Authentications and encryptions are essential in the fast growing hardware devices used for security applications in the Internet of Things (IoT) era. However, existing cryptographic solutions may not be applicable to such IoT devices because of their extremely limited resources such as computation power and battery. On the other hand, hardware primitives have been proposed as a promising alternative for IoT applications because of their low cost and high energy efficiency. In this paper, we propose a novel application of the well-studied Physical Unclonable Function (PUF), where we use PUF circuit to improve the entropy of an input source. As an example, we develop a PUF-based password strength enhancer (PUF-PassSE) that can significantly improve the security, measured by the entropy, of the human-generated passwords. The proposed design is prototyped on the Nexys 4 DDR FPGA board to demonstrate its low hardware overhead, which ensures its applicability in IoT applications.