X-ray CT (Computed Tomography) is a technique that can generate cross-sectional images of an object. This technology has spread rapidly in the field of diagnosis and inspection, and have provided significant benefits for our lives. However, when metal objects are present in the X-ray projection, a strong radial noise called metal artifacts arises and makes it difficult to diagnose and inspect. Even though various studies to reduce metal artifacts have been conducted, the problem is not resolved yet. In order to reduce these artifacts, we have proposed new algorithms based on the FBP(Filtered back projection), which remains the most widely used reconstruction method in CT. On the other hand, iterative reconstruction (IR) has emerged as a new CT reconstruction technique in recent years. IR requires a lot of computation time, however, it enables reduction in pixel noise or artifacts. In this paper, we proposed a new reconstruction algorithm to reduce metal artifacts based on IR. The algorithm utilizes the energy information to apply the exact successive approximation. Validation was performed using a simple experimental phantom. Our results demonstrate that the new algorithm effectively reduces metal artifacts in certain situations.