scaling of hollow cathode magnetrons for metal … · • hollow cathode magnetrons (hcm) are...
TRANSCRIPT
-
GEC99C01
SCALING OF HOLLOW CATHODE MAGNETRONS FORMETAL DEPOSITIONa)
Gabriel Fontb)
Novellus Systems, Inc.San Jose, CA, 95134 USA
and
Mark J. KushnerUniversity of Illinois
Dept. of Electrical and Computer EngineeringUrbana, IL, 61801, USA
October 1999
a) Work supported by Novellus and SRCb) Present Address: University of Texas at San Antonio
-
AGENDA
GEC99C15
University of Illinois Optical and Discharge Physics
• Introduction to Ionized Metal PVD and Hollow Cathode Magnetrons
• Description of Model
• Plasma properties of IMPVD of Cu using a HCM
• Comparison to Experiments
• Deposition and Target Erosion
• Concluding Remarks
-
COPPER INTERCONNECT WIRING
GECA99C12
University of Illinois Optical and Discharge Physics
• The levels of interconnect wiring in microelectronics will increase to 8-9over the next decade producing unacceptable signal propogation delays.
• Innovative remedies such as copper wiring are being implemented.
Ref: IBM Microelectronics
-
UNIVERSITY OF ILLINOISOPTICAL AND DISCHARGE PHYSICS
IONIZED METAL PHYSICAL VAPOR DEPOSITION (IMPVD)
CACEM98M03
• In IMPVD, a second plasma source is used to ionize a large fraction of the the sputtered metal atoms prior to reaching the substrate.
• Typical Conditions: • 10-30 mTorr Ar buffer • 100s V bias on target • 100s W - a few kW ICP • 10s V bias on substrate
TARGET(Cathode)
MAGNETS
ANODESHIELDS
PLASMA ION
WAFER
INDUCTIVELYCOUPLEDCOILS
SUBSTRATE
SECONDARYPLASMA
BIAS
e + M > M+ + 2e
NEUTRALTARGET ATOMS
+
-
UNIVERSITY OF ILLINOISOPTICAL AND DISCHARGE PHYSICS
IMPVD DEPOSITION PROFILES
CACEM98M04
• In IMPVD, a large fraction of the atoms arriving at the substrate are ionized.
• Applying a bias to the substrate narrows the angular distribution of the metal ions.
• The anisotropic deposition flux enables deep vias and trenches to be uniformly filled.
SiO2
METALATOMS
METALIONS
METAL
-
HOLLOW CATHODE MAGNETRONS
GEC99C02
University of Illinois Optical and Discharge Physics
•• Hollow Cathode Magnetrons (HCM) are typically cylindrical devices withinverted metal cups and floating shields. Production tools have 300 mmsubstates.
MAGNETS
IRON PLATE
IRON RING
CATHODE
SHIELD (FLOATING)
SUBSTRATE (FLOATING)
SHIELD (GROUND)
GROUND
GAS INLET
PUMPRADIUS (cm)
0 3 6 912 369 120
HE
IGH
T (c
m)
10
5
15
20
25
35
30
• Typical Operating Conditions:
• Pressure: 5-10 mTorr
• B-field: 100-300 G atcathode
• Voltage: 300-400 V
• Substrate: Floating or biased
• Buffer gas: Ar, 50-100 sccm
• Power: 2-5 kW
-
HOLLOW CATHODE MAGNETRONS-IMPVD
GEC99C03
University of Illinois Optical and Discharge Physics
•• HCM - IMPVD devices use magnetic confinement in the cathode for high iondensities and sputtering, and cusp B-fields to focus the plasma.
ION CONFINEMENT AND SPUTTERING
PLASMA FOCUSING
DIFFUSION (SOME GUIDING)
• B-Field Vectors • B-Field Magnitude
-
HYBRID PLASMA EQUIPMENT MODEL
GECA99C13
University of Illinois Optical and Discharge Physics
LONG MEAN FREE PATH (SPUTTER)
Φ(r,θ,z,φ) V(rf), V(dc)
PLASMA CHEMISTRY MONTE CARLO SIMULATION
ETCH PROFILE MODULE
FLUXES
E(r,θ,z,φ),
B(r,θ,z,φ)ELECTRO-
MAGNETICS MODULE
CIRCUIT MODULE
I,V (coils) E
FLUID- KINETICS
SIMULATION
HYDRO- DYNAMICS MODULE
ENERGY EQUATIONS
SHEATH MODULE
ELECTRON BEAM
MODULE
ELECTRON MONTE CARLO
SIMULATION
ELECTRON ENERGY EQN./
BOLTZMANN MODULE
NON- COLLISIONAL
HEATING
SIMPLE CIRCUIT
MAGNETO- STATICS MODULE
= 2-D ONLY = 2-D and 3-D
EXTERNAL CIRCUIT MODULE
MESO- SCALE
MODULE
VPEM: SENSORS, CONTROLLER, ACTUATORS
SURFACE CHEMISTRY
MODULEFDTD
MICROWAVE MODULE
Bs(r,θ,z)
S(r,θ,z), Te(r,θ,z)
µ(r,θ,z)
Es(r,θ,z,φ) N(r,θ,z)
Φ(r,θ,z)
R(r,θ,z)
Φ(r,θ,z)
R(r,θ,z)
J(r,θ,z,φ) I(coils), σ(r,θ,z)
Es(r,θ,z,φ)
v(r,θ,z) S(r,θ,z)
-
PHYSICS MODELS USED IN HCM SIMULATIONS
GECA99C14
University of Illinois Optical and Discharge Physics
• Monte Carlo secondary electrons emitted from cathode
• Fluid bulk electrons (Te from energy equation with Boltzmann derivedtransport coefficients)
• Continuity, momentum and energy for all heavy species (multi-fluid, slipand temperature jump boundary conditions)
• Long-mean-free path transport for sputtered atoms with sputter-heating
• Implicit Poisson solution simultaneous with continuity and momentum forelectrons
• Species in Model:
e, Ar, Ar(4s), Ar+, Cu(2S), Cu(2D), Cu(2P), Cu+
-
UNIVERSITY OF ILLINOISOPTICAL AND DISCHARGE PHYSICS
DESCRIPTION OF SPUTTERING MODEL
PEUG99M03
• Energy of the emitted atoms (E) obeys the cascade distribution, an approximation to Thompson’s law for Ei ≈ 100’s eV:
F(E) =2 1+
EbΛE i
2EbE
Eb + E( )3 , E ≤ ΛE i
0, E > ΛEi
where
Λ = 4m imT / m i + mT( )2
subscripts: b ~ binding, i ~ ion, T ~ target.
• The sampling of sputtered atom energy E from the cascade distribution gives
E =EbΛE i RN
Eb + ΛE i 1− RN( )
where RN is a random number in interval [0,1].
ion flux
fast neutral flux
θ
ArAl Al
Target
wafer
-
HCM- Cu IMPVD: ELECTRON DENSITY, TEMPERATURE
GECA99C04
University of Illinois Optical and Discharge Physics
• E-Density(6.6 x 1012 cm-3)
• E-Temperature(4.8 eV)
• Electron densities in excess of1012 cm-3 are generated inside andin the throat of the cathode.
• Fractional ionizations are ≤≤ 10%.
• Electron temperatures peak at 4-5eV in the cathode, and are a feweV downstream.
• Ar, 6 mTorr, 150 sccm, 325 V, 160 G
-
HCM- Cu IMPVD: ELECTRON SOURCES
GECA99C05
University of Illinois Optical and Discharge Physics
• E-beam(3 x 1017 cm-3s-1)
• Bulk(5 x 1018 cm-3s-1)
• Secondary electron emission andbeam ionization produce electronsources in the cathode.
• Ionization in the bulk plasma,however, is the major electronsource.
• Ar, 6 mTorr, 150 sccm, 325 V, 160 G
-
HCM- Cu DENSITIES
GECA99C06
University of Illinois Optical and Discharge Physics
• Cu(1.9 x 1013 cm-3)
• Cu+
(1.5 x 1012 cm-3)
• Neutral copper [Cu(2S) + Cu(2D)]undergoes rapid ionization andrarefaction in the throat of thecathode.
• The majority of neutral copper isin metastable Cu(2D).
• Fractional ionization of Cu in thebulk is 10's %.
• Ar, 6 mTorr, 150 sccm, 325 V, 160 G
-
HCM- GAS TEMPERATURE
GECA99C07
University of Illinois Optical and Discharge Physics
• Average Gas Temperature (max = 1530 K)
• Gas heating occurs dominantly by symmetriccharge exchange between Ar neutrals andions, and produces significant rarefaction.
• Slip between neutrals and disparate rates ofcharge exchange produce differences in theirmaximum temperatures:
Ar: 1546 K Cu: 900 K
• Ar, 6 mTorr, 150 sccm, 325 V, 160 G
-
ELECTRON DENSITY AND TEMPERATURE vs EXPERIMENT
GECA99C09
University of Illinois Optical and Discharge Physics
MODEL
EXPERIMENT
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0 2 4 6 8 10 12
RADIUS (cm)
EXPERIMENT
MODEL
ELE
CT
RO
N T
EM
PE
RA
TU
RE
(eV
)
• Electron Density • Electron Temperature
• Langmuir probe measurements of electron density (2.8 cm abovesubstrate) show densities and temperatures more highly peaked on axis.
• The model underpredicts the "jetting" of plasma from the throat of thecathode which likely results from long-mean-free path effects not capturedby the ion model.
• Ar, 6 mTorr, 150 sccm, 325 V, 160 G
-
ELECTRON DENSITY, TEMPERATURE vs BIAS
GECA99C18
University of Illinois Optical and Discharge Physics
• HCM devices have significantly "less steep" I-V characteristics comparedto conventional magnetron discharges.
• For example, downstream plasma densities and power deposition scalenearly linearly with bias. Electron temperatures are weak functions ofbias.
0.0
0.4
0.8
1.2
1.6
2.0
2.4
2.8
3.2
320 340 360 380 400 420 440
CATHODE BIAS (v)
EXPERIMENT
MODEL
EL
EC
TR
ON
T
EM
PE
RA
TU
RE
(e
V)
0
5
10
15
20
25
320 340 360 380 400 420 440
CATHODE BIAS (v)
EXPERIMENT
MODEL
ELE
CT
RO
N D
EN
SIT
Y (
101
1 cm
-3)
-
HCM- ION AND CU FLUXES TO THE SUBSTRATE
GECA99C08
University of Illinois Optical and Discharge Physics
10
8
6
4
2
0
FLU
X (
101
7 c
m-2
s-1
)
Ar+
Cu (Direct)
Cu+ Cu(2D)
Cu(2S)
• Ion flux to the substrate isdominated by Ar+
• Direct non-thermal Cudominates the Cu fluxes tothe substrate.
• The Cu flux is 25-30%ionized.
• Ar, 6 mTorr, 150 sccm, 325 V, 160 G
-
Cu DEPOSITION RATE
GECA99C10
University of Illinois Optical and Discharge Physics
MODEL
EXPERIMENT ( x 1.33)
• Cu Deposition Rate (A/min)
• The off-axis peak in directsputter flux produces an off-axis peak in the depositionrate.
• Ar, 6 mTorr, 150 sccm, 325 V, 160 G
-
TARGET EROSION
GECA99C11
University of Illinois Optical and Discharge Physics
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0 5 10 15
EXPERIMENT MODEL
POSITION ALONG TARGET (cm)
EROSION
DEPOSITION
CORNER
0
L
HCM TARGETB-FIELD
• Magnetron trapping and the resulting large ion fluxes to the target occursat the side walls.
• Net deposition of sputtered Cu occurs on the end walls while the sidewalls experience net erosion.
• Ar, 6 mTorr, 150 sccm, 325 V, 160 G
-
CONCLUDING REMARKS
GEC99C16
University of Illinois Optical and Discharge Physics
• Hollow Cathode Magnetrons (HCM) are capable of producing metaldeposition with plasma densities of 1012 cm-3.
• The HCM operates somewhat as a remote plasma source with moderateelectron temperatures near the substrate compared to the cathode interior.
• The configuration of the magnetic field is important in at least tworespects:
• Jetting of plasma at throat of cathode• Erosion profile inside the cathode
• Small HCMs having higher power densities experience significantrarefaction which ultimately limits their capability to produce high plasmadensities and deposition rates.
• Large HCMs (10-20 cm diameter) which avoid these problems have beendesigned and built based on these scalings studies.