Activities in electrolyte solutions by molecular ...horsch/pdt/slides/2016_AIChE_OPAS.pdf ·...
Transcript of Activities in electrolyte solutions by molecular ...horsch/pdt/slides/2016_AIChE_OPAS.pdf ·...
Activities in electrolytesolutions by molecular simulation
of the osmotic pressure
M. T. Horsch, M. Kohns, S. Reiser, M. Schappals, and H. Hasse
Laboratory of Engineering ThermodynamicsUniversity of Kaiserslautern, Germany
2016 AIChE Annual MeetingSan Francisco, November 17, 2016
Osmotic pressure for activity of solvents
2Nov 17, 2016
In a dense liquid, it is hard to com-pute the chemical potential (or acti-vity) by inserting test particles.
OPAS method: Compute the osmo-tic pressure from the force acting ona virtual semipermeable membrane.
External harmonic potentialconstraining the solute inside.
No external potential for the solvent.
Π = P in – P
out
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Osmotic pressure for activity of solvents
3Nov 17, 2016
Step 1: Pseudo-NPT run, where the barostat is adjusted to regulate P
out.Step 2: NVT simulation.
OPAS method: Compute the osmo-tic pressure from the force acting ona virtual semipermeable membrane.
Solvent activity (Raoult normaliza-tion) from osmotic pressure:
Density and compressibility of thepure solvent are easy to compute.
RT lnasolvin = −∫P
out
P in
dP v solvpure ≈ −Πv solv
pure
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
http://www.ms-2.de/
Validation of the OPAS method
4Nov 17, 2016
Test case: Mixture of argon (LJ) and oxygen (2CLJQ).
Ar is treated as solvent and moves freely through the simulation volume.The solute O
2 is confined by the membranes to the inner compartment.
T = 140 K
oxygen
argon
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Validation of the OPAS method
5Nov 17, 2016
T = 140 K
oxygen
argon
Test case: Mixture of argon (LJ) and oxygen (2CLJQ).
Osmotic pressure from force acting on the membranes: 6.33 ± 0.06 MPa.Osmotic pressure from difference between compartments: 6.1 ± 0.2 MPa.
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Aqueous alkali halide salt solutions: Model
6Nov 17, 2016
Ions
1 Lennard-Jones
1 point charge
Molecular models:
Water (here: SPC/E)
1 Lennard-Jones
3 partial charges
--++
Li
+ Na
+ K
+ Rb
+ Cs
+
F
– LiF NaF KF RbF CsF
Cl
– LiCl NaCl KCl RbCl CsCl
Br
– LiBr NaBr KBr RbBr CsBr
I
– LiI NaI KI RbI CsI
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Adjusting the model to experimental data
7Nov 17, 2016
Lennard-Jones size parameter σ Lennard-Jones energy parameter ε
m
Density ratio “solution : pure” Self-diffusion coefficients
P = 100 kPaT = 298 K
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Validation: Non-aqueous solutions
8Nov 17, 2016
Density of methanolic electrolyte solutions at T = 298 K and P = 1 bar:
• Experimental data (this work)
• Simulation
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Water activity in electrolyte solutions
9Nov 17, 2016
Validation of OPAS simulation resultsagainst literature data for aqueousNaCl solution using
● the SPC/E water model,● the Joung-Cheatham NaCl model.
OPAS simulation results (this work)
Correlation to present OPAS results
Moučka et al. (sim.), SPC/E + JC
Hamer and Wu (exp.)
T = 298.2 KPout = 1 bar
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Salt activity coefficient in aqueous solution
10Nov 17, 2016
Exp.
OPAS (this work)
Moučka et al.
Mester andPanagiotopoulos
Chemical potentials of the soluteand the solvent are related by theGibbs-Duhem equation.
Chemical potential of the saltexpressed as activity coefficient(Henry normalization).
T = 298.2 KPout = 1 bar
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Salt activity coefficient in aqueous solution
11Nov 17, 2016
Series of own models for
● alkali cations● and halide anions,
collectively adjusted to
● reduced solution densities● and self-diffusion coefficients.
Present simulation results Mester and Panagiotopoulos
Validation against:
Correlation to experimental data by Hamer and Wu
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Salt activity coefficient in aqueous solution
12Nov 17, 2016 Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Model reparameterization
13Nov 17, 2016
Revised σ(K
+) = 2.06 Å (before: 2.77 Å), σ(F
–) = 3.94 Å (before: 3.66 Å).
For a specific combination of ions, the accuracy for the chemical potentialcan be improved without a loss in accuracy for the density.
KF(aq)T = 298 KP = 1 bar
T = 293 KP = 1 barrevised model
previous model
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse
Conclusion
14Nov 17, 2016
In previous work of our group, LJ + point charge models of the alkali and halide ions were parameterized to the density of aqueous salt solutions.
The models were validated against own experimental (density) data for non-aqueous solutions and for aqueous solutions at different conditions.
By OPAS simulation with virtual semipermeable membranes, the influ-ence of salts on the activity of the solvent can be determined with a reduced effort, compared with methods based on particle insertion.
Activity coefficients of the solute alkali halide salt can be obtained by Gibbs-Duhem integration.
In some cases, the models agree well with experimental activity data. In other cases, the agreement can be improved substantally.
Martin Horsch, Maximilian Kohns, Steffen Reiser, Michael Schappals, and Hans Hasse