仿生資料庫

Database of nanoparticle decorated SnS₂ CO₂ Reduction efficiency (六層仿生粒子層層披覆之二硫化錫基板減碳資料庫)

Introduction

The SnS₂ substrate covered with six layers of bionic particles shows excellent carbon dioxide capture ability, which can be used in green technology to reduce the carbon dioxide in the atmosphere. The CO₂ consumption rate of nanoparticle-modified SnS₂ will be affected by the morphology of the surface nanoparticles while the morphology of each layer of nanoparticle is controlled by the experimental parameters. According to the experimental parameters of each layer, this database selects the key parameters including (1) the photoreduction time of the 1st layer, (2) the amount of HAuCl₄ in the 2nd layer, (3) the electrophoresis time of the 3rd layer, and (4) the electrophoresis time of the 4rd layer, (5) the amount of silver nitrate used in the 5th layer and (6) the concentration of silver nitrate in the 6th layer. These parameters are set as the setting conditions. The machine learning model will provide the CO₂ capture efficiency of SnS₂ substrate covered with six layers of nanoparticles under different setting conditions. The optimized model generated after 1,000,000 learning cycles has a root mean square error (RMSE) of less than 0.002 and false rate less than 0.01.

六層仿生粒子層層披覆之二硫化錫基板具有優秀的二氧化碳捕捉能力，可應用於綠色科技，減緩二氧化碳於大氣中的含量。奈米粒子修飾二氧化硫之二氧化碳消耗率會受到表面奈米粒子之形貌所影響，而各層奈米粒子之形貌則由實驗參數所調控，因此可藉由調控各層製程的關鍵參數來控制材料的二氧化碳捕捉效率。本資料庫針對各層實驗參數，分別選擇 (1)第一層之光還原時間，(2)第二層之四氯金酸用量，(3)第三層之電泳時間，(4)第四層之電泳時間，(5)第五層之硝酸銀用量與(6)第六層之硝酸銀濃度做為設定條件，以基因演算法類神經網路(GANNs)進行機器學習與模型構建，提供在不同參數條件下，六層仿生粒子層層披覆之二硫化錫基板的二氧化碳捕捉效率。經過1,000,000次學習循環生成之優化模型具有小於0.002之方均根錯誤值(RMSE)，誤判率亦小於0.01。