This paper introduces an enhanced approach to loop closure detection in Simultaneous Localization and Mapping (SLAM) by integrating the Faster-LIO framework with Stable Triangle Descriptors (STD). SLAM, essential in autonomous driving and augmented reality, often encounters cumulative errors affecting mapping accuracy. Traditional detection methods based on sensor data like camera images and LiDAR point clouds struggle with environmental changes that alter scene appearance and geometry. Our approach utilizes LiDAR sensors and STD, exploiting the geometric stability of triangles to maintain robustness against rotational and translational changes. The process involves storing triangle descriptors from key frames in a hash table within the Faster-LIO framework, a voxel-based LiDAR-Inertial Odometry optimized for efficiency and speed. These descriptors are then matched across frames using a voting mechanism to ensure reliable loop closure detection. Validation on the KITTI dataset and a proprietary subterranean parking garage dataset demonstrates that this integration not only enhances loop closure detection but also simplifies computational demands by avoiding complex tree structures. This method shows promise for broader applications in robotics and autonomous systems, with future research focusing on refining the descriptor and expanding its applicability to other sensor modalities.