Tutorial Pickard
-
Upload
ellintirta -
Category
Documents
-
view
221 -
download
0
Transcript of Tutorial Pickard
-
8/3/2019 Tutorial Pickard
1/26
GIPAW
A first rinci les theor of solid state NMR
Chris J PickardSchool of Physics and Astronomy
University of St Andrews
-
8/3/2019 Tutorial Pickard
2/26
2
n = 2d sin
Experiment
Theory
Structure
Theinverseproblem
-
8/3/2019 Tutorial Pickard
3/26
Magnetic resonance
NMR EPR
shielding, EFGs, g-tensor, hyperfine
-
8/3/2019 Tutorial Pickard
4/26
What kind of theory?
The trained spectroscopist
(Semi-)empirical computational methods
Interpolative
First principles or ab initio
Extrapolative and predictive
-
8/3/2019 Tutorial Pickard
5/26
Magnetic resonance is
quantum mechanical
Bint(
r)
Bext
J(r)
B = 0J
Born-Oppenheimer approximation: only treat electrons quantum mechanically
-
8/3/2019 Tutorial Pickard
6/26
Quantum mechanics
is difficult
|000+ b |001+ c |011+ d |111+
e |110+ f |100+ g |101+ h |010
|0+ |1|0
|1One particle, two states
Three particles, two states
(1 |0+ 1 |1)(2 |0+ 2 |1)(3 |0+ 3 |1)
Number of coefficients goes as 2N
In a material, the number of states and particles is very large
-
8/3/2019 Tutorial Pickard
7/26
Density functional theoryHohenberg-Kohn
E[] = F[] + Vext(r)(r)d3r
Kohn-Sham
(r) =
n
n
(r)n(r)Ts[] = 12
n
n
2nd3r
F[] = Ts[] +
(r)(r)
|r r|d3rd
3r
+ Exc[]
Changes the problem to the easy independent
particle option
-
8/3/2019 Tutorial Pickard
8/26
Density functional theory
But we still dont know whatExc[] is
The local density approximation
Exc[] = xc()(r)d3r
This is first principles
Exc[] = xc(,)(r)d3r
Generalised gradient approximations
These may or may not be first principles
-
8/3/2019 Tutorial Pickard
9/26
Extended Systems
Glass Ice Polymer
Simple crystal Zeolite Solution
-
8/3/2019 Tutorial Pickard
10/26
Cluster approximations
The only option for traditional methods
But you have to worry about:
making the cluster small
terminating the cluster
long range electrostatics
-
8/3/2019 Tutorial Pickard
11/26
Periodic boundary conditionsAl Bandstructure
1st Brillouin zone
Use the translational
symmetry of the crystal
nk(
r
) =eikr
u
nk(r
)
Blochs theorem
-
8/3/2019 Tutorial Pickard
12/26
Supercells
-
8/3/2019 Tutorial Pickard
13/26
Plane waves
ei(k+G)
r
Basis functions
Unlike atom centred atomic orbitals or gaussian
functions, there is no origin to a plane wave
There is no gauge origin problem
1
2 |k+G
|
2
Ecut
The quality of the basis is controlled by a single number
A Fourier series
-
8/3/2019 Tutorial Pickard
14/26
Pseudopotentials
Na
The core electrons are frozen and the valence
orbitals smoothed within the core radius
All-electron potential
Pseudopotential
rc
All-electron wavefunction
Pseudowavefunction
-
8/3/2019 Tutorial Pickard
15/26
A pseudopotential theory
The core electrons contribution to the
shielding must not be chemically sensitive
We must fix up the wavefunction near the nucleus
| = T | where T = 1 +
n
(|n |n)n|
Projector augmented waves
These break gauge invariance
-
8/3/2019 Tutorial Pickard
16/26
GIPAW
Gauge Including Projector Augmented Waves
PAW with gauge invariancerestored
Position operator rdescribed aslong wavelength limit of sin(qr)
Scalar relativistic effects includedusing ZORA
Response calculated in terms ofoccupied orbitals only (not sumover all states)
Implemented in: Paratec, (NMR-)CASTEP
-
8/3/2019 Tutorial Pickard
17/26
Doing calculations
Converge
Basis set
Brillouin zone integration
Supercell size
ChooseComputer
Code
DFT functionalPseudopotentials
100 atoms on a PC
200 atoms on a cluster
1000 atoms on a supercomputer?
My preferences
CASTEP, the GGA-PBE functional and ultrasoft pseudopotentials
-
8/3/2019 Tutorial Pickard
18/26
Structure
Where do you get your structure from?
The phase space of possible structures is large
Many applications start with X-ray structures
CAUTION: Check forces on atoms
Protons are typically misplaced
NMR chemical shifts are verysensitive to structure
-
8/3/2019 Tutorial Pickard
19/26
Dynamics
Calculations are typically performed at 0K
Experiments are typically performed at 300K
CAUTION: the atoms are really moving
This might be tackled via lattice dynamicscalculations or molecular dynamics simulations
An area of active research
-
8/3/2019 Tutorial Pickard
20/26
Density functional theory
Structures are typically good to 1-2%
Chemical shifts might be good to 2% of the shift range
An Inverse Lottery
When things gowrong they can go
very wrong
It is normally good
-
8/3/2019 Tutorial Pickard
21/26
Applications
Minerals
Amino acids
Protein fragmentsPharmaceutical
polymorphs
Boron carbidesGlasses
Bone
-
8/3/2019 Tutorial Pickard
22/26
Structure determination
Poster 342: Lyndon Emsley et al
Rose: from database
-L-Aspartyl-L-alanine
Orange: protons relaxed usingCASTEP
C
H
-
8/3/2019 Tutorial Pickard
23/26
Structure determination
PSD
PSD + CASTEP
Poster 342: Lyndon Emsley et al
H
C
-
8/3/2019 Tutorial Pickard
24/26
Conclusion
Good theory makesexperiments better
What do theorists get out of this?
NMR gives us a wealth of high quality data
which challenges us to improve our theories
We have quite a good theory- but it could be better
-
8/3/2019 Tutorial Pickard
25/26
Bibliography
Solid State PhysicsAshcroft andMermin Brooks Cole (1976)
Electronic Structure: basic theory and practical methods Martin Cambridge University Press (2004)
Iterative minimization techniques for ab initio total-energy
calculations: molecular dynamics and conjugate gradients
Payne et al Rev. Mod. Phys. 64, 1045 - 1097 (1992)
Calculation of NMR andEPR parameters Kaupp et al Wiley-VCH (2004)
All-electron magnetic response with pseudopotentials: NMRchemical shifts
Pickard and Mauri Phys. Rev. B. 63, 245101 (2001)
First principles methods using CASTEP Clark et alZeitschrift fur Kristallographie, 220,
567-570 (2005)
http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075
Direct link to Clarket al
http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075http://www.atypon-link.com/OLD/doi/abs/10.1524/zkri.220.5.567.65075 -
8/3/2019 Tutorial Pickard
26/26
Acknowledgments
Theorists Experimentalists Castep
Francesco Mauri
Jonthan Yates
Sian Joyce
Anne-Christine Uldry
Robin Harris
Sharon AshbrookSteven Brown
Lyndon Emsley
Melinda Duer
Ray Dupree
Ian Farnan
The CDG
Stewart Clark
Matt Probert
Phil HasnipMatt Segall
Keith Refson
Mike Payne
Accelrys
Victor MilmanGeorge Fitzgerald
http://www.castep.org/
http://www.castep.org/http://www.castep.org/http://www.castep.org/