Force convergence problem in ion adsorption case.

[GoodZhenLi]

Prblem statement
I built a Zn (101) slab model (the 24 angstrom vaccumn layer is along with y axis where the slab is located at the center of the whole cell on the direction of y axis) with $SO_4^{2-}$ adsorbate to calculate the adsorption energy. But it is hard to get a convergent relaxation (the largest force is around 0.2 eV/A) results. Here is my INPUT files. Can someone give me some advice?

INPUT
suffix ABACUS # 输出文件名称后缀
calculation relax # 结构弛豫
nelec_delta 2
basis_type lcao # 使用LCAO基组，对于大体系更高效
ecutwfc 100 # 平面波动能截断半径 (Ry)
scf_thr 2e-6 # 电子自洽收敛标准 (Ry)
scf_nmax 60 # 电子自洽最大迭代步数
chg_extrap atomic # 电荷密度外推方法
force_thr_eV 0.02 # 原子力收敛标准 (eV/Å)
relax_nmax 200 # 离子弛豫最大步数
out_stru 1 # 输出弛豫后的结构
relax_cg_thr 0.10
relax_method cg
kspacing 0.14
kpar 4
symmetry 0
dft_functional pbe
#vdw_method d3_0
efield_flag 1
dip_cor_flag 1
efield_dir 1
#Parameters (7.Compensating charge)
gate_flag 1
zgate 0.9
imp_sol 1 # Turn on implicit solvation
eb_k 80 # Dielectric constant of water

[LKFEIYI]

1. The implicit solvation model implemented by ABACUS is only compatible with a charge-neutral system. When calculating charged species, the solvation energies are incorrect.
2. Do not use "Compensating charge" in such a system, because the charged plate will attract the charged species. You may see the SO4-2 approaching the gate. You can refer to the discussion here: Reliability of ABACUS in charged system #5645 (comment)

[GoodZhenLi]

Thanks for your helpful advice, but I still suffuer force convergence problem. I guess it may come from the irregular Zn (101) surface and unsuitable initial configuration.