Any parameters are better than no parameters. Part of using a FF is checking that results make sense, so if you can find something in this bare market for anions, try it. Proof will be in the pudding.
LigParGen can only generate ions between -2 and +2 charge, so $\ce{PO4^3-}$ is out. When I tried with $\ce{SO4^2-}$, it did not work. $\ce{H2SO4}$ also did not work. I then tried $\ce{H3PO4}$ (via SMILES) and it(LigParGen) did not work. I then tried $\ce{H3PO4}$ via pdb file, it worked. I tried $\ce{CO3^2-}$, it did not work, but $\ce{H2CO3}$ did work. I have attached the $\ce{H3PO4}$ and $\ce{H2CO3}$ pdbs so you can remake them if you wish, but also attach the GROMACS topology files (actually I have pasted them at the bottom... I don't know how to attach them...).
I did a quick and dirty geometry optimization in Avogadro using GAFF, so, we know that GAFF at least can do the anions(Avogadro says the molecular formula includes the hydrogens, so I am not sure if it is really doing the anion? I removed hydrogens anyways. Atomic properties does not list hydrogens...), so I would suggest looking into GAFF along with OPLS/AA.
My suggestion is to try using GAFF to get the bond/angle parameters (maybe GAFF[Antechamber] will even do the charges too, but I wouldn't want to use them). I would suggest generating MBIS partial charges. If GAFF will do the anions, then use the LJ parameters it generates. If it does not, generate the neutral molecules, for instance, $\ce{H3PO4}$ and $\ce{H2SO4}$ and "borrow" the LJ parameters as well as bond/angle parameters. You can tweak the bond/angle parameters using QM if you wish. Then get partial charges...I would use a partitioning method over a random number generator like RESP, but regardless, you can generate the partial charges using QM programs. Gaussian will do ESP/Hirshfeld/CM5 and many others. HORTON will do MBIS for you.
It is hopefully worth making the forcefield in this somewhat Frankenstein manner if you can't find one. The anions are small so you should be able to "tweak" parameters to make it work for the property you are interested in. Only trial and error will tell :S
$\ce{H3PO4}$ PDB file
COMPND UNNAMED
AUTHOR GENERATED BY OPEN BABEL 2.3.90
HETATM 1 P UNL 1 -1.464 0.745 -2.751 1.00 0.00 P
HETATM 2 O UNL 1 -2.421 1.312 -3.725 1.00 0.00 O
HETATM 3 O UNL 1 -1.527 -0.862 -2.535 1.00 0.00 O
HETATM 4 H UNL 1 -0.784 -1.161 -1.978 1.00 0.00 H
HETATM 5 O UNL 1 -1.530 1.339 -1.276 1.00 0.00 O
HETATM 6 O UNL 1 0.076 0.988 -3.056 1.00 0.00 O
HETATM 7 H UNL 1 -2.453 1.221 -1.091 1.00 0.00 H
HETATM 8 H UNL 1 0.145 0.489 -3.857 1.00 0.00 H
CONECT 1 2 3 5 6
CONECT 2 1
CONECT 3 1 4
CONECT 4 3
CONECT 5 1 7
CONECT 6 1 8
CONECT 7 5
CONECT 8 6
MASTER 0 0 0 0 0 0 0 0 8 0 8 0
END
OPLS/AA FF for $\ce{H3PO4}$
;
; GENERATED BY LigParGen Server
; Jorgensen Lab @ Yale University
;
[ atomtypes ]
opls_806 H806 1.0080 0.000 A 0.00000E+00 0.00000E+00
opls_801 O801 15.9990 0.000 A 2.96000E-01 8.78640E-01
opls_800 P800 30.9738 0.000 A 3.74000E-01 8.36800E-01
opls_807 H807 1.0080 0.000 A 0.00000E+00 0.00000E+00
opls_805 O805 15.9990 0.000 A 3.12000E-01 7.11280E-01
opls_802 O802 15.9990 0.000 A 3.12000E-01 7.11280E-01
opls_804 O804 15.9990 0.000 A 3.12000E-01 7.11280E-01
opls_803 H803 1.0080 0.000 A 0.00000E+00 0.00000E+00
[ moleculetype ]
; Name nrexcl
UNL 3
[ atoms ]
; nr type resnr residue atom cgnr charge mass
1 opls_800 1 UNL P00 1 2.8980 30.9738
2 opls_801 1 UNL O01 1 -1.2051 15.9990
3 opls_802 1 UNL O02 1 -1.0619 15.9990
4 opls_803 1 UNL H03 1 0.4728 1.0080
5 opls_804 1 UNL O04 1 -1.0008 15.9990
6 opls_805 1 UNL O05 1 -0.9984 15.9990
7 opls_806 1 UNL H06 1 0.4508 1.0080
8 opls_807 1 UNL H07 1 0.4445 1.0080
[ bonds ]
2 1 1 0.1480 439320.000
3 1 1 0.1610 192464.000
4 3 1 0.0945 462750.400
5 1 1 0.1610 192464.000
6 1 1 0.1610 192464.000
7 5 1 0.0945 462750.400
8 6 1 0.0945 462750.400
[ angles ]
; ai aj ak funct c0 c1 c2 c3
2 1 3 1 108.230 836.800
1 3 4 1 108.500 460.240
2 1 5 1 108.230 836.800
2 1 6 1 108.230 836.800
1 5 7 1 108.500 460.240
1 6 8 1 108.500 460.240
3 1 5 1 102.600 376.560
5 1 6 1 102.600 376.560
3 1 6 1 102.600 376.560
[ dihedrals ]
; IMPROPER DIHEDRAL ANGLES
; ai aj ak al funct c0 c1 c2 c3 c4 c5
[ dihedrals ]
; PROPER DIHEDRAL ANGLES
; ai aj ak al funct c0 c1 c2 c3 c4 c5
4 3 1 2 3 0.000 0.000 0.000 -0.000 -0.000 0.000
7 5 1 2 3 0.000 0.000 0.000 -0.000 -0.000 0.000
8 6 1 2 3 0.000 0.000 0.000 -0.000 -0.000 0.000
7 5 1 3 3 0.000 0.000 0.000 -0.000 -0.000 0.000
8 6 1 5 3 0.000 0.000 0.000 -0.000 -0.000 0.000
8 6 1 3 3 0.000 0.000 0.000 -0.000 -0.000 0.000
7 5 1 6 3 0.000 0.000 0.000 -0.000 -0.000 0.000
5 1 3 4 3 0.000 0.000 0.000 -0.000 -0.000 0.000
6 1 3 4 3 0.000 0.000 0.000 -0.000 -0.000 0.000
[ pairs ]
2 4 1
4 5 1
2 7 1
4 6 1
3 7 1
2 8 1
3 8 1
6 7 1
5 8 1
$\ce{H2CO3}$ PDB file
COMPND UNNAMED
AUTHOR GENERATED BY OPEN BABEL 2.3.90
HETATM 1 C UNL 1 0.943 0.994 -1.701 1.00 0.00 C
HETATM 2 O UNL 1 1.658 0.127 -2.342 1.00 0.00 O
HETATM 3 O UNL 1 1.315 2.096 -2.237 1.00 0.00 O
HETATM 4 H UNL 1 1.969 -0.472 -1.669 1.00 0.00 H
HETATM 5 H UNL 1 1.933 1.686 -2.870 1.00 0.00 H
HETATM 6 O UNL 1 0.632 0.886 -0.535 1.00 0.00 O
CONECT 1 2 3 6
CONECT 2 1 4
CONECT 3 1 5
CONECT 4 2
CONECT 5 3
CONECT 6 1
MASTER 0 0 0 0 0 0 0 0 6 0 6 0
END
$\ce{H2CO3}$ Topology file
;
; GENERATED BY LigParGen Server
; Jorgensen Lab @ Yale University
;
[ atomtypes ]
opls_804 H804 1.0080 0.000 A 0.00000E+00 0.00000E+00
opls_801 O801 15.9990 0.000 A 3.12000E-01 7.11280E-01
opls_805 O805 15.9990 0.000 A 2.96000E-01 8.78640E-01
opls_802 O802 15.9990 0.000 A 3.12000E-01 7.11280E-01
opls_803 H803 1.0080 0.000 A 0.00000E+00 0.00000E+00
opls_800 C800 12.0110 0.000 A 3.55000E-01 2.92880E-01
[ moleculetype ]
; Name nrexcl
UNL 3
[ atoms ]
; nr type resnr residue atom cgnr charge mass
1 opls_800 1 UNL C00 1 0.6339 12.0110
2 opls_801 1 UNL O01 1 -0.6044 15.9990
3 opls_802 1 UNL O02 1 -0.4751 15.9990
4 opls_803 1 UNL H03 1 0.4774 1.0080
5 opls_804 1 UNL H04 1 0.475 1.0080
6 opls_805 1 UNL O05 1 -0.5067 15.9990
[ bonds ]
2 1 1 0.1364 376560.000
3 1 1 0.1364 376560.000
4 2 1 0.0945 462750.400
5 3 1 0.0945 462750.400
6 1 1 0.1229 476976.000
[ angles ]
; ai aj ak funct c0 c1 c2 c3
2 1 3 1 118.180 584.923
1 2 4 1 113.000 292.880
1 3 5 1 113.000 292.880
2 1 6 1 121.000 669.440
3 1 6 1 121.000 669.440
[ dihedrals ]
; IMPROPER DIHEDRAL ANGLES
; ai aj ak al funct c0 c1 c2 c3 c4 c5
6 1 2 3 4 180.000 43.932 2
[ dihedrals ]
; PROPER DIHEDRAL ANGLES
; ai aj ak al funct c0 c1 c2 c3 c4 c5
4 2 1 3 3 23.012 0.000 -23.012 -0.000 -0.000 0.000
5 3 1 2 3 23.012 0.000 -23.012 -0.000 -0.000 0.000
6 1 2 4 3 23.012 0.000 -23.012 -0.000 -0.000 0.000
6 1 3 5 3 23.012 0.000 -23.012 -0.000 -0.000 0.000
[ pairs ]
3 4 1
2 5 1
4 6 1
5 6 1
