Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

download Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

of 108

Transcript of Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    1/108

    Semiconductor and SuperconductorPhysics

    Erkki Lahderanta

    Lectures follow the books:

    Puolijohdeteknologian perusteet (Sinkkonen) and

    A. L. Rose-Innes and E. H. Rhoderick: Introduction to Superconductivity

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    2/108

    Contents

    1 Classical conductor 3

    1.1 Electrical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

    1.1.1 Conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

    1.1.2 Hall effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.1.3 Frequency dependence of conductivity . . . . . . . . . . . . . . 8

    1.2 Optical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

    2 Quantum mechanics 15

    2.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

    2.2 Schrodinger equation (1926) . . . . . . . . . . . . . . . . . . . . . . . . 20

    2.3 Classical limit (macroscopic particles) . . . . . . . . . . . . . . . . . . . 21

    2.4 Solution of Schrodinger Equation . . . . . . . . . . . . . . . . . . . . . 25

    3 Structure of material 29

    3.1 Atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

    3.2 Atomic bonds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

    3.3 Solid material and crystal structure . . . . . . . . . . . . . . . . . . . . 31

    3.4 Lattice vibrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

    4 Free-electron model of metals 34

    4.1 Free electron gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

    4.2 Thermal emission current . . . . . . . . . . . . . . . . . . . . . . . . . 404.3 Field emission current . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

    5 Energy bands 45

    5.1 Formation of bands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

    5.1.1 Periodical potential . . . . . . . . . . . . . . . . . . . . . . . . . 48

    5.2 Models for free energy bands . . . . . . . . . . . . . . . . . . . . . . . . 52

    5.2.1 Free electron model . . . . . . . . . . . . . . . . . . . . . . . . . 52

    5.2.2 Kronig-Penney-model (1930) . . . . . . . . . . . . . . . . . . . . 54

    5.3 Electrons movement in the energy band . . . . . . . . . . . . . . . . . . 57

    1

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    3/108

    5.3.1 Semiclassical equation of motion . . . . . . . . . . . . . . . . . . 57

    5.3.2 Electrical conductivity . . . . . . . . . . . . . . . . . . . . . . . 63

    5.3.3 Charge carriers in semiconductor: Holes . . . . . . . . . . . . . 64

    5.4 Energy bands for various materials . . . . . . . . . . . . . . . . . . . . 665.4.1 Metal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

    5.4.2 Semiconductors and insulators . . . . . . . . . . . . . . . . . . . 67

    5.5 Interaction between electron and wave . . . . . . . . . . . . . . . . . . 68

    5.5.1 Disturbation theory . . . . . . . . . . . . . . . . . . . . . . . . . 68

    5.5.2 Optical properties of semiconductor . . . . . . . . . . . . . . . . 71

    6 Semiconductor physics 74

    6.1 Density of charge carriers . . . . . . . . . . . . . . . . . . . . . . . . . . 74

    6.1.1 Density of states in energy band . . . . . . . . . . . . . . . . . . 746.1.2 Electron and hole densities . . . . . . . . . . . . . . . . . . . . . 76

    6.1.3 Doped semiconductor . . . . . . . . . . . . . . . . . . . . . . . . 78

    6.1.4 Charge carrier density of doped semiconductor . . . . . . . . . . 80

    6.2 Mobility, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

    6.3 Generation and recombination of charge carriers . . . . . . . . . . . . . 85

    6.4 Diffusion current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

    6.5 Current caused by temperature gradient . . . . . . . . . . . . . . . . . 90

    6.5.1 Current density . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

    6.5.2 Density of heat flow rate (lampovirrantiheys) . . . . . . . . . . . 91

    7 Preparation of semiconducting materials 93

    7.1 Crystal growing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

    7.2 Selective doping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

    7.3 Diffusion and ion-implantation . . . . . . . . . . . . . . . . . . . . . . . 95

    7.4 Oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

    7.5 CVD-growing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

    7.6 Metal growing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

    7.7 MBE-method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

    8 Spintronics 102

    8.1 Magnetic semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . 102

    8.2 CMR and GMR phenomenon . . . . . . . . . . . . . . . . . . . . . . . 103

    8.3 Multilayer structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

    8.4 Spintronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

    2

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    4/108

    Chapter 1

    Classical conductor

    Electrons in metal. Free electron gas and positive metal ions.Current carrier react on electromagnetic field ELECTRICAL and OPTICAL proper-ties.

    Density of atoms 5 1022 atoms/cm3 dense medium.Medium Crystal lattice periodic potential modify properties of current carriersand gives energy bands.

    Temperature Thermal vibrations.

    1.1 Electrical properties1.1.1 Conductivity

    Figure 1.1

    E= 0motion + collision + motion + collision + . . .Short motion is linear and to arbitrary direction Brownian motion.

    Thermodynamic stabilitystatistical methods, specially AVERAGE.Velocity distribution of a classic particle: Maxwell-Boltzmann distribution

    f(vx, vy, vz) =Nexp

    EkinkT

    = Nexp

    12

    m(v2x+v2y+ v

    2z)

    kBT

    (1.1)

    3

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    5/108

    N = normalization constant;

    vx, vy, vz = velocity inx, y and z direction;

    m= mass of particle;

    kB = Boltzmann constant = 1.380 622 1023 J K1 = 8.6173 105 eV K1.

    IfT= 300 K kBT = 0.0258 eV (small).(1.1) is symmetricvelocity average = 0 particle drift and electric current = 0.Thermodynamics and statistical physicsAverage kinetic energy per direction

    12

    mv2 always positive

    = Exkin = Eykin = Ezkin =1

    2 kBT

    3 dimensionEkin = 32 kBTmost commonspeed = thermal speed vT.

    1

    2 mv2T =

    3

    2 kBT vT =

    3 kBT

    m (1.2)

    Numerical values m = 9.1 1031 kg

    T= 293 K vT 107cm

    s

    = 105m

    s

    = 100km

    s

    (1mm in 10ns)

    Electric field E= 0 F = qEEquation of motion

    F =ma mdvdt

    = qE (1.3)

    E= constant v(t) =v(t0) qm

    E(t t0) (1.4)t0= initial moment and v(t0) = initial velocity.

    Speed depends linearly on Eand on (t t0).Ift0= previous collision, then (1.4) is working up to next collision. For every electron

    we have own (1.4) and own t0. Average of (1.4) over collision times t gives

    v

    v (1.4)

    = qm

    E

    t t0

    = qm

    E vD (1.5)

    =

    t t0

    = average time between collision = relaxation time.

    In (1.5) is taken into account v(t0) = random

    v(t0)

    = 0.

    Average velocity vD, proportional to electrical field E

    vD = drift velocity.

    4

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    6/108

    ELECTRIC CURRENT

    Figure 1.2

    We investigate time period electrons move distance x= vDElectron density n across the area Ais moving amount of electric charge

    Q= qnxA Electric current density J

    J= I

    A=

    Q/

    A =

    qnxA

    A =

    qnvD

    (1.5)=

    q2 n

    m E E (1.6)

    Here is definition of conductivity

    q2

    m n (1.7)

    This gives resistivity

    1 (1.8)Ohmic law (1.6): Electric current density is directly proportional to electric field

    strength

    Proportionality constant depends on

    n= density of current carriers and = relaxation time (= collision frequency)

    Numerical values. Metal n amount of atoms 5 1022 electrons/cm3 106 1 cm1

    (1.7) 1013 s.Free path, l = free distance = how long is average distance between the collisions.

    Thermal speed =vT = 105 m s1 l= vT = 105 m s1 1013 s = 10 nm (293 K).

    Much less than size of component.

    MOBILITY,

    Average velocity (1.5) vD can be written

    |vD

    |=E (1.9)

    5

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    7/108

    (1.5) gives

    =q

    m (1.10)

    which is MOBILITY of current carrier.Numerical values: 1013 s mobility 200 cm2 V1 s1.Average velocity, vD, is much less than thermal velocity, vT.

    IfE= 10 V cm1 = 1kV m1 vD = 20 m s1, but vT 105 m s1.Now we try to take vD from equation of motion.

    ma= F, i.e. m dvdt

    = qEcould give vD, but collisions??Drude model: we take collision into account with additional friction term.Drude model. Collision (i.e. scattering) is limiting F = qEto times with t < Similar result if we add a term which is similar to friction, i.e. term proportional to

    velocity,vDma= F Ffriction mdvD

    dt = qE vD (1.11)

    Friction-term taken from physics of fluid dynamics (viscosity). In steady state (tbig)

    dvDdt

    = 0 (1.5) = m

    (1.12)

    Solution of Eq. (1.11) ifEis switched ON when t= 0

    vD(t) =qE

    m 1 expt (1.13)

    Figure 1.3

    Now we see why is called as relaxation time:

    After a change (e.g. field ON or OFF), the velocityvD becomes stable after one collision,

    that is after t .Equation (1.11) is called Hydrodynamic model, or Drude model. Charges behave like

    particles in liquidresult is obtained from average and viscosity.

    6

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    8/108

    1.1.2 Hall effect

    =(n, ), meaning depends on mobility, , and electron density, n.

    How to measure n? Answer: B.

    Lorentz-force

    F = qv B Fv and FB (1.14)Total force F = qE+ qv BIn material appears small voltage perpendicular () to direction of current Fvand magnetic field

    FB.

    Figure 1.4

    This voltage is called Hall-voltage.

    Drude model m dvD

    dt

    (1.11)(1.14)= m

    vD q(E+vD B) (1.15)

    Steady state situation dvDdt

    = 0 and current densityJ (1.6)

    = qvDn

    E= 1

    J RHJ B (1.16)

    where

    RH= 1

    nq = Hall-constant (1.17)

    Eq. (1.16) has 2 parts

    1. Usual ohmic voltage parallel to current

    EL= 1

    J=

    vD

    (1.18)

    2. EH= field perpendicular to current:

    EH= RHJ B = vD B (1.19)

    7

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    9/108

    In steady state this EH compensates the Lorentz-ForceHall-voltage=

    UH=EH H= 1qn

    I BW

    (1.20)

    In measurements longitudinal voltage gives

    UL= EL L= L I

    W H (1.21)

    In measurements perpendicular voltage, Hall-voltage, givesn, density of current carries.

    These together gives mobility, .

    Ratio ofEH and EL (1.18) and (1.19)

    EHEL

    =B (1.22)

    Numerical values. IfB = 1 T = 1 V s m2 and = 600 cm2 V1 s1 EHEL

    0.06.

    1.1.3 Frequency dependence of conductivity

    Electron collision to positive atoms different time windows give different results.Apply electric field

    E=E0exp(it) (1.23)

    E0= amplitude

    = angular frequency (2 = )

    this gives average velocity, drift velocity, from (1.4) and conductivity

    vD=q

    m

    1

    1 i

    E (1.24)

    () = DC1 i (1.25)

    where DC is given in (1.7).

    AC-behaviour by replacing 1i in DC-behaviour

    8

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    10/108

    (1.25) in 2 parts

    Re[()] =DC/(1 +22)

    Im[()] = DC/(1 +2

    2

    )

    (1.26)

    Figure 1.5

    Low frequency

    < 1

    = DCHigh frequency ( 1) Re 0 like 1

    2.

    Im-part grows linearily with , when 1

    , maximum at = 1

    .

    Im> Re when > 1

    wire behaves like impedance. Maxwell-equation

    Jtot = J+ E

    t (1.27)

    J= usual current density with J=() E

    E

    t = dieletric displacement current (siirrosvirta)

    (1.25) means that resistance and inductance are in series

    R=

    DC A (1.28)

    L=

    DC A = m

    q2n

    A (1.29)

    A= cross-section; = length.

    Second term in Maxwell equation (1.27) is connected to capacitance, so called geomet-

    rical capacitancein parallel

    C=A

    (1.30)

    Conductor, wire, is LC-circuit with losses

    9

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    11/108

    Figure 1.6

    Geometrical capacitance is charged when voltage is applied.

    Electrons behave in inductive way. Resistance from collision (t ).Resonance frequency ofLC-circuit =

    p = 1

    LC=

    q2n

    m (1.31)

    Name: PLASMA FREQUENCY

    1.2 Optical properties

    Real part of conductivity meanslosses and damping of the wave =() Differentfrequencies give different reflection and propagation

    Maxwell E= Bt

    (1.32)

    1

    B =J+ Et

    (1.33)

    = permittivity;

    = permeability.

    Ohm law

    J=E (1.34)

    If additional chargeElectric field

    E= (1.35)

    where = charge density.

    4 equationsElectromagnetic fields.Wave equation. Start with neutral and homogenous conductor (= 0)

    Calculation2E= E

    t +

    2E

    t2 (1.38)

    Solution is

    E=E0exp[i(k r t)], which is a wave (1.39)

    10

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    12/108

    k r t = phase angle (r, t), E0= amplitude, k= wave vector.Phase angleis constant ifrk, meaning in direction perpendicular to propagation ofthe wave.

    In direction of wave propagation (rk) phase angle = const if

    r=

    k t r0 (t= time)

    Interpretation: In direction of propagation the phase moves with velocity, so called

    phase velocity (vaihenopeus) vp.

    Figure 1.7

    vp=r

    t =

    k (1.40)

    During distance of wave length , phase angle change by 2 = 2 r k+ (t)

    =0= 2 k= 2

    k=2

    (1.41)

    i.e. lenght of wave vector is connected to

    Question: When (1.39) is solution of Maxwell equations?

    (1.32) B =B 0exp[i(k r t)] (1.42)with amplitudeB 0 =

    1

    k E0 (1.43)

    In other words: Magnetic field component in plane wave is perpendicular to Eandk.

    Moreover, because charge density = 0 (1.35) E= 0 and therefore also

    k E= 0 (1.44)

    In electromagnetic wave B and Earein comparison to propagationExist 2 directions of polarization.

    Wave equation (1.38) requires (calculations)

    k2 =i( i) (1.45)

    11

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    13/108

    Figure 1.8

    This is called dispersion relation and it gives k= k().

    Dispersion = wave beam splits to wave components.

    Tangential components(direction of surface) is continuous

    incident angle = angle of reflection; =

    snell lawk sin = k

    sin

    (1.46)

    ifk

    =k

    () wave beam will split to components.

    Figure 1.9

    INSULATOR. = 0, and are real constants (non-imaginary)

    (1.45) gives k=

    1

    v

    (1.47)

    where v 1

    = velocity (1.48)

    v usually is written as

    v= c

    n (1.49)

    with c= velocity in vacuum; n= index of refraction (how thick optically).

    n=

    00=

    c

    v (1.50)

    12

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    14/108

    (1.47) can be written also in the form

    k= n

    c (1.51)

    Snell law (1.46) can be written in the form

    n sin = n

    sin

    (1.52)

    CONDUCTOR. = const, real number

    k (1.45)

    = 1

    v

    1 +

    i

    1/2(1.53)

    Wave numberk: Imaginary part gives attenuation, damping (k2 =2 +i).

    Low frequency

    Imk=

    1

    21

    (1.54)

    whereis called skin depth (Finnish languagepenetration depth)High frequency is complex, similar to (1.25)

    Figure 1.10

    If 1 Eq. (1.53) (is the time between collision).If 1 Im ReConductor behave like insulator and wave numberbecomes

    k=1

    v

    1

    p

    21/2(1.55)

    wherep is plasma frequency (1.31).

    If > p k is real (non-imaginary) and NO DAMPING of the wave.

    Interpretation: Free electrons cannot follow the electromagnetic fieldsX-rayspenetrate the material.

    13

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    15/108

    In metals p 1015 s1 (2000 nm?).

    Reflectance. From air to material with index of refraction = n electric field components

    E0 = incoming, E1= reflected, E2 = refracted. Tangential components continuous

    E0+E1 = E2 (1.56)

    Figure 1.11

    Magnetic field components in similar way B0+B1 = B2

    (1.43) B0 = 1

    k E0

    (1.50) k= n

    c

    B0 proportional to n E0 (1.56)=E0 E1 = nE2 (1.57)

    calculations= E1E0 = 1n1+n and E2E0 = 2n+1

    Coefficient of reflection =

    R =intensity of reflecting waveintensity of incoming wave

    =v0|E1|2v0|E0|2 =

    1 n1 +n

    2(1.58)

    Coefficient of trasmission =

    T =

    intensity of trasmission

    intensity of incoming wave

    =v0n

    n2|E2|2

    v0|E0|2 =

    4n

    (1 +n)2(1.59)

    (v0= velocity of incoming wave, in vacuum).

    More high frequency X-rays ( 0.1 nm) Macroscopic quantities, such asdielectricity constant, not enough.

    Reflection from atoms, not from surfaceBragg law

    2d sin = integer (1.60)

    14

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    16/108

    Chapter 2

    Quantum mechanics

    First experimental results, which were not possible to explain classical physics.

    Figure 2.1

    2.1 History

    A. Black body radiation Classical model: Energy densisty of radiationE 2.Experimentally 2 dependence only with small , with high frequency intensity

    0, when increasesClassical model: Radiation energyE is continuous. In temperatureT average

    15

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    17/108

    energy for waveform=

    E =

    0 dE E exp(E/kt)

    0 dEexp(E/kt) =kBT = const (2.1)

    Density of waveform 2 Energy density 2.PLANCK(1901): Energy exchange between electromagnetic wave and material

    happens as portions, quantas, not continuous and size of energy quanta .Meaning, energy of generated irradiation has form

    E=n (2.2)

    with n= integer and

    = constant.In this case average will be discrete (

    0 )

    E= n=0 n exp(n/kt)n=0exp(n/kt)

    =

    exp

    kBT

    1

    (2.3)

    Energy quantization explained spectrum of radiation. = Planck constant = h2

    .

    Irradiation contain photons and each photon has energy = Amount ofphotons in a wave =

    n = 1exp

    kBT

    1

    (2.4)

    This has a name Bose-Einstein distribution and it describe behaviour of Bosons

    (spin is integer number. Spin halfFermion).

    Figure 2.2: Experimental arrangement and explanation

    16

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    18/108

    B. Photoelectric effect No current until > 0 = threshold frequency. Light

    intensity influence only on size of current.

    Threshold frequency not possible to understand with classical model.

    Planck idea about quantization of energy in lightEinstein explained thresholdfrequency (1905):

    When frequency is enough, only then energy is able to lift electron to height

    ofW0

    0= workfunction.

    turned out to be the same constant as in black body radiation. Energy that

    remain after 0, will be kinetic energy

    W0 = 12

    mv2 (2.5)

    C. Specific heat of solid state materials (phonon) Classical physics: Atom vi-

    brating around equilibrium positionEnergy of Harmonic oscillator

    E=12

    M v2x+1

    2x2 (2.6)

    M= mass, = string constant.

    If classicalMaxwell-Boltzmann distribution gives energy averageE =kBT (2.7)

    This is xdirection. 3-dimensional E= 3 kBTand final resultSpecific heat =cv =

    ET

    = 3 kB per particle (2.8)

    In high temperatures OK, in low temperatures experiments show that cv 0Einstein explained (1906) low temperature behaviour: Energy for vibrating atom

    is also quantized:

    E=n (2.9)Quantum is called PHONON

    D. Bohr model (1913) Classical mechanics: Particle energy is continuous, therefore

    infinite amount of orbits.

    Experiments: Emission and absorption spectrum not continuous, sharp lines.

    17

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    19/108

    Figure 2.3

    Bohr (1913): Angular momentum is quantized

    L= n (2.11)

    Only fixed orbits are accepted.Circular motion: v=r p= mv = mr L= pr = mr2

    mr2 =n (2.12)

    Additionally F =ma

    14E0

    q2r2

    =m2r (2.13)

    Possible orbits

    En= e4m

    2(4E02)2 1

    n2 (2.14)

    rn=4E02

    e2m n2 (2.15)

    Valid for atoms with one electron (hydrogen-like). n= quantum number.

    E. Compton scattering (Phys. Rev. 22, 409 (1923))

    Experiment: X-rays into electrons.

    Interaction between photon and electron is like collision of two particles. Also

    conservation of momentum.

    Classical physics: light has energy (but no particles and therefore no momentum)

    Quantum mechanics: light consists of photons and photons have momentum.

    18

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    20/108

    For photon, with m0 = 0 and velocity =c, we haveE=pc andE=

    p=

    c =

    h

    with h= 2 (2.16)

    Photoelectric effect: All photon energy to electron, photon disappears.

    Compton scattering: Part of the photon energy to electron, amount of photons

    remain constant.

    F. Wave nature of electron (de Broglie, PhD-thesis 1924)

    Photon has wave nature and also particle behaviour.

    de Broglie: Also particle have both behaviours, including wave nature. Photon

    p= h , de Broglie: this is valid also for particles

    =h

    p =

    h

    mv (2.17)

    Also energy similar to photonE=

    =E

    (2.18)

    Additional result: Simple interpretation of Bohr quantization rule: (2.17) and

    (2.18)Length of the orbit must be

    2r = n (2.19)

    Figure 2.4

    Question: de Broglie is theoretical idea. What about experiments?

    Answer: 1927 Yes, Electron beam to crystal gave scattering according to Bragg

    law

    2d sin = n (2.20)

    19

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    21/108

    2.2 Schrodinger equation (1926)

    On atomic level the measuring accuracy is not enough for usualmeasuring.

    Tunneling microscope: r is ok, but tis not ok.

    Fast lasers (attolaser,

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    22/108

    2.3 Classical limit (macroscopic particles)

    Quantum mechanics is for atom-size objects. For macroscopic particles the results must

    be similar to classical mechanics. We describe macroscopic particles with wavefunction:

    First we think about free microscopic particle. 1D Schrodinger gives wavefunction with

    normalization:

    (x, t) = 1

    Lexp

    i

    kx E(k)t

    (2.33)

    where k = wavenumber, L = size of volume (1D) and has been replaced by E(k)

    ,

    where

    E(k) = energy = 2k2

    2m (2.34)

    Example:

    Probability distribution = ||2 = = 1L

    = const.

    Linear combination of (2.33)

    (x, t) =

    dk a(k)exp

    i

    kx E(k)t

    (2.35)

    is also solution of Schrodinger. We call this wave packet.

    What isa(k)?

    It is simply multiplication function. Wave packet (2.35) is solution of Schrodinger and

    it describes a particle with all a(k)functions.We can describe classical particle by choosing

    a(k) =

    1, when

    k0 k2

    < k V0, now V1< V0).

    Intermediate result

    =k tankL2 (2.81)

    or

    = k cotkL

    2 (2.82)

    where definitionsk=

    2m2

    E and =

    2m2

    (V0 E) (see (2.77)).V0 big kL2 = n2 (standing wave), meaning

    k=n

    L (n N) (2.83)

    Final result: Energy levels

    En= 22n2

    2mL2 (n N) (2.84)

    These are deep well energy levels (V0 very high)

    IfV0< Energy levels little below (2.84).

    28

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    30/108

    Chapter 3

    Structure of material

    3.1 Atom

    Bohr model is rather simple. More detailed with Schrodinger equation. Hydrogen like

    atom possible to describe. Schrodinger equation

    2

    2m2 e

    2

    40r

    (r) = E(r) (3.1)

    gives solution

    (r) = nlml(r) =Rnl(r)Ymll (, ) (3.2)

    In spherical coordinates (r) separates in two parts: one part depending on distance

    and one part depending on direction. Variables for spherical coordinates r, and .

    Figure 3.1

    Electron energy in hydrogen atom in state n:

    En= e2m

    2(4E0)2 1

    n2 = 13.6 eV

    n2 (3.3)

    29

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    31/108

    In solution(3.2)indicesn, l,ml are called quantum numbers, they characterize different

    solutions.

    For hydrogen the energy depends only on main quantum numbern. In more complicated

    atoms the electron energy depends also on l and ml

    l= orbital quantum number = 0, 1, 2, . . . , n 1 (3.4)

    describes angular momentum of electron.

    ml = magnetic quantum number = 0, 1, 2, . . . , l (3.5)

    describeszcomponent of angular momentum of electron.Angular momentumL = p

    ract as operator and spherical harmonic functionYml

    l

    (, )

    fulfill:

    L2Ymll = l(l+ 1)Ymll (3.6)

    LzYmll = mlY

    mll (3.7)

    Here is more clearly seen that l is connected to magnitude ofL andml is connected to

    zcomponent ofL.Question:Where iszaxis of an atom?

    Answer: Need external interaction, meaning a magnetic field zaxis is definited indirection ofBName for ml = magnetic quantum number.More heavy atomsMany particle problem and correction terms. Energy dependsalso on l E= EnlCorrection terms: In Schrodinger equation is second derivate. Not exact solution, only

    approximations. In magnetic field energy depends also on ml (Zeeman spliting).

    Names and spectroscopic notations

    Occupation of electron shell nlx

    x= amount of electrons in subshell nl

    n= main quantum number, n= 1, 2, 3...

    Orbital quantum number notation with letter

    l= 0 s-statel= 1 p-statel= 2 d-statel= 3

    f-state

    30

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    32/108

    Example 1.

    1s2 describes electron shell of He

    Example 2.

    Every shell with own notation. Larger atommore shells 1s2

    2s2

    2p6

    is spectroscopicnotation of electron shell for Ne.

    3.2 Atomic bonds

    A. Ionic bond. Coulomb interaction.

    B. Covalent bond. Neighbouring atoms have common electrons.

    C. Metallic bond. Electron gas, electron sea, about 1 electron from each atom.8-N-rule.

    3.3 Solid material and crystal structure

    Lattice points are similar: Each of them have similar environment. Lattice vector:

    R= n1a1+n2a2+n3a3 (3.8)

    ai = basis of lattice (i= 1, 2, 3) and ni are integers.Exist 14 different lattice structures, for example 3 different cubic structures

    Miller indices. Macroscopically crystal looks like planes. Mathematical name for

    different planes = Miller indices. Described in lectures.

    Direction corresponding to plane: Similar numerical values for Miller indices, but

    different kind of brackets; (nkl)= plane and [nkl] direction(nkl).

    31

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    33/108

    3.4 Lattice vibrations

    Figure 3.2

    Average total energy per atom = 3kBT(3 dimensions). Vibration of atoms is connected

    to each other, modelling with ideas from springs, collective behaviour. Nearest neighbour

    interactionequation of motion for atom i (ma= F)

    M d2ui

    dt2 =[(ui+1 ui) (ui ui1)] =[ui+1+ui1 2ui] (3.9)

    M= mass of atom i

    ui= deviation from stable position

    = elasticconstant = stringconstant

    Figure 3.3

    Differential equation (3.9) has solution as plane wave

    un=A exp[i(kna t)] (3.10)

    Inserting (3.10) into (3.9)It is required

    =

    M2sin

    ka

    2

    (3.11)

    This has a namedispersion relation, frequency as function ofk. With long wave lengths

    32

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    34/108

    Figure 3.4

    (3.10) is elastic wave, meaning sound wave. long k small. When k 0 we have(3.11) in form

    = M

    2sinka2 ka

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    35/108

    Chapter 4

    Free-electron model of metals

    In this chapter will be combined Fermi-Dirac distribution with Schrodinger equation.Results will support earlier results of classical conductor.

    Figure 4.1

    4.1 Free electron gas

    Electrons are completely free inside the metal. No interactions. Infinite potential-energy

    barriers in edge of the metal.Schrodinger equation has solution as plane wave

    = 1

    Vexp(i k r) (4.1)

    whereV= volume of the metal. Putting (4.1)to Schrodinger equation will give equation

    betweenE and k:E(h) =

    2k2

    2m (4.2)

    HereE is kinetic energy.

    34

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    36/108

    Figure 4.2

    Figure 4.3

    Size of the metalL L L, withL macroscopic. Periodic boundary condition: wave

    function similar in opposite edges:

    (x= 0, y , z ) = (x= L, y,z) kxL= 2 n (n N)

    (In reality = 0 in the edges because infinite energy barrier in the edge. This is difficult

    to calculate and it is enough to have periodic boundary condition).

    Periodic boundary condition in 3D

    k=2

    L nxi +nyj+nzk

    (4.3)

    Difference between neighbouring k-values = 2Lonek-vector, meaning one STATE,

    occupies in k-space a volume

    Vk=

    2

    L

    3=

    (2)3

    V (4.4)

    Density of states is constant in kspace.Density of state g(E) (How many places for electrons?) Amount of states, dN,

    35

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    37/108

    Figure 4.4

    between energy valuesE E+ dEper unit volume:

    dN=g(E) dE= 2V

    VkVk

    (4.4)=

    1

    2k2 dk

    (4.2)=

    1

    22

    2m

    h2

    3/2 EdE (4.5)

    In calculations in more details.

    Vk = volume in k-space betweenE andE+dE.Numerical values g(E) 1022 states/(eVcm3).(4.5)

    g(

    E) = 1

    22 2m2 3/2

    E

    (density of states for free electron)

    Figure 4.5

    Fermi-Dirac distribution

    Maxwell-Boltzmann distribution: Classical, particle density small, individual particle

    possible to indentify.

    Pauli exclusion principle

    Electrons are identical (not possible to identify)

    36

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    38/108

    In state with energy = E, the amount of electrons is on average

    f(E) 1expEEF

    kBT + 1

    (4.6)

    This is the possibility that state, with energy = E, is occupied by an electron.Fermi-energyEF

    1. Parameter. Strong connection with particle density. Depends on temperature.

    2. In zero temperature electron exist up toEF. Electrons take lowest energy.

    3. New electron appearing in the systemIt has energy EF.Analysing the distribution

    A. States with low energy are all occupied

    E EF EF E>> kBT f(E) = 1 = 100%

    B. States with high energy are occupied according to Maxwell-Boltzmann

    E EF EEIkT = EkT f(E) = exp(E/kT)| exp(. . .) 1 exp(. . .) + 1exp(. . .).

    Fermi-Dirac distribution changes strongly only nearEF:Practical estimation for width is

    4kBT.

    In metal n 1022 el/cm3 and density of states 1022 states/(eVcm3) EF feweVbutkT 0.02 Also in room temperature approximation

    f(E) =1, whenE< EF0, whenE> EF (4.7)

    Figure 4.6

    Approximation(4.7)possible to calculate density of electrons, n, as a function ofenergyE

    n=

    0

    g(E)f(E) dE (4.5)= 122

    2m

    2

    3/2 0

    Ef(E) dE (4.7)=

    = 122

    2m23/2 EF

    0

    EdE integrate= 132

    2mh2

    EF3/2 (4.8)37

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    39/108

    Information about where is Fermi-level, meaning we get Fermi-energy EF dependenceon electron density n

    EF = 2

    2m(32n)2/3 (4.9)

    States are occupied up toEF.EF in other unit:

    - Fermi wave vector, meaning radius os Fermi-sphere

    kF = (32n)1/3 (4.10)

    - Electron wave length on the Fermi-surface

    F =2kF

    = 2 3n1/3 (4.11)

    - Fermi velocity, meaning electron velocity on Fermi-surface

    vF = kF

    m (4.12)

    Numerical values:

    vF 10classical thermal velocity and F 2distance between electrons.Example: Specific heat

    C=U

    T (4.13)

    Classical U=3

    2kBT and Cclassical =

    3

    2kB per electron (4.14)

    Experimentally electron contribution smaller.

    Explanation: Fermi-Dirac distribution Only those electrons that are nearEF canabsorb energy. Specific heat due to these electrons only. Width of the area where

    changes 4kBT Classical result must be scaled by 4kBTEF

    Cel 32

    4kBTEF kB (4.15)

    38

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    40/108

    kBT EF Cel Cclassical, meaning specific heat much less than classical prediction.

    Figure 4.7

    39

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    41/108

    4.2 Thermal emission current

    Figure 4.8

    Electrons in potential well. Zero level for potential energy: Electrons free from the well.

    Photoelectric effect: Fermi level distance from zero level is =q =W= Workfunction.

    Distribution of electron velocity

    Electron has state (k) and momentum p= hk, meaning velocity =

    v= p

    m

    k

    m (4.16)

    One state occupy in k-space a volume (Eq. (4.4)) Vk = (2)3

    V Amount of states dN

    per unit volume with wave vector in k-volume d3k

    dN= 2 d3k

    Vk/V = 2

    d3k

    (2)3/V/V =

    1

    43d3k (4.17)

    Multiplication 2 from spin.

    Fermi-Dirac distributionf(E) density dn of electrons with wave vector in k-volumed3k

    dn= f(E) dN (4.17)

    =

    1

    431

    exp EEFkT+ 1 d3k (4.18)

    v= m

    k d3v= m

    3d3k and kinetic energyE= 2k2

    2m = 1

    2mv2

    dn= 1

    43

    m

    3 d3vexp

    12mv

    2EFkBT

    )

    + 1(4.19)

    This is so called thermal emission = current of electrons from metal to vacuum = out

    from the potential well.

    40

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    42/108

    Electrons needE= 12

    mv2 > EF+q Calculation of current density

    J=

    dn=

    [Eq. (4.19)] dvxdvydvz

    Integral from 12

    mv2x= EF+q, meaning integral from = vx =

    2(EF+q)m

    .

    Additionally approximationq kT Maxwell-Boltzmann Final result for currentdensity (thermal)

    J=C T2 exp

    qkBT

    (4.24)

    with

    C= Richardson constant =qmk2B223

    = 120A cm2 K2 (4.25)

    More clear presentation for current density (4.24) with n= density of electrons

    J= qveffn expq

    kBT.

    (4.26)

    This equation has more clear interpretation:

    Probability isexpqkBT

    that electron has enough big energy, q. These electrons move

    with effective velocity

    veff=3

    8 kBT

    EF

    2

    vF (4.27)

    where Fermi velocity is decreased by a factor

    kBTEF2

    . Probability expq

    kBT

    is very

    small in room temperatureTo have current it is needed heating.Extreme example wolframq =4.5 eV Needs very high temperature On the otherhand it can stand high temperature (light bulb).

    Numerical values for wolfram: T= 2500K J= 3 mA cm2.Quantum mechanical correction

    Part of electrons are reflected even if enough energy. Current density even more small.

    41

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    43/108

    Figure 4.9

    4.3 Field emission current

    TEMPERATUREElectric field instead.

    (a) (b)

    Figure 4.10

    In chapter 4.2 electrons go out from metal with help of temperature. Now we see how

    electrons go out form metal with help of electric field.

    Electric field E >0 toxdirectionElectron has potential energy

    V = qEx (4.29)

    Potential barrier with heightEF+q.In quantum mechanics barrier replaced by but mathematics is similar Tunneling

    42

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    44/108

    probability (2.79), meaning current, meaning field emission current

    T(E) =exp(2L) = exp22m

    2 1/2 (EF+q E)3/2

    qE (4.30)whereE= kinetic energy.Remarks:

    1 Tunneling probability increases whenE increases.

    2 Tunneling only from those states that are occupied

    2a Occupied states exist up toEF

    2b AboveEFstates are occupied with probability exp EkBT= smallTunneling mainly form Fermi-level E= EF Current is

    J T(E) exp2

    2m

    2

    1/2(q)3/2

    qE

    (4.20)

    Does not depend on temperature.

    Question: When current flows?

    Answer: Compare Eq. (4.20)to Eq. (4.26)where probability isexp qkBT Tunnelingcurrent if

    2

    2m

    2

    1/2 (q)3/2qE

    >

    qkBT

    (4.21)

    Need electric field

    E >kBT

    q 2

    2mq

    2

    1/2(4.22)

    Numerical values: Ifq =4.0 eV and T = 300KE >5 106 V cm1 (Comparablewith electric break down 3 MV m1 = 0.03 106 V cm1)

    Practical example: Tunneling microscope. G. Binning and H. Rohrer (1982, Nobel1986).

    (4.30) J exp(L)

    Resolution

    Vertical direction 0.1 AHorizontal direction 1 A

    Tunneling, feed back circuit, scanning the sample surface. STM and AFM. Feedback

    voltage is the measuring result.

    43

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    45/108

    Figure 4.11

    44

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    46/108

    Chapter 5

    Energy bands

    Up to now two extreme cases

    A. Electrons inside the atom. Localized. Sharp energy levels and line spectrum.

    Describes behaviour of insulators.

    B. Free electrons in metal, chapter 4. Delocalized. EnergyE = 2k22m

    , energy is

    continuos meaning all energy values are permitted (if the metal piece is macroscopic.

    See calculations).

    Between these two extreme cases is SEMICONDUCTOR: The energy is continuos

    only with some energy values= ENERGY BANDS. Between energy bands existFORBIDDEN BAND, energy gap, with no allowed states for electron.

    Roughly:

    Electric properties of semiconductor Free-electron behaviour.

    Optical properties of semiconductor Behaviour inside energy gap.

    5.1 Formation of bands

    Figure 5.1

    45

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    47/108

    Start with metal. One atom, for example Na. Then many Na atoms and crystalPotential is sum of individual atoms. When atoms come near, potential barrier between

    the atoms is less than energy of 3s-electrons 3selectron is FREEMETAL.

    Figure 5.2

    In energy-axis appear band widening of sharp line. In this picture 5 atoms and 5

    electrons with 3s energy. In reality approximately 1023 electrons (about 1 mol).

    Spinevery electron has 2 possibilities, meaning 2 allowed states.Band has double amount of states and band is half filled.Fermi energyEF in the middle of the band, INSIDE THE BAND.

    No 3s-electron in potential well Standing wave with0 = 2aand E0 =

    22

    2ma2 . TwoatomsTwo electrons and two potential wells.

    Figure 5.3

    If these two atoms are near each otherone energy well, standing wave2(2a) =n (5.1)

    meaning 1 = 4a and 2 = 2a. Also two energy levelsEn E0n2 , meaningE1 =E0 andE2= 14E0.Finally, many Na atoms and many potential wells: Nwells near to each other.

    46

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    48/108

    Figure 5.4

    Standing wave N(2a) =n (n= 1, 2,...,N) (5.2)

    Energy levels

    En=

    1

    n2

    E0 (n= 1, 2,...,N) (5.3)

    Formation of bands in covalent material

    Previous analysis was metal. Now we take covalent material. For example silicon

    (1s22s22p6)3s23p2.

    Outermost electrons are 3sand 3pstates. Totally 4 electrons4 wavefunctions.

    A. First we think that Si-atoms are far from each other (but with correct crystal

    structure, DIA)Atoms are separated and energy levels are sharp

    Figure 5.5

    1. B. Atoms closer, lattice constant smaller. Potential barrier between the atoms

    reduces (compare with metal)electrons will spread to neighbouring atoms andappears covalent bond, valence bond. Sharp energy levels widen to be bands.

    C. Atoms more closer 3s2 and 3p2 bands overlap giving one wide band similar tometal. Total amount of states = 8 = 2

    spin

    ( 13s

    + 33p

    ) per atom.

    D. Atoms more and more closer One wide band deviates to two parts. Theseparts have equal amount of states. Lower band is VALENCE BAND. Upper band

    47

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    49/108

    Figure 5.6

    is CONDUCTION BAND. In between NO ALLOWED STATES. Width = energy

    gap =EgFORBIDDEN GAP. In reality the lattice constant gives situation D.

    Figure 5.7

    In valence band the total amount of states (with N= amount of atoms) = 12 (total)

    = 12 [2 (1 + 3) N] = 4N 4Nelectrons and 4N statesvalence band is full

    and conduction band is empty.

    For siliconEg = 1.1 eV.Eg kBT at room temperatureno electrons in conduction band.Interesting electrons (for example possible current carriers) are near the edges, meaningnear to valence band maximum and conduction band minimum. This is correct

    regardeless the bands are several eV wide.

    5.1.1 Periodical potential

    Lattice symmetry: Periodicity

    V(r+R) =V(r) (5.4)

    withR = n1a1 +n2a2 +n3a3 = lattice vector. a1, a2, a3 = unit vector, ni Z. Quantum

    48

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    50/108

    Figure 5.8

    number in the case of atom n,l,ml. Now in the case of periodical potential new quantum

    numberk. Very useful when talking about energy bands. In physics sense k is wavenumber of electron.

    Symmetry. Solution of Schrodinger equation, wavefunction (x), describes electron

    and is connected to a specific energy valueE. We move the coordinate system by onestep (a): x =x+a and results should remain similar because V(x+a) =V(x)

    d(x)dx

    = d(x+a)

    dx and

    d2(x)dx2

    = d2(x+a)

    dx2 (5.6)

    Also wave function (x) is solution of Schrodinger equation

    (x) and (x) are

    linearily dependent:

    (x+a) = (x) =c(x) (5.7)

    Normalization

    ||2 = 1 |c|2 = 1 c= exp(i) =ei (5.8)

    Nonreal wave equation can be given in two parts

    (x) =ei(x)

    R(x) (5.9)

    where Phase function, (x), Amplitude function, R(x), are real.

    Periodical potential (symmetry)Physical properties are periodic|(x)|2 = ei(x)R(x)2 = |R(x)|2 R is periodic. Periodicity and (5.9) calculation=

    (x+a) =ei[(x+a)(x)](x) (5.10)

    Comparing this with (5.7) and (5.8) c has phase angle

    = (x+a) (x) (5.11)

    49

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    51/108

    This was result of moving one step. Then we move 2 steps: x=x + 2a result will be

    2= (x+ 2a) (x) (5.12)

    And general case with n steps

    n= (x+na) (x) (5.13)

    LINEARITY. Phase and phase function are linear vs. position

    (x) =k x(k= const.) (5.14)

    And finally, we write wave function (5.9) in general form

    (x) =eikxR(x) (5.15)

    This is Bloch theorem (1928). Wave function(5.15) has name Bloch function. Felix

    Bloch, PhD 1928 about quantum theory of solids. Nobel 1952 about developing the

    nuclear magnetic resonance method.

    We put Bloch function (5.15) into Schrodinger equationwe get wave equation forR(x), for the periodic amplitude function

    2

    2md

    2

    R(x)dx2

    ik2

    mdRdx

    + 2k22m

    +V(x)R(x) = ER(x) (5.16)Wave function depends on ksolutionR(x) depends on kenergyEdepends on k

    (x) = nk=eikxRnk(x) (5.17)

    E= En(k) (5.18)

    k appears in a same way like quantum number of atom.

    k = wave number. Very near to each other. Solution (orR) can be understoodanalyzed as a function ofk.

    n is index for energy band. Energies with one nvalue form energy band curve as afunction ofk, meaning an energy band.

    Next: in 3 dimensions.

    3 dimensional analysis. Bloch function (5.15) will be in form

    (r) = nk(r) =eikrUnk(r) (5.19)

    k= wave vector.Unk = function with periodicity of the lattice: Unk(r+R) =Unk(r) with R= lattice

    50

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    52/108

    vector.

    Again move the coordinate system with a step of lattice vectorFinal result in 3Dfor Bloch function

    nk(r+R) =eik

    R

    nk(r) (5.20)

    and energy values becomes k-dependent

    E= E(k) (5.21)

    This was traslation symmetry, when moving the coordinate system. Similar way would

    be obtained result when using lattice rotation symmetry or inversion symmetry:

    En k En k (5.22)

    Possiblek-values

    First 1 dimension. Boundary conditions, size of the crystal = L = N aand wave function

    must fulfill (x+N a) = (x) We get

    k2

    N a n(n= 0, 1, 2, . . . , N 1) (5.24)

    k is wave number and (5.24) is similar to Bohr quantization for free electron.

    Figure 5.9

    Ifn in (5.24) gets values n= 0, 1, 2, . . . , N 1 All possible phase angles in (5.20).All possible phase anglesAll possible possibilitiesamount ofk-values =N=

    amount of atoms. kmax=

    2

    a and distance between neighbouringkvalues = 2

    L =

    2

    Na .UsuallyN= 0, 1, 2, . . . , N 1 is replaced by n= 0, 1, 2, . . . , N

    2

    a k

    a .

    3dimensionalFirst usual cubic structure (SC). All coordinate axis in the same way In k-spacepossiblekvalues form cubic lattice with length of crystal face = 2

    L . Full length of the

    crystal = 2a

    . k-vector has totallyNdifferent values. This has a name First Brillouin

    zone, 1st BZ. If original crystal is not simple cubic1st BZ will be different. Also inthese cases k-vector has N values. k-vector takes in k-space a volume

    Vk=(2)3

    V (5.25)

    51

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    53/108

    Figure 5.10: 1st BZ of SC-crystal

    Similar to (4.4) for free electron.

    Figure 5.11: 1st BZ of FCC-crystal. Symmetry points and lines are noted in standard way

    5.2 Models for free energy bands

    5.2.1 Free electron model

    Old information: wave function

    (x) = 1

    Leiqx (5.26)

    Energy

    E= 2k2

    2m (5.27)

    52

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    54/108

    For wave number we use symbol q. Divide in 2 parts:

    q= k +K (5.28)

    where k= wave number inside 1st BZ and

    K= integer 2a

    (5.29)

    (x) =eikx 1

    L eiKx = nk (5.30)

    Meaning periodic amplitude function R(x) is described byeikx and index n, that labels

    the bands, is in form ofk

    Energy bands

    E= En(q) = 2

    2m(k+K)2 (5.31)

    OriginalEn(k) values are moved by value ofK Energies are inside 1st BZ. Physicalresults remains unchanged.

    Figure 5.12

    53

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    55/108

    Figure 5.13

    5.2.2 Kronig-Penney-model (1930)

    We describe lattice by rectangle potential, name Kronig-penney-potential. Calculate

    (x) in different parts. Result:

    0=A exp(ik0x) +B exp(ik0x); 0 x a1=Cexp(ik1x) +D exp(ik1x); a x a +b (5.32)2=Eexp(ik2x) +Fexp(ik2x); a +b x 2a +b

    k0 = k2=

    2mE2

    and k1 =

    2m

    (E V0)2

    (5.33)

    It is possible to calculate coefficients A, B , C, . . . and to obtain condition for existence.

    We will not calculate here, but we make simplification

    Barrier height V0 and width b 0 in such way that area remains: meaningdelta-functions, -function

    V(x) =n

    V0b(x na) (5.41)

    Coefficient A, B , C , D, . . . will have solution only if

    Figure 5.14

    Psin aa

    + cos a= cos ka (5.42)

    54

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    56/108

    where

    P =ma

    2V0b (5.43)

    and

    = 2mE2

    Right-hand side of (5.42) only cos this is between [1, +1] also left-hand side

    Figure 5.15: Left-hand side of (5.42)

    must be between [1, +1].Exist forbidden values for and therefore for energy E, values where A, B , C , . . .notexist and not existENERGY GAPS.

    Maximum of each band in place a = n

    En= 22n22ma2

    (5.44)

    These maxima in same level as free electron energies. Large potential: Areafunction=V0b . Eq(5.42)has solutionsin = 0 and final result will be sharp energy levelslike isolated atoms.

    Area of-function =V0blarge, but < Final result will be

    En(k) = 22n2

    2ma2 1 W

    n[1 (1)n cos ka]

    (5.47)

    where width of the belt, W, should not be large.

    55

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    57/108

    Figure 5.16

    This shows that sharp energy levels start to develop to be wide, to be belt. This

    approximation has name tight binding approximation.

    Weak potential. Other extreme case, electrons almost free. V0b small. Approximations

    and series expansion and mathematics

    En(k) (5.43)= 22

    2m =

    2

    2m(20+ 20) (5.50)

    where0 =k+n 2

    a

    , meaning k when 1st BZ and n 2a

    addition for other Brillouin

    zone. In Eq. (5.50) is small correction because of the potential. Without potentialnoandEn=

    2202m

    = 2k2

    2m. Values for in extremes of energy band (edge or center of

    1st BZ)

    = 02P0a2

    (5.51)

    Two values in edge of 1st BZ and two values in center of 1st BZ. These give corrections

    to energy

    E=022P

    ma2(5.43)

    = 2V0ba

    (5.52)

    In extremes of energy band, appears a gap 2V0 ba , meaning forbidden gap = Eg = 2V0 ba .Expanded description of band: wave number can change by multiple of 2

    asimilar

    result.

    56

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    58/108

    Figure 5.17

    Inversion symmetry:E(k) = E(k) because cos is symmetric.Derivative antisymmetric:

    dE(k)dk

    = dE(k)dk

    (5.54)

    Derivative=velocity Full energy band has antysimmetric velocity distribution (laterin details).

    5.3 Electrons movement in the energy band

    5.3.1 Semiclassical equation of motion

    Bloch-function and quantum mechanics is difficultBetter to build semiclassical.Particle is decribed by wave packet. Electron stay inside one energy band. Wave packet:

    - One energy band.

    - From Bloch function.

    (x, t) =Uk(x)

    dka(k)exp

    i

    kx E(k

    )t

    (5.55)

    General case for electron in energy band.

    Simplifications

    57

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    59/108

    Figure 5.18

    A. Multiplication functiona(k) thin in k-space. Upper graph.

    k2a

    , meaning width of BZ (5.56)

    B. In the area ofk, the amplitude functionUk , amplitude of

    Uk Uk = const. (5.57)

    wave packet

    (x, t) =Uk(x)

    dka(k)exp

    i

    kx E(k

    )t

    (5.58)

    Important result

    1. Wave packet is superposition of plane waves.

    2. Same form as for free electron (Eq. (2.35))

    58

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    60/108

    3. BUT. Wave packet must be multiplied with periodic amplitude function from

    Bloch-function menaning envelope of plane wave superposition modulatesthe periodic partUk(x). Lower graph.

    Width of the wave packet (compare with chapter 2) =

    x =2

    k (5.59)

    and x lattice spacing. x a, but small in the sense that it describes point likeparticle.

    Energy in wave packet is energy of energy band. Periodic potential changes dynamics

    of particle E(k) no longer quadratic.Velocity will be similar to Eq. (2.39)

    v=1

    dE(k)dk

    (5.60)

    and in vector form in 3D

    v= 1

    kE(k) (5.61)

    For equation of motion,F =ma, compare the equation of motion in electromagnetic

    field

    dk

    dt =

    q(E+v

    B) (5.62)

    Take derivative of (5.61)

    a= dv

    dt

    (5.61)= k

    1

    kE(k)

    dk

    dt

    (5.62)=

    1

    2kkE(k) (q)[E+v B] (5.63)

    F =ma

    a= 1m F gives idea about mass, effective mass m (reciprocal tensor)

    1

    m =

    1

    2

    d2

    dk2E(k) (5.64)

    in 1D: 1m

    = 12

    d2

    dk2E(k)

    Mass, i.e. effective mass, depends on shape (curvature) of energy bandEffectivemass depends on material.

    Eq. (5.61) and (5.63) give path of current carriers, if shape ofE(k) is knownSemiclassical equation of motion.

    Not need to know periodical potential, nor Bloch function,E(k) is enough. UsuallyE(k) can be modelled using a few parameters.

    59

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    61/108

    Example. Kronig-Penney tight binding approximation (compare (5.47))

    E(k) =W(1 cos ka) (5.65)

    v(k) (5.61)

    = W a

    sin ka (5.66)

    1 For small k v(k) is linear like in the case of free electron (see Fig. 5.20).

    2 There exist maximum in v(k) (see Fig. 5.20).

    3 v(k) = 0 in the border of 1st BZ.

    Reason: v(k) = 0 antisymmetric

    v k=

    a= v k=

    a

    v(k) = 0 in BZ border

    Figure 5.19: a)Energy band b)Velocity v(k) c)Effective mass m(k) for tight binding ap-proximation

    Effective mass (5.64)

    m = 2d2E/dk2

    = 2

    W a2 cos ka (5.67)

    60

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    62/108

    (See Fig 5.20)

    with small k-values m = 2

    Wa2

    m can be different from m, depends on width Wof the energy band.

    k increases m increases. m whenk

    2a . m negative whenk >

    2a (See Fig5.20) (mass

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    63/108

    Figure 5.21: Electron moving in electric field inside one energy band. Dashed and dottedline describes free electron

    and kinetic energy

    Ekin= 2q(t)2

    2m (5.72)

    Tight binding beltvelocity (5.66) v(k) = W

    sin ka Integral gives location. k(t) istaken from (5.69)

    x(t) =

    t0

    v[k(t)] dt (5.66)

    = . . .= W

    q|E|

    1 cos

    1

    q|E|at

    (5.73)

    cos oscillates also in x-space

    Frequency = =1

    q

    |E

    |a=

    2

    T

    (5.74)

    Amplitude of oscillations =A = W

    q|E| (5.75)

    These are called Bloch oscillations:

    E= 105 V cm1

    W= 1 eV

    Amplitude 0.1 m

    Electric field causes (aiheuttaa) oscillating current. (Strange). This is because of

    periodic potential, which makes electrons to behave atomic-like.Bloch oscillations never observed experimentally.

    62

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    64/108

    Scattering cut electron movement before electron goes through all 1st BZ.

    Scattering has relaxation time 1014 . . 1 013 s 1012 s.Extreme experiment: Artificial superlattice with a 5 nm Even in this case no clear

    observation.

    5.3.2 Electrical conductivity

    ScatteringDrude model ((1.11), electrons flow like viscous liquid)Now we insert friction force to equation of motion of wave vector

    (5.62) dkdt

    = k q[E+v B] (5.76)

    = relaxation time of scattering.If only electric field (B = 0)solution (with k= k0 when t = 0)

    k(t) =k0exp

    t

    1

    qE

    1 exp

    t

    (5.77)

    t < No scattering and behaviour similar to Eq. (5.69) that is k k0

    t > k(t ) 1

    qE (5.78)

    All electrons in the energy band feel change ofk by an amount

    k >1

    q|E|

    Total velocity and current = 0 (J= 0) in a full energy band.

    Reason: Energy band inversion symmetric in k-space (E(k) =E(k)) Popula-tion probability (miehitystodennakoisyys) inversion symmetric. Additionally velocity

    antisymmetric (v(k) = v(k))Total velocity, of all electron, is = 0.IfE

    = 0

    electrons move by k= 1

    hqE. If energy band full

    Population probability

    does not change J= 0 Electrical conductivity exist only if there is energy band with incomplete occupation.

    Only in this case inversion symmetry is lost and J= 0. J dissymmetry of velocitydistribution.

    k small J q 2kv(k0) = 2q2E dE(k0)dk0

    (5.79)

    Based on energy band structure, it is possible to distinguish between conductor and

    insulator.

    63

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    65/108

    Figure 5.22: Electric field and scattering Population becomes dissymetric. Must havepartly filled energy band (metal)

    5.3.3 Charge carriers in semiconductor: Holes

    Maximum of valence band and minimum of conduction band are important: The charge

    carriers are here.

    In Fig 5.23 is direct energy gap: Maximum of valence band and the minimum of

    conduction band in the same place in k-space. Other possibility: indirect energy gap.

    Figure 5.23

    Not big influence in electrical properties but optically important.

    In the minimum of conduction band the amount of electrons 1020 cm2 non dense(harva) charge carrier gas Maxwell-Boltzmann distribution. Taylor series for energy

    64

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    66/108

    band-curve

    E(k) = E(k0) +12(k k0)

    2E(k)2kk

    k=k0

    h(k k0) (5.80)

    k0= location of energy minimum.

    Electron motion with semiclassical equation of motion. For example effective mass m

    with Eq. (5.64). Valence band negative curvature m negative. But it is enough tochange the sign of the chargemotion of hole. New particle: Positive chargePositive mass m (not need m negative).Hole. Properties opposite to electron, but moves like an electron.

    E(k) graph for hole by changing E and k k.

    Ehole= Ea= Eelectron= Ee (5.81)khole=ka = kelectron= ke (5.82)

    For example velocity of hole =

    Figure 5.24

    va=1

    Eaka

    =1

    Eeke =ve (5.83)

    velocities are similar.Wave vector for hole (5.62)

    dkadt

    = dkedt

    = |q|[E+V B]= |q|[E+V B] (5.84)65

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    67/108

    This shows that hole has positive charge. Moreover: (5.81) and (5.82) hole haspositive m.

    Electric current in valence band

    I= |q| i occupied

    V(ki) (5.85)

    i occupied statesI= |q|

    holes

    V(ki) (5.87)

    Conclusion

    1. For electric current it is enough to analyze empty state and to forget other states,

    namely in valence band to forget states that are full.Holes in thin energy layerenough to know m.

    2. Exist electrons and holes. Both form non-dense (harva) charge carrier gas. Charge

    carriers behave like free particles, but m m

    3. Classical free particle model (Returning back to simple ideas after long and

    difficult discussion)

    5.4 Energy bands for various materials

    5.4.1 Metal

    1Dsimple parabola (Fig 5.25) usually FCC. 1DMore complicated energy bandstructure (Fig 5.26). Scaling with

    E0= 2k20

    2m =

    (2)2

    2ma2 (5.88)

    which is energy for electron having wave number k0= 2a.Numerical value: a 0.5 nm E 6.0 eV. In Fig 5.26 the horizontal lines showFermi-energy (EF) for different amount of electrons per unit cell (= valence-number)Analyze of Fig 5.26

    Valence z= 1 lowest energy band half empty charge carriers are usual electrons.For example earth metals Li, Na, Ka,. . . and noble metals Cu, Ag, Au.

    Valence z= 2 (Cd, Zn) EFin places where degeneracy and system is sensitive to

    the influence of periodical potential. In kplace the lowest energy band is empty.

    66

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    68/108

    Figure 5.25

    In L-place next energy band is filled upCharge carriers are holes near k-placeof 1st BZ and charge carriers electrons near L-place of 1st BZ.

    Earlier in lectures were described meaning ofk andL: Places in 3- dimensional

    Brillouin zone, in 3-dimensional reciprocal space.

    Valence z= 3 (Al)Lowest energy band is full. Second and third partly occupied.One example. Energy bands fo Cu. Electron structure (Ar)3d104s. Colour

    comes from optical transition from 3d-belts toEF. Difference 2 . . 3 eV which isred-yellow wavelength.

    5.4.2 Semiconductors and insulators

    Ge, Si, GaAs. Most important semiconductors. Valence band similar: each has

    maximum in the center of 1st BZ.

    Two valence bands, different curvature (m) meaning heavy and light band.

    GaAs direct energy gap. (Place of conduction band minimum)

    Si has conduction band minimum near BZ border

    Ge has conduction band minimum in BZ border

    indirect

    Rotation symmetry. Energy bands have similar rotational symmetry like the crystal

    lattice.

    Si has 6 minima (minimum in [100] direction) and Ge has 8 minima (minimumin [111] direction). Equal-energy surface: GaAs sphere but Si and Ge spheroid

    Curvature different m has 2 components: Longitudinal and transversal.

    67

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    69/108

    Figure 5.26

    Insulator. Similar to semiconductor, butEg wide.SiO2 is most important in technical applications. Amorphic (non crystalline).

    5.5 Interaction between electron and wave

    5.5.1 Disturbation theory

    Earlier electromagnetic field,Econstant,

    optical and

    kBT

    Semiclassical

    equations of motion.

    Now we investigate eV (Semiconductor) kBT.Schrodinger equation ((2.27)) H0 =i

    t

    .

    Here H0 is operator, H0= 22m 2

    x2+V(x)

    Giving for free electron

    (r, t) = exp

    i

    ki r Eit

    (5.91)

    Interaction with wave means that now we have in Schrodinger-equation one additional

    68

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    70/108

    Figure 5.27

    INTERACTION-term(H0+ H

    ) =i

    t (5.90)

    This additional interaction term

    H = A ei(qrt) +ei(qrt) (5.89)is like an disturbation for electron movement ((r, t)).

    Eventually mathematics shows that disturbation H gives additional, new plane wave

    components with

    Ef= Ei (5.93)kf=ki q (5.94)

    whereEf and kfare final energy and wave vector andEi and ki are initial, beforeinteraction. qis the wave vector of the wave (for example photon).

    Disturbation makes electron to have transition from initial state i to final state f. Eq.

    (5.93) and (5.94) are clearly the conservation laws for energy and momentum. Eq.

    (5.94) gives momentum if multiplied by .

    means disappearance (+) and appearance (-) of an energy quanta. These can becalled absorption/annihilation (+) and emission/creation (-).

    For electromagnetic wave is photon. v = f Energy and momentum are con-nected

    = vk (5.95)

    69

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    71/108

    Figure 5.28

    This can be understood as photon energy band curve v = 3 108 m s1 = 1 eVwhen = 1.24 m and q= 5.1 104 cm1Comparison to wave vector of an electron:

    q2

    =

    2

    a

    =ka

    =k a k (5.96)

    Photon wave vector q 0 in comparison to electron wave vector k, becausea 5 104.

    Electron wave vector remain unchanged in optical transition.

    PHONON. v vsound 1000ms1 105 times less than light velocity.(5.95)Phonon energy very small (

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    72/108

    Figure 5.29

    5.5.2 Optical properties of semiconductor

    Energy structure gives clear and easy explanation.

    Direct energy gap (For example GaAs).

    > Eg gives allways transition. One electron jump from valence band to conductionband and one photon disappears. Limit for optical absorption clearly from

    Eg.

    Inirect energy gap (Si, Ge, etc.)

    For transition is needed also a phonon (k= 0) Transition probability smaller(because more complicated).

    In practical situation light with Eg is absorbed to GaAs about 100 times strongerthan to Si. Many applications in components.

    71

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    73/108

    Figure 5.30

    72

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    74/108

    Figure 5.31: Energy-wave vector diagram for photon and phonon

    Figure 5.32

    73

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    75/108

    Chapter 6

    Semiconductor physics

    - Amount of charge carriers.

    - How to influence to the amount of charge carriers.

    - How well they move.

    6.1 Density of charge carriers

    6.1.1 Density of states in energy band

    Charge carriers are near the max of valence band and min of conduction band. Equal

    energy-surface is ellipsoid. Energy in conduction band

    E(k) = Ec+3

    i=1

    2(ki koi)22mii

    (6.1)

    Ec= Energy of the minimum in conduction bandkoi= location of

    Ec in k

    axis

    mii= effective mass in iaxis direction

    Ifmii scalarEq. (6.1) looks like free electron energy.Density of states for free electron Eq. (4.5)Now density of states =

    gc(E) = 122

    2m

    2

    3/2

    E Ec (6.2)

    Here mis replaced by m andE

    is replaced by (

    E Ec).

    74

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    76/108

    Figure 6.1: Near to the min of conduction band

    Ifmii values are different m is average. One can get:

    m= (m11m22m

    33)

    1/3 (6.3)

    If more than one energy minimum (6.2)must be multiplied by the amount of minima.For valence band in similar way

    gv(E) = 122

    2m

    2

    3/2Ev E (6.4)

    Ev = Energy of the maximum in valence band

    mh = average effective mass for the hole

    75

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    77/108

    Figure 6.2: Near to the max of valence band.

    6.1.2 Electron and hole densities

    Density of states increases asE Fermi energyEFdefines how big amount of charge

    carriers fit into the energy band. UsuallyEF inside forbidden band and usually

    Ec EF kBT (6.5)

    EF

    Ev

    kBT (6.6)

    This kind of semiconductor is called non-degenerated semiconductor (degeneroitumaton)

    In this case charge carriers form non-dense (harva) gas Fermi-DiracMaxwell-

    Figure 6.3: Density of states for valence band and conduction band

    76

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    78/108

    Boltzmann distribution

    f(E) = 1eEEFkBT

    + 1

    calculations fe(E) = 1eEEFkBT

    =e (EEF)

    kBT (6.8)

    Similar approximation in valence band fh(E) =e(EFE)

    kBT .

    Now we can calculate current carrier density with integral

    n=

    Ec

    g(E)f(E) dE= (6.2)(6.8). . . =Nc e(EcEF)

    kBT (6.12)

    Nc= 2

    mekBT

    22

    3/2= effective density of states in conduction band (6.13)

    In integral is used 0

    xex dx= 2

    Nc Amount of statesvolume within thin energy layer ofkBT aboveEc.Numerical values:

    m = m

    T= 300 K

    Nc= 10

    19 cm3

    For holes in similar way:

    p= Ev

    gv(E)fh(E) dE=Nve(EFEv)

    kBT (6.14)

    whereNv is similar to Nc but with mh (i.e. effective density of states in valence band).

    (6.12) and (6.14) are basic equations for charge carrier densities for non-degenerated

    semiconductors.

    Multiply (6.12)(6.14)np- rule (EFdisappear, supistuu).

    np= NcNvexp

    Eg2kBT

    = n2i (6.15)

    For pure clean semiconductor (without doping): electrons are from valence bandn= p (6.16)

    n= p = ni=

    NcNvexp

    Eg2kBT

    (6.17)

    For example Si: ni 1010 cm3. Depends exponentially on T. Very small.

    (6.12)

    (6.14) EF = EFi =1

    2(Ec+ Ev) few eV

    3

    4 kBT0.02 eV ln

    mem

    h 1

    2(Ec+ Ev) (6.18)

    77

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    79/108

    MeaningEFapproximately in the middle of forbidden band.

    6.1.3 Doped semiconductor

    Changing of charge carrier concentration by doping. Basis for semiconductor technology.

    Enough 106 doping, every 1 000 000th atom.

    DONOR: Impurity that has one electron more than host atoms (other atoms) in crystal.

    This electron becomes free, rise to conduction band.

    ACCEPTOR: Impurity that has one electron less than host atoms. Takes one electron

    (accepts electron) from enviroment and the result is a hole in valence band.

    Donor increases density of electrons and acceptor increases density of holes.

    For Si donor impurities are V-group atoms (P, As, Sb) and acceptor impurities are

    III-group atoms (B, In, Ga, Al). For compound semiconductor (yhdistepuolijohde), e.g.GaAs, situation is more complex: Donor impurities are VI-group atoms and acceptor

    impurities are II-group atoms. IV-group atoms can be donor or acceptor, depending on

    if impurity replaces Ga or As. This depends on concentration.

    Energy of electron of donor-atom Take one donor-atom. Donor will result in one elec-

    tron plus positive im. Weak attraction force

    Ed = Ec+ 2k2

    2m e

    2

    4r (6.19)

    Small = dielectric constant of the lattice.

    2 first terms: Electron in conduction band.

    Third term Coulomb potential, r > lattice constant.

    Appears energy level belowEc+ 2k22m . System similar to Hydrogen atomEnergy

    Figure 6.4: Energy level belowEc+ 2k22m

    Ed Ec= 13.6 eV m

    m

    0

    2(6.20)

    78

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    80/108

    Radius

    r = 0.53 Am

    m

    0

    = average of distance (6.21)

    Here 13.6 eV and 0.53A areE and r for ground state of hydrogen

    Figure 6.5: Hydrogen atom

    0 10

    m

    m 0.1

    ionization energy = Ed Ec 0.01 eVand orbit radius 50 AkBT 0.025 eV >Ed Ec Electron rises from donor state to conduction band. and 0 are dielctric constant for the lattice and for vacuum.

    If 0

    only 5 Ed Ec 0.04 eV not suitable for electronics components.For acceptor: Similar hydrogen-like system, Eq. (6.20) and (6.21). Hole disen-

    gages(vapautuu) to valence band.

    Figure 6.6: Energy level aboveEv

    Both energy levels are near to the energy bands. Important energy levels. Exist also

    energy levels deep inside the forbidden band:

    Reason: Large crystal imperfections. These energy levels are not good.

    Deep levels and shallow levels. Impurity levels rather easy to measure (luminescence)

    Exist good information.Purity is very important. CLEAN. Reason: 106 doping would be enough.

    Density of crystal imperfections and unwanted impurities must bedoping. If notwrong energy levels (virhetiloja) inside energy gap.

    79

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    81/108

    Figure 6.7: Impurity states and ionization energies for Ge,Si and GaAs. D and A meandonor/acceptor state.

    Nowaday the best progress exist with Si: Unwanted impurities in the level of1013 at/cm3. This shows the level of quality.

    In lectures analyze of Fig. 6.7.

    6.1.4 Charge carrier density of doped semiconductor

    Basic detail: in room temperature kT > Ed Ec and kT > Ea Ev

    n= density of electrons = density of donor impurities = Nd

    p= density of holes = density of acceptor impurities = Na

    This is the influence of doping. Now we look this in more details.

    In Fig 6.8 is shown energies as function of location. E= 0 Energy levels are descibedby horizontal lines.First: ntype semiconductors (with donors).Occupation probability by using Fermi-Dirac-distribution: Probability to have electron

    in energy levelEd (i.e. donor state)

    f(Ed) = 1eEdEFkT

    + 1(6.22)

    80

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    82/108

    Figure 6.8: Energy band diagram

    Ionizated donor atoms with density

    N+d = [1 f(Ed)] Nd = Nd

    eEdEFkT

    + 1(6.23)

    where total amount of donor atoms = Nd. If Fermi-energy EFfar from Ed EFEd 0 N+d =Nd meaning all donor atoms ionizated.Acceptors in similar way. Ionizates acceptor atoms with density

    Na =f(Ea)Na= Na

    eEaEF

    kT

    + 1(6.24)

    IfEF far fromEa Ea EF 0 Na =Na meaning all acceptor are ionizated.Calculate Fermi-energyEF.Earlier n= p and Eq. (6.12) and (6.14). Now charge neutrality

    n+Na =p+N+d (6.25)

    Using this equation, it is possible to calculateEF and n, p, Na, Nd.Calculation rather complex. Approximative calculation in book and here. First n-type

    semiconductor (Donors, and ifNa then anyhow Nd Na). In very high temperaturesnand pare high, meaning Na and Nd.In very high temperatures Na and Nd have no influence toEF.In high temperatureEF in center of forbidden gap.Temperature limit T0 where influence of doping disappears: Ni(T0) =

    12

    Nd. One get:

    T0= Ed

    2kBln 2NcNvNd

    (6.28)

    whereNc andNv are effective density of states in conduction band and valence band

    81

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    83/108

    (Eq. (6.13)).

    IfNd = 1017 cm3 T0 = 800 K for Si (far above room temperature).

    Below this temperature limit one gets density of electrons and holes

    n= Nd Na majority charge carriers

    and p= n2i

    Nd Na minority charge carriers

    (6.29)

    Result

    1. Amount of dominant doping defines the density of dominant charge carriers (now

    electrons).

    2. Minority charge carrier density is proportional to 1doping density

    and

    ni(internal

    density of charge carriers in undoped semiconductor).

    3. Fermi energyEFmoves towards the majority band (now conduction band).

    Now with n-type material one gets:

    EF = Ec kBT ln NcNd Na (6.30)

    New temperature limit T1: whenEF increases up toEc

    T1= (Ec Ed)

    kBln

    NcNdNa

    (6.31)IfNd = 10

    17 cm3 T1 = 40 K for Si. Donors are no longer ionizated and electrondensityn Nd (frozen)ptype semiconductor. Similar results. Acceptor-doping is dominant. At very hightemperaturesp= n = ni.

    Room temperature range: P =Na and n= n2iNa

    .

    Very low temperature: Charge carrier densities go to zero. Temperature dependence ofcharge carrier (hole) density and temperature dependence of Fermi-energyEF similarto Fig. 6.9.

    Ifvery strong doping in semiconductor; Doping more than effective density of states

    (Nc or Nv, Eq. (6.13)).

    Must be doping 1019 . . 1 021 cm3.Usually is used sentence: Semiconductor is degenerated

    Fermi-energy is inside a band Nd EFinside conduction bandNa EFinside valence band82

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    84/108

    (a) Fermi-energy in Si as a function

    ofT

    whenN

    d =N

    a = 1016

    cm3

    (b) Electrodensity in Si when Nd= 1016 cm3

    Figure 6.9

    In this case behaviour is like ntype orptype metal. This possibility is used in placeswith the contact.

    6.2 Mobility,

    Mobility of charge carriers is limited by scattering. In principle all disturbances inperiodic potential. Two most important: Phonon (lattice vibrations) and impurities in

    doping.

    Phonon scattering Similar to photon Fig. 5.30.

    Figure 6.10: Phonon scattering

    83

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    85/108

    Relaxation timephononeventually one gets

    phonon=qphonon

    m (m)5/2T3/2 (6.35)

    phonon= relaxation time = time between collision to phonon.

    Mobilityphonon increases when temperature decreases.

    Mobilityphonon increases also when effective mass decreases.

    Impurity scattering

    Scattering from impurities is calculated like collision to obstacle (este) with an area, in

    this case so called scattering cross-section.

    Figure 6.11: Cross-section area of impurity scattering

    Eventually one gets for mobility limited by impurities:

    Imp= qImp

    m T

    3/2

    mNImp

    (6.39)

    Imp= relaxation time = time between collisions to impurities.

    NImp= Amount of impurities.

    In this case mobility Imp increases when temperature increases.

    MobilityImp increases when effective mass m decreases.

    In practical cases exist both: Phonon and impurities. Indipendent of each othersScattering probabilities can added together

    1

    =

    1

    phonon+

    1

    Imp(6.40)

    1

    =

    1

    phonon+

    1

    Imp(6.41)

    This has a name: Mathiessen rule. In low temperatures mobility is limited by impurity

    scattering.

    84

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    86/108

    Mobility increases with increasing temperature up to70 . . 100 Kwhen phonon scattering

    becomes more important. In room temperature mobility is limited by phonon scattering

    and decreases with temperature increases.

    In Fig. 6.12 is mobility vs. concentration when T= 300K.

    Figure 6.12

    Impurity scattering starts to influence when concentration becomes 1016 cm3.Electron mobility in GaAs very good because effective mass m very small (In GaAsm

    mc= 0.07).

    6.3 Generation and recombination of charge carri-

    ers

    Amount of charge carriers can be influenced by temperature, doping AND BY LIGHT.

    Electron is lifted to conduction band using photon energy Electron-hole-pair.This is GENERATION OF CHARGE CARRIERS

    RECOMBINATION: Electron from conductivity band goes down to valence band into

    an empty state.

    Meaning electron goes to a hole.

    85

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    87/108

    Figure 6.13

    In this process, recombination, is emitted one photon; energy difference, E, whichbecomes free, is used for .

    Recombination also through impurity levels which are inside energy gap. Specially deep

    levels are important. Deep levels are called trapping states (loukkutilat), or sometimes

    as recombination center.If energy gap is direct, then recombination is usually directly to valence band. Not

    need deep levels. If the energy gap is indirect, then recombination takes place mainly

    via deep level. For example in silicon:

    G= generation rate (generaatiovilkkaus) =

    = transition rate from valence band to conduction band =

    = transition/time and volume

    R= recombination rate == transition rate form conduction band to valence band

    Generation because of external electromagnetic radiation, but also because of black

    body radiation.

    R= A np (6.42)Reason: Recombination proportional to occupied initial states in conduction band

    (n) and to empty final states in valence band (p), meaning product np. A describes

    transition probability. Material parameter.

    In similar way

    G= B NvNc (6.43)

    B = transition probability, depending on excitation activity

    Nv = density of states in full valence band

    Nc= density of states in empty conduction band

    Thermodynamic equilibrium: No external radiation, only black-body radiation

    G0

    and R0 where R0=A n0p0.

    86

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    88/108

    Equilibrium G0=R0 and therefore

    G0=R0=A n0p0 (6.15)= An2i (6.44)

    In general case: Net recombination =

    R= R R0 (6.44)= A(np n2i ) (6.45)

    External optical radiationGeneration Goptcharge carrier density changes

    n

    t =Gopt R= Gopt A(np n2i ) =

    p

    t (6.46)

    This gives movement (transition) of charge carriers between the energy bands. For

    example ntype semiconductor;

    n= n0+ n and p= p0+ p (6.47)

    Recombination in n-p pairs, also generation in n-ppairs n= p. Additionally

    n n0 and p n (6.48)

    Recombination Rcan be approximated

    R= A(np n2i ) =. . . An0p= 1

    pp (6.49)

    where time constant p = 1An0

    = Recombination lifetime for minority charge carriers.

    Radiationmore charge carriers.Steady state

    tstabilizes. (6.46) and (6.49) Photoconductivity

    =qpp+qnn= q(p+n)pGopt (6.52)

    Switch OFF radiationRecombination of extra charge carriers

    (6.46) p(t) = n(t) = p(0) expt

    p

    (6.53)

    This shows that p is lifetime of minority charge carriers, can be measured using

    resistivity (conductivity) measurement.

    Fig. 6.14 Shows lifetime in silicon. With usual doping 1016 cm3 10 s.In GaAs direct gap direct recombination and 1 ns with doping level of

    1017

    cm3

    .Unequilibrium between energy bands. Appears because of radiation. Time constant =

    87

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    89/108

    Figure 6.14: Lifetime of minority, charge carriers vs. amount of majority charge carriers

    lifetime of recombination. Many orders longer than scattering relaxation, meaning slow.

    Unequilibrium is described by quasi Fermi level:

    density of electrons and holes

    n= Ncexp

    Ec EFn

    kT

    (6.54)

    p= Nvexp

    EFp Ev

    kT

    (6.55)

    whereEFn andEFp are quasi Fermi level for electrons and holes. If thermodynamicequilibrium EFn = EFp = EF.Usually for majority charge carriers quasi Fermi level not change;EFn = EF.But for minority charge carriers (p p0,p0 small) quasi Fermi level moves towardsthe bands (ifp valence band).

    6.4 Diffusion current

    In semiconductors often density gradient of current carrierssmall current, diffusioncurrent. Analysis of diffusion gives eventually total current density (left-right + right-

    left)Jdiffusion=qDn

    n

    x (6.62)

    where diffusion constant

    Dn=1

    3v2T=

    kBT

    m =

    kBT

    q n (6.63)

    88

  • 8/10/2019 Semiconductor and superconductor physics %2528korjattu%2529 (1).pdf

    90/108

    Figure 6.15: Gradient of charge carrier density

    here

    vT = Thermal velocity =

    3 kBT

    m

    = relaxation time = time between collisions

    Dn=kBT

    q n is called Einstein relation

    One example: Electron density depends exponentially on location: n expxLn

    ,Ln=

    profile parameterJdiffusion= qDn

    n

    x = qD

    nLn n (6.64)

    general formulation of current density = J=qvnn average velocity, now diffusionvelocity

    vn =DnLn

    (6.65)

    If not ex