跳到主要內容 :::

年度	113	專案性質	實驗性質
專案類別	模場試驗	研究主題	整治
申請機構	國立中興大學	申請系所	環境教育暨永續科技研發中心
專案主持人	張書奇	職等/職稱	副教授
專案中文名稱	以負碳為基礎之污染底泥整治模場試驗
中文關鍵字	負碳、多氯聯苯、底泥、整治、模場試驗
專案英文名稱	A pilot study of contaminated sediment remediation based on carbon-negative technology
英文關鍵字	Carbon negative; Polychlorinated biphenyls; Sediment; Remediation; Pilot study
執行金額	1,870,000元
執行期間	2024/12/1 至 2026/11/30
計畫中文摘要	工業發達國家之底泥污染情況相當普遍，河川下游之底泥與河口附近海水底泥均已遭受嚴重污染，其中尤以疏水性有機物及重金屬污染最為嚴重，而台灣之底泥污染物檢測濃度在全世界相關文獻報導中屢屢名列前茅。如二仁溪與南崁溪底泥中之多氯聯苯（polychlorinated biphenyls, PCBs）。這些污染物藉由食物鏈破壞生態系及影響人體健康，急需發展經濟有效之整治技術。近年來，國際間研究已經顯示厭氧生物整治因會產生溫室效應相當高的甲烷氣體，而被視為較不永續之整治技術，且循環經濟與 2050 淨零排放議題甚囂塵上，本年度計畫則是採用本土優養化藻類作為生物整治助劑進行底泥中 Aroclor 1254 污染整治技術研發，並結合人工智慧之機器學習技術，將整治技術研發所產生之大量數據進行機器學習，有助於現場整治技術提升，一舉達成循環經濟、永續整治、減少碳排與優養化整治的多重目標。本計畫為 2 年期計畫，第 1 年計畫為實驗室規模，第 2 年為現地模場試驗。目前執行進度均符合預定期程。以目前成果而言，以藻類為助劑之田口試驗結果顯示：以小球藻為助劑下，最佳條件為低溫培養（15°C）、中性 pH 值（pH 7.0）、低有機質（0.03%）與低藻類基質濃度（50 mg/L），且以 pH 值影響最顯著。功能性基因 rdh12 與 Aroclor 1254 移除率相關性最高，可做為後續模場試驗之指標。降解表現最佳的 G1 可以少數優勢菌群，在低多樣性條件下達成最佳降解效率；對菌群結構影響最大菌種為 Citrobacter freundii 8090 ，其 次為 Desulfomicrobium norvigecum ，再 其次 為 Trichococcus patagoniensis。對菌群結構影響最大之環境因子為底泥有機質，其次為 pH 值。由功能預測可知中性偏低 pH 值有利於木質素分解與芳香族化合物降解。機器學習以 LASSO 模式之表現最佳，其訓練階段與測試階段之 R2、均可達 1.00000，其次為 Elastic Net，重要性最高前 20 名菌種中除 Stutzerimonas chloritidismutans、 Stutzerimonas stutzeri 及 Enterobacter soli ATCC BAA-2102 可能與PCBs 降解直接相關，其他菌種多為發酵產氫菌與產酸菌。藻類助劑相較以往之大豆油乳化液，不僅具備較低的生命週期碳排放量，且能協助解決優養化問題並實現循環經濟，具發展潛力與實務應用價值。第 1 年計畫預期產出 2 篇 SCI 期刊論文，2 篇研討會論文，其中 2 篇研討會論文已經完成 2025 環工年會投稿並獲口頭發表，2 篇 SCI 期刊論文預計在 2026/5/15 前完成。
計畫英文摘要	Sediment pollution is widespread in industrialized regions. Sediments in the lower reaches of rivers and near estuaries are often severely contaminated, with hydrophobic organic compounds and heavy metals being the most prevalent pollutants. In Taiwan, sediment contamination levels consistently rank among the highest reported worldwide. For instance, polychlorinated biphenyls (PCBs) have been detected in the sediments of the Erren and Nankan Rivers. These pollutants pose serious threats to ecosystems and human health through bioaccumulation in the food chain, highlighting the urgent need for cost-effective remediation technologies. Recent international studies have shown that anaerobic bioremediation, although effective, is considered less sustainable due to methane emissions, a potent greenhouse gas. Meanwhile, the concepts of a circular economy and the global 2050 net-zero emissions target have gained increasing attention. This project aims to develop a remediation technology for Aroclor 1254-contaminated sediments using native eutrophic algae as bioremediation adjuvants. By integrating artificial intelligence and machine learning, large datasets generated during the remediation process will be analyzed to optimize field-scale applications. The proposed approach simultaneously addresses multiple goals: promoting a circular economy, achieving sustainable remediation, reducing carbon emissions, and mitigating eutrophication. This two-year project involves laboratory-scale experiments in the first year and field trials in the second year. Progress to date is on schedule. Preliminary Taguchi experiments using algal biomass identified optimal conditions for Chlorella vulgaris as an adjuvant: low temperature (15 °C), neutral pH (7.0), low organic matter content (0.03%), and a low algal substrate concentration (50 mg/L), with pH being the most influential factor. Functional genes ardA and rdh12 correlated strongly with Aroclor 1254 removal rates and can serve as indicators for future field applications. Among tested algal biomasses, G1 demonstrated the highest degradation efficiency under low-diversity conditions dominated by a small bacterial population. The key bacterial species influencing community structure were Citrobacter freundii 8090, Desulfomicrobium norvegicum, and Trichococcus patagoniensis. Environmental factors shaping biomass composition were primarily sediment organic matter, followed by pH. Functional prediction analyses suggested that neutral to slightly acidic conditions favor lignin decomposition and aromatic compound degradation. Among the machine learning models tested, the LASSO model performed best, achieving an R² of 1.0000 in both training and testing phases, followed by the Elastic Net model. Of the 20 most influential bacterial species, Stutzerimonas chloritidismutans, Stutzerimonas stutzeri, and Enterobacter soli (ATCC BAA-2102) are potentially involved in PCB degradation, while most others are hydrogen- and acid-producing bacteria. Compared with conventional soybean oil emulsions, algae-based additives offer lower life-cycle carbon emissions and help alleviate eutrophication, demonstrating strong potential for practical application within a circular economy framework. In this first project year, two SCI-indexed journal articles and two conference papers are planned. Two conference papers have already been submitted and presented at the 2025 Environmental Engineering Annual Meeting, and the two SCI papers are expected to be completed by May 15, 2026.