SPUTTER HEATING IN IONIZED METAL PHYSICAL VAPOR DEPOSITION +

UNIVERSITY OF ILLINOISOPTICAL AND DISCHARGE PHYSICS

SPUTTER HEATING IN IONIZED METAL PHYSICALVAPOR DEPOSITION+

ICOPS99_00 TITLE

Junqing Lu* and Mark Kushner**

*Department of Mechanical and Industrial Engineering

**Department of Electrical and Computer Engineering

University of Illinois at Urbana-Champaign

June 1999

+Supported by SRC, TAZ


AGENDA

ICOPS99_01 AGENDA

• Introduction to IMPVD

• Description of sputter model

• Overview of Hybrid Plasma Equipment Model

• Validation of sputter model

• Results and discussions:• Al target, with and without sputter heating• Comparison between Al and Cu target, with sputter heating

• Summary


IONIZED METAL PHYSICAL VAPOR DEPOSITION (IMPVD)

ICOPS99_02 REACTOR

• 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.

• 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

+


SPUTTER GAS HEATING IN IMPVD

ICOPS99_03 INTRO

• An important process in IMPVD reactors is gas heating due to momentum and energy transfer from sputtered metal atoms to background gas atoms.

• The degree of gas heating is dependent on the magnetron power, ICP power, and sputter yield.

• This gas heating affects the background gas density, ion flux to the target, and subsequently the 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).


SCHEMATIC OF 2-D/3-D HYBRID PLASMA EQUIPMENT MODEL

SRCRM9801

S(r,z)E (r,z)s

v(r,z)

PLASMA CHEMISTRY MONTE CARLO SIMULATION

ETCHPROFILEMODULE

FLUXES

E (r,z, ),

B(r,z, )

Θ φ

φ

ELECTRO-MAGNETICS

MODULE

CIRCUITMODULE

I,V (coils) EΘ

S(r,z), T (r,z) (r,z)µ

e

FLUID-KINETICS

SIMULATION

HYDRO-DYNAMICSMODULE

ENERGYEQUATIONS

SHEATHMODULE

N(r,z) E (r,z, )s

φ(r,z), I (coils), J(r,z, )σ φ

ELECTRONBEAM

MODULE

ELECTRONMONTE CARLOSIMULATION

ELECTRONENERGY EQN./BOLTZMANN

MODULE

NON-COLLISIONAL

HEATINGLONG MEANFREE PATH(SPUTTER)

SIMPLECIRCUIT

MAGNETO-STATICSMODULE

B(r,z)

= 2-D ONLY

= 2-D and 3-D

EXTERNALCIRCUITMODULE

MESO-SCALE

MODULE

(r,z, )φΦ

R

(r,z, )φΦ

V(rf),V(dc)

VPEM: SENSORS, CONTROLLER, ACTUATORS

SURFACECHEMISTRY

MODULE

(r,z, )φΦ

RFDTD

MICROWAVEMODULE

ICOPS99_04 HPEM


IMPROVEMENT TO SPUTTER ALGORITHMS

ICOPS99_05 IMPROVE

• The energy-dependent yield is computed from: • Experimental data* (preferred), • Semi-Analytical expression**. *Masunami et al., At. Data Nucl. Data Tables 31, 1 (1984) .**Mahan and Vantomme, JVST A 15, 1976 (1997).

• Energy of the emitted atoms obeys a cascade distribution (Thompson’s law for Eion ≈ 100’s eV):

F(E) =2 1+

Eb

ΛEi

2EbE

Eb + E( )3 for E ≤ ΛEi

0 for E > ΛEi

• To better model the IMPVD process, the following improvements have been made to the sputter algorithms in the HPEM:

• Ion energy-dependent yield• Ion energy-dependent kinetic energy of the emitted atom• Momentum and energy transfer between the sputtered atoms and the background gas atoms

ΛEi: maximum recoil energy.

0 100 200 300 400 500 6000.0

0.2

0.4

0.6

0.8

1.0

1.2

Ar Ion Energy, eV

Experiment

Analytical


OVERVIEW OF SPUTTER MODEL

ICOPS99_06 SPUTTER

• The sputter model employs a kinetic Monte Carlo approach:

• Sputter rate = Yield • (ion flux + fast neutral flux) • Sputtered metal atoms are emitted from the target with a cosine distribution, and a kinetic energy having a cascade distribution.

• Collisions of the sputtered atoms with the background gas atoms are assumed to be elastic.

ion flux

fast neutral flux

θ

ArAl Al

Target

wafer


OVERVIEW OF SPUTTER MODEL (Continued)

ICOPS99_07 SPUTTER2

• Recording of metal atoms: • Thermalized --> Green’s Function • In-flight --> local density • Incorporation of statistics into fluid equations:

• Quantities of interest generated:

• Metal atom densities in the plasma, • Metal flux to the wafer, • Gas heating terms.

Rate of Changeof Momenta and

EnergySourceTerms

ThermalizedMetal Atoms

FluidEquations


MODEL VALIDATION: Al IMPVD

ICOPS99_08 VALID

• HPEM IMPVD model was validated by comparing with experiment*.

*Dickson and Hopwood, J. Vac. Sci. Technol. A 15(4), 1997, p. 2307

• Operating conditions:

• ~ 240 W magnetron. • 0 V on wafer.

• Al density (1011 cm-3) at radius of 4 cm:

Dis. (cm) HPEM EXPERIMENT2 3.4 4.0-7.54 1.7 3.0-5.08 1.5 1.5

10 1.5 1.0-1.5

6.0E11

1.0E10

1.0E8

Thermal Al Density (cm -3) In-Flight Al Density (cm -3)

Al

RADIUS (cm)0 5 10 15515 10

400 W ICP Power30 mTorr Ar

MagnetAl

Substrate

Target to waferdistance = 12 cm

Inlet

Outlet

Coils

Target


SPUTTER HEATING: Ar DENSITY

ICOPS99_09 AR

• The IMPVD reactor of interest: external coil reduces erosion, larger target area increases sputtered atom flux.

• Operating conditions:

• 0.5 kW ICP • 1.0 kW magnetron• 30 V rf on substrate

• 30 mTorr

• The minimum Ar density with heating is 75% of that without heating.

• The bias at the target is: -177 V with heating, -168 V without heating.

• The Ar+ density decreases due to sputter heating. More voltage is required to maintain the 1 kW magnetron power with a lower ion current.

RADIUS (cm)

With Sputter Heating Without Sputter Heating

0 5 10 1551015

Ar, cm-3

6.5E14

2.0E14

5.0E13

Coils

FaradayShield


SPUTTER HEATING: ELECTRON TEMPERATURE

ICOPS99_10 TE

• The electron temperature profiles are similar, and > 3 eV throughout the plasma regions.

• The maximum electron temperature below the target is caused by energetic electrons from secondary emission.

• The relatively high electron temperature next to the Faraday shield is due to the skin effect in ICP power deposition.

RADIUS (cm)


0 5 10 1551015

Te, eV

5.3

3.2

1.5


SPUTTER HEATING: ELECTRON DENSITY

ICOPS99_11 E

• The electron density profiles are also similar, indicating the sputter heating does not significantly affect the electron density.

• The high electron density below the target is due to magnetron confinement.

• The maximum electron density is off-center, due to the magnetron effect.

RADIUS (cm)


0 5 10 1551015

e, cm-3

4.6E11

7.0E10

1.0E10


SPUTTER HEATING: Ar+ FLUX

ICOPS99_12 ARPFLUX

• The Ar+ flux is high at the magnetron confinement region.

• The maximum flux below the target generates sputter track on the magnetron target.

• The Ar+ flux above the substrate is less than half of that below the target because the rf bias on the substrate is only 1/5 of the dc bias on target.

RADIUS (cm)


0 5 10 1551015

Ar+ Flux,

1/cm-2-s

1.6E17

6.0E16

-3.0E16


SPUTTER HEATING: Al DENSITY

ICOPS99_13 AL

• The maximum Al density with sputter heating is half that without heating, though its gradient to the substrate is smaller.

• The magnetron power is 1 kW in both cases, so the sputtered atom fluxes and Al inventory should be approximately the same.

RADIUS (cm)


0 5 10 1551015

Al, cm-3

1.7E12

4.0E10

1.0E9

• Sputter heating redistributes Al in the reactor to conserve the inventory.

10 9

10 10

10 11

10 12

10 13

0 2 4 6 8 10 12 14 16Distance Below Target (cm)

Radius = 6 cm

Without Sputter Heating

With Sputter Heating


SPUTTER HEATING: Al+ DENSITY

ICOPS99_14 ALP

• The maximum Al+ density is approximately the same.

• The Al+ distribution is determined by the Al atom distribution and the mean free path for ionization.

• The background gas density is reduced by 25% by sputter heating, so the mean free path for Al ionization increases.

• Because the metal flux to wafer consists mostly of ions, the depositing metal flux with heating is about 15-25% larger.

RADIUS (cm)


0 5 10 1551015

Al+, cm-3

5.2E10

6.0E9

5.0E8


Ar AND e DENSITIES vs MAGNETRON AND ICP POWER

ICOPS99_15 POWER

• The trends shown agree with expectation: more power, more heating, more rarefaction and more ionization.

• The Ar and e densities change linearly with magnetron power, to first order. More nonlinearity occurs at low ICP powers.

0.0

1.0

2.0

3.0

4.0

5.0

0.0 0.5 1.0 1.5 2.0

Magnetron Power (kW)

ICP Power

0.5 kW

1.0 kW

1.5 kW

0.0

1.0

2.0

3.0

4.0

5.0

6.0

0.0 0.5 1.0 1.5 2.0


ICP Power

1.5 kW

1.0 kW

0.5 kW


Ar AND e DENSITIES vs MAGNETRON AND ICP POWER

ICOPS99_16_AR_E_CU_AL

• The Ar density for the Cu target is smaller than that for Al target. The sputter yield for Cu is twice that of Al, hence Ar gas is more rarefied for Cu target.

• The electron densities for the two targets are approximately the same, because electron density is strongly affetced by magnetron and ICP powers, and insensitive to gas rarefaction.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0.0 0.5 1.0 1.5 2.0


0.5 kW ICP power

Al Target

Cu Target

0.0

1.0

2.0

3.0

4.0

5.0

0.0 0.5 1.0 1.5 2.0

Magnetron Power (KW)

0.5 kW ICP power

Al Target

Cu Target


SUMMARY OF IMPVD STUDY

ICOPS99_17 CONCLU

• The magnetron confinement of the electrons leads to higher ionization of background gas atoms below the target.

• The background gas density decreases due to sputter heating, which redistributes the metal species inventory.

• Sputter heating increases as the yield of the target material increases.

• The electron density scales with the ICP and the magnetron power, and is insensitive to the sputter heating.

SPUTTER HEATING IN IONIZED METAL PHYSICAL VAPOR DEPOSITION +

Documents

Transcript of SPUTTER HEATING IN IONIZED METAL PHYSICAL VAPOR DEPOSITION +