TIPS3p-PPPM Water Model in Electrostatics

• TIPS3p-PPPM is a rigid three-site water model that utilizes PPPM for long-range electrostatics, retaining CHARMM-compatible parameters despite overestimated dielectric constants.
• The model’s simulation protocol, featuring 3000 water molecules and long-range electrostatics treatment, achieves internal consistency with continuum theories while highlighting quantitative deviations from experiments.
• Comparative studies show that TIPS3p-PPPM exhibits overscreening at charged interfaces, leading to lower near-surface counter-ion densities, yet its MD results can align with Poisson–Boltzmann theory when using model-derived inputs.

Searching arXiv for the specified paper and closely related PPPM/LAMMPS references. TIPS3p-PPPM is a rigid three-site water model designation used for a specific simulation protocol in which TIPS3p-type water is evaluated with long-range particle-particle particle-mesh electrostatics rather than truncated Coulomb interactions. In comparative molecular dynamics simulations of a charged silica/aqueous NaCl interface, it was retained not because it was the most accurate bulk model, but because it is widely used in practice, especially with the CHARMM force-field family for biological systems. Within that study, TIPS3p-PPPM occupies a distinctive position: it is practically important and internally self-consistent, yet it systematically overestimates the dielectric constant and therefore yields quantitatively different interfacial ion distributions from better bulk-calibrated models (Tavakol et al., 4 Aug 2025).

1. Definition and nomenclature

In the comparative study of charged solid–liquid interfaces, TIPS3p-PPPM denotes the TIPS3p water model with long-range electrostatics treated using PPPM. The designation is operational rather than parametric in a broad force-field sense: the model retains TIPS3p-type water geometry and Lennard-Jones/charge parameters, while changing Coulomb long-range handling relative to plain TIPS3P as represented in the paper’s Table 1. The model is described as a 3-site, rigid water model with $\sigma_{OO}=3.1507$, $\varepsilon_{OO}=0.152$, $q_H=0.417$, O–H bond length $=0.958$, H–O–H angle $=104^\circ$, “Flexible? No,” “LJ/Coul: Yes,” and “PPPM: Yes” (Tavakol et al., 4 Aug 2025).

The study’s selection logic is explicit. Eleven water models were screened for bulk dielectric behavior, and four were carried forward for charged-interface simulations: SPC/Fw, H2O/DC, TIP3P-ST, and TIPS3p-PPPM. TIPS3p-PPPM was retained “due to its widespread [use] in the field,” while TIPS3p is described as “mainly used with the CHARMM forcefield for biological systems.” This makes TIPS3p-PPPM a model of practical relevance rather than a top-ranked bulk dielectric model (Tavakol et al., 4 Aug 2025).

A useful contextual distinction is that PPPM in molecular simulation literature is not uniquely tied to one interaction class. In LAMMPS, PPPM is a standard long-range electrostatics solver for charged systems (McDoniel et al., 2017), while a separate PPPM Ewald implementation also exists for long-range dispersion $r^{-6}$ interactions (Isele-Holder et al., 2012). In the TIPS3p-PPPM nomenclature of the charged-silica study, the designation refers to long-range electrostatic treatment via PPPM rather than dispersion PPPM (Tavakol et al., 4 Aug 2025).

Parameter	TIPS3p-PPPM
Site model	3-site
Flexibility	Rigid
$\sigma_{OO}$	3.1507
$\varepsilon_{OO}$	0.152
$q_H$	0.417
O–H bond length	0.958
H–O–H angle	$104^\circ$
LJ/Coul	Yes
PPPM	Yes

2. Bulk dielectric behavior

The central bulk diagnostic in the study is the dielectric constant. For pure water, TIPS3p-PPPM yields $\varepsilon_{OO}=0.152$0, whereas experiment is about $\varepsilon_{OO}=0.152$1 at $\varepsilon_{OO}=0.152$2. Plain TIPS3p without PPPM gives $\varepsilon_{OO}=0.152$3. The switch to PPPM therefore slightly reduces the predicted dielectric constant, but does not remove the model’s systematic overestimation. Relative to the experimental value, the discrepancy is roughly $\varepsilon_{OO}=0.152$4 dielectric units, or about $\varepsilon_{OO}=0.152$5 (Tavakol et al., 4 Aug 2025).

Among the four interface-tested models, TIPS3p-PPPM is the least accurate for pure-water dielectric response. The reported values are SPC/Fw: $\varepsilon_{OO}=0.152$6, H2O/DC: $\varepsilon_{OO}=0.152$7, TIP3P-ST: $\varepsilon_{OO}=0.152$8, and TIPS3p-PPPM: $\varepsilon_{OO}=0.152$9. The models identified as closest to experiment in the broader eleven-model screening were SPC/Fw, H2O/DC, TIP3P-ST, OPC3, and FBA/e; TIPS3p-PPPM is not in that subset (Tavakol et al., 4 Aug 2025).

The dielectric constant was computed from dipole fluctuations using

$q_H=0.417$0

with

$q_H=0.417$1

The surrounding text identifies $q_H=0.417$2, $q_H=0.417$3, $q_H=0.417$4, $q_H=0.417$5, $q_H=0.417$6, and $q_H=0.417$7 as the total dipole moment, vacuum permittivity, volume, Boltzmann constant, temperature, and ensemble average. The study emphasizes that dielectric-constant estimation via this fluctuation formula converges slowly, which is one reason it cautions against over-interpreting small inter-model differences (Tavakol et al., 4 Aug 2025).

For saline water, the selected models all reproduce the qualitative decrease of dielectric constant with increasing NaCl concentration, but all still overestimate it to some extent. TIPS3p-PPPM is reported to overestimate the dielectric constant the most at every concentration tested. At $q_H=0.417$8 NaCl, it predicts a dielectric constant almost twice the experimental value, with experiment reported as about $q_H=0.417$9. The paper does not tabulate exact dielectric values beyond pure water; the following values are explicitly identified as figure-based estimates from Fig. 2b: about $=0.958$0 at $=0.958$1, $=0.958$2 at $=0.958$3, $=0.958$4 at $=0.958$5, $=0.958$6 at $=0.958$7, $=0.958$8 at $=0.958$9, and $=104^\circ$0 at $=104^\circ$1. These estimates are interpretively useful but not exact tabulated data (Tavakol et al., 4 Aug 2025).

The bulk dielectric calculations used boxes containing 3000 water molecules in $=104^\circ$2, and no explicit finite-size correction is reported. This matters because the study’s conclusions about TIPS3p-PPPM are framed around a clearly quantified bulk bias rather than an inferred deficiency (Tavakol et al., 4 Aug 2025).

3. Interfacial ion distributions at charged silica

At the negatively charged silica/NaCl interface, TIPS3p-PPPM reproduces the expected qualitative electric double-layer structure: co-ions are depleted near the surface, counter-ions accumulate near the surface, and lower salt concentration yields a longer-ranged diffuse layer. The model therefore captures the basic interfacial physics at the level of sign and layering trends (Tavakol et al., 4 Aug 2025).

Its quantitative behavior differs from SPC/Fw, H2O/DC, and TIP3P-ST. TIPS3p-PPPM consistently underestimates counter-ion density close to the Stern layer, and at low salt it does not show the near-surface counter-ion overestimation relative to analytical theory that appears for the other models. The study attributes this to overscreening caused by the model’s excessively large dielectric constant. It states that replacing the experimental dielectric constant in the analytical model with the TIPS3P-PPPM-derived dielectric constant leads to good agreement between MD and analytical results, implying that the higher dielectric constant causes overscreening of electrostatic interactions and hence lower counter-ion density near the surface (Tavakol et al., 4 Aug 2025).

The free-energy minimum position is reported as nearly model-independent. For all models and concentrations it lies at about $=104^\circ$3 from the surface in the paper’s coordinate convention, and this position defines the Stern layer boundary. A plausible implication is that the Stern-layer location is controlled more strongly by the ion–surface environment than by differences among these water models. The paper itself does not provide a separate analysis of water density layering, orientation, or hydrogen-bond structuring for TIPS3p-PPPM (Tavakol et al., 4 Aug 2025).

TIPS3p-PPPM also underestimates adsorbed ionic charge in the Stern layer relative to the other three interface-tested models. At $=104^\circ$4, all models except TIPS3P-PPPM converge to approximately the same Stern-layer total charge, around $=104^\circ$5, whereas the silica surface charge per side is $=104^\circ$6. TIPS3p-PPPM remains below that common value. The paper does not tabulate exact TIPS3p-PPPM Stern-layer charges at intermediate concentrations; any more specific values are therefore figure-based estimates only (Tavakol et al., 4 Aug 2025).

At higher salt, random ion pairing becomes important for all models, including TIPS3p-PPPM. The paper gives explicit pair counts for SPC/Fw and TIPS3p-PPPM: at $=104^\circ$7, SPC/Fw has $=104^\circ$8 ion pairs and TIPS3p-PPPM has $=104^\circ$9; at $r^{-6}$0, SPC/Fw has $r^{-6}$1 pairs and TIPS3p-PPPM has $r^{-6}$2. These random pairs reduce reproducibility and contribute to deviations from Poisson–Boltzmann theory at elevated concentration (Tavakol et al., 4 Aug 2025).

4. Continuum comparison and “intrinsic consistency”

A major conclusion of the study is that TIPS3p-PPPM can be reconciled with continuum electrostatics if the continuum model uses the dielectric response and Stern-layer charge actually realized in the MD simulation. The analytical comparison employs Gouy–Chapman/Poisson–Boltzmann relations, including the Debye length

$r^{-6}$3

surface potential

$r^{-6}$4

potential profile

$r^{-6}$5

and ion density profile

$r^{-6}$6

with the minus/plus sign taken for cations/anions as described in the text (Tavakol et al., 4 Aug 2025).

For TIPS3p-PPPM, the key contrast is between using experimental $r^{-6}$7 and using the model’s own MD-derived $r^{-6}$8. If the analytical model uses the experimental dielectric constant, the TIPS3p-PPPM MD predictions deviate more than those of the other three selected water models. If the analytical model instead uses the MD-derived dielectric constant for TIPS3p-PPPM, agreement becomes good. This is the basis for the model’s “intrinsic consistency” in the paper’s interpretation (Tavakol et al., 4 Aug 2025).

The free-energy analysis is expressed through

$r^{-6}$9

$\sigma_{OO}$0

$\sigma_{OO}$1

For TIPS3p-PPPM, the counter-ion free-energy minimum becomes deeper as salt concentration increases, as it does for the other models. Compared with Gouy–Chapman theory using experimental $\sigma_{OO}$2, however, the free-energy minimum deviates by about $\sigma_{OO}$3. Using the MD-derived dielectric constant removes the noticeable discrepancy (Tavakol et al., 4 Aug 2025).

This result is important because it qualifies a common misconception. The paper does not conclude that a water model with a substantially biased bulk dielectric constant is unusable at charged interfaces. Rather, it concludes that continuum PB/Gouy–Chapman theory can still describe the MD free-energy minimum if one supplies two MD-derived inputs: the total ionic charge in the Stern layer and the dielectric constant realized by that MD model. The abstract generalizes this by stating that the consistency stands even for water models with a dielectric constant off by $\sigma_{OO}$4, although the pure-water discrepancy for TIPS3p-PPPM itself is larger than that benchmark (Tavakol et al., 4 Aug 2025).

The same section also clarifies the high-salt limit of this agreement. For salt concentrations higher than $\sigma_{OO}$5 NaCl, random ion–ion pairs limit both reproducibility of the MD results and applicability of the analytical method. Deviations at high concentration are therefore not attributable solely to the water model; they also reflect physics absent from standard PB theory (Tavakol et al., 4 Aug 2025).

5. Simulation protocol and methodological context

The simulations were carried out in LAMMPS, version 23 Jun 2022, with OVITO, Matplotlib, and dedicated C++ and Python analysis codes. Thermodynamic conditions were 300 K and 1 atm, controlled with a Nose–Hoover thermostat and barostat. Long-range electrostatics were treated with PPPM, while nonbonded and short-range electrostatics were truncated using the CHARMM potential cutoff function, with force and energy smoothed to zero between inner and outer radii. The exact PPPM grid, accuracy, and cutoff radii are not specified in the excerpted methods text (Tavakol et al., 4 Aug 2025).

The charged interface was a silica Q3 surface from the interface forcefield database, assumed pH 7, with a surface density of 0.67 SiO$\sigma_{OO}$6 Na$\sigma_{OO}$7 groups per $\sigma_{OO}$8. The silica slab thickness was 2.3 nm, the lateral cross-section was $\sigma_{OO}$9, and the box length normal to the slab was 16.5 nm. Periodic boundaries yield effectively infinite silica laterally and two interfaces normal to the slab. The paper refers to a silica surface charge of about 0.08 C/m$\varepsilon_{OO}$0 and $\varepsilon_{OO}$1 per side in the simulation cell (Tavakol et al., 4 Aug 2025).

The ion parameterization is especially relevant for reproducing TIPS3p-PPPM results. For SPC/Fw, H2O/DC, and TIP3-ST, the authors used Na$\varepsilon_{OO}$2/Cl$\varepsilon_{OO}$3 parameters calibrated for SPC/e because model-specific ion parameters were unavailable. For TIPS3p-PPPM, Na$\varepsilon_{OO}$4 and Cl$\varepsilon_{OO}$5 parameters were taken from the CHARMM forcefield. This makes TIPS3p-PPPM not only a water-model choice but also a specific water–ion parameter ecosystem within the study (Tavakol et al., 4 Aug 2025).

Sampling followed a common sequence: random placement of water and ions, energy minimization, several short equilibration runs, and production simulation. Reported production lengths were 50 ns for bulk water and 30 ns for dielectric calculations and SLI simulations. For dielectric calculations, salt concentrations tested for the four selected models were 0, 0.5, 1.0, 1.5, 2.0, and 2.5 M NaCl (Tavakol et al., 4 Aug 2025).

6. Interpretation, scope, and limitations

TIPS3p-PPPM is not presented as the best-performing water model in the study. On bulk dielectric response it is the clear outlier among the four interface-tested models, and at the charged silica surface it produces lower near-surface counter-ion density and lower Stern-layer adsorption than SPC/Fw, H2O/DC, and TIP3P-ST. If the goal is quantitatively accurate dielectric behavior of saline water, the paper does not identify TIPS3p-PPPM as the strongest choice (Tavakol et al., 4 Aug 2025).

Its practical value lies elsewhere. The model is widely used, especially in CHARMM-style simulation ecosystems, and the paper shows that it remains internally coherent in the sense that its interfacial MD results can be interpreted consistently with continuum theory when the continuum inputs are taken from the same MD model. In that restricted but important sense, it functions as a reliable qualitative-to-semiquantitative model for charged interfaces, provided one does not substitute experimental dielectric data into the analysis and then expect exact agreement (Tavakol et al., 4 Aug 2025).

Two misconceptions are explicitly corrected by the study. First, switching plain TIPS3p to PPPM does not “fix” the model’s dielectric constant; it lowers the pure-water dielectric constant from $\varepsilon_{OO}$6 to $\varepsilon_{OO}$7, but the result is still substantially too high (Tavakol et al., 4 Aug 2025). Second, a water model with an imperfect dielectric constant is not automatically analytically useless. TIPS3p-PPPM is the study’s clearest case showing that continuum free-energy agreement can remain good if the model-specific Stern-layer charge and dielectric constant are used (Tavakol et al., 4 Aug 2025).

The principal limitations are also clear. Above about $\varepsilon_{OO}$8 NaCl, random ion-pair formation reduces reproducibility and undermines direct Gouy–Chapman comparison. The study also does not provide a direct structural analysis of interfacial water layering, orientation, or hydrogen-bonding for TIPS3p-PPPM, so its assessment is strongest for dielectric response, ion distributions, Stern-layer charging, and free-energy consistency rather than for a full microscopic characterization of interfacial water structure (Tavakol et al., 4 Aug 2025).

Taken together, TIPS3p-PPPM is best understood as a practically important and methodologically revealing benchmark: a widely used rigid three-site water model whose bulk dielectric inaccuracy produces overscreening and weaker near-surface counter-ion adsorption, but whose interfacial behavior can nonetheless be rationalized within Poisson–Boltzmann/Gouy–Chapman theory when interpreted with the dielectric constant and Stern-layer charge that the MD model itself realizes (Tavakol et al., 4 Aug 2025).