ObjectivePolarization imaging reveals distinct targets that are difficult to distinguish in conventional optical imaging by capturing the polarization or depolarization characteristics of light after it is reflected, refracted, or scattered by different objects. To achieve better polarization imaging, various polarization imaging systems with different structures and principles have been developed to obtain and utilize polarization information, such as polarization angle and degree of polarization. Among these, focal plane polarization imaging systems inscribe four polarization directions into each pixel region, simultaneously acquiring polarization intensities from different directions. Compared with other polarization sensor image acquisition methods, this system is compact in structure and offers strong real-time performance. However, the fabrication of focal plane polarization imaging sensors currently depends on high-precision electron beam lithography equipment. By contrast, nanoimprint lithography (NIL) provides a low-cost and efficient approach to directly integrate micro-nano polarizer arrays onto the surface of image sensors. The integration technology of nanograting arrays onto image sensors mainly consists of two parts: the fabrication of flexible UV nanoimprint composite masks and the alignment imprinting onto image sensors. This paper focuses on the cross-scale alignment issue of micron-scale patterns with millimeter-scale templates during the CMOS image sensor alignment imprinting process.
MethodsBased on an analysis of the impact of alignment deviations on the polarization performance of micro-polarizers during the fabrication process (
Fig.4-
Fig.5), a microscopic optical image-based alignment method is proposed to ensure pixel alignment between the micro-polarizers on the composite mask and the image sensors (
Fig.8). A high-precision alignment device was developed, which can be structurally divided into three components: a microscopic vision module, an alignment adjustment module, and an image processing module (
Fig.9). During the alignment process, the system does not rely on high-precision alignment marks. The microscopic vision module uses high-magnification microscopic imaging to capture the microstructures of the millimeter-sized nanoimprint template and the sensor surface. The image processing module employs Hough transform to identify the linear features of the contours of the micro-polarizer array and the pixel array, and an outlier iterative elimination algorithm (
Fig.11) is used to accurately compute the orientation and position information of the target array. Finally, the alignment adjustment module performs alignment adjustments using a precision micro-motion platform.
Results and DiscussionsA testing device is established to evaluate the alignment performance of the integrated micro-polarizer array sensor (
Fig.12). The sensor's response to specific linearly polarized light is observed, demonstrating consistency with the designed performance of the micro-polarizer array (
Fig.13). A further quantitative analysis is conducted on the uniformity of the pixel response to linearly polarized light for pixels with the same polarization sensitivity on the integrated sensor. The standard deviation of pixel responses across different regions of the aligned micro-polarizer array sensor ranges within the range of 15 to 40. In contrast, for an unaligned micro-polarizer array sensor, the standard deviation of pixel responses across different regions is around 100 (
Fig.16). This indicates that the overall polarization response uniformity of the sensor is significantly improved after alignment. Pixel alignment is achieved within a polarized pixel range of 500×500, reducing the effective pixel loss caused by crosstalk. Furthermore, an outdoor experimental setup was set up to detect the distribution of polarized light in the sky under natural conditions (
Fig.17). The polarization angle diagram calculated based on the light intensity output from the sensor can clearly identify the solar meridian and distinguish buildings at the edges of the image (
Fig.18).
ConclusionsAiming at the cross-scale alignment challenges in the integration technology of micro-polarizer array sensors using nanoimprint lithography, an alignment scheme was designed. This scheme avoids the use of high-precision alignment marks and directly captures the geometric features of the surface microstructures of the alignment object and target using microscopic optical imaging. The probabilistic Hough transform algorithm and the outlier iterative elimination algorithm are employed to ensure the efficient and accurate extraction of geometric features, achieving an overall torsional error of no more than 0.02° and pixel dislocation error within 20% of the pixel size. The final micro-polarizer array sensor demonstrates uniform and highly consistent pixel responses in the same polarization-sensitive direction. Pixel alignment is achieved within a polarized pixel range of 500×500, allowing more effective utilization of all the polarization information obtained by the pixels.