Document not found! Please try again

Intro to Bandstructure Theory of Solids

7 downloads 0 Views 3MB Size Report
Nov 7, 2017 - William Penney. (1909-1991) ..... Tipler, P. A.; Llewellyn, R. A. “Modern Physics”, 5th Ed. (2008), p. 443. 11 / 07 / 2017 ... HH – Heavy Hole Band.
Dr. Walczak’s Group Meeting 2017

“Intro to Bandstructure Theory of Solids” OBJECTIVES  To present a Kronig-Penney model as one of the exactly solvable problems in quantum mechanics.  To present plots of E-k diagrams and discuss the concepts of effective mass and density of states.

 To explain transport and optical properties of bulk and 2D materials within the banstructure theory.

Dr. Kamil Walczak Department of Chemistry and Physical Sciences, Pace University 1 Pace Plaza, New York City, NY 10038

Dr. Walczak’s Research Group Principal Investigator: Dr. Kamil Walczak Student:

Topics:

Rita Aghjayan

“Thermal Rectifiers”, “Relaxation Processes”

Arthur Luniewski

“Molecular Noise”, “High-Intensity Heat Fluxes”

David Saroka

“Tunneling of Heat”, “Wave-Packet Approach”

Joanna Dyrkacz

“Inelastic Heat Flow”, “Electron-Phonon Effects”

Luke Shapiro

“Thermal Memristors”, “Hysteretic Loops”

Hunter Tonn

“Quantum Interference”, “Tunneling of Heat”

Erica Butts

“Neuronal Networks”, “Memristive Ion Channels”

Omar Tsoutiev

“Thermal Noise”, “Quantum-Interference Devices”

Kamil Walczak

11 / 07 / 2017

Periodic Structure of Crystal Lattice We have to solve the stationary problem of a particle in a one-dimensional potential created by ions in the periodic structure of crystal lattice. (arrangement of ions)

V0 - depth of potential well

V0

2a - width of potential well 2b - distance between wells

V  V0

L  2a  2b - periodicity

The regular potential inside the periodic lattice can be simplified by a series of identical one-dimensional potential barriers of rectangular shapes. Kamil Walczak

11 / 07 / 2017

Kronig-Penney Model (1931) Periodic crystal lattice of solids is modeled as ideal quantum-mechanical system that consists of an infinite periodic array of rectangular potential barriers!

Difficulties to include local defects (such as: chemical impurities, dislocations), interactions between electrons and other (quasi)particles!

Ralph Kronig (1904-1995)

The Kronig-Penney model qualitatively describes formation of band structure in solids, but due to unrealistic simplifications, it can not be used for any quantitative analysis!

William Penney (1909-1991)

R.L. Kronig & W.G. Penney, Proc. Roy. Soc. (London) A 130, pp. 499-513 (1931). Kamil Walczak

11 / 07 / 2017

Schrödinger Wave Equation Goal: to find energy levels and the corresponding wavefunctions for an electron in a periodic crystal lattice (solve the eigenvalue problem for energy). Time-independent Schrödinger wave equation:

 2 d2   V( x )   ( x )  E  ( x )  2  2m dx  V(x)  V(x  L) The wavefunction of a particle placed in a periodic potential is a product of a periodic Bloch function (that has the same periodicity as the potential) and a plane wave:

(x)  u(x) exp  ikx  Bloch function

u(x)  u(x  L) Kamil Walczak

Plane wave (free evolution) 11 / 07 / 2017

Bloch’s Theorem (1928)

|  ( x ) |2

Charge distribution corresponding to standing waves in 1D crystal.

“When I started to think about it, I felt that the main problem was to explain how the electrons could sneak by all the ions in a metal… By straight Fourier analysis I found to my delight that the wave differed from the plane wave of free electrons only by a periodic modulation.”

Felix Bloch (1905-1983)

F. Bloch, Z. Physik 52, pp. 555-600 (1928). Kamil Walczak

11 / 07 / 2017

Periodicity Due to periodic boundary condition, the following Schrödinger equation is satisfied:

 2 d2   V ( x )   ( x  L)  E  ( x  L)  2  2m dx  Both wavefunctions can differ only by constant and complex factor:

(x  L)   (x)

(x  nL)  n (x)



  ei

is real

|  |n  1

   

V( x )

a

a

a  2b

x (simplified potential)

 V0 Kamil Walczak

11 / 07 / 2017

Specific Solutions (1) in the area  a  x  a

 2 d 2 (x)   E  V0  ( x ) 2 2m dx 2m(E  V0 ) q 2 (2) in the area a  x  a  2b

 2 d 2 (x)   E  ( x ) 2 2m dx   Kamil Walczak

2mE 2

and for energy  V0  E  0

1 (x)  A eiqx  B eiqx (wave vector)

and for energy  V0  E  0

2 (x)  C e x  D e x (wave vector) 11 / 07 / 2017

Boundary Conditions Specific boundary conditions for x = a:

1 (x  a )  2 (x  a )

A eiqa  B eiqa  C e a  D e a

d1 ( x ) d2 ( x )  dx x a dx x a

iqA eiqa  iqB eiqa  C e a  D e a

Specific boundary conditions for x = a + 2b:

2 (a  2b)  1 (a  2b)  ei 1 (x  a )

d2 ( x ) d1 ( x ) d1 ( x )   ei dx a  2 b dx a  2 b dx x  a

 iqA e



C e (a 2b)  D e (a 2b)  A eiqa  B eiqa ei C e (a 2b)  D e (a 2b) Kamil Walczak

iqa

involving Bloch’s theorem



 iqB eiqa ei 11 / 07 / 2017

Linear Algebra: Special Condition Our specific boundary conditions can be written in the form of homogeneous system of four linear equations for the wavefunction’s coefficients:

 eiqa  iqa iqe   ei (  qa)  i (  qa) iqe

e iqa  iqe iqa ei (   qa)

 e a  e a  e (a 2b)

 iqe i (   qa)

 e  ( a  2 b )

 e  a  A   0       a e  B    0   e   ( a  2 b )  C   0           ( a  2 b )  e  D   0 

From linear algebra we know that non-trivial solution (where all coefficients are equal to zero) can be obtained when matrix determinant is equal to zero:

 eiqa  iqa iqe det  i (  qa) e  i (  qa) iqe Kamil Walczak

e iqa  iqe iqa ei (   qa)

 e a  e a  e (a 2b)

 iqe i (   qa)

 e  ( a  2 b )

 e  a   e  a  0 (a 2b)  e (a 2b)  e 

(*)

11 / 07 / 2017

Further Transformations Extension within the top row of the analyzed fourth-order determinant:

 iqe iqa eiqa ei (  qa)  iqe i (  qa)

iqe iqa  e a ei (  qa) iqe i (  qa)

 e a  e (a 2b)  e  ( a  2 b )

 iqe iqa ei (  qa)  iqe i (   qa)

iqe iqa  e a e  a e  a  e (a 2b)  e(a 2b)  e   ( a  2 b )  e iqa ei (  qa) iqe i (  qa)  e  ( a  2 b ) e   ( a  2 b ) e   ( a  2 b )

iqe iqa e  a  e   ( a  2 b )  e  a ei (  qa) iqe i (  qa) e   ( a  2 b )

 iqe iqa ei (  qa)  iqe i (  qa)

 e a  e (a 2b)  0  e  ( a  2 b )

Extension within the top row of the analyzed third-order determinants (Sarrus): (a 2b)   e eiqa   iqe iqa (a 2b)    e 

Kamil Walczak

 e (a 2b) e

(a 2b)

 e a

ei (  qa)  iqe

i (   qa)

 e (a 2b) e

(a 2b)

 ... 11 / 07 / 2017

Further Transformations ...  e  a

ei (  qa)  iqe i (  qa)

 e

a

 e  ( a  2 b )  iqa  iqa  e  ( a  2 b )  e   ( a  2 b )   e  iqe (a 2b)  (a 2b) (a 2b)   e   e  e  

ei (  qa) iqe

 e (a 2b)

i (  qa)

e

(a 2b)

 e

i (   qa)  e  e a  iqe iqa i (   qa)   iqe 

 e  a

ei (  qa)

 a

iqe i (  qa)

 e (a 2b) e   ( a  2 b )

ei (  qa)

ei (  qa)

iqe i (  qa)

 iqe i (  qa)

 iqe

iqa

ei (  qa) iqe

i (  qa)

 iqe iqa

ei (  qa)

 e (a 2b)

iqe i (  qa)

e   ( a  2 b )

i (   qa)   e   e  a  iqe iqa i (   qa)    iqe  

 e (a 2b)  e

 e (a 2b)   (a 2b)   e 

(a 2b)

 e

a

 e (a 2b)  e  ( a  2 b )

ei (  qa)

ei (  qa)

iqe i (  qa)

 iqe i (  qa)

 0  

…we can continue analysis by expanding determinants and keep reducing terms… Kamil Walczak

11 / 07 / 2017

Characteristic Function …or we can use “Mathematica” to solve the equation (*)…

 eiqa   iqe iqa M :  i (  qa)  e  iqe i (  qa) 

e iqa  iqe iqa ei (   qa)

 e a  e a  e (a 2b)

 iqe i (   qa)

 e  ( a  2 b )

 e  a    a e  ; (a 2b)  e  (a 2b)  e 

Solve[Det[M]  0, ] Using trigonometric identities to simplify expressions, we obtain the relation:

2  q 2 cosh2b  cos2qa   sinh 2b sin 2qa   cos  2q  1  f (E)  1 Kamil Walczak

(characteristic function) 11 / 07 / 2017

New Variables New representations as inputs in characteristic function:

2mEa 2 2b  2   2i x 2  2mEa 2 2mV0a 2 2qa  2    2i x   2 2     2mEa 2   2 x      2 2 2  q  x     2mEa  2mV0a 

  2mEa 2  2mV0a 2  q 2  x       2 2    x   2mEa 

New variables:

2mEa 2 x 2 2mV0 a 2  2

  b/a

2  q 2 1   q  1  x x           2q 2 q  2 x  x  Kamil Walczak

11 / 07 / 2017

Characteristic Function: Plot We can use “Mathematica” to define and plot characteristic function:

f [x _,  _,  _] : Cosh[2i x ]Cos[2i x   ]



 

1 x x     Sinh 2i x Sin 2i x   2  x  x 

  6,   16



Procedure in “Mathematica”:

Plot[f [x,6,16],{x,0,1.6}] f (x) allowed bands forbidden bands

x Kamil Walczak

11 / 07 / 2017

Bandstructure Theory of Solids   7,   6

Procedure in “Mathematica”:

Plot[f [x,7,6],{x,0,1.6}] f (x) allowed bands forbidden bands

x Bandstructure theory of solids models the behavior of electrons in solids via the existence of allowed and forbidden energy bands. A solid creates a large number of closely spaced energy levels (molecular orbitals) which are usually overlapped due to broadening effects experienced by them. Bandstructure theory of solids dictates optical and transport properties of solids. Kamil Walczak

11 / 07 / 2017

Formation of Bandstructures (Mesoscopic limit) Discrete electronic structure of atoms/molecules

~ 1020 atoms Periodic arrangement in energy favorable distance

Continuous electronic structure of solids/materials

Physisorption (≈ 0.01 ÷ 0.1 eV) Chemisorption (≈ 0.5 ÷ 0.6 eV)

E

Bonding controls electronic bandstructure of materials and the associated bandgaps!

rC rP Kamil Walczak

r 11 / 07 / 2017

Formation of Bandstructures Z6

Carbon (C) 2

[He] 2s 2p

2

Z  14

Closely-spaced energy levels

~ 1023 eV

Silicon (Si) [ Ne] 3s 2 3p 2 Z  32

Germanium (Ge) [Ar] 3d10 4s 2 4p 2

Splitting of the 2s and 2p states of carbon (diamond), the 3s and 3p states of bulk silicon or the 4s and 4p states of bulk germanium! Tipler, P. A.; Llewellyn, R. A. “Modern Physics”, 5th Ed. (2008), p. 443.

11 / 07 / 2017

Classification of Solids INSULATOR SEMICONDUCTOR

EF

CONDUCTOR

E G  2 eV

EF k BT  E G  2 eV

Metal

Semi-metal

Fermi energy is “hypothetical” energy level up to which all accessible energy levels below are occupied by electrons, while energy levels above are empty (at OK). Kamil Walczak

11 / 07 / 2017

Kronig-Penney Model: E-k Diagrams “Extended Zone Scheme”

“Reduced Zone Scheme”

 2k 2 E 2m eff

E

E

k 0

 L

2 L

3 L

k 

 L

0



 L

Looking at E-k diagrams, we conclude that in certain situations related to solids, “nearly free electron approximation” works quite well! Kamil Walczak

11 / 07 / 2017

Parabolic Band Approximation HH – Heavy Hole Band LH – Light Hole Band SOH – Split-Off Hole Band

Local approximation for energy:

 2k 2 E(k )  E 0  2m eff Definition of an effective mass:

  E(k )      2  k  2

m eff

Kamil Walczak

1

2

11 / 07 / 2017

Kronig-Penney Model: Dirac Comb Dirac delta function limit for potential barriers:

V0  ,

a  0,

V0a  const  

Dirac delta function limit for potential barriers:

2b  0



sinh2b  2b





cosh2b  1

2  q 2 q  2q 2

Under such conditions, the characteristic functions is defined as follows:

cos2qa  

 2qa sin 2qa   cos  2

 1  f (E)  1 Kamil Walczak

(characteristic function) 11 / 07 / 2017

Specific Solutions & Boundary Conditions The specific solutions in distinguished areas for positive energy E > 0:

1 (x)  A eiqx  B eiqx

q  2m(E  V0 ) /  2

2 (x)  C eikx  D eikx

k  2mE /  2

By applying boundary conditions from “slide 9” together with Bloch’s theorem, we obtain the system of four equations for four unknown coefficients:

A eiqa  B eiqa  C eika  D eika iqA eiqa  iqB eiqa  ikC eika  ikD eika

 iqA e



C eik (a 2b)  D eik (a 2b)  A eiqa  B eiqa ei ikC eik (a 2b)  ikD eik (a 2b) Kamil Walczak

iqa



 iqB eiqa ei 11 / 07 / 2017

Characteristic Function Our specific boundary conditions can be written in the form of homogeneous system of four linear equations for the wavefunction’s coefficients:

 eiqa  iqa iqe   ei (  qa)  i (  qa) iqe

e iqa  iqe iqa ei (   qa)

 eika  ike ika  eik ( a  2 b )

 iqe i (   qa)

 ike ik ( a  2 b )

 e ika  A   0      ika ike  B    0   e ik ( a  2 b )  C   0         ik ( a  2 b )  ike  D   0 

Non-trivial solution exists when matrix determinant is equal to zero, and from this condition we obtain the following relation:

2  q 2 cos2b  cos2qa   sin 2b sin 2qa   cos  2q  1  g(E)  1 Kamil Walczak

(characteristic function) 11 / 07 / 2017

New Variables New representations as inputs in characteristic function:

2mEa 2 2b  2  2 x 2  2mEa 2 2mV0a 2 2qa  2   2 x 2 2     2mEa 2   2 x     2 2 2  q x    2mEa  2mV0a 

 2mEa 2  2mV0a 2   2  q x     2 2    x   2mEa 

New variables:

2mEa 2 x 2 2mV0 a 2  2

  b/a

2  q 2 1   q  1  x x          2q 2 q  2 x x  Kamil Walczak

11 / 07 / 2017

Characteristic Function: Plot We can use “Mathematica” to define and plot characteristic function:

g[x _,  _,  _] : Cos[2 x ]Cos[2 x   ]



 

1 x x     Sin 2 x Sin 2 x   2  x x 

  5,   15



Procedure in “Mathematica”:

Plot[g[x,5,15],{x,0,2.1}] g( x ) allowed bands forbidden bands

x Kamil Walczak

11 / 07 / 2017

Ohm’s Physics (1827) The First Ohm’s Law: the electric current I flowing through two points of a conductor is equal to an applied voltage V (potential difference) between those two points:

IG V A

L The Second Ohm’s Law: the electric conductance G is proportional to the cross-sectional area A for the charge flow and inversely proportional to the length of the conductor L:

G

 Kamil Walczak

A L

Georg Ohm (1789-1854)

- material-dependent electrical conductivity 11 / 07 / 2017

Conducting Properties of Materials

Insulators: impede the free flow of electrons through the crystal structure. Injected charge usually remains in the initial location (no global charge redistribution is allowed). Semiconductors: thermal energy supports a relatively small amount of electrons (and holes) in their free motion within the sample. As temperature increases, the electrical resistance of the sample decreases. Conductors: contain a lot of electrons moving freely within the sample. Injected charge is quickly distributed across the surface of the object. The electrical resistance increases linearly with temperature. Superconductors: collective behavior of electron pairs which are in quantum state. Those low-temperature metals are characterized by “zero” resistance.

Kamil Walczak

11 / 07 / 2017

Classification of Bulk Materials Semiconductors

Insulators

Conductors

Electrical Conductivity (S/m)

108

1012 Perovskites Cuprates

10 4

Mercury Copper Graphene

100 Germanium

10 4 Boron Silicon

Diamond Air Rubber Glass

Teflon

1024 1020 1016 1012 10 8

The best electrical insulators are teflon and vacuum, while the best conductors are superconductors below critical temperature, graphene, silver, and copper. The most important semiconductor from technological point of view is silicone. Kamil Walczak

11 / 07 / 2017

Density Functional Theory (DFT) The Kohn-Sham single-electron equation:

 2    VKS ( r ) n ( x )  E n n ( x )   2m  The K-S potential is the sum of external, Hartree and exchange-correlation terms:

      E [n ] q n( r ' ) VKS ( r )  Vext ( r )  VH ( r )  VXC ( r )  Vext ( r )     d 3r '  XC | r  r '| n ( r )  

2

Since potential is a unique function of electron density, so is any other observable: N   2 n ( r )   | n ( r ) | n 1



with

 3 n (  r )d r  N



The total energy associated with electrons is a functional of electron density:

Eel [n]  TKS[n]  U H [n]  E XC [n] Problem: LDA-DFT calculations underestimate bandgaps typically by 30-50%! Kamil Walczak

11 / 07 / 2017

Self-Consistent DFT Scheme    E [n ] q n( r ' ) VKS ( r )  Vext ( r )     d 3r '  XC | r  r '| n ( r ) 2

Guess Initial Density

Create KS Potential

 2   out out    V ( r )  ( x )  E  KS n n (x)   n  2m 

Solve KS Equations N/2   n ( r )  2 | nout ( r ) |2 n 1

Obtain New Density

E tot NO Kamil Walczak

Converged?

YES

  3  TKS[n]   Vext ( r )n( r )d r

 U H [n]  E XC [n] 11 / 07 / 2017

Fixing Bandgap Problem: GWA-DFT Exchange-correlation term:

Experimental Bandgap [eV]

8

  VXC ( r )n ( r )  

   3   ( r , r ' )n ( r ' )d r '

6



4

Self-energy term:

      ( r , r ' )  iG ( r , r ' )W( r , r ' )

2 Screened Coulomb interaction: (red) LDA-DFT (blue) GWA-DFT

0 0

6 2 4 Theoretical Bandgap [eV]

Schilfgaarde et al., Phys. Rev. Lett. 96, 226402 (2006).

8

     VC ( r , r ' ' ) 3 W( r , r ' )     d r ' ' ( r ' ' , r ' ) 

Dyson equation for propagator:

G  G 0  G 0G 11 / 07 / 2017

E-k Diagram & DOS: Bulk Silicon

Kamil Walczak

11 / 07 / 2017

Bandstructure of Bulk Silicon: Details Temperature-dependent indirect bandgap (temperature in K):

4.37 104 T 2 E G  1.17  T  636 Six equivalent valleys at conduction bandedge:

Kamil Walczak

11 / 07 / 2017

Effective Mass / Bare Mass

Effective Mass Vs. Bandgap

0.2

0.1

0.0 0

1

2 Bandgap [eV]

3

Concluding Remark: in general, the larger the bandgap in the semiconductor, the higher the value of the effective mass associated with charge carriers! Kamil Walczak

11 / 07 / 2017

E-k Diagram & DOS: Bulk Copper

Kamil Walczak

11 / 07 / 2017

Spin-DOS: Copper, Cobalt, Iron









Kamil Walczak





11 / 07 / 2017

E-k Diagram & DOS: Graphene (2D)

Dirac cone

Kamil Walczak

11 / 07 / 2017

Hybridization & E-k Diagram: Graphene (2D)

sp2 hybridization

conduction

Relativistic Wave Equation:

valence

    F  k    E(k )       Dirac cone

Kamil Walczak

11 / 07 / 2017

Bandstructures of 2D Materials

Energy [eV]

Silicene

Germanene

10 8 6 4 2 0

2 4 6 8  10 

M K k-points

P. Miro et. al. Chem. Soc. Rev. 43, pp. 6537-6554 (2014).



M K k-points

 11 / 07 / 2017

Bandstructure Calculations Among the standard methods used to calculate bandstructure of solids are: Kronig-Penney Model with nearly free electron approximation, where interactions between electrons are completely neglected and Bloch’s theorem is applied. Density Functional Theory as a first-principles method addressing electron-electron interaction problem via exchange-correlation term in functional of electron density. Theory of Propagators in which many-particle effects (correlations) are included via self-consistent computation of Green’s functions satisfying Dyson equation. Tight-Binding Approach is based on the assumption that electrons in solid are highly localized near atomic sites and LCAO with Wannier wave functions is applicable. The “k dot p” Method which is based on perturbation theory formulated in terms of energies and wave functions exactly at the Gamma point (where k = 0). Kamil Walczak

11 / 07 / 2017

Closing Remarks Bandstructure theory is based on a few assumptions that system is homogeneous and impurity-free, macroscopically large (composed of 1023 atoms or more), while electrons do not interact dynamically with other electrons, lattice vibrations, etc. Electronic structures of solids show bands of allowed energies, being separated by zones of forbidden energies as a consequence of periodic structure of solids! Bandstructure of a solid dictates its optical and transport properties! Due to applicability of Bloch’s theorem, the unbound electrons in solids can be treated as nearly free particles which do not contribute to resistance (there is no scattering on periodic potential created by crystal ions)!

Impurities and interfaces modify bulk bandstructures (surface and dopant states, local charges and band bending), while conventional bandstructure theory is not applicable to strongly correlated systems (Mott insulators, Hubbard model)! Bandstructure theory is only an approximation to the quantum state of a solid! Kamil Walczak

11 / 07 / 2017