Generally, I had no great success with QST2 nor QST3

The best practice to locate a TS is

a) perform a relaxed scan along the reaction coordinate (bond) under study

you could use something like

# opt=modredundant uMN15L/6-31+g(d,p)

Then add the details at the bottom of the file

B 1 2 S 20 0.100000

b) Take the highest energy point from step a and save it as input file for your TS search , then locate the TS using

# opt=(calcfc,ts,noeigen) uMN15L/6-31+g(d,p) freq

noeigen ensures that you will have only one imag freq

Also, take a look at

https://doi.org/10.1039/D3RE00036B

https://doi.org/10.1021/acs.jpca.8b08071

https://doi.org/10.1039/C7CP06082C

https://doi.org/10.1039/C6RA07761G

https://doi.org/10.1021/jp3087122

Good luck