image: Figure | Principle and system configuration of L²-CPI. a, The schematic diagram illustrating the difference between L2-CPI and the conventional QPI. b, The relationship between resolution and FOV in conventional QPI and L2-CPI. 0.61λ represents the theoretical maximum achievable resolution limit by conventional dry optical microscopy systems. The blue box delineates the range of field of view (FOV) and resolution capabilities attainable with objective lenses of varying magnifications. In contrast, the gray curve illustrates the FOV limits imposed by these lenses at different magnifications. c, The optical configuration of L2-CPI. d, The schematic diagram illustrating the data acquisition process of L2-CPI. The gray structures on the left represent the sample, which consists of four letters: “HUST.” The purple boxes on the sample represent the imaging area, which corresponds to the camera's sensing area. As the sample moves along the scanning direction, many interferograms are captured. See the three representative interferograms corresponding to the sample's three positions along the scanning direction (i.e., x-axis) in Fig. 1d. The three red lines in the three interferograms correspond to precisely the same positions (i.e., the cross-sections along the y-axis) on the sample’s surface. For a given cross-section on the sample’s surface, its corresponding interferogram lines (i.e., the red lines) in all the interferograms are extracted to form a sampling sequence (i.e., see the dark green box marked by “sampling sequence” in Fig. 1d) in first-come-first-served order. The phase-retrieval algorithm can then be applied to the sampling sequence to recover the phase corresponding to the cross-section on the sample’s surface.
Credit: Jinlong Zhu et al.
The rapid advancement of semiconductor manufacturing and precision medicine demands imaging tools capable of capturing nanometer-scale details across large areas. Quantitative Phase Imaging (QPI) has emerged as a premier solution for observing transparent biological cells and measuring the 3D topography of silicon wafers. However, conventional QPI systems face a fundamental "bottleneck": high-magnification objectives provide high resolution but only a tiny field of view. To image a whole wafer or a large tissue sample, researchers must use "step-and-repeat" stitching, a process prone to mechanical misalignments and phase discontinuities that compromise data integrity.
In a new paper published in Light: Advanced Manufacturing, a team of scientists led by Professor Jinlong Zhu from the School of Mechanical Science and Engineering at Huazhong University of Science and Technology, China, has developed a novel computational imaging architecture called Lateral Line-scan Computational Phase Imaging (L2-CPI). This system extends the conventional limits of optical microscopy by enabling continuous, high-resolution phase imaging across an arbitrarily large field of view, limited only by the travel range of the scanning stage.
The L2-CPI system is centered around a Linnik-type interferometric configuration that utilizes the sample’s lateral motion to generate phase shifts. Unlike traditional methods that require the sample to stop for every frame, L2-CPI captures data "on the fly." The team integrated a Dynamic Compensation System (DCS) to suppress mechanical vibrations in real time and employed a robust three-parameter cosine-fitting algorithm to retrieve phase information with high precision. These scientists summarize the operational principle of their system:
"We designed a lateral scanning architecture that serves three purposes in one: (1) to achieve diffraction-limited resolution across an arbitrary field of view without the need for image stitching; (2) to eliminate 'proximal errors' typically found in step-scan systems through a continuous pixelated acquisition strategy; and (3) to ensure high-fidelity 3D reconstruction even in the presence of industrial-level mechanical vibrations."
"Because our method retrieves phase information from a massive number of data points during the scan, it is significantly more robust against noise and environmental disturbances than traditional phase-shifting techniques," they added.
The researchers demonstrated the power of L2-CPI by inspecting patterned defect array wafer and microlens arrays. The system successfully identified sub-wavelength defects—such as bridge and cutting defects — in an intentionally fabricated defect array on a wafer with a 60 nm critical dimension, which are typically invisible to conventional intensity-based imaging systems. Based on our simulations, L²-CPI is expected to retain detectability even when the critical dimension is scaled down to 15 nm.
"The presented technique provides a high-throughput, non-destructive solution for large-scale nanometrology. It will be particularly transformative for the semiconductor industry, where rapid and accurate defect detection is critical for yield enhancement. Beyond the cleanroom, we foresee L2-CPI opening new avenues in digital pathology and automated biological screening, allowing researchers to 'see' large-scale life processes with unprecedented detail," the scientists forecast.
Journal
Light: Advanced Manufacturing
Article Title
L²-CPI: high-resolution computational phase imaging with an arbitrary field of view