30th ICPIG, August 28th – September 2nd 2011, Belfast, Northern Ireland, UK
D13
Dissociation efficiency of NF3 in a capacitively coupled plasma: a
phenomenological parameterization
G. F. Leu1, P. Grünenfelder1, P. Modrzynski1
1
P
P
P
OC Oerlikon Solar, Hauptstrasse 1a, 9477, Trübbach Switzerland
One of the challenges in the industrial relevant PECVD processes is the reproducibility
requirement. A possible solution for increasing the reproducibility is the use of so called cleaning
plasma after every deposition, i.e. plasma which removes the deposited layer from the reactor
walls. Fluorine containing plasma, as for example NF3, is used in case of Silicon based coating.
The present paper investigates the dissociation efficiency in a capacitively coupled NF3 plasma.
Mass spectrometry is used to determine the partial density of plasma components in the pump
inlet. A Yasuda-like phenomenological parameterisation allows the better understanding of the
phenomena in plasma.
1. Introduction
A PECVD process will generally coat not only
the substrate but also the reactor. The next
deposition will start in this case in a different reactor
and will have different properties. The so called run
to run reproducibility can be increased if the
deposition on the reactor walls is removed before
every new process.
Fluorine containing plasmas are successfully
used for cleaning after Silicon based depositions [1].
Atomic or molecular Fluorine reacts at the reactor
walls with the deposited Silicon, forms Silicon
Tetrafluoride which is then evacuated through the
vacuum pumps. The creation rate of Fluorine atoms,
i.e. the dissociation efficiency of a cleaning plasma,
is therefore one of the most interesting parameters
from technological point of view.
The first step in optimizing a cleaning process is
the optimization of the creation of atomic Fluorine
in a not deposited reactor.
The present work analysis the dissociation
efficiency of NF3 in a capacitively coupled plasma
produced in a large area reactor.
2. Experimental setup
The experimental device is extensively described
elsewhere [1]. A rough drawing and a brief
description of the reactor are provided below (see
Fig. 1).
The reactor is an isothermal rectangular
parallelepiped with the dimension 1.4 m x 1.2 m x
0.028 m. The gas flow is controlled with mass flow
controllers (MFC) and the gas mixture is prepared
outside the reactor. The gas is let into reactor via a
gas shower to ensure a homogeneous distribution.
The upper electrode is powered at 40.67 MHz and
has a concave shape to compensate the wave effects
at this excitation frequency. The power ranges from
few hundreds of W up to 6 kW. The gas is pumped
out through symmetric lateral ports. The pressure is
controlled with a system consisting of a butterfly
valve and a pressure gauge. A quadrupole mass
analyser (QMA) was mounted at the inlet of the
pump.
MFC NF3
MFC Ar
Gas shower
Reactor
Pumping lines
Butterfly
valve
Pressure
gauge
QMA
Vacuum
pump
Fig. 1 – Experimental device
QMA curves were taken for different flows of
pure gases without plasma (NF3, N2, F2, Ar). The
30th ICPIG, August 28th – September 2nd 2011, Belfast, Northern Ireland, UK
signal intensity was then scaled to the outlet pressure
and consequently to the outlet flow of the given gas.
NF3 plasma was the produced for different
combinations Power/ Inlet flow. The butterfly valve
was kept completely open. QMA spectra were taken
and the outlet flow of NF3, F2 and N2 were inferred.
A similar parameterization is found for the
pressure at the pump inlet. The ratio between the
pressure in presence and in absence of the plasma
depends only of the composite parameter Power/
Flow.
3. Experimental results
The results of the measurements of NF3 outflow
are depicted in Fig. 2
Fig.4 – pressure parameterization
Fig.2 – NF3 outflow
As expected the NF3 outflow decreases with
increasing applied power and increases with
increasing inlet flow.
It is remarkable that measurements taken for
different flows and powers can be put together in
only one parameterized curve. On the X axis is
represented the ratio between applied power and
NF3 inlet flow, and on the Y axis the ratio between
NF3 outflow in the presence and in the absence of
the plasma.
Fig.3 – NF3 parameterized outflow
For parameter P/F < 2 W/ sccm, the increase of
the applied power or a decrease of the NF3 inlet
flow lead to an increase of the dissociation
efficiency. At about 2 W/ sccm only 20% of the inlet
NF3 can be found at the reactor outlet. 80% of the
NF3 was dissociated.
For P/F smaller than 2W/sccm the pressure
increases with the composite parameter, For P/F
greater than 2W/sccm the pressure remains almost
constant.
Similar behaviours are obtained for the other
plasma components.
The qualitative interpretation is very similar with
that one given by Yasuda [2] for his "similarity law"
in plasma polymerisation.
There are two regions in the space phase: an
energy reach and a precursor reach one. If the
discharge is in the energy reach region, the system
reached its dissociation limits; further increase of
applied power does not increase the dissociation
degree. If the plasma is in the precursor reach
region, only some of the molecules are dissociated,
an increase of the applied power leads to an increase
of the dissociation degree.
4. References
[1] D. Chaudhary et al, 24th European
Photovoltaic Solar Energy Conference, 21-25
September 2009, Hamburg, Germany
[2] H. Yasuda, Glow Discharge Polymerisation,
Journal of polymer Science/Macromolecular
Rewiews, 16, 1 (1981), 199-293