Unraveling phosphorus enrichment, transformation, and optimal process strategies in sewage sludge pyrolysis using a...

Global phosphorus (P) resources are facing a depletion crisis, and pyrolysis of P-rich sewage sludge (SS) offers significant resource potential. Optimizing pyrolysis conditions remains key yet challenging for enhancing P retention and bioavailability. This study conducted a correlation-prediction-causation integrated framework (CPCIF) to investigate how heating temperature (HT), heating rate (HR), and retention time (RT) influence total P enrichment rate (BTPE), relative inorganic P transformation rate (BITP), and relative apatite P transformation rate (BAIP) from SS to biochar during pyrolysis. Different HT exhibit significant zonal effects on the linear relationships, with high temperatures promoting BAIP but inhibiting BTPE. A Bayesian optimization-Random Forest model accurately predicted all targets (R² = 0.80-0.93). However, factors like H/C, N/C, and pH confounded the relationships between pyrolysis parameters and targets, leading machine learning to underestimate the actual contributions. Double machine learning corrected these biases, revealing negative average treatment effects (ATEs) of HT, HR, and RT on BTPE (-0.190 to -0.006), and a strong positive effect of HR on BAIP. Conditional ATEs demonstrated that HT-HR and HT-RT interactions shape P outcomes. CPCIF attribution analysis via entropy weight and Technique for Order Preference by Similarity to an Ideal Solution highlighted that causal components dominate the weights for all targets (45.5-61.1%), with HT and HR being key control variables for P enrichment and transformation, respectively. A two-stage pyrolysis strategy (low-to-high HT and HR) is proposed to optimize both BTPE and BAIP. This study offers theoretical support and technical pathways for SS resource utilization and sustainable P management.