Getting lower density and voids inside box during NPT run

GROMACS version:2021.2
GROMACS modification: No
Dear Developers and Users,
I am simulating some polymer chains. I am getting a lower value for density in comparison to experimental density during the NPT run. When I am visualizing the system, there are some voids are there inside the box. How can I remove those empty spaces? I am attaching the mdp file script

integrator = md ; leap-frog integrator
nsteps = 10000000 ; 10ns
dt = 0.0001 ; 1 fs
; Output control
nstenergy = 10000 ; save energies every 1.0 ps
nstlog = 10000 ; update log file every 1.0 ps
nstxout-compressed = 10000 ; save coordinates every 1.0 ps
; Bond parameters
continuation = yes ; continuing from NVT
constraint_algorithm = lincs ; holonomic constraints
constraints = h-bonds ; bonds to H are constrained
lincs_iter = 1 ; accuracy of LINCS
lincs_order = 4 ; also related to accuracy
; Neighbor searching and vdW
cutoff-scheme = Verlet
ns_type = grid ; search neighboring grid cells
nstlist = 20 ; largely irrelevant with Verlet
rlist = 1.2
vdwtype = cutoff
vdw-modifier = force-switch
rvdw-switch = 1.0
rvdw = 1.2 ; short-range van der Waals cutoff (in nm)
; Electrostatics
coulombtype = PME ; Particle Mesh Ewald for long-range electrostatics
rcoulomb = 1.2
pme_order = 4 ; cubic interpolation
fourierspacing = 0.16 ; grid spacing for FFT
; Temperature coupling
tcoupl = V-rescale ; modified Berendsen thermostat
tc-grps = system ; two coupling groups - more accurate
tau_t = 0.1 ; time constant, in ps
ref_t = 425 ; reference temperature, one for each group, in K
; Pressure coupling
pcoupl = berendsen;Parrinello-Rahman ; pressure coupling is on for NPT
pcoupltype = semiisotropic
; uniform scaling of box vectors
tau_p = 1.0 ; time constant, in ps
ref_p = 1.0 1.0 ; reference pressure, in bar
compressibility = 4.5e-5 0.0 ; isothermal compressibility of water, bar^-1
refcoord_scaling = com
; Periodic boundary conditions
pbc = xyz ; 3-D PBC
; Dispersion correction is not used for proteins with the C36 additive FF
DispCorr = EnerPres
; Velocity generation
gen_vel = no ; velocity generation off after NVT

Hi,

Just a observation: you couple the pressure in x/y direction while z is kept constant. Is this what you want?
This setting does not allow to relax z direction and that may be the source of the gaps in the system (just a guess).
Best regards
Alessandra