Precise and spatially/temporally resolved temperature sensing is crucial across various fields, including industrial manufacturing, environmental monitoring, and healthcare. Real-time temperature monitoring in biological systems is especially important for point-of-care diagnostics and treatment. While thermoelectric or thermoresistive sensors are commonly used, their wired electrical designs are susceptible to electromagnetic interference, particularly during MRI. Optical-based sensors offer advantages such as remote detection, minimal invasiveness, immunity to electromagnetic interference, and high resolution. These sensors can be based on luminescence intensity, wavelength, peak width, or decay lifetime. Infrared thermometers and imagers provide spatially resolved temperature information non-contactly but only measure surface temperatures and are influenced by surface emissivity. Optical cavity-based thermal sensors offer high precision but rely on sophisticated spectrometric systems, limiting biomedical applications. Liquid crystals show temperature-dependent color changes but are affected by environmental factors. Photoluminescent (PL) materials offer visualized temperature information with high spatiotemporal resolution. Upconversion processes, transforming near-infrared (NIR) photons to visible light within the biological transparency window, are particularly advantageous for biomedical applications due to reduced autofluorescence and improved tissue penetration. Lanthanide-based nanoparticles and optoelectronic NIR-to-visible upconversion devices based on semiconductor heterostructures have shown promise. This study investigates the temperature-dependent PL characteristics of an optoelectronic upconversion device and demonstrates its capabilities for thermal sensing, revealing that the thermal-dependent PL emission is determined by the band properties of semiconductor materials and the integrated device circuit architecture. The researchers demonstrate spatially resolved temperature mapping with patterned device arrays and *in vitro*/ *in vivo* applications using fiber optics, dynamically monitoring human exhalation and deep brain temperature variations in animals.

The introduction provides a comprehensive overview of existing temperature sensing technologies, highlighting their limitations and advantages. It discusses the challenges associated with traditional methods like thermoelectric and thermoresistive sensors, focusing on their susceptibility to electromagnetic interference. The review then explores optical-based sensing modalities, emphasizing the advantages of infrared thermometers, optical cavity-based sensors, and liquid crystals. The limitations of each method are clearly outlined. A detailed discussion of photoluminescent materials and their temperature-dependent properties sets the stage for the introduction of upconversion processes as a superior approach, especially for biomedical applications. The advantages of upconversion, including reduced autofluorescence and improved tissue penetration, are emphasized. The review cites relevant studies on lanthanide-based nanoparticles and optoelectronic NIR-to-visible upconversion devices, positioning the current research within the existing body of knowledge.

The study employed a fully integrated optoelectronic upconversion device consisting of a GaAs-based double junction photodiode and an InGaP-based LED connected in series. The device's structure was characterized using scanning electron microscopy (SEM). The temperature-dependent photoluminescence (PL) emission was evaluated under steady-state NIR excitation, with spectral measurements conducted using a fluorescence microscope equipped spectrometer. A temperature range from 25 °C to 90 °C was used. The intensity-temperature sensitivity and spectrum-temperature sensitivity were determined from measurements of 10 devices. Theoretical calculations based on the detailed balance model for diodes and the Varshni expression were used to validate the experimental findings. The noise associated with the spectral readings was analyzed to determine temperature resolution. To understand the thermal sensing mechanism, the thermal behaviors of individual InGaP LEDs and GaAs photodiodes were investigated separately. Their electroluminescence (EL) spectra and external quantum efficiency (EQE) were measured at different temperatures. The efficiency drops for the LED, photodiode, and upconversion device were compared. The current-voltage characteristics of the LED and photodiode were analyzed at various temperatures to understand the device's operating conditions. The ultrafast PL dynamics of the devices were also noted. Device arrays were fabricated and used to demonstrate spatially resolved temperature mapping. The devices were integrated with fiber optics using epitaxial lift-off and transfer printing techniques, creating fiber-coupled thermal sensors. The stability of the fiber-coupled devices was evaluated. *In vitro* experiments monitored exhaled breath temperature, and *in vivo* experiments measured deep brain temperature fluctuations in animals. A thermocouple was used for calibration and comparison in both cases.

The optoelectronic upconversion device demonstrated a temperature-dependent PL emission with an intensity-temperature sensitivity of ~1.5% °C⁻¹ and a spectrum-temperature sensitivity of ~0.18 nm °C⁻¹. The experimental results showed quantitative agreement with theoretical calculations. The temperature resolution, based on standard deviation of normalized intensity and peak wavelength shift, was determined at various temperatures. The analysis of individual InGaP LEDs and GaAs photodiodes revealed that the intensity-temperature dependence of the upconversion device is determined by the efficiency drops of both the LED and the photodiode, while the spectrum-temperature sensitivity is mainly determined by the InGaP bandgap narrowing. The operating conditions of the series-connected LED and photodiode determine the overall current and emission intensity. Pattened device arrays enabled spatially resolved temperature mapping. The integration of the devices with fiber optics resulted in stable, portable sensors suitable for biomedical applications. *In vitro* experiments successfully monitored exhaled breath temperature changes, with a high correlation (R² = 0.90) between the sensor's PL signal and thermocouple measurements. *In vivo* experiments demonstrated the ability to monitor deep brain temperature fluctuations.

The findings demonstrate the successful development and characterization of a novel optoelectronic thermometer based on microscale infrared-to-visible upconversion devices. The device's high sensitivity, spatial resolution, and biocompatibility make it suitable for a wide range of applications, including biomedical imaging and monitoring. The detailed analysis of the device's sensing mechanism provides a deeper understanding of the interplay between the optoelectronic properties of the semiconductor materials and the integrated circuit architecture. The ability to perform spatially resolved temperature mapping opens up possibilities for high-resolution thermal imaging. The integration with fiber optics enhances the device's versatility, allowing for minimally invasive measurements in dynamic environments. The successful *in vivo* demonstration of deep brain temperature monitoring highlights the potential of this technology for neuroscience research and clinical applications. The correlation between the device's signal and thermocouple measurements validates its accuracy and reliability.

This study successfully demonstrated a novel optoelectronic thermometer based on microscale infrared-to-visible upconversion devices, showcasing high sensitivity, spatial resolution, and biocompatibility. The device's operation was systematically characterized, linking its performance to semiconductor band structure and circuit design. Successful *in vitro* and *in vivo* applications highlight its potential in various fields. Future research could explore miniaturization for improved spatial resolution, integration with advanced imaging modalities, and expanded biomedical applications.

The current study primarily focuses on temperature ranges relevant to biological applications (25-90 °C). Extending the operational temperature range may require modifications to the device materials and design. While the spatial resolution is high, the device array's resolution is currently limited by the device pitch. Further optimization of lithographic processes could improve this. The *in vivo* experiments were conducted on animals. Further research is needed to validate the device's performance in human subjects.