Challenges in CDFT Geometry Optimization Using aug-cc-pVDZ

I performed CDFT geometry optimization and frequency calculations using the wB97X-D/6-31G** level of theory, and the calculations converged successfully without any imaginary frequencies. However, when I switched to wB97X-D/aug-cc-pVDZ and started from the previously optimized geometry, the calculation struggled to converge, particularly in the CDFT iterations, even though I used the same charge constraints as before. I was wondering what could be causing this convergence issue with the larger basis set.

$molecule
-1 2
C      -1.42918000    -0.96239000     0.76804000
C      -2.38999000    -0.71226000     1.76600000
C      -3.70560000    -1.14913000     1.62427000
C      -4.13143000    -1.85459000     0.47942000
C      -3.17186000    -2.05506000    -0.54573000
C      -1.84751000    -1.61710000    -0.39853000
C      -5.53834000    -2.41110000     0.37868000
C      -6.40499000    -2.42655000     1.50529000
C      -7.68584000    -2.97900000     1.42789000
C      -8.14370000    -3.52641000     0.23498000
C      -7.32045000    -3.52269000    -0.88879000
C      -6.03398000    -2.97553000    -0.82153000
H      -2.13738000    -0.20363000     2.68385000
H      -4.36374000    -0.93104000     2.44836000
H      -3.40477000    -2.58650000    -1.45246000
H      -1.13542000    -1.83127000    -1.18664000
H      -6.11684000    -2.04115000     2.46722000
H      -8.32276000    -2.98808000     2.30432000
H      -9.13551000    -3.95709000     0.18212000
H      -7.67909000    -3.95077000    -1.81548000
H      -5.45029000    -3.00755000    -1.72665000
C       1.95962000     1.08866000     0.40783000
C       2.73669000     0.08622000     1.27473000
C       1.95006000    -0.21545000     2.56750000
C       0.41894000    -0.16419000     2.37837000
C       0.04958000    -0.62625000     0.96187000
C       0.63046000     0.39117000    -0.08663000
H       1.67782000     1.93542000     1.07410000
H       3.75275000     0.41047000     1.57880000
H       2.86275000    -0.83570000     0.66318000
H       2.28153000    -1.22044000     2.90823000
H       2.19384000     0.51358000     3.37034000
H       0.04345000     0.87780000     2.49219000
H      -0.02615000    -0.78895000     3.18227000
H       0.58406000    -1.59745000     0.84816000
H       0.78550000    -0.20734000    -1.00771000
H      -0.09173000     1.20507000    -0.31605000
C       7.16138000     3.97122000    -1.47806000
C       6.97234000     3.86451000    -2.79543000
C       5.72940000     3.23890000    -3.30226000
C       4.79891000     2.76549000    -2.47038000
C       5.01576000     2.88113000    -1.01928000
C       6.12582000     3.46041000    -0.55808000
C       3.68281000     2.18694000    -2.91535000
C       2.68230000     1.60368000    -1.98745000
C       2.88208000     1.68365000    -0.65803000
C       4.10272000     2.39727000    -0.18276000
H       8.06545000     4.42568000    -1.08431000
H       7.72337000     4.23243000    -3.48720000
H       5.60345000     3.16492000    -4.37954000
H       6.30668000     3.55641000     0.51078000
H       3.49643000     2.09661000    -3.98304000
H       1.81844000     1.11273000    -2.41511000
H       4.24931000     2.50295000     0.89024000
$end

$rem
JOB_TYPE              Optimization
METHOD                wB97X-D
BASIS                 6-31G(d,p)
SCF_CONVERGENCE       8
SCF_MAX_CYCLES        500
GEOM_OPT_MAX_CYCLES   200
SYMMETRY              FALSE
SYM_IGNORE            TRUE
GUI                   2
CDFT                  TRUE
CDFT_PRINT            TRUE
CDFT_BECKE_POP        TRUE
CDFT_THRESH           8
BECKE_SHIFT           UNSHIFTED
XC_GRID               2
$end

$cdft
1.0
1  1  21
$end

@@@

$molecule
read
$end

$rem
JOB_TYPE              Freq
METHOD                wB97X-D
BASIS                 6-31G(d,p)
SCF_CONVERGENCE       8
SYMMETRY              FALSE
SYM_IGNORE            TRUE
SCF_GUESS             READ
CDFT                  TRUE
CDFT_PRINT            TRUE
CDFT_BECKE_POP        TRUE
CDFT_THRESH           8
BECKE_SHIFT           UNSHIFTED
XC_GRID               2
MEM_TOTAL             248000
MEM_STATIC            4000
$end

$cdft
1.0
1  1  21
$end

wB97X-D/aug-cc-pVDZ theory: I tried with just charge constraint and also adding spin constraints. But in both cases, it did not converge.

$molecule
-1 2
C      -1.05752000    -1.39603000     0.86890000
C      -1.98340000    -1.07935000     1.87509000
C      -3.34662000    -1.24635000     1.73247000
C      -3.93541000    -1.76228000     0.52891000
C      -2.97880000    -2.08119000    -0.49354000
C      -1.62407000    -1.89930000    -0.31619000
C      -5.34379000    -1.95340000     0.36716000
C      -6.31069000    -1.47018000     1.31826000
C      -7.66613000    -1.66360000     1.15836000
C      -8.19340000    -2.35200000     0.05048000
C      -7.27318000    -2.83557000    -0.89836000
C      -5.91515000    -2.64747000    -0.75702000
H      -1.58975000    -0.70472000     2.81804000
H      -3.98378000    -1.00946000     2.57968000
H      -3.32679000    -2.46107000    -1.44983000
H      -0.94209000    -2.15306000    -1.12869000
H      -5.97198000    -0.90580000     2.18253000
H      -8.34434000    -1.26407000     1.91282000
H      -9.26240000    -2.50178000    -0.06982000
H      -7.63806000    -3.37920000    -1.77023000
H      -5.25798000    -3.06873000    -1.51271000
C       2.32798000     0.56276000     0.74525000
C       3.14409000    -0.64894000     1.22539000
C       2.49538000    -1.29493000     2.46371000
C       0.95405000    -1.19003000     2.45442000
C       0.44023000    -1.18585000     1.00607000
C       0.89635000     0.13018000     0.31945000
H       2.23755000     1.24048000     1.60426000
H       4.17184000    -0.34247000     1.44840000
H       3.21508000    -1.37812000     0.40891000
H       2.79317000    -2.34859000     2.51161000
H       2.88750000    -0.82228000     3.37161000
H       0.64954000    -0.26026000     2.95605000
H       0.52972000    -2.01873000     3.03046000
H       0.95180000    -2.01863000     0.48928000
H       0.84777000     0.00310000    -0.76789000
H       0.20224000     0.94114000     0.56352000
C       5.43790000     5.25355000    -1.94193000
C       5.64548000     4.63013000    -3.19452000
C       5.16854000     3.36453000    -3.42033000
C       4.46393000     2.66271000    -2.40928000
C       4.25120000     3.28906000    -1.15122000
C       4.75746000     4.59930000    -0.94725000
C       3.95839000     1.35053000    -2.60502000
C       3.28103000     0.70388000    -1.60857000
C       3.05532000     1.31836000    -0.34497000
C       3.54208000     2.58620000    -0.14148000
H       5.81806000     6.25633000    -1.77219000
H       6.18181000     5.15878000    -3.97642000
H       5.32100000     2.88069000    -4.38152000
H       4.59600000     5.07771000     0.01512000
H       4.11384000     0.86582000    -3.56523000
H       2.90138000    -0.29899000    -1.78267000
H       3.38478000     3.07583000     0.81721000
$end

$rem
JOB_TYPE              Optimization
METHOD                wB97X-D
BASIS                 aug-cc-pVDZ
SCF_CONVERGENCE       8
SCF_MAX_CYCLES        500
GEOM_OPT_MAX_CYCLES   200
SYMMETRY              FALSE
SYM_IGNORE            TRUE
GUI                   2
CDFT                  TRUE
CDFT_PRINT            TRUE
CDFT_BECKE_POP        TRUE
BECKE_SHIFT           UNSHIFTED
XC_GRID               2
MEM_TOTAL             248000
MEM_STATIC            4000
$end

$cdft
1.0
1  1  21
1.0
1  1  21  s
$end

@@@

$molecule
read
$end

$rem
JOB_TYPE              Freq
METHOD                wB97X-D
BASIS                 aug-cc-pVDZ
SCF_CONVERGENCE       8
SYMMETRY              FALSE
SYM_IGNORE            TRUE
SCF_GUESS             READ
CDFT                  TRUE
CDFT_PRINT            TRUE
CDFT_BECKE_POP        TRUE
BECKE_SHIFT           UNSHIFTED
XC_GRID               2
MEM_TOTAL             248000
MEM_STATIC            4000
$end

$cdft
1.0
1  1  21
1.0
1  1  21  s
$end

It also doesn’t converge for me with aug-cc-pVDZ, even with a EML(99,590) grid. May need to consider a better guess (e.g., read MOs from previous small-basis job using BASIS2) or else use a different convergence algorithm. There’s no guarantee that it can be made to converge with the constraints.