Nowadays, there is an increasing demand for images with high definition and fine textures, but images captured in natural scenes usually suffer from complicated blurry artifacts, caused mostly by object motion or camera shaking. Since these annoying artifacts greatly decrease image visual quality, deblurring algorithms have been proposed from various aspects. However, most energy-optimizationbased algorithms rely heavily on blur kernel priors, and some learning-based methods either adopt pixelwise loss function or ignore global structural information. Therefore, we propose an image deblurring algorithm based on recurrent conditional generative adversarial network (RCGAN), in which the scalerecurrent generator extracts sequence spatio-temporal features and reconstructs sharp images in a coarseto- fine scheme. To thoroughly evaluate the global and local generator performance, we further propose a receptive field recurrent discriminator. Besides, the discriminator takes blurry images as conditions, which help to differentiate reconstructed images from real sharp ones. Last but not least, since the gradients are vanishing when training generator with the output of discriminator, a progressive loss function is proposed to enhance the gradients in back-propagation and to take full advantages of discriminative features. Extensive experiments prove the superiority of RCGAN over state-of-the-art algorithms both qualitatively and quantitatively.

Reliable characterization techniques to accurately quantify the metallization-induced recombination losses as well as contact resistivity losses of screen-printed cells are crucial for successful optimization of the contact grid design. Previously, the dark saturation current density at the contact ( ${J}_{\rm 0c}$ ) is often assumed to be constant for different finger width. Similarly, impact of finger width on contact resistivity ( ${\rho }_{\rm c}$ ) is rarely reported. Therefore, we performed a comprehensive evaluation of ${J}_{\rm 0c}$ and ${\rho }_{\rm c}$ as a function of finger width, spacing as well as firing temperature. We found out that ${J}_{\rm 0c}$ increases from $\approx$ 2000 to $\approx$ 8100 $\text{fA/cm}^{2}$ , when the finger width increases from 60 to 400 $\mu$ m; and ${\rho }_{\rm c}$ decrease from 7.2 to 2.2 $\text{m}{\Omega }\cdot \text{cm}^{2}$ when using a wide-TLM rather than a narrow-TLM structure, for samples fired at 840 $^{\circ }$ C. Based on our cross-sectional and top-down scanning electron microscopy images, we believe that the physical root cause can be explained by the difference in the microstructure formed at the metal–silicon interface during the firing process for the screen-printed contacts.

This paper presents an easy-to-implement, but rigorous method to determine the dark saturation current densities J _01m and J _02m under the metallized region of a silicon solar cell contact. The method is based on an autofitting procedure by the Griddler artificial intelligence finite-element simulation program to measured photoluminescence images of samples with especially designed test structures. It is found that a deeper p-n junction strongly reduces the recombination losses at the metal-silicon interface. By carefully tailoring the boron diffusion profile, the J _01m is reduced from 839.2 to 274.2 fA/cm ² .