Ohmic contacts

Common techniques to make ohmic

contacts

Ohmic contacts should be

Choose metal so that its work function

Fmetal is close to that of semiconductors
Fsemi (thermal ionic)
Insert thin layer of narrow bandgap
material between metal and
semiconductor
Increase the doping level near the
semiconductor surface as high as
possible (tunneling assisted)
Low contact resistance (< 10-6 cm2)
Thermally stable (does not degrade at
elevated temperature or react with
oxygen), which requires no phase
change or no phase change leading to
high resistance
Smooth morphology
Compatibility with the whole device
process

Both semiconductors and metal

sources should be CLEAN!

Michaelson, IBM J. of R&D, 1978

Slide # 1

Common Techniques for Ohmic contacts

(i) Ohmic contact by band alignment
Evac

Evac
EF

EF

M < +Ec-EF = S
For n-type
semiconductor.
Reverse for p-type

B = S - M
n doped
B = band bending

(ii) Ohmic contact by high doping

Usually for compound semiconductors
the ohmic contact by band alignment is
hard to realize due to surface states and
Fermi pinning. For p-type, the problem
is caused by unavailability of metals
with large enough work function
High n-type doping required for ohmic
contacts to n-type semiconductors,
which can also be realized by interfacial
layer reaction chemistry

n-doped

n+ doped

Electrons from
conduction band
can move very
easily to the
metal and vice
versa by
tunneling
Slide # 2

Ohmic on n-GaN

Possible metals: Ag, Nb, Ti, Al, In, Ta, Cr

Ag: poor adhesion

Nb: extremely easy to oxidize thus difficult to
process
Ti: formation of TiN (intermetallic) and high N
vacancies in GaN -> good! But easy to
oxidize need a stable cap like Au
Al: formation of AlN (not intermetallic) and high
N vacancies in GaN -> ok! Also easy to
oxidize Au cap is necessary
In: most popularly used for quick contacts
Ta: studied by Qiao et al. 5.7e10-7 cm2 on
AlGaN/GaN (2001); but others could not
reproduce the results
Others: also studied but not as good as the
one below
As deposited or alloy: generally alloyed unless
doping near the surface is very high!

Popular schemes: Ti/Al/Ni/Au

Ti/Al bilayer: formation of N vacancies, TiN,

Al3Ti (thermally very stable ); but ratio of Ti/Al
has to be carefully controlled (~1/2.5) 
Add high conductive and protective layer of Au,
but Au diffuses easily
Add Ni as diffusion barrier (decent, other
metals were tried, Pd and Pt were worse)
State-of-art: 0.1-0.2 cm2 (~ 10-8 cm2 )

Lim et al, APL 78, 3797(2001)

Liu et al, Solid State Electronics, 42, 677(1998)

Slide # 3

Schottky contacts
Schottky contacts are
formed when

Evac

Evac

M
 Doping in the
EF
semiconductor is not very
high i.e. > ~5x1018 cm-3
 The metal work function is
greater than the n- type
M > +Ec-EF = S
semiconductor work
For n-type semiconductor and
function
reverse for p-type
 The metal work function is
lower than p-type
Schottky contact Electrons from
Bn =
semiconductor work
conduction band
M -
function
or in the metal
faces barrier to
 Very high density of
free movement,
surface states pinning the
n doped
Fermi level at the surface
and tunneling is
w.r.t. the conduction band
also not easy
(Example: GaAs)
Slide # 4

Conduction mechanisms in schottky contacts

Thermionic emission
Electrons emit over the barrier
Low probability of direct tunneling
Valid for low doping (ND < ~ 1017 cm-3)

Thermionic-field emission
Electrons use thermal energy to tunnel trough the
thin barrier in the upper end of the conduction
band
Valid for intermediate doping (~ 1017 cm-3 < ND <
~ 1018 cm-3)

Field emission
Direct tunneling, as depletion region is very narrow
Valid for heavy doping (ND > ~ 1018 cm-3); almost
ohmic

Leakage current
High probability of defect-assisted tunneling and
simple conduction
Occurs in poor material/interface quality; dislocations

Slide # 5

Thermionic emission current: Schottky diode

I-V characteristics
Typical I-V
characteristics

Forward bias

Reverse bias

Schottky diode I-V equation:

J = J0 (e qV / kT 1), where J0 is the
saturation current density given by

q
J o = A*T 2 exp Bn T = temperature, A* = effective Richardsons constant
kT
6
Slide #

Schottky on n-GaN

Experimentally shown
very weak surface
pinning
Surface cleanness has
been heavily
investigated, however

Thermal stability is
IMPORTANT

External cleaning is
generally sufficient to
achieve decent
Schottky
Leakage is largely due
to dislocations

Ni does not react with

GaN below ~ 600 C
Pd reacts with GaN at ~
400-500 C
W and Rd ~ 600 C

The higher the schottky barrier, the lower

the leakage current
Using polarization in nitrides i.e.
GaN/AlGaN/GaN structure, the schottky
barrier can be made larger

Liu et al, Solid State Electronics, 42, 677(1998)

Slide # 7

Electrical properties of dislocations in MBE-grown n-GaN

(Ed Yu --- UCSD)
Pure screw dislocations can be highly conductive in MBE-grown n-GaN:
topography:

1m

current:

1m

[E. J. Miller, D. M. Schaadt, E. T. Yu, C. Poblenz, C. Elsass, J. Speck, J. Appl. Phys. 91, 9821 (2002).]

Edge and mixed dislocations typically contain negative charge in dislocation core:
high current leakage
in Schottky contacts
Screw: conducting, uncharged
Edge: nonconducting , e charge
Mixed: nonconducting, e charge

scattering, local
carrier depletion

[B.S. Simpkins, E.T. Yu, P. Waltereit, J.S. Speck, J. Appl. Phys. 94, 1448 (2003).]

Slide # 8

Mitigation of dislocation-induced leakage

currents in MBE n-GaN (Ed Yu, UCSD)
AFM

AFM

2m

2m

current

current

AFM

1m

current
~10A

1m

with electrochemical
process

NaOH solution V = 30V

pH = 13.1
I ~ 1-10mA
T = 30C
t = 1000s

~100pA

b = 0.800.02V
n = 1.740.01

b = 0.860.02V
n = 1.130.02

AFM
unmodified
area = 1.2310-4cm2

1m

[E. J. Miller, D. M. Schaadt, E. T. Yu, P. Waltereit, C. Poblenz, and J. S. Speck, Appl. Phys. Lett. 82, 1293 (2003).]

Slide # 9

Ohmic to p-GaN

Similar techniques like

ohmic to n-GaN have
been tried, but:

P-GaN/Ni/Au annealed in
air (N2/O2) proved to be
one of the best:

rC~ 10-6 cm2

Why?

rC~ 10-3 cm2

After annealing, new

phases form: NiO, NiGa-O with Au particles,
GaN
NiO is p-semiconductor
with high Ni vacancies

Continuing challenges:

Transparency to visible
and UV
Ohmic to p-AlGaN
Tunneling junction
contacts
Ho et al. JAP 85, 4491 (1999)

Slide # 10

Another Contact Metal for p-GaN

The absence of a metal with a sufficiently high work function. The band gap of
GaN is 3.4 eV, and the electron affinity is 4.1 eV, but metal work functions are
typically ~ 5 eV
The relatively low hole concentrations in p-GaN due to the deep ionization level of
the Mg acceptor ~170 meV
The tendency for the preferential loss of nitrogen from the GaN surface during
processing, which may produce surface conversion to n-type conductivity.

TEM
image

Palladium gallide creates Ga vacancies that

reduce contact resistances
Temperature and time of anneal also important
Slide # 11

Schottky to p-GaN

Schottky (Ni) on as grown GaN:Mg (MOCVD) --- quasi-ohmic (higher Mg near the surface?)
Schottky (Ni) on etched GaN:Mg --- rectifying (tunneling and defect-assisted tunneling still
significant thus it is difficult to extract barrier height and Richardson constant from I-V)

Slide # 12

Schottky contact characterization

Current-Voltage (IV) measurements
*

J0 = A T e

q Bn
kT

. VF vs. J intercept gives J0 and Bn

kT A*T 2

ln
=
q J 0

Capacitance-Voltage (CV) measurements

1
q s N D
2 s
2 (Bn V )
(Bn V ) C =
W=
C
2( Bn V )
qN D

So the intercept of 1/C2 vs. V gives the barrier height

Photoelectric measurements (by photon incident

on the schottky contact; this is very accurate)
Photocurrent R is related to the barrier height as
R ~ hv qBn So the intercept gives the barrier height

Slide # 13

Evaporation systems
Contact Metallization (Ti, Al, Ni, Au etc)
Metal Electron-Beam Evaporation System

Sample

Target Metal Source

with e-beam

Rapid Thermal Annealing System

from 20 oC to 1000 oC in seconds

Slide # 14

Ohmic contacts: n-type or undoped nitride

Standard recipe for ohmic contact:
Ti/Al/Ti/Au or Ti/Al/Ni/Au deposition. Ti/Al thickness ratio is important
Annealing at 800 900 C for about 1 min for alloying. Alloying temperature
and alloying time are important factors controlling contact resistance.

Ti/Al/Ni/Au

Since TiN and AlN are formed by reaction

between the nitride layers and Ti or Al, Nvacancies are created, which can dope the
contact region and create ohmic contact

Slide # 15

Specific contact resistivity and sheet resistance

For any semiconductor device there are two main resistances:
Contact resistance
Semiconductor resistance

d
Z
t

Product of contact resistance Rc and area

A is called specific contact resistivity c:
c =

1
J
V
V =0

( . cm2)

(Can also be expressed in terms of . mm)

Semiconductor layer resistivity :

1
=
en

. cm)

Sometimes semiconductor resistance is

expressed in terms of sheet resistance sh

=
sh =
t (e n ) t

(/ )

The total semiconductor resistance is then

given by
d

1
d
()
Rs = dx =
A0
Zt
Slide # 16

Ohmic contact characterization:

Transmission line method (TLM)
I(x)
L

I(x+x)
shx/Z

V(x)

c/(Zx)

V(x +x)

dI
V ( x) Z
=
dx
c
I
I
dV
= s = sh
dx
Zt
Z

d 2 I I ( x)
c
2 = 2 , where LT =
sh
dx
LT

The solution for I(x) is given as:

is called the transfer length.

I ( x ) = Ae x / LT + Be x / LT

Now putting the boundary condition I(x = L) = 0, and finding the

solution for V(x), we can find the contact resistance as the ratio of the
input voltage and input current as: RC = V (x = 0 ) I ( x = 0)
Slide # 17

Transmission line method (TLM) II

L
coth
The contact resistance Rc is then given by: RC =
ZLT
LT

For L >>LT , we have, RC = C

Ohmics
ZLT

Z
When the following conditions are further
d
satisfied, d << Z and t << LT (to avoid
t
current spreading in the sides or into the film),
sh = /t
sh d
Then, RTot = 2 Rc + Rs = 2 Rc +
Z
Putting Rtot = 0, and using the relation c = Rc LT Z , we have,

d ( RT = 0) = 2 LT . So, the transfer length can be found from the

intercept of the total resistance on the x-axis.
Note that the contact resistivity is not given by the product of the
contact resistance and the total contact area, but by the product of
contact resistance, width Z, and transfer length LT.

Slide # 18

Measurement technique
100

Typical measurement set up

B1205 UV LED
n-TLM

Current, mA

50

4um
6um
8um
10um
12um
14um
16um

-50

-100
-2

-1

Voltage, V
40

Slope =
sh/Z

30

What is wrong in
this measurement?

Resistance, Ohm

Plot of total resistance

vs. distance

B1205 UV LED
n-TLM

20

Y =14.51607+1.13839 X
Rc=7.258
LT=6.373um

10

c=6.93*10-5-cm2

Rsh=170.7/sq
0

10

15

gap, um

Slide # 19

20