SPUTTER-WIND HEATING IN IONIZED METAL PVD+
Junqing Lu* and Mark Kushner**
*Department of Mechanical and Industrial Engineering
**Department of Electrical and Computer Engineering
University of Illinois at Urbana-Champaign
October 1999
+Supported by SRC, TAZ
AVS99_00 TITLE
AGENDA
• Introduction to IMPVD
• Overview of Hybrid Plasma Equipment Model
• Description of sputter model
• Validation of sputter model
• Results and discussions for Al IMPVD
• Investigating the effects of sputter heating by comparing to the results
without sputter heating
• Plasma properties
• Depositing Al fluxes
• Ar and electron densities as functions of magnetron and ICP power
• Summary
AVS99_01 AGENDA
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
IONIZED METAL PHYSICAL VAPOR DEPOSITION (IMPVD)
• Ionized Metal PVD (IMPVD) is being developed to fill deep vias and trenches for
interconnect, and for deposition of seed layers and diffusion barriers.
• In IMPVD, a second plasma source is used to ionize a large fraction of the
the sputtered metal atoms prior to reaching the substrate.
MAGNETS
TARGET
(Cathode)
ANODE
SHIELDS
• 10-30 mTorr
Ar buffer
PLASMA
INDUCTIVELY
COUPLED
COILS
• Typical Conditions:
SECONDARY
PLASMA
ION
+
NEUTRAL
TARGET ATOMS
Ar+ + M > Ar + M+
e + M > M+ + 2e
WAFER
• 100s V bias on
target
• 100s W - a few
kW ICP
• 10s V bias on
substrate
SUBSTRATE
BIAS
AVS99_02 REACTOR
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
DEPOSITION PROFILES: ATOMS vs IONS
• Ions are able to fill deep trenches because their spread in angles is narrowed
by the rf bias.
METAL
ATOMS
METAL
IONS
METAL
VOID
METAL
SiO2
SiO2
RF
BIAS
AVS_03 PROFILE
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
SPUTTER GAS HEATING IN IMPVD
• In IMPVD processes, two types of atoms produced in the sputtering process
transfer momentum and energy to background gas atoms, (sputter heating)
• Sputtered metal atoms
• Reflected neutral atoms produced by the incident ions
• The degree of sputter heating increases with:
• Magnetron power
• Sputter yield
• Collision cross section of the gas
• This sputter heating affects
• Background gas density
• Ion flux to the target
• Sputtered atom flux and the depositing metal flux
• To investigate the effects of sputter heating, we incorporated a sputter algorithm
into a Hybrid Plasma Equipment Model (HPEM).
AVS99_04 INTRO
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
SCHEMATIC OF 2-D/3-D HYBRID PLASMA EQUIPMENT MODEL
E s (r,z)
v(r,z)
MAGNETOSTATICS
MODULE
CIRCUIT
MODULE
S(r,z)
B(r,z)
EΘ (r,z, φ ),
B(r,z, φ )
I,V (coils)
EΘ
ELECTROMAGNETICS
MODULE
PLASMA CHEMISTRY
MONTE CARLO
SIMULATION
FLUXES
= 2-D ONLY
ETCH
PROFILE
MODULE
= 2-D and 3-D
FLUIDKINETICS
SIMULATION
ELECTRON
MONTE CARLO
SIMULATION
HYDRODYNAMICS
MODULE
ELECTRON
BEAM
MODULE
ELECTRON
ENERGY EQN./
BOLTZMANN
MODULE
NONCOLLISIONAL
HEATING
FDTD
MICROWAVE
MODULE
S(r,z), T e (r,z)
µ (r,z)
ENERGY
EQUATIONS
Φ (r,z,φ )
V(rf),V(dc)
Φ (r,z,φ )
R
N(r,z)
E (r,z, φ )
s
MESOSCALE
MODULE
SHEATH
MODULE
LONG MEAN
FREE PATH
(SPUTTER)
Φ (r,z,φ )
R
σ (r,z), I (coils), J(r,z,φ )
EXTERNAL
CIRCUIT
MODULE
SURFACE
CHEMISTRY
MODULE
SIMPLE
CIRCUIT
VPEM: SENSORS, CONTROLLER, ACTUATORS
SRCRM9801
AVS99_05 HPEM
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
IMPROVEMENT TO SPUTTER ALGORITHMS
• To better model the IMPVD process, the following improvements have been made
to the sputter algorithms in the HPEM
• Ion energy-dependent yield for sputtered atoms
• Ion energy-dependent kinetic energy for sputtered and reflected atoms
• Momentum and energy transfer from sputtered and reflected atoms to the
background gas atoms
• The energy-dependent yield of the sputtered atoms is*
αQKs n (ε)
2.8
0.42
1−
E
/
E
(
th
i ) , Ei > E th
Us (1+ 0.3U s se (ε))
Y(E i ) = 
0, Ei ≤ E th
*Masunami et al., At. Data Nucl. Data Tables 31, 1 (1984) .
• The effective yield of the reflected neutrals
• 0.9 for high kinetic energy
1.2
0.8
0.4
0
0
200
400
800
Incident Ar + Energy (eV)
• 0.1 for thermal energy
AVS_06 IMPROVE
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
ENERGY DISTRIBUTIONS
OF THE SPUTTERED AND THE REFLECTED ATOMS
• Energy of the emitted atoms obeys a cascade distribution:
(Thompson’s law for Ei ≈ 100’s eV):
2
 

U
UsE
21+ s 
, E ≤ ΛEi
F(E) =   ΛEi  (Us + E) 3

0, E > ΛE i
Us
surface binding energy.
ΛEi maximum recoil energy.
• Reflected neutral energies are obtained from TRIM* and MD simulations.
*Ar+ incident on Al, David Ruzic, Depart. of Nuclear Engineering, UIUC.
• Kinetic energies of reflected neutrals are curve fitted into thermal
accommodation coefficient (α) vs. incident ion energy.
α=
AVS_07 ENERGY
Ei − Er
Ei − E T
• For Ar+ incident on Al target, α ≈ 0.95.
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
OVERVIEW OF SPUTTER MODEL
Target
• The sputter model employs a kinetic
Monte Carlo approach:
θ
• Sputter rate =
Yield • (ion flux + fast neutral flux)
Al
Al
Ar
• Sputtered atoms and reflected
neutrals are emitted with a cosine
distribution in angle.
• Collisions of sputtered atoms and
reflected neutrals with the background
gas are assumed to be elastic.
AVS99_08 SPUTTER
wafer
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
OVERVIEW OF SPUTTER MODEL (Continued)
• Recording of sputtered metal atoms and reflected neutrals
• Thermalized
Green’s Function
• In-flight
local density
• Incorporation of statistics into fluid equations
Rate of Change
of Momenta and
Energy
Source
Terms
Fluid
Equations
Thermalized
Atoms
• Quantities of interest generated
• Metal atom densities in the plasma
• Metal flux to the wafer
• Gas heating terms
AVS99_09 SPUTTER2
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
MODEL VALIDATION: Al IMPVD
• HPEM IMPVD model was
validated by comparing with
experiment (Dickson and Hopwood, J.
Vac. Sci. Technol. A 15(4), 1997, p. 2307)
Thermal Al Density (cm -3 ) In-Flight Al Density (cm -3 )
Target to wafer
distance = 12 cm
400 W ICP
240 W Magnetron
30 mTorr Ar
Al Al
Target
Magnet
6.0E11
Inlet
• Good agreement with
experimental data:
Coils
• V-I characteristic
HPEM
Voltage (V)
255
Current (A)
0.94
1.0E10
EXP.
240
1.0
1.0E8
Substrate
15
10
5
Outlet
0
RADIUS (cm)
5
10
15
• Ionization fraction of Al at r = 4 cm
Distance (cm)
10 (plasma)
12 (flux)
AVS99_10 VALID
HPEM
15%
73%
EXPERIMENT
10-15%
70%
• Predicted and measured Al
densities = 1011 cm-3
at r = 4 cm, 8-10 cm below
target.
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
SPUTTER HEATING: Ar DENSITY
• Modified TEL IMPVD tool.
• Operating conditions
Without Sputter Heating
With Sputter Heating
Target to wafer
distance = 15 cm
• 0.5 kW ICP
• 1.0 kW magnetron
• 30 V rf on substrate
• 30 mTorr Ar
6.8E14
Magnet
Al Target
Ar, cm-3
Inlet
• The minimum Ar density
Coils
4.6E14
• Decreases by 30% with
sputter heating
• Occurs below target due to
sputter heating and charge
exchange
• Contribution to sputter heating
Faraday
Shield
2.5E14
Outlet
Substrate
15
10
5
0
5
10
15
RADIUS (cm)
• Reflected neutrals 2/3.
• Sputtered metals atoms 1/3.
AVS99_11 AR
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
MAGNETIC FIELD, Ar+ FLUX AND DENSITY
Without
Sputter Heating
With Sputter Heating
B-field (G)
0
5
10
RADIUS (cm)
15
Ar+ (cm-3)
500
6.6E11
25
3.4E11
1
0.2E11
15
10
5
0
5
10
15
RADIUS (cm)
• Exponential decay of the magnetic field
away from the magnets.
• The target voltage is -178 V with
sputter heating, and -168 V without.
• Peaks of Ar+ fluxes due to the magnetic
cups.
• Decreasing Ar+ density below
target leads to a lower ion current
and a higher voltage for the same
magnetron power
• Sputter heating increases the Ar+ density
near the center of reactor.
AVS99_12 ARPFLUX
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
SPUTTER HEATING: ELECTRON TEMPERATURE
Without Sputter Heating
With Sputter Heating
• Te(max) below the target is
caused by energetic secondary
electrons and joule heating.
Te, eV
5.3
• Te decreases by 1 eV with
sputter heating. This agrees
with observations (Dickson, Qian,
and Hopwood, JVST A 15 (2), 340 (1997)).
2.6
• The electron temperature is
high near the coils, due to
inductive heating.
0
15
10
5
0
5
10
15
RADIUS (cm)
AVS99_13 TE
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
SPUTTER HEATING: Al DENSITY
• The maximum Al density with
sputter heating is only 1/3 of that
without, due to the longer mean
free path.
Without Sputter Heating
With Sputter Heating
Al, cm-3
1.7E12
• Since the magnetron power is the
same, the amount of sputtered Al
atoms is about equal in both cases.
4.0E10
• Sputter heating redistributes Al in
the reactor to conserve the total
inventory of Al atoms.
1.0E9
• The Al density with sputter heating
decays much slower from the
target to the substrate.
15
10
5
0
5
10
15
RADIUS (cm)
AVS99_14 AL
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
SPUTTER HEATING: Al+ DENSITY
• The Al+ density with sputter heating has a broader peak and decreases slower
toward the substrate, due to the longer mean free path in a more rarefied gas.
• Note the depletion of Al+ ions by the target bias with sputter heating.
With Sputter Heating
Without Sputter Heating
Al+, cm-3
5.2E10
6.0E9
5.0E8
15
10
5
0
5
10
15
RADIUS (cm)
AVS99_15 ALP
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
SPUTTER HEATING: DEPOSITING Al FLUX
• The total depositing metal flux consists mostly of Al+ and thermal Al atoms .
• The direct Al flux is negligible because of long throw distance (15 cm,
~10 mfp), and Al* is depleted by ionization.
• The magnitude of the Al flux with sputter heating is >2 times that without,
while the ionization fraction decreases to 67-86% from above 90%.
• Without sputter heating
• With sputter heating
3
3
Total
2
2
Al+
Total
1
1
Thermal Al
Al+
0
0
2
4
6
Radius (cm)
AVS99_16 AL_FLUX
8
10
Thermal Al
0
0
2
4
6
8
10
Radius (cm)
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
Ar DENSITIES vs MAGNETRON AND ICP POWER
• Both the minimum and the reactor averaged Ar densities decrease with increasing
magnetron power due to increasing sputter heating.
• The minimum Ar densities converge at high magnetron power
• The specific power density decreases due to longer stopping distance.
• The minimum gas density occurs right below the target, and the heat loss to
the target increases as the gas temperature increases.
• Ar - Minimum
5
• Ar - Average
5
0.5 kW ICP
4
4
0.5 kW ICP
1.0 kW
3
2
1.0 kW
3
1.5 kW
2
1.5 kW
1
1
0
0
0.5
1.0
1.5
Magnetron Power (kW)
AVS99_17 POWER
2.0
0
0
0.5
1.0
1.5
2.0
Magnetron Power (kW)
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
ELECTRON DENSITY vs MAGNETRON AND ICP POWER
• The electron density significantly increases when magnetron power is increased
from zero due to low-ionization potential metal atoms .
• As the magnetron power further increases, the electron density saturates due to
decreasing Al ionization fraction.
• At constant magnetron power, electron density increases linearly with ICP power.
8
1.5 kW
6
1.0 kW
• Average electron
density
4
0.5 kW ICP
2
0
0
0.5
1.0
1.5
2.0
Magnetron Power (kW)
AVS99_18 POWER_E
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS
SUMMARY OF SPUTTER HEATING STUDY
• Sputter heating from both sputtered metal atoms and reflected neutrals
significantly rarefies the background gas, thus increasing the mean free path
for transport of the sputtered atoms and redistributing metal species in the
reactor.
• Sputter heating decreases the ionization fraction of the depositing metal flux,
but increases its magnitude, similar to operating at lower pressure.
• Sputter heating should NOT be neglected in the modeling of IMPVD processes.
• The addition of small amount of metal atoms with low ionization potential
significantly increases the electron density.
• Electron density tends to saturate with increasing magnetron power since gas
rarefaction reduces the ionization fraction of the metal atoms.
AVS99_19 CONCLU
UNIVERSITY OF ILLINOIS
OPTICAL AND DISCHARGE PHYSICS