Conversion of carbon dioxide (CO2) with the help of an appropriate electrocatalyst with high stability, low onset potential, and exceptional selectivity is still one of the great tasks in the electrocatalytic reduction of CO2 to valuable chemicals. Herein, by means of systematic first-principles simulations, we investigate the CO2 reduction reaction (CO2RR) activity of zirconium-based single-, double-, and triple-atom (Zr n @C2N; n = 1–3) catalysts anchored on a graphitic carbon-nitride monolayer. In tune with the Sabatier principle, our results reveal that a moderate CO2 binding is vital for a low onset potential for the CO2RR. Consequently, based on rigorous free energy calculations, the Zr-based single-atom catalyst (SAC) is found to be most effective to convert CO2 to valuable products such as HCOOH and CH3OH. It is worth noting that CO2 reduction to HCOOH is spontaneous via the *HCOO intermediate on Zr1@C2N and involves a low onset potential of −0.23 V with respect to the reversible hydrogen electrode from the *COOH intermediate. Among all the catalysts evaluated computationally, the Zr SAC further reveals the lowest onset potential of −0.89 V for CH3OH formation. The results show that the Zr-based catalysts especially Zr1@C2N are found to effectively suppress the competitive hydrogen evolution reaction and promote the CO2RR. Moreover, all three catalysts exhibit high kinetic and thermal stability with negligible distortion due to which their structures can be retained very well up to 600 K. Thus, the current work may provide effective catalyst-design strategies for enhancing the electrocatalytic CO2RR performance of Zr-based materials.

Document deskewing is a fundamental problem in document image processing. While existing methods have limitations, such as Hough Line Transformation that can deskew images upside down, and Deep Learning models that require huge amounts of human labour and computational resources and still fail to deskew while taking care of orientation, OCR-based methods also struggle to read text when it is tilted. In this paper, we propose a novel, simple, cost-effective deep learning method for fixing the skew and orientation of documents. Our approach reduces the search space for the machine learning model to predict whether an image is upside down or not, avoiding the huge search space of predicting an angle between 0 and 360. We finetuned a MobileNetV2 model, which was pre-trained on imagenet, using only 1000 images and achieved good results. This method is useful for automation-based tasks, such as data extraction using OCR technology, and can greatly reduce manual labour.

No abstract