Thin Solid Films 418 (2002) 151–155

In situ growth rate measurements for plasma processing of opaque

materials
A. Salifu, G. Zhang, Edward A. Evans*
Department of Chemical Engineering, University of Akron, Akron, OH 44325-3906, USA

Received 2 August 2002; accepted 3 August 2002

Abstract

The accurate measurement of growth and etch rates during plasma processing is difficult, if not impossible, to achieve for
opaque, three-dimensional substrates. Sensing small weight changes that correspond to discontinuous film formation on a substrate
is also difficult with traditional optical methods. Using a gravimetric recording microbalance, we have successfully measured
small weight changes of aluminum samples during the pre-cleaning and nitriding stages in low temperature plasma. The data will
show that the microbalance can provide useful information about the rates of various processes within a plasma. The data can be
interpreted using reaction–diffusion equations to determine the relevant rate parameters for a plasma process. This approach can
be extended to the study of many materials including oxides, metals, and nitrides subjected to various processes including etching
and deposition.
䊚 2002 Elsevier Science B.V. All rights reserved.

Keywords: Microbalance; Plasma processing and deposition; Nitrides; Diffusion; Plasma chemistry

1. Introduction techniques are most sensitive if the substrate is atomi-

In situ gravimetric measurements during plasma proc-
essing can provide accurate rate data for processes
occurring on many types and shapes of substrates.
Plasma processing of materials is a common technique
but it is difficult to characterize accurately. Plasma
processing involves a complex interplay between gas
phase reactions, ion dynamics, surface reactions, and
diffusion w1,2x. Accurate measurement of the parameters
that describe the reactions and transport of species within
plasma processes is critical to continued process optim-
ization w1x. To determine these parameters directly with-
in a plasma system for opaque materials is not possible
using current optical technology.
Several in situ diagnostic techniques are used to
determine growth and or etch rates during thin film
processing w3x. Each technique provides useful infor-
mation about the growth of thin films. The optical
*Corresponding author.
E-mail address: evanse@uakron.edu (E.A. Evans).
cally flat and has different optical properties than the
film being deposited. Films being deposited must be at
least semitransparent. Under these conditions, current
optical techniques can sense film thicknesses down to a
monolayer. The optical techniques are not suitable for
measuring growth rates on rough or three-dimensional
substrates with much accuracy. Optical techniques can-
not be used to sense changes that do not result in distinct
interfaces. Finally, optical techniques do not measure
thicknesses or mass directly. The data obtained from
optical techniques must be interpreted using models
based on ideal systems. The only way to measure growth
rates directly on any substrate is by using a gravimetric
technique. Gravimetric techniques can be used to deter-
mine growth rates on smooth, rough, two-, or three-
dimensional substrates with an accuracy that surpasses
the optical techniques. Gravimetric techniques are inva-
sive and are limited to low mass substrates for the
highest sensitivity. The gravimetric microbalance
approach provides a suitable in situ technique for sen-
sitive detection of growth rate.

0040-6090/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 0 4 0 - 6 0 9 0 Ž 0 2 . 0 0 6 9 6 - X
152 A. Salifu et al. / Thin Solid Films 418 (2002) 151–155

these problems we have set up a vertical, inductively

coupled system with a low aspect ratio to make direct
weight measurements during plasma processing. The
microbalance measurements can be used to determine
process parameters such as rate constants and diffusivi-
ties. These parameters can then be used to refine models
of the growth processes.

2. Experimental description

The depositionyprocessing system is an axisymmetric,

inductively coupled plasma reactor (Fig. 1). The quartz
tube reactor is 60 cm long and has a diameter of 15
cm. Both ends of the 11 l quartz tube are seated into
aluminum base plates in shallow circular grooves with
square cross section. Both ends of the quartz tube are
sealed against the aluminum end plates using a viton o-
ring around the outside diameter of the tube compressed
with an aluminum ring (Manitou Systems) (Fig. 2). All
other flanges are MDC Kwik-Flange fittings and the
tube fittings are Swagelock. The reactor is enclosed by
a second quartz tube to allow for wall-temperature
control during a run. The entire reactor is shielded to
prevent radio frequency (RF) electromagnetic radiation
leakage into the lab. The plasma power supply (ENI)
and matching network are coupled to the plasma cham-
ber through a silver plated copper coil. The power
supply can deliver up to 1.5 kW of power at 13.56
Fig. 1. Schematic of inductively coupled plasma reactor with in situ MHz. An inductively coupled plasma allows for stable
microbalance. The microbalance is used to measure substrate weight plasma formation that is not perturbed by the substrate
during a run. if the substrate is small with respect to the plasma
Gravimetric microbalances have been used for deter-
mining the kinetics of thin film deposition in thermally
activated chemical vapor deposition (CVD) processes.
Within a stable flow environment the gravimetric micro-
balance provides the most accurate and direct measure-
ment of processing rates. A gravimetric microbalance
has been used to determine the kinetics of graphite
deposition and carbon infiltration into carbon–carbon
composites w4x. Gravimetric microbalances have also
been used to study low-pressure CVD of tungsten w5x.
Harris w6,7x, Angus w8,9x, and Evans w10x have used the
gravimetric microbalance extensively for determining
the kinetics of diamond film growth. From this work a
fundamental understanding of the stages involved in
diamond growth were determined; the three stages were
observed to be carburization, film nucleation, and
growth.
The gravimetric microbalance has not been used
previously in a plasma enhanced CVD process. Plasma
systems are difficult to combine with the gravimetric
technique because the substrate must hang freely, the Fig. 2. Schematic of sealing mechanism for the ends of the quartz
tubes. Both ends of the quartz tube are sealed against the aluminum
role of the substrate in plasma stability should be end plates using a viton o-ring around the outside diameter of the tube
minimized, and the flow around the substrate should be compressed with an aluminum ring. Schematic is a radial slice show-
stable to achieve precise measurements. To overcome ing half of the symmetric reactor.
 A. Salifu et al. / Thin Solid Films 418 (2002) 151–155 153

Table 1
Operating conditions for plasma nitriding w11x

Sputter cleaning Nitriding

Gas used Ar Ar and nitrogen
Pressure (Pa) 100–133 100–133
Flow rate 10–20 sccm 10–20 sccm total
Power (W) 200 200–400
Thermocouple temperature (8C) 190 190
Substrate mass (g) 1 1

Table 2
Dimensionless groups describing the inductively coupled plasma reactor

Dimensionless Definition Physical Order

group interpretation
Knudsen lyL Ratio of mean free path to characteristic length 1=10y4
Grashof (bgL3DT)yv 2 Ratio of inertial and buoyant forces to viscous forces 1=10y3
Peclet (mass) VLyD Ratio of mass flux by convection to mass flux by diffusion 1
Rayleigh (bgL3DT)yav Ratio of thermal flux by free convection to thermal flux by conduction 1=10y3
Froude V 2ygL Ratio of inertial forces to gravity forces 1=10y4
l is the mean free path of the gas, L is the radius the reactor, b is the thermal expansion coefficient of the gas, g is the gravitational constant,
DT is the temperature difference across the radius, v is the kinematic viscosity, V is the velocity of the gas, D is the diffusivity of the gas, and
a is the ratio of the thermal conductivity to the product of the heat capacity and the density.

To determine the rate of material removal (sputtering)

Fig. 3. Weight of aluminum sample measured during plasma nitriding
using an in situ Cahn microbalance.
volume. The operating pressure, the flow rate, and the
supplied power control the plasma characteristics such
as volume, intensity, and shape. The coil length is equal
to one-quarter the wavelength of the RF power. The
length of the coil will allow for resonant mode operation
when the capacitive component of the plasma potential
is reduced using a Faraday shield. In the resonant mode
a much higher active species concentration can be
achieved over a wide pressure range. The reactor design,
therefore, allows for the generation of active species
over a wide range of pressures and temperatures.
or deposition, a Cahn D200 microbalance was used to
make direct in situ measurements of the substrate weight.
To demonstrate the utility of this system we measured
the nitriding rate of aluminum at low plasma powers.
The samples were hung directly from the balance using
either nichrome or quartz wire (Polymicro). With a
sample weight of less than 1.5 g the Cahn balance can
sense weight changes on the 0.1 mg level. Aluminum
substrates were cut from 5 mm diameter aluminum rod
or thin foil. The microbalance was used, as described
below, to record weight vs. time of the samples during
the cleaning and nitriding of the aluminum substrates.
The weight vs. time data is used to calculate rates. The
average of at least 60 rate measurements taken over a
10-min period is reported. The nitriding procedure fol-
lowed that which was described in the literature for
other plasma aluminum nitriding processes w11x; typical
conditions are given in Table 1. The aluminum sample
was first cleaned with isopropyl alcohol and then dried
in air. The sample was then hung down the axis of the
chamber. Prior to exposing the samples to the nitrogen
plasma, samples were typically cleaned in a 100 Pa
argon plasma at 200 W for 15 min. The samples were
then exposed to a 100 Pa nitrogen plasma sustained at
200 W for at least 5 min and no more than 8 h. Table
2 shows the value of the dimensionless groups that
characterize the system at these conditions.
3. Results
From Table 2 it is reasonable to assume that the flow
within the chamber is not subject to roll cells (Grashof)
154 A. Salifu et al. / Thin Solid Films 418 (2002) 151–155
Fig. 4. Nitriding rate as a function of plasma power. The ratio of
nitrogen to argon in the feed gas was 2:1. The total pressure was 128
Pa.
and that the mass transport is dominated by diffusion
(Peclet). These conditions favor stable microbalance
measurements. Fig. 3 shows the weight of a sample
with respect to time for a typical nitriding experiment.
When either the argon or the nitrogen plasma is initiated
there is an apparent rapid weight increase followed by
the onset of etching or nitriding; the apparent increase
is due to drag on the sample created by the change in
the gas flow. This apparent weight increase also occurs
during thermal processes w9x. During the argon pre-
cleaning step there is a weight loss. This weight loss
occurs for several minutes. When the sample weight
becomes stable, the argon plasma is turned off and the
system is allowed to cool. Nitrogen is then flowed into
the chamber causing a dramatic swing in the apparent
sample weight. When the nitrogen plasma is initiated
an apparent weight increase is observed. The weight of
the sample does not return to its previous value because
the pressure within the system increases slightly. In the
nitriding step, a weight gain is observed for up to 10
min. X-ray photoelectron spectroscopy analysis of the
samples shows that nitrogen is incorporated into the
aluminum sample and the nitride is formed w12x. After
the 10-min weight gain the weight levels off and there
is no more significant weight gain for up to 8 h of
nitriding at 200 W. To estimate the noise in the data,
the average and standard deviation of both the weight
and the rate of change of the weight with respect to
time is determined. The data are only considered useful
if the standard deviation is less than 10% of the average
value. For example, in a low power plasma the average
rate of deposition is on the order of 3=10y3 mgymin
with a standard deviation of 1=10y4 mgymin; using
the average density of aluminum nitride (3.3=103 mgy
cm3) and the surface area of the sample, a rate of
3=10y3 mgymin corresponds to a linear growth rate of
8=10y7 cmymin (0.5 mmyh). These measurements
demonstrate the sensitivity of the microbalance tech-
nique for three-dimensional substrates.
The microbalance data allow one to precisely deter-
mine rates of mass change, characteristic times of the
diffusion process, and the effect of process changes.
During the nitriding of aluminum in a low temperature
plasma, the microbalance data indicate that after a short
time there is a diffusion barrier formed at the outermost
surface cutting off any further supply of nitrogen. This
trend can be interpreted using reaction–diffusion equa-
tions to determine the effective diffusivity and reaction
rate constant for aluminum nitride formation w12x. At a
temperature of 190 8C we have estimated that the
diffusivity of nitrogen in aluminum is 7.3=10y13 cm2 y
s w12x.
The rate at which aluminum is nitrided depends on
the process conditions. The nitriding rate is determined
from the data taken after the large swing in the sample
weight that occurs after the nitrogen plasma has been
initiated (see the nitriding step in Fig. 1). Fig. 4 shows
the nitriding rate in a plasma with a ratio of nitrogen to
argon of 2:1. Fig. 4 shows that as plasma power is
increased the nitriding rate increases up to a power of
250 W. As power is increased above 250 W the nitriding
rate decreases. The nitriding rate goes through a maxi-
mum with respect to power corresponding to a change
in the balance of deposition and etching. The different
regimes are a result of changes in the species distribu-
tion, which results from changes in the plasma condi-
tions w13x. The nitriding rate is a function of both the
plasma power and the plasma composition. Fig. 5 shows
the dependence of the nitriding rate on composition of
the gas entering the plasma chamber. There is also a
maximum in the nitriding rate with respect to composi-
tion. Following the trend from pure nitrogen to pure
argon shows that the addition of argon increases the
nitriding rate; pure argon plasma results in an etching
environment for the aluminum plugs. Considering the
Fig. 5. Nitriding rate as a function of plasma composition. The total
pressure was 128 Pa and the power was 250 W.
 A. Salifu et al. / Thin Solid Films 418 (2002) 151–155
data in Figs. 4 and 5, it is evident that the argon in the
mixture is acting to remove the diffusion barrier, or at
least the surface species that interfere with the diffusion
process. It is also likely that argon increases the rate of
generation of active nitrogen species within the plasma
w13x.
4. Conclusion
To further refine growth processes for advanced mate-
rials will require more accurate measurements of the
parameters that characterize these processes. We have
demonstrated that in situ rate measurements can be
made in a plasma environment using a gravimetric
recording microbalance. The microbalance technique can
be used to measure the parameters characterizing these
processes for a wide range of material systems. We have
presented data that illustrate the sensitivity and useful-
ness of this approach. For plasma nitriding of aluminum,
there is a maximum in the nitriding rate with respect to
plasma power and composition. This maximum is the
result of the competition between nitriding and etching
of the sample. Analytical and numerical models will be
developed to increase the understanding of the plasma
nitriding process.
155
Acknowledgments
The authors would like to thank The University of
Akron for support of this research project through start-
up funds and an internal faculty research grant.
References
w1x M.A. Lieberman, A.J. Lichtenberg, Principles of Plasma Dis-
charges and Materials Processing, Wiley, New York, 1994.
w2x E. Meeks, P. Ho, Thin Solid Films 365 (2000) 334.
w3x D.A. Glocker, S. Ismat Shah (Eds.), Handbook of Thin Film
Process Technology, IOP Publishing, 1995, pp. D03–D09.
w4x S. Bammidipati, G.D. Stewart, J.R. Elliot, S.A. Gokoglu, M.J.
Purdy, AIChE J. 42 (1996) 3123.
w5x J.A.M. Ammerlaan, P.J. Van der Put, J. Schoonman, J. Appl.
Phys. 73 (1993) 4631.
w6x S.J. Harris, A.M. Weiner, Thin Solid Films 212 (1992) 201.
w7x S.J. Harris, A.M. Weiner, J. Appl. Phys. 75 (1994) 5026.
w8x S.P. Chauhan, J.C. Angus, N.C. Gardner, J. Appl. Phys. 47
(1976) 4746.
w9x E.A. Evans, J.C. Angus, Diamond Relat. Mater. 5 (1996) 200.
w10x E.A. Evans, Ph.D. Thesis, Case Western Reserve University
(1998).
w11x H.R. Stock, C. Jarms, F. Seidel, J.E. Doring, Surf. Coat.
Technol. 94–95 (1997) 247.
w12x R. Evans, A. Salifu, G. Zhang, E. Evans, S. Hariharan, G.
Young, Surf. Coat. Technol. 157 (2002) 59.
w13x N. Duez, B. Mutel, O. Dessaux, P. Goudmand, J. Grimblot,
Surf. Coat. Technol. 125 (2000) 79.