Efficient determination of the structure of (molecular) liquids based on
diffraction experiments
(using Reverse Monte Carlo modeling)
László Pusztai
Institute for Solid State Physics and Optics, Wigner Research Centre for Physics,
Hungarian Academy of Sciences
pusztai.laszlo@wigner.mta.hu
The
BEGINNINGS...
Elemental liquids (1)
51 liquids altogether...
Some are ‘simple’, some
are not so simple...
Elemental liquids (2): simple
32 simple elemental liquids (liquid ‘good’ metals, noble
liquids):
By and large, the only
difference is particle size.
Molten salts (1)
A new feature here: 2 components, 3 partials.
Heavy use of isotopic substitution neutron diffraction data (Cl and
Ni isotopes – good, old
days...).
Molten salts (2): intermediate range
A densely packed, uniform (BaCl
2, RIGHT) vs. a more open, more uneven (ZnCl
2, LEFT) structure. The
tendency towards covalency is obvious in molten ZnCl
2.
Liquid sulphur
???
!!!
CONSTRAINTS
Since, in general, there are many-many-many possible structures that are consistent with
results of diffraction (and EXAFS) experiments...
...we MUST MEANINGFULLY
NARROW the (configuration)
space of solutions !!
CONSTRAINTS
• Density
• Particle sizes (‘packing fraction’)
Geometrical constraints (possibly based on other experimental evidence):
number of neighbours; bond angles.
Diffraction data
CONSTRAINTS
(available in the present software, RMC++/RMC_POT)
• ND, XRD, EXAFS data
• Coordination number (specific, average,
‘cumulative’)
• Angular distribution (‘bond angles’)
• ‘fixed neighbours’ (mostly for molecular systems,
but for covalent glasses, it may also be OK)
Liquid sulphur (revisited)
YES!
One F(Q),
THREE different angular distributions!
!!!
Liquid sulphur (2)
Diffraction data CANNOT tell if l-S is (a) atomic; (b) chain-
like; or (c) ring-like.
The
PAST 15 YEARS...
...the rise of
MOLECULAR
LIQUIDS
WHY (1)
are molecular liquids important?
WATER &Co. !!!!!!
+ „multisale customers”:
- near-critical systems
- polymers (and soft matter in general) - colloidal aggregates
- Etc...
WHY(2)
molecular liquids?
Don’t we know everything ???
Well...
Just think of liquid CCl 4 ...
THE MAIN ISSUE HERE
(molecules! possibly, multiple length scales!):
WHAT ARE THE PARTICLES TO CONSIDER
(and MOVE) ??
HOW TO HANDLE MOLECULES?
Rigid: not good...
Must be flexible (a) FNC
(b) real intramolecular potentials
( RMC_POT)
„Fixed Neighbours Constraints”, FNC FNC-s are essentially neighbour
lists: lists of atoms that have to be connected throughout the entire
calculations (within prescribed distance ranges).
(RMCProfile: a similar approach introduced.)
Simple concept – GREAT use!!
Evrard et al.,J. Phys.: Condens. Matter17, S1 (2005)
FNC-s at work: XCl 4 liquids, a classic example (1)
HARD SPHERE-LIKE REFERENCE SYSTEMS !!
A classic example (2): XCl 4 liquids
HSMC REFERENCE SYSTEMS; simple orientational
correlation functions (by R. Rey).
These are (the last???) systems that may be understood very well
(‘fully???’) by
experiments and RMC
modeling.
The
PRESENT...
...the rise of COMPLEX molecular liquids:
approach towards MD.
A strongly coupled application of RMC and molecular dynamics (MD)
simulations for understanding the
structure of complex systems
WHY MD/MC SIMULATION?
To access information that is not available otherwise (like via experiments).
A change of paradigm (for me…): since experimental evidence is (very...) limited for the most important
bit, the H-bond
It is MD that will provide ‘the final solution’ for
complex disordered systems ( BIOLOGY)
How do we know that a(n MD/MC/AIMD) simulation is ‘good’?
COMPARE simulation results with EXPERIMENTal data.
WHAT results and HOW ... ????
VALIDATION OF SIMULATION RESULTS
pRDF’s from MD
Total S(Q)
from diffraction
A structural modell via RMC
that is(??) consistent with BOTH.
What if you do RMC only?
(CsCl solutions in water)
2 datasets, 10 prdf’s
=>
Too many possibilities
We MUST include extra information!!!
(Whose reliability can be checked simultaneously!)
Mile et al.,J. Phys. Chem B, 113, 10760-10769 (2009)
MD + RMC
(15 molar% CsCl solution in water)
Mile et al.,J. Phys. Chem B, 113, 10760-10769 (2009)
What have we learnt from the MD+RMC combination
concerning water&Co.??
• We have a clue which water potentials are
‘OK’ for pure ambient liquid water STRUCTURE.
• We now know that a major problem while
simulating aqueous solutions is the water
potential – (probably) need a new one!
BEYOND WATER
Pure alcohols: do we know everything?
-1 0 1 2 3 4 5 6
0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9
Series2 Series3 Series4 Series5 Series6 Series7 Series8 Series9 Series10 Series11
From methanol to decanol: increasing supramolecular organisation?
X-ray diffraction data; WHAT ABOUT NEUTRONS????
Pure alcohols: micro- and mesoscopic structure of MeOH, EtOH, PropOH
MD??? How come...??
Involving interatomic potentials
Even closer interplay with molecular dynamics simulations
RMC_POT
Gereben O, Pusztai L; Journal of Computational Chemistry; Volume 33, Issue 29, 5 November 2012, Pages: 2285–2291 (2012)
( CREDIT TO EPSR here –
AKS&Co. started to play with potentials AND
structural modelling.)
The
FUTURE...
...understanding H- bonded liquids
( BIOLOGY).
Involving interatomic potentials : :: : RMC_POT
Gereben O, Pusztai L; Journal of Computational Chemistry; Volume 33, Issue 29, 5 November 2012, Pages: 2285–2291 (2012)
GROMACS philosophy GROMACS-like file structure
Interatomic potentials of any complexity handling molecules of any complexity !
Complex molecules complicated application
(probably inevitable).
Involving interatomic potentials : :: : RMC_POT
Description of molecules of any complexity; for instance, dihedrals may be:
a)Proper dihedral angle, b) improper dihedral for
rings, c) planar group and d) chiral centre.
Involving interatomic potentials :: : : RMC_POT for dimethyl-trisulfide
-2 -1.5 -1 -0.5 0 0.5 1
0 5 10 15
Q (Å-1)
S(Q) .
experimental Fm_fq_NB
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
1 2 3 r (Å) 4 5 6
S -H g (r ) .
MD FNC_fq Fm_fq
Fm_fq_gr_NB
CH3 – S – S – S – CH3
Gereben O, Pusztai L; Journal of Computational Chemistry; Volume 33, Issue 29, 5 November 2012, Pages: 2285–2291 (2012)
Alcohol-water mixtures: a widely misunderstood class of liquids?
Letters to Nature
Nature416, 829-832 (25 April 2002) | doi:10.1038/416829a; Received 26 October 2001; Accepted 4 March 2002
Molecular segregation observed in a concentrated alcohol–water solution S. Dixit1, J. Crain1, W. C. K. Poon1, J. L. Finney2 & A. K. Soper3
When a simple alcohol such as methanol or ethanol is mixed with water1, 2, the entropy of the system increases far less than expected for an ideal solution of
randomly mixed molecules3. This well-known effect has been attributed to hydrophobic headgroups creating ice-like or clathrate-like structures in the surrounding water4, although experimental support for this hypothesis is scarce5, 6,
7. In fact, an increasing amount of experimental and theoretical work suggests that the hydrophobic headgroups of alcohol molecules in aqueous solution cluster together2, 8, 9, 10. However, a consistent description of the details of this self- association is lacking11, 12, 13. Here we use neutron diffraction with isotope substitution to probe the molecular-scale structure of a concentrated alcohol–water
mixture (7:3 molar ratio). Our data indicate that most of the water molecules exist as small hydrogen-bonded strings and clusters in a 'fluid' of close-packed methyl
groups, with water clusters bridging neighbouring methanol hydroxyl groups through hydrogen bonding. This behaviour suggests that the anomalous thermodynamics of water–alcohol systems arises from incomplete mixing at the molecular level and from retention of remnants of the three-dimensional hydrogen-
bonded network structure of bulk water.