Search for: All records

Fabrics and fibrous materials offer a soft, porous, and flexible substrate for microelectromechanical systems (MEMS) packaging in breathable, wearable formats that allow airflow. Device-on-fiber systems require developments in the field of E-Textiles including smart fibers, functional fiber intersections, textile circuit routing, and alignment methods that adapt to irregular materials. In this paper, we demonstrate a MEMS-on-fabric layout workflow that obtains fiber intersection locations from high-resolution fabric images. We implement an image processing algorithm to drive the MEMS layout software, creating an individualized MEMS “gripper” layout designed to grasp fibers on a specific fabric substrate during a wafer-to-fabric parallel transfer step. The efficiency of the algorithm in terms of a number of intersections identified on the complete image is analyzed. The specifications of the MEMS layout design such as the length of the MEMS gripper, spatial distribution, and orientation are derivable from the MATLAB routine implemented on the image. Furthermore, the alignment procedure, tolerance, and hardware setup for the alignment method of the framed sample fabric to the wafer processed using the custom gripper layout are discussed along with the challenges of the release of MEMS devices from the Si substrate to the fabric substrate. 

Electronic textile (E-textile) research requires an understanding of the mechanical properties of fabric substrates used to build and support electronics. Because fibers are often non-uniform and fabrics are easily deformed, locating fiber junctions on the irregular surface is challenging, yet is essential for packaging electronics on textiles at the resolution of single fibers that deliver power and signals. In this paper, we demonstrate the need to identify fiber junctions in a task where microelectromechanical structures (MEMS) are integrated on fabrics. We discuss the benefits of fiber-aligned placement compared with random placement. Thereafter we compare three image processing algorithms to extract fiber junction locations from sample fabric images. The Hough line transform algorithm implemented in MATLAB derives line segments from the image to model the fibers, identifying crossings by the intersections of the line segments. The binary image analysis algorithm implemented in MATLAB searches the image for unique patterns of 1s and 0s that represent the fiber intersection. The pattern matching algorithm implemented in Vision Assistant - LabVIEW, uses a pyramid value correlation function to match a reference template to the remainder of the fabric image to identify the crossings. Of the three algorithms, the binary image analysis method had the highest accuracy, while the pattern matching algorithm was fastest.