Sweat Fingerprint Identification Technology Based on Ion Dynamics: Research on Inkjet-Printed Organic Electrochemical Transistor Arrays

Academic Background

Sweat, as a non-invasive biomarker, contains rich physiological information that can reflect human health conditions, such as hydration balance and disease markers. However, sweat has complex components, including various ions and molecules. Traditional sweat monitoring devices typically rely on sensors with specific bio-recognition elements (e.g., ion-selective membranes and enzymes), which require complex chemical modifications to selectively bind specific ions or molecules. However, these complex chemical modification processes can lead to signal drift and interference, limiting their broad application. To address this issue, researchers proposed a sweat fingerprint identification strategy based on ion dynamics, combining inkjet-printed Organic Electrochemical Transistor (OECT) arrays and artificial intelligence (AI) algorithms to achieve efficient detection and analysis of sweat components.

Source of the Paper

This research was jointly conducted by Yuanlong Shao from Peking University, Lizhen Huang from Soochow University, and Ming Wang from Fudan University, among others, and was published on April 18, 2025, in the journal Device. The paper is titled “An Inkjet-Printable Organic Electrochemical Transistor Array with Differentiated Ion Dynamics for Sweat Fingerprint Identification.” The research was supported by the National Natural Science Foundation of China, among other projects.

Research Process and Results

1. Preparation of Inkjet-Printed Porous Films

The study first prepared OECT arrays with differentiated ion dynamics using inkjet printing technology. The researchers optimized the printing process by using a composite ink comprising PEDOT:PSS, high-boiling-point ethylene glycol (EG), and the surfactant Triton X-100. Through non-contact additive manufacturing, the ink was jet-printed onto a substrate to form porous films. By controlling the substrate temperature and droplet spacing, high-precision (<70 µm) PEDOT:PSS patterns were achieved. To regulate the microstructure of the films, the researchers also triggered local aggregation and phase separation of PEDOT:PSS through solvent evaporation discrepancies, forming 3D-interconnected porous (3D-IP) films, and tubular porous (TP) films through post-treatment with dimethyl sulfoxide (DMSO).

2. Construction and Electrochemical Characterization of the OECT Array

Based on the 3D-IP and TP films, the researchers constructed a fully inkjet-printed OECT array. Through electrochemical testing, they found that 3D-IP-OECTs exhibited faster ion intercalation dynamics, while TP-OECTs showed slower ion intercalation behavior. The differences in ion dynamics of these two films were utilized to construct an ion dynamics-differentiated sensing array, achieving differentiated dynamic electrochemical responses to sweat components. Additionally, the study employed a Convolutional Neural Network (CNN) model to classify sweat fingerprints, successfully distinguishing six artificial sweat samples and four real sweat samples with accuracies of 95.0% and 98.0%, respectively.

3. Development of a Flexible Sweat Sensing System

The research further developed a flexible wireless multi-channel sweat sensing system, including an inkjet-printed OECT array, a microfluidic module, an adapter circuit board, a wireless multi-channel testing circuit board, and a mobile application. This system can collect sweat fingerprint data in real-time and transmit the data wirelessly to a mobile app for display. The system exhibits excellent flexibility, adapting to deformations on different parts of the body, and showed no significant degradation in sensing performance after 600 sweat measurements under environmental conditions.

4. AI-Assisted Sweat Fingerprint Identification

To extract effective information from complex sweat fingerprints, the researchers adopted AI algorithms for data analysis. Using a CNN model, they successfully extracted 10 discrete time-point data from the current-time (I-t) curves of eight sensors, forming an 8×10 sweat fingerprint pattern for training the AI model. After 300 training epochs, the model achieved 95.0% accuracy in identifying artificial sweat samples and 98.0% accuracy in identifying real sweat samples. Additionally, the researchers used Shapley Additive Explanation (SHAP) analysis to evaluate the contribution of each sensor to the model, finding that each sensor contained non-overlapping information beneficial to the model’s predictions, thereby enhancing the model’s anti-interference capability and robustness.

Conclusion and Significance

The study proposed a sweat fingerprint identification strategy based on ion dynamics, achieving efficient detection and analysis of sweat components through inkjet-printed OECT arrays combined with AI algorithms. This technology provides a non-invasive solution for health monitoring, disease identification, and potentially individual identification. The stability of OECTs benefits from the intrinsic ion dynamics design rather than relying on external modifications, effectively reducing signal interference. The study also demonstrated the scalability and cost-effectiveness of inkjet printing technology, offering potential for developing smaller wearable devices in the future.

Research Highlights

1. Innovative Ion Dynamics Strategy: By designing OECT channels with different microstructures, differentiated dynamic electrochemical responses to sweat components were achieved.
2. Inkjet Printing Technology: High-precision OECT arrays were prepared using inkjet printing technology, showcasing its potential in wearable devices.
3. AI-Assisted Sweat Identification: CNN models successfully distinguished multiple sweat samples, demonstrating the value of AI algorithms in decoupling complex biological signals.
4. Flexible Wireless Sensing System: A flexible wireless multi-channel sweat sensing system was developed, enabling real-time monitoring and analysis of sweat components.

This research not only advances the scientific development of sweat sensing technology but also provides new tools for non-invasive health monitoring, holding significant clinical and market potential.