I suggest using tigher SCF and opt thresholds, and slightly adjusting structure along imaginary mode, and then retry. Try you best to eliminate any imaginary frequency. (However, if you make sure that those groups have little impact on the result of your interest, you may simply replace them by hydrogen atoms...)

Dear Prof. Tian Lu,
Thank you for the information. I don't have access to Gaussian 16.
All the five imaginary frequencies belong to the rotation of individual methyl groups of t-butyl group or whole t-butyl group rotations only. Can I ignore them? If I run the ESD(ABS) for Full AH UV-vis spectrum or ESD(FLUOR), I need ESHessian. If ES Hessian derived from the geometry having multiple imaginary frequencies won't give error in my calculation?
Here is the output:
-----------------------
VIBRATIONAL FREQUENCIES
-----------------------

Scaling factor for frequencies =  1.000000000  (already applied!)

     0:       0.00 cm**-1
     1:       0.00 cm**-1
     2:       0.00 cm**-1
     3:       0.00 cm**-1
     4:       0.00 cm**-1
     5:       0.00 cm**-1
     6:    -144.10 cm**-1  ***imaginary mode***
     7:     -96.33 cm**-1  ***imaginary mode***
     8:     -71.26 cm**-1  ***imaginary mode***
     9:     -61.98 cm**-1  ***imaginary mode***
    10:     -40.99 cm**-1  ***imaginary mode***
    11:      17.69 cm**-1

There is no evident problem in your keywords. I suggest carefully checking atomic motions of the imaginary frequencies to evaluate the possible reasons. (or using Gaussian 16 instead, which is more robust and has analytical Hessian of TDDFT)

I am trying to optimize the S1 excited state geometry (input: GS geometry coordinates) with the following input.

"! LibXC(B3LYP) D3BJ RIJCOSX def2-SVP TightSCF TightOPT DEFGRID3 OPT FREQ
! PrintBasis MiniPrint PrintMOs
! AIM xyzfile pdbfile

%pal
   nprocs 32
end

%maxcore 20000

%TDDFT
NROOTS 5
IROOT 1
FOLLOWIROOT TRUE
TDA FALSE
END

*xyz 0 1"

I ended up 5 imaginary frequencies. Am I doing any mistake. Any guidance in this problem is appreciated.
Thank you