introduction to molecular dynamics simulations...
TRANSCRIPT
![Page 1: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/1.jpg)
SMR/1831-1
Spring College on Water in Physics, Chemistry and Biology
10 - 21 April 2007
Introduction to molecular dynamics simulations methods
Mauro BOEROCenter for Computational Sciences,
University of Tsukuba,Ibaraki 305-8577, Japan
![Page 2: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/2.jpg)
1
Lecture 1: Introduction to molecular dynamics - from classical force fields to first principles
Mauro Boero
Center for Computational Sciences, University of Tsukuba,
1-1-1 Tennodai, Tsukuba, Ibaraki 305-8577 Japan
![Page 3: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/3.jpg)
2
Outline of the lectures• Molecular Dynamics (MD):
Newtonian and Lagrangean dynamics
Basic algorithms
Time averages and ensemble averages
• Density Functional Theory (DFT) based MD:Brief review of DFT and Born-Oppenheimer approximation
Car-Parrinello Molecular Dynamics (CPMD)
Practical implementation
• Reactive CPMD:General problem in chemical reactions
Blue Moon ensemble approach
Metadynamics
• Water as a catalyst:Catalytic properties of supercritical water
Synthetic reactions in supercritical H2O – Beckmann rearrangement
![Page 4: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/4.jpg)
3
Molecular Dynamics (MD)
The aim of Molecular Dynamics (MD) is to study a system of interacting particles (solid, liquid, gas) by recreating it on a computer in a way as close as possible to nature and by simulating its dynamics over a physical length of time
relevant to the properties of interest.
Any MD method is an iterative numerical scheme forsolving some equations of motion (EOM), coded in a computer program, that represent the physical evolution of the system under study
![Page 5: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/5.jpg)
4
F = m a
e-( E/kT)ab initio
moleculardynamics
Monte Carlo
mesoscale continuum
Time and Length scales in simulations
Length Scale (m)
Tim
e S
cale
(s)
10-10 10-8 10-6 10-4
10-6
10-8
10-12
![Page 6: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/6.jpg)
5
Classical Molecular Dynamics (MD)
The objects described by Molecular Dynamics (MD) are particles(atoms, molecules, polymers) represented as deterministic variableshaving positions and velocities.
The Cartesian positions (x1,y1,z1),…,(x N,yN,zN) of a system of Nparticles can be denoted by RI = (xI,yI,zI) where I = 1,…,N
These particles interact via a given function of the positions RI, the potential V(RI). The forces fI on each particle are simply thegradients (derivative) of this potential,
I
II
V
R
Rf
)(
and the analytical 3xN dimensional function fI is called force field
![Page 7: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/7.jpg)
6
MD Simulations: how to construct a Force Field?
We have to consider all the interactions that we want to study
intermolecularinteractions
intramolecularnonbondedinteractions
torsion
bond stretch
bending
2021 )(),( IJIJIJIJIJ rrkkrv
2021 )()( IJKIJK kv
v( IJKL)
![Page 8: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/8.jpg)
7
MD Simulations: How do we get the parameterskIJ, k , etc… for the force field ?
From experimentsmolecular and bulk properties
X-ray / neutron scattering structure factors
isotopic substitutions, etc…
From ab initio calculationsmolecular and cluster properties
static geometry optimization (minima, saddle points)
first principles methods
![Page 9: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/9.jpg)
8
“Eppur si muove…” (Galileo Galilei, 1564-1642)
Atoms are generally not fixed at a given position RI, but theymove due to e.g finite temperature T > 0 K, collisions, externalfields, etc…so they are rather described by dynamical variablesRI(t). The evolution of RI(t) for each atom I=1,…,N is described in analytical mechanics by a Lagrangean(*)
I
K
I
IIII VMVTL RRRR1
2
2
1,
(*) after Joseph Louis Lagrange, Torino (Italy), 1736-1813.See Mechanique analytique and Miscellanea Taurinensia,
1766-1773 (5 volumes).
![Page 10: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/10.jpg)
9
The Euler-Lagrange equations give us the equations of motion (EOM) of Classical MD
The atoms move from a position RI(t) to a new position RI(t+dt)via standard dynamics:
0II
LL
td
d
RR
or, more explicitly, the old good Newton equation
IIII VM fRR )(
where MI is the mass of the Ith particle of our system moving inthe force field fI
![Page 11: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/11.jpg)
10
Numerical integration of the EOM: Verlet algorithm
To solve numerically the EOM, the first step is a discretization
of the time t in terms of small increments called time steps of(arbitrary) length t. The system then passes across a set of timeordered configurations
… RI(tm-1) RI(tm) RI(tm+1) …
separated in time by t = tm-tm-1 = tm+1-tm = …An easy but successful integrator in MD is the Verlet algorithm;if we make a Taylor expansion of RI(t+ t) and RI(t- t) we get
432
)(6
)(2
)()()( tOtt
tM
tttttt II
I
III bfvRR
432
)(6
)(2
)()()( tOtt
tM
tttttt II
I
III bfvRR
![Page 12: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/12.jpg)
11
We use the standard notation
By simply summing up the two Taylor expansions, we get
42
)()()(2)( tOtM
tttttt I
I
III fRRR
IIIIII RbRaRv ,,
III M af
and this has an accuracy which is third order in time.The velocity results, in a similar way, as
and the EOM
4)()(2
)()( tOtttM
tttt II
I
II ffvv
and this is what we call the velocity Verlet algorithm, a standard in any MD computer code.
![Page 13: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/13.jpg)
12
Thermodynamic control methods: constant temperature (I)
From a statistical mechanics point of view, the temperature T of asystem of particles having masses MI and moving with velocities vI
is simply given by the average
TkNM
B
N
I
II
2
3
21
2v
It seems than that if we can control in some way the velocities vI
then we can keep under control the related parameter T.From statistical mechanics textbooks, we know that this control is operated by a thermostat, ideally an auxiliary system with infinite
degrees of freedom in contact with our system of particles and acting as a heat reservoir.
![Page 14: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/14.jpg)
13
Thermodynamic control methods: constant temperature (II)
An original idea of S. Nosé, extended by W.G. Hoover, was that we do not really need infinite degrees of freedom, but only one.This additional degree of freedom takes care of the proper scaling of the velocities, reproduces the canonical ensemble (N,V,T) and reduces the problem of the thermostat to deterministic dynamical equations.
This single degree of freedom, s, does a rescaling of the velocities vI
during the simulation,
and is similar to a friction force fcI=-spI if s > 0 or to a heating
process if s < 0.
III s vvR
![Page 15: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/15.jpg)
14
Thermodynamic control methods: constant temperature (III)
The new single degree of freedom s can be added to the Lagrangean of the molecular dynamics as a new variational variable. The extended Lagrangean then reads
The new variable s(t) has a kinetic term in which Q represents the fictitious mass, basically the time scaling of the motion of s(t) with respect to the motion of RI(t). The velocities are rescaled as svI andthe virtual potential for s(t) is given by the Boltzmann-like canonical term (3N + 1) kBT ln s, where 3N + 1 are all the degrees of freedom: 3N for the atomic coordinates plus one for the new variable s.
sTkNsQVsML BI
N
I
II ln)13(2
1
2
1 2
1
22RR
![Page 16: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/16.jpg)
15
Thermodynamic control methods: constant temperature (IV)
The (coupled) Euler-Lagrange equations of motion, under the rescaling of the Nosé-Hoover thermostat become then
III
II Ms
s
sM R
fR 2
2
N
I
BII TkNMss
sQ1
2 )13(1 2
R
and these can be solved numerically with the Verlet algorithm. Note that now we have two more input parameters: the target temperature T and the Nosé-Hoover thermostat mass Q.
![Page 17: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/17.jpg)
16
What can we get out of a MD simulation ?
• Dynamical averages = statistical averages of several physical quantities: total and free energy, molecular velocity distributions, etc.
• Radial distribution function / pair correlation functions gij(r)and angular distribution functions
• Temperature, pressure (stress tensor), crystal and non-crystal phases, etc…
• Diffusion coefficients
• Vibrational spectra and normal modes
• etc…
Let’s see each point one by one and how we can extract these
data from a MD trajectory.
![Page 18: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/18.jpg)
17
Connection to Statistical Mechanics:time averages and ensemble averages
A successful MD run gives you, in output, a time-ordered sequenceof positions RI(tm) and velocities vI(tm) at the discrete time pointstm = m t, m=1,…,M for a total simulation time t = M t.
We call this sequence trajectory and the trajectory represents the set of configurations “visited” by our system during the dynamics, i.e. during its motion under the action of the force field that you selected.
So, what do we do with such a sequences of numbers ?
We can use this discrete trajectory to visualize the motion of the particles on most of the graphical PCs, workstations…
![Page 19: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/19.jpg)
18
…and what do we see on our computer ?
Example: the case of water… a droplet of H2O molecules atT = 300 K and density of 1 g/cm3
![Page 20: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/20.jpg)
19
Connection to Statistical Mechanics:time averages and ensemble averages
…but this is just a visual inspection (= no or very little science).In the end we want always to compute ensemble averages or, within the ergodic principle, time averages of some function A(RI(t),vI(t))of the positions and velocities, or, equivalently A(RI(t),pI(t)) of the positions and momenta. These averages are quantities that can be measured experimentally.
Ergodic principle: if the dynamics is long enough so that our system can explore its whole phase space {RI,pI}, then time averages are identical to ensemble averages, which means, if
M
m
mm
t
tEnsemblettA
MdtttA
tAA
10
)(),(1
)(),(1
pRpR
t
![Page 21: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/21.jpg)
20
Connection to Statistical Mechanics:a trivial example – Temperature and Total Energy
VTNkVtMtVtME B
N
I
II
N
I
IIIt
Tot
2
3)]([
2
1)()]([
2
1
1
2
1
2vRv
T
ETot
0dt
dETot
![Page 22: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/22.jpg)
21
Connection to Statistical Mechanics:static properties
If our system has an average density ( = 1 g/cm3 for H2O), thisdensity fluctuates as a function of the distance from a given point…
The density at a given radial distance r depends on whether or notanother particle is present at r and is a measure of the structure of the system. The adimensional function g(r) is called radial distribution
function: It is zero at r = 0 (two particles cannot occupy the same place) and because in the whole volume (r) = .1)(lim rg
r
(r) = g(r)
From a MD trajectory :N
JI
IJrrN
rg1
)(
![Page 23: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/23.jpg)
22
Connection to Statistical Mechanics: extracting properties from simulations
Summarizing:
static propertiessuch as structure, pressure etc.. are obtained from g(r)pair (radial) distributionfunctions
O-O
H-H
O-H
See P. L. Silvestrelli and M. Parrinello, J. Chem. Phys. 111, 3572 (1999)
![Page 24: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/24.jpg)
23
Connection to Statistical Mechanics:dynamic properties
The dynamics of the system can be measured from the displacementof its particles in time or, equivalently, from the velocities. An easyquantity to compute is the mean square displacement (MSD)
N
I
II tN
tR1
22 )0()(1
)( RR
In general, for a solid R2(t) is small and almost constant in time (non-diffusive regime).Instead, in a liquid or a gas it grows roughly linearly (diffusive regime)
)0(6)( 22 RtDtR
D is the (self)diffusion coefficient to be compared with experiments
DtRtd
d)(
6
1 2
![Page 25: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/25.jpg)
24
Connection to Statistical Mechanics:dynamic properties
The velocities provided by the MD can be used in a mathematicallyequivalent way since
t
III tdtt0
)()0()( vRR
and if we evaluate the time derivative of the MSD,
)()(2)()()(0
2 tttdtttdtdtd
dtR
td
d t
o
t t
ovvvv
using the invariance of the origin of time. So, finally,
)0()(2)0()(2 vvvv ttdtttdt
o
t
o
t
ttdD0
)0()(3
1vv
![Page 26: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/26.jpg)
25
Connection to Statistical Mechanics:dynamic properties
The quantity
is called velocity autocorrelation function and is useful also to compute the vibrational spectrum of the system
ti
oettdI )0()( vv
N
I
II tN
t1
)0()(1
)0()( vvvv
as a simple Fourier transform from the time domain to the frequencydomain.
![Page 27: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/27.jpg)
26
Connection to Statistical Mechanics:dynamic properties - Vibrational spectrum of water
bending
stretching
![Page 28: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/28.jpg)
27
Summarizing:
dynamic and transport propertiesare obtained from time correlation functions
Connection to Statistical Mechanics:extracting properties from simulations
MD (good)simulation
Experiment
2.8 0.5 2.4
Example: self diffusion coefficient of water
D (x10-5 cm2/s)
![Page 29: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/29.jpg)
28
Suggestions for further readings:
• Understanding Molecular Simulations, From Algorithms to Applications,D. Frenkel and B. Smit, Academic Press, San Diego, 1996 • Computer Simulation of Liquids, M. P. Allen and D. J. Tildesley, Clarendon Press, Oxford, 1987• An Introduction to Computational Physics, T. Pang, Cambridge University Press, Cambridge, 1997• http://www.fz-juelich.de/nic-series/volume23/volume23.html(freely downloadable)
•S. Nosé, Mol. Phys. 52, 255 (1984)•S. Nosé, J. Chem. Phys. 81, 511 (1984)•W. G. Hoover, Phys. Rev. A 31, 1695 (1985)•H. C. Andersen, J. Chem. Phys. 72, 2384 (1980)•M. Parrinello and A. Rahman, Phys. Rev. Lett. 45, 1196 (1980)•M. Parrinello and A. Rahman, J. Appl. Phys. 52, 7182 (1981)
![Page 30: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/30.jpg)
29
First Principles MD : instead of looking for a potential V(RI),we try to include quantum electrons and classical nuclei and to
compute forces from fundamental quantum mechanics
electron-electron
ion-ion
electron- ion
electron- ion
i(x)
RI
Beside the atom-atom (ion-ion) interaction, new ingredients must be included
![Page 31: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/31.jpg)
30
Density Functional Theory: brief review
• Define the electronic density (x) as a superposition of single particle Kohn-Sham (KS) orbitals
• Write the total energy functional as
E[ i,RI] = Ek + EH + Exc + Eps + EM
i.e. sum of electron-electron + electron-ion +
ion-ion interaction
![Page 32: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/32.jpg)
31
Density Functional Theory: brief review
• Electron-electron interaction:
• Ek = kinetic energy, EH = Coulomb interaction,
Exc = exchange-correlation interaction (our
way to write what we do not know:
the many-body interaction)
![Page 33: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/33.jpg)
32
Density Functional Theory: brief review
• Electron-ion interaction:
the core-valence interaction is described by pseudopotentials
• Ion-ion interaction:
![Page 34: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/34.jpg)
33
HOMO-3
-25.50 eV
HOMO-2
-13.21 eV
HOMO-1
-9.38 eV
HOMO
-7.26 eV
H2O: electronic (KS) states
LUMO
-1.00 eVLUMO+1
+0.50 eV
LUMO+2
+1.10 eV
O-H = 0.966 A
HOH = 104.3o
![Page 35: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/35.jpg)
34
Electronic structure from QM: the case of water
Lone pairs bond
![Page 36: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/36.jpg)
35
Car-Parrinello Molecular Dynamics• In 1985, Roberto Car (Princeton University) and Michele Parrinello (Swiss
Federal Institute of Technology) proposed to use the fundamental quantum mechanics to replace the classical V(RI) with E[ i,RI]
• And extend the classical MD Lagrangean by adding the electronic degrees of freedom i plus a constraint for keeping the wavefunctions orthogonality
• And any external additional variables q (e.g. thermostats, stress, etc.)
![Page 37: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/37.jpg)
36
Car-Parrinello Molecular Dynamics
• Solve the related Euler-Lagrange EOM
- R. Car, M. Parrinello, Phys. Rev. Lett. 55, 2471 (1985)
- M. Parrinello, Comp. in Sci. & Eng. 2, 22 (2000)
Temperature control,
pressure, control, etc.
![Page 38: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/38.jpg)
37
BO surface
CP trajectory
BO trajectory
The difference between the CP trajectories RICP(t) and the
Born-Oppenheimer (BO) ones RIBO(t) is bound by
| RICP(t) - RI
BO(t)| < C 1/2
(C > 0) if 020HOMOLUMO
See F.A. Bornemann and C. Schütte, Numerische Mathematik vol.78,N. 3, p. 359-376 (1998)
![Page 39: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/39.jpg)
38
Practical implementation• To implement the CP-EOM numerically, the KS orbitals are
generally expanded in plane waves
i(x) = G ci(G) eiGx
• G are the reciprocal space vectors. The Hilbert space spanned by PWs is truncated to a suitable cut-off Ecut such that
G2/2 < Ecut
• PWs have so far been the most successful basis set, in particular for extended systems requiring periodic boundary conditions
• The basis set accuracy can be systematically improved in a fullyvariational way
• PWs do not depend on atomic positions (no Pulay forces)
• However they are not localized and can be inefficient, e.g., for small molecules or clusters in a large simulation cell
![Page 40: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/40.jpg)
39
R space G space
E2cut > E1
cut
Plane wave expansion: i(x) = G ci(G) eiGx
For each electron i=1,…,N , G=1,…,M are the reciprocal space vectors. The Hilbert space spanned by PWs is truncated to a cut-off Gcut
2/2 < Ecut
E1cut
![Page 41: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/41.jpg)
40
Practical implementation
• Verlet’s algorithm on EOM gives
( / t2) [ | i(t+ t)> + | i(t- t)> - 2 | i(t)>] = -(HCP+ ) | i(t)>or, in G-space,
• The ionic degrees of freedom RI(t) are updated at a rate (speed) t while the electronic degrees of freedom | i(t)> are updated at
a rate t/ ( t ~ 5 a.u., ~ 500 a.u.), hence they are much slower (adiabatic) with respect to the ions.
• The parameter that controls the adiabaticity (BO) is
• However this t is driven by QM and is much shorter than t inclassical MD
j
jiji
CP
iii ccHtcttcttct
)()(),(2),(),(2
GGGGGGGG
![Page 42: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/42.jpg)
41
What can we get out of a CPMD simulation ?
• Dynamical averages of several physical quantities: total and free energy, molecular velocity distributions, etc.
• Radial distribution function / pair correlation functions gij(r)
• Diffusion coefficients and vibrational spectra
…as in classical MD.
So what’s new ?
• Electronic structure evolution during the dynamics
• Infrared (from dipole autocorrelation function) and Ramanspectra
• NMR chemical shifts and hyperfine ESR parameters
• Breaking and formation of chemical bonds if the barrier is zero or very small (~ kBT)
![Page 43: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/43.jpg)
42
Suggestions for further readings:
• T. Pang, An Introduction to Computational Physics, Cambridge University Press, Cambridge, 1997• R. Car and M. Parrinello, “Unified approach for molecular dynamics and density-functional theory”. Phys. Rev. Lett. 55, 2471-2474 (1985) • http://www.fz-juelich.de/nic-series/Volume3/Volume3.html(freely downloadable)
![Page 44: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/44.jpg)
43
Lecture 2: Reactive first principles molecular dynamics
Mauro Boero
Center for Computational Sciences, University of Tsukuba,
1-1-1 Tennodai, Tsukuba, Ibaraki 305-8577 Japan
![Page 45: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/45.jpg)
44
An easy “reaction”: breaking and formation of H-bonds
When water aggregates to form a liquid state (or solid ice), molecules stick together by forming hydrogen bonds (H-bonds):
+
+
+ - ’
- ’
+
H-bond
The energy required to break one H-bond is 2.6 kcal/mol in normal liquid water*. These small barriers can easily be overcome in anyMD simulation and well reproduced also by classical force fields.
*From low-frequency Raman scattering; G.E. Walrafen et al., J. Chem. Phys. 85,6970 (1986)
![Page 46: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/46.jpg)
45
…and possible H-bond configurations in H2O
Linear H-bond (LHB)Dipole Moment = 2.90 D
Cyclic H-bond (CHB)Dipole Moment = 0.23
Bifurcated H-bond (BHB)Dipole Moment = 4.38 D
2-fold H-bond (2fHB)Dipole Moment = 1. 80 D
![Page 47: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/47.jpg)
46
When do we see these H-bonds ?
Average number of the different H-bond configurations per 32 molecules of H2O at different thermodynamic conditions
Configuration =0.32 g/cm3
T = 647 K
scH2O @ lowdensity
=0.73 g/cm3
T = 653 K
scH2O @ high density
=1.00 g/cm3
T = 300 K
Normal liquid water
CHB 0.615 ± 0.013 1.863 ± 0.021 0.015 ± 0.001
BHB 0.452 ± 0.008 0.320 ± 0.011 0.000
2fHB 0.260 ± 0.012 1.601 ± 0.016 0.525 ± 0.011
E(CHB-LHB) = 2.2 kcal/mol (local minimum)E(BHB-LHB) = 2.8 kcal/mol (local minimum)
![Page 48: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/48.jpg)
47
An example of dynamical chemical bond breaking and
formation: first principles MD of the diffusion of protons in water (electronic structure becomes important…)
H3O+
H3O+
H3O+
H3O+
H3O+
The presence of a hydrogen bond network
linking water molecules is crucial to propagation
R. Pomès and B. Roux, J. Phys. Chem. 100, 2519 (1996)
![Page 49: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/49.jpg)
48
A proton in normal liquid
water:= 1.0 g/cm3
T = 300 K
simulation time displayed: 0.6 ps
total simulationtime: 19.1 ps
![Page 50: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/50.jpg)
49
The excess proton makes a continuous switch betweenEigen H9O4
+ complex and Zundel H5O2+ shared proton
structure, forming and breaking bonds
Eigen Zundel
H-bond -bond
![Page 51: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/51.jpg)
50
Proton diffusion: a couple of observations
• A simple first principles MD seems to be able to simulate the basic chemistry process for a proton diffusion: the Grotthus switchingfrom a bond (O-H) to a hydrogen bond (H-bond)• It seems than that if we can control in some way the temperature of the water, we can speed up (heating) or slow down (cooling) such a process [Note: this is the idea behind the use of supercritial water
that we shall discuss later]• Indeed, for simple chemical reactions that occur with little (fewkBT) or no energy barrier, thermal fluctuations are enough to activate the process• Yet, kBT = 26 meV = 0.60 kcal/mol at T = 300 K
…so let’s try to heat up our system from 300 K to 673 K*
* Typical supercritical temperature
![Page 52: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/52.jpg)
51
High temperature: = 1.0 g/cm3
T = 673 K
simulation time displayed: 0.6 ps(total simulation
time: 20.0 ps)
the H+ diffusionbecomes faster
![Page 53: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/53.jpg)
52
Theoretical diffusion coefficients computed from the velocity-velocity autocorrelation function. D and from the mean square displacement, DMSD.The experimental proton diffusion coefficient is D
exp, diffusion and Dself is
the water self-diffusion coefficient.
[a] R. A. Robinson and R. H. Stokes, Electrolyte Solutions, Butterworths, London, 1959.
However this is just a lucky case… in general a chemical reaction requires
the overcoming of a much higher energy barrier.
(g cm-3) T (K) D (cm2/s) DMSD (cm2/s) Dexp (cm2/s) Dself (cm2/s)
1.0 300 13.0 x 10-5 15.0 x 10-5 9.3 x 10-5 [a] 2.5 x 10-5
1.0 673 70.0 x 10-5 62.0 x 10-5 33.0 x 10-5
![Page 54: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/54.jpg)
53
From a general reactant A to a product B, we have to climb above a (more or less) high mountain
• A general chemical reaction starts from reactants A and goes into products B
• The system spends most of the time either in A and in B
• …but in between, for a short time, a barrier is overcome and atomic andelectronic modificationsoccur
• Time scale: Tk
F
molBe
*
~ ~*F
![Page 55: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/55.jpg)
54
How to simulate a chemical reaction ?
A + B
AB
• The reaction A+B AB passes across a rare event (transition state) that is difficult to see
• CP molecular dynamics time scale is of the order of some tens of ps and classical simulations are of the order of few ns, but still insufficient… we cannot wait forever…
![Page 56: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/56.jpg)
55
init final
A+B
AB
Select the reaction coordinate to be sampled, e.g. theinteratomic distance = |RA – RB|Add to the CP lagrangean LCP the holonomic constraintLCP LCP + ( - 0) ( = Lagrange multiplier)Compute the ensemble (time) average < >
E( )
Blue Moon ensemble theory (I)
why ?
![Page 57: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/57.jpg)
56
Blue Moon ensemble theory (II)From the standard definition of free energy
F = -kBT ln <exp{-[H CP + ( - 0)]/kBT}>
using exp(x) ~ 1+x+O(x2) and ln(1+x) ~ x - O(x2) we get
TkTkTkF BB
CP
B /)(/H1ln 0
)(H 00CP
0HCP
00000 ,0
)(H/)(H 00CP
B
CP
B TkTk
but because HCP is a constant of motion
and
because 0 is the average value of (and a constant !)
![Page 58: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/58.jpg)
57
Blue Moon ensemble theory (III)
…so the variation of the free energy become simply
If we now integrate the average constraint force along the sampled reaction path between init and final , we finally arrive at the following expression for the free energy difference between initial (reactant) and final (product) states
FF
![Page 59: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/59.jpg)
58
Metadynamics (I): reactive CPMD with several collective variables
MtN ,...,,..., 101 RR
1) The atomic and electronic configuration of our initial system isgiven by a set of variables
2) And we assume that some known functions of few of them (collective variables) are necessary and sufficient to describethe process we are interested in
MNns iI ,,...,1;R
3) …so that the FES is a function of these changing variables
ntstF ,...,1)()()( sss
![Page 60: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/60.jpg)
59
• For reconstructing F(s) in many dimensions as a function of one or more collective variables.
• Used as a tool for escaping local (free energy) minima, i.e. accelerating rare events, reconstructing the free energy in the selected interval of reaction coordinates (not everywhere !).
• Artificial dynamics in the space of a few collective variables [1]• The CPMD dynamics is biased by a history-dependent potential constructed as a sum of Gaussians [2]. • The history dependent potential compensates, step after step, the underlying free energy surface [3,4].
[1] I. Kevrekidis et al, Comput. Chem. Eng. (2002)[2] T. Huber et al, J. Comput. (1994)[3] F. Wang and D. Landau, Phys. Rev. Lett. (2001)[4] E. Darve and A. Pohorille, J. Chem. Phys. (2001)
What is it used for ?
Metadynamics (II)
![Page 61: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/61.jpg)
60
Metadynamics (III): how does it work ?
•Put a “small” Gaussian•The dynamics brings you to the closest local minimum of F(s) plus the sum of all the Gaussians
F(s)
s
![Page 62: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/62.jpg)
61
In the one dimensional example shown, the system freely moves in a potential well (driven by MD).
• Adding a penalty potential in the region that has been already explored forces the system to move out of that region,
• …but always choosing the minimum energy path, i.e. the most natural path that brings it out of the well.
• Providing a properly shaped penalty potential, the dynamicsis guaranteed to be smooth
• Therefore the systems explores the whole well, until it finds the lowest barrier to escape.
Metadynamics (IV): summarizing…
![Page 63: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/63.jpg)
62
History-dependent
potential
Fictitious kinetic energy
Restrain potential: coupling fast and slow variables
(k /M ) « I
Metadynamics (V): plugging the method in CPMD – yet another extended
Lagrangean
![Page 64: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/64.jpg)
63
We use (again) the velocity Verlet algorithm to solve the EOM
and we have two new contributions to the force
Metadynamics (VI): Equations of motion for the collective variables
![Page 65: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/65.jpg)
64
1. The parameter t in s (t) is not a real time, but rather an index labeling the configurations thatexplore the FES
s || = s (t+ t) - s (t) can be arbitrarily large (coarse-grained metadynamics) and gives the accuracy in the sampling of the FES along the trajectory described by the s (t) variables
3. The dynamics at each fixed s is a true dynamicsused to explore the local portion of FES
4. The FES is smoother than the PES, being dependent on a reduced number of variables
Metadynamics (VII): note that…
![Page 66: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/66.jpg)
65
4||
2
121
2
2
21
)(
)()(exp
)(exp,
i
iii
tt
ii
ssWtV
i
sssssss
where the Dirac has been expressed in the approximateGaussian form
ttx
x )(2
1exp
2
12
2
s
and the discrete time step t must be such that
1s
CPMD ttCPMD time step Highest oscillationfrequency of s
Metadynamics (VIII): Discrete form of V(s,t) used in the implementation
![Page 67: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/67.jpg)
66
When the (meta)dynamics is over and the walker has explored all the portion of the {s} phase space available, we have completed our job (at large t) and filled all the local minima,then the shape of V(s,t) is similar to the FES apart from a sign and an arbitrary additive constant
.)(),(lim constFtVt
ss
In practice: the number of gaussians required to fill a minimumis proportional to (1/ )n (n = dimensionality of the problem) and
5.0/2/122/1 feW
Metadynamics (IX): FES reconstruction, i.e. what V(s,t) is used for
![Page 68: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/68.jpg)
67
Metadynamics (X): Example of FES exploration by two collective variables
![Page 69: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/69.jpg)
68
F(s1,s2)(kcal/mol)
0
-5
-10
s1,s2 (Å)1 1.5 2 2.5 3 3.5 4
ATP
ADP
F# = FATP-FTS = 3.0 kcal/molF = FATP-FADP = 6.9 kcal/mol
(exp. 7.1 kcal/mol)
-2
-4
-6
-8
-10
reactant
F(s1,s2) (kcal/mol)
0
-5
-10
1.02.0
product
s1 (Å)3.0
4.0
4.03.0
2.0s2 (Å)
(Example from a true simulation on ATP hydrolysis)
![Page 70: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/70.jpg)
69
• Parameters are generally system dependent
• The collective variables s must be identified by the user
• …and must be able to discriminate between initialand final state
• …and must keep into account all the slow degreesof freedom
• Although the trajectory generated describes the most probable reaction pathway, it is not the true dynamics
Metadynamics (XI): caveat
![Page 71: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/71.jpg)
70
Note: connection between metadynamics and Blue Moon
In the case of a single collective variable (reaction coordinate), =1
0),( tsV• Do not update the Gaussians:
• Do not move dynamically s(t):
),()(2
1
2
1 22 tsVssksM IR
0s
2)( ss IR
where = k/2.…Then we are back to a Blue Moon formulation.
}update manually
![Page 72: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/72.jpg)
71
Further readings:
• M. Sprik, G. Ciccotti, J. Chem. Phys. 109, 7737 (1998): Blue Moon method• M. B. et al., J. Chem. Phys.112, 9549 (2000): limitations of Blue Moon• van Gunsteren et al. J. Comput. 8, 695 (1994): adaptive elevated potential• H. Grubmüller Phys. Rev. E 52, 2893 (1997): flooding potential • D. Chandler et al. J. Chem. Phys. 108, 1964 (1988): path sampling• D. Passerone and M. Parrinello, Phys. Rev. Lett. 87, 8302 (2001): action derived reaction path• A. Laio and M. Parrinello, Proc. Nat’l. Ac. Sci. USA 99, 12562 (2002): coarse grained non-Markovian (meta)dynamics • M. Iannuzzi et al., Phys. Rev. Lett. 90, 238302 (2003): CPMD metadynamics• M. B. et al., J. Chem. Theory Comput. 1, 925 (2005); CPMD metadynamics
![Page 73: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/73.jpg)
72
Lecture 3: Water as a catalyst
Mauro Boero
Center for Computational Sciences, University of Tsukuba,
1-1-1 Tennodai, Tsukuba, Ibaraki 305-8577 Japan
![Page 74: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/74.jpg)
73
Main interests in scH2O• Water above its critical point T=654 K, P=22.1 MPa,
=0.32 g/cm3 is found in nature inside the earth mantel and in the “black jets” of deep oceanic trenches (like the Japan trench).
• scH2O has important technological applications:
i) it is an advanced tool for the treatment of hazardous wastes such as dioxin and PCB [1]
ii) it promotes non-catalytic chemical reactions and can replace non-polar solvent and acid catalysts which are environmentally dangerous [2]
• [1] J. W. Tester et al., ACS Symp. Series, Emerging Technologies for Hazardous Waste Management, vol. III, ACS, Washington DC, 1991
• [2] D. Bröll et al., Angew. Chem. Int. Ed. 38, 2999 (1999)
![Page 75: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/75.jpg)
74
Hydrogen-bond network in water
Normal water
T = 300 K = 1.00 g/cm3
Supercritical water:
Disrupted H-bond NW
T = 653 K= 0.73 g/cm3
scH2O
T > 647 K
P > 22.1 MPa
![Page 76: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/76.jpg)
75
scH2O : = 0.66 g/cm3 T =
673 K
simulation time displayed: 0.6 ps
total simulationtime: 18.4 ps
![Page 77: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/77.jpg)
76
Supercritical water: = 0.66 g/cm3, T = 673 K
Eigen (H7O3)+ unstable (H5O2)
+ Zundel (H5O2)+
A more complex and varied scenario arises in SCW and the proton displaces faster than in the normal liquid state
![Page 78: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/78.jpg)
77
Eigen complex pair correlation functions
![Page 79: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/79.jpg)
78
In general a proton H+ is transferred to a H2O molecule but it is a different one that subsequently leaves the H3O
+
H+ + H-O-H H-OH2 H-O-H + H+
We then follow the center of mass of the Eigen/Zundel complex:
)(1
)(1
i
N
I
II
EZ
iEZ tMM
tEZ
RR
where NEZ is the number of atoms forming the Eigen or Zundel complex, respectively. MEZ is of course the related total massIn practice, REZ(ti) corresponds to the position of the central O atom in the case of Eigen complex and to the position of the shared proton in the case ofZundel complex
~ REZ(ti)
~ REZ(tj)
Computing the diffusion of protons water:
![Page 80: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/80.jpg)
79
• We then constructed the velocity of the migration of the structural defect as
ii
iEZiEZiEZ
tt
ttt
1
1 )()()(
RRv
• Then computed the diffusion coefficient, by using the Einstein relationexpressed in terms of the Green-Kubo formula, assuming this velocitydefinition in the autocorrelation function
t
EZEZ tD0
03
1vv
being t half of the simulation time (~ 9 ps) and as
2)0()(lim
6
1EZEZ
t
MSD tdt
dD RR
Computing the diffusion of protons water:
![Page 81: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/81.jpg)
80
[g cm-3] T [K] D [cm2 s-1] DMSD [cm2 s-1] Dexp
[cm2 s-1] Dself [cm2 s-1]
1.00 300 13.0 x 10-5 15.0 x 10-5 9.3 x 10-5 [a] 2.5 x 10-5
1.00 673 70.0 x 10-5 62.0 x 10-5 - 33.0 x 10-5
0.66 673 54.0 x 10-5 55.0 x 10-5 49.0 x 10-5[b] 54.0 x 10-5
0.32 673 37.0 x 10-5 41.0 x 10-5 17.0 x 10-5[b] 177.0 x 10-5
[a] R. A. Robinson and R. H. Stokes, Electrolyte Solutions, Butterworths, London, 1959.[b] P. C. Ho, H. Bianchi, D. A. Palmer and R. H. Wood, J. Sol. Chem. 2000, 29, 217. The experimental value closest to 0.66 g cm-3 is = 0.63 g cm-3 and T = 623.15 K
Let’s complete our table of theoretical diffusion coefficients computed from the velocity-velocity autocorrelation function. D and from the mean square displacement, DMSD. The experimental proton diffusion coefficient is D
exp, diffusion and Dself is the water self-diffusion coefficient.
![Page 82: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/82.jpg)
81
Beckmann rearrangement for -Caprolactam
1. Commercially important for production of synthetic fibers (nylon 6). About 4 million tons per year.
2. Known to be catalyzed only by strong acids in conventional non-aqueous systems
3. Formation of byproducts (ammonium sulfate, (NH4)2SO4) oflow commercial value in acid catalyst: byproducts = 1.7 products (in weight)
4. Environmentally harmful: acid wastes are produced
Points 2, 3 and 4 and related problems can be eliminated
in scH2O: no acid required & no byproducts.
See Y. Ikushima et al., Angew. Chem. Int. Ed. 38, 2999 (1999);
J. Am. Chem. Soc. 122, 1908 (2000).
![Page 83: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/83.jpg)
82
Beckmann rearrangement reaction
in strong acid
and
supercritical H2O
hydrolysis in H2O
and
superheated H2O
protonattackto O
protonattackto N
![Page 84: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/84.jpg)
83
Schematic outline of reaction mechanism
+ H2OR—C—R’
N+
H+
R—N C—R’
HO
R—N C—R’
HOOH
H
R—C—R’
N—OH
oxime
R—N—C—R’
OH
amide
H2O
R—N C+—R’
OHH
![Page 85: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/85.jpg)
84
Which are the important ingredients that make
water special at supercritical conditions ?
• Proton attack is the trigger (experimental outcome !)
fast proton diffusion
difference in hydration
between O and N
High efficiency :
High selectivity :
![Page 86: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/86.jpg)
85
Computational details
• 1 cyclohexanone oxime immersed in a system of 60 H2O molecules
• DFT-based first principles molecular dynamics within BLYPgradient corrected approach
• Systems: T= 673 K, =0.66 g/cm3, L=14.262 Å supercritical
T= 673 K, =1.00 g/cm3, L=12.430 Å super heated
T= 300 K, =1.00 g/cm3, L=12.430 Å normal
• Temperature control via Nosé-Hoover thermostat chain
• Integration step of 4.0 a.u. (0.0967 fs) and fictitious electronic mass of 400 a.u.
• Reaction path sampling within both metadynamics and Blue Moon
![Page 87: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/87.jpg)
86
Cyclohexanone-oxyme in scH2O:• The energy barrier seems rather high and the reaction pathway not unique
• The reaction is generally acid catalyzed, hence protons are expected to be essential in triggering the process
• At supercritical conditions, however, the Kw of water increase, hence H+
and OH- can be around in the solvent in non-negligible concentration
• And small amounts of weak acids greatly enhance reaction rates
![Page 88: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/88.jpg)
87
Cyclohexanone-oxyme in scH2O in presence ofan excess proton H+:
H+
T = 673 K
![Page 89: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/89.jpg)
88
Cyclohexanone-oxyme in scH2O (+ H+)
R—C—R’
N—OH
H+
R—C—R’
N+
+ H2O
R—N C+—R’
A very small activation barrier
(about 1 kcal/mol) is required
for the N insertion process.
![Page 90: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/90.jpg)
89
~ 0.9 ps in scH2O
~ 5.2 ps in n-H2O
Contrary to normalliquid water, scH2Oaccelerates selectivelythe formation of the first intermediate
![Page 91: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/91.jpg)
90
~ 3.4 ps at 673 K
Superheated water:( = 1.00 g/cm3 )
Proton attacks N
Cycrohexanon
(byproduct) formation
![Page 92: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/92.jpg)
91
Efficient reaction in scH2Odue to fast
proton diffusion and acid properties of
the (broken) Eigen-Zundel complexes
Acid catalyst ?
![Page 93: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/93.jpg)
92
Proton diffusion in ordinary liquid and supercritical water
Is the proton diffusion slowed down in
scH2O ? Not really…
• Hydrogen bond network is
disrupted in SCW.
![Page 94: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/94.jpg)
93
Proton diffusion
normal water and supercritical water
Proton (structural defect) diffusion coefficient estimation in the 3 systems:
SystemDiffusion constant
D (cm2/s)Hydrogen bond
network
n-H2O(normal water)
15.0 x 10-5 continuous
Superheated
H2O62.0 x 10-5
continuous
(fast switch)
scH2O
(supercritical water)55.0 x 10-5 disrupted
In scH2Othe network is disrupted and the motion occurs in sub-networks that join and break apart rapidly due to density fluctuations; two diffusion regimes are cooperating: hydrodynamics (vehicular) and Grotthus
![Page 95: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/95.jpg)
94
complex formation in the process of proton diffusion in normal water
The excess proton makes a continuous switch between EigenH9O4
+ complex and Zundel H5O2+ shared proton structure
and turns out to be rather stabilized. Life time ~ 1.0 ps.
Eigen Zundel
![Page 96: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/96.jpg)
95
Unstable complex in scH2O ( = 0.66 g/cm3)
The Eigen H9O4+ complex is incomplete (H7O3
+) most of the time due to the disrupted H-bond network and the stabilization of the excess proton in the H-bond sub-network is reduced.
Pseudo-Eigen Zundel
life time ~ 0.1 ps
![Page 97: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/97.jpg)
96
Selective reaction in scH2O due to
different solvation of O and N
Reaction selectivity ?
![Page 98: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/98.jpg)
97
H+
T = 673 K
wet
dry
Cyclohexanone-oxyme in scH2O (+ H+):the selectivity
![Page 99: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/99.jpg)
98
Comparison of the pair correlation functions
(a) = n-H2O, (b) = superhetaed H2O, (c) = scH2O
Depletion(hole around N)
![Page 100: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/100.jpg)
99
+ H2OR—C—R’
N+
H+
R—N C—R’
HO
R—C—R’
N—OH
oxime
H2O
R—N C+—R’
OHH
R—N C+—R’
…and now the second step: C-O bond formation
![Page 101: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/101.jpg)
100
…and now the second step: Blue Moon approach
E = 5.9 kcal/mol
F = 5.1 kcal/mol
Approach of an H2Omolecule, = |Owat-C|
![Page 102: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/102.jpg)
101
+ H2OR—C—R’
N+
H+
R—N C—R’
HO
R—N C—R’
HOOH
H
R—C—R’
N—OH
oxime
R—N—C—R’
OH
amide
H2O
R—N C+—R’
OHH
The last step:eventually the -caprolactam
![Page 103: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/103.jpg)
102
The final step: Solvent-solute proton exchange via metadynamics
• Collective variables used:
s1(N-Hw) = coordination number of N with H of water molecules of scH2O
s2(Ow-H) = coordination number of O of scH2O with H of the solute
• M = 40.0 a.u. for both s1(N-Hw) and s2(Ow-H)
k = 0.6 a.u.
• V(s ,t) updated every 100 CPMD steps
1.0s
![Page 104: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/104.jpg)
103
Proton exchange with water
![Page 105: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/105.jpg)
104
s1(t)
s2(t)
![Page 106: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/106.jpg)
105
Total energy: a very rugged landscape
![Page 107: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/107.jpg)
106
Free energy surface: a less rugged landscape
s1 = 0.5
s1 = 1.0
![Page 108: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/108.jpg)
107
Water solvation and proton diffusion:what did we learn ?
• Proton diffuses with proton wire mechanism mediated by H bond. For an atom to be attacked by a proton, the atom has to be solvated.
• In normal water and superheated water, both O and N are solvated by water molecules. byproducts
• In supercritical water, only O is solvated and N is dry. desiredproducts only
• Moreover, solvation water around N prevents N insertion into ring.
R—C—R’
N+
R—C—R’
N+
H O
HscH2O normal H2O
super heated H2O
![Page 109: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/109.jpg)
108
Further readings:
• D. Bröll et al., Angew. Chem. Int. Ed. 38, 2999 (1999)• Y. Ikushima et al., Angew. Chem. Int. Ed. 38, 2910 (1999)• Y. Ikushima et al., J. Am. Chem. Soc. 122, 1908 (2000)• E. Fois, M. Sprik and M. Parrinello, Chem. Phys. Lett. 223, 411 (1994)• M.B., K. Terakura, T. Ikeshoji, C.C. Liew and M. Parrinello, Phys. Rev. Lett.
85, 3245 (2000)• M. B., K. Terakura, T. Ikeshoji, C.C. Liew and M. Parrinello, J. Chem. Phys.115, 2219 (2001)• M. B., M. Parrinello, K. Terakura, T. Ikeshoji and C.C. Liew, Phys. Rev. Lett.90, 226403 (2003) • M.B., T. Ikeshoji, C.C. Liew, K. Terakura and M. Parrinello, J. Am. Chem.
Soc. 126, 6280 (2004)• M.B., T. Ikeshoji and K. Terakura, ChemPhysChem 6, 1775 (2005)
![Page 110: Introduction to molecular dynamics simulations methods.indico.ictp.it/event/a06184/session/7/contribution/5/material/0/0.pdf · 3 Molecular Dynamics (MD) The aim of Molecular Dynamics](https://reader030.vdocuments.us/reader030/viewer/2022040410/5ecd254a5f998d2c8838de60/html5/thumbnails/110.jpg)
109
Acknowledgements
• Michele Parrinello, ETHZ-USI and Pisa University
• Roberto Car, Princeton University
• Kiyoyuki Terakura, JAIST, AIST and Hokkaido University
• Michiel Sprik, Cambridge University
• Pier Luigi Silvestrelli, Padova University
• Alessandro Laio, SISSA, Trieste
• Marcella Iannuzzi, Zurich Univeristy
• Michael L. Klein, University of Pennsylvania
• Giacinto Scoles, ICTP, Trieste
• Erio Tosatti, SISSA, Trieste
• Roger Rousseau, SISSA, Trieste
• Sandro Scandolo, SISSA, Trieste
…all of you for listening despite the rough, incomplete and approximate
level of these lectures.