The Since their inception, phishing emails have become a significant threat in cyberspace, favored by many attackers as one of their preferred techniques. Phishing emails typically employ psychological manipulation techniques from social engineering, aiming to persuade victims to perform actions expected by attackers. These tactics include creating a sense of urgency, offering false attractive rewards, and intimidation, thereby coercing victims into unknowingly revealing personal information or clicking on malicious links. In this study, we introduce an advanced method that combines Large Language Models (LLMs) with sentiment analysis features. Utilizing prompt engineering techniques, we guide large language models to accurately identify and judge the content of emails, effectively detecting phishing attempts. We employed publicly available datasets commonly used in previous research for testing, and our experimental results demonstrate our method's significant detection capability. Moreover, our research not only highlights the superiority of LLMs in identifying phishing emails but also clarifies their practical application in email security. However, we also address the limitations of this approach and discuss important directions for future developments in phishing email detection technology.