Development and characterization of flexible carbon temperature sensors: a study on different inks and sensor designs

Several industries and applications rely on the accurate monitoring of temperature. Consequently, the development and integration of reliable and flexible temperature sensors that are able to conform to different surfaces is desirable. Despite being a known fact that the used materials, chosen technologies, sensor size, and design tend to impact the performance of said sensors, not many studies focus on the analysis of the influence of these factors. As a result, this work intended to explore how the performance of carbon-based screen-printed temperature sensors could be tuned or optimized by changing the ink and the design of the printed sensors. Therefore, two commercial carbon inks and three different designs were used to create the herein-studied temperature sensors. Firstly, to explore potential differences between the used inks, their morphologic, chemical, and electrical properties were evaluated, allowing further insights to be gathered regarding the filler type, composition, and bulk resistivity. After printing the sensors, they were also submitted to electrical impedance spectroscopy, which allowed for a circuit equivalent model to be proposed for each sensor type, unraveling how current flew across the printed materials. Then, a full factorial statistical analysis was outlined and conducted to uncover which combination of factors allowed to optimize the sensors’ performance, i.e. lower initial resistance, higher thermal coefficient of resistance (TCR), and higher correlation coefficient (R2) of the responses. To gather the experimental data, the electrical response of the sensors (resistance variation) to increasing temperature was collected. As a result of these experiments, it was revealed that ink 1 (based upon carbon nanoparticles) allowed to obtain less resistive sensors and, even though ink 2 (based upon carbon nanotubes) presented higher sensitivity, ink 1 resulted in higher linearity of the sensors. Overall, the factor design had less influence on the results than the factor ink. Notwithstanding, design 1 could be used to obtain sensors with lower native resistance and higher linearity. To sum up, using ink 1 combined with design 1 (fewer hoops and thicker line width) allowed to obtain temperature sensors with average resistance at room temperature of 7.8 kΩ, a TCR of 1.13 %.ºC−1, and R2 of 0.99, which means the herein studied sensors present a very promising behavior when compared to the sensors in the preexisting literature. Thus, the proposed optimized sensors presented attractive characteristics as flexible printed temperature sensors.