Introduction to Plasma Discharges for Material Processing

Preview:

Citation preview

Introduction to Plasma Dischargesfor Material Processing

Andrea Macchiwww.df.unipi.it/∼macchi

polyLAB, INFM-CNR, University of Pisa, Italy

– p.1/28

References[1] J. Reece Roth, Industrial Plasma Engineering, Voll.1–2

(Institute of Physics Publishing, 2004)[2] M. A. Lieberman and A. J. Lichtenberg, Principles of

Plasma Discharges and Materials Processing, SecondEdition (John Wiley and Sons, 2005)

[3] K. Ostrikov, “Colloquium: Reactive plasmas as aversatile nanofabrication tool”, Rev. Mod. Phys 77, 489(2005)

[4] C.-M. Chan, T.-M- Ko, H. Hiraoka, “Polymer surfacemodification by plasmas and photons”, Surf. ScienceRep. 24, 1 (1996)

[5] A. Macchi, Appunti su scariche di plasma perapplicazioni tecnologiche, and references therein(2005–, in progress!)

– p.2/28

Example 0: nature

Principle of plasma chem-istry: discharge createschemically reactive species(either ions or neutrals)Ligthning creates ozone,fixed nitrogen and light a

asee e.g. M. A. Uman, The lightning discharge (Dover, 2001)

– p.3/28

Example 1: etching on Silicon

Scheme of directionalplasma etching on the sur-face of Si, for the productionof microchips

(see [1], p.2) Result example inmonochrystalline Si(dimensions: 0.2× 4 µm)

– p.4/28

Example 2: Carbon nanostructures

Scanning electron micrograph [3] ofa) Silicon surface before hydrocarbon plasma processing

b) nanotips c) nanopyramids d) nanowalls– p.5/28

A list of possible applications- Microelectronics etching, deposition, oxidation,implantation, passivation- Liquid crystal display and solar cell depositions- Aerospace and automotive ceramic and metal coatings,films, paints- Metallurgical melting, refining, welding, cutting, hardening- Ceramics synthesis, ultrapure powders, nanopowders- Food packaging permeability barriers- Textile adhesion treatments- Medical materials bio-compatibility treatments,sterilization, cleaning- Architectural and automotive glass coatings- . . .

– p.6/28

Why plasmas?In a discharge, energetic particles (electrons or ions)drive chemical reactions with high activation energywhich may be problematic to obtain in other waysAn important issue is often the absence of thermalequilibrium: active species particles may have highenergy but the overall energy density is low (example:avoid heating surfaces to high temperatures)plasma chemistry is activated and controlled by electricfields; plasma processes are much more clean than wet(chemical) ones!

– p.7/28

Characterization of discharge plasmas

� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �

� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � �

plasmagas

V

Background gas pressure P ≈ 10−4 ÷ 1 atmFree electron density ne ≈ 108 ÷ 1014 cm−3 (Z∗ ¿ 1)Electron Temperature Te ≈ 1÷ 10 eVDevice size L ≈ 10÷ 102 cmAC frequency: ω ≈ 10 Mhz÷ 10 Ghz (rf to µWs)Surface interaction with walls (electrodes, processingmaterial)

– p.8/28

Elementary discharge model

� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �� � �

x

V

0

n

� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �

+++

x

V

Vp

0

n

λs

E

ne

Z∗ni

A neutral plasma withne = Z∗ni is created be-tween grounded walls att = 0

Electrons drift to thewalls, creating a negativesheath where ions areaccelerated

Ions bombard the walls with energy Ei = Vp(plasma potential)

– p.9/28

Sheath modelingSheath modeling (even in DC and thermal equilibriumconditions) is an old, classical, and still discussedproblem of plasma physics a

For plasma etching and nanofabrication, the desiredregime is that of non-collisional, high voltage (eVs À Te),DC sheaths, to have well-oriented ion fluxes on thesurfaceDepending on the application, modeling of AC sheaths,collisional sheaths, . . . , is desirable

a[see e.g. Riemann et al, Plasma Phys. Control. Fusion 47, 1949 (2005)]

– p.10/28

Child-Bohm’s sheath modeling

Assumptions: eVs À Te, ne ¿ ni

λs: sheath thickness V (x): electric potentialJ0: ion current V0: bias potential of the wall

λs =

(

4√2ε0

9en

)1/2

T−1/4e V

3/40

V (x) = −V0

(

x

λs

)4/3

J0 = Zen0vb = Zen0

ZTeMi

See e.g. [1], vol.I, par.9.4.5

– p.11/28

Plasma anisotropy is the key to etchingDischarge creates ions→ sheath accelerates ions→ ionscatalyze surface reaction creating volatile specie

Example: 2XeF2 + Si Ar+

→ 2Xe + SiF4 ↑

– p.12/28

Plasma treatment of surfacesThe surface energy may be modified by plasma processingin various ways (depending on exposure time and otherparameters)

plasma cleaning (attive/passive): removal of surfacecontaminantsattachment of polar groups (C=O, OH, COOH) yieldinghydrogen bondingrotation of bulk polar group by the electric fields at thesurfacesurface etching

– p.13/28

Plasma depositionPlasma Chemical Vapor Deposition (PCVD) is used toproduce thin film with special properties.In a discharge, the feedstock gas id dissociated intomolecular fragments and eventually monomers whichrecombine of the surface of the substrate (plasmapolymerization).The physical details are now well understood, butseveral “recipes” (chemical composition, plasmaparameters, reactor geometry . . . ) for thin filmfabrication have been found.

– p.14/28

Plasma-based nanofabrication

Building Units (BUs) are- generated in plasmadischarge bulk- accelerated (whencharged) in plasmasheath- transported to the sur-face- sometimes able todrive surface activation(no external heatingnecessary)

– p.15/28

The role of sheath fieldsThe electric field E in a collisionless, high-voltage sheathcontrols

the flux of ions to the surfacethe landing energy (just that required for activation)the orientation (normal to the surface)

“Lucky” effect: lines of E convergetowards sharp ends→ nano-focusing of BUsto desired sites(example: current distributionfrom simulations ofcarbon nanotips fabrication) [3]

– p.16/28

Other nanofabrication examples

++

+_

+

++

_F

F

Quantum dots produced by plasmaRF sputtering [3]

Silicon nanowiresgrown by chargednanocluster BUs ina plasma: electricalinduction favors one-dimensional growth[3]

– p.17/28

Atmospheric plasma dischargesPlasma processing at 1 atm is desirable for industrialapplications because

vacuum operation not needed: reduce costscontinuous (instead of batch) processing possible:increase productivityhigher flux of particles possible: reduce processing time

Drawbacks:

- plasma control is more complex- higher breakdown voltage is required

– p.18/28

Paschen law of breakdown

Vb =Cpd

ln[Apd/ ln(1 + 1/γ)]= f(pd)

Vb: voltage, p: pressure, d: electrode spacing,γ: secondary emission coefficent from cathode

Optimal condition forionizing collisions

eE`c = Iz ; E = Vb/d ,

`c ∝ n−1 ∝ p−1

Secondary emissionprovides positive feedback

– p.19/28

Dielectric barrier discharge (DBD)A dielectric layer (barrier ) is posedover (both) electrode(s) to:- provide a non-dissipative currentlimiter (ballast)- distribute charges over the surface

dg

dDε

Vext

J

Several DBD schemes exist:

a) simple planar cellb-c) coplanarelectrodes celld) “packed bed”reactor

– p.20/28

DBD reactors for bulk processing

Parallel plate reactorfor non-adherent websprocessing

Cylindrical annulus reactorwith rotating cylinderfor adherent web processing

– p.21/28

DBD phenomenologyThe discharge is filamentary andshort-lived:- microdischarge initiated by avalancheionization- electrons on avalanche head reachthe barrier depositing surface charge- electric field is quenched; dischargestops

++

_ _ _ _ _ _ __

++

+

+ +

+

� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �

� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �

� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �

+

_

+++

+++

++

_ ____

__ _

� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �

� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �� � � � � � � � � � � � � � �

+

_

E

Typical microdischarge parameters@ p = 1 atm:τ ≈ 10÷ 100 ns ne ≈ 1014 ÷ 1015 cm−3

Te ≈ 1÷ 10 eV rs ≈ 100 µmQ ≈ 1 pC U ≈ 1 µJE ≈ 104 V cm−1

– p.22/28

Simple model of DBD

Consider singlefilament and anequivalent cir-cuit:

+V

R(t)Cg

CD

Sc = πr2

s , Sd = πr2

0Cd = εSd/dd , Cg = εSd/dg

Deposited charge Q and absorbed energy Uass do notdepend on R(t) = ρ(t)dg/Sc:

Q ' εEb(dg/dd)Sdep , Uass =1

2

C2

d

Cg + CdV 2

min , Vmin ' Ebdg

Duration τ ' ρ̄ε(dg/dd)(Sd/Sc) ' Eb(ε/enevd)(Sd/Sc)

– p.23/28

Collective effects in DBDSpace-charge fields drive ionization waves: streamersCompetition of microdischarges for available dielectric sur-face yields spatial (anti)correlation

Interaction betweenmicrodischarges leads to 2Dpattern formation:similarity with Bènard convec-tion cells

– p.24/28

Atmospheric diffuse (“glow”) discharges- Issue: transition from filamentary to uniform discharge- Simple criterion: avalanche electron “heads” must

overlap before streamer phase

Spherical head model:rh ≈

√za`d (diffusion), Na ≈ eαza,

Ea ≈ 1

4πε0Naer2

h

.= Eb for streamers

→ yields “critical distance” zcr

Heads must overlap:ne0 > r−3

cr = (`dzcr)−3/2

ne0: “preionization” density

__ _

++ +__ _

++ +

__ _

++ +

__ _

++ +

__ _

++ +

� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �

� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �

� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �

� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � � � � � �

E

– p.25/28

An example: OAUGDPTM

The “One Atmosphere Uniform Glow Discharge Plasma”scheme has been patented and registered by Roth [1].Concept: there is a frequency range (RF) in which ions aretrapped between electrodes (planar, symmetric, capacitiveDBD configuration) and electrons are not.

x̄osc,i ≤ d/2 x̄osc,e ≥ d/2

2eV̄

miνcid2≤ ω ≤ 2eV̄

meνced2

Outside the frequency range: filamentation instabilitiesUnderlying physics (and chemistry?) needs to be studiedApplications: etching, surface treatment, sterilization, . . .

– p.26/28

Dual frequency discharges

� � � � � � � � � � � � � �� � � � � � � � � � � � � �� � � � � � � � � � � � � �

� � � � � � � � � � � � � �� � � � � � � � � � � � � �� � � � � � � � � � � � � �

� � � � � � � � � � � � � �� � � � � � � � � � � � � �� � � � � � � � � � � � � �� � � � � � � � � � � � � �V1, ω1

V2, ω2

E

Purpose: enhanced control ofion flux for etching (Lieberman)

Plasma density and ion energyare controlled indipendently

ne ∝ Pabs ∝ ω2V , Ei ∝ V = Vh + Vl

ω2

hVh À ω2

l Vl → Vh controls ne Vl À Vh → Vl controls EiDetailed scheme involves the physics of double AC sheathstructure, stochastic and anomalous skin effect heating,electromagnetic surface wave coupling of RF power to theplasma, . . .

– p.27/28

ConclusionsPlasma discharges work very well :-) . . .although we do not fully understand why :-(Recipes for specific applications can be foundCurrent research has lot of issues: empirical applicationand optimization, enhanced control of plasmaparameters, numerical modeling, understanding ofbasic physicsLet’s start the discussion!

– p.28/28

Recommended