Validation of an ICP model for Low Pressure

High Density Chlorine Plasma

A. Bhoj, K. Shah and M. Megahed
ESI Group, 6767 Old Madison Pike, Huntsville, Alabama 35806, USA

Abstract: In this paper, results from an axisymmetric inductively coupled plasma (ICP) mod-
el for a Chlorine discharge in a modified GEC reference cell are compared to experimental
measurements reported in the literature. In the model, the solution of electromagnetic fields,
induction heating, electron and heavy species transport, plasma chemistry and surface kinetics
is addressed. The radial distribution of ion densities computed at different horizontal planes in
the cell compare well with reported measurements.

Keywords: ICP, Low Pressure Plasma, Plasma simulation, Multiphysics, Chlorine Plasma,
Electronegative chemistry

1. Introduction A section through the GEC reference cell is shown in

Plasma deposition and etching of thin films is widely Fig 1. The dimensions shown are those most pertinent to
used in production of integrated circuits. Chemistries that the numerical model.
use Cl2 are among those commonly used for plasma etch-
ing. Energetic ions in these discharges play a crucial role
in the etching process. Reliable modeling tools that ad-
dress the multi-disciplinary nature of plasma discharges
can help in understanding and improving the design and
operation of such reactors.
Fluid models are commonly used in industry and ac-
count for varying species properties, flow, heat transfer
and radiation, charged particle transport and finally the
detailed chemistry of reactions taking place both in the
discharge and on the reactor walls. The current paper dis-
cusses numerical results obtained using a multi-physics
commercial fluid code, CFD-ACE+®, in a GEC reference
cell modified for inductively coupled plasma. A descrip-
tion of the GEC reference cell is given in Section 2. The
model and reaction chemistry are discussed in Section 3
and results are presented in Section 4. Our concluding
remarks are in Section 5. Fig. 1 Section through GEC Reference Cell [1, 2]

2. GEC Reference Cell 3. Model Description and Reaction Mechanism

The Gaseous Electronics Conference reference cell
configuration is well documented [1]. The inductive coil
is a five turn, planar coil constructed from 1/8 in. diameter
oxygen-free high-conductivity copper tubing, 11 cm in
diameter. The coil is separated from the plasma by a 1 cm
thick quartz window. The distance from the window to the
lower electrode is 3.8 cm. The lower electrode can be
grounded, or biased with rf or dc voltage. Due to the de-
sign of the ICP source, the clear view between the ICP
antenna and the lower electrode is approximately 13 cm
in the radial dimension and 3.1 cm in the axial dimension.
The lower electrode and inductive coil are water cooled.
The well documented setup makes it suitable for model
validation.
In this axisymmetric model, a two-dimensional compu-
tational mixed grid is used containing both structured and
unstructured elements for a total of 22588 cells. Struc-
tured grids are used in most parts of the computational
domain; within the quartz window an unstructured grid is
used as shown in Fig. 2. A finer resolution is chosen in the
discharge region between the dielectric window and the
lower electrode in order to resolve gradients in plasma
characteristics. All surfaces bounding the plasma, except
the quartz window, are metallic.
A detailed description of the equations that are solved
in this model can be found elsewhere [3 – 5] and only a
brief description of some of these is given in the follow-
ing paragraphs.
 the metallic surfaces [7] than at the quartz window sur-
face [8] by an order of magnitude. They are not precisely
known and are essentially tuning parameters.
In the next section, results from the simulations, includ-
ing the spatial distribution of power and species densities
are discussed. The radial distribution of Cl+ is compared
to experimental measurements of LIF data for the nor-
malized radial Cl+ density distribution at different hori-
zontal locations above the electrode [2]. For the simula-
tion discussed here the computational time is about 8
hours on a single processor 3 GHz machine.

Fig. 2 The model is 2-d axisymmetric and the computa-
tional grid contains mixed element types.
4. Results
The base case conditions considered are chamber pres-
sure of 2.6 Pa, 185 W inductive power and 5 sccm flow
rate for Chlorine. The transient simulation option march-
ing to steady state conditions is used. Inductive heating
leads to a peak power density of 2.2106 Wm-3 in the re-
gion below the quartz window.

For ICP plasmas, quasi-neutrality is assumed and the

electron number density is obtained algebraically from the
solution of ion densities. Charged species fluxes are
computed using the ion momentum equations for heavy
species and the drift-diffusion approximation for electrons.
The electron temperature is found from the electron en-
ergy balance accounting for the energy transfer due to
convective flux and thermal diffusion, energy absorbed by
electrons due to inductive heating and energy loss due to
Fig. 3 Inductive heating peaks below the window.
the inelastic electron impact reactions.
The ambipolar electric field in the plasma is obtained
Electrons are heated to a peak temperature of 3.7 eV in
using the quasi-neutrality constraint and current conserva-
the regions below the quartz window. The dissociation of
tion. With the ICP model, the coil current is adjusted so
Cl2 leads to an abundance of Cl radicals as shown in Fig.
that the specified target power to the plasma is attained.
4 with a peak value of 1.451020 m-3 closer to the quartz
Neutral transport is addressed using the Navier-Stokes
window due to the lower rate of surface recombination. In
equations. Since the operating pressure is fairly low, slip
this region, the ratio of Cl atoms to Cl2 (nCl/nCl2) peaks at
wall conditions were employed. A single fluid approach is
about 0.225.
used to describe the gas temperature with both the ions
and neutrals assumed to be at a single temperature and
described by a single enthalpy balance equation.
The objective here was to compare the ICP model re-
sults to experimental data as reported in the literature by
Johannes et al [2] for Chlorine discharges. The species
considered are Cl2, Cl, Cl-, Cl+, Cl2+ and electrons. The
gas-phase reaction mechanism consists of electron impact
reactions leading to dissociation, ionization and detach-
ment and heavy particle reactions including charge ex-
change reactions and ion-ion neutralization with rates
taken from Ref. 2. An additional reaction not mentioned
in [2] and considered here is the charge exchange reaction
Fig. 4 Cl density is higher near the window due to lower
of Cl+ with Cl2, taken from [6]. The surface reactions used
surface recombination to form Cl2.
for the wafer and the chamber walls are simply recombi-
nation of Cl radicals and neutralization of ions. The coef-
Due to the electronegative nature of Cl species, the
ficient of recombination of Cl radicals used was higher on
density of Cl- ions is a factor of 5 higher than electrons,
with a peak density of 3.31017 m-3. The negative ions
tend to peak closer to the region of the quartz window
where Cl radicals are in abundance as shown in Fig. 4.
Fig. 4 Cl- density peaks closer to where Cl density is high.
The background gas temperature rises to 530 K, due to
contributions from both Joule heating and the heat of
gas-phase reactions in the discharge.
Fig. 5 Gas heating raises temperature to 530 K.
The density of the dominant positive ion, Cl2+ is shown
in Fig. 6 below. In addition to electron impact reactions,
charge exchange with Cl+ increases the Cl2+ density.
Fig. 6 Density contours of Cl2+ ions are similar to Cl-.
Contours for Cl+ ion densities are shown in Fig. 7.
While the peak density is about an order of magnitude
lower than Cl2+ density, the spatial distribution in the
radial direction is remarkably different. This is further
illustrated in Fig. 8, where the experimental distributions
of normalized Cl+ density from Ref. 2 are plotted with
modeling data. The LIF experimental data in [2] has the
radial variation in Cl+ density at different horizontal
planes - 2.7 cm, 1.7 cm and 0.7 cm above the lower elec-
trode. At 2.7 cm, the normalized Cl+ density drops off by
a factor of 50 with increasing radius. This factor reduces
to about 7 at the 1.7 cm plane but increases again to about
16 at the 0.7 cm plane.
Fig. 7 Steep gradients in the density contours of Cl+ ions.
Fig. 8 Comparison of normalized Cl+ ion density to expe-
rimental data in Ref. 2.
As shown in Fig.8, modeling results slightly
over-predict experimental measurement at the 2.7 cm ho-
rizontal plane but capture them well at the 1.7 cm plane.
At the 0.7 cm plane, the results under-predict experimen-
tal data at lower radii. Overall, the model results track the
experimental trends well for these conditions.
5. Summary
Modeling results for a low pressure, inductively
coupled Chlorine discharge in a modified GEC reference
cell configuration were presented. To address the low
pressure conditions, slip boundary conditions were em-
ployed for gas flow. Ion momentum equations were
solved for transport of heavy charged species. The elec-
tronegative chemistry of this discharge results in a higher
Cl- ion density than electrons. Modeling results for the
radial distribution of Cl+ densities at different horizontal
planes compared well with available experimental data
for these conditions.
References

[1] P. Miller, G. Hebner, K. Greenberg, P. Pochan and B.

Aragon, J. Res. Natl. Inst. Stand. Technol., 100, 427,
1995.
[2] J. Johannes, T. Bartel, G. A. Hebner and J. Woodworth,
J. Electrochem. Soc., 144, 2448 (1997).
[3] S. T. Seo, Yong H. Lee, Kwang S. Lee, D. R. Yang and
B.K. Choi, Run-to-run Control of An Inductively
Coupled Plasma Etching of SiO2: Constructor of
Process Simulator with a CFD Code, ICCAS, June
2005.
[4] ESI Group, CFD-ACE+® User Manuals (2008),
www.esi-cfd.com.
[5] D. Bose, T. R. Govindan and M. Meyyappan, J. Elec-
trochem. Soc., 146, 2705 (1999).
[6] P. Subramonium and M. J. Kushner, J. Vac. Sci.
Technol. A, 20, 325 (2002).
[7] G. P. Kota, J. W. Coburn and D.B. Graves, J. Vac. Sci.
Techno. A, 16, 270 (1998).
[8] J. vanRoosmalen, J. A. G. Baggerman, S. J. H. Brader,
Dry Etching for VLSI, pp 56, Plenum Press, New
York (1991).