MODELING OF SEASONING OF REACTORS: EFFECTS OF ION ENERGY DISTRIBUTIONS TO CHAMBER WALLS* Ankur Agarwal a) and Mark J. Kushner b) a) Department of Chemical and Biomolecular Engineering University of Illinois, Urbana, IL 61801, USA aagarwl3@uiuc.edu b) Department of Electrical and Computer Engineering, Ames, IA 50011, USA mjk@iastate.edu http://uigelz.ece.iastate.edu 53 rd AVS Symposium, November 2006 *Work supported by the SRC and NSF
AGENDA Seasoning of plasma reactors Approach and Methodology Hybrid Plasma Equipment Model Surface Chemistry Model Si etching in Ar/Cl 2 Effect of seasoning reactor walls on etch rates Sputtering of quartz window Concluding Remarks ANKUR_AVS06SS_Agenda
SEASONING OF PLASMA REACTORS Deposition on reactor walls during a process changes surface reactivity (e.g., seasoning) Ref: E.S. Aydil et al., JES 150, G418 (2003) Seasoning changes reactive fluxes to substrate. To control wafer-to-wafer variability: Clean the seasoned chamber following each wafer. Season the chamber prior to process step. ANKUR_AVS06SS_01 Ref: G. Cunge et al., PSST 14, S42 (2005)
SEASONING: CHALLENGES Understanding the role that chamber condition plays in manufacturing variability is important. Modeling Challenges: Energetic Etch products; products not thermalized Ion energy distributions are different on different surfaces of reactor. Reactivity of surface changes during a process as well as run-to-run. In this talk, seasoning of reactor will be computationally investigated while accounting for variation of IEDs and reactivity on all surfaces. ANKUR_AVS06SS_02
HYBRID PLASMA EQUIPMENT MODEL (HPEM) Electromagnetics Module: Antenna generated electric and magnetic fields Electron Energy Transport Module: Beam and bulk generated sources and transport coefficients. Fluid Kinetics Module: Electron and Heavy Particle Transport, Poisson s equation Plasma Chemistry MC Module: IEADs to surfaces Surface Chemistry Module: Surface coverage and reactive sticking coefficients. ANKUR_AVS06SS_03
SURFACE CHEMISTRY MODULE (SCM) Surface Chemistry Module (SCM) is an integrated module in HPEM to address surface reactions on substrate and walls of the reactor. The SCM: Uses the reactant fluxes from the plasma modules. Applies a user defined reaction mechanism Updates surface sticking and product reflection coefficients for the plasma species. Calculates surface coverages and etch rates. ANKUR_AVS06SS_04
SURFACE REACTION PROBABILITIES: IMPLEMENTATION Surface reaction probabilities are a function ion energies. Processes with threshold energies are extremely sensitive to details of IEADs: ε ave < ε th but rate could still be finite. Compute IEADs to all surfaces in reactor continuously during time evolution of process. Obtain surface probabilities vs position from convolution with p(x,ε) p( x) = f ions f ( ε, x) p( ε ) dε ions ( ε, x) dε ANKUR_AVS06SS_05
Si ETCHING IN Ar/Cl 2 : WAFER SURFACE MECHANISM Cl adsorbs on forming SiCl x passivation layer. Cl(g) + Si(s) SiCl(s), p=0.99 Cl(g) + SiCl n (s) SiCl n+1 (s) p=0.2 Ions etch passivation (for 200eV). Cl + (g) + SiCl(s) SiCl 2 (g), p=0.3 Cl + (g) + SiCl 3 (s) SiCl 4 (g), p=0.6 M + (g) + SiCl x (s) SiCl x (g) p=0.6 Etch products further passivates, creating etch blocks. SiCl 2 (g) + Si(s) Si 2 Cl 2 (s) p=0.3 SiCl 2 (g) + SiCl n (s) Si 2 Cl n+2 ANKUR_AVS06SS_06 p=0.1-0.2
Si ETCHING IN Ar/Cl 2 : WALL SURFACE MECHANISM On chamber walls SiCl 2 (g) + W(s) SiCl 2 (s) p=0.2 SiCl 2 (g) + SiCl 2 (s) (no reaction) M + (g) + SiCl 2 (s) SiCl(s) + Cl(g) p=0.6 M + (g) + SiCl 2 (s) SiCl 2 (g) + W(s) p=0.6 Passivated walls effect reactivity of Cl. Cl(g) + W(s) Cl(s) p=0.1 Cl(g) + Cl(a) Cl 2 (g) + W(s) p=0.1 Cl(g) + SiCl 2 (s) (no reaction) ANKUR_AVS06SS_07
Si ETCHING IN Ar/Cl 2 Seasoning investigated for Si etch products in Ar/Cl 2. Base case conditions: Ar/Cl 2 = 90/10, 100 sccm 15 mtorr, 300 W 50 V bias at 5 MHz Silicon etching by chlorine is the source SiCl x. Transport of SiCl x results in deposition (and further sputter/etch) on other surfaces. ANKUR_AVS06SS_08
Si ETCHING IN Ar/Cl 2 : REACTANT FLUXES Dominant ions are Ar + and Cl + due to dissociation of Cl 2. Dominant neutrals are Cl, SiCl 2 and SiCl 4. SiCl 2 is potentially reactive with surfaces; SiCl 4 is not. Ar/Cl 2 =90/10, 100 sccm, 15 mtorr, 300 W, 50 V at 5 MHz. ANKUR_AVS06SS_09
Si ETCH: ION ENERGY ANGULAR DISTRIBUTIONS Ion energies on wafer are 50-75 ev, elevated above applied voltage due to dc bias. Ion energies on other reactor walls peak at time averaged floating plasma potential (40 V). Reactivity of wafer and walls differ due to differences in threshold energies and IEDs Ar/Cl 2 = 90/10, 100 sccm, 15 mtorr, 300 W, 50 V at 5 MHz ANKUR_AVS06SS_10
SURFACE COVERAGES: WAFER Surface site coverages depend on bias voltage. Surface coverage of SiCl 2 and SiCl 3 decreases with increasing bias. More rapid removal of higher chlorinated sites. SiCl fraction increases as removal rate of higher chlorinated sites uncovers native Si. Ar/Cl 2 =90/10, 100 sccm, 15 mtorr, 300 W Surface coverages after 1 min. ANKUR_AVS06SS_11
SURFACE COVERAGES: REACTOR WALLS Surface coverage of walls depend on wafer surface kinetics. Low biases: Etch products not significant in the bulk plasma. Sites dominated by Cl(s). High biases: Large production of etch products produce higher SiCl and SiCl 2 surface coverage. Etch products surface coverage ultimately saturate with time at about 0.25-0.30. Surface coverages after 1 min. Ar/Cl 2 =90/10, 100 sccm, 15 mtorr, 300 W ANKUR_AVS06SS_12
SEASONING EFFECT: ETCH RATE Si etch for 1 min for each wafer. Etch rate in seasoned chamber decreases. Passivation of walls by SiCl 2 decreases further reactivity of SiCl 2 increasing density in plasma. Wafer SiCl 2 passivates wafer SiCl x sites forming Si 2 Cl y etch blocks. SiCl 2 (g) + SiCl n (s) Si 2 Cl n+2 (s) Ions removes Si 2 Cl y with no net contribution to etch rate. Rate of change of etch rate decreases with number of wafers; chamber wall conditions stabilize. ANKUR_AVS06SS_13 Ar/Cl 2 =90/10, 100 sccm, 10 mtorr, 300 W, 75 V at 5 MHz
SEASONED CHAMBER ETCH RATE: VOLTAGE Si etch rates decrease in seasoned chamber. With additional wafers, etch rates stabilize as chamber reaches final state. Etch rate stabilizes sooner at higher voltages. Higher etch rates and more etch products season chamber faster. Larger ion energies remove overlying Si 2 Cl n more rapidly. In spite of lower reactivity of Cl on walls (and larger Cl in plasma), etch rates decrease due to site blockage. Ar/Cl 2 =90/10, 100 sccm, 10 mtorr, 300 W ANKUR_AVS06SS_14
Si ETCH: QUARTZ SPUTTER Capacitive coupling of coils was included to facilitate sputtering of O atoms from the quartz window. Base case conditions: Ar/Cl 2 = 90/10, 100 sccm 15 mtorr, 300 W 50 V substrate bias at 5 MHz The voltage at the turns was fixed at 300 V (turn 1) 400 V (2) 500 V (3) to emulate inductive voltage drop across coil. ANKUR_AVS06SS_15
QUARTZ SPUTTER: SURFACE MECHANISM Ion sputtering of top quartz window is the source of O in the plasma. M + (g) + Quartz(s) O(g) p=1.0 at 200 ev (ε th = 60 ev) O in bulk plasma may under go electron impact reactions (e.g., ionization). Surface reaction mechanism (any occurrence of Si x Cl y ) O(g) + Si x Cl y (s) SiOCl(s) p=1.0 M + (g) + Si x OCl y (s) Si x Cl y (s) + O(g) + M(g) p=1.0 at 200 ev (ε th = 60 ev) ANKUR_AVS06SS_16
IEADs: CAPACITIVE COUPLING Ion induced sputtering of O from quartz through capacitive coupling of coils. Total IEADs peaks are near the sputter threshold of quartz. O species adsorb on the wafer, forming Si x OCl y etch blocks. Energetic ion bombardment on the wafer may remove the etch block. Ar/Cl 2 =90/10, 100 sccm, 15 mtorr, 300 W, 50 V bias at 5 MHz ANKUR_AVS06SS_17
SiOCl PASSIVATION: WAFER and METAL WALL For t < 0, ICP only with capacitive coupling produces O atoms from quartz and SiOCl etch blocks on wafer. At t=0, bias is applied, removing SiOCl etch blocks and increasing density of etch products. O atoms adsorb on etch product passivated sidewalls as SiOCl. ANKUR_AVS06SS_18
QUARTZ SPUTTER: IMPACT ON ETCH RATE Capacitive coupling (O atom sputtering) reduces etch rates at low voltages. O atoms adsorb on wafer forming SiOCl (an etch block) which is not removed. Large bias voltages increases ion energies and sputters the etch block restoring higher etch rates. Pathway to decrease in etch rate O(g) + Si x Cl y (s) SiOCl(s) ANKUR_AVS06SS_19 p=1.0 M + (g) + SiOCl(s) SiCl(s) + O(g) p=1.0 at 200 ev (ε th = 60 ev) Ar/Cl 2 =90/10, 100 sccm, 15 mtorr, 300 W
CHANGE IN CAPACITIVE COUPLING The degree of capacitive coupling from the coils was reduced by moving coils above the dielectric window. Reduced capacitive coupling reduces ion energies onto the quartz (less sputtering). ANKUR_AVS06SS_20
CAPACITIVE COUPLING: ETCH RATE Reduction in capacitive coupling Lower ion energies Sputter O atoms density lower. O atom adsorption on wafer forming SiOCl are less rate-limiting. At higher voltages, the etch rate increases due to less energy absorbing O in the plasma. Ion flux increases Ar/Cl 2 =90/10, 100 sccm, 15 mtorr, 300 W ANKUR_AVS06SS_21
CONCLUDING REMARKS Chamber seasoning was investigated in Si etch using Ar/Cl 2 plasmas. IED integrated with Surface Chemistry Model to update wall recombination coefficients; change surface loss rates. Etch rates decreased in a seasoned chamber. Seasoned reactor increases SiCl 2 flux back to wafer. Feedback of etch products (SiCl 2 ) from the plasma form Si x Cl y etch blocks. Removal of Si x Cl y does not contribute to etch rate. Sputtering of quartz window is a source of O atoms. O atoms adsorb on wafer forming etch block. Reduction of etch rates at low biases At large bias, SiOCl removed. ANKUR_AVS06SS_22