To obtain dI/dV vs V you need to run STM in spectroscopy mode. Some of the STMs provides dI/dV vs V (bias) plot, and some others convert I-V plots into dI/dV vs V using its software. If none of these is available in your system, you have to post process your IV data.
Around EF, I and V are very small, obviously. Therefore you may get artifacts there when I approaches 0 faster than V does because dI/dV/(I/V) contains a "division by almost zero". Maybe it is sufficient to just smooth your raw I/V data before calculating the derivative and performing the division.
Hi, Wang Zhenyu if your system has spectroscopy mode it is advisable to run the system in the spectroscopy mode to obtain dI/dV/(I/V) data. But if you want to calculate dI/dV/(I/V) from the I-V data you need to smoothen the raw data although in my experience both data will be different. if you want you can check this paper.
Article Two-dimensional Nanohybrid of MoS2 and Rose Bengal: Facile S...