Gaussian 03 Online Manual
Last update: 1 August 2007


This keyword requests that a calculation be performed in the presence of a solvent, using one of the following models:

  • The Onsager model [281,282,283,284,565,566], which places the solute in a spherical cavity within the solvent reaction field.

  • Polarizable Continuum (PCM) models in which the cavity is created via a series of overlapping spheres, initially devised by Tomasi and coworkers [285,286,287,288,289,290,291,292,293,295,576]. The current implementation is the work of Barone and coworkers [285,286,287,297,299,300,301,302,303] and Tomasi, Mennucci and coworkers [293,294,296,298].

  • A static isodensity surface polarized continuum model (IPCM) [307].

  • A Self-Consistent Isodensity PCM (SCI-PCM) model [307].


Keywords and options specifying details for a PCM calculation (SCRF=PCM, CPCM or IEFPCM) may be specified in an additional blank-line terminated input section provided that the Read option is also specified. Keywords within this section follow general Gaussian input rules. The available keywords are listed in a separate subsection following the examples.


For the Onsager model (SCRF=Dipole), the solute radius in Angstroms and the dielectric constant of the solvent are read as two free-format real numbers on one line from the input stream. A suitable solute radius is computed by a gas-phase molecular volume calculation (in a separate job step); see the discussion of the Volume keyword.


For the IPCM and SCI-PCM models, the input consists of a line specifying the dielectric constant of the solvent and an optional isodensity value (the default for the latter is 0.0004).


Selects the solvent in which the calculation is to be performed. Note that the solvent may also be specified in the input stream in various ways for the different SCRF methods. If unspecified, the solvent defaults to water. Item is a solvent name chosen from the following list:

  • Water or H2O: ε=78.39

  • Acetonitrile or CH3CN: ε=36.64

  • DiMethylSulfoxide or DMSO: ε=46.7

  • Methanol or CH3OH: ε=32.63

  • Ethanol or CH3CH2OH: ε=24.55

  • Isoquinoline: ε=10.43

  • Quinoline: ε=9.03

  • Chloroform or CHCl3: ε=4.9

  • Ether or DiEthylEther or CH3CH2OCH2CH3: ε=4.335

  • DiChloroMethane or MethyleneChloride or CH2Cl2: ε=8.93

  • DiChloroEthane or CH2ClCH2Cl: ε=10.36

  • CarbonTetrachloride or CCl4: ε=2.228

  • Benzene or C6H6: ε=2.247

  • Toluene or C6H5CH3: ε=2.379

  • ChloroBenzene or C6H5Cl: ε=5.621

  • NitroMethane or CH3NO2: ε=38.2

  • Heptane or C7H16: ε=1.92

  • CycloHexane or C6H12: ε=2.023

  • Aniline or C5H5NH2: ε=6.89

  • Acetone or CH3COCH3: ε=20.7

  • TetraHydroFuran or THF: ε=7.58

  • DiMethylSulfoxide or DMSO or CH3SOCH3: ε=46.7

  • Argon or Ar: ε=1.43

  • Krypton or Kr: ε=1.519

  • Xenon or Xe: ε=1.706

We list the ε values here for convenience, but be aware it is only one of many internal parameters used to define solvents. Thus, simply changing the ε value will not define a new solvent properly.


For quantum mechanical calculations, performs a reaction field calculation using the IEF-PCM model [288,290,293] (see below). This is the default. Note that this option has changed in meaning with respect to Gaussian 98. Also, some details of the formalism and the implementation have changed, as described in [302].

Perform a PCM calculation using the integral equation formalism model [288,293,294,295]. The model of Chipman [568] is closely related to this earlier one [569].

Note that if IEF-PCM is used for an anisotropic or ionic solvent, then items in the PCM input section must be used to select the anisotropic and ionic dielectric models for these types of solvents, using the Read option (see below).

Perform a PCM calculation using the CPCM polarizable conductor calculation model [292,303].

Perform an Onsager model reaction field calculation.

Perform an IPCM model reaction field calculation. Isodensity is a synonym for IPCM.

Perform an SCI-PCM model reaction field calculation: perform an SCRF calculation using a cavity determined self-consistently from an isodensity surface.


Sets the value for the solute radius in the route section (rather than reading it from the input stream). If this option is included, then Solvent or Dielectric must also be included.

Sets the value for the dielectric constant of the solvent. This option overrides Solvent if both are specified.


Indicates that a separate section of keywords and options providing calculation parameters should be read from the input stream (as described above).

Pick up SCRF information from the checkpoint file, but also read modifications from the input stream.


Use Vne basins for the numerical integration.

Use density basins for the numerical integration. The job may fail if non-nuclear attractors are present.


Force the use of the density matrix in evaluating the density.

Force the use of MOs in evaluating the density.

Use the gas phase isodensity surface to define the cavity rather than solving for the surface self-consistently. This is mainly a debugging option.

The PCM models are available for semi-empirical, HF, DFT, MP2, MP3, MP4(SDQ), QCISD, CCD, CCSD, CASSCF, CIS, TD, CID and CISD energies and HF, DFT, MP2, CIS and CASSCF gradients.

IEFPCM and PCM may be used to compute frequencies for the methods listed for gradients.

Int=AM1 must be used in the route section if SCRF AM1 is specified.

The solvent reaction field for PCM MP2 calculations is equilibrated to the solute electronic density obtained at the SCF level. Note that ΔGsolvation=EPCM-MP2EMP2 cannot be obtained using the PCM SCFVac option, but must be obtained by comparing the results of two separate calculations, performed in gas-phase and in solvent.

CIS PCM [298] and TD PCM [300] calculations are by default non-equilibrium calculations with respect to the polarization process between the solvent reaction field and the charge density of the electronic state indicated in the input (where the ground state is the default). However, equilibrium CIS PCM calculations are the default for geometry optimizations.

By default, CASSCF PCM [297] calculations corresponds to an equilibrium calculation with respect to the solvent reaction field- solute electronic density polarization process. Calculation of non equilibrium solute-solvent interaction involving two different electronic states (e.g. the initial and final states of a vertical transition) can be performed using the NonEq=type PCM keyword, in two separate job steps (see the PCM input section below).

The IPCM model is available for HF, DFT, MP2, MP3, MP4(SDQ), QCISD, CCD, CCSD, CID, and CISD energies only.

The SCI-PCM model is available for HF and DFT energies and optimizations and numerical frequencies.

The Onsager model is available for HF, DFT, MP2, MP3, MP4(SDQ), QCISD, CCD, CCSD, CID, and CISD energies, and for HF and DFT optimizations and frequency calculations.

The Opt Freq keyword combination may not be used in SCRF=Onsager calculations.

SCRF=PCM and SCRF=IPCM jobs can be restarted from the read-write file by using the Restart keyword in the job's route section. SCRF=SCIPCM calculations which fail during the SCF iterations should be restarted via the SCF=Restart keyword.

Volume, SCF

PCM Energy. Energy output from the SCRF models other than Onsager appears in the normal way within the output file, followed by additional information about the calculation. For example, here is the section of the output file containing the predicted energy from a PCM calculation:

SCF Done:  E(RHF) =  -98.569083211     A.U. after    5 cycles 
          Convg  =    0.4249D-05             -V/T =  2.0033 
          S**2   =   0.0000 
Variational PCM results
<psi(f)|   H   |psi(f)>                 (a.u.) =     -98.568013 
<psi(f)|H+V(f)/2|psi(f)>                (a.u.) =     -98.569083
Total free energy in solution: 
 with all non electrostatic terms       (a.u.) =     -98.573228
(Polarized solute)-Solvent               (kcal/mol) =      -3.27 
Cavitation energy                        (kcal/mol) =       5.34 
Dispersion energy                        (kcal/mol) =      -3.08 
Repulsion energy                         (kcal/mol) =       0.34 
Total non electrostatic                  (kcal/mol) =       2.60  

Additional output lines may appear when various PCM options are included.

The total energy in solution is the sum of the SCF energy and all of the non-electrostatic energy terms (both are highlighted in the output). Note that the PCM results also include the dipole moment in the gas phase and in solution (not shown here), and the various components of the predicted SCRF energy.

For all iterative SCRF methods, note that the energy to use is the one preceding the Convergence achieved message (i.e., the one from the final iteration of the SCRF method).

Onsager Energy. The energy computed by an Onsager SCRF calculation appears in the output file as follows:

    Total energy (include solvent energy) =   -74.95061789532

Additional Keywords for PCM Calculations

Additional input keywords may be specified for PCM SCRF calculations. They are placed in a separate input section, as in this example:

# HF/6-31++G(d,p) SCF=Tight SCRF=(PCM,Read,Solvent=Cyclohexane) Test       

PCM SP calculation on hydrogen fluoride  
F 1 R       


This Gaussian job performs a PCM energy calculation on the molecule HF using the solvent cyclohexane. The calculation is performed at a temperature of 300 K using a scaling factor for all atoms except acidic hydrogens of 1.21 and a value of 70 tesserae per sphere. The final input section ends as usual with a blank line.

The following keywords are available for controlling PCM calculations (arranged in groups of related items):


The solvent for the PCM calculation may be specified using the normal Solvent option to the SCRF keyword. The solvent name keyword or ID number may also be placed within the PCM input section. Alternatively, the EPS and RSOLV keywords may be used in the PCM input section to define a solvent explicitly:

Dielectric constant of the solvent.

Solvent radius in Angstroms.

Density of the solvent expressed as number of particles per unit volume, in Å-3.

Optional value for the dielectric constant at infinite frequency.

Note that if any of these parameters are specified, the others default to the values for water, and so you will probably want to set all of them appropriately.


Skip the calculation of dispersion solute-solvent interaction energy.

Skip the calculation of repulsion solute-solvent interaction energy.

Skip the calculation of the cavitation energy.

By default, non-electrostatic energy contributions are computed and printed, but they are not added into the energy and its derivatives during geometry optimizations. The keywords DDis, DRep, and DCav may be used to include them for the rare cases where the non-electrostatic energy terms are known to affect the geometry. Such cases will require care during optimization, and the optimization process may be trickier and more lengthy.

Performs the gas phase calculation before that in solution. It allows for the calculation of ΔGsolvation, the variation of the dipole moment in solution and so on, but only by HF or DFT methods. NOSCFVAC is the default. The recommended radii for this calculation type are the United Atom Topological Model applied on radii optimized for the HF/6-31G(d) level of theory (specified with RADII=UAHF).

Performs analysis of the solute solvent interaction energy in terms of atomic or atomic groups additive contributions. This analysis involves a fitting of atomic charges to the molecular electrostatic potential in solution.

Compute the electrostatic energy gradients neglecting the geometrical contributions (i.e. at "fixed cavity"). MobGrd is the default.

Compute the electrostatic energy second derivatives neglecting the geometrical contributions (i.e. at "fixed cavity"). MobHss is the default.

Solve the PCM electrostatic problem to calculate polarization charges through a linear scaling iterative method using a Jacobi-like scheme.

Solve the PCM electrostatic problem to calculate polarization charges through an inversion matrix algorithm. This is the default.

Specify the maximum number of iterations allowed to the iterative solution of the electrostatic problem. 200 is the default.

Set the convergence threshold for the iterative calculations of the PCM polarization charges to 10-N or to one of the following predefined types: VeryTight (10-12), Tight (10-9) and Sleazy (10-6). Default convergence values are QConv=Tight for PCM energy calculation and QCONV =VeryTight for PCM energy gradients calculations.

Skip the DIIS algorithm for the iterative solution of the PCM problem when the Jacobi scheme is exploited.

Number of vectors used in the DIIS extrapolation

Turn off the use of the Fast Multipole Method in the iterative solution. FMM is the default.

Set the degree of the polynomial for the electrostatic potential multipole expansion in the FMM. 6 is the default.

Set the length in Angstroms of the FMM box. 6.0 is the default.

Set the preconditioner type for the PCM iterative solution. 0 means no preconditioner. 1 corresponds to a simple Jacobi preconditioner, while 2 is a preconditioner based on the correlation considered only for charges located on the same sphere. 2 is the default.

Set the iterative algorithm to a stabilized biconjugate gradient. The DIIS option is not allowed with this keyword, and the algorithm defaults to Jacobi when it is used.

Set the iterative algorithm to a squared conjugate gradient. This is the default for CPCM calculations.

Set the iterative algorithm to a conjugate gradient.

The ICOMP keyword, formerly used to specify the charge compensation mode, is no longer needed and is deprecated.


Performs a PCM calculation for anisotropic solvent according to the IEF-PCM formalism. The 3-rank symmetric tensor representing the dielectric constant must be specified as the values for these six additional keywords: EPSX, EPSY, EPSZ, EUPHI, EUTHE, and EUPSI (all of them take a parameter: e.g., EPSX=value).

Performs a PCM calculation for ionic solution according to the IEF-PCM formalism. The ionic strength in mol/dm3 Å2 has to be specified as the value to the keyword DISM.


By default, the program builds up the cavity using the United Atom (UA0) model, i.e. by putting a sphere around each solute heavy atom: hydrogen atoms are enclosed in the sphere of the atom to which they are bonded. There are three UA models available (see below).

The cavity can be extensively modified in the PCM input section: putting spheres around specified hydrogens, changing sphere parameters and the general cavity topology, adding extra spheres to the cavity built by default, and so on. The whole molecular cavity can be also provided by the user in the input section.

Indicates the topological model and/or the set of atomic radii used. Available models and sets are:

UA0: Use the United Atom Topological Model applied on atomic radii of the UFF force field.

UAHF: Use the United Atom Topological Model applied on radii optimized for the HF/6-31G(d) level of theory. These are the recommended radii for for the calculation of ΔGsolvation via the SCFVAC PCM keyword.

UAKS: Use the United Atom Topological Model applied on radii optimized for the PBE0/6-31G(d) level of theory.

UFF: Use radii from the UFF force field. Hydrogens have individual spheres (explicit hydrogens).

PAULING: Use the Pauling (actually Merz-Kollman) atomic radii (explicit hydrogens).

BONDI: Use the Bondi's atomic radii (explicit hydrogens).

When using the UA0 model, places an individual sphere on the hydrogen at the Nth position in the atoms list.


When using the UA0 model, put individual spheres on acidic hydrogens (those bonded to N, O, S, P, Cl and F atoms).

Specify the electrostatic scaling factor by which the sphere radius is multiplied. The default value is 1.2.

Specify the type of molecular surface representing the solute-solvent boundary. Available options are:

SES: Solvent Excluding Surface. The surface is generated by the atomic or group spheres and by the spheres created automatically to smooth the surface ("added spheres"). This is the default for electrostatic contribution.

VDW: Van der Waals surface. Uses unscaled atomic radii and skip the generation of "added spheres" to smooth the surface.

SAS: Solvent Accessible Surface. The radius of the solvent is added to the unscaled radii of atoms and/or atomic groups.

Avoid the generation of added spheres to smooth the cavity surface. ADDSPH is the default.

Alter parameters for one or more spheres. The modified spheres can be indicated in the PCM input using the following format:

atom_number radius [alpha] 

Add N user-defined spheres to the cavity. Parameters of the spheres can be indicated using the following format:


X Y Z radius [alpha]                                                   X,Y,Z are the Cartesian coords. in the standard orientation. 

The cavity is built just from the N spheres provided by the user, specified on lines of the following format:

atom_number radius [alpha]      
X Y Z radius [alpha]                                                   X,Y,Z are the Cartesian coords. in the standard orientation. 

Do not impose the molecular symmetry to the cavity. SymmCav is the default.

Specify the overlap index between two interlocking spheres [569]. Decreasing this index results in a smaller number of added spheres. The default value is 0.89.

Set the minimum radius in Angstroms for SES added spheres. Increasing this value results in a smaller number of added spheres. The default value is 0.2.

Set the average area of the tesserae generated on each sphere in the cavity surface, in units of Å2 (area=0.2 is the default value). Reducing this value results in a finer surface discretization. Values suggested as the best compromise between accuracy and numerical stability range between 0.2 and 0.4, or even larger for molecular mechanics calculations.


Threshold to discard small tesserae (the default is 10-4 Å2).


Threshold to discard short edges in a tessera (the default is 5.0*10-7 Angstroms).


Create the file describing the cavity. This file contains input for the GeomView program (see which can be used to visualize the molecular cavity.

Include the descriptions and values of all the internal PCM parameters in the Gaussian log file.