This paper presents a case study on a process excursion where a subtle defect spray with twelve pairs of defects aggregated flow pattern on the front side of the wafer. The defect of interest is molten tungsten (W) balls which are generated in a dielectric etch chamber caused by plasma arcing between one part of the etch chamber and the dissimilar W film remaining on the wafer bevel. Observations of defect scans and date review collection show that the molten W balls never distribute randomly far from the arc crater in dielectric etching, but spray out over the surface of the wafer in subsequent oxygen ash processing. A possible formation mechanism for the defect spray is proposed. The two primary assumptions involved in the formation of such spray are non-uniform conductive oxygen fluid flowing over the surface of the wafer, and charged molten W balls being moved on the surface of the wafer. To eliminate the defect spray, several promising solutions are identified. This case study confirms that high conductivity p/p + -type Si substrate or the use of a different model of ash chamber with more uniform optimal gas flow distribution will improve yield.