Silicon photonic ring resonator thermometers have been shown to provide temperature measurements with a 10 mK accuracy. In this work we identify and quantify the intrinsic on-chip impairments that may limit further improvement in temperature measurement accuracy. The impairments arise from optically induced changes in the waveguide effective index, and from back-reflections and scattering at defects and interfaces inside the ring cavity and along the path between light source and detector. These impairments are characterized for 220 x 500 nm Si waveguide rings by experimental measurement in a calibrated temperature bath and by phenomenological models of ring response. At different optical power levels both positive and negative light induced resonance shifts are observed. For a ring with L = 100 um cavity length, the self-heating induced resonance red shift can alter the temperature reading by 200 mK at 1 mW incident power, while a small blue shift is observed below 100 uW. The effect of self-heating is shown to be effectively suppressed by choosing longer ring cavities. Scattering and back-reflections often produce split and distorted resonance line shapes. Although these distortions can vary with resonance order, they are almost completely invariant with temperature for a given resonance and do not lead to measurement errors in themselves. The effect of line shape distortions can largely be mitigated by tracking only selected resonance orders with negligible shape distortion, and by measuring the resonance minimum wavelength directly, rather than attempting to fit the entire resonance line shape. The results demonstrate the temperature error due to these impairments can be limited to below the 3 mK level through appropriate design choices and measurement procedures.