催化学报

催化学报

• •    

双相界面工程: 一种机器学习指导的优化FCC油浆选择性氧化脱硫策略

邢肖肖a,1, 吴沛文b,1, 邹义儒b, 高兆增b, 俞镇东b, 唐旻萌b,*, 巢艳红a,*, 朱文帅b,c*, 刘植昌b, 徐春明b   

  1. a中国石油大学(北京)理学院, 重油加工国家重点实验室, 北京 102249;
    b中国石油大学(北京)化学工程与环境学院, 重油加工国家重点实验室, 北京 102249
  • 收稿日期:2025-08-22 接受日期:2025-08-22
  • 通讯作者: * 电子信箱: mmtang@cup.edu.cn (唐旻萌), chaoyh@cup.edu.cn (巢艳红), zhuws@cup.edu.cn (朱文帅).
  • 作者简介:1共同第一作者.
  • 基金资助:
    国家重点研发计划(2022YFE0208300); 国家杰出青年科学基金(22425808); 国家自然科学基金(22508419, 22578187, 22178154); 中国博士后科学基金(2024M753617); 中国石油大学(北京)自然科学基金(2462022YJRC002, 2462024XKBH002, 2462025BJRC002); 江苏省自然科学基金(BK20230068); 碳中和研究院基金(CNIF20240103).

Biphasic interface engineering: A machine learning-guided strategy for optimizing selective oxidative desulfurization of FCC slurry oil

Xiaoxiao Xinga,1, Peiwen Wub,1, Yiru Zoub, Zhaozeng Gaob, Zhendong Yub, Minmeng Tangb,*, Yanhong Chaoa,*, Wenshuai Zhub,c,*, Zhichang Liub, Chunming Xub   

  1. aState Key Laboratory of Heavy Oil Processing, College of Science, China University of Petroleum-Beijing, Beijing 102249, China;
    bState Key Laboratory of Heavy Oil Processing, College of Chemical Engineering and Environment, China University of Petroleum-Beijing, Beijing 102249, China;
    cShandong Key Laboratory of Green Electricity&Hydrogen Science and Technology, Shandong Institute of Petroleum and Chemical Technology, Dongying 257061, China
  • Received:2025-08-22 Accepted:2025-08-22
  • Contact: * E-mail: mmtang@cup.edu.cn (M. Tang), chaoyh@cup.edu.cn (Y. Chao), zhuws@cup.edu.cn (W. Zhu).
  • About author:1Contributed equally to this work.
  • Supported by:
    National Key R&D Program of China (2022YFE0208300), the National Science Foundation for Distinguished Young Scholars (22425808), the National Natural Science Foundation of China (22508419, 22578187, 22178154), the China Postdoctoral Science Foundation (2024M753617), the Science Foundation of China University of Petroleum, Beijing (2462022YJRC002, 2462024XKBH002, 2462025BJRC002), the Natural Science Foundation of Jiangsu Province (BK20230068), and the Carbon Neutrality Research Institute Fund (CNIF20240103).

PDF

10

摘要： 催化裂化油浆富含3‒5环芳烃, 是制备高价值针状焦的理想原料, 但其硫含量普遍较高, 高温碳化过程中会造成气体溢出影响针状焦的各向异性和机械强度. 加氢脱硫需在> 420 °C、> 4 MPa下断裂C-S键, 脱硫的同时不可避免会造成芳环饱和, 降低油浆作为针状焦原料的适用性. 氧化脱硫可在常温常压下将加氢难除去的噻吩类硫化物氧化为砜类, 再经极性溶剂萃取脱除. 然而, 油浆体系存在硫种类更加复杂、黏度高以及需在大量芳烃的情况下实现硫化合物选择性氧化去除, 因此, 氧化脱硫的难点在于设计高活性和高选择性催化剂.
本研究首先对油浆的物化性质进行了分析, 油浆中芳烃含量高达68.85 wt%, 主要的硫化物为苯并萘并噻吩及其衍生物. 采用正庚烷对油浆进行稀释以增强流动性, 将粘度从2324.5降至1.9 mPa·s (313 K). 其次结合油浆氧化脱硫过程中存在的关键挑战, 创新性地提出了一种双相界面调控策略. 通过将多金属氧酸盐与有机阳离子改性剂进行分子组装, 构建了一类兼具催化与界面调控功能的多功能催化剂. 该催化剂在H2O2介导的氧化反应中, 能够稳定大量油水界面, 发挥类表面活性剂的作用, 显著强化了反应物在水油两相的传质过程, 并且通过对催化系统中亲水亲油微环境的精确调控, 解决了高芳烃油浆中芳烃保留与硫去除的关键矛盾. 与传统的分子乳化剂不同, 这些材料在体系中仍以湿润固态存在. 因此, 仅通过简单的物理方法, 就可以很容易地使其沉淀回收. 该工作系统研究了影响脱硫性能的关键因素如反应温度、催化剂用量、氧硫比、循环性能以及活性自由基类型. 采用了随机森林模型对这些变量的相对重要性进行排序. 在此基础上, 还应用贝叶斯算法对反应条件进行优化, 仅通过十次迭代确立了最优工艺参数, 实现了在保留核心原料3‒4环芳烃的情况下将硫含量从1.6 wt%降至0.34 wt%, 显著降低了实验成本. 与加氢工艺相比, 该氧化脱硫技术反应更温和, 脱硫效率更高. 机理研究表明, 其卓越的脱硫活性源于超氧自由基激活催化剂分子中的W=O末端氧, 进而形成高活性的动态过氧物种.
综上, 本文以创新的双相界面催化策略, 为高芳烃含量油浆的脱硫提供了一条技术创新且切实可行的途径, 在温和条件下实现硫的高效脱除与芳环结构的保留, 为重质油浆的高值化清洁利用提供了新思路, 并为后续工业应用开发提供了参考.

关键词: 催化裂化油浆, 氧化脱硫, 界面催化, 针状焦, 机器学习

Abstract: The non-destructive desulfurization of aromatic structures is crucial for the high-value utilization of FCC slurry oil. Hydrodesulfurization causes aromatic saturation, impairing the suitability of slurry oil as needle coke feedstock. Therefore, developing methods capable of selective desulfurization while preserving aromatics is essential. Herein, we address the critical challenges impeding the application of oxidative desulfurization (ODS) to slurry oil, specifically its complex composition, high sulfur content, prohibitively high viscosity, and inefficient oil-water interfacial mass transfer. An innovative ODS strategy based on biphasic interface regulation was proposed. By constructing a catalytic system through the combination of polyoxometalate and organic cationic modifiers to stabilize the oil-water interface, enhanced mass transfer efficiency was achieved. These catalysts function as surfactant-like homogeneous catalysts during H2O2 mediated oxidation, while enabling rapid separation after reaction. Systematic model system studies identified catalysts with exceptional sulfur-oxidation selectivity, operating via dynamic peroxo-species formation from terminal oxygen of W=O activation by superoxide radicals. Deployment in real slurry oil under Bayesian-optimized conditions reduced sulfur content from 1.60 wt% to 0.34 wt% while completely preserving the core feedstock components 3-4 ring aromatic components and maintaining 86.4% slurry recovery. This research provides a technologically innovative and practically viable pathway for desulfurization of slurry oils with remaining high aromatic contents.

Key words: FCC slurry oil, Oxidative desulfurization, Interface catalysis, Needle coke, Machine learning