When I run solution blank (2% HNO3), everything looks good. When it comes to air blank, Mn intensity is really high. I have tried to clean spray chamber but the problem still remains.
https://www.perkinelmer.com/CMSResources/Images/44-74379ATL_TableOfPolyatomicInterferences.pdf list 10 possible polyatomic interferences for Mn. These include 40Ar14N1H , and 40Ar15N . As both Ar and N are present in the plasma (N from the atmosphere) you are possibly measuring these transient molecular species.
Is the air blank intended to detect these molecules so that you can adjust plasma settings and detector setting to reduce their occurrence?
Before you start looking for a more complex cause of the Mn presence signal, it is better to be sure that it is not contamination. Use a different burner, preferably not used at all or for solutions with Mn.
If it is not a contamination, then the cause is air (N2 approx. 80%) and water vapor introduced to the plasma (the chamber is moist inside and this water evaporates). The signal could therefore come from 40Ar14N1H as well as other complex ions (see Matthew's explanation). Try to keep the plasma pure argon without sucking or blowing air (pump stop). If the signal drops significantly, we have an explanation. Such effects will not make it difficult for you to measure manganese signals, except at extremely low concentrations.