Basic Properties of Compound

Semiconductors

Lattice and Band Structure

(Swaminathan Ch 1)

1
 III-V Compound Semiconductor 化合物半導體

III IV V

Diamond and Zincblende

Wurtzite
2
 Semiconductors
Amorphous semiconductor  Single crystal
 Poly crystal
 Amorphous
 Pseudomorphic
 Metamorphic
Short-range order (a reasonably well-defined
bond length, bond angle, and coordination
number)
Long-range order (remote portions exhibit
correlated behavior)

• sc-Si (singlecrystalline Si; grain size g >10 cm)

• mc-Si (multicrystalline Si; g = 1–100 mm)
• pc-Si (polycrystalline Si; g = 1–1000 μm, no amorphous tissue)
• μc-Si:H (hydrogenated microcrystalline or, equivalently, nanocrystalline Si; g < 1 μm,
3
material contains amorphous tissue).
Pseduomorphic and metamorphic

TEM showing
dislocations

4
ABABAB --
Hexagonal

ABCABC –
Cubic FCC

5
ABABAB -- Hexagonal

ABCABC – Cubic FCC

6
 FCC

Hull, introduction to dislocations, BH 7

Hexagonal structure

8
 Five common tetrahedral structure types

C. Hammond, The basics of

crystallography and diffraction

9
 SiC
ploytypes

C. Hammond, The basics

of crystallography and
diffraction

10
Crystal structure and lattice constant

11
Miller indices

12
Notations for direction and plane
[ … ] indicates a direction: [111], [112] …
( … ) indicates a plane: (100), (110)
< … > indicates a family of directions: <110>
represents directions [110], [101], [011], …
{ … } indicates a family of planes.

13
Cubic system

14
(111)A and (111)B in Zincblende

15
How to identify the orientation of wafers
EJ Major Flat
US Major Flat

Minor
Flat

For wafers in size larger than 3 inches, notches are

used to replace the flats.

16
 Hexagonal system

a1 : [2 -1 -1 0]
a2 : [-1 2 -1 0]
a3 : [-1 -1 2 0]
17
 Wurtzite
structure

Phillips, Bonds and bands in

semiconductors, p.72, 2010. 18
 Internal parameter u and c/a
b

Bernardini, PRB 56, R10024, 1997. u = b/c

For perfect tetrahedral coordination,
u = 3/8 = 0.375

a (Å) c (Å) c/a

AlN 3.112 4.982 1.6009
GaN 3.1896 5.1855 1.6258
InN 3.548 5.760 1.6234

Adachi, properties of semiconductor alloys, p.17 19

Covalent and ionic character

Cohesive energy is the energy to

break the atoms of the solid into
isolated atomic species. In
general, the cohesive energy
decreases with increasing ionicity.

20
 Energy band structure
H (r )  E (r )
p2
H  V(r )
2m 0

Bloch’s theorem
 (r )  exp(ik  r ) U k (r )
U k (r )  U k (r  T)

Nearly free-electron approximation

The effective mass is defined as the reciprocal of the
curvature of E-k diagram
h2 2 2 2
 ( 2  2  2 ) (r )  E (r )
2m x
*
y z
21
 k·p method
 p2 h h 2k 2 
  k p   V(r ) U k (r )  E(k ) U k (r )
 2m 0 m 0 2m 0 
(H 0  H1  H 2  H 3 ) U k  EU k
p2
H0   V(r ) Hamiltonian of the unperturbed problem
2m 0
h2
H1  kp Spin independent perturbation
m0
h2
H2  2 2
(V  p)   Spin dependent problem
4m 0 c
Usually very small compared to H2 since the main
2
h contribution to the spin-orbit interatction comes
H3  2 2
(V  k )   from the core region where V and p are very large.
4m 0 c 22
Nonparabolicity and effective mass from
Kane’s model
Nonparabolic correction (to k4)

m* 2 h 2 k 2 2 3E g  4  2 / E g
2
h 2k 2
E(k )   (1  ) ( )
2m *
m0 2m *
(E g  )(3E g  2)

Effective mass

1 1 2P 2  2 1 
  2 
m *
m 0 3h  E g E g   
where P is the momentum matrix element between valence and conduction
band. For GaAs, 2m0P2/ħ2 = 22.71 eV, Eg=1.5192 eV, Δ=0.341 eV.
cf. E. O. Kane, J. Phys. Chem. Solids 1, 249, 1957.
23
1st Brillouin zone for
cubic semiconductor

24
GaAs energy band structure

25
Ge, Si and GaAs

26
3D constant energy surfaces

27
4 for Cu
1.49 for Si

28
10-6 eV/bar

29
30
Varshni Equation

T 2
E g ( T )  E g ( 0) 
T 

Varshni Equation:
 is close to Debye temperature
Y. P. Varshni, Physica, 34, 149,
1967

31
Mechanisms for energy gap variation:
(1) Temperature dependent dilatation of
the lattice. The effect is linear at high
temperature and accounts for only a
fraction about 0.25 of the total
variation.
(2) Temperature dependent electron
lattice interaction. For
T << q Eg  T2
T >> q Eg  T
where q is Debye temperature

T 2
E g (T)  E g (0) 
T 

32
 Bose-Einstein Expression
2a B
E g (T )  E g (0)  q / T
e 1
where aB represents the strength of electron-phonon
interaction and q corresponds to the mean phonon energy
in units of temperature.

This equation contains Bose-Einstein statistical factor for

phonon emission and absorption.

L. Vina, PRB 30, 1979.

2024/2/20 33
Positive temperature coefficient for energy gap

 Ordinary semiconductors such as Si, GaAs and so on

have a negative temperature coefficient.
 The temperature coefficient of some narrow gap
materials, such as HgTe (Zincblende) and PbSe/PbTe
(Lead salt) is positive.
 The interband electron-phonon interaction in the
narrow-gap semiconductors leads to band mixing in
addition to the Kane mixing and to an additional
renormalization of the energies of the electron bands.
(Guenzer PRB3, 4655)
34
Effective mass

35
Constant energy surfaces

36
Conduction band effective mass is a
function of doping. Nonparabolicity
becomes significant as the doping
level increases and the band
becomes filled.

37
Compound Semiconductor
Technology
光電半導體技術
EE5115
921 U7150

1
Basic Properties of Compound
Semiconductors

Alloy

(Swaminathan Ch 1)

S. Adachi, Properties of Semiconductor Alloys, Wiley, 2009.

Vurgaftman, JAP 89, 5815, 2001.

2
Group III atoms and group V
atoms are seated in their own
sublattices. Misplaced atoms are
called anti-site defects.

3
Band gap energy versus lattice constant

4
Ordering and disordering
in InGaP

S. Froyen, APL 68(20), 2852, 1996.

5
Ordering structures in SiGe
RS1: each plane along a particular
[111] direction consists of either all
Si or all Ge atoms. (Si-Ge)-(Ge-Si)

RS2: Similar to RS1 but in (Si-Si)-

(Ge-Ge) sequence.

MS1: The sequence is (Si-M)-(Ge-M)

where M represents a mixed plane
with the average Si0.5Ge0.5
composition.

RS3: The sequence is four SiGe

layers with different parameters a, b,
g, and d, which represent the
occupation probabilities of Ge in
each plane.

6
PRB51, 10947
Vegard’s law for lattice constant
Vegard proposed that the
lattice constants of solid
solutions should vary linearly
with composition in 1921.
L. Vegard, Z. Physik 5, 17
(1921).
Professor Lars Vegard
(1880/02/03-1963/12/21)
Department of Physics
University of Oslo, Norway
Vegard is known for his studies of
isomorphous compounds (Vegards law) and
for his low temperature studies of the crystal
structure of elements (N2 and O2), and
compounds that are gaseous at room
temperature. 13 physics students graduated
on projects in crystallography from 1925 to
1941 and published a paper with Vegard.

7
 Vegard’s limit and Pauling’s limit
for bond length

Linus Pauling

Extended X-ray absorption fine structure

2024/2/20 8
JAP 57, 5188
 The atoms on the mixed
sublattice (A and B) are
fixed at positions
determined by VCA,
while the atoms of the
common element (C) are
allowed to relax in each
tetrahedron to minimize
the total strain. JAP60,
3850.

2024/2/20 9
 Bond distortion in InAsPSb
Binary InSb
Binary InP
Sb
P
Sb
P
In In
P Sb
P
Sb

P Alloy InAsPSb
 Lattice mismatch (bond length mismatch)
between InP and InSb is 10.4%. P
 Bonds in alloy resemble their original In
length in binaries and thus undergo
stretching and bending distortion. Sb
 The strong distortion could affect the
properties of the alloy. Sb
10
 Bowing
parameter c
for band gap
PRB 1, 3351
ci : dielectric method based on
virtual crystal method
ce: disorder

E 0 ( x )  a  bx  cx 2
E 0 (0)  a
E 0 (1)  a  b  c
E 0 ( x )  xE 0 (1)  (1  x )E 0 (0)  x (1  x )c
11
 Effects of chemical and
structural disorder

Molecular coherent-potential approximation (MCPA)

12
PRL 52, 77, 1984
 Failure of VCA in alloys’ band gap
The failure of virtual crystal approximation (VCA) is due to atomic scale
disorder, which is attributed to:
 Bimodal distribution of bond length
Extended X-ray absorption fine-structure measurement showed that
the In-As and Ga-As bond lengths in InGaAs were almost equal to
their bond lengths in InAs and GaAs, irrespective of the In
composition. (PRL 46, 1412)
 Fluctuations in electron density at the sites of the mixed
sublattice
The charge density variation on the mixed sublattice is proportional
to the difference in electronegativity between the two mixed anions.

Electronegativity is a chemical property that describes the ability of an atom (or, more rarely, a
functional group) to attract electrons (or electron density) towards itself in a covalent bond
M. Hanini, dilute nitride semiconductors, Ch 6, pp. 182-185.
 Material parameters of alloy
P(A x B1 x C)  xPAC  (1  x )PBC
P(A x B1 x C y D1 y ) 
xyPAC  (1  x ) yPBC  x (1  y)PAD  (1  x )(1  y)PBD

If the parameters for ternary alloys AxB1-xC, AxB1-xD,

ACyD1-y , BCyD1-y are available then

P(A x B1 x C y D1 y ) 

x (1  x )[(1  y)PABD  yPABC ]  y(1  y)[xPACD  (1  x )PBCD ]
x (1  x )  y(1  y)

15
 Quaternary

III-III-V-V: InxGa1-xAsyP1-y
III-III-III-V: AlxInyGa1-x-yP
III-V-V-V: InAsxPySb1-x-y

16
 Band gap determination

E g (A x B1 x C y D1 y )  xE ACD  (1  x )E BCD  

  x (1  x )[(1  y)c ABD  ycABC ]  y(1  y)[ xc ACD  (1  x )c BCD ]

17
18
InAsPSb energy gap and lattice constant contours
InSb
0.00 1.00

0.2
0.25 0.75

0.3
C1533

0.50 0.4
0.50
C1759 0.5

0.6

0.75 0.25

1.00 0.00
InP
0.00 0.25 0.50 0.75 1.00
InAs
19
Pentanary (Quinternary)materials
InP
5.8657 Å

GaP
5.4505 Å

InAs
6.0583 Å

GaAs InSb
5.6532 Å
6.4794 Å
z

y x GaSb
6.0959 Å 20
c. f. Sminov, Lancaster Univ.
21
22
23
24
25
For the lattice-matched InGaAsP

E g ( y)  1.35  0.775 y  0.149 y 2

For the quaternary alloy

E O ( x, y)  a  by  c III x (1  x )  c V y(1  y)

For the lattice-matched InGaAsP

c( y)  0.219 y  0.149 y 2
General case of arbitrary x and y:
InGaAsP
E g ( x , y)  1.35  0.668x  1.068 y  0.758x 2
 0.078 y 2  0.069 xy  0.322 x 2 y  0.03xy 2

26
27
28
29
30
 Thermal resistivity

W ( x )  xWAC  (1  x ) WBC  c A  B x (1  x )
31
Thermal expansion coefficient

Pietsch and Marlow found that the

thermal expansion coefficient of the
relaxed lattice is about 30% smaller
than the values shown in Fig. 1.23.

32
Band lineup

33
34
35
 Band edge alignment

Tiwari, APL 60, 630, 1992. 36

37
 Basic Properties of Compound
Semiconductors

Structural Properties: Strain and its

effect on energy gap

Swaminathan Ch 1
Kittel SSP Ch 3

1
Strain and stress
Kittel SSP Ch3

2
Elastic strain

\
xz

Taylor series expansion

eij: strain components

eij deformation

3
 Stress tensor

 xx  xy  xz 
  The stress vector at any point associated with a
ij   yx  yy  yz 
plane of normal vector e can be expressed as a
  zx  zy  zz  function of the stress vectors on the planes

perpendicular to the coordinate axes.

Tx   xx  xy  xz  e x 
T     yy

 yz  e y 
 y   yx
 Tz    zx  zy  zz   e z 

ij stress
 Matrices of elastic stiffness and compliance
constants for cubic semiconductor
 xx  C11 C12 C12   e xx 
    e 
   12
yy C C11 C12 0   yy 
  zz  C12 C12 C11   e zz 
    C 44 0 0  2e yz 
  
yz
 
 zx   0 0 C 44 0  2e zx 
   0 0 C  2e 
 xy  44   xy 

 e xx  S11 S12 S12  xx 

e     
S S
 yy   12 11 12 S 0   yy 
 ezz  S12 S12 S11   zz 
2e    S44 0 0  yz 
 yz
   
2ezx   0 0 S44 0  zx 
2e   0 0 S   
 xy   44   xy 
 S11  S12
C11 
S
S12
C12  
S
1
C 44 
S44
S  (S11  S12 )(S11  2S12 )
6
 Young’s modulus and Poisson’s
ratio for cubic crystals

1 S44 2 2 2 2 2 2
 S11  2(S11  S12  )(l1 l 2  l 2 l3  l1 l3 )
E 2
S44 2 2 2 2 2 2
S12  (S11  S12  )(l1 m1  l 2 m 2  l3 m 3 )
 2
S44 2 2 2 2 2 2
S11  2(S11  S12  )(l1 l 2  l 2 l3  l1 l3 )
2
For <100> axes
1 Young’s modulus is the stress
 S11 divided by strain
E
Poisson's ratio is the ratio of transverse contraction strain to
S
   12 longitudinal extension strain in the direction of stretching force.
S11 Tensile deformation is considered positive and compressive
deformation is considered negative. The definition of Poisson's ratio
contains a minus sign so that normal materials have a positive ratio. 7
Young’s modulus and Poisson ratio
 e  S11 S12 S12   0 
 e   S S S 0
   12 11 12   

e z  S12 S12 S11  

e  S12
1
e z  S11  
E

1 S12 C12
 S11  
E S11 C11  C12

8
 Bulk modulus
1 e  S11 S12 S12   
B e   S S11 S12   
3(S11  2S12 )    12
E e  S12 S12 S11   
B
3(1  2)

Cij Sound velocity

e  S11  2S12 
v
  is density V 1
 3e  3S11  2S12   
V B

E v  b(S11  S12 ) under [100] stress

 d 
E v   S44  under [111] stress
2 3
9
 Deformation potential constants

1 dE g
a   (C11  2C12 )
3 dp

I. Vurgaftman, JAP 89, 5815.

10
 The fail of coherent:
relaxed with dislocations
critical thickness
D. L. Smith Thin film deposition: principles
and practice, Ch6
K. N. Tu, Electronic thin film science for
electrical engineers and material scientists, Ch7

11
Coherent Growth

12
Dislocation
 Misfit
dislocation

l=60º
(1/2)a<011>60º
JAP68, R31
Threading dislocation
17
 b
e
2r
b

2r

E e  e 2 Bh f: mismatch

b 2   r   b   hc  
Ed   ln   1 hc  ln b   1
4(1  )   b   8(1   )f    
King-Ning Tu, Electronic thin film science for electrical engineers and material
18
scientists, Ch. 7, Macmillan 1987.
 Matthews-Blakeslee model
(Force-balance model)

b(1   cos 2 )   h c  
hc   ln   1
2f (1  ) cos l   b  

where
b: Burger’s vector (usually ~a/(2)1/2)
: Possion’s ratio
l: the angle between the slip direction and that direction in the film plane which
is perpendicular to the line of intersection of the slip plane and the interface (for
threading dislocations in GaAs/GaAsP cosl = 0.5)
: the angle between the dislocation line and its Burgers vector (for threading
dislocations in GaAs/GaAsP cos = 0.5)
J. M. Matthews and A. E. Blakeslee, J. of crystal growth 27, 118 (1974) .
19
20
21
Strained layer and its
electronic structure

22
Biaxial Stress

23
 Biaxial strain on (001) plane

 x   C11 C12 C12  e x 

    C C C  e 
 y   12 11 12   y 
 0  C12 C12 C11   e z 
e
eZ
  x  y
e  ex  ey

  (C11  C12 )e  C12e z

0  2C12e  C11 e z
z
2C12
ez   e
C11

24
 Strain and Lattice Mismatch
Strain Lattice mismatch
as a f a f
e 1  1
af as as
az a z a z
ez  1  1
af as as
2C12 2C12
ez   e (a z  a f )   (a s  a f )
C11 C11

 Strain’s calculation is based on free-standing lattice constants.

 In most papers including Wei’s PRB paper, tensile strain is positive and
compressive strain is negative.
 Lattice mismatch’s calculation is based on the lattice constant of substrate. In
XRD measurement, the Bragg’s angles of epilayer and substrate are compared.
Therefore, the measurement gives the information of mismatch.
25
Find lattice mismatch
2C12
(a z  a f )   (a s  a f )
C11

a z  a f  ( a z  a s )  (a s  a f )
2C12
a z  a s  (1  )(a f  a s )
C11

C11
af  as  (a z  a s )
C11  2C12
(C11  C12 )  C12 1 
 (a z  a s )  (a z  a s )
(C11  C12 )  C12 1 

26
Bragg’s law
2 d sinq = nl

27
28
Strain-compensated Superlattice

e A d A  e Bd B  0
aS  a A aS  a B
dA  dB  0
aA aB
dA  dB
aS 
dA dB

aA aB

29
30
Strain Hamiltonian for conduction band and
valence band

H sc  a c (e xx  e yy  e zz )

H sv  a v (e xx  e yy  e zz )
  2 0 0   1 0 0 1 0 0  
      
 b v  0 1 0 e xx   0  2 0 e yy   0 1 0 e zz 
 0 0 1  0 0 1
 
0 0  2 
  
 0  1 0  0 0 0   0 0  1 
       
 3d v   1 0 0 e xy   0 0  1e yz   0 0 0 e zx 
 0 0 0   0 1 0 
 
 1 0 0  
  

Strain theory: S. H. Wei, PRB 49, 14337, 1994.

Parameters: I. Vurgaftman, JAP 89, 5815, 2001. 31
 (001) Coherent strain
H sv  a v (2e  e z )
  2 0 0   1 0 0   1 0 0  
      
 b v  0 1 0 e   0  2 0 e   0 1 0 e z 
 0 0 1   0 0 1   0 0  2  
1 0 0 
 
 a v (2e  e z )  b v (e z  e) 0 1 0 
 0 0  2
 
In this case, E1 and E2 are degenerate, while E3 is separated from E1 and
E2 by the crystal field splitting, which is given by,

C11  2C12
 b  3b v (e z  e)  3b v e
C11
C11  C12
a v ( 2e  e z )  2a v e
C11

32
 Spin Hamiltonian
0  i 1
 
 i 0  i 
 SO  0 1 i 

H SO 
3  1 0 i 
  i  i 0 
 
1 i 0 


 b  i SO 0 0 0  SO 
 
 i SO b 0 0 0  i SO 
1  0 0  2 b   SO i SO 0 

H H H 
SO S

3 0 0   SO b i SO 0 
 0 0  i SO  i SO b 0 

 i SO 0 0 0  2 b 
 SO

 2 b  l   SO i SO
  SO b  l i SO  0 where SO is the spin-orbit splitting.
 i SO  i SO b  l (cf Wei, PRB 49, 14337.)
33
 Strained Valence Bands
If spin-orbit coupling Hamiltonian is also considered, the results are,
C11  C12
E C  E C ( 0)  2a e
C11
1
E HH  E v  ( SO   b )
3
1 1 8
E LH  E v  (  SO   b )  ( SO   b ) 2   SO  b
6 2 3
1 1 8
E SO  E v  ( SO   b )  ( SO   b ) 2   SO  b
6 2 3

where SO is the spin-orbit splitting. (cf Wei, PRB 49, 14337.)

34
 Strained Valence Bands
If the strain splitting is small enough,
1 8 1 so 2 2 8
(so  s ) 2  so s    2so s  s  so s
2 3 2 3
1 so 2 2 so s s2 so 2 s
       1
2 3 2 3 so
so  1 s  so s
 1  
so 

2  3   2 6

then we have

so s 1
E HH  E v   E HH  E v  (so  s )
3 3 3
so s 1 1 8
E LH  E v   E LH  E v  (so  s )  (so  s ) 2  so s
3 3 6 2 3
2so 1 1 8
E SO  E v  E SO  E v  (so  s )  (so  s ) 2  so s
3 6 2 3

35
 Strained Valence Bands

so s
E HH  E v  
C11  C12 3 3
E C  E C ( 0)  2a e
C11 so s
E LH  E v  
3 3
2so
E SO  E v 
3

 C  C12 C  2C12 
E GHH   2a 11  b 11 e
 C11 C11 
 C  C12 C  2C12 
E GLH   2a 11  b 11 e
 C11 C11 

36
In here, e is mismatch,
and is negative for tensile
strain.
cf: Katz, InP and related
materials, Artech House 37
38
 (111)Coherent Strain
For (111) strain, we have exx= eyy= ezz, and exy= eyz= ezx , the Hamiltonian
becomes

 0  1  1
s  
H v  3a v e xx  3d v   1 0  1e xy
 1 1 0 
 

Eigenvalues are -2, 1, 1. Again, E1 and E2 are degenerate, while E3 is

separated from E1 and E2 by the crystal field splitting, which is given by,

s  3 3d v e xy

39
Compound Semiconductor
Technology
光電半導體技術
EE5115
921 U7150

1
Basic Properties of Compound
Semiconductors

Optical Properties

(Swaminathan Ch 1)

2
 E  
 dH Optical Properties
c dt
4  dE E  E 0 exp{i[ t  ( )]}
x
H  E
c c dt v
H  0 d 2E  dE  d 2 E
2
 2 4  2 2
E  0 dx c dt c dt
2   2
 2  i 2 4  2 
v c c
 d
E   (  H) 1 4 
c dt  i  2
v 2
c 2
c
 dE  d 2 E 1  (4 / )
  2 4  2 2  2 i
c dt c dt v 2
c c2
    E  (  E)   2 E 1 n 2  k 2 i 2nk
2
 2
 2
d 2E  dE  d 2 E v c c
2
 2 4  2 2
dx c dt c dt J. I. Pankove, optical processes in
semiconductors, p.87, 1971. 3
 Optical Properties
x
E  E 0 exp{i[ t  ( )]}
v
n c  n  ik
c 1 n ik
v   
nc v c c
n ik
E  E 0 exp{i[ t  (  ) x ]}
c c
inx kx
E  E 0 exp(it ) exp( ) exp( )
c c
P( x ) E 2 ( x ) 2kx
  exp(  )  exp(x )
P(0) E (0)2
c
2 k n: refractive index

c k: extinction coefficient
4
Optical Properties Moss rule:
An empirical relation
between the index of
n c  n  ik
refraction and the energy
 c  1  i 2 band gap of a semiconductor
1  n 2  k 2 ,  2  2nk


2k n 4 E g  77
c
(n  1) 2  k 2 range of n:
R
(n  1) 2  k 2
440 > n4 > 30
(n  1) 2
R 4.58 > n > 2.34
(n  1) 2

5
 2
(0)  LO 
  
()  TO 

6
Refractive index in III-V
The refractive index of GaAs is
governed by a strong resonance,
which is attribute to the E1+D1
transition, and two weak maxima at
E0 and E0+D0. E0 is the gap energy
G1c – G15v, and E1 is attributed to
L1c-L3v transition.

7
JAP 53, 655
Refractive index

JAP 53, 655

8
Sellmeier equation for refractive index

IEE Proc.-Optoelectron. 142, 37 (1995).

Origin Sellmeier equation:

9
 Absorption spectra
Absorption coefficient 
P(x) = P(x=0) exp(-x)

10
 2
(0)  LO 
  
()  TO 

11
Absorption above and below band edge

12
Dielectric constant model

The fitting uses the interpolation given

by Eq. (1.10) (Swaminathan Ch1). The
quantity (-1)/(+2) is used as the
parameter.

Clausius-Mosotti equation:
(Feymann Physics vol.2 Ch32)

n 2  1 Nq e2 fk
3 2 
n  2 m 0
k  2  i   2
k 0k

13
Dielectric constant model

Adachi proposed a generalized model of dielectric

constants of semiconductors based on models of
interband transitions. According to this model the real
part of the dielectric constant is expressed as a
function of photon energy E as
 3

 1  E g
 2

1 ()  A 0 f ( x )    f ( x 0 )   B0
 2  E g  D 0  
 
[2  (1  x ) 2  (1  x ) 2 ]
1 1

f (x) 
x2
x  E / Eg
E
x0 
Eg  D0
A 0 ( y)  8.40  3.40 y
B0 ( y)  6.60  3.40 y

Adachi, JAP 53, 5863 14

Dielectric constant model

EdE0
n 2 1 
E 02  E 2

E E E 2
 E 4  2 E 2
 E 2
 E 2

n 1 
2 d
 3 
d
ln  0 g

  
 Eg  E
2 2
E0 E0 
E
  3 2d 2
2E 0 ( E 0  E g )

EdE0
n 2 1 
E 02  E 2

15
 References
C. Kittel, introduction to solid state physics, 8th
Ed. Ch 14.
G. A. Abstreiter, M. Cardona, and A. Pinczuk,
Light scattering by free carrier excitation in
semiconductors, in Light scattering in solids IV
Electrical Properties
 Donors and Acceptors of III-V compounds

Be B C N O

Mg Al Si P S

… Zn Ga Ge As Se

… Cd In Sn Sb Te

… Hg Tl Pb Bi Po
Commonly used Impurities in III-V Compounds
Material Technology N-type P-type SEMI

GaAs/InP MBE Si Be
GaAs MOCVD Si, Ge Zn
GaAs/InP LPE Sn, Te, Si Ge, Si, Zn
GaAs Diffusion Zn
GaAs Implantation Si, S, Se, Te Zn, Cd, Be, Mg H, O
GaAs Bulk Growth Si Zn AsGa, Cr
InP Diffusion Cd
InP Bulk Growth Sn, S Zn Fe
InP MOCVD Si Zn Fe
InSb Diffusion Cd
GaN MBE/MOCVD Si Mg, Zn
   
Hall Effect F  qv  B
 
J  qnv

E H  R H JB
R H IB
VH 
d
rn
RH  
qn
rp
RH 
qp

Note: r is a constant that for spherical energy bands is given by 3/8; but more generally, it
lies somewhere between 0.5 and 1.5. For simplicity, it is often taken as 1, and indeed, if the
field is high enough, r=1.
Hall coefficient RH = r/
van der Pauw method
Resistivity and Mobility of Si
 Electron mobility for GaAs

Stillman and Wolfe, Thin Solid Films 31, 69, 1976.

Rode and Knight, PRB 3, 1971.

 Carrier Density
E top

n  N(E)F(E)dE
EC
1
2 (E  E C ) 2
3
N(E)  M c ( m de ) 2

 2
h 3

1
F(E) 
 E  EF 
1  exp 
 kT 

2  EF  EC 
n  NC F1  
 2  kT 
3

 2m dc kT  2
N C  2 2  MC
 h 
 E  EC 
n  N C exp F 
 kT 

Kittle, SSP Ch. 8

Sze, Physics of semiconductor devices, p.17
 Fermi-Dirac Integral

2  EF  EC 
n  NC F1  
 
2
kT 
3

 2m dc kT 
2

N C  2 2  MC
 h 
 E  EC 
n  N C exp F 
 kT 
 Charge neutrality
n  N C e  ( E C  E F ) / kT
p  N v e ( E F  E v ) / kT
N D N 0D 1
 1 
ND N D 1  2e ( E F  E D ) / kT
N A 1

N A 1  2e ( E A  E F ) / kT
n  N A  p  N D
NA ND
N C e ( E C  E F ) / kT   N v e ( E F  E v ) / kT 
1  2e ( E A  E F ) / kT 1  2e ( E F  E D ) / kT
Shockley diagram
 Carrier density as a function of reciprocal
temperature in high purity GaAs epilayers

Thin solid films 31, 69.

The deviation of the experimental data from the calculated curves at low temperatures
for samples 1-3 is due to two-band conduction involving the conduction band and an
impurity band which results from overlap of the impurity ground state wavefunctions
in these more heavily doped samples
 Band renormalization
1 1 1
DE G  AN  BN  CN
3 4 2

JAP 68 3747
Alloy scattering
Modulation doped structure
mobility without impurity scattering
 Phonon vibration mode
Raman scattering and IR absorption
Phonon modes

Zone center

Zone edge
Raman scattering in alloy
Bulk Growth
 Ga 鎵 (Gallium)

 Gallium was discovered by Lecoq de Boisbaudran in

1875. (Latin Gallia – France, gallus -- rooster)
 The abundance of Ga in the Earth’s crust is ~16.9 ppm.。
 The only two economic sources for Ga are as by product
of Al and Zn.
 MP: 29.78oC, BP: 2237oC
 The density of liquid state is higher than that of solid
state by ~3%. Corrosive.
 In 銦 (Indium)
In 1867 the German chemists Ferdinand Reich and
Hieronymous Theodor Richter discovered In.
Indigo (420-440 nm)靛藍
The abundance of In in the Earth crust is 0.25 ppm.
MP: 156.6oC, BP: 2072oC
Low melting alloy, semiconductor, dopants, mirror,
ITO。
MP is a defining fixed point on the international
temperature scale ITS-90。
 P 磷(Phosphorus)
 Phosphorus has several forms (allotropes) that have
different properties. The two most common allotropes are
white phosphorus and red phosphorus
 White phosphorus contains a P4 tetrahedron structure unit.
This P4 starts decomposing to P2 molecules at 800ºC. The
MP and BP of white phosphorus are 44.2°C and 280.5ºC,
respectively. It ignites at about 30ºC.
 Red phosphorus has a chain-like structure, which is much
more stable than white phosphorus. Its sublimation
temperature is 416 - 590ºC. Red phosphorus does not catch
fire below 260ºC.
 As砷(Arsenic)
 Arsenic exhibits three allotropies, grey, yellow, and
black arsenic. Grey arsenic adopts a layered structure,
a ruffled analogue of graphite, consisting of many
interlocked six-membered rings.
 硫化物有雄黃(realgar, a-As4S4) 又名雞冠石 硫化
砷，以及雌黃 (yellow orpiment, As2S3)
 氫化物為AsH3，無色劇毒。
 砷加熱後形成As2O3，為砒霜或白砷。
 用於殺蟲劑、農藥或特殊藥品。
Phase diagram of GaAs
Gibb’s free energy of
component i in the solution

 i   i0  RT ln  i X i

Simple solution model

G E  a(T )X i (1  X i )
G E
  G  XC
E
A
E
 a(T)X C2
X C
 EA  a(T)(1  X A ) 2
RT ln  A  a(T )(1  X A ) 2
RT ln  A X A  RT ln X A  a(T)(1  X A ) 2
The Nobel Prize in Physics 1946
"for the invention of an apparatus to produce extremely high
pressures, and for the discoveries he made therewith in the field
of high pressure physics"

Percy Williams Bridgman

USA
Harvard University
Cambridge, MA, USA
b. 1882
d. 1961
 LEC crystal pulling
Liquid encapsulated Czochralski crystal pulling
Professor Jan Czochralski, second from left, in a Warsaw University of
Technology lab in the nineteen thirties (courtesy of Z. Czochralska); Polish State
President, I. Moscicki - first from right.
 Crystalline Imperfections
Point defects
EL2 in GaAs
Dislocations
Stacking faults
ABCBABCABC extrinsic stacking fault
ABCA CABC intrinsic stacking fault
Twins
ABCABCBACBACBA
 Point defects
Foreign impurities: Donors and acceptors
Vacancies: sites which in the ideal structure
should be occupied by atoms but are not.
Interstitials: atoms in sites where they should
not be.
Misplaced atoms: atoms in sites which in the
ideal crystal are assigned to atoms of different
type.
Interstitials
 Dislocations
 Edge dislocation
 Screw dislocation
 Threading dislocation
Dislocation generation mechanisms

 Nonuniform heat flow during solidification and the

ensuing thermal stresses causing plastic deformation
 Condensation of excess point defects present near the
growth temperature to form prismatic dislocation loops
 Propagation of dislocations from defective seed crystal
or punching out of dislocations from foreign particles
and inclusions
EPD (Etch pits density)
In large diameter (>20 mm) LEC crystals,
dislocation generation is primarily caused by
thermal stresses because of large radial and
axial temperature gradients.
 Method to achieve low EPD
Use dopants which harden the semiconductor,
that is, increase the critical resolved shear stress
(CRSS).
GaAs: Si, Te
InP: S, Ge, Zn.
Isovalent impurities is preferred. In in GaAs, Sb,
Ga, As in InP.
 Distribution coefficient k
k = cs/cl, where cs and cl denote concentrations
of the impurity in the solid which is in
equilibrium with the melt.
The concentration in the crystal obeys the
normal freezing law
c = c0 k (1-g)k-1
where c0 is the initial concentration in the melt
and g is the fraction of the melt solidified.
 Distribution coefficient k
(1  g)c l  dg  c l  k
dc l   cl g dg 1-g
1  g  dg
(1  g)c l  dg  c l  k  c l (1  g  dg)

1  g  dg cs cl
dg  c l  (1  k ) dg  c l  (1  k )
 
1  g  dg 1 g
dc l d (1  g )
 (1  k )
cl 1 g
 c l (g)  c 0 (1  g) k 1
 c s (g)  c 0 k (1  g) k 1
Vacuum Technology
History
 Vacuum Unit

1 mbar = 1 dyne/cm2
10 mbar = 1 Nt/m2

1 Pa (Pascal) = 1 Nt/m2
1 mbar = 100 Pa （毫巴與百帕）
1 atm = 1,013 mbar

1 atm = 760 torr

1 mm = 1 mm Hg = 1 mtorr

1 mbar  0.75 torr

 Nature Vacuum

“Low Vacuum Technology”

Mosquitoes ?
Human 300 torr (by suction)
Octopus 100 torr

Vacuum in Space
Sea level 760 torr
~100 km one decade/15 km
10-3 torr @ 90 km
100 ~ 400 km one decade/100 ~ 200 km
10-10 torr @ 1,000 km
 Maxwell and Boltzmann Distribution

The fraction number of moleculars in the velocity

range between v and v+dv is,

1 dn 4 m 3/ 2 2 mv 2
fv   ( ) v exp( )
n dv  2kT 2kT

where n is the concentration, v is the velocity, m is

the mass, k is Boltzmann’s constant, and T is absolute
temperature in K.
 Molecular Drag

As a molecule collides a surface, experimental evidences show

that the molecule condenses on the surface, rests on it a given
time, and then it is reevaporated in a direction which is
independent of that of incidence.
If the surface is moving, it can transfer a velocity component to
the molecule.
 Gas Flow in viscous flow region

Reynold number: Re = rvD/

where r is gas density, v is the velocity, D is the
diameter of tube, and  is the viscosity.

Turbulent flow Re > 2,100

Q > 200 D

Laminar flow Re < 1,100

Q < 100 D

Q in torr l/s and D in cm.

 Gas Flow

Knudson number: D/l

where D is the diameter of tube and
l is the mean free path of moleculars

Viscous flow D/l >100

Intermediate flow 100>D/l >1

Molecular flow D/l <1

PV = n R T S’p
Q = C (P1-P2) Sp
P1 P2
Conductance: C (liter/sec)
Throughput: Q (torr liter/sec)
Q
Ultimate pressure: P = le/ Sp
Pumping speed: Sp (liter/sec)
Q Q
Leak rate: le (torr liter/sec) Q  C(P1  P2 )  C( '  )
Sp Sp
1 1 1 1 1 1
 '   '  
C Sp Sp S p C Sp
 Vapor in vacuum system
 Any liquid surface inside the vacuum system is a source
of vapor.
 As long as any liquid remains in the system, the
minimum pressure attainable is the vapor pressure of
that liquid at the existing temperature.
 If the vapor is compressed as a result of pumping or
handling operations, its pressure will increase only up to
the vapor pressure. (gas ballast)
 Reduction in the temperature reduces the vapor
pressure of any vapors. (Cold traps, cryogenic pumps)
 Rate of evaporation

Clausius-Clapeyron
equation

B
log Pv  A   C log T
T
M
W  5.83  10  2 Pv f
T

PV in torr
W in g/cm2/s
f: sticking coefficient
 Solubility and permeation
The passage of a gas into, through and out of a
solid barrier.
The process involves:
adsorption of gas on the surface
dissolution of gas in the outside surface
diffusion of gas across the solid
desorption of gas out of the surface
 Sorption
Adsorption: process whereby molecules are
attracted to and become attached to the surface
of a solid.
Physisorption or chemisorption
Absorption: gas which enters into the solid in
much the same manner as gas dissolving in a
liquid.
Adatom or admolecule
 Desorption and outgassing
 The gas which was
previously ad- or
absorbed begins to
desorb, i. e., to
leave the material.
 The generation of
gas resulting from
the desorption is
known as
outgassing.
 Pumps and Gauges
Foreline/Roughing pump Low vacuum gauge
• Rotary/Mechanical pump • McLeod Hg gauge
• Diaphragm pump • Gessiler tube
• Sorption pump • TC/Pirani gauge
• Roots pump • Baratron gauge

High vacuum pump

• Diffusion pump
High vacuum gauge
• Turbomolecular pump
• Ion gauge
• Molecular drag pump
• Cold cathode gauge
• Ion pump
• Ti sublimation pump
• Cryo-pump
Diaphragm pump

Pfeiffer MVP 040 diaphragm

pump

Exhaust pressure: 1100 mbar

Nominal pumping speed at 60 Hz:
2.5 m3/h
Ultimate pressure < 4 mbar
Sorption
pump
 Diffusion pump

Maximum pumping speed

Smax = 3.64(T/M)1/2A liter/sec T: temperature in K
S=HSmax M: molecular weight in g
H= 0.3~0.45 (Ho-factor)
Cold traps
Turbo-molecular Pumps
 2
1
3

1 Wake vacuum area

2 Area that allows gas particles from upper region to enter into
3 Area that allows gas particles from bottom region to enter into

Area 2 > Area 1

Because the Blade faces down, the gas particles trapped on the Blade surface have
more probability to be emitted to the bottom region. In order to reach a balance, the
pressure in the upper region must be lower than that in the bottom region. The ratio
is called compression ratio, usually ~10.
Molecular Drag Pump
Ion pumps
Diode pump and Triode pump
Argon pumping mechanism
 Titanium
sublimation pump
(TSP)
 Cryopump
A cryopump is a vacuum pump that traps
gases and vapors by condensing them on a
cold surface. They are only effective on
selective gases, depending on the freezing
and boiling points of the gas relative to the
cryopump's temperature.

80K: water, hydrocarbon

15K: N2, O2, Ar
Charcoal: He, H, Ne
 CTI cryopump
CTI cryopump consists of a cold head and a vacuum vessel. An 80K condensing array, a 15K array,
cold head station heaters, and an 80K radiation shield are located in the vacuum vessel. The cold
station heaters and 15K array are secured to the cold head, which is welded to the vacuum vessel. The
cold head provides cooling to the three arrays.
The cold head consists of a two-stage cold head cylinder (part of the vacuum vessel) and drive unit
displacer assembly, that together produce closed-cycle refrigeration at temperatures that range from 60
to 120K for the first-stage cold station to 10 to 20K for the second-stage cold station, depending on
operating conditions. Within the drive unit displacer assembly, the drive unit actuates the displacer-
regenerator assembly located in the cold head cylinder and thereby controls the flow of helium into
the cold head. Within the drive unit are located the crankcase and drive motor, which is a direct-drive
constant-speed motor, operating at 72 rpm on 60Hz power and 60 rpm on 50 Hz power.
During operation, high pressure helium from the compressor enters the cold head at the helium
supply connector, and flows through the displacer-regenerator assembly, crankcase, and motor
housing before exiting through the helium gas return connector and returning to the compressor.
Helium expansion in the displacer-regenerator assembly provides cooling at the first and second stage
cold stations.
The 80K condensing array, as shown in Figure 1-3, condenses water and hydrocarbon vapors. The
15K array condenses nitrogen, oxygen, and argon while the specially processed charcoal of this array
traps helium, hydrogen, and neon.
Baratron Gauge
Conflat flange
Before bake-out
After bake-out
Effect of LN2 Filling

Before LN2

After LN2
Effect of TSP firing

Before TSP firing

TSP is firing

After TSP firing

 Liquid phase epitaxy

 LPE
 Phase diagram
 Gibb’s free energy
 Simple solution model
 Immiscibility (phase separation)
 Epitaxial Growth
Liquid phase epitaxy (LPE)
Molecular beam epitaxy (MBE)
Vapour phase epitaxy (VPE)
Chemical vapour deposition (CVD)
Phase diagram of GaAs

Gibb’s free energy of

component i in the solution

 i   i0  RT ln  i X i

Simple solution model

G E  (T )X i (1  X i )
G E
  G  XC
E
A
E
 (T)X C2
X C
 EA  (T)(1  X A ) 2
RT ln  A  (T )(1  X A ) 2
RT ln  A X A  RT ln X A  (T)(1  X A ) 2
 Thermodynamic Principles
Enthalpy:
Chemical equilibrium or nearly equilibrium
between melting and solidification H=Q+PV
Gibb’s free energy:
A(l)  B(l)  AB(s)
G=H-TS
 A (T )   B (T )   AB (T )
S is entropy
Gibb’s free energy (or chemical potential)
for pure material and solution

a is activity, X is mole fraction, and  is the extra free

 i    RT ln a i
0
i energy due to interaction between components in
  i0  RT ln  i X i the solution and is called activity coefficient.

 0A  RT ln  A X A   0B  RT ln  B X B   0AB
 0  (RT ln  A X A  RT ln  B X B )  RT ln( A X A  B X B )
 Enthalpy:
Thermodynamic Principles H=Q+PV
For constant pressure and composition: Gibb’s free energy:

T2
G=H-TS
 i0 (T2 )   i0 (T1 )    Si0 (T)dT S is entropy
T1

For a reversible process

CP Cp is the heat capacity at constant pressure. Cp=dQ/dT.
dS  dT
T
T
C 0P ,i
Si0 (T)  Si0 (T2 )   dT
T2
T
T2 T
C 0P ,i
 (T2 )   (T1 )  S (T2 )(T1  T2 )   
0
i
0
i
0
i '
dT 'dT
T1 T2
T Neglect the Cp term.
T2 T
C 0P ,i cf O. Kubaschewski,
 (T2 )   (T1 )  S (T2 )(T1  T2 )   
0 0 0
dT 'dT
i
T' Metallurgical
T1 T2
thermochemistry, pp. 392,
 Si0 (T2 )(T1  T2 ) Oxford, 1967.
 0 (TF )   0 (T1 )  Si0 (TF )(T1  TF )
 Enthalpy:
Thermodynamic Principles H=Q+PV
Gibb’s free energy:
G=H-TS
S is entropy
 0 (TF )   0 (T)  Si0 (TF )(T  TF )

 RTF ln[ A (TF )  B (TF ) / 4]  RT ln[  A (T )  B (T )X A X B ]   Si0 (TF )(TF  T )

Strictly regular solution model for binary solution

G E  (T)X A X B  (T)X B (1  X B )
G E
  G  XB
E
A
E
 (T)X 2B
X B
 EA  (T)(1  X A ) 2
RT ln  A  (T)(1  X A ) 2
RT ln  A X A  RT ln X A  (T)(1  X A ) 2

(TF )
RTF ln[ A (TF )]  RTF ln[  B (TF )] 
4
RT ln[ A (T )]  (T )X 2B
RT ln[ B (T )]  (T)X 2A
(T)  a  bT
Thermodynamic Principles

 0 (TF )   0 (T )  Si0 (TF )(T  TF )

 RTF ln[  A (TF )  B (TF ) / 4]  RT ln[ A (T )  B (T )X A X B ]   Si0 (TF )(TF  T )

(TF )
  RTF ln 4  (T)(X 2A  X 2B )  RT ln(X A X B )  Si0 (TF )(TF  T )
2
1 b
(T )(X 2A  X 2B  )  RT ln(4X A X B )  [Si0 (TF )  R ln 4  ](TF  T )  0
2 2
1
(T )(X 2A  X 2B  )  RT ln(4X A X B )  SF (TF  T)  0
2

XA  XB  1
2
 1
2(T ) X A    RT ln[4X A (1  X A )]  SF (TF  T )  0
 2
 Lattice match at
growth temperature

Thermal expansion
coefficient
InP: 4.5610-6/C
In0.53Ga0.47As: 5.6610-6/C

Lattice match at growth

temperature is free from
misfit dislocations.
HL1

Lattice pulling effect

G. B. Stringfellow, JAP 43, 3455

投影片 18

HL1 Hao-Hsiung Lin, 2023/2/24

 Immiscibility
Gibb’s free energy of the product (epilayer) A1-xBxC1-yDy
F  F( x , y)
 1 (1  x )(1  y)  2 (1  x ) y  3 x (1  y)  4 xy
 13 (1  x ) x (1  y)   24 (1  x ) xy  12 (1  x ) y(1  y)   34 x (1  y) y
 RT{(1  x ) ln(1  x )  x ln x  (1  y) ln(1  y)  y ln y}
where  is the sum of the first and second nearest neighbor interaction energy of i-
th binary compound and  is the pseudobinary interaction parameter.
(Onabe, JJAP, 22, 663, 1983)
Free energy per mole can be expressed in the DLP model as
(Stringfellow, J. Crystal Growth, 27, 21, 1974)

K
GS   2 .5
 RT[ x ln x  (1  x ) ln(1  x )  y ln y  (1  y) ln(1  y)]
a0

K=1.15107 cal/mole Å2.5

immiscibility
 Miscibility gap

Spinodal pts
(inflexion pts)
C

B’
A
B
Binodal pts

 Point A: tangent line is always below the curve.

==> stable
 Point B: tangent line is below the curve locally,
but intersects the curve at point B’. ==>
metastable
Meijering, Philips Res Rep, 333, 1950.  Point C: tangent line is above the curve locally.
==> unstable
(between two spinodal points)
 Spinodal points

Spinodal pts
(inflexion pts)

Binodal pts

Meijering, Philips Res Rep, 333, 1950.

In the case of ternary alloy, there is only one composition and the spinodal points
are the inflexion points which are defined as

 2G(x )
0
x 2
In the case of quaternary alloy, there are two compositions and the spinodal points
are defined as follows by setting the determinant of Hessian matrix to zero:
2
 G ( x , y)  G ( x, y)  G ( x , y) 
2 2
    0
x 2
y 2
 xy 
 Immiscibility of InPSb
alloy
1200
Tcrit = 1319K Spinodal pts
Binodal curve
(inflexion pts)
1000
o
942 C
Temperature ( C)

800 Binodal pts

o

Spinodal curve
• A large miscibility gap in the InP1-xSbx
600
o alloy extending from x=0.02 to 0.96 at
527 C
450oC was reported and it is difficult to
grow single phase InPSb crystal lattice
o
450 C
400
matched to InAs substrate.

200
• Critical temperature 1319K is very close
to the melting point of InP (1335K),
and far above that of InSb (800K).
0
0.0 0.2 0.4 0.6 0.8 1.0 • Paper reports indicated that InPSb can
Sb mole fraction in InPSb be grown by MOCVD for all values in
spite of the miscibility gap.

Spinodal and Binodal curves calculated by DLP

model
Substrate effect in Free Energy diagram

As suggested by Stringfellow a
strain energy contribution
should be added to the GFE
in case of nearly lattice-
matched growth. This term is
proportional to Da2, where
Da is the lattice mismatch.
Convexity can thus be locally
obtained even inside the
unstable region. This is true
even if the strain energy itself
is very small compared to the
total mixing GFE (only the
second derivative determines
Quillec, Appl.Phys.Lett., 40, 325, 1982. stability).
 Strain effect on Phase
Spinodal pts
(inflextion pts) Separation
Free energy per mole can be expressed in the DLP model as
(Stringfellow, J. Crystal Growth, 27, 21, 1974)

Binodal pts K
GS   2.5
 RT[ x ln x  (1  x ) ln(1  x )]
15000 a0
2nd derivative of free energy, (G )" (cal/mole)

Strain energy can be expressed as

10000
(Nahory, J. Electrochem Soc., 125, 1053, 1978)
S

1  a
5000 G strain  2 Vm ( ) 2
1  a
0
  12 C 44

This calculation indicates that coherent strain

-5000
with strain energy energy can drive the epilayers out of unstable
region. However, the effect can only be applied
without strain energy
-10000
on layer thinner than its critical thickness. For
0.0 0.2 0.4 0.6 0.8 1.0 bulk materials, a strain accommodation factor
Sb composition of InPSb should be considered.

2nd derivative of Free energy with and without considering

the coherent strain on InAs substrate at 450°C
Molecular Beam Epitaxy

2024/02/19
Molecular Beam Epitaxy
 MBE system
 Multi-chamber vacuum UHV system
Deposition chamber/Preparation chamber/Load
lock/Analysis chamber
 Pumping system
ion pump/turbo pump/diffusion pump/cryopump
 Cryopanel and liquid nitrogen
 Deposition: Knuson cells
Thermal cell, cracking cell, hot-lip cell, cold-lip cell
 Substrate heater
 Ion gauge/RGA/Rheed
VG V150 MBE for production
MBE in SPACE

Compound Semiconductors, 2, 1, 8, 1995

Wake Vacuum
Ion gauge
Shutter
 Rate of evaporation

Clausius-Clapeyron
equation

B
log Pv  A   C log T
T
M
W  5.83  10  2 Pv f
T
 A B T (dPv/dT)/Pv
B Al 11.79 1.594104 1185oC 0.0172/oC
log Pv  A   C log T
T Ga 11.41 1.384104 950oC 0.0213/oC
M
W  5.83 10  2 Pv f In 11.23 1.248104 812oC 0.0244/oC
T
Si 12.72 2.13104 1150oC 0.0242/oC
PV in mtorr Be 12.01 1.647104 800oC 0.0329/oC
W in g/cm2/s Sb2 11.15 8.63103 300oC 0.0605/oC
 8  hc
( )d   d
Pyrometer
hc
5 ( e  kT
 1)
-4
10

-5
10
1000K

(T)d
10
-6 900K

800K
-7
10

-8
10
0 1000 2000 3000 4000 5000
(nm)
Pyrometer oscillation
Molecular Beam Epitaxy

2024/02/20
 A B T (dPv/dT)/Pv
B Al 11.79 1.594104 1185oC 0.0172/oC
log Pv  A   C log T
T Ga 11.41 1.384104 950oC 0.0213/oC
M
W  5.83 10  2 Pv f In 11.23 1.248104 812oC 0.0244/oC
T
Si 12.72 2.13104 1150oC 0.0242/oC
PV in mtorr Be 12.01 1.647104 800oC 0.0329/oC
W in g/cm2/s Sb2 11.15 8.63103 300oC 0.0605/oC
Group III atoms Group V atoms

JAP 74, 255

Arthur’s investigation showed:
1. Ga has a sticking coefficient of
unity below 750K.
2. At 885K, Ga has a surface lifetime
of 10 sec.
3. As has negligible sticking coefficient
unless Ga atoms are present on the
surface.
At ~700oC, Ga and As
are congruently
sublimated.
As T>500oC, P is
seriously sublimated
from InP and In is left
on the surface because
of the large discrepancy
in their pressures.
 Growth rate and Composition Control
Grp III – in the region of unity sticking coeff,
Grp III flux defines the growth rate of the alloy.
GR = Grp III flux / Grp III concentration
IIIA-IIIB-V alloy:
GR(IIIA-IIIB-V) = GR(IIIA-V)+GR(IIIB-V)
Mole fraction x for IIIA-V
x = GR(IIIA-V)/GR(IIIA-IIIB-V)
 Growth rate and Composition Control
III-VA-VB alloy:
Growth rate is controlled by Grp III flux.
Grp Vs’ sticking coeff<1. They compete each
other and mole fractions are defined by their
incorporation rates.
There are two growth models for multi-Grp-V
alloys.
Kinetic model
Thermodynamic model
Unit vectors in real space and reciprocal space

C Real n (r )   n G exp(iG  r )
space G


B
 
 B C
a  2   
 A  B C
A
 
 C A

c Reciprocal b  2   
space BC A

 
b  AB
c  2   
CAB

a
 n (r )   n G exp(iG  r )
G

Scattering amplitude:

F   dVn ( r ) exp[i(k  k )  r ]   dVn (r ) exp(ik  r )

   dVn G exp[i(G  k )  r ]
G
Rheed Pattern
Atomic layer epitaxy (ALE)
Migration enhanced epitaxy (MEE)
Phase-locked
epitaxy (PLE)
Vapor Phase Epitaxy

2020/07/09
 Halide and Hydride VPE

Main Chemical Reaction:

As4 + 4 GaCl + 2H2  4 GaAs + 4 HCl
Halide process
4 AsCl3 + 6H2  As4 + 12 HCl
HCl + Ga  GaCl + ½ H2
Hydride process
4 AsH3  As4 + 6H2
HCl + Ga  GaCl + ½ H2
 Trichloride VPE

The main disadvantage related to trichloride VPE is that it is difficult to grow III-V alloys
by that means. For example it is not possible to grow AlGaAs since the growth
temperature of AlAs, 1100C, differs too much from that of GaAs, 750 C. Another
disadvantage is that both GaAs and InP substrates must be misoriented by 3-6 degrees
away from (100) to attain specular epitaxial layer surfaces. Growth facets occur otherwise.
Wall deposits have plagued crystal growers because such deposits uncontrollably modify
the partial pressures of reactants in the growth zone.
Intentional Doping in VPE grown GaAs

N type:
H2S, H2Se, SiH4, GeH4

P type:
Zn (in a separate furnace)
 Residual impurity in VPE grown GaAs

Residual impurity: Si from quartz tube

Quartz reacts with HCl and produces chlorosilanes

SiO2 + (4-n) H2 + n HCl  SiH4 - nCln + 2 H2O

Chlorosilanes then transport to GaAs surface and

deposit Si on GaAs
(n-2) H2 + SiH4 - nCln  Si (in GaAs) + n HCl

Increase the flowrate of HCl can reduce the incorporation

reaction of Si, and results in high purity GaAs.
Halide system: 1013 ~ 1014/cm3 m77K = 200,000 cm2/Vs
Hydride system: 1014/cm3 m77K = 160,000 ~ 200,000 cm2/Vs
Since gas velocity must be zero at the surface of a growing layer, a layer-like region above
the substrate in which the gas velocity is small is expected. The thickness of the so-called
boundary layer can be altered by varying the tilt angle of the substrate.
 Steady-state growth
 Mass-transport limited j  h( Ng  N0 )
j  kN 0
 Reaction-rate limited
k  k 0 e  E a / RT
adsorption, desorption,
dx j N g  hk 
surface reaction    
dt n n hk

k  h Reaction-rate limited
dx j kN g
 
dt n n

h  k Mass-transport limited
dx j hN g
 
dt n n
 At high temperature,
even though the system
is usually in the mass-
Endothermic 吸熱 exothermic 放熱
transfer-limited region,
Mass transfer the reaction itself is the
region rate limiter, and causes
Kinetic a decrease in the
control region growth rate with
increasing temperature.
GaAs halide and
hydride processes
exhibit this
characteristics
Crystal growth of InGaAsP

High temperature →

P4+4HCl+4InP ↔ 4InCl+2P4+2H2
Hydride VPE
In(l) + HCl ↔ InCl + (1/2)H2
 MOVPE
2011/01/05
References:

The technique was pioneered by Manasevit in 1968.

H. M. Mansevit, Appl. Phys. Lett. 11, 156 (1968).

M. J. Ludowise (HP Lab), Metalorganic chemical vapor deposition of

III-V semiconductors, J. Appl. Phys. 58, R31, 1985.

D. Schmitz (Aixtron GmbH), Low-pressure MOVPE of InP-based

compound semiconductors. (Ch. 4 of InP and related materials, edited
by A. Katz, Artech House, Boston, 1992)

G. B. Stringfellow, Organometallic vapor-phase epitaxy (2nd Ed. Elsevier

Inc., 1999)
Methyl (Me) -CH3
Ethyl (Et) -CH2CH3
Isopropyl (iPr) –CH(CH3)2
Tertiarybutyl (iBu) -C(CH3)3
Ga(CH3)3 +AsH3 ==> GaAs + CH4

In(CH3)3 + PH3 ==> InP + CH4

 MOCVD growth mechanism
Gas (precursors) transport (fluid mechanics)
Boundary layer diffusion
Precursors cracking and absorption
Adatoms migration, desorption and
incorporation
By-products desorption
By-products diffusion and transport
Veeco E475
Commercial MOCVD
 Step motion

Figure 3.10. Schematiccross-sectionsof (a) [110] and (b) [110] steps:

underlying monolayer, ( • ) ;growing monolayer, (O). (After Asai , J. Crys.
Growth 80 425, 1987)
Figure 3.8. Lateral spread and vertical thickness Figure 3.9. Lateral and vertical growth rates as
as a function of growth time for GaAs layers a function of growth temperature for GaAs
grown by OMVPE. (After Asai, J. Crys. layers grown by OMVPE. (After Asai, J. Crys.
Growth 80 425, 1987) Growth 80 425, 1987)
Residual carrier
density
Low V/III ratio ==> p-type
High V/III ratio ==> n-type

Reduce V/III ratio ==> increase density

of group V vacancies

The result suggests that the carriers are

from group IV residual impurities.

Group IV on group V site ==> p-type

Group IV on group III site ==> n-type
The drop in doping efficiency with x is attributed
to the increase in the acceptor binding energy.