Introduction

In semiconductor technology, the achievement of a totally dry

device fabrication scheme is considered to bring a significant
improvement in parameter control, reproducibility, reliability, and
production yields. Plasma oxidation is one such dry and cold
process which allows the growth of high integrity thin film oxides
in the presence of oxygen plasma. The physical properties of
such an oxide are similar to thermally grown one, but the
advantage that it presents us with is the low temperature
processing.
Primarily, the particles in the plasma can be categorised into
positive (O+, O2+, etc.), neutral (O radical) and negative (O, O2,
O2, etc.) charge states. As shown in Fig. 11, all three types of
particles may reach the electrode. The positive ions mainly
contribute to thesputteringoftheoxide/substratesurface,as they
are repelled by the anode. It is shown in previous experiments
that oxygen radical concentration has no correlation with the
oxide growth rate.2 The mechanism proposed in most of the
literaturesuggests that the electrons attractedtowardstheanode
dissociate the O2 molecule into negative oxygen ions which then
diffusetotheSiSiO2interface.2,3

Constructionofthemodel
In plasmaanodization of silicon,therateofoxidationiscontrolled
by the flux of the oxidising species assumed to be the O ion
across the SiO2/Si interface. The rate of change of O ions wrt
timeisgivenas:
N(x,t)
2N(x,t)
N(x,t)
t = D x2 E(x, t) x + U (x, t)
where N(x,t) is the O concentration at time t and at distance x
from the SiO2plasma interface, D and arediffusion coefficient
and mobility of O ions respectively, E(x,t) is the electricfieldand
U(x,t) is therateofreactionofOionsatadistancexandattimet
asshown in Fig. 2.3 Then, the rate of oxide growth(thicknessw)
depends on the flux of O ions across the SiSiO2interfacei.e.at
x=w,whichcanbemathematicallywrittenas,
dw = 1 (w, t) ,where
dt
nO

(x, t) = D N(x,t)
x + E(x, t)N(x, t)
where nO is the number density of O atoms in SiO2, (x,t) is the
fluxofOionsatatdistancexandtimet.
In typical experimental conditions, the total constant current
density isaround 30300 mA cm2, a field E of ~106 Vcm1 and a
substrate temperature of 400600 o C. The exact electric fieldcan
be found out by solving the Poissons equation in the oxide. But
on finding out an (acceptable) approximate solution of the
Poissons equation, one can see that the gradient of the electric
field across the oxide is ~104 V cm2 which contributes negligibly
to the change in electric field over practical oxide thickness of
104cm. Moreover, the voltage across the electrodes follows the
oxidethicknessaccordingtoexperimentalobservations,which

suggests that the electric field (voltage/thickness ratio) can be

considered approximately constant throughout the oxide
thicknessforeaseofanalysis.
Next, we need to consider the loss in O concentration as they
travel through the oxide. The e concentration is almost constant
astheyzipthroughtheoxideinatimescaleof~1011s( = w/eE
) for a thickness of 104 cm. Thus, the decrease in O
concentration at the SiSiO2 interface seems a very
straightforward explanation for the decrease in oxide growth rate
i.e. dw/dt.TakingalltheOlossmechanismsintoaccount suchas
electron detachment, O vacancy traps and bulk defects in SiO2,
combination with holes formed by UV photons getting emitted in
theplasma,etc.,thereactionrateofOionscanbewrittenas,
U (x, t) = kN(x, t)
Furthermore,inanodizationexperimentsthediffusivetransportis
negligiblecomparedtothedriftduetotheelectricfield.So,we
mayaswellneglectthediffusionterm.Equatingthetimerateof
changeofOionsto0insteadystategivesustheirconcentration
profilethroughtheSiO2layer.
N(x,t)
N(x,t)
t = E x kN(x, t) = 0

TheaboveequationyieldsN(x)(showninFig.33)as,
N (x) = N oexp( xk
E ) withNo=numberofionsatx=0
fromwhich(x,t)canbecalculatedtogetdw/dtas,
dw = EN exp( xk )
o
dt
E

Solvingtheaboveequationwithinitialconditionw(0)=0gives,
Nok
w(t) = E
k ln(1 + nO t)

Challenges
Oneofthemajorchallengesthatisquitetalkedabout inliterature,
isthe problem of the damaging action of the high energy ions to
the oxide film. High energy ions may cause sputtering of oxide
film and retard film growth. To overcome this issue, we need a
precisecontroloverthebombardingenergywhichshouldbesuch
that it isnottoolowthatsurfaceactivationdoesnotoccurandnot
toohighthathighenergyionsdamageouroxidefilm.
Moreover,contaminationoftheoxidebyelectrodematerialisoften
observed especially in highvoltage DC glow discharges, even if
the sample is not directly facing the cathode. It has been
observed that the metal used as cathode plays an active role in
the quality/nature of oxide formed. A proposed solution to this
problem could be to use carbon cathode, but it is almost
impossibletoanodizeusingcarboncathodes.
Sometimes it isalsoobservedthatwhileusingalowpressureAC
glowdischargeusingeitherRF (afewMHz)ormicrowave (afew
GHz), the SiO2 grown possesses a high density of electronic
trapswhich can only be annealed at hightemperature(~1000oC)
thus negating the advantage of the lowtemperature growth
process.


Figure1.Variousparticlesimpingingonthesurfaceofthegrownoxide

Figure2.Thecoordinatexintheoxidelayeroxidethicknesswisafunctionoftime.

Figure3.O ionconcentrationprofileintheSiO2layerfortwodifferentreactiontimes.Fullline:profileatt1
dashedline:extensionofprofileatt2


[1]Gourrier,S.,andM.Bacal."Reviewofoxideformationinaplasma."Plasmachemistryandplasma
processing1.3(1981):218.
[2]Hasegawa,Isahiro,TakeshiYamauchi,andHideoSugai."MechanismofOxidationofSiSurfaces
ExposedtoO2/ArMicrowaveExcitedPlasma."Japanesejournalofappliedphysics46.1R(2007):98.
[3]Peeters,Jozef,andLiLi."Anewmodelfortheplasmaanodizationofsiliconatconstantcurrent."Journal
ofappliedphysics72.2(1992):719724.